Relaxation and electron transfer dynamics in bare and DTDCI sensitized MoS2 nanoclusters

Article abstract of DOI:10.1063/1.1289765

The trapping dynamics of photogenerated electron and holes in MoS₂ nanoclusters using time resolved emission polarization and absorption spectroscopies were studied. The results indicated that emission from MoS₂ band edge state were polarized and that from trapped electrons and holes were unpolarized. Results obtained from MoS₂/DTDCI (diethylthiodicarbocyanine iodide) showed that electron emission occurs with a time constant of 12 ps.

Full text of DOI:10.1063/1.1289765

Relaxation and electron transfer dynamics in bare and DTDCI sensitized MoS 2
nanoclusters
V. Chikan, M. R. Waterland, J. M. Huang, and D. F. Kelley
Citation: The Journal of Chemical Physics 113, 5448 (2000); doi: 10.1063/1.1289765
View online: http://dx.doi.org/10.1063/1.1289765
View Table of Contents: http://scitation.aip.org/content/aip/journal/jcp/113/13?ver=pdfcov
Published by the AIP Publishing
Articles you may be interested in
Controlled growth of vertically aligned MoO3 nanoflakes by plasma assisted paste sublimation process
J. Appl. Phys. 114, 184310 (2013); 10.1063/1.4830278
Investigation of the photoluminescence properties of thermochemically synthesized CdS nanocrystals
AIP Advances 1, 012113 (2011); 10.1063/1.3562892
Carrier relaxation dynamics in annealed and hydrogenated ( Ga In ) ( N As ) ∕ Ga As quantum wells
Appl. Phys. Lett. 87, 252111 (2005); 10.1063/1.2149154
Electron and hole trapping in WS 2 nanoclusters
J. Chem. Phys. 113, 793 (2000); 10.1063/1.481854
Carrier trapping due to Fe 3+ /Fe 2+ in epitaxial InP
Appl. Phys. Lett. 70, 3374 (1997); 10.1063/1.119175
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
155.33.16.124 On: Sun, 30 Nov 2014 05:24:31

JOURNAL OF CHEMICAL PHYSICS
VOLUME 113, NUMBER 13
1 OCTOBER 2000
Relaxation and electron transfer dynamics in bare and DTDCI sensitized
MoS2 nanoclusters
V. Chikan, M. R. Waterland, J. M. Huang,a) and D. F. Kelleyb)
Department of Chemistry, Kansas State University, Manhattan, Kansas 66506-3701
͑Received 8 May 2000; accepted 5 July 2000͒
The trapping dynamics of photogenerated electrons and holes in MoS2 nanoclusters has been studied
using time resolved emission polarization and absorption spectroscopies. These results are
compared to absorption kinetics obtained on MoS2 nanoclusters with adsorbed DTDCI
͑DTDCIϵdiethylthiodicarbocyanine iodide͒ dye. The results indicate that emission from the MoS2
band edge state is polarized, while emission from trapped electrons and holes is unpolarized. This
polarization difference is used to obtain the electron and hole trapping times and values of 275 ps
and 42 ps, respectively, are obtained. Decays having the same time constants are observed in the
transient absorption results. The results obtained on the MoS2 /DTDCI system show that electron
injection occurs with a time constant of 12 ps. These kinetics also show a 225 ps decay component
which is assigned to electron trapping and reverse electron transfer. The 225 ps decay time along
with the 275 ps trapping time indicates that reverse electron transfer from the conduction band has
a time constant of about 1.2 ns. Following trapping, reverse electron transfer from deep traps is
slow. © 2000 American Institute of Physics. ͓S0021-9606͑00͒70337-0͔
I. INTRODUCTION
and hole traps. Thus nanometer sized semiconductor par-
ticles may have electronic structures which are vastly differ-
ent than that of the bulk semiconductor, and can have corre-
spondingly different photophysical and electronic properties.
The finite relaxation rates of both electrons and holes be-
tween the delocalized and the localized ͑trap͒ states can dra-
matically affect the dynamics following direct photoexcita-
tion or excitation of an adsorbed sensitizing dye.
There has recently been great interest in the sensitization
of nanoporous semiconductor films and of semiconductor
nanoclusters by adsorbed dyes.1 Sensitization occurs when a
dye absorbs light and injects an electron into the conduction
band of a large band gap semiconductor. Much of this inter-
est results from the potential application of these systems in
solar energy conversion.2–6 Sensitization of nanoporous
films exploits the fact that these materials have very high
surface-to-volume ratios. In addition to high surface to vol-
ume ratios, some types of semiconductor nanoclusters and
nanoporous films of these semiconductors exhibit quantum
confinement of photogenerated electrons and holes. As a re-
sult of quantum confinement, the properties of the nanoclus-
ters can be very different than the properties of the bulk.
Potential applications of these materials may also exploit
these differences.
Quantum confinement can have large effects on the
spectroscopy and photophysics of semiconductor
nanoclusters.7–10 The finite particle size can cause an in-
crease in the band gap energy and cause the dense continua
of valence and conduction band electronic states to become a
set of delocalized, discrete states. In sufficiently small par-
ticles, the valence and conduction bands are replaced by sets
of well-separated electronic states. In addition to these delo-
calized electronic states, semiconductor nanoclusters can
also have a high density of localized surface states: electron
The transition metal dichalcogenides, specifically MX2
͑MϭMo, W and XϭS, Se͒ are of considerable interest in
photoelectrochemical cells, principally because they are
highly photostable.11 Of these materials, nanoclusters of
MoS2 have received the most attention. Their synthesis,
characterization, and several aspects of their spectroscopy
have been reported.12 MoS2 is a layered material, consisting
of covalently bound S–Mo–S trilayers separated by large
van der Waals gaps. The molybdenum atoms are in trigonal
bipyramidal sites between hexagonally close packed sulfur
planes.13,14 MoS2 nanoclusters are routinely synthesized us-
ing room temperature inverse micelle methods.12 X-ray dif-
fraction results show that aggregates of the particles have the
same crystal structure as bulk MoS2.12a While there is no
direct evidence regarding the morphology of these particles,
comparison with analogous nanoclusters of materials having
very similar bulk properties can yield some insight into this
question. MoSe2, WS2, and WSe2 have the same crystal
structure as bulk MoS2 and form comparably sized
nanoclusters.13–17 Electron diffraction results show that
MoSe2, WS2, and WSe2 nanoclusters consist of single
X–M–X trilayers having the same crystal structure as the
bulk material.15,17 It is reasonable to assume that this is also
the case for the smallest MoS2 nanoclusters.
a͒
Present address: NSF/Center for Micro-Engineered Materials, 203 Farris
Engineering Center, University of New Mexico, Albuquerque, NM 87131.
b͒
Author to whom correspondence should be addressed: Electronic mail:
dfkelley@ksu.edu
0021-9606/2000/113(13)/5448/9/$17.00
5448
© 2000 American Institute of Physics
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
155.33.16.124 On: Sun, 30 Nov 2014 05:24:31

5449
J. Chem. Phys., Vol. 113, No. 13, 1 October 2000
Electron transfer dynamics in MoS2 clusters
In addition to structural and static spectroscopic studies,
several types of dynamical studies have also been performed
on MoS2 nanoclusters.18–21 Specifically, dynamical studies
of the recombination of trapped electrons and holes18 as well
as interfacial electron transfer involving adsorbed electron
acceptors have been reported.19 In addition, some of the in-
terfacial electron transfer dynamics involving adsorbed dyes
have also been studied.20,21
In this paper, we examine the dynamics associated with
direct photoexcitation of the nanoclusters and with electron
injection from an adsorbed sensitizing dye, DTDCI
͑DTDCIϵdiethylthiodicarbocyanine iodide͒. These two sets
of dynamics are considered together because both produce a
conduction band electron which can subsequently undergo
trapping and recombination with the hole or the oxidized
dye. In the present study, time resolved emission polarization
and adsorption measurements are used to determine these
dynamics. We show that the band edge emission is polarized
while emission from trapped electrons and holes is unpolar-
ized and exploit these polarization differences to determine
the trapping rates. However, trapping of either electrons or
holes could result in partial or complete emission depolariza-
tion. Thus the observed depolarization dynamics may be pri-
marily due to trapping of either carrier. This ambiguity can
be resolved by the comparison of the emission depolarization
and transient absorption results. This is because while both
nanocluster and dye excitation result in injection of an elec-
tron into the nanocluster conduction band and subsequent
electron trapping, no hole is produced and so there are no
hole dynamics in the dye excitation case. The combination of
these experiments permits determination of kinetics of elec-
tron injection, electron trapping, reverse ͑nanocluster to oxi-
dized dye͒ electron transfer, and hole trapping.
The studies presented here use some of the same meth-
ods and are quite complementary to our recent studies of the
electron and hole trapping dynamics in WS2 nanoclusters. In
previous publications15,16 we have shown that just as in the
MoS2 case, WS2 bandedge emission is polarized while emis-
sion from trapped electrons and holes has little or no polar-
ization. As in the present case, the emission depolarization
undergoes a biphasic decay, and there is an ambiguity re-
garding which component corresponds to electron versus
hole trapping. In the WS2 case, this ambiguity was resolved
by comparison of the dynamics in the presence and in the
absence of adsorbed electron acceptors. The presence of the
electron acceptors partially quenches the population of con-
duction band electrons, thereby allowing assignment of the
electron trapping dynamics in the observed depolarization
kinetics. We found that electron and hole trapping occur with
330 ps and 28 ps time constants, respectively. Comparable
time scales are also observed in the MoS2 case reported here.
The results presented here also allow us to comment on the
spectroscopy of MoS2 nanoclusters.
FIG. 1. Static absorption spectra of MoS2 nanoclusters ͑dot-dash curve͒ and
MoS2 nanoclusters with adsorbed DTDCI dye ͑solid curve͒ in toluene. Also
shown is an emission spectrum of MoS2 nanoclusters in acetonitrile ͑dashed
curve͒.
present study. Within these limitations, the results reported
in the previous study are completely consistent with those
reported here. However, the electron trapping process was
not resolved and what is shown below to be reverse electron
transfer from the conduction band was previously assigned
to reverse electron transfer from shallow trap states.
II. EXPERIMENT
Molybdenum͑IV͒sulfide nanoclusters are synthesized us-
ing the same methods used in previous studies.12,18,19 Briefly,
this synthesis is as follows. MoCl4 is dissolved in a degassed
ternary tridodecylmethyl ammonium iodide ͑TDAI͒/hexanol/
octane ͑8/8/84 by weight͒ inverse micelle solution, at a con-
centration of 1.0ϫ10Ϫ3 M. Typical sample volumes are
10–20 mL. Four molar equivalents of H2S are added via
gas-tight syringe to these solutions. Reaction occurs within a
few minutes. Following reaction, the inverse micelle solution
along with the MoS2 nanoclusters are extracted with an equal
volume of acetonitrile. The nanoclusters partition between
the nonpolar octane phase and the polar acetonitrile/TDAI/
hexanol phase. The two phases show nearly identical absorp-
tion spectra, with a sharp absorption maximum at 364 nm
͑see Fig. 1͒. MoS2 nanoclusters having a 364 nm absorption
maximum have been characterized and have been reported to
have an average diameter of 3.0 nm. In agreement with these
results, we have recently performed rotational diffusion mea-
surements on these particles and find an average diameter of
about 3.5 nm. These rotational diffusion measurements have
characterized the size distributions of these and several other
types of nanoclusters, and will be reported in a later paper.
The acetonitrile phase was used for emission polarization
studies. Samples of MoS2 nanoclusters with adsorbed
DTDCI in toluene, hexane, or hexadecane were prepared by
initially dissolving the dye in the nanocluster acetonitrile so-
lution. This permits the DTDCI to attach to the nanoclusters.
The MoS2 /DTDCI results presented here can be com-
pared to those obtained in our previous study of this
system.20 In that study, MoS2 /DTDCI was examined using
transient absorption spectroscopy. However, the temporal
resolution was 35 ps ͑compared to subpicosecond resolution
here͒ and the signal to noise ratio was lower than in the
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
155.33.16.124 On: Sun, 30 Nov 2014 05:24:31

5450
J. Chem. Phys., Vol. 113, No. 13, 1 October 2000
Chikan et al.
The solvent is pumped off and the nanoclusters are then re-
dissolved in hydrocarbon solvent. Bare nanocluster samples
in toluene ͑without the dye͒ are prepared in exactly the same
way.
1.0. The sample solutions are continuously stirred to avoid
local heating effects. In all cases, reducing the intensity of
the pump beam by a factor of 5 lowers the amplitude of the
signal, but otherwise has no effect on the observed kinetics.
The sample is probed with a 10 nm slice of a white light
continuum. Wavelength selection is accomplished using 10
nm bandpass interference filters. Continuum generation is
achieved by focusing a small fraction ͑4%, ca. 32 ␮J per
pulse͒ of the 775 nm pulses onto a sapphire plate and then
re-collimating with an achromat. A beam splitter separates
the white light continuum into two components that form the
reference and probe beams. Residual 775 nm from the con-
tinuum generation stage is removed prior to the sample. Re-
sidual pump energy is removed after the sample with a suit-
able dichroic mirror. In the present configuration, there is no
compensation for group velocity dispersion, and the instru-
ment response function is about 500 fs. Detection of the
probe and reference beams is achieved with silicon photo-
diodes ͑UDT Detectors, model PIN-10D͒ biased at Ϫ12 V.
The voltage from the photodiodes is amplified by a factor of
either 100 or 200 and integrated with a Stanford Research
Systems boxcar gated integrator. The resulting signals are
input into a 16 bit A/D converter ͑Analog Devices, model
PCI-MIO-16XE-50͒ in the backplane of a Pentium II PC
running LabView. Each laser shot is integrated and digitized
and all signal averaging is performed with home-written
LabView software in the PC. Typically, a total of 5000 laser
shots ͑20 scansϫ250 shots per point͒ are acquired at each
delay stage position.
Molybdenum͑IV͒chloride was synthesized by refluxing
MoCl5 in tetrachloroethylene in an evacuated and sealed ves-
sel for three days. The resulting product was repeatedly
washed with carbon tetrachloride in an inert atmosphere
glovebox to remove starting material. TDAI was repeatedly
washed with ether, and stored in the glovebox in the dark.
Octane was distilled from sodium, in a nitrogen atmosphere.
Acetonitrile was distilled from phosphorous pentoxide in a
nitrogen atmosphere. Hexanol was distilled from iodine-
activated magnesium in a nitrogen atmosphere. DTDCI was
obtained from Exciton and used without further purification.
Time resolved emission results were obtained by time-
correlated single photon counting, using an apparatus that
has been previously described.22 In this apparatus, the
sample is excited with 7 ps pulses from the second harmonic
of a synch-pumped, cavity dumped cw dye laser ͑Spectra-
Physics 3500͒, which is pumped by a cw mode locked YAG
laser ͑Spectra-Physics 3800͒. Detection is accomplished with
a Hamamatsu 6 ␮MCP PMT and time correlated single pho-
ton counting electronics. Wavelength selection is accom-
1
4
plished using a m monochromator with a 150 groove/mm
grating. The emission kinetics reported here were taken at
450 nm, with a bandwidth of about 10 nm. In all cases, the
excitation wavelength was 312 nm with pulse energies of
about 1.0 nJ focused to a spot size of about 0.5 mm. This
combination of low pulse energies and relatively large spot
sizes results in fluences that are sufficiently low that multi-
photon excitations are completely avoided. Polarized emis-
sion detection was accomplished using a 1 cm Glan–Taylor
excitation polarizer and a Polaroid emission polarizer in a
collinear geometry. Appropriate corrections were made for
the polarization dependent throughput of the detection sys-
tem.
The femtosecond transient absorption apparatus is based
on a Clark-MXR CPA-2001 laser system. The CPA-2001
utilizes an Erbium doped fiber as the gain medium in a
mode-locked fiber laser. The mode-locked output of the Er-
bium fiber oscillator, centered at 1550 nm, is pulse com-
pressed and frequency-doubled to generate seed pulses for
the Ti:Sapphire regenerative amplifier. This amplifier is
pumped by a Q-switched, frequency-doubled Nd:YAG,
which provides 8 W of 532 nm light at a repetition rate of 1
kHz. A feedback loop provides long-term Nd:YAG power
stability. The resultant output is 800 ␮J, 130 fs 775 nm
pulses, at 1 kHz. This output is frequency doubled to 387.5
nm to provide sample excitation of the bare nanocluster
samples. Alternatively, the 775 nm output is used to pump an
optical parametric amplifier ͑Clark-MXR, Vis-OPA͒. This
provides continuously tunable pump pulses with 10 ␮J per
pulse over the range from 420 nm to Ͼ660 nm. Nanocluster/
dye samples were excited at 540 nm. Pump intensities are
typically 20 ␮J/pulse ͑387.5 nm͒ or 10 ␮J/pulse ͑540 nm͒,
and were typically focused to 0.5–1.0 mm spot sizes on a 1
cm thick sample cell. Sample concentrations are adjusted
such that the absorbance at the excitation wavelength is 0.5–
III. RESULTS AND DISCUSSION
A. Emission depolarization kinetics
The absorption spectra of MoS2 nanoclusters with and
without adsorbed DTDCI in room temperature toluene are
shown in Fig. 1. The bare nanoclusters show an almost iden-
tical absorption spectrum in acetonitrile. The emission spec-
trum of MoS2 nanoclusters in acetonitrile is also shown in
Fig. 1. Following excitation with polarized 312 nm light, the
resulting emission is polarized. The emission anisotropy is
given by23
rϭ IparϪIper͒/ Iparϩ2Iper͒,
͑1͒
͑
͑
where Ipar and Iper are the emission intensities having polar-
izations parallel and perpendicular to that of the absorbed
light, respectively. An experimental plot of the time-
dependent 450 nm emission anisotropy of bare MoS2 nano-
clusters in acetonitrile is shown in Fig. 2. Photoselection
theory shows that following excitation with polarized light, a
linear oscillator exhibits an emission anisotropy of 0.40. The
anisotropy starts out with a value close to 0.40 and subse-
quently decays, indicating that photoexcitation produces a
state that is close to being a linear oscillator. Figure 2 shows
that the anisotropy has an initial fast, almost pulse width
limited decay, followed by a slower decay, on the hundreds
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
155.33.16.124 On: Sun, 30 Nov 2014 05:24:31

5451
J. Chem. Phys., Vol. 113, No. 13, 1 October 2000
Electron transfer dynamics in MoS2 clusters
FIG. 3. Experimental plot of polarized emission intensity (5/2(IparϪIper),
closed circles͒ and of unpolarized emission intensity (3/2(3IperϪIpar), open
circles͒ versus time for MoS2 nanoclusters in acetonitrile. Also shown is a
calculated triexponential decay which is convoluted with the instrument
response function. The calculated curve corresponds to decay components
and relative amplitudes of 42 ps ͑60%͒, 275 ps ͑35%͒ and 3.0 ns ͑5%͒.
FIG. 2. Experimental plot of emission anisotropy versus time for MoS2
nanoclusters in acetonitrile. The excitation and detection wavelength were
312 nm and 450 nm, respectively.
of picoseconds time scale. Figure 2 also shows that there is
also a very small anisotropy which decays on a longer time
scale.
I
perϭIper, rϭ0.0ϩIper, rϭ0.40 .
These polarization results allow us to comment on the
spectroscopy of MoS2 nanoclusters, specifically, the assign-
ment of the 364 nm peak. The 364 and 294 nm absorption
peaks have previously been assigned12 as correlating to the A
and B excitons, respectively, in bulk MoS2. The A and B
exciton transitions are polarized in the plane of the
nanoclusters,24 perpendicular to the crystallographic c-axis.
As such, the maximum anisotropy from excitation of the
either the A or B exciton transition is 0.10. The observation
of more polarized emission (rХ0.40 at tϭ0) indicates that
this assignment is not correct and that the 364 nm peak cor-
relates to a c-axis polarized transition of bulk MoS2. Several
such transitions have been observed at energies somewhat
higher than the A exciton,25,26 but no definite assignment can
be made at this point. This result indicates that the in plane
transitions are subject to a larger quantum confinement effect
than the out of plane transitions. Bulk MoS2 has only a rela-
tively weak interactions between S–Mo–S trilayers ͑along
the c-axis͒, so this result is not too surprising. A similar
conclusion was made in the WS2 nanocluster case.15,16
Analysis of the anisotropy decay, specifically the pro-
cesses resulting in emission depolarization, is facilitated by
plotting the results in terms of the kinetics of the polarized
and unpolarized emission components. This corresponds to a
model in which there are two emitting states, a linearly po-
larized (rϭ0.40) and an unpolarized, isotropic (rϭ0.0) os-
cillator. It is possible to relate the experimentally observed
parallel and perpendicular emission intensities to the total
emission intensities from the polarized and the unpolarized
oscillators. Specifically,
Using Eq. ͑1͒ and an anisotropy value of 0.40 for the polar-
ized oscillator, we get that
I
par, rϭ0.40ϭ3Iper, rϭ0.40
.
͑2b͒
For the unpolarized emission, Iper, rϭ0.0ϭIpar, rϭ0.0 . The total
population in each state is proportional to its total (Ipar
ϩ2Iper) time-dependent emission intensity. The total time-
dependent polarized emission intensity is given by
I
tot, rϭ0.40ϭ(Ipar, rϭ0.40ϩ2Iper, rϭ0.40)ϭ5/2(IparϪIper),
the total unpolarized emission intensity is given by
tot, rϭ0.0ϭ(Ipar, rϭ0.0ϩ2Iper, rϭ0.0)ϭ3/2(3IperϪIpar). Plots of
and
I
these quantities are shown in Fig. 3. In the simplest case,
these plots give the time dependent relative populations of
the polarized and unpolarized emitting states. A state having
emission which is partially polarized will contribute to the
apparent intensities of both the polarized and unpolarized
components. The extent of the contribution into each com-
ponent will depend on the extent to which the emission is
polarized. The strongly biphasic decays of the anisotropy
͑Fig. 2͒ and the polarized emission component in Fig. 3 in-
dicate that there is an intermediate state which gives partially
polarized emission. The emission anisotropy of this state can
be estimated by extrapolating the slowly decaying ͑hundreds
of picoseconds͒ component of the emission anisotropy to t
ϭ0, and a value of about rϭ0.12 is obtained. With the in-
clusion of this state, we get that the polarized emission in-
tensity is related to the population in the rϭ0.40 oscillator
plus some component of the rϭ0.12 oscillator. Similarly, the
unpolarized emission intensity is related to the population in
the isotropic (rϭ0.0) oscillator plus some component of the
rϭ0.12 oscillator. Specifically,
I
parϭIpar, rϭ0.0ϩIpar, rϭ0.40
and
͑2a͒
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
155.33.16.124 On: Sun, 30 Nov 2014 05:24:31

5452
J. Chem. Phys., Vol. 113, No. 13, 1 October 2000
Chikan et al.
The initial polarized emission is assigned to the band-
edge state and the unpolarized emission is assigned to emis-
sion from trapped electrons and holes. The long-time multi-
exponential decay of the unpolarized emission is associated
with the distribution of trapped electron-hole separations,
and has been previously discussed for both MoS2 and WS2
nanoclusters.15,19 The intermediate state is assigned to
trapped holes and conduction band electrons. With these as-
signments, the unpolarized emission is due to trapped holes
recombining with trapped electrons and to a lesser extent,
conduction band electrons, and the polarized emission is due
to conduction band electrons recombining with valence band
holes, and to a lesser extent, trapped holes. The extent to
which the band edge, trapped hole/conduction band electron,
and trapped electron and hole states contribute to the polar-
ized and unpolarized emission components is given by Eq.
͑3͒. Thus the 42 ps and 275 ps depolarization components
are assigned to hole and electron trapping, respectively. It
follows that the bandedge emission is polarized as a linear
oscillator, rϭ0.40, and that hole trapping significantly, but
not completely, depolarizes this emission. Radiative recom-
bination of a trapped hole and a conduction band electron
gives somewhat polarized (rϭ0.12) emission. This situation
is depicted in Scheme 1.
FIG. 4. Experimental plots of the absorption intensity versus time for MoS2
nanoclusters in toluene. The excitation and probe wavelengths were 387.5
and 700 nm, respectively. A and B display the same results, but on different
time scales. Also shown is a calculated curve corresponding to decay com-
ponents of 42 ps ͑61%͒, 275 ps ͑17%͒ and 3.0 ns ͑22%͒.
The assignment of comparatively rapid hole trapping fol-
lowed by slower electron trapping ͑rather than the other way
around͒ is established by comparison to the transient absorp-
tion results obtained on MoS2 nanoclusters with and without
adsorbed dyes. These results are discussed below. The small
͑ϳ5%͒, long lived component of polarized emission may be
due to nanoparticles lacking trap states, a small polarization
from trapped electrons and holes, or thermal population of
the conduction band from trapped electrons. These possible
contributions to the long lived polarized emission are not
mutually exclusive, and we find that the exact amount of this
component varies from sample to sample. In the case of WS2
nanoclusters, it was shown that the electron traps are very
shallow and the hole traps are much deeper. Thus in the WS2
case, a similar long lived polarized emission component is at
least in part due to thermal repopulation of the conduction
band by trapped electrons.16 No effort to further analyze this
component is made here.
5/2 IparϪIper͒ϭItot, rϭ0.40ϩ0.3Itot, rϭ0.12
,
͑3a͒
͑3b͒
͑
3/2 3IperϪIpar͒ϭItot, rϭ0.0ϩ0.7Itot, rϭ0.12
.
͑
To the extent that the oscillator strengths and spectral char-
acteristics of these oscillators are the same, the relative emis-
sion intensities shown in Fig. 3 are related to the relative
populations. Figure 3 shows that the polarized ͓5/2(Ipar
ϪIper)͔ kinetics exhibit an instrument limited rise followed
by a triexponential decay. These kinetics are fit to compo-
nents having decay times and relative amplitudes of 42 ps
͑65%͒, 275 ps ͑30%͒ and 3.0 ns ͑5%͒. The corresponding
unpolarized emission kinetics are also shown in Fig. 3. There
are two kinetic components having finite rise times followed
by a multiexponential decay. The slower of the two rise
times occurs on the timescale of a few hundred picoseconds,
and is associated with the 275 ps decay of the polarized
emission. Most of the unpolarized emission has a rise time
which matches the fast ͑42 ps͒ decay of the polarized emis-
sion.
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
155.33.16.124 On: Sun, 30 Nov 2014 05:24:31

5453
J. Chem. Phys., Vol. 113, No. 13, 1 October 2000
Electron transfer dynamics in MoS2 clusters
FIG. 6. Same as Fig. 5, except the detection wavelength was 700 nm.
FIG. 5. Experimental plots of the absorption intensity versus time for MoS2
nanoclusters with adsorbed DTDCI in toluene. The excitation and detection
wavelengths were 540 and 660 nm, respectively. A and B display the same
results, but on different time scales. Also shown is a calculated curve cor-
responding to decay components of 12 ps, 225 ps, 1.7 ns and a long lived
absorption.
The relative amplitudes of these transients indicate that hole
trapping results in considerable loss of 700 nm absorption,
and that electron trapping also results in loss of 700 nm
absorption, but to a lesser extent.
The 660 nm and 700 nm transient absorption kinetics of
MoS2 nanoclusters with adsorbed DTDCI dye following 540
nm excitation are shown in Figs. 5 and 6, respectively. Sev-
eral species can absorb in this wavelength range and may
therefore contribute to the observed kinetics. DTDCI has a
significant ground state absorption at 660 and 700 nm ͑see
Fig. 1͒ and the net absorbance change following photoexci-
tation is negative; the ground state dye is bleached. In addi-
tion to the ground state bleach, there can be absorption from
the excited state dye, the dye cation, and the conduction band
or trapped electron. Following the pulse width limited onset
of the bleach, the 660 nm kinetics show a subsequent small
amplitude, 12 ps increase in the absorption ͑a bleach de-
crease͒ followed by a slower bleach decay. The 700 nm ki-
netics also show a small amplitude initial 12 ps transient, but
at this wavelength it is an absorption decrease ͑an apparent
increase of the bleach͒. Figure 4 shows that the bare MoS2
kinetics exhibit no 12 ps transient, so this kinetic component
is assigned to an interaction of the nanocluster with the dye.
Cyanine dyes are known to inject electrons into semiconduc-
B. Transient absorption kinetics
The 700 nm transient absorption kinetics of MoS2 nano-
clusters following 387.5 nm excitation are shown in Fig. 4.
Several kinetic components can be seen. There is a fast tran-
sient observed in the first 10 ps. This consists of an approxi-
mately 1.5 ps rise followed by a 2.5–3.0 ps decay. This
transient is associated with relaxation to the bandedge, indi-
cating that the state excited with 387.5 light is not the band
edge state. This suggests that the emitting bandedge state
corresponds to a weak absorption buried under the red edge
of the 364 nm absorption. The emission polarization results
indicate that the excited and emitting states are both polar-
ized parallel the c-axis but no detailed assignment of these
states can be made on the basis of the results presented here.
The fast transients are followed by a decay having com-
ponents of 42 ps ͑61%͒, 275 ps ͑17%͒ and 3 ns ͑22%͒. A
calculated curve corresponding to these values is also shown
in Fig. 4. These decay times correspond to the times ob-
served in the polarized emission kinetics discussed above.
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
155.33.16.124 On: Sun, 30 Nov 2014 05:24:31

5454
J. Chem. Phys., Vol. 113, No. 13, 1 October 2000
Chikan et al.
tors and it is quite reasonable to assign the 12 ps transient to
electron injection from the photoexcited dye to the nanoclus-
ter conduction band. With this assignment, it follows that the
excited state dye has less absorbance than the dye cation and
the conduction band electron at 660 nm. The opposite is true
at 700 nm. Almost identical results are obtained in hexane
and in hexadecane solvents.
tion band electron. The 660 and 700 nm extinction coeffi-
cients of the ground state dye are very large and partial re-
population of the ground state dye is expected to result in a
net increase of the absorption, consistent with the observed
decays. Thus reverse ͑conduction band to oxidized dye͒ elec-
tron transfer occurs along with electron trapping. These dy-
namics are summarized in Scheme 2.
Cyanine dyes are known to undergo excited state
isomerization,27–29 with a rate that depends on the solvent
viscosity. This isomerization results in a time dependent in-
crease in the observed bleach throughout the 600–700 nm
region for free DTDCI in alcohols. Otherwise stated,
throughout this spectral region, the excited state dye has
more absorbance prior to, compared to following isomeriza-
tion. We find that the isomerization rates vary from about 30
ps in ethanol to about 80 ps in an ethanol/ethylene glycol
mixture. Identical 12 ps transients are obtained for
MoS2 /DTDCI in hexane and hexadecane solvents, which
have viscosities differing by about a factor of 10. Thus no
solvent viscosity dependence of the 12 ps transient seen in
the MoS2 /DTDCI system, indicating that it not associated
with DTDCI isomerization. If isomerization were involved
in this transient, then a large solvent viscosity effect would
be expected. The photoisomerization involves torsion of the
methylene chain and we conclude that this does not occur
when the dye is adsorbed on the nanoparticles.
Following the 12 ps transient, at both wavelengths there
is a slower, 225 ps decay as well as a long lived, nanosecond
absorbance change. The most important thing to note in Figs.
5 and 6 is that no 42 ps component is observed in the
MoS2/dye transient absorption kinetics. This establishes that
the 42 ps transients seen in the polarized emission kinetics
and in the bare nanocluster absorption kinetics are not related
to the electron dynamics and may therefore be assigned to
hole trapping.
Several kinetic components can be extracted by fitting
the decays in Figs. 5 and 6. This fitting is complicated by the
observation that these samples show weak DTDCI emission,
about 12% of the emission intensity seen in ethanol solution.
This indicates that about 12% of the dyes do not undergo
electron transfer but simply radiate, presumably with their
1.7 ns solution phase lifetime. The 660 nm calculated curve
also has a 12% component of 1.7 ns bleach decay associated
with dyes that do not inject. Inclusion of this 1.7 ns ͑12%͒
component assumes that the extinction coefficient of the ex-
cited state dye is small compared to that of the ground state
dye at these wavelengths. DTDCI has significant emission
intensity at 700 nm, and the 700 nm kinetics show a some-
what larger 1.7 ns component due to stimulated emission.
Despite this complication, both 660 nm and 700 nm curves
clearly show a 225 ps component. The 225 ps bleach recov-
ery transient observed here is significantly shorter than the
275 ps decays observed in the emission polarization and the
bare nanocluster absorption kinetics. It is also of interest to
note that electron trapping causes an absorption decrease in
the bare nanocluster case ͑Fig. 4͒, while the 225 ps transient
in the nanocluster/dye case is an increase of the absorbance
͑i.e., a decrease in the intensity of the bleach͒. This indicates
partial repopulation of the ground state dye from the conduc-
From the difference of the bare nanocluster and nanocluster/
dye conduction band decays, it is possible to estimate the
rates of conduction band reverse electron transfer, kret
.
These values ͑275 ps and 225 ps, respectively͒ give an esti-
Ϫ1
ret
mate for k of about 1.2 ns. This value comes from taking
the difference of two separate rates and is therefore subject to
considerable uncertainty. The above rate constants indicate
that about 20% of the conduction band electrons undergo
reverse electron transfer, while about 80% get trapped. Fol-
lowing electron trapping, reverse electron transfer from trap
states, krЈet , occurs on a longer time scale and is not apparent
in the kinetics presented here.
The results presented here show that electron trapping
occurs much more slowly than hole trapping. It may be pos-
sible to rationalize this result in terms of the trapping reor-
ganization energy and the extent of electron-phonon cou-
pling. The small magnitude of electron-phonon coupling
results from the nature of the bonding in MX2-type
compounds.24 Optical excitation for these materials is largely
a metal centered d-d, nonbonding-to-nonbonding, transition.
As a result, photoexcitation produces very little lattice dis-
tortion and there is very little electron-phonon coupling as-
sociated with band gap excitation. If electron or hole trap-
ping significantly distorts the local lattice, then there is a
considerable reorganization energy associated with this pro-
cess. In the case of WS2 nanoclusters, we recently showed
that electron traps are very shallow ͑a few hundred wave
numbers͒, and hole traps are considerably deeper ͑a few
thousand wave numbers͒.15,16 In a simple exciton model of
the traps, this difference is due to the different carrier effec-
tive masses and the same situation is probably the case in
MoS2 nanoclusters. We speculate that in the case of electron
trapping, the shallow traps along with the lattice distortion
result in small Franck–Condon factors for electron trapping.
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
155.33.16.124 On: Sun, 30 Nov 2014 05:24:31

5455
J. Chem. Phys., Vol. 113, No. 13, 1 October 2000
Electron transfer dynamics in MoS2 clusters
In other words, the significant reorganization energy and
small trapping energy results in an energetic barrier to elec-
tron trapping. Hole traps are much deeper and the barrier
may be smaller in this case. This is basically a Marcus
Theory consideration.30 The barrier for electron or hole trap-
ping is given by
͑3͒ Electron injection from adsorbed DTDCI to the MoS2
nanocluster conduction band occurs with a 12 ps time
constant following dye photoexcitation. This is followed
by a 225 ps decay which is assigned to electron trapping
and reverse electron transfer from the conduction band.
The presence of this 225 ps decay and the absence of a
42 ps decay permits the assignment of the electron trap-
ping decay component seen in the polarized emission
and bare nanocluster transient absorption results. From
these rates, the reverse electron transfer time is inferred
to be about 1.2 ns. This means that about 20% of the
conduction band electrons undergo reverse electron
transfer, while about 80% get trapped.
⌬G†ϭ␭/4 1ϩ⌬G/␭͒2,
͑4͒
͑
where ␭ is the reorganization energy and ⌬G is the trapping
energy. Equation ͑4͒ shows that if ␭ is comparatively large
and ⌬G is small, then there is a significant barrier to trap-
ping. As the trapping energy gets larger and approaches the
same magnitude as the reorganization energy, the barrier gets
smaller and the trapping rate gets larger. We also note that
the electron and hole trapping times reported here are con-
siderably longer than what is observed in other nanocluster
systems such as SnO2, TiO2, Fe2O3, and most notably, CdS
and CdSe.31–40 We speculate that the reason that slow trap-
ping is observed in MoS2 and WS2 nanoclusters is related to
the weak electron-phonon coupling in these materials. How-
ever, electronic factors may also be important in determining
the trapping rates and the slow trapping observed in MoS2
and WS2 nanoclusters may also be due to electronic overlap
considerations. The roles of electronic and vibrational factors
in controlling trapping rates in MoS2 and WS2 nanoclusters
are not clear at this point.
͑4͒ Following electron trapping, reverse electron transfer
from these deep traps is slow and occurs on the nanosec-
ond time scale.
ACKNOWLEDGMENTS
This work was supported by a grant from the Depart-
ment of Energy. M.R.W. also thanks the Foundation for Re-
search, Science and Technology of New Zealand for a New
Zealand Science and Technology Postdoctoral Fellowship
͑Contract No. KSU-901͒.
There have been numerous studies of electron injection
from adsorbed dyes into semiconductors. The injection rates
depend on the energetics, the extent of the electronic cou-
pling and the density of conduction band electronic states at
the injection energy. A wide range of electron transfer times
are observed, ranging from less than a picosecond to hun-
dreds of picoseconds. We are currently studying the electron
injection rates from DTDCI and other cyanine dyes in
closely related nanocluster systems to assess the roles of
each of these factors in determining the electron injection
rates. Those results will be reported in a later paper.
1 For
a
recent and comprehensive review see: Semiconductor
Nanoclusters—Physical, Chemical and Catalytic Aspects, edited by P. V.
Kamat and D. Meisel ͑Elsevier, New York, 1997͒.
2 A. Hagfelt and M. Gratzel, Chem. Rev. 95, 49 ͑1995͒.
3 M. K. Nazeeruddin, A. Kay, I. Rodicio, R. Humprhy-Baker, E. Muller, P.
Liska, N. Vlachopoulos, and M. Gratzel, J. Am. Chem. Soc. 115, 6382
͑1993͒.
4 T. Hannappel, B. Burfeindt, W. Storck, and F. Willig, J. Phys. Chem. 101,
6799 ͑1997͒.
5 S. K. Deb, R. Ellingson, S. Ferrere, A. J. Frank, B. A. Gregg, A. J. Nozik,
N. Park, and G. Schlichthorl, presented at the 2nd World Conference on
Photovoltaic Solar Energy Conversion, July 1998, Vienna, Austria.
6 S. K. Deb, S. Ferrere, A. J. Frank, S. K. Huang, A. J. Nozik, G. Schlich-
thorl, and A. Zaban, presented at the 26th IEEE Photovoltaic Specialists
Conference, October, 1997, Anaheim, California.
7 L. E. Brus, J. Phys. Chem. 90, 2555 ͑1986͒.
8 R. Rosetti, J. E. Ellison, J. M. Gibson, and L. E. Brus, J. Chem. Phys. 80,
4464 ͑1984͒.
IV. CONCLUSIONS
9 M. V. Rama Krishna and R. A. Friesner, J. Chem. Phys. 95, 8309 ͑1991͒.
10 P. E. Lippens and M. Lannoo, Phys. Rev. B 39, 10935 ͑1989͒.
11 H. Gerischer, in Topics in Applied Physics, Vol. 31: Solar Energy Con-
version, edited by B. O. Seraphin ͑Springer, Berlin, 1979͒.
Several conclusions can be drawn from the results pre-
sented here.
12
͑1͒ MoS2 nanoclusters exhibit polarized emission from the
bandedge state and unpolarized emission from trapped
electrons and holes. This polarization difference can be
used to determine the rates of electron and hole trapping.
Such determinations are not possible from the total ͑un-
polarized͒ emission kinetics. We find that hole and elec-
tron trapping times are 42 ps and 275 ps, respectively.
͑2͒ The same decay times seen in the polarized emission
kinetics are also seen in the transient absorption kinetics.
In addition, faster rise and decay components are also
seen in the transient absorption results. These transients
correspond to relaxation to the band edge state, but no
definite assignments can be made at this time.
͑a͒ J. P. Wilcoxon and G. A. Samara, Phys. Rev. B 51, 7299 ͑1995͒; ͑b͒ J.
P. Wilcoxon, P. P. Newcomer, and G. A. Samara, J. Appl. Phys. 81, 7934
͑1997͒.
13 F. Levy, Crystallography and Crystal Chemistry of Materials with Lay-
ered Structures ͑Reidel, Holland, 1976͒, p. 2.
14 Electrons and Phonons in Layered Crystal Structures, edited by T. J.
Wieting and M. Schluter ͑Reidel, Holland, 1979͒.
15 J. M. Huang, R. Laitinen, and D. F. Kelley, Phys. Rev. B ͑in press͒.
16 J. M. Huang and D. F. Kelley, J. Chem. Phys. 113, 793 ͑2000͒.
17 J. M. Huang and D. F. Kelley, Chem. Mater. ͑in press͒.
18 F. Parsapour, D. F. Kelley, S. Craft, and J. P. Wilcoxon, J. Chem. Phys.
104, 4978 ͑1996͒.
19 R. Doolen, R. Latinen, F. Parsapour, and D. F. Kelley, J. Phys. Chem. B
102, 3906 ͑1998͒.
20 C. I. Butoi, B. T. Langdon, and D. F. Kelley, J. Phys. Chem. B 102, 9635
͑1998͒.
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
155.33.16.124 On: Sun, 30 Nov 2014 05:24:31

5456
J. Chem. Phys., Vol. 113, No. 13, 1 October 2000
Chikan et al.
21 B. T. Langdon, V. J. MacKenzie, D. J. Asunskis, and D. F. Kelley, J.
Phys. Chem. B 103, 11176 ͑1999͒.
31 V. I. Klimov, Ch. J. Schwarz, D. W. McBranch, C. A. Leatherdale, and M.
G. Bawendi, Phys. Rev. B 60, R2177 ͑1999͒.
22 M. R. Nimlos, M. A. Young, E. R. Bernstein, and D. F. Kelley, J. Chem.
Phys. 91, 5268 ͑1989͒.
32 V. I. Klimov and D. W. McBranch, Phys. Rev. Lett. 80, 4028 ͑1998͒.
33 N. P. Ernsting, M. Kaschke, H. Weller, and L. Katsikas, J. Opt. Soc. Am.
B 7, 1630 ͑1990͒; Y. Kayanuma, Phys. Rev. B 38, 9797 ͑1988͒.
34 P. Guyot-Sionnest, M. Shim, C. Matranga, and M. Hines, Phys. Rev. B
60, R2181 ͑1999͒.
23 A. Albrecht, J. Mol. Spectrosc. 6, 84 ͑1961͒.
24 R. Coehoorn, C. Haas, J. Dijkstra, C. J. F. Flipse, R. A. deGroot, and A.
Wold, Phys. Rev. B 35, 6195 ͑1987͒; R. Coehoorn, C. Haas, and R. A.
deGroot, ibid. 35, 6203 ͑1987͒.
35 P. Guyot-Sionnest and M. A. Hines, Appl. Phys. Lett. 72, 686 ͑1998͒.
36 D. E. Skinner, D. P. Colombo, Jr., J. J. Cavaleri, and R. M. Bowman, J.
Phys. Chem. 99, 7853 ͑1995͒.
25 W. Y. Liang, J. Phys. C 6, 551 ͑1973͒.
26 A. R. Beal, W. Y. Liang, and H. P. Hughes, J. Phys. C 9, 2449 ͑1976͒; A.
R. Beal and H. P. Hughes, ibid, 12, 881 ͑1979͒.
37 J. J. Cavaleri, D. P. Colombo, Jr., and R. M. Bowman, J. Phys. Chem.
102, 1341 ͑1998͒.
27 Y. Onganer, M. Yni, D. R. Bessire, and E. L. Quitevis, J. Phys. Chem. 97,
2344 ͑1993͒.
38 J. Z. Zhang, R. H. O’Neil, T. Roberti, J. L. McGowen, and J. E. Evans,
Chem. Phys. Lett. 218, 479 ͑1994͒.
28 R. E. Di Paolo, L. B. Scaffardi, R. Duchowicz, and G. M. Bilmes, J. Phys.
Chem. 99, 13796 ͑1995͒.
39 J. Z. Zhang, R. H. O’Neil, and T. W. Roberti, J. Phys. Chem. 98, 3859
͑1994͒.
29 A. K. Chibsov, G. V. Zakharova, H. Gorner, Y. A. Sogulyaev, I. L. Mush-
kalo, and A. I. Tolmachev, J. Phys. Chem. 99, 886 ͑1995͒.
30 R. A. Marcus, J. Chem. Phys. 43, 679 ͑1965͒; Annu. Rev. Phys. Chem.
15, 155 ͑1964͒; J. Jortner and M. Bixon, J. Chem. Phys. 88, 167 ͑1988͒.
40 N. J. Cherepy, D. B. Liston, J. A. Lovejoy, H. Deng, and J. Z. Zhang, J.
Chem. Phys. B 102, 770 ͑1998͒.
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
155.33.16.124 On: Sun, 30 Nov 2014 05:24:31