Proceedings of the 11th Italian Conference

Sensors and Microsystems

This page intentionally left blank
 Proceedings of the 11th Italian Conference

iensors and Microsystems

lecce, Italy 8-10 February

Editors

P Siciliano
CNR-IMM, Italy

S Capone
CNR-IMM, Italy

C Di Natale
University of Rome "Tor Vergata',' Italy

A Amico
University of Rome "Tor Vergata',' Italy

world scientific
NEW J E R S E Y L O N D O N S I N G A P O R E B E I J I N G S H A N G H A I H O N G K O N G T A I P E I C H E N N A I
Published by
World Scientific Publishing Co. Re. Ltd.
5 Toh Tuck Link, Singapore 596224
USA oflce: 27 Warren Street, Suite 401-402, Hackensack, NJ 07601
UK ofice: 57 Shelton Street, Covent Garden, London WC2H 9HE

British Library Cataloguing-in-PublicationData

A catalogue record for this book is available from the British Library

SENSORS AND MICROSYSTEMS

Proceedings of the 11th Italian Conference
Copyright 0 2008 by World Scientific Publishing Co. Re. Ltd.
All rights reserved. This book, or parts tliereoj may not be reproduced in any form or by any means,
electronic or mechanical, including photocopying, recording or any information storage and retrieval
system now known or to be invented, without written permission from the Publisher.

For photocopying of material in this volume, please pay a copying fee through the Copyright
Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, USA. In this case permission to
photocopy is not required from the publisher.

ISBN-13 978-981-279-338-6
ISBN-I0 981-279-338-0

Printed in Singapore by World Scientific Printers (S) Pte Ltd

 Foreword

This book contains a selection of the contributions presented and discussed

during the 11 AISEM (Associazione Italiana Sensori e Microsistemi) National
Conference on Sensors and Microsystems, held in Lecce, Italy, on February 8-10,
2006.
The Conference has been organized by the Institute for Microelectronics and
Microsystems (IMM-CNR) of the National Council of Research. In comparison
with the other editions this one has been organized having in mind the idea to
emphasize the scientific talent of the young researchers, who are considered to be
the future resources of the Science, in general, and of Sensors and Microsystems
area, in particular. To this aim, the poster and oral presentations, included
invited, after an evaluation and selection based on the scientific contents, were
strongly recommended by young researchers. Even if this edition of the
Conference could be remembered with the slogan AISEM for Younger, it will
anyway provide, as usual, a strong opportunity for the presentation of updates
results to both theoretical and applied research in the field of Sensors and
Microsystems bringing together scientists from academic institutions and
industry.
We would like to thank all the people who gave their contributions for the
successful organisation of the Conference, in particular the members of the
Steering and Scientific Committees of AISEM and mainly the members of the
Local Scientific and Organising Committees belonging to the Sensors and
Microsystems staff of IMM-CNR in Lecce.
Particularly thanks are due to the various institutions and companies, which
sponsors the event giving their support for the organisation of the Conference.

Pietro Siciliano
Simonetta Capone
Corrado Di Natale
Arnaldo DAmico

V
This page intentionally left blank
 THE llTw NATIONAL CONFERENCE
ON SENSORS AND MICROSYSTEMS
Lecce, Italy
8-10 February, 2006

Conference chair
Pietro Siciliano, IMM-CNR ,Lecce

STEERING COMMITTEE SCIENTIFIC COMMITTEE

A. DAmico Univ. Roma Tor Vergata F. Davide Telecom Italia
L. Campanella Univ. Roma LaSapienza A. Diligent1 Univ. Pisa
P. De Gasperis CNR-IMM, Roma C. Di Natale Univ. Roma Tor Vergata
C. Mari Univ. Milano L. Dori CNR-IMM, Bologna
G. Martinelli Univ Ferrara G. Faglia Univ. Brescia
U. Mastromatteo ST Microelectronics C. Malvicino CRFIAT, Orbassano (TO)
A.G. Mignani CNR-IFAC, Firenze M . Mascini Univ. Firenze
M. Prudenziati Univ. Modena B. Morten Univ. Modena
G. Sberveglieri Univ. Brescia G. Palleschi Univ. Roma Tor Vergata
P. Siciliano CNR-IMM, L a c e F. Villa ST Microelectronics
G . Soncini Univ. Trento M. Zen ITC-IRST, Trento

LOCAL ORGANISATION
Local scientific committee Local Organisation committee
Simonetta Capone Francesco Buccolieri
Cosimo Distante Angiola Forleo
Mauro Epifani Luca Francioso
Roberto Rella Alessandro Leone
Andrea Baschirottto Maria Grazia Manera
Jolanda Spadavecchia
Antonella M. Taurino
Dominique S. Presicce
Marzia Zuppa

vii
This page intentionally left blank
 CONTENTS

CHEMICALS SENSORS

F. Damato, P. Mazzinghi, S. Viciani, P.W. Werle

Tunable Diode Laser Spectrometers (TDLSs) as Airborne In-situ Sensors for
Stratospheric Trace Gases..................................................................................... 1

D. Monti, G. Ercolani, R. Paolesse, M. Scarselli, P. Castrucci, M. De Crescenzi

Self-organization of Tetraphenylporphyrinon HOPG. Unprecedented Double-
Layer Pattern Driven by Adsorption Compression. ............................................ ..6

A. Bonavita, G.Neri, G. Micali, G. Riuo, N. Pinna, M.Niederberger

A Room Temperature Pt/ln203-basedOxygen Sensor........................................ 13

E. Sgreccia, A. Macagnano, E. Dalcanale, C. DiNatale, R. Paolesse

A Study of Mass Effect in a Cavitand Supramolecular Sensor ........................... 18

A. Cusano, A. Iadicicco, P. Pilla, L. Contessa, A. Cutolo, S. Carnpopiano,

M. Giordano
Polymer Coated Long Period Gratings: New Perspectives as High Sensitivity
Chemical Sensors................................................................................................ 23

M. Consales, A. Cutolo, A. Cusano, S. Carnpopiano, M. Penza, P. Aversa,

G. Cassano, M. Giordano
Optical Fiber Sensors for Hydrogen Detection at Cryogenic Temperatures .......28

G. Barillaro, A. Diligenti, L. M. Strambini

Silicon pn Mesa Junction with a Porous Layer as Sensing Element ..................33

A. De Girolamo Del Mauro, L. Quercia, F, Loffredo, G. Di Francia

Influence o f Solvent on Thin Film Composite Morphologies and Sensing
Properties ............................................................................................................ 38

G. Neri, A. Bonavita, G. Micali, F. A. Deorsola, I. Amato, B. De Benedetti

Gas Sensing Properties of Nanostructured Titania: A Comparative Study
of Different Preparation Methods ...................................................................... .43

ix
X

S. Kandasamy, A. Trinchi, W. Wlodarsky, E. Comini, G. Sberveglieri,

L. Pandolji, S. Kaciulis
Hydrocarbon Sensing Device Based on Binary Metal Oxides ............................ 48

A. ladicicco, D. Paladino, A. Cutolo, A. Cusano, S. Campopiano, M. Giordano

Multi-Cavity Structured FBGs: Towards Advanced Chemical Sensing
Devices ............................................................................................................... 53

A.G. Mignani, L. Ciaccheri

Belgian Beer Mapping and Digital Fingerprinting using Color and Turbidity
Assessment......................................................................................................... .58

C. Cucci, A.G. Mignani, C. DallAsta, G. Galaverna, A. Dossena,

R. Marchelli, R. Pela
A Portable Fluorometer for the Rapid Screening of M1 Aflatoxin in Milk ........63

A.G. Mignani, L. Ciaccheri, A.A. Mencaglia, R. Paolesse, C. Di Natale,

A. Del Nobile, R. Benedetto, A. Mentana
Quality Monitoring of Extra-virgin Olive Oil using an Optical Sensor ..............68

R. Bernini, F. Mottola, A, Minardo, L. Zeni

Spectral Interrogation of Optical Metal-Cladding Waveguides or Chemical
Sensing.............................................................................................................. ..73

M. Buzzolani, M. Butturi, C. Malagh, G. Martinelli, B. Vendemmiati

Application of Quality Standard for the Planning, Design and Realisation
of Thick Film Gas Sensors .................................................................................. 78

M. Pisco, M. Consales, A. Cutolo, A. Cusano, S.Campopiano, M. Giordano,

M .Penza, P. Aversa, G. Cassano
A Photonic Bandgap Fiber Sensor Based on Single Walled Carbon
Nanotubes as Sensing Material .......................................................................... .81

L. Longo, G. Vasapollo, R. A. Picca, C. Malitesta

Synthesis and Selective Nucleoside Recognition of a New Substituted
Zinc-phthalocyanine........................................................................................... .86

V. Biagiotti, F. Valentini, D. Moscone, G. Palleschi

Synthesis and Characterization of Nanomaterial as Sensitive Platform to
Assemble Selective Chemical Sensors for NO2 Detection .................................. 91
 xi

S. Capone, P. Siciliano, A.M. Taurino, I. Hotovy, V. Rehacek

Modified p-type NiO Thin Films using Pt- and Au-overlayers for Hz Sensing...97

V. Marotta, S. Orlando, B. Mitu

Multilayered Metal Oxide Thin Film Gas Sensors Obtained by Conventional
and RF Plasma-assisted Laser Ablation ............................................................ 102

M. Toneuer, A. Quaranta, G. Della Mea, G. Maggioni, S. Carturan

A New Competitive Technique for the Production of Porphyrin Thin Films:
The Glow Discharge Induced Sublimation (GDS)............................................ 107

S. Carturan, M. Bonafini, G. Maggioni, A. Quaranta, M. Tonezzer,

G. Della Mea, G. Mattei, C. de Julidn Fernandez, P. Mazzoldi
Optical Sensing to Organic Vapors of Fluorinated Polyimide Nanocomposites
containing Silver Nanoclusters ......................................................................... 112

PHYSICAL SENSORS

P. Bruschi, N. Nizza, M. Piotto, M. Schipani

Solid State Directional Anemometer for Harsh Environments ......................... 117

C. Calaza, N. Viarani, G. Pedretti, M.Gottardi, V. Zunini, A. Simoni

A Focal Plane Array based on CMOS Uncooled Thermoelectric Infrared
Detectors ........................................................................................................... 122

A. Baschirotto, E. Melissano, P. Siciliano, E. Dallago, P . Malcovati,

M. Marchesi , G. Venchi
A 2-D Planar Microfluxgate with Sputtered Core............................................. 127

L. Ciccarelli, M. Medugno
An Optical Probe for RF Electric Field ............................................................ 132

S. DAddio, A. Cusano, A. Tutolo, S. Campopiano, M. Giordano,

M. Balbi, S. Balzarini
Coated Fiber Bragg Grating As High Sensitivity Hydrophone ......................... 137

M. A. Ferrara, L. Sirleto, G. Messina, M. G. Donato, L. Rotiroti, I. Rendina

Strain Measurements in Porous Silicon by Raman Scattering .......................... 142
 xii

C. Ambrosino, P. Capoluongo, D. Davino, C. Visone, A. Cutolo,

A. Cusano, S. Campopiano, M. Giordano
Fiber Bragg Gratings and Magnetic Shape Memory Alloys: A Novel High
Sensitivity Magnetic Sensor.............................................................................. 147

F. Mancarella, A. Roncaglia, M . Passini, G. C. Cardinali, M. Severi

A Computationally Efficient Model for Thermal Sensors Simulation ..............152

L. Rossi, G. Breglio, A. Irace, P. Spirit0

Characterization of a Point-Wise Close Electric Field Sampling System
Exploiting the Electro-Optic Effect .................................................................. 157

D. Caputo, G. de Cesare, A. Nascetti, R. Scipinotti

On The Realization of Chromium Silicide Stress Sensor.................................. 162

BIOSENSORS

F. Olimpico, M. Pighini, S. Bernardi, A. Scarpa and S. Greco

Nanogravimetric Immunodetection of Gliadin by pLibra Balance...................167

R. Bernini, E. De Nuccio, A. Minardo, L. Zeni, P. M. Sarro

Integrated Silicon Micro Flow Cytometer based on Hollow Arrow
Waveguides...................................................................................................... .I72

S. Kaciulis, A. Mezzi, A. Galdikas, V. Bukauskas, A. Mironas, A. setkus,

V. Luurinavitius, R. MeSkys, J. Razumiene
Properties of Heme c Containing Enzyme Layer Self-assembled on Planar
Metallic Electrodes ........................................................................................... 177

D. Caputo, G. de Cesare, A. Nascetti, R. Scipinotti, R. Negri, A. Tarquini

Detection of Labelled DNA Based on Amorphous Silicon Devices .................182

L. Campanella, E. Martini, D. Lelo, M. Tomassetti

Carbarnate, Triazinic and Benzotriazinic Pesticide Analysis using an
Inhibition Tyrosinase Organic Phase Enzyme Sensor...................................... .187

D. Dell 'Atti, M . Minunni, S. Tombelli, M. Mascini, A. Cavazzana,

M. Zuvaglia, G. Bevilacqua
DNA-based Piezoelectric Biosensor for Human Papilloma Virus (HPV)
Detection........................................................................................................... 193
 xiii

M. Del Carlo, M. Mascini, D. Compagnone, A. Visconti

User Friendly Electrochemical Hand-Held Device for Durum
Wheat Safety..................................................................................................... 198

L. Campanella, E. Martini, M. Tomassetti

Determination of Immunoglobulin G in Human Serum by Means of an
Immunosensor fitted with an Enzymatic Probe as Detector and a Clark
Electrode as Transducer.................................................................................... 203

M.G. Manera, J. Spadavecchia, R. Rella, A. Leone, D. DellAtti,

M. Minunni, M. Mascini
SPR Imaging Technique: A Study on DNA-DNA Biological Interactions.......209

MICRO- AND NANO-TECHNOLOGY: MATERIALS AND PROCESS

R. Ainu, G. Barlocchi, M. Cattaneo, P. Corona, A. Fischetti, M. Marchi,

U. Mastromatteo, M. Palmieri, M. Pesaturo, F. Villa, A. Vinay
Lab-on-chip Integrated Genetic Analysis.......................................................... 217

S. Zampolli, I. Elmi, G.C. Cardinali, M. Severi, E. Dalcanale, P. Betti

Use of Cavitands as Selective Adsorbing Materials in GC-like Air
Quality Monitoring Microsystems .................................................................. ..222

L. De Stefano, I. Rea, L. Rotiroti, K.Malecki, I. Rendina, L. Moretti,

F. G. Della Corte
Fast Optical Detection of Chemical Substances in Integrated
Silicon-glass Chip ........................................................................................... ..227

M. Scandiuuo, D. Stoppa, A. Simoni, L. Pancheri, G.-F. Dalla Betta

CMOS Single Photon Avalanche Diode for Imaging Applications ..................232

M. Bellei, R. Gaddi, A. Gnudi

Design of a MEMS Test-Structure for Stress Characterization by
Capacitive Read-out........................................................................................ ..237

A. Castaldo, E. Massera, L. Quercia, G. Di Francia

Controlling Sensing Properties of Composite Polymers using Different
Conductive Nanoparticles............................................................................... ..242
xiv

W. Vastarella, M. De Leo, J. Maly, A. Masci, L.M. Moretto, R. Pilloton

Biosensors Based on Nano Electrode Ensemble and Screen Printed
Substrates......................................................................................................... .247

G. Barillaro, A. Nannini
Buried Microchannels by ElectrochemicalEtching of Silicon......................... .252

F. Siviero, N. Coppedk, L. Aversa, M. Nardi, A. Pallaoro, T. Toccoli,

R. Verucchi, S. Iannotta
Gas Sensing Properties of Metal-oxidelorganic Hybrid Materials Grown
by Supersonic Beam Deposition ....................................................................... 257

V. La Ferrara, I. Nasti, E. Massera, B. Alfano, G. Di Francia

Focused Ion Beam in Gas Sensor Technology.................................................. 262

F. Mancarella, A. Roncaglia, S. Cristiani, G. C. Cardinali, M. Severi

Process Dependence of Doped Polysilicon Thermal Conductivity for
Thermal MEMS Applications .......................................................................... .267

M. Grassi, P. Malcovati, S. Capone, L. Francioso, P. Siciliano, A. Baschirotto

Gas Sensing System Consisting in MOX-based Microsensors Interfaced
to a Novel Integrated 5-decade Dynamic Range Front-end .............................. 272

V. La Ferrara, I. Nasti, E. Massera, G. Di Francia

Polymeric Replicas from Porous Silicon Template (and Its Response
to Vapour)......................................................................................................... 278

C. Bonomo, L. Fortuna, P. Giannone, S. Graziani, S. Strazzeri

A Software Tool for the Simulation of IPMC Actuators and Sensors...............282

R. A. Picca, C. Malitesta, D. Manno

Synthesis and Characterization of Imprinted TiOz Nanoparticles.
Preliminary results ........................................................................................... .287

M. Schipani, T. Ungaretti, P. Bruschi, N. Nizza

A Single Chip General Purpose Circuit for Sensor Interfacing and
Characterization............................................................................................... .292

F. Dalena, G. Roselli, P. Flora

A Three-Axis Low-g Ultracompact Linear Accelerometer .............................. .297
 xv

F. Lo Castro, A. DAmico, G. Brambilla

Vehicle Health and Comfort Level Measurement: A Preliminary Study ..........302

F. Lo Castro, P. Bisegna
Forced Damping Vibration of a Cantilever Beam............................................. 307

W. Vastarella, M. Ilie, L. Nardi, A. Masci, R. Pilloton, E. Cianci,

S. Quaresima, A. Coppa, V. Foglietti
Improved Capillaries Connection to a Continuous Flow Glass
Micromachined Micro-cell ............................................................................... 3 12

SENSOR ARRAY AND MULTISENSORIALSYSTEMS

S. Centi, S. Luschi, I. Palchetti, M.Mascini

Simultaneous Detection of Polychlorinated Biphenyls (PCBS) by Means
of an Electrochemical Disposable Multiarray Sensor ....................................... 3 17

G. Pioggia, F. Di Francesco, G. Campobello, G. Cannath, N. Donato,

A. Bonavita, G. Micali, G. Neri
Integrated Microgravimetric Set-up for Chemical Sensing............................. ..322

S. De Vito, E. Massera, L. Quercia, G. Di Francia

In-Situ Monitoring of Volcanic Gases at Solfatara Crater with Hybrid
e-nose ............................................................................................................... .328

M.C. Carotta, E. Ferrari, S. Gherardi, C. Malag2, M. Piga, G. Martinelli,

A. Rosignoli
Monitoring of Airborne Fine Particles and NOx Beside a Major Road ............334

G. Ferri, V. Stornelli, F. Ruscitti, C. Falconi, A. DAmico

A Dynamic-element Matched CCII for Smart Sensors ................................... ..339

C. Moriconi, M. Santoro
Multisensor Layout for Robots in Hostile Environments.................................. 344

SENSOR NETWORK AND DATA ANALYSIS FOR SENSORS

D. Blasi, V. Cacace, L. Casone, M. Riuello

Wireless Sensor Networks Overview, Looking at the ZIGBEE Architecture...349
xvi

M. Conti, E. Gregari, G. Valente, G. Anastasi, C. Spagoni

Motes Sensor Networks in Dynamic Scenarios: A Performance Study
for Pervasive Applications in Urban Environments ........................................ ..354

M. Rizzello, A. Gentile, F. Versori

Portable Software Tools for Configuring a IEEE 802.15.4 WSN
with Mobile Devices ......................................................................................... 359

M. Cicioni, A. Scorzoni, F. Alimenti, P. Placidi, L. Roselli, S. Zampolli,

I. Elmi, G.C. Cardinali, M. Severi, S. Marco, J.M. Gdmez, F. Palacio,
B. Mauolai, A. Mondini, V. Raffa, V.Mattoli, P. Dario, R. Ingles Bort,
J. L. Ramirez, E. Llobet, E. Abad, T. Becker
Low-Power Sensor Front-End Electronics With WID Communication
Capabilities For Food Logistic Datalogging Applications.............................. ..364

M. Medugno
A Neural Spectral Classifier for Optical Sensors .............................................. 369

G. Barillaro, P. Bruschi, R.G. Garroppo, S. Giordano, A. Molfese, F. Pieri

Design of a Wireless Sensor Network for Urban Pollution Monitoring ...........374

M. Zuppa, A.M. Taurino, C. Distante, D. Presicce, S. Capone,

L. Francioso, P. Siciliano
Application of Multi-sensor Miniaturized System for Olive Oil Evaluation.....379

G. Ferri, V. Stornelli, A. De Marcellis F. Mancini

Low Voltage Low Power Improved CCII Based Interface for the Measure
and Heating Control of Resistive Sensors........................................................ .384
 TUNABLE DIODE LASER SPECTROMETERS (TDLS'S) AS
AIRBORNE IN-SITU SENSORS
FOR STRATOSPHERIC TRACE GASES

F. D'AMATO, P. MAZZINGHI, S. VICIANI, P.W. WERLE

CNR - Istituto Nazionale di Ottica Applicata,L.go E. Fermi 6
50125 Firenze, Italy

In this paper two spectrometers will be described, based on tunable diode lasers, suitable for
operation on board of a stratospheric aircraft. These instruments are used to cany out
measurements about production, transformation and transportation of atmosphenc
constituents. The principles of operation, the environmental constraints, the technological
solutions and some results will be shown.

1. Introduction
The knowledge of our atmosphere is a fundamental issue for both
understanding some important phenomena, like greenhouse effect, ozone
depletion, climate changes, and adopting the necessary countermeasures. While
most of the antropogenic emissions occur close to the ground, some natural
important productions of gaseous species are distributed all along the height of
the troposphere: volcanic emissions of HC1 and HF, nitrogen oxides from
lightening, etc. A major role in the distribution of the gaseous species in the
atmosphere is played by the tropopause, the thermal inversion layer which
divides the troposphere from the stratosphere. There is presently one aircraft
only, which can carry a payload of 1500 kg up to a height of 21000 m. It's the
Myasischev M55 "Geophysica", a Russian plane built for military purposes, and
then transferred to a private company and equipped to host several instruments,
based on different principles, which can measure different atmospheric
parameters (dust, water content, ozone, NO,, CFC's, etc.). There are many
constraints for devices to be mounted on board of such an aircraft:
1. The power supply on board is either 28 VDC or 110 VAC, 400 Hz. There
are limitations on the available power, whose stability is not the same as in
one's laboratory.
2. Pressure ranges from 1000 to 50 mBar, temperature from -70 to +50 "C.
3. All instruments must pass the EM1 test suitable for this kind of platform.

1
2

4. Mechanical design must keep into account vibrations, and instruments

must pass severe vibration tests.

2. The instruments
Both instruments are based on the Beer-Lambert law:

,
I,,, = I,, eWaL

where IOU,,
I,, are the powers of the light beam respectively at the exit and at the
entrance of the sample, L is the length of the sample and:

a=S.g(v).n.

In (2) S is the strength of the absorption, g is the absorption profile vs

fiequency and n is the density of the absorbing molecules. Once v and L are
fixed, the density n can be retrieved from the absorption measurements,
provided that temperature and pressure are known. The optical pathlength
should be maximized, but a compromise must be found with the transmission
and the dimensions of the multipass cell. When the absorption is of the order of
1%, the detection technique of direct absorption can be easily employed, when
the absorbance is lower than this value it is usual to adopt more effective
detection techniques, in order to enhance the S/N ratio. Now we enter into the
details of the two instruments, for which two quite different layouts have been
selected.

2.1. Measurement of CO
The starting point for the measurement of CO is the analysis of its inkared
absorption spectrum. This species has its fundamental absorption band around
2140 cm-' (4.67 pm). In this region lead salts diode lasers are available, and the
main interferences come from water, carbon dioxide and nitrogen oxide ( N 2 0 ) .
The use of lead salts lasers results in a serious constraint to the experimental
apparatus. This sources require cryogenic temperatures for operation, and the
same holds for the detectors in the middle infiared. The simplest solution is a
Dewar, filled with liquid nitrogen, containing both laser and detector. In order
to ensure the stability of the boiling point of nitrogen, the Dewar must be
pressurized. The optics layout of this instrument is shown in Figure 1 [l]. The
white disk in the center of Fig. 1 is the cover of the Dewar. The laser beam
exits fiom the upper leR side. Part of the beam is reflected by a beam splitter.
 3

Figure 1. Photo of the optical layout of the CO analyzer.

This beam reaches a reference detector at room temperature, aRer passing

across a cell filled with pure CO and a ZnSe etalon, for relative frequency
reference. The main part of the beam enters a multipass cell (36 m pathlength),
and then is focused onto a detector, placed in the same Dewar. The air flow
through the multipass cell is driven by a small pump, which operates only at
altitudes greater than 5000 m, in order to avoid blowing dust or raindrops into
the cell.
The detection technique is direct absorption. The laser frequency is
scanned across an absorption line at 1 kHz repetition frequency. The signals
from the two detectors are digitized with 12 bit vertical resolution, and
averaged over 2000 scans. The data processing is carried out off-line, after
landing.
A typical result, obtained during a measurement campaign in Brazil in
JanuaryFebruary 2005, is shown in Fig. 2. Such results are related to those
coming from other i n s t ~ e n t sdetecting
, ozone, NO,, etc.

easurement of ~ e t ~ a n e
The instrument for methane detection has been designed [2] by keeping into
account some specific constraints: the volume and the shape of the available
space, the maximum weight and power consumption.
4

Figure 2a. CO mixing ratio vs time during a flight in Aragatuba (BR) in February 2005.

temperature (K)

co mixing ratio (*lo)

Figure 2b. Profiles of CO mixing ratio and temperature w t h height. The horizontal line corresponds to
the tropopause.

For the limitations above it was not possible to have a pressurized vessel
and to use liquid nitrogen, so that a room temperature diode laser had to be
employed. The optical setup is very similar to that of the CO analyzer but, due
to the low strength of the absorption lines in the near inffared, a more effective
 5

detection technique, the so called two-tone fiequency modulation [3], had to be

adopted. A typical result is shown in Fig. 3.
Aracatuba, Geophysica Flight (Fl). 01/02/05

Figure 3 CH4 mixing ratio vs time during a fllght in Araqatuba (BR) in February 2005.

4. Conclusions
Diode laser based analyzer have proved to be usehl for fast and sensitive
detection of trace gases in the lower stratosphere. The data obtained during the
measurement campaigns are being used for environmental analysis.

5. Acknowledgments
This activity has been supported by the European Community through contract
TROCCINOX.

References
G. Toci, P. Mazzinghi, B. Mielke and L. Stefanutti: "An Airborne Diode
Laser Spectrometer for the Simultaneous Measurement of H20 and HN03
Content of Stratospheric Cirrus Clouds", Opt. & Las. in Eng. 37, pp. 459-
480 (2002)
F. D'Amato, P. Mazzinghi and F. Castagnoli: Methane analyzer based on
It

TDL's for measurements in the lower stratosphere: design and laboratory

tests", Appl. Phys. B 75, 195-202 (2002)
G.R. Janik, C.B. Carlisle, and T.F. Gallagher: "Two-tone fiequency-
modulation spectroscopy", J. Opt. SOC.Am. B3, pp. 1070-1074 (1986)
 Self-organization of Tetraphenylporphyrin on HOPG.
Unprecedented Double-Layer Pattern Driven by Adsorption
Compression

Donato Monti, Gianfranco Ercolani, Roberto Paolesse

Dipartimento di Scienze e Tecnologie Chimiche, Universith di Tor Vergata, Via
della Ricerca Scientifica, Roma, Italy. E-mail: monti @stc.unirorna2.it

Manuela Scarselli, Paola Castrucci, Maurizio De Crescenzi

Dipartimento di Fisica, Universith di Tor Vergata, Via della Ricerca Scientijica,
Roma, Italy.

Summary
Combined Scanning Tunneling Microscopy (STM), X-Ray photoelectron spectroscopy
(XPS), Reflection Electron Energy Loss Spectroscopy (REELS) studies have been
performed ~n sttu under ultra high vacuum condition (UHV), on tetraphenylporphyrin
(HzTPP) molecules sublimated on highly oriented pyrolitic graphite (HOPG). The results
obtained indicate the formation of a well-ordered two-layer structure of HzTPP molecules.
The first adlayer appears as a densely packed uniform background and closely bound to the
substrate as indicated by XPS and REELS measurements. STh4 images of the second layer,
lying above the first one, consists of HzTPP molecules arranged according to a quasi-
hexagonal lattice with a periodicity of 3.0 nm. This loose unusual arrangement has been
investigated by molecular mechanics modeling of the system (MMFF94 force field). The
computational results suggest that the second layer, lacking the driving force provided by
the interaction with the substrate, would be less densely packed. In particular, each molecule
of the second layer, with respect to the arrangement of the first layer, would have its nearest
neighbors missing so as to avoid adsorption compression. Accordingly the periodicity of the
second layer would be twice that calculated for the first layer. The MMFF94 calculations
match perfectly with the observed periodicity lending support to the title hypothesis.

1 Introduction
The quest for morphologically defined molecular assemblies, with controlled
size and shape, onto solid surfaces, is a mandatory issue not only for
fundamental advancements in materials science [ 11 but also for the
development of molecule-based nano-devices and electronics [ 2 ] . In this
respect porphyrin derivatives and related congeners are playing an
increasingly important role as molecular components of optical devices [3],
chemical sensors [4], and functional supramolecular materials [ 5 ] . Several
strategies can be followed for the deposition of porphyrins on solid surfaces
with controlled features; mostly important are self-assembly of monolayers
on gold [6], Langmuir-Blodgett or Langmuir-Schaefer techniques [7], liquid-

6
 7

phase deposition [8] and vacuum sublimation [9]. Scanning tunneling

microscopy (STM) is one of the preferred experimental techniques to
directly investigate the adlayers, as it allows direct assessing of the position,
the orientation and the packing arrangement of single or complex molecules
deposited on solid substrates and the
correlation of the morphological C ~ H ~
information with their electronic
properties [lo]. To the best of our
knowledge, previous studies dealing with
porphyrin adlayers, involved either simple
metalloporphyrins] or free-base porphyrin
derivatives bearing ad-hoc functional
groups, such as peripheral alkyl chains CsH
[8,11], or carboxylic substituents, which
exert stabilization and ordering by Van der Waals forces or hydrogen
bonding, respectively. We focused on the study of the simple 5,10,15,20-
tetraphenyl-2 lH,23H-porphine (i.e. tetraphenylporphyrin, HZTPP hereafter),
sublimated under UHV conditions on HOPG, through in situ STM, X-ray
photoelectron spectroscopy (XPS), and Reflection Electron Energy Loss
Spectroscopy (REELS). The results, reported here, are discussed in the light
of molecular modeling calculations that nicely agree with the information
obtained from STM images.

2. Results and Discussion

Porphyrin molecules deposited on
HOPG by UHV sublimation have been
initially studied by STM to investigate
spatial arrangement. At very low
exposure time, only isolated molecules
have been imaged. By increasing the ,
quantity of deposited H2TPPs extended
areas on the graphite surface result
modified by the presence of ordered
molecular adlayers extending over areas
of hundreds of squared nanometers
(STM image in Figure 1). The mean
diameter (1.8 nm) and height (0.35 nm)
of each bright spot have been evaluated by averaging the values obtained
from profile plots like that reported in Figure 1. The spot diameter matches
perfectly the distance of 1.77 nm separating the hydrogens in the para
positions of opposite phenyl rings of the H2TPP structure, as obtained by
hybrid density-functional theory geometry optimization carried out at the
B3LYP/6-31G(d) level of theory 112,131. The periodicity of the molecules in
8

the adlayer turns out to be 3.0 nm, indicating that the Van der Waals surface
of each molecule does not come into close contact with those of its nearest
neighbors, the minimum separation distance being larger than about 1 nm.
XPS investigation after H ~ T P P
sublimation was mainly focused
on the binding energy of 1s
core level of nitrogen, as
reported in Figure 2, evidences
two main features at 399.5 eV
and 397.6 eV, for the pyrrole
type nitrogens and for the
imine-type ones, respectively,
as previously reported for
similar free-base porphyrin I
derivatives [ 141. The energy 405 400 395
Binding Energy (eV)
difference between the N l s
pyrrole type and imine-type energies is about 1.9 eV. An additional structure
has been detected at 401.6 eV that can be associated to a strong interaction
between carbon atoms of the graphite surface and the HzTTP molecule. This
assumption can be supported by comparing the reflection energy loss spectra
obtained from clean and modified graphite, as reported in Figure 3-Inset.
The most significant n* and B* transitions of graphitic carbon result
significantly modified after molecules deposition [ 151. In particular the
original energy position of the two transitions located at about 5.8 eV and
23.9 eV in the HOPG spectrum are shifted at 6.7 eV and 19.6 eV,
respectively. The REELS spectrum of the graphite after H2TPP deposition
shows an additional n-n* feature located at around 3.0 eV (Figure 2 , Inset).
This new contribution, which is strongly related to the transition involving
HOMO and LUMO molecular states of the porphyrins, can be associated to
the optical absorption, usually referred as Soret band, that for these
molecules occurs at around 2.8 eV. In summary these results indicate the
formation of a well-ordered two-layer structure of HzTPP molecules on the
HOPG surface. The first layer that appears as a uniform background in the
STM experiments, about 0.5 nm thick, would be closely bound to the
substrate. Although STM failed to give any structural detail of this layer, it
showed the ordering of the second layer. In particular it showed that the
constituent HzTPP molecules are arranged according to a quasi-hexagonal
lattice with a periodicity of 3.0 nm indicating that a minimum distance of
about 1 nm separates molecules.
These findings, that to the best of our knowledge has not precedents in the
porphyrin literature, is surprising; it poses the question of how molecules that do
not come in close contact can form such a regular two-dimensional lattice. In
order to rationalize the observed results we modelled the system by molecular
 9

mechanics (MM) calculations by using the Merck Molecular Force Field

(MMFF94) developed by Halgren [16]. Our working hypothesis was based on
the following reasoning: close-packing of the first layer of HzTPP molecules on
the graphite surface must be driven primarily by molecule-substrate interactions
(adsorption energy) and secondarily by molecule-molecule interactions (two-
dimensional crystallization energy). One would normally expect that both
interactions are attractive, however if attractions of the molecules to the surface is
very strong, molecules are forced to pack to reach greater densities despite this
would lead to repulsive forces among them (adsorption compression). This
arrangement would affect the packing of the second layer. If attractions of the
molecules to the first layer are not as strong as those to the substrate, molecules
would lack the driving force to form a second layer as compressed as the first
one. This would result in a second layer having the same pattern of the first one,
in which, however, the nearest neighbours of each molecule are missing so as to
avoid compression (Figure 3).
We first modelled the
adsorption of one HzTPP
molecule on a graphene sheet
of the shape of a rhombus
consisting of 30x30 fused
benzenic rings by the
MMFF94 force field. The
calculated MMFF94 binding
energy is -33.2 kcal mol-'.
This value, which is significantly large, could be even underestimated. Indeed
MMFF94 evaluates intermolecular interactions by Van der Waals and
electrostatic interactions [ 161, ignoring, as all the MM methods, possible orbital
interactions between molecules, which would provide an additional stabilization.
In fact X P S and REELS measurements have shown that significant orbital
interactions occur between adsorbed HzTPP and graphite, indicating an
additional electronic stabilization. Considering that the adsorption energy of an
isolated molecule is -33.2 kcal mol-', the two-dimensional crystallization energy
of the four molecules is about -10 kcal mol-', showing favourable attractive
interactions among them. Of course such a model is devoid of adsorption
compression because the surface available to the molecules was in no way
limited. Successively we modelled the absorption of one molecule of HzTPP
above the compressed unit cell. We carried out a constrained optimization in
which the positions of the atoms constituting the compressed unit cell and the
graphene layer were frozen, whereas the HzTPP molecule constituting the second
layer was free to vary its geometry and position [12]. The calculated binding
energy of the HzTPP molecule constituting the second layer is -26.8 kcal mol-',
which is, as expected, smaller than that of a molecule of the first layer (-33.2 kcal
10

mol-'). Apart from the obvious limitations of molecular mechanics, the

accordance between the calculated periodicity of the assembly and that measured
on the STM images is exceptionally good. Indeed the distance between the
porphyrins of the second layer, calculated as twice the distance separating the
porphyrins in the compressed layer below, is 3.02 nm to be compared with the
experimental distance of ca. 3.0 nm. This accordance strongly supports the
hypothesis of adsorption compression of the first layer of HzTPP molecules as
the ultimate reason for the observed ordering of the second layer. The calculated
vertical distance of the second layer from the graphene sheet is 0.94 nm which is
in fair accordance with the average thick of the two layers (background + spot

height) of ca. 0.85 nm obtained from STM profile plots (Figure 1). The
optimized final structure of the system is shown in Figure 4.

Fig. 4. Optimrsed structure of the first layer (tube) and the second layer (space filling) of HzTPP
molecules adsorbed on graphite (not shown for clarity).

3 Experimental
STM Studies. Freshly cleaved highly oriented pyrolitic graphite has been
introduced in the UHV chamber (base pressure low lo-'' Torr) and vacuum
sublimation has been performed by heating a tungsten filament previously
dipped in a 1% H2TPP chloroform solution. During the sublimation cycles,
the base pressure was kept as low as 10'' Torr and the H2TPPs deposition
rate was estimated from the exposure time and partial pressure, calibrated
during preliminary experiments. The sample was characterized in situ by X-
Ray photoelectron spectrosocopy (XPS), Reflection Electron Energy Loss
Spectroscopy (REELS) and scanning tunneling microscopy (STM). Scanning
tunneling microscopy imaging was performed at room temperature by using
an Omicron-STM system. Electrochemically etched tungsten tips were
employed for the experiments. STM calibration was done by comparing
 11

images of molecular adsorbates with atomically resolved ones of graphite.

All STM images were unfiltered apart from rigid plane subtraction.
Molecular Modeling Calculations. The Merck Molecular Force Field
(MMFF94) developed by Halgren was the method of choice as its efficiency
in modeling porphyrin systems had already been remarked in the literature
[ 171. MMFF94 calculations were run in vacuo, with default parameters and
convergence criteria as implemented in the Spartan molecular modeling
package (Wavefunction, Inc., Irvine, CA), at the B3LYP/6-3 1G(d) level of
theory [13,18]. Graphite was modeled by a graphene sheet of the shape of a
rhombus consisting of 30x30 fused benzenic rings with dangling bonds
terminated by hydrogen atoms (formula C1920H122). Two-dimensional lattices
were built with the program GaussView 3.09 (Gaussian, Inc., Pittsburgh,
PA). Hybrid density-functional calculations were carried out with the
Gaussian 03 program [ 181.

References
1. C. F. J. Fad, M. Antonietti, Adv. Muter., 2003, 15, 673.
2. J. M. Tour, Acc. Chem. Res., 2000, 33, 791; (b) C. Joachim, J. K.
Gimzewski, A. Aviram, Nature, 2000,408,541.
3. M. P. Debreczeny, W. A. Svec, M. R. Wasielewski, Science, 1996,274,584.
4. R. Paolesse, F. Mandoj, A. Marini, C. Di Natale, Porphyrin-Based Chemical
Sensors in Encyclopedia of Nanoscience and Nanotechnology, Nalwa, H.S.,
Ed., American Scientific Publishers CA, 2004, 9,21.
5. C. F. Van Nostrum, R. J. M. Nolte, Chem. Commun., 1996,2385.
6. (a) I. Goldberg, Chem. Commun., 2005, 1243; (b) J.-M. Lehn, Science,
2002, 295, 2400; (c) D. N. Reinhoudt, M. C. Crego-Calama, Science, 2002,
295,2403.
7. A. Ulmann, An Introduction of Ultrathin Organic Films, from Lungrnuir-
Blodgett Films to SelfAssembly, Academic Press, San Diego, CA, 1991.
8. X. Qui, C. Wang, Q. Zeng, B. Xu, S. Yin, H. Wang, S. Xu, C.Bai, J. Am.
Chem. SOC., 2000,122,5550.
9. T. Yokoyama, S. Yokoyama, T. Kamidado, Y. Okuno, S. Mashiko, Nature
2001,413,619.
10 Wiesendanger (Ed.), Scanning Probe Microscopy and Spectroscopy,
Cambridge University Press, Cambridge, 1994.
11 S. De Feyter and F. C. De Schryver J. Phys. Chem. B 2005,109,4290.
12. A. D. Becke, J. Chem. Phys. 1993,98,5648.
13. G. Polzonetti, A. Ferri, M. V. Russo, G. Lucci, S. Licoccia, R. Paolesse, J.
Vac. Sci. Technol. A 1999,17, 832.
14. A. Hoffman, G.L.Nyberg, J. Liesegang, Phys. Rev. B 1992, 45,567.
15. M. Gouterman, The Porphyrins, 1978, 3, 1, Dolphin D., Ed., Academic,
N.Y.
12

16. T. A. Halgren, J. Comput. Chem. 1996,17,490.

17. a) X. Huang, N. Fujioka, G. Pescitelli, F. E. Koehn, T. R. Williamson, K.
Nakanishi, N. Berova, J. Am. Chem. SOC. 2002, 124, 10320. (b) G.
Pescitelli, S. Gabriel, Y. Wang, J. Fleischhauer, R. W. Woody, N. Berova, J.
Am. Chem. SOC.2003,125,7613.
18. Gaussian 03, Revision B.O1, Gaussian, Inc., Wallingford CT, 2004.

Acknowledgments
The Authors wish to thank MIUR-FISR "Sensori Ottici ed Elettroottici per
Applicazioni Industriali ed Ambientali" (Project SAIA 2003), and MIUR-FIRB
(Project No: RJ3NEOlKZZM) for funding.
 A room temperature Pthn203-based oxygen sensor

A. Bonavita, G.Neri, G. Micali, G. Rizzo

Dept of Industrial Chemistry and Materials Engineering, University of Messina,
C/da di Dio, Vill. S.Agata,
98166 Messina, Italy
[email protected]

N. Pinna
Martin-Luther- Universitat Halle- Wittenberg, Institut fur Anorganische Chemie,
Kurt-Mothes-Strasse 2
06120 Halle (Saale), Germany

M. Niederberger
Ma-Planck-Institute of Colloids and Interfaces, I4424 Potsdam, Germany

Summary
In this work the effects of Pt-doping and thermal treatments on nanostmctured Ina03
powders obtained via nonaqueous sol-gel method were studied with the aim to promote the
room temperature sensing characteristics towards oxygen of resistive sensors bascd on thick
films of Pt-doped nanostructured indium oxide powders.
The tested devices have shown very promising sensing Characteristics for the O2
monitoring at room temperature. A synergic effect between the small grain size of In203 and
the presence of Pt as dopant, opportunely activated via a thermal trcatmcnt, is likely at the
origin of the behaviour observed.

1 Introduction
Oxygen sensing in the low temperature range find its main applicative fields
on medical-diagnostic and waste managements. In despite of the obvious
advantages given by the use of resistive devices at low working temperature,
principally represented by low costs, simple use and assembly and longer
lifetimes, few works are reported in literature concerning the application of
semiconductor-baseddevices for application at near room temperature [ 1-31.
In this work we present a study on the room temperature response towards
oxygen of sensing devices based on Pt-doped nanostructured indium oxide.
In203is reported to be sensitive to many gases. It is a wide band-gap material,
Eg-3.70 eV, whose microstructural, electrical and sensing characteristics have
been well studied [4-71. When prepared in oxygen-deficient form, it reaches n-
type doping level due to oxygen vacancies. Further, in order to enhance the
sensitivity and get lower operating temperature, In203 can be opportunely

13
14

prepared in a nanostructured form and/or doped with suitable metallic promoters

[&I. The promoting effects on gas sensing characteristics obtainable by reducing
the grain size in the nanometric order of magnitude and by doping with noble
metal are, in facts, well known. Here we report a study focused on indium oxide
nanopowders prepared via a nonaqueous sol-gel method and successively doped
with Pt by impregnation, focusing our attention on the effect of metal doping and
thermal treatments.

Experimental
tal
In203 nanopowders were synthesised via a nonaqueous sol-gel method
involving the solvothermal reaction of indium (111) isopropoxide with benzyl
alcohol [8]. The dopant, Pt, has been dispersed on the surface of the
semiconducting metal oxide by wetness impregnation, contacting the In203
nanopowders with the proper amount of an aqueous solution of H2PtC16. The
preparative and doping routes are schematically reported in Figure1 .

Fig.1 Preparation and doping routes of In20, nanopowders.

The nanopowders have been finally dried and fired at controlled temperature
(between 150 and 450 "C) in air.
XRD and I-LRTEM characterization of the synthesized powders were carried
out by a Italstructure diffractometer mod. APD2000 and a Jeol JEM 2010
microscope.
Sensing tests were preformed on devices obtained by mixing 1n203or Pt-
doped In203 powders with water to create a paste which was successively
screen-printed on standard alumina substrates ( 3 x 6 m 2 ) supplied with comb-
like electrodes and a Pt-heater. The devices were dried and then treated in air for
one hour in order to stabilize the film texture and microstructure.
 15

Resistance measurements were carried out by a Agilent 34970A data

acquisition switch unit computer linked, maintaining the sensors under
controlled stream of dry nitrogen, and pulsing different 0 2 concentrations
ranging from 2.5 to 20%.

3 Results and Discussion

Previous results of microstructural characterization showed a narrow grain
size distribution for the In203powders synthesized via the nonaqueous sol-gel
method. The mean grain size, calculated by the Sherrer equation and confirmed
by TEM analysis (Fig. 2), resulted around 25 nm. The successive impregnation
and thermal treatments don't modified the In203 microstructure. After the
thermal treatment at 350 "C also occurred the decomposition of the Pt precursor
salt on the surface of the semiconductor oxide: H2PtC16 -+Pt + H2 + 3C12, as
well visible in the TEM micrograph.

Fig.2 TEA4 image of the 3%Pt- In203 powder calcined at 350 "C.

Sensing tests have shown that at room temperature pure In203is not sensitive
to oxygen, and Pt addition is necessary to promote the response to 0 2 . Figure 3
shows the response dynamic and the calibration curve relative to tests carried out
at room temperature with the Pt-doped sensor. Response and recovery times are
around 15 min.
16

I
0 40 80 120 160
Time (min)

Fig.3 Transrent response at room temperature (a) and calibration curve (b) ofthe Pt InzOj-bused
sensor

Sensing tests have also shown that the pretreatment temperature strongly
influence the sensor response. Sensitivity towards oxygen at room temperature
has shown a maximum on the sensors treated at 350 "C (Fig. 4). On the basis of
characterization results this was directly related to the formation on the surface
of the indium oxide grains of small metallic platinum particles.

10 I . , .
, . , , .
, . ,. , . .
100 150 200 250 300 350 400 450 5 0
Annealing Temperature ("C )

Fig.4 Effect of the annealing temperature ofthe Pt/ln203film on the response to 20% of0,.
 17

Conclusions
Nanostructured In203 prepared via a non aqueous sol-gel method is a
promising material for low temperature O2 monitoring. However to promote the
response at room temperature doping with Pt is necessary. The annealing
temperature at which occurs the decomposition of the Pt precursor influences
strongly the sensing performances of Pt/ InzO3. Therefore an oxygen sensing
device operating at room temperature with better performances respect to the
existing ones has been fabricated by using Inz03 nanopowders opportunely
activated with Pt.

References
[I] F. Chaabouni, M. Abaab, B. Rezig, Sensors and Actuators B 100 (2004)
200-204
[2] Y. Hu, 0. K. Tan, J. S. Pan, X. Yao, J. Phys. Chem. B 108 (2004) 11214-
11218.
[3] Y. Hu, 0. K. Tan, W. Cao, W. Zhu, Ceram. Int. 30 (2004) 1819-1822.
[4] M. Bender, N. Katsarakis, E. Gagaoudakis, E. Hourdakis, E. Douloufakis, V.
Cimalla and G. Kiriakidis J. Appl. Phys. 90 (2001) 5382.
[5] A. Gurlo, M. Ivanovskaya, N. Barsan, M. Schweizer-Berberich, U. Weimar,
W. Gopel, A. Dieguez, Sens. Actuators B 44 (1997) 327.
[6] E. Comini, M. Ferroni, V. Guidi, G. Faglia, G. Martinelli and G.
Sberveglieri, Sens. Actuators B 84 (2002) 26.
[7] K. Zakrzewska, Thin Solid Film 391 (2001) 229; G. Korotcenkov,
A.Cerneavschi, V. Brinzari, A. Comet, J. Morante, A. Cabot, J. Arbiol, Sens.
Actuators B 84 (2002) 37.
[8] M.I. Baraton in Nanostructured Materials: Science and Technology, Nato
Advanced Study Institute, Kluver Academic Publisher, Dordrecht, (1998),
303
[9] N. Pinna, G. Neri, M. Antonietti, M. Niederberger, Angew. Chem. Int. Ed.
43 (2004) 4345.

Acknowledgments
The authors gratefully acknowledge the financial support for this work by
MIUR under the framework of the FIRB-SQUAREproject.
 A study of mass effect in a cavitand supramolecular sensor

E. Sgreccia, A. Macamano
lstituto per la Microelettronica e Microsistemi-Roma, CNR, Via Fosso del
Cavaliere 100, 00133, Roma
[email protected]

E. Dalcanale
Dipartimento di Chimica Organica ed Industriale, Universitir di Parma, Parco
Area delle Scienze 17 A, 431 00, Parma

C. Di Natale
Dipartimento di Ingegneria Elettronica, Universitd Tor Vergata, Via del
Politecnico 1, 00133, Roma

R. Paolesse
Dipartimento di Scienze e Tecnologie Chimiche, Universita Tor Vergata, Via
della Ricerca Scientifica, 00133, Roma

Summary
The sensing properties of functionalized cavitand have been studied by thin-films
coating TSMRs by measuring their responses towards analytes having different
functional groups. We have studied sensing features in term of sensitivity and
selectivity and the contribute of viscoelasticity to responses by two growing thickness
layers

1 Introduction
In the last few years a great variety of absorbent materials have been
synthesized and investigated to improve chemical sensors performances and to
develop someone new in order to monitoring specific environments. The
structure of a generic chemical sensor can ideally be divided in two subunits: the
sensing material and the transducer. The sensing material interacts with the
chemical species present in the environment by changing some of its
physicochemical properties, while the transducer transforms these variations into
a readable signal, generally an electrical signal. The sensor sensitivity,
reproducibility and selectivity strictly depend on sensing materials properties.
Advances in supramolecular chemistry offer many opportunities to design
and prepare molecules endowed with binding site where multiple interactions can

18
 19

cooperate to selectively bind a given class of substances; in particular cavitands

are molecules where multiple binding interactions are encoded in a single
cavitand receptor.
In previous work [13, resorcinarene framework was functionalized, introducing
alkyl chains at the lower rim and phosphate or phosphonate substituents at the
upper rim, and was used as sensitive film High selectivity towards short-chain
linear alcohols was obtained for molecules having an Hbond acceptor P=O
group, oriented inward with respect to the cavity, as a bringing unit at the upper
rim of receptor. The purpose of the present work has been to study the sensing
features, especially the selectively towards alcohols, of resorcinarene molecule
having four P=O groups oriented inward increasing the film thickness.
Sometimes mass sensors responses is due not only to the mass effect, but
also to the viscoelastic properties. In fact mass sensors response is due not only
to the mass effect, but also to the film viscoelastic properties. Viscoelastic
materials respond to external forces in intermediate manner between elastic solids
and viscous liquids, so an acoustic wave that travels in a coated device
penetrates into the adjacent (sensitive) film and translates and deforms the film.
TSMR reacts to sensitive film deformation as a frequency shift, so it is not yet
possible to distinguish between mass and viscoelastic effect [3]. However,
electrical parameters analysis can help us to know sensitive film elastic
characteristics, especially the plot of the phase of electrical impedance.

2 Experimental
The molecular structure of the compound used as sensing material is reported
in Figure I Resorcinarene molecules were functionalized by introducing alkyl
chains at the lower rim and four P=O groups, oriented inward with respect to the
cavity, at the upper rim of receptor.

Fig1 Resorcinarene molecure functionalized by four P=O

groups, oriented inward with respect to the cavity
20

Sensors were fabricated using AT-cut quartzes (Electroquarz, Milano) having

fundamental frequency of 20 MHz. Two amount of sensitive films (10 kHz and 60
kHz frequency variation) were deposited by spray casting on both sides of QMBs
fiom M CHC13 solution. In this way mass transducers sensors have been
obtained, where the frequency shift is linearly correlated to the mass loading
according to the Sauerbrey equation [ 2 ] :
Af = -k,&n
The sensors have been exposed to alcohols (ethanol and isopropanol),
aromatics (benzene and toluene), amines (triethylamine) and hydrocarbons
(hexane) (reagent grade Aldrich).

3 Results and Discussion

Exposure to ethanol vapours shows (Figure 2 (a)) a Langmuir-like responses;
the shape of the curves is the result of a complexation between the analyte and
the cavitand receptor. The interaction between the film and the analyte doesn't
change with the mss increasing but enhance responses and sensitivity (Figure 2
(b)). Sensitivity ( S = (a(Af)/ac)) has been calculated, at 50 ppm concentration,
fitting the response curves by the Freundlich model

(where k , is the Sauerbrey costant of the quartz, Mwis the molecular weight, NA
is the Avogadro number, c is the concentration, Gand n are the Freundlich
parameters).

Fig.2 (a) Sensors responses versus increasing ethanol concentration; (b) Sensitivity of
different thick layers to ethanol calculated at S O ppm concentration

Experimental results show a strong contribute of the mass effect for other
volatiles too (Figure 3 and Figure 4): higher responses have been obtained not
 21

only to alcohols, but also to aromatics, hydrocarbons and amines (Figure 3).
Responses to amines are very high respect to ones other volatiles and sensitivity
normalized to molecular weight (l), calculated at 500 ppm concentration, to amines
is comparable to one short-chain alcohols (ethanol). Mass increasing doesnt
change sensitivity pattern (Figure 4), but increases its value; so the mass
increasing of sensitive film improves sensors performances.

-1 7

Fig.3 Response curves of I 0 kHz and 60 kHz mass deposited versus volatiles

Fig.4 Sensitivity (normalized to molecular weight) at 5 0 0 p p m concentration

Sensitive film viscoelastic properties have been investigated by HF8753A

Network Analyzer, measuring reflection coefficient. Reflection coefficient (S, J is
related to electrical impedance by
22

where Zois 50 Q.
Figure 5 shows the phase of the electrical impedance; the shape of this
parameter doesnt change with the mass loading and the maximum value doesnt
change too. The increasing of the mass loading slightly reduces the width of the
resonance, probably its due to the deposition technique. However mass loading
doesnt introduces viscoelastic effect, so the responses observed in previous
figures are due only to interactions between cavitands and vapours.

Fig.5 Phase of electrical impedance before and after sensitivefilm deposition

4 Conclusions
The introduction of four P=O group at the upper rim of resorcinarene
framework have been enhanced responses to alcohols, especially to linear short-
chain alcohols. In the same time, the modification of the molecule introduced a
worse selectivity: sensitivity to amines are comparable to ones alcohols.
The growing of thickness of the sensitive film doesnt introduce the
increasing of dispersion interactions, so the shape of curves doesnt change, but
the responses are enhanced for every volatile examined. For this reason the
sensitivity values are increased, while the pattern of its doesnt change.
The increasing of the thickness doesnt introduce viscoelastic effects; in fact
electrical parameters and electrical impedance are slightly diminished, so the
results obtained are due only to the interactions between sensitive film receptors
and volatiles .

References
[l] R. Paolesse, C. Di Natale, S. Nardis, A. Macagnano, A. DAmico, R. Pialli, E.
Dalcanale, Chemistry - A European Journal, 9 (2003), 5388
[2] F. Sauerbrey, Zeitschrift fur Physik, 155 (1959), 206
[3] R. Lucklum, P. Hauptmann, Sensors and Actuators B, 70 (2000), 30
[4] R. Lucklum, C. Behling, P. Hauptmann, Sensors and Actuators B, 65 (2000), 277
 Polymer Coated Long Period Gratings:
New Perspectives as High Sensitivity Chemical Sensors

A. Cusano, A. Iadicicco, P. Pilla, L. Contessa, A. Cutolo

Engineering Department, University of Sannio, Benevento Italy
a.cusano(~unisa~Inio.i~

S. Campopiano
Departmentfor Technologies, University of Naples Parthenope, Napoli, Italy

M. Giordano
Institute of Composite Biomedical Materials, National Research Council,
Napoli, Italy

Summary
In this work, long period gratings coated with polymeric high refractive index (HRI) thin
layers are proposed as high sensitive opto-chemical sensors with tailored sensitivity profile.
The deposition of thin coatings with high refractive index (compared to the cladding one)
induces strong changes in the field distribution related to the cladding modes. The main
effect is a significant change in the sensitivity characteristic to surrounding refractive index
(SRI) and to overlay refractive index. Differently from the standard LPG, where the
sensitivity to SRI exhibits an exponential form, in HRI coated LPG the sensitivity shows a
resonant-like profile. Since the resonant-peak can be centered around the selected SRI by
acting on the overlay features (refractive index and thickness), this effect can be efficiently
used to realize high performance opto-chemical sensors. In order to investigate the sensor
performances the SRI sensitivity of attenuation bands related to different cladding modes
have been experimentally tested. Here &form Syndiotactic Polystyrene (sPS) was selected as
nanoscale HRI polymeric layer.

1 Introduction
In-fiber long period gratings (LPGs) which couple light from the core mode
to cladding modes have been widely studied for applications in fiber-optic
communications such as band rejection filters [l]. On the other side, their high
sensitivity to temperature and strain and surrounding refractive index (SRI) has
led to the development of LPG-based sensors [2-31.
Although a great effort has been carried out to improve the sensitivity to SRI
of the LPG by several researcher groups, the LPG sensor exhibits a monotonic
behaviour reaching the maximum value as SRI approaches the silica index [4].
In the last years, new configurations, based on LPGs coated with nano-size
overlay with refractive index higher than the fiber one have been widely
investigated [5-71. As an HRI overlay is cast along a standard LPG, a lowering
of the modes bounding factor is induced, leading to strong changes in the field

23
24

distribution related to the cladding modes. In particular, for a proper

combination of the HRI overlay features (thickness and refractive index) and
SRI the lowest order cladding mode (LPoz)becomes guided within the overlay
and a re-organization of all the cladding modes occurs. As matter of fact, the
HRI coated LPG exhibits a new spectral behaviour, in terms of sensitivity
profile, depending on the overlay features (thickness and refractive index) and
SRI. For fixed overlay features, the SRI sensitivity exhibits a resonant-like shape
versus the SRI. In addition the resonant-peak can be tuned around the desired
SRI range by acting on the overlay refractive index or thickness. On the other
side, a similar behaviour can be achieved with regards to overlay features when
SRI is fixed. In this paper an exhaustive experimental characterization of the
changes in sensitivity profile versus SRI as direct consequence of the HRI
coating, is presented. The new sensitivity characteristics make the HRI coated
LPG an attractive structure to realize high sensitivity chemical-sensor.

2 Theoretical approach
As well known, the transmission spectrum of a standard LPG consists of a
series of attenuation bands related to the power coupling of the several cladding
modes. Since the cladding modes are bounded by the cladding surrounding
medium interface, the LPG exhibits high sensitivity to SRI. In addition, in
standard LPG the sensitivity increases as the SRI approaches the silica index and
as the cladding mode order increases.
When azimuthally symmetric nano-scale HRl coatings are deposited along
LPGs, changes in the distribution of cladding modes occur, depending on the
layer features (refractive index and thickness) and on the SRI. For a fixed
overlay thickness and refractive index, the effective refractive indices of the
cladding modes increase with the SRI [7] until a critical value is reached, where
the neffrelated to the lowest order cladding mode (LPoZ) overcomes the refractive
index of the cladding. Successive increase of the SRI leads the LPO2mode to be
guided in the overlay area exhibiting an evanescent behaviour in the core and
cladding area. As direct consequence a re-organization of the other cladding
modes aimed to recovery the original field distribution in the fiber occurs. This
means that the effective index of the ithmode shifts to match the index of the (i-
l)thmode.
As matter of fact, the effect in the transmitted spectrum is the total
weakening of the attenuation band related to the LPO2cladding mode and a quasi
simultaneous shift of the attenuation bands due to higher order cladding modes:
the peak loss related to cladding mode LPo3 shifts to spectral position of the
attenuation band associated to LPoz before the mode transition and so on.
Since high shift in the attenuation band are achieved in small SRI range,
strong changes in the sensitivity profile are expected. In particular, the
sensitivity characteristic of every attenuation bands exhibits a resonant-profile
 25

centered on the SRI responsible of the LPoz transition, leading to a high

sensitivity gain.
The key feature of the HRI coated LPG relies on the possibility to tune the
SRI range where the transition occurs by acting on the overlay thickness. In fact,
since a thicker overlay induces an increase of the effective refractive indices of
every cladding mode, the transition region is reached for lower SRI.
From these considerations, a properly coated LPG can be the base structure
to realize high performance in-fiber chemical sensors.

-E
1500.
V

V
V

Surrounding Refractive Index

Figure I Sensitivity to the SRI of the LPOb LPo7 and LPo8cladding mode in the bare LPG.

3 Experimental results
For the experimental test aimed to prove the sensitivity changes as direct
consequence of the HRI coating, a standard LPG with pitch of 340pm was
selected. The sensitivity related to the LPos, LP07 and LPo6 cladding modes are
plotted in fig. 1, showing the standard monotonic behaviour versus the SRI
where the maximum value is reached for SRI approaching the silica index.
Here, the Syndiotactic Polystyrene (sPS) in &form, demonstrating a
refractive index of 1.58 as proved in previous work [7], is used as HRI coating.
To this aim, a simple experimental setup for test and coating deposition, reported
in detail elsewhere, was used [6-71.
Fig. 2 shows the transmitted spectra of a 140nm coated LPG for different
values of SRI. As observable, the quasi simultaneous transition of the LP07 and
LP06 cladding mode is evident. In particular as the SRI reaches the critical value,
in this case approximately 1.44, a larger shift in all the attenuation bands central
wavelengths modulated by decreasing in the amplitude of the attenuation bands
peaks occurs. In addition, after the transition all the attenuation bands exhibits a
blue shift able to quasi completely recover the spectral configuration before the
transition.
Due to the reorganization of the cladding modes, the sensitivity to SRI of the
central wavelengths of the attenuation bands exhibit a totally new shape as
shown in fig. 3. As it is observable, the sensitivity of every attenuation bands in
the HRI coated LPG exhibit a resonant-like shape centered at SRI where the
26

transition is reached. The experimental results reveal a sensitivity gain, in the

transition region, of 4, 7 and 10 for the LPos, LPo7 and LPo6 cladding mode,
respectively. To better understand the dependence of the resonant-like sensitivity
to the related cladding mode the fig. 4 plots the resonant-peaks and the related
SFUs versus the investigated cladding mode.

IV n=l 431 VII n = l . 4 6

I
1250 1300 1350 1400 1450
Wavelenght [nm]
Figure 2. Transmission spectra of a 140nm sPS coated LPG f o r dvferent values of SRI in the range
I 33-1.472. The LPOJcladding modes were marked with JI, JII, ..., JVIl where the apices are referred
to different SRI.

- 4000

3000
Y

1.34 1.36 1.38 1.4 1.42 1.44 1.46

Surrounding Refractive Index
Figure 3 Sensitivity to the SRI of the LPo6,LPo7and LPos cladding modes in I40 nm HRI coated
LPG.

' " " " " 0._

c5>
x .>
22
to LP06 LP07 LP08

._ 1 , 4 4 4 1
2c
1.435
1.43
LP06 LP07 LP08
Figure 4 (a): Maximum sensitivity and (b) SRI related to maximum sensitivity versus cladding mode.
As matter of fact the sensitivity strong increases with the cladding mode
while the SRI related to the maximum sensitivity slightly changes.
 27

Based on these results the HRI coated LPG could be the base structure for
new LPG based chemical sensor. It could be efficiently used to realize high
sensitive refractive index measurements in the selected SRI range by a proper
chosen of the coating features. On the other side, since high sensitivity to
overlay refractive index is achieved in the transition region, as demonstrated in
previous work [S-91, this structure can be successfully used for selective
chemical sensor as a sensitive material is selected as HRI coating.

Conclusion
In this work, an experimental characterization of the HRI coated LPG aimed
to investigate the changes in the sensitivity has been carried out. The presence of
a HRI coating along a standard LPG leads to a new profile of the sensitivity to
SRI demonstrating a resonant-like shape. As matter of fact strong sensitivity
gain in a selected SRI range can be efficiently achieved by a proper chosen of
the overlay features. Based on the attractive sensitivity characteristic this
structure can be the base for high performance chemical sensor.

References
[I] K. W. Chung, S. Yin, Analysis of widely tunable long-period grating by use of an
ultrathin cladding layer and higher-order cladding mode coupling, Opt. Lett. 29,
812-814,2004.
[2] V. Bhatia, Applications of long-period gratings to single and multi-parameter
sensing, Opt. Express 4 , 4 5 7 4 6 , 1999.
[3] C. Y. Lin, L. A. Wang, and G. W. Chern, Corrugated Long Period Fiber Gratings as
Strain, Torsion, and Bending Sensors, J. Lightwave Technol. 19, 1 159-1 168,2001.
[4] S.W. James and R.P. Tatam, Optical fibre long-period grating sensors:
characteristics and applications, Meas.Sci. technol. 14, R49-R61, 2003.
[5] I. Del Villar, I.R. Matias, F.J. Arregui and P. Lalanne, Optimization of sensitivity in
Long Period Fiber Gratings with overlay deposition, Opt. Exp. 13, 56-69,2005.
[6] A. Cusano, A. Iadicicco, P. Pilla, L. Contessa, S. Campopiano, A. Tutolo, M.
Giordano, Cladding Modes Re-Organization in High Refractive Index Coated Long
Period Gratings: Effects on The Refractive Index Sensitivty, Optics Letters, Vol. 30,
No. 19, pp. 2536-2538, October 2005.
A. Cusano, A. Iadicicco, P. Pilla, L. Contessa, S. Campopiano, A. Cutolo, M.
Giordano, I Mode Transition in High Refractive Index Coated Long Period
Gratings, Optics Express 19, Vol. 14, No. 1, January 2006.
A. Cusano, P. Pilla, L. Contessa, A. Iadicicco, S. Campopiano, A. Tutolo, M.
Giordano, High Sensitivity Optical Chemo-Sensor Based on Coated Long Period
Gratings for Sub ppm Chemical Detection in Water, Applied Physic Letters, Vol. 87,
234105,2005.
[9] A. Cusano, A. Iadicicco, P. Pilla, L. Contessa, S. Campopiano, A. Cutolo, M.
Giordano, G. Guerra, Coated Long Period Fiber Gratings as High Sensitivity Opto-
Chemical Sensors, IEEE Journal of Lightwave Technology, Vol. 24, No. 4, April
2006.
 Optical Fiber Sensors For Hydrogen Detection At Cryogenic
Temperatures
M. Consales, A. Cutolo, A. Cusano
Optoelectronic Division, Engineering Dept., University of Sannio, Benevento, Italy
a.cusano(4unisannio.it
S. Campopiano
Department for Technologies, University of Naples Parthenope, Naples, Italy
M. Penza, P. Aversa, G. Cassano
ENEA, C.R. Brindisi, Materials and New Technologies Unit, Brindisi, Italy
M. Giordano
Institute for Composite and Biomedical Materials, CNR, Naples, Italy

Abstract
In this contribution, preliminary results on the capability of Silica Optical Fiber (SOF)
sensors, coated by Single-Walled Carbon Nanotubes (SWCNTs) based sensitive material, to
perform hydrogen detection at cryogenic temperatures are presented. The optical
configuration exploited is the one of a low finesse Fabry-Perot interferometer, with a thin
film of SWCNTs deposited onto the distal end of standard optical fibers [1,2] Changes in
the thickness and complex dielectric constant of the sensitive layer as a consequence of its
interaction with the analyte molecules lead to changes in the fiber-film reflectance,
indicating the presence of hydrogen in the test ambient. Preliminary results obtained from
hydrogen detection testing, carried out at a temperature as low as -160C (113 K),
demonstrate the excellent potentiality of the proposed configuration to sense very IOW
percentages of gaseous hydrogen (4%) in air.

1 Introduction
The hydrogen detection has been widely studied by the researchers in the last
few decades. In 1981 Lundstrom introduced a hydrogen sensor based on thin
films of Palladium, deposited on the top of the channel of a MOSFET device
[3,4]. Among the big amount of sensing devices, fiber-optic detectors have
strongly attracted the attention of the scientific community in this field,
especially because of their high sensitivity and lack of sparking possibilities,
guaranteed by the removal of all electrical power from the test site. To date
several configurations of fiber-optic hydrogen sensors have been proposed,
based on different transducing principles and materials [5-91.
Recent reports of high reversible adsorption of molecular hydrogen in carbon
nanotubes and graphitic nanofibers [ 10-111 have stimulated tremendous interests
in the research community to exploit the novel carbon materials as ideal
candidates for hydrogen storage devices. Due to their hollow structure and high
specific surface area, the interactions between carbon nanotubes and gaseous
molecules attracted much experimental and theoretical interest.
The integration between SWCNTs as nanoscale sensitive material and optical
fiber technologies has already been carried out by the authors for the developing

28
 29

of SOF sensors capable of Volatile Organic Compounds (VOCs) detection with

sub-ppm resolution, good repeatability and fast response [ 1,2,12].
Here, preliminary results on the capability of SOF sensors, coated by two
different kinds of SWCNTs-based sensitive materials, to perform hydrogen
detection at cryogenic temperatures are presented. The proposed optical
configuration is essentially based on a low finesse Fabry-Perot interferometer,
realized by means of the deposition of both Closed-ends (CeSWCNTs) and
Opened-ends SWCNTs (OeSWCNTs) on the distal end of a properly prepared
single-mode SOF. The sensitive material is transferred upon the optical
transducer by means of the well known Langmuir-Blodgett (LB) deposition
technique [2,12]. Single wavelength reflectance measurements have been carried
out in order to investigate the capability of the proposed SOF sensors to detect
low percentages (4%) of gaseous hydrogen at cryogenic temperature.
Experimental results demonstrated the excellent potentiality of the proposed
configuration to sense very low concentrations of hydrogen at the very low
temperature of - 160C. Especially CeSWCNTs demonstrated good sensitivity,
fast response times together with good recovery and reversibility features.
2 Principle of Operation
The key point of a SOF sensor relies on the deposition of a sensing overlay
on the distal ends of standard optical fibers. When the light propagating within
the fiber encounters the fiber-film interface, the amount of reflected power
depends on the optical and geometrical properties of the CNTs-based overlay.
Consequently, any change occurring within the sensitive layer is able to induce a
corresponding change in the film reflectance (ARFif,,,), as expressed in the
following equation:
f
A RFilm = (A &Film A dFilm
3 (1)
where EF,~,,, and dF,l,,,are the complex dielectric constant and the thickness of the
CNTs-based film. When the optical probe is exposed to gaseous hydrogen, its
molecules are adsorbed within the sensitive layer, changing its complex
dielectric hnction and thus the optical signal reflected at the fiber film interface.
3 Sensors Fabrication
Thin films of CeSWCNTs and OeSWCNTs with different thicknesses were
transferred upon the distal end of single-mode optical fibers by the molecularly
engineered LB deposition technique [ 121. Standard silica optical fibers (SMF-
28), with core and cladding diameters respectively of 9 pm and 125 pm,have
been exploited. The investigation of the capability of the SWCNTs-based SOF
sensors to detect gaseous hydrogen at cryogenic temperatures, has been carried
out by means of three SOF probes, two coated by 2 and 6 monolayers of
CeSWCNTs (called 2-CeSWCNTs and 6_CeSWCNTs, respectively) and one
coated by 2 monolayers of OeSWCNTs (called 2-OeSWCNTs).
30

eri~entalSet-up
In Fig. 1.a is reported the exploited SOF sensors interrogation system, capable to
continuously monitor the fiber-sensitive layer interface reflectance [ 1,2]. A 2x2
directional coupler redirects the light coming from a Superl~inescentLight
Emitting Diode (SLED) towards two arms: the first one is directly connected
with the photoreceiver2,while the second one is connected with the sensor head.
The amount of light reflected by the sensitive interface comes back through the
same fiber and is collected by the photoreceiver,. The optoelectronic sensor
output is given by:
I=---'sign,/
VSource
- '''Film (2)

where k is a constant taking into account all the setup parameters, VSignalis the
signal proportional to the source power and to the film reflectance (RF,~,,,)and
Vsourceis the signal proportional just to the power emitted by the source. In this
configuration, any effect able to modify the complex dielectric function of the
sensitive layer changes the normalized output of the optical sensor, and eventual
fluctuations of the power emitted by the SLED do not affect the reflectance
measurements. In addition, in order to increase the signal to noise ratio, a
synchronous detection scheme has been implemented with the light source
externally amplitude modulated at 500 Hz and the sensor outputs retrieved by
using a dual channel lock-in amplifier. Furthermore, an optical switch has been
exploited in order to test all the three realized samples with a Time Division
Multiplexing (TDM) approach. The overall set up was properly engineered in
order to provide robust, compact and portable equipment.

Figure 1. Schematic view of the (a) SOF sensors interrogation system and @) experimental set-up
used for the detection of cryogenic hydrogen.

The hydrogen detection at cryogenic temperatures has been performed by

flowing argon gas with hydrogen (5%) in a proper designed cylindrical chamber,
where the optical fiber sensors have been located. In order to reach very low
temperatures (1 13 K) the chamber has been inserted in a tank containing liquid
nitrogen, as evidenced in Fig 1.b. The temperature monitoring has been carried
 31

out by means of a copperlcopper nickel thermocouple located close to the optical

sensors heads. Nitrogen gas has been continuously injected in the cylindrical
chamber in order to reduce the humidity content before the strong temperature
decreasing due to the insertion of the chamber into the cryogenic tank. Nitrogen
has also been mixed with the argon and hydrogen gas in order to expose the SOF
sensors to an hydrogen concentration lower than 5%.
5 Results
Preliminary results obtained from the low temperatures hydrogen detection
measurements are shown in Fig. 2, where the responses of the sensors (a)
2_CeSWCNTs, (b) 6-CeSWCNTs and (c) 2_0eSWCNTs, exposed to
decreasing concentration pulses of hydrogen at a temperature of 113K are
reported.

Figure 2. Responses to decreasing concentrationpulses of gaseous hydrogen (<5%), at 113 K, of three SOF
sensors: (a) 2_CeSWChTs, (b) 6-CeSWcNTs and (c) 2-0eSWCNTs

As it can be observed from Fig. 3.a, where the normalized reflectance

changes ARFllm/RFllm of the three tested SOF sensors are reported, all the optical
probes provides a significant change of the optoelectronic sensor output as a
consequence of the adsorption of gaseous hydrogen within the sensitive
materials. In addition, good sensitivity and complete recovery of the nitrogen
signal after hydrogen concentrations pulses have been observed for the
CeSWCNTs sensors, while good sensitivity, but no recovery of the steady-state
value has been observed for the OeSWCNTs-based sensor. Furthermore, a study
of the sensors response and recovery times reveals that the probes
2-CeSWCNTs and 6-CeSWCNTs are characterized by the same response times,
which are approx. 3 minutes, but by slightly different recovery times, which are
approx. 9 and 11 minutes, respectively (Fig. 3.b). Instead, 2-0eSWCNTs
showed a response time (approx. 5 min.) higher than that obtained with the
CeSWCNTs-based material. This could be attributed to the fact that, while in the
case of the CeSWCNTs the hydrogen molecules adsorption takes place only on
the surface of the single carbon tubes and in the interstitial sites between tubes,
in the case of the counterpart OeSWCNTs, the hydrogen molecules adsorption
takes place also inside the carbon tubes, requiring, for this reason, more time for
the establishment of the equilibrium condition with the external environment.
32

1 H2Pulse

-6.8
-5.3 .10-3
-3.6.10-3

Figure 3. (a) Normalized reflectance changes ARF,I,,,/RF,~,and @) mean response and recovery times
for the three tested SOF sensors.

6 Conclusions
In this work have been reported preliminary results on the capability of SOF
sensors, coated by thin films of two different types of SWCNTs, Closed-ends
SWCNTs (CeSWCNTs) and Opened-ends SWCNTs (OeSWCNTs), to detect
very low hydrogen percentages at cryogenic temperatures. The Langmuir-
Blodgett technique has been used for the deposition upon the distal end of
standard optical fibers of thin film of the aforementioned sensitive materials. The
fabricated optical probes have been exposed at cryogenic temperatures to
gaseous hydrogen with concentration lower than 5%. Experimental results
demonstrated the excellent potentiality of the proposed configuration to sense
very low percentages of gaseous hydrogen at cryogenic temperatures as low as
113K (-160C ). As matter of fact, the OeSWCNTs-based sensor is characterized
by high sensitivity but no recovery of the steady-state vaIue after the hydrogen
adsorption, whereas the CeSWCNTs-based probes exhibited good sensitivity
and complete recovery of the nitrogen signal after both hydrogen concentration
pulses. Finally, it has been found that the response times of the latter ones are
much lower than that of the counterpart OeSWCNTs-based transducer.
References
[ l ] M.Penza et al., 2004, Applied Physics Letters, 85, (12), 2378-2381.
[2] M.Penza et al., 2005, Sensors and Actuators. B, 111-1 12, 171-180.
[3] Lundstrom et al, 1975, Appl.Phys.Lett, 26,55-57
[4] Lundstrom, 1981, Sensors and Actuators, 1 403426.
[5] M.A.Butler, 1991, J. Electrochem. SOC.138, L46
161 M.A.Butler, 1994, Sensors and Actuators B 22, 155.
[7] H.Raether, 1988, Springer-Verlag, Berlin, p. 21.
[8] B.Sutapun et al, 1999. Sensors and Actuators B 60,27.
[9] Villatoro et al, 2001, Electr. Lett., 37, 1011.
[ 101 AC.Dillon et al., 1997, Nature;386:377-379.
[I I] C.Liu et al., 1999, Science, 286:1127-9.
[I21 M.Penza et al., 2005, Nanotechnology 16,2536-2547.
 Silicon p'n mesa junction with a porous layer
as sensing element

G. Barillaro, A. Diligenti, L. M. Strambini

Dipartimento di Ingegneria dell 'Informazione: Elettronica, Informatica,

Telecomunicazioni, Universitb di Pisa, via G. Caruso 16, 56126 Pisa-Italy.
8.barillaro(&et. unini .it

Summary
The structure of silicon solid-state devices can be modified by locally converting crystalline
silicon into porous silicon, so that their I-V characteristics become sensitive to gases in the
surrounding atmosphere. Following this approach, the surface of a p f n junction has been
provided, by means of electrochemical etching, with a porous layer. The resulting device has
been tested as sensor for isopropanol vapours. The fabrication process and the electrical
characterization of the device are presented and discussed.

1 Introduction
Air monitoring in the environment, industrial installations and hospitals is one of
challenges of the modem society. To achieve this goal solid-state sensors, small
and low cost, are required. Several materials, such as semiconductor oxides,
have been proposed in last years for sensor fabrication; among them, porous
silicon (PS) [ 11 is one of most promising for its high chemical reactivity at room
temperature. However, the integration on the same chip of a sensor hlly made
with PS, along with the necessary conditioningldriving electronic circuitry, is
difficult for two reasons: i) the fragility of the material, which limits the use of
lithographic processes, and ii) the lack of knowledge on the behaviour of
PS/metal contacts.
To avoid these drawbacks, recently a new approach for the integration of PS-
based gas sensors using industrial processes has been proposed [2, 31. In this
approach, differently from PS sensors reported in literature, adsorption of
molecules into the porous silicon layer is exploited to modify the electrical
properties of a crystalline device, such as a FET, for which the PS layer
constitutes the sensible part.
In this work, by using the same approach, a crystalline p'n silicon diode
surrounded by a PS layer (see Fig. 1) is proposed as gas sensor. Adsorption of
molecules in the PS layer modifies the electrical properties of the PS and/or of
the PS-crystalline silicon (c-Si) interface and in turn the diode current.

33
34

2. Sensor Fabrication
The fabrication process schematically consists of the following steps: 1) boron
implantation on a n-type wafer ((100) oriented, N O about ~ m - ~and
) , drive-in
to form a p layer, 2.3 pm deep with a surface doping of lo ~ m -2)~ aluminum
;
thermal evaporation and patterning ( lstmask) to provide the anode contact of the
diode. On the same chip several circular contacts with different diameters d were
defined (d = 500, 200, 100 pm). 3) Aluminum evaporation on the back of the
sample (cathode contact). The cathode is the same for all the diodes. 4) Anode
protection by using a patterned photoresist layer (2ndmask), hardbaked at 140 C
for 30 minutes; this step is essential in order to make the resist layer able to
withstand the electrochemical etching in HF, which is necessary to the PS
formation. The final step was the selective anodization of the p and n type
material through the photoresist-free spaces (fig. le) to form the PS layer
surrounding the pn diode. The electrochemical process was performed at a
current densityj = 30 mA/cm2 and for a time of 50 s in a solution HF(48
%):CH3CH20W1:1 under illumination conditions. Finally, the samples were
rinsed in acetone, in CH3CH20H(ethanol) and dried in N2 flow.
A schematic drawing of the device at the end of the fabrication sequence is
shown in figure 1. In a typical diode the porous film has a thickness of about 3
pm. The whole fabrication process requires only two masks and few
technological steps.

3. Sensor Characterization in Isopropanol Vapours

In order to investigate the behaviour of the diode as gas sensor, electrical
measurements were performed in dry nitrogen and in isopropanol (ISO), the
latter properly diluted in N2, used as carrier gas, to obtain a given concentration;

Fig. I Schematic view of the PS surrounded pn diode gas sensor. The cylinder is crystalline
silicon
 35

the experimental set-up is described in ref. [2]. All the measurements were
performed at room temperature.
The I-V DC characteristic in NZ for ap'n junction with a diameter of 200 pm, is
shown in Figure 2.
Since the DC equivalent circuit of the sensor can be represented as a pn diode
and a parallel conductive path, constituted by the PS film, any current variation,
at a given bias, induced by the gas adsorption could be ascribed both to the diode
andor the resistor. However, since the variations are sensibly different for
forward and reverse polarization at the same absolute voltage value, it is
reasonable to suppose that the main contribution is given by the diode.
The main sensor features resulting- from the characterization are:
both the forward and the reverse current, respectively IF and IB show a
noticeable variation when I S 0 vapours at 10,000 ppm are injected into the
test chamber; moreover, the initial current value 1, (reference current in N2)
is completely restored as soon as IS0 vapours are removed;
the transient behaviour is quite fast, some tens of seconds (see Figure 3),
and can be related to 1) the time required to filVempty the test chamber and
2) the diffusion time of molecules toward the diode depletion zone. Because
of the volume - 100 cm3 - of the test chamber, it is reasonable that the main
contribution to the transient time is due to the filling/empting time; an
improvement of the sensor response time is thus expected by reducing the
chamber volume.
the forward current variation AIF = 1 - I. at a fixed IS0 concentration can
result >O or <O, depending on the forward polarization voltage value;

-10 -8 -6 -4 -2 0 2
Voltage (Volt)
Fig. 2 Typical I-V characierisirc in niirogen of a PS surrounded p'n diode gas sensor with a
diameier of 200 ,um.
36

0.0

h -200.On
P2
% 4000n
E
9
+ -600.0n
2
5 -800.0n
0
-1.op
lsopropanol 10000 ppm
-1.2p, . , . , . , . , . , . , . , . , . , . , .
0 5 10 15 20 25 30 35 40 45 50
Time (minutes)
Fig. 3 Transient behaviour of a PS surrounded p'n diode gas sensor f o r dferent reverse bias
values in presence of rsopropanol vapours.

4) the reverse current variation AZR (alwaysIM,I > 0 ) at a fixed I S 0

concentration depends on the reverse bias value and can be as high as
several orders of magnitude, when a proper reverse voltage is chosen. In
Figure 3 the IR(Q behaviour of a diode driven by lo4 ppm I S 0 pulses is
reported for different reverse bias values. It is interesting to note that the
current variation noticeably increases as the reverse bias is increased, so that
changing the reverse bias value is a feasible way to tune the sensor
sensitivity.

The results l), 3), 4) could be explained if one supposes that the interaction of
the adsorbed gas with the localized states on the lateral surface of the diode,
changes the surface band structure; the situation is analogous to that of a gate-
controlled diode [4], but in this case the gate is floating.

4. Conclusions
The presented device, owing to its high sensitivity and simplicity of fabrication,
seems to be suitable for the production of low-cost sensors, and could be
integrated along with the necessary control electronics, since the formation of
the PS layer occurs at the end of the fabrication process. The approach used,
whose validity has been also demonstrated with different devices [2, 31, could be
applied to any silicon device, giving rise to a new class of integrable sensors. In
this case the test gas was isopropanol, but it is expected, on the basis of results
obtained on different sensors, based on the same principle [ 5 ] , that the diode
 37

results very sensitive for NO*, which constitutes a polluting gas of high
environmental interest.

Acknowledgments
The authors wish to thank STMicroelectronics (Cornaredo, Milan, Italy) for
having supported this work, which was performed in the frame of a cooperation
STMicroelectronicsAJniversity of Pisa.

References
[l] M. Tucci, V. La Ferrara , M. Della Noce, E.Massera, L. Quercia, J. Non-
Crystalline Solids 338-340, 776 (2004).
[2] G. Barillaro, A. Nannini, F. Pieri, Sens. Actuators B 93, 263 (2003).
[3] G. Barillaro, A. Nannini, G. Marola, L. M. Strambini, Sens. Actuators B 105,
278 (2005).
[4] A. S. Groove, Physics And Technology of Semiconductor Devices, John
Wiley & Sons, Inc. USA (1967), p. 289.
[5] G. Barillaro, A. Diligenti, A. Nannini, L. M. Strambini, E. Comini, G.
Sberveglieri, IEEE Sensors Journal 6, 19 (2006).
 Influence of solvent on thin film composite morphologies and
sensing properties

A. De Girolamo Del Mauro, L. Ouercia, F. Loffredo, G. Di Francia.

Research Centre Portici, ENEA, Naples, Italy
auercia @ Dortici.enea.it

Summary
In this work sensors based on thin composite films characterized by different contents of
carbon black (5-40% by weight) in polymer matnx (Poly (2 hydroxy-ethyl-metacrylate)are
fabricated and characterized. These films are obtained using various orgamc solvents. We
study the samples by dynanuc light scattering, optical and scanning electronic microscopy
and thermogravimetric analyses. The hypothesis used in order to describe qualitatively the
working mechanism of these sensors considers that polymer matrix swells when exposed to
organic vapor and the swelling decreases the connectivity between the conductive filler
particles. Testing to different VOCs the sensor devices, will show the influence of different
morphology on the characteristics of the sensors responses.

1. Introduction
During the past few decades, electrically conductive polymeric composites
have found use in a variety of applications. Binary composite systems,
comprised of conductive filler (e.g. carbon black (CB), metal powders, graphite
fibers, etc.) in a polymer matrix, create a material that is tough, flexible and
electrically conductive. These unique materials are ideally suited for antistatic
layers [ 1,2], electromagnetic interference shielding [3], chemical vapor sensors
[4-71, and thermal resistors [8].
In previous works we have studied the influence of filler characteristics [ 9 ] ,
the thin film morphology and fabrication process [lo] on the sensing device
performances in order to gain more information on the detailed working
mechanism of real devices.
Instead, in present study we have prepared different film polymer composites
of poly(2-hydroxy-ethyl-methacrylate) (PHEMA) with CB (PHEMNCB) using
four organic solvents, characterizing the relative sensing devices. PHEMA has
been known for its hydrophilic properties and suitability for this kind of sensors
[ I l l and CB is already optimized for applications in conductive polymeric
composite. When exposed to vapor phase analytes these conducting composites
exhibit good sensitivity through an increase in their resistance. The hypothesis
used in order to describe qualitatively the working mechanism of these sensors
considers that polymer matrix swells when exposed to organic vapor and the
swelling decreases the connectivity between the conductive filler particles [ 121.

38
 39

art
2. Experimental part
The sensing films were prepared by dissolving the polymer P m M A in the
suitable solvents and dispersing the filler, in various weight ratios (table 1), in
the solution (0.5-0.8% by weight) by ultrasonic bath. All suspensions were
deposited by spin coating or casting on different substrates. The four solvents
used to prepare the sensors are reported in Table 1.

(wc)
m-cresol (MC) 60i40

I (-)Ethyl L-lactate(ETLAC) I 80i20

PHEMNCB I ethanol (ETOH) I 90i10

1,l ,I ,3,3,3 Hexafluoro-2-propanol

(HFIP)

Table I : Solvents used to prepare the composite films

The thin film polymeric composites obtained have been characterized out by
Tencor P10 surface profiler, SEM and optical analysis. To determine the filler
contents of the composites, thermogravimetric analyses (TGA) were performed
on TA instruments 2950 equipment at a heating rate of 20C mid' (from 25C
to 800C). The sensing devices responses to organic vapors have been studied
using a Gas Sensor Characterization System (GSCS) already described [13].

3. Results and discussion

Fig. 1 shows the optical micrographies of PHEMAICB composites obtained
depositing by spinning on glass substrates different dispersions of CB in
PHEMA solution characterized by different solvent but the same initial percent
ratio between polymer and filler weights (80/20 weight % ).

Fig. I : Optical micrographies ( 1 0 0 ~of) compositefilms surfaces PHEMNC3 prepared with:

A ) MC; 3) ETOH; C)E T U C ; D ) HIFP.
40

The morphologies of the thin film composites result varied depending both
on the specific interaction between polymer/CB/solvent in the liquid dispersion
and the preparation conditions (spinning parameters and substrate/dispersion
interaction).
As expected, H I P , being a good solvent to disperse CB, allows to obtain the
most homogeneous thin film composites characterized by the smallest size of the
CB agglomerated (fig.lD).
The preparation of thin film composites starting by a biphasic dispersion
does not allow to forecast the effective CB content in the thin film. In order to
determine experimentally the total filler contents of the composite, we have used
thermogravimetric analyses (TGA). In Fig. 2 we compare the results obtained by
TGA for composites prepared with ETLAC and HIFP.

TcmpLratiu c {*C> I einlicrLiIure(.C.]

80/20 weight% in dispersion 60/40 weight% in dispersion

90/10 weight% in dispersion 80/20 weight% in dispersion
95/5 weight% in dispersion 90/10 weight% in dispersion
PHEMA PHEMA
Fig.2: TGA of PHEMAKB compositefilms prepared with :A ) ETLAC; B) HIFP.
For composite films, the CB content is indicated on curves

It is interesting to note the CB content in the thin film composite can be up to

three times higher than in the dispersion using ETLAC as solvent while it is
practically the same using HFIP as solvent. Therefore, it is possible to vary the
CB content in the polymeric films only changing the solvent. Furthermore, using
HFIP as solvent it has been possible to prepare conducting composite thin films
in a wider range of CB content.
All composite thin films, with a CB content ranging from 5% to 40% by
weight, have shown a conductivity suited to become sensing elements. The
PHEMNCB sensors are tested to different VOCs type (acetone, methanol,
ethanol, acetic acid and humidity). In general we have observed that the
sensitivities of devices obtained using different solvents for their preparation
does not change as much as we expected considering their morphologies (fig.1)
and relative TGA analysis (fig.2). As an example, we show the responses of
composite films PHEMNCB (80/20 weight%), prepared with ETLAC (A) and
 41

HIFP (B), to vapors of ethanol (fig.3). Both devices show comparable

sensitivities, however the dynamical responses change strongly especially
regarding response times and reversibility . The sensing device obtained using
HFIP as solvent, characterized by the more interesting morphology and more
controllable preparation conditions, is affected by longer response times and
reduced reversibility. This suggests how critical is the choice of the solvent used
for the sensing film preparation: morphology, flexibility of preparation
conditions and sensing characteristics must be evaluated in order to find the best
compromise for the specific application.

For small relative swelling AVN <<1 (where AV is the volume increase due
to swelling) we have modeled the sensors responses by the equation:
AR- tq5 AV
(1)
R (4-4cc) v
where $ is the volume fraction of CB and $c is the volume fkaction at
percolation threshold and t is the critical exponent as described by G.E.M.
equation [ 131.
We point out that, in principle, are expected little variations of all parameters
in equation (1) out of $ , the volume fraction, for all PHEMA composite
sensors, obtained with different solvents. On the other side, all sensors show
similar responses in spite of different total CB content as measured by TGA.
This suggests that Q is similar for all samples and only a small percentage of CB
in the composite films contributes to determine the "effective" value of $.

4. Conclusions
The solvent chosen to prepare the thin film composites has shown a great
influence on their morphology. As expected, the more homogeneous
42

morphology can be obtained by spinning the finest CB/ polymer dispersion

(obtained for the PHEMNCB system with HFIP solvent ).
Thermogravimetric analysis has pointed out that in general carbon black
weight percentage vary strongly between the initial liquid dispersion and the
resulting composite film.
The composite films obtained using the HFIP as solvent have kept the same
carbon black weight percentage of the dispersion. Thank to this it has been
possible to prepare sensing devices with a wide and more controlled range of
carbon black weight percentage.
Sensitivities to different VOCs of sensing devices prepared with different
solvents does not change significantly, as expected by their morphologies and
carbon black contents. However, PHEMNCB composite films obtained with
HFP, characterized by the better homogeneous morphology, are less reversible
and show longer response times.
In summary, the choice of better solvent to make sensors based on
polymer/carbon black composites films must be related to the application,
because it is a compromise between fabrication process and sensor performances
requirements.

References
[ I ] Bigg, D. M.; Stutz,D. E. Polym Comp 1983,4,40.
[21 Reboul, J . P. in Carbon Black-Polymer Composites; Sichel, K., Ed.; Marcel Dekker: New York,
1982, Chapter 3.
[31 Sau, K. P.; Chaki, T, Chakraborty, A.; Khastgir, D.; Plast Rubber Comps Process Appl.; 1997,
26,291.
[4] Lundberg, B.; Sundqvist, B.; J. App" Phys. 1986,60, 1074.
[5] Lewis, N. S.; Lonergan, M. C.; Severin, E. J.; Doleman, B. J.; Grubbs, R. H. SPIE 1996, 3079,
660.
[6] Ruschau, G. R.; Newnham, R. E.; Runt, J.; Smith, B.; Sens. Actuators 1989, 20, 269.
[7] Talik, P.; Zabkowska-Waclawek, M.; Waclawek, W.; J. Muter. Sci. 1992, 27, 6807.
[8] Moffat, D. M.; Runt, J. P.; Halliyal, A,; Newnham., R. E. J. Muter Sci. 1989, 24,609.
[9] Quercia L.; Loffredo F.; Di Francia G.; Sens. Actuators B 2005, 109, 153.
[lo] Quercia L.; Loffredo F.; Bombace M.; Nasti I.; Di Francia G.; Sens. Actuators B Vol.111-112
(2005). 166-170.
[ l l ] Chatzandroulis S.; Tegou E.; Goustouridis D. ; Polymenakos S. ; Tsoukalas D ; Sensor and
Actuators 2004, 103, 392-396.
[12] Doleman B. J.; Severin E. J.; Beaber S A.; Grubbs R. H., Lewis N. S.; Chem. Muter. 1996, 8,
2298.
[I31 McLachLan D. S.; Solid State Comm. 1989,72,831.
 Gas sensing properties of nanostructured titania: a
comparative study of different preparation methods

G.Neri*, A. Bonavita, G . Micali

Dept. of Industrial Chemistry and Materials Engineering, University of Messina,
C/da di Dio, Vill. S.Agata,
98 166 Messina, Italy
neri @ ingegneria.unime.it

F. A. Deorsola, I. Amato, B. De Benedetti,

Dept. of Materials Science and Materials Engineering, Corso Duca degli
Abruzzi, 24, 10129 Torino, Italy

Summary
This paper reports a study concerning the preparation of nanopowders of titania,
TiOz, through two novel processes of synthesis, namely gel-combustion and reactive
microemulsion. The microstructural properties of products obtained and their behaviour as
1-sensors for automotive exhaust gas have been evaluated. Titania synthesised by reactive
microemulsion has shown the better microstructural (lower grain size, higher surface area)
and oxygen sensing properties.

1 Introduction
Nanostructured Ti02 is a promising material for gas-sensing applications
[I]. For example, h sensors used actually are electrochemical devices based on
stabilized zirconia [2]. The lambda factor regulates the air-to-fuel ratio to
improve engine combustion and, therefore, optimize the conversion of
deleterious combustion products such as CO, hydrocarbons and NOx. The right
value of lambda factor for better engines performances is around one,
corresponding to partial pressure of oxygen ranging from 0,2kPa to IkPa.
However, planar resistive sensors which use Ti02 as sensing element present
some advantages and therefore they are actually the subject of numerous
investigation [3].
To develop titania with better oxygen sensing characteristics, the
process of synthesis is of fundamental importance because it determines the grain
size, structure and morphology of the obtained material, all factors affecting
strongly the sensing properties. Here, we focused our attention on two novel
processes to synthesize TiOz nanopowders: gel-combustion and reactive
microemulsion. Gel-combustion is an innovative and low-cost process combining
chemical gelation and combustion processes. The synthesis through reactive
microemulsion is instead based on the precipitation of particles from a solution,

43
44

containing low-cost precursor, previously emulsified in an immiscible liquid.

These materials have been tested for oxygen detection as lambda sensor for
automotive exhaust gas at high temperature.

2 TiOz nanopowders synthesis

Flow-charts showing the main paths in the preparation of Ti02 by gel-
combustion and reactive microemulsion are reported in the Schemes 1 and 2,
respectively.

Ti[ OCH(CH;)2]+ (CH;):CHOH

Scheme 1. Flow-chart of the gel-combustion process

Ti02 gel-combustion nanopowders were synthesized starting from

titanium (1V)isopropoxide (Ti(OCH2(CH3),),), isopropanol and hydrogen
peroxide as raw materials. A solution of titanium isopropoxide in isopropanol
(50/50 vol) was prepared first. Then, hydrogen peroxide was added drop by drop
under continuous stirring. The addition of H202 initiated the exothermic reaction,
that can be supposed divided in two step: i ) hydrolysis of titanium isopropoxide
forming titanium hydroxide Ti(OH)4, and i i ) oxidation of this product with a
significant increase of temperature and gas release [4]. The resulting product
obtained is dried at 80C, ground in a mortar, has the appearance of a highly
porous and softly agglomerated nanostructured powder.
In the reactive microemulsion synthesis, an aqueous solution of titanium
chloride Tic14 was prepared. It was added to a paraffin oil containing a small
amount of non ionic emulsifier, creating a water-in-oil emulsion in which each
single aqueous droplet contained the TiC14 solution. Each individual droplet
acted as a reactor, where the precipitation of spherical and nanosized titania
powders occurred by means of the thermal instability of the Ti precursor at high
temperature [5].During the precipitation, each particle is coated by an emulsifier
 45

film preventing the agglomeration [ 6 ] . Once the process was completed, the
precipitate was centrifuged, washed with isopropanol and dried at 80C.

$1 1-1 stable emulsion

Scheme 2. Flow-chart of the microemulsion process

3 Nanopowders charcterization

TEM analysis of as synthesized titania through gel-combustion and

reactive microemulsion showed particles with very small particle size (Fig. 1). In
according, XRD analysis of microemulsion titania showed the formation only of
the anatase phase (Fig. 2a), with an average grain size of about 6 nm. Specific
surface area was consequently very high (about 280 m2/g).

Fig. 1 TEM micrograph of the synthesized Ti02 nanopowders prepared by reactive microemulsion.

After treatment at high temperature the rutile phase appeared in the

XRD pattern of treated powders (Fig.2b). Moreover a re-crystallization of the
46

powders also occurred. For example, titania synthetized by reactive

microemulsion treated at 600C, showed an average size of 30 nm and a specific
surface of 150 m2/g.

A? 5u 'xi

?B I"]

Fig. 2. XRD analysis of Ti02 nanopowders prepared by reactive microemulsion a ) as synthesized;

b ) treated at 600C.

A commercial titania (Aldrich), used as reference, was also

characterized. It has a mean particles size of an order of magnitude higher than
the gel-combustion and reactive microemulsion synthesized by us.

4 Sensing test

Our synthesised titania samples have shown a higher resistance

compared to the commercial TiOz at all working temperature. This is related to
the smaller particle size of the synthesized samples. In despite of this, titania
prepared through the reactive microemulsion was found more effective for
detection of low concentrations of 0 2 at high temperature, showing a higher
response in the range 0,2 -1 kPa of oxygen pressure partial than other samples
(Tab. 1).

Lambda factor 0 2 Gel-combustion Microemulsion Commercial

(I) &Pa) Resistance (Q) Resistance (Q) Resistance (Q)
0.99 0.2 l.o.los 8.4.1O4 2.040~
1.01 0.5 1.240~ 1.0.105 2.3.103
1.03 1 1.3.10' 1.240~ 2.540~

Tab. 1. Lambda factor, partial pressure and resistive values of titania samples.
 47

2.4 1 ' " " ' ' I '

I microemulsion 1 / I--- microemdsion 1

2,2

2 2
U

2I 1,8
p
:/

9
a:
1.6
li
1.4

0,1 1 0,8 0,9 1.0 1,l 1 2 1,3 1,4 1,5

Pressure Partilal O,(KPa) Lambda factor

Figs. 3(a) Calibration curve at 650Cfor low oxygen concenrration, (b) Resistance dependence vs
lambda factor at 650C.

Figs. 3a and 3b show the calibration curve at 650C and the variation of
the resistance as a function of the h factor. Oxygen sensors based on
microemulsion titania have shown also a fast dynamic response (see insert in Fig.
3a), a properties highly required for air to fuel regulation of combustion engines.

4 Conclusion

Nanosized titanium oxide powders prepared through gel-combustion and

precipitation in reactive microemulsion show promising properties for oxygen
detection at high temperature. Particularly, reactive microemulsion give titania
with better microstructural (lower grain size, higher surface area) and oxygen
sensing properties.

References

[ l ] J. W. Fergus, J. Mater. Sci. 38,4259 (2003).

[2] J. Riegel, H. Wiedenmann, H. Neumann, Sol. State Ionics 152,783 (2002).
[3] N. Savage, B. Chwieroth, A. Ginwalla, B.R. Patton, S.A. Akbar, P.K. Dutta,
Sens. Actuators B 79, 17(2001).
[4] M. R. Ayers, A. J. Hunt, Mater. Lett. 34,290 (1998).
[5] H. Herring, R. Hempelmann, Mater. Lett. 27,287 (1996).
161 R. Zhang, L. Gao, Mater. Res. Bull. 37,1659 (2002).

Acknowledgments
The authors gratefully acknowledge the financial support for this work by
MIUR under the framework of FIRB-SQUAREproject.
 Hydrocarbon Sensing Device Based on Binary Metal Oxides
S. Kandasamy', A. Trinchi2,W. Wlodarsky'.
'School of Electrical and Computer Engineering, RMIT Universiv, GPO Box
2476V, Melbourne, VIC 3001, AUSTRALIA
2
Imbedded Intelligence Team, CSIRO Manufacturing and Infrastructure
Technologv, Highett, Victoria, AUSTRALIA
E. Comini3,G. Sberveglieri3
'CNR-INFM - Sensor Laboratory, Department of Physics and Chemistryfor
Materials Engineering. University of Brescia, ITALY
L. Pandolfi4, S. Kaciulis4
4
ISMN-CNR, Monterotondo (RM), ITALY

Summary
In this work, chemical and morphological properties of Ti-W-0 mixed oxide, employed as
material for hydrocarbon sensing devices, has been investigated. The reactive mixed oxide
(Ti-W-0) was deposited by r. f. magnetron sputtering from a pure T i m target on the
polished side of n-type 6H-Sic wafers. For the formation of the Schottky contact, a Pt layer
of 100 nm was deposited on top of the oxide layer. The samples before and after thermal
treatments were investigated by XPS and FE-SEM techniques.
The changes in the chemical composition and element distribution through the oxide film
thickness, caused by thermal treatments, have been revealed. The morphological structure
has been also investigated before and after the thermal treatments. The sensing performance
of the Pt/(Ti-W-O)/SiC MROSiC device has been investigated as a function of temperature
and concentration of propene in synthetic air.

1 Introduction
The prevention of the health risks due to the atmospheric pollution is one of
the most important challenges in the countries with high industrial production.
Nowadays the international rules are imposing the monitoring of hydrogen and
hydrocarbons emitted during energy-intensive industrial processes.
Single metal oxides and their various combinations have shown particular
sensitivity for H2 and hydrocarbons. Bulk and surface properties of oxide film,
such as band gap, electronic structure, position of Fermi level, etc., are
determining its interaction with gases. Therefore, a contemplated choice of the
mixed metals enables to obtain a sensing material where one of the above
mentioned properties can be emphasized and employed for the gas monitoring.
Moreover, the necessity of direct pollution monitoring frequently requires of
the devices working in critical conditions, such as high temperature and
corrosive environment. Metal - reactive oxide - silicon carbide (MROSiC)
devices have been indicated as one of the solutions for the reliable sensing in
industrial environment [ 1-21.

48
 49

The Ti-W-0 films (as deposited, after annealing at high temperature and
after gas sensing test), used in MROSiC sensing devices, were characterized by
means of X-ray Photoelectron Spectroscopy ( X P S ) and Field Emission Scanning
Electron Microscopy (SEM-FEG).

eri
2 Experimental
Mixed metal oxide films were deposited on the polished side of N-doped (n-
type) 6"-Sic wafers. Ohmic contact was formed on the unpolished side of Sic
by depositing Ti and Pt, 100 nm each. A layer of Pt with a thickness of 100 nm
was deposited on top of the reactive oxide film, forming the Schottky contact
(Fig. 1).

Fig.1 Schematic of the MROSiC device.

The mixed oxide was deposited using r. f. magnetron sputtering from a pure
T i m alloy target (20% and 80% in weight) under reactive atmosphere (Ar : 0 2
at 8x10" mbar). The substrate was kept at 300 "C to promote the formation of
the film and to favour the filling of oxygen deficiencies.
For the gas sensing test, the operating temperature was varied between room
temperature and 750 "C. Propene (C3H6)and hydrogen (Ha), at a constant flow
rate of 200 mlfmin, were used as analyte gases after the sensor's stabilization in
synthetic air. Their concentration was varied in the range of 0.06 to 1%. The
response was measured as the voltage change, while the sensors were operated at
constant bias current of 10 PA. This voltage shift was recorded by using a HP
33 104A multi-meter. The current-voltage (I- v) characteristics were obtained
using a Tektronix 571 curve tracer.
Selected-area XPS (SAXPS) measurements and Ar' ion sputtering have been
performed in a VG Escalab MkII spectrometer (5 channeltron detection system;
unmonochromatized source of A1 Ka). The electrostatic lenses were operated in
selected-area mode, providing photoelectrons collection from the sample area of
about 0.3 mm2.The Ar' ion beam of 2.0 keV energy was rastered over an area of
1 x 1 mm2 and operated at a current density j = 3 pNcm2.
The morphology of the samples has been investigated by using a LEO 1530
SEM-FEG instrument.
50

3 Results and Discussion

The Schottky contact and the reactive oxide film of the MROSiC gas sensor
were investigated after its preparation (as deposited) and after the treatments at
450 OC for 1 h and at 600 C for 2 h. The X P S depth profiles of two different
areas of the as deposited sample are shown in Figg. 2-3.
The depth profile performed on the TiWO film (Fig. 2) showed an uniform
distribution of both the metals (Ti, W) through the film. The average atomic
ratio of W and Ti was about 2.5. The oxygen amount decreased during the
profiling due to the well-known effect of preferential sputtering. The rapid
decrease of the oxygen and metals concentrations reveals a sharp interface
between the oxide film and S i c substrate.
80

70

60

5O
$ 60
5 40 50
30 4 40

20
30
20
10
10
0 0
0 20 40 60 80 100 120 0 20 40 SO 80 100 120 140 0
spvnerlnp tlme (me) SpuUerinp Time (rnin)
Fig.2 As deposrtedsample - depth profile out of Fig.3 As deposited sample - depth profile on the
the Pi contact. Pi contact.
80
TI-W oxide
1-
M ol,l*nmo*sn
80

30

20
20
10
10
0 0
0 20 40 SO SO 100 (20 140 180
spunsrlnp Tlms (min)

Fig.4 Annealed sample - depth profile out of the Fig.5 Annealed sample - depth profile on the Pt
Pt contact. contact.

The depth profile, performed on the Pt contact, is presented in Fig. 3, where

can be seen the multilayer structure of the sample: Pt/Ti-W-O/SiC. Also in this
profile was revealed an uniform composition of the layers with steep interfaces.
The depth profiles of the annealed sample are shown in the Figg. 4-5. As it is
possible to note, the W distribution through the films thickness is not uniform.
Moreover, the W : Ti ratio varies from 0.5 at the surface to 1.6 at the substrate. It
is possible to suppose that the annealing was causing the segregation of W with
 51

its consequent sublimation from the surface. The depth profile, performed on Pt
contact, reveals a different picture. Due to the presence of Pt layer, the migration
of W is directed not towards the surface but to the substrate. The broader
interfaces Pt/oxide and oxidelsubstrate confirm the thermal diffusion of the
metals towards the substrate. The results, obtained for the tested sample, were
similar to the ones of the annealed sample. In the depth profile, performed on the
Pt contact, the X P S signal of Pt was present until the substrate. The high
working temperature caused an intermixing of Pt layer and oxide film.
On the surface of all investigated samples were present the metal ions of Tii4
and W+6.In the film's volume, the presence of W sub-oxides was revealed. The
decrease of 0 concentration together with the change of the W chemical state
could be attributed to the ion bombardment (31. However, at the Pt /oxide
interface, the W 4f line-shape is different from the one at the oxidelsubstrate
interface. This result indicates that the presence of W sub-oxides is partially
caused by the thermal processes in the MORSiC sample during the treatments.

Fig.6 As deposited sample Fig. 7 Annealed sample

The changes in the sample morphology after the treatment are shown in Figg.
6-7. The as-deposited film consists of amorphous layer with symmetrically
shaped nano-crystallites sized of about 200 nm and located on the surface. The
heat treatment causes the formation of a nanosized polycrystalline film [Z]. The
size of the crystallites is increased and their shape becomes asymmetrical.
The gas sensing tests were performed using C3H6 and H2 in air. The sensor
presented a maximum voltage shift of 2.52 V when exposed to 1% C3H6 at 420
"C. The sensitivity was decreasing with increasing temperature. In the case of
1% Hz, a maximum shift of 1.15 V was registered at 200 "C, whereas it was only
0.46 V at 420 "C. The response to 1% C3H6is approximately 5 times higher than
to the same concentration of H2.
Fig. 8 shows the dynamic response-recovery characteristics of the sensor at
420 "C. The voltage shifts of about 2.52 and 0.35V were observed for 1 and
0.06% of C3H6, respectively. The sensor exhibits fast response time (-15 to 20
s), and a longer recovery time (- 350 s). In the detection of hydrocarbons, the
52

slow recovery time may not pose a practical problem. The relation between the
voltage shift and the C3H6 gas concentration is fairly linear, indicating a large
operating dynamic range. The I-V characteristics of the device are shown in Fig.
9. A shift in the I-V curves was observed when 1% of H2 and C3H6 were
introduced into synthetic air. At 420 C and lOpA, the shifts of 2.6 and 0.46 V
were observed for 1% C3& and Hz, respectively.
8 __
.
0 Synthetlcalr
100 A l%Propene A *
+ l%Hydrogen A
80 A
E 7 A - 0
r *

g
% 6

5
0 1100 2200 3300 4400 5500 0 3 6 9 1 2 1 5

Time [+I Vdtage M

Fig.8 The response-recover), characteristics of Fig.9 Comparison of the sensors response to
the sensor different ambient atmospheres

4 Conclusions

The SAXPS investigation has revealed the elemental distribution through the
whole thickness of the sensor. Also the changes in the chemical composition and
in the morphology of the samples after the thermal treatments have been
determined. The results of gas sensing tests and micro-chemical investigations
demonstrated the suitability of mixed Ti-W oxide for the application in MORSiC
gas sensing devices.

Acknowledgments

The authors are grateful to Dr. G.M. Ingo (ISMN-CNR) for the acquisition of
SEM-FEG images.

References
[ l ] C. V. Gopal Reddy, W. Cao, 0. K. Tan, W. Zhu and S. A. Akbar, Selective
detection of ethanol vapor using xTi02-(l-x)WO3 based sensor, Sens. Actuators
B, V O ~94,. pp. 99-102,2003.
[2] E. Comini, G. Sberveglieri and V. Guidi, Ti-W-0 sputtered thin film as n-
or p-type gas sensors, Sens. Actuators B, vol. 70, pp. 108-1 14,2000.
[3] S. Kaciulis, G. Mattogno, Characterization of thin-film devices for gas
sensing Surf. Interface Anal., vol. 30, pp. 502-506, 2000.
 Multi-Cavity Structured FBGs:
Towards Advanced Chemical Sensing Devices

A. Iadicicco, D. Paladino, A. Cutolo, A. Cusano*

Engineering Department, University of Sannio, Benevento, Italy
a.cusano(&nisannio.i t

S. Campopiano
Department for Technologies, University of Naples Parthenope, Napoli, Itab

M. Giordano
Institute of Composite Biomedical Materials, National Research Council,
Napoli, Italy

Summary
This work deals with micro-structured fiber Bragg grating (MSFBG) involving two-dcfect.
Every perturbation consists in a localized stripping of the cladding layer, leading to a
distribute phase shift on the guided mode. As consequence one or more defect states are
induced inside the original grating bandgap. On the other hand, the spectral features strongly
depend on the perturbations features (length and depth), refractive indices surrounding the
thinned areas and the geometrical distribution of the perturbations along the structure. Based
on the high sensitivity to the external medium refractive index, this structure exhibits
attractive features for communication and sensing applications. This work is focused on the
possibility to use the two-dcfect MSFBG for dual-SRI measurements by a single sensing
element. This approach could be a suitable solution when a large number of spatial locations
have to be monitored.

1 Introduction
Recently, the authors proposed and experimentally realized a novel
configuration of in-fibre periodic structure employing a localized defect along a
standard fiber Bragg grating (FBG) [l-21. The perturbation consists in a
localized stripping of the cladding layer positioned in the centre of the grating.
The defect along the periodic structure induces changes in the core propagation
[3] and thus a phase-shift on the guided mode [l], leading to the formation of a
defect state inside the original bandgap. The length and the depth of the induced
perturbation as well as the surrounding refractive index (SRI) define the spectral
response of the micro-structured FBG (MSFBG) in terms of position of the
defect state inside the band-gap. Differently from standard phase shift grating
(PSG) [4], here the spectral position of the defect state is strongly related to the
SRI. This effect has been efficiently used for high resolution refractive
measurements [2]. Successively the authors report a preliminary study on a two-
defect structure, revealing attractive spectral features for sensing and
communication applications. In this work, the spectral characteristics of a multi-

53
54

defect MSFBG as well as its potentialities for multi-SRI measurement by a

single sensing element are numerically investigated.

2 TwoDefect
efect s structured
t r ~ e t u r ~FFBG
d
According to one-dimensional photonic band-gap [6] and in-fibre phase shift
grating (PSG) [4] theories, more perturbations along a periodic structure permit
more defect states to be located inside the band-gap. Fig. 1 plots a schematic
diagram of a FBG with two cladding stripped regions. LTh and D T are ~ the
thinned region length and diameter, respectively, while the subscript 1 and 2
indicate the first and the second perturbation (from left to right). In addition, L1
and L3 indicate the left and right unperturbed grating regions lengths while Lz is
the length of the unperturbed region between the thinned areas. As demonstrated
in previous work [5], in the multi-defect MSFBG, the spatial distribution of the
defects plays a fundamental role on spectral response, in terms of shape and
spectral position of the defect states. In particular, the formation of a single
defect state with flat band or two defect states is ruled by the ratio of the
unperturbed grating length. Here, the geometrical parameters of the investigated
structures are selected in order to achieve two-defect state inside the original
bandgap.

+---+-
J", LlP*l "0 J ??re Ls

Fig. I Schematic diagram of the two cavities MSFBG (not in scale)

Numerical analysis based on multilayer approach and doubly cladding fiber

model allows an accurate spectral investigation. Here SMF-28 optical fibre
features have been used 221. Fig. 2.a shows the reflected spectra of a MSFBG
with two perturbations consisting in totally cladding stripping (DThl=DThZ=DCore)
for different SRIs and in case of Ll=L3=2000pm, LThl=LThZ=500pm and
Lz=lOOOpm. Note that the same SRI for both thinned regions has been taken into
account. The selected grating exhibits a core refractive index modulation with
maximum amplitude An=7~10-~, leading to a central wavelength hB=1550.08m
and a maximum reflectivity of about 99%. As expected, more defect states are
induced inside the spectral band-gap. Moreover, SRI changes are able to
strongly influence the spectral response of the device, in terms of the defect state
shift, leading to the possibility to realize advanced all-fiber devices.
 55

- SRI
~

--- SRII =41

I
~
- SRI

Wavelength [nm]
Fig. 2 Spectral response of a iwo-defect MSFBGfor d#erent SRI in case offulf etching,
L1=L3=2000pm,L~~=Ln2=50Oprn and L~=1000pm.

3 Sensing Applications
In this section, a preliminary study on the possibility to use a single two-
defect MSFBG for dual-SRI measurements, by monitoring the shifts of the
central wavelengths of the defect states, is presented. When a large number of
spatial locations have to be monitored this approach is the suitable solution.

Wavelength [nm]
Fig. 3 Spectral response of asymmetric Fig.4 Shfts ofthe central wavelengths of both
two-defect MSFBG defect states versus as An, and A n 2 move in a
range of 5.10".
The investigated structure is characterized by L1=L3=1650pm,
L~hl'L~h2'950pm, L2=800pm, Dnl=DTh2=DCore.The SRIS in the two thinned
regions move in different ranges: in particular, the SRI in the perturbation 1
moves around 1.42, while the other one around 1.40. An, and An2 denote the SRI
changes in the perturbation 1 and 2, respectively. Fig. 3 shows the spectral
response of the MSFBG in case of unperturbed SRIs. The selected geometrical
parameters induce two-defect states inside the spectral band-gap [ 5 ] , centered at
hl=1549.935nm and h2=1550.116nm and ruled by the phase shift of both
perturbations. Here, weak defect states are achieved since an asymmetric
structure (different SRIs) is taken into account.
56

The fig. 4 shows that the shifts in the central wavelengths of the two defect
states exhibit a quasi linear behaviour versus SRI changes, if Anl and An2 move
in a small range. Based on this result and supposing that the effects of Anl and
An2 can be separated, we can write the wavelength shifts as:
A A , z k,An, + k,An,
{ A A 2 E k,An, + kiAn,
(1)
where Ah1 and Ah2 are the central wavelength shifts of the left and right defect
state, respectively, and kl, k2, kl and k2 are constants indicating the sensitivity
of the two defect states central wavelengths versus the SRI variations in the two
perturbations. Assuming that kl, k2, kl and k2 are known, eqs. (1) can be easily
solved for An1 and An2. From the fig. 4 the following coefficients are estimated:
kl=2.753nm, k2=7.014nm, kl=4.037nm and k72=1.799nm. In case of
asymmetric structure, the eqs. (1) form a well-conditioned system, leading to the
following solution:
k i A A n ,- k , A A ,
An,s =
k,kl - k;k,
k l A A, - k,A A,
An,, =
klk5 - k;k,
where Anls and A n 2 s are the estimated SRIs variations. They can be slightly
different from the An, and An2 due to the linear approximation and to the
separation of the two perturbations effects proposed by eqs. (1). This means that
the residuals Rl=Anls-Anl and R2=An2s-An2 are strongly dependent on the An,
and An2 variation ranges, thus the approach advised in eqs. (1) is correct as both
An, and Anz move in a small range.

An,[xlO-] An,[xlO-] An,[xlO-] *

Fig. 5 Residual R ~ = A n , s d n (a)
, ~ n d R ~ = A n ~ ~@)Aversus
n 2 An, and An2 moving in a range of 54,7-
around 1.42 and 1.40, respectively

In order to prove the capability to demodulate the SRI variations in both

thinned regions through the linear approximation proposed in eqs. (I), different
variations of An1 and An2 from 0 to 5.10-3are taken into account and the induced
Ahl and Ah2 are numerically computed. Then Anls and An2s, are calculated by
eq. (2). Figs. 5.a and 5.b report the residuals RI and R2 versus the simultaneous
variations of Anl and An2 in the range Ot5.10 around 1.42 and 1.40
 57

respectively, revealing that the maximum residuals are Rlmax=1.7.10-4and

R2,ax=3.8-10~4.
As Anl and Anz move in a range of lo around 1.42 and 1.40, respectively,
the higher errors strongly decreases to R1,,=7.0.10-6 and R~,ax=1.5~10-5. It is
obvious that the decrease in the residuals is related to a better linear
approximation of the Ah, and Ah2 as the SRIs move in a smaller range. Based on
these results a single multi-defect MSFBG packaged by a properly micro-fluidic
holder can be efficiently used for multi-SRI measurements.

4 Conclusions
Based on the obtained results a single multi-defect MSFBG packaged by a
properly micro-fluidic holder can be efficiently used for multi-SRI
measurements. Note that the small investigated SRI ranges are not limitative for
the proposed sensor configuration in case of chemical applications, where low
concentration (with the order of ppm) of pollutants needs to be monitored.
The study of MSFBGs with more than two defects is currently under
investigation, while, differently from previous works, a new fabrication
technique based on UV laser micromachining is actually explored for the first
multi-defect MSFBG prototype realization.

References
A. Iadicicco, S. Campopiano, A. Cutolo, M. Giordano, A. Cusano, Micro-
Structured Fiber Bragg Gratings: Analysis and Fabrication, IEE Electronic Letters,
Vol. 41, No. 8, April 2005.
A. Iadicicco, S. Campopiano, A. Cutolo, M. Giordano, A. Cusano, Refractive
Index Sensor Based on Micro-Structured Fiber Bragg Grating, IEEE Photonics
Technology Letters Vol. 17, No. 6, June 2005.
A. Iadicicco, A. Cusano, A. Cutolo, R. Bernini, M. Giordano, Thinned Fiber Bragg
Gratings as High Sensitivity Refractive Index Sensor, IEEE Photonics Technology
Letters, Vol. 16, No. 4, 1 149-1151, April 2004.
L. Wei, J.W.Y. Lit, Phase Shifted Bragg Grating Filters with Symmetrical
Structures, Journal of Lightwave Technology, Vol. 15, No. 8, August 1997.
A. Cusano, A. Iadicicco, S. Campopiano, M. Giordano, A. Cutolo, Thinned and
Micro-Structured Fiber Bragg Gratings: Towards New All Fiber High Sensitivity
Chemical Sensors, Journal of Optics A: Pure and Applied Optics Vol. 7, pp. 734-
74 1,2005.
I. Del Villar, I. R. Matias, F. J. Arregui, Fiber-optic Multiple-Wavelength Filter
Based on One-Dimensional Photonic Bandgap Structures With Defects, Journal of
Lightwave Technology, Vol. 22, No. 6, June 2004.
 Belgian beer mapping and digital fingerprinting
using color and turbidity assessment

A.G. Mignani, L. Ciaccheri

CNR-IFAC, Sesto Fiorentino (FI), Italy
[email protected]

Summary
Multi-wavelength and multi-angle absorption spectroscopy performed in the visible spectral
range, that IS, scattered colorimetry, has been used to map a library of 25 diverse and
commercially-available Belgian light beers. The resulting map shows four main clusters,
relative to blonde, amber, red, and weiss beers, respectively, demonstrating that scattered
colonmetry is a valid method for beer authentication and fingerprinting.

Introduction
Beer is one of the Belgiums greatest specialities. There are more than 100
different breweries producing beers with unusual tastes [1-41. In addition to
having a different taste, each beer has a different and distinctive appearance,
which is caused not only by the color but also by the turbidity content.
Consequently, the beers appearance can be considered as a product fingerprint.
Suitable methods and technologies for product qualification, authentication, and
certification are advisable in order to protect both producers and consumers.
The brewing industry, both the artesan breweries and the large-scale
complexes, make use of specially-equipped laboratories for the assessment of
beer quality, so that the production process can be optimized. Chemicals
responsible for flavor and nutritional aspects, which are of particular interest to
beer drinkers are monitored. Other parameters, such as allergens and
microbiological organisms, are also measured, in order to guarantee quality and
safety during the production process and packaging phase [5, 61.
This paper presents a technique that takes into account the whole content of
color and turbidity of beer. Instead of measuring color and/or turbidity only,
some indicators are measured that consider the global contribution of both color
and turbidity. This is a physical information which can be considered to be
complementary with respect to that provided by using standard analytical
techniques. The result of this study is the mapping of 25 diverse and
commercially-available Belgian light beers. The mapping can be considered as a
digital fingerprinting of the beer samples.

58
 59

Scattered colorimetry
Multi-wavelength and multi-angle absorption spectroscopy performed in the
visible spectral range, which we call 'scattered colorimetry' or 'spectral
nephelometry', is a novel optical method for simultaneously monitoring the
whole content of color and turbidity in liquids.

The instrumentation used for scattered colorimetry is an optoelectronic device

for measuring the absorption spectra of samples at different angles. It consists of
four white-light LEDs spanning the 450-650 nm spectral range, with a
minia~rizedoptical fiber spectrometer that serves as a detector (Figure 1 -left).
The sources, which can either be fitted to the probe or guided by optical fibers,
are located at an angle of 0", 30, 60", and 90" with relation to the detector
(Figure 1-right) [7, 81. The LEDs are switched on sequentially in order to
measure the transmitted and scattered spectra. Because scattering is a
wavelength-dependent phenomenon, the spectrum transmitted in turbid media,
which mainly provides information regarding color, is also affected by turbidity.
On the other hand, the scattered spectra, which mainly provides information on
turbidity, is also affected by color. Given the spectrometer's spectral resolution,
the sample of liquid under test can be characterized by means of a maximum of
184 spectral values coming from the four absorption spectra, each of which
consists of 46 wavelengths.
Multivariate data processing is used to achieve a reduction in data dimensionality
so as to better extract information significant for sample identification [9-121.
The spectral data are usually processed using principal component analysis
(PCA) or linear discriminant analysis (LDA), which provide the coordinates for
identifying the samples on two- or three-dimensional maps. When several
samples of liquids are analyzed, the maps are populated by point clusters, each of
which groups the samples in terms of the similarities in their color and turbidity.

Figure I Working principle of scattered colorimetry (lefr)and optical probe (right)

60

Table I List and characteristics of Belgian beers measured

using the scattered colorimetry technique

Beer mapping and fingerprinting

A library of 50 diverse and commercially-available Belgian light beers was
considered, as listed in Table 1. These beers differ not only in terms of color and
turbidity, but also in their alcoholic content. Brown beers were not considered
because of their high absorbance, which results in a poor signal-to-noise ratio.
The beer samples were measured using the scattered colorimetry technique, and
Figure 2 shows the absorption spectra measured at 0" and 90". The spectral data
 61

were processed by means of PCA, and Figure 3 shows the beer mapping in the
PC1-2 subspace. Figure 3-left shows the mapping according to the symbols that
were assigned to the beer types, while Figure 3-right highlights the four main
clusters according to type: blonde, amber, red, and weiss. The PCl and PC2
coordinates, which represent the digital fingerprinting of the measured beers, are
listed in Table 1.

Transmission Spectra ( 0") Scattering

_ Spectra
. ( 90" )
""P.
0 07

0 06 Shon Dash = Ambcr

ShonDash= Amber 0OO't
06
0 05
long Dash=Bbnd&Wess
0 04

0 03

0 02

0 01

450 500 550 600 650

Wavelength (nm) Wavelength (nm)

10- , 15

0 10
5'
w v 5

*
-
P- 0 -
0
vv
G O
\v P-
-
'D 8 vv
V
V -
'0 -5

El
r-4 -5.
:-lo
E 0
0 D.
V Blonde -15
-10.
-20
0
-15' ' -25

Conclusions
Using scattered colorimetry and PCA data processing, we accurately and
non-subjectively mapped different kinds of Belgian beers and grouped them
62

according to assigned classes. The sharp clustering of blonde, amber, red, and
weiss types demonstrates that scattered colorimetry and PCA data mining provide
a valid method for beer authentication and fingerprinting. The implementation of
an online system to be used in one or more production stages is possible. This
could be designed as a multipoint sensing device controlled by a single
electrooptic unit. A further implementation of this technique is a multispectral
analysis in the UV, near- and mid-infrared spectral regions and the correlation of
the spectral data to the most important chemical quantities.

References
1. M. Jacksons, Great Beers of Belgium, Media Marketing Communication SA,
Antwerpen, 2001.
2. T. Webb, Good Beer Guide to Belgium, Gardners Books, Eastbourne, 2005.
3. B. Dubrulle, Guide to Belgian Beers, NeoCity Publishing, Ghent, 2005.
4. http://belgianstyle.com/mmguide/.
5. http://www.ebc-nl.com/ and http://www.asbcnet.org/ .
6. Y.H. Hui, Data Sourcebook for Food Scientists and Technologists, John
Wiley&Sons, 1991.
7. A.G. Mignani, L. Ciaccheri, P.R. Smith, A. Cimato, C. Attilio, R. Huertas, M.
Melgosa Latorre, A.C. Bertho, B. ORourke, N.D. McMillan, Scattered
colorimetry and multivariate data processing as an objective tool for liquid
mapping, SPIE vol. 5855 17h International Conference on Optical Fibre
Sensors, pp. 38-41,2005.
8. A.G. Mignani, L. Ciaccheri, A. Cimato, C. Attilio, P.R. Smith, Spectral
nephelometry for the geographic classification of Italian extra virgin olive oils,
Sensors and Actuators B, vol. 111-112, pp. 363-369,2005.
9. LA. Cove, J.W. McNicol, The use of principal component in the analysis of the
near infrared spectra, Applied Spectroscopy, vol. 39, 1985, pp. 257-266.
10. B.G.M. Vandeginste, D.L. Massart, L.C.M. Buydens, S. De Jong, D.J. Lewi, J.
Smeyers-Verbeke, Handbook of Chemometrics and Qualimetrics, Elsevier
Science BV, Amsterdam, 1998.
11. M.J. Adams, Chemometric in Analytical Spectroscopy, Royal Society of
Chemistry, Cambridge UK, 1995.
12. T.M. Cover, P.E. Hart, Nearest neighbor pattern classification, IEEE
Transaction on Information Theory, vol. IT-13, 1967, pp. 21-27.
 A portable fluorometer for the rapid screening of
M1 aflatoxin in milk

C. Cucci, A.G. Mignani

CNR-IFAC, Sesto Fiorentino (FI), Italy
c .cucci@ ifac.cnr.it

C. DallAsta, G. Galaverna, A. Dossena, R. Marchelli, R. Pela

Department of Organic and Industrial Chemistry, University of Parma, Italy

Summary
A compact fluorometric sensor equipped wlth a LED source and a high sensitivity PMT
detector has been implemented for the selective detection of native fluorescence of aflatoxin
AFMl in liquid solutions. This compact and easy-to-handle device is addressed to the rapid
monitoring of AFMl in milk, enabling the detection of concentrations up to the legal limt,
which is 50 ppt. The system is suitable for preliminary screening at the earlier stages of the
industrial process, and makes it possible to discard contaminated milk stocks before their
inclusion in the production chain.

Introduction
Aflatoxins are fungal metabolites produced by widespread moulds
(Aspergillus species), which can colonise foodstuffs and feed. Among the
different aflatoxins identified, four types (aflatoxin B1, B2, G1 and G2) are of
main interest, since they have been classified as Group 1 carcinogens by the
IARC [I]. Aflatoxin B 1 (AFBI), normally predominant in agricultural products,
has the highest toxic potential [2, 31. Its main metabolite is AFMI, also known as
milk toxin. AFMl is indeed found in milk of lactating animals following their
ingestion of AFBl-contaminated feed [4,51. AFMl is considered to be a highly
hazardous food contaminant, due to the fundamental role of milk in the human
diet, especially as infant nourishment.
At international level, the EU has the most rigorous regulations concerning
mycotoxins in food [6]. As regards AFMl in milk, the maximum legal limit is set
at 0.05 pg /Kg (50 ppt) for all EU Member States, and guidelines suggest to
allowing Member Countries to adopt lower limits (25 ppf) for baby food [ 7 ] .
Nowadays, different methods, based on HPLC, TLC or immunoassay
methods (ELISA), are well-established for detection of AFMl in milk [8, 9, 10,
111. These procedures require trained staff and/or expensive equipments, and are
not suitable for large-scale screenings. In addition, the routine laboratory controls
can be insufficient for coping with sporadic aflatoxins emergencies, that are
linked to particular environmentalheasonal conditions which favour the fungal

63
64

invasion of animal feeds and agricultural products. New low-cost screening

methodologies suitable for rapid and large-scale tests on milk are therefore still
greatly needed.
In the present work, a prototype of a compact and easy-to-handle device for
the rapid detection of AFMl at concentrations up to the legal limit (50 ppt) is
presented. The system is a portable fluorometer based on a LED source and an
extremely sensitive PMT detector, which make possible the direct identification
of low concentrations of AFMl. This system is intended to be used for quick
threshold indications and as an early warning system, so as to rapidly single
out risMalarm situations. Three levels of contamination are thus identified: 1)
alarm 2) alert for baby-food and 3) safe. Although tested up to now only on
synthetic solutions of AFMI, this methodology is highly promising in a
perspective of large-scale controls on the milk industrial chain. Indeed, it would
make it possible to simply discard contaminated milk stocks before their
inclusion in the production chain.

The instrumentation
The instrumentation consisted of a commercially-available compact device
for fluorescence analysis [ 121, which had been specifically re-adapted and
optimised for the selective detection of AFh4l in liquid samples.
As shown in Figure 1-left, the instrument made use of a LED source and a
highly sensitive PMT detector that made it possible to measure extremely low
fluorescence signals. The source and the detector were equipped by means of
band pass filters, which provided an optimal optical isolation between the
illumination and detection spectral bands. The optical system was PC-driven, and
the measurement procedure could be suitably automated by using the specific
software. Although the instrument was equipped by means of a quartz cuvette, it
could be easily implemented by means of an optical fiber strand, as shown in
Figure 1-right.
The optical system consisted of inter-changeable modules, which made the
fluorometer highly versatile and adaptable for different applications. In the
present work, the LED and the filters of the original instrument have been
replaced with new units optimised to match the fluorescence bands of AFMl
aflatoxin [ 13, 141. The technical specifications of the optical components used
were the following: LED source: Roithner Lasertechnik, UVLED365-10
(h=365nm, 1.4 mW at 20 mA, lo, TO-46 ball lens); Excitation filter: Roithner
Lasertechnik, RLT-365-12-A (center band: 365nm); Emission filter: Edmund
Optics, Filt. 450nm Broad PO-016746 (center band 450nm); Cell: Quartz ultra-
micro cuvette, Hellma 105.251-QS, volume: 45 pl.
 65

band-pass band-pass filter

LED LED fiber optic probe for

cuvette uorescence measurements
source souroe

Figure I Schematic diagram of the Juorometer (left) and its possible imple~ntationusing fiber
optics (right)

Preliminary tests were performed to check the stability of the system, the
entity of the background signal, and the possible noise contributions to the
fluorescence signal, under typical measurement conditions. The background
signal due to the pure solvent was estimated by using bi-distilled water, ethanol
and acetone. With the selected optical configuration, the lowest background
signal was obtained using bi-distilled water, which was about 280ort100 counts.
Bi-distilled water was therefore the solvent selected for the preparation of A M 1
samples.

Experimental
eri results
res~lts
The experimental tests were aimed at measuring the titration curve for the
A m 1 aqueous solutions in the 0-125 ppt range, step 25ppt, which is the interval
of interest for practical applications. All the samples were prepared by diluting a
standard solution of AFMl (1 ppm; Suppl.: BIOPURE Tulln, Austria). It should
be emphasized that the fluorescence signals were directly detected in the
solutions prepared at the concentrations reported, without any pre-concentration
of the samples, as usually is required in HPLC measurements.
The high sensitivity of the fluorometer had the major drawback of the high
susceptibility of the quantitative result to several sources of uncertainty in the
measurement and in the sample preparation. In addition to the practical difficulty
of obtaining identical replicas of a sample, especially in the case of very low
AFMl concentrations, other sources of uncertainty were present, such as
cleanliness of the cell, and impurities or air bubbles in the liquid. Consequently,
in order to ensure measurement reproducibility, an experimental protocol was
established for: a) sample preparation, b) the method of injecting samples into the
cell, aimed at avoiding the formation of micro-air bubbles; c) cell cleaning after
measurement; d) check of the cell cleanliness by measuring the reference liquid
(pure water).
66

Six series of samples covering the 0-125 ppt range were prepared and
measured in independent sessions. The experimental results, together with the
error bars, are shown in Figure 2. Each point of the curve was calculated as mean
value over six measurements.
Since the main interest was to develop a reliable screening device able to
rapidly single out situations of potential risk, starting from the titration curve
three main thresholds were identified:
- level 1 - safe, A w l < 25 ppt: [2000-29001 photo-counts;
- level 2 - alert for baby-food, 25 ppt < AFMl c 50 ppt: [2900-37701
photo-counts;
- level 3 - alarm, AFMl > 50 ppt: over 3770 photo-counts.
Thus, based on this simplified reference scale, the fluorometer could be used
as the starting point for tree decisions when measuring fluorescence of
unknown samples: if the result is level 1, the sample is considered safe and the
control chain ends at this point; if the fluorescence value falls in level 2, the
sample is potentially contaminated, and more thorough analyses are needed; if
the result is level 3, the sample belongs to a stock contaminated over the admitted
level, and the stock should be discarded.

Conclusions
A portable fluorometer was proposed for the rapid and straightforward
detection of low concentrations of AFMl, without any need of pre-concentration
of the sample. The technical features of the fluorometer were presented, with a
special focus on its potentialities as an early warning system, which makes it
possible to distinguish between safe and potentially contaminated samples,
according to the legal definition.
Although still limited to tests with synthetic solutions of AFMl , the results
obtained encourage us to regard t h s prototype as the starting point for a new
low-cost technology for the rapid screening of AFMl in milk. Further
development could involve a procedure suitable for non-trained users based on
low-cost materials, such as disposable plastic cells.

Acknowledgments
Project MIUR-FIRB #RBNEOl KZZM Biosens is gratefully acknowledged for
the partial financial support. The authors would like to thank Mr. Franco Cosi for
the technical assistance.
 67

6500 1_7
.....-..
...... .
.....
..
I
A !
6000
f 41
5500
EB
3s
5000
9Q
4500 =.5Q P P ~

4wo * I

3500

3000

2500

2000

Figure 2 Titration curve for AFMl aqueous solutions in the 0-125 p p t concentration range

References

IARC Monographs on the Evaluation of Carcinogenic Risks to Humans. International

Agency for Research on Cancer, Lyon, 56 (1993).
P.M. Scott, J. Assoc. 08 Anal. Chem. 70, (1987), 276.
J. Leitao, G. De Saint Balnquat, J. R. Bailly, Ch. Paillas: J . Chromatogr. A, 253,
(1998), 229-234.
R. Alleroft, R. B. A. Carnaghan, Vet. Rec. 65, (1963), 259.
H.P. Van Egmond, Introduction in Mycotoxins Dairyproducts, Elsevier Applied
Science, London 1998.
FAO, Worldwide regulations for mycotoxins 1995 - A compendium. FA0 Food and
Nutrition Paper, 64 Rome, ISSN0254-4725 (1 997).
European Committee Regulation No. 1525, Bruxelles, 16 July 1998.
W.Th. Kok, Th.C.H. van Neer, W.A. Traag, L.G.M.Th Tuinstra; J . Chromatogr, 367,
(1986), 231-236.
M.J. Sheperd, J. Gilbert; Food Add. Contam. 1 (1984), 322.
l o H. Joshua, J . Chromatogr. A 654, (1993), 247-254.
0. J. Francis, G.P. Kirschenheuter, G.M. Ware, A.S. Barman, S.S. Kuan, J . Assoc. 08
Anal. Chem. 7 , (1988), 725-728.
l 2 PMT-FL fluorometer, Fialab Instruments Inc., Bellevue WA, http://www.fialab.com/ .
l 3 J. Chelkowski, Photochem. Photobiol. 20, (1974), 279-280.
I4 W.E. Steiner, K. Brunshweiler, E. Leimbacher, R. Schneider: J. Agric. Food Chem. 40,
(1992), 2453-2457.
 Quality monitoring of extra-virgin olive oil
using an optical sensor

A.G. Mignani, L. Ciaccheri, A.A. Mencaglia

CNR-IFAC, Sesto Fiorentino (FI), Italy
1.ciaccheri@ ifac.cnr.it

R. Paolesse, C. Di Natale
Universita di Roma Tor Vergata , Dipartimento di Scienze e Tecnologie
Chimiche, Dipartimento di Ingegneria Elettronica, Roma, ltaly

A. Del Nobile, R. Benedetto, A. Mentana

Universita di Foggia, Dipartimento di Scienze degli Alimenti, Foggia, Italy

Summary
An optical sensor for the detection of olive oil aroma is presented. It is capable of
distinguishing different agcing levels of extra-virgin olive oils, and shows effective potential
for achieving a non destructive olfactory perception of oil ageing. The sensor is an optical
scanner, fitted with an array of metalloporphyrin-based sensors. The scanner provides
exposure of the sensors to the flow of the oil vapor being tested, and their sequential spectral
interrogation. Spectral data are then processed using chernornetric methodologies.

Introduction
Extra-virgin olive oil is usually bottled so that a head space is left between the
oil surface and the cap. During the oils shelf life, this head space becomes
saturated by oil vapors. These vapors are true indicators of the quality of the oil,
since the aroma of the oil changes drastically to rancid because of ageing or
improper conservation. Several sensors have been proposed for this purpose, and
are usually based on conductive polymers which simulate the olfactory
perception [l-31. A non destructive system would be advisable to detect the state
of the oil aroma during its shelf life without opening the bottle.
This work presents preliminary testings of an optical sensor for oil aroma
detection that potentially offers the implementation of a smart cap, to be used as
a non invasive system for oil aroma detection. The sensor presented here makes
use of metalloporphyrin-based materials as oil quality indicators. Since these
materials change color and, consequently, the absorption spectrum when exposed
to oil vapors, absorption spectroscopy was used for the implementation of an
optical sensor. An optical scanner fitted with the array of metalloporphyrin-based

68
 69

sensors was implemented, whch provided exposure of the sensors to the flow of
the oil vapor being tested and their sequential spectral interrogation. Since
metalloporphyrins are highly sensitive to a wide range of volatile compounds, but
are lacking in selectivity, the overall array response was processed by means of
chemometric methodologies.
T b s system is capable of distinguishing different ageing levels of extra-virgin
olive oils and shows effective potential for achieving a non invasive olfactory
perception of oil ageing.

The optical scanner

A fiber optic scanner capable of carrying out absorption spectroscopy was
implemented in order to address an array of metalloporphyrin-based sensors
operating in the visible spectral range. The scanner consisted of three
independent modules (Figure 1) [41:
1) An absorption spectroscopy module for spectral interrogation of the
sensor array, which was a custom-made fiber optic spectrophotometer
capable of achieving transmission measurements in the 380-780 nm spectral
range, with a resolution of 10 nm [ 5 ] .
2) A mechanical module consisting of a gear that housed the sensors and
provided them with exposure to the oil vapor being tested, and their
sequential interrogation. The gear was a revolving platform rotated by means
of a step motor. The revolving platform housed 16 replaceable glass disks
arranged around a ring. One disk was taken as a reference for signal
normalization. The other 15 disks were the sensors, which were spray-coated
by means of metalloporphyrins based on cobalt, iron, copper, manganese,
zinc and tin. The platform was fitted into a flow cell suitable for vapor flow.
The optical fibers for illumination and detection, which were coupled to the
spectrophotometer, were fitted to the mechanical module in a fixed position,
so that they could perform transmission measurements of all glass disks
while the revolving platform rotated. In practice, the spectral interrogation of
the whole sensor array was achieved while the oil vapor flowed through the
cell. Gaskets were used in many parts of the cell to provide gas proofing.
3) A software module, which was a Labview-driven interface capable of
managing the optical and mechanical modules and of recording the spectral
response of all the sensors.

The metalloporphyrin-based sensors are highly sensitive to volatile compounds

[6], but are lacking in selectivity. Consequently, the overall sensor response was
70

processed by using chemometric methodologies in order to achieve qualitative

information similar to that obtained by means of olfactory perception.

Miniaturizedspeclrophotorneter
for aosorption specrroscopy

Flow cell with

electromechanicalscanner for PG Labview and data processing
sensor array interrogation

Figure 1 The optical scannerfor spectral interrogation of the metalloporphyrin-based sensor-array

Experimental results
ental re$~lt$
Two types of extra-virgin olive oils from the Italian region of Puglia,
Peranzana and Taggiasca, respectively, were artificially aged by exposure to
100C for 50 hours. The vapors of the original oil and the aged samples were
analyzed by means of HPLC measurements, revealing a total content of 45-50
ppm of n-hexaldehyde, octyl-aldehyde, nonanal, and heptaldehyde vapors.
For the optical experiments, a glass vial was used to contain the oil sample
being tested: the vial was partially filled with the oil. The vapor generated by the
oil in the vial head space was carried to the flow cell containing the optical
sensors by means of nitrogen. The metalloporphyrin-based sensor-array was
spectrally interrogated when exposed to new and oxidized oils, and PCA
processing of sensor-array spectra was computed [7]. The PCA processing
linearly combined the spectral data of the 15 sensors characterizing the oil aroma
in order to produce new variables or principal components (PCs). High-order
PCs had little weight in characterizing the samples, and could be disregarded
with little loss of information. In practice, only PCs 1, and 2 were found to be
significant for sample identification. Therefore, the two-dimensional map in the
PC1-2 subspace was populated by point clusters, each of which grouped the oil
samples in terms of the similarities of their aroma. Figure 2 shows the resultant
PCA map. Oil OA and Oil 8A are the Peranzana oil samples, new and aged,
respectively. Oil OB and Oil 8B are the Taggiasca oil samples, new and aged,
respectively. This figure shows also that the metalloporphyrin-based sensor array
 71

was capable of distinguishing between new and oxidized oils for both types of
oils.

Conclusions
A metalloporphyrin-based sensor-array has successfully been used to assess
the aroma of new and artificially-aged extra-virgin olive oils. When exposed to
the vapor of the oil under test, the absorption spectra of the sensor array was
measured by means of a fiber optic-based scanner. A chemometric methodology
was used in order to achieve qualitative information similar to that obtained by
means of olfactory perception. These preliminary results show that it is possible
to implement a smart cap capable of revealing the rancidity of the oil contained
in a bottle without having to open it. The smart cap should be designed as a
suitably sandwiched sensor array which is in contact to the oil vapors on one
side, while allowing reflectance spectroscopy measurements on the other. A cap
of this type could follow the oil aroma during the bottles shelf life, thus
revealing the onset of rancidity.

0.4

0.2 oxiding v
a O~IOA
V 0118A
h

8 0
2
N
v
N -0.2
2
-0.4

-0.6
-1 -0.5 0 0.5 1
PC 1 ( 55.7 % )

Figure 2 Metalloporphyrin-based sensor-array response: Oil OA and Oil 8A are the Peranzana
oil samples, new and aged, respectively; Oil OB and Oil 8B are the Taggiasca oil samples
new and aged, respectively
72

Acknowledgments
Project MIUR-FIRB #RBNEOl KZZM Biosens is gratefully acknowledged
for the partial financial support. The authors would like to thank Mr. Franc0 Cosi
for the technical assistance.

References
1. Guadarrama, M.L. Rodriguez-MCndez, J.A. de Saja, J.L. Nos, J.M. Olias,
Array of sensors based on conducting polymers for the quality control of the
aroma of the virgin olive oil, Sens. Act. B , vol. 69, n. 3, 2000, pp. 276- 282.
2. J. Martinez, G. Pioggia, M.L. Rodriguez-MCndez, J.A. De Saja, Electronic
nose for the quality control of the olive oil aroma discrimination of quality,
variety of olive and geographic origin, Proc. ISOENO2 - 9TH International
Symposium Olfaction & Electronic Nose, Eds. Arnaldo DAmico, C. Di Natale,
ISBN 88-7999-450-6. Aracne Editrice Srl, Roma, 2003, pp. 220-223.
3. S. Buratti, S . Benedetti, M.S. Cosio, Evaluation by an electronic nose of olive
oil oxidation during storage, Italian J . Food Sci., vol. 17, n. 2, 2005, pp. 203-
210.
4. R. Paolesse, C. Di Natale, A. Macagnano, S. Nardis, A. DAmico, A.A.
Mencaglia, P. Bizzmi, A.G. Mignani, Fiber optic multimeter for interrogating
an array of absorption-based optochemical sensors, Proc. SPIE vol. 5270
Environmental Monitoring and Remediation I l l , 2004, pp. 140-143.
5. A.A. Mencaglia, A.G. Mignani, Optical fiber instrumentation for online
absorption and reflection spectroscopy, Proc. SPIE vol. 4763 European
Workshop on Smart Structures in Engineering and Technology, 2003, pp. 248-
251.
6. The Porphyrins, D. Dolphin Ed., vol. 111, Academic Press, New York, 1978.
7. I.A. Cove, J.W. McNicol, The use of Principal Component in the analysis of
the near infrared spectra, Appl. Spectr., vol. 39, 1985, pp. 257-266.
 Spectral Interrogation of Optical Metal-Cladding Waveguides
for Chemical Sensing

R. Bernini
IREA- CNR, Via Diocleziano 328, 80124 Napoli, Italy
bernini .r @ irea.cN.it
F. Mottola, A. Minardo, L. Zeni
Dipartimento di Ingegneria dellhformazione, Seconda Universita di Napoli,
via Roma, 29, 81031 Aversa (CE), Italy

Summary
An integrated optical sensor, based on a Metal-Cladding Leaky Waveguide is proposed for
chemical and biological applications.The proposed sensor permits to measure the refractive
index of a liquid placed on its surface by simply monitoring the reflected optical spectrum.
The device has been realized by using polymeric materials and the experimental results
confirm the numerical simulations. The sensor exhibits sensitivity, calculated for water as
the external medium, higher than 2000 nm.

1 Introduction
Optical chemical sensors have been widely studied in the past two decades.
The interest for optical sensors is due to several advantages provided by optical
methods, such as flexibility, high sensitivity, small size, low price, potential of
mass production and the absence of electromagnetic interference. Typically,
these sensors are based on the interaction of the evanescent part of a guided
mode with the sensing medium [ 11. In a three-layered waveguide (substrate,
guiding layer, external medium), the evanescent field is in contact with the
external medium; a change in the external medium leading to a variation of the
evanescent field properties. As a result, modal properties change. These changes
are translated into a variation of the effective refractive index andor the losses of
the waveguide.
Evanescent field techniques have been employed to detect various particles in
water, such as in Reverse Symmetry Waveguide [ 2 ] , Resonant Mirror [3], and
Surface Plasmon Resonance [4].
Also metal-cladding leaky waveguides (MCLWs) have recently attracted a
great interest in chemical and biosensing applications, because of their enhanced
sensing capabilities with respect to traditional waveguides based on high
refractive index guiding films [ S ] . Enhanced sensitivity comes from the deeper
penetration of the evanescent film inside the sensing medium, as the metal film
pushes the waveguide mode further into the sensing region.
Typically, MCLWs are interrogated by a monochromatic source, e. g. a laser,
measuring the reflected spectrum in function of the angle of incidence. In this
work we present a spectral interrogation of MCLW, where the angle of incidence

73
74

of the source is kept constant while measuring the reflected intensity as a function
of the source wavelength.

2 Theoretical background
Optical Metal-cladding leaky waveguides (MCLW) are leaky waveguides in
which a metal cladding, inserted between the guiding layer (lower refractive
index) and the substrate (higher refractive index), improves the confinement of
light in the core region, so that the light lost at each reflection with the metal
cladding is reduced.
Metal-cladding waveguide can be used as a chemical sensor, relying on the
interaction between the evanescent field of the guided mode, and the analyte. By
a suitable design, the penetration depth of the evanescent field in the external
medium is large, compared with both conventional dielectric waveguides and
surface plasmon resonance [ 6 ] , resulting in a stronger interaction with the
analyte. In Figure 3, a typical modal profile of the fundamental TE mode of a
MCLW is shown; as it can be seen the evanescent field of the mode has a quite
deep penetration in the sensed medium. This can be useful e.g. in the sensing of
particles larger than 1pm diameter, such as bacteria or spores.
Typically, a MCLW sensor is operated in reflection mode, in which the
waveguide is illuminated from the bottom side, by means of the Kretschmann
configuration, while measuring the reflected intensity as a function of the angle
of incidence. This results in a clear dip in the reflected intensity at an angle of
incidence that matches the waveguide mode angle [ 7 ] . Here we present a novel
interrogation method for MCLW based sensors, in which the angle of incidence
of light is kept constant while varying the wavelength of the incident light. In this
case, the reflectance is measured as a function of wavelength. This configuration
presents the advantage of not requiring a high-precision goniometer to evaluate
the angle.

I
a 4
It

05 1 1.5 2
LaLwm-m
Fig. 3 - Mode profile of a MCLW, evaluaied at a wavelength of 633' n m
 75

Experimental results
The sensor consists of a glass substrate, whose refractive index is nl=1.51, on
which a gold thin film (nz=O.12+3.29i, d2=30nm), acting as a metal-cladding, is
sputtered. The guiding layer is deposited by means of the spin coating technique,
and consists of a PMMA film (n3=1.49, d3=300nm). All refractive indices are
evaluated at a wavelength of 633nm.
In Figure 4, the scheme of the experimental setup is shown. Reflection
spectrum measurements were carried out by coupling the light at an angle of
64.5" from the substrate, by means of a coupling prism. A white lamp, followed
by a collimator and a polarizer, was used as the source, whereas a
spectrophotometer was employed for the acquisition of the reflected spectrum.

MCLW sensor

(white lamp)

Fig. 4 - Experimental setup

Fig. 5 -Reflected spectra'for the MCLWat three different values of the external refractive
indexes (no=1.333; 1.337; 1.341)
76

In Fig. 5 the reflected spectra of the MCLW are shown for three different
external refractive indexes (b=1.333; 1.337; 1.341).
As the refractive index of the external medium q, changes, so does the
wavelength at which light is coupled in the waveguide. Hence, the position of the
dip of the spectrum is a measurement of the external medium refractive index,
within an opportune range. The dip in reflectivity is sharper, compared with
surface plasmon resonance. Defining the sensitivity as the ratio between the
change in resonant wavelength and the spectral width of the dip, the improvement
of MCLW, compared with SPR, can be up to approximatively of a factor of 6
[71.
In Figure 6, the refractive index of the sensed medium n,, as a function of the
resonant wavelength (i.e. the dip position of the reflected spectrum) hR is shown
for both theoretical and experimental data. As it can be seen, there is a good
agreement. The sensitivity S, = aAR/an, calculated for a refractive index of the
external medium ~ = 1 . 3 3 3 0is higher than 2000 nm.

1.346

1.344- + --
-Theoretical

D P positionM
Fig. 6 - Analyte refractive index infunction ofthe resonant wavelength of the MCLW.
Comparison between experimental and theoretical data

4 Conclusion
Metal cladding leaky waveguides have been presented as chemical sensors.
The sensor is realized using gold as metal-cladding, and PMMA as the guiding
layer of the MCLW. The validity of the sensor has been tested by evaluating the
spectral response for different refractive indexes of the external medium, and the
experimental data are in good agreement with the theoretical ones. The proposed
sensor can be used for a number of applications, such as chemical sensing, by
 77

using chemically selective polymers, biological sensing of large particles and

cells, and for highly efficient fluorescence detection.

References
[ 11 P. V. Lambeck, Sens. Act. B 8, 103-1 16, 1992.
[2] R. Horvath, L. R. Lindvold, N. B. Larsen, Appl. Phys. B 74, 383-393,2002
[3] R. Cush, J. M. Cronin, W. J. Stewart, C. H. Maule, J. 0. Molloy, N. J.
Goddard, part 1-11,Biosens. Bioelectron. 8, 347-354, 1993.
[4] J. Homola, S. S. Yee, G. Gaugltitz, Sens. Act. B 54, 3-15 1999.
[5] M. Zourob, N. J. Goddard, Biosensors and Bioelectronics 20, 1718-1727,
2005.

[6] M. Zourob, S. Mohr, P. R. Fielden, N. J. Goddard, Sens. Act. B 94, 304-312,

2003.
[7] N. Skivesen, R. Horvath, H. C. Pedersen, Sens. Act. B 106,668-676,2005.
 Application of Quality Standard for the planning, design and
realisation of thick film gas sensors

M. BuzzolaniI*, M. Butturi,
C. Mala&, G. Martinelli, B. Vendemiati

University of Ferrara, Polo Tecnologico - via Saragat, 1 - 44100 Ferrara, Italy

[email protected]

Summary
The Sensor and Semiconductor Laboratory (SSL) of the Department of Physics of the
University of Ferrara has implemented a Quality Management System (QMS) about
processes of planning, development and production of thick-film gas sensors and
photovoltaic cells. Since March 2006 SSL has also certified the process of managing
and monitoring of Project.
The applied International Standard, UNI EN IS0 9001 :2000, represents the standard of
reference to match the requirements of Quality Management Systems. Espccially thc
documentation (and in particular the Quality Manual) of an organization should be
configured in a way that is appropriate to its unique activities.
The management system of the SSL, based on the scheme Plan, Do, Check-Act, allows
the monitoring of the laboratory activities to get continual improvement.
The principles of quality have been applied for the first time to the research activity in
the University, and they turned out to be simple but important tools for the success of
the system. The international standard applied to SSL is based on four main elements
which are: manager responsibility, resource management, design and development (and
the management of projects), and measurement and analysis.

1. Introduction
The Project to certify the Sensor and Semiconductor Laboratory according to
the Normative Vision 2000, started in 2000. The Top Management chose to
improve the Quality Policy through the principles of Customer Satisfaction,
Efficiency and Excellence of processes.
The main goal of the SSL has been defining an internal method of working to
coordinating the several activities of the teamwork. In particular to promote
effectiveness and efficiency to the management of the projects there has been
important spread of quality culture and an approach to goals. To get good
results it is necessary to apply a systematic approach to design and development
and pay particular attention to the key processes of management and planning

78
 79

projects. To provide an effective framework for development of activities the

Laboratory should plan:

1. Objectives
2. Activities and Processes (of the different projects)
3. Resources
4. Inputs and Outputs
5. Review of design and development of prototypes
6. Monitoring of the projects

Y
CUSTOMER CUSTOMER

INTERESTED PART/
NEEDS
(partners,
laboratories I /
LEJ I
;.
INPUT
Develooment

Manaaement
of Proicts
OUTPUT
PRODUCT/SERVICE
(senson, cells, know
how, information )

Figure 1. SCHEME "Plan-Do-Check-Act'' (PDCA) applied to QMS of SENSOR AND

SEMICONDUCTOR LABORATORY

2. Application
The SSL has applied the principles of the standard IS0 9001:2000 to the main
processes such as management of instruments and facilities, purchasing,
management of human resources etc. Particular attention has been paid to the
activities of design and manufacturing of thick film gas sensors. The Top
Management of the Laboratory decided to apply the IS0 9001:2000 (standard to
get the certification) in 2000. Since September 2001 the SSL has certified the
processes of design and manufacturing of thick film gas sensors and
80

photovoltaic cells. The main processes about planning and realization of thick
film gas sensors are described in the Quality Manual (QM) [3] and in the
procedures of the organization. The QM expresses the policy, the mission and
the vision of the Laboratory while the procedures describe the main operational
processes (both general and specific). As a consequence there are procedures
that describe supporting activities (purchasing, nonconformity, management of
human resources etc.) and procedures that specify peculiar processes (sensors
planning, sensors development and production etc.).
The process of monitoring of the projects is described in the QM and the main
activities are expressed in the Project Management Procedure. The structure
of this procedure is:

- scope
- application field
- standard references
- modus operandi
- flow-chart
- attachments

Since March 2006 the SSL has certified the process of Managing and
Monitoring Projects.

3. Conclusions
In a context of scarce resources it is important to realize projects in an effective
way to show customers and working partners the capacities of the Lab to get
the expected goals. So the development of a common working method and a
continual monitoring of the activities (from the financial and technical point of
view) allows one a simplification of the work and reduces time and risk.
Creating a project-management procedure represents an important strategic
choice for an organization as a laboratory of research that collaborates with
industrial partners and important national and international organization of
research.

References
[ l ] IS0 9001:2000 http://www.iso.org
[2] IS0 9000:2000 http://www.iso.org
[3] QM LSS ED. 2005 - University of Ferrara, Department of Physics
 A Photonic Bandgap Fiber Sensor Based On Single Walled
Carbon Nanotubes As Sensing Material
M.Pisco, M.Consales, A.Cutolo, A.Cusano*
Optoelectronic Division - Engineering Department, University of Sannio,
Corso Garibaldi 107, 82100 Benevento, Itab, [email protected]
S .Campopiano
Department for Technologies, University of Naples Parthenope,
Via Medina 40, 80131 Napoli, Italy
M .Giordano
Institute of Composite and Biomedical Materials, CNR, 80124, Naples, Italy
M.Penza, P.Aversa, GCassano
ENEA, C.R. Brindisi, Materials and New Technologies Unit, 72100 Brindisi,
Italy

Summary
In this work, the feasibility of using Hollow-core Optical Fibers (HOFs) coated by carbon
nanotubes for gas detection applications has been investigated. Single Walled Carbon
NanoTubes (SWCNTs) overlays have been deposited by the Langmuir-Blodgett technique
onto HOFs. The success of the SWCNTs deposition onto the HOF has been confirmed by
Scanning Electron Microscope observations carried out on the samples after the deposition
procedure. Reflectance characterization of the obtained sensing probes has been carried out
in the range 1530-1570nm and an inverse Fourier transform based method has been
employed in order to estimate the degree of penetration of the carbon nanotubes inside the
hollow structure. Finally the sensing capability of the HOF sensors has been investigated by
exposure in a test chamber to Tetrahydrofuran. The preliminary results obtained demonstrate
the capability of the novel HOF sensor to perform chemical detection of volatile organic
compounds showing a good sensitivity and fast response times.

1 Introduction
Since the first experimental demonstration of Photonic Bandgap (PBG)
guidance in Hollow core Optical Fibers (HOFs) in 1998 by Knight et al. [ 11, the
optical properties of such fibers have attracted significant attention in the
scientific community and several sensing applications using the HOFs have been
developed [2-51. Furthermore, in the recent years, the sensing properties of the
carbon nanotubes deposited onto a singlemode standard optical fiber (SOF)
configured in a reflectometric sensor system have been widely investigated [6-81
demonstrating their capability to perform chemical detection of volatile organic
compounds (VOCs) at room temperature. As well known, the sensitivity of a
refractometric based thin film sensor can be enhanced by increasing the

81
82

refractive index mismatch at the interface between the fiber core and the film.
Hence, the HOFs, presenting a lower effective refractive index of the
propagating mode, can be a promising substitute of the SOF based sensors. In
addition, by filling the air holes of the HOF with a sensitive material, an high
interaction between the light and the sensing material is expected.
In this work, a novel HOF chemical sensor is presented. The sensor probe is
constituted by an HOF covered and partially filled with Single Walled Carbon
NanoTubes (SWCNTs) as sensing material. The sensing system principle of
operation relies on the measurement of the light intensity reflected at the
interface between the fiber and the sensing material. Since the presence of THF
modifies the optical features of the carbon nanotubes, the sensing material
dielectric function variations, as well as the losses due to the fiber PBG
modification, lead to a reflectance modification of the sensing overlayer.

2 Sensor fabrication and characterization

The Langmuir-Blodgett (LB) technique [6-81 has been chosen to transfer
SWCNTs monolayers onto the HOF. Once the HOF end has been properly cut
by using a precision cleaver to obtain a planar cross-section, the LB deposition
technique has been used in order to deposit twenty monolayers of SWCNTs onto
an HOF section with a length of about 9 cm.
After the deposition procedure, the HOF has been spliced at the input end to
a singlemode SOF terminated by a fiber connector in order to improve the
coupling efficiency with the source. The two fibers have been spliced by using a
commercial electrical arc splicing system (Fujikura FSM-50s). The fiber ends
have been properly cleaved and cleaned, then they have been aligned and
pressed against each other using the splicer precision motors. In order to avoid
the collapse of air holes, a series of arcs with short duration (100ms) and high
power (80bit power) has been forced. The splicing losses were estimated to be
about 1dB and are mainly caused by mode field mismatch and light reflection
from the silica-air interface. The bonding achieved between the two fibers has
been enforced by using an heating protection sleeve.
In order to investigate the spectral features of the sensing probe, the
reflectance at the SOF-HOF interface has been measured within the HOF
bandwidth. To the aim, a Superhminescent Light Emitting Diode (SLED) with
36nm bandwidth around 1556nm and an optical spectrum analyzer (Ando
AQ63 15A) were used.
The spectrum of the signal reflected by the HOF sensor probe has been
normalized with respect to the reflected spectrum of the same HOF without
carbon nanotubes overlayer. The retrieved reflectance is reported in figure l(a).
Several interference fringes have been manifested. In order to determine the
 83

degree of penetration of carbon nanotubes within the hollow structure, the

Inverse Fourier Transform (IFT) of the reflectance has been performed.

-1 0,
- 0 006

-
r_
R -1.2 'E
k8 -1.4 (I)
v
0.004
C =
-a
3 5 0.002
'
G
-1.6

-1.6
4
0.000
-2.0
1530 1540 1550 -70 0.00 0.05 0.10 0.15 0.20 0.25
Wavelength (nrn) 2nd (m)
(4 (6)
Fig.1 Reflectance of the HOF coated by 20 monolayers of SWCNTs (a)
IFT of the HOF sensor reflectance (b)

In figure l(b) the result of the IFT on the reflectance of the sensing probe
has been reported. The spectrum exhibits several peaks corresponding to the
axes coordinates 2cm, lscrn, 1832cm and 1839cm. The harmonic components,
corresponding to a fixed z=2wd, represent the reflected power contribution due
to the power portion going through a path 2d long in a medium with mean
refractive index n. It is worth noting that the harmonic contribution in the low
region of the z-domain has been attributed to the source ripple and to the
intermodal interference within the HOF. The peaks at 2cm, 18cm, 1832cm and
20cm reveal the presence of a double resonant cavity.

Fig.2 AFM image of an hollow fiber (a) and

SEMimage of the HOF coated by I 0 monolayers of SWCNTs @)

The one is composed by the HOF presenting a length of 9cm. The other one,
with a 2nd-lenght of 2cm, represents the resonant cavity constituted by the
carbon nanohibes penetrated within the hollow structure during the LB
deposition. In spite of the difficulty to offer an reliable estimation of the
refractive index of the SWCNTs agglomerate, it is possible to deduce that the
sensitive material has been able to adhere on the fiber glass substrate and to
penetrate inside the hollow structure along a 0.5-lcm length. The residual peaks,
18&9cm, are attributed to the HOF higher order modes. In order to investigate
the morphological characteristic of the fabricated HOF sensor, a SEM have been
84

used. In figure 2 the AFM image of the HOF before the deposition and the SEM
image after the deposition of 10 monolayer of SWCNTs are showed
demonstrating the success of the SWCNTs deposition onto the HOF.

3 Testing and results

In order to test the sensing performances of the fabricated sensor, reflectance
measurements have been carried out by using the interrogation system [8]
schematized in Fig. 3 (a). The optoelectronic sensor output, consisting in the
ratio between the reflected signal and the one corresponding to the power source
monitoring, is proportional to interface reflectance [81.
The optical sensor has been located in a test chamber for THF exposure
measurements. Dry air has been used as reference gas and carrier gas to transport
the VOCs of THF with different concentration pulses. The test cell containing
the sensor had a volume of 1200ml whereas the total flow rate for each exposure
has been kept constant at 1000 ml/min. The gas flow rate has been controlled by
a mass flowmeter driven by a controller-unit. The VOCs vapors have been
generated by the bubbling method with a thermostated flask containing the
liquid analyte.

mmm

150 200 250 300 350

Time (min)

Fig.3 Experimental setup (a) Time responses of HOF sensor coated by 20 monolayers of SWCNTs.
exposed to 3 d$ferent concentrations of THF vapors (3)

The optical response of the SWCNTs-based sensor to 30 minute pulses of

THF vapors at three decreasing concentrations is showed in figure 3 (b). The
signal decreasing upon exposures to THF vapors can be attribuited to a change
of the SWCNTs complex refractive index. Moreover, as the carbon nanotubes
filled the cladding holes of the HOF, a modification of the mode propagation
characteristic occurs and could be responsible for the optical signal decreasing.
The revealed sensitivity for low concentrations is -1.2.10-5ppm-' and a limit of
detection (LOD) of 40ppm with the exploited instrumentations is allowed,
whereas, according to the european commission directive 2000/39/EC, the THF
permissible exposure limits (PELS) are 50ppm in a long time (8 hours) and
lOOppm in a short time (15 minutes) exposure. The time response also
 85

demonstrates a weak dnft due to thermal changes occurring in the test chamber
during VOC exposure. Finally, good results have been obtained in terms of fast
adsorption and desorption kinetics. In particular, response times less than 15
minutes and recovery times less than 30 minutes have been observed.

4 Conclusions
An HOF coated and partially filled with SWCNTs has been used for the first
time for VOCs detection. The success of the SWCNTs deposition onto the HOF
by LB technique has been confirmed by the SEM image of the sensing probe.
Spectral characterization has been carried out in the range 1530-1570nm
revealing a degree of penetration of the carbon nanotubes inside the hollow
structure of 0.5-lcm. Finally, the fabricated sensor has been located in a test
chamber and employed in a refractometric system involving single wavelength
reflectance measurements in order to test its sensing performances towards THF
vapors. The preliminary results obtained demonstrate the capability of the HOF
sensor to perform chemical detection of THF at room temperature with a
sensitivity of -l.2.lO- ppm-, a LOD of 40ppm and response times of a few
minutes.

References
1. Knight, J. C., Birks, T. A., Russell P St, J. and Atkin, D. M., Opt. Lett.,
21(19), (1996), pp. 1547-9
2. J. C. Knight, Nature, vol. 424, (2003), pp 847-85 1
3. T. Ritari, J. Tuominen, H. Ludvigsen, J. C. Petersen, T. Sorensen, T.P.
Hansen, H.R. Simonsen, Optic Express, vol. 12, (2004), pp. 4080-4087
4. T.M. Monro, D.J. Richardson and P.J. Bennett, Electronics Letters, (1999),
Vol. 35, NO. 14, ppll88-1189
5. T. T. Larsen, A. Bjarklev, D. S. Hermann, J. Broeng, Optics Express, Vol.
11, NO. 2, (2003), pp.2589-2596.
6. M. Penza, G. Cassano, P. Aversa, A. Cusano, A. Cutolo, M. Giordano, L.
Nicolais, Nanotechnology, 16 (2005) 2536-2547.
7. M. Penza, G. Cassano, P. Aversa, F. Antolini, A. Cusano, M. Consales, M.
Giordano, L.Nicolais, Sens. Actuators B 111-112 (2005) 171-180.
8. M. Penza, G. Cassano, P. Aversa, F. Antolini, A. Cusano, A. Cutolo. M.
Giordano, L. Nicolais, Applied Physics Letters, vol. 85, ni 12, (2004) pp.
2379-2381
 Synthesis and selective nucleoside recognition of a new
substituted zinc-phthalocyanine

Luigia Longo, GiuseDpe Vasapollo

Dept. Ingegneria dell 'Innovazione, University of Lecce, Lecce, Italy
giiiscppe.vasapoIlo~~unile.il

Rosaria Anna Picca, Cosimino Malitesta

Dept. Scienza dei Materiali, University of Lecce. Lecce, Italy

Summary
A new zinc-phthalocyanine peripherally substituted with methacryloyl groups was
synthesized to be employed as functional monomer in the formation of molecularly
imprinted polymers as nucleoside receptors. The binding affinity and selectivity of the
synthesized phthalocyanine towards nucleosides were evaluated by UV-vis titration
experiments in CH2C12at 298 K. The binding constant (k) and Gibbs free energy changes (-
AGO) were calculated according to the modified Benesi-Hildebrand equation. Binding
experiments showed that K, of phthalocyanine with tri-0-acetyladenosine (TOAA) is 1.35 x
104, 500 times that of phthalocyanine with tri-0-acetyluridine (TOAU), indicating a high
selectivity of the synthesized phthalocyanine derivative.

1 Introduction
The development of synthetic receptors that recognize nucleotide bases and
their derivatives is an important area in chemistry today. The literature provides
many examples of artificial molecule receptors for each of the common
nucleoside bases [ 1-71. Many of these receptors showed selectivity in nucleoside
binding based on the positioning of interactive functional groups. A promising
approach to design synthetic receptors is the molecular imprinting technique [8].
The procedure involves the formation of a template-monomer complex mediated
by specific interactions, followed by polymerization. After the removal of the
template, complementary recognition sites remain in the polymer network.
Commercially available monomers such as methacrylic acid (MAA) have been
widely used, making polymer preparation a simple and facile process. An
approach based on cooperative interactions [2] could enable a better control in
the formation of high-affinity binding sites for each corresponding template
minimizing at the same time the inherent non-specific binding properties
common in non-covalent imprinted polymers.
We describe in this work the synthesis and characterization of a new zinc-
phthalocyanine peripherally substituted with methacryloyl groups (compound 5,
Scheme 1) to be employed as functional monomer in the formation of

86
 87

molecularly imprinted polymers (MIPS) as nucleoside receptors. The synthesis

was conducted by using precursors bearing the functional groups required to
obtain a three-dimensional binding site on the phthalocyanine plane in cross-
linked polymers to which the nucleotide could be specifically bound via
coordination by the phthalocyanine metal center and hydrogen
bonding/electrostatic interaction by modifiers linked with the phthalocyanine.
Cooperative interactions approach in MIP formation, already proposed [2], is for
the first time thought out for phthalocyanine compounds.
The binding affinity and selectivity of the synthesized phthalocyanine
towards nucleosides were evaluated by UV-vis titration experiments. An
organic soluble nucleoside derivative, tri-0-acetyladenosine (TOAA), was
utilized as prospective template, since it contains the 2-aminopyridine
substructure previously found to be important for polymer imprinting with
nucleotide derivatives [7]. Tri-0-acetyluridine (TOAU), not containing the 2-
aminopyridine moiety, was also tested for comparison [7].
R

ZDCH,C'W~
I-butanal

Scheme I Synthesis of compound 5

2 Experimental Section
Materials and methods. All chemicals were obtained commercially and
used without further purification. 'H and I3C-NMR spectra were recorded at
room temperature on a Bruker Avance 400 instrument operating at 400 h4Hz
with TMS as internal standard. Mass spectra were measured on an Agilent 1100
88

Series LCMSD system equipped with an electrospray ionization interface (ESI).

FT-IR and UV-vis spectra were obtained with a Jasco FT/IR-660 plus and a
Cary 100 Scan UV/vis spectrophotometers,respectively.
Synthesis of 4-(4-hydroxyphenoxy)phthalonitrile (1). 1.4 g (0.01 mol) of
finely ground anhydrous K2CO3 was added gradually (280 mg at intervals of 30
min) under nitrogen atmosphere to a stirred solution of 2 g (0.01 mol) of 4-
nitrophthalonitrile and 1.1 g (0.01 mol) of hydroquinone dissolved in 65 ml of
dry DMSO. The reaction mixture was stirred for 24 h at room temperature under
nitrogen, then it was filtered, added to ca. 50 ml of water, extracted with ethyl
ether and dried on anhydrous sodium sulphate. The organic solvent was
evaporated under reduced pressure and the crude product was purified by
column chromatography (silica, ethyl acetate/petroleum ether 4:6) and recovered
in 94 % yield. 'H-NMR (DMSO-d6): 6 9.64 (s, lH), 8.04 (d, J= 9.7 Hz, lH),
7.66 (d, J= 2.4 Hz, lH), 7.27 (9, J= 8.9 Hz, 1H) 7.02-6.99 (m, 2H),6.85-6.83
(m, 2H) ppm. I3C-NMR (DMSO-d,): 6 162.6, 155.7, 145.7, 136.6, 122.2, 122.1,
121.3, 117.1, 116.9, 116.4, 115.9, 107.7 ppm. FT-IR (ATR system): 3419, 2233,
1593, 1562, 1505, 1485, 1247, 1195, 1089,952,836 cm-'. MS: 53 (79), 63 (50),
76 (29), 81 (loo), 88 (13), 100 (78), 109 (67), 115 (lo), 127 (70), 236 (18) m/z
(%).
Synthesis of tetra(4-hydroxyphenoxy)phthalocyanine (2).
A mixture of 0,6 g (2,54 mmol) of 1 and 0,07 g (0,64 mmol) of hydroquinone
was pounded in a mortar and put in a Pyrex tube. The mixture was h s e d by
gentle heating to the melting point without vacuum, cooled, sealed under
vacuum and reacted at 180 "C for 20 h. The crude blue-green solid was washed
with ethyl ether and then purified by column chromatography (silica, THF/ n-
hexane 8:2). The product was recovered in 20% yield. FT-IR (ATR system): v
3627,3291,2954,2914,2871, 1615, 1506,1473, 1433, 1215, 1150, 1094, 1011,
867 cm-'. UV-vis (THF): A, 286, 339, 606, 638, 667, 703 nm. ESI-MS: calc.
M 946, obs. [M+l]+947 m/z.
Synthesis of tetra(4-methacryloyloxyphenoxy)phthalocyanine (3).
A mixture of 0,3 g (0,32 mmol) of 2, 0.26 g (2.5 mmol) of triethylamine and
0.26 g (2.5 mmol) of methacryloyl chloride in 15 ml of ethyl ether was stirred
for 18 h at room temperature under nitrogen atmosphere. The reaction mixture
was added to water, extracted with CHzCl:! and dried on anhydrous sodium
sulphate. The crude product, obtained after evaporation of the solvent under
reduced pressure, was purified by washing with n-hexane and recovered in 70%
yield. FT-IR (ATR system): v 2957, 1735, 1616, 1499, 1474, 1320, 1187, 1127,
1012 cm-I. UV-vis (CHC13): A,,, 285, 342, 606, 638, 665, 700 nm. ESI-MS:
calc. M 1218, obs. [M+1]' 1219 m/z.
Synthesis of Zn(I1) tetra(4-hydroxyphenoxy)phthalocyanine (4).
A mixture of 0,15 g (0,123 mmol) of 3 and 0,045 g (0,246 mmol) of zinc(I1)
acetate in 5 ml of 1-butanol was refluxed for 20 h under nitrogen atmosphere.
 89

After cooling to room temperature the dark green solid product was filtered off,
washed with n-hexane and dried under vacuum, giving the pure product in 90%
yield. FT-IR (ATR system): v 3374, 2955, 2913, 2870, 1608, 1506, 1473, 1395,
1339, 1220, 1084, 1044, 950 cm-'. UV-vis [(THF): A,, 282, 351, 610, 677 nm.
ESI-MS: calc. M 1008, obs. [M+l]' 1009 m/z .
Synthesis of Zn(I1) tetra(4-methacry1oyloxyphenoxy)phthalocyanine (5).
A mixture of 0,28 g (0,282 mmol) of 4, 0.24 g (2.26 mmol) of triethylamine
and 0.23 g (2.26 mmol) of methacryloyl chloride in 14 ml of ethyl ether was
stirred for 18 h at room temperature under nitrogen atmosphere. The reaction
mixture was added to water, extracted with CHzClz and dried on anhydrous
sodium sulphate. The crude product, obtained after evaporation of the solvent
under reduced pressure, was purified by washing with n-hexane and ethyl ether
and recovered in 28% yield. FT-IR (ATR system): v 2957, 1735, 1616, 1499,
1239, 1187, 1126, 1044, 947 cm-'. UV-vis (CHC13):A, 281, 347,679 nm. ESI-
MS: calc. M 1280, obs. [M+l]+ 1281 m/z.
UV-Visible titrations. Small aliquots (10 pl) of the solution of nucleoside in
CHzClz at 298 K were added to 5 x M of 5 in CHzClZ measuring the
absorbance at 680 nm until no longer increased. Solutions 5 x M of TOAA
and 0.5 M of TOAU were utilized. The effect of dilution caused by the addition
of nucleoside solutions on the absorption values was taken into account during
analysis and corrected.

3 Results and Discussion

The metal phthalocyanine 5 was synthesized according to the Scheme 1. The
analytical characterization confirmed the identity of this compound (see
Experimental section), UV-visible titration of phthalocyanine 5 with nucleoside
molecules was conducted at 680 nm in CH2C12 at 298 K. The association
constants of the complexes of 5 with nucleosides were determined by the
modified Benesi-Hildebrand equation, Eq. 1, where [HI0 and [GI0 refer to the
total concentration of 5 and nucleoside, respectively, AE is the change in molar
extinction coefficient between the free and complexed phthalocyanine, and AA
denotes the absorption changes at 680 nm of the phthalocyanine on addition of
nucleoside [9, lo]. For all nucleosides examined, plots of calculated [Hl0[G],JAA
values as a function of [GI, values gave excellent linear relationships, supporting
the 1:l complex formation. The binding constants (K,) and Gibbs free energy
changes (-AGO)of 5 with nucleoside molecules, obtained from usual curve fitting
analyses of observed absorbance changes, are summarized in Table 1.

CHloCGlo - 1
+ -[GI,
AA AEK, AE
90

Table I Binding constants (KJ andfree energy of complexation (-AGO)for the 1.1 complexes
between nucleoside molecules and 5 in CHIC12 at 298 K

nucleoside K, (dm3 mot') -AGO(KJ mot')

TOAA (1.35 f 0.5) x lo4 23.5
TOAU 27 f 2.5 8.2

4 Conclusion
A new Zn phthalocyanine derivative (compound 5) to be employed as
functional monomer in the formation of molecularly imprinted polymers as
nucleoside receptors was prepared. The binding affinity and selectivity of 5
towards different nucleosides (TOAA and TOAU) were evaluated by UV-vis
tritation experiments, which indicated that phthalocyanine 5 bound TOAA most
strongly showing in this way a high selectivity.

References
1. K.J. Shea, D.A. Spivak, B. Sellergren, J. Am. Chem. Soc., 115 (1993) 3368
2. J. Matsui, M. Higashi, T. Takeuchi, J. Am. Chem. SOC.,122 (2000) 5218
3. D.A. Spivak, K.J. Shea, Anal. Chim. Acta, 435 (2001) 65
4. D. Spivak, M.A. Gilmore, K.J. Shea, J. Am. Chem. Soc., 119 (1997) 4388
5. K. Yano, K. Tanabe, T. Takeuchi, J. Matsui, K. Ikebukuro, I. Karube, Anal.
Chim. Acta, 363 (1998), 111
6. H. Tsunemori, K. Araki, K. Uezu, M. Goto, S. Furusaki, Bioseparation, 10
(2002) 3 15
7. D.A Spivak, K.J. Shea, Macromolecules, 31 (1998) 2160
8. M.J. Syu, J.H. Deng, Y.M. Nian, Anal. Chim. Acta, 504 (2004) 167
9. Y. Turgut, E. Sahin, M. Togrul, H. Hosgoren, Tetra: Asymm., 15 (2004)
1583
10. H.A. Benesi, J.H. Hildebrand, J. Am. Chem. Soc., 71 (1949) 2703

Acknowledgements
The work was supported by PRIN 2004 - MIUR (No 2004034021-003).
 Synthesis and characterization of nanomaterial as sensitive
platform to assemble selective chemical sensors for NOz-
detection

Vanessa Bianiotti, Federica Valentini, Danila Moscone, Giuseppe Palleschi,

Universith degli studi di Roma Tor Vergata, Dipartimento di Scienze e
Tecnologie Chimiche, via della Ricerca SAcientifzca 1, 00133 Roma (Italy)
vanessa.biaaiotti~~uiiiroma2.i(

Summary
Platinum electrodes were modified by electropolymerized polymer nanotubule nets and were
applied to nitrite detection in drinking water. Several analytical parameters were investigated
such as: different monomers (1,2- 1,3- I,4-DAB, pyrrole, o-anisidine, ]&DAN),
permeability toward nitrites and other interferences, permselectivty toward nitrites. The best
performances were obtained with a Poly(l,2-DAB) nanotubule net and the assembled sensor
was characterized morphologically by scanning electron microscope and electrochemically
by cyclic voltammetry and amperometry coupled to flow injection analysis (FIA), in terms
of linear range of concentration (10-1000 pM), limit of detection (2 pM), reproducibility
(RSD%: 1.5) and linear regression (y/pA=l. 14 x/pM+2.6)

1 Introduction
Nitrite is a very interesting analytical substrate because of its prevalence in
industrially produced meats as a preserving agent and also appearance builder.
But, when ingested it can react with hemoglobin, oxidizing it to metahemoglobin
and with amines, converting them into nitrosamines, which are well-known
carcinogens. The environmental impact from the build-up of high nitrite and
nitrate concentrations, owing to their use as fertilizers, and the problems caused
by the contamination of water sources used for human consumption, are also
important concerns. Nitrites are electroactive species readily oxidized at
platinum electrodes polarized at + 900 mV vs Ag/AgCl. Since the potential for
nitrite oxidation is high, many other electroactive compounds present in complex
media can interfere in the nitrite analysis. To avoid interference effects, platinum
electrodes are often covered by membranes or electropolymerized films.In this
work platinum electrodes were covered by nanostructured polymeric films:
poly( 1,2- 1,3- 1,Cdiaminobenzene) [poly( 1,2- 1,3- 1,4-DAB)], poly( 1,8-
diaminonaphtalene) [poly(1,8-DAN)], poly(o-anisidine) and poly(pyrro1e) were
investigated. A study of permeability of electropolymerized films toward nitrites
and some common interferences, was performed. For this purpose three target
molecules which can be oxidized at the same working potential, were selected:
ascorbic acid (AA), phenol (PhOH) and sulfite

91
92

2 Experimental

2.1 Electrode polishing

The platinum electrode surfaces were polished with alumina powder (A1203,
Buehler, Evanston, IL) having different particle size: 1, 0.3 and 0.05 pm
before use. After rinsing with distilled water, the electrodes were pre-treated
by potential cycling in 0.5 M HzSO4 from -0.2 to +1.2 vs Ag/AgCl at a scan
rate of 20 mV/s, until no changes were observed in the cyclic
voltammograms [I].

2.2 Synthesis of polymeric nanotubule nets

Polymeric nanotubule were produced by the template synthesis [2]
approach. A metallic Pt foil (1 cm) has been used as working electrode. The
polycarbonate nanoporous particle track-etched membranes were used as a
template membrane for the synthesis of polymer nanostructured nets and they
were assembled as working electrode in a conventional one-compartment cell
on a platinum foil. The electropolymerization was performed, at room
temperature using platinum as counter electrode, and an Ag/AgCl as
reference electrode. All the monomers were used at a concentration of 5 mM
and dissolved in 0.01 M HC104 + 0.1 M NaC104, as supporting electrolyte.
Solutions were deoxygenated with Nz for 15 min just before the
electropolymerization experiment. Different potentials were applied for 250 s
during the chronocoulometric experiment, depending on the monomer: 0.6 V
for the three diaminobenzene isomers, 0.75 V for 1,8-DAN, 0.8 V for pyrrole
and o-anisidine. Then, the template was dissolved by dichloromethane and
the polymer nanotubules were placed directly on the platinum electrode
surface, by drop coating. For the preliminary investigations 50 nm pore of
PC template membrane, were used for all the monomers. Then the pore
diameter was varied in order to evaluate the size exclusion effect of the
nanostructured polymeric film on nitrite ion permeability.

2.3 Study of permeability and permselectivity

The permeability toward nitrites and several common interferences such as
ascorbic acid, sulfites and phenols, was evaluated by cyclic voltammetry [3].
The potential was cycled between 0.2 V and 1.2 V at scan rate of 20 mV/s.
All these analytes were used at a concentration of 40 mM, in 0.1 M acetate
buffer pH 4 and the oxidation current peaks of nitrite and interferences were
compared.
 93

2.4 FIA procedure

For amperometric measurements in a flow injection analysis (FIA) the
optimised parameters were found to be: flow-rate 0.6 mumin, applied
potential + 0.9 V vs Ag/AgCl, and 0.1 M acetate buffer, pH 4 as carrier
solution. Standard nitrite solutions were prepared in 0.1 M acetate buffer pH
4 and then injected into a 250 pl loop, using a six-way injection valve.

3 Results and discussion

3.1 Study of permeability

The goal of this work is to investigate various polymeric barriers that could
be used to assemble a sensor that will be highly permeable to nitrite but
exclude as much as possible potential interference. For this purpose, in this
work study about permeability toward nitrites and some common
interferences, was performed. Film permeability was evaluated measuring the
peak current by cyclic voltammetry for each analyte. Then the Permeability
(P%) was calculated using the equation 1, as described in [I]:

where I bare is the peak current recorded at bare electrode and I film is the peak
current at the modified platinum electrode. The permeability results obtained
for platinum electrodes modified by nanostructured polymeric net are

Table 1. Permeabilw ofpolymeric nanotubule nets

Almost all of the films exhibit high permeability toward nitrite, but also
toward interferents. A better behaviour was obtained using poly( 1,2-DAB)
nanotubule net, considering that it represented a compromise between good
permeability for nitrite with a significant repulsion of interferents.

3.2 Study of permselectivity

Since a film that comparably reduces the response to analyte and potential
interferents offers no advantages over a bare platinum electrode, the
94

d e t e ~ n a t i o nof the permselectivity represents an important factor for

evaluation of the sensor performance. It is defined as the ratio between nitrite
permeability and interferent permeability, as reported in the equation 2:
P = P% nitntes 1 P% interferences (2)
Fig. 1 shows the results obtained synthesizing polymeric nanostructures
using polycarbonate membrane having 50 nm-pore size. The best
performances in terms of permselectivity were obtained using poly( 1,2-
DAB) nanotubule net. The higher permselectivity values are probably related
to denser distribution of these structures that is directly related to an easier
process of electropolymerization. Infact small pore sizes of template
(nanometer cut off) play an important role during the monomer diffusion
process making easily 1,2-DAl3 electropolymerization, because only one
amino group is involved. This effect is strictly related to poly(1,2-DAl3)
molecular structure if compared to the other polymers having complex
molecular architectures both in terms of the different substituent positions on
the aromatic ring (poly(l,3-DAB) and poly(1,CDAB)) and a major number
of the aromatic rings in the structure (poly(1,%DAN)). For this purpose, to
confirm this hypothesis related to the permselectivity performances of the
poly( 1,2-DAB) nanotubules and their nanostructured morphology, an
additional study was performed using different pore diameters of
polycarbonate membrane during the poly( 1,2-DAB) nanotubules growth.
The best analytical responses were obtained with nanotubules having 30 m
of diameter (Table 2) considering a compromise between a good
permeability and a good permselectivity, respectively. This nanometer cut off
minimizes the major permeability toward interferences, typical of the largest
pore diameter of the template membrane, giving an easier access for the
polymer diffusion into the template pores compared to the smaller pore
diameter template.
 95

Pore size diameter (nm) P nit/ P AA P nit / P sulfite P nit/ P PhOH

15 0.88 3.57 1.12
30 10.4 0.95 0.68
50 7.61 3.30 1.46
100 1.55 2.24 1.28

3.3 Morphological characterization

Poly( 1,2-DAB) nanotubule nets synthesized by the template method
dissolving the membrane in dichloromethane, showed a typical filled
cylindrical morphology which was investigated using a field emission type
scanning electron microscope (FE-SEM). [4]. This coating provides a large
surface area per unit volume, so porous layers were obtained and this is in
agreement with high permeability values calculated for these nets.

3.4 FIA procedure

Having identified the best nanostructured polymeric film synthesized on the
probe surface, it was analytically characterized in a FIA analysis system for the
amperometric detection of nitrite. The best nanostructured sensor, assembled
using poly(l,2-DAB) nets , showed good linearity over the range 10-1000 pM, a
linear regression equation of JIpA = 0.20 x/pM + 2.8 and a detection limit of 2
pM &OD= 3sh) The reproducibility in terms of relative standard deviation is
0.09 % on slope values and 1.5 % on bias values for n=3.

4 Conclusion

In this work platinum electrodes were modified by polymeric barriers

synthesized by template synthesis technique to have a selective sensor for
nitrite detection. The best resulting sensor was obtained using poly( 1,2-DAB)
nanotubule nets and it has the sensitivity required for implementation of Italian
legal limits for nitrite in drinking water. Further investigations will be carried out
using other synthesis approach.

References
[l] M. Badea, A. Amine, G. Palleschi, D. Moscone, G. Volpe, A. Curulli,
New electrochemical sensors for detection of nitrites and nitrates, Journal of
Electroanalytical Chemistry 509 (2001) 66-72
[2] A.Curulli, F. Valentini, S. Orlanducci, M.L. Terranova, C. Paoletti, G.
Palleschi, Electrosynthesis of non conventional-polymer nanotubules: a new
nanostructured material for analytical applications, Sensors and Actuators B 100
(2004) 65-71
96

[3] L.J. Murphy,

Reduction of interference response at a hydrogen peroxide detecting electrode
using electropolymerised films of substituted naphtalenes, Anal. Chem. 70
(1998) 2928-2935
[4] F.Valentini, A. Salis, A. Curulli, G. Palleschl,
Chemical reversibility and stable low-potential NADH detection with
nonconventional conducting polymer nanotubule modified glassy carbon
electrodes, Anal. Chem. 76 (2004) 3244-3248

Acknowledgments

The authors wish to thank the MIUR target project FIRE3 2001
n.RBNEOIMBTC-002 for financial support.
Modified p-type NiO thin films using Pt- and Au-overlayers for
Ht sensing

S. Cauone, P. Siciliano, A.M. Taurino

Institute of Microelectronics and Microsystems I. M.M.-C.N. R., Sezione di Lecce,
Via Monteroni, 73100 Lecce, Italy
[email protected]

I. Hotovy, V. Rehacek
Department of Microelectronics, Slovak University of Technology, Ilkovicova 3,
812 19 Bratislava, Slovakia

Summary
Nanocrystalline NiO thin films were deposited by dc reactive magnetron sputtering in a mixture of
oxygen and argon and subsequently coated by Au on a NiO film surface. Very thin Pt and Au
overlayers with a thickness of about 1, 5 and 7 nm have been prepared by magnetron sputtering.
Then, the surface modified NiO films have been analysed by TEM, EDX and SEM. NiO thin films
showed a polycrystalline structure with the size of nanocrystals ranging kom a few nanometers to 10
nm. Electrical responses of NiO-based structure towards hydrogen have been measured.

1. Introduction
Nickel oxide (NiO), is the metal oxide usually taken as a model for p-type
materials. Its is an attractive material well known for its chemical stability as
well as for its excellent optical and electrical properties. Indeed, NiO thin films
have been studied for applications in electrochromic devices and also as
functional layers for solar cells [1,2]. In particular, the field of gas sensing has
benefited from the production of prospective materials characterized by a high
surface-to-volume ratio. The gas-sensing properties of metal oxides are more or
less related to the material surface, to an high porosity and to a nanostructure
with small particles. Also, these properties can be essentially improved by doping
of the surfaces of the sensing layers by catalytic metals as platinum and gold (Pt
and Au) [3,4]. Sputtering is a deposition technology widely used both for the
preparation of metal oxide thin films and promoter overlayers. In our previous
work, we analysed the influence of the addition of a very thin Pt overlayer on the
gas-sensitive properties of sputtered NiO thin films towards Hz [ 5 ] . In this work,
we continued this research by comparing the effect of two different metallic
overlayers, i.e. platinum and gold (Pt and Au) on the Hz-sensing properties of
sputtered NiO-based gas sensors.

97
98

eri
We prepared NiO thin films by dc reactive sputtering from a metallic Ni target
in a mixture of oxygen and argon. A sputtering power of 600 W was used. Both
the inert argon flow and reactive oxygen flow were controlled by mass flow
controllers. The relative partial pressure of oxygen in the reactive mixture 02-Ar
was 20 %. The total gas pressure was kept at 0.5 Pa. We have successfully
prepared nanocrystalline NiO thin films with the mean crystal size of -10 nm.
Next, we modified them with overlayers of metallic promoters. Both Pt and Au
thin layers have been sputtered on the surface of the NiO films by varying the
sputtering parameters. In particular, the thickness of the catalytic layers were
varied by controlling the sputtering time. As measured by AFM, Pt films (1, 5
and 7 nm thick) and Au films (1, 5 and 7 nm thick) resulted. Moreover, a suitable
photolitho~aphicmask was used for the deposition of the Pt and Au catalytic
layers. The geometry of the photolitho~aphicmask consists in a squared grid by
which Pt (or Au) squares, 100 pm x 100 pm sized and -40pm spaced, have been
realized, Gas sensors based on such Pt- and Au-modified NiO films were realized
using rough alumina substrates (3 x 3 mrn2 squared) equipped with interdigital
capacitor electrode-structure (IDC) and Pt-microheater. In fig. 1 a top view of the
final structure of the sensor device is shown. The structural features of the films
were investigated by means of a Tecnai 20 S-TWIN transmission electron
microscope (TEM) operated at 200 kV. It is equipped with energy dispersive
Xray (EDX) facility for high resolution chemical analysis. Selected electron
dif~actionpatterns have been recorded together with bright and dark field images
of the film structure.

(4 (b)
Fig.1 a) SEM image of the sensor surface and b} image of the device bounded on a standard
support.

3 Results and discussion

The structural properties of the NiO thin films have been examined using TEM
(fig.2a). Identification of the deposited films was based on the observed electron
diffraction patterns. The measured lattice spacings are shown in fig.2b.
Comparisons can he made with the tabulated d-spacing for the cubic NiO phase
(PDF Number 4-835) and this provides further evidence for the formation of this
 99

oxide. The diffraction pattern was of a continuous ring type indicating a

polycrystalline film. The grains were not oriented homogeneously, but into
certain prominent directions. TEM observations (fig.2) of unmodified NiO films
confirmed that the films were formed by nanocrystals and showed a fine-grained
structure. The size of the nanocrystals ranges from a few nanometres to 10
nanometers depending on the position in the film. We can see that the samples
contain small grains that are partially bonded into clusters.

(a) (6)
Fig.2 a) TEM image of the NiO film after deposition, b) results of the diflraction patterns
obtained by TEM.

TEM images of the NiO film with Au film is not continuous and close. It does
not cover the NiO nanocrystalline surface completely and the Au sputtered atoms
create areas of islands and conglomerates. Au nanoparticles tend to form circle
shaped closed clusters in diameter of several tens nanometers. The presence of
Au was confirmed by EDX analysis recorded during TEM observation. The Au
clusters were homogeneously dispersed on the NiO surface 161. Also Pt film
resulted discontinuos and porous, Pt sputtered atoms form areas of islands and
conglomerates [ 5 ] .
The gas-sensing properties of the Pt- and Au-modified NiO films towards
hydrogen have been compared. The sensor performances resulted a function both
of the catalytic material and the thickness and morphology of promoter layer. As
example, in fig.3a the response curve of three NiO based sensors (i.e. unmodified
and surface modified with Pt and Au 1 nm thick layer) towards 1000 ppm H2 in
dry air vs. working temperature is reported. Both Pt- and Au-modified NiO
sensors showed higher response as compared to unmodified NiO sensor thus
confirming the promoting catalytic effect of the metallic layer. At lower
temperature the Pt-modified NiO sensor seems to prevail showing a first
maximum at -200 O C , while at higher T the Au-modified sensor showed higher
response with an optimum behaviour at -350 O C . In fig.3b the dynamic response
of the above mentioned three sensors at 350 O C is showed.
100

0 1 2 3 4 5 6 7 8 9 1011121314

Time (h)

(a) (b)
Fig.3 a) Response to 1000 ppm HZ for three NiO based sensors (i.e. unmodified and surface
modified with Pt and Au 1 nm thick layer); b) corresponding dynamic curve at different Hz
concentrations at T=350 "C.

At increasing the thickness of the Pt and Au overlayers, the response to Hz

resulted still enhanced by the catalytic activity of the Pt and Au films, but some
new characteristics appeared due to the higher amount of sputtered metals but
also to different morphologies of the two catalytic layers subjected to changes
when the sensor working temperature increases (up to 400 "C) (fig.4). This effect
was confirmed by SEM analysis (fig.5).

Fig.4 Response to 1000 ppm Hz for three NiO based sensors (ie. unmodified and surface modified
with Pt and Au 7 nm thick layer).

Conclusions
NiO thin film modified with Au and Pt overlayer (1, 5, 7 nm thick) were
deposited by dc reactive magnetron sputtering. Both Pt- and Au-modified NiO
sensors (Pt ad Au film 1 nm thick) showed higher response as compared to
unmodified NiO sensor thus confirming the promoting catalytic effect of the thin
metallic overlayer. Both Pt- and Au-modified NiO sensors (Pt ad Au film 1 nm
thick) showed higher response as compared to unmodified NiO sensor thus
confirming the promoting catalytic effect of the thin metallic overlayer. They
 101

showed high responses and sensitivities to medium concentration (500-5000

ppm) of H2 in air; by taking into account that the lower explosion limit (LEL)
range of H2 is 40,000 ppm, these results make the Au- and Pt-modified NiO
sensors promising candidates for implementing hydrogen leak detector.
The H2 sensing properties have been discussed also with reference to the
thickness and the morphology of the Au and Pt thin layers. The sensor response
to H2 resulted to be strongly dependent not only on the grain size of Pt or Au
cluster islands but also on the porosity of the catalytic film on the NiO surface.
At lower temperature the Pt-modified ( I nm) NiO sensor seems to prevail
showing a first maximum at -200 "C, while at higher T the Au-modified sensor
showed higher response with an optimum behaviour at -350 "C.

u overlayer Au overlayer
(1 nm thick) (7 nm thick)

Fig.4 SEM image of Au-modified (1 and 7 nm thick) NiO surface for a sample (a) as-deposited,
(b) a sample after sensor measurement at 400C (b). The images are compared with the surface of the
NiO film on the left.

efere~ces
[l] Magana C, Acosta D, Martinez A, Ortega J 2006 Solar Energy 80 161
121 Wang Y, Zhang Y, Liu H, Yu S , Qin Q 2003 Electrochimica Acta 48 4253
[3] Schweizer-Berberich M, Zheng J, Weimar U, Gopel W, Barsan N, Pentia E,
Tomescu A 1996, Sensors and Actuators B 3 1 7 1
[4] Steffes H, Imawan C, Solzbacher F, Obermeier E 2001 Sensors and Actuators
B 78 106
[5] I. Hotovy, J. Huran, P. Siciliano, S . Capone, L. Spiess, V. Rehacek, Sensors
and Actuators B 103 (2004) 300-31 1.
[6] I. Hotovy, D. Donoval, J. Huran, S. Hascik, L. Spiess, M. Gubisch, S.
Capone, Czechoslovak Journal of Physics, Vol. 56 (2006), Suppl. B, 1192-1198.
 Multilayered metal oxide thin film gas sensors obtained by
conventional and RF plasma-assisted laser ablation

V. Marotta, S. Orlando
CNR - Istituto di Metodologie Inorganiche e dei Plasmi - Sezione di Potenza,
Zona Industriale, 1-85050 Tito Scalo (PZ),Italy
[email protected]

B . Mitu
National Institute for Lasers, Plasma and Radiation Physics,
P.O. Box MG 36 Magurele, Bucharest, 77125 Romania

Summary
Reactive pulsed laser ablation has been established as a feasible method for thin film deposition
of a large variety of compounds with simple or complex stoichiometry such as oxides, nitrides,
semiconductors and superconductors. This technique relies on photoablation of pure elements, or a
mixture of materials, with simultaneous exposure to a reactive atmosphere. The reactivity of the
plasma plume can be enhanced if the ambient gas is excited, for example by an additional RF
discharge. In the case of oxides, reactions between the laser vaporized metals and oxygen lead to the
formation of intermediate complexes and finally to oxide thin films.
Sn02 -based multilayered thin films of In203, have been deposited by conventional and RF
plasma-assisted Reactive Pulsed Laser Ablation, with the aim to evaluate their behaviour as toxic gas
sensors. The depositions have been canied out by a frequency doubled Nd-YAG laser (h=532 nm,
c = 7 ns) on Si(100) substrates, in 0 2 atmosphere. The thin films have been characterized by X-Ray
Diffraction and Electrical Resistance measurements. A comparison of the electrical response of the
simple (indium oxide, tin oxide) and multilayered oxides to toxic gas (nitric oxide) has been
performed. The influence on the structural and electrical properties of the deposition parameters,
such as substrate temperature and RF power, is reported.

1 Introduction
Studies on semiconductor oxides are growing very fast due to the successful
involvement of these materials as detecting devices. Metal oxide coatings are
widely used as transparent resistive heaters, electrical electrodes for flat panel
display and electrochromic devices, antireflecting and high refraction index
materials for optical applications, and gas sensors. These thin films are usually
produced by Chemical Vapor Deposition [ 11, Microwave discharge, Molecular
Beam Epitaxy [2], sputtering [3] and sol-gel process [4]. Another attractive
choice for the preparation of these films is the reactive pulsed laser ablation and
deposition (RPLAD) [ 5 ] . In our laboratory, pulsed laser deposition (PLD) has
been already applied successfully in the past to deposit many materials such as
semi- and super-conductors, nitrides and oxides [6-81. This technique relies on
photoablation of pure elements, or a mixture of materials, with simultaneous

102
 103

exposure to a reactive atmosphere. Neutral and ionized species are formed in the
laser ablated plume. In the case of oxides, reactions between the laser vaporized
metals and oxygen lead to the formation of intermediate complexes [9,10] and
finally, when collected on suitable substrates, to oxide thin films. Gas sensors
based on semiconducting oxide thin films present some advantages with respect
of other types of gas sensors. To detect small amounts of a reactive gas in air, the
surface reactions are much more relevant than the bulk changes, so that the
specific surface of sensitive elements must be as large as possible. The RPLAD
technique is adequate for satisfying such a primary requirement. In addition to
high sensitivity, thin film sensors have high stability combined with simple
construction at low cost. In this paper, the RPLAD deposition of indium oxide
and tin oxide thin films, and multilayers of these two compounds, has been
performed in order to evaluate the behaviour of such films as toxic gas (NO)
sensors. The deposited films have been characterized by Electric Resistance
measurements to understand the influence of physical parameters such as the
substrate temperature and the RF power applied to generate reactive oxygen
plasma.

2. Experimental
The experimental set-up has been described in detail elsewhere [7]. In the
experiments here reported, the multi-port stainless steel chamber has been
pumped down to 10" Pa, whereas the depositions were performed in static O2
atmosphere (pressure 100 Pa). Commercial metallic targets of indium and tin
have been mounted to the rotating holder (2 rpm). The Quantel Nd:YAG 581
laser (h=532 nm, z = 7 ns, repetition rate = 10 Hz, fluence = 8 J/cm2) was
impinging on the target at 45" with respect to the normal at constant fluence of 8
J/cm2. The gaseous species were collected on heated Si (100) substrates
(temperature up to 1000 K) positioned 5 cm far away from the target in the on-
axis configuration. The PLD set up has been improved by employing a RF
generator (13.56 MHz EN1 Model OEM-6A). The substrate holder is surrounded
by an isolated stainless steel ring concentric to the substrate holder and connected
to the RF generator through a customized matching unit [6,8]. Thin films of
indium oxide and tin oxide, and two-layered films (Sn - In) have been deposited
on Silicon (100) substrates by reaction of laser ablated metallic targets of indium
and tin in presence of oxygen, at different substrate temperatures [7]. X ray
spectra were detected by a Rigaku Miniflex diffractometer using the K a line of a
Cu target (h= 0.154056 nm) as an X ray source; step scan 0.02"/sec. A four-
contact probe, based on a Keithley 224 Programmable Current Source and a
Keithley 182 Sensitive Digital Voltmeter, has been utilized to evaluate the
resistance variation of thin films when exposed to 1000 Pa of 5% NO in He gas
104

mixture after evacuation. The resistance measurements have been performed at

room temperature and in the dark to avoid any influence due to the light [ 111.

3. Results and discussion

The improvement of the film surface smoothness has been achieved either by
increasing the deposition temperature or by assisting the conventional PLD with
the RF plasma. These evidences had been reported in a previous paper [ 121.

Figure 1. Comparison of the XRD patterns of Sn02-1n203 bilayers deposited:

(left)by RF plasma assisted PLD at 700K and 9OOK subsirate iemperature respectively;
(right) at 900K subsirate temperature by conveniional PLD and RFplasma-assisted PLD.

X-Ray Diffraction (XRD) analysis showed polycrystalline structures either in

indium oxide and tin oxide single layer depositions and in bi-layered films
[9,10]. XRD data of samples deposited by conventional PLD, at low deposition
temperature (below 600 K), show predominant diffraction peaks corresponding
to metallic phases for both indium and tin thin films, and also for multilayers
(In203-Sn02 and Sn02-In203). Increasing the deposition temperature, the
oxidation becomes more evident but some "metallic" peaks still testify a further
uncomplete oxidation. Only at 900 K deposition temperature almost all the peaks
connected to metal phases disappear. Analogous behaviour has been found in the
samples produced with the RF oxygen plasma. Comparing the two sets of
samples the peaks enhancement due to the RF plasma is evident, even if the
major effect is due to the deposition temperature. We summarize the results in
Figure 1: on the left are reported the patterns of multilayered (Sn02-In203)
samples produced at two different substrate temperatures (700 and 900 K) by
employing the RF oxygen plasma. The diffraction peaks corresponding to h203
and SnOz are much more evident when the deposition temeperature is higher.
Also some metallic peaks, in particular corresponding to tetragonal tin - which is
the first layer onto the substrate - become negligeable. Even if the In (101)
 105

metallic peak does not seems to be very much affected by the temperature, the
great enhancement of almost all the oxide diffraction peaks reveal a huge
improvement of the oxidation also for the indium oxide top layer.
The effect of the presence of the RF plasma during the deposition is shown in
Figure 1 (right): the XRD spectra of multilayered (Sn02-In203) samples
deposited at 900 K substrate temperature are compared in order to evidence the
improvement of the oxides' formation induced by the RF plasma. The
crystallinity enhancement is noteworthy especially for the indium oxide top layer
considering that often is quite difficult to grow the second layer in a well oriented
crystalline phase. In particular, the enhancement of the In203 (222) peak is
considerable, even if it is overlapping the Sn (200) peak. The relative intensity of
the last one should be a little bit higher than the Sn (101) diffraction peak, close
to it. Unfortunately, the In (101) diffraction peak of the conventional PLD sample
cannot be evaluated because of the overlap to another diffraction peak related to
the superficial silicon oxides present on the substrate.

la . . . . . . . . . . . . . . . . . . . . . . , . . . . I .
12 ~..,....,....,...--r....I....,..
comparison In vs In+Sn ccinparlsn Sn v8 Sndn
1

- 0
0 10 20 30 40 50

Figure 2. Comparison of the normalized resistances. asfunction of time after exposure to NO toxic
gas, of: (left) indium oxide single layer and bilayered In203-Sn02;
{right) tin oxide single layer and bilayered Sn02-ln203.

When the toxic gas is adsorbed by the film, the surface resistance changes
depending upon the variation of the free carriers involved in the adsorption. In
Figure 2, the normalized resistance variations of the films produced by RF
plasma assisted PLD at high deposition temperature (900 K) have been shown
when exposed to NO toxic gas. On the left (Figure 2) are reported the behaviours
of the single layer indium oxide film and of the indium oxide - tin oxide bilayer.
Analogous behaviours are shown, on the right, for the tin oxide single layer film
and for the tin oxide - indium oxide bilayer.
106

4. Conclusions

In this paper we report preliminary results on the resistance variation, when

exposed to NO toxic gas, of thin films (indium oxide, tin oxide, and bilayers
Sn02-In203 and In203-Sn02).The sensors were grown on Si (100) substrates
by reactive pulsed laser ablation and deposition of metal targets (indium and tin)
in presence of oxygen also activated by a RF generator. The XRD
characterization has shown the structural correlation with the deposition
temperature and with the presence of oxygen plasma, generated by a RF source.
Both the single layers of indium oxide and tin oxide work well as toxic gas
detectors. An improvement due to the multilayered deposition has been
ascertained because the bilayered films have shown a faster time response
velocity and a larger resistance variation with respect to single oxide thin films.
Work is in progress to better understand the influence of other deposition
parameter, such as laser fluence and wavelength.

Acknowledgments

One of the authors (B. M.) gratefully acknowledges NATO-CNR Advanced

Fellowships Programme 2002 (Grant n. 215,358) for financial support. This
work was partially supported by the Italian research ministry MIUR under the
project MIAO.

References

[I] E. N. Eremin, L. I. Nekrasov, E. A. Rubtsova, M. V. Belova, V. L. Ivanter, L. N. Zacharov, L. N.

Petukhov, Russ. J. Phys. Chem., 56 (1982) 788.
[2] S. Nakamura, Y. Harada, M. Seno, Appl. Phys. Lett., 58 (1991) 2021.
[3] G. Sberveglieri, P. Benussi, G.Coccoli, S. Groppelli, P. Nelli, Thin Solid Fzlms, 186 (1990) 349.
[4] R. Rella, P. Siciliano, L. Vasanelli, C. Gerardi, A. Licciulli, J . Appl. Phys., 83 (1998) 2369.
[5] D. B. Chrisey, G. K. Hubler (Eds.), Pulsed Laser Deposition of Thin Films,John Wiley & Sons,
Inc., New York, 1994.
[6] A. Giardini, V. Marotta, S. Orlando, G.P. Pansi, Surf: and Coat. Technol. 151-152 (2002) 316.
[7] V. Marotta, S. Orlando, G.P. Parisi, and A. Giardini, Appl. Surf: sci., 154-155 (2000) 640.
[8] V. Marotta, S. Orlando, G.P. Parisi, A. Giardini, Appl. Phys. A, 69 [Suppl.](1999) S675.
[9] R. Teghil, V. Marotta, A. Giardini Guidoni, T. M. Di Palma, C. Flamini, Appl. Surf: Sci., 138-
139 (1999) 522.
[lo] A. Mele, D. Consalvo, D. Stranges, A. Giardini Guidoni, R. Teghil, Int. J. Mass Spectrom. Ion
Process, 95 (1990) 359.
[l 11 A. Giardini Guidoni, V. Marotta, S. Orlando, G . P. Parisi, Int. J. of Photoenergy 3 (2001) 213.
[ 121 V. Marotta, S. Orlando, G.P. Parisi, A. Santagata, Appl. Surf: Sci., 208-209 (2003) 575.
 A new competitive technique for the production of porphyrin
thin films: the Glow Discharge induced Sublimation (GDS)

M. Tonezzer, A. Quaranta, G. Della Mea

Dipartimento di Ingegneria dei Materiali e Tecnologie Industriali, Universitir di
Trento, via Mesiano 77, 38050 Povo(TN), Itab
[email protected]. it

G. Maggioni, S. Carturan
Universitci di Padova, c/o INFN-LNL Viale dell Universitir,2- 35020 Legnaro,
Italy

Summary
Thin porphyrin assemblies, widely used as sensing materials in different kinds of transducers,
are usually produced through chemical deposition techniques. In this work a new physical
technique named Glow-Discharge-induced Sublimation (GDS) has been employed for the first
time for the production of cobalt 5,10,15,20 meso-tetraphenyl porphyrin (CoTPP) thin films.
For the sake of comparison, CoTPP coatings have been also produced by vacuum evaporation
and spin coating procedures. The inspection of the surface morphology with a Scanning
Electron Microscope evidenced a higher surface roughness of GDS samples with respect to
films deposited by means of conventionally used procedures. Optical measurements performed
in atmospheres containing ethyl alcohol vapours highlight that GDS films yield more than ten
times intense responses in comparison with the films deposited through the commonly used
techniques, very fast response times (t50 = 12 sec and t90 = 72 sec) and fast and complete
recovery.

1 Introduction
Porphyrins are widespread compounds in nature, where they play essential
functions for life [l]. A great number of features makes porphyrins eligible as
good sensing material able to detect the volatile organic compounds (VOCs)
present in the enviroment [2]. Porphyrins are stable compounds and their
properties can be finely tuned by simple modifications of their basic molecular
structure increasing thus the versatility of these molecules. In order to be
exploited as sensing materials, porphyrins compounds are usually deposited as
thin solid films through chemical techniques (solvent casting, Langmuir-
Blodgett, spin coating) [3].Fewer efforts have been carried out to produce thin
solid films by means of physical technique in spite of the fact that these
techniques assure several advantages as great reproducibility, more uniformity
and stricter control of the film thickness in comparison with standard chemical
techniques. Moreover physical techniques, producing thin solid films without
using any extraneous compound, allow to produce samples characterized by very

107
108

high purity. The material purity is expected to play a very important role in gas
sensing field because the traces of retained solvent can partially hinder the
interaction between the film and the analyte molecules. Recent optical
measurements demonstrated that vacuum evaporated thin porphyrin films exhibit
higher responses towards analyte molecules than the chemically deposited ones
[4]. This paper reports the employment, for the first time, of a new physical
technique developed at LNL and named Glow-Discharge-induced Sublimation
(GDS) for the deposition of thin cobalt 5,10,15,20 meso-tetraphenyl porphyrin
(CoTPP). GDS technique is based on the use of a weakly ionized glow discharge
produced in a standard radio frequency magnetron sputtering equipment: low
energy (E < 1keV) He ions impinge on solid organic powder leading to the
sublimation of the organic molecules and to their condensation onto the
substrate. GDS has recently showed to place as an attractive deposition
technique for the production of thin organic films for gas sensing applications
allowing to produce samples characterized both by high purity and especially by
extraordinarily large surface area to bulk ratios [5,6]. It is worth to note that this
last feature represents a very important characteristic in gas sensing field because
a large surface area improves the interaction between film and analyte molecules
and therefore the sensing capability of the films. Thin CoTPP films were also
deposited through spin coating (SPIN) and vacuum evaporation (VE)
techniques, for the sake of comparison. In order to study and compare the
behaviours of the different samples, optical measurements in presence of various
atmospheres of ethyl alcohol (EtOH) were performed.

Experimental
CoTPP compound is commercially available (Acros Organics) and was used
directly without some further process.
This material was deposited in the form of thin films onto clean quartz
substrates by means of GDS technique. GDS set-up consisted of a vacuum
chamber evacuated by a turbomolecular pump to a base pressure of Pa. The
glow discharge was sustained by a I-in. cylindrical magnetron sputtering source
connected to a radio frequency power generator (600 W, 13.56 MHz) through a
matching box. The CoTPP powder was put on the surface of an aluminum target
and placed on the sputtering source. The glow discharge feed gas was helium
(99.9999 %), whose pressure inside the chamber was measured through a
capacitance gauge. Typical values of rf power, target DC self-bias, and working
pressure were in the ranges 10 + 20 W, -20 + 300 V, and 20.0 f 0.1 Pa,
respectively. CoTPP thin films were also deposited through vacuum evaporation
and spin-coating techniques.
Optical absorbance spectra were acquired utilizing an experimental apparatus
in which the measure chamber, placed inside a Spectrophotometer V-570 Jasco,
 109

is connected to two mass flow controllers suited to assure an accurate real-time

control of the vapour concentrations into the chamber during the spectra
collections. This apparatus allows to record both dynamic behaviours at a fixed
absorbance wavelength and complete absorbance spectra in different vapow
concentration atmospheres.

3. Results and discussion

SEM pictures o f SPIN, VE and GDS samples o f CoTPP compounds,

reported in Figure 1, point out three different morphologies depending on the
three different deposition techniques.

Fig. I : SEM images of SPIN (A), VE (B)) and GDS (C)films of CoTPP.

In particular VE and SPIN samples show surfaces characterized by flat

morphology and very low surface area to bulk ratios; GDS sample, on the
110

contrary, is characterized by extraordinary rough surfaces and by very large

surface areahulk ratios.
In order to study and compare the responses of the three CoTPP samples
(SPIN, VE and GDS) upon exposure to various ethyl alcohol (EtOH)
atmospheres, the absorbance variation AA was utilized:

dA =A, - A ,

where At and Ao are respectively the Soret band (h = 440 nm) absorbance
intensities of the samples under EtOH and nitrogen atmospheres.
A comparison of the evolution of the B band (h = 440 nm) intensity of the
three CoTPP samples upon exposure to 1000-4500 ppm of EtOH as a function
of time is shown in Fig. 2.

: '2300 ppm
I '

i 3000 ppm
' I :
; 4500 p
' n:
---
I . I

0.0151 : A
n
5
n

8 I

s.
I ' I ' I ' I ' I . I ' I ' I

0 10 20 30 40 50 60 70 80
Time (min)

Fig.2. Concentration dependence of SPIN (a), VE (5) and GDS (c)jlms of CoTPP compound upon
exposures to EtOH atmospheres over the range llO0 - 4500ppm.

As can be seen, the kinetics of the gas exposure are characterized by a fast
increase of the band intensity followed by a slow increase of this intensity until
saturation values are reached. As the ETOH vapour stream is switched off, a dry
nitrogen flush is activated. This rapidly initiates recovery and after few seconds
the original Soret band intensities have been restored.
It is worth to point out that the optical behaviours of all the samples are
characterized by a complete and fast recovery: this experimental result becomes
particularly interesting taking into account that all the measurements have been
 111

performed at room temperature.

Fig. 2 points out the more than ten times sensing capability of the GDS
sample in comparison with the SPIN and the VE ones in the EtOH detection:
these larger GDS intensities can be attributed both to the high purity and
especially to the extraordinarily high surface area to volume/ratio of GDS
assemblies.
In order to characterize the speed of the response the parameters t50 and t90
(defined, respectively, as the time taken for the signal intensity to reach the 50%
and the 90% its final saturated value) are usually used. These parameters,
respectively tso in the order of ten seconds and t90 in the order of one minute,
reflect that the GDS films of CoTPP behave as fast sensors. In particular t50 has
been found to be at least half of those measured for similar chemically deposited
porphyrin assemblies (for SPIN sample of CoTPP compound t50 = 23 sec) and
considerably smaller than those values typically reported for different molecular
systems.
From these preliminary tests it is possible to hold that GDS places itself as a
very promising technique for the deposition of thin CoTPP solid film for gas
sensing applications allowing to produce samples characterized by very fast
responses, complete recovery and extraordinarily larger intensity and sensitivity
in comparison with the samples deposited by means of the standard deposition
techniques.

Acknowledgements
This research was financially supported by the Fifth Commission of Istituto
Nazionale di Fisica Nucleare (DEGIMON project).

References
[I]D. Dolphin (Ed.), The Porphyrins, VI(a), VII(b), Academic Press, New York, 1978.
[2] N.A. Rakow et al. - Nature, 406 (2000) 710.
[3] A. DAmico et al. Sens. Actuators B 65 (2000) 209.
[4] M. Tonezzer et a1 - presented at Eurosensor XIX, 11th - 14th September 2005,
submitted Sens Actuators B.
[5] G. Maggioni et al. Chem. Mater., 17 (2005) 1895.
[6] G. Maggioni et al. Surf. Coat. Techno]., 200 (2005) 476.
 Optical Sensing to Organic Vapors of Fluorinated Polyimide
Nanocomposites containing Silver Nanoclusters

S. Carturan, M. Bonafini, G. Maggioni,

I.N.F.N.- Laboratori Nazionali di Legnaro, Viale dell 'Universith 2, 35020
Legnaro, Italy, carturan@lnl. infn. it

A. Quaranta, M. Tonezzer, G. Della Mea

University of Trento, Department of Materials Engineering and Industrial
Technologies, via Mesiano 77, 38050 (TN), Italy

G. Mattei, C. de Julian Fernandez, P. Mazzoldi

University of Padova, Department of Physics, Via Marzolo 8, 35123 Padova,
ItaZy

Summary
Polyimide-silver nanocomposite films were prepared by surface modification with KOH aqueous
solution followed by K-assisted Ag doping and thermal reduction in hydrogen-containing
atmosphere. The film porosity was controlled by dispersing into the polymer the compound
azodicarbonamide, which decomposes and desorbs dunng thermal treatment, thus generating a more
porous structure After heat treatment in reducing atmosphere, the films present mono-dispersed
metallic nanoparticles with average size of about 5 nm and a narrow size distribution. The exposure
to water, acetone and ethanol vapors causes absorbance changes of the plasmon absorption peak of
silver, which are ascribed to the changes of the refractive index of the embedding medium. The
reversible absorbance changes and the fast response (40-160 s) over a concentration range from 800
to 88000 ppm make these nanocomposite films useful for optical detection of organic vapors.

1 Introduction
Optical gas sensors have been attracting growing interest owing to several
advantages over the conventional electrical gas sensors [1,2]. In fact, they are
unaffected by electromagnetic noise and fire ignition and can be used for remote
control and information through an optical fibre network. In particular, materials
whose optical absorption properties reversibly change in the presence of
atmospheric gases and moisture are considerably attractive for use as
optochemical sensors.
As known, dielectric materials containing nanoclusters of Cu, Ag and Au
present optical absorption bands, named plasmon absorption bands, which
depend both on the particle size and on the refractive index of the surrounding
medium. Since the adsorption of gas molecules modifies the dielectric properties
of the nanoparticle-surrounding medium, the optical absorption change can be
correlated to the kind and amount of adsorbed gas.

112
 113

This paper deals with a study of the sensing capability towards organic
vapors of nanocomposite materials constituted by Ag nanoparticles embedded in
transparent polyimide thin films obtained by ion-exchange technique followed
by annealing in reducing atmosphere [3,4].
2. Experimental
The chemical structures of the starting polyimides 6FDA-DAD and BPDA-
3F are sketched in Fig. 1. The synthetic procedure for their preparation and their
physical properties have been reported elsewhere [ 5 ] . Thin films about 1 pn
thick were obtained by spinning a solution of each polyimide (10% wt. solids) in
N-methylpyrrolidone on silicon and silica substrates. Azodicarbonamide (ADC)
was mixed to the polyimide-containing solution at two different concentrations:
5 wt. % and 15 wt. %. The samples were heated in air up to 250 "C for 1 hour in
order to effect the thermolysis and removal of ADC compound. Afterwards the
films were immersed in KOH 5M for time intervals ranging from 5 s up to 1 min
at 45C. Both polyimides react with KOH to give the salt potassium polyamate,
as reported in Fig. 1. Subsequently, the samples were immersed in a solution of
AgN03 50mM at 45"C, thus undergoing the ion-exchange process between
potassium and silver ions. Thermal reduction of silver and particles aggregation
were attained by annealing the samples in reducing atmosphere (Hz 12% - Ar) at
350C for 30 min.

Fig.1 Chemical structures of 6FDA-DAD and BPDA-3F polyimides and alkaline hydrolysis
reaction scheme.
Optical UV-Vis absorption spectra were recorded by using a Jasco V530
spectrometer. Nitrogen physisorption experiments were carried out on polyimide
films with and without ADC at 77 K on an ASAP 2020 Micromeritics
sorptometer. The specific surface area was calculated by the BET equation. The
silver cluster dimension and distribution were investigated by cross sectional
transmission electron microscopy (TEM), carried out with a field emission gun
FEI TECNAI F20 (S)TEM microscope operating at 200 kV.
3. Results and Discussion
A strong increase of the films porosity is achieved by including ADC into the
two polyimides, as evidenced by nitrogen physisorption measurements. In fact,
114

the 6FDA-DAD doped with 5 % of ADC exhibits an increase of effective

surface area from 1.7 m2/g, for the undoped polymer, to 2 1.7 m2/g.
As to the optical transparency of polymer films, after the ion-exchange
procedure the samples are still transparent, but after the heat treatment in
reducing atmosphere they exhibit an intense yellow colour. The optical
absorption spectra, shown in Fig. 2, exhibit an absorption peak at 430 nm for
6FDA-DAD and at 440 nm for BPDA-3F. As can be observed, the intensity of
the plasmon peak increases at increasing ion exchange time duration. On the
other hand, 6FDA-DAD thin film treated in KOH for 1 and AgN03 for 2
exhibits a plasmon absorption feature of lower intensity with respect to shorter
treatment times. This deviation from the expected trend can be ascribed to the
thinning of the pristine film during the prolonged KOH treatment.

Fig.2 Optical absorption spectra of the BPDA-3F and 6FDA-DAD thin PIms treated at
different times in KOH and AgNOj aqueous solutions.
The presence of silver clusters is confirmed by TEM measurements, not
reported. Along the doped layer a uniform distribution of spherical silver
clusters has been observed, whose mean size was evaluated to be <D> = 5.1 nm.
By fluxing saturated vapors of water, ethanol and acetone, the plasmon
absorption peak exhibits a change of both intensity and position, induced by the
local change of the refractive index of the medium surrounding the nano-sized
metal clusters. Figure 3 displays the absorbance changes AA defined as Aanalyte-
of the sample BPDA-3F doped with 15 % ADC and treated for 1 and 3
Anitrogen
in KOH and AgN03 aqueous solutions, respectively. As can be observed the
variation is more pronounced in the presence of water saturated vapors and
decreases with acetone and ethanol.
 115

010, . I , 1 . I , ,

I
350 400 450 500 550
Wavelength (nm)
Fig.3 Absorbance variation M in presence of saturated vapors of water, ethanol and
acetonefor the BPDA-3F thinfilm. The treatment parameters are detailed in the graph.
For a better evaluation of the sensing capability of the nanocomposites
towards the different analytes, a kinetic response study has been performed, by
exposing the samples to the vapors at concentrations ranging from 800 to 88000
ppm. The maximum available concentration depends on the vapor pressure of
the different analytes at room temperature. Figure 4 displays, as an example, the
kinetic responses to the three analytes of the 6FDA-DAD thin filmdoped with 5
% ADC and treated for 30 in KOH and for 1 in AgNO3 aqueous solutions.
6FOA-OAO+AOC 5%
0.03
.... .
ETHANOL
9600 PPm 7700 ppm 6400 ppm 3400 ppm

0.03- 88000 ppm ACETONE

8 0.02-
0.01 -
0.00-
WATER

I . , . , .
0 1000 2000 3000 4000 51 30
Time (sec)
Fig.4 Dynamic absorbance variation in presence of vapors at diflerent concentrations of the
6FDA-DAD thin film doped with 5 % of ADC and heated for 30 seconds and I minute in
KOH and AgNOj aqueous solutions. respectively.
The data have been collected by measuring the absorbance changes at 422
nm, with a bandwidth of 5 nm, every 5 seconds. From the kinetic response
curves the response and recovery times have been also calculated. The response
times range from 40 to 160 s and the recovery times from 140 to 460 s. These
116

relatively low values point out the fast permeation of analyte molecules through
the polyimide matrix, which easily reach the cluster proximity.
4. Conclusions
Silver-nanoclusters-containing fluorinated polyimide films have been
produced by using the ion exchange procedure followed by thermal treatment,
The porosity of the polymer matrix has been enhanced by adding to the solved
polyimide small amounts of azodicarbonamide, which thermally decomposes
thus increasing the free surface area of the polymer films up to more than
tenfold. The distribution of silver nanoclusters in the film is homogeneous, as
shown by TEM micrographs, and the average nanocluster size is 5.1 nm.
The optical sensing capability of silver-polyimide nanocomposite materials
have been tested upon exposure to vapors of water, ethanol and acetone. The
analyte/film interaction produces a change of the nanoparticle plasmon
absorption peak owing to a change of the average refractive index of the
environing medium. Measurements of the absorbance changes at a fixed
wavelength show that the nanocomposite films respond to all the analytes at
concentrations ranging from 800 to 88000 ppm. Response times ranging from
0.7 to 2.0 min and recovery times in the order of few minutes have been
obtained for all the nanocomposite films.

Acknowledgements
This research was financially supported by the Fifth Commission of Istituto
Nazionale di Fisica Nucleare (DEGIMON project).

References
[ I ] K. Eguchi, Optical gas sensors, in: G. Sberveglieri (Ed.), Gas sensors, Kluwer Academic
Publisher, Dordrecht, 1992, pp. 307-328.
[2] W.R Seitz, Chemical sensors based on fiber optics, Anal. Chem. 56 (1984) 16A-34A.
[3] K. Akamatsu, K. Nakahashi, S. Ikeda, H. Mawahne, Fabrication and structural characterization
of nanocomposites consisting of Ni nanoparticles dispersed in polyimide films, Eur. Phys. J. D 24
(2003) pp. 377-380.
[4] S. Ikeda, K. Akamatsu, H. Nawafune, T. Nishino, S. Deki, Formation and growth of copper
nanoparticles from ion-doped precursor polyimide layers, J. Phys. Chem. B 108 (2004) pp. 15599-
15607.
[ 5 ] S. Carturan, A Quaranta, A. Vomiero, M. Bonafini, G. Maggioni, and G. Della Mea, Polyimide-
Based Scintillators studied by Ion Beam Induced Luminescence, IEEE Trans. Nucl. Sci. 52(3) (2005)
748-75 1.
 SOLID STATE DIRECTIONAL ANEMOMETER FOR HARSH
ENVIRONMENTS

P.BRUSCHI, N.NIZZA
Dipanimento di Ingegneria dellhfomazione, via G. Caruso 16, 56122 Pisa, Italy.
[email protected]

M. PIOTTO
IEIIT-Pisa CNR, via G. Caruso 16, 56122 Pisa, Italy.

M. SCHIPANI
STMicroelectronics, via Tolomeo I , 20010 Comaredo (MI), Italy.

An anemometer consisting in two silicon chip pairs glued to a thin ceramic plate is
presented. The four chips include a precise temperature sensor and a polysilicon heater and
the two pairs are positioned along orthogonal axes in order to measure both the modulus and
the direction of the wind. The ceramic plate protects the chips from contamination
maintaining, at the same time, the thermal contact between the wind and the sensors.
Preliminary charactensation of the device is presented and discussed.

1. Introduction
The knowledge of wind distribution is a key information in many
application fields including meteorology, aeronautics, energy conversion,
controlling systems and environmental monitoring. The dispersion of a
poisonous agent after a catastrophic leakage from an industrial plant or simply
after a period of intense traffic in a town can be predict more accurately if the
real distribution of the wind flows is known. Simple numerical models, in fact,
begin to fail in urban environments with a complex geometry of building
structures and street configuration [ 1,2] or in a forest with an irregular geometry
[3]. Therefore, experimental measurements of the wind intensity and direction
are necessary both to validate the existing numerical models and to generate new
ones. In particular, the complicated nature of wind distribution inside an urban
area requires a dense network of sensors to provide a knowledge of the wind
distribution. The sensors should be (i) miniaturized to both reduce the visual
impact and increase the network density, (ii) low power demanding to decrease
the routine maintenance operations of the network and (iii) suitable for harsh
environments.
In this work we present a thermal anemometer based on a silicon chip,
previously used in innovative single sensor flow-meters 141,which is equipped
with a precise temperature sensor and a polysilicon resistor, used as a heater. The

117
118

device is made up of (i) two chip pairs to sense the orthogonal components of the
wind and (ii) a thin ceramic plate that protects the chips from the air stream.

esign and s ~ u l a t i o n

In order to understand the operating principle of the sensor the simple

configuration of Figure 1 (a) can be considered. The two elements, represented
as squares in the figure, are composed of a heater and a temperature sensor,
measuring the element temperature.

Figure 1. Two element (a) and four element (b) anemometer configuration.

The temperature of the elements are indicated by TI and T,, while P is the
power delivered to the heaters, identical for the two elements. Let us start by
assuming that mismatch errors are not present. In this condition, if both elements
are fed by the same heating power and the wind speed is zero, than Tl=T2.
Conversely, if a wind speed component along the s, direction is present, than
heat transfers from the upwind to the downwind element through forced
convection. Experiments performed with similar arrangements IS] showed that
the temperature difference created by the air flow is nearly proportional to the
cosine of the angle a between the wind direction and the s, axis. Unfortunately,
the dependence on the modulus of the wind speed is more complicated and non
linear so that an empirical calibration is generally required. In order to detect
both the wind modulus and direction, two one-dimensional structures with
orthogonal axes s, and sy can be combined to form the four element sensor of
figure 1 (b). For such an arrangement the following equations can be reasonably
hypothesized:

T2 -Z',' = A(v)cos(a), T4 - T3 = A(v)cos(a - YO") (1)

where A(v) is a function of the wind speed v.

The proposed device is built according to this four element configuration.
 119

The elements consist in silicon chips including a precise temperature sensor

based on the hVBEprinciple (sensitivity 3.3 mVK) and a polysilicon heater. The
chips have been fabricated using the Bipolar-CMOS-DMOS process BCD3s of
croelectronics. More details about these elements are given in Ref. 141. The
chips are glued to a ceramic-epoxy laminate (RO 4003), 0.25 m thick,
characterized by a 0.6 WK"m-' thermal conductivity, roughly six times the
conductivity of standard laminates for printed circuit boards. The laminate is
provided with a copper layer (35 pm thick) the geometry of which was defined
with standard photolitho~aphicprocess to obtain the interconnects for the chips.
The laminate is mounted on top of a stainless steel cylinder as shown in
figure 2(a). In this way the four chips, forming the structure of figure l(b), are in
thermal contact with the air stream through the thin laminate, which, at the same
time, protects the chips from contamination, dust or moisture. A photograph of
the laminate taken from the chip side is shown in figure 2 (b).

memhrane
30 mm
wmd
'A

Figure 2. Final configuration of the sensor probe (a) and optical photograph of the laminate with the
chip and copper interconnect layer@).

riving and ut circuit

The interface circuit for the sensor described in the previous section has
been designed on the basis of simulations performed using the finite element
environment FEMLABTM.A particularly relevant result is shown in figure 3,
representing the case of wind direction along the s, axis (a=rO). The upwind and
downwind temperatures, Tup and Tdown,respectively, and their difference is
reported. It can be observed that, as the wind speed increases from zero, Tdown
actually increases and Tup decreases, in conformity with the considerations of
previous section. However, this trend stops just for very low wind speeds and
also TdoW begins to decrease (overall cooling). The result is a non monotonic
behaviour of the output signal (AT=Tdown-Tup) that reduces the valid wind speeds
to a range clearly unsuitable for most applications. In order to solve this problem,
a driving strategy that stabilizes the average temperature of the 4 sensor elements
has been adopted. In this way, considering a simpler two elements configuration,
if one sensor temperature decreases, the other one should necessarily increase.
120

This result has been obtained by means of the circuit of figure 4. The four
voltages V1-4, linearly dependent on the corresponding sensor element
temperatures, are amplified by the four blocks indicated with A (based on the
quad op-amp LT1114) and their average is computed by resistors R.

22
20
I-
18
t-?
1201 / ----- -1 14
100
1. , / ,
-02 0 0 02
, , , , .
04 06 08 10 12
, , , . , , ,
14
$
16
wind speed (mk)
Figure 3. Finite element simulations showing the overall cooling phenomenon.

The average temperature signal (V,,) is compared with the desired set-point
(VsET) and the difference, amplified by the instrumentation amplifier ICl
(AD620),is used to drive the heaters of the four chips, connected in parallel. IC2
is a power op-amp (L272M). The diode D1 is necessary to prevent the heater
voltage from getting negative and cause instability.

Figure 4.Circuit used to implement the constant average temperature method.

3. Experimental results
The effectiveness of the proposed driving strategy is shown in figure 5 (a)
where the sensor responses, obtained with the constant average temperature and
the traditional constant power methods, are compared. The experiments were
performed in a purposely built wind tunnel consisting of a pipe (diameter 120
mm) equipped with a fan and a reference anemometer. The output signal Vx is
proportional to the temperature difference T2-T1(see figure 1) while the offset
 121

Vx(0) has been subtracted. We note that in conditions of constant power, the
sensor output actually saturates and tends to invert slope as predicted by the
simulations, while with the constant average driving the saturation does not
occur.
The effect of varying the angle is represented in figure 5 (b), where the
signals VX and Vy, proportional to the temperature differences (Tz-T1)and (T4-
T3), respectively, are shown for a fixed wind speed. A large offset (not cancelled
in these curves) can be observed for the Vx data. The offset is mainly due to non
symmetrical thermal conductivities deriving from the uncertainty on the glue
layer thlckness between the chips and the laminate and the non symmetrical
copper interconnection layout. The presence of offsets is typical of these systems
and has to be acquired during calibration in order to be cancelled by signal
processing. Solid lines represent the best fit operated using sinusoidal functions.
A good agreement with the prediction of Eq. (1) can be observed.

30
-E 20
10
% O ------___-- _ - - _ _ _ _ _ _ _ _
> >- -10
-20
0 tu -30
0.0 0 2 0 4 0.6 0 8 1.0 1.2 1.4 -50 0 50 100 150 200 250 300 350 400
wind speed ( d s ) angle (deg)

Figure 5. Sensor response with constant power and constant average heater driving (a) and
dependence of the output signals on wind direction for V = 1.4 m/s (b).

Acknowledgments
The authors thank STMicroelectronics of Cornaredo (Milan, Italy) for
fabricating the chips used in this work.

References
1. A. Dobre, S. J. Arnold, R. J. Smalley, J. W. D. Boddy, .J. F. Barlow, A. S.
Tomlin and S. E. Belcher, Atmospheric Environment 39,4647 (2005).
2. J. W. D. Boddy, R. J. Smalley, N. S. Dixon, J. E. Tate and A. S . Tomlin,
Atmospheric Environment 39,31477 (2005).
3. M. Molder, L. Klemedtsson and A. Lindroth, Agricultural and Forest
Meteorology 127,203 (2004).
4. P. Bruschi, D. Navanini and M. Piotto, Sensors and Actuators A 110, 269
(2004).
5. B. W. Van Oudheusden and A. W. Van Herwaarden, Sensors and Actuators
A 21-23,425 (1990).
A Focal Plane Array based on CMOS Uncooled Thermoelectric
Infrared Detectors

C. Calaza, N. Viarani, G. Pedretti, M. Gottardi, V. Zanini, A. Simoni

ITC-irst, via Sommarive, 16, 38050 Povo (TN),Italy. [email protected]

Summary
This paper descnbes an uncooled infrared thermal imager fabncated with a standard CMOS
process (AMS 0.8pm) and a compatible front-side bulk micromachining post-process based
on a TMAH solution. The fabncation approach does not involve any material deposition,
lithography step or particular etch-stop technique after the CMOS process flow, so that the
imager cost is almost equal to that of the CMOS chip cost. The infrared imager is composed
of a focal plane array (FPA) with 16x16 thermopile pixels monolithically integrated with the
addressing and readout electronics. The optical measurements performed with the fabricated
devices have shown a responsivity S of 15.0 V M , a noise equivalent power NEP of 1.37
nW/HzJn and a normalized detectlvity D* of 1.05.10' crn.HzJn.W' for the infrared pixel. The
readout channel features a maximum gain of 85dB with an equivalent input noise of 22
nV/HzIn. An infrared imager based on the FPA has been build and thermal imaging has been
demonstrated

1 Thermopile pixel
Figure 1 shows a micrograph of the fabricated FPA. Each pixel consists of a
thermally-isolated micromachined membrane suspended by two arms that
contain an embedded thermopile temperature sensor made up of
polysilicordaluminium thermocouples. The absorbed infrared radiation heats up
this membrane causing a temperature rise that is measured by the thermopile
sensor. The thermal isolation is attained by etching away the silicon substrate
under the pixel structure by means of a front-side bulk micromachining post-
process based on a TMAH solution [I]. This etching process has been carried
out with the single CMOS dies and does not require a particular etch stop
technique. The etch openings are readily obtained with the CMOS
photolithographic steps, by arranging the several opening masks (active area,
via, contact and pad) on top of each other.
The thermopile is made up of two 5 thermocouple series connected between
them by an additional thermocouple shared by the two support arms. The pixel
output voltage VT is given by VT=N.~.AT,where N, a and AT denote the
number of thermocouples, the Seebeck coefficient of the materials and the
temperature difference between hot and cold junctions. The optimization of the
performance achievable with this CMOS-MEMS approach for a given pixel

122
 123

pitch has been carried out using the finite element method. A FEM model of the
pixel has been used to calculate the temperature distribution caused by a heat
load that emulates the absorption of infrared radiation. The different
performance parameters can be derived using this temperature to evaluate the
thermopile output voltage using the previous equation.

Fig.1 Micrograph view of the fabricated FPA

The geometry selected improves the responsivity attainable with a certain

pixel size by increasing the absorbing membrane area while preserving a good
thermal isolation between the thermopile hot and cold junctions. The pixel
structure is outlined by two etching openings with two different sections: the
main one (20pm width) is intended to expose the silicon substrate during the
etching process, while the narrow one (only 2pm) is intended to lengthen the
support arms with a minimum loss of absorbing area. The use of only two
support arms also allows the arrangement of the pixels in the same column on a
shared etched cavity increasing the overall FPA fill factor. Table 1 summarizes
the main parameters that characterize the FPA pixel. The pixel includes a heating
resistor intended to implement a self-test function that allows an electrical test of
the FPA without need of specific infrared equipment [ 2 ] .

2 Readout channel
el
Since the voltage levels self-generated by the thermopile pixels are in the
range of few pV the implemented readout channel consists of a low-noise
voltage amplifier with high gain that can be tuned for different operation
conditions (from 50dB to 85dB). The readout circuit makes use of the chopper
principle and the correlated double sampling technique to reduce the amplifier
offset level and the noise floor [3]. The addressing of the FPA is carried out by
124

means of two independent decoders (rows and columns) that control the set of
MOS switches integrated in the silicon rims between adjacent columns to
connect the thermopile and the heater ends of the selected pixel to the associated
column lines. The addresses are externally supplied so that the pixels can be
randomly accessed or can be sequentially scanned to obtain an infrared frame.

FPA pixel characteristics

Active area 20.6 x lo3pm2
Fill factor 0.56
Pixel pitch 228 x 220 p n
Arm length / width 173 x 31 p n
Structure thickness From 3.2 to 4.1 pm
Thermocouple number 11
Thermopile resistance 40.1 kR
Seebeck coefficient 143 pV/K

Table 1 Main characteristic parameters of the FPA pixel

3 Experimental results
A specific test structure [4] has been included into the FPA chip in order to
obtain a measurement of the Seebeck coefficient of the n-polysilicodmetal
thermocouples (a=143pVIK). The system (pixel and readout channel) has been
electrically assessed using the embedded pixel self-test capability, which allows
the emulation of the infrared radiation absorption by dissipating a known power
with the embedded heating resistor. The voltage generated by the thermopile can
be calculated from the measured readout output and together with the applied
heating power provides an estimation of the ideal pixel responsivity (100%
absorptance), 26.7 V N . This output voltage allows the estimation of the
temperature difference between the hot and cold junctions, AT=V,IJNaab, that
can be used to estimate the thermal conductivity of the pixel, which is found to
be 5.88.10'5 WIK.
The FPA has been optically characterized in the 8-14 pm band, usually used
for infrared thermal imaging. A rather high absorptance is expected in this band
due to the presence of the Si-N bonds in the passivation layer molecules, which
have vibrational and rotational states with associated wavelengths in this band.
The optical measurements have been carried out using a Scitec IR-21V infrared
emitter as source, an Oriel pyroelectric sensor as calibrated reference detector
and two equal band pass filters to restrict the wavelength range in both detectors.
A reflected image of the infrared emitter has been projected onto the detectors
measuring plane using a gold coated parabolic mirror. Figure 2 shows the
infrared beam profile measured by the fabricated 16x16 FPA. The output signal
obtained with both detectors has been used to evaluate the pixel optical
responsivity, 15.0 V N , which corresponds to a mean absorptance of 56% in the
 125

band between 8 and 14 pm. The noise equivalent power NEP is measured to be
1.37 nW/Ilz"*, which corresponds to a normalized detectivity D* of 1.05.107
cm.Hz'".W'*. The relative spectral responsivity has been measured including a
monochromator into the optical setup. The pixel shows a regular absorption in
this band with an absorption peak at 8.8pm due to the presence of polysilicon
and metal layers in the support arms.

Fig. 2 Projile of the test infrared beam measured with the fabricated 16x16 FPA

ager
4 Thermal imager
A thermal imager has been created using one of the fabricated FPA chips
together with a housing that includes the infrared imaging optics and a computer
that is used to control the operation of the demonstrator (address generation) and
to acquire and display the image data. The housing is basically an aluminium
holder that contains an IR lens casting the image of the thermal scene onto the
sensor array. The lens has a diameter of 7.25 mm, a focal length of 8.7 mm, an f-
number nf of 1.2 and is covered by an antireflection coating that provides a
t r a n s ~ s s i o nof 94% in the infrared band between 8 and 12 pm.
The signals from the individual pixels are multiplexed and amplified by the
FPA on-chip circuitry. The integrated row and column decoders are controlled
through the pixel address supplied by the computer. For the acquisition of a
complete image all pixel addresses are sequentially supplied and the respective
pixel signals are synchronously converted with the computer acquisition
hardware. The A/D converter in the acquisition hardware measures the output
value by integrating the signal during 20 ms, discarding its values during the
first 5 ms after an address change in order to let the amplifier output signal settle
to the new value. This integration scheme is used to suppress the noise and
signal components with frequencies above 50 Hz. With this timing method the
pixel acquisition lasts for 25 ms, while the acquisition of a complete image
126

frame with 256 pixels requires 6.4 s, which corresponds to a frame rate of 0.16
Hz. This frame rate can be increased by reducing either the delay or the
integration time, however, a shorter delay will increase the electrical cross-talk
between pixels, and a decreased integration time will deteriorate the NETD due
to the larger signal bandwidth and increased noise.

Fig. 3 Thermal image of a person, acquired with the thermal imager developed at ITC-irst: (a)
Unprocessed image, 16x16; (b)the image after a bilinear interpolation between pixels, 64x64.

Figure 3 shows an image acquired with the thermal imager based on the
CMOS FPA and the previously described setup. The picture shows a person,
head and chest with one arm up, at a distance of approximately 2 meters. The
blue region corresponds to the cold background at ambient temperature, while
red regions correspond to the hottest parts, 13C above this ambient temperature.
The NETD achieved with this setup is 490 mK, for a signal bandwidth of 50 Hz.

Reference
I. TIMA laboratory, CMP Multi-Project Circuits Service. http://crn~.irn~~.f~
2. B. Charlot, F. Parrain, S. Mir, B. Courtois, Proceedings DTIP2001, Cannes-
Mandelieu, France, April 2001,96
3. D.A. Johns, I<. Martin, Analog Integrated Circuit Design, John Wiley and
Sons, 1997
on Arx, 0. Paul, H. Baltes, IEEE Trans. Semiconductor ~anu~acturing,
201 (1997)

Acknowledgments
ents
This work has been supported by the Provincia Autonoma di Trento, through
the funded project DIRMI.
 A 2-D PLANAR MICROFLUXGATE WITH SPUTTERED
CORE

Andrea Baschirotto
Dept. of Innovation Engineering, University of Lecce, 73 100 Lecce - Italy
Enrico Melissano, Pietro Siciliano
Microelectronics and Microsystems Institute of the National Council of Research
(IMM-CNR), 73 100 Lecce - Italy
Enrico Dallago, Piero Malcovati, Marco Marchesi , Giuseppe Venchi
Power Electronics Laboratory, Department of Electrical Engineering, University
of Pavia, 27 100 Pavia - Italy

Summary
In this paper a double axis planar micro-Fluxgate magnetic scnsor is presented. The
ferromagnetic core material, the Vitrovac 6025 X, has been deposited on top of the coils
with the DC-magnetron sputtering technique, a new type of procedure with respect the
existing solutions in the field of Fluxgate sensor. This procedure allows us to obtain a core
with the good magnetic properties of the amorphous ferromagnetic material, typical of a core
25 pm thick, but with a thickness of only 1 pm, typical of a core electrodcpositcd. The
micro-Fluxgate has been realized in a 0.5 pm CMOS process using copper mctal lines to
rcalize the excitation coil and aluminium metal lines for the sensing coils. Applying a
triangular excitation current of 18 mA peak at 100 kHz the magnetic sensitivity achieved is
about 0.45 mV/pT, suitable for detecting the Earth's magnetic field (i50 pT), while the
linearity error is 1.15% of the full scale. The maximum angle error of the sensor, evaluating
the Earth magnetic field, is 3.4". The power consumption of the sensor is about 13.7 mW.

Introduction
Many applications require low-cost and low-power highly sensitive
miniaturized magnetic sensors to locally and accurately measure weak magnetic
fields in a plane [l, 21. For this task integrated devices appear to be the best
solution. Unfortunately, none of the existing integrated magnetic sensors, such
as magnetoresistive [3] or Hall sensors [4], fulfils simultaneously all the desired
requirements due to their poor magnetic field measurement accuracy.
Innovations are expected in the integrated sensors based on the Fluxgate
principle.
Following the experience in [ 5 , 61, the sensor presented in this paper is a
micro-Fluxgate, developed in a CMOS process, where the ferromagnetic core is
realized as a post processing step with the DC-magnetron sputtering technology.
This technique allows us to realize a core with the good magnetic properties of
the amorphous ferromagnetic material and with a very small thickness (about 1
pm). At this time, in literature it is possible to find solutions with

127
128

electrodeposited magnetic

asic sensor structure

Fig. 1 shows a photo of the micro-Fluxgate sensor, realized in a 0.5 pm
CMOS integrated circuit process. The planar excitation coil, which has been
realized in copper, is characterized by 5.5 pm thickness, 71 turns and 12 pm
pitch (8 pm metal width and 4 pm of spacing between two metals). The total
area for the excitation coil is 1760 x 1760 pm and its resistance is about of
123.4 a.
The geometrical parameters of the sensing coil, realized in aluminum, are:
1 pm thickness, 1.4 pm metal width and 1.6 pm of spacing between two metals.
The total area for the sensing coils, consisting of 66 turns, is 650 x 650 pm2 and
their resistance is about of 1.84 kS2.

Fig. I Photograph of the micro-integrated Fluxgate realized (with the sputteredferromagnetic core
of l p m thick).

The ferromagnetic cross-shaped core (Fig. l), realized with the Vitrovac
6025 X [7], has been obtained with the DC-magnetron sputtering technique. The
thickness of the sputtered core, of about I pm, has been evaluated with a SEM
microscope.
In order to optimize the DC-magnetron sputtering process and to verify that
the material is not degraded by the deposition process, we performed several
measurements of the magnetic properties of the deposited material, with a
system of coils similar to the loop tracer reported in (81. A comparison between
the sputtered and the as-cast ferromagnetic material hysteresis is shown in Fig. 2.
It is evident that the two measured loops are different from the loop given by the
supplier (Vacuuschmelze). These differences can be explained as follows:
1. In our system the magnetic circuit is open (the sputtered core is not a toroid)
 129

Naggtic Fiekl [Nd

Fig. 2 Comparison between the hysteresis loops of the Vitrovac sputtered and as-cast (dashed
line).

Because of this only a comparison between the B-H loop of the sputtered
material and the as-cast material is possible. From this point of view, the two
loops look similar, resulting in nearly equivalent magnetic properties. The
fundamental difference is the coercitivity field, that for the deposited core is
greater, about 100 N m . This difference is in agreement with the results
described in [9]. The increase of the coercitivity after deposition of the material
will result in a slight increase of the current required in the excitation coil to
achieve the saturation of the core. The measured magnetic properties of the
deposited material are anyway suitable for the correct operation of the Fluxgate
sensor with only a slight increase of the power consumption.

Measurement and results

The characterization of the sensor has been realized using a universal source
(HP3245A) to supply the excitation coil. The frequency of excitation has been
fixed to 100 lcHz and the excitation current has been set to 18 mApe*, which is
the value necessary for the core saturation.
The fundamental component of the differential output voltage, after an
amplification of about 33 dB, and imposing an external magnetic field with a
pair of Helmholtz coils, has been evaluated using a spectrum analyzer
(HP3589A). This voltage is plotted in Fig. 3 against the external magnetic field.
The sensor shows a linearity error of about 1.15% of the full scale in the range of
$50 pT with a sensitivity of about 0.45 mV/pT.
To evaluate the possibility of using this integrated micro-Fluxgate sensor as a
compass, the voltage output has been measured while rotating the sensor in a
130

horizontal plane. Fig. 4 shows the output voltage for each axis of sensitivity
exposed to a magnetic induction of about 20 pT.The angle error is smaller than
3.4" and includes signal non-linearity, hysteresis and noise. The power
consumption of the sensor, considering the resistance of the excitation coil of
! and 18 mApe*of triangular current excitation, is about 13.7 mW. At the
123.4 2
moment this power consumption is not comparable with the other existing
solutions because it is evaluated for the sensor alone. In this case the circuitry
has not been realized yet and therefore its power dissipation was not taken into
account.
--~-
"
OMi*
~---;Ii
+1
,- ,1
- I
I
~~~~

I
I -.LL
-,

Flw -w bTI
Fig. 3 Voltage output form the sensor rmposing diferent value for the external magnetic field.

Fig. 4 Angle response for the two measurement axes (X and r).

Conclusions
In this work a micro-Fluxgate with ferromagnetic core sputtered has been
presented. This technique allows us to reduce the thickness of the core and
 131

therefore the power consumption maintaining the good magnetic properties of

the amorphous ferromagnetic material, the Vitrovac 6025 X, used as target.
The experimental results presented show that the sensor has good sensitivity,
0.45 mV/pT, and linearity, with an error of about 1.IS%, in a range of A50 pT.
The measurement error when measuring the magnetic field angle is smaller than
3.4 and includes signal non-linearity, hysteresis and noise. The power
consumption, referred only to the sensor, is actually about of 13.7 mW. The
magnetic properties of the sputtered core could be improved with suitable
treatments, e.g with an annealing after its deposition; hence we also expect a
reduction of the power consumption.

Acknowledgment
The authors would like to thank ing. Marco Morelli (STMicroelectronics) for
the support in the micro-integrated coils realization and Vacuumschmelze,
Hanau Germany, for VitrovacB samples and support.

Refrences
[ 11 Pave1 Ripka, Magnetic Sensors and magnetometers, Artech House
Boston, London (2001).
[2] M. Vopalensky, P. Ripka, A. Platil, Precise magnetic sensors Sensors and
Actuators A 106 (2003) pp. 38 - 42.
[3] C.H. Smith, R.W. Schneider, Magnetic field sensing utilizing GMR
materials, Sensor Review 18 1998 23&236.
[4] R. Popovic, W. Heidenreich, Magnetogalvanic sensors, In: R. Boll, K.J.
Overshott, Eds., Magnetic Sensors, vol. 5, Sensors, A Comprehensive
Survey, VCH, 1989, pp. 43-95.
[5] A. Baschirotto, E. Dallago, P. Malcovati, M. Marchesi, G . Venchi, From a
PCB Fluxgate to an integrated micro Fluxgate magnetic sensor, Proceeding
of Instrumentation and Measurement Technology Conference (IMTC-2005),
Ottawa (Canada) 17 - 19 May 2005, pp. 1756-1760;
[6] A. Baschirotto, F. Borghetti, E. Dallago, P. Malcovati, M. Marchesi, E.
Melissano, G. Venchi and P.Siciliano, Fluxgate magnetic sensor and front-
end circuitry in a micro-integrated system, EUROSENSORS XIX,
Barcelona (Spain) 11-14 September 2005;
[7] Vacuumschmelze Hanau Germany, Amorphous metals VITROVAC
www.vacuumschme1ze.de;
[8] P.T. Squire, S.M. Sheard, C.H. Carter and M.R.J. Gibbs, Digital M-H
plotter or low-coercivity metallic glasses, J.Phys. E: Sci Instrum. 21
(1988), pp. 1167-1172.
[9] M. Ali Growth and Study of Magnetostrictive FeSiBC Thin Films for
Device Applications, University of Sheffield (1999) Ph.D thesis
 An optical probe for RF electric field

Lucio Ciccarelli, Mario Medugno

I.MM - C.N.R., Neaples. Italy
ILucio.Ciccarelli, Mario.Medugnol(ii)na.imm.cnr.il

Summary
Electromagnetic field (EMF) sensors are widely used in several industrial, scientific and
medical applications and for measurements of fields yield by a huge set of communication
devices. We focuses on some experiences on the use of an integrated optical device which
implements a Mach-Zehnder interferometer, and on its further development as an
electromagnetic field sensor. The probe is mainly made by non conductive materials, making
such device suitable for experimental measurement of electromagnetic fields in near field
region (or Fresnels one) of transmitting antennas. Here no simple theory is available in
order to evaluate the fields and mutual coupling between antennas and standard probes
strongly affect the measurements: the optical probe avoids the coupling of the fields with
metallic structures and the loss of antenna calibration which typically yield measurement
errors.

1 Introduction
Standard expensive sensors probes for electric and magnetic field
measurements [l], also in the Fresnel region, are often equipped with optical
transmission lines to avoid coupling with the measuring field. Their heads
generally need both power source and electronics for: i) conditioning and
processing of the signals received by an external antenna; ii) electro-optical (EO)
conversion to eventually drive an optical link with the measure equipment.
The mutual coupling of EMF with the standard probe scheme is the source of
precision loss, which can strongly be reduced by a completely optical probe.
Recently have been developed new EMF sensors made by dielectric materials to
overcame this problem.

2 The principle of an optical EMF

The probe is an optical driven device with input-output optical fibers
connecting it to a laser source and a photodetector. A very tiny resistively loaded
antenna with a flat response in a wide band, yields the electric input for an
integrated Mach-Zehnder interferometer. The optical probe is implemented on a
L i m o 3 substrate as sketched in the Figure 1.
132
 133

T-

Figure 1: Optical EMFprobe scheme

Applying an electric field to the electrodes, a phase shift r arises in the arms
of the interferometer yielded by the electro-optic effect of the LiNb03
underlying the waveguide channels; this produces an amplitude modulation in
the optical output. Optical probes result in small dimension devices mainly made
by non metallic parts, thereby dramatically less invasive than standard EMF
probes.

2.1 The optical interferometer

The base component of the optical probe is a symmetric Mach-Zehnder
interferometer. It was developed on a LiNb03 substrate in order to exploit the
Pockel effect which results in a linear change of the refractive index due to an
external E-field. The substrate has X-cut, Y-propagating waveguides, as the
extraordinary index of the Z axis ne is lower than the ordinary index no of the X
axis, where the relatively large electro-optic coefficient is r33= 32,2 p m N . The
phase shift is I-= -T V rj3 L n,$/;E G E L , where L is the electrodes lenght and
GELtheir spacing. Planar and channel optical waveguides (green ellipses in the
Figure) were manufactured by using high energy ion implantation technique of
oxygen ions at 5 MeV [2].

3 Optical EMF probe characterization and experiments

In this section we characterize electrically and optically the integrated optical
device, and we propose a model suitable for implementing non-invasive probe in
a range of several Gigahertz.
The Mach-Zehnder interferometer is used as an optical modulator of the
input laser source, therefore the device can be modeled by the electric equivalent
circuit shown in Figure 2. Here the generator is the antenna (which voltage
equals the field E times the antenna effective length he) with two series
134

impedances: the antenna impedance Z, and antenna loading ZI. The modulator
impedance has an optical input port and an output optical port both characterized
by two efficiency parameters qi, qo, therefore it converts the electric signal at
the electrodes into an optical signal according to the Thevenin complex
equations and the optical power budget.

From LASER

. To PD
1

I I

Fig. 2 Equivalent circuif of the EMFprobe

The interferometer electrodes have a push-pull configuration such that a

dual phase change is induced in the arms, yielding a doubled Pockel effect. The
electrodes, which parameters are in Table 1, are supplied with the signal induced
by a suitable resistively loaded dipole antenna, implementing a progressive wave
in order to suppress the typical resonance of metal dipoles. The design of a such
tapered antenna is derived according to the Wu-King approximation [3,4] and is
characterized by an impedance distribution:

Z(X) = 60g(h, k, a) /(h/2 - Ixv where:

g(h, k, a) =2 { [sinh-1 h/a - Ci (2ka, kh) f j [2(1-e-jkh) / hk - Si (2ka, kh) ]

and h is the dipole length, A is the wavelength, k =2n/A is the wavenumber, a is

the thickness of the antenna, Ci is the cosine integral and Si is the sine integral.
Z(x) can be realized as a pure resistance because when a < I the imaginary part
of g(h, k, a) is smaller then the real part of g(h, k, a). For this kind of antenna
Z
l ,l the module of the antenna impedance at the gap, is a monotone decreasing
function of the frequency and can be shown that the peak corresponding to the
resonance frequency are smoothed with respect to one occurring in an ordinary
metal dipole.
Using this approximation we realized a triangular shaped resistively loaded
antenna 8 x 0.35cm, increasing the frequency range of linearity for the EMF
probe. The antenna elements are made by the following process: 1) Deposition
of a 1.4 pm film of I T 0 by sputtering on the substrate (1 mm glass); 2)
Deposition by spinning of reversible photoresist; 3) Lithography of the I T 0 film
 135

by the antenna element mask; 4) Subsequent development of photoresist

(negative process); 5) Wet etching of IT0 elements by HCl.

Electrodes length 5,O e-02 [m]

Electrodes thickness 1,4 e-06 [m]
M~dulatorcapacity 2,2 e-11 [F]
Modulator resistance 1,0 e+8 [ohm]

Table I Optical probe electrical characterization.

The MZ based optical probe has been preliminary tested with different laser
source: a 2 mW in fiber laser diode (LD) at 1300 nm, a 150 mW in fiber LD at
980 nm, a 100 mW in fiber LD at 830 nm and an He-Ne in fiber coupled laser at
630 nm; the optical power was variable for each LD. After several
measurements in laboratory and in anechoic chamber with an open-ended wave
guide, we experimented a random power loss frequently dependent on optical
device and fibers stress. Such power loss fluctuations were due to damages in
fiber lines which were optically detected next to the pigtails of the
interferometer.

Fig. 3 Probe responsef o r ELF and VHF band

The measured total loss was greater (63 dE3) than the calculated one by the
optical power budget below; new fibers have been soldered and the power
fluctuations were reduced at all. Photodetectors suitable for the sources spectrum
and power were tested. We used at first a Thorlabs DET 410 photodetector
electrically terminated with different RF shielded resistive loads RL for probe
exposition in the range ELF to VHF band (30 Hz-150 MHz). We experienced
several critical problems with such device: the conversion slope was not constant
vs frequency with upper limit at few MHz (dependent on RL), the battery charge
state was critical for interferometric and V, measurements (i.e. the voltage
yielding the hi2 phase shift). In order to resolve these problems we used a
136

Newport amplified photodetector with a wider band (up to 1,5 GHz) and a
conversion gain of 200 V/W, necessary to match the optical power budget used
in this experimental phase, that is:

LT= L m f LFC f LS = 3 dB/cm x 7crn +2 x 3 f 2 x 0,s = 28 dB

where LT is the sum of losses due to Mach-Zehnder interferometer, the two

Fiber Coupling and the two Solderings of optical fibers.
In the Figure 3 is shown electrical outputs of the probe exposed to Extremely
Low Frequency (ELF) composed by the fields of power supply at 50 Hz and an
additive electrical field supplied by an square wave at 200 Hz (on the left). The
environmental 50 Hz field was measured with a standard probe (PMM
mod.8053A) and it was 300 mV/m. A small exposition cell has been realized in
order to generate stronger fields at higher frequencies and to mitigate the
external noise. In this way we obtained cleaner interference figures as showed in
Figure 3. Sensitivity was estimated in order of 100mVIm.

4 Conclusions
The proposed EMF probe configuration is suitable for implementing low cost
optical chips for Industrial, Medical, Scientific applications e.g. antenna radar
systems for geophysical applications which have to reach tens of meters under
the soil, spatial applications where integrated optical devices are suitable, and
biomedical applications which requires non invasive sensors. As further
development, the antenna probe must be integrated into the interferometer
substrate in order to a) optimize the implementation process, b) minimize the
electrical coupling with the field.

5 References
[I] Kanda M 1993 Standard probes for electromagnetic field measurements
IEEE Trans. Antennas andpropag. 4 1 10 1349-64
[2] Bentini G G, Bianconi M, Chiarini M, Correra L, Macagnani P, Tamarri
F and Guzzi R 2002 Theory and design of an integrated optical sensor
based on a Mach-Zehnder interferometer Sensors and Microsystems
Proceedings of 7th italian conference, World Scientific 3 18-25
[3] Wu T T and King R W P 1965 The cylindrical antenna with nonreflecting
resistive loading IEEE Trans. Antennas andpropag. AP-13 3 369-73
[4] Shen L C and King R W P 1965 Corrections to The cylindrical antenna
with nonreflecting resistive loading IEEE Trans. Antennas andpropag.
AP-13 6 998.
 Coated Fiber Bragg Grating As High Sensitivity Hydrophone

Salvatore DAddio, Andrea Cusano, Antonello Cutolo

Engineering Dept., University of Sannio, Corso Garibaldi I07 82IOO Benevento
Italy
[email protected]

Stefania Campopiano
Department for Technologies, University of Naples Parthenope, Napoli, Italy

Michele Giordano
Institute for Composite and Biomedical Materials, National Research Council
Piazzale Vincenzo Tecchio, 80 - 80125 Napoli, Italy

Mario Balbi, Sergio Balzarini

WhiteheadAlenia Sistemi Subacquei, via Monteruscello 75, Pozzuoli, Italy

Summ ary
In this work, a new fiber optic hydrophone based on the intensity modulation of the laser
light in an FBG (Fiber Bragg Grating) under the influence of the sound pressure is
presented. In order to increase the sensitivity, the FBG has been coated with a proper
material. The minimum detectable acoustic pressure is found to be about 10Pa. In
comparison with conventional fiber optic hydrophones based on bare FBGs, an increase of
sensitivity of three order of magnitude is obtained. In addition, the developed FBG
hydrophone detects an acoustic field in water with excellent performances: linear response
and wide dynamic range. Thanks to the simplicity of its operating principle, geometry and
interrogation scheme, this type of hydrophone can provide a pratical alternative to the
current state of the art in piezoelectric hydrophone technology.

1 Introduction
In the last years, fiber optic hydrophones have been attracting considerable
interest due to their potential advantages, such as immunity to electromagnetic
interference, remote sensing, stability in harsh environments, wide dynamic
range. Several fiber optic sensing schemes have been reported, including
interferometric techniques [ 1-31, polarimetric techniques [4-61 and fiber grating
techniques [7-91. While a conventional fiber optic hydrophone is composed by
an ordinary optical fiber and relies on the phase shift of the laser light
propagating through the sensing fiber under the influence of the sound pressure,
the operating principle of a FBG-based hydrophone is typically based on the
intensity modulation of the laser light due to the shift of transmission power
spectrum curve of the sensing element under the influence of the acoustic field.

137
138

The important feature of the intensity-modulated FBG-based hydrophone is its

simplicity. Since the detection is carried out by the measurement of light
intensity, which is generally easier than that of a shift in wavelength or phase,
the structure of an FBG-based hydrophone is expected to be more compact and
simpler compared to a conventional optical fiber hydrophone. Unfortunately, the
low sensitivity to acoustic pressure of this class of sensors will limit their use in
underwater applications, where piezoelectric transducers are widely used despite
their dimensions, complex signal processing and electronic front-end and
difficult multiplexing. In this work, a novel fiber optic hydrophone that uses a
fiber Bragg grating embedded in a material of low elastic modulus is presented.
For a given acoustic pressure, the basic effect of the FBG coating is to
enhance the dynamic strain experienced by the sensor of a factor given by the
ratio between the fiber and the coating elastic modulus.

2 Principle of operation
In this section, the interaction between a uniform Bragg grating of length L,
written into the core of a standard single-mode fiber, and an incident acoustic
wave is presented. For a spatial uniform sound pressure P(t)=psin(wst) around
the FBG (where p and ws are the amplitude and angular frequency of the sound
pressure, respectively), the corresponding normalized Bragg wavelength shift
AhB/hBis given by ref. 9:

where n=1.465, E=70GPa, v=0.17 and pll=O.121 and p12=0.270 are the effective
refractive index of the guided mode, the Youngs modulus, Poisson ratio and the
elasto-optic coefficients of the optical fiber, respectively. Thus, the spectral
response of the FBG moves without changing its shape at the same frequency of
the applied acoustic pressure. For a GE-doped FBG at 1550 nm, AhB/AP was
measured as -3x 10 nm/MPa over a pressure range of 70Mpa [lo]. This means
that with interrogation units able to perform wavelength shift measurements with
a resolution of 0. lppm in the investigated acoustic frequency range, an acoustic
pressure limit of detection of tens of KPa can be obtained.
When the optical fiber is coated with plastic material, it exhibits some order
of magnitude increase in its pressure sensitivity [ 11- 121.
Indeed, according to the Hocker analysis [13], if the FBG is coated with a
thick layer of polymer, the normalized wavelength pressure sensitivity is given
by:
 139

where E,,, and v,,, are the Youngs modulus and the Poisson ratio of the coating,
respectively.
Therefore, it can be seen from eq.(2) that for coatings with small Youngs
modulus compared with the fiber one, the wavelength pressure sensitivity of the
FBG can be increased significantly. This means that by a proper choice of the
coating elastic properties, shape and dimension, excellent pressure sensitivity
gain is possible.

3 Experimental set-up
As mentioned in the previous section, the wavelength shift of the spectrum
curve of the FBG is directly related to the incident sound pressure. Based on this
line of argument, a low cost intensity based demodulation technique that relies
on the use of a narrowband source and a broadband FBG is used. In the
transmission mode and by working on the edge of the grating spectrum, the
transmitted optical power P, is directly related to the sound pressure according to
the following expression:

where P, is the incident optical power, To is the transmission value at FWHM,

dT/ako and represent the edge slope of the grating spectral response and
the wavelength sensitivity to the pressure, respectively. Yet, in order to achieve
the maximum sensitivity and dynamic range, the laser wavelength should be set
in correspondence of the full width at half maximum (FWHM), on either the
longer or shorter wavelength side of the spectrum curve. It is seen from eq.(3)
that the ac component of the transmitted light power is proportional to the sound
pressure experienced by the FBG. Thus, the detection of the light with a
photodiode provides an electrical output directly proportional to the acoustic
field in the water. From the resulting temporal waveform, the amplitude, the
frequency and the phase of the field can easily be measured after a proper
calibration procedure.
The experimental set-up is shown in Figure 1. The acoustic field is generated
by a PZT acoustic transducer, immersed in a water tank of 1lm long, 5m large
and 7m deep together with the reference PZT hydrophone and the hydrophone
under test, at a distance of l m and 2.2m from the PZT transducer, respectively.
A computer-controlled scanning stage that allows independent translations in X,
Y, and Z directions and rotations about the vertical axis is used to place the
optical fiber hydrophone. The utilized fiber Bragg grating (characterized by a
140

central wavelength of 1547.62nm and a FWHM of 0.45nm) was embedded in a

material, the Damival 13650 Gecalsthom, of spherical geometry of diameter of
D=4.4 cm exhibiting a lower value of elastic modulus than the fiber. The optical
source was a stable narrowband wavelength tunable laser, its output wavelength
is tuned to the center of the slope of the transmission spectrum curve of the FBG,
at 1547.85nm, in order to achieve the better performance of the interrogation
system. The power of light is selected at the value of 8mW. The wavelength
stability was less than lpm. The laser light intensity modulated by the FBG
subjected to the sound pressure is detected by a photodiode. Two optical
isolators, the first at the output of the laser, the latter at the input of the
photoreceiver, are inserted in order to stabilize the sensor output signal. The
resolution of the system was estimated to be slightly less than lpm in the
frequency range of interest. The outputs of the optical fiber and electric
hydrophone were collected in a PC by an AID acquisition board.

/
FBG / I PZT acousnc
sensor Reference PZT Acusuc wave transducer
hydrophone

Fig.1 Lateral view of the experimental set-up

4 Results and discussion

In this section, we report the results of measurements carried out to evaluate
the performances of the new optical fiber hydrophone and to compare the
obtained performances with a reference PZT hydrophone. In particular, in figure
2-left, a comparison between the typical temporal response of the sensor FBG
under test and the reference hydrophone PZT to a sound pressure pulse at the
frequency of 4.6KHz is shown. It can be seen that the FBG hydrophone operates
as good as the piezoelectric technology. The phase difference between the two
responses is due to the different distance from the acoustic source. Using the
measured signal-to-noise ratio from the FFT of sensor response, the minimum
detectable pressure level was estimated to be about 10Pa. In figure 2-right, the
sensitivity curve of the FBG sensor as function of the the sound pressure
frequency is plotted. It can be seen that the FBG hydrophone presents a
decreasing response in the range from low frequencies up to about 16KHz. For
 141

upper frequencies the signal to noise ratio approaches the unity around the value
of -235dBreV/pPa.

Fig.2 Left: Typical temporal response of the sensor FBG under test (upper) and the reference
hydrophone PZT (lower) to a soundpressure pulse of 2KPa at the frequency of 4.6Kuz
Right: the calculated sensitivity curve of the FBG hydrophone

In conclusion, preliminary results demonstrate performances that could be

really competitive with conventional and widely used piezoelectric based
hydrophones for underwater applications.

References
13 J. A. Bucaro et al., Appl. Opt., vol. 16, no. 7, pp. 1761, Jul. 1977.
21 P. C. Beard et al., Electron. Lett., vol. 33, no. 9, pp. 801, Apr. 1997.
31 C. Koch, Appl. Opt,, vol. 38, no. 13, pp. 2812, May 1999.
41 R. P. De Paula et al., IEEE J. Quantum Electron., vol. 18, n. 4, pp. 680, 1982.
51 H. L.W. Chan et al., J. Appl. Phys., vol. 66, no. 4, pp. 1565, Aug. 1989.
61 NK. S. Chiang et al., J. Lightw. Technol., vol. 8, no. 8, pp. 1221, 1990.
71 E. Fisher et al., Applied Optics Vol. 37, No. 34, Dec. 1998
81 N. Takahashi et al., Ultrasonics 38, 581 (2000)
91 D.C. Betz et al., J.l of Smart Materials and Structures, v01.12, pp.122, (2003).
101 Xu M.G. et al., Electron. Lett. 29 398-9 (1993)
113 J. A. Bucaro et al., Applied Optics, Vol. 18, Issue 6, pp. 938-941 (1979)
121 G. B. Hocker, Applied Optics, Vol. 18, Issue 21, pp. 3679-3681 (1979)
131 Hocker G.B Appl. Opt. vo1.18 pp.1445 (1979)
 Strain Measurements in Porous Silicon by Raman Scattering

M. A. FERRARA92,L. SIRLETO, G. MESSINA2,M. G. DONAT02,

L. ROTIROTI AND I. RENDINA
Istitutoper la Microelettronica e Microsistemi - CNR
Via P. Castellino 111 - 80131 Napoli
2
DIMET - Universitri Mediterranea Reggio Calabria,

at itonella.fkrrara(4na.iinm.cnr.it

Summary
Strains produce a modification of the vibration properties of materials, which can be
observed by Raman scattering: a compressive stress results in an increase of the Raman
frequency shift, while tensile stress results in a decrease. On this line of arguments, in this
work, we investigate adsorption strains in porous silicon by Raman spectroscopy.
The adsorption of chemical species induces a compressive strain in the porous silicon due to
the action of the molecular forces, as a consequences a reversible blue-shift of the Raman
spectra of the sample has been observed.

Introduction
Porous silicon (PS) is a material of great interest for the study of adsorption
phenomena. The structure of PS may be spongelike, moreover, the high
specific area and the great reactivity of the porous layer allow its use as a base
for detection of vapors and liquids [ 1,2].
For sample of porous silicon, strain depends on various parameters, like the
porosity of the porous silicon layer, the doping type and the level of the silicon
substrate. Measurements of strain have been carried out with high resolution
X-ray diffraction apparatus [3]. Due to the large internal surface, the strain is
also very sensitive to oxidation, and the presence of the fluid or to
impregnation of various substance. For example in Ref. [4]has been proved
that when a p+ type porous silicon layer is filled by a liquid, a small increase of
strain is observed for pentane and ethanol. On the contrary, for vapour
adsorption a small decrease of strain has been observed when capillary
condensation occurs in the pores [ 5 ] . In both cases X-ray diffraction has been
used in order to measure strains induced by vapour adsorption in porous
silicon.
Raman scattering can be used to measure strain induced in substrate, in fact
an increase of the Raman frequency shift is observed when a compressive
stress is applied, while tensile stress results in a decrease [ 6 ] .

142
 143

In this paper, we study adsorption strains in porous silicon by using

spontaneous Raman scattering. We prove that when the PS structure is exposed
to saturated vapor of isopropanol or ethanol, a reversible blue shift of the
Raman spectra is observed.

Spontaneous Raman Emission

Porous silicon is composed of wires and/or dots of nonuniform dimensions.
If the size of the particle reduce to the order of nm, the wave function of
optical phonons will non longer be a plane wave. The localization of wave
function leads a relaxation in the selection rule of wave vector conservation.
Not only the phonons with zero wave vector q=O, but also those with q>O take
part in the Raman scattering process, resulting in the red shift of the peak
position and the broadening of the peak width. The line-shape asymmetry
increases as the incident photon energy is increased [7-151.
A quantitative model, developed by Campbell and Fauchet, calculates that
the peak position mainly depends on the number of atoms included in a cluster,
while the width of spectra depends on the shape of crystallites [13,14]. If PS is
modeled as an assembly of quantum wires, the phonon confinement is assumed
to be two dimensional, while if the PS is modelled as an assembly of quantum
dots, the confinement is three dimensional. The weight factor of the phonon
wave function is chosen to be a Gaussian h c t i o n as follows:

where L is the correlation length related with the size of dots. Square of
Fourier transform is given by:

The first-order Raman spectrum I(o) is thus given by:

144

where q is expressed in units of 2 and a=0.54 is the lattice constant of

a
silicon, r is the natural line width for c-Si at room temperature (3.5 cm-I) and
w (q ) is the dispersion relation for optical phonons in c-Si [ 121 which can be
taken according to:

w ( q ) = w, .-120(;)2 (4)

where w,,= 520 cm-' and qo = 2 n / a 0

Experimental Results
The PS sample used in this experiment is a porous silicon monolayer
obtained by electro-chemical etching on p+ type (p=8-12& cm) standard
silicon wafer. The total porous silicon sample thickness is 20 micron and the
porosity is 70%.
The Raman scattering measurements were carried out, at room temperature,
by using a Jobin Yvon Ramanor U- 1000 double monochromator, equipped
with a microscope Olympus BX40 for micro-Raman sampling and an
electrically cooled Hamamatsu R943-02 photomultiplier for photon-counting
detection. A Coherent Innova 70 Ar+ laser, operating at 514.5 nm wavelength
was utilized as excitation source. In order to prevent laser-annealing effects,
the laser power was about 2 mW at the sample surface. Using a 50X objective
having long focal distance, the laser beam was focused to a diameter of about 1
pm. Its position on the sample surface could be monitored with a video
camera. All components of the micro-Raman spectrometer were fixed on a
vibration damped optical table.
The location of the Raman peaks and the shape of crystallites are obtained
fitting the experimental results by a curve obtained with the model proposed by
Campbell and Fauchet formulated in (3). We start our investigation measuring
the normalised Raman spectra in unperturbed porous silicon monolayer. The
results are shown in Fig. 1; the measured Raman peak is at about 519.7 cm-I,
corresponding to a red-shift of 15.7 THz with respect to the pump wavelength.
We note that the observations are in excellent agreement with the value of the
optical phonon frequency in porous silicon [ 12,151. The estimated crystal size
is of 6.OE-7 cm.
In Fig. 2 are shown the experimental Raman spectra in PS sample after
exposure to isopropanol (a) and ethanol (b). Fitting the experimental results
 145

with the model developed by Campbell and Fauchet [13,14], Raman peaks are
obtained at about 520,2 cm-' both for isopropanol and for ethanol. Therefore,
we measure a shift of Raman spectra with respect to the unperturbed case of
about 0.5 cm-' both for isopropanol and for ethanol. These shifts are in
agreement with the compressive strain obtained by Dolino et al. by X-rays
diffraction [3]. We note that the Raman shift is reversible.
In both cases, the strain is compressive because an increase of the Raman
frequency shift is observed.

07
m
I

$ 06
D
05
:
:
2 04
E
2 03

02

01

0
Raman shiff [cm-11

Figure I : Unperturbed Raman spectra in porous silicon sample.

(a) @I
Figure 2 Spontaneous Raman spectra in porous silicon sample ufler exposure to (a)
isopropanol and (b) ethanol.

Conclusions
In this paper, adsorption strains in porous silicon by Raman measurements
has been investigated. The experimental results prove that when the porous
146

silicon structure is exposed to isopropanol or ethanol a reversible increase of

the Raman frequency is observed.

References
[ l ] A. G. Cullins, L. T. Canham and P. D. J. Calcott, J. Appl. Phys. 82, 909-
965, 1997.
[2] W. Thei B., Surf. Sci. Report 29,91-192, 1997
[3] D. Bellet, G. Dolino, Thin Solid Films 276, 1-6 (1995)
[4] D. Bellet, G. Dolino, Phys. Rev. B 50, 17162-5 (1994)
[5] G. Dolino, D. Bellet, C. Faivre, Phys. Rev. B 54, 17919-30(1996)
[6] I. De Wolf, Spectroscopy Europe 15/2 (2003)
[7] J. M. Perez., J. Villalobos, P. McNeilll, J. Prasad, R. Cheek and J. Kelber,
J.P. Estrera, P.D. Stevens and R. Glosser, Appl. Physics Lett. 61 ( 5 ) , 563-
565,1992.
[8] G. Mariotto, F. Ziglio, F. L. Freie Jr., Journal of non cristalline solids 192
- 193,253-257, 1995.
[9] Z. Sui., P. P. Leong, and I. P. Herman, Appl. Physics Lett. 60 (17), 2086-
2088, 1992.
[I 01 S. L. Zhang, Y. Hou and K. S. Ho, B. Qian and S. Cail, J. Appl. Phys. 72
(9), 4469-4471, 1992.
[ll]J. D. Moreno, F. Agullo-Rueda, E. Montoya, M. L. Marcos and J.
Gonzales-Velasco, R. Guerrero-Lemus and J. Martinez-Duart, Appl. Phys.
Lett. 71 (13),2166-2168, 1997.
[12]F. Kozlowski, W. Lang, Appl. Phys. Lett, 5401-5408, 1992.
[ 131H. Ritcher, Z. P. Wang and L. Ley, Solid State Communications 39, 625-
629, 1981.
[ 141I. H. Campbell and P. M. Fauchet, Solid State Communications 58, 739-
741, 1986.
[15] R. Tsu, H. Shen and M. Dutta, Appl. Physics Lett. 60 (l), 112-1 14, 1992.
 Fiber Bragg Gratings and Magnetic Shape Memory alloys: a
Novel High Sensitivity Magnetic Sensor

C. Ambrosino, P. Capoluongo, D. Davino, C . Visone, A. Cutolo, A. Cusano

Eng. Dep. of the Univ. of Sannio, Benevento ITALY
S . CamDopiano
Dept. for Techn. of the Univ. of Naples Parthenope Naples ITALY
stefania.camnpopiano(ir),unisannio.it

M. Giordano
Inst. for Composite and Biomedical Materials, Nat. Res. Council, Naples ITALY

Summary
In this work a novel magnetic field sensor based on Fiber Bragg Gratings (FBG) is
demonstrated. Up to now, FBGs based magnetic sensors have been proposed by using
Terfenol, metallic alloy as transducer Here, a different configuration is proposed employing
a new class of magnetic active materials, Magnetic Shape Memory alloys (MSMs) with a
very giant magnetostriction coefficient. A FBG has been bonded on a MSM sample and
sensor characterization has been carried out. Experimental results and comparison with
Terfenol based FBG sensor performances are reported, showing, for MSM, a sensitivity
(A&&)/(AH/H,) of 1.81*10-4, where hB is the Bragg wavelength and H, the mean
magnetic field in the considered range.

1 Introduction
In last years a great number of applications of magnetic sensors, spreading
from electrical power industry to control engineering and robotics have gained
great interest [ 1-31. For all these applications optical current sensors are a
winning choice, due to their many advantages compared to conventional iron-
core current transformers, including their immunity to electromagnetic
interference, high dynamic range, easily integration within structures and
competitive costs [4]. In particular, among optics devices, FBGs exhibit good
features in terms of sensitivity, resolution and bandwidth [S]. In general, the
common FOSs used to detect magnetic field are based on the Faraday effect, but
due to the low value of the Verdet coefficient, they showed very low sensitivity.
In order to overcome this drawback, FBGs based sensors exploit different
transducers. Recently, FBG based magnetic field and current sensors have been
demonstrated [6,7], exploiting Terfenol-D magnetostrictive rod as elasto-
magnetic transducer. Unfortunately, Terfenol-D works for low magnetic field
achieving a non linear range of response already around few tens of kA/m and a
maximum strain of 0.24% [S]. Along this line of argument, here, a novel FBG
based sensor is proposed employing a new class of magnetic active material,
Magnetic Shape Memory alloys (MSMAs) [9-121. This material presents a

147
148

magnetostriction coefficient that is almost two order of magnitude higher than

Terfenol one (up to 10%) [9] and so it can reach a very high sensitivity. The
sensor is formed by a FBG bonded on the MSMA stick. When the active
material is subjected to magnetic field there is a redistribution of its internal
martensitic twin variants, resulting in the change of the element shape. This
change in the active material produces strain that is transferred to the bonded
FBG, which in turn changes its resonance wavelength. The FBG interrogation
system used for this sensor relies on a low cost ratiometric technique based on
the use of a broadband light source and an in fiber grating optical filter, able to
convert wavelength shifts in amplitude modulation [ 13,141. In the following,
experimental results of magnetic fields detection in the range of 0-200 kA/m are
reported.
2 Experimental set-up
All reported tests have been performed on a sample of 3 x 5 ~ 2 0mm3 stick of
MSMA in martensitic state. The elasto-magnetic characteristics of such types of
materials are reported in [9] with a threshold field value around 250 W m and a
saturation value of 500 W m . After a suitable bonding protocol choice, a FBG
has been bonded with cyanoacrilate glue oriented along the principal axis of the
transducer. In addition, in order to obtain the reversibility of the magnetic -field-
induced strain, the material was pre-stressed by springs or in other way [9-111.
Fig. I . A schematic diagram of the set up

-
Comtxessive macbne
1 50nmfib.er
r H 0 h a e e eratine
I H
H

e
1 Maenetic fieldH
- PC
- FASM samnle
FBG inte ocation
&ern

In this case, the sample was placed in a tension-compression testing system

(Lloyd Instruments LRXPlus 5kN) between two aluminium rods. The
compressive stress was applied along the [ 1001 crystallographic direction while
the magnetic field was successively applied perpendicular to the stress using a
custom electromagnet opportunely fed. With this actual configuration, however,
a maximum magnetic field approaching the threshold value can be obtained. So
the performances retrieved are referred to the worst case. Improved
performances would be available if higher magnetic fields are involved in.
Magnetization was measured in the direction of the applied magnetic field using
a Hall effect based gauss-meter (Lakeshore 475 DSP), while the strain induced
on the fiber optic sensor was measured by using the technique based on the
 149

optical filtering combined with broadband interrogation described in [13,14]. In

particular, the optoelectronic setup used for the proposed magnetic sensor relies
on the interrogation optic sensor system described in [13,14], with an optic filter
with a nominal linearity range of lOnm (centered at 1550nm), a super
luminescent diode with 40nm FWHM bandwidth (at 1550nm) and a FBG with a
central wavelength, I,, free at 1548 nm and a 0.736 nm FWHM bandwidth.
Previously, a proper calibration procedure has been carried out, , in order to
obtain the FBG response expressed directly in terms of wavelength shift. The
schematic diagram of the whole set up described above is represented in Figure
1.
3 Results and discussion
3.1 Magnetic field sensor and sensitivity

In order to characterize the sensing system, several tests have been

performed. The magneto-elastic active material's response also depends on the
compressive stress applied to it. Infact, the higher is the level of pre-stress and
the lower is the occurred strain. So several tests with different pre-stress levels
have been performed. For each one, a sinusoidal magnetic field with amplitude
around the threshold value for MSM sample has been applied to the structure in
order to explore the range [0-3001 W m .

."...
?MI .. .
. .. .. . _ _ _ _...... . -

" . ~ C * . + ^ ~

Trme [tee1 --I

(a) (b)

Fig. 2. (a) Gaussmeter signal @) Fiber Bragg wavelength shij?

Related Gauss-meter and FBG responses have been acquired using a data
acquisition (DAQ) system. Here, results for a 15N of pre-stress for which the
best results have been obtained limited by the limited magnetic field range used,
are showed.
In Figure 2 are reported the time history of the gaussmeter and the FBG response
expressed directly as wavelength shift. It is worth to note from FBG response
that the MSM sample acts always in a tensile stress state. Even though, before
magnetic field applications, the FBG was initially in a compressive state (due to
150

the 15 N pre-stress) this state has been chosen as baseline. Indeed, the curve in
Figure 2 (b) starts from the point f i g = 0 and the first path represents the first
magnetization curve for the material.
In Figure 3. the sensor characteristics has been reported. From the figure, it can
be noticed that the sensor response results linear almost up to 100 kA/m and a
sensitivity (AhB/AB)/(AH/Hm)of 1.81* lo4 for applied fields smaller than 100 W m
has been estimated. Here, H, is the mean magnetic field in the considered
range.The minimum detectable value for AH/Hm, under these conditions, is
3.56*102.

Fig.3. Magnetic sensor characteristic Fig. 4. System strain field characteristic

3.2 System characterization

Furthermore, the same setup has been used for the whole system
characterization. In Figure 4, the system strain-field characteristic is shown.
The curve starts from the point AhB =0, H=O and the first path represents the first
magnetization curve for the material, then follows the typical butterfly closed-
loop shape of a hysteretic material. It is worth noting that the first path is outside
of the closed-loop because the latter is a minor-loop of the limit-cycle.
4 Comparison with Terfenol-based sensor
Using the same interrogation sensor system exploited for MSM
characterization, tests have been also carried out in the case of Terfenol D rods.
Also Terfenol needs a pre-stress and in this case, a maximum magnetic field of
approximatively 12 kA/m was applied. In fact Terfenol works correctly for
lower magnetic field than MSM achieving a saturating response for few tens of
W m . Several tests have been performed with different level of pre-stress even
if the maximum strain was obtained and showed for a level of pre-stress of 43N.
A sensitivity (AABlhB)/(AWHm)of 5.1* 10 has been estimated. The results are in
agreement with data previously reported in literature [S]. Comparing the
obtained characteristics and relative sensitivities of MSM and Terfenol based
sensors, its worth to note that similar sensitivity can be achieved by using
MSMA magnetic materials but with a significantly extended dynamic range.
Moreover it can be seen that forcing higher magnetic fields, a great improvement
 151

in the magneto-striction coefficient and thus in the sensor sensitivity can be

achieved.
5 Conclusion
The exploitation of a FBG magnetic field sensor has been demonstrated. The
FBG has been bonded to a pre-stressed MSM sample. The interrogation
optoelectronic system used for the proposed magnetic sensor relies on a low cost
ratiometric technique based on optical filtering combined with broadband
interrogation. With this set up, a sensitivity (Ahs/aB)/(AH/Hm) of approx. 2* has
been achieved even if MSM material has not been exploited up its real
possibilities. Comparing results obtained using also Terfenol D, it appears that
despite working in the worst case for MSM with magnetic field values lower
than 300 W m , MSM and Terfenol have similar performances. But increasing
magnetic field values the situation change overworking for MSM a
magnetostriction coefficient two order of magnitude higher than Terfenol one.
Therefore, in order to further increase the MSM sensitivity and exploit all the
linear range, a development of magnetic set up with magnetic field up to
5OOWm has been planned.
References
[ 1J J. E. Lenz, Proc. of the IEEE, Vol 78, No 6, p. 973 - 989, 1990.
[2] T. Bosselmann, Proc. SPIE Vol. 5855 pp 188-193,2005
[3] K. Bohnert et al. Proc. of SPIE Vol. 5855 pp 210-213,2205
[4] T. Wang, et al., IEEE Trans. on Instr. and Meas., vol. 50, no. 3, 2001.
[5] B. Culshaw and J. Dakin, Artech House inc., Nonvood, 1997.
[6] J. Mora et al. IEEE Phot. Techn. Letters, Vol. 12, No. 12, pp. 1680 - 1682,2000.
[7] D. Satpathi et al. IEEE Sens. Joum., Vol. 5, No. 5 Oct. 2005 pp. 1057-1065
[8] M.Pasquale, Sens. and Act., 2003 pp 142-148
[9] J. Tellinen et al., 8th Intern. Conf. ACTUATOR02, Bremen, Germany, 2002.
[lo] 0. Heczko, Journ. of Magnetism and Magnetic Materials (2005) 787-794
[ 1 I] A. A. Likhachev et al. Mat. Science and Eng. 378 (2004) 513-5 18
[ 121 0. Heczko, L. Straka, Materials Science. and Eng. A 378 (2004) 394-398
[ 131 A.Cusano et al. Sensors and Actuators, 110 (1-3) 276-28 1,2003
[I41 A. Cusano et al. Optical Engineering, (accepted)
 A COMPUTATIONALLY EFFICIENT MODEL FOR THERMAL
SENSORS SIMULATION

F. MANCARELLA, A. RONCAGLIA, M. PASSINI,

G. C. CARDINALI, M. SEVEN
IMMBologna, CNR, Via Gobetti 101
Bologna, I-40129, Italy

Thermal sensors based on silicon micromachining have many applications in thc MEMS
field. In order to dcsign such dcvices for an optimized performance a flcxible simulation
model is needed. We propose here a computationally efficient modcling technique for
computer aided dcsign of micro thermal sensors.

1. Introduction
Realizing thermal microsystems, such as micro hotplates, bolometers,
thermopiles and so on, typically requires several design and fabrication cycles
until the specifications are fulfilled. An accurate thermal model, that permits to
estimate the temperature distribution on the structures, could appreciably reduce
the fabrication and test time, and, consequently, the overall realization cost. On
the other hand, thermal simulations of complex three-dimensional devices are
usually very time-consuming. However, for devices realized on thin membranes,
it is possible to introduce proper approximations that permit to treat the problem
by a two-dimensional PDE model.
In the following, this modeling approach is described and validated on
measurements performed on polysilicon-based thermopiles, realized on a thin
dielectric suspended membranes.

2. PDEmodel
For thin-film microstructures, like those presented in this work, the ratio of
lateral extent to thickness exceeds unity by at least one order of magnitude.
Hence, temperature gradients perpendicular to the membrane plane can be
neglected resulting in a two-dimensional thermal model.
The two-dimensional thermal model adopted is the following:

1
(T - Te) + (T - Te) - k(T - Te) - 4oT3( E , + E , )(T - Te) + -Po =0 (11
ax2 a Ad Ad

This work is supported by Netgas project- EC V framework program, proposal IST-2001-37802

152
 153

where T(x,y) is the unknown temperature distribution on the device, T, the room
temperature, h the heat transfer coeMicient through air, E, and EI the emissivities
of the upper and lower faces of the device respectively, 0 the Stephan-
Boltzmann constant, Po the heating power surface density, d the thickness of the
membrane and h its average thermal conductivity calculated as:

being h, the thermal conductivity of the s-th membrane component layer and I,
its thickness.

3. Finite element simulations and results

For simplified geometries, Eq. (1) can be analytically solved, however for the
cases in which the analytical method fails due to more complicated geometries,
like for instance the device reported in Figure. 1, a finite element (FEM)
simulation is needed. To this purpose, the simple model expressed by Eq. (1)
has been implemented in a commercial FEM simulator (FemLab [l]).
In order to validate the 2-d model, the simulated results have been
compared with the electrical measurements performed on infrared detectors,
realized by a CMOS-compatible technology [2]. In these devices,

I'
x
A

Thermal T = Tc = 300 K insulator

insulator
Figure 0. Thermopile layout for Finite Element Simulation

thermocouples fabricated with heavily doped polysilicon and aluminum thin

films are employed, disposed on a dielectric suspended membrane for thermal
isolation from the substrate. The device response has been measured at the
wafer-level by emulating the incoming infrared radiation with a polysilicon
heater placed on the membrane center.
154

The thermal PDE problem expressed by Eq. 1 has been solved together
with a model of the electrical field and current (Laplace equation with ohmic
electrical transport). In doing this, the two differential equations turn out to be
coupled by means of the polysilicon resistivity temperature dependence p(T),
and by the heat source Po generated on the polysilicon heater by Joule effect,
expressed as

where VV is the electrical potential gradient in the heater structure, p the

temperature coefficient of the polysilicon resistance (TCR), po the resistivity at
the room temperature T, and d the membrane thickness .
As far as the model boundary conditions are concerned, it has been assumed that
the cold junctions of the thermocouples are at the same constant temperature of
the silicon bulk T,. The effect of the air surrounding the 2-d boundary of the
thermopile has been neglected. This assumption is reasonable, since air thermal
conductivity is at least one order of magnitude lower with respect to that of the
other materials. Therefore, the device boundary is supposed to be thermally
isolated where in contact with air.

Table 1 . Thermal, electrical and geometrical parameters used in the FEM simulations

The electrical, thermal and geometrical parameters used in the simulations

are reported in Table 1. All have been measured using specific test structures on
 155

the samples employed in the experiments, except the convective coefficient h,

treated as a fitting parameter.
I -

--

Figure 2. Double-polysilicon thermopile: thermocouple cross-section (left), thermopile layout

(center) and estimated temperature distribution (right).

Figure 3. Ah-poly thermopile: thermocouple cross-section (left), thermopile layout (center) and
estimated temperature distribution (right).

The value of h was estimated by matching simulation and measurement

results on a double-polysilicon thermopile, chosen as a reference device (see
Figure 2). Subsequently, the extracted h value has been used to simulate other
devices with different kind of thermoelements (Figure 3 in example), but having
the same membrane geometry and, consequently, the same features concerning
heat exchange through air.
In these simulations, the radiative heat loss has been linearized, since the
peak temperature difference typically reached on the membrane center during
both the test and the normal operation of the devices is a few hundreds mK
above the room temperature, It is rather clear that this approximation is not
correct if a high temperature device (such as a hotplate) has to be simulated.
In Figures 2, 3 the simulated thermopile layouts and the temperature
distribution obtained on the membrane using the proposed model may be
observed. The best match between the measured and simulated results for the
device in Figure 2 has been obtained for a value the convective coefficient h
equal to 325 W/m2K. By using this value, we also calculated the temperature
156

Double polysilicon thermopile Al/n-poly thermopile

8 1.4
- Simulation -Simulation
h
* Measurement
I .2
* Measurement

5 =, sE ox
-
E1
>= .i

L
I

1, .zu .in 60 80 IrJll tm iw o 20 JO 60 EO IOD ita 14u

Ph*ltes (pw) Phca,, (PW

Figure 4.Comparison between measured and simulated responsivity for a double polysilicon
(left) and Al/poly (right) thermopile.

distribution and the responsivity of an Al/polysilicon thermopile (Figure 3). The

comparison of simulation and measurements for the latter case yielded the result
reported in Figure 4 (right side), in which the good agreement observed seems
to confirm the appropriateness of both the model and the parameters employed.

4. Conclusions
A simulation model useful for designing thermal microsensors based on thin
membranes has been proposed and validated with measurements performed on
micromachined IR detectors. A good agreement between measurements and
simulations has been achieved on different device types using a unique value of
the fit parameter employed to describe the heat exchange through air.

References

1. http://wwwfemlab.com
2. A. Roncaglia, F. Mancarella, G. C. Cardinali, F. Tamarri, M. Severi,
Eurosensors XZXProceedings, Barcelona, Spain, Sept. 11-14, p. MC13, (2005).
Characterization of a Point-Wise Close Electric Field Sampling
System exploiting the Electro-Optic Effect

Lucio Rossi, G. Breglio, A. Irace, P. Spirito

Dept. of Electronic Engineering and Telecommunications, University Federico
II, Napoli, Italy
I.rossi @unina.it

Summary
The linear (Pockels) electro-optic effect that many bi-refringent crystals show can be
exploited to realize sensing devices that investigate the intensity and shape of the electric
field close to integrated circuits and devices, Non-invasiveness, wide bandwidth, linearity
and small spatial resolution are the main features that a sensor based on such a principle can
achieve. In this work we aim to give a first characterization of a step-by-step electro-optical
probe which measures the variation of light polarization induced by a lithium niobate
crystal (realized by SELEX S.I.)immersed in the electric field provided by a high frequency
Integrated Circuit.

1 Introduction
The modern design of RF and Microwave devices is constantly driven by the
pushing need of smaller dimensions, light weight and reduced power
consumption, but with the shrinking dimensions the devices testing becomes an
issue as the design feedback itself. To overcome these difficulties and
considering that modern day automatic network analyzers are, if not bandwidth
limited, at least lacking flexibility since measurement is restricted to some
particular ports only, a near electric field point-wise measurement system with
peculiar features of non-invasiveness and wide bandwidth is proposed.
This system is based on the Electro-Optic effect which, being caused by a
distortion in the electrons motion inside the atoms or molecules of a substance, is
very fast a can be used for sensing purposes in the GHz bandwidth. Another
advantage beyond doubt is the non-invasiveness of the system due to the fact that
the crystals (or the polymers) used have a low permittivity and interfere as little
as possible with the field to be measured.

2 Theory behind the instrument: The Electro-OpticEffect

Some anisotropic crystals which are lacking a centre of symmetry exhibit the
linear electro-optic effect which can be explained as a modification either in the
form or the orientation of the optic indicatrix when an electric field is applied

157
158

across them. When this happens, some additional terms appear in the index
ellipsoid equation:

The Pockels (linear) electro-optic effect modifies the value of the coefficients in
the above equation. An electro-optic tensor - (6x3) can be introduced to relate
each one of the coefficients with the applied electric field:

3
Ai=x<jEj Ai=A
j=1

Consider for example the LiNb03 crystal in the conventional reference system.
The values of the non-zero elements in the electro-optic tensor are given in Table
2.1

r12 8.6

Table 2.1 Electro-optic coefficients for the LiNbO3 crystal. LF and H F stand for low and high
frequency respectively

If we have a reference system in which x is orthogonal to m and apply the electric

field along the y axis the index ellipsoid changes as follows:

This is the so-called transverse configuration which can be exploited for

modulation and with some additional element for sensing purposes.
To understand how this is possible, consider the light propagating along the z
direction, the indexes of refractions of the two allowed polarizations after
developing in series up to the linear term are:
 159

= n o +ynor22Ey
1 3

(2.4)

[;- .)i
1

n , = ,+r 22 E = n o - -1n r03 22 E y

This means that the two polarizations of a wave propagating along the z axis
undergo the following total phase difference:

Inserting two polarizing plates one before and the other after the crystal and 90"
rotated with respect to the other, the light intensity at the output is the one
showed in Figure 2.1

',
,
' j
I
4
...,
.
, ,
j
- , <
, .' ; !
i ' ~

I/<" '
,

, '. . i
V/Z
, , . . i
Figure 2. I Modulation with the electro-optic effect

For sensing purposes this characteristic has to be moved left in order to obtain to
have the linear region around the 0 V point. This can be performed using a
couple of retarding plates as showed in the next section.

3 The Optical Chain

The Optical chain is showed in Figure 3.1. An 835 nm pulsed laser goes through
a polarizing beam splitter to obtain a linear polarization and probe the input
power. Thanks to two retarding plates we gain two degrees of freedom to adapt
160

the polarization of the light to the crystal features in order to obtain maximum
linearity and signal swing from the electro-optic effect.

Figure 3.1 The optical chain

The beam then crosses the electro-optic crystal and is focused on top of the DUT
where is reflected back in the optical chain. The light finally reaches the beam
splitter again, where the component reflected by 90" is probed by a second
photodiode PD2.

Figure 3.2 Input/Outpu~characteristic of the optical chain

The effect of inserting the two retarding plates can be calculated using the Jones
formalism and is showed in Figure 3.2.
As it is possible to notice from Figure 3.2 not only the I n p u ~ O u ~characteristic
ut
is now centered around the 0 V point, but the periodicity is also reduced by a
factor two which results in a steeper slope in the linear range so that a better
sensitivity is achieved.
 161

4 Measurments

/
/,
I
/
/
/I

,/

Figure 4.1 Measured I/O characteristic with a Q-switching Pockels cell and small signal
characteristic of x-cut 2 x 2 ~ 1
mm crystals

The first lithium niobate crystal that was used inside the optical chain was a
Pockels cell meant for laser Q-switching (C1041 Dorotek) with considerable
dimensions. The static characteristic has been measured in this case and it is
showed in figure 4.1. A Vx of around 1.2 kV was measured and it is in good
agreement with the theory. The Vx periodicity is also in agreement with the fact
that light crosses the LiNb03 crystal two times.
Some other smaller ( 2 x 2 ~ 1mm) crystals have (realized by Selex S.I.) have been
characterized in this system. They were x-cut and their small signal behavior in
the range of 0-40 Vpp applied to the crystal at a frequency of 26 kHz was
measured. In order to retrieve the noisy signal a Lock-in amplifier (Stanford
Research SR844) tuned at the frequency of the laser pulses was used. The results
of these measurements are showed in Fig. 4.1.

REFERENCES
[I]. R. Hofmann, H.-J. Pfleiderer Electro-optic Sampling System for the Testing of High-
speed Integrated Circuits Using a Free Running Solid-state Laser JOURNAL OF
LIGHTWAVE TECHNOLOGY, VOL. 14, NO. 8, 1788-1793 (AUGUST 1996)
[2]. J. Allam, C. L. Yuca and I. R. A. Cleaver Quantitative electro-optic sampling for high-
speed charactenzation of passive and active devices 0-7803-5298-X/99 IEEE (1999)
[3]. K. S. Giboney, S. T. Allen, M. J. Rodwell, J. E. Bowers Picosecond Measurements by
Free Running Electro-Optic Sampling IEEE Photonics Technology Letters Vol. 6, No.
11, (Nov. 1994)
[4]. G. R. Lin, Y. ChangA Novel Electro-Optic Sampling system by Using Delay-Time
Tunable Pulsed Laser Sources 0-7803-5947-W00/$10.0002000 IEEE
[5]. R A Dudley, A G Roddie, D J Bannister ELECTRO-OPTIC PROBING OF
MICROWAVE CIRCUlTS TH3A-2
 On The Realization Of Chromium Silicide Stress Sensor

D. Caputo, G. de Cesare, A. Nascetti, R. Scipinotti

Department of Electronic Engineering, University of Rome La Sapienza ,
via Eudossiana, I 8 00184 Rome (Italy)

Summary
In this paper, a very thin and high conductivity chromium silicide film, formed on a n-type
amorphous silicon layer, is utilized as active material in a stress sensor. The sensor can be
deposited directly on the stressed materials without additional packaging, avoiding the use
of the adhesive resin which usually affects the device performances.
Details of fabrication and characterization of sensors grown on different kind of substrates
(glass, metal, ceramic and plastic) suitable for mechanical stress and pressure measurement
are reported.

1 Introduction
Mechanical stress is an important physical parameter in various areas of
application, such as industrial automation, mechanics, automotive and medical
engineering [1][2]. In particular, strain gauges are used in a number of advanced
applications such as non destructive manipulation of the objects with robotic
hands [3].
The common approach to stress measurement is essentially based on the use
of piezoresistors realized in different technologies. This approach has different
limitations: large area, small sensitivity and high temperature drift. On the other
hand the use of silicon thin film technology at low deposition temperature (<150
C) using novel substrates, like plastic, has recently received great attention due
the new large area applications.
Here, we present a thin film stress sensor where the active material is a very
thin and high conductivity chromium silicide film formed on a n-type amorphous
silicon layer. An important advantage of our device with respect to the existing
approaches is due to the low temperature fabrication process, which can be
lower than 100C for all the technological steps. Furthermore, the sensor
element can be deposited directly on the stressed materials without additional
packaging, avoiding the use of the adhesive resin which usually affects the
device performances. The presented sensor has been used in two different
applications: to measure a force applied on glass or metal substrates and a
vacuum pressure induced on ceramic or plastic membranes.

162
 163

rica~onand o~eration
Figure 1 shows the structure of the sensor element.

Fig. I: Structure of the s&ss sensor

The first step of the sensor fabrication process is the deposition of a 500 nm
n-type amorphous silicon film by PECVD in a high vacuum deposition system,
with the following recipe: gas flow of pure silane 40 sccm; silane diluited
phosphine 10 sccm; pressure 300 mTorr; power density 40 mW/cm2; deposition
temperature 100 "C.
A three-layer chromium-aluminum-chromium (151300115 nm) is then
vacuum evaporated and patterned by photolithographic process and chemical
etching, in order to form the metal contacts. At the interface between amorphous
silicon and chromium, a chromium silicide (CrSi) layer is formed.
The CrSi acts as active layer by shunting the resistance of the amorphous
layer in the sensor structure. In fact its electrical conductivity is about two orders
of magnitude higher than the n-doped a-Si layer ( 2 . 5 ~ 1 0CX'cm-')
-~ [4].
Finally, a reactive ion etching (RIE) has been performed, in an
02( 1Osccm)/CF4(100 sccm) gas mixture, to define the active area of the sensor.
The presented sensor has been deposited on different substrates: glass, metal,
ceramic and plastic and has been utilized for force and pressure measurements.
From an electrical point of view, the sensor can be considered as a simple
bridge of resistances between four metal contacts, where the resistances are due
to the conductivity of the amorphous siliconichromium silicide stacked films. In
the operating condition, two contacts of the bridge are used to apply the bias
current to the sensing element, while the other two, orthogonal to the previous
ones, provide an output voltage proportional to the anisotropic modification of
the resistivity induced by the mechanical deformation. The output voltage of the
sensor has been monitored by a very high input impedance CMOS amplifier
(LMC6032 National Semiconductor),with a gain equal to 100.
164

3 Force measurement
In order to generate shear stresses the substrate on which the sensor is
fabricated has been clamped at one side and a vertical force has been applied to
other side. Figure 2 reports the output voltage of two sensors, fabricated on
Corning Glass 7059, with different orientation of the sensor contact with respect
to the direction of applied stress 0,.As expected, the slope of the output voltage
vs the bending momentum of the sensor A is much lower of the output voltage
o f sensor 3, whose bridge resistances are rotated by +45" with respect to the
applied strain (see inset in Figure 2). Sensor B shows a sensitivity around
0.4V/(Nm). For both devices we observe an excellent linearity in the
investigated range. This property is due to the deposition of the device directly
on the substrate, since the linearity in the sensor response is usually affected by
the adhesive resin used to fix the sensor element [5].
For the sensor grown on the metal substrate, we deposited a 500 nm thick
silicon nitride film as isolation layer between metal and amorphous silicon layer,
because the very low transverse resistivity of the doped amorphous silicon
directly connects the electrodes with the substrate. Two different substrate
thicknesses (0.8 mm and 1.5 mm) have been used. Equal applied forces
determine different deformations of the two substrates. Results are reported in
Figure 3, where as expected we observe a much higher sensitivity of the sensor
grown on the thinnest substrate. In particular, in agreement with theory, we
found that the ratio of the two slopes is very close to the ratio of the cube of the
thicknesses of the two substrates and equal to 1.1 and 8 pV/g for the 1.5 and 0.8
mm substrates, respectively.

Fig. 2: Output voE?age vs the bending Fig. 3: Output voltage of the sensor vs
momentum of two sensors, with weigh?applied on steel substrates with
dferen? orientation of the contacls dferen? ihicknesses.
with respect to ?heappfiedstress.
 165

ent
Two sensors (1 mm area) have been fabricated on a ceramic membrane,
with 9 mtn diameter and 0.6 mtn thickness. One sensor is close to the center, the
second is positioned at half radius of the diaphragm (see inset in Figure 4). The
ceramic membrane has been positioned as a cap of a volume that has been
emptied by a rotary vacuum pump. The membrane deformation, proportional to
the pressure, has been monitored measuring the sensor output voltages, which
are reported in Figure 4. The gains of the two sensors are equal to 14.5mVlmBar
and 3.75mV/mBar. The different slopes are, obviously, ascribed to the different
deformations of the membrane at the sensor positions.
As a final application as pressure sensor, we report in Figure 5 the output
voltage of a sensor (1 mm area) fabricated on Kapton tape. As shown in the
inset, the tape has been attached to a glass substrate to avoid plastic deformation
during fabrication steps. We observe that, the output voltage of the sensor shows
good linearity as a fimction of vacuum pressure in the investigated range.

Fig. 4: Output voltage vspressure of two sensors, with dzfferentposition on the ceramic
membrane
166

Fig. 5: output voltage vs pressure of a sensor deposited on Kapton membrane,

~o~clusio~
Different applications of a thin film stress sensor have been presented. The
active layer is a very thin silicide chromium. layer grown on a n-type a-Si:H
region. Details of fabrication and characterization results on different substrates
(glass, steel and ceramic) have been reported, showing the suitability of our
sensor for both force and pressure measurements.

eferences
[l] A. Bicchi, J.K. Salisbury, D. L. Brock; A. I. Memo No. 1262, MIT
Cambridge, MA, Oct 1990.
[2] A. Cicchetti, A. Eusebi, C. Melchiorri, G. Vassurra; 7th Int. Conf. On Adv,
Robotic, ICAR95, Spain, Sept, 1995.
131 J. Butterfass, M. Grebenstein, H. Liu, G. Hirzinger; ICRAO1, IEEE Int.
Conf. On Robotic and Automation, Korea, May 2001.
(41 R.A. Street, in Hydrogenated amorphous silicon Cambridge University
Press, (1991)
[5] L. Biagiotti, M. Gavesi, C. Melchiorri and B. Riccti; Proceedings of IEEE
International Conference on Robotics and Automation 2 (2002) 1655.
Nanogravimetric Immunodetection of Gliadin by pLibra Balance

F. Olimpico, M. Pighini, S. Bernardi, A. Scarpa and S. Greco,

Technobiochip Scrl, Via Carrara 12/A 04100 Latina (Italy)
[email protected]

Summary
A nanogravimetric piezoelectric crystal-based immunoassay for the detection of gliadin is
here described. Gliadin is the ethanol-soluble fraction of gluten, the protein content in
wheat, barley, rye and oats, and it is responsible for gluten intolerance called Coeliac
Disease, an autoimmune disease, characterised by damage to all or part of the villi lining the
small intestine. The method for gliadin detection utilized a nanogravimetric balance
(Technobiochips FLibra), where the piezoelectnc crystal quartzes were functionalized with
antibodies specific to gliadin. We functionalized the quartzes surface by using different
strategies, alone or in combination (e.g y-aminopropyltriethoxysilane (APTES)+3,3 -
Dithiodipropionic acid di (N-succinimide ester) (DTSP), Poly-L-Lysine or Protein A).
Moreover, we have compared the binding of two anti-gliadin antibodies, Sigma and PN3,
against both standard IRMM or Sigma purified gliadin APTES+DTSP and Poly-L-Lysine
treatments were the best functionalization methods for Sigma or PN3 antibody binding,
respectively. In conclusion, obtained data revealed that the nanogravimetric method is able
to detect gliadin with a good sensitivity, so providing a useful test for the presence of toxic
gluten.

Introduction
Coeliac disease (CD) is an inflammatory disease of the upper small
intestine, characterised by villous atrophy resulting from gluten ingestion in
genetically susceptible individuals, where association with particular HLA-DR and
DQ types is well established [1,2]. Gluten represents the cereal storage proteins
found in wheat, barley rye and oats. These proteins fall into four groups: albumins,
globulins, glutenins and prolamins. Wheat, barley rye and oats prolamins (known
respectively as gliadins, hordeins, secalins and avenins) represent the alcohol-soluble
fraction of storage proteins and are responsible for triggering the disease. Gliadins
may be subdivided according to their relative electrophoretic mobility into
a,b, yand w subfractions [3,4]. Interaction of these proteins with the intestinal
mucosa leads to permanent mucosal damage and causes an immune reaction in
several tissues and organs, leading to a large number of complications, such as
anaemia, osteoporosis, infertility, neurological disorders and skin disorders [5].
WHOFA0 CODEX ALIMENTANUS has not yet produced a guidelines
concerning the definitive acceptable levels and methods for gluten measurement, due

167
168

to the current lack of a golden standard method for the detection of CD-toxic gluten
suitable for all types and states of foods.
This study was aimed at the design and production of a immuno-
nanogravimetric sensor. The nanogravimetric method is based on piezoelectric quartz
device, which offers most of the properties required for an ideal immunosensor in a
wide range of applications, such as in clinical diagnostics, food quality control and
environmental monitoring. The potential of piezoelectric devices is based on the
relationship between mass deposited on quartz surface and the resonant frequency
changes accordingly to the Sauerbrey and Kanazawas formulas [6,7]. Detection of
purified gliadin has been achieved by immobilizing anti-gliadin antibodies onto gold
electrodes of quartz crystals surface. We functionalized the quartzes surface by using
different strategies, alone or in combination, (e.g. APTES+DTSP,Poly-L-Lysine or
Protein A). Moreover, we have compared the binding of both two anti-gliadin
antibodies, Sigma and PN3, and standard IRMM or Sigma purified Gliadin from
wheat.

Materials and Methods

Reagents
Mouse Polyclonal anti-Gliadin Antibody (Ab) was obtained from SIGMA
(Milan, Italy) and Mouse Monoclonal Anti-Gliadin PN3 Ab was kindly provided by
KCL (Kings College London). Other reagents were purchased from Sigma (Milan,
Italy). The quartz crystals, 10 MHz AT-cuts, gold electrodes, were obtained from
International Crystal Manufacturing Company (Ohio, USA).
Apparatus
PLibra is a Quartz Crystal Microbalance (QCM) produced by
Technobiochip Scrl (Latina, Italy), composed by P Main Unit and a Cell Base Unit
which contains two oscillators and equipped with two low-volume flow-through cells
for real-time measurements. PLibra output data are elaborated by the LibraVIEW
software.

Fig. I : Quartz Crystul Microbalance jLibru

 169

Experimental protocol
After cleaning with piranha solution (HzSO4:HzO2 7:3 v/v) and
hctionalizations, quartzes were placed in a static flow-cell when basal frequency
oscillation (AF) was stabilizated by buffer adding (Phosphate Buffered Saline, PBS,
lOmM pH 7.4), the antibody solution (SIGMA or PN3 0.1 mg/ml) was flown and the
surface saturated with a BSA solution (0.1 mg/ml). After PBS washing, gliadin
solutions (SIGMA or IRMM 0.1 to 0.05 mg/ml in PBS) were flown and AF of
antigen-antibody interaction recorded. All the experiments were performed as
triplicates.
Strategies of Antibodies Immobilization
Different strategies in order to firmly stabilize antibody onto quartzes
surface were used:
1, Lipoic Acid deposition by Langmuir-Blodgett techniques: APTES quartzes were
functionalized with 10 layers of lipoic acid with a commercially available KSV LB-
5000 instrument. Formed films were compressed at 10 mm/min to produce surface
pressure of 11 mN/m.
2. Traut 's reagent: antibody solution was incubated for 1 hour with Traut's reagent,
which reacts with primary amines to introduce sulfhydryl groups. After purification
using a Sephadex 25G column, the fractions showing maximum absorbance at 280
nm have been pooled and the free sulfhydryl groups produced have been titrated by
Ellman's method.
3. APTES + DTSP-SAM (self-assembled monolayer): quartzes were dipped in 2%
APTES solution for 20 minutes, baked at 12OoC, sonicated in a water bath for 2
minutes and finally washed with ultra pure water. After silanization, they were
dipped into 10 mM DTSP in DMSO and incubated for 30 minutes at room
temperature and then washed with distilled water.
4. Poly-L-lysine+DTSP-SAM: quartzes were dipped into a poly-L-Lysine solution
(poly-L-lysine:H20 1:l v/v) for 15 minutes, placed in oven at 60C for 20 minutes
and air dried. Thus, modified quartzes were dipped into a 10 mM DTSP solution in
DMSO for 30 minutes and washed with distilled water.
5. APTES) +DTSP-SAM+ProteinA: APTES-DTSP-modified quartzes were placed in
a static flow-cell and incubated with Protein A (Img/ml).

Results and Discussion

Several approaches for achieving firmly andor oriented antibody
immobilization have been studied and applied to piezoelectric immunosensor
development.
First approach used was a Langmuir-Blodgett deposition. This technique is
useful for the preparation of solid-supported highly organized films assembled at the
170

liquid-gas interface, and enables the precise control of the monolayer thickness. A
very low oscillation frequency decrease of 1812.2 Hz was observed subsequently to
the interaction between gliadin and Ab bound to ten layers of lipoic acid deposited
by LB technique (data not shown). Secondly, an oscillation frequency decrease of
2513.2 Hz was observed after interaction with thiolated antibody (data not shown).
The very low antigen binding results previously obtained, aimed us to perform SAM
covalent strategies.

SIGMA nb PM3 Ab

&PIES t DTSP Pdy-L-Lysine t APES t DTSP t APTES t DTSP Poly-L-Lysine t APTES t DTSP t
DTSP ProteinA DTSP Protein A

Fig. 2 Means of dijjierence of a s c i l ~ a t i o n ~ e q u e(N)

n ~ a$er SIGMA (A) or PN3 (B) anti-gliadin Ab
binding to quartz by using SIGMA (purple bars) or IRMM (red bars) gliadins are represented as
histograms. Experimeim were pei$ornred in triplicates and standard deviations shown.

As shown in Figure 2, an overall high affinity of SIGMA antibody for

gliadin (Sigma>>>IRMM) than the PN3 antibody, was observed, and among the
different immobilization methods, a better binding of SIGMA antibody was observed
on the APTES+DTSP-modified quartz surface rather than on the Poly-L-lysine one.
Conversely, PN3 antibody showed a greater affinity for the Poly-L-lysine modified
quartz surface rather than for the other two methods.
These preliminary experiments have shown that nanogravimetric sensor
might be useful for the detection of gliadin in foods and SAM strategies are the best
methods to bind the antibody onto crystal surface.

~~~nowledg~ents
The present study is part of the CD-Chef FP5 EC project, which regards the
development of a disposable microsystem with integrated optimal extraction and
 171

detection for precise and accurate quantitation of coeliac disease toxic gluten in all
types of foods.

References
[ l ] Sollid LM, Markussen G, Ek J, Gjerde H, Vartdal F, Thorsby E. Evidence for a
primary association of celiac disease to a particular HLA-DQ alphaheta heterodimer.
JExp Med, 1989; 169:345-350.
[2] Lundin KE, Scott H, Hansen T, Paulsen G, Halstensen TS, Fausa 0. Gliadin-
specific, HLA-DQ(a1pha 1*0501,beta 1*0201) restricted T cells isolate d from the
small intestinal mucosa of celiac disease patients. JExp Mid, 1993; 178: 187-196
[3] Ejllnckd AJC. N-terminal amino acid sequencing of prolamins of wheat and
related species. Nature, 1979; 282527-529.
[4] Peter R Shewry ASTDDK. Cereal proteins and coeliac disease. In: Michael N
Marsh, Coeliac Disease. Oxford: Blackwell Scientific Publications, 1992; 305-348.
[5] Van de Kamer JH WHDW. Coeliac Disease IV. An investigation into the
injurious constituents of wheat in connection with their action on patients with
coeliac disease. Acta Paediatr, 1953; 42:223-23 1.
[6] Sauerbrey GZ. Use of quartz crystal vibrator for weighting thin films on a
microbalance. 2 Phys 1959; 155:20&222.
[7] Kanazawa KK, Gordon JG. Frequency of a quartz microbalance in contact with
liquid. Anal Chem 1985; 57:1770-1771.
 INTEGRATED SILICON MICRO FLOW CYTOMETER
BASED ON HOLLOW ARROW WAVEGUIDES

R. Bernini
ZREA-CNR, Via Diocleziano 328, 80124 Napoli, Italy
bernini .r @irea.cnr.it

E. De Nuccio, A. Minardo, L. Zeni

DII, Seconda Universita di Napoli, Via Roma 29, 81031 Aversa, Ztaly

P. M. Sarro
ECTM-DIMES, TUDelf. NL-2600 GB Delf, The Netherlands

Summary
An integrated silicon micro flow cytometer based on hollow core AntiResonant Reflecting
Optical Waveguide (ARROW) is proposed. The hollow core ARROW permits to confine
both the cells to be analysed and the excitation light to carry out fluorescent measurements.
This solution permits to design the device in order to enhance the pump source/cells
interaction. In this paper, the operation principles, the main fabrication steps and wine
experimental tests are shown.

1 Introduction
The flow cytometer is a device that realises multi-parametric measurements
on single cells or particles [l]. Typically, the cells are forced to travel in line by
the hydrodynamic focussing effect. In this way, the cells arrive at the detection
region, where multi-parametric analysis is carried out by observing both the
scattered light and the emitted fluorescence. The conventional bench top flow
cytometer presents an optical detection instruments arranged in delicate optical
alignment methods and apparatus. Moreover, by using bulk optical device, the
flow cytometer is expensive and not portable instrument[l].
Recently, the integration of optical waveguides with microfluidic devices for
optical detection has been widely investigated. In this line of argument, the
capability of AntiResonant Reflecting Optical Waveguides (ARROWS) to
confine the light in media with low refractive index and their possibility to be
realized with hollow core appear an optimal starting point toward the realization
of the micro flow cytometer. In fact, the hollow core ARROW offers the
possibility to confine both the light and the liquids in the core region. In this way,
it is possible to couple the light directly inside the probe liquid, so as to enhance
light-cells interaction.
In this paper, a integrated micro flow cytometer is proposed. The device is
realised using standard silicon technology and offers the possibility to integrate
the microfluidic part with the microelectronic devices necessary for the detection.

172
 173

In order to validate the correct hehaviour of the device, the hydrodynamic

focussing phenomenon investigation and the fluorescence measurements are
carried out.

2 Principle and design

The scheme of the proposed integrated micro flow cytometer is shown in
Figure l(a). The sample liquid, containing the cells or particle under analysis, is
injected into the central inlet. By tuning opportunely the flow rates of the sheath
liquid, the hydrodynamic focusing effect reduces in diameter the central sample
flow, forcing the cell to travel in line. The marked cells/particles intercept the
pump light source which travels collinearly along the path, so the fluorochromes
are excited to a higher energy state. This energy is released as a photon of light
with specific spectral properties unique to different fluorochromes. The
fluorescent emission light is collected by the two arms placed orthogonally to the
flow, so as to increase the signalfnoise ratio. The collected light is coupled into
the optical fibers in order to he analysed.

Sheath liquid

Fluorescence

Fig.1 Scheme of the proposedjlow cytometer (a).Section of the hollow core ARROW waveguide (b)

The structure is based on integrated hollow core ARROW waveguide (Figure

l(b)), that presents a low refractive index n, core region and two cladding layers
designed to form a high reflectivity Fabry-Perot antiresonant cavity. This solution
permits to confine both pump source and the liquid with the cells under analysis,
so as to reduce in complexity the detection system by using the same channel to
excite and to focus the cells.
In order to minimize the optical losses, the antiresonant condition has to be
achieved. For a fixed core width d, and core refractive index &, in the equivalent
one-dimensional vertical structure, this condition is guaranteed for the cladding
layers thicknesses [Z]:
174

where A is the operating wavelength.

3 Device fabrication
The micro flow cytometer is realised in standard silicon technology by using
two silicon wafers bonded together (Figure l(b)). By photolithography, the
pattern of the structure is impressed on the bottom wafer. The anisotropic dry
etching permits to realize the channels (the depth is set to 150pm and the width is
200pm). The channel dimensions are chosen to permit easily the insertion and
consequently the alignment of both the excitation and collection fibers. The used
materials are silicon dioxide (SiO;?)as second cladding that presents a refractive
index n2=1.4S7 at the wavelength h=633nm, and silicon nitride (Si3N4)as first
cladding with a refractive index n1=2.227. By using the core refractive index
nc=1.333 (water) and dc=lSOpm in eq. 1, the thickness values are dl=266nm for
N=l and d2=266nm for N=O. The interferometric layers are deposed by LPCVD
at 800".After that, the bottom and the upper wafers are joined together by direct
wafer bonding. Finally, the excitation and collection fibers are inserted in the
respective locations and there sealed by optical glue.

4 Results and discussion

The hydrodynamic focusing effect is achieved by tuning the flow rate of the
sheath liquid respect to the sample liquid flow rate. Finite Elements Method
( E M ) has been used to solve the fluidodynamics equations of the structure, with
the aim of finding the optimal ratio between core flow rate and shield flow rate
(m).
In Figure 2, the concentration profile of an aqueous solution in diffusing
water shows the hydrodynamic focussing phenomenon simulated by solving the
Navier-Stokes and Diffusion-Convection coupled equations. The values for the
flow rates at the inlets are SmVh for the central liquid and 15mlh for the focusing
liquids (FRR=3). In this case we have a Flow Rate Ratio, the physical effect is
clear and the calculated width of the focused stream is about 2 0 ~ m .
 175

Fig2 The simulated ~ydrofocussingphenomenon by calculation of the concentration profile of an

aqueous solution in water dijjiusing

An experimental observation of the hydrofocussing effect was conducted by

injecting a fluorescent dye (a water solution of resorufin) into the core of the
structure, while using de-ionized water as the sheath liquid. Results are shown in
Figure 3. In this case, the silicon top wafer was substituted by a thin glass slide in
order to observe the emitted fluorescence by a fluorescence microscope. By
setting FRR=3, with a central liquid flow rate equal to SmVh, the focussed stream
width value is about 24pm, that is the typical dimension of the cell of interest,
finding a good agreement with the numerical simulation.

Fig.3 Experimentai resultsfor the hydrofocusingprocess

The testing of the detection system has been carried out by considering the
device realised by silicon wafer/silicon wafer bonding. The measurements were
performed by using a pump source laser (h=532nm), an optical filter stopping the
wavelengths lower than 532nm, a femtowatt photodiode and an oscilloscope. Due
to the low-pass band behaviour of the photodiode, the absolute velocities of the
liquids have been reduced, so to be reduced the cells occurrence. By setting the
sample flow rate to lml/h and the FRR=3, the results of the test are shown in
Figure 4. The intensity's peaks of the collected emitted fluorescence represent the
176

cell event. It is to underline that at low flow rates, the sample stream becomes
unstable due to the large pressure difference between the sample and sheath flows
near the channel inlets, resulting in discontinuous liquid motion. The
discontinuous sample flow causes particles to become trapped in the inlet
reservoir, which is one of the major reasons for the low cell rate in the capturing
[31.
1 ,
008-

007-
I
m
006-
t=

- 005-
a,
004-

60 70 80 90 100 110 120 130 140

Time [sec]
Fig. 4 Collected fluorescence intensity of the marked cells

In conclusion, a novel micro flow cytometer has been proposed. The device is
based on the hollow core ARROW waveguide, realised in standard silicon
technology. The correct behaviour of the device has been investigated, showing
good results. The technological choices are motivated observing the capability of
the hollow core antiresonant optical waveguides to confine both the sample liquid
and the pump source, and the possibility to integrate microelectronic devices with
standard IC process.

References
[ 11 Gwo-Bin L e a , , Che-Hsin Linb, Guan-Liang Changc, Sensors and
Actuators A, 103, (2003), 165-170.
[2] R. Bernini, S. Campopiano and L. Zeni, IEEE J. Selected Topics in
Quantum Elect., 8, 106, (2002).
[3] Yi-Chung Tung a, Min Zhang a, Chih-Ting Lin b, Katsuo Kurabayashi
a,c, Steven J. Skerlos, Sensors and Actuators B, 98, (2004), pp. 356-367.

Acknowledgments
The authors would like to thank the IC Process Group of DIMES for
technical assistance.
Properties of heme c containing enzyme layer self-assembled on
planar metallic electrodes

S . Kaciulis, A. Mezzi
Istituto per lo Studio dei Materiali Nanostructurati ISMNCNR,
P.O. Box 10. I-0001 6 Monterotondo Stazione ( R W , Italy
mezzi@,mlib.cnr.it

A. Galdikas, V. Bukauskas, A. Mironas, A. Setkus

SerniconductorPhysics Institute, A. Gostauto I I , Vilnius, Lithuania

V. LaurinaviEius, R. MeSkys,J. Razumiene

Institute of Biochemistq Mokslininku 12, Vilnius, Lithuania

Summary
Electrical properties of symmetric metal-enzyme (pyrroloquinoline quinonc-dcpcndcnt and
heme c containing alcohol dehydrogenase) junctions were experimentally invcstigatcd in
planar structures in gaseous surrounding. Two types of the ADH molecules (typc I1 and 111)
have been used in the experiments. Gas sensitive state in the junctions is induccd by
temporal connection of an external dc-voltage to the electrodes. The properties of the
structures, determined by a relationship between the electron transfer and chcmical processcs
in metal-ADH junction, have been analyzed. Alcohol sensitive characteristics and thc XPS
chemical composition of molecular layers, consisting of pure ADH-I1 and ADH-111, have
been compared.

1 Introduction
In the most developments, the biosensors are based on electron transfer from
enzyme to electrode stimulated by enzyme catalyzed biochemical reaction [ 1-31.
Practically applicable devices are frequently developed long before
comprehending their functioning mechanism. Nevertheless, the understanding of
these processes can be valuable for the description of the biosensors principal
limitations and also for its further improvement.
Present study is focused on an original family of the sensors based on dry
enzyme layer that respond to vapours in air. This report presents novel results of
the study focused on the two forms of alcohol dehydrogenase (ADH) and
describes the formation and the properties of a pure molecular layer.

177
178

2 Experimental
Soluble quinohemoprotein ADH type (QH ADH-11) from Pseudornonus
Putida HK 5 and membrane-bound quinohemoprotein ADH type (QH ADH-111)
from Gluconobacter sp 33 were used for dry molecular layers in our study. The
two proteins contain heme c and pyrroloquinoline quinine cofactor (PQQ). The
dry layers with QH ADH-I1 and QH ADH-111 were deposited from individual
colloidal solutions based on tris-HC1 and potassium phosphate buffers,
respectively. A layer has self-assembled on a substrate with planar metallic
electrodes after drying at ambient temperature in air. Planar strip-like electrodes
were shaped from thin films of different pure metals by standard
photolithography on an insulating base-plate (e.g., glass). Pt, Au, Ag and Ni
have been used for the electrodes. The typical gap between the electrodes was of
about 50 pm.
DC-voltage drop was measured on the load resistance connected in series to
the Me/ADWMe structure after the ADH-layers were dried in ambient
atmosphere. During the dc-electrical tests, an original method was used for the
activation of the ADH molecules. The method is analogous to an electrical shock
and was proposed by us recently [4]. For the activation, the planar Me/ADWMe
structures were temporary (100-300 seconds) connected to an external power
source. After the activation, the signal proportional to the current was measured
in the circuit without an external power source. The sequence of the procedures
was cycled for a long time and analogous results were obtained in each cycle.
XPS investigations were carried out by means of an VG ESCALAB M k I1
spectrometer equipped with A1 K a source and five-channeltron detection
system. In order to avoid the X-ray damage, the sample holder during the
measurements was cooled with liquid nitrogen.

3 Results and Discussions

After the activation, the junctions generated an electric signal in response to
alcohol vapour in air. Typical variation of the signal in response to a pulse of
ethanol and 1,2-propanediol is illustrated in Fig. 1. The activation voltage was
U,, = 4.24 V at T = 22 "C and relative humidity of air of 43 %. This response
was recorded in unlimited number of the cycles, each of which consisted of four
periods, namely the activation, waiting, responding and restoring. The sequence
of the measured signals was described in detail in our previous report [4]. It
follows from Fig. 1 that the kinetics of the response signal is individual in the
ADH-I1 (2) and ADH-I11 (1) dry-layers. The difference between the signals is
described by A, and Ab that are a delay between the rises and the falls of
response signals respectively in the two ADH types. Another distinctive
characteristic of the signal is a spike in the response signal of ADH-111 layers
 179

after the clean air was let to flow through the sensor chamber. The origin of the
differences in Fig. 1 is not clear at present state of our investigations.

3.01
2.5

0 50 100 150 200 250

tm sI

Fig. I . Time dependence of the electrical signal measuredfor ADH-111 ( I ) and ADH-11
(2) in response to a pulse of ethanol vapour (4.4 %) in air.

The responses of the Me/ADWMe structures were proportional to the partial

pressure of ethanol vapour in air. The proportionality is illustrated in Fig. 2 by
typical dependences of the response of the ADH-I1 dry-layers (two sensors) on
the amount of 1,2-propanediol(l) and ethanol (2) vapour in air (T,= 21 "C; RH
= 43 %) after the activation with external voltage Ueo= 4.2 V.

000

Fig. 2. Dependences of the response signal ofADH-11 basedstructures to 1,2-

propanediol(1) and ethanol (2).
180

The dependences in Fig. 2 can be described by a power law U, pb with b u -

0.8 (line 1) and 1.5 (line 2). Analogous results were obtained for ADH-I11 layers.
It was ascertained that the response of a dry enzyme layer is dependent on
the residual compounds from the buffer. An influence of these compounds on
the response of the dry enzyme layers is illustrated in Fig. 3. The dependences of
the response on the activation voltage are plotted in the figures. It was found that
the effect of the residual compounds is significant at low Ueo.

Fig. 3 The response at various activation voltages Ue0temporary applied to the ADH-III (1,I 3 and
ADH-II(2.23 layers before (1,2) andajier (I :23 washing: (a) to ethanol vapour (345OOppm); (b)
to I,2-propanediol vapour (I 000ppm).

From the sensitivity tests it was concluded that the selectivity of the pure
ADH based dry layers is higher than those containing residual compounds from
the buffer. The most evident change in the selectivity is shown in Fig. 3b. There
was no response to 1,2-propanediol in pure ADH-I1 dry layers, while the
removal of the residual compounds from ADH-I11 layers increased significantly
the response signal to 1,2-propanediol(see 2' in Fig. 3b).
As it was mentioned in our previous publication [ 5 ] , the enzymes can be
characterized by spectral fingerprints, when they are investigated by X P S .
Therefore, the shape of C 1s signal is typical for ADH-I1 and ADH-I11 dry
layers, composed from the three components, positioned at BE = 285.0, 286.5
and 288.1 eV and assigned to C - C, C - OH and C = 0, C - NO, respectively
(Fig. 4). The differences in the C 1s spectra of two enzymes can be explained by
the presence of buffer residuum and some non-controllable contaminants. When
the buffer residuum was removed by washing the samples in distilled water, the
C 1s spectra of both enzymes became identical. The complete removal of the
residuum from the ADH-I11 layer is demonstrated by the K 2p signal detected in
the as-deposited sample, while it was absent in the washed sample (spectra 1 and
2 in Fig. 4a). At the same time, the C 1s spectrum was unchanged, what indicates
that the buffer removal was performed without the modification of the enzyme's
structure.
 181

In addition, the XPS analysis proved that the metal electrodes are still
completely covered by enzyme layer after rinsing the samples with distilled
water.

Fig. 4. XPSspectra registered before ( I , ] 7 and ajier (2,2 7 the removal of buffer residuum
from ADH-III (a) andADH-II @) layers by washing in distilledwater.

Acknowledgments
The work in the Semiconductor Physics Institute and the Institute of
Biochemistry was carried out in the frame of the joint project BIOHEMAS
supported by the Lithuanian State Science and Studies Foundation.
Dr. A. Mezzi acknowledges the support of the European Center of
Excellence PRAMA.

References
1. W. Schuhmann, Rev. Molecular Biotechnology, 82 (2002), p. 425.
2. M. Aslam, N.K. Chaki, J Sharma, K. Vijayamohanan, Current App. Phys., 3
(2003). p. 115.
3. B. Bhushan, D.R. Tokachichu, M.T. Keener, S.C. Lee, Acta Biomaterialia,
1 (2005), 327.
4. A. Setkus, J. Razumiene, A. Galdikas, V. LaurinaviCius, R. MeSkys, A .
Mironas, Sens. Actuators B: Chem., 95 (2003), 344.
5. A. Curulli, A. CusmB, S. Kaciulis, G. Padeletti, L. Pandolfi, F. Valentini, M.
Viticoli, Surf: Interface Anal., 38 (2006), 478.
 Detection of labelled DNA based on
amorphous silicon devices

D. Caputo, G. de Cesare, A. Nascetti, R. Scipinotti

Department of Electronic Engineering, University of Rome "La Sapienza ", via
Eudossiana, 18 00184 Rome (Italy)

R. Negri, A. Tarquini
Laboratory of Functional Genomics and Proteomics of Model Systems, University of
Rome "La Sapienza", via dei Sardi, 70, Rome (Italy)

Summary
In this paper, we present an alternative fluorochrome labeled DNA detection system based
on the use of an amorphous silicon photosensor. The sensor is an n-i-p stacked structure
whose photo-response has been optimized by using a numerical device simulator. The
optimization process was performed as a trade off between the spectral selectivity and the
dark current of the device in order to increase the signal-to-noise ratio.
It has been observed that the system detection limit lies around 3 nmol/l. Taking into account
the spccific activity of labeled DNA in the solution, it corresponds to a minimal surface
density of fluorochrome around 50 fmol/cm2. This result is very encouraging because it is
obtained with a low cost system, where the use of optics for focusing and filtering of both
exciting and emitting radiation is avoided.

1 Introduction
Microarray technology (or DNA chip) allows the parallel study of many
different DNA sequences in a single hybridization experiment [l] [2] [3]. The
quantification of the hybridization is mainly based on the detection of
fluorescent molecules bound to the DNA target [2]. Fluorescence is detected by
laser scanners, CCD cameras, or microscopy [4], which produce an image of the
microarray [4][5]. However, most of these systems are quite expensive and need
complex optics for focusing and filtering of the emitted light.
Recently [6][7], amorphous silicon (a-Si:H) sensors have been developed as
alternative detectors in order to take advantage of its on-chip integration
possibility. In this paper, we present an alternative fluorochrome labeled DNA
detection system based on the use of an amorphous silicon (a-Si:H) sensor,
where the need of complex optics is overcome.

2 Labeling the DNA

The DNA molecules have been labeled with the fluorochrome Alexa Fluor
350, whose excitation (open squares) and emission (open circles) spectra are
reported in Fig. 1. We observe that this fluorochrome is excited by UV radiation

182
 183

with an excitation peak at Lx=340 nm and re-emits visible spectra with an

emission peak at h,,=442 nm.

Alexa Excitation
Alexa Emission -
0
QEExperiment .
0
-QE Modeling x
-08 2
a,
I!
-0 6 5
-

m
-5
c
40 - 04 6
i
i a
m
a, 20 - 02 z!
6 E
. b
z
0.. - 00

200 250 300 350 400 450 500 550 600 650 700 750 800 850
Wavelength (nm)

Fig. 1: Emission (open circles) and excitatron (open squares) specira of thejluorochrome Alexa 350.
Measured(so1id circles) and modeled(so1id line) QE of the sensor are also reported.

24 bases long DNA oligonucleotides, carrying an amine group at the 5

extremity were labeled by reacting them with a succimidyl ester of Alexa Fluor
350. As an alternative, a double stranded molecules, produced by Polymerase
Chain Reaction (PCR), were labeled using a substrate of labeling reaction
solution containing 1 pl of aminoallyl mix lOOx composed by 25mM dATP,
25mM dCTP, 25mM dGTP, lOmM dTTP, 15mM dUTP. The 3-aminoallyl-
dUTP has the function to replace the Timine bases in the new DNA chain and to
make a coupling point with the labeling fluorochrome.
Labeled molecules were purified by electrophoresis method and analysed by
spectrophotometry in order to determine DNA concentration and specific
activity. In the different experiments specific activities in the range of 05-1.5
molecules of fluorochrome/molecule of DNA have been estimated. Labelled
DNA preparations of known specific activity were subsequently used to test the
sensitivity of the detection system.

3 Photodetector fabrication and characterization

The sensor is an n-i-p stacked structure, deposited on a glass substrate
covered by a transparent conductive oxide (TCO). In order to detect the light
emitted by the fluorochrome, the detectors quantum efficiency should match the
emission spectra of the Alexa Fluor 350 as much as possible. The optimization
of the spectral selectivity, however, should take into account the photocurrent as
184

well as the dark current of the device in order to achieve the actual signal-to-
noise ratio.
Modeling of device performances has been carried out by a numerical device
simulator [8]. In the optimization process we considered the optical properties of
the glass and the TCO. The spectral transmittance of a TCO covered glass
substrate is like to a high-pass filter for the wavelengths of the incident light.
The cutoff wavelength occurs at 390 nm, wavelength positioned between the
peaks of the excitation and emission spectra of the fluorochrome. This has an
important consequence in the detection system, because the sensor detects only
the light emitted by the Alexa Fluor 350 and not the UV excitation radiation,
filtered by the TCO covered glass.
Basing on simulation results and taking into account the growth rate in the
chambers of our Plasma Enhanced Chemical Vapor Deposition (PECVD)
system, we deposited a n-i-p device with the deposition parameters reported in
Table 1 corresponding to the following optimized layer thicknesses: 25, 100 and
25 nm for the n, i and p zones, respectively.

SIH4 CH4 PH, B z H ~ PD PRF TD to Thickness

Layer
(sccm) (sccm) (sccm) (sccm) (Tom) (mW/cm2) (C) (sec) (A)
d 40 __ 10 -- 300 30 200 120 250
I 40 __ __ -- 300 30 180 600 1000

p+ 40 __ -_ 3 300 30 160 180 250

In Fig. 1 the normalized quantum efficiency (QE) curve (solid line) as

obtained from simulations and the normalized measured QE (solid circles) of the
device are shown. Even though the peak of the device QE does not occurs at k,,,
the thickness of the deposited layers ensures a good trade-off between
photogenerated carrier collection and the requirement of a low dark current.
Circle shaped samples with area equal to 3.14 111111 have been characterized
measuring dark current and photocurrent at hem. The lamp intensity at 440 nm
was 15.7 pW. The device dark current in operating condition (Vblas-lOOpV) was
in the order of 5 4 J 0 A/cm2and its responsivity around 145 mA/W.

4 Discussion
The detection system where our detector is included is depicted in Fig. 2. UV
radiation is incident on the labeled DNA, dropped on the same glass substrate
but on the other side with respect to the sensor.
Spatial confinement of the DNA solution has been achieved by
manufacturing a Kapton microtrap 350 pm deep. The radiation re-emitted by the
 185

fluorochrome passes through the substrate and is absorbed by the sensor, which
is aligned with the DNA solution.

Kapton rn

Fig. 2: Experimental set-up ofDNA detection system. The light is impinging on the solution
containing DNA dropped on the top of substrate.

The selected excitation wavelength for fluorochrome excitation was

he,=253.4nm, because this wavelength ensures a good excitation efficiency and
is very well filtered by both glass substrate and TCO. The UV lamp intensity at
253.4 nm was in order of 3 pW. The photo-detector is therefore illuminated only
by the emission wavelength and the photocurrent generated inside the sensor is
directly proportional to the amount of labeled DNA.
Sensor photocurrent due to fluorochrome fluorescence is reported as solid
circles in Fig. 3 versus DNA concentration.

1 10 100 1000
DNA concentration (nmol/l)
Fig. 3. Measured sensor photocurrent (symbols) as a function of DNA concentration. The ofset
correction allows to obtain the excellent lineariv shown by the solid line.
186

These data have been calculated performing the correction of the

photocurrent offset due to glass substrate fluorescence when the UV radiation is
impinging on a solution without DNA. The slope-1 solid line in the double
logarithmic plot of Fig. 3 serves as guide for the eyes to show the excellent
linearity of the detector response over more than two orders of magnitude. We
also observe that the detection limit of our system lies around 3 nmol/l. From the
measured values of dark and photo current, we deduce that this limit can not be
ascribed only to the current shot noise, but has to be related to the experimental
set-up.
Finally, taking into account the specific activity of labeled DNA in the
solution, we calculated a minimal detectable surface density of fluorochrome
around 50 fmol/cm2.

5 Conclusions
An alternative fluorochrome labeled DNA detection system based on the use
of an amorphous silicon sensor has been presented. The sensor has been
deposited on a glass substrate avoiding the use of optics for focusing and
filtering of both exciting and emitting radiation. The photodiode has been
optimized by using a numerical device simulator taking into account both the
spectral selectivity and signal-to-noise ratio. The detector response has been
measured as a function of DNA concentration. An excellent linearity has been
found over more than two orders of magnitude. DNA concentrations as low as 3
nmoVl have been measured and a minimal surface density of fluorochrome of 50
fmol/cm2has been estimated.

References
[l] M. Schena, D. Shalon, R.W. Davis, P.O. Brown, Science 270 (1995) 467.
[2] D.Meldrum, Genome Res, 21 (1999) 20.
[3] S.K. Moore, IEEE Spectrum, 3 (2001) 54.
[4] R.M. Ostroff, D. Hopkins, A.B. Haeberli, W. Baouchi, B. Polisky, Clin.
Chem., 45 (1999) 1659.
[5] J.B. Lamture, K.L. Beattie., B.E: Burke, M.D. Eggers, D.J. Ehrlich, R.
Fowler, M.A. Hollis, B.B. Kosicki. R.K. Reich. S.R. Smith, Nucleic
Acids Res. 22 (1 994) 2 12 1.
[6] F. Fixe, V. Chu, D. M. F. Prazeres, J.. P. Conde, Nucleic Acids Research
32 (2004) e70.
[7] T. Kamei, B.M. Paegel, J.R. Scherer, A.M. Skelley, R.A. Street,R.A.
Mathies, Anal. Chem. 75 (2003) 5300.
[S] D. Caputo, U. Forghieri, F. Palma, Journal of Applied Physics 82 (1997)
733.
[9] D. Caputo, Solar Energy Materials and Solar Cells 59 (1999) 289.
 Carbamate, triazinic and benzotriazinic pesticide analysis
using an inhibition tyrosinase organic phase enzyme sensor

Luigi Campanella, Elisabetta Martini, Dalina Lelo, Mauro Tomassetti

Department of Chemistry, University of Rome La Sapienza, P.le A . Moro 5, 00185

Rome Itah
luigi.campanella@uniromal .it; mauro.tomassetti@uniromal .it

Summary
Tyi-osinase enzyme inhibition has recently been used with increasing frequency to determine
different types of pesticides, in particular those of the triazinic type. Starting from this
premise, the present authors, who originally developed inhibition biosensors based on the
enzymes butyrylcholinesterase + choline oxidase to determine organophosphorus and
carbamate pesticides, also as OPEE devices (i.e. organic phase enzyme electrodes), more
recently suggested a new tyrosinase inhibition OPEE suitable for determining triazinic
pesticides. In the present research several triazinic pestides have been tested using this
inhibition OPEE. In addition they indicated how, using this biosensor, also benzotriazinic and
carbamate pesticides can be detected up to a concentration of the order of 10-8-10-9mol I-.

1 Introduction
Several inhibition biosensors have been constructed for pesticide
determination [I-31. The majority are based on the inhibition of the enzyme
acetylcholinesterase or of the enzyme butyrylcholinesterase. Also our research
group on several occasions has undertaken the development of this type of
inhibition biosensor [4-61.
Moreover, the recent development of OPEEs (Organic Phase Enzyme
Electrodes) [7, 81 has provided a way of working round one of the greatest
difficulties often found in the biosensor analysis of pesticides which is related to
the poor solubility in water shown by many of these compounds. In this context
we recently showed how [9, 101, in its inhibition OPEE configuration, the
tyrosinase biosensor allows triazine pesticides to be determined when operating
in organic solvent (chloroform), in which the pesticides tested are much more
soluble than in water.
This reduces the risk of underestimating the concentration of these pesticides
due to the difficulty of dissolving them completely in the aqueous phase. The
method was first optimized [9], and then some triazine pesticides (Atrazine,
Atrazine-desethyl, Atraton, Propazine) were tested [9, 101. The method has been
extended to other triazines (Simazine and Ter-buthylazine) and to another
important class of pesticides, i.e. benzotriazines; one of these 1, 2, 3-

187
188

benzotriazinic pesticides which also have a phosphorodithioate group, i.e.

azinphos-methyl, has been tested by tyrosinase inhibition OPEE.
Since bibliographic references [ 111 show that also carbarnates exert an
inhibiting action on the tyrosinase also a carbamate pesticide (Pirimicarb) was
tested.

2 Methods
In practice the enzymatic inhibition tyrosinase biosensor operating in non-
aqueous solvent (OPEE) described in detail in previous papers [7-91 was used in
order to determine several phytopharmaceuticals having a triazinic,
benzotriazinic or carbamate structure (Fig. 1).
The inhibition biosensor calibration curve at growing pesticide concentrations
is obtained by recording the percentage inhibition of biosensor response as a
function of the pesticide concentration in the solution in which the biosensor has
been immersed for a given period of time (incubation time) before substrate
(phenol) addition. This response is then compared with the biosensor response to
the phenol in the absence of pesticide. For the calibration curves, as required,
standard solutions of pesticides in water-saturated chloroform from 2 . 0 ~ 1 0 - ~
-'
mmol 1-' to 10 mmol 1 are usually used.
In practice, in order to perform each measurement, the biosensor is dipped
into a thermostated cell containing 10 ml of organic solvent (water-saturated
chloroform) kept under constant magnetic stirring and the signal is allowed to
stabilize. Then 200 pl of a solution of phenol 2.5x10-' mol 1-' in water-saturated
chloroform (i.e. concentration previously optimized [9]) is added and the signal
variation measured. The measuring cell is rinsed and dried and a new measure
performed by washing and then dipping the biosensor in 10 ml of one of the
various water-saturated chloroform atrazine-containing solutions prepared to
build the calibration curve, then waiting for a period of time equal to the
incubation time of 20 min (time optimized in previous research [9]), while the
enzyme and the inhibitor remain in contact, before adding 200 p1 of a phenol
solution 2.5x10-' moll-' in chloroform.
After adding the phenol solution the signal variation is recorded and
compared with the one previously obtained. In constructing the calibration curve
the final pesticide concentration is placed on the X-axis and the following ratio
on the Y-axis:
Ai% = (Sc - Sd) x 100
sc
Where:
Ai% = percent inhibition value
Sc = signal variation due to the addition of phenol in the absence of pesticide
in solution.
 189

Sd = signal variation on addition of phenol after the biosensor has been

"incubated" in the presence of pesticide.
All the triazine, benzotriazine and carbamate pesticides tested were
determined using this operating procedure.

SIMAZINE TERBUTHYLAZINE

NY" MIEt
Y" NHEt
AZINPHOS-METHYL PIRIMICARB

V y
II
N
I
\
CH 2- S-
Is
P-
I
OMe
MezN-
"+
C -0
I
MP
--
0 OMe II
0
Fig. I - biazinic, benzotriazinic and carbarnate pesticides analyzed

3 Results

The calibration curves of several triazine and benzotriazine pesticides tested

in water-saturated chloroform are shown in graphic form in figures 2(a), 3(a) and
4(a). As can be seen, in all cases a logarithmic trend is obtained that is typical of
an inhibition biosensor response, which may thus be converted into a linear trend
by transferring the data to a semilogarithmic diagram (see figures 2(b), 3(b) and
4(b)). Tables 1, 2 and 3 show the equations referring to the straight lines obtained
by transferring the data to a semilogarithmic diagram, as well as the main
analytical data found.
Figures 5(a) and 5(b) respectively show calibration curves (both logarithmic
trend and trend converted into linear form using a semilogarithmic diagram) of a
typical carbamate pesticide, while the relative equation and analytical data are
summarized in table 4.
190

0 2 4 6 0 10
Islmulnol (mmol t']
Fig. 2 (a) Calibration curve (Simazine) in water-saturated chloroform, and (b) the s rme
calibration curve data as in (a) transferred to a semilogarithmic scale,

y = (5.8i2.3) Log(x) + (71.M8.9)

r2 = 0.9483 (1 - a ) = 0.90 t = 2.365
x = concentration ; y =percent inhibition (AI%)
Linear range from 2.0 x 10-5 mmo11-1 to 10 mmo11-1.
LOD = 1.o x 1o - ~~ O I-' I

80 80

70 .
v

;;:/
-B
a 30-

20 -

10 -
- 0

Fig. 3 - (a) Calibration curve (terbuthylazine) in water-saturated chloroform, and (b) the same
calibration curve data as in fa). transferred to a semiloaarithmic scale.

y = (4.212.1) Log(x) + (62.3*8.8)

r2 = 0.9527 (1 - a ) = 0.90 t = 2.447
x = concentration ( m o l l - 1 ) ; y =percent inhibition (AI%)
Linear range from 2.0 x 1o - mmol ~ 1-1 to I o mmo11-1.
LOD = 1.O x 1O4 mmol 1.'
"Pooled SD%" = 8 %

Tab 2. Analytical data relative to the calibration curve offig. 3 (b)

 191

100

80

-5
P
60

f
5 40

d
20

0 2 4 6 6 4 4 3 - 2 - 1 0 1
[Arinphos-msthyl (mmoll') Log CPznphao-mA)1] rnrml 1 %

:ig. 4 - (a) Calibration curve (Azinphos-methy[)in water-saturated chloroform, and ( the

.~nmnp ~ dntn n r ...
rnlihrntinn r m v ~ ~ trnnsfPrrPd to -.--
-.,, ..---.-J
in 1n1 n .semiln~arithmic
~ -
.. . .........scale.

y = (4.h1.3) Log(x) + (108.1 k10.9)

r2 = 0.9797 (1 - a) = 0.90 t = 2.446
. .. (rnrnoi
x = concentrarion ~ 1 3 - 1 ,
I . . ~,~
*~... , * r n / \
1 a ) ; y =percent innloition ( . w - / ~ J
~~~

Linear range from 2.0 x 10-6 mmol 1-1 to 5 mmo11-1.

I nn = I n I n-6mmnt
YVY I." 1" ,
1-1
>,"I.".

"Pooled SD%" = 9 %

Tab 3 . Analytical data relative to the calibration curve o f j g . 4 (6)

ig. 5 - (a) Calibration curve (Pirimicarb) in water-saturated chloroform, and (b) the same
data of calibration curve in (a), transferred in a semilogarithmic scale.

y = (6.2i0.5) Log(x) + (81.8 k8.6)

r2 = 0.9739 (1 - a) = 0.90 t = 2.356
x = concentration ( m o l l - 1 ) ; y =percent inhibition (AI%)
Linear range from 2.0 x 10-5 rnrno11-1 to 5 m m o ~1-1.
LOD = 1.o x 1o - ~~ O I-' I
"Pooled SD%"= 1 1%

Tab 4: Analytical data relative to the calibration curve o f j g . 5 (b)

192

4 Conclusion
The results obtained indicate the complete validity of the method. The
tyrosinase biosensor, in the inhibition OPEE configuration, allowed triazine,
benzotriazine and carbamate pesticides to be determined when operating in
organic solvent (chloroform), in which the pesticides tested are much more
soluble than in water.
Also in this case the results using the same operating procedure as in the case
of the inhibition OPEE based on the bienzymatic pair butyrylcholinesterase +
choline oxidase [ 12, 131 were completely positive, while as solvent, instead of
the water-saturated chlorofodhexane mixture (50% v/v), previously used for
the butyrylcholinesterase + cholinase oxidase OPEE, this time only water
saturated chloroform was used in view of the good solubility of the substrate
(phenol) in this solvent.
References
[I] Besombes J.L., Cosnier S., Labbe P., Reverdy G. (1995), Anal. Chim. Acta 31 1. 255-263.
[2] McArdle A,, Persaud K.C. (1993), Analyst 118: 419-423.
[3] Mazzei F., Botre F., Lorenti G., Simonetti G., Porcelli F., Scibona G., Botre C. (1995), Anal.
Chim. Acta 3 16: 79-82.
[4] Campanella L., Cocco R., Sammartino M. P., Tomassetti M. (1992), Sci. Tot. Environ.,
123/124: 1-16.
[5] Campanella L., Achilli M., Sammartino M. P., Tomassetti M. (1991), Bioelect. Bioenerg., 26:
237-249.
[6] Campanella L., Colapicchioni C, Favero G., Sammartino M. P., Tomassetti M. (1996), Sens.
Act. B 33: 25-33.
[7] Campanella L., Favero G., Sammartino M. P., Tomassetti M. (1994), Talanta 41: 1015-1023.
[8] Campanella L., Favero G., Sammartino M. P., Tomassetti M. (1998), Talanta 46: 595-606
[9] Campanella L., Bonanni A.,Martini E., Todini N., Tomassetti M. (2005), Sens. Act. B111-112:
502-514.
[lo] Campanella L., Dragone R., Dalina L., Martini E., Tomassetti M. (2005), Anal. Bioanal. Chem.
(in press).
[ I l l Lee K.H., Koketsu M., Choi S. Y., Lee K.J., Lee P., Ishihara H., Kim S.Y. (ZOOS), Chem
Pharm. Bull., 53(7): 747-749.
[I21 Campanella L., De Luca S., Sammartino M. P., Tomassetti M. (1999), Anal. Chim. Acta 385:
59-71.
[13] Campanella L., Persi L., Sammartino M.P., Tomassetti M., Zannella S. (ZOOO), Ann. Chim. 90:
35-49.

Acknowledgments
This work was funded by FIRB (Fondi di Investimento per la Ricerca di Base) - Bando
2001 "Microsenson e nanofiltri per il controllo di aggregati di molecole e di microrganismi in
fase liquida: applicazione alla potabilizzazione dell'acqua".
 DNA-BASED PIEZOELECTRIC BIOSENSOR FOR Human
Papilloma Virus @PV) DETECTION

D. Dell'Atti, M. Minunni*, S. Tombelli, M. Mascini

Universita degli Studi di Firenze, Dipartimento di Chimica,e- mail
corresponding author: minunni@,unifi.it

A. Cavazzana, M. Zavaglia, G . Bevilacqua

Universitd degli Studi di Pisa, Dipartimento di Oncologia, dei trapianti
e delle nuove tecnologie in medicina

Summary
HPV infection is one of the most common diseases sexually transmitted. HPV DNA is
present in all the lesions associated to cervical cancer.
The aim of this work is the development of piezoelectric DNA biosensors to detect
viral DNA as an alternative method respect to traditional diagnostic techniques based
on cytological screenings (PAP TEST) or on molecular biology tests.

1 Introduction
The carcinoma of the uterine cervix is the second most common type of cancer
worldwide among women [I]. In the European Union more of 25000 new cases are
diagnosed each year (3.8% of all women neoplasia) with a mortality rate near to
50% [2]. Molecular and epidemiologic studies have provided unequivocal
evidences of a tight link between the cervical carcinoma and the infection with
Human Papilloma Virus (HPV).
HPV is a double-strand DNA virus that belongs to the papovavirus family.
More than 200 HPV types have been identified and at least 80 different HPV
genomes have been sequenced. Moreover 30 HPVs are associated at cancer and
they are subdivided in low and high risk to develop the uterine cervix carcinoma.
Low-risk HF'Vs (6, 11) generate benign lesions or warts and high-risk HPVs (16,
18) have potential to progress into cancer. Traditionally, the conventional method
to detect the presence of the virus is the PAP TEST, but this cytological
screening has some limitations such as reading errors of cytological preparation.
Moreover it is a prognostic method since it supplies only a data of the actual
situation of the organ. Today it is common to support the cytological screening

193
194

with techniques of molecular biology in order to detect directly the viral DNA
(PCR ELISA assay, HPV DNA test, Hybrid Capture assay).
The aim of this work was to develop DNA piezoelectric biosensors to detect
HPV and its genotypes. DNA piezoelectric biosensors are rapid, inexpensive,
label free analytical methods and they could represent a promising alternative to
current HPV DNA screening technologies. Three DNA biosensors have been
developed and optimized. Two biosensors were designed in order to obtain a
device for simultaneous detection of different HPV genotypes. The target
sequences mapped into the gene for L1 region (viral protein) of the viral capside.
The first biosensor was developed using a 11-mer probe common to all the
genotypes. The second one has a 3 1 bases probe which consists of a conserved
region coincident with the 11-mer probe and a degenerated region. This latter
consists in a mixture of probes differing only in few bases, able to catch and
hybridise to different HPV genotypes. In particular the 31-mer probe was chosen
aiming to the detection of: 6, 11, 16, 18 and 33 strains, the most interesting strains
for screening. The last developed biosensor was specific for HPV 18, that is
considered a high risk genotype. The hybridisation reaction was first studied with
synthetic oligonucleotides and biosensors were optimised in the main analytical
parameters.

2. Materials and methods

9.5 MHz AT-Cut quartz crystals (14 mm) with gold evaporated (42.6 mm2area)
on both sides were purchased by International Crystal Manufacturing (USA).
The measurements were conducted in a methacrylate cell where only one side of
the crystal was in contact with the solution. The quartz crystal analyser used for
the measurements was the QCMagic analyser by Elbatech (Marciana, Livorno,
Italy).
The reagents for the probe immobilisation are: 11-mercaptoundecanol from
Sigma (Milan, Italy), Dextran 500 from Amersham Biosciences (Uppsala, Sweden),
(+)/-epichlorohydrin and N-hydroxysucciminide from Fluka (Milan, Italy). Ethanol
and all the reagents for the buffers were purchased from Merck (Italy).
Two different buffers were used: immobilisation buffer (NaCl 300 mM,
Na2HP0420 mM, EDTA 0.1 mM, pH 7.4) and hybridisation buffer (NaCl 150 mM,
Na2HP0420 mM, EDTA 0.1 mM, pH 7.4). Oligonucleotides were purchased from
MWG Biotech (Milan, Italy). The base sequences of the 5- biotinylated probes
and of the complementary targets are shown in the table 1.
 195

HPV3 1 deg 3 lbases 5-biotin-ttt gtt act gt(gt) gt(at) gat ac(ct) ac (at) cgc
agt a-3

I HPVl 1 1lbases I 5-biotin -ttt gtt act gt-3 I

HPV 18 20bases 5-GTA CAG GAG ACT GTG TAG AA-3

HPV3 1 deg 3 lbases 5-tac tgc g(at)g t(ag)g tat c(at)a c(ac)a cag taa caa a-3

HPVll 1 lbases 5- aca gta aca aa -3

Tabl.

Probe immobilisation: Before the isnmobilisation of the probe, the crystals

were washed in a boiling solution of H202(30%), NH3(30%) and milliQ water in a
1:l:S ratio for ten minutes and then rinsed with milliQ water. The biotinylated
probe was immobilised via biotin-streptavidin binding on the gold sensor surface
previously modified with thiol/dextra~streptavidinas reported in Tombelli et al.
~31.
Hvbridisation reaction: Once the probe was immobilised on the gold surface,
the hybridisation reaction with synthetic oligonucleotides in hybridisation buffer
solution was conducted by adding to the sensor cell 100 pL of the
oligonucleotide solution at different concentrations in the range 0.06-1 pM. The
reaction was monitored for 10 minutes, the solution was then removed and the
surface washed with the same hybridisation buffer to eliminate the unbound
oligonucleotide. After each cycle of hybridisation, the single stranded probe on
the crystal surface was regenerated by 1 min treatment with HCl 1 mM solution.

3 Results and discussion

The biotinylated probes were successfully immobilised on the sensor surfaces

and this was in agreement with our previous work conducted with different
biotinylated probes [3,4].
Three calibration curves with synthetic oligonucleotides were obtained using
respectively the 3 1-mer probe and 1 1-mer probe, able to detect the HPV presence.
196

The results are shown in figure 1. It is clear that the highest signal was recorded
when the 3 1-mer probe hybridises the target with higher molecular weight, 3 1-mer
target (calibration curve A). The lowest signals were observed when the 11-mer
probe hybridises the target with lower molecular weight, 11-mer (calibration curve
C). In between the hybridisation signals obtained with 11-mer probe and target
31-mer (calibration curve B) were found.
Moreover a calibration curve was also obtained for biosensor HPV18, using
the HPV 18 probe (calibration curve D). (Figure 2).

-120

-100

-
-E!2 -80

5 -60

2
5 -40
U

;-20
0

I 0 01 02 03 04 05
Conc. (pM)
06 07 08 09 1

Fig. 1: Calibration curves obtained respectively with: A) probe 3lmer- target 3lmer. B)
probe Ilmer-target 3lmer; C) probe 1 lmer-target l l m e r

-900 -1
-800.

-700 .
0 -60.0.

0 01 02 03 04 0.5 06 07 08 09 1
cons. (pM)

Fig. 2:D)Calibration curve obtained with probe specific forHPV 18 and its complementary
target. Concentration range: 0.06-IpM
 197

For each sensor the specificity was tested adding 1 pM 26- mer solution of
non-complementary sequence, used as negative control. No remarkable
hybridisation signal ms obtained, in accordance with the specificity of the
sensors (shift <3 Hz). Reproducibility, expressed as average coefficient of
variation (CV% ), was evaluated for the different sensors .The CV% value are
respectively: calibration curve A lo%, calibration curve B 20%, calibration curve
C 28%, calibration curve D 11%. The masurements were always repeated in
triplicates.
As the best results were obtained when the 3lmer probe was immobilised
(calibration curve A, Wh,, lo%), this biosensor was chosen for the
simultaneous detection of different HPV genotypes.

4. Conclusion
The developed sensors optimised with synthetic oligonucleotides demonstrated
the possibility to obtain an HPV biosensor able to detect the virus presence and,
eventually to biosensor to perform also the relative genotyping, by coupling it to
a second biosensor specific for HPV 18 genotype.
Future work will be directed in testing these biosensors with real samples such as
amplified DNA (by polymerase chain reaction) extracted from scraping, aiming to
demonstrate the potentialities of these devices as an alternative method with
respect to screening technologies currently in use. The application of such
devices will be first as qualitative screening for the virus presence (yes or not
HPV) by using the degenerated 31-mer probe common to all the genotypes; then
if the first answer is positive for the virus presence, further HPV genotyping with
the relative specific biosensor to apply in series to the first one will be performed.
We have here shown the HPV 18 successfully applied. Biosensor development
for the second high risk genotype for cervical cancer (HFV 16) is in progress.

References
[I] Parkin DM, Bray F, et al. Int J Cancer 2001,94:153-156
[2] Black RJ,Bray F, et al. Eur J Cancer 1997 33: 1075
[3] Tombelli, S., Mascini, M., Braccini, L., Anichini, M., Turner, A.P.F.,
2000.Biosens. Bioelectron. 15 (7-8), 363-370.
[4] Minunni, M., Tombelli, S., Scielzi, R., Mannelli, I., Mascini, M.,
Gaudiano, C., 2003. Anal. Chim. Acta 481 (l), 55-64
 User Friendly Electrochemical Hand-Held Device For Durum
Wheat Safety

Michele Del Carlo, Marcello Mascini,

Dario Compagnone
Department of Food Science, University of Teramo, Teramo
mdelcarlo($.unite.it

Angelo Visconti
Institute of Sciences of Food Production, ISPA-CNR, Baril

Summary
In this report we describe the development of an integrated device for different analytes
threatening durum wheat safety. Particularly we have developed diverse electrochemical
biosensors, based on screen printed electrodes for the detection of organophosphate
pesticides (dichlorvos and pirimiphos methyl), toxins (ochratoxin) and PCR amplified DNA
from Fusurium culmorum. All the analytes were detected below the maximum residue level,
when established by the European Union. A dedicated software was also developed to
simplify the implementation of the procedure and data interpretation.

1 Introduction
Durum wheat safety is affected by different threats comprising both abiotic
and biotic agents. Organophosphorus compounds (OP) are widely used in
agricultural practices. They inhibit acetylcholinesterase (AChE) [ 11. This can
lead to clinical implications and high acute toxicity [2]. Among OPs Dichlorvos
and Pirimiphos methyl are important for the durum wheat industry [3]. European
Union regulation foresees a maximum residue limit (h4RL) for dichlorvos in
durum wheat at 2.0 pglg [4-51. Among biotic agents fungi that produce toxins
are a widespread threat. Fusarium species, (e.g Fusarium culmorum) are
universal fungal contaminants of cereals [6]. Their early identification is crucial
in order to avoid crop losses and protect consumer health. Other fungi
belonging to Aspergillus and Penicillum species can contaminate cereals
resulting in a possible contamination with ochratoxin A (OTA). Recently, the
European Commission fixed the maximum levels for OTA at 5 pgkg in cereals.
In this report we describe the application of biosensors to the determination
of such contaminants in durum wheat samples. All the reported applications
have been developed using a PalmSens hand-held potentiostat (PalmSens,
Amsterdam, The Netherlands). Finally, a dedicated software was also developed
in order to enable the end user to carry out the different analysis regardless his
technical training and skills.

198
 199

2 Detection procedures
Screen printed electrochemical sensors for dichlowos andpirimiphos-methyl
Both the determination reported rely on the inhibition activity of OPs
pesticides towards AChE combined with the electrochemical detection of
choline at a modified screen printed choline oxidase biosensor. Standard and
sample extract solutions were analyzed: first, a blank sample extract was
measured, and then the current of either the standard or the sample extract was
measured. The inhibition was calculated as: I (%) = 100 *(lo - IJIO.

Screen printed electrochemical sensors for the detection of ochratoxin

Competitive electrochemical enzyme-linked immunosorbent assays based on
disposable screen printed electrodes have been developed for quantitative
determination of ochratoxin A (OTA).
Indirect and direct format of immunosensors based assay were developed. 6
p1 of OTA-BSA in buffer (indirect) and 6 pl of goat IgG (anti-mouse IgG)
(direct) in buffer were dispensed on the screen printed working electrodes, PVA
solution was used to block the surface. 6 p1 of OTA monoclonal antibody were
added to the electrode surface for 30 min at room temperature. Binding or
competition was run with 6 p1 of OTA-AP conjugate or conjugate + standard
for 30 min at room temperature. Washing was then carried out. The activity of
the label enzyme was measured electrochemically by addition of 100 pl of
substrate solution (5 mg/ml 1-naphthyl phosphate in DEA buffer; prepared
daily), for 2 min at room temperature. The product, 1-naphthol, was detected by
DPV.

Screen printed electrochemical sensors for the detection of Fusarium sp DNA

DNA electrochemical sensors were developed for the detection of specific
sequences of Fusarium Culmorum. The sensing principle of the application was
based on the affinity interaction between complementary strand of nucleic acids:
the probe was immobilised on the sensor surface and the target was free in
solution. The hybridisation was measured using square wave voltammetry
(SWV). The label-free detection was done by monitoring the guanine oxidation
peak. This approach rely on the use of inosine-modified (guanine-free) probes.
In fact, the non electroactive base inosine still forms a specific base-pair with the
cytosine residue. This results in a flat baseline for the probe-modified electrode.
The hybridisation was detected by the appearance of the guanine oxidation peak
of the target sequence. The specificity of the genosensors was tested.

3 Results
Two different extraction approaches have been optimised for dichlorvos: the
former using whole wheat kernels and aqueous extraction and the latter using
200

grounded samples and hexane extraction. A typical electrochemical signal as

obtained by the instrument is shown in figure 1.
, )*

Figure I : @pica/chronoamperogramobtained in the ~nh~bition

assay

In the first approach, the method was calibrated both in buffer and durum
wheat extract. The detection limits in durum wheat samples were 0.45 mgkg for
the eeAChE and 0.07 mgkg for rAChE. These characteristics allowed the
detection of contaminated samples at the legal maximum residue limit, which is
2 mgkg. Moreover fortified samples of d u r n wheat were obtained with both
dichlorvos and the commercial product Didivane, which contains dichlorvos as
active molecule. At all the tested levels, the occurrence of contaminant was
detected with an average recovery of 75%. The total assay time, including the
extraction step, was 30 min. Samples (11755) fortified at different levels (4, 2, 1,
and 0.5 mgkg) were extracted and analyzed with the proposed electrochemical
assay using both eeAChE and rAChE. The incomplete recovery (75%) may be
due to concurrent causes, such as the use of whole kernels as sample, water-
based extraction solvent, and adsorption exerted by the matrix.
In order to apply the electrochemical method to ground wheat samples, as
generally required in food analysis, a nonaqueous extraction solvent was
required. To perform electrochemical analysis with the choline oxidase
biosensor, dichlorvos needed to be transferred to PBS solution, thus avoiding
any electrochemical interference by organic solvents. Therefore, the filtered
hexane extract was submitted to liquid-liquid partitioning with PBS and the
upper organic layer was removed by evaporation. Dichlorvos was easily
measured in ground wheat by electrochemical bioassay at levels as low as 0.05
pg/g. The mean recoveries of dichlorvos ranged from 97 to 108%, RSDr values
ranged from 5.5 to 7.8%, and RSDR values ranged fiom 9.9 to 17.6%. A good
correlation between dichlorvos concentrations obtained by electrochemical
biosensor and GC analysis was also found ( ~ 0 . 9 9 1 9 ) .
The electrochemical assay for the detection of AChE inhibitors above
outlined has been also optimised for the detection of pirimiphos methyl in durum
wheat. A procedure for the oxidation of pirymiphos methyl via N-
bromosuccinimide (NBS) and AChE inhibition was optimised in buffer solution
for reagents concentration and inhibition time. As a compromise, between
 201

analytical performance and overall assay length, a 10 min. incubation of NBS

and AA and 30 min. AChE incubation was selected. Calibration were performed
both in buffer and in sample extract. The intra electrode CV was between 1.6
and 15.0, the inter electrode CV% was between 4.6 and 15. The detection limit
(LOD) was 38 ng/mL, and the 150% was 360 ng/ml. The calibrations were
slightly affected by the sample matrix resulting in a increased LOD (65-133
ng/ml) and 150% (640-1650 ng/ml). Spiked samples were prepared at the EU
regulated level (5 mg/Kg) and analysed resulting in a recovery of 70.3%.
Competitive electrochemical enzyme-linked immunosorbent assays based on
disposable screen printed electrodes have been developed for quantitative
determination of ochratoxin A (OTA). The assays were calibrated using
monoclonal antibodies in the direct and indirect format. OTA working range, I50
and detection limits were 0.05-2.5 pg L'' and 0.1-7.5 pg L-', 0.35 pg L'' and
0.93 pg L-', 60 pg L-' and 120 pg L-' in the direct and indirect assay format,
respectively. The immunosensor in the direct format was selected for the
determination of OTA in wheat. Samples were extracted with aqueous
acetonitrile and the extract analyzed directly by the assay without clean-up. The
I50 in real samples was 0.2 pg L-' corresponding to 1.6 pgkg in the wheat
sample with a detection limit of 0.4 pglkg. Within- and between-assay
variability were less than 5 and lo%, respectively. A good correlation (r =
0.9992) was found by comparative analysis of naturally contaminated wheat
samples using this assay and an HPLC/immunoaffinity clean-up method based
on the AOAC Official Method 2000.03 for the determination of OTA in barley.
As long as concern DNA detection 4 different probes were used to obtain 4
sensors that were tested for selectivity and specificity with synthetic sequences
and PCR samples. The DNA biosensors were used to test PCR samples. The
results exhibited a dose dependent response up to 7 pg/mL, for higher
concentrations an hook effect was observed. In order to avoid false negatives a
control procedure was introduced. The PCR samples were split in two aliquots,
one of which was thermally denatured, for the analysis. It was evaluated that if
the denaturatednon denaturated was higher that 3, the sample were to be
considered as positive.

1 2 3 4 5 6 7 % 9 10
TZP.DIPCP .-I.

Figure 2: results obtained with one of the selected DNA probes with PCR samples
202

Particularly, interesting results were obtained using the probes

complementary to the region 3 and 4. The screen printed electrode modified with
the probe complementary to region 3 exhibited good repeatability (CV within
20% on 3 consecutive measurements) with no measurable analytical signal in
the presence of different amounts of non-complementary PCR sample.

4 Dedicated software
The software has an user friendly interface that inform the user on the
procedural steps that must be,performed (e.g add 20 p1 of Vial A to Vial B, mix
and transfer on the sensor surface) allowing the realisation of the measurement
even to not-trained personnel. All the internal controls result in a pass or fail
message as well as all the incubation times are controlled by the software. At the
end of every protocols the user receive a screen message that contains an self
explanative result via a colow: code according to a traffic light scheme (Figure

Figure 3: typical sof&varefi.ameobtained with a dichlorvos positive sample

References
[ 17 U.S. FDA. Pesticide Monitoring Database 1996-1999;
http://www.cfsan.fda.gov/
[2] Council Directive 801778lEEC (OJ L229, p 11,30/0S/l980) of 15 July 1980.
[3] 5a d d , D. Hartley, J. Kennedy, Insecticides and Acaricides. In European
Directory of Agrochemical Products, 2nd ed.; The Royal Society of Chemistry:
London, U.K.,; 3 (1986) 99-247
[4] Commission Directive 2001/57/EC of 25 July 2001.
[5] W.J. Donarsky, D.P. Dumas, D.P.Heitmeyer, V.E. Lewis, F.M. Raushel,
Archives of Biochemistry and Biophysics, 227 (1989) 4650-4655
[6] A. Bottalico, G . Perrone, European Journal of Plant Pathology, 108 (2002)
61 1 -624

Acknowledgement
This research was supported by MIUR project no. N12792 472 (SINSIAF)
 Determination of immunoglobulin G in human serum by
means of an immunosensor fitted with an enzymatic probe as
detector and a Clark electrode as transducer

Luigi Campanella, Elisabetta Martini, Mauro Tomassetti

Department of Chemistry, University of Rome ILaSapienza, P.le A. Moro 5, 00185

Rome Italy
[email protected];mauro.tomassetti@uniromal .it

Summary

A new immunosensor is developed to measure human immunoglobulin G . The method is

competitive in the heterogeneous phase and separative. The greatest novelty lies in the
use of a tyrosinase biosensor as detector and a Clark electrode as transducer. The
enzymatic marker is alkaline phosphatase. Using this new immunosensor it was possible
to determine anti-HIgG up to a concentration in the order of 10. mol I-. The present
note makes a detailed analytical comparison between this new immunobiosensor and a
similar potentiometric detection immunosensor proposed by us in a previous work.

1 Introduction

In previous work [ 11 we constructed and characterized an electrochemical

immunosensor for the determination of antibodies of human immunoglobulin G,
i.e. anti-HIgG (human anti-immunoglobulin G). This immunosensor was based
on a potentiometric gaseous diffusion electrode for NH3 as transducer and a
polymeric membrane overlapping the head of the electrode, on which human
immunoglobulin G (HIgG) was immobilized. Furthermore, urease was used as
enzyme marker for the antibody (anti-HIgG). The immunosensor was
characterized by a limit of detection (LOD) of 5 x moll- and by a linearity
interval between 9 x and 6.5 x lo4 mol 1- of anti-HIgG. We are currently
continuing this investigation, above all by developing a measuring system with a
lower limit of detection. For this purpose a pre-concentration system by
immunoprecipitation was initially developed, which, although tedious,
successfblly attained the goal set with an LOD of lO~-lO~mol 1-I. The
potentiometric immunosensor method was then used to determine
immunoglobulin G in human serum. It was found however that the presence of
urea in the serum can be the cause of a small but non negligible interference,
especially at low immunoglobulin G concentrations. To overcome this problem
we then constructed a completely different and highly innovative immunosensor
that used the alkaline phosphatase enzyme as marker, sodium phenylphosphate
as substrate and a tyrosinase biosensor as detector. In this case the transducer
was a Clark type amperometric electrode. Clearly the construction geometry of

203
204

this kind makes the imrnunosensor extremely selective. After optimizing the
competitive measurement procedure also in this case, it was used to determine
the anti-HIgG, with a LOD comparable to the one described above for the
potentiometric measurement using the preconcentration method.

2 Experimental

Materials and Apparatus

The amperometric measurements were performed in a 25 ml thermostatted

glass cell kept under constant stirring. The Clark electrode was supplied by
Amel (mod.332) and the amperometric measures were performed using an
oximeter (Amel model. 360), connected to a recorder ( M L mod. 868). Anti-
HIgG phosphatase alkaline conjugated, Anti-HIgG, HIgG and tyrosinase
enzyme were supply by Sigma-Aldrich, Immobilon membrane by Millipore,
LJK.

3 Methods
s

Method I (determination of anti-Hi@ with potentiometric immunosensor)

(Figure I )

Fig. I - Diagram ofpotentiometric immunosensor used

The Immobilon membrane, on which the HIgG was immobilized, was fixed
onto the head of potentiometric gas diffusion electrode for N&. Before
measurement, the potentiometr~cimmunosensor was immersed in 0.1 mol I-
(PK 8.0), containing 0.05 % Tween-20 by weight and 2.5% BSA by weight
(bovine albumin was used to minimize non specific adsorption on the
membrane). The anti-HIgG to be determined, together with a fixed
concentration of the anti-HIgG urease conjugated in 0.1 mol 1- Tris-NC1 buffer
solution, pH 8.0 was then incubated. The labelled and unlabelled antibody
competed to bond with the HIgG present on the membrane. Urea (the enzyme
 205

substrate) was added to the renewed solution, after washing all the anti-HlgG
not bound to the HIgG out of the immunosensor using the same buffer. A
calibration curve was used to determine the concen~ationof the unknown anti-
HgG. The sequence followed in measuring the anti-HIgG by this 1st method is
schematized in figure 2.

I Immobilizationof the HIgG on preactivated membrane Immobilon fucd on the head o f

potentiometric gas diffusion electrode for NH, . I
0.05 %Tween- min was used to

Washing thc membrane with the same buffer solution to remove all
the not hound anti-HI&.

NH, was produced from the enzyme reaction and was detected by potentiometricsensor.
The signal measured was correlated to the quantity of unlabelled anti-IiIgG.

-
Fig.2 Determination of antibody (anti-HIgG) by potentiometric immunosensor, using a NHj
electrode as transducer. Tesi geometry: competition for HIgG immobilized on Immobilon
membrane, between anti-HIgG urease conjugated and anti-HIgG non conjugated.

Method 2 (determination
-~ of anti-HIgG with immuno-biosensor)
- (Fi ure 3)

electrode fix 0, )

gas-permeable menibiane

Immohilved HIgG on
Imrnohilon membrane

Fig. 3 -Diagram of immuno-biosensor used

In practice, three membranes were mounted on the head of the electrode, in

the following order: gas-permeable membrane, dialysis membrane, TAC
membrane with tyrosinase enzyme and Immobilon membrane with HlgG
alkaline phosphatase immobilized on it. The membranes were kept in place by a
nylon net and an O-ring. Prior to measurement the immuno-biosensor was
206

dipped into a Tris-HC1 buffer solution, 0.1 mol 1-I (PH 8.0), containing 0.05 %
Tween-20 by weight and 2.5% BSA by weight (bovine albumin was used to
minimize non specific adsorption on the membrane). The anti-HIgG to be
determined was then added together with a fixed concentration (0.1 mol 1-I) of
alkaline phosphatase conjugated anti-HIgG to the Tris-HC1 buffer solution, pH
8.0. The enzyme-conjugated antibody competes with the non-conjugated
antibody to bind with the HIgG immobilized on the membrane. After washing
with the same buffer solution to remove all the unbound anti-HIgG, the specific
enzyme substrate, in this case sodium phenyl-phosphate, was added to the new
buffer solution in which the immunosensor was dipped, under stirring. The
measured signal correlates inversely with the quantity of anti-HIgG to be
measured. A calibration curve was constructed and used to determine the
unknown concentration of anti-HIgG contained in the sample. The sequence for
measuring the anti-HIgG by this 2nd method is schematized in figure 4.
lmmohtllration ofthe H I S on preactivated membrane Immobtlon,
fixed to the Clark amperometnc electrode

Washing the membrane with Tris-HCI buffer solution 0.1 mol 1_' @H 8.0),
contaming 0.05 % Tween-20 by weight and 2.5% BSA by weight (bovme albumm
was used to mmunue non sneclfic adsomtion on the membrane)

1
The anti-HlgG alkalme phosphatase wzyme-conjugated competes with the non-conjugated antibody (anti-HIS)
to blnd with the HlgG nnmohilued on the membrane.

8
Washmg the membrane with the same buffer solution to remove all the not
bound anti-HIS,

1
Addmg the speciiic enzyme substrate, m this case sodium phenCphosphate, to the new buffer
solution m which nnmunosemor was dipped, under sturmg

I Akalme ohosohatase allowed the release of ohenol to the sodium nhend-ohosohate. I

~ _________ ~~~

Lastly tyrosmase enzyme catalized the reaction in which phenol was transform m o-qumone, and
oxygen m solution is consumed The oxygen loss was checked by Clark electrode.

Fig. 4 - Determination of antibody (anti-HlgGj by immuno-biosensor, using a Clark electrode as

transducer and a tyrosinase biosensor as detector. Test geometry: competition for HIgG
immobilized on Immobilon membrane, between anti-HIgG alkaline phosphatase conjugated and
anti-HlgG non conjugated.

4 Results and discussion

Figures 5 and 6 respectively show the semilogarithmic diagram calibration

curves obtained for the determination of the anti-HIgG with the potentiometric
immunosensor (fig. 1) and the immuno-biosensor (fig. 3). Table 1 compares the
 207

54-

\:I:-eI 48 -

46 -

44 ,..! I I I I I ,,., , .. I I ,,,, I . . . ..

Determination of by Determination anti-HlgC by means

of immuno-biosensor, that uses as
means of potentiometric
detector a tyrosinase enzyme sensor.
immunosensor.
Geometry of the test: competition
Geometry of the test:
Methods
bemeen free anti-IIIgC, conjugated the *'gCt On the
with ureaSe and not conjugated, for between anti-H1gC
conjugated with the alkaline
the "IgC immobilized On the
fosfatase and not conjugated, both
membrane of immunosensor
free in solution.
Employed Membrane Immobilon Immobilon
Regression equation
(Y=a.u , X= moI I-') Y = -0.99 (M.06) lOgX + 46.2 (+0.7) Y = -0.86 (*0.04) log X + 1.33 (+0.02)
level of confidence (n - v ) = 6 ; (t = 2.26) (n - v ) = 7 ,(t = 2.36)
(I- a)= 0.95 ;
Linear range (mol r') (0.08-6.6) x lo4 (0.26 - 13) x lo-"
Correlation coefficient 0.9510 0.9831
Repeatability of the
5.9 42
Measurement (as pooled SD%)
Low detection limit
0.04 x lod 0.13 x lo-''
(LOD) (mol I-')
Instrumental response time
(minuts) 50 20
Lifetime
(number of successive measures) 3 2-3

-
Table I Analytical characterization and comparison of immunosensor methods f o r determining
anti-HIgG which respectively use as transducer a potentiometric sensor f o r NH, and an
amperornetric gas diflusion sensor for 0 2 0.e. tyrosinase biosensor as detector).
208

1 0 1
Log [anbHlgG] x 10-'* mol 1_'
Fig. 6 - Calibration C U F and confidence intervalfor the anti-HIgG. Immuno-biosensor response as
a function of growing anti-HIgG concentrations using the Immobilon membrane.

5 Conclusions

With the new amperometric immunobiosensor, which uses a tyrosinase

biosensor as detector, the limit of detection is about three decades lower than the
potentiometric immunosensor. Immunoglobulin G detection using the immuno-
biosensor also displays a better precision RSD% = 4.2 instead of 5.9 and an
enhanced correlation coefficient for the calibration curve.

Found Experimental A% Eqerimental A%

after 0.5 ml urea
immunoglobulin G after 0.5 ml urea mol 1'1 addition
Human serum concentration lo-' mol I-' addition
(using
(mol I-'), without (using
potentiometric
urea addition immunobiosensor)
immunosensor)
Diluted 1: 1000 1.3~10'~ +2.76 -25.60
Diluted 1:10000 1.8~10~'" +].lo -22.63
Diluted 1:100000 1.7~10~" -0.25 -28.26

Above all it became possible to determine immunoglobulin G in human serum

without the slightest interference by urea or other species contained in the
human serum, also at low immunoglobulin G concentration level (see table 2).
References
[l]. L. Campanella, R. Attioli, C. Colapicchioni, M. Tomassetti; Sensors and Actuators B 55
(1999). 23.

Acknowledgments
This work was financed by FIRB (fondi di Investimento per la Ricerca di Base)- Bando
2001"Microsensori e nonofiltri per il controllo di aggregati di molecole e di microorganismi in fase
liquid: applicazione alla potabilizzazione dell'acqua".
 SPR IMAGING TECHNIQUE:
A STUDY ON DNA-DNA BIOLOGICAL
INTERACTIONS
M.G. Maneva''2, J. Spadavecchia', R. Rella', A. Leone',
IMM - CNR, Via Provinciale per Arnesano, 73100 Lecce,
I

mariagrazia.manera @le.imm.cnr.it
'ISUFI-lstituto superiore di formazione interdisciplinare, Universita degli studi
di Lecce, Viale Gallipoli, 73100 Lace

D. Dell 'Atti3,M . Minunni3, M. Mascini3

'Dipartimenzo di Chimica, Universitcidegli szudi di Firenze, Via della Lastruecia
3, 50019 Sesto Fiorentino ( Fi)

Summary
Surface Plasmon Resonance (SPR) imaging is a label free method that can be used in an
array format for the detection of biomolecular interactions. In this work the interaction
between oligonucleotide DNA probes, immobilized onto suitable photolithography
patterned gold substrates and their complementary strands was monitored by means of
changes in the optical properties of the surface. The investigation involved the DNA probe
specific for the 35s promoter, a marker of the presence of genetically modified organisms
(GMOs).

1 Introduction
Surface Plasmon Resonance (SPR) technique is a surface sensitive optical
method for the characterization of ultra thin films adsorbed onto gold or other
noble metal surfaces [ 11. The interaction processes taking place between
oligonucleotides immobilised onto suitable photolithography patterned gold
substrates and their complementary strands are beginning to assume an important
role in biotechnological applications and in particular in the realization of
biosensors for specific applications like food control, medicine, environmental
control [2]. The possibility to monitor the nucleic acids hybridisation process by
a specific experimental set-up allows the detection of a complementary sequence
in a sample of unknown DNA, by immobilising a suitable probe [3,4].
Traditional techniques for the analysis of specific hybridisation require the use of

209
210

labels, such as radioactive isotopes or fluorescent molecules . They make use of

dangerous and expensive reagents, consequently, great attention has been
devoted to the development of new, simpler and faster, label free biosensors.
Specific variations in mass, generated at the surface of a sensing layer and due to
the hybridisation reaction, have been detected by using a quartz crystal
microbalance (QCM) as a transductor [5]. Also changes in surface optical
properties due to unlabeled DNA hybridisation have been observed by using the
surface plasmon resonance technique (SPR) [ 6 ] .Our work deals with the use of
this latter approach for DNA sensing applications. The surface plasmon
resonance phenomenon is the result of the interaction between electromagnetic
waves of the incident light and the free electron systems in a conducting surface
layer like a metal. The use of SPR technique to probe surface interactions is
advantageous, since it is able to monitor any dynamic process occurring at the
metal surface, such as adsorption or degradation, rapidly. It can give information
on the rate and the extent of adsorption, enabling the determination of dielectric
properties, the associatioddissociation kinetics and affinity constants of specific
interactions.
Two different experimental systems for the excitation of surface plasmons were
developed by Otto and Kretschmann. The latter is generally used in the designs
of most SPR instruments and it is adopted in this work. The position of the
resonance angle is sensitive to the refractive index immediately adjacent to the
metal surface. A change in the refractive index of the surface generates different
SPR responses. In this paper we describe the preparation of specific substrates
for SPR imaging experiments suitable to host sequences of DNA in array
configuration. The details of the adopted SPR experimental set-up were already
described elsewhere [7]. We have investigated the presence of a target analyte,
the P35S oligonucleotides, specific for the 35s promoter, a marker of the
presence of genetically modified organisms (GMOs) inserted in the host genome
to regulate the transgene expression, The probe 35s was thiolated in 5 and
immobilized by self-assembling monolayer technique (SAM) onto gold surfaces.
Notwithstanding we are interested in the realization of a specific SPR imaging
experimental set-up suitable to observe DNA hybridisation, the DNA sequences
adopted in this work being significant in the control of GMOs.

2 Experimental
Preparation of patterned substrates: Gold patterned substrates were prepared
starting from 25x25 mm2 SFlO slabs. A Cr/Au (2nm / 50nm) metallic multilayer
was deposited on the slabs by e-beam evaporation. An accurate deposition
system calibration has been previously performed by using test samples in order
to obtain a strict control on the deposited metal thickness to improve SPR
evidence. A bright field mask having arrays of circles with different diameter
 211

(from 50 to 200 pm) has been used to expose a positive photoresist. A 7.5 nm Cr
film has been successively deposited on the sample surface and with good arrays
have been obtained by the classical lift-off technique. Thiolated oligonucleotide
probes (HS-ssDNA) used in our work are 25 mer for GMO (5-HS-(CH2)6-GGC
CAT CGT TGA AGA TGC CTC TGCC-3).The oligonucleotides probes were
immobilised onto each single gold trap (150 pM in size) by the self assembling
monolayer formation (SAM). For this purpose a 5p1 of lpM HS-ssDNA solution
was prepared (in Na2HP04 0,02mM Tris-lOmM NaCl buffer, pH 7,2) and
droplets of this solution were spotted onto the surface of the realised substrate
and left for 2h at room temperature in a humid environment to prevent the spots
from drying out. Then the substrate was immersed into a solution of MCH(
mercaptoethanol 10-4M) for l h to eliminate the non specific adsorption of the
oligonucleotide bases on the gold surface; successively the substrate was soaked
in deionised water for some minutes and rinsed with water to remove unbound
probes.

SPR-Imaging apparatus: SPR studies were performed on a specially designed

system exploiting the Kretschmann configuration. In order to obtain SPR images,
a collimated beam generated by a LED source (1=660 nm, spectral width D1=20
nm) illuminates the functionalized gold sensor surface which is in contact with
the solution, through a coupling prism and under a variable incident angle
controlled by a rotating platform. The reflected light is detected at a fixed angle
with a CCD camera which produces an SPR image, then transferred to a
computer for analysis. The data could be plotted as spatial intensity maps of the
SPR substrate surface, where an increase in intensity indicates an increase in SPR
response. SPR image arises from variations in the reflected light intensity from
different parts of the sample. These variations are created by changes in the
dielectric at the interface with the metal layer; specifically binding of organic
matter to the transducer surface increases the local refractive index, allowing to
quantify interactions between an immobilised and a free molecular species. The
images presents 480x512 raw resolution at 8bpp (grey-scale levels) and were
acquired by using a commercial software package (Spiricon Laser Beam
Diagnostics). All experiments were carried out at room temperature. The in situ
investigated region of the gold surface was about 2x2 mm2, which was large
enough to monitor several immobilised oligonucleotide probe traps
simultaneously.

3 Results and Discussion

Figure1a shows the pseudo-colours SPR image corresponding to a magnification
of the region of the sample containing each gold traps after immobilization of the
thiolated probes.
212

Figure 1: Pseudo-colours SPR-Image of the gold traps after immobilization (a)

of the P35S probe and after hybridization of the target sequence with the
immobilized P35S probe (b).

Our aim was to obtain SPR images of the array and to monitor the
immobilization and hybridisation processes in real time. Consequently, the
experimental set-up demonstrates the potentiality of the proposed technique to
detect the DNA-DNA interaction and to test a preliminary threshold of sensitivity
of this interaction.
An array consisting of the same thio-oligonucleotides probes were immobilized
onto all the different gold traps. After the immobilization of the oligonucleotide
probes, the biochip was placed in a suitable cell (75 p1 in volume). The efficiency
of the immobilized active probes was tested in dynamic liquid phase in order to
demonstrate the hybridization process taking place with the complementa~
strand present in the injected solution. Figure l b puts in evidence the variation in
the SPR image morphology (variation in the grey-scale levels inside the gold
traps) of the investigated region of the sample after twenty minutes from the
immersion in a flow of buffer solution containing the DNA target.
Each step involved in the biosensing process has been followed by SPR
measurements in conventional scanning angle configuration and by imaging
methods.
 213

IP I

AU
immbb
passiv
hybrid u
regeneraz

3 3 5 4 0 4 2 4 4 4 6 4 8
external incident angle (deg)

Figure 2: Scanning-angle SPR measurements of each step of the biosensing

process.

Figure 2a reports the scanning-angle SPR measurements related to each step; the
immobilization of the probes onto the Au substrates results in a shift toward
higher angles of the minimum of the SPR curve as predicted by the SPR theory,
when a layer is adsorbed onto the metal film. A second step, is the surface
passivation, conducted with the mercapto-ethanol solution to eliminate any
possible aspecific adsorption onto the gold surface. The passivation step gave a
negligible shift in the SPR curve reported in Figure 2. Finally the hybridisation
reaction with complementary target was performed and, in this case, it results in a
shift of the resonance minimum toward higher values, demonstrating that the
surface binding between the probe and its complementary sequence causes an
increase of the refractive index and thickness of the layer above the metal film.
Afterwards, the regeneration process performed to dissociate the hybrid probe-
target allowing a second measurement cycle on the same surface, causes the
dissociation of the target leaving the surface and then, consequently, the return of
the SPR minimum to smaller angles. Finally a negative control, consisting in non
complementary DNA oligonucleotides, was injected into the cell. No significant
signal was observed in this case as shown in Figure 2 and in Figure 3.
214

Figure 3 shows the SPR images relative to the mentioned steps ( i ~ o b i l i s a t i o n ,

hybridisation, regeneration, negative control injection). The change in reflectance
(represented by a change in colours), can be observed at each step; the variations
are due to the different chemical processes occurring at the gold metal surfaces,
demonstrating that different events can be monitored by this SPR imaging
technique.

Figure 3: SPR images acquired after: (a) probe i ~ o b i l i z a t i o nin buffer

solution; b) hybridisation in buffer solution; c) regeneration; d) negative
control with a non complementary target.

Figure 4 reports the Absolute Difference image at grey-scale levels of the images
recorded after and before the hybridisation reaction, in order to roughly evaluate
how the hybridisation process occurs. In other words, the activity i s performed to
estimate the binding of the target sequence and its distribution in each well. In
order to improve the qualitative process of measurement of the hybridisation
effect, an automatic procedure has been used to align the input images allowing
the correct evaluation of the absolute difference. On the other hand, by using the
proposed automatic calibration tool, a generic pixel at location (ij) in the image
obtained before hybridisation is exactly related to the corresponding pixel in the
image evaluated after the hybridisation. This step is crucial since just one offset
pixel in the absolute difference gives rise to apparent faults during the
interpretation of the output image. The code for the evaluation of the absolute
difference image has been developed in Mathworks Matlab@7 (Figure 4) with
Image Processing Toolbox 5.2.
 215

Figure 4: SPR difference image : abs(post binding image - after binding image).

-----?7

Figure 5: Histogram at gray-scale levels of the (a) immobilization image and

(b) hybridization image.

Figure 5 shows histograms of grey-levels (a global operator) related to images

acquired before (a) and after (b) hybridization. We can note that the image
acquired after hybridization presents a shift in the grey-levels distribution
demons~atingthe occurrence of the binding event. The presence of variations in
the distribution of the images demonstrates that the reflectance of the surface
increases; this phenomenon is due to the binding events of the target sequence to
the immobilized probe according to the previous qualitative considerations.
216

Conclusions

A SPR- based platform has been used in order to obtain images of self assembled
thio-oligonucleotides onto suitably patterned gold surfaces. The detection of
hybridization events of complement DNA (target) onto the surface-immobilized
DNA (probes) has been realized. The hybridization resulted in an increase in the
local index of refraction, which in turns led to a change in the reflectivity from
the surface that is monitored by a CCD camera. This work demonstrated the
potentiality of the SPR imaging technique, complementary to the scanning SPR
apparatus, in monitor in situ DNA adsorption onto noble metal surfaces.
Important controls are included to test for non-specific binding of target DNA to
the substrate and to verify that the observed DNADNA interactions are sequence
specific.

Acknowledgments
The authors are grateful to F. Casino and C. Martucci for technical assistance
during the realization of the experimental set-up.

References
J. Homola, S.S. Yee, G. Gauglitz, Sens.Actuators B 1999, 54, 3-15.
K. L. Beattie, W. G. Beattie, L. Ming, S. L. Turner, R. Coral-Vazquez, D.D.
Smith; P.M. McIntre, D.D Dao,. Clin Chem.l995,41(5), 700-706.
V. Dugas, G. Depret , Y. Chevalier, X. Nesme, E. Souteryand, Sensors and
Actuator B: Chemical(2004) B 101(1-2), 112-121
T. Livache, E. Maillart, N. Lassalle, P. Mailley, B. Corso, P. Guedon, A.
Roget, Y.Lavy Journal of Pharmaceutical and Biomedical Analysis, 32
(2003) 687,
F. Caruso, E. Rodda, D. N. .Furlong, K., Niikura, Y. Okahata, Anal. Chem.
69 1997,2043-2049.
Christina Boozer, Qiuming yu, Shengfu Chen, Chi-Ying lee, Jiri Homola,
Sinclair S.Yee, Shaoyi Jiang. Sensors and Actuators B 90 2003, 22-30.
R. Rella, J. Spadavecchia, M.G. Manera, P. Siciliano, A. Santino, G, Mita.
Biosensors and Biolectronics, 20(2004) 1140-1148.
 LAB-ON-CHIP INTEGRATED GENETIC ANALYSIS

R.Aina, G.Barlocchi, M.Cattaneo, PCorona, A. Fischetti, M.Marchi,

U. Mastromatteo", M. Palmieri, M. Pesaturo, F. Villa, A. Vinay

STMicroelectronics, Via Tolomeo, 1- 20010 Comaredo (MI)

Abstract
STMicroelectronicshas leveraged its long-standing experience in microfluidics and silicon
integration of miniaturized components to address the grooving demand for fast, convenient
and low cost DNA analysis: the result is the In - Check product platform, the first
commercial chip that combines the capability of different common laboratory instruments
into a single, self-containedunit. The ST platform is based on a silicon chip that integrates all
the functions needed to identify given oligonucleotide sequences in a sample, including
microfluidic handling, a miniaturized PCR reactor and a custom microarray. The chip is
mounted on a plastic slide that provides the necessary mechanical, thermal, electrical and
fluidic connections. In addition to the slide itself a dedicated set of instruments is provided
usas outside the traditional lab setting can easily be trained and operate the platform to
generate molecular diagnostic information where it is most useful. Due to a fast time-to-
answer, portability and panel of probes, the In-Check platform is particularly suited for
diagnostic applications in which it is useful to simultaneously screen a number of sequences.
Additional pathogens can be added to a panel without any additional cost.

1. Introduction

Genetic assay have an enormous scope of applications in

biotecnology and medicine, ranging from agriculture and farming to
the detection of pathogens in foods to genetic diagnostics on human
subjects. Recently, there has been much interest in the
implementation of microfluidic devices for genetic tests. Genetic
analysis, in fact, are powerful technology drivers and excellent
candidate applications for miniaturization technologies because the
demand for inexpensive genetic information is essentially
unlimited, and the cost and time for the diagnostic decrease
with sample volume. In addition, since manufacturing techniques
are similar to those used for conventional integrated circuits it is

217
 218

possible to leverage four decades of know-how and use existing

production facilities; besides genetic assay can benefit from the
automation and control provided by miniature electronic devices.
To address the growing demand for fast, convenient and low cost
DNA analysis STMicroelectronics has leveraged its long-standing
experience in microfluidics and silicon-based integration of
miniaturized components: the result is the In-Check product platform,
the first commercial chip that combines the capabilities of different
common laboratory instruments into a single, self-contained unit.
Here in the following a full description of In-Check product platform
is given.

2. ST In -Check Platform Technology

The heart of In-Check platform is a silicon chip that integrates all

the functions needed to identify given oligonucleotide sequences in a
sample, including microfluidic handling, a miniaturized PCR reactor
consisting of a bank of twelve triangular channels, each 150-by-ZOO
pm, buried in silicon and a custom microarray. The chip is mounted
on a lx 3 plastic slide that provides the necessary mechanical,
thermal, electrical and fluidic connections. The chip is designed to
handle small sample volumes (2-8 pl), and does not require highly
trained personnel to be operated.
The PCR processor can be loaded directly and sealed to prevent
evaporation during temperature cycles. Acting as a fully integrated
PCR reactor, the chip allow users to run fully customized
amplification processes. All chemical reactions occur on the biochip
and, because the cartridge that carries it is disposable, this technique
is free of the cross-contamination risk of conventional
instruments.The In-Check platform, based on a silicon lab-on-chip
device, accelerates the PCR speed by at least 3 times, compared to
the same protocol running on a conventional thermocycler. Besides,
thanks to the outstanding temperature control of the PCR solution,
the In-Check platform ensures an extremely high specificity of the
amplified products compare to a conventional thermocycler.
 219

The surface in contact with the fluid are biocompatible and do not
inhibit polymerase.

Fig. 1 - Sample is loaded on a chip using a standard micropipette tip. On - Chip

PCR results are equivalent to thermocycler in less than 30% of the time and
on-board microarray delivers superior contrast and spot definition for
reliable data interpretation.

i~roarraythat is used
receive a pre-spotted
to perform a given analysis, or can alternatively spot oligos of their
choice. The detection is based on standard fluorescent labelling. Up
to 200 individual spots allow even some of the most complex
diagnostic applications to be placed on the microarray, including
control probes and duplicates for redundancy and statistical reliability
of data. Single nucleotide mismatches can be detected and will result
in a negative outcome on a given spot.

3. ST In - Check Platform Instruments

str ts

The platform Instruments include a Multiplex Thermal Control

System and a Compact Optical Reader.The first one is an integrated
 220

compact high precision erat ) which

drives and fully controls the In-Check chip. The TCS is a completely
independent modular system that manages up to five different tests
and protocols simultaneously.

Fig. 2- LC02 Disposable Card and Compact Optical Reader

Compared to a conventional thermocycler and hybrydization station,

the user can start the tests independently without having to wait to
collect the complete set of samples. Thanks to the ST microcontroller
based embedded architecture the temperature accuracy is within +/-
0.1%. An interesting feature permits to control the ramping speed
during the temperature transition to fully optimize PCR protocols.
The system is modular and can be scaled to accommodate the need of
both large diagnostic setting and point-of-care applications.
The second instrument is a Compact Optical Reader which detects
and analyzes the microarray in a few seconds. Autofocus, automatic
optics setting and microarray tool analysis make this instrument
particularly suited for non trained personnel and point-of-care
applications. At the same time, the advanced mode option allows to
change the hardware parameters and automatically collect the
complete set of spots data to satisfy the most demanding expert user.
 221

4. ST In - Check Platform: The Present Applications and

Next Evolution

Today due to a fast time-to-answer, portability and panel of probes,

the In-Check platform is particularly suited for diagnostic
applications in which it is useful to simultaneously screen a number
of sequences. Additional pathogens can be added to a panel without
any additional cost. Moreover, the simplification of operations
needed, the self-contained nature of the chip and the reduced
footprint of the instrumentation make it an ideal platform to develop
Point-of-care applications.
For the future the goal is to integrate in the current platform the
functional block for sample preparation. In order to include in a
silicon Lab On Chip this functional block it has been developed a
device based on dielectrophoresis (DEP) able to sort, isolate and
burst target cells from clinically relevant samples. Although the
physics may appear complex, the technology needed to accomplish
DEP is straightforward and relatively inexpensive.
Because the DEP force is proportional to two independent term& is
possible to exploit this versatility applying appropriate electrical
signals to the electrodes and so optimize each application in
accordance to the dielectric characteristic of the particles. The
developed device has been designed for the separation of white blood
cells in a diluited row blood sample and then lead these cells in an
electroporation section to perform lysis for the genetic information
extraction.

References

- K.B.Mullis et al. The Polymerase Chain Reaction Boston, MA:

" "

Birkhauser, 1994.

- G.Barlocchi, P. Corona, U. Mastromatteo, F.F.Villa : Silicon

"

Micromachining for Lab On Chip " .Sensors and Microsystems

5thItalian Conference - Lecce, Italy, 12-16 February 2000.
 Use of Cavitands as Selective Adsorbing Materials in
GC-like Air Quality Monitoring Microsystems
S. ZamDolli, I. Elmi, G.C. Cardinali, M. Severi
CNR - IMM Bologna, Italy, zampolli@bo. imm.cnr.it
E. Dalcanale, P. Betti
Dipartimento di Chimica Organica e Industriale, Universitri di Parma, Italy

Summary
The use of cavitands as innovative pre-concentration materials for outdoor air quality
monitoring applications was studied. With respect to conventional adsorbing materials,
cavitands feature lower desorption temperatures and a selective pre-concentration behaviour
due to the specificity of the host-guest interactions. These characteristics make them
particularly suitable for integration in simplified GC-like gas scnsing systems based on
micromachined components. The detection limit of a miniaturized GC system with a MOX
sensor used as detector will be reported.

1 Introduction
Lack of selectivity of Metal Oxide Semiconductor (MOX) gas sensors is one
of the major issues in applications where single compounds within complex
mixtures must be detected and quantified. Some of the compounds typically
found in air quality samples, like e.g. water vapour, though being harmless to
population, are present at high concentrations and can vary rapidly with time,
producing significant shifts in the response of MOX sensors. Other compounds,
like e.g. benzene, are toxic or even carcinogenic at ppb concentrations.
Optimization of sensing materials and sensor operating conditions to be used in
arrays often allows to reach the selectivity necessary for classification purposes,
but quantitative analyses of mixtures are generally not possible.
In previous works [ 1,2], the authors have shown how the use of miniaturized
silicon micromachined gas-chromatographic (GC) separation columns,
combined with innovative injection procedures, can be exploited to increase the
selectivity of MOX sensors for quantitative monitoring of ppb-level VOC
concentrations in indoor air samples.
For outdoor air quality monitoring applications, the concentrations of
benzene, toluene, ethylbenzene and xylene to be quantified are in the sub-ppb
range, at least one order of magnitude lower than for indoor air quality. In
classical GC analyses, purge-and-trap pre-concentration procedures are often
used to increase the quantity of pollutants injected into the analyzing systems,
allowing to reach the necessary sensitivities. However, the materials typically
used in purge-and-trap configurations (e.g. Tenax@) have several constraints
which do not allow their use in simplified micromachined GC systems, like e.g.

222
 223

the high temperatures necessary for desorption and trap purging and the need for
inert desorption carrier gases.
In this work, the use of quinoxaline-bridged cavitands as innovative,
selective pre-concentration materials for miniaturized GC-like applications is
proposed. Used in purge-and-trap configuration, benzene, toluene, ethylbenzene
and xylene are desorbed in air at temperatures lower than 100C, allowing for
the use of simplified injection procedures without the need for carrier gas
bottles. The selectivity of the pre-concentration eliminates non-aromatic
compounds from the sample mixture, enhancing the effectiveness of the GC
separation procedure. Furthermore, by optimizing the desorption kinetics,
cavitands are expected to provide selective release of the target analytes at
different temperatures, allowing for the combination of pre-concentration and
separation effects in a single device.

2 Experimental
2.1 Cavitands

For this work we have chosen a quinoxaline-bridged cavitand (QxCav) having a

cavity of 8.3 A depth which completely engulfs aromatic guests, providing
additional weak CH-A interactions with the cavity walls. In particular the
complexation properties of QxCav towards aromatic compounds have been
demonstrated both in solution [3] and in the gas phase [4].

Fig. 1: sketch andX-ray structure of QxCavC6(I)*PhF(side vim)

Cavitand 1 (see fig. 1) has been prepared following a published procedure [3].
To be used as a trap for purge and trap gas chromatography analysis, the
cavitand has been crystallized from acetone and the resulting crystalline powder
has been passed through 35-60 mesh sieves to obtain a material having
homogenous granulometry.
224

2.2 Characterization of cavitands

The experimental setup used for the characterization of the cavitands is

shown in figure 2. A gas distribution system comprising several mass flow
controllers connected to certified bottles provides a calibrated mixture of
aromatic volatile organic compounds at different concentrations eluted in
synthetic air, which is sampled by the cavitands in a purge and trap
configuration.
/ - - - - - - - -__ \Gas distiibution b ? s I s ( i

I
I Fur -
I
I [B]= 500 ppb -

Fig. 2: Characterization system usedfor cavitand testing

During sampling the cavitands are kept at room temperature. A MOX sensor
is mounted in a low-volume chamber and connected to the outlet of the cavitand
cartridge. The absorption efficiency is of the cavitands is measured by
comparing the sensor response in case of a cartridge filled with cavitands with
the response in case of an empty cartridge. After a certain sampling time (1 5 or
20 minutes), the cavitands are heated with a linear temperature ramp from room
temperature to 80C in 25 minutes, and a MOX sensor is used as detector of the
released volatiles.
In figure 3a the typical sensor response during absorption of benzene is
shown. In red the sensor response in case of an empty cartridge is reported,
representing the sensor response to the injection of 20 ppb of benzene for
15 minutes. The black plot represents the sensor response when the cartridge is
filled with 54 mg of QxCav in a mesh 35-60. The comparison of these responses
clearly shows the high absorption efficiency of the considered cavitands.
In figure 3b the sensor responses during the desorption temperature ramp are
compared. The black plot shows the sensor response to the compounds released
after a 20 minutes absorption of 20 ppb of benzene, while the red plot refers to
the release of a mixture of 5 ppb of toluene and 5 ppb of m-xylene.
As can be disclosed from these plots, in both cases the release has a
maximum peak at around 70"C, which is a temperature low enough to be easily
implemented in a miniaturized and portable purge and trap system. On the other
hand, the temperature at which the release occurs is very similar for all of the
considered compounds, therefore the selectivity needs to be enhanced for a
reliable quantification of the single aromatic compounds.
 225

a) b)8 80

8 -20 min [B]= 20 ppb 70

-20 min [TIXI = 10 p

6 60
6
h
50
;3i- 4 $ 4
v v

- - 40 I-

2 2 30
20
0 0
0 5 10 15 20 0 10 20 30
Time (min) Time (min)
Fig. 3: Cavitanh absorption eflciency (a) and desolption kinetics (5)

ncreae of selectivi~:GC system architecture

To increase the selectivity, a GC-like approach is proposed, as shown in

recent publications [ 1,2]. ARer the cavitands cartridge, a micromachined GC
separation column is inserted before the MOX sensor used as detector (fig. 4).
After sampling and release of the volatile organic compounds by the cavitands,
the pre-concentrated mixture is separated in time by the stationary phase of the
GC column.
G& diitribuliori systim

Fig. 4: Selectivity increase: GC separation column

In figure 5a) the chromatograms relative to different concentrations of

benzene, toluene and xylene sampled for 25 minutes are shown. While the short
GC column used in this case does not allow for a sufficient separation of the
toluene peak, the different peak heights for benzene concentrations as low as
1 ppb can be easily disclosed, as shown in the calibration curve in figure 5b.

3 Conclusions

The use of cavitands as selective pre-concentration materials in a

miniaturized GC like outdoor air quality monitoring was studied. The results
show the feasibility of ppb level benzene monitoring using a MOX sensor as
detector and a micromachined GC separation column for selectivity
226

enhancement. The absorption efficiency and desorption kinetics of cavitands are

compatible with the integration in a miniaturized system, and micromachining of
a cavitand purge and trap cartridge will be investigated. Furthermore the use of
new bridging groups for the cavitands will be studied, aiming at selective
desorption kinetics of the different aromatic compounds in order to reach a
temperature-selective
desorption technique without the necessity of a GC separation c o l m .
1 ' " " " ' ' "

b)
6-

3
v
/@
-c.
c
.-
4-
/@
a,
m ;
.c /"
2;/@
0 " " " " " "

eferences
[l] S. Zampolli, S. Nicoletti, I. Elmi, G.C. Cardianali, L. Dori, 9th Italian
Conference on Sensors and Microsystems AISEM 2004, Trento, Italy
[2] S. Zampolli, I. Elmi, J. Sti,irmam, S. Nicoletti, L. Dori, C.C. Cardinali,
Sensors and Actuators B, 105 (2005) 400-406
[3] P. Soncini, S. Bonsignore, E. Dalcanale and F. Ugozzoli, J. Org. Chem.
1992,57,4608
[4] M. Vincenti, E. Dalcanale, P. Soncini and G. Guglielmetti, J. Am. Chem.
Soc., 1990, 112,445

Acknowledgments

This work was hnded by the MIUR projects FISR "Microsistemi per
applicazioni ambientali ed agroalimentari" and FIRB RBNEO1 9 T W .
 Fast optical detection of chemical substances in integrated
silicon-glass chip

L. DE STEFANO, I. REA, L. ROTIROTI, K. MALECIU AND I.

RENDINA
Institute for Microelectronics and Microsystems - National Council of
Research
Via P. Castellino I l l - 80131 Napoli

L. MORETTI, F. G. DELLA CORTE

DIMET Mediterranea University of Reggio Calabria

Localith Feo di Vito, Reggio Calabria, Italy

lzrca.c/e.sfcfbno@,na.imm.c171:it

Summary
In this communication, a silicon-glass microfluidic chip for the realization of optical sensing
components in lab-on-chip applications has been demonstrated. The technology of Anodic Bonding,
of great importance for the integration of microfluidic functions, has been used for the fabrication of
chemical microsensors. A cross injection valve has been integrated with the porous silicon optical
transducer on the prefabricated glass wafer with microchannels, so ensuring sample injection into the
stream of carrier gas. A pneumatic actuation has been used for the valve control.
The porous silicon (Psi) is the optical transducer material. Due to its porous structure it has a very
large specific surface area of the order of 200-500 mz ~ r n . ~so, that an effective interaction with
several substances is assured. The optical read out of the Psi layer has been realized by means of an
external optical system. The obtained preliminary results show that very short response time allow to
classify this simple analytic opto-chemical chip in the area of advanced sensors for chemical
analyzers

Introduction
There is a compelling need for fully integrated and networked
environmental sensor solutions that provide for both early warning type
detectors, coupled with sample collection, and confirmatory detectors that can
provide accurate species identification. In addition to environmental sensors,
the development of easily deployable medical instruments capable of
determining whether an individual has been exposed to a biological pathogen
and the specific nature of the pathogen are equally critical to the health
monitoring. Micro-opto-electrical-mechanical systems (MOEMS) and micro-
total-analysis systems have unique features fulfilling these critical needs. The
fast and intensive development of lab-on-chip for sensing application is due to
several factors essential for routine analytical determinations, such as very
limited sample consumption and the short analysis time which can be obtained
by measuring a transient signal in a flow-through detector.

227
228

The AI3 technique easily allows the encapsulation and sealing of three-
dimensional microfluidic structures, such as channels, chambers, cavities and
other complex gas or liquid routes on silicon chips [l-21. Additionally, glass
transparency at optical wavelengths enables simple, but highly accurate,
alignment of pre-patterned or structured glass and silicon wafers.
On the other hand, porous silicon (Psi) can be employed as a transducer
element due to its sponge-like nanostructure: it is able to effectively interact
with several chemicals and biological molecules. On exposure to bio-chemical
substances, several physical quantities of the Psi, such as refractive index,
photoluminescence, and electrical conductivity, change drastically. In
environmental monitoring applications, optical read-out techniques are of
particular interest mainly because they do not require electric contacts that may
cause explosions or fire in dangerous environment, and also allow wireless
remote interrogation. We have recently proposed several Psi optical sensors
with monolayer as well as multilayer structure, based on the refractive index
change induced in different ways by chemicals (such as hydrocarbons,
flammable substances, pesticides, etc.) and biological species (DNA and
enzymes) [3-51.
In this work we have joined some typical microelectronic fabrication
technologies together with the Psi transducer features and the microfluidic
injection of the analytes to realize a new microminiaturized opto-chemical
analyzer. The Flow Injection technique is commonly exploited to perform
different chemical and biological diagnosis in a wide range of sensing
applications. On the other hand, porous silicon is receiving great attention due
its sponge-like structure allowing a very effective interaction with several
chemicals and biological molecules. Moreover, we have characterized the
sensor dynamic by time-resolved measurements on exposure to several volatile
organic compounds.

Experimental results
The scheme of the complete optical microsystem and the read-out experimental
set-up is shown in Figure 1.
The reaction micro chamber with the Psi on the bottom was produced by a two-
step electrochemical etch of a <loo> silicon wafer, p+ type, using an HF/EtOH
(30:70) solution. The first step creates a thick sacrificial layer of high porosity
Psi which has been completely removed by rinsing the samples in a NaOH
solution. The result is an empty chamber into the silicon substrate with depth of
150 pm.
 229

Fig. I Scheme of the complete optical microsystem and of the read-out experimental set-up.

--+
Waste

Fig. 2 Scheme of thejlow injection analysis principle.

Since the diameter of the porous silicon area is always about I cm, the total
volume of the analysis chamber is 12 pL. By using a fresh etching HFEtOH
(30:70) solution, without moving the chip from the electrochemical cell, a
second etch step (150 mA/cm2 for 45 s) produces Psi monolayer on the bottom
of approximately 6 pm thickness and a porosity of 70 %.
A Borofloat 33 type glass has been chosen as a chip cover. In order to feed gas
or liquid substances, inlet and outlet channels have been mechanically drilled in
the top glass wafer, on the opposite sides of the porous region.
In Figure 2 the principle of the flow injection analysis, which is a widespread
microfluidic technique in the lab-on-chip world, is shown: a continuous flow of
nitrogen carriers the analyte from the tank source into the Psi microchamber
which acts both as the reaction chamber, since it allows the capillary
condensation of the analyte, and as the detector, since it allows the optical
transduction of the gas-material interaction. A pneumatic actuator has been used
for the injection valve control. We have optimised the frequency of the valve
opening (100 mHz) and the pressure of the carrier searching for the fastest time
response of the sensor device.
We have performed time-resolved measurement in order to characterize the
sensor dynamic: using the laser beam from an IR source, we have measured, as a
function of time, the signal of a receiving photodetector before, during and after
the exposure to some organic volatile compounds (Isopropanol, Ethanol and
230

Methanol). On exposure to the vapows, due to the phenomenon of capillary

condensation, the average refractive index of the layer increases, and, as a
consequence the optical thickness of the porous silicon layer also increases. The
result of a time-resolved measurement is shown in Figure 3: in the case of
Isopropanol, we have obtained a response time to the solvent presence which is
156 ms, while the signal returns to its original value in shorter time but always in
the order of milliseconds (qeC=24ms).
On exposure to Isopropanol

-Carrier burst
-- -PSI signal
0-
.. __ . . - n4 . .. . . ... . .
8-
I
I T,*,p = 156 rns
I
6- I I T,*== 24 ms
I I
I I
4- I I
I I
I I
2- 1

,,/.;;.
I
. . c .
.:-.>,L.lJ' 1 ,,;'I?*
0 7 . . . I . ,- , ' .
00 05 1.0 15 20 2 5
Time (sec)

Fig. 3 Time resolved measurement in case of Isopropanol. The optical response and the time
reference of the carrier have been shiftedfor the sake of clarity.

The response time depends not only on the physical phenomena involved (i.e.,
equilibrium between adsorption and desorption of each substances in the Psi
layer) but also on the geometry of the test chamber and on the measurement
procedure. In the same experimental way, we have verified the stability and
repeatability of the sensor on several pulses of the gas analyte (Fig. 4)

On exposure to Isopropanol
12

0 2 4 6 8 0
Time (sac)
Fig. 4 Repeatability and reversibillty of the gas sensing measurements
 231

Conclusions
The integration of a PSI sensor element and the micromechanical
technologies leaded to a compact microsystem capable to discriminate very
small amount of pure organic compound. A very fast response signal has been
observed when a short injection time has been applied to introduce solvents
vapors into the system. We have characterized the sensor dynamic by time-
resolved measurement during exposure to Isopropanol. The data reveal fast
analysis time and the complete reversibility of the sensing mechanism.

References
[l] B. Puers, E. Peeters, A Van Den Bossche, A. Sansen, Sens. Actuat. A, 21-
23 (1990), pp 108-114.
[2] Ljibisa Ristic, Sensors Technology and Devices, pp. 207, 1994.
[3] L. De Stefano, I. Rendina, L. Moretti, A.M. Rossi, S. Tundo, Applied
Optics, 43/1, 167-172,2004.
[4] L. De Stefano, I. Rendina, L. Moretti, A.M. Rossi, A. Lamberti, P. Arcari,
0. Longo, M. Rocchia, IEEE Trans. on Nanotech. 3/1,49-54,2004.
[5] L. Rotiroti, L. De Stefano, L. Moretti, A. Piccolo, I. Rendina, A. M. Rossi,
Biosen. Bioelec. 20/10,2136-2139, 2005.
 CMOS SINGLE PHOTON AVALANCHE DIODE FOR
IMAGING APPLICATIONS

M.Scandiuzzo,D. Stoppa, A. Simoni

ITC-irst, via Sommarive, 18, 38050 Povo di Trento, Italy
scandiuzzo@ itc.it

L. Pancheri, G.-F. Dalla Betta

DIT, Universitir di Trento, Via Sommarive 14, 38050 Povo di Trento, Italy

Summary
This paper reports on a novel distance sensor employing an avalanche photodiode (APD)
operating in a single photon counting configuration for Time-Of-Flight 3D imaging
applications. An active pixel containing a single photon detector and a read-out channel has
been designed. A 64 pixel linear array has been fabricated in a conventional industrial high-
voltage 0.8-pm CMOS technology, together with an APD test structure. The pixel size is
38x180 pm2 and the power consumption is 15pW. A characterization of the sensor is
reported.

1 Introduction
Recently, three dimentional (3D) measuring systems, known also as range
finders, have emerged in many application fields, like safety, surveillance and
robotic guidance, where specific patterns obstacles and moving elements should
be recognized in real time by using small and cheap systems. Several scannerless
range finders based on the time of Flight (TOF)technique have been recently
proposed [1,2,3,4]. In order to employ the direct TOF with accuracies in the cm
range, high detector sensitivities and picosecond time discrimination capability
are needed. Due to its high sensitivity, a Single Photon Avalanche Diode (SPAD)
can be a suitable device for direct TOF range finding.
A SPAD is basically an APD biased beyond its breakdown voltage to be
operated in the Geiger mode. In this configuration, single photon detection can be
achieved, provided that a suitable quenching mechanism is implemented. The
feasibility of SPADs in conventional CMOS and BiCMOS technology, as
demonstrated in the past few years [5-61, opens the way to the realization of low-
cost and high-performance fully integrated systems for high sensitivity imaging.
In this paper a pixel which includes a SPAD and a readout circuit for photon
arrival time estimation is proposed. The pixel contains a SPAD, a quenching

232
 233

circuit, a comparator, an arrival-time estimation circuit and an averaging circuit

to increase the distance measurement resolution.
In order to validate the operation principle, a test chip consisting of 64-pixel
linear array has been fabricated in a high-voltage 0.8ym CMOS technology. An
electro-optical characterization of the test structures and of the array has been
carried out and is presented here. The sub-Geiger responsivity and the dark count
rate in the Geiger mode have been measured, assuring the suitability of the
device for the application here proposed. Moreover experimental results of 3D
images are reported with a precision better than 2mm with only 100 frames
averaging.

2. Pixel Architecture

The block diagram of the proposed pixel is sketched in Fig. la. It consists of
a SPAD, a quenching resistor %uench to limit the avalanche current, a voltage
comparator for the avalanche event detection and an analog time counter, which
measures the arrival time of the received light pulse. To reduce the influence of
jitter noise and to limit the effect of thermally-generated spurious events, the
measurement is repeated many times and the arrival-time information is averaged
by means of two circuits dedicated to the count of low and high numbers of
events, respectively. In this way the time resolution of the system is increased by
a factor Npln, where N, is the number of pulses spread on the scene, but the
overall measurement time is kept low because the read-out of the pixel array is
not performed at every measurement cycle. This solution becomes of paramount
importance as the number of pixels increases.
To avoid the use of high-voltage transistors, expensive in terms of area
occupation, the cathode of the SPAD is connected through the quenching resistor
to Vdd,and the biasing above the breakdown voltage IVbl=27V of the SPAD is
assured by means of the external line VSpd.biased at a very negative voltage
(about -25 V).

3. SPAD

A schematic cross section of the fabricated Avalanche Photodiodes is shown

in Fig. lb. In order to avoid edge breakdown, a guard-ring surrounding the p+
implantation has been implemented using a p-tub layer inside a deep ntub
available in this fabrication processes. The active area is defined by means of an
optical window opened in the metal light shield only in correspondence with the
region where avalanche multiplication occurs. The SPADs implemented in the
pixels have a circular active area with a diameter of 5 pn.
234

Photon Arrival-Time

4 .....-- ---.__
-I_____.-.--------_
--..-...-.
(b) .,/
.
-
I

metal2 Active
-. Illera?. %,*

4'
.' shield area shield *- *.

;"' c thode"',,

Fig. I . (a)Pixel block diagram and (b)APD cross section

An APD test structure, having a circular active area with 100 pn diameter,
has been fabricated in the same chip, in order to test the characteristics of the
device. An electro-optical characterization of the APD test structure has been
carried out. The device has a breakdown voltage of 26.8 V at ambient
temperature. The spectral responsivity has been measured at room temperature
and is shown in Fig. 2 at OV bias and with a reverse bias just below breakdown
voltage. The multiplication gain is about 15 with a 26 V reverse bias voltage at
the light wavelength A= 550 nm
In order to measure the dark count rate, the APD has been biased with a 200
kOhm quenching resistor, thus obtaining a passively quenched SPAD with about
1 ps dead time. The dark count rate was measured at different temperatures and
different excess bias voltages, and its value normalized for the area of the SPAD
included in the pixels is shown in Fig. 3.The dark count rate approximately
doubles every 10 "C, indicating an SRH dominated dark count.
 235

--- VB'OV
-Vf2b V

0
Y)
c
0

:: 01
e

0 01
3
400 500 600 700 8110 900

Wavelength [nm]

Fig. 3. Dark count rate as afunctron of the

Fig. 2. APD spectral responsiviw
temperaturef o r dfferent excess bias voltages

The reported results indicate that the 5-pm diameter SPAD included in the
pixel should have a dark count rate of about 100 Hz at room temperature with 3
V excess bias voltage. This is a very good value, since in a measuring time-
window, some tens of nanoseconds long, the probability of having spurious
events caused by thermally generated dark counts will be very low.

4. Experimental Results

The distance measurement of a cooperative target in the 2m-5m range,

obtained after pixel calibration, exhibiting a precision better than _+0.75%.
This value could be further improved, at the expense of a reduced frame rate, if
an external averaging operation is implemented. Extrapolating the achieved
experimental results for a 64x64-pixel array would give a resolution of lmm with
a real time operation (3Ofps) capability.
Preliminary examples of 3D scene acquired through a vertical scanning (64
frames), and illuminated using an uncolimmated, 905-nm, 100-ns pulsed laser
source having a mean power of about 250mW, is illustrated in Fig. 4. The insets
(a-b) shown the intensity map (2D image), while the measured 3-D data, coded
both in the z-axis and in grey-level intensity, are represented in the insets (c-d)
with two different perspective views. The two objects are placed in front of plane
at 50cm from the sensor. It is worth noting that the resolution of the reported 3D
images is mainly limited by the preliminary optical setup.
236

Fig. 4: Examples of2D and 3 0 images.

Conclusions

A novel pixel architecture has been presented, which includes a SPAD and a
read-out channel in the same pixel. Besides the SPAD, a dedicated quenching
circuit, a voltage comparator for the avalanche detection, a time-to-voltage
converter and a filtering stage for the extraction of the mean time value are
integrated in each pixel. Experimental results confirm the validity of the design
approach. The measurement of non-cooperative targets has been successfully
performed with a precision of some millimetres and a pixel-to-pixel uniformity
of few centimetres mainly limited by the optical setup used for preliminary tests

References

R. Lange, P. Seitz, IEEE Journal of Quantum Electronics, 37 (2001), 390

D. Stoppa, L. Viarani, A. Simoni, L. Gonzo, M. Malfatti, G. Pedretti, Proc. of European Solid-
State Circuit Conference, (2004), 419.
R. Jeremias, W. Brockherde, G. Doemens, B. Hosticka, L. Listl, P. Mengel, ISSCC Digest of
Technical papers, (Feb 2001), 252.
C. L. Nicfass, A. Rochas, P. A. Besse, E. Charbon, ISSCC Digest of Technical papers, (Feb.
2004), 120.
A. Rochas, A. R. Paucbard, P.A. Besse, D. Pantic, Z. Prijic and R. S. Popovic, IEEE Trans. on
Electron Devices, 49 (2002). 387.
A. Biber, P. Seitz, H. Jackel, IEEE Trans. on Electron Devices, 47 (ZOOO), 2241.
Design of a MEMS Test-Structure for Stress Characterization
by Capacitive Read-out

Marco Bellei, Roberto Gaddi, Antonio Gnudi

ARCES-DEIS University of Bologna, vide Risorgimento 2, 40136 Bologna, Italy
mbcllei@,dcis.unibo.it

Summary
A MEMS (Micro-ElectroMechanical System) test-structure for the characterization of the
planar stress induced in structural layer deposition of a surface micromachining process is
here presented. A classical rotating needle design has been adopted and extended with
electrical capacitive transduction of the rotation angle, in order to allow automated
measurements. Moreover, stress extraction is performed by digital fashion, leading to a
potential improved robustness versus capacitive parasitics effects. Electromechanical design
aspects and preliminary results obtained by finite element simulations are discussed. An
average device resolution of f 1 8 MPa has been obtained for the extracted planar stress in a
gold structural layer.

1 Introduction
It is well-known from planar microelectronics that process steps induce a
mechanical stress in deposited layers. Although the film deposition should be
stress-controlled for all the layers, in a micromachining process the knowledge of
the in-plane stress induced by fabrication is a key aspect especially in structural
parts, such as suspended plates and beams. The parasitic effects of an unexpected
stress can affect the device performances and sometimes dramatically impair its
electromechanical behavior. Several characterization techniques have been
proposed. Electrical measurements could be generally less accurate than their
alternative counterparts [1,2,3,4,5], but they do not need non standard
instrumentations and they are suitable for automated measurements, if an
automatic prober is available. Among electrical techniques, three challenging
categories have been proposed: electrostatically actuated microstructures [6, 71;
resonating beams (the operation principle of which can be found in [S]); comb-
drive microstructures with in-plane displacement and analog capacitive read-out
[9]. These methodologies exhibit potential drawbacks due to a quite complex
modeling for achieving high accuracy and they are sensitive to process
parameters. In addition, the first and the second can be exploited for tensile stress
only, while the third extracts the stress in an analog fashion by transducing the
strain into small capacitance variations. Moreover, since it exploits sidewall
capacitances, it generally requires fabrication process with high-aspect ratio and
the structural parts could be sensitive to processes feturing high residual stress
gradients.

237
238

In this work we would like to introduce a novel methodology for stress estimation
based on electrical measurements by avoiding the above-mentioned potential
problems. The more interesting potential benefit of the presented test-structure is
the digital read-out since good accuracy and precision in the measured
capacitance values are not relevant: the fundamental requirement is to
discriminate about a threshold value, as in digital electronic circuits.
From a mechanical point of view, it should be noted that the presented device is
properly a strain sensor. Nevertheless, since we assume the knowledge of the
strain-stress relation under elastic regime, generally we will refer directly to the
stress as the unknown parameter.

2 Design
esi aspects
ects
We present a test-structure design in which the stress-induced rotation of a
classical lancet or needle [ 101 is measured by digital capacitive detection instead
of traditional optical inspection. Figure 1 shows the layout image of the device.
After the release step, the residual stress in the suspended structure will induce a
strain in the stress beams and thus a clockwise (negative) or counter-clockwise
(positive) rotation for compressive or tensile stress respectively. The relationship
between rotation angle and planar stress can be extracted by employing a non-
linear finite element analysis (FEA) or, for rough estimation, by analytical
physical models based on Hookes law.

Figure1 Layout image of the designed test-structure @lack: suspended structure; grey: probe pads
and underlying electrodes).
 239

100 3.5E-07
80
3.OE-07 -
60
40 2.5E-07 8
cd

Q)
20 2.OE-07 2E
m o
5 1.5E-07
Y

p
d -20 U
0 .3

,$
.3

U
-40 1.0E-07
2 -60 5
5.0E-08
-80
-100 O.OE+OO
-250 -150 -50 50 150 250
planar stress [MPa]
Figure 2 Mechanical response of the test-structure and its derivative. ParabolicPtting functions
are also shown bothfor tensile (yt) and compressive (yc) stresses.

Figure 2 shows the above-mentioned relationship together with its derivative.

The decreasing derivative magnitude for increasing values of the stress
magnitude is mainly due to the non-ideal hinges which connect the stress beams
to the suspended needle. Its stiffness induces an unwanted increased bending of
the stress beams for increasing rotations. Thus, the total force induced by stress
will have a component also in the lateral direction, which does not contribute to
rotation. The structural layer is made of gold and we assume a nominal Youngs
modulus of 80 GPa and Poissons ratio of 0.42. For further aspects on the
fabrication process, refer to [ l l ] . The stress beams are 265 pm long and 10 pm
wide.
The rotation angle has been quantized thanks to N radial adjacent electrodes
alternatively placed under the suspended needle on the left and on the right, as
shown in Figure 1. For lowering the quantization error, the electrodes width and
distance must be chosen at the minimum size allowed by the fabrication process.
By taking into account the maximum rotation angle of f 8 0 d a d consequently
obtained given the stress range of interest, a number N=9 can be used. Thus, the
discretized rotation angle can be obtained by measuring the nine capacitances
between the G pad (connected to the needle) and each of the pads (0, lP, ...,4,
1M, . .., 4 M } connected to the read-out electrodes. These capacitances assume a
high (H) value for the electrodes located just underneath the needle and a low (L)
value for all remaining pads. If a threshold between H and L capacitance values
is defined as C~=O.25xC,,,, where C,, is the maximum capacitance
corresponding to the perfect superposition between the suspended needle and a
read-out electrode, either one or at most two measured capacitances can assume
240

the H-state. In the former situation the suspended needle is basically overlapped
with one electrode only, while in the latter it features a partial overlap with two
adjacend electrodes (left- and right-side placed respectively). In this preliminary
design, the capacitance has been modeled as a parallel plate capacitor. C,, and
Ch must be measured by appropriate reference capacitive structures. After device
fabrication, a more efficient methodology for L-H discrimination could be
studied and thus adopted.
In order to obtain the best swing between L and H value, a larger capacitance
signal in H-state can be obtained by removing the extra air-gap through
electrostatic actuation of the suspended structure. For this purpose, two actuation
electrodes are introduced (see A pad in Fig. 1). In this situation, a C , value of
300 fF has been obtained by FEA. Moreover, we observe that this operation
becomes necessary if the suspended layer is affected by high positive stress
gradient, since the distance between the long needles and the underlying
electrodes could lower the capacitive signal below the parasitics threshold. By
finite element electromechanical simulations on the actuated structure affected by
stress gradient up to 5 MPdpm, we observed a C,,, value down to 90 fF,which
is widely acceptable.

3 Resolution and read-out table

In the considered stress range, the corresponding rotation angle range is nearly
80+80=160 mRad (see fig. 2). Since 8 (+1 for zero stress) electrodes have been
used for discretization and partial overlap of two electrodes is also exploited, the
angle resolution is 160/(8x2) =10 d a d . By employing the approximated linear
characteristic, an average value of nearly f 1 8 MPa can be estimated.
By taking into account the capacitance-discretized angle configurations and the
stress-angle non-linear relationship 8=8(0), a truthhead-out table can be obtained
(Table 3). Since the derivative of the mechanical characteristic is not constant, it
is not possible to express the resolution in a compact form and thus stress
corresponding to rotation angles has been written as a set of intervals.
Finally, by measuring the nine capacitance values on the read-out pads, we
expect to find only one of the considered configurations which corresponds to the
unknown planar tensile stress within a certain error.
The maximum rotation angle is imposed by the expected stress magnitude range,
with a theoretical limit value here assumed to be approximately [-250, 2501 MPa.
Although the stress range observed by wafer curvature is smaller for the gold-
alloy deposited by previous electroplating setup of the process [12], a potential
wider range could be observed through this structure, if required.
 241

L L L L L L H H L 50h5 138+167
L L L L L L L H L 60*5 167~195
L L L L L L L H H 70+5 195i222
L L L L L L L L H 80*5 2221248

Table 3 Truthhead-out table ofthe designed test-structure for the case of tensile stress.

4 Conclusions

A test-structure design for residual tensile/compressive planar stress

characterization by electrical measurements has been presented. The potential
benefits are the robustness of digital extraction, the electromechanical
model/design simplicity and thus the reduced sensitivity to technology
parameters. Moreover, the possibility of optical inspection can be also exploited.
The main drawback is the intrinsic low resolution. Nonetheless, a more complex
design based on mechanical amplifications [13,14] can lead to a better resolution.
In addition, whenever data on plastic regime will be available after a full
characterization of the gold layer, a new study including non-elastic deformation
should be performed.

References
[I] H.Guckel, etAl., IEEE Trans. on Electron Dev., 35 n. 6 (1988), 800
[2] W.Fang, J.A.Wickert , J. Micromech. Microeng., 6 (1996), 301
[3] Y.-H.Min, Y .-K.Kim, , J. Micromech. Microeng., 10 (2000) , 3 14
[4] M.W.Denhoff, J. Micromech Microeng., 13 (2003) ,686
[5] X.Y.Ye et Al., Sensors and Actuators A, 54 (1996),750
[6] K.Najafi, K.Suzuki, IEEE Int Workshop on MEMS '89 (Jan. 1989), 96
[7] P.M.Osterberg, S.D.Sentuna, J. of Microelectromechanical Systems, 6 n.2 (1997), 107
[8] A.Prak, et Al., J. of Microelectromechanical Systems, 1 (1992), 179
[9] L.L.Chu et Al., J. of Microelectromechanical Systems, 11 n.5 (2002), 489
[lo] L.B.Wilner, E E E Solid State Sens. and Act. Workshop (June 1992), 76
[I I ] F.Giacomozzi et Al., Workshop on Millimeterwave Comm. MEMSWAVE (July 2004), C25
[12] B.Margesin et Al., Proc. of DTIP of MEMS&MOEMS (May 2003), 402
[I31 Y.B.Gianchandani, J.Najafi, J. of Microelectromechanical Systems, 5 n.l (1996), 52
[I41 A.Bagolini, B.Margesin, et Al., Sensors and Actuators A, 115(2004), 494
 Controlling sensing properties of composite polymers using
different conductive nanoparticles

Anna Castaldo, Ettore Massera, Luigi Quercia, Girolamo Di Francia

ENEA Centro di ricerche di Portici Via vecchio Macello, Loc. Granatello, Portici (Na)
80055
[email protected]

Summary
We report on a new class of polymeric sensors based on a properly selected
polysilsesquioxane filled with different conductive powders obtained by ball milling These
sensors exhibit excellent sensing properties with respect to both polar and apolar analytes
and do not base their operating mechanism on the swelling. In fact, exposure to water
molecules or hexane vapours determines an increase of conductivity. In particular the
response in the full RH% range for devices obtained with different filler (Zn-Graphite, Zn,
Cu) is of four magnitude orders variation between 0% and 100% RH in the conductivity of
the sensors.

1 Introduction
In the past many papers have appeared in literature describing the use of
electronic nose based on array of sensors, each one showing low selectivity,
used to characterize analyte mixtures [ 11. Often in such devices polymer-carbon
black composites have played a very important role in spite of their simple
manufacture and cheapness [ 2 ] .The mechanism of response in polymer carbon-
black composite sensing films is, at its simplest level, based on film swelling.
Vapours sorbed in the film cause the film to swell, disrupting conductive
pathways by pushing carbon particles apart, thus increasing the electric
resistance of the film itself. For the first polymers used in electronic nose
applications, chemical functionality was very rarely considered. Clearly,
polymer selection greatly depends on the suite of target compounds. If target
compounds represent a broad distribution of chemical functionalities, then a
broad distribution of polymers should be necessary. Unfortunately the use of a
single filler, i.e. carbon black, could preclude the possibility to treat given
classes of polymers, due to possible incompatibilities with solvents and
dispersing agents for that kind of filler. For this reason we are at present
investigating various conductive fillers, e.g. noble metals or alloys intermetallic

242
 243

compounds, whose processing is compatible with polymers, designed to sense

specific analytes, as thioles, or amines.
In this work we consider various different nanometric fillers obtained by
mechanical ball milling (BM), starting from graphite, copper, silicon [3] and
their alloys. It is important to note that BM permits to obtain nanocrystalline
metal alloys with properties that are undoubtedly quite different from
equilibrium intermetallics. Moreover, nanoscale in the sensors is known to have
an high impact on the performance [4]. Polymers investigated as matrix in this
study belong to an emerging class of three-dimensional polyhedral silicon
polymers, named polysilsesquioxanes, containing nanosized cages. Such hybrid
amphiphilic materials with nano-size interfaces between organic and inorganic
domains may offer good opportunities to create composites with unique
properties [ 5 ] , [ 6 ] , [ 7 ] . In particular in this study we use the Poly
[(propylmethacryl-heptaisobutyl-POSS)-co-(n-butylmethacrylate)] shown in
Figure 1, whose functionality classification in base to a definition of Ryan and
colleague reported in a NASA bulletin on composite for electronic-nose [S] is
WD-HBB. Heptaisobutyl and butyl chains grant for a weak dipolar (WD)
character, while siloxane and methacrylic group furnish hydrogen-bond basic
(HBB) properties. This, in general, makes such a polymer a versatile matrix for
polymer composite sensors and, as a matter of fact, excellent sensing properties
of a polysilsesquioxane with respect to hydrogen and methane have been already
reported [9]. Here, we have investigated the effects of alternative fillers in the
same POSS polymer matrix. In particular nano-fillers have been fabricated by
ball milling starting from both n-type and p-type doped crystalline silicon also as
graphite-silicon alloys while as far as metallic fillers are concerned, we have
already obtained interesting results with copper and zinc submicrometric
conductive powders that we have tested vs ammonia and apolar vapours.

3 Results and Discussion

The incorporation of the polyhedral oligomeric silsesquioxanes or POSS into
polymeric matrices by covalent bonds has received a considerable amount of
attention. POSS have a compact hybrid structure with an inorganic core made up
of silicon and oxygen (SiO 1.5)" externally surrounded by organic tails. The
interaction between the organic tails and the polymeric matrix controls the POSS
degree of dispersion in the medium, i.e. its compatibility and thus the final
properties. The feasibility of controlling the arm number, arm length, the arm
functionality makes POSS topologically ideal for the preparation of
nanocomposite materials. Hybrid materials, second generation composite, have
excellent properties, high porosity, very good thermal and mechanical
properties, low dielectric constant. Promising candidates for sensor materials, in
our opinion, are methacrylate based POSS, versatile matrix for the sensing of
244

both polar and apolar analytes. On the base of the above reported physico-
chemical properties we assumed as starting hypothesis for this work that a
similar polymer could swell when exposed to analytes while selectivity could be
conferred to the sensor by dispersing in the polymer secondary fillers home
made by ball milling techniques, tailored for the best dispersion in the polymer
itself.

i ~ ~ . c -ocyccy)zcHH,
. ; t t ~ C ~ - ~

0 "I
, 0
Y

Fig. I Poly ((propylmethacryl-hept~isobulyl-POSS)-co-~n-bu~lmet~~ry~te)]

We excluded carbon black for the difficulty to obtain good dispersion in the
polymer solvent and worked to synthesize carbonaceous nanofillers, like
graphitic powder or nanostructured metals like zinc and copper. Figure 2 shows
EM images of some of the fillers used in this study. They are graphite, zinc,
copper-graphite. We have synthesized also silicon, silicon-graphite, zinc-
graphite submicrometric powders.

Fig.2 Graphite, zinc, copper-graphite.

Several PO based nanocomposites with a graphite content ranging between

10% to 20 % in weight were obtained and sensor devices were fabricated by
means of spinning deposition on Alumina substrates with interdigitated Au
contacts. In Fig.3 the response to water vapour in the 30%-70% range of relative
humidity of the device based on a 20% graphite nanocomposite film is reported.
 245

As it is possible to observe, the device response is quite high and the sensor
rapidly and reversibly follows RH% variations, increasing its conductivity in
presence of water molecules.
This unusual behaviour suggests that sensor device does not work on the
swelling mechanism (decrease of conductivity in presence of the analyte). It is
important to note that we also tested the POSS without any fillers and even in
this case we observed a small increase of its conductivity in presence of water
molecules. This furthermore points towards a more complicated mechanism
involving POSS cages. In our opinion secondary fillers ameliorate the charge
transfer between the cages as it is suggested by the fact thet we note that even
with metallic filler performance with respect to relative humidity is very good
(see Fig.4).

I I
I4 15 16 17 18 19 20 21 22

Time hours

Fip. 3 Device resvonse to RH of the pravhite filled POSS

Another interesting aspect of POSS based device is the very large amount
of conductivity variation. In Fig. 4 is reported the response in the full RH%
range for devices obtained with different filler (Zn-Graphite, Zn, Cu) where the
four magnitude orders variation between 0% and 100% RH in the conductivity
of the sensors is shown. Sensors devices have been also tested in presence of
apolar vapours, like hexane, to verify the versatility of the matrix. In Fig. 5
response to hexane in anhydrous ambient of different devices is reported.

o i , . ,
o r n m m w m
RH%

Fig 4 Peiformance in the humidity sensing of devices in logarithmic scale.

246

It is possible to observe, even in this case the increase of the conductivity in

presence of the analyte.

Rswda all'ssano

1 , . , , , , , I
10 m YI 40 50
x-- ?...."I

Fig.5 Response to hexane of various devices.

Work is in progress to comprehend operating mechanism of such new class of

polymeric sensors.
4 Conclusions
In this work we present polymeric nanocomposite sensors, based on a POSS
matrix properly selected and nanostructured compatible fillers, that exhibit
excellent sensing properties and could be used in the sensing of both polar
and apolar analytes. In particular we obtain new RH% devices with different
filler (Zn-Graphite, Zn, Cu), whose conductivity varies of four magnitude
orders between 0% and 100% RH.

References
[l IDeisingh A.K., Stone D.C., Thompson M., 1ntem.J. Food Science and Tech. 39,
(2004) 587-604.
[2] Quercia L.; Loffredo F.; Di Francia G.; Sensors and Actuators B, 109, (2005),153.
[3] Castaldo A., Di Francia G., Massera E., Quercia L., Fameli G., Delli Veneri P., Book
of Abstracts AlSEM 2005 p. 53.
[4] Di Francia G., Quercia L., Rea I., Maddalena P, Lettieri S., Sensors and Actuators B,
111-112,(2005), 117-124.
[5] Patel RR, Mohanraj, Pittmann CU, Journal of Polymer Science Part B Polymer
Physics 44 (1) (2006) 234-248.
[6] Striolo A., McCabe C, Cummings PT, Journal of Physical Chemistry B 109 (30),
(2005), 14300-14307.
[7] Zheng L., Waddon A. J, Farris R. J., Bryan Coughlin E. Macromolecules, 35, (2005)
2375-2379.
[8] Ryan M. A. Shevade, A.V., Zhou H., Homer M.L. NASA Bulletin oct.2004 714-719.
[9] Fujimoto, C. H.; Hickner, M. A.; Cornelius, C. J.; Loy, D .A. Macromolecules (2005)
in Press.
 BIOSENSORS BASED ON NAN0 ELECTRODE
ENSEMBLE AND SCREEN PRINTED SUBSTRATES

W.VASTARELLA", M.DE LEO^, J.MALY"", A.MASCI",

L.M.MORETTO~,R.PILLOTON"*
a*ENEA,SP061, Via Anguillarese 301, 00060 Santa Maria di Galeria, Rome (I)
Dipartimento di Chimica Fisica, UniversitG Ca ' Foscari, Venice (I)
Dep. of Biology, University of J.E.Purkyne, 40001 Usti nad Labem (CZ)
* [email protected]
Summary
Metal nanowires were synthesized according to an electroless deposition method in polycarbonate
template, obtaining nano electrodes ensembles with specific features and well defined diamctcr size.
Nan0 electrode ensembles were coupled with disposable screen printed substrate in order to increase
the sensitivity of the commonly used carbon graphite working electrode for electrochemical sensing.
Furthermore, worlang surface modification provided the basis for original protein immobilization
procedures that otherwise would not be allowed.

Abbreviations. ADS: physical adsorption; Au: gold; CA: chrono-amperometiy; CYS: cysteamine;
FIA: flow injection analysis; GA. glutaraldehyde, GOx, glucose oxidase; 1.o.d.: limit of detection;
PB: phosphate buffer 0 I M (pH=6.8); PC: polycarbonate; RE.printed Ag/AgCl pseudo-reference
electrode; RSD: relative standard deviation, SAM: self-assembled monolayer; SPE: screen prrnted
electrode; SPS screen printed substrate, WE: working electrode.

1. Introduction
Nanostructured materials have proven as one of the most powerful tools in new
technologies and research, due to their absolutely peculiar properties at nanometer size
scale. Many studies have shown that optical, mechanical, photo-catalytic and transport
properties drastically changes, depending on quantum size effect, as the mean diameter
of the particles is in the exciton size regime (i.e. 1Onm) [l-61. Both metallic and
semiconductor nanosized materials have found large applications in biochemistry,
bioanalytical techniques as well as luminescence, catalysis, optoelectronics and
photochemistry [7-lo]. The unique chemical and physical properties of metal, oxide and
semiconductor nanoparticles make them extremely suitable also for designing new and
improved sensing devices, especially electrochemical sensors and biosensors. The
important functions of nanostructured particles include the immobilization of
biomolecules, the catalysis of electrochemical reactions, the enhancement of electron
transfer mechanism between electrode surfaces and proteins, labeling of biomolecules
and even acting as reactant. The topic task is still the capability to assemble the
biochemical recognizing system on such sensors, by controlling at the same time the
surface activation (i.e. using SAM), the chemical-physical modification of the substrate
and the immobilization of biomolecules [ 1 1-20].

247
248

Recently, nanoelectrode ensembles

(NEEs) with specific features were
synthesized in PC template, according
to an electroless deposition method [21-
231. Nanowires of Au were obtained
with the expected diameter size (Fig.1)
and coupled with carbon SPE. The
association of Au NEE with graphite
SPS is aimed to couple the high
electroanalytical sensitivities, deriving
from the nanosized properties, with the
feasibility and versatility of screen
printing technology in easy to be used Figure I : Scanning electron microscopy
sensor fabrications. NEE based sensors (SEW image of Au nanowires deposited
and biosensors were electrochemically into nanoporous PC membranes (nominal
compared to unmodified carbon or pore size: 30 nm)
conventional Au SPEs.

2 Experimental

2.1- Preparation of Screen Printed Substrates

Conducting and insulators inks were printed on 0.3-0.5 mm thick polyvinyl-chloride
(PVC) substrate using a HTIO FleischleTMscreen printer (Brackenheim, Germany).
Silver and carbon-graphite pastes for the conducting paths and WE, AgfAgCl for RE and
insulator pastes were all from GWENT Electronics Materials IncTM.New templates for
an optimized planar concentric lay-out and specific three-electrode probes were
previously designed. SPEs based on conductive pastes from G W N T were shown to be
useful for batch or flow-through measurements where a high quality of the
electrochemical response is required.

2.2- Synt~esisof Au nunowires

Amongst the plenty of methods to synthesize nanosized materials, the formation of
metal, semiconductor, oxide and polymer nanowires or nanotubules via templates
represents one of the most simple and rapid one, especially when using nanoporous PC
or alumina membranes as template system. Electroless deposition of Au was performed
into track etched PC membrane (Uniporem, nominal pore size 30 nm, thickness 10 pm)
according to Menon et al. [21]. Track-etched polymer membranes are preferred for NEE
fabrication over alumina membranes because of their smaller pore densities and their
lower fragility. The original synthetic route was slightly varied, by immerging the
membrane 40 minutes for the surface activation by Sn (tin chloride solution was from
Sigma AldrichTM),15 minutes for Ag nanoparticles deposition (silver nitrate from Sigma
AldrichTM)and using AuCIi solution (Sigma AldrichTM)for final Au deposition by
means of galvanic displacement [22,23]. SEM observations, as the one reported in figure
1, show the successfil formation of 30 nm diameter Au nanowires with a good size
distribution, after removing the front side of the membrane from the Au layer grown on
the template surface.
 249

2.3- Nanoelectrode ensembles on screen printed substrate

Au nanowires modified PC membranes can be either used as platform for immobilization
of biomolecules or as free standing NEEs, i.e. a large assembly of very small
ultramicrolectrodes confined in a rather small space. In each case, NEEs can exhibit three
distinct voltammetric response regimes, depending on the scan rate and reciprocal
distance between the nanoelectrode elements, which is a function of the pore density of
the template [24]: total overlap regime, pure radial regime and linear regime.
It was demonstrated that for electroanalytical applications the total overlap regime is the
most advantageous one because of the higher faradaic-to-capacitive current ratios [21].
Without giving a detailed description of theoretical equations supporting the
electrochemical behaviour of NEEs, it should be sufficient to know that total overlap
regime, usually observed at NEEs fabricated from commercial track-etched membranes,
presents a signal to noise ratio in the order of 10 respect to a conventional Au macro
electrode with the same geometrical area [2 1,241.
In order to assemble our nanoelectrodes based membrane with SPS (obtaining
NEEEPS), Au NEE membrane was one side peeled and soaked only on the rear side into
a wet graphite ink pad. Either a vacuum pen or a vacuum-controlled silicone rubber tube
were conveniently used for placing the inked membrane on the graphite WE. Afterwards,
the device was completed by printing insulator and RE layers as in common screen
printing procedure. In each case, the active WE area is defined by the insulator geometry
(2.53 mm).

2.4- Analytical procedure for NEEBPS characterization

A home made flow cell was used for the characterization of NEEBPS and successively,
for the amperometric detection of glucose with a Glucose oxidase (GOx) based
biosensor. The microcell was tailored with appropriate dimensions and inner electrical
connections, in order to tightly lodge the screen printed substrate on the bottom surface
of the Perspex block. The NEE/SPS was screwed between these two polymer blocks,
leaving two holes in the upper side equipped for connection to the sample loop (I25 pL)
via Teflon tubing. The peristaltic pump (Minipuls-3 from GilsonTM)was used to propel
solution along a FIA system. PB was used as a carrier in all the following experiments.
Amperometric experiments performed under batch and FIA conditions were conducted
with the AutolabTMpotentiostat PGSTATlO and GPES electrochemical analysis system
(ECOchemie).

3. Results
3.1- Assays on NEE/SPSglucose sensor
The aim of this preliminary work was to evaluate the FIA response to a substrate sample
after immobilization of a specific enzyme on NEE/SPS. GOx immobilization was used as
a model system to test feasibility of the sensing probe and analytical performances of
nanoelectrodes assembled at a SPE. The preparation of NEEs, was followed by the SAM
formation and subsequent covalent attachment of the enzyme. A well experimented
methodology which has been used in our laboratory, was chosen as immobilization
method of GOx on Au element:
250

a) cysteamine was assembled on Au nanodisks either electrochemically (20 s of 21 mM

CYS growth at +700 mV vs. RE) or chemically (immerging for 16 hours the NEE/SPS in
a solution of 2 1 mM CYS);
b) glutaraldehyde (GA) was chosen as a coupling agent; SAM/NEE/SPS was immersed
for 1 hour in a 12.5 % v/v solution of GA;
c) GOx buffered solution (6 mg mL") was dropped on GA/SAM/NEE/SPS and left to be
covalently attached thanks to the primary amine groups of the enzyme. The non-
covalently bound enzyme as
well as the excess of GA 80
~

I
was easily removed by a I 1 20
three-fold washing process
with buffer solution. Figure
2 shows the signal to
background ratio (dots) as
well as the response of
current (squares) and
background (triangles) at 0
different applied potential, 0 8 * 1 . 0
c
due to the
200 300 400 500 600 700 800
oxidation/reduction of the
Potential (mV)
produced H202, when
multiple injections of M Figure 2: Hydrodynamic voltammogram under FIA
glucose sample were conditions of GOx mod$ed NEE/SPS. Flow rate: 0.4
performed on the same mL min-', appliedpotential: +600 m V. Grey triangles:
GOx/SAMNEE/SPS. background signal; squares: current signal, black
The unmodified NEEiSPS dots: signal to background ratio
-
did not respond to glucose,
confirming that non-specific oxidation of glucose at the Au surface does not take place.
The best signal to noise ratio was achieved in this sensor within the range +500 and +600
mV.
The response of the biosensor to various glucose concentration was investigated between
7.5 M and 3.1 M (R2=99.3%), which is within the analytical range for blood
monitoring. At higher glucose concentration, where saturation of enzymatic active sites
takes place, the response is no more linear. The glucose injection led to well defined
peaks with a response time of approximately 20 s, and return time of 50 s. The sensitivity
for GOx/SAM/NEE/SPS was calculated from the slope of the calibration curve and was
34.6 nA mM-' glucose, limit of detection (1.o.d.) of 1.5 M calculated by Zund Meier
method [26].The linear response can be estimated by using a Michaelis Menten analysis
of the calibration plot: an apparent constant of 14.9 mM was obtained after extrapolation
of the Lineaweaver-Burk relation and the value was significantly lower than that of the
native enzyme (33 mM) [27]. High operational stability was also achieved within 48
hours of continuous glucose injection. These analytical data were compared with those
available from literature in similar enzyme systems based on Au NEEs [28]: our sensors
are valuable in their features at least for some of these parameters, such as linear range
(up to 30mM), RSD (3.9%), 1.o.d. (0.15 mM).
 251

4. Conclusions
Using an electroless procedure for the deposition of Au nanowires into nanoporous PC
template membrane, nanoelectrodes ensembles (NEEs) with interesting mechanical,
electrical and optical properties, as well a s unique surface chemistry were prepared.
Fabrication of SPS in association with these nanosized component resulted in an original
system of NEEstSPS for biosensor applications. NEEsISPS were tested in a glucose
oxidase based sensor under FIA. High sensitivity, stability, and linearity of the
electrochemical response of the overall system were successfully achieved. According to
this novel approach, the advantage of nanostructured material properties (such as higher
signal to background ratio than conventional electrode, capability to control the property
depending on the nanoparticles diameter size) was further enhanced by the typical
features of thick film technology in screen printing production (such as disposability,
flexibility, durability of the product).
Acknowledgements
These experiments have been supported by FIRB, the Italian Fund for Fundamental
Research and by CoSMiC, the ENEA target Project on Biosensors and Bioelectronics -
www.hiosensinp. twt
References
1)M. Nirmal, D.J. Nornis, M. Kuno, M.G. Bawendi, A.L. Efros, M. Roscn, Phys. Rev. Lett., 75
(1995) 3728-3731.
2)A.P. Alivisatos, Science, 271 (1996) 933.
3)C.M. Lieber, SobdSiate Commun., 107 (1998) 607.
4)V. Albe, C. Jouanin, D Bertho, J. Cryst. Growth, 185 (1998) 388.
5)R.E. Smalley, B.1. Yakobson, SolidState Commun., 107 (1998) 597.
6)A J. Williamson, A. Zunger, Phys. Rev. B-Condens Matter, 59 (1999) 15819.
7)M. Bruchez, M. Moronne, P. Gin, S. Weiss, A P. Alivisatos, Science, 281 (1998) 2013.
8)V.L. Colvin, A N . Goldstein, A.P. Alivisatos, J. Am Chem. Soc, 114 (1992) 5221.
9)H. Henglcin, Chem. Rev, 89 (1989) 186 1.
1O)I. Klcps, A Angelescu, M. Miu, Materials Science andEngineering, C 19 (2002) 219.
11)N. Dharmaraj, H.C. Park, B.M. Lee, P. Viswanathamurthi, H.Y. Kim, D.R. Lcc, Inorganrc
Chemistry Communications, 7 (2004) 431.
12)s. Valizadeh, J.M. George, P. Leisner, L. Hultman, Elechochimica Acta, 47 (2001) 865
13)I. Willner, E Katz, Angav. Chem. Int. Ed., 39 (2000) 1180.
14)R.A. Kamin, G.S. Wilson, Anal. Chem., 52 (1980) 1198.
15)C.A. Koval, F.C. Anson, Anal. Chem., 50 (1978) 223.
16)T. Nakanishi, B. Ohtani, K. Uosaki, Jpn. J. Appl. Phys., 36 (1997) 4053.
17)s. Ogawa, F.F. Fan, A J. Bard, J. Phys. Chem., 99 (1 995) 1 1182.
18)C.J. Zhong, W.X. Zheng, F.L. Leibowitz, Electrochemistry Communrcations, 1 (1999) 72.
19)F C. Meldrum, J Flath, W. Knoll, Langmuir, 13 (1997) 2033.
20)A. Badia, L Cuccia, L. Demers, F. M o m , R.B. Lennox, J. Am Chem. Soc., 119 (1997) 2682.
21)V P Menon, C.R Martin, AndChem , 6 0 (1995) 1920
22)B. Brunetti, P. Ugo, L.M. Moretto, C.R. Martin, J. Elechoanal. Chem., 491 (2000) 166.
23)L.M. Moretto, N. Pepe, P. Ugo, Talanta, 62 (2003) 1055
24)J. C. Hulteen, V.P. Mcnon, C.R. Martin, J. Chem SOC.Faradq Trans., 92 (1996) 4029.
25)F. Cheng, L. D. Whiteley, C.R. Martin, Anal. Chem., 61 (1989) 762.
26)P.C Meier, R.E. Zund, Siatisrrcal Methods rn Analytical Chemistry, J.D Winefordncr (Ed.), J.
Wiley & Sons Inc., New York (1993) 87.
27)B Swoboda, V. Massey,J. Biol. Chem ,240 (1965) 2209.
28)M. Delvaux, S. Demoustier-Champagne, A. Walcarius, Elechoanalysrs, 16/3 (2004) 190.
 Buried Microchannels by Electrochemical Etching of Silicon

G. Barillaro, A. Nannini
Dipartimento di Ingegneria dell 'Informazione: Elettronica, Informatica,
Telecomunicazioni, Universita di Pisu, via G. Caruso 16, 56126 Pisa-Italy
g.barillarokdiet.unipi.it

Summary
In this work, a simple, one musk process for the fabrication of buried microchannels into
silicon substrates is presented. The process flow consists of two main technological steps: 1)
fabrication of high aspect ratio meander-shaped structures into a n-type silicon substrate by
using a photo-electrochemical etching technique; 2) thermal oxidation to merge the top of
grooved structures and leave a buried channel at the bottom. The fabrication process is
detailed and experimental results are reported and discussed.

1 Introduction
In last years, a number of processes for the fabrication of microchannels into
different substrates (i.e. silicon, quartz, plastic) has been proposed.
Microchannels, besides from being essential components of any microfluidic
system, in which they act as connection parts between pumps, valves, etc., are
used as separation columns for gas and liquid chromatography, as heat
exchangers in microreactors or electronic chip cooling.
Three main approaches are today employed for microchannel fabrication: 1)
wafer-to-wafer bonding techniques (i.e. anodic or fusion bonding) [ 13; the main
drawbacks of such processes are the wafer-to-wafer misalignment and micro
voids introduced during the bonding process. 2) Surface micromachining
techniques, which use a thin film as a structural part, deposited upon a suitable
sacrificial layer which is dissolved to obtain the microchannel [2]. The limits of
such an approach are the dimension of fabricated channels, usually restricted to
few microns by the maximum thickness of the sacrificial layer. 3) Bulk
micromachining, exploiting both dry and wet (anisotropic and isotropic) silicon
etching techniques to obtain microchannels grooved into the silicon substrate
[3]. The channels are usually sealed by means of thermal oxidation. This
approach is, however, quite complex requiring both dry and wet etching steps.
In this work an original process for the fabrication of buried silicon
microchannels with high aspect ratio is reported. The process exploits the
electrochemical etching of silicon to fabricate high aspect ratio meander-shaped
trenches, that are then sealed by thermal oxidation. The fabrication process is
straightforward and requires only one mask.

252
 253

2 Fabrication process
A schematic view of main technological steps of the buried microchannel
fabrication process is reported in Fig. 1.
The starting material was a n-type silicon wafer with resistivity of 2.2-
4 Qcm, 550 pm thick, (100) oriented, single-side polished. A 100 nm thick
silicon dioxide layer was grown by means of a thermal dry oxidation (1 h at
1050 "C). A standard lithographic step was used define in the oxide layer several
meander-shaped patterns with different pitch and dimension. A KOH etching
was then employed to transfer the pattern into the silicon, producing in this way
the notches which were exploited as seed for the next silicon electrochemical
etching.
Photo-electrochemical etching in a HF-based solution [4] was then performed
and high aspect ratio trenches were fabricated (Fig.la). The etching current was
properly changed during the electrochemical process from a given initial value

Fig.1 Main technological steps of the buried microchannel fabrication process: a)

fabrication of meander-shaped trenches by electrochemicaletching of silicon; b) sealing of the
top of etched trenches by thermal oxidation.
254

to a higher one in order to increase the channel size over a given depth. Actually,
varying the etching current is a feasible way to change the porosity (i.e. the ratio
between the etched silicon mass and the total silicon mass) and in turn the lateral
size the etched structure.
A wet thermal oxidation was finally employed to seal the top of fabricated
trenches and leave a buried pipe at the bottom. In fact, as the lateral size of
trenches changes perpendicularly to the wafer surface, the outgrowing oxide
layer only seals the top part of the structure, where the gap between adjacent
silicon walls is thinner, and a buried path is obtained at the bottom (Fig. lb).

e r i ~ e n t aresults
l and dis~Mssion
The buried microchannel fabrication process is mainly based on the photo-
electrochemical etching (PEE) of a n-type silicon substrate in a HF-based
solution
Since 1990, the PEE of a n-type silicon wafer is a well known method for
fabricating deep regular macropore array with high aspect ratio (over 200) and
pore diameter in the range of the micrometer [5, 61. An array of pre-defined
notches on the substrate surface, produced by standard lithography and
subsequent KOH etch, is in this case required to restraint the macropores growth
in pre-determined positions during the etching process.
Recently, it has been demonstrated that the PEE can also be employed to etch

Fig. 2 S ~ ~ cr # s s - s e c ti #(lefi)
n and top view (right) of the meander shaped trenchesfabricated by
electrochemical etching. It is visible the change in the lateral size of the structure over a given
depth (lefi image).
 255

Fig. 3 SEM cross-section (Zefl) and top view (right) of the buried microchannels obtained at the
end of the thermal oxidation step. It is clear that while the top of the structure is sealed by the
outgrowing oxide, a buriedpipe is left at the bottom (lefl image).

silicon microstructures with a high degree of freedom about the shape, size,
porosity and pitch [4, 71. In fact, changing the predefined geometry on the
substrate surface it is a feasible way to vary the shape of the microstructure
resulting from the PEE. In this way, squared and circular spirals, pillars, tubes,
walls, and other structures with high aspect ratio, were fabricated on the same
silicon die.
In this work, by defining a meander-shaped pattern on the silicon surface,
meander-shaped trenches were grooved into the chosen substrate by using PEE.
A SEM top view and the cross-section of a typical sample after the
electrochemical etching step are shown in Fig. 2. In this case, an etching current
of 30 mA, co~espond~ng to a porosity of 50%, for 5 minutes was used to etch
the top part of the structure; the current was then increased to a value of 42 mA
for 15 minutes to obtain a porosity of 70% at the bottom. According to the
etching current variation, the lateral size of the etched structure increases over a
given depth, as can be seen from the cross-section of Fig.2: trenches are about 2
pm wide at the top and about 3 psn at the bottom.
Fig. 3 shows a SEM top view and the cross-section of a typical sample after
the thermal oxidation sealing step. The steam oxidation process was performed
at 1050 "C for 4 h. It is clear that while the top part of the structure (low porosity
region) is completely sealed by the outgrowing oxide layer, a buried path is
256

obtained at the bottom (high porosity region). The resulting buried pipe is about
20 pm high and 1 pm wide, corresponding to an aspect ratio of 20.

4 Conclusions
An original, one mask process for the fabrication o f buried microchannels into
silicon substrates was reported in this work. The process is based on the photo-
electrochemical etching o f the silicon, used to produce high-aspect ratio
meander-shaped trenches with a size variation along the direction perpendicular
to the wafer surface. This variation allows to seal the top part of trenches by
means of thermal oxidation, while leaving a buried channel at the bottom. Future
work will be devoted to the optimisation of the channel geometry, definition of
inlet and outlet o f the channel, and filling tests.

References
[ 11 G. Perret, J. Boussey, C. Schaeffer, M. Coyaud, IEEE Trans. on Components
and Packaging Technologies, 23, 665 (2000).
[2] W. J. Nam, S. Bae, A. K. Kalkan, S. J. Fonash, J. Vac. Sci. Technol. A, 19
(4), 1229 (2001).
[3] C. Rusu, R. van't Oever,M. J. de Boer, H.V. Jansen, J.W. Berenschot, M.L.
Bennink, J.S. Kanger, B.G. de Grooth, M. Elwenspoek, J. Greve, J. Brugger, A.
van den Berg, J. of Microelectromechanical Systems, 10,238 (2001).
[4]G. Barillaro, A. Nannini, M. Piotto, Sens. Act. A, 102, 195 (2002).
[5] V. Lehmann and H. FO11, J. Electrochem. SOC.,137, 653 (1990)
[6] V. Lehmann, J. Electrochem. Soc.,l40,2836 (1993).
[7] G. Barillaro, P. Bruschi, A. Diligenti, A. Nannini, Physica status solidi (C), 2
(9), 3198 (2005).
 Gas Sensing Properties of Metal-oxide/Organic Hybrid
Materials grown by Supersonic Beam Deposition
F. Siviero, N. Coppedb, L. Aversa, M. Nardi, A. Pallaoro, T. Toccoli, R.
Verucchi, and S . Iannotta
IFN-CNR Istituto di Fotonica e Nanotecnologie - Sezione ITC di Trento, Via
Sommarive 18, 38050 Povo di Trento, Italy.- [email protected]

A.M. Taurino, P. Siciliano

IMM-CNR Istituto per la Microelettronica e Microsistemi - Via per Arnesano
73100 Lecce. Italy

Summary
Hybrid organic-inorganic gas sensing devices have been developed by means of supersonic
beam co-deposition of titania clusters and copper phthalocyanine. This technique allows
controlling the cluster size and structure as well as the kinetic properties of organic
precursors, thus enabling both the room-temperature growth of nanocrystalline Ti02 and its
functionalization. Particularly the high kinetic energy transferred to the organic molecules is
exploited to enhance reactivity at the interface between the two counterparts, resulting in the
synthesis of a new material with interesting gas sensing properties. The hybrid devices show
improved performances as to stability and sensitivity towards oxidizing and even non-
oxidizing gases, which IS a remarkable result for metal-phthalocyanine based sensors. The
role of the organic-inorganic interface in the development of such characteristics is briefly
discussed.

1. Introduction
The synthesis and characterization of new functional hybrid nanomaterials is
a rapidly evolving field of research, which is constantly enlarging its range of
applications [ 11. Particularly for gas sensing, the functionalization of inorganic
materials with organic species is a very promising approach to produce new
generations of devices overcoming the limits due to the limited selectivity of
metal oxides and to the low stability of organic molecules. Among different
techniques for the synthesis of nanostructures, the supersonic beam deposition of
molecules [2] and clusters [3] allows unique control on the properties of the
precursors, on their mutual interaction and the formation of interfaces. This is
achieved by controlling key parameters as the kinetic energy and momentum of
the particles, as well as cluster size and crystalline structure.
We report here about the development and the response characterization of
hybrid devices made of alternating layers of copper phthalocyanine and cluster-
assembled titania. The results point out how this original approach to the
synthesis of such hybrid nanostructures brings to the development of novel
sensing properties.

257
258

2. Experimental

The deposition of the sensing layers was performed in an UHV apparatus

equipped with two supersonic molecular beam sources (the one for the organics
and the cluster source), a quartz microbalance and a Jobin-Yvon in-situ
ellipsometer. The deposition chamber is connected via an UHV transfer line to
the analysis one, which hosts facilities for electron spectroscopy and surface
characterization ( X P S , U P S , Auger, LEED). This experimental setup allows the
co-deposition and the in situ characterization of hybrid thin films and interfaces.
The source for organic species consists of a quartz tube ending with a nozzle
(50-100 pm) in which powders of organic species are sublimated by Joule
heating and carried out of the source in the supersonic expansion of the carrier
gas. Acting on gas inlet pressure and source's temperature we tune the kinetic
energy of the precursors: in the case of CuPc, helium is used and the kinetic
energy gained by the molecules is about 18 eV, as measured by time-of-flight
mass spectrometry. The TiOz cluster deposition is performed with the Pulsed
Microplasma Cluster Source [4], which is based on the ablation of a titanium rod
by means of a pulsed plasma discharge, followed by condensation of clusters
from the gas phase in helium plus 0.1% oxygen. Aggregates dispersed in the
supersonic beam show a mass distribution ranging from a few to several hundreds
of atoms: the maximum and the relative abundance of heavier and lighter species
of such distribution can be controlled by changing the operation parameters
(mainly discharge timing), or by aerodynamic separation effects. Mass selection
is a very powerful tool in order to determine the properties of the nanostructured
film, since a correlation between size and crystalline phase has been observed.
The sensors were prepared using alumina substrates with 7 pairs of
interdigitated gold electrodes (200 pm spacing) and platinum heaters on the back.
The test of gas sensing devices was carried out with small concentrations of high-
purity methanol and NOz, which were diluted in dry air with precision MKS mass
flow controllers. The current flowing in the films at the fixed voltage of 4 V was
measured by a Keithley 65 17A sourcemeter.

3. Results and discussion

We have explored two kinds of different device architectures: the deposition
of copper phthalocyanine on Ti02 (CuPc/Ti02 sensor) and vice versa
(TiOz/CuPc). The thickness of the layers was estimated by measuring the
deposition rate with the quartz microbalance to be about 80 nm for the oxide and
60 nm for the organic film. The performances of the hybrid sensors have been
compared with the ones of pure CuPc devices deposited with the same technique.
Figure 1 shows the response to 10 ppm of methanol, defined as the ratio between
the change in current A1 and the current value I, reached in presence of the gas.
 259

At the temperatures of 155C for CuPc and CuPc/Ti02 sensors and 130C for
Ti02/CuPc, we observe a p-type response typical of organic materials and the
baseline currents 10 in pure dry air are reported in Table 1. It can be noted how
the hybrid films show a similar conductivity, which is roughly one order of
magnitude lower than that of the pure organic film. At this temperature the
conductivity of the hybrids is due to the organic layer, since the oxide has a much
higher resistance. In the case of CuPc deposited on the oxide, the organic film
has a very defected structure induced by the high surface roughness of
nanostructured Ti02, which has a granular structure with grains typically in the 5-
30 nm range and peak to peak corrugation of several tens of run [5].On the other
hand for the opposite architecture the cluster deposition is likely to damage the
soft organic layer causing a decrease in its overall conductivity.

0.0 -

0.1 -

0.2 -

0.3 - ... -.:.&<..$f-

. -CuPc 155C
-:
.r 0 CuPcTTiO, 155C
t - TiOJCuPc 130C
0.4 1 . i * i . i i 1 . 1

Fig. 1 Response of pure CuPc and hybrid sensors to 10 ppm of methanol. Arrows indicate the
beginning and the end of gas exposure.

Organic devices are found to have a low sensitivity to methanol as usually

reported in literature for reducing gases [6], while the hybrid sensors show a
much higher response. This result can be explained envisaging a large
contribution to the response from the interface region between CuPc and the
oxide. We have strong indications that the kinetic energy supplied to the
precursors is able to enhance the reactivity at interfaces, overcoming large
260

activation barriers, for this kind of material [7] and for different systems [8].
Anyhow the presence of chemical interactions between the organic and inorganic
counterparts is likely to be the origin of the sensitivity of hybrids to the reducing
gas, by virtue of a charge transfer mechanism from TiOz to CuPc.

1 -CuPc 155C
3- 0 CuPcTTiO, 155C
- TiOJCuPc 130C

-
1
0

a 2-
a,
v)
C
0
0.
v)
a, 1-
[r

0-
t
1 I I 1 I I

Fig. 2 Typical sensors response to 10 ppm of nitrogen dioxide.

Upon interaction with an oxidizing specie such as NOz, the response of the
pure CuPc sensors is clearly not as low as in the previous case, anyway the
hybrids still show higher response (defined as AH& as reported in Figure 2. In
this case hybrid devices demonstrate also better recovery performances with
respect to the pure organic, for which the poisoning effect of NO2 is evident.
Table 1 reports the percentage of the current change recovered in a period
equivalent to that of gas exposure (about 22 min). For methanol we dont observe
such a significant difference in the recovery properties of the different sensors as
in the case of nitrogen dioxide, also taking into account that the Ti02/CuFc
device is operated at a lower temperature.
 261

Table 1. For each type of sensor we report the current b flowing in the films in pure dry air and the
percentage of the change in current recovered after about 22 min from the exposure to methanol and

Sensor
Methanol
9.44 x 10-
CuPc/TiOz 8.2 x 1 0
6.0 x

Conclusions
Novel hybrid nanotitania-copper phthalocyanine resistive gas sensors have
been developed by jointly using supersonic beams of molecules and clusters. The
properties of the interface between the two materials as synthesized by this
technique are at the basis of the development of novel sensing properties, such as
the ability to respond to reducing gases at low operating temperatures, where
pure Ti02 has no sensing activity and pure CuPc shows quite poor performances.
We envisage new classes of gas-sensing devices could be developed on the basis
of these results.

Acknowledgments
This work is financially supported by MIUR - FIRB Program - project
SQUARE (RBNEOlY8C3-007) and FUR - Provincia Autonoma di Trento -
projects RASO2 and NanoCOSHy.

References
[I] C.Sanchez, B.Julian, P.Belleville, M.Popal1, J. Mater. Chem., 15 (2005) 3559.
[2] S. Iannotta and T. Toccoli, Journal of Polymer Science B 41,2501, 2003.
[3] P.Milani, S.Iannotta, Cluster beam synthesis of nanostructured materials, Springer-
Verlag, Berlin-Heidelberg, 1999.
[4] E. Barborini, P. Piseri, P. Milani, J. Physics D: Appl. Phys. 32, L105, 1999.
[5] T.Toccoli, S.Capone, L.Guerini, M.Anderle, A.Boschetti, E.Iacob, V.Micheli,
P.Siciliano and S.Iannotta, IEEE Sens.J., 3 (2003) 199.
[6] C.Maleysson, M.Passare, J.P.Blanc, V.Battut, J.P.Germain, A.Pauly, V.Demarne,
A.Grise1, C.Tiret, R.Planade, Sens. and Act. B, 26-27 (1995) 144.
[7] N.Copped2, L.Aversa, M.Nardi, A.Pallaoro, F.Siviero, T.Toccoli, R.Verucchi,
EJacob, V.Micheli, M.Anderle, A.M.Taurino, PSiciliano and S. Iannotta, X AISEM
Conference Proceedings, World Scientific Publications, 2005.
[8] L.Aversa, R.Verucchi, G.Ciullo, P.Moras, M.Pedio, L.Ferrari, A.Pesci and S.Iannotta,
App. Surf. Sci., 184 (2001) 350.
 Focused Ion Beam in Gas Sensor Technology

Vera La Ferrara, Ivana Nasti, Ettore Massera, Brigida Alfano and

Girolamo Di Francia
ENEA, Centro Ricerche Portici Localit2 Granatello 80055 Portici (NA)
[email protected]

Summary
Focused Ion Beam is used to realize two gas sensing devices machining as substrate a
silicon/Si,Na wafer and using single walled carbon nanotube ropes as sensitive element.
Ionic beam is applied both for milling substrate and for platinum deposition. Sensors are
characterized at room temperature introducing NH3 and NO2 gases in test chamber. A well
defined resistance change is observed

1 Introduction
Recently, nanoparticles and nanowires are receiving an increasing attention
in the field of chemical sensors. High surface to volume ratio and enhanced
reactivity of nanosensors [l] improve device response, even at room
temperature, and limit the device power consumption. It has been reported that
when metallic nanowires or nanoparticles are exposed to specific analyte
molecules, a change in their electrical resistance is recorded [2]. Analogous
effects have been observed for single walled carbon nanotubes [3,4] and for
metal oxide semiconducting nanostructures [5-71.
In order to overcome the obvious complexity related to nanodevice fabrication,
several special tools have been, and still are, under investigation.
In this respect, one of the most promising apparatus is the Focused Ion Beam
(FIB). The use of high energy FIB is an already well established technique in
microelectronics and micro-fabrication for mask-less patterning. Moreover
milling and deposition by FIB have already found applications in micro-
electromechanical and micro-optomechanical systems, micro-sensors and
actuators [8,9]. Dual Beam (FEI - Quanta 200 3D) system, located at ENEA
laboratory, is composed of an electron column (Scanning Electron Microscopy -
SEM), a Ga ions column (Focused Ion Beam - FIB) and a gas injector system (a
Pt organometallic precursor). By means of this apparatus imaging, patterning
deposition and etching of both conductive or isolating materials is possible.
In this work we report first results on the realization of gas sensing devices
obtained fabricating by FIB patterning Si/Si3N4substrate and using single walled
carbon nanotube (SWCNT) ropes as chemical sensitive interfaces. Sensors are
analyzed at room temperature using NH3 and NO2 as testing gases.

262
 263

2 Experimental

The first gas sensor device is realized starting with a silicon substrate coated
with a Si3N4layer (200 nm deposited by PECVD). By e-beam evaporator, one
pad, 200 pm width, is deposited on silicon substrate placing on two metallic
layers of Cr (20 nm thickness) and Au (280 nm thickness). By means of FIB,
under 30 KeV accelerating voltage, 30 pA emission current, 1 ps dwell time
without overlap for the digitised ion beam position, a 1 pm wide channel milling
CrIAu pad has been realized obtaining two separated rnicroelectrodes on which
SWCNTs have been subsequently deposited.
At this aim, SWCNTs have been dispersed in dimethylformamide (DMIj) to
debundle nanotubes and form an uniform suspension of nanotube ropes with
average diameter of 50 nm and 3-4 pm length. Few drops of the suspension have
been deposited between two CrlAu electrodes. One single nanotube is found by
SEM and contacted by FIB depositing platinum using injector of organometallic
precursor (figure 1).

Fig. 1: (a) SEM image of carbon nanotubes on a FIB milled channel (1 pm width) @) Scheme of
device: silicon substrate coated with Si3N4 on which two Cr/Au electrodes are realized. One single
rope, placed on it, is contacted by platinum deposited by FIB

The second gas sensing device has been fabricated depositing SWCNTs ropes
on silicon substrate coated with a 100 nm Si3N4layer, PECVD deposited.
The substrate surface has been explored by means of FIB, selecting one CNT
rope, suitable for the subsequent fabrication process. Under 30 KeV accelerating
voltage, 10 pA emission current and 0.1 ps dwell time with 50% overlap for the
digitised ion beam position, the CNT rope has been contacted realizing a
platinum electrode at each of the rope ends. Each electrode, about 300
264

nanometer in diameter, have a length of a few microns, ending in contacting

gold pads (0.25 cm), previously deposited by e-beam evaporation.
In figure 2 we show the nanotube device fabricated according to the above
process.

Fig. 2: (a) Ion image of carbon nanotube on siliconlSigN4device, Single nanotube has been contacted
by two Pt wires deposited by FIB using Pt organometallic precursor (b) Device scheme

3 Results and discussion

After FIB micro-machining, a volt-amperometric technique, at constant bias,
is employed for sensor dc electrical characterization in a controlled gas-flow
environment, pre-mixed with dry carrier in the desired percentage by mass flow
meters and continuously controlled by means of an in-line FTIR. Total gas flow
is set to 500 sccm. For the measurements here reported, certified bottles
containing mixtures of 30 ppm of nitrogen dioxide in synthetic air and 500ppm
of ammonia in synthetic air, have been used [101.
Results are reported in figures 3 and 4. Devices have been DC biased at 0,lVoIt.
Graphs show that both the sensor devices respond to NH3 via an increase of
resistance while NOz results in a decrease of their resistance. Such a different
behavior in reducing and oxidizing environments has been already reported for
similar materials [ 1I] and claimed to be due to the interaction between nanotube
devices and gas molecules, via nanosized local defects on CNT. Although it is
evident a resistance change when devices are in presence of gas, current doesnt
recover when synthetic air is again introduced in the test chamber.
 265

Fig. 3: Sensor output current is reported versus the measurement time under 500ppm of NH, in
synthetic air. (a) Gas sensor device response (see scheme reported in figure 1); (b) Gas sensor device
response (see scheme reported in figure 2)

Fig. 4: Sensor output current is reported versus the measurement time under 30ppm of NOz in
synthetic air.. (a) Gas sensor device response (see scheme reported in figure 1); (b) Gas sensor device
response (see scheme reported in figure 2)

In table 1 the relative current variations is reported in order to compare devices

behaviour.

device variation in NH3 (!A) variation in NO2 (!+A)

First type 16 20
Second type 1 0.6
266

two CrlAu electrodes cannot be ruled out. On the contrary the second sensor
response is certainly due to just one SWCNTs rope as it is evident in figure 2.

4 Conclusions
We have shown our first attempts to use FIB as a tool to fabricate
nanosensing devices. Both for patterning and for soldering, suitable operating
conditions have been found for the realization of two sensing devices based on
CNT wires. The devices respond both to NH3 and to NO2 although not
reversibly. Further investigations are necessary to enhance desorption
mechanism.

References
[I] G. Di Francia, L. Quercia, I. Rea, P. Maddalena, S. Lettieri, Sensor and
Actuators B, 111-1 12 (2005) 117
[2] F. Favier, E.C. Walter, M.P. Zach, T. Benter, R.M. Penner, Science,
293 (2001) 2227
[3] P.G. Collins, K. Bradley, M. Ishigami, A. Zettl, Science 287 (2000)
1801
[4] J. Kong, H.J. Dai, J. Phys. Chem. B, 105 (2001) 2890
[5] E. Comini, G. Faglia, G. Sberveglieri, D Calestani, L. Zanotti and M.
Zha, Sensor and Actuators B: Chemical 111-112 (2005) 2.
[6] E. Comini, G. Faglia, G. Sberveglieri, Z. Pan, Z.L. Wang, Applied
Physics Letters, 8 1 (2002) 1869
[7] X.Y.Kong, Y. Ding, R.S. Yang, Z.L. Wang, Scince, 303 (2004) 1348
[8] R. Puers, S. Reyntjens, D. De Bruyker, Sensors and Actuators A, 97
(2002) 208
[9] A.A.Tseng, J. Micromech. Microeng., 14 (2004) R15
[lO]L. Quercia, F. Cerullo, V. La Ferrara, G. Di Francia, C. Baratto,
G.Faglia, Phys. Stat. Sol. (a) 182, (2000) 473
[ 111K.S. Lee, H.J. Kim, J.E. Fischer, A. Thess and R.E. Smalley, Nature,
388 (1997) 255.
 PROCESS DEPENDENCE OF DOPED POLYSILICON
THERMAL CONDUCTIVITY FOR THERMAL MEMS
APPLICATIONS'

F. MANCARELLA, A. RONCAGLIA, S. CRISTIANI,

G. C. CARDINALI, M. SEVERI
IMMBologna, CNR, Via Gobetti 101
Bologna, I-40129, Italy

Thermal conductivity measurement by micromachined test structures permits to evaluate

the microscale variations of thermal conductivity on different devices. We used this
technique to measure polysilicon thermal conductivity with particular concern to
investigating its dependence on annealing temperature and doping type.

1. Introduction
Doped polysilicon layers are widely employed in fabricating thermal
microsystems such as bolometers, micro hotplates or thermopiles thanks to their
thermal and electrical properties. Since the performances of such devices are
strongly influenced by heat conduction, a precise characterization of polysilicon
thermal conductivity is of paramount importance.
In order to determine the thermal conductivity (K) of thin films,
microstructures like those proposed in [l] can be adopted. In this work, such
microstructures have been adopted in order to investigate the dependence of
polysilicon thermal conductivity on the type of doping adopted (p or n-type) and
the annealing conditions.

2. The test structure

The device consists of a dielectric cantilever micromachined on a silicon
substrate by anisotropic etching. It is 200 pm wide and 150 pm long. The tip of
the cantilever can be heated by a heavily-doped polysilicon resistor and its
temperature measured by the variation of a second resistor with different
geometry. Both are connected to bonding pads by platinum or aluminum
interconnections, placed over the suspended dielectric arms. The cantilever-type
membranes employed in the test structures have been released from the
underneath silicon substrate by anisotropic etching using a TMAH solution.

' This work is supported by Netgas project- EC V framework program, proposal IST-2001-37802
267
268

Figure 1 reports a microphotograph of a released device and a layout detail

in which the two resistors used for heating and temperature monitoring (needed
for the measurement as explained in section 3) are highlighted.
The measurement principle relies on the fact that, when a Joule power P is
dissipated on the heater, the temperature of the cantilever tip is increased by AT
with respect to the substrate. At the steady-state, this phenomenon depends on
the balance between the heat produced and the one dispersed towards the colder
surrounding environment along the cantilever, the four arms, and the air. The
heating power P and the temperature difference AT are thus related by:
P
G=G,+G,+G --
--AT

\ f
a_.-..-
Figure 1. Micrograph of a released micromachined test structure for thermal conductivity
measurement (left) and a detail of the cantilever (right).

where G, G,, G, denote the thermal conductances of the whole structure, the
four arms and the cantilever respectively, while Gh is the one due to heat
exchange through air. G, is given by

where the summation runs over the component layers of the cantilevers, and K,,
d,, w and L indicate their respective thermal conductivity, thickness, width and
length.
As proposed in [ 11, the thermal conductance of a particular layer is obtained
by subtracting the overall measured conductance of the corresponding cantilever
from a reference one. In order to do that, different structures have been realized
like those represented in Figure 2, which differ in the cantilever composition.
Structure (a) is the reference one, composed by a plain supporting membrane (a
stack of LPCVD silicon oxidehitride). In the other devices, a layer of p- or n-
type polysilicon is added to the supporting membrane (see the cross sections of
the different cantilevers reported in Figure 3).
 269

Since the presence of an additional film increases the thermal conductance

of the cantilever sandwich, the K value of the layer considered can be calculated
by subtraction of the reference structure from the modified structure as
described in the following.

Figure 2. Micromachined test structures for thermal conductivity measurement: reference cantilever
(a), n-polysilicon (b), p-polysilicon (c).

~~~~~~~~ ~~~~~~~

Figure 3. Schematic cross-sections of microcantilevers used to determine the thermal conductivities

of thin films. Reference structure (left), modified structures for n-polysilicon (center) and p-
polysilicon (right) thin films.

Once the thermal conductances of the two structures have been determined
with the method described in Section 3, the thermal conductivity of the added
layer is calculated as

where dpoly,wpolyand L are the thickness, width and length respectively of the
layer of interest.
In both cases, the measured overall thermal conductance also includes a
contribution due to heat convection, than can be cancelled thanks to the method
used, provided it is maintained constant from one measurement to another.

33. The measurement method

A scheme of the measurement setup is reported in Figure 4. The reference
cantilever is heated biasing the resistor, placed over the free end of the structure,
by imposing a current Ib. In this way a temperature difference AT is created
between the tip of the cantilever and the silicon bulk.
270

This AT is estimated by measuring the resistance change of the thermistor

(fabricated with polysilicon, as well). In order to eliminate the offset due to the
i n s t ~ e n t a t i o n that
, can be deleterious due to the low temperature difference
measured, the following procedure has been adopted. For a fixed value of the
heater current Ih, the thermistor current is ramped up from Ith-AIth to Ith+AIth
(indicated as Irampin the figure), Ith being a reference value. From the slope of
the resulting Vth(Ith) curve, an offset-compensated &, value is obtained for
different Ih values, and the temperature difference AT is extracted from the
previously calibrated TCR (temperature coefficient of resistance) value for the
polysilicon thermistor.

Figure 4.Electrical probe setup used in the thermal conductivity measurement.

Consequently, the thermal conductance can be calculated by the following

expression:
G =P h
AT
where Ph = 12%is the power dissipated by the heater.
Typical results obtained on a reference cantilever (Gmemb)are reported in Table
1.

HEATER THERMISTOR AT Gmemb

(PA) (C) (EiWlmK)
0 10-15
125 10-15 1.63 34.92
250 10-15 7.32 34.98
375 10-15 16.58 34.76
500 10-15 29.52 34.70

The described technique has been applied to measure the thermal

conductivity of heavily doped polysilicon layers implanted with boron and
phosphorus respectively, and annealed at different temperatures. The
 271

corresponding calculated thermal conductivities K are shown in Table 2. From

the results it is apparent that n-type polysilicon thermal conductivity does not
depend strongly on annealing, differently than p-type polysilicon. This can be
probably ascribed to the different grain growth kinetics in the two layers. In fact
phosphorus enhances grain growth even at low temperatures, and this would
explain the higher thermal conductivity of n-type polysilicon even for the 900
"C temperature annealing.
Table 2. Measured thermal conductivities on polysilicon samples annealed at different tempcratures.

implant dose
anneal. temp. P K
LAYER
(cm-2) ("C) (pacm) (W/mK)
900 2.15 23.44
n-poly 3.3~10'~ 1000 1.71 25.33
1100 1.56 26.17
900 6.46 9.14
p-poly 6 . 6 loi5
~ 1000 3.90 10.46
I I I

1100 I 1.79 I 16.88

4. Conclusions
Test microstructures have been use to estimate the microscale variation of n-
and p-type heavily doped polysilicon thermal conductivity as a function of the
annealing temperatures. The results point out that n-type polysilicon thermal
conductivity does not depend strongly on annealing, differently than p-type
polysilicon, probably due to the different grain growth kinetics in the two layers.
The microstructures and measurement method described could also be
adopted to estimate the thermal conductivity of other microelectronic materials,
such as silicon dioxide, nitride and metal films.

References

1. M. von Am, 0. Paul, H. Baltes, Journal of Microelectromechanical Systems

9, 136 (2000).
2. A. McConnell, S. Uma, K. E. Goodson, Journal of Microelectromechanical
Systems 10,360 (2001).
 GAS SENSING SYSTEM CONSISTING IN
MOX-BASED MICROSENSORS
INTERFACED TO A NOVEL INTEGRATED
5-DECADE DYNAMIC RANGE FRONT-END
Marco Grassi, Piero Malcovati
Department of Electrical Engineering, University of Pavia
[email protected]

Simonetta Capone, Luca Francioso, Pietro Siciliano

Research Center IMM-CNR, Lecce
[email protected], [email protected]

Andrea Baschirotto
Department of Innovation Engineering, University of Lecce

Summary
Research on gas sensors aims to improve the performance, minimize size, power
consumption and cost of such devices. For implementing gas sensing microsystems
with the above mentioned features, the micromodule approach, where the sensors and
the read-out electronics are manufactured on different chips and mounted in the same
package or on the same substrate, appears well promising and versatile. The proposed
system, realized for environmental monitoring purpose, consists of an integrated tin-
oxide Sn02 resistive sensor over micromachined substrate followed by a 0.35pm
CMOS auto-scaling read-out circuit. The interface circuit, which is actually a trans-
resistance variable gain amplifier connected to an incremental N D converter, after the
initial calibration procedure, required to minimize offset and gain inter-scale error,
achieves a worst case accuracy of about 0.1% over a range of more than 5 decades
(100n-20MQ)

1. Introduction
Gas sensors are widely used for several applications, ranging from
environmental and air-quality monitoring to automotive and industrial control.
Recently, especially in view of legislative initiatives aimed to the reduction of
pollution and of human exposure to dangerous gasses, hand-held systems for gas
sensing are becoming quite important, thus pushing in the direction of improving
the performance, minimizing size, power consumption and cost of such
systems[l][2]. A gas-sensing microsystem consists of three main parts that are
subject to continuous development from researchers: an array of gas sensors, an
electronic read-out circuit and a pattern recognition data processor. The
improvements in the overall gas-sensing microsystem are obtained by improving
the single parts and/or their coordination. Micro-membrane resistive gas sensors
have been recently developed[3] and appear to be the most promising sensing
devices for portable applications, since their thermal time constant is small

272
enough to reach the operating temperature by means of a low-power integrated
heater, and to operate the device at different temperatures in time and in
space[4][5]. On the other hand, the research is also concentrated on developing
read-out electronic circuits in IC form, which are mandatory for realizing
portable gas sensing systems. Their compactness is extremely interesting since it
is possible include in the same IC both the conditioning of the signal coming
from the sensor and the signal digitalization. In this way the read-out electronic
circuit can be directly connected with the block implementing the pattern
recognition digital signal processing algorithms for the final detection of the
target gases. These digital blocks are the objective also of important research
activity. For instance, recently developed dynamic pattern recognition
techniques extract important information also from the derivative of the sensor
response[6]. Therefore they require an electronic interface circuit sufficiently
fast to preserve the transient data, i.e. that provides at least 10 accurate samples
of the resistance value of every element of the array per second. Thus both the
sensor micromodule and the IC-interface, reported in Fig. 1 should be fast
enough to grant sufficient dynamic data for this kind of feature extraction
method.

Fig. 1. Block diagram of the proposed gas sensing Microsystem.

2. Gas sensor development

According to the main guidelines of research on gas sensors, miniaturized
and low power consumption microsensors have been designed and fabricated.
These are chemoresistive gas sensors based on Metal Oxide thin films (MOX)
acting as gas sensing elements, realized with the sol-gel technique and deposited
by spin coating on silicon micromachined substrates equipped with integrated
heater/thermometer and interdigitated electrodes. Since these types of gas
sensors are based on chemiadsorption and charge transfer processes between the
gas molecules and MOX film causing a simple electrical resistance variation of
the gas sensing element, they are characterized by a real functioning easiness. To
this important feature, they join enhanced gas sensing properties due to the
nanocrystalline sol-gel MOX films, small size and reduced power consumption
due to the use of Si-micromachined substrates[3]. The developed substrates are
Si micro-hotplates (2x2 mm2 sized) with a central area constisting of a thin
dielectric membrane of silicon nitride (Si3N4)(-0.8 pm thick and -700x700pm2
lateral sized), suspended and supported by a silicon bulk rim. Inside the central
area, a smaller region (-300x300pm2 sized), the active area, on which the
sensing element is deposited, was realized. Following a vertical approach, a
274

buried platinum heaterhhermometer and IDC electrodes have been integrated in

the active area in different layers of the dielectric membrane. The
implementation of a thermometer in the Pt heater structure allows the
simultaneous measurement of the working temperature of the sensor. The Pt
interdigitated electrodes, deposited by RF sputtering, consist of n=5 fingers
50pm spaced, 2 0 p wide and 3OOp-n long; the use of interdigitated electrodes
with a large length-to-width ratio allows even the evaluation of sensing films
with very high sheet resistivity. The dielectric membrane provides the thermal
insulation between the active area, heated up to the high working temperatures
of the sensor (20O+40O0C),and the external Si-bulk fiame, that remains at room
temperature. In such a way the low thermal conductivity and inertia of the
micromembrane allows a low power consumption (in the range of 20+50mW)
and the possibility to operate not at constant but periodically variable
temperature (pulsed or oscillating), being the latter a measurement methodology
which improves the sensor selectivity. Fig. 2 shows a cross section scheme of
the designed Si-micromachined
substrates.

Silico
Silico nitride(Si,NJ
Platinu

- LTO (Siq)
Meta oxid (MOX)
Gold wir
- 4
07

f t t
'-

KOH
Fig. 2. Cross section scheme of the designed Si-micromachinedsubshares.

3. The IC interface
The read-out circuit[ lo] is composed of a single-ended continuous-time
programmable trans-resistance amplifier that converts the CWent Isens=VREF/Rsens
flowing through the sensor into a voltage and of a fully-differential switched-
capacitor oversampled 13-bit incremental A D converter that digitizes the output
of the first stage. In addition two 8-bit DACs and a DSP section allow us to
reconfigure the input stage features to match sensor specifications. The gas
sensor, modeled as a resistance R,,,,, is connected between a reference voltage
VREFand the operational amplifier (opamp) virtual ground, biased at v'. A
current I,,,, is then flowing across the sensor given by: I,e,,=(V~~-V+)/RSenS,
being the nominal voltage drop across the sensor 500mV. The sensor resistance
possible values range in the 5.2-decade extremely large interval [ 100Q-20MQ],
that would give a comparable and, then, extremely large dynamic range at the
opamp output voltage. This would result to be impracticable. In order to
 275

accommodate this large R,,,, range, the feedback resistance Rf is adjusted in

order to be comparable to Kens.In this way the output swing is comparable to
VREFand can be managed by the opamp. This Rf programming is done by
splitting the resistance values into 10 scales, each of them covering about half
decade. These different scales are generated by realizing the feedback resistor
with an array of different possible Rf to be properly connected for each scale.
The use of different scales requires to face the problems arising from the
mismatches (mainly in terms of offset and gain error) between consecutive
scales. A calibration technique is used for this purpose, which exploits a partial
overlap of adjacent scales of about a quarter of decade. Notice that the above
approach of adjusting the Rf value to match the R,,, value does not give a full
resistance matching, due to quantized values of Rf. Any mismatch between Rf
and R,,,, corresponds to a deviation of the output voltage and then limits the
accuracy. Fig. 3 shows the complete read-out schematic. In this circuit, the
regulated current source used to compensate inter-scale system offset mismatch
is realized with an 8-bit buffered resistive DAC (DAC1 in the schematic) and a
programmable resistor RDAC,that also needs to be selected from an array. In this
design we decided to have always RDAC=Rfin order to keep the opamp working
with gain and feedback factors of the same order of magnitude over the entire
dynamic range and to guarantee a good integrated component matching for RDAC
and Rf. Notice that Ical and I,,,, may range from 25nA to 5mA. Furthermore, by
regulating the sensor voltage reference (VREF)through an additional buffered
DAC (DAC2), it is possible to correct separately the gain-error of each of the 10
scales available in the circuit. As shown in Fig. 3, the two 8-bit DIA converters
(DAC1, DAC2) and the two selector circuits for Rf and RDAC(SEL1, SEL2) are
all controlled by a common digital unit, whose tasks are of course the choice of
the current measurement range and the actual correction of offset and gain error
for each scale using the "calibration words" determined during initial setup
phase.

VREF
\ L..

f12/~RDAca,~bu

-"" DACl
1C.l
'
Clock :

to RESISTOR
@
ARRAYS and DAC 2
, tt.I
DIGITAL

fl
-ONEN'

WORD
CALlBR4noN

Fig. 3. Integrated re-conjgurable front-end channel.

276

4. Electrical and chemical measurements

The electrical characterization of the interface circuit accuracy has been
performed testing separately every single one of the ten available partially
overlapped scales and then by re-constructing the full system transfer function
by canceling inter-scale offset and gain error mismatch. Thus in Fig. 4, the
relative error in the resistance value measurement, performed over the complete
5.3 decades dynamic range, may be plotted. These data underlines that the
accuracy of the front-end circuit in terms of relative precision of resistance value
measurement is better than 0.1% for almost the whole input dynamic range,
while the root mean square (RMS) electrical error in resistance sensing is better
than 250ppm. Several chemical measurements with different gas types (as
methane and ethanol) by using the designed micromodule system were
performed. A calibration gas sensing test-bench for electrical-conductance
measurements in controlled atmosphere is used for the experiments. In
particular, the microsensor is introduced into a stainless-steel-made cylindrical
test cell, connected to a multichannel mass flow programmer/display (MKS
mod.647B) driving three separate gas-channels connected to four distinct analog
mass flow controller (MFCs of MKS mod.ll79A). Dry air is used both as
reference gas and as diluting gas to obtain gaseous mixtures in air at different
concentrations. Certified bottles of synthetic air and methane (2000 ppm in dry
air) are used; whereas, for ethanol, different vapor concentrations were obtained
by diluting in air the saturated vapors coming from a ethanol bubbler. The
electrical characteristics of the developed microsensors are obtained measuring
by prototype front-end circuit connected to the sensor device. A photo of the
instrumentation setup is reported in Fig. 5 . For the gas sensing tests, undoped
and Pd-doped (with 5% weight of palladium) tin oxide based microsensors are
used. Fig. 6 shows the system transient response varying ethanol concentration,
from 4000ppmvol to 64OOppmVol. The coating used in these particular
measurements is a Pd-doped SnOz thin layer with a working temperature of the
hotplate of 350C.
0 150

0.125

Gg o"oo
-
>
-
I

B
0.075

0.050

0.025

2 1 4 5 6 7
10 10 10 10 10 10 1o8
Sensor Resistance [Ohm]

Fig. 4. Relative error of ,R measurement [ I 00!2-20M!2].

 277

Fig. 5. Complete ins~mentationusedfor chemical measurements.

D (0 20
-
30 10
wmni1-
w

Fig. 6. System transient response to ethanol.

70

eferences
Portable system for ambient gas monitoring with smart AD front-end improving sensors resolution and
accuracy, PRIN 2003091427, [ h ~ : / / i m s . ~ i p v , i ~ p r i n 0funded
3 ] , by Italian Government.
D. Barrettino, M. Graf, S. Hafizovic, S. Taschini, C. Hagleitner, A. Hierlemann, H. Bakes, A single-chip
CMOS micro-hotplate array for hazardous-gas detection and material characterization, IEEE ISSCC
proceedings, pp. 312-313,2004.
S. Capone, P. Siciliano, Gas Sensors from Nanostructured Metal Oxides, Encyclopedia of Nanoscience
and Nanotechnology, Vol. 3, pp. 769-804,2004.
G. Ferri, N. Guemni, V. Stornelli, A temperature control system for integrated resistive gas sensor arrays,
SPIE VLSI Circuit and Systems 11, Vol. 5837, pp. 972-982,2005.
General Olfaction and Sensing on a European Level, GOSPEL Network, [hrcp://w\lrw.gospel-network.org],
funded by European Community.
E. Martinelli, G. Pennazza, C. Falconi, A. DAmico, C. Di Natale, Comparison between two alternative
feature extraction methods for chemical sensor array, World Scientific AISEM proceedings, pp. 334-339,
2003.
M.Grassi, P.Malcovati, G. De Iaco, A.Basehirotto, An Integrated Wide-Range Resistance-to-Time
Converter With Decoupled Oseillator, AISEM book of abstracts, pp.172,2005.
A. Flammini, D. Marioli, A. Taroni, A low-cost interface to high value resistive sensors varying over a wide
range, IEEE IMTC proceedings, pp. 726-731,2003,
D. Barrettino, M. Graf, W. H. Song, K. Kirstein, A. Hierlernann, Hotplate-based monolithic CMOS
microsystems for gas detection and material characterization for operating temperatures up to 5OO/spl
deg/C, IEEE Joumal of Solid State Circuits, vo1.39, n. 7, pp. 1202-1207,2004.
[lo] P. Malcovati, M. Grassi, F. Borghetti, V. Ferragina, A. Baschirotto, Design and Characterization of a 5-
Decade Range Integrated Resistive Gas Sensor Interface with 13-Bit AD Converter, IEEE Sensors
proceedings, in press, 2005.
 Polymeric Replicas from Porous Silicon Template (and Its
Response to Vapour)

Vera La Ferrara, Ivana Nasti, Ettore Massera, and Girolamo Di Francia

ENEA, Centro Ricerche Portici Localitri Granatello 80055 Portici @A)
vera.lafeirara@,portici.enea.i t

Summary
Polystyrene replica is realized preparing polystyrene and depositing the polymer onto a
nanostructured porous silicon matrix used as a template. The thin polymer film is peeled off
from the substrate and exposed to organic vapours. Swelling affects the optical reflectance
and, by a specially adjusted set-up, the kinetic of the vapour desorption is measured in ordcr
to investigate replicas as drug delivery systems.

1 Introduction
Nanometer sized structures are emerging for the development of future
devices in several different fields. Polymeric materials have, in fact, many
characteristics that make them ideally suited for defining nano-electromechanical
structures even with low-cost processes such as replication and templating.
Flexible polymers have been actually used as replicas of nanostructured template
to fabricate sensors, deformable and tunable opticals fibers and biocompatible
materials. They have good mechanical property (Youngs modulus can be tuned
over two orders of magnitude), are cheap and have the property to tight with
silicon and glass permitting to design hybrid devices that could contain silicon
electronics, light sources and detectors [ 11. Moreover porous polymers are quite
attractive due to their porous size related tunable diffusion kinetic . As a matter
of fact, swelling-controlled release systems are actually used as drug release
systems in a wide range of applications from simple nasal to buccal to more
complex bioadhesive uses [2]. Recently, porous polymer nanostructures
produced from an oxidized porous silicon template have demonstrated their
effectiveness for drug delivery applications [3]. Because of the tunability of
porous silicon surface, it is possible to fabricate several polymer replicas. In this
way, it is possible to transfer the mechanical properties of porous silicon to a
variety of organic polymers, tuning their drug release time.
In this work we have realized porous polystyrene replicas using as a template a
nanostructured porous silicon (PS) matrix. To investigate the feasibility of such
replicas as drug delivery systems, the thin polymer film has been peeled off from
the substrate and exposed to organic vapours. A new optical measurement

278
 279

method based on the modification of the optical reflectance induced by the

polymer swelling is also proposed in order to measure the desorption kinetic.

2 Experimental
To fabricate the porous silicon masters, PS samples have been prepared by
anodic etch of n-type <100>-oriented Si with 1 ohm*cm resistivity in a solution
50% aqueous W:isopropilic alcohol (7:3 by volume) with a current density of
40 diem' for 10 minutes. Samples have been partially oxidized in ambient air
for about one month. Polymer has been prepared using a solution of polystyrene
in toluene obtained on a rotating plate at 25C for about 1 hour. The porous
polymer replica has been prepared by means of polystyreneltoluene solution cast
into oxidized porous silicon and dried in ambient air for two days.
The porous silicon template has been immersed in W for lh. Resulting
composite porous silicodpolystyrene has been mechanically peeled off from the
porous substrate and then washed in 10% NaOH to remove nanoporous phase on
polymeric replica surface. Porous polystyrene replica, so obtained, has a
thickness of 300 nm. A polystyrene film, with the same thickness, peeled off
from silicon substrate, has been fabricated and measured to compare the
chemical and morphological properties with porous polystyrene replica ones. In
figure 1 we show a SEM image of the porous polystyrene replica.

Fig. 1 SEM imagc of porous polystyrene replica after removal fiom porous silicon template and
washing in NaOH.

In order to characterize polystyrene replicas as controlled release devices,

polystyrene films have been put in a test chamber where acetone saturated
vapour (20%) mixed with nitrogen has been introduced as swelling agent.
Optical reflectance has been measured by means of an optical fiber (illuminating
samples with an halogen lamp) and collecting data at a fixed wavelength (631
nm) with a CCD spectrometer (figure 2). Total gas flow has been set to 500
sccm in all the performed tests.
280

Fig. 2 Experimental set up for optical measurements. Acetone saturated vapour mixed with nitrogen
has been introduced in the test chamber controlling total gas flow by a mass flow meter. An optical
fiber illuminates samples and its optical reflectance is collected by a CCD spectrometer.

esults and discussion

Optical reflectance dynamic response of polymeric replica has been recorded
switching from nitrogen to acetone saturated vapour for both the polystyrene
film and the porous polystyrene replica (figure 3).

1851 nitrogen
I
roo.
0.85-

OW-

I
Iif +
i
nltr
0.70- I

0.65-

(4 @)

Fig3 Reflectance measurement of (a) the polystyrene film and (h) the porous polystyrene replica
switching from nitrogen to acetone saturated vapours.

Measurements show that the mechanism of acetone release of polystyrene

replicas is different respect to polystyrene one.
 281

At this aim, the desorption kinetic has been investigated analyzing reflectance
decreasing, obtained when nitrogen is introduced in test chamber. Results are
shown in figure 4.

256- k, = 23 2
,'

zi;
~ ~ ' 8
200- B,'
7,'
8,"
150- ,'
,/' "
U 2
2 1W- /* ,,'8 M,/k = k,' t""' + k,' t

50- ,F'
f' 1w -

0- ,,'
/ 0- k,=81

1 . , . , . I . , . , . ! . / .

As expected, measurements show that mechanism of acetone release is strictly

Fickian only for porous polystyrene replica. In figure 4a it is reported linear
behavior of acetone release for polystyrene film. In figure 4b it is evident a
Fickian behavior for acetone release kinetic. Peppas' equation [4] reported in
figure and used for data fit, describes the fractional drug release. Different
constant k are probably due to morphology of porous polystyrene replica.

4 Conclusions
We have shown how it is possible to transfer porous silicon morphology onto
polystyrene and obtain polymeric replicas capable to tune the kinetic of chemical
release when exposed to organic vapour.

References
[l] S.R.Quake and A. Scherer, Science, vol290 (2003) 1536-1540
[2] N.A. Peppas and P. Colombo, J. of Controlled Release, 45 (1997) 35
[3] Y. Y. Li, F. Cunin, J. R. Link, T. Gao, R. E. Betts, S. H. Reiver, V.
Chin, S. N. Bhatia, M. J. Sailor, Science 299 (2003) 2047
[4] N.A. Peppas, P. Bures, W. Leobandung and H. Ichikawa, Europ. J. of
Pharmaceutics and Biophaxmaceutics, 50 (2000) 27
 A Software Tool for the Simulation of IPMC Actuators and
Sensors

C . Bonomo, L. Fortuna, P. Giannone, S . Graziani, S . Strazzeri

Dipartimento di Ingegneria Elettrica Elettronica e dei Sistemi
University of Catania
VialeA. Doria, 6 - 95125 - Catania, ITALY
Tel. +39 095 7382327-Fax: +39 095 330793
pietro.giannone@,diees.unict.it

D. Harvey, G.S. Virk

School of Mechanical Engineering
University of Leeds
WoodhouseLane, Leeds, West Yorkshire, UK. LS2 9JT
Tel. +44 113 343 2156- Fax: +44 113 343 2150

Summary
Ionic Polymer Metal Composites (IPMCs) are emerging materials made of an ionic polymcr
membrane metallized on both sides with a noble metal. They can work either as a low-
voltage-activated motion actuator or as a motion sensor. When an electric field is applied
across the IPMC thickness a deformation occurs, while if they arc dcformcd they generate a
detectable voltage.
IPMCs have the additional benefits of being pliable lightweight, soft and so forth. These
benefits make IPMCs suitable for applications in very different fields including robotics,
aerospace, biomedical, etc.
In this paper the authors introduce a software tool to simulate the behaviour of both IPMC
sensors and actuators working in air. The software tool is completed by a graphical user
interface (GUI) developed in Matlab@. The GUI has been developed in order to give a
useful tools that allow the user to understand the behaviour and the parameters involved in
the phenomena. In the GUI, parameters affecting the sensing and the actuating properties of
IPMC-based transducers arc taken into account together with their dynamic characteristics

1 Introduction
Ionic Polymer Metal Composites (IPMCs) are emerging materials used to
realize motion actuators and sensors [1],[2]. In the former case a voltage input is
able to cause the membrane to bend while in the latter case by bending an IPMC
membrane, a voltage output is obtained. Figures l a and lb show the basic
working principles of an IPMC respectively as actuator and as sensor.

282
 283

Several applications can be found in the literature regarding these materials

~3l,l?I7[51.

Fig.1 (a) The IPMC working as actuator and @) as sensor.

Based on the analytical models introduced in [7] and [XI, a software tool for
the simulation of the IPMC behaviour, both as motion actuator and sensor, has
been implemented. The models, on which the software tool is based, are
organized into blocks. As far as the IPMC actuator is concerned, the model is
divided in to a cascade of two sub-blocks, the first is devoted to predicting the
absorbed current as a function of the input voltage, and the second that
transduces the absorbed current either to a deflection or in a force, as requested
by the user.
As for the sensor model, the software for the sensor simulation has been
implemented by a single block able to convert the mechanical stimulus into an
electrical signal.

2 The IPMC
C ~Working nasactuator
o r ~ as gactuator
As largely reported in the literature [l], [5], [6], the IPMC mechanical
behaviour is produced basically from the charge accumulation. Starting from this
assertion, the solution to adopt two sub-blocks allowed the estimation of the
current absorbed by applying a voltage across the electrodes. The current is then
converted to a mechanical reaction by using an electromechanicalmodel.
The electrical transduction was obtained by using an equivalent nonlinear circuit
in order to estimate the non-linear behaviour between the applied voltage and the
absorbed current [7]. The electromechanical transduction is obtained by using
the well-know Euler-Bernoulli theory of the pinned beam and an adaptation of
the theory of the piezoelectric coupling effect. A scheme of the model to
implement the proposed software is reported in Figure 2.
The first step to carry out is the identification of the parameters in the
model for different IPMC samples, in order to create a database. The
284

identification is performed by testing the samples, acquiring the necessary data

and optimising the parameters so that the model matches the response.

f Electromechanical model

Elertried tmnsduction

Fig2 Scheme of the electromechanical model.

2.1 The actuator software

The software tool is very easy to use: the GUI appears clear, and each
parameter is well explained. The graphic interface shown in Figure 3 gives the
reader an idea on how it is possible to predict the nonlinear behaviour of the
IPMC at low frequency.

Fig.3 The GCJI of the software toolproposed for the actuator.

The software interface requires both the dimensions of the IPMC, which
desired behaviour is to be simulated and the applied voltage signal. On the right
 285

hand side of the panel the signals involved in the simulation process are shown.
Those signals are respectively: the applied voltage, the absorbed current, the free
deflection, the blocking force and the voltamogram.
If a voltage sweep frequency is selected as input, the software is able to
estimate also the resonance frequency and the amplitude of the deflection at that
frequency.

3 The IPMC Working as sensor

As far as the IPMC working as sensor is concerned, a linear model was
suggested based, as previously mentioned, on the Euler-Bernoulli theory of the
pinned beam and on an adaptation of the theory of the piezoelectric coupling
effect [8]. The scheme of the electromechanical model used to implement the
simulation software, together with the involved quantities, is reported in Figure
4.

Fig.4 Scheme ofthe mechano-electrical model.

e $ e ~ s o $oftware
r
The GUI of the sensor can be opened directly by the window shown in
Figure 3, in particular it is sufficient to select the sensor button to switch in the
sensor simulation mode. Figure 5 shows the software interface to simulate the
IPMC sensor behaviour. In this case the input signal is not a voltage but a
deflection.
A typical simulation phase is also reported in Figure 5 where the input signal
is a frequency sweep deflection. As it can be seen from the same figure, the
sensor response depends on the frequency of the input signal. In particular, the
sensor shows a typical band pass behaviour.
Acknowledgements
The research activity presented in the paper is supported by the EU under the V1
Framework project ISAMCO NMP2-CT-2003-505275; the membranes used in
the mentioned measurement surveys were supplied by CCR (Consorzio Catania
Ricerche), as part of the project collaboration activity.
286

Fig.S The GUI of the software tool proposedfor the sensor.

e~erences
111 M. Shahinpoor and K. J Kim, Ionic polymer-metal composites: I.
Fundamentals, Smart Materials and Structures 10, 200 1, pp 8 19-833.
P I M.Shainpoor, Y.Bar-Cohen, J.O. Simpson, and J. Smith, Ionic Polymer-
Metal Composites (IPMC) as Biomimetic Sensors, Actuators, and Artificial
Muscle - A Review, Int. J. Smart Materials and Structures, Vol. 7, 1998, pp.
R15-R30.
131 Shahinpoor M.,Kirn J. K., S., IV. Industrial and medical applications, Smart
Materials and Structures 14 (2005) 197-2 14
[41 Tung S., Witherspoon S. R., Roe L. A., Silano A., Maynard D. P., Ferraro
N., A MEMS-based flexible sensor and actuator system for space inflatable
structures, Smart Materials and Structures 10 (2001) 1230-1239
151 Bar-Cohen, Electroactive Polymers As Artificial Muscles-Capabilities,
Potentials And Challenges, HANDBOOK ON B I O M I ~ T I C SYoshihito
,
Osada (Chief Ed.), Section 11, in Chapter 8, Motion paper##134,
publisher: NTS Inc., Aug. 2000.
161 S. Nemat Nasser, J. Y. Li, Electromechanical response of Ionic Polymer
Metal Composites, Journal of Applied Physics, Vol. 87 (2000), no 7,
pp.3321-3331.
171 C. Bonomo, L. Fortuna, P. Giannone, S. Graziani, An Electric Circuit to
model Ionic Polymer-Metal Composites as Actuator, IEEE Transactions on
Circuits & Systems Vol. 53 No. 2 2006.
[XI C. Bonomo, L. Fortuna, P. Giannone, S. Graziani and S. Strazzeri, A model
for ionic polymer metal composites as sensors, Smart Materials and
Structures 2006, Vol. 15, pp.749-758.
 Synthesis and Characterization of Imprinted Ti02
Nanoparticles. Preliminary results

Rosaria Anna Picca, Cosimino Malitesta

Laboratorio di Chimica Analitica, Dipartimento di Scienza dei Materiali,
Universita di Lecce, Via Arnesano, 73100 Lecce
[email protected]

Daniela Manno
Laboratorio di Microscopia elettronica, Dipartimento di Scienza dei Materiali,
Universith di Lecce, Via Arnesano, 73100 Lecce

Summary
The synthesis of Ti02 nanoparticles (Ti02-NPs) in the presence of a target analyte (I-
tyrosine) has been developed to prepare imprinted nanostructured materials. The effective
growth of Ti02-NPs and their size distribution in stable dispersions has been evaluated by
HRTEM analysis. UV-Vis spectroscopic characterization has been performed to study
interactions between Ti02-NPsand I-tyrosine.

Introduction
Molecular imprinting [11 has spread out as a viable technique to realize high
selective recognition sites for a target molecule. Imprinted polymers (MIPs) are
employed for several analytical applications as they can mimic bioreceptor
behavior while improving stability and reducing costs [l]. In particular, they
were also proposed as active layers in sensors to detect different analytes such as
drugs, amino acids, toxins, pesticides [2]. However, conventional MIPs suffer of
some disadvantages: heterogeneous site affinity (non-covalent approach),
incomplete template removal, reduced number of active sites and difficult access
to them [3, 41. Thus, alternative imprinting methodologies and materials have
been developed to overcome these limitations [4, 51.
Inorganic oxides appear suitable candidates for imprinting procedures: in
fact, they possess rigidity, high surface area and porosity [ 6 ] .Among them, SiOz
and Ti02 3-D matrices are easily realized via sol-gel process. In last years, other
techniques have been applied to prepare imprinted thin films directly on a
substrate (for example, the surface of a transducer): surface sol-gel process 781,
liquid phase deposition [8], spin coating [9]. As a result, recognition sites are
present in a high number and located near or at the surface, thus mass transport
is favored enhancing sensor performances.

287
288

Nanoparticles (NPs) possess these properties in even higher degree leading to

the development of sensors based on ZrOz-NPs [lo] and even on single-
nanoparticle [ 1I]. On the basis of these considerations, it appears very attractive
to conjugate the selectivity conferred by imprinting technology and the
sensitivity of NPs in sensors.
To our best knowledge, no work has been presented on imprinting of TiOz-
NPs. In this paper it is reported for the first time the synthesis of imprinted TiOz-
NPs. The selected template is Z-tyrosine and growth of TiOz-NPs is performed by
room temperature hydrolysis of titanyl oxalate complex as illustrated in our
previous work [12]. In fact, adding Ti (IV) salt to an Z-tyrosine (Tyr) solution
could lead to the formation of colloidal titania imprinted by this amino acid. This
target molecule was chosen as it plays a central role in biological systems
because it is not only the precursor of dopamine, etc. but its metabolism is also
involved in some diseases (Parkinsons, atherosclerosis, etc.) [ 131. HRTEM and
UV-Vis spectroscopy characterizations are presented to assess NPs formation
and to evaluate occurrence of imprinting effects.

~xperi~ental
KzTiO(Cz04)*2Hz0 and 1-tyrosine were purchased fiom CAFU,OERBA
Analytical Grade, while phenol (> 99%), HC104 and NaOH from Sigma. All
chemicals were used as received. All solutions were prepared in ultrapure water
(Milli-Q, Millipore, 18.2 MCi*cm).
UV-Vis measurements were performed with a Varian Cary 50 UV-VIS
spectrophotometercomputer-controlled by CaryWin-UV software.
A JEOL 2010 microscope was used at an acceleration voltage of 160 kV for
HRTEM analysis.

Results and discussion

HRTEM analysis (Figure 1) was carried out to assess NPs growth in the
presence of Tyr. and their size distribution A mean diameter of 2 nm was found.

Fig.1 HRTEM micrographs and size distribution of NPs grownfrom Q r 2.5+1U4M with titanyl
oxalate 1u4M
 289

As already shown [ 121, TiOz-NPs growth in titanyl aqueous solutions can be

followed by UV-Vis spectroscopy as they present an absorption band between
250-350 IUII.
As also Tyr shows two absorption peaks (Figure 2 - Tyr (0 min)) at 225 and
275 nm, spectroscopic characterization was employed to record time evolution
(Figure 2) of Ti02-NPs-Tyr system.
4.0

3.5

3.0

2.5

4 2.0

1.5

1.0
350 464
0.5 Ilnm

0.0
200 250 300 350 400
X/nm
Fig.2 UVspecha of 2.5*104 A4 Tyr solution with titanyl oxalate IQ' M.T i 4 spectrum after 24 hours
growth and TiO2-NPs-Tyrspectrum after 28 days in the m e t

As it can be noticed, the usual appearance of TiOZband is observed, while

Tyr absorption disappears after 28 days, leaving a spectrum quite similar to TiOl
alone (see the inset). This evidence supports the hypothesis of tyrosine inclusion
in TiOz-NPs during their growth. The possible interactions of tyrosine
carboxylic moieties with TiOz surface [ 141 are not evident in this case, because a
350 nm red shift in the TiOz band did not occur.
It was also evaluated if pH environment can influence the formation of TiOz-
NPs (in titanyl and/or titanylhyrosine systems).
Extremely acidic or basic media (HC104 2 M, NaOH 1 M) cause inhibition of
nanoparticle nucleation in both systems. No change is observed when titanyl is
added to acid (see figure 3) or basic (data not shown) solutions of Tyr. On the
other hand, acidification (or alcalinization) of Ti02-Tyr samples (3 hours aged)
does not alter UV spectra evolution confirming that NPs nucleation, not growth,
is inhibited by extreme pHs.
290

iy-acid
imin -
15min -
30min -
~h .___
3h -
q h -
4ah -

I I I I I
200 250 300 350 400

h/nm
Fig.3 UVspecira of 2 . ~ 1 0M. ~Tyr solution in HC104 2 Mbefore andafrer adding I U 4Mtitanyl

As Ti02 photocatalytic effects [15] are well known, it is important to

estimate sunlight influence on TiOz-NPs formation. Thus, some aliquots, both
for titanyl and titanyVTyr systems, were stored in the dark and exposed to light
only during UV measurements. The recorded UV spectra indicate that radlation
does not affect colloids.
Formation of Ti02-NPs in presence of phenol was investigated to assess the
role of -OH phenolic group in imprinting by Tyr. Figure 4 shows that increasing
in absorption occurs within 24 h and it depends on Ti02-NPs growth as no
significant modification of phenol spectrum is evident.
phenol
5 min
lh
3 1 .- ... 3 h

- . ..
~ 10 davs

I I I I I
200 250 300 350 400
hlnm
Fig.4 UVspectra of 2.5*10-4Mphenol solution before (yellow) and afrer addiction of I F 4 M
tiranyl oxalate
 291

As a result, it can be concluded that phenolic hydroxyl group is likely not

responsible for binding with titania.

Conclusions
The first example of synthesis of TiOz-NPs imprinted for I-tyrosine has been
proposed, based on hydrolysis of a titanyl complex in Tyr aqueous solutions.
HRTEM characterization has been developed to confirm Ti02-NPs formation.
Spectroscopic investigation of Ti02-Tyr system has revealed that recorded
spectra are not the sum of single species absorption profiles but there is a
significant time evolution. In particular, while TiOz-NPs grow also in this case,
Tyr absorption decreases during time until disappearing after almost one month.
Extreme pH conditions inhibit NPs nucleation also in the titanyl/Tyr system. The
use of phenol as template has demonstrated that phenolic moiety does not
interact with titania. These preliminary results on spectroscopic analysis indicate
that inclusion (hopefully imprinting), rather than adsorption, of amino acid
molecules into TiOz-NPs occurs. Further characterizations (Resonance Raman
Spectroscopy, ATR-IR) will be performed to fully understand interactions and
separative techniques (Flow Field Fractionation) will be applied to isolate
imprinted NPs for the preparation of nanostructured thin films as recognition
elements in piezoelectric sensors.

References
1. K. Haupt, Chemical Communications, (2003), 171
2. S.A. Piletsky, S. Subrahmanyam, A.P.F. Turner, Sensor Review, 21 (2001), 292
3. N. Sallacan, M. Zayats, T. Bourenko, A.B. Kharitonov, I. Willner, Analytical Chemistry,
74 (2002), 702
4. S.C. Zimmerman, M.S. Wendland, N.A. Rakow, I. Zharov, K.S. Suslick, Nature, 418
(2002), 399
5. A. Katz, M.E. Davis, Nature, 403 (ZOOO), 286
6. D.Y. Sasaki, Molecularly Imprinted Polymers Man-made mimics of antibodies and their
applications in analytical chemistry, B. Sellergren (Editor), Elsevier (2001) Amsterdam
(The Netherlands) 213
7. S.-W. Lee, I. Ichinose, T. Kunitake, Langmuir, 14 (1998), 2857
8. L. Feng, Y. Liu, J. Hu, Langmuir, 20 (2004), 1786
9. M. Hashizume, T. Kunitake, Langmuir, 19 (2003), 10172
10. G. Liu, Y. Lin, Analytical Chemistry, 77 (2005), 5894
11. S.-K. Eah, H.M. Jaeger, N.F. Scherer, G.P. Wiederrecht, X.-M. Lin, Applied Physics
Letters, 86 (2005), 03 190211
12. C. Malitesta, A. Tepore, L. Valli, A. Genga, T. Siciliano, Thin Solid Films, 422 (2002),
1I9
13. G.G. Huang, J. Yang, Biosensors and Bioelectronics, 21 (2005), 408
14. T. Rajh, J.M. Nedeljkovic, L.X. Chen, 0. Poluektov, M.C. Thurnauer, Journal of Physical
Chemistry B, 103 (1999), 3515
15. T. Cassagneau, J.H. Fendler, T.E. Mallouk, Langmuir, 16 (2000), 241
 A SINGLE CHIP GENERAL PURPOSE CIRCUIT FOR
SENSOR INTERFACING AND CHARACTERIZATION

M. Schipani, T. Ungaretti
STMicroelectronics, via Tolomeo I , 20010 Cornaredo (MI) - Itab,
e-mail: monica.schipani@iet. unipi.it

P. Bruschi, N. Nizza
Dipartimento di Ingegneria dell 'Informazione, Universitb di Pisa,
via G. Caruso, 56122 Pisa - Italy

Summary
A single-chip general purpose interface for MEMS sensors has been introduced. The
capacitive readout circuit, consisting in a fully differential switched capacitor circuit
implementing CDS and Chopper Stabilization Techniques is designed and illustrated. The
circuit can be tuned to work with sensors of nominal capacitance values ranging from 200fF
to 3pF, with a maximum differences between the two sensor elements of 500fF. The
effectiveness of the input common mode stabilization circuit is demonstrated by means of
electrical simulations.

Introduction
MEMS sensors are very interesting devices thanks to their microscopic
dimensions, high sensitivity and low power consumption. The mechanical sensor
can be integrated with its electronic readout interface, obtaining a complete
system on chip (or system in package), the capabilities of which make MEMS
suitable for a wide range of application areas such as, for example, automotive
industry, consumer electronics and industrial processes.
For each particular application, requiring a specific sensor, the electronic
readout interface is designed according to the sensor characteristics and this
increases the costs to produce a complete system. A valid solution to this
problem is represented by a general purpose interface, capable of reading out
different type of sensors. Such an interface represents also a good platform to
quickly validate a new sensor architecture, reducing time to market.
A large variety of integrated sensors falls within the two main categories of
capacitive and resistive sensors. Examples are accelerometers, pressure sensors
and microphones. In this work, an interface with flexible read out channels for
both capacitive and resistive sensors is presented.
A general overview of the whole chip is introduced, describing the various
specifications of interest and how to meet them in terms of block diagram. Then
we focus on the capacitive readout circuit that is a discrete-time (switched

292
 293

capacitor) circuit, implementing a C N (capacitance-to-voltage) conversion. This

solution, differently from other techniques such as capacitance-to-time or
capacitance-to-frequency conversion, allows to have both analogue and digital
output, increasing the versatility of the interface. In this paper, the switched
capacitor front end is presented in details, with particular emphasis on the input
common mode control circuit. In particular the beneficial effect in terms of
offset and sensitivity to parasitic capacitance mismatches reduction is shown.

Description of the general purpose interface

The schematic block diagram of the general purpose interface is represented
in Fig. 1. Different readout channels can be multiplexed to ensure the
compatibility with the different type of sensor. Sigma-delta and S A R analog-to-
digital converters for either precision and velocity performance, respectively,
will be included. The interface is equipped with the testing and trimming
circuitry, the voltage and current references and the generators of all the required
clock signals. The interface has been designed to be able to communicate with
low cost microcontroller with no need of external components so also the
circuitry implementing SPI and 12Cprotocols will be included.
- _.
capacitive readout
interface
mux mux - ADC
10 bit SAR

-
16 bitE.4
interface

;ensor
driving channel
temperature
sensor

current interface circuit phase

reference

Fig. I Schematic block of general purpose sensor interface

The capacitive readout channel is a fully differential switched capacitor

circuit implementing Correlated Double Sampling (CDS) [ 13 [2] and Chopper
Stabilization Techniques. This system, represented in Fig. 2, works with a three
phase clock in order to cancel offset and l/f noise. Careful design of the fully
differential amplifier has been carried out to minimize the thermal noise. This is
required since thermal noise represents the main source of uncertainty in CDS
systems because of the foldover mechanism.
294

An internal trimming circuit provide for sensor capacitance mismatch

calibration.
The gain and the bandwidth of the system are widely programmable to better
meet applications requirements.
According with common MEMS sensors characteristic, the nominal value of
input capacitances Cs, and Cs2 can vary from 200fF to 3pF with a maximum
difference of 500fF (absolute value).

Fig. 2 Capacitive readout circuit

The chosen architecture requires the input common mode (ICM) to be

constant to avoid gain, offset and non linearity errors, thus an input common-
mode feedback (ICMFB) circuit has been included. The values of Cxand CFB
have to be chosen satisfying the equation
CAN + cxvx + ~ F B Y C h 4 F B= O
where Cxis the mean value of the two sensor capacitances.
In order to make ICMFB efficient for the whole sensor capacitance range, all
capacitor used in ICM control circuit has to be opportunely programmable.
Therefore, capacitive networks similar to that represented in Fig. 3 are used.

Fig. 3 Programmable capacitive network

The amplifier used in ICMFB is a folded cascode with an input differential

pair (see figure 4) that senses the variation of input common mode voltage with
respect of a voltage reference.
 295

Fig. 4 Input differential pair of the ampli3er used in ICMFB

Simulation results
Fig. 4 shows the output waveform of the main blocks of the circuit. The input
signal is a 3V square wave. The mean value of sensor capacitances Cs is 3 0 0 s
with a difference of 1fF. An offset of lOmV has been introduced by means of a
voltage source in series with the GAIN amplifier inputs.

Fig. 4 Output waveform of the main block of the circuit

In the figure, the three working clock phases are highlighted. After the reset
phase $I,, offsets and low frequency noise are stored during the phase +2 and then
the input signal is amplified during the phase $I3. The valid output value is the
sampled at the end of phase $I3.
The effect of programming the ICMFB capacitors is shown in Fig.5. The
sensor capacitances Csl and Cs2 are set either to 1.3pF (so there is no input
signal) and unbalanced parasitic capacitance are placed at the input nodes stat1
and stat2. Under this conditions an undesired output signal is due to the input
common mode voltage variation.
296

looooo 0000000001 00010 00011 00100 00101 00110 00111 00111 00111 00111
, . , . , . , . , . , I

180 200 220 240 260 280 300

Time (ps)

Fig. 5 The effeci of programming ICMFB capacitors

The progressive variation of the code that sets the capacitors towards the
optimum value is shown on the bottom. The upper plot represents the input
common mode voltage while the lower plot is the output voltage. As expected,
for the correct code value the input common mode voltage is constant and there
is no output error, so the ICMFB works properly.

References
1. Baschirotto, A. Gola, E.Chiesa, E. Lasalandra, F. Pasolini, M. Tronconi
and T. Ungaretti, A +lg Dual-Axis Linear Accelerometer in a
Standard 0.5pm CMOS Technology for High-Sensitivity Applications
IEEE Journal of Solid State Circuits, vol. 38, no. 7, July 2003.

2. M. Lemkin, B.E. Boser, A Three-Axis Micromachined Accelerometer

with a CMOS Position-Sense Interface and Digital Offset-Trim
Electronics IEEE Journal of Solid State Circuits, vol. 34, no. 4, April
1999.

3. J. Zhang, K. Zhang, Z. Wang, and A. Mason, A Universal Micro-

Sensor Interface Chip with Network Communication Bus and Highly-
Programmable Sensor Readout, Midwest Symposium on Circuits and
Systems, vol. 2, pp. 246-249, August 2002.
 A Three-Axis Low-g Ultracompact Linear Accelerometer

F. Dalena, G. Roselli, P. Flora

STMicroelectronics
c/o Distretto Tecnologico, via per Amesano Km 5, 73100 Lecce, Italy
Jfrancesco.dalena, niuditta.roselli, pasciuale.flora1@ st.com

Summary
This paper describes a three-axis low-g ultracompact linear accelerometer that includes a
sensing element and IC interface processing the information from the sensing element and
providing an analog signal to the external world. The ultracompact accelerometer has been
designed for consumer market where the chip size becomes smaller and smaller. The proposed
device is a three-axis accelerometer with a LGA package of 5x3xlmm. Thanks to the very
small size of the device its main applications are: gaming, virtual reality, ffee-fall detection,
mobile terminals, appliance and robotics.

1 Introduction
As with any technology, certain market demands must to be attained prior to
widespread adoption in a broad range of end applications. In the case of MEMS
(Micro Electrical Mechanical System) technology, low cost and higher
sensitivities (new awareness of position and movements detected with higher
accuracy) had to be achieved.
Today, communication, consumer and industrial markets are considered
emerging markets for h4EMS devices and will be the key drivers for their
success.
Capacitive accelerometers, and in general capacitive microsensors, have the
main advantages of low-power, high-sensitivity, relatively simple structure,
inherently low temperature sensitivity, and is easily integrated with CMOS.
This paper describes a new three axis linear accelerometer in a LGA 3 x 5 ~ 1
mm package. Thanks to the ultracompact dimension it is possible the use of this
accelerometer in various consumer market application like: gaming, virtual
reality, free-fall detection for data protection, picture browsing, mobile
terminals, appliance and robotics [ 13.
The device contains two dies: 1) sensing element; 2) IC interface.
The electrical signal generated by mechanical part is very small so the IC
interface must have very low noise and no offset.

297
298

The IC interface is factory-calibrated for sensitivity, Zero-g level, internal

oscillator, currentlvoltage references, output resistor and the trimming values
are stored inside the device by a non volatile structure.
The three axis accelerometer works with a power supply from 2,16V to 5,5V
with a current consumption of 640 FA at 2,4V; it is able to sense k2g at full
scale with an acceleration noise density of 50pg /a.
2 Mechanical Sensor
Fig. 1 shows the sensor equivalent electrical circuit: it consists of two
sensing capacitors C, and Cs2 between the single rotor and each of the two
stators. Some additional parasitic capacitors are present between each stator and
ground. These capacitors are due to the mechanical die, IC interface and
interconnections between two dies.
When acceleration is applied to the sensor the proof mass displaces from its
nominal value, causing an imbalance in the capacitive half-bridge Fig. 1.
So an external acceleration causes a reduction in the value of one
capacitance and improves the value of the other capacitance. This differential
capacitive sensor has the important property of reducing the entire common
mode signal that the mechanical part could sense. This imbalance is measured
using charge integrator in response to a voltage pulse applied to the sense
capacitor.

3 stator2
Fig.1 Electrical schematic of the mechanical part
 299

3 ASIC
Interface circuitry is an important part of integrated inertial sensors. The
design of interface circuits with high sensitivity, low noise and large dynamic
range is very challenging. Prior interface circuit designs for capacitive sensing
were mainly focused on high sensitivity and low electronic noise, but low power
consumption has not been achieved. However, in consumer applications, low
power operation is always a predominant requirement.
The differential capacitive sensor needed a fully differential IC interface to
take advantage of the differential capacitive sensor structure. This means a
bigger chip area but also a lot of advantages like: power supply

Fig.2 ASIC Block Diagram

The accelerometer measures changes in a capacitive half-bridge to detect

deflections of a proof mass, which result from the acceleration input [2]. When
an acceleration is applied to the sensor, the proof mass displaces from its
nominal value, causing an imbalance in the capacitive half-bridge. This
imbalance is measured using charge integrator in response to a voltage pulse
applied to the sense capacitor. This capacitive position sensing has a low
intrinsic temperature coefficient, can be highly sensitive, and is easily integrated
with CMOS [3]. The sensing element is manufactured using a dedicated process
called THELMA developed by ST to produce inertial sensors and actuators in
silicon. The IC interface is manufactured using a CMOS process that allows
high level of integration to design a dedicated circuit, which is trimmed to
better match the sensing element characteristics.
300

The measure chain blocks Fig.2 realize a low-noise capacitive amplifier

that implements a Correlated Double Sampler (CDS) at its output to cancel the
offset and l/f noise.
A fully differential structure [4] is used to read the charge difference which
flows through the sense capacitors in response to a voltage step. The differential
structure has several other important advantages, including improved power-
supply rejection ratio and first-order rejection of common-mode errors, such as
switch charge injection and substrate noise. The output stage is composed of a
multiplexer, controlled by three different digital phases necessary for the TDM
(Time Division Multiplexing) technique, and three S&H, working in the same
manner for X, Y, Z , channels and converting the differential signal into a
single-ended one, compatible with the external world.
In addition, the device provides an embedded multiplexer to allow the
redirection of either the analog output signals or an auxiliary input signal onto a
single pin for operation with a single channel A/D converter. For instance, an
application could be the temperature measurement if a signal is applied to the
TEMP PAD and it is redirect on MUX output PAD.
All the analog parameters are ratiometric to the supply. This allows that
increasing or decreasing the voltage supply, the sensitivity and the offset will
increase or decrease linearly.
The IC interface is factory-calibrated for sensitivity, Zero-g level, internal
oscillator, currentholtage references and the trimming values are stored inside
the device by a non volatile structure.
The only external components are the capacitors at the output pins to
implement low-pass filtering for antialiasing and noise reduction. The cut-off
frequency (ft) of the external filters is:
1
f, =
-
2n * R,, Cload( x , Y , 2)
The internal filtering resistor (ROUT)has a nominal value of 32ka and it is
factory-calibrated at 10%. The cut-off frequency is function of the external
capacitance:

This means that large capacitor values correspond to high measurement

resolution.
A very useful feature to test the sensor mobility and the front-end has been
included in the ASIC. It is called self-test and it is governed by ST pad. The
ST pad is put to gnd in normal mode. When the ST pad is put to Vdd an
actuation force is applied to the sensor simulating an input force. This produces
a change in the DC level of the analog output of the device.
 30 1

The microsystem [l] exhibits a r2g full-scale, operates from a minimum

power supply of 2,16V and with a current consumption of 650pA. It is capable
of measuring accelerations over a maximum bandwidth of 2.0 kHz on a
temperature range of -40 C to +85 C. The acceleration noise density is equal
to 50pg I&.

4 Conclusion
In this paper a complete fully-integrated three-axis linear accelerometer in a
LGA 3 x 5 ~ 1mm package has been presented. The accelerometer has three
ratiometric analog outputs and one multiplexed output to interface the sensor
with a single channel A D converter. The device match three power supply
standard: 2,4V; 3,3V; 5V.
Thanks to the ultracompact dimension it is possible the use of this
accelerometer in various consumer market application like: gaming, virtual
reality, free-fall detection for data protection, picture browsing, mobile
terminals, appliance and robotics

Acknowledgments
The authors would like to thanks F. Guglielmo and A. Castelfranco for the
careful design of the device layout, and G . Tripoli for the contribution in order
to realize the ASIC design.

References
[l] LIS302ALB Data Sheet.
[2] M.Lemkin, Micro Accelerometer Design with Digital Feedback Control, doctoral dissertation,
Univ. of Califomia at Berkeley, 1997.
[3] C. Contiero, B. Murari, B. Vigna, Progress in Power IC and MEMS, analog Technologies to
interface the real world, 2004 proceedings of IEEE.
[4] N.Yazdi, H.Kulah, K.Najafi, Precision Readout Circuits for Capacitive Microaccelerometers,
Sensor, 2004 Froceedings of IEEE.
[5] M.Lemkin, B.E.Boser, A Three-Axis Micromachined Accelerometer with a CMOS Position-
Sense Interfaceand Digital Offset-Trim Electronics, IEEE Journal of solid state circuit, vol. 34,
no. 4,April 1999.
 Vehicle Health and Comfort Level Measurement:
A Preliminary Study

Fabio Lo Castro, Arnaldo DAmico

Electronic Department, University of Rome Tor Vergata,
Via del Politecnico 1, 00133 -Rome, Italy
Losastro 0 ing .unirorna2.it

Giovanni Brambilla
CNR, Istituto di Acustica OM Corbino I.

Via del Fosso del Cavaliere100, 00133 -Rome, Italy

Summary
In this paper is presented a system able to measure the vibrations inside the car, formed by
a PC, an acquisition board and two MEMS tri-axial accelerometers as sensors.
We tested the device on a vehicle measuring the acceleration, driving along different types
of street as highway and street,.
Subsequently the accelerations have been weighted with curves given by the normative in
order to evaluate the comfort and the risk level.

1 Introduction
The fatigue is a problem in all 24-hour operations, especially during driving.
Our biological clocks switch the brain automatically to low levels of alertness
after an heavy workload or during the night to induce sleepiness.
Vehicle accidents due to driver error and fatigue are a significant cause of
death, accident and injury. The studies on the safety drive, specially by the
automotive industry, try to reduce these accidents acting on different ways that
can be divided in three safety Ievels: prevention, warning and alert.

Fig.1 Frequency weightfilters: Wb Wd and Wj

302
 303

This paper has been focused on the prevention, studying the vehicle health
and comfort levels, related to vibrations inside the car. Analyzing the frequency
spectrum of the vibration it is possible to find the cause of discomfort.
In the literature it is explained how the human sensitivity to vibration is
highly frequency-dependant, as shown in figure 2. The effect of frequency is
reflected in the principal frequency weightings labelled Wk, Wd and Wf, that are
used to evaluate the comfort and the sickness levels.
Table 1 resume the corresponding acceleration weighting values to the risk
levels, normalized to eight hours, and the comfort levels.
Eyeball
lnwocdar Head
(wid mode)
(20-30Hz)

Threshold Level rms A(8) = 0.25 d s 2

Chwf wall Action Level rms A(8) = 0 , 5 d s 2
Limit value rms A(8) = 0 , 7 d s 2

Am,
(5-IUHz)

-eHg& comfort levels

A a, < 0.31.5 d s 2 I not uncomfortable

I 1.25 < a, < 2.5 d s 2 I

I
very uncomfortable I
a, > 2 . 5 d s2 extremely uncomfortable
I

Fig.2 Simpl@ed human body sub-systems. Table 1 Normalized acceleration values to

Each body sub-system h a s a resonance eight hours versus risk levels and r.m.s.
frequency band. acceleration values corresponding to the
comfort Levels

2 Experimental

We have developed a hardware prototype (see figure 3), driven by a Personal

Computer, that is able to evaluate the Health Level and the Comfort Level of the
vehicle expressed as weighted RMS acceleration value, according to the
European directive I S 0 263 1- 1.
304

ACC.l XYZ -SEAT

1
. NOTEBOOK

ACC.2 XYZ . S T E E ~ ~ ~ ~ E E L

Fig.3 Acquisition system installed on the car

The IS0 2631-1 directive gives only tables with values for the Wk, Wd, Wf so
we have used an approximation function of the third order quasi-least-square
which has an r.m.s. errors at the one-third octave frequencies less than 0.08 [I],
as shown in figure 1.
The following vertical acceleration filter Wk and horizontal acceleration filter
Wd have been used to evaluate the comfort level:

.03+
w.,= s38 0+78.92s2
~ 989.0s+
~ 0.02108
+ 2412s+5614

4.55+ 6.026s + 7.725

w -- s31+15.02s2
~~
+51.63s+47.61

In order to evaluate the motion sickness the following filter has been used:

0.05726~3+3.876s
w'-- s3+ 4.263~2
+ 4.777s + 4.396
The frequency ranges considered are:

- 0.5 Hz to 80 Hz for health, comfort and perception;

- 0.1 Hz to 0.5 Hz for motion sickness;
 305

Different paths as highway, road and street, have been chosen to study the
vehicle vibrations due both to the ground and to the vehicle speed.
Two Tri-axial accelerometer (LIS3L02AS4) with a range of & 2g as maximum
acceleration have been used to test the vibration levels inside a car. The first
accelerometer was placed on the dashboard and the second one was placed on
the driver seat.

Fig.4 Acquisition system used to evaluate the Comfort and Health level of the vehicles.

Table 2 shows the frequency responses of the vehicle used, captured along a
highway and a street. The Sensor was a tri-axial capacitive accelerometer model
STM LIS3L02AS4 mounted in a deforrnable pad that follows the seat contour
place on the seat surface. The data recorded by the PC have been filtered to
evaluate the comfort and sick level, as shown in figure 4. Both levels show that
the vibration on the highway are higher than street vibration due to the speed and
the state of the shock absorbers.

I Seat comfort I I Sickness I

Level Measurement [m/s2] Level Measurement [m/s2]
KMSa,, I airect hrphway KMS a,,> I street I
highway
a, 0.55 I 0.72 a,
I

I 0.31 I
I

0.41
I A@)
I
I 0.16
I
1 0.21
Table 2 Comparison of the differentsensors versus time need to damp vibrations

Figure 5 shows the frequency spectrum versus time during a test along the
highway where two peaks are always visible, one centred between 1Hz and 2 Hz
and the other at around 10 Hz. The temporal mean of the spectrum is shown in
figure 6.
306

Fig.5 Spectrogram of a test performed along a Fig.6 Mean frequency spectrum of a test
highway. performed along a street.

Two single tri-axial accelerometer, placed on the seat surface and on the
dashboard, have been used to measure the vibrations inside the vehicle. The
data, recorded by the PC, have been filtered by a third order approximation
equation of the IS0 2631-1 acceleration frequency weight curves to estimate the
Comfort and Sickness Levels of a car.

efere~ces

1. L. Zuo, S.A. Nayfeh ,Journal of Sound and Vibration 265 (2003), 459-465.
2. Magnus Wollstrom, TRITA - FKT Report 1998:44, ISSN 1103 - 470X, IS
KT~KT~--98/44--SE
3. Borje Rehn, Umed University Medical Dissertations - New Series No. 852 -
ISSN 0346-6612 - ISBN 91-7305-517-4
4. F. Benvenuti, S. Giambattistelli, 0. Nicolini, I. Pinto, E. Cini, etc , ISPESL-
"Linee guida per la valutazione del rischio da vibrazioni negli ambienti di
lavoro".
 Forced Damping Vibration of a Cantilever Beam
Fabio Lo Castro
Electronic Department, University of Rome Tor Vergata
Via del Politecnico I , 00133 -Rome, Italy
Lo.castroOing.uniroma2.it

Paolo Bisegna
Department of Civil Engineering, University of Rome Tor Vergata
Via del Politecnico I , 00133 -Rome, Italy

Summary
Forced damping vibration strategies have a paramount relevance in all those cases where
surface vibrations represent disturbances.
Knowing that surfaces show different resonant modes of vibration related to their geometry
we have selected a cantilever beam that has only one principal resonant frequency. Therefore
we have developed a circuit that, using the opposite phase signals technique, reduces the
oscillation time of the cantilever beam.
Different sensors as accelerometer, piezoelectric, infrared proximity detector have been used
to read out the vibration of the cantilever.

1 Introduction
The system under study has been a steel cantilever beam (250mm length,
40mm width and 1.5mm thick) with two identical piezoelectric wafers, model
number QPl5W made by ACX, applied on both sides close to the constraint.
One piezoelectric wafer has been used to detect the vibration and the second
one to generate the vibrations or to damp the cantilever oscillation, respectively
when driven by a sine wave or when the control system is enabled.
A Labview program controlled the beginning and the end of the solicitation
and acquired the signal read out by the sensor. To compare the efficiency of the
piezoelectric wafer we have been used other types of sensors, as infrared, model
OPB704W made by OPTEK, and quartdshear accelerometer, model K-SHEAR
number 8702B100 (8614) made by Kistler .

2 Experimental
The block diagram in figure 1 shows the architecture of the experiment. The PC
generates a sine wave that drives the actuator, the sensor detect the vibrations,
that are shifted and amplified by the control board, and applied to the actuator in
order to damp the oscillation.

307
308

Fig.1 System block diagram

The pictures in figure 2 and 3 show the cantilever beam with the piezoelectric
wafers near the constraint while the accelerometer or the IR sensor is placed on
the end of the free edge of the cantilever beam.

Fig.2 The cantilever under test Fig.3 The IR sensor placed on the end of
thefree edge

The principle used to reduce the vibrations (see figure 4) is that the sum of
opposite phase sine waves, with the same amplitude, is a null signal.

Fig.4 The closed loop system

 309

It is necessary to shift the phase of the signal read out by sensor because it
changes according to the position of the sensor, respect to the origin of the
oscillation, and the gain of the amplifier.
In this way a PID control has been designed with the possibility to select the
single parameters of the controller.

3 Results and discussion

In this experiment has been measured the time need to damp the vibration until
the 95% of the initial amplitude, value established by our limit of detection
under which noise and signal was unrecognised.
Four sensodactuator couples have been tested, called NULL, ACC, IR and
PIEZO:

NULL: Free damping oscillation; the cantilever was first excited by

first piezoelectric wafer (actuator) and then left free to oscillate. No
control loop was applied. The sensor was the second piezoelectric
wafer.
ACC: damping circuit realized with the couple
accelerometer/piezoelectric wafer with the piezoelectric used as
actuator
IR: damping circuit realized with the couple IR sensor/piezoelectric
wafer with the piezoelectric used as actuator.
PIEZO: damping circuit realized with two piezoelectric wafers, the
first used as sensor and the second as actuator.

In the last experiment the actuator was first driven by the Pc to excite the
cantilever and then it was switched to the control board to damp the oscillation.
The results of the tests are shown on table 1 while figure 5 and 6 shows
respectively the transient of a free oscillation and of a damping oscillation of the
PIEZO couple.

Sensor Type Damping factors

50% 90% 95 %
[SI [SI [SI
NULL 2.7 10 14
ACC 0.06 0.20 0.26
IR 0.04 0.16 0.23
PIEZO 0.08 0.30 0.40
Table I Comparison of the different sensors versus time need to damp vibrations
310

During the experiment the inconvenience of the accelerometer used was the low
sensitivity, in fact it was been placed near the end of the free edge of the
cantilever where there was the maximum displacement of the cantilever beam.
The sensitivity of the infrared system was better than the accelerometer one,
even though changing its distance from the constraint,.
The only requirement of the IR sensor respect to the other sensors is that it has to
be placed on another system stand respect the cantilever.
This is an advantage for a very small cantilever, or where the mass of the
accelerometer is heavy.
In fact we have measured a change in the resonance frequencies of the cantilever
with and without the accelerometer from 21.1 Hz to 18.7 Hz for the first
resonance frequency and from 127.7Nz to 123.0 Hz for the secondary resonance
frequency.

"4

B 02

ji $1

1 .*2

04

t IS1 14

Fig.3 Free damping vibration Fig.4 Forced damping vibration

During the development of the control circuit of the IR Emitter and Receiver to
make the IR sensor, particular attention has been made to reduce the interference
with other light sources, so that the sensor could be able used in a open
environment.

4 Conclusions

In this paper a cantilever has been considered as a single dimension system

so that only one sensor, as an accelerometer or a piezoelectric wafer or an
infrared device, has been used to drive a single piezoelectric actuator to damp
the vibration.
A PID control unit has been developed to drive the actuator, changing the
parameters according to the sensors characteristics.
Furthermore it has been developed an IR sensor as vibration sensor to be
tested with the accelerometer and the Piezoelectric wafer.
All couples sensor/actuator have worked properly as shown in the table 1.
 311

References

1. E.F. Crawley and J. de Luis - AIAA Journal, vol. 25, 1987, 1373--1385.
2. G. Caruso - Smart Materials and Structures, vol. 10(5), 2001, 1059--1068
3. F. Lo Castro, G. Caruso, P. Bisegna - AISEM 2004,257-261
4. Vibration Control of Plates using Periodically Distributed Shunted
Piezoelectric Patches.: Mohammad Tawfik AboElSooud, Ph.D. Proposal
5. Structural vibration suppression by concurrent piezoelectric sensor and
actuator.: S M Yang and C A Jeng - 1996 Smart Mater. Struct. 5 806-813
 IMPROVED CAPILLARIES CONNECTIONTO A CONTINUOUS
F'LOWGLASSMlcRoMAcHINEDMIcRo-cELL

w. VASTARELLA", M.ILIE",~,L. NARDI", A. MASCI", R. PILLOTON"

"ENEA, Via Anguillarese,301,00060 Santa Maria di Galeria, Rome, Italy
blJniv. Politehnica Bucuresti, LAPI, P.O. Box 15-135, Bucharest, Romania
e-mail corresponding author fifihaela.Ilie(C!cu,sacciu.etieu. it

E. CIANCI, S.QUARESIMA, A. COPPA, V. FOGLIETTI

CNWIFN, Via Cineto Romano 42, Rome, Italy

SUMMARY
Long term-functional tests of previously developed fluidic connections to a continuous flow
micro-cell revealed troubles such as bubble formation and analyte loss. In order to overcome
these drawbacks alternative technical solutions are investigated concerning both the materials
and the flow-pattern transferring as well as the assembly technique. Thus, the sand blasting of
borofloat glass -as well as the lithography of photosensitive glass (F0TURAN)- are used for
generating the openings, araldite is used as a glue and a sandwich technique is approached in
order to obtain the reservoir with openings for the capillaries. Functional tests are in progress
with in-house built borosilicate glass adaptors.

1 Introduction
An important step towards total analysis systems miniaturization is represented by
the appropriate fluidic connections. A continuous flow micro-cell has been
previously obtained by the authors and tested' in an analytical system. The micro-
cell consisted of a planar array of Au working electrodes, covered by a glass
reservoir (600 nl capacity), connected with perpendicular capillaries and provided
with a reference electrode (Ag/AgCl wire). The micro-reservoir has been
micromachined in borofloat@ glass by means of optical lithography combined
with wet etching'. The fluidic connections aim to insure the turbulence free fluid
flow and the reliability while assembling /disassembling the capillaries. The first
technical solution adopted' for these fluidic connections was based on the use of
elastic silicone materials, for both adaptors and gluing; the seal was made liquid-
proof by using additional conical adaptors. Long term tests at 10 @min showed
the accumulation of bubbles in the reservoir as well as a low reliability of the
connection and its dependency on the position of the pump (either at the entrance

312
 313

or at the exit of the reservoir). Each constructive part was analysed and
morphologically characterized from the point of view of surface smoothness and
contact surface compatibility.
A second technical solution is presented in this work, in order to minimize the
drawbacks of the first one. Other methods and materials are investigated to obtain
the openings: a) sand blasting of borofloatm glass and b) optical lithography &
wet etching of photosensitive glass (Foturanm), respectively. The openings and
the reservoir are obtained on different slides of glass, having a thickness of 0.4
m; the necessary parts are assembled only after the capillaries are fixed on the
cover provided with in- house-made glass adaptors. Polyimide and respectively
araldite have been used as adhesive.

e analyiof the ~revioulyobtained device

The parts of the connection have been investigated: surface and walls of the
adaptors, openings, gluing material, reliability of the adhesion of silicone adaptors
on glass. As it was previously shown2,the wet etching of glass has been optimized
to obtain a very low roughness and an isotropic process. In these conditions the
diameter of the opening has a minimum value in the middle of its transversal
section thus deviating the flow lines. The coupling of the capillaries has been
performed' using a multi-part silicone adaptor made of components with different
shapes (Figla). The cutting of these adaptors generated radial (Fig.lb) or coiled
(Fig. lc) micro-channels that enabled the glue (silicone) to advance towards the
functional opening down to the analyte flow zone, and to act as an additional
irregularity resulting in a further deviation of the flow lines.

Fig.1 a) multi-part adaptor (mag.X2); 6) radial and c) coiled channels on the adaptors surfaces (X20)

We suppose the lowering of the connections reliability is due to the mechanical

aging of the gluing material. Thus, two main causes have been identified for
bubbles formation and connection reliability: on one side the surface irregularities
of both adaptors and openings and on the other side the glue aging.
314

roving t e ~ h n i ~ elution
al
A simpler rhombus shape of the reservoir showed better results than the previous
one. The profile of the transversal section of the openings has been drastically
improved by means of the anisotropy of the new involved technological processes:
wet etching of photosensitive glass Foturan@03and, respectively, sand blasting of
borofloat@glass. Besides, a sandwich solution has been approached for obtaining
the reservoir (Fig. 2.a): the electrodes array (1) is covered with the pool (2) and
the openings (3) that are obtained on different glass slides (each of 8x8x0.4 mz).
The double side isotropic etching of borosilicate glass in HF 10% vol. is replaced
by the double side etching of FoturanB glass @reviouslyexposed with 100 Jlcm
UV radiation at a wavelength of 3 10 nm and cured one hour at 500C and then at
600C) ( detail of the opening in Fig.2.b). This anisotropic process provides an
aspect ratio of 1:30 on each of the two slides. Thus, as can be seen in the exploded
image (Fig.2.c), the reservoir consists of two pieces (2,3) glued together
meanwhile the multipart assembly has been replaced by only one piece of
borosilicate glass capillary (4), micromachined using small scale glass blowing.

2 a
Y

h d

Fig.2 The s a n h i c h solution a) transversal section ofthe microcell b) detail of the opening;
(x.700); c) exploded image of the reservoir; d)sandblasted opening [XIO)

Differently from other proposed solutions4 the additional structure does not
belong to the reservoir body. The main problem was the global and local
roughness of the photosensitive glass due to thermal induced stress in the
 315

unexposed areas. An improvement has been obtained when dehydrating5the glass

at 250C for one hour immediately after the etching. Higher temperature
treatment6 at 570C eliminated the micro-strains induced in the bulk of the
material but induced macro strains at global level.
The openings in borofloat@glass were also performed by one side sand blasting 2
min at 6 bar with FEPA P120 A1203 particles. The masking layer was the
RapidMask HDTMdry processing , self adhesive photo resist film with a dry
thickness of 50pm optically patterned, exposed in a mask aligner with 2J/cm2at a
wavelength of 436 nm. Although the high anisotropy of the process one can notice
the roughness of the surface around the border (Fig.2.d). The polyimide coated
capillary has been inserted in the adaptor and the parts have been glued over the

a b

Fig.3. Capillaries connections: a) polyimide coated capillaries with external diamefers of 400pm
respectively 700p are glued inside the adaptor using polyimide; b)fhe tejlon tubes inserted over the
adaptors

opening by means of fast-cured polyimide that rends them fix and sealed (fig.3.a).
The fixed capillaries have a length of several centimeters in order to avoid bending
stress of the glued zone and in the same time to allow a safe connection to the
external pipes by means of additional house-made conical couplers. There is
practically no danger of blocking either the capillary or the opening with the
adhesive material because the capillary is cut after the gluing. Several tens of
capillaries with inner diameters in the range 0.4mm-0.7mm have been assembled
and tested at low bending stress. Half of them were stripped oE. Higher stress-
resistance has been obtained when araldite has been used to glue the adaptors on
the covers and teflon tubes were used instead of capillaries. The tubes were not
glued at all but simply heated and inserted over the adaptors (fig. 3.b) .
316

4. Conclusions
Previously obtained fluidic connections have been tested by means of long term
continuous flow functioning during which bubbles formation and analyte loss
have been noticed due to irregular profile of the openings obtained by double-side
isotropic etching of borosilicate glass (aspect ratio 1:l). The profile has been
smoothened and the aspect ratio became 1:30 when using double-side etching of
photosensitive glass Foturano. The pool and the openings have been obtained
separately and glued after etching. In order to increase the reliability of the
connection the couple of silicone adaptors has been replaced by one single glass
adaptor traditionally micromachined by glass blowing. Flexible teflon tubes were
mounted over the glass adaptors in order to enable the analyte flow and this
connection ended to be more reliable than the previous one obtained with fused
silica capillaries inserted and fixed with polyimide inside of the glass adaptors.
Araldite and polyimide have been experimented as gluing materials and the first
one showed a higher reliability. An increasing of the local and global roughness
has been noticed in both sand blasting and photosensitive glass processing. In the
first case the resist /process has to be improved, while in the second one the
dehydrating of glass4 at 250C has been performed with pretty good results.
Further efforts are dedicated to improve the surface quality of the micromachined
glass and the alignment of the cover with respect to the reservoir. The functional
tests are in progress.
Acknowledgements
The authors acknowledge Progetto 5% Nanotecnologie of MURST, Italian Found for
Research (FISR), Theme 4Optical and Electro-Optical Sensors, Italian Found for
Fundamental Research (FIRB) and Cosmic (ENEA target project on Biosensors and
Bioelectronics) for financially sustaining this work. Special thanks are addressed to Mrs.
Emilia Castagna (ENEA-InternationalRelation Office) for her efficient logistic support.
References
1. J. Maly, M. Ilie, V. Foglietti, E. Cianci, A. Minotti, L. Nardi, A. Masci, W.
Vastarella, R. Pilloton, Sensors and Actuators B 111-1 12 (2005) 3 17-322
2. M.Ilie, V. Foglietti, E. Cianci, A. Minotti, Advanced Topics in
Optoelectronics, Microelectronics and Nanotechnologies, Proceedings of
SPIE V01.5227 (2003), pp. 3 18-321
3. M. Ilie, V. Foglietti, E. Cianci, A. Minotti, E. Vasile, L. Nardi, J. Maly, R.
Pilloton, A. Masci, Sensors and Microsystems, Ferrara, Italy 2004, 72-77
4. B. L. Gray, D. Jaeggi, e.a., Sensors and actuators 77(1999) 57-65
5. K. Grob, Making and Manipulating Capillary Columns for Gas Chromato-
graphy, Huethig Heidelberg 1986, ISBN 3-7785-13 12-5, pp.105-120
6. Y.Cheng, K. Sugioka, et al., RIKEN REVIEW no.50, Jan. 2003
 SIMULTANEOUS DETECTION OF POLYCHLORINATED
BIPHENYLS (PCBS) BY MEANS OF AN
ELECTROCHEMICALDISPOSABLE MULTIARRAY
SENSOR

Sonia Centi, Serena Laschi, Ilaria Palchetti, Marco Mascini

Dipartimento di Chimica, Universita degli Studi di Firenze, Via della
Lastruccia 3, 50019 Sesto Fiorentino, Firenze, Italia,
e-mail corresponding author:sonia.centiOimifi.it

Summary
Polychlorinated biphenyls (PCBs) constitute a well-known ecological problem. In addition to
the conventional techniques used for the analysis of PCBs, generally based on gas
chromatography coupled with mass spectrometry (GCMS), alternative methods based on
immunoassays provide inexpensive and rapid screening techniques for both laboratory and
field analysis. In this work disposable antibody-coated magnetic beads coupled to disposable
screen-printed electrochemical arrays were used to develop an immunosensor for detection of
some coplanar PCB congeners. Compared to single electrodes, arrays have the advantage of
testing many samples simultaneously. The advantage of using screen-printed arrays (based on
four-working electrodes and a reference electrode) is that they can be mass-produced at a low
cost and thus be disposable. The proposed immunosensor is based on a direct competitive
immunoassay scheme involving magnetic beads as solid phase and screen-printed arrays as
transducers for the evaluation of the extent of the immunochemical reaction. Alkaline
Phosphatase (AP) 6 used as enzyme label and the detection of the reaction product is
performed by using Differential Pulse Voltammetry (DPV). This immunoassay format could
be easily applied to the detection of PCBs in many matrices such as food or sediment/soil.

1 Introduction
Polychlorinated biphenyls (PCBs) are among the 16 chemicals designated as
perskitent organic pollutants (POPS). They are highly stable compounds that
persist in the environment, accumulate in the fatty tissues of most living
organisms since they are lipophilic, and are toxic to humans and wildlife. Their
chemical stability makes PCBs one of the most widespread environmental
pollutants.
There are 209 PCB congeners, which are divided into three classes based upon
orientation of the chlorine molecules, i.e., coplanar, mono-ortho coplanar, and
non-coplanar. Some of the them (the so-called coplanar congeners), in addition to
be stereochemically similar to the planar 2,3,7,8-tetrachlorodibenzo-p-dioxin

317
318

(2,3,7,8-TCDD), show also a biochemical activity and toxicity similar to these

compounds, for this reason they are indicated as dioxin-like molecules.
A disposable electrochemical immunosensor based on magnetic beads and
carbon array is proposed as a screening device for the rapid detection of coplanar
PCBs. Antibody-coated magnetic beads used as solid phase for the
immunochemical test, coupled with carbon screen-printed arrays as
electrochemical transducers were reported as new interesting tool n affnity-
based biosensors [ 1,2].

2. Materials and Methods

Electrochemical measurements were performed using CHI 1030 multichannel
potentiostat coupled with CHI 5.5 software (CH Instrument inc.). Carbon screen-
printed electrochemical array were printed in house using a High Performance
Multi Purpose Precision Screen Printer DEK 245 (DEK, Weymouth, UK), using
different thermoplastic inks. The cell consisted of four graphite working
electrodes with a diameter of 2 mm, a graphite counter electrode and a silver
pseudo-reference electrode. A connector for the array was used coupled to a
magnetic bar. Differential Pulse Voltammehy (DPV) was used as electrochemical
technique with the following parameters: range potential 0/+600 mV, step
potential 7 mV, modulation amplitude 70 rnV, interval time 0.1 sec..
A competitive assay was developed, following the scheme reported in figure 1.
Magnetic beads protein G coated were modified by immobilisation of antibodies
against PCB169. The beads obtained were incubated with a optimised dilution of
the tracer (PCB169-AP) and PCBs present in the sample. After molecular
recognition, the extent of the affinity reaction was evaluated by addition of a
enzymatic substrate, which was transformed in a electroactive product and
oxidized at the electrode surface using a useful potential value.
< I!$ .SiICP( I1 0 hl.,gr,il,~.bi.,.cpirdiri lir<v<c<l

coatmg Incubation

Affinity reaction

Figure 1: Schematic representation of a competitive assay and of the electrochemical

measurement on carbon screen-printed array.
 319

3 Procedure
3.1 Beads preparation and antibody immobilisation
The magnetic beads coupled with protein G were washed with 0.1 M sodium
phosphate solution pH= 5 before use to remove the NaN3 preservative. A
suspension of 10 pL of the beads was introduced in a tube containing 500 pL of
antibody (IgG anti-PCB169) solution 100 pg/mL prepared in 0.1 M sodium
phosphate solution pH= 5. After 20 minutes incubation time, the tube was
positioned on a magnet holding block, the supernatant was removed and the
beads were washed twice with 500 pL of washing solution (0.1 M sodium
phosphate solution pH= 5). Antibody-coated beads could be prepared in
advance and kept at +4"C.

3.2 Affinity reaction and electrochemicalmeasurement

50 pL of suspension containing antibody-coated beads were mixed with 940 pL
of sample solution; 10 pL of the tracer solution (PCB169-AP) was added to this
mixture. After 20 minutes incubation time, the beads were magnetically separated
and the supernatant was removed. After two washing steps, the beads were re-
suspended in 100 pL of working assay buffer and 10 pL of the suspension were
transferred onto the surface of each working electrode of the array. To better
localise the beads onto the electrode, the magnet holding block was placed on the
bottom part of the electrode. Then 300 pL of the enzymatic substrate solution (a-
naphtyl phosphate 1 mg/mL in DEA buffer) were deposited on the screen-printed
array, closing the electrochemical cell. After 5 minutes, the enzymatic product was
finally determined by DPV.
The results were reported as &/B, percentage units, where B, is the signal in
presence of competitor and Bo is measured without competitor.

4. Samples analysis
1 g of non contaminated marine sediment samples (collected and characterised
by GC-ECD in Fisheries Research Service-Marine Laboratoxy, Aberdeen,
Scotland) were spiked before the extraction by addition of 1 mL of a standard
congener solution and kept in contact for 16 hours. Then a sonication assisted
extraction was carried out [3].

4.1 Extraction procedure

0.5 g of marine sediment sample were added to 10 mL of methanol and after a
short mixing time (5 minutes), the mixture was sonicated for 2 minutes using an
ultrasonic probe. After 2 minutes to allow for the sedimentation of the solid, the
sample was filtered using a nitrocellulose 0.45 pm filter and then was diluted 100
320

times by addition of a solution containing magnetic antibody-coated beads and

the tracer at the optimised dilution. Then, the competition reaction was performed
as described in section 3.2.
5. Results and Discussion
5.1 Immunosensor application to PCB standard solutions analysis
The immunosensor developed using the antibody against PCB 169 (IgG anti-
PCB169) and the tracer PCB169-AF' was applied to the detection of the coplanar
congeners PCB169 and PCB126. Chlibration curves are shown in Figure 2: they
exhibited sigmoidal shapes typical of a competitive immunoassay. The different
sensitivities measured were due to the different affinities of the antibody against
the two coplanar congeners. This consideration was demonstrated by the ICso
values of each congener (0.2 ng/mL for PCB169 and 2.6 ng/mL for PCB126) and by
the different values of detection limit (6.8.10-* n g / d for PCB169 and 7.6.10-l
ng/mL for PCB126). IC50 is defined as the target analyte concentration which
inhibits the assay by 50% and detection limit has been calculated by evaluation
of the mean of the blank solution response (in absence of competition) minus two
times the standard deviation.

125
100
17 V PCB77
I PCB169
e 75
#m 50

25

0 I I I I I
blank -1 0.1 1.1 2.1
log [PCB congener concentration(nghnL)]

Figure 2: Calibration curves for PCB169 and 126 performed by the electrochemical
immunosensor using IgG anti-PCBl69 as antibody and PCB169-AP as tracer. IgG anti-
PCB169 100 pg/mL.

5.2 Analysis of marine sediment samples

The results obtained from the analysis of marine sediment samples are reported
in Table 1. The BJB, values were calculated by considering the responses of the
spiked marine sediment samples and that of the corresponding un-spiked sample
(blank). The same PCB solutions were also prepared by spiking the buffer and
also in this case the responses were compared with the corresponding blank (un-
spiked buffer). As it can be seen in table 1, a similar behaviour was obtained for
both extracted samples and samples prepared in hffer, all solutions gave a
decrease of signal with respect to the corresponding blank. For both cases the
 321

signal percentage obtained for R, S and T samples, which contained the same
congener (PCB169) but at different concentrations 9.1, 1 and 5 pgKg) was
different. The signal percentage measured for the sample T was lower than the
value obtained for the sample S and R, which contained the same congener
(pCB169) but a lower concentration (respectively, 0.1 and 1 pgKg versus 5
pg/Kg). Different signal percentages were measured for the S and U samples,
containing different congeners (PCB 169 and PCB 126) but at the same
concentration.
Table 1: Results obtained from the analysis of marine sediment samples, extracted by the
sonication assisted method. A comparison with the signal obtained in the buffer for the same
concentrations is reported.

Sample Spiked Spiked Buffer Marine

congener concentration B,/B, (%) sediment
@g/Kg) B,/B, (Yo)
PCB 169 0.1 +
75 2 74f2
S I PCB169 1 121+4 128f2
PCB 169 5 8k1 6+2
PCB126 5 44+4 4056

6 Conclusions
A sensitive and reproducible immunosensor for the detection of some coplanar
PCB was developed. The assay scheme was based on magnetic beads coupled to
disposable and low cost sensor. Different dioxin-like congeners could be detected
in short time at ppb level. The possibility to perform a screening measurement
that allows to know in short time if a sample is contaminated or no, is important
when a big number of samples have to be analysed; therefore these experiments
could be a valid analytical tool to carry out a fast screening of many samples,
analysing only positive samples with traditional methods in order to have a exact
quantification of PCB concentration.

References
[l] Palecek E., Fojta M., Jelen F., Bioelectrochemistry, 56,2002,85-90.
[2] Centi S., Laschi S., Franek M., Mascini M., Anal. Chim. A d a , 538, 2005, 205-
212.
[3] H. Okuno, B. Yim, Y . Mizukoshi, Y. Nagata and Y. Maeda, Ultrasonics
Sonochemistry 7 (2000), 261-264.
 Integrated Microgravimetric Set-up for Chemical Sensing
G. P i o a
Centro Interdipartimentale di Ricerca E. Piaggio , Universita di Pisa
F. Di Francesco
Istituto di Fisiologia Clinica, CNR, Pisa
G. Campobello, G. Cannati, N. Donato
Dip. di Fisica della Materia e Tecnologie Fisiche Avanzate, Univ. di Messina
A. Bonavita, G. Micali, G. Neri
Dip.di Chimica Industriale ed Ingegneria dei Materiali, Universita di Messina
Summary
It is a well known fact that stability and reproducibility of measurement are crucial topics
for chemical sensing. Most problems are intrinsically associated with the sensor chemico-
physical characteristics and with the fabrication techniques, but often the measurement set-
up must carefully be designed. In this paper we addressed the development of an integrated
microgravimetric set-up able to perform fast and accurate measurements with an array of
quartz crystal microbalances (QCMs). In order to guarantee measurement accuracy,
reliability and repeatability, the sampling system was designed to control and optimise
factors capable of influencing the generation of sensor transduction signals. The system
comprises a dedicated electronic interface for measuring the QCM resonance frequency by
using a novel technique. This technique allows reducing the measurement time while
maintaining a high frequency resolution. The system was simulated in the range of
frequencies covered by prototypal PVA-coated QCMs. The resonance frequency was
estimated for a six elements sensor array with a resolution lower than 0.2Hz when the input
signal presents a SNR near 40dB at a lsec scan rate.

1 Introduction
The piezoelectric effect in coated quartz crystals can be exploited as a sensing
mechanism for the direct monitoring of organic compounds in gaseous and liquid
phase. A quartz sensor, also known as quartz crystal microbalance (QCM),
consists of a quartz disc with electrodes patterned on the opposite sides, one of
which covered with a sensing layer capable of suitable interactions with the
analyte of interest. The application of an oscillating electric field across the
electrodes triggers mechanical vibrations, whose fiequency is related to the
acoustic load of the sensor. A one-dimensional transmission-line model can be
used to model the behaviour of such a sensor, by analyzing the effect on wave
propagation of the absorption and diffusion of analytes into the viscoelastic
coating. The use of QCM in liquids is considerably harder than in gaseous media.
This happens because the observed fiequency shift is caused not only by the mass
effect, but also by the viscous coupling with liquid at the quartz surface, that
depends on the surface roughness, the surface stress and possible changes in the

322
 323

liquid properties. Moreover, changes in the viscoelastic properties of the sensing

layer (swelling and/or softening, cross linking or phase transitions) may generate
additional contributions to frequency shift which cannot be easily distinguished
from those related to mass changes. The low sensitivity and the insufficient
specificity can be considered drawbacks of microgravimetric measurements, but
the ongoing research on new materials and dedicated processing architectures
continues to offer solutions to these problems, widening the application field of
QCM sensor arrays.
In this paper authors addressed the development of an integrated
microgravirnetric set-up able to convey in optimal conditions the analyte of
interest and perform fast and accurate measurements of the minimum detectable
mass change with QCM sensor array. In order to guarantee measurement
accuracy, reliability and repeatability, the sampling system was realised to
control and optimise factors capable of influencing the generation of sensor
transduction signal, and a new method proposed by Campobello et al. [ 11 for fast
and accurate frequency shift estimation was adopted.

2 The microgravimetric set-up

The microgravimetric set-up was composed of a hydraulic and an electronic
section, the former consisting of a sampling system and an exposure chamber, the
latter of an oscillator and sub-section for resonance frequency estimation. The
sampling system was composed of a line of chromatographic air with a controlled
flow and an in-line humidity generator. Inert PTFE tubing and fittings were used
to send different concentrations of water vapour into the exposure chamber,
where QCMs were housed. A valve allowed to switch the system between state 1
(QCMs flushed with chromatographic air, baseline acquisition and cleaning) and
state 2 (exposure of QCMs to humid air). The measurement protocol consists of
three phases for each experiment: baseline acquisition; exposure; desorption and
cleaning.

2.1 The exposure chamber

To expose the array of QCMs to a chemical mixture in optimal conditions,
the chamber was designed to:
to simultaneously expose all the sensors under the same conditions
to accurately reproduce the shape of the input concentration signal
to obtain the same concentration profile in repeated measurements
to obtain short analyte concentration rise/fall times
to avoid memory effects and sample dilution
to use only a small amount of sample
A homogeneous flow with low speed gradient, no recirculating zones or
stagnant regions and the same local concentration of volatiles over each sensor
324

was obtained by choosing a radial geometry for the chamber, whose dedicated
deflector allowed homogeneous flow conditions with low velocity gradients
(figure 1).

Figure 1 - Exposure chamber for QCM sensor array; a) 3D drawing, b) a photo

2.2 The electronic section

In order to study the response of quartz crystal resonators, the complete
frequency response in a range centered on the resonance frequency can be
monitored. As an example, lOMHz AT-cut quartz were spin-coated by polyvinyl
alcohol (PVA), a highly hydrophilic polymer. The surface wettability and
easiness of functionalizationmake it of great interest for chemical and biological
sensing. The frequency response of such PVA-coated quartz crystals at different
humidity levels was investigated. If a purely gravimetric changes occurred, a
frequency shift with no change in peak admittance would result. This case is
limited to thin coatings rigidly coupled to the underlying crystal and not for
polymers. Prototypal PVA-coated crystals show a viscoelastic behaviour
resulting in changes in peak admittance, as a consequence of an energy loss
(damping) (figure 2a). In particular, humidity induces a swelling and softening in
the material. Figure 2b shows the resonance frequency shift at different humidity
levels.

m2Mm *4261((10 ss*moo iiPSIOB0 sPu1mo - -/

600
Frequency (Hi] ?4 RH
a) b)
Figure 2 - a) Changes in peak admittance and b) resonance frequency shift of prototypal PVA-
coated crystals at different humidity levels
A network analyzer was used to acquire the complete frequency response.
Such instruments are interesting for laboratory use and research but are expensive
 325

and voluminous. An oscillating circuit is more practical, since the crystal can be
incorporated into the circuit with a quasi-digital output signal and a frequency
counter can be employed to estimate the resonance frequency. In case of pure
mass accumulation, if the mass variation is uniformly distributed throughout the
crystal (uniform thickness and density of the film) and the shear velocity is
identical in the film and in the crystal, the Sauerbrey equation [2] is valid and
quantifies the frequency shift due to a variation of the deposited mass. Factors
such as mechanical clamping, damping in the electrical circuit, temperature
variations are not considered, hence calibration methods are needed. In case of
pure mass accumulation, the oscillator circuit must work at a constant phase of
the quartz resonator and sensor resolution and accuracy depend on the device and
method for frequency estimation. Let A be the area of the quartz crystal in cm2,
M the mass attached on the crystal surface and f4 the resonance frequency;
considering the density of quartz (2.648 g ~ m -and ~ ) the propagation rate of the
acoustic wave ( 3 . 3 3 6 ~ 1 0 5cm s-I), it can be obtained:
AM
A f = -2,26.
f,'-A
A frequency shift of 1 Hz corresponds to 40pg/mm2 of mass variation.
Commercial systems are designed to reliably measure mass changes up to 100 p
g, whereas the minimum declared detectable mass change is around 10pg/mm2,
whch corresponds to a frequency shift of 0.25Hz. Viscoelastic coatings on
hydrophilic or hydrophobic surfaces may be difficult to model, and the change in
their properties cannot be distinguished in a simple way from the mass effect
within the sensor response. Furthermore, until recently, it was generally
perceived that viscoelasticity would result in such a severe damping to prevent
oscillation of the crystal, until Kanazawa and Gordon [3] and Bruckenstein and
Shay [4] showed that the energy loss is not so severe as to preclude oscillation.
In general the sensor transfer function can be determined by a one-
dimensional transmission-line model and in particular by the acoustic load
concept (ALC) [S], even though on the basis of a number of calibration
measurements. In the ALC model the acoustic load impedance carries all the
changes in the sensitive coating, i.e. pure mass accumulation, mass accumulation
accompanied by material property changes, or material property changes induced
by chemical or physical effects. Furthermore, the ALC model directly relates the
acoustic load impedance to the resonance frequency shift, Af, and the motional
resistance change, AR or similar values representing acoustic energy dissipation.
In particular, Lucklum et al. [6] have found that the resonance frequency is to be
measured at the maximum real part of quartz admittance and the motional
resistance at the reciprocal of the maximum real part of quartz admittance. The
maximum of the real part of quartz admittance does not coincide with a constant
quartz phase. Bomgraber et al. [7] have solved this issue realising an automatic
gain-controlled oscillator, by modifying a well proven oscillator circuit [8]. It
326

follows that high efficient and accurate electronic oscillating circuits and methods
for resonance frequency shift estimation are mandatory.
Frequency estimation is a wide studied problem and several methods exist in
literature [9]. In particular, as regards the accuracy, a lower bound was proved in
the case of Additive White Gaussian Noise, knows as Cramer-Rao lower bound,
by means of it, the accuracy can be related on three key factors: sampling
frequency, Signal-to-Noise Ratio (SNR) and number of available samples [lo].
However, other parameters can be taken into account in QCM measuring, i.e. the
elaboration time and the bandwidth.
In this work a new method proposed by Campobello et al. [l] for fast and
accurate frequency shift estimation was adopted. The method can estimate the
frequency starting from samplings of a sinusoidal signal. An oscillator circuit
with a sinusoidal signal at the resonance frequency as the only output was
adopted. The main characteristic of such a method is its intrinsic low complexity
and use which enable an easy setting of the right balance between speed,
accuracy and bandwidth relating to the particular QCM sensor under test. The
output sinusoidal signal is processed by a dedicated electronic interface (figure 3)
composed of the following main subsections:
Reference signal (white quartz or signal generator)
Mixer (for down conversion)
Filter (for noise reduction and anti-aliasing)
ADC (acquisition board)
DSP (elaboration implemented by hw or sw algorithm)

The use of a white quartz as reference allows the frequency shift caused by
temperature variations to be compensated placing inside the measurement
chamber both crystals. The system was simulated in the range of frequencies
covered by six prototypal PVA-coated QCMs (40kHz) at a scan rate of lsec. In
figure 4 the maximum absolute error is reported. It can be noticed the frequency
is estimated with a resolution lower than 0.2Hz when the input signal presents a
SNR near 40dB. These results are very interesting for biomedical applications,
where the analyte concentrations are extremely small.
 327

I
3 35 LI 5 55 6 65 I
r, I,*!
Figure 4 - The maximum absolute error in frequency estimation
Conclusions
In this paper we proposed and discussed an instrument for fast and accurate
measurements with an array of quartz crystal microbalances (QCMs). The
sampling system, the exposure chamber and the dedicated electronic interface
were detailed. A novel method for frequency estimation allows the measurement
time to be reduced while a high frequency resolution is maintained. Taking into
account prototypal PVA-coated QCMs, the resonance frequency was estimated
with a resolution lower than O.2Hz with an input signal SNR near 40dB at a lsec
scan rate for a six elements sensor array. These results are promising for the
application of such a system in biomedical research where measurements at very
low concentrations are mandatory.
References
1. G. Campobello, N. Donato, A. La Corte, G. Pioggia, A New Method for Fast and Accurate
Frequency Estimation, IEEE Transactions on Signal Processing, submitted.
2. C K. OSullivan, G.G. Guilbault, Commercial quartz crystal microbalances - theory and
applications, Biosensors & Bioelectronics 14 (1999) 663470.
3. K.K. Kanasawa, J.G. Gordon, Frequency of a Quartz Crystal Microbalance in Contact with
Liquid, Anal. Chem. 57 (1985) 1770-1771.
4. S. Bruckenstein, M. Shay, Dual quartz microbalance oscillator circuit, Anal. Chem. 66 (1994)
1847-1855.
5. R. Lucklum, P. Hauptman, The generalized acoustic load concept for QCM-Mass sensitivity,
Viscoelasticity, and other phenomena, Proceedings of Eurosensors H I (2002) 7-8.
6. R. Lucklum, C. Behling, R.W. Cernosek, S.J. Martin, Determination of complex shear modulus
with thickness shear resonators, J. Phys D: Appl. Phys. 30 (1997) 346-356.
7. R Borngraber, J. Schroder, R. Lucklum, P. Hauptmann, Is an oscillator-based measurement
adequate in a liquid environment?, IEEE Transactions on Ultrasonics, Ferroelectrics and
Frequency Control, 49(9) (2002) 1254-1259.
8. F. Eichelbaum, R. Borngraber, J. Schroder, R. Lucklum, P. Hauptmann, Interface circuits for
quartz-crystal-microbalance sensors, Review of Scientific Instruments, 70(5) 1999.
9. S. Kandeepan, S. Reisenfeld, DSP Based Frequency Estimation Techniques and Their Relative
Performance, CSNDSP (2002).
10. D.C. Rife, R.R. Boorstyn, Single-Tone Parameter Estimation from Discrete-Time
Observations, IEEE Transactions on Information Theory, IT-20 (5) (1974).
 In-Situ monitoring of Volcanic Gases at Solfatara crater with
hybrid e-nose

S. De Vito, E. Massera, L. Quercia, G. Di Francia

ENEA - C.R. Portici, Via VecchioMacello s.n.c.,80055, Portici A), Italy,
[email protected]

summary
The analysis of fumarolic gases, released in high activity volcanic areas, can provide
significant information on the status of these areas. Variation in composition of this gases
can, infact, be correlated with volcanoes dynamics. This variations are typically due to
chemical interactions between magma and the different layers of rocks and aquifers with
which it came in touch during its movements toward surface. In this work, we propose
the use of an hybrid e-nose, hosting both commercial and ENEA developed gas sensors,
for the analysis of fumarolic gases within the solfatara crater. This approach has shown
to be capable to provide both qualitative and quantitative information that by the use of
pattern recognition techniques and neural regression schemes could be used to extract
significant indexes on undergoing volcano dynamics.

1. Introduction
Understanding volcanoes dynamic has always fascinating researchers in the
effort to model and forecast dangers of high activities periods. These events pose
relevant threats to human beings, since they produce severe impact on human
settlements. Given the impossibility to control volcanoes dynamics, prediction
capabilities are the only way to effectively contain damages. As such, these
capabilities are highly needed especially when high population density
settlements are located in the nearside of an high activity volcanic area. Here,
surveillance networks typically rely, at present, on the predominant use of
geophysical methods of analysis. During the last decade, geochemical methods
have been studied and the analysis of fumarolic gases, released in high activity
volcanic areas, revealed capable to provide significant information on the status
of these areas. Variation in composition of these gases, in fact, seems to be
correlated to volcanoes dynamics. These variations are typically due to changes
in chemical interactions between magma and the different layers of rocks and
aquifers it can find in its movement toward the surface and the ratio between
particular species (e.g H20 to C0 2 ratio) is considered an important fingerprint

328
 329

of the volcanic activity [ 11. Increase in H 2 0 concentration is in fact considered to

be related to the heat flow caused by magma movements, whether the C 0 2
concentration, although relatively high in volcanic gases, can have limited
relation with magma movements being less reactive than other magma
components [ 2 ] .Furthermore, some fumarolic gases can have a relevant impact
on local climate and also on life conditions of human. For this reasons, volcanic
gases are periodically monitored, usually with analysis carried out ex-situ with
laboratory scale equipment using techniques such as mass spectometry and gas
chromatography. Unfortunately, for such methodologies the characteristic time,
that recently showed up as an important feature for fumarolic gas analysis [3], is
very coarse. Furthermore, direct spring sampling in this harsh environment can
be very dangerous for humans and measurements systems or, in some cases, even
impracticable. In the last few years, the need for on-site and continuous
monitoring with unattended operation has been constantly growing and some
application of remote sensing techniques have been developed for use in this
scenario [4]. Optical methods, for example, seems to offer a practicable way, but
they are affected by some disadvantages such as beam scattering due to high
water vapour content and beam position change due to temperature fluctuation of
fumarolic gases.
In this paper we propose the use of a portable e-nose for the analysis of
fumarolic gases at Solfatara crater. Solfatara crater is located within Campi
Flegrei area, probably the most complex volcanic structure on the earth surface.
We believe that e-Noses could represent a valid answer to these needs thanks to
the abilities to discriminate between different mixture patterns [5] or to actively
determine single analyte concentrations in complex mixtures such as volcanic
gases [61[71. When equipped with the right sensor array and coupled with an
opportune data analysis subsystem e-noses can, in fact, easily meet any
significative time resolution constraint in this applicative scenario.

2. The proposed system

During last years, we designed and developed a portable e-nose prototype,
named NASECANETM,based on a high modular three stage architecture. The
first stage host a ten slots sensor array. The developed prototype can host an
array of sensors with different transducing mechanism both commercial or
developed on purpose at the ENEA laboratories. Sensors can be feeded with a
pump based subsystem or let operate in open air. The second stage host a
precision electrometer while the third stage includes data processing subsystem,
communication subsystem and a battery providing power to the entire system.
The MCU based data processing subsystem has been chosen to be capable, in the
near future, of running dedicated sensor fusion algorithms in order to make the e-
nose capable to work out classificatiodregression schemes as a standalone unit.
330

Furthermore the modular design allows for replacement of the c o ~ u n i c a t i o n

subsystem at the moment based on RS-232, with a wireless module either
Bluetooth or WiFi.

Figure 1: NASECANE prototype e-nose (a). Sampling locations within Solfatara crater :
0,
Sauna FangaiaA and the control site@. ( b)

~ntal
Located inside the Campi Flegrei voIcanic district, the Solfatara crater i s the
site of the most conspicuous superficial thermal anomaly in the area; volcanic
gases are emitted within the entire crater, by means of several fumaroles
surrounding the main one, called Bocca Grande. If sufficiently accurate the
analysis of such gases may provide information on the temperature and pressure
conditions in the hydrothermal reservoir feeding the fumaroles and i t s
relationship with the underlying magma body [2]. In table 1, we report type and
characteristic of the selected sensors.
Table I . The sensors array selected for the~marolicgases analysis.
Range Resolution Manufact.
CL2 0,2-20 ppm 0,02 COTS
co 1- 1000 ppm 1 COTS
H2S 0,3-100 0,05 COTS
PPm
NO 3-250 ppm 0,2 COTS
NO2 0,2-20 ppm 0,02 COTS
SO2 5-2000 ppm 2 COTS
ENEAS ENEA~
ENEA2 ENEA3
RH 5-95% RH 3% COTS
TEMP (-SO)-5O0C 02 COTS
Elect. Cell.
Porous Silicon Membrane
Inverse Diode Porous Silicon Heterojunction
 33 1

The selection of the sensor array has been driven by the known composition
of fumarolic gases and by the need to combine both commercial and ENEA
developed sensors so to obtain a deeper insight on their dynamic. We conducted
our investigations using the e-nose, analyzing the gas emitted at several locations
within the Solfatara crater (see Fig. I), positioning the prototype at a maximum
distance of 5 meters from the fumaroles (Bocca Grande, Sauna) or inside the
areas interested by diffuse degassing (Fangaia). Previously a short campaign
had been conducted in order to assess the influence of weather conditions,
measurements sites selection etc.

4. Results
Data analysis has been performed on sampled data by means of statistical
(PCA) and pattern recognition techniques. The e-nose detected significative
differences among fingerprints of the different sources within the Solfatara crater
and between in-Solfatara samples and the control site. A clear separation of site-
specific clusters is achievable by means of a PCA plot using two or three
principal components. Responses analysis has shown the presence of near-
constant characteristic ratios between many couples of specific sensor responses,
regardless of the variable measurement conditions (wind intensity and direction).
Some of these ratios are site specific while some others are specific of the entire
Solfatara crater with respect to the control site.

Figure 2: 3D-PCA scatter plot of normalized data (a); NO versus H2S sensors responses plot (6)
(V). BG1= Bocca Grande I , BG2= Bocca Grande 2, SAN = Sauna, FNG= Fangaia, CST =
Control Site

As an example, in Fig. 2 (b) we show the presence of a site specific x-axis

intersection of regression line and very similar angular coefficient over the entire
crater when looking at H2S and NO COTS sensors responses. Analyzing
different couples of sensor responses, we also found specific angular coefficient
for each of the 4 Solfatara measurement sites.
Furthermore, we performed regression analysis by the use of neural networks
(3 layer Back Propagation Networks) in order to highlight the capability of the
332

two ENEA sensors to predict the response of specific COTS sensors and
ultimately estimate specific gas concentration by means of neural sensor fusion4.
The ENEA developed sensors revealed to be extremely good at estimate the
commercial NO sensor response reaching values of less than 1% when evaluating
both MAE (Mean Absolute Error) and STD (Error Standard Deviation) with
respect to the range of sensor responses. We also evaluated the same
performance indexes for different BPNs trained to estimate any other commercial
sensor response when feeded with couple of other sensor instantaneous response.
A subset of obtained results is presented in table 2. All BPNs have been trained
using 75% of samples set and tested on the remaining 25% adopting a 4-fold
cross-validation scheme.

Estim. Sl S2 MAE STD

NO2 ENEAl ENEA2 0.045 0.07
H2S co so2 0.095 0.08
H2S ENEAl ENEA2 0.045 0.055
co ENEA 1 ENEA2
CL2 ENEAl ENEA2
0.09
0.045
0.10
0.055 I
s o 2 ENEA 1 ENEA2 0.05 0.04
RH ENEAl ENEA2 0.06 0.05
NO ENEAl ENEAZ <o.01 <o.01
NO co CL2 0.07 0.095
NO H2S RH 0.055 0.055
NO H2S ENEAl 0.06 0.05
-
NO H2S--- co 0.06 0.095 ~1 23s ae a* \2 0 n: ,. S8 .

Table 2: Performance of BPN based regression Figure 2: Normalized regression plot for the
estimating COTS sensor response using a subset estimation of NO (a) and CL2 (b) using
(actually two) of the remaining sensor responses. ENEAl, ENEA2 sensor responses. On the X
axis the commercial sensor outputs is
reported, Y axis shows the BPN response
estimation.

As it can be observed,, performance indexes reach values that can be

considered sufficient for the implementation for the e-nose of an instrument fault
accommodation scheme (IFA). IFAs can provide graceful performance
degradation instead of a catastrophic one, when one of the sensor of a sensing
apparatus became faulty, see for example [8]. This scheme can be implemented
by means of multiple sensor fusion subsystems trained to operate on a subset of
working sensor responses for the estimation of the faulty sensor response. In this

Network has been feeded with Enea developed sensors instantaneous response. 15 sigmoidal
tangent units have been used in the hidden layer, a single linear unit represented the output layer.
 333

case, we show that a set of BPN properly trained can estimate a faulty sensor
response by knowing the instantaneous response of only two of the working
sensors set.

5. Conclusions and Further works

We have tested a prototype e-nose in a complex operative scenario such as
the Solfatara crater. The response quality allows for the extraction of interesting
features with respect to the geochemical point of view. Further works are
required in order to exploit e-nose fingerprint relationships with other significant
geochemical and geophisycal parameters, and so effectively extend classic
methods of geochemical analysis.
We have also obtained significative information about ENEA developed
sensors array subset. ENEA sensors, when coupled with a sensor fusion
subsystem, have proven to be very effective in estimating commercial NO sensor
response. The intrinsic nature of Solfatara fumarolic gases emission,
characterized by known species correlations, and the properties of selected
hybrid sensor array made possibile to reach a very significant robustness to
sensor faults, the response of a faulty sensor being guessed with sufficient
accuracy by the use of a neural based IFA subsystem.

Acknowledgments
The authors wish to thanks Vulcano Solfatara s.r.1. for the precious
collaboration and kindness.

References
1. D. Tedesco, JC Sabroux, Bull. Volcanol. 49, 381-387. (1987).
2. G. Chiodini, M. Todesco et al. Geophys. Res. Lett. 30, 1434 (2003).
3. M. Zimmer, J. Erzinger, J. Volcanol. Geotherm. Res.125, 25-38 (2003).
4. A. Rocco, et al., AppI. Phys. B 78, 235-240 (2004).
5. T. C. Pearce, et al., Eds., Wiley, (2002) (ISBN: 3-527-30358-8).
6. M. Pardo et al., Sensors and Actuators B 65 267-269, (2000).
7. S. De Vito et al., Proc. of Eurosensors XIX, (2005). EA available on CD.
8. D. Capriglione et al., Proc. of 13" ICIAP, LNCS 3617,1117-1125 (2005).
Monitoring of Airborne Fine Particles and NOx Beside a Major
Road

M. C. Carotta, E. Ferrari, S. Gherardi, C. Malagu, M. Piga, G. Martinelli and A.

Rosignoli
CNR-INFM-Physics Department, University of Ferrara, via Saragat I , Ferrara,
44100, ItaZy
[email protected] nfn. i t

Summary
In this paper a correlation between nitrogen oxides (NO,), and diesel particles was performed. New
monitoring units constituted by an array of chemoresistive gas sensors capable of detecting CO, NO,
e 0 3 were located at two different heights to study the gas vertical distributions. Diesel particles were
sampled directly from the exhaust pipe of diesel car to assess the structure, dimension and shape of
fresh particles. While a second sampling was carried out close to a busy road in the centre of the city
to assess the contribution at the airborne urban particulate of particles exhausted by the diesel
vehicles. The particles were sampled by a multistage cascade impactor and characterized by different
analytical methods (SEM, EDS).

1 Introduction
In the last 20 years, in Italy, fleet of cars has increased of 60%, and nowadays
the diesel cars represent the 58% of the vehicles sold in comparison with the
40% of gasoline powered vehicles. The reason of the diesel powered vehicles
booming market is that the diesel engine use up about less then 10-15% of
energy per kilometre with respect to gasoline and the fuel is cheaper than
gasoline. The pollution from vehicles arises from the by-products of combustion
process (emitted via exhaust system) and from evaporation of the fuel itself.
These emissions are composed from volatile organic compounds (VOCs) and
they are produced because the fuel is not completely burnt (oxidized) during the
combustion. NOx result from the oxidation of nitrogen at high temperature and
pressure in the combustion chamber. CO occurs when carbon in the fuel is
partially oxidized rather than fully oxidized to COz. PM is produced from the
incomplete combustion of fuels, additives in fuels and lubricants, w o n material
that accumulate in the engine lubricants and those coarse particles brakes, tires
and dust emission from road. [ 11
Despite the significant improvement in diesel engine design and efforts made to
reduce emissions in general, heavy duty diesel fuelled vehicles in average emit
the half of CO but 6 and 8 times of particle mass and NOx, respectively, more
than a gasoline fuelled vehicles. Therefore the diesel vehicles are one of the most
important producers of PM and NOx in urban environment.

334
 335

The diesel particle exhaust are constituted by carbonaceous particles (soots) and
inorganic particles (ashes), that are generated by mechanisms such as molecular
growth, condensation, coagulation, agglomeration and oxidation of molecular
precursors in the combustion chamber. Traditionally, the assessment of diesel
exhaust particles has been based on mass concentration, but this may not give an
accurate valuation of the diesel aerosol exposure. By number count, most diesel
aerosol is found in nuclei mode with a particle diameter less than 50 nm and a
number median diameter between 15 and 20 nm when measured under
atmospheric conditions. However, most of the aerosol mass lies in the
accumulation mode with particle size ranging from 50 nm to 1 pm and the
median aerodynamic diameter between 100 to 200 nm.[2]
NOx play an important role in photochemical reaction cycles, leading to smog in
the urban troposphere, depletion of stratospheric ozone and acid rain. In
photochemical smog formation certain gas phase reactions are promoted by
ultraviolet light.
Aim of this work is to correlate the NOx and diesel particle sources through a set
of local measurements performed with portable and versatile devices: small gas
monitoring units based on solid state gas sensors to measure NOx concentrations
and portable multistage cascade impactor to collect particulate matter.

2 Experimental
The experiment was divided in two parts: in the first one diesel and airborne
particulate matter were sampled and in the second one NOx concentrations were
measured. The results were compared to assess the correlation between the gas
and diesel particles, The diesel particles were sampled directly from the exhaust
pipe of a diesel car (Ope1 Astra, 2002, Euro 111) at a distance of 30 cm, the
engine was operated at idle condition. The airborne particle sampling point was
close to a busy interception and located at the heights of 3 and 8 meters above
the street on the windowsill and about 10 meters from the crossroads.
The particles were sampled with a Portable Cascade Impactor Sampler (PCIS),
that separates the particulate in five granulometric fractions and at every stage it
was placed a polycarbonate filer where particles were collected. The stage cut-
off diameters are 10-2.5, 2.5-1.0, 1.0-0.5, 0.5-0.25, <0.25 pm, respectively, for
the five stages of the impactor. The two final stages particles (<0.5 pm) are
interesting for this work because they are mostly generated by primary
combustion sources and are characteristic of relatively fresh emissions such as
those dominating in a typical urban area [3]. The particle composition and the
morphology were determined by scanning electron microscope (SEM) equipped
with energy electron spectrometry (EDS). The voltage for the energy dispersive
X-Ray analysis was 20 KeV and the time of X-Ray collection was 100 s. Each
filter was covered whit gold before SEM observation.
336

Near the particle collection points units for gas pollutants monitoring were
located. They consisted of an array of chemoresistive gas sensors capable of
detecting CO, NOx and ozone. The thick-film gas sensors were developed and
fabricated by the Sensor and Semiconductor Laboratory of the University of
Ferrara.

3 Particle Sampling and Analysis

The soot particles appear as agglomerates composed of small, basic particles that
are spherical or nearly spherical with varying diameter from 10 to 80 nm. The
primary particles are constituted of graphite. During the agglomeration of soot,
inorganic ultrafine particles are included into the growing clusters, while
condensable unburned hydrocarbons along with water-solved inorganic
precursors easily adsorb onto the extended surfaces of porous clusters [ lb].
The SEM observations revealed the heterogeneity, in dimensions and shape, of
airborne particles. However, the particles collected in the last two filters of the
impactor seem to belong to two species: nanometric sphere agglomerates and
single spheres with diameter ranging from 100 up to 300 nm. The agglomerate
particles (see Figure la) were similar in structure and dimensions to those
observed in diesel engine samples (see Figure lb). Single spherical particles, as
in Figure lc, were found only in the airborne particles; their origin could be
ascribed to a different combustion source. The microanalyses carried out on this
particles revealed a high content of carbon (about 78%), oxygen and, in some, a
small amounts of Zn and Na.

Fig.1 a) Soot airborneparticle; b) Soot particle exhausted by diesel car; c) Ash airborneparticle
 337

a. b.
Fig.2 Airborne collected onfilter D af 8m (a) and 3m @).
The vertical distribution was studied sampling the airborne particles for 15
minutes at two levels ( 3 and 8 m). The particles number was greater at lower
level with respect to the upper level (see Figure 2). This fact was due to
vehicular source and the break up of road surface that contribute to the
generation of fine and coarse particles, respectively, at road level, while finest
particles fraction, lighter than coarse fraction, can reach greater heights before
deposition.

Monitoring of Nox concentrations

The concentrations of NOx was measured in January, at the same time of
particulate samplings. NOx was detected by a W03 sensor doped with
manganese. The sensors were calibrated on the concentrations measured with
chemiluminecent spectroscopy device located in the monitoring unit of the
regional environment protection agency (ARPA). The NOx concentration,
measured at the two levels during the first 19 days of January 2006, showed
higher values at lower level, nearest to the vehicular source of the pollutant, as
for particulate.
The gas concentration follows a vertical as a daily dispersion, showing low
concentrations at night or at dawn and high concentrations at diurnal hours, with
peak values that correspond to moments of intense traffic. This it clearly seen in
Figure 3 , in which the daily hourly concentrations, of two subsequent weeks
were compared: the first was a holiday week, characterized by low traffic except
Friday, Epiphany day, hence low NOx concentrations; during the second the
concentrations were higher because the normal activities (school, works, etc.)
restarted, so the vehicular traffic increased whit NOx concentration. Especially
at 8.00 am, 14 pm and 20 pm higher values were recorded nearest to the road.
During the nigh time the vertical difference of concentrations was slight, except
when an inversion of concentration sometimes occurs for short time intervals.
Such an inversion is probably due to dispersion of pollutants from the bottom to
the top or the lack of vehicular sources during the night revelled the presence of
338

pollution sources at highest level, usually covered by traffic, as domestic and

I
00 00 00 04 00 00 08 00 00 12 00 00 160000 20 00 00 24 00 00

January 2-82006

500

400

.
7
E
-

Cn
2
300

8z 200

100

00-00 00 04 00 00 08 00 00 12 00 00 16 00 00 20 00 00 24 00 00

January 9-15 2006

Fig.3 Comparison of daily NOx concenhations measured in two consecutive weeks

It turns out from previous observations that the finer fraction of particulate
matter (with diameter less than 1 pm), collected near the street, is composed
mainly by soot particles recognized as diesel particles.
The vertical distribution of particulate matter highlighted greater concentrations
of soot particles at lower level, near the pollution source. This fact represents a
strong evidence that the source is traffic; in this situation, the dependence on
height of finer particles and NOx can be related.

References
1. M. Yip Wong, Exposure assessment in a bus way canyon,pp27-28
http://www.sbg.ac.at/ipk/avstudio/pierofun/exotica/thesis-mari.pdf.
2. Health Effect Institute, Research Directions to Improve Estimates of Human
Exposure and Risk from Diesel Exhaust, p. 154.
3. C . Misraa, M. Singha, S. Shena, C. Sioutasa, P. M. Hall, Development and
evaluation of a personal cascade impactor sampler (PCIS), Aerosol Science 33
(2002) 1027-1047.
 A DYNAMIC-ELEMENT MATCHED CCII
FOR SMART SENSORS

GiuseDne Ferri, Vincenzo Stornelli, Andrea De Marcellis, Christian Falconi*

and Arnaldo DAmico*

Dipartimento di Ingegneria Elettrica, Universita dell Aquila,Montelzrco di Roio,

67I 00 L Zquila, Italy, e-mail:,fkwi(din,c.mrisw.it
Dipartirnento di Ingegneria Elettronica, Universita di Tor Vergata, Rome, Italy

Summary
In this paper the dynamic element matching (DEM) technique is applied to a second gcneration
current convcyor (CCII) in order to compensate the input offsct and l/f noise voltages. Apart from a
rcsidual clock-fccdthrough error, a significant reduction in the offsct voltage between CClI X and Y
nodcs is achicvcd with a slight incrcasc in circuitry complcxity. As an examplc. the dynankclcnicnt
matchcd CCII has bccn uscd for thc design of a rcsistivc scnsor intcrfacc. Simulation rcsults confirm
thcorctical rcsults.

1 Introduction
Smart sensors have both the sensing element and the electronic interface
integrated in the same chip, resulting in many advantages, such as low cost, high
performance, low power dissipation, low susceptibility to interference, small size
and weight. For low cost reasons, CMOS smart sensors are generally preferable;
unfortunately CMOS amplifiers exhibit rather high input offset and noise
voltages, especially in low-voltage low-power applications. Consequently, in
CMOS interfaces for smart sensors it is often essential to compensate the input
offset and l/f noise voltages of the amplifiers; since static offset compensation
techniques require calibration and may not compensate time varying errors (such
as drift and l/f noise), only dynamic techniques are sujtable for CMOS smart
sensors. Circuits using dynamic techniques may be grouped in three main classes
[I 3: autozero circuits (AZCs), chopper circuits (CHCs) and circuits which use
dynamic element matching (DEM). In AZCs the error sampled in a first phase is
used to reduce the error during the second phase. CHCs modulate the signal
twice and the input offset voltage only once; as a consequence, the signal is
modulated to higher frequency and then demodulated back to the base band,
while the input offset voltage is only modulated to higher frequency and may be
removed by averaging (low pass filtering in continuous time systems or digital
averaging in non continuous time systems). However, since in symmetric (or
quasi-symmetric) circuit topologies offset is (mainly) due to mismatch of a
few coniponents, it may also be reduced by dynamically interchanging those
components; this strategy is known as dynamic element matching (DEM). In this

339
340

paper we have applied the DEM technique to a second generation current

conveyor (CCII) [2],which can be used as an active element in resistive sensor
interface [3-51.

2 Second Generation Current Conveyor (CCII)

Second generation current conveyors (CCIIs) are current-mode basic blocks
[2-51 which, in numerous applications, can favourably replace operational
amplifiers, both in linear and nonlinear contexts. In an ideal CCII device (see
Figure l), if a voltage is applied at Y node, an equal voltage is produced at X
node and the current flowing into X node is equal or opposite to the current
flowing inlo Z node. Moreover, Z and Y nodes show infinite impedances and
X node shows zero impedance. Positive (CCII+) and negative (CCII-) current
conveyors are respectively obtained for 1%= Ix and Iz = - Ix.

TJJ
VY T vx

1-. -1-1
Fig.2 CCII-DEM input stage block scheme

Typically, in CCIls, a unity voltage transfer fimction V x N y is ensured by

implementing a differential transistor input pair, but a residual offset due to
components mismatch affects the follower operation.

3 Dynamic-element matched CCIl

We have applied the DEM technique to a second generation current conveyor
(CCII) resulting, to the best of our knowledge, in the first ever reported offset
(and l/f noise) compensated CCI1. In Figure 2 the input stage of the dynamic-
element matched CCII is shown: the DEM technique is applied by means of the
 341

chopper switches (CHI-CH2). Figure 3 shows the complete transistor level

schematic of the proposed CCII-DEM. Ql and Q2 are the input transistor pair,
biased by Irep Q5,Q6. TrdnSiStOrS Qj, Q4 constitute the active load of the input
pair, while Q7, Qx and Q g , Q/o are the output stages (XZnodes respectively).
Finally, transistors Q,/-QSNform the dynamic switches clocked by two opposite
non-overlapping phases.

X
-3

L li; Q6 L i

Figure 4 shows the input offset voltage of the proposed CCII-DEM as a function
of the X-node load resistance. The systematic offset of the designed CCII is
about 40 pV; a random offset has been simulated through the insertion of a DC
voltage between the gate of the transistor Ql and the correspondent terminal of
the block CHI (see Figure 2).
Figure 5 shows a typical CCII offset behaviour vs. time domain when a
systematic offset of 2 mV has been applied at Y node: as shown CCII offset has
been reduced to 120yV.
342

-2mV

15mV

0 8mV

100 1000

Fig.4 CCII-DEM qfJsetsinitlulation results: a,forced oj5et has been applied afier compensation of the
i n p t switch

Fig.5 Qfliel coinpensated CCll when a systmratic ofset of 2m V is applied a/ Y node

4 cc for sensor interface applications

In Figure 6 a CCII resistive sensor interface is depicted [4-51. This scheme
utilizes two CCII-t- and one CCII- . The sensor is modelled by the resistance Rp;
if we call x its relative variation, the output voltage V,,,, is directly proportional to
x plus cornponcnts due to the offset o f the implemented CCIIs (Vo
indicates the i-th CCII); eq.( l), obtained designing VNEpz=VltEp.RdR, and
RZ=R~=RZJ, shows how low CCII offset values give a better sensor estimation by
eliminating the reidted terms. This interface has been both simulated with a
classical second generation CCIl and with the here proposed CCII-DEM. In both
solutions a random offset has been introduced in the current conveyors.
Simulation results has shown, as expected, a significant improvement in Rp
 343

estimation accuracy, when a CCII-DEM is used. due to the quasi-cancellation of

the offset tenns in eq( 1).

Fig6 CCII seiwor interface

5 Conclusions
In this paper we have applied the DEM technique to a second generation
current conveyor (CCIl). Simulation results, apart from residual clock-
feedthrough error, show a significant reduction in the offset voltage between
CCII X and Y nodes with a slight increase in circuiky complexity. The DEM-
CCII has been applied in the design of a CCll based resistive sensor interfaces,
significantly improving its accuracy.

References
I. C. Falconi, C. Di Natale, A. DAmico, M. Faccio, Electronic interface for the
accurate read-out of resistive sensors in low voltage-low power integrated
systems, Sensors and Actuators A, 2004.
2. G. Ferri, N. Guerrini, Low voltage low power CMOS currcnt conveyors,
Kluwer Academic Publisher, Boston, 2003, 726 pp., ISBN 1-4020-7486-7.
3. JSainitier, M. Puig-Vidal, S. A. Bota, C. Rubio, S. K Siskos, T. Laopoulos, A
Current-Mode Interface Circuit for a Piezoresistive Pressure Scnsor, IEEE
Trans. on Instrumentation and Measurements, Vol. 47, No.3, June 1998,
pp. 708-7 10.
4. C. Cantalini, G. Ferri, N. Guerrini, S. Santucci, A low voltage low power
current mode gas sensor integrated interface, ICM Proc. 2004, Tunisia.
5. G. Ferri, N. Guerrini. S. Del Re, F. Mancini CCII-based integrated gas sensor
interfaces. Sensors And Microsystems. Proc. AISEM (Italian Conference on
Sensors and Microsystems), 2004, pp. 424-429.
 Multisensor Layout for Robots in Hostile Environments

Claudio Moriconi, Marco Santoro

Department of Robotics, ENEA Research Center Casaccia, Via Anguillarese
301, 00060 Rome, Italy
[email protected]

Summary
Robots operating in harsh and hostile environments need reliable multisensorial systems able to
automatically acquire as much information as possible. Sensory data from a range of disparate and
independent multiple sensors have to produce an improved model or estimate of the domain of
interest. Therefore the automated intelligent combination of data from multiple sensors can derive
less ambiguousiuncertain information about the desired state.
E N E A , the Italian National Agency for New Technologtes, Energy and the Environment, is
developing a multisensorial layout for robots operating in such cnvironmcnts.

Introduction
A multisensorial distributed unit contains several discrete sensors over a
definite surface areaholume (Fig. 1).

- __

Fig. 1 General layout of a rnultisensorial distributed unit

344
 345

The Department of Robotics (http://wwwrob.brindisi.enea.it/) at ENEA

Research Center Casaccia is currently developing a multisensor device able

to measure and acquire different physical quantities in order to allow pattern

recognition of the gathered observations i.e. their classification and description.
The sensor system currently contains
30 Platinum Resistance Temperature Detectors of lOOQ@OC nominal
resistance (PTIOO);
0 30 Silicon PIN Photodiodes sensitive to visible and infrared radiation
(peak sensitivity at 900nm and spectral bandwidth range b 5=600 ...
1050nm);
0 Pressure Pad based on a foam material whereby light introduced to the
foam is scattered in a manner dependent on the force applied to the
surface of the foam.

How the multisensor device works

60 pressure sensors are spaced evenly (Fig. 2) throughout the area of the
touchpad (210mm X 350 mm). Each individual pressure sensing element
consists of a tiny light source and a tiny detector imbedded in a thin cellular
elastomer (Fig. 3). As pressure is applied to the elastomer in the region of the
sensor, the cell size distribution is altered along with the bulk optical scattering
properties of the material in that area. The light detector measures the change in
scattered light intensity at the sensor, and interprets this as a change in pressure.

Fig.2 Layout of the pressure pad

346

Fig. 3 The pressure pad

Optical fibers laminated into the structure are used as the nervous system, to
deliver transmi~edlight to the location of the sensors, and to deliver received
light back from the sensor. It is convenient and less expensive to use these fibers
rather than individual light sources and detectors actually located within the
material.

The light source is a light emitting diode (LED) connected to a bundle of

fibers, and the light detector is an array of photodiodes connected to a second set
of fibers. A microcontroller converts the signals hom the detectors to digital
signals, conditions those signals, and computes the required pressure information
or decision data using a set of algorithms implemented in firmware.
This pressure system is unaffected by electromagnetic interference, since light
intensity is used for measuring the pressure applied, but the foam layers have to
be protected by an outer covering material (Nylon New Perlite) that avoids outer
light to enter the optic fibers (Fig. 4).
 347

Fig. 4 Nylon cover, beneath the pressure pad. It completely covers the pad when pressures are being
detected

This layout is being completed by further sensor layers. 30 evenly spaced

Platinum Resistance Temperature Detectors of 1OOs2@OC nominal resistance
(PT100) and 30 evenly spaced Silicon PIN Photodiodes (Fig. 5) are being placed
over the pressure pad. These additional layers allow the detection of temperature
and visible and infiared radiation.

Fig. 5 Platinum Resistance Temperature Detector (Pt loo), range -50 .+ +600 "C @e$) and Silicon
PIN Photodiode, range /lo s=600 I O5Onm (right)
..I
348

Some inertial sensors (Fig. 6) will be inserted inside the pad in order to detect
low-frequency vibrations.

Fig. 6 fNERTfAL SENSOR: 3-Axis 2g/6g Linear Accelerometer

Partitioned data clustering algorithms are being developed with the aim of
classifying detected data. A complete pattern recognition system allows
improving the understanding of the information acquired.
 WIRELESS SENSOR NETWORKS OVERVIEW,
LOOKING AT THE ZIGBEE ARCHITECTURE

Danilo Blasi, Vincenzo Cacace, Luca Casone, Marco Rizzello

STMicroelectronics c/o Distretto Tecnologico-ISUFI (Pal.B, p.t.)
Via per Arnesano 16, 73100 - Lecce - Italy
(danilo.blasi, vincenzo.cacace, luca.casone, rnarco.riuello) @ st. corn

Summary
The Wireless Sensor Networks (WSN) are an emerging application field that promises wide
support to the interaction between people and their surroundings. Because of the exeemely
variable nature of t h s interactlon, the WSN topic is still in its stage of development and rises
several challenges, which span om the design of a smart device h a t allows easy and reliable
access to the environmental characteristics, to the creation of a wireless network of such devices
that cooperate to perform complex tasks. Besides, the pressing market request for low
cost/power/dimensionproducts emphasizes the need for standards against the use of expensive
proprietary solutions adopted so far. In this paper, we will discuss the most important features
of WSN, introducing the IEEE 802.15.4 and ZigBee architecture, showing how their key
aspects fit the WSN requirements.

1 Introduction to WSN
In the recent years, improvements in embedded systems and integrated radio
circuits have made available a new kind of applications: the Wireless Sensor
Networks.
The potential of the WSN concept [ 11 simply lays in the three words composing
the acronym itself Wireless puts the focus on the freedom that the elimination
of wires gives in terms of mobility handling and ease of the system installation;
Sensor reflects the capability of micro-lnano-technology to provide means to
perceive and interact - in a wide sense - with the world; Networks gives
emphasis on the possibility to build up systems whose functional capabilities are
given by a plurality of devices possibly distributed over large areas.
WSN are usually characterized by a high number of very low cost devices,
densely deployed, with limited energy and computational resources. A possible
simplified functional architecture of a generic WSN device is shown in Fig. 1.
Five basic components are illustrated: the transducing unit (XDCR, including
multiple sensors andor actuators), the micro controller unit (MCU), the
memory unit, the radio interface and the power unit. The device may also
integrate two additional components such as a location finding system (LOC)
and a mobilizer (MOB): they are special subsystems providing information
about the devices position and allowing the devices mobility, respectively.

349
350

Fig.1 ~ e ~ n c t ~ oarchitecture
~ nal of a WSN device

Altogether, the networks members are able to monitor andlor control a wide
variety of environmental conditions that include, but are not limited to,
temperature, humidity, vehicular movement, lightening conditions, pressure.
In effects, the great interest WSN are gaining is essentially related to the nearly
infinite range of their applications. In the simplest form, a WSN may include
deviceslnodes exchanging data with a single base station, which is in charge of
synchronizing and managing all the network transmissions. In such an
infrastructur~network organization, however, the covered area and the
manageable number of members are respectively limited by the wireless link
range and the base stations elaboration capabilities. Clearly, a larger area may
be monitored by deploying multiple base stations and making them
interconnected (through either wired or wireless links), but this is an expensive
solution due to the infrastructures implementation. An alternative, challenging
approach - which would also potentially boost the system scalability - is to
manage communications in a distributed and in~astruc~reless way (otherwise
said ad hoc mode). Such networks are composed by nodes only, possibly with
different kind of functionalities and capabilities (i.e., heterogeneous networks)
not including base stations. Nodes are required to be self-configuring and self-
organizing to participate in distributed and multi-hop communications, where
fixed infrastructure are unnecessary. This way, distributed measurement and
control systems can be more easily built up, even over wide geographic area. It
has to be pointed out that the realization of such systems has to deal with low-
cost nodes: this implies limited energy and scarce processing resources. The
multihop communications well fit with the first constraint, as they are expected
to consume less energy than the traditional single hop ones; conversely, multi-
hopping requires a higher number of nodes that execute advanced networking
protocols to face the topology changes mainly due to nodes failures. On the
 351

other side, the high number of devices favours both the robustness (if a node got
failed, another would take its place) and information reliability (the same
information can be gathered from multiple, different sources). The redundancy
of correlated information (i.e. produced by multiple sensor devices that are
reporting of a common phenomenon) is usually considered also in the relaying
process: in particular, rather than separately propagating high correlated data
items towards their intended consumers, they may be aggregated, depending on
the target application, as they are flowing through the network, so that
considerable energy and bandwidth savings can be achieved. This is the main
idea behind a well-know data centric technique called in-network aggregation
[2, 31.

2. Many applications, few standards: IEEE 802.15.4 and ZigBee

As already said, WSN may effectively support many new application areas. For
example, given their characteristics of rapid deployment, self-organization and
fault tolerance, they are very valuable in emergency scenarios (e.g., search and
rescue, battlefield, etc.). Other typical application scenarios include industrial
and building automation, environmental control, health care, urban traffic
control, interactive museums, and so on.
This variety of WSN domains may make very hard the design of a unique
general-purpose HW/SW platform. In spite of producers and vendors concrete
actions aiming at developing largely reusable and interoperable components
and/or devices, the sensor and control arena is still mostly composed by many
proprietary solutions that are almost incompatible.
However, one of the proposed standards well suited for WSN is the IEEE
802.15.4 [4] for Low-Rate Wireless Personal Area Networks. It belongs to the
IEEE 802.x family and, such as, defines specifications for the ISO/OSI Medium
Access Control (MAC) sub-layer and the Physical Layer (PHY) (see Fig. 2). Its
main target is low data rate wireless connectivity (up to 250 Kbps, in 27
channels of ISM band) among low power, low cost, low dimension devices,
typically operating in a personal operating space (POS) within tens of meters.
The specifications establish a rough classification of devices, distinguishing
between Reduced Function Devices (RFD) and Full Function Devices (EFD).
The former identified as the networks users or leaf nodes, are meant to be
devices with poor resources. The latter, implementing the full set of networking
functionalities (e.g. coordination, addresses assignment, packet relaying, etc.),
can be identified as the WPAN coordinators and are supposed to have higher
resources. Following this classification, devices are allowed to establish
communications, according to two possible network configurations: star
(RFD/FFD devices can communicate through a FFD intermediate central
coordinator) and peer-tepeer (FFD devices can directly communicate,
352

without mediators). To manage networks deployed over wider areas (e.g. multi-
hop relaying), protocols above layer two of ISO/OSI stack (see Fig. 2) have to be
defined. The ZigBee Alliance [ 5 ] is just carrying out this definition process atop
the IEEE standard.
The ZigBee Alliance is an association of Companies working together to enable
reliable, cost-effective, low-power, wirelessly networked, monitoring and
control products based on an open global standard. The Alliance is not pushing
a technology itself; it is rather providing a standardized base set of solutions for
sensor and control systems. The ZigBee membership includes promoters (like
Philips, Motorola) and participants (like S T M i ~ ~ l ~ t r o n i c s ) .
The ZigBee specifications concern the networking and security protocols, the
application framework and services for different application profiles. Other
addressed functionalities relate to (i) discovery and management of multi-hop
paths, (ii) reliable transfers (based on handshaking procedures), (iii) devices
description according to supported application profiles, (iv) wide support to
application layer (e.g. devices and services discovery).
ZigBee was ratified in December 2004; since then, a growing number of
companies interested in integrating ZigBee technology within their products has
appeared on the market: so that a steady growth in ZigBee-compliant devices
availability is expected in the near future. Nevertheless, different issues are still
either not clarified or not solved at all. Just a new ZigBee version is under
definition to cope with, among all, (i) devices mobility support, (ii) datagrams
~agmentation,(iii) multicast traffic management. Other topics are still not
faced up to, such as mechanisms for knowing nodes physical locations and

Fig.2 - The IEEE 802.1.5.4 and ZgBee protocol stack

 353

position-aware protocols.
Together with IEEE 802.15.4standard, ZigBee was expected - by its promoters
- to become a fully defined protocol stack for WSN, leaving to system
developers the realization of the application software only. However, ZigBee
already does not appear to be able to support all WSN possible applications,
even if the definition of new application profiles is in progress. For example, the
address centric ZigBee/IEEE 802.15.4framework does not seem to well match
with data centric WSN, like those considered in the well known Smart Dust
Project [6].Moreover, the production and deployment costs are still too high.
All in all, there is still room for R&D activities in the field of WSN.

4 Conclusion
According to recent marketing reports on the factors influencing the WSN
market [7, 81, the success of competing WSN solutions will be based on
customer demand for reliability, simplicity, price and availability, with the risk
to obtain too slow standardization efforts and many prompt proprietary
solutions. On the other side, experience demonstrated that an accepted
industrial standard is better than no standard at all but, in its turn, it will be
always determined by wide industrial and market consensus. For sure, even in
the absence of consolidated standards, there will be an explosive and pervasive
adoption of WSN and their applications in the near future!

References
[ 11 I.F.Akyildiz, M.C. Vuran, 0. B. Akan, W. Su, Wireless Sensor Networks:
A Survey Revisited, Computer Networks Journal (2005)
[2] B. Krishnamachari, D. Estrin, and S. Wicker, Modelling Data-Centric
Routing in Wireless Sensor Networks, Proc. of IEEE Infocom, 2002.
[3] C. Intanagonwiwat, R. Govindan, D. Estrin, J. Heidemann, and F. Silva,
Directed Diffusion for Wireless Sensor Networking, Proc. of IEEE/ACM
Transactions on Networking, 11(1), 2003.
[4] http://standards.ieee,org/getieee802/download802.15.4-2003 .pdf
[5] http://www.zigbee.org
[6]J.M. Kahn, R.H. Katz, and K.S.J. Pister, Next century challenges: mobile
networking for smart dust, Proc. of the ACM MobiCom, pp. 271-278,
1999.
[7] K. West, Wireless Sensor Technology and Market Tracking Service:
ZigBee, Zwave, Insteon, RFID, IEEE 802.15.4and their Competition,
Report abstract WTRS, 2005.
[ 81 J. Dexheimer and Dr. R. Hannemann, Investment Opportunities in Sensor
Networking: Dust, Hype, Fuzz, and Reality, First Analysis with Physical
Sciences Inc. and D13 Partners, 2004.
MOTES SENSOR NETWORKS IN DYNAMIC SCENARIOS:
A PERFORMANCE STUDY FOR PERVASIVE
APPLICATIONS IN URBAN ENVIRONMENTS

M. CONTT, E. GREGOFU AND G. VALENTE

Institute for Informatics and Telematics, Italian National Research Council,
Via G. Momzzi 1,
56124, Pzsa, Italy
E-mail:{firstname.lastname} @zit.cnr.it

G. ANASTASI AND C. SPAGONI

Information Engineering Department, University of Pisa,
Via Diotisalvi 2,
56122, Pzsa, Italy
E-mail: (firstname.lastname) @aet.unipi.it

The use of mobile agents for data collection tasks in sensor networks may have
significant advantages over traditional ad hoc sensor networks. Such agents, also
known as Mobile Ubiquitous LAN Extensions (MULEs), have already been de-
scribed and evaluated in literature by the means of simulation and numerical anal-
ysis. In this paper we investigate the performance of the data MULE model and its
ability t o support sensor-based applications in a real urban environment through
an experimental analysis. Specifically, we use a testbed based on Berkeley motes,
and analyze the impact of parameters like the MULEs speed and the distance
between the static node and the MULEs line of motion on the contact time. The
results obtained are very promising and show that the MULE architecture is really
suitable for a large set of sensor-based applications in urban environments.

1. Introduction
As discussed in, in a delay-tolerant scenario a three-tier MULE architec-
ture provides the benefits of both the infrastructure and the infrastructure-
less approach to connectivity, in terms of sensor energy consumption, data
success rate and infrastructure cost. It comprises of a three-tier layered
abstraction that can be adapted to different situations and needs: (i) a top
tier for WAN connection; (ii) a middle tier of MULEs; (iii) a bottom tier
of static wireless sensor nodes.
The top tier (i) is composed of Access Points and is meant to be set

354
 355

up at convenient locations, where reliable network connectivity and power

are guaranteed. Access Points should communicate with a central data
warehouse t o synchronize the data they collect.
The intermediate layer of mobile MULE nodes (ii) is characterized by
large storage capacities (relative to sensors), renewable power, and the abil-
ity t o communicate with both sensors and networked Access Points. As a
result of their motion, MULEs collect and store data from the sensors.
The bottom tier (iii) normally consists of randomly distributed wireless
sensor nodes. In our case however, the placement is strategic depending
on the purpose of the application. Anyway, the work performed by sensor
nodes has t o be minimal, as their constraint on resource usage is the highest
of all tiers.
The three-tier architecture supports increased reliability as the redun-
dant Access Points and multiple MULEs create a fault-tolerant system
where failures can only lead to reduced data success rate and increased
laten~y~?~.

2. Experimental Environment and Methodology

Our experimental testbed is based on Mica2 Motes with the TinyOS oper-
ating system and MIB510 sinks.
We ran our experiments in a big parking area a t the CNR campus a t
Pisa: the actors were a mobile sensor node (the MULE), trying to collect
the data packets sent by a static sensor node (see Fig. 1).
In our experiments we measured the system's performance in terms of
contact time.
In many papers the radio range R of the static node is assumed t o be
fixed. Hence, the packet loss experienced by the mobile MULE is assumed
t o be 100% when the distance between the mobile and the static node is
greater than R, and immediately drops to 0% as soon as the MULE enters
the radio range of the static node. Under this hypothesis, the contact time

static node
,D' t
DY

Figure 1. Testbed environment

356

is defined in4 as the time interval during which the mobile node is within the
radio range of the static node. The contact time can be derived analytically
in terms of the radio range R, the distance between nodes, and the MULEs
speed4. Accordingly, the number of packets N,, correctly received by the
MULE can be easily derived based on the contact time and the bit rate of
the wireless link (which is assumed constant as well).
The above communication model is simple and analytically tractable.
However, it is not completely realistic. We considered a different, more re-
alistic, definition of contact time. In our analysis the contact time is defined
as the total time during which the packet loss experienced by the MULE
is below a given threshold (e.g., 15%). The contact time thus gives an in-
dication of the amount of time during which the MULE is able to receive
data from the static node with a given communication quality (expressed
in terms of packet loss).
We evaluated the effect of the MULEs speed and the perpendicular
distance between the static node and the MULEs line of motion (D,in
Fig. 1) on the contact time. We replicated all the experiments several
times to increase the accuracy of the results. The results presented below
are averaged on all the replicas. A more detailed view of the results is
available in .
3. Experimental Results
As mentioned above, in an urban environment the role of MULE may be
played by different actors (e.g., persons, buses, cabs) and, hence, MULEs
may have different speeds. Therefore, we decided t o divide our analysis
into two parts. First, we considered a low mobility scenario assuming that
MULEs are pedestrians and, hence, their speed is limited in the range [l,
21 m/s. In the second part, we considered a high mobility scenario where
MULEs are buses (or cabs) and their speed varies in the range [20, 401
km/h.

3.1. Low Mobility Scenario

The low mobility scenario is characterized by a limited range of speeds
(from 1 to 2 m/s). In a such a scenario, in addition to the MULEs speed,
the distance between the nodes, and the duty cycle of the static node may
impact both the contact time and the total number of packets received by
the MULE. Fig. 2(a) shows the packet loss as a function of the distance
between the nodes. In our experiments we considered three different val-
 357

ues for the perpendicular distance, D,, between the static node and the
MULEs line of motion: 15, 25 and 35 m. For the sake of clarity in Fig.
2(a) the curve at 25 m is omitted (the results for 25 m are in between those
for 15 and 35 m, respectively). From the packet loss curve we can easily
derive an estimate for the contact time. Assuming a threshold value for
the packet loss of 15% we have approximately the following values for the
contact time: 61 s (15 m), 36 s (25 m), and 12 s (35 m).

(a) Impact of distance in low mobility (b) Impact of the MULEs speed in high
scenarios. mobility scenarios.

Figure 2. Packet loss and contact time in low and high mobility scenarios.

From the above results it emerges that the perpendicular distance D ,

has a strong impact on the contact time.

3.2. High Mobility Scenario

A MULE scenario in an urban environment can theoretically involve mobile
nodes attached to anything. For example we can think of buses acting as
MULES, as they are intrinsically characterized by coordination and attitude
to full coverage of a zone. Based on the above remarks, we extended our
analysis t o a high mobility scenario where the MULE moves a t speeds
typical of vehicles in residential areas (20 to 40 km/h).
By the time we were planning those experiments we observed that, a t
these speeds, when D, > 15 m the communication becomes very difficult
and unreliable. Therefore, we decided to run this second set of experiments
at a distance D, = 15 m.
Figure 2(b) shows the packet loss as a function of the distance D,
measured along the MULEs line of motion, for different MULEs speeds.
As expected, the contact time becomes shorter and shorter as the MULEs
358

speed increases (Fig. 2(b)). With a MULE moving at 20 km/h, t h e two

nodes are able t o communicate with a packet loss lower than 15% for no
more than 10 meters (-4 < D, < 6), resulting in a contact time reduced
t o around 2 s. In the case of a MULE moving at 40 km/h the contact time
is approximately 0 s. However, even with such low values of the contact
time, the number of packets successfully transferred t o the MULE is not
negligible.

4. Conclusions
Based on the above results we can conclude that the MULE architecture
is really suitable for a rich set of environmental monitoring applications in
urban environments, both in low and high mobility scenarios. Our analysis
was based on Mica2 Berkeley motes. We can reasonably expect t h a t with
future advancements in sensor networks technology, it will be possible t o
achieve even better results that those presented in this paper.

Acknowledgments
This work has been carried out under the financial support of the FET-IST
HAGGLE project and the Italian Ministry for Education and Research
(MIUR) in the framework of the FIRB-PERF and FIRB-VICOM projects.

References
1. R. C. Shah, S. Roy, S. Jain, and W. Brunette. Data MULES: Modeling and
Analysis of a Three-tier Architecture for Sparse Sensor Networks. Elseuaer Ad
HOCNetworks Journal, uol. 1, assues 2-3, 2003.
2. M. Ho and K. Fall. Poster: Delay Tolerant Networking for Sensor Networks.
In Proc. Farst IEEE Conference on Sensor and Ad HOCCommunications and
Networks (SECON), Santa Clara, CA, USA, 2004.
3. D. Jea, A. Somasundra, and M. Srivastava. Multiple Controlled Mobile
Elements (Data Mules) for Data Collection in Sensor Networks. In Proc.
IEEE International Conference on Distributed Computing in Sensor Systems
(DCOSS), Marina del Rey, CA, USA, 2005.
4. S. Jain, R. C. Shah, G. Borriello, W. Brunette, and S. Roy. Exploiting Mobility
for Energy Efficient Data Collection in Sensor Networks. In Proc. Modeling and
Optamizataon i n Mobale, Ad HOCand Wireless Networks (WaOpt), Cambridge,
UK, 2004.
5. G. Anastasi, M. Conti, E. Gregori, C. Spagoni, and G. Valente. Motes Sen-
sor Networks in Dynamic Scenarios: an Experimental Study for Pervasive
Applications in Urban Environments. Journal of Ubiquitous Compuntzng and
Intelligence (American Scientific Publishers), uol. 1, issue 1, 2006.
 PORTABLE SOFTWARE TOOLS FOR CONFIGURING A
IEEE 802.15.4 WSN WITH MOBILE DEVICES
Marco Rizzello
STMicroelectronics, SST Advanced System Technology, WSNplatform team
Viaper Arnesano 16, c/o Complesso Garrisi, Building B, 73100 Lecce, Italy
marco.rizzello~,s~.,st.com
Anna Gentile, Federico Vergori
Universitri degli Studi di Lecce, DII, Facolta di Ing. Informatica, Lecce, Itab
anna.pentile(iunile.it

Summary
This paper presents a framework for setting up a Wireless Sensor Network (WSN) by means
of a mobile device like a PDA, through an IEEE 802.15.4 wireless link. This problem of
installing a network of small sensing devices in a general environment is often referred to as
Commissioning practice, especially in the industrial world. In order to setup such a
network, there is the need to face firstly with the network connections establishment, and
when the topology is well formed, then it turns the time to setup the application related
parameters, as the trigger levels, policy rules for actuators, timers and so on. A serial
communication protocol has been designed and introduced between an IEEE 802.15.4 node
and a PDA, in order to allow to the PDA to enter in the WSN and to tune the network and
applications knobs. Once setup is done, the application could be started and the status of
sensors and actuators IS updated on the PDA. The protocol supports only star topology. A
small demo has been succeeded by using few nodes, a commercial PDA and prototypal
IEEE 802.15.4 boards.

1 Introduction
Wireless Sensor Networks are the new frontier of computing; typically we
suppose they are built up with hundreds or even thousands of small, battery-
powered devices, scattered throughout a target environment that can be indoor as
for Home/Building/Industrial AutomationlMonitoring,or outdoor as for Military
or Environmental applications.
To make feasible the actual deployment of these systems, there is the need to
lower the costs both of the single device and of the overall network
infrastructure; the former is primarily a problem of technology maturity but
requires also a general design of the single node; the latter is a question about the
smartness of the network.
Our work focuses on the second aspect, experiencing the design and the
realization of a software tool that can ease the creation of compelling
applications for WSN, facing the problems tied to initial setup both of the
networking layer, the core of the operation, and of the applications related
issues; all in all, trying to generalize these setup procedures, despite the several

359
360

scenarios that can be imagined for each application field. In particular, we

implemented these tools upon an IEEE 802.15.4 compliant WSN, facing
problems related to this architecture and protocol stack, briefly described here.

2 IEEE 802.15.4/ZipBee
ZigBee is today the only standard-based technology that addresses the needs
of a great part of WSN applications and it was designed to satisfy the markets
needs for a cost-effective, standard-based low-data-rate wireless network,
supporting low power consumption, security and reliability [1][1]. To this aim,
the ZigBee Alliance, an industry-based working group (www.zigbee.org) to
which STMicroelectronics adheres, is developing standardized application
software on top of the IEEE 802.15.4-2003 wireless standard [2], by working
closely with the IEEE to ensure an integrated, complete, and interoperable
network, providing interoperability certification testing of 802.15.4 systems that
include the ZigBee software layer. It may be helpful to think of IEEE 802.15.4
as the physical radio specifications and ZigBee as the logical network and
application software, as Fig I illustrates.

Fig I The ZigBeeprotocolstack

ZigBee-compliant products operate in unlicensed bands, including the

worldwide 2.4GHz (16 channels), with a maximum raw data-throughput rates of
25OKbps, and an achievable transmission distance expected to range from 10 to
75m, depending on power output and environmental characteristics.
ZigBee networks are typically built with a combination of three device types:
* the network coordinator (PAN - Personal Area Network - Coordinator), the
most sophisticated device, that requires the greatest amount of memory and
computing power in order to maintain overall network knowledge;
 361

the full function device (FFD), that supports all 802.15.4 features; it is
responsible for network router functions, but it could also be used in network-
edge devices;
the reduced function device (RFD), that carries limited functionality to lower
cost and complexity. Its generally found in network-edge devices (where the
network touches the real world, by means of sensors), as slave of an FFD.
Ultra-low power consumption is the way ZigBee technology promotes a long
lifetime for devices with non-rechargeable batteries, trying to conserve the
power of the slave nodes, For most of the time, a slave device is in deep-sleep
mode and wakes up only for a fraction of a second to confirm its presence in the
network (the transition time from sleep mode to active state is around 15ms and
new slave enumeration typically takes just 30ms). ZigBee networks can use
beacon or non-beacon operation modes, while security and data integrity are key
benefits of the ZigBee technology, leveraging the security model of the IEEE
802.15.4 MAC sub-layer which specifies four security services, and
implementing a special cross-layer in the ZigBee protocol stack.
A general ZigBee devicelnode is essentially made of a MCU with memory, a
low-power Radio, a battery and one or more transducers that determines the
functional behaviours of each node; but, despite their simple and low-count
element composition, the nodes have to perform complex functionalities, given
that the strength of the system has to be put in the network and not in the single
device, so a tool to configure this sort of distributed engine is needed.
In the scientific community a general property of WSN is that often named
self-healing or self-organization, borrowed and inherited by the Ad Hoc
Networks studies, where it is intended that the network must be able to operate
under very dynamic conditions in an unattended mode, referring especially to the
network layer functions (topology management, addressing, routing and so on).
From the point of view of the system integrator and of the end user, that is
our main interest in this work, the network will operate in three different states,
specifically start-up, steady, and failure state.

3 Our Software Tool

With this software tool we address the first two stages, ignoring the
maintenance of the network and failure issues. In practice, it has been designed
by defining procedures to setup and manage a ZigBee network, providing for
network topology control, node addressing and application properties (user
entered description, locatiodtagging information and grouping) and node
binding, ignoring for the moment security-related issues, as well as network
management support (reconfiguration, change of topology and device roles,
network diagnosis).
The services offered by this software will aim to make ZigBee-WSN quickly
programmable (initial setup), checkable and usable (steady state). Initial setup
362

includes both network and application parameters (e.g. sampling time, standby
time).
Functionalities offered by this tool are divided into three levels: NWK, ZDO
and AF. NWK level includes network scanning and associations; services of
ZDO level allow user make device and service discovery (e.g. number, address,
type and features of all devices of the network); AF level includes all
functionalities that allow monitoring and control of a Home Automation
environment.
We developed our software upon the Java 2 Micro Edition Platform,
following the Connected Device Configurations (CDC) guidelines, as we think it
will be useful to deploy this tool on mobile devices (cell phones and PDA),
without worrying of the different platforms, being the Java technology portable
and more common in these types of devices. We used a commercial PDA and
the Chipcon CC2420 Development Board Kit for tests, in a Building Automation
scenario, realizing the monitoring of the environment in our laboratories as
shown in Fig 2.

Fig 2: Home Automation testbedfor the development tool

unication ~rotocol
The communication between PDA and PDA-Node is based on a client-server
model through a RS232 link. Client is represented by PDA, by which the user
can make requests to PDA-Node that is the server; PDA-Node is a kind of proxy
between the PDA and the coordinator of the ZigBee network. Communication is
performed using a variable length packet consisting of the fields shown in Fig 3.
The packet is made of a SOP byte used to characterize the incoming data
transmission, a Header of a fixed length, a variable length Payload and a FCS
that is used for checksum.
 363

Fig 3 The packetframe

The Header field is made of 3 sub-fields: Seq num, Command Id Field and
Length. The Seq-num sub-field, 1 byte long, specifies the sequential number of
each packet, allowing to link each request with the corresponding response. The
Command Id Field sub-field, 2 bytes long, is made of 3 sub-fields: Cmd Type,
Cmd Level and Cmd Id. The Cmd Type sub-field, 4 bits long, specifies if the
command is a request, a response, a subscription or a debug message; the Cmd
Level sub-field, 4 bits long too, defines the application domain (AF, ZDO or
NWK Level); the Cmd Id sub-field, 1 byte long, specifies the command type as
regards an application domain. The Length sub-field, 1 byte long, expresses
the number of bytes of Payload.
The Payload field has both a variable frame and a variable length due to the
packet type. The FCS field, 1 byte long, is used for checksum to identify errors
in packet transmission.

Conclusions
The main goal of this work has been the implementation of the IEEE
802.15.4 RF capabilities on a PDA node that could be used to setup and monitor
a WSN. To this extent a serial communication protocol has been implemented
between the PDA itself and a IEEE 802.15.4 prototypal node. Both network and
application parameters could be adjusted in order to control the monitoring
application. A small demo with few nodes has been succeeded.

References
[ 11 ZigBee Specification, Version 1.O, 14 December 2004.
[2] IEEE Std. 802.15.4, Wireless Medium Access Control (MAC) and Physical
Layer (PHY) Specifications for Low-Rate Wireless Personal Area Networks
(LR-WPAN), 1 October 2003.

Acknowledgments
This work has been performed within the project FIRB ccMiTeTIV -
MIcroTEchnologies for Virtual Immersive Telepresencen founded by Minister0
dellIstruzione dellUniversita e della Ricerca under PNR 2001-2003,
 Low-Power Sensor Front-End Electronics
With RFID Communication Capabilities
For Food Logistic Datalogging Applications

M. Cicioni*, A. Scorzoni, F. Alimenti, P. Placidi, L. Roselli

Dipartimento di Ingegneria Elettronica e dell'lnformazione, via G.Duranti 93,
06125 Perugia, Italy
michele.cicioni@,diei.unipg.it, Ph: +39-075-5853644, Fax: +39-075-5853654

S. Zampolli, I. Elmi, G.C. Cardinali, M. Severi

CNR-IMMBologna, via P. Gobetti 101, 40129 Bologna, Italy

S. Marco, J.M. Gbmez, F. Palacio

Universitat de Barcelona, Departament d'Electronica, Marti i Franqub I ,
08028 Barcelona, Spain

B. Mazzolai, A. Mondini, V. Raffa, V. Mattoli, P. Dario

CRIM - Research Center in Microengineering, Scuola Superiore Sant'Anna,
Via Rinaldo Piaggio, 34, 56025 Pontedera (PI), Italy

R. Ingles Bort, J. L. Ramirez, E. Llobet

Universitat Rovira i Virgili, Dept. of Electronic Engineering, Avda. Pabos
Catalans 26, Campus Sescelades, 4300 7 Tarragona Spain

E. Abad
Fundacidn Tekniker, Spain

T. Becker
EADS Deutschland GmbH, Munchen (Germany)

Summary
In this work, we describe a low-power transponder based on a Texas Instruments MSP430
microcontroller which interfaces both with sensors and with a passive, 13.56 MHz WID
front-end. This prototype features commercial RH, temperature light sensors. We tested the
correct operation of the transponder using a commercial reader compliant to the ISO-15693
directives. The implemented commands are correctly managed using a pulse position coding
1 out of 4, a 100% ASK modulation from reader to transponder and a high data rate from
transponder to reader.

364
 365

uctio
Current and future concerns related to agrofood safety and quality will
increasingly require a multidisciplinary and universal approach based on the
massive use of simple detection systems able to be used near to the foodstuff.
The technology used nowadays to assess food safety and quality relies on lab
solutions that are bulky, costly, punctual and time consuming. Innovative
approaches will comply, through the development of innovative Micro and Nan0
Technology solutions, with the needs of ubiquity, low cost and low power, fast
response, simple use and fully interconnection to the decisional bodies. In the
framework of the GoodFood FP6 Integrated Project, a flexible transponder gas
sensing system with WID comunication capabilities for food logistic
applications is being proposed. The transponder, besides storing dynamic
traceability information, features physical sensors (relative humidity (RE?),
temperature, light) and, depending on the application scenario, will also feature
Metal-Oxide chemical gas sensors. The envisaged GoodFood-WP6 scenario
imagines the RFID transponder and the Reader immersed in the so-called
Ambient Intelligence (AmI) for ubiquitous sensing in the frame of increased
food safety and quality.
In this work, we describe a low-power transponder based on a Texas Instruments
MSP430 microcontroller [ 11 which interfaces both with sensors and with the
passive, 13.56 MHz RFID front-end communication electronics. This prototype
features a Sensirion SHTI5 RH and T sensor and a commercial light sensor,
while low-power MOX sensor readout is planned for the next versions. The
transponder is designed to give information on food evolution during transport.
After being programed with data logging parameters at the beginning of the
transport phase, it should store information from the various sensors during
transport and this information should be read on arrival of the food at final
destination.

Fig. I Block diagram of the transponder

366

Architecture of the transponder

Figure 1 shows a block diagram of the transponder. The microcontroller
MSP430F 169 is connected to the SHTl5 through a two-wire bus which provides
temperature and relative humidity measurements in digital form. Data
communication is implemented using a proprietary pseudo 12Cprotocol [2].
The light sensor is composed of a visible to IR sensitive diode [3], the electrical
current it produces is a function of the amount of incident radiation, then the
current signal is converted to voltage signal using a low power operational
amplifier. The acquisition is performed using the internal ADC of the
microcontroller.

Communication
In W I D systems, the reader radiates an electromagnetic signal that is also used
by the transponder to send the information back to the reader using a load
modulation technique. The communication between the transponder and the
reader presently uses a subset of ISO-15693 directives. Focusing on the
transponder, the reception of incoming commands is based on a ASK
modulation with pulse position coding, this signal is demodulated by monitoring
the state of a digital signal produced by a counter in the clock generator block.
This signal is periodic except in the time slot where the modulation is actuated.
Conversely, the transmission from the transponder to the reader is based on a
FSK load modulation with two subcarriers. A suitable modulation circuit driven
by a a programmable logic device accomplish this task.

Antenna
The antenna is built on DuPontTMPyraluxB AP flexible substrate featuring a
thickness of 50.8 pm, 3.4 relative dielectric constant, 0.008 dissipation factor.
The substrate is bonded to a 18 pm thick copper foil. The antenna works in near
 367

field operating region and so the coupling between the devices is mainly
magnetic. In order for the RFID transponder to receive the needed energy from
the eIec~omagneticsignal, a suitable voltage level should be present at the
antenna. To increase this level, the resonance frequency of the antenna must be
tuned at the operative frequency. RF simulations were performed using AsiticTM
and AWR Microwave OfficeTMin order to design the best antenna that could fit
a credit card size. In particular, Asitic returned a more correct value of the
antenna inductance, while Microwave Office featured was better suited for
calculating the complete behavior of this type of structure at the relevant
frequencies. The resulting antenna is a five loop planar inductor, featuring a self-
resonance frequency higher than 13.56 MHz, due to the presence of embedded
parasitic capacitances. Figure 2 shows the equivalent circuit of the designed
antenna. To decrease the resonance frequency towards the actual operating
frequency we paralleled the inductor with a capacitor C,,, of suitable value. This
capacitor was integrated in the antenna, as shown in Figure 3. Because a very
precise estimation of the parasitic capacitance requires a very good processing
control, we built a modular tuning capacitance (Figure 2) in order to be able to
fine tune the resonance frequency after processing.

2.5 PF

Fig. 2 Equivalent circuit of the designed antenna.

Fig. 3 (a) Layout and (b) physical realization of theflexible antenna

368

Transponder operation
The reader-to-transponder and transponder-to-reader communication are
managed by a Gate Array Logic (GAL) and by the microcontroller. The reader-
to-transponder communication includes a hardware counter which helps the
microcontroller to detect the 100% ASK modulations, while the transponder-to-
reader communication includes the generation of the frequencyf,/28, wheref, is
the carrier frequency, the Manchester encoding and the modulation with two
subcarriers. Presently the microcontroller code allows the reader to perform ISO-
15693 read single block and write single block operations. We tested the
correct operation of the transponder using a commercial reader compliant to the
ISO- 15693 directives. The implemented commands are correctly managed using
a pulse position coding 1 out of 4, a 100% ASK modulation from reader to
transponder and a high data rate from transponder to reader.

Conclusions
In this work we presented a low-power transponder based on a Texas
Instruments MSP430 microcontroller interfaced both with sensors and with a
passive, 13.56 MHz W I D front-end. This prototype features commercial RH,
temperature light sensors. We tested the correct operation of the transponder
using a commercial reader compliant to the ISO-15693 directives. The
implemented commands are correctly managed using a pulse position coding 1
out of 4, a 100% ASK modulation from reader to transponder and a high data
rate from transponder to reader.

References
1. MSP430xlxx Family users Guide, Texas Instruments (SLAU049E)
2. SHTlx / SHT7x, Humidity & Tempcrature Sensor, datasheet
3. SLCD-61NII Solderable Planar Photodiode, datasheet

Acknowledgments
We would like to acknowledge Luca Abbati and Marco Gubbiotti for their
commitment during their Laurea project.
 A Neural Spectral Classifier for Optical Sensors

Mario Medugno
I.M.M. - C.N.R., Neaples, Italy
niario.medugno(i$imn.cnr.it

Summary
Multispectral remote sensing deviccs have the potential for large scale integration and can be
successfully used in environmcntal monitoring; in particular optical sensors could embed
microelectronic circuitry in order to perform signal preconditioning and more complex classification
tasks for specialized applications. We propose a neural approach for supervised classification of the
spectral signature of electrically scnsed optical emissions, since the neural networks can be trained
for the correct classification without any predetermined logic or programming burden. The s o h a r e
and hardware implementation of a such smart optical sensor for the spectral signature recognition is
discussed and the classification results for fire detection are presented.

1 Introduction
Automatic surveillance systems on wide territorial areas are infrastructures of
growing interest, motivated by natural and anthropic disasters, which alas
determine huge people injury, ecological and economical losses. Efficient
environmental monitoring systems can mainly accomplished by two integrated
electronic subsystems, namely sensors and radio communication networks.
Monitoring actions are based on environmental remote sensing (over or outside
the earth surface) by different optical sensors, usually producing large images
made by one or more picture elements for different specb-a1 channels (eg. see
Mivis and Modis systems); however each data sensor generally need several
processing steps in order to extract the required information among noisy or
misleading data.
Depending on the nature of monitoring system, the qualitative and quantitative
interpretation of remotely sensed signals requires an ad-hoc choice of computer
assisted steps for data pre-processing, feature extraction, classification. Data
processing is generally not on-line with the data acquisition process whether in
active or passive multispectral image sensors (independently from platform
type). Nevertheless real-time data processing aimed to provide the need
information is a must for special purpose sensors or in applications as motion
control, transportation systems, robotics, critical environments monitoring;
therefore low cost-definition smart sensors for in-situ, optical detection are
investigated.

369
370

2 Microelectronic Technologies and Sensor Data Processing

Two evolutionary paths are, in my opinion, emerging in the sensor fields due
to their integration with embedded processing units: smart sensors and System
on Chip (SOC); the former sensor type include procedures for calibration,
protocols for automatic interface and setting conforming to the standard
proposed in [ 11 or CMOS sensors with self-contained signal processing circuitry,
or VLSI with optically responsive pixel cells as proposed in [2, 31. The latter
sensor type include system integrated sensor with multiple specialized
processing units which perform in parallel different tasks on different data (i.e. a
MIMD paradigm); several SOC for radio communication, physical, chemical
and bio sensing have been recently proposed.
Recent microelectronic advances as the single electron transistors and CMOL
technology promise transistors implementation below the lOnm scale, giving
therefore a possible path to continue in the next decades the trend of Moore low
[4]; as showed therein, such technology is based on a nano-wire crossbar
architecture which straightforwardly permits the implementation of high
connectivity neuromorphic networks for large pattern classification based on
Hopfield model with Hebbian rule. Nevertheless large networks of Hopfield
networks were implemented on massively parallel computers [5] showing good
pattern learning and noisy auto-association but introducing some drawbacks as
spurius states and not yielding any categorized output [6]. Here we propose a
data processing approach, oriented to fast and robust pattern classification of
generic sensors.

2.1 Neural Techniques of Data Processing and Data Classification

We use an Artificial Neural Net (ANN) instead of a statistic approach to data
classification because their major advantage is the ability to generalize, in the
sense that a trained net could classify data from the same class as the learning
data set that it has never seen before. The implemented ANN is based on
robust statistical bases of non parametric techniques proposed by Parzen, which
overcome the Hopfield networks drawbacks; our network in fact can be
employed as a pattern classifier, and is composed by simple computing node.

2.2 Artificial Neural Network

An ANN is a layered network structure made by simple nodes (neurons, in a
real NN) processing data (the electric signals of firing neurons) yield by
unidirectional interconnection channels (the synapses). The basic structure of a
node and of a simple network is sketched in the Figure 1. A node is
 371

characterized by an activation function which describes the combining of input

signal, the output function which describes the response to the input stimuli, and
the re-configurable channel weights showed by numbers in the Figure 1. Since
the first ANN proposal of Mc Culloch and Pitts, a large number of network
topologies and learning methods for specific network training have been
proposed [7]. In this paper we have considered Probabilistic Neural Network
(PNN) for spectral signature classification; a PNN is a three layered network

to orher w t s

output finchon

actiratioii firncuan

which differs from Figure 1 as weights only connect two successive layers.

Fig. I Basic structure o f a Node and ANN

3 A Generalizing Classifier by Probabilistic Neural Network

Probabilistic Neural Network (PNN) are able to estimate the unknown
probability density functions for each category, and then use it for pattern
classification. This method works fine also with multi-modal density by varying
a window parameter h, as complex decision regions Rn with volume Vn can be
realized by superposition of simple window functions having the form
p(z)=(1/2~~2exp(-z2/2).If we assume that the region Rn of the n-th class is an
a-dimensional hypercube of volume Vn = (hn)a, it can be shown that the n-th
estimate of the density function pa($ for the samples x l . . x b , assumed
independent and identically distributed, is:

when hn goes to zero, p n (x) become a super-position of Dirac delta

functions centered at the sample points. Summarizing, the PNN input layer
normalize the input pattern, the inner layer computes pn (x) and the output layer
selects the class with maximum probability. Hardware implementation of PNN
372

have been suggested in [S] due to their relatively simple computation

requirements arising from not stringent space and time complexities.
As first step the PNN has been dimensioned, therefore the number of neurons
for each layer has been chosen. This is an application dependent choice, as the
number a of neurons in the input layer depends on the dimension of the input
units (the spectral channels), the number b of neurons in the intermediate layer
depends on the number of patterns on which the estimate is based, the number c
of neurons in output layer depends on the number of classes on which a generic
patterns can belong. Such application dependency has to be supported by a
general purpose re-configurable hardware architecture.

3.1 Preprocessing and Feature Estraction for Optical Fire Detection

We need to detect the spectral signature pattern with a non-imaging optical
sensor. The fire spectrum was early measured in [9] and different reference
flames are proposed recently for flame detection. We moreover suppose the
optical sensor has to detect the wider possible spectra for fires in natural
environment; note that the sunlight reflections can generate strong response in
VIS and infrared NIR to TIR bands, 1.e. the range 750-10.000nm. Therefore we
choose to dispose a setup with single pixel, large FOV multispectral sensors in
UV, and two NIR bands able to capture typical C02 and flambe spectral peaks.
These signals are preprocessed by low noise amplifiers and a converter to
translate the typical frequency response of the UV sensors based on gas
discharge tube. UV-NIR signals are transformed in digital numbers and are used
to compute also a NDVI index; these are the input for the successive supervised
classification by PNN.

3.2 Classifier Training

The ANN is trained with a reduced set of realized patterns, a subset among all
the possible patterns generable in special applications. In order to reach the best
generalization, the dataset can be divided in i) a training set the user has to issue
for the learning phase, during which the error is minimized, ii) a validation set
the user uses for estimate the classification performance of the ANN over
patterns not included in the training set, iii) a test set for performance evaluation
purposes of the ANN. Some time it could happen that a small training error does
not guarantee a low test error.

4 System Implementation and Results

The multispectral array sensor has been built by discrete optical sensor and
preprocessing board by Hamamatsu; the PNN has been implemented by Matlab
 373

for the preliminary simulated fire tests. A first hardware prototype is based on
Microchip PIC 12F675, a versatile MCU with 8-bit ADC and digital output.
Preliminary tests show acceptable sensibility and classification for in-situ fire
detection, and false alarm rate with butane and propane flames (2 cm diameter
size at 40 cm) is less than 10 % in sunlight conditions.

5 Conclusions
The proposed neural classifier for multispectral optical sensor has been tested
for a fire detection application by optical sensor, and its sensibility and false
alarm rate can be increased as fixther developments. The proposed data process
method can be usefully employed as a general tool for the implementation of
smart optical sensors, It has been shown its potential for an hardware
implementation, integrating optical sensors and the present CMOS
programmable technology or the future ULSI devices based on CMOL
technologies.

6 References
[l] IEEE 1451.3 Standard for a Smart Transducer Interface for Sensor and
Actuators-Digital Communication and Transducer Electronic Data Sheet
(TEDS) Form. Dis.Multi.Sys.
[2] H. Djahanshahi et alii, Neural-based Smart CMOS Sensor for On-Line
Pattern Classification Applications, ISCAS 96 International symposium
on circuit & systems, pp. 384-387, 1996.
[3] Chiu-Hung Cheng et alii, In the blink of a silicon eye, IEEE Circuits and
Devices Magazine, vol 17, 3, pp. 20-32, May 2001
[4] J.H. Lee and K.K. Likharev, CMol CrossNets as Pattern Classifier, in
IWANN 2005 LNCS35 12, pp.446-454, Springer Verlag 2005
[5] M. Mastroianni, M. Medugno, U. Scafuri, Realizzazione di reti neurali in
ambienti paralleli, in Applicazioni dellintelligenza artijkiale
nellingegneria, ed. Liguori (1993) 115- 124
[6] A. Iazzetta, M. Medugno, U. Scafuri, Autoassociative memory simulation
on a transputer network, in Eurosim 92 Simulation Congress, Capri I
29.9.1992-2.10.1992, ed. North Holland, (1993) 3 19-324
[7] J. Hertz, A. Krogh, R.G. Palmer, Introduction to the theory of neural
computation, Addison-Wesley Publishing Company
[8] D. F. Speckt, Probabilistic Neural Network, Neural Networks, 3( 1):109-
118,1990
[9] A. R. Jones, Flame Failures detection and Modern Boilers, J. Phys. E :
Sci. Instrum., 21,pp. 921-928, 1988.
 Design of a wireless sensor network for urban pollution
monitoring

G. Barillaro, P. Bruschi, R.G. Garroppo, S. Giordano, A. Molfese, F. Pieri

Dipartimento di Ingegneria dell 'Informazione, Universitd di Pisa, Italy
f.pieri($iet.unipi.it

Summary
Monitoring of air pollution is an essential clement for the management of urban areas, and
today it is mostly implemented using fixed stations. In this work, we examine the design of
an air pollution monitoring solution based on a wireless sensor network, and present some
of the issues that must be confronted by the designer. The presented implementation is
based on off-the-shelf networking hardware and custom developed sensors. The final system
will include a porous silicon based NO2 sensor, a solid-state, low power anemometer, and
temperature sensors. Specific attention is devoted, as is essential in a wireless sensor
network, to power consumption issues. It will be shown the power requirements of the
sensors often dwarf those of the network connectivity, so that smart sensor sampling
strategies arc as important as communication strategies for proper power management.

Introduction
Public monitoring of several different pollutants (NO,, CO, ozone, to name
the most common) and other environmental variables in urban area is an essential
element of the management of modern cities.
Present technical approaches to this task (i.e. sparse, mostly fixed, bulky
monitoring stations) lack the desirable pervasiveness and promptness of
response. The typical monitoring network of m e d i d l a r g e Italian cities is
composed by less that ten stations. As a result, accurate information on the spatial
distribution and time evolution of the pollutants, which is important to reduce the
disruptive impact of public countermeasures (as is the block of the circulation of
private vehicles), is not available. In this respect, the measurement of the wind
speed and direction is important to anticipate the pollution evolution. As the wind
is strongly influenced by the geometry of surrounding buildings [l], it should be
monitored on a very small spatial scale, and not by present-day very coarse
networks.
To handle these issues, Wireless Sensor Networks [2] based on many small,
autonomous nodes, implementing both the sensing and connectivity functions,
are promising. However, extreme care in planning the measurement and
communication strategies must be used. Frequently, this aspect is not considered

374
 375

Fig.I System architecture.

in depth in the literature, that often refers to abstract network models and is more
focused on the power optimisation of the communication part [3].

Architecture
For this project, commercially available hardware - Crossbow Technology
MICA2 comm~ication boards (the motes), developed at the UCB [4] -
(primarily used to implement the connectivity) will be integrated with custom-
developed state-of-the-art solid state sensors and their conditioning electronics.
Each network node is based on a h4PR400 mote, including an RI;transceiver (the
Chipcon CC 1000), working at 9 15MHz, and an Atmel microcontroller (the
Atmega 128L). Each mote can be interfaced with any of several sensor boards
(not used in this project), including sensors for temperature, humidity,
acceleration, etc. To allow the development of custom sensor boards, an
acquisition board (the MDA300), including several 12-bit ADCs and digital I/O
channels, is also available. The MDA300 will be used to interface our sensors
(described in the following section) with the network. Custom boards, carrying
the sensors, will be fabricated and connected to the MDA300 to be used for
sensor reading and signal conditioning. The boards are designed to be compact,
low-voltagellow-power circuits, with very low idle power consumption. The
whole node (mote + MDA300 + sensor board) is powered by two standard AA
batteries. A gateway node (the MIB510) is used for interconnection to a PC and
mote programming (fig. 1).
376

Fig. 2 Photos of the NO, PS-based sensors in a standard T O 8 p a c k a g e ~ e fand

~ , of the 4-chip
anemometer prototype, implemented on a standard PCB (right). The PCD diameter is 35 mm.

~e~sors
The monitored variables will be NOz concentration, wind speed and direction,
temperature, and humidity.
Porous silicon (PS) based sensors [ S ] (fig. 2, left) will be used to measure the
concentration of NOz. With respect to commercially available NO2 sensors, our
PS-based sensors have the great advantage of ambient temperature operation,
which implies reduced power consumption. The effect of humidity, which is a
known interfering variable for NO2 sensors, will be compensated with a humidity
sensor, built-in on the MDA300 board.
The anemometer, based on the differential temperature principle, is designed
with a standard VLSI technology, already used to fabricate a single sensor Row-
meter [ 6 ] . The first prototype [7] is based on a 4-chip assembly (fig. 2, right).
This prototype suffers of a high power requirements, as heating of the chips is
necessary for its operation. However a fully integrated, single chip version (under
design), should result in a very compact, low power device.
Finally, the temperature sensor will be a solid-state standard device based on
the AVh principle 191. Expected energy requirements for the various sensors
are summarized in table 1, where the acquisition energy is the total energy
required to obtain a valid reading from the sensor. While the prototype
anemometer has by far the largest energy requirements, in the final version the
NO2 will be the most important energy user, due to its (relatively) long
acquisition time.
 377

required acquisition acquisition

Sensor type
power (mW) time (s) energy (mJ)
Anemometer (bulk version) 500 35 17500
(Anemometer (integratedversion) 1 15 1 0.200 1 3 1
PS-based NOz sensor 4 10 40
Solid state temperature sensor 2.5 0.050 0.125
Hurniditv sensor (built-in) 2.5 4 10
Table I Power requirements of the sensors.

Power budget
Extreme care in planning the measurement and communication strategies
must be used, especially because the power consumption of the sensors can be a
significant part of the total power budget. For estimation of the power
requirements, the approach developed in [3] was followed.
The power source is a pair of AA 1.5 V batteries, for which a capacity of
2500 mAh was assumed. The node is supposed to transmit each time the sensors
are sampled, with a rate given in the table below. Between transmissions, the
node samples the communication channel at a fixed time interval (100 ms for the
data in the table), to check for possible transmissions, If a neighbouring node is
transmitting, the node receives the data and (possibly) retransmits it to other
nodes, depending on the communication strategy. The number of neighbour
nodes was set to five. Our data suggests that an important contribution to the
overall power consumption is given by the power requirements of the sensors.
Therefore, the measurement and communication approaches should be oriented
to the reduction of the sample rate of the sensors, implementing dynamic
strategies related to the current environmental conditions, Several approaches are
possible. A first one can be based on the correlation of pollutant concentration
with wind data. As it is known that strong winds reduce the risk of high pollutant
concentration, the sensor sample rate (and power consumption) can be reduced
when strong winds are present. Another possible approach is to increase the
sample rate of pollutant concentration sensors only when the warning threshold
concentration is approached. The effectivity of such approaches will be verified
with on-field deployment of the network.
378

sample rate battery life

Active sensors
(hr-) (days)
Bulk anemometer only 10 <7
Integrated anemometer only 10 1580
NOz sensor 10 1050
All (int. anemometer, N02, temp., humidity) 10 950
All (int. anemometer, N02, temp., humidity) 20 550

Acknowledgments
This work is financially supported by Fondazione Cassa di Risparmio di Pisa
under project RESPIRO. The sensors design is performed by Universita di Pisa,
while their fabrication is based in part on STMicroelectronics technologies, made
available in the framework of the joint Universita di Pisa-STMicroelectronics
R&D Center.

References
[ I ] J. W. D. Boddy, R. J. Smalley, N. S. Dixon, J. E. Tate, A. S. Tomlin, Atmos-
pheric Environment, 39 (2005), 3 147.
[2] W. Tsujita, A. Yoshino, H. Ishida, T. Moriizumi, Sensors & Actuators B,
B110 (2005), 304.
[3] J. Polastre, J. Hill, D. Culler, in Proceedings of the Second ACM Conference
on Embedded Networked Sensor Systems (SenSys 04), Baltimore, USA,
2004.
[4] http:llwww.xbow.com
[S] G. Barillaro, A. Diligenti, A. Nannini, L. M. Strambini, E.Comini, G. Sber-
veglieri, IEEE Sensors Journal, 6 (2006), 19.
[6] P. Bruschi, D. Navarrini, M. Piotto, Sensors and Actuators A, 110 (2004),
269.
[7] P. Bruschi, N. Nizza, M. Piotto, M. Schipani, XI Conferenza Annuale
dellAssociazione Italiana Sensori e Microsistemi (AISEM 2006), Lecce,
Italy, 2006.
[8] R. J. Baker, H. W. Li, D. E. Boyce, CMOS - Circuit Design, Layout and
Simulation, IEEE Press, New York (1998) 474.
 Application of Multi-sensor Miniaturized System for
Olive Oil Evaluation

M. Zuppa, A.M. Taurino, C. Distante, S. Capone, L. Francioso, D. Presicce,

P. Siciliano
Microelectronic and Microsystem Institute - CNR Lecce,
[email protected]

Summary
The detection of aroma volatile compounds emitted by extra virgin olive oils (EVOOs)
IS of key importance in the quality control of this product. Physicakhemical techniques
(GC, GCMS, HPLC) and sensory analysis (panel test) are the classical methods used for
this purpose, but they are expensive, time consuming, and do not allow on-line
measurements.
This work shows the performance of an electronic nose, based on pure and doped
S n 0 2 sol-gel thin films, used to analyze different single-cultivar extra-virgine olive oil
samples.

1. Introduction
A primary ingredient of the Mediterranean diet is Extra Virgin Olive Oil, which
is also one of the most important products of the Italian agricultural industry.
Extra Virgin Olive Oils (EVOO) taste and aroma are closely related both to some
non volatile compounds and to a number of volatile compounds.
Moreover, many types of EVOO are produced by mixing oils from different
cultivar, that is, from different variety of tree species coming from different
geographical origin area of cultivation. Each type of blend is characterized not
only by a different smell but also by different taste and color. The organoleptic
properties of oils also depend on the particular blend.
In the Italian region, the presence of typical varieties, the peculiar microclimatic
conditions and precise olive orchard management lead to the production of very
valuable olive oil with a distinctive taste. Seasonality, water availability and
temperatures clearly influencing the ripening process.
Another very important aspect is the ageing of olive oils. In fact, organoleptic
properties changes completely during the storage period of olive oils having a
strong degradation. Therefore, the detection of aroma volatile compounds
emitted by EVOOs is of key importance in the quality control of this product.
Physical-chemical techniques (GC,GCWS, HPLC) and sensory analysis (panel
test) are the classical methods used for this purpose, but they are expensive, time
consuming and do not allow on-line measurements.
Electronic Nose instrumentation has advanced rapidly during the past ten years
with a lot of successful applications in food and drinks industry applications.
This paper reports the performance of a metal oxide sol-gel thin films based
Electronic Nose, developed at the sensors laboratory of CNR-IMM Lecce,
sensing different singlecultivar EVOOs produced by Tuscan producers and

379
380

following the ageing of these olive oil samples. So that a technique of

classification based on rn-SOM neural network in combination of the
propabilistic neural network was applied to classify the EVOO aroma sensed by
the multisensor system.
In order to know the real headspace of olive oil samples sensed by
multisensor system, analytical technique like Headspace Solid Phase Micro
Extraction \ Gas Chromatography \ Mass Spectrometry (HS-SPME\GC\MS)
analysis was applied to the analysis of volatiles compounds in EVOOs samples.

2. Experimental set-up
Four samples of different Tuscan single-cultivar EVOOs, labeled Seggianese,
Scarlinese, Grappolo and Mugnai, were analyzed by means of two different
techniques. The complete set of measurement was repeated three times with an
interval of 30 days.
The multi-sensor array was composed by six different micro-sensors, which had
different sensing layer. They were pure and Ni, Os, Rh, Pd-doped tin oxide sol
gel thin films. For the preparation of the pure Sn02 sols we started from
anhydrous SnC14 as precursor. For the preparation of the doping solution a
prescribed amount of suitable solutions, with an atomic ratio X/N = 0.05 (X= Ni,
Os, Rh, Pd respectively), were added to the Sn02 sols. Details of the preparation
of the solution are reported elsewhere [ 11. The films were deposited, by means of
spin coating, onto A1203 substrates pre-arranged for cutting and obtaining (2x2)
mm2 single samples. After the spinning the sensitive layers were thermally
annealed at 500C. The alumina substrates were equipped, on the back side, with
TiPt meander as heating element. Moreover, on the front side, onto the sensing
layer, inter-digitated Ti/Au electrodes were deposited. Finally, the devices were
mounted onto TO39 socket.
During the operation, all the sensors were heated at the operating temperature,
which in this case was 250"C, by supplying a given voltage to the heating
element. The sensor responses towards the volatile compounds of the different
olive oil samples were carried out by applying a constant voltage of 2 V between
the electrodes and measuring the current flowing through the sensors by means
of an electrometer Keithley 65 17A type equipped with a multiplexer module.
Concerning the experimental set-up for the measurements, the baseline was
acquired in a dry air -nitrogen atmosphere in a continuous total flow of 100 sccm
(standard cubic centimeter per minute) (50 sccm air and 50 sccm nitrogen), while
for the measurement 10 ml of sample in a 20 ml vial kept at a temperature of
21"C, was stripped by means of a deviation of the only 50 sccm nitrogen flow
for 20 min, maintaining the other 50 sccm of dry air constant. In this way, the
volatile compounds were directly transferred by the carrier gas into the sensor
chamber. All fluxes were controlled by means of mass flow meters and a mass
flow controller (mod. MKS 647B).
 381

All the process was controlled by a PC by means of National Instruments

software.
In the case of HS-SPME\GCWlS analysis, SPME 50pm CAR (Carboxen)-PDMS
(polydimethylsiloxane) (black code, Supelco) fibers were used and exposed to
lOml of oil sample in 20 ml gastight vials at 21OC. After this, fiber was desorbed
in a splitkplitless injector, equipped with deactivated SPME glass inserts, and
analyses were carried out on a 30m x 250 pm ID x 0.25 pm HP INNOVAX
polyetilenglicole column. A GC system HP 6890 Series, Agilent Technologies,
was coupled with HP 5973 mass selective Detector, Agilent Technologies.
Injection was split 5:l with 5 min relay time. GC conditions were: 40C for 5
min, 150C at 4"C/min, then 250C at 15"C/min; injector temperature was
250"C, helium flow was lml/min. Transfer line was held at 260C. Spectra were
produced in the electron impact mode at 70 eV. Compounds were identified by
comparison with NIST reference spectra. Mass range was 30-350 amu, solvent
delay time was 4 min, threshold 150 and scan rate 4,45 scan s-I.

3. Methods and Results

The Principal Component Analysis was used for the initial examination of the
whole dataset. The figure 1 shows the score plot for the first two principal
components (PCs), reflecting a not clear clustering in particular for data coming
from the third period of maturing. A technique of classification based on m-SOM
neural network in combination of the probabilistic neural network (PNN) (Figure
2) was applied to classify the EVOO aroma sensed by the multisensor system.
The sensor responses were processed by means of the Discrete Wavelet
Transform and subsequently Principal Components Analysis to extract useful
features [l]. The next step of data analysis was to discriminate different single-
cultivar olive oil samples by applying a neural architecture (m-SOM) of m self-
organizing maps, where m was the number of single cultivar olive oil groups to
be recognized. The processing was mainly made in two phases: training and
testing. In the training phase, each map was trained with observations belonging
to the same single-cultivar olive oil group. Once the maps were self-organized, a
refining process was performed by using Learning Vector Quantisation (LVQ)
algorithm to reduce the high uncertainty accumulated at the borders of two or
more maps. In the testing phase the unknown measurement vector x was
presented to the network. The Euclidean distance measure was computed over all
neurons of the maps, and the winner map (the one with the nearest neuron or
with minimum distance) was considered as the estimated recognized olive oil
class [2].
Once the selection of single cultivar olive oil was performed, the single cultivar
olive oil samples of different ageing period were discriminated. The codebook
vectors of each single map, which were closest to the input vectors, fed up the
probabilistic neural network to determinate the period of storage of the oil
382

sample [3]. In order to test the capability of the classification technique a cross-
validation method was performed by selecting randomly a certain number of
measurement vectors from whole set as training data of the mSOM-PNN
classifier. This procedure was repeated for different times and the average error
rate was 8%.

4. Conclusions
In this work four different single-cultivar Tuscan Extra-Virgin Olive Oils were
analyzed by means of Electronic Nose. By means of Electronic Nose it was
possible to separate among clusters of different EVOOs and of different cultivar,
while with the HS-SPME\GC\MS we obtained the chemical map of the different
samples.
Even if the PCA clustering results are not completely satisfactory, the mSOM-
PNN classifier allows a good classification The used technique required a small
train data set and at the same time it showed a good performance in terms of
errors committed in recognizing the given data set in comparison with others
supervised technique of classification as k or PNN.

W I

Fig. 1 PCA score plot of whole Electronic Nose data set

 383

Fig. 2 mSOM-PNN classifier; input vector x fed up mSom network to assign the single-cultivar olive
oil cluss . The codebook vector of the winner map fed up the PNN algorithm to assign the period of
maturing.

efgre~cgs
[I] Distante C. , Leo M., Siciliano P., Persaud K.C. On the study of feature
extraction methods for an electronic nose Sensors and Actuators B 87 (2002)
274-288
[2] C. Di Natale, A. Macagnano, A. DAmico, F. Davide, Electronic nose
modelling and data analysis by seEf organizing neural network,
Meas. Sci. Technol. 8 (1997) 1236-1243.
[3] Zheng Hai, Jun Wang Electronic nose and data analysis for detection of
maize oil adulteration in sesame oil Sensors and Actuators B 119 (2006)
449-455

The authors thanks prof. Cimato of CNR - Tree and timber Institute for olive
oil samples.
 LOW VOLTAGE LOW POWER IMPROVED CCII BASED
INTERFACE FOR MEASURE AND HEATING CONTROL
OF RESISTIVE SENSORS

Giuseppe Ferri, Vincenzo Stornelli, Andrea De Marcellis, Fabrizio Mancini

Dipartimento di Ingegneria Elettrica e dell 'lnjormazione, Universitd.
dell'Aquila, Monteluco di Roio, 67100 L 'Aqziila,Italy, fcrri(u~in~.univa4.il

Summary
In this papcr, wc prcscnt a low-voltagc (*0.75V) low-power (2OOyW) currcnt-modc bascd
uitcgratcd inlcrftacc for thc measure and hcating control of rcsistivc gas scnsors. It is formed
by two rcsistances, two improved CCIIs, which have bccn dcvelopcd to achicvc rcduccd
parasitic coinponetits at its tcrininals, and a logic control block which both cbtitiiatcs thc
rcsistancc sensor value (through a rcsistancc-to-voltagc convcrsion) and controls the
tcmpcraturc of thc wholc system.

1 Introduction
The interfacing of resistive sensors, a s sensitive elements, with a suitable
integrated circuit is a fundamental characteristic in smart sensors. CMOS
technology is widely used, because it allows to match the scduction of costs of
the silicon technology with the possibility of designing new low voltage and low
power interfaces, typically supplied by a battery. Recently, also the current-mode
approach, which considers the current as reference signal, has been applied in
sensor interface [ 1-21. Second generation current convcyor (CCII) is the current-
mode basic block. In this paper, we present a low-voltage (50.75V) low-power
(200pW) current-mode based integrated interface, using two improved CCIls,
which have been developed to achieve reduced parasitic components at their
terminals.

2 Improved CClI topology

Second generation current conveyors (CCIIs) are current-mode basic blocks
utilized in numerous applications, both in linear and nonlincar contexts, which
sometimes can excellently substitute the traditional operational amplifier [ M I .
In fig. I the block scheme of a CCII device is reported: If a voltage is applicd at
Y node, an equal voltage will be obtained at X node while the current flowing
into X node is either equal or opposite to the current tlowing into Z node.

384
 385

VY i
Figure I . CCII sclicniufic /dock.

Ideal CCII behaviour is the following:

where Z and Y nodes show infinite impedances and X node shows zero
impedance. Positive (CCII+) and negative (CCII-) current conveyors are
respcclively obtained for IZ=Ix and 17= -Ix. Non ideal CCIls present the
following relations:

where typically a and p are very close to 1 . Generally, a unity voltage transfer
function Vx/Vy is ensured by implementing a differential transistor input pair,
when large transistors output resistances are required in order to obtain an a
parameter dependent only on the transconductance ratio of the input pair itself.
Concerning the current transfer function, unfortunately it is not independent
from the load connected to Z node. If this load impedance is negligible with
respect to the transistor output resistances, p parameter is very close to its ideal
unitary value. In summary, non ideality problems are related to the parasitic
impedances at CCll lerminals that have to be necessarily laken inlo account in a
large number o f low voltage low power applications. For this reason we have
developed a novel CCI I topology which presents reduced parasitic components
at X and Z nodes (11. Fig. 2 shows the schematic of the improved CCII at
transistor level while in fig. 3 its layout is depicted.
386

Figure 2: Improved CCII schematic at transistor level.

Figure 3: Inproved CCII layou/.

The parasitic impedance at X node, that is required lo be low, is inversely

proportional to the input transistors g,. A small input resistance requires large
values o f g,,, so a trade-off between X node parasitic resistance and power
consLiinption has made. Concerning the Z parasitic impedance, it has been
ensured to be high by the transistor output resistances at Z-node. The proposed
circuit works at a low supply voltage, rt0.75V, and has a very low power
consumption, about 200yW. In Tab. I the main CCil characteristics are
reported.
 387

Value
It 0.75V

3dB Bandwidth 16 MHz

Dynamic Range - 1Vh0.9V
Biasing Currents 10 pA
Voltage Gain (alfa) I .oo
Current Gain (beta) 1 .OO (Rx=Rz=lOKSZ)
X Parasitic Resistance 11.4 Q
Z Parasitic Resistance 7.2 Mi2
Layout area 0.13 mm'

3 CCII-Based Interface
Fig. 4 shows the considered CCII-based interface where the sensor is modelled
by resistance R,,,,,.

The output vollage Y,,,,is linearly proportional to R,,,,, variations. In fact,

considering ideal CCIIs, we can write:

The first lemi of ( 5 ) is linearly proportional to sensor variation and, through the
value of Rcon,,.,can be used to maximize circuit sensitivity, while the second one
388

), fixed by a suitable choice of K.,),lt,.,

gives a constant current (I,,,,= Vco,,t~Rc,,,lt~
through the logic block which is also used for heating the sensor itself. In this
way, the problem of the system temperature control and R,,,,, estimation has been
here reduced to the measure of the output voltage dropped across the sensor
resistance. This task is made easy thanks to the second CCIl that acts as a
voltage buffer decoupling the logic block from the system. The delivered current
must be such that the temperature has to remain constant. The suitable control
logic evaluates the sensor resistance through a resistance to voltage conversion
and generates a feedback signal, which maintains the sensor at the desired
temperature. Designing improved CCli blocks avoids a bad estimation for R,s,,,,
determined by temperature and power supply variations of the whole system.
Fig. 5 shows the heating current variation at different temperatures for R,,,,,
value ranging between IOKQ and 100KQ. The heating current, I,,,, remains
constant enough in the considered range, even if, for high temperature values,
the circuit does not maintain this behaviour. Therefore, in the future, a particular
attention has to be made to achieve an improved interface, which allows to
reduce the sensitivity to the temperature variations of the whole device.

-10
-20
40
-60
-80

10 loo h e n s [Kn]

Figure 5. I,,,, ctitretit vs. R,,,,, ul diferent r e , ~ the bottom, I 0,200,40,600,800CJ

t e i t y ~ e r o i ~ ~froin

References
1. G.Fcrri, V.Stornclli, M. Fragnoli, An intcgratcd improvcd CCIl topology for rcsistivc scnsor
applications, Analog lntcgratcd Circuits and Signal Proccssing, acccptcd for publication, 2006.
2. G.Pcrri, N. Gucrrini, S. Dcl Rc, F. Mancini Currcnt Mode Bascd Intcgratcd Gas Sensor
Intcrfaccs, Scnsors And Microaystcms, Proc. AISEM 2004, pp. 424-429.
3. K.C.Smith, A.Scdra, The currcnt-conveyor- a ncw circuit building block, IEEE Proc 56, 1968,
pp 1386-1369.
4. A. Scdra. K. Smith, A Sccond-Gcncration Currcnt Convcyor and Its Applications, IEEE Trans.
Circuit Thcory, 17, 1970, pp. 132-133.
5. G Fcrri, N Gucrrini, Low voltagc low powcr CMOS current conveyors, Kluwer Academic
Puhlishcr, Boston.2003.226 pp , ISBN 1-4020-7486-7
6. R.Mita, G Palumbo, S.Pcnnisi, CMOS CCIIt With High Currcnt -Driving Capability, IEEE
Trans. on Circuits and Systcms 11, Vol 50. Issuc 4, April 2003, pp 187 190.
This page intentionally left blank