Acoustically Levitated Droplets

A Contactless Sampling Method

for Fluorescence Studies
JORK LEITERER,a MARKUS GRABOLLE,a KNUT RURACK,a UTE RESCH-GENGER,a
JAN ZIEGLER,b THOMAS NANN,b AND ULRICH PANNEa
a
Federal Institute for Materials Research and Testing (BAM), Berlin, Germany
b
University of East Anglia, School of Chemical Sciences, Norwich, United Kingdom

Acoustic levitation is used as a new tool to study concentration-dependent processes in fluorescence

spectroscopy. With this technique, small amounts of liquid and solid samples can be measured
without the need for sample supports or containers, which often limits signal acquisition and
can even alter sample properties due to interactions with the support material. We demonstrate
that, because of the small sample volume, fluorescence measurements at high concentrations
of an organic dye are possible without the limitation of inner-filter effects, which hamper such
experiments in conventional, cuvette-based measurements. Furthermore, we show that acoustic
levitation of liquid samples provides an experimentally simple way to study distance-dependent
fluorescence modulations in semiconductor nanocrystals. The evaporation of the solvent during
levitation leads to a continuous increase of solute concentration and can easily be monitored by
laser-induced fluorescence.

Key words: acoustic levitation; dyes; energy transfer; fluorescence; quantum dots; nanocrystals;
ultrasonic trap

Introduction condensed (flowing) phases by means of a stationary

ultrasonic field. For that purpose, a concave reflector is
Many of today’s analytical problems are charac- adjusted concentrically at a distance of several multiple
terized by small sample volumes and can be solved half-wavelengths of the so-called sonotrode that emits
only through a corresponding miniaturization of the an ultrasonic wave. As a result of multiple reflections
analytical instrumentation.1,2 In principle, analytical between sonotrode and reflector, a standing wave is
methods that are based on spectroscopic techniques generated (FIG. 1). In all the equally spaced nodes of
are sensitive enough for the analysis of small sample the sound’s pressure wave, solid or liquid samples with
amounts, but handling of small volumes is inherently effective diameters of less than half a wavelength can
difficult because of desorption of contaminants from be levitated and held on a constant position by the in-
and adsorption of analytes at the walls of containers. terplay of axial radiation pressure and radial Bernoulli
Furthermore, miniaturized systems that rely, for in- stress. The latter effect is caused by an increase in ve-
stance, on analytical chips often suffer from optical in- locity of air streams and a corresponding decrease in
terferences at chip surfaces, and for micro total analysis pressure when the droplet moves away from the levi-
systems, analyte adsorption at interfaces can hamper tation axis. Levitation of samples with densities up to
the reliability of the detected signal.3 the heaviest solid (iridium, ρ = 22.6 g cm−3 ) or liquid
Acoustic levitation is a powerful tool to circumvent element (mercury, ρ = 13.6 g cm−3 ) are possible with
such drawbacks.4–6 This method allows the contact- this technique.7 An advantage of acoustic levitation
free handling of solid and liquid samples in a gaseous is that the samples do not have to meet special re-
environment as well as the trapping of solid objects in quirements, such as those for magnetic or electrostatic
levitation techniques. Moreover, when working with
liquids in a gaseous environment, one can directly in-
ject liquid samples into the nodes of the standing sound
Address for correspondence: Jork Leiterer, Dipl.-Phys., Div. I.3, Federal
wave with a pipette. The volume of these droplets is
Institute for Materials Research and Testing (BAM), Richard-Willstätter-
Str. 11, 12489 Berlin-Adlershof, Germany. Fax: +49(30)-8104-1137. typically 5 nL−5 μL, corresponding to a diameter of
[email protected] 0.2–2 mm.
Ann. N.Y. Acad. Sci. 1130: 78–84 (2008). 
C 2008 New York Academy of Sciences.
doi: 10.1196/annals.1430.039 78
Leiterer et al.: Ultrasonic Trap for Sampling in Fluorometry 79

single-molecule level,24 mainly laser-induced fluores-

cence techniques are used. Besides outstanding sen-
sitivity, laser-induced fluorescence is also the method
of choice to obtain detailed information on morpho-
logical and general physical (e.g., temperature) pa-
rameters of the levitated liquid sample during anal-
ysis.22,23,25 Because evaporation of the solvent during
levitation gradually decreases the droplet’s volume, ul-
FIGURE 1. Principle of operation and instrumental trasonic traps as sample compartments offer the pos-
setup of the acoustic levitator used in our studies. The trap sibility of conveniently monitoring chemical and phys-
is operated at a frequency of 58 kHz and a sound level
ical processes as a function of concentration, such as
of approximately 160 dB compensates for gravity. Labeled
beads of polystyrene hover in the central nodes of the stand- aggregation phenomena, in a dynamic and continu-
ing acoustic wave to demonstrate the levitation of multiple ous fashion. Acoustic sample levitation thus also en-
samples. (In color in Annals online.) ables the investigation of distance-dependent effects,
such as energy transfer or other processes that lead
On the basis of these features, acoustic levitation has to fluorescence quenching, without interference of un-
been used in various applications in the past decade. desired background signal due to scattered excitation
For instance, the classic strategy of handling objects in light or autofluorescence from the sample container
the gas phase has been used for comparatively simple itself.
purposes, such as analyte enrichment8 and pretreat- Our interest in coupling fluorescence spectroscopy
ment9 of liquid samples, as well as for sophisticated and acoustic levitation is based primarily on the study
studies of biochemical reactions10 in or the molecular of concentration-dependent phenomena in samples
structure of living cells.11 Studies of trapped particulate containing different types of fluorescent labels. For
objects in (mainly capillary) flow systems have recently this purpose, we investigated the changes in fluores-
been realized for the detection of proteins,12 the in vitro cence intensity, band shape, and maximum that take
sensing of toxicants,13 and the manipulation of cells.14 place in transparent liquid solutions of a fluorescent
Furthermore, the combination of X-ray diffraction dye (fluorescein 27 in 0.1 mol/L NaOH) and core–
or scattering techniques and ultrasonic levitation per- shell nanocrystals (CdSe/ZnS quantum dots) that both
mits gaining deeper insight into solidification and show rather resonant fluorescence bands. Fluorescein
crystallization processes. The evaporation of the sol- is one of the most frequently used traditional fluo-
vent during levitation gradually decreases the droplet’s rescent labels, and quantum dots are strongly gain-
volume, yet the contactless nature of the method ing in importance in imaging applications that can
avoids nucleation at foreign templates, such as surfaces rely on labels that are significantly larger than single
of the sample compartment. Interesting results have molecules.26
thus been obtained for inorganic salts,15 proteins,16,17 The instrumental setup for fluorescence measure-
and alloys.18 ments in combination with the acoustic levitator used
by us here is relatively simple. The levitated droplet is
excited with a laser beam of approximately 5-mm di-
Acoustic Levitation and Fluorometry ameter, thus illuminating more than the volume of the
droplet with homogenously distributed intensity. The
Optical spectroscopic techniques are particularly fluorescence signal is collected in right-angle geometry
suited to be used with acoustic traps to mon- and, after collimation by a lens system into an optical
itor small levitated objects. Accordingly, most of fiber, is detected by a spectrometer. To know the actual
the studies that use acoustic levitation in analytical concentration of the solute in the evaporating droplet
chemistry are based on optical detection. Other than at a certain stage of the experiment, one must measure
ultraviolet/visible spectrophotometry19,20 and Raman the droplet volume continuously during evaporation.
spectroscopy,11,20,21 luminescence spectroscopy is es- This goal is accomplished by monitoring the shadow of
pecially a tool of choice here. For luminescence spec- the droplet that is produced by a telecentrical infrared
troscopy, organic dyes, such as pyrene,10 fluorescein,12 beam with a camera (PICTOR VC2068/E; FiberVi-
or rhodamine B,14 as well as fluorescent beads12,13 or sion GmbH, Würselen, Germany). Because the size
rare-earth emitters,22,23 served as the reporter species. of the shadow of the droplet is proportional to its
To fully exploit the potential of fluorometry in this cross-sectional area, and under the assumption that
field, that is, to be able to reach detection limits at the the droplet possesses rotational symmetry around the
80 Annals of the New York Academy of Sciences

FIGURE 2. Normalized absorption and fluorescence spectra of fluorescein 27 in 0.1 mol/L NaOH at
10−6 mol/L (left ) and fluorescence spectra of the dye as a function of concentration (right ) as measured in
a 10-mm cuvette (excitation at 488 nm). For better illustration, the spectra in the right part are normalized
at 550 nm. (In color in Annals online.)

axis of levitation, the volume can be calculated and for 13 nm to the red, and a decrease of the fluores-
directly yields the increase in concentration as the ra- cence signal is observed. In the right part of FIGURE 2,
tio of the initial to the actual droplet volume. For the the fluorescence spectra are normalized at 550 nm,
investigation of distance-dependent phenomena, the where the emission is not disturbed by reabsorption
mean interparticle
√ distance D can be approximated because fluorescein does not absorb at wavelengths
by D = 1/ 3 n , where n is the number of particles per longer than 540 nm (FIG. 2, left).
volume unit. FIGURE 3 compares the results of the concentration-
dependent measurements in a cuvette with the cor-
responding measurement on a levitated droplet.
Concentration-dependent Fluorescence
Whereas reabsorption effects at concentrations above
of an Organic Dye in Solution 10−6 mol/L lead to a pronounced displacement of the
emission maximum for the cuvette-based experiments,
Fluorescein derivatives are among the most widely
no wavelength shifts are recognized upon increasing
used fluorescent dyes because they show favorably high
the concentration of the dye up to 10−4 mol/L in the
molar absorption coefficient and fluorescence quan-
levitated droplet. Thus, measurements at such high dye
tum yields. However, their absorption and fluores-
concentrations are not interfered with by the inner-
cence behaviors critically depend on pH,27 and they
filter effect when ultrasonic traps are used for the han-
are known to adsorb to surfaces, potentially under-
dling of a sample.
going dimerization.28 Thus, to be able to study the
influence of the sample handling technique on the
concentration-dependent emission signal of an organic Concentration-dependent
dye with a small Stokes shift, one must account for Luminescence of Nanocrystals
both potential drawbacks. The pH problem can be in Solution
circumvented when controlled basic conditions (e.g.,
0.1 mol/L NaOH) are used to hold fluorescein in its Semiconductor nanocrystals are roughly spherical
monomeric, dianionic state so that self-aggregation is particles with a diameter of about 2–10 nm that typ-
avoided in liquid aqueous solution at submillimolar ically consist of a luminescent core of a low-bandgap
concentrations. The second drawback is avoided sim- material, a protection shell of a few monolayers of a
ply by the nature of the contactless sampling technique high-bandgap material, and organic ligands attached
used here. to the outer surface, which render the particles sol-
FIGURE 2 depicts the spectroscopic characteristics uble. The nanocrystals generally show a broad ab-
of a dilute solution of fluorescein as well as the dra- sorption spectrum and a narrow, symmetric, and vir-
matic effect of reabsorption on the fluorescence spec- tually Gauss-shaped emission spectrum the width of
trum when the concentration of the dye is increased which depends on the width of the particle size distri-
in a conventional fluorometer-based experiment using bution (FIG. 4). The spectral position of the emission
10-mm cuvettes. Upon increasing the concentration and the onset of the absorption bands can easily be
from 10−9 to 10−4 mol/L, the shape of the spec- tuned by the (preparation controlled) size of the parti-
trum changes significantly, with the maximum shifting cles. Both fluorescence and absorption bands shift to
Leiterer et al.: Ultrasonic Trap for Sampling in Fluorometry 81

FIGURE 3. Fluorescence spectra obtained for fluorescein as a function of concentration using acoustic levitation (left ).
Position of the emission maximum of fluorescein as a function of dye concentration (right ) as measured in conventional
cuvettes (squares) and levitated droplets (circles). Inset: levitated droplet of fluorescein in NaOH upon irradiation with a
blue laser (λ = 488 nm); the primary beam is scattered mainly at the equator near the surface. (In color in Annals online.)

cuvette can also complicate the measurement. Layer-

based experiments, on the other hand, offer small-
sample films, thus avoiding inner-filter effects, but can
be influenced by scattering and interference at the sup-
porting surface. An acoustic levitator seems so to be
better suited as a contactless sample holder.

Energy Transfer between Nanocrystals

The particles used here are CdSe/ZnS nanocrystals

capped with a mixture of Tri-n-octylphosphine oxide
(TOPO), Tri-n-octyl phosphine (TOP), and Hexadecy-
FIGURE 4. Normalized optical spectra of quantum dots lamine (HDA) organic ligands.31 FIGURE 4 shows the
in solution. The absorption decreases steadily from 400 to
absorption and emission spectra in chloroform. The
640 nm, with the intensity at only one-thousandth of its maxi-
mal value. The narrow fluorescence spectrum is found in the environmental temperature was 298 K at a median
range of 550–650 nm, with a maximum at 509 nm. Inset: humidity of 50%, and the particles investigated here
chemical architecture of the stabilized nanocrystals. possess a fluorescence quantum yield of approximately

longer wavelengths with increasing particle diameter.

Detailed descriptions of semiconductor nanocrystals
can be found in the literature.29,30
Concentration-dependent studies of the lumines-
cence of an ensemble of quantum dots can reveal
information on distance-dependent photophysical in-
teractions such as energy transfer when particles
approach in a shrinking volume. Using ultrasoni-
cally trapped samples for such purposes has advan-
tages compared with cuvette- and layer-based exper-
iments. In a cuvette, at concentrations greater than
10−6 mol/L, reabsorption effects can modulate the re-
sults in a similar way as described earlier for the organic FIGURE 5. Fluorescence spectra of CdSe/ZnS nanocry-
dye. Moreover, because of the specific chemical nature stals in CHCl 3 as a function of levitation time (excitation at
of the nanoparticles, adsorption at the walls of the 436 nm).
82 Annals of the New York Academy of Sciences

ergy transfer. The net effect is a redshift of the emission

of the nanocrystal ensemble upon prolonged levitation
and concentration of the sample.
As shown in FIGURE 4, the overlap of the absorption
and emission spectra seems sufficient to allow for ef-
fective energy transfer between the quantum dots. The
calculated overlap integral of 9.3 × 1015 mol L−1 cm−1
nm4 derived from the spectra at dilute concentrations
(see Ref. 32 for formula) is more than 10 times higher
than that of widely used organic dye donor–acceptor
pairs, such as 4.9 × 1014 mol L−1 cm−1 nm4 for the
common FITC–TRITC (fluorescein isothiocyanate–
tetramethylrhodamine isothiocyanate).33 These favor-
FIGURE 6. Fluorescence intensity and emission maxi-
able FRET features are based mainly on the high mo-
mum of CdSe/ZnS nanocrystals in CHCl 3 as a function of
levitation time (excitation at 436 nm).
lar absorption coefficient of CdSe particles of about
15 × 104 cm−1 mol L−1 at the first excitonic maximum
(λ = 570 nm).34 The Förster radius for energy transfer
20%. A droplet of 5 μL containing particles at a con- between the CdSe particles was calculated to 6 nm (see
centration of 3.8 × 10−7 mol/L was directly injected Ref. 32 for formula), which corresponds well with the
into the ultrasonic field by means of a microtip, and estimated inter particle distance at high sample con-
the sample was excited at 436 nm by an argon-ion centration (<10 nm) mentioned above. A similar shift
laser. Fluorescence spectra were taken continually ev- of the emission maximum to longer wavelengths due
ery few seconds. During levitation, the volume of the to energy transfer has been observed by other groups
droplet decreases because of the evaporation of the for thin layers of nanocrystals of different sizes.35–37
solvent, reducing the mean particle–particle distance.
FIGURE 5 shows the emission spectra recorded during Conclusion
the evaporation process.
The center of the emission shifts to longer wave- In an acoustic levitator, it is possible to carry out flu-
lengths with increasing sample concentration (FIG. 6). orescence measurements at high concentrations with-
At the same time, the emission intensity is reduced out disturbance by inner filter effects (reabsorption),
(FIG. 6). The latter effect can be attributed to a gen- which often limit such experiments. Monitoring the
eral fluorescence self-quenching effect caused by the evaporation of the solvent with laser-induced fluores-
interparticle energy transfer. The underlying mech- cence techniques makes it possible to monitor volume
anism of the wavelength shift, however, can be changes over three orders of magnitude, thus continu-
rationalized in terms of a resonance energy transfer ously increasing the sample concentration in a straight-
between particles of different sizes. With reducing the forward way to investigate concentration-dependent
sample volume, the mean particle–particle distance effects. For this report, we studied two types of emit-
is reduced and eventually drops to values below ap- ters. For the organic dye fluorescein, the dye molecules
proximately 10 nm. At distances below this value, the remain in monomeric form in the solution through-
particles can exchange excitation energy due to res- out the concentration range studied and shifts of the
onance energy transfer (Förster or fluorescence reso- emission band position are not noticed. This finding
nance energy transfer [FRET]).32 A prerequisite for suggests that at a concentration of 1 × 10−4 mol/L,
this energy transfer is a spectral overlap of the emis- the distance between the individual dye molecules is
sion of the donor with the absorption of the acceptor. still larger than the effective Förster radius necessary
Because the particles in a nanocrystalline ensemble are to activate the FRET process. On the other hand, stud-
not uniform in size but display a certain size distribu- ies of the larger semiconductor nanocrystals revealed
tion, as reflected by the width of the emission band a sufficient reduction in interparticle distance to in-
(vide ante), energy transfer is expected to occur mainly voke FRET, reflected by the observed redshift of the
from smaller, shorter wavelength–emitting particles to fluorescence band.
bigger, longer wavelength–absorbing particles. These In contrast to similar experiments with nanocrystals
bigger particles reemit part of the transferred excita- based on layered samples,38 in our approach, a pos-
tion energy as fluorescence, whereas the emission of the sible influence of the surface on the observed spectral
smaller particles is partly quenched because of the en- changes is avoided and the change in concentration
Leiterer et al.: Ultrasonic Trap for Sampling in Fluorometry 83

can be continuously monitored over several orders of 13. MORGAN, J. et al. 2004. Manipulation of in vitro toxicant
magnitude. Future work aims to improve the dynamic sensors in an ultrasonic standing wave. Toxicol. In Vitro
working range for continuous analysis. Moreover, fur- 18: 115–120.
14. EVANDER, M. et al. 2007. Noninvasive acoustic cell trapping
ther experiments will focus on energy transfer between
in a microfluidic perfusion system for online bioassays.
two nanocrystals of distinctly different sizes, on the in- Anal. Chem. 79: 2984–2991.
teraction between fluorophores and metallic nanopar- 15. LEITERER, J. et al. 2006. The use of an acoustic levitator to
ticles, and on agglomeration studies with quantum dots follow crystallization in small droplets by energy-dispersive
and dyes in colloidal environments. X-ray diffraction. J. Appl. Crystallogr. 39: 771–773.
16. CHUNG, S.K. & E.H. TRINH. 1998. Containerless protein
crystal growth in rotating levitated drops. J. Cryst. Growth
Acknowledgments 194: 384–397.
17. SANTESSON, S. et al. 2003. Screening of nucleation condi-
We thank M. Jakobi for programming the size de- tions using levitated drops for protein crystallization. Anal.
termination system and Prof. A. F. Thünemann for Chem. 75: 1733–1740.
fruitful discussions. 18. XIE, W.J. et al. 2002. Eutectic growth under acoustic levi-
tation conditions. Phys. Rev. E Stat. Nonlin. Soft Matter
Phys. 66 (6 Pt 1): 061601.
Conflict of Interest 19. ROHLING, O., C. WEITKAMP & B. NEIDHART. 2000. Exper-
imental setup for the determination of analytes contained
The authors declare no conflicts of interest. in ultrasonically levitated drops. Fresenius J. Anal. Chem.
368: 125–129.
20. DAVIES, A.N. et al. 2000. Acoustic trap for simplified micro-
References sample handling in laser spectroscopy. Appl. Spectrosc.
54: 1831–1836.
1. JANASEK, D., J. FRANZKE & A. MANZ. 2006. Scaling and the 21. SANTESSON, S. et al. 2003. Airborne chemistry coupled to
design of miniaturized chemical-analysis systems. Nature Raman spectroscopy. Anal. Chem. 75: 2177–2180.
442: 374–380. 22. SEAVER, M. & J.R. PEELE. 1990. Noncontact fluorescence
2. KRICKA, L.J. et al. 2005. Miniaturized detection technology thermometry of acoustically levitated waterdrops. Appl.
in molecular diagnostics. Expert Rev. Mol. Diagn. 5: 549– Opt. 29: 4956–4961.
559. 23. OMRANE, A., G. SARNER & M. ALDEN. 2004. 2D-
3. Symposium P: Materials and Strategies for Lab-on-a-Chip– temperature imaging of single droplets and sprays using
Biological Analysis, Microfactories, and Fluidic Assembly thermographic phosphors. Appl. Phys. B 79: 431–434.
of Nanostructures. S. GREGO et al. 2007. Chairs. Presented 24. DE MELLO, A.J. 2003. Seeing single molecules. Lab Chip 3:
at 2007 MRS Spring Meeting. San Francisco, CA, April 29N–34N.
9–13, 2007. 25. OMRANE, A. et al. 2004. Laser techniques in acoustically
4. PRIEGO-CAPOTE, F. & L. DE CASTRO. 2006. Ultrasound- levitated micro droplets. Lab Chip 4: 287–291.
assisted levitation: Lab-on-a-drop. Trends Anal. Chem. 26. WANG, F. et al. 2006. Luminescent nanomaterials for biolog-
25: 856–867. ical labelling. Nanotechnology 17: R1–R13.
5. SANTESSON, S. & S. NILSSON. 2004. Airborne chemistry: 27. SEYBOLD, P.G., M. GOUTERMAN & J. CALLIS. 1969. Calori-
acoustic levitation in chemical analysis. Anal. Bioanal. metric, photometric and lifetime determinations of fluo-
Chem. 378: 1704–1709. rescence yields of fluorescein dyes. Photochem. Photobiol.
6. WELTER, E. & B. NEIDHART. 1997. Acoustically levitated 9: 229–242.
droplets—a new tool for micro and trace analysis. Frese- 28. CHARREYRE, M.T. et al. 1995. Fluorescence energy trans-
nius J. Anal. Chem. 357: 345–350. fer from fluorescein to tetramethylrhodamine covalently
7. XIE, W.J. et al. 2002. Levitation of iridium and liquid mercury bound to the surface of polystyrene latex particles. Lang-
by ultrasound. Phys. Rev. Lett. 89: 104304. muir 11: 2423–2428.
8. PETERSSON, M. et al. 1998. Sample enrichment in a sin- 29. ALIVISATOS, A.P. 1996. Semiconductor clusters, nanocrys-
gle levitated droplet for capillary electrophoresis. J. Chro- tals, and quantum dots. Science 271: 933–937.
matogr. B 714: 39–46. 30. GAPONENKO, S.V. 1998. Optical Properties of Semiconduc-
9. EBERHARDT, R. & B. NEIDHART. 1999. Acoustic levitation tor Nanocrystals. Cambridge Studies in Modern Optics.
device for sample pretreatment in microanalysis and trace Cambridge University Press. Cambridge, UK.
analysis. Fresenius J. Anal. Chem. 365: 475–479. 31. ZIEGLER, J. et al. 2007. High-quality ZnS shells for CdSe
10. SANTESSON, S. et al. 2000. Airborne cell analysis. Anal. nanoparticles: rapid microwave synthesis. Langmuir 23:
Chem. 72: 3412–3418. 7751–7759.
11. PUSKAR, L. et al. 2007. Raman acoustic levitation spec- 32. LAKOWICZ, J.R. 2006. Principles of Fluorescence Spec-
troscopy of red blood cells and Plasmodium falciparum troscopy. Springer. Berlin.
trophozoites. Lab Chip 7: 1125–1131. 33. HEIDECKER, M., Y. YANMARRIOTT & G. MARRIOTT. 1995.
12. WIKLUND, M. et al. 2003. Ultrasonic-trap-enhanced selectiv- Proximity relationships and structural dynamics of the
ity in capillary electrophoresis. Ultrasonics 41: 329–333. phalloidin binding site of actin filaments in solution and on
84 Annals of the New York Academy of Sciences

single actin filaments on heavy meromyosin. Biochemistry 36. FRANZL, T. et al. 2004. Fast energy transfer in layer-by-layer
34: 11017–11025. assembled CdTe nanocrystal bilayers. Appl. Phys. Lett.
34. YU, W.W. et al. 2003. Experimental determination of the ex- 84: 2904–2906.
tinction coefficient of CdTe, CdSe, and CdS nanocrystals. 37. XU, L. et al. 2007. Luminescence and resonant energy trans-
Chem. Mater. 15: 2854–2860. fer of two sizes of CdTe quantum dots embedded in gelatin
35. CROOKER, S.A. et al. 2002. Spectrally resolved dynamics of films. J. Mater. Sci. 42: 9696–9699.
energy transfer in quantum-dot assemblies: towards engi- 38. FRANZL, T. et al. 2004. Exciton recycling in graded
neered energy flows in artificial materials. Phys. Rev. Lett. gap nanocrystal structures. Nano Lett. 4: 1599–
89: 186802. 1603.