Upconversion luminescence of CdTe nanoparticles

Article abstract of DOI:10.1103/PhysRevB.71.165304

Efficient upconversion luminescence is observed from CdTe nanoparticles in solution and precipitated as solids. In the solids, the upconversion luminescence spectrum is significantly red shifted relative to the photoluminescence spectrum, whereas in solution, there is very little spectral shift. The upconversion luminescence exhibits a near-quadratic laser power dependence, both at room temperature and at 10 K. Both the upconversion and photoluminescence show similar decay dynamics with the solid samples showing shorter lifetimes compared to the solutions. This lifetime shortening is attributed to surface-state quenching. These results indicate that two-photon excitation is the likely upconversion excitation mechanism in these particles and that phonon-populated trap states do not contribute to the upconversion.

Full text of DOI:10.1103/PhysRevB.71.165304

PHYSICAL REVIEW B 71, 165304 ͑2005͒
Upconversion luminescence of CdTe nanoparticles
1
2,
1
3
3
Alan G. Joly, Wei Chen, * David E. McCready, Jan-Olle Malm, and Jan-Olov Bovin
1
Pacific Northwest National Laboratory, P.O. Box 999, Richland, Washington 99352, USA
2
Nomadics, Inc., 1024 South Innovation Way, Stillwater, Oklahoma 74074, USA
3
Materials Chemistry, Center for Chemistry and Chemical Engineering, Lund University, P.O. Box 124, SE-22100, Lund, Sweden
Received 5 July 2004; revised manuscript received 1 November 2004; published 6 April 2005͒
͑
Efficient upconversion luminescence is observed from CdTe nanoparticles in solution and precipitated as
solids. In the solids, the upconversion luminescence spectrum is significantly red shifted relative to the pho-
toluminescence spectrum, whereas in solution, there is very little spectral shift. The upconversion lumines-
cence exhibits a near-quadratic laser power dependence, both at room temperature and at 10 K. Both the
upconversion and photoluminescence show similar decay dynamics with the solid samples showing shorter
lifetimes compared to the solutions. This lifetime shortening is attributed to surface-state quenching. These
results indicate that two-photon excitation is the likely upconversion excitation mechanism in these particles
and that phonon-populated trap states do not contribute to the upconversion.
DOI: 10.1103/PhysRevB.71.165304
PACS number͑s͒: 78.55.Et
I. INTRODUCTION
only requiring a continuous-wave ͑cw͒ source such as a
He-Ne laser or a Xe lamp.
Recent interest in a number of applications including laser
1
Nanoparticles or quantum dots have the potential to be-
come a new class of fluorescent probes for many biological
and biomedical applications, especially cellular imaging.2
Many recent publications have been dedicated to the appli-
technology, three-dimensional microfabrication and optical
2
,3
displays and optical limiting,4 imaging
storage,
4–29
5
6
techniques, and biological caging has renewed interest in
upconversion luminescence.7 Upconversion luminescence
͑
UCL͒ is luminescence in which the emitted wavelength is
cations of luminescent nanoparticles for biological or bio-
medical imaging or probing.2
4–29
In these reports, the lumi-
shorter ͑higher in energy͒ than the excitation wavelength in
contrast to photoluminescence ͑PL͒, where the emitted
wavelength is lower in energy than the excitation photons.
Upconversion luminescence has been readily observed in
nescence images are recorded with an optical microscope or
confocal microscope based on nanoparticle
For biological applications, upconversion
a
2
6–29
fluorescence.
2
+
bulk semiconductors such as ZnS:Mn ͑Ref. 8͒ as well as in
luminescence techniques have several advantages over fluo-
rescence because autofluorescence ͑background lumines-
cence from cell proteins͒ can be avoided using upconversion
9
–11
12,13
14–16
porous silicon,
CdS nanoparticles,
CdTe,
CdSe
1
7
and InP colloidal nanoparticles,
CdSe/ZnS quantum
1
8,19
20–23
5
dots,
and III-V quantum dots.
Auger recombination,
measurements. Consequently, more sensitive and higher-
1
0,22
12,13,21,23
two-photon absorption,
resolution imaging or detection may be obtained. Upconver-
sion luminescence of nanoparticles also has the potential to
be useful in display technology, as well as memory and light-
ing applications. Thus, new materials and techniques for un-
derstanding the underlying physics of these processes is ex-
tremely valuable.
In this paper, we report our new observations on the up-
conversion luminescence of CdTe nanoparticle solid and so-
lution samples based on the measurements of the power de-
pendence and decay dynamics. Our observations reveal that
the upconversion luminescence in the solution and solid
samples occurs by two-photon excitation. In the solid
samples, the upconversion luminescence contains a greater
contribution from either larger particles or surface states rela-
tive to the photoluminescence.
and thermally assisted surface-state processes14–19 have been
proposed to explain these observations. The Auger process
involves transfer of energy from an excited electron-hole pair
upon recombination to another electron or hole, creating a
highly excited carrier. This carrier is then available for re-
combination at a higher energy than the original excitation
wavelength. Two-photon absorption may occur in one of two
ways: The process may proceed through an intermediate
state within the band gap and is then termed two-step two-
photon absorption ͑TSTPA͒ or else the process may proceed
through a virtual intermediate state ͑TPA͒. The process oc-
curring with a virtual intermediate state is significantly
weaker and often requires higher excitation powers. Both the
Auger and TPA processes are inherently nonlinear in nature,
requiring the initial photon to populate an intermediate state
and then excitation of this intermediate state to a higher ex-
cited state via either Auger or a second photon excitation.
Upconversion via surface-state processes involves thermally
populated defect states which absorb a single photon, leading
to higher-energy luminescence. These processes show single-
photon power dependences and a temperature dependence
characterized by increasing UCL intensity at increasing tem-
peratures. These processes are also extremely efficient, often
II. EXPERIMENT
Cadmium perchlorate hydrate ͑Aldrich͒, aluminum tellu-
ride ͑99.5% pure, Gerac͒, thioglycolic ͑mercaptoacetic͒ acid
͑Aldrich͒, L-cysteine ͑99% pure, Alfa͒, and sulfuric acid
͑95% pure, Aldrich͒ were used as received. CdTe nanopar-
ticles were prepared by the reaction of precursors containing
cadmium perchlorate hydrate and hydrogen telluride ͑H Te͒
2
1098-0121/2005/71͑16͒/165304͑9͒/$23.00
165304-1
©2005 The American Physical Society

JOLY et al.
PHYSICAL REVIEW B 71, 165304 ͑2005͒
2
+
under vigorous stirring. The Cd -containing solution was
*
prepared as follows: 0.7311 g of Cd͑CIO ͒ H O was dis-
4
2
2
solved in 125 ml of water. Then 0.25 ml of thioglycolic acid
͑
TGA͒ or 0.4 g of L-cysteine was added to the solution and
the pH adjusted to ϳ11 by the addition of 0.1 Mol NaOH.
The solution was then purged with nitrogen for at least 30
min. H Te gas was generated by the chemical reaction of
2
excess aluminum telluride with 0.5 Mol sulfuric acid in an
inert atmosphere ͑nitrogen͒ and was combined with the
above solution containing Cd2 ions using the setup as de-
scribed in Ref. 30. After the completion of the reaction a
yellow ͑TGA͒ or green ͑L-cysteine͒ solution of CdTe nano-
crystal nuclei was obtained. This solution was then refluxed
at 100 °C to promote crystal growth. During the growth pro-
cess, fractions with nanoparticles of different sizes were ex-
tracted and stored at 4 °C in the dark. The solid samples were
made from the solution by adding acetone and the precipita-
tion was carried out at 4 °C in order to avoid oxidation of the
nanoparticles. The experiments in this paper were carried out
on two solution samples ͑orange and green͒ and three solid
samples ͑yellow, orange, and red͒.
+
FIG. 1. HRTEM image of CdTe nanoparticles orange particles
showing an average size between 3 and 4 nm.
͑for photoluminescence͒ was directed onto the particles
and the emission was collected at right angles and focused
into a streak camera ͑Hamamatsu C5680͒. Suitable bandpass
and cutoff filters were used to collect the luminescence at
different wavelengths. The time resolution was determined to
be about 200 ps full width at half maximum ͑FWHM͒ using
a standard scattering material.
The phase identity and average crystallite size of the
nanoparticles were determined by x-ray powder diffraction
͑
XRPD͒ and high-resolution transmission electron micros-
copy ͑HRTEM͒. The diffraction apparatus was a Philips
X’Pert MRD system ͑PW3040/00 type͒ equipped with vari-
able slits and a Cu x-ray source ͑␭=0.15406 nm͒ operated at
III. RESULTS AND DISCUSSION
A. Solution samples
1
.8 kV. The study specimens were ground in an agate mortar
The methods for making CdTe nanoparticles are well de-
veloped. The sizes of the particles in this paper were con-
trolled by the reaction time as described in Refs. 30–32. To
confirm the formation of CdTe nanoparticles, HRTEM im-
ages were used to observe the structure, size, and shape of
the particles. The particle size of the orange nanoparticle
solution sample is around 3–4 nm, as estimated from HR-
TEM images ͑Fig. 1͒. As observed in Fig. 1, most nanopar-
ticles are spherical in shape. The ͓111͔ lattice spacing of the
particles is estimated to be about 0.36 from the HRTEM
images. This is in good agreement with the ͓111͔ spacing of
cubic CdTe ͓0.374 nm ͑Ref. 33͔͒.
and pestle, then mounted in an off-axis quartz ͑001͒, front-
loading, cavity-type holder which produced minimal back-
ground signal. The XRPD scan range was 10.00°–80.00° 2␪
and the scan rate varied from 0.6 to 1.5° /min. Analysis
of the experimental XRPD patterns was accomplished
using the program JADE ͑Materials Data, Inc., Livermore,
CA͒ and the Powder Diffraction File database ͑International
Centre for Diffraction Data, Newtown Square, PA͒.
The nanoparticles in solution were brought onto holey
carbon-covered copper grids for HRTEM observations.
The HRTEM images of the particles were obtained with an
electron microscope ͑400 kV͒ with a structural resolution of
Figure 2 displays the absorption, photoluminescence, and
upconversion emission spectra of the green and orange solu-
tion samples. Both the green and the orange solutions have
0.16 nm.
Room-temperature optical absorption spectra were taken
with a Hewlett-Packard HP8453 spectrophotometer. The
photoluminescence excitation and emission were recorded
on a SPEX FLUOROLOG fluorescence spectrophotometer.
The upconversion emission spectra and power dependences
were collected using a nanosecond optical parametric
oscillator and amplifer ͑Spectra-Physics MOPO-730͒ operat-
ing at a 10-Hz repetition rate and tunable between
440 nm and 1800 nm. The laser output was directed
onto the particles and emission was collected at right angles
to the excitation and focused into a 1/8-m monochromator
equipped with a gated-intensified charge-coupled-device
͑CCD͒ detector. The power dependences were measured
by integrating the area under the luminescence as a function
of input power. The lifetimes were collected using the
output of a femtosecond regeneratively amplified titanium-
:
sapphire laser system operating at 1 kHz. The 150-fs
FIG. 2. Optical absorption ͑dashed line͒, photoluminescence
emission ͑gray line͒, and upconversion emission ͑black line͒ of
green ͑2.5 nm͒ and orange ͑3.5 nm͒ CdTe particle solutions.
pulses of this laser at either 830 nm ͑for upconversion
luminescence͒ or else the second harmonic at 415 nm
165304-2

UPCONVERSION LUMINESCENCE OF…
PHYSICAL REVIEW B 71, 165304 ͑2005͒
FIG. 4. Photoluminescence emission spectra of the orange CdTe
particle solution ͑dash͒ and solid ͑solid͒ samples.
FIG. 3. Optical absorption ͑dashed line͒, photoluminescence
emission ͑solid line͒, and upconversion emission ͑dotted line͒ of the
yellow ͑3 nm͒, orange ͑3.5 nm͒, and red ͑5 nm͒ CdTe particle solid
samples.
B. Solid samples
Figure 3 displays the photoluminescence and upconver-
sion luminescence spectra of the three solid samples ͑yellow,
orange, and red͒ along with the absorption spectra of the
solution samples from which the solid samples were precipi-
tated. All three solutions have pronounced absorptions peak-
ing at 522 nm, 563 nm, and 585 nm, which are blueshifted
from the energy gap of bulk CdTe at 860 nm ͑Ref. 34͒ as a
result of quantum size confinement. Accordingly, the particle
sizes are estimated to be around 3, 3.5, and 5 nm for the three
samples.
The orange solid sample precipitated from solution dis-
plays a PL spectrum almost identical to the orange solution
͑Fig. 4͒. However, in the solid sample the upconversion lu-
minescence maximum is about 16 nm redshifted from that of
the photoluminescence emission maximum, unlike the solu-
tion sample which shows a negligible difference between the
PL and UCL emission spectra. Similar results are observed
for the yellow and red solid samples. For the yellow sample,
the upconversion luminescence is about 21.5 nm and for the
red solid sample the upconversion emission is about 17 nm
pronounced absorption peaking at 515 nm and 563 nm, re-
spectively, which are blueshifted from the energy gap of bulk
CdTe ͓860 nm ͑Ref. 34͔͒. The absorption maximum and
edges shift to shorter wavelengths with decreasing size as a
consequence of quantum size confinement. The sizes of the
nanoparticles may be estimated using the effective mass
3
5
approximation and the shift of the absorption edge. The
particle sizes are estimated to be around 2.5 nm for the green
and 3.5 nm for the orange solutions, respectively, which are
in good agreement with the results from HRTEM measure-
ments.
The photoluminescence and upconversion luminescence
spectra are also displayed in Fig. 2. The PL of the green
solution peaks at 532 nm and the UCL peaks at 537 nm, 5
nm redshifted from the PL emission maximum. For the or-
ange solution, both the PL and UCL have emission maxima
at the same wavelength, 600 nm.
FIG. 5. X-ray powder diffraction patterns of
the orange, red, and yellow CdTe particle solid
samples.
165304-3

JOLY et al.
PHYSICAL REVIEW B 71, 165304 ͑2005͒
TABLE I. Absorption ͑ABS͒, photoluminescence ͑PL͒, and upconversion luminescence ͑UCL͒ peaks of
CdTe nanoparticles.
Samples
Size ͑nm͒
ABS ͑nm͒
PL ͑nm͒
UCL͑nm͒
Stokes ͑nm͒
PL-UCL͑nm͒
Green solution
Orange solution
Yellow solid
Orange solid
Red solid
2.5
3.5
3
515
563
522
563
585
532
600
577
602
630
537
600
17
37
55
39
45
5
0
598.5
618
21.5
17
17
3.5
5
647
redshifted from that of the photoluminescence emission
maximum.
commonly accepted that in CdTe nanoparticles, larger par-
ticles show emission at longer ͑redder͒ wavelengths and
smaller particles emit at shorter wavelengths.3
0–32,35,36
In ad-
The emission spectrum of the solid orange sample is al-
most identical to that of the solution sample. This indicates
that the particle size in the solid sample is similar to the
particle size in the solution from which the solid sample is
precipitated. For the solid samples the upconversion emis-
sion peak is significantly more redshifted from the photolu-
minescence emission peak than for the solution samples. In
order to better understand the nature of the solid samples,
x-ray powder diffraction measurements were performed on
the three samples. Figure 5 shows the XRPD patterns of the
orange, red, and yellow solid samples. Three reflection peaks
corresponding to the ͑111͒, ͑220͒, and ͑311͒ planes of cubic
zinc-blende CdTe ͑PDF No. 15-770͒ are observed. While the
XRPD results give information about the average crystallite
size, this is not necessarily the same as the particle size. It is
dition, the HRTEM measurement provides information about
the particle size that corresponds well with the optical mea-
surements. From the XRPD results, the samples are observed
to have varying degrees of crystallinity. The orange particles
show the highest degree of crystallinity with an average crys-
tallite size of about 2.3 nm. This is similar to the particle size
as determined by HRTEM and optical measurements and im-
plies that the particles are mostly single phase. The red par-
ticles display both a weakly crystalline component and a
broader background. The broad background arises from par-
ticles with crystallite sizes smaller than the particle size;
therefore, the particles cannot be considered single phase al-
though they still retain a degree of crystallinity. The yellow
particles show only broad diffraction peaks; therefore, the
average crystal domain size is significantly less than the par-
ticle diameter. From these measurements, we can conclude
that the orange particles are mostly monocrystalline, the red
particles are less crystalline, and the yellow particles are the
least crystalline. Thus, although Scherrer’s equation predicts
FIG. 7. A schematic illustration of various processes that result
in upconversion of the incident photons. ͑a͒ represents two-step
two-photon absorption through a real intermediate state while ͑b͒
represents two-photon absorption through a virtual intermediate. ͑c͒
displays upconversion from thermally populated intragap states. ͑d͒
Auger-type upconversion requires two excited carriers which may
interact such that the energy of recombination of the first excited
carrier is transferred to the other excited carrier. This carrier may
then diffuse to a different region ͑material͒ in a heterostructure and
result in luminescence with an energy characteristic of the larger
band-gap material.
FIG. 6. A semilogarithmic base of 10 ͑log ͒ plot of the power
10
dependences of the upconversion emission intensity on laser power
at both room temperature and 10 K for CdTe solution and solid
samples.
165304-4

UPCONVERSION LUMINESCENCE OF…
PHYSICAL REVIEW B 71, 165304 ͑2005͒
power dependences were performed over one order of mag-
nitude, beginning at the detection limit on the low-power
K
end. Using the relation of Iϳpower , where I is the intensity
of the luminescence and “power” represents the input laser
excitation power, the values of K are about 1.9–2.1 for the
solution samples and are 1.5–1.9 for the solid samples.
Therefore the UCL mechanism is largely two-photon in na-
ture, although the yellow solid particles show some varia-
tion. Figure 7 displays mechanisms commonly employed to
explain upconversion luminescence in nanostructured sys-
tems. Both TSTPA and TSA ͓Figs. 7͑a͒ and 7͑b͔͒ are ex-
pected to show a quadratic power dependence as observed
here. In recent investigations of UCL in CdTe nanoparticles,
a linear excitation laser power dependence of the UCL inten-
sity was observed.1
4–16
In addition, extremely low excitation
powers were used to excite the UCL and the UCL in these
particles increased with increasing temperature. From these
observations, the authors concluded that the UCL originates
in thermally populated surface states1
4–16
schematically de-
picted in Fig. 7͑c͒. Auger-type excitation processes as de-
picted in Fig. 7͑d͒ may also lead to a quadratic power depen-
dence on the incident photon power. The Auger-type process
is essentially an energy transfer process between an excited
electron-hole pair and a second excited carrier. The energy
from recombination of the excited electon-hole pair may be
transferred to the second excited carrier, resulting in an in-
crease of this carrier’s energy. This carrier is then available
for recombination resulting in luminescence at a higher en-
ergy than the input photon.
FIG. 8. Photoluminescence ͑solid line, excited at 403 nm͒ and
upconversion ͑dashed line, excited at 805 nm͒ decay curves of
green and orange CdTe nanoparticle solutions.
Auger-type upconversion occurs most frequently in
heterostructures or quantum well systems. This process
requires that the incident photon energy span the band gap of
at least one of the materials to form excitons in the smaller-
band-gap material ͓Fig. 7͑d͔͒. The Auger process results in a
high-energy carrier that can diffuse to the larger-band-gap
material where it can result in frequency upconverted lumi-
nescence from the larger-band-gap material ͓Fig. 7͑d͔͒. Such
that the smallest crystallites should have the broadest diffrac-
3
7
tion peaks, it cannot be used to give reliable values for
particle size because of the varying degrees of crystallinity of
the different samples.
Interestingly, the conclusion from the XRPD measure-
ments is in agreement with the optical observations. The
Stokes shift of the emission maximum from the absorption
peak is a parameter that can reflect the particle quality. Gen-
erally, the Stokes shift is greater if the nanoparticles contain
a higher concentration of surface states or defects because
these surface states or defects are often the origin of the
luminescence and shift the emission to longer wavelength.36
As shown in Table I, the yellow sample has a larger Stokes
shift ͑55 nm͒ than the red sample ͑45 nm͒ and the red sample
has a larger Stokes shift than the orange sample ͑39 nm͒.
This is consistent with the conclusion from the XRPD
observations.
a mechanism has been used to explain UCL in GaAs
22
-
GaInP interfaces. Quadratic Auger-type processes require
2
that electron-hole pairs be formed with the initial photon.
In the CdTe nanoparticles reported here, the incident photons
are not energetic enough to produce electon-hole pairs
or excitons. Therefore, Auger-type processes may be ruled
out as a possible mechanism to explain the observed
upconversion.
Although the solution power dependences are quadratic,
the solids show a slightly smaller value. The nonquadratic
power dependences may result from either competition be-
tween linear and quadratic processes or else saturation of the
nonlinear process. Linear processes as depicted in Fig. 7͑c͒
require the participation of phonons to result in upconverted
photons. In order to check whether the UCL arises from pho-
non populated states, the power dependences of the yellow
and red samples were measured at 10 K and are displayed in
Fig. 6. No variation in the power law is observed at this
temperature and the UCL intensity is observed to increase at
lower temperatures, leading to the conclusion that phonon-
C. Upconversion mechanism
Several publications have been dedicated to the upconver-
sion from semiconductor nanoparticles; however, the mecha-
nisms for upconversion luminescence in semiconductor
nanoparticles are still under debate. Often the dependence of
the UCL on the input photon power dependence can yield
some insight into the mechanism. A semilogarithmic plot dis-
playing the excitation laser power dependences of the upcon-
version luminescence intensity are displayed in Fig. 6 for the
green solution, orange solution, yellow solid, orange solid,
and red solid samples, respectively. The excitation laser
populated states do not contribute to the UCL as has been
reported in other CdTe nanoparticles.1
4–16
Therefore, it is
likely that the slightly lower-than-quadratic power depen-
165304-5

JOLY et al.
PHYSICAL REVIEW B 71, 165304 ͑2005͒
FIG. 9. Photoluminescence ͑solid line, excited
at 403 nm͒ and upconversion ͑dashed line, ex-
cited at 805 nm͒ decay curves of yellow, orange,
and red CdTe particle solid samples.
dences in some of the solid samples originates from satura-
tion of the nonlinear excitation. Semiconductor nanoparticles
such as CdSe-ZnS are known to have extremely large two-
photon cross sections and therefore saturate extremely
easily. Attempts to reduce the input laser power in order to
avoid saturation resulted in undetectable signal levels. There-
fore it is reasonable to conclude that the slightly smaller-
than-quadratic power dependences are due to saturation of a
two-photon excitation.
Two-photon excitation may be accomplished through a
real midgap state or else through a virtual intermediate
state. Both processes would show similar quadratic
laser power dependences. In TSTPA, the excitation process
is determined by the combined excitation cross sections
of the excitation from the ground state to the intermediate
state and the excitation from the intermediate state to
the final state. The overall excitation efficiency is governed
by these cross sections and the lifetime of the intermediate
state. Because the intermediate state lifetime can be
fairly long ͑nsec, µsec, or even msec͒, TSTPA can usually
be accomplished with low-average-power continuous-wave
laser systems. In contrast, TPA requires high peak power
because the virtual intermediate state has an extremely
short lifetime ͑fsec͒ and a large number of photons/sec is
required to achieve excitation. Therefore, short pulses ͑nsec,
psec, or fsec͒ are usually needed to accomplish efficient ex-
citation.
In order to ascertain whether the excitation is through a
real or virtual state, the UCL from the CdTe solutions was
measured under both cw and fsec-pulsed conditions using
the same excitation wavelength in both cases. Following
fsec-pulsed excitation, the UCL is easily observable; how-
ever, following cw excitation, no detectable UCL was ob-
served from any of the CdTe samples investigated. There-
fore, it is reasonable to conclude that the excitation
mechanism is through two-photon excitation via a virtual
intermediate state.
Luminescence lifetimes of both the PL and UCL mea-
sured at or near the peak emission are displayed in Figs. 8
and 9, with the results tabulated in Table II. The lifetimes
reported in Table II for the solid samples are the longest-
decay-component result of multiexponential fitting to the de-
cay curves. In addition to this component, decay lifetimes on
the picosecond ͑ϳ200 psec͒ time scale as well as intermedi-
3
8
165304-6

UPCONVERSION LUMINESCENCE OF…
PHYSICAL REVIEW B 71, 165304 ͑2005͒
TABLE II. Luminescence and upconversion lifetimes of CdTe nanoparticles.
a
a
Samples
Size ͑nm͒
PL Lifetime ͑nsec͒
UCL Lifetime ͑nsec͒
Green solution
Orange solution
Yellow solid
Orange solid
Red solid
2.5
3.5
3
32
23
16
3.5
14
31
27
16
2
3.5
5
14
aLifetime reported is the result of fitting the longest exponential decay of the data.
ate nsec components are present in some of the samples. In
general, the luminescence from the solution samples is
nearly single exponential, whereas the luminescence from
the solid samples is biexponential or multiexponential. This
is likely because the solid particles have less passivated sur-
faces due to the removal of some of the stabilizers as a result
of adding acetone during the precipitation. Multiexponential
behavior is often interpreted as arising from either a distri-
bution of decay times or else to a combination of recombi-
nation from band-edge states, trap states, and surface states.
In any case, the influence of surface and defect states likely
results in the highly nonexponential nature of the solid-
sample decay curves.
excitation energies; the fact that the UCL spectrum is
redshifted argues that different subsets of particles are
excited in each case. For instance, surface or defect states of
slightly lower energy than the band edge may be preferen-
tially excited in the two-photon excited upconversion relative
to one-photon excitation because the excitation cross-section
to these states may be greater. This effect can be enhanced
in the solid particles as the surface capping is of lower
quality because of the removal of capping agents during
precipitation in acetone. Therefore, the UCL may selectively
excite the most defected particles within the sample. The
fact that the UCL power dependence is slightly smaller
than two may be due to saturation of this subset. There is
another possible cause of this redshift in the UCL spectrum
relative to the PL spectrum. The UCL may be composed
of luminescence from slightly larger particles, which
would show redshifted luminescence. The two-photon
excitation cross section to the larger particles may be slightly
greater therefore, the UCL selectively displays this subset.
Unfortunately, the experimental data do not allow a unique
determination of the states involved in the upconversion
in this case.
Our observations indicate that the upconversion lumines-
cence in semiconductor nanoparticles is related to the qual-
ity, particularly surface characteristics. This is likely why
different results are reported by different researchers. For the
CdTe nanoparticles reported here, the upconversion is due to
two-photon excitation. The demonstration of two-photon ex-
citation for upconversion in nanoparticle solutions is of sig-
nificant importantance for applications, particularly for bio-
logical imaging, because two-photon optical imaging has
several obvious advantages over fluorescence imaging.5
Two-photon excitation minimizes tissue photodamage, pho-
totoxicity, and photobleaching as it limits the region of pho-
tointeraction to a subfemtoliter volume at the focal point.
Two-photon excitation wavelengths are typically about
double one-photon excitation wavelengths. This wide sepa-
ration between excitation and emission spectra ensures that
the excitation light and the Raman scattering can be rejected
while filtering out a minimum of fluorescence photons. More
importantly, advantages arise from the use of infrared wave-
lengths, thus avoiding tissue autofluorescence and increasing
the tissue penetration depth.5
Perhaps the most significant lifetime results are the differ-
ences between the solution and solid samples and the signifi-
cant shortening in the orange solid-sample lifetime relative
to the other particles. Recent results also confirm lifetimes
between 1 and 30 nsec for CdTe nanoparticles in
3
9,40
solution.
The shortening of the solid lifetimes relative to
solution may be due to a higher density of surface states
present in the solid samples. The additional shortening of the
orange solid lifetime is not clear. Bulk indium-doped single
crystal CdTe has a very short lifetime, on the order of 200
psec ͑Ref. 41͒. The orange sample is the most crystalline of
the samples as demonstrated by the XRPD measurements.
This may be the reason that the lifetime is significantly
shorter in this sample.
For the CdTe nanoparticle solid samples, the emission
lifetimes excited at 403 nm are almost identical as those
obtained at one-half the excitation energy ͑805 nm excita-
tion͒. ͑Fig. 9 and Table II͒. These results along with the
power dependence favor the mechanism of two-photon ab-
sorption for the quadratic energy contribution to the UCL
intensity. For the solid samples, the upconversion emission
spectra excited at 791 nm are about 17–22 nm redshifted
from photoluminescence spectra obtained at 396 nm excita-
tion ͑Fig. 2 and Table I͒. Recent observations of UCL from
1
4–16
CdTe quantum dots.
dots,
as well as CdSe and InP quantum
1
4,15,17
show a similar redshift of the UCL spectrum rela-
tive to the PL spectrum. A model involving thermally popu-
lated surface or intra-band-gap states followed by resonant
absorption has been used to explain the UCL process in these
particles.
The dramatic shift in the UCL spectra of the solid samples
relative to the PL spectra argues that surface or defect
states may be involved in the UCL process in the solid
samples. Both one- and two-photon absorption ͑PL and
UCL, respectively͒ ultimately result in identical final
IV. SUMMARY
In summary, efficient upconversion luminescence from
CdTe nanoparticles in solution and precipitated as solids has
165304-7

JOLY et al.
PHYSICAL REVIEW B 71, 165304 ͑2005͒
been observed. The upconversion luminescence is shifted to
longer wavelengths relative to the photoluminescence in the
solid samples. The upconversion luminescence shows a non-
linear near-quadratic power dependence at both room tem-
perature and 10 K, leading to the conclusion that thermally
populated surface states are not responsible for the upconver-
sion. Instead, two-photon absorption is likely the dominant
mechanism responsible for upconversion excitation in these
samples.
and No. 1R43CA110091-01͒, and the Department of Energy
͑DOE, Grant No. DE-FG02-04ER84023͒ for grants. Part
of the research described in this paper was performed at the
W.R. Wiley Environmental Molecular Sciences Laboratory,
a national scientific user facility sponsored by the Depart-
ment of Energy’s Office of Biological and Environmental
Research and located at the Pacific Northwest National
Laboratory ͑PNNL͒. PNNL is operated by Battelle for
the U.S. Department of Energy under Contract No. DE-
AC06-76RLO1830. J.-O.M. and J.-O.B. would like to thank
the Swedish Natural Science Research Council and the
Foundation for Strategic Research of Sweden for financial
support.
ACKNOWLEDGMENTS
W.C. would like to thank Nomadics, Inc., the National
Institute of Health ͑NIH, Grant No. 1 R43 CA112756-01
*
Corresponding author. Electronic address: wchen@nomadics.com
S. A. Filonovich, M. J. Gomes, D. V. Talapin, A. L. Rogach, and
A. Eychmuller, Opt. Spectrosc. 94, 921 ͑2003͒.
Y. P. Rakovich, J. F. Donegan, S. A. Filonovich, M. J. Gomes, D.
V. Talapin, A. L. Rogach, and A. Eychmuller, Physica E ͑Am-
sterdam͒ 17, 99 ͑2003͒.
C. Kammerer, G. Cassabois, C. Voisin, C. Delalande, and P.
Roussignol, Phys. Rev. Lett. 87, 207401 ͑2001͒.
P. P. Paskov, P. O. Holtz, B. Monemar, J. M. Garcia, W. V. Schoe-
nfeld, and P. M. Petroff, Appl. Phys. Lett. 77 812 ͑2000͒.
F. A. J. M. Driessen, Appl. Phys. Lett. 67, 2813 ͑1995͒.
Z. P. Su, K. L. Teo, P. Y. Yu, and K. Uchida, Solid State Commun.
99, 933 ͑1996͒.
1
E. Downing, L. Hesselink, J. Ralston, and R. M. Macfarlane,
Science 273, 1185 ͑1996͒.
D. A. Parthenopoulos and P. M. Rentzepis, Science 245, 843
1
9
2
͑
1989͒.
3
4
20
21
M. P. Hehlen, H. U. Gudel, Q. Shu, and S. C. Rand, J. Chem.
Phys. 104, 1232 ͑1996͒.
A. A. Said, C. Wamsely, D. J. Hagan, E. W. van Stryland, B. A.
Reihardt, P. Roderer, and A. G. Dillard, Chem. Phys. Lett. 228,
2
2
6
46 ͑1994͒.
5
6
23
M. D. Cahalan, I. Parker, S. H. Wei, and M. J. Miller, Nat. Rev.
Immunology 2, 872 ͑2002͒.
D. L. Petti, S. S. H. Wang, K. R. Gee, and G. J. Augustine,
Neuron 19, 465 ͑1997͒.
F. Auzel, Chem. Rev. ͑Washington, D.C.͒ 104, 139 ͑2004͒.
W. Chen, A. G. Joly, and J. Z. Zhang, Phys. Rev. B 64,
24
M. Bruchez, M. Moronne, P. Gin, S. Weiss, and A. P. Alivisatos,
Science 281, 2013 ͑1998͒.
7
8
25
26
W. Chan and S. M. Nie, Science 281, 2016 ͑1998͒.
B. Dubertret, P. Skourides, D. J. Norris, V. Noireaux, A. H. Bri-
vanlou, and A. Libchaber, Science 298, 1759 ͑2002͒.
S. Kim, Y. T. Lim, E. G. Soltesz, E. G. Soltesz, A. M. De Grand,
J. Lee, A. Nakayama, J. A. Parker, T. Mihaljevic, R. G. Lau-
rence, D. M. Dor, L. H. Cohn, M. G. Bawendi, and J. V. Fran-
gioni, Nat. Biotechnol. 22, 93, ͑2004͒.
0
41202͑R͒ ͑2001͒.
9
0
27
L. A. Golovan, A. A. Goncharov, V. Y. Timoshenko, A. P. Sh-
kurinov, P. K. Kashkarov, and N. I. Koroteev, JETP Lett. 68,
7
J. Diener, D. Kovalev, H. Heckler, G. Polisski, N. Kunzner, F.
Koch, Al. L. Efros, and M. Rosen, Opt. Mater. ͑Amsterdam,
Neth.͒ 17, 135 ͑2001͒.
70 ͑1998͒.
1
2
8
B. Ballou, B. C. Lagerholm, L. A. Ernst, M. P. Bruchez, and A. S.
Waggoner, Bioconjugate Chem. 15, 79 ͑2004͒.
1
1
29
N. Kunzner, D. Kovalev, H. Heckler, J. Diener, G. Polisski, F.
N. Morgan, S. English, W. Chen, V. Chernomordik, A. Russo, P.
Smith, and A. Gandjbakhche, Acad. Radiol. 12, 313 ͑2005͒.
N. Gaponik, D. Talapin, A. L. Rogach, K. Hoppe, E. V.
Shevchenko, A. Kornowski, A. Eychmuller, and H. Weller, J.
Phys. Chem. B 106, 7177 ͑2002͒.
D. Talapin, S. Haubold, A. L. Rogach, A. Kornowski, M. Haase,
and H. Weller, J. Phys. Chem. B 105, 2260 ͑2001͒.
W. Yu, Y. A. Wang, and X. Peng, Chem. Mater. 15, 4300 ͑2003͒.
A. L. Rogach, L. Katsikas, A. Kornowski, D. Su, A. Eychmuller,
and H. Weller, Ber. Bunsenges. Phys. Chem. 100, 1772 ͑1996͒.
C. Kittel, in Introduction to Solid State Physics, 6th Edition
͑Wiley, New York, 1986͒, p. 185.
Y A. I. Ekimov, AL. L. Efros, and A. A. Onuschenko, Solid State
Commun. 56, 921 ͑1985͒.
W. Chen, in Handbook of Nanostructured Materials and Nano-
technology, editted by H. S. Nalwa ͑Academic Press, San Diego,
2000͒.
Koch, Al. L. Efros, and M. Rosen, Phys. Status Solidi B 224, 21
3
0
͑
2001͒.
1
1
1
2
3
4
S. A. Blanton, A. Dehestani, P. V. Lin, and P. Guyot-Sionnest,
Chem. Phys. Lett. 229, 317 ͑1994͒.
31
32
J. R. Lakowicz, I. Gryczynski, G. Piszczek, and C. J. Murphy, J.
Phys. Chem. B 106, 5365 ͑2002͒.
Y. P. Rakovich, S. A. Filonovich, M. J. Gomes, J. F. Donegan, D.
V. Talapin, A. L. Rogach, and A. Eychmuller, Phys. Status So-
lidi B 229, 449 ͑2002͒.
Y. P. Rakovich, A. A. Gladyshchuk, K. I. Rusakov, S. A. Filonov-
ich, M. J. Gomes, D. V. Talapin, A. L. Rogach, and A. Eych-
muller, J. Appl. Spectrosc. 69, 444 ͑2002͒.
X. Wang, W. Yu, J. Zhang, J. Aldana, X. Peng, and M. Xiao,
Phys. Rev. B 68, 125318 ͑2003͒.
E. Poles, D. C. Selmarten, O. I. Micic, and A. J. Nozik, Appl.
Phys. Lett. 75, 971 ͑1999͒.
K. I. Rusakov, A. A. Gladyshchuk, Y. P. Rakovich, J. F. Donegan,
3
3
1
5
34
3
3
5
6
1
1
1
6
7
8
37
Handbook of X-Rays, edited by E. F. Kaelble. ͑McGraw-Hill,
165304-8

UPCONVERSION LUMINESCENCE OF…
New York, 1967͒.
PHYSICAL REVIEW B 71, 165304 ͑2005͒
Mello Donega, Nano Lett. 3, 503 ͑2003͒.
3
8
40
D. R. Larson, W. R. Zipfel, R. M. Williams, S. W. Clark, M. P.
S. F. Wuister, F. van Driel, and A. Meijerink, Phys. Chem. Chem.
Bruchez, F. W. Wise, and W. W. Webb, Science 300, 1434
Phys. 5, 1253 ͑2003͒.
G. Ghislotti, D. Ielmini, E. Riedo, M. Martinelli, and M. Della-
giovanna Solid State Commun. 111, 211 ͑1999͒.
4
1
͑
2003͒.
3
9
S. F. Wuister, I. Swart, F. van Driel, S. G. Hickey, and C. de
165304-9