EFFECT OF THE NATURE OF BOTH TOP AND BOTTOM ELETRODES ON THE DIELECTRIC AND FERROELECTRIC PROPERTIES OF PZT THIN

FILM CAPACITORS
Nossikpendou Sama *a), Rachid Bouregba +, Caroline Soyer *, Denis Remiens *
* IEMN, DOAE, MIMM group, Cite Scientifique, 59655 Villeneuve d Ascq Cedex, France.
+ Laboratoire CRISMAT-ENSICAEN, CNRS UMR 6508,

Boulevard du Marchal Juin, 14050 Caen Cedex France.

Outline

 Introduction  Films growth  Model of capacitor and experimental work  Results  Summary and conclusions

Introduction

Many studies have been made on PZT thin film capacitors to investigate the degradation of dielectric and ferroelectric properties when the thickness is reduced. reduced. The aim of this study is to show that this degradation results from interface effects. This is achieved by using effects. electrodes with different nature: metal (Pt) and oxide nature: (LaNiO3). This study also aims at discriminating which of the top or degradation. the bottom interface mainly contributes to this degradation.

Introduction
To this end, dielectric and ferroelectric measurements were carried out on PZT54/46 films in four different structures, PZT54/ each one with four different thicknesses (200, 400, 600 and 200, 400, 800 nm). nm).
Pt LNO Pt LNO

PZT
Pt

PZT
Pt

PZT
LNO

PZT
LNO

substrate

substrate

substrate

substrate

Films growth

Bottom electrodes are Pt/TiOx/SiO2/Si and LNO/SiO2/Si deposited by sputtering, Pt/TiOx electrode is 120 nm thick and LNO one, 250 nm. nm. Then PZT is deposited on them by sputtering and annealed at 625 625C. Finally, Pt and LNO top electrodes, with respectively 120 nm and 200 nm thicknesses, are deposited on PZT by sputtering. sputtering. Four different kind of structures were grown, each one with 4 different thicknesses : 200, 400, 600 and 800 nm (a total of 200, 400, 16 structures available). available).

Films growth

XRD spectra of PZT/Pt

XRD spectra of PZT/LNO

PZT/Pt shows a <110> preferential orientation while PZT/LNO has <100> 110> 100> orientation. orientation. In both cases the intensity of the peaks increases with the thickness of the film, without changing its texture. texture.

PZT/Pt

PZT/LNO

PZT/Pt has larger grains than PZT/LNO

Experimental work xperimental

The PZT capacitors are modeled as stacked structure with a bulk ferroelectric layer sandwiched between two nonferroelectric space charge layer at each surface [1]. This model enables to determine some data related to the interfaces and the bulk. bulk. Based on this model, dielectric and ferroelectric measurements were performed in order to characterize the properties of the films with different electrodes and with various thicknesses. thicknesses.
[1] R. Bouregba, G. Le Rhun, G. Poullain, and G. Leclerc, J. Appl. Phys. 99, 034102 (2006).

Results
Dielectric measurements ( )
d (nm) Pt-PZTPt-PZTPt 422 476 606 709 LNO-PZTLNO-PZTPt(bottom) 717 755 797 928 Pt-PZTPt-PZTLNO(bottom) 445 651 803 815 LNO-PZTLNO-PZTLNO 1028 1078 1087 1185

200 400 600 800

always increases with the thickness of the PZT film but this trend is less pronounced for the LNO/PZT/LNO structure. substitution of Pt by LNO as top electrode systematically leads to a significant improvement of the dielectric constant. Such substitution appears less drastic when performed at the bottom electrode. For a same top electrode, (100) oriented PZT on LNO has better than (110) 110) oriented PZT on Pt. Finally, for a given thickness, Pt. is higher for the LNO/PZT/LNO/SiO2/Si structure and weaker for that Pt/PZT/Pt/TiOx/SiO2/Si one.

Results
1/Cmeas (d) for all the structures
In the framework of the stacked capacitor model with depressed interface layers : 1/Cmeas = 1/Cf + 1/Ci = (d/ 0/ f/A) + 1/Ci , Ci is the capacitance related to both interfaces. interfaces.[remarque : t es pas oblig de l crire si tu le dis!]

Results
Determination of bulk permittivity and interface capacitance for all the structures
PtPt-Pt Slope 1/
0

LNO-PZTLNO-PZTPt(Bottom) 6.10-6 1070 0,91

Pt-PZTPt-PZTLNOLNO-LNO LNO(Bottom) 4.10-6 1600 0,333 4.10-6 1600 3,333

* S [(pF.nm)-1] 6.10-6
f

1070 0,435

Ci (nF)=1/y(0)

Confirmation of the first observations : - Strong contribution of the Pt top electrode in the degradation of the dielectric properties (materialized by the smallest interface capacitance) - For a given bottom electrode, the same value of f is found whatever the nature of the top electrode : consistency of the model as f is mainly determined by the crystallographic orientation of the film - PZT on LNO exhibits larger f than PZT on Pt : consistent with the fact that (100) 100) orientation is preferable to achieve better dielectric properties - Moreover, values of similar composition (#1000 1600) f (#1000 and 1600) are consistent with those of bulk PZT for a

Results
Ferroelectric loops as a function of thickness for a given electrode configuration
All the ferroelectric loops have been measured for a constant maximum polarisation (30 C/cm2) in accordance with the model.

 PZT on Pt

Results  PZT on LNO

Results
Comparison of the ferroelectric loops for a given thickness

Results
Main prediction of the model The model gives the following relation between the electric field applied to the whole structure, the electric field in the ferroelectric layer and the electric displacement :

AD (  Vbi ) Ci E ! Ef  d
Case of special interest : coercive field Ec obtained when D=0 D=0

Vbi Ec ! E fc  d
The model predicts a linear variation of Ec vs (1/d) which the slope and the ordinate origin give interface potential Vbi and the coercive field in the bulk ferroelectric layer Efc, respectively

Results
Coercive fields Ec (1/d)

 PZT on Pt

Results  PZT on LNO

Results
PtPt-Pt
Efc+ (kV/cm)=y(0) Efc- (kV/cm)=y(0) LNO-PZTLNO-PZTPt(Bottom) Pt-PZTPt-PZTLNO(Bottom)

LNOLNO-LNO 23,1 -32,2

27,37 -17,3

25,65 -28,9

25,15 -19,6 19,

The coercive field of the bulk Efc deducted from the curves before is about 25 kV/cm and not very different for PZT/Pt and PZT/LNO. This value is coherent with the PZT/LNO. massive PZT one. one.

PtPt-Pt Vbi+ (V) VbiVbi- (V) 3,87 3,68

LNO-PZTLNO-PZTPt(Bottom)

Pt-PZTPt-PZTLNO(Bottom)

LNOLNO-LNO 0,21 0,14

0,031 0,34

3,43 3,18

Vbi is much lower when LNO replaces Pt as top electrode.

Summary and conclusions

1 - For a given bottom electrode, significantly increased and Ec significantly decreased when LNO replaces Pt as top electrode 2 - Thickness dependence of much less significant with LNO as top electrode but not totally absent (finite value of interface capacitance) 3 - Thickness dependence of Ec almost non-existent with LNO as top nonelectrode 4 - The nature of the bottom electrode (LNO or Pt) has only a minor influence in terms of thickness effect. Bottom electrode seems not to degrade the effect. quality of the interface, though it strongly influences the permittivity in terms of crystallographic orientation (better with <100> oriented PZT than with 100> <110> oriented PZT) 110>

Summary and conclusions

1 - General consideration : according to this model the thickness dependence of may be interpreted as due to presence of finite interface capacitance Ci, but the thickness dependence of Ec requires Ci to be charged (potential barrier Vbi) ! Hence the only notion of passive or dead layer model is not sufficient 2 - LNO top electrode Interface capacitance Ci compared to Pt

and interface potential barrier Vbi

3 - Oxyde electrode only at top surface of the film Elimination of thickness dependence of both dielectric and ferroelectric properties 4 - Measurements are in progress to investigate the effect of Pt and LNO as top and bottom electrodes upon the fatigue mechanism