Molecular clusters of binary and ternary mercury chalcogenides: Colloidal synthesis, characterization, and optical spectra

Article abstract of DOI:10.1021/jp0274684

A series of binary (HgSe) and ternary (HgSe_1-xS_x) mercury chalcogenide clusters are synthesized utilizing a colloidal technique involving the phase separation of metal and chalcogen precursors in the presence of strong Hg(II) coordinating ligands. The clusters vary in size between 2 and 3 nm and possess the cubic zinc blende structure of the bulk. Energy-dispersive X-ray measurements show that the composition of the ternary material can be varied throughout the entire composition range from HgS to HgSe. In all cases, the linear absorption of these binary and ternary species is narrow with well-resolved transitions at both the band edge and at higher energies. Complementary band edge emission is also observed with no apparent deep trap emission. Size- and composition (x)-dependent optical properties of these clusters are investigated using photoluminescence (PL) and photoluminescence excitation (PLE) spectra. In the case of HgSe clusters, the size-dependent behavior of up to four excited states is followed. For HgSe_1-xS_x clusters, where x varies from 0 to 1, a size/composition-dependent progression of up to five excited states is observed.

Full text of DOI:10.1021/jp0274684

5758
J. Phys. Chem. B 2003, 107, 5758-5767
Molecular Clusters of Binary and Ternary Mercury Chalcogenides: Colloidal Synthesis,
Characterization, and Optical Spectra
Masaru Kuno,*,† Keith A. Higginson, Syed B. Qadri,‡ Mohammad Yousuf,§ Sang Ho Lee,†
Benjamin L. Davis,| and Hedi Mattoussi†
NaVal Research Laboratory, 4555 OVerlook AVenue SW, Washington, DC 20375
ReceiVed: NoVember 15, 2002; In Final Form: March 25, 2003
A series of binary (HgSe) and ternary (HgSe1-xSx) mercury chalcogenide clusters are synthesized utilizing a
colloidal technique involving the phase separation of metal and chalcogen precursors in the presence of strong
Hg(II) coordinating ligands. The clusters vary in size between 2 and 3 nm and possess the cubic zinc blende
structure of the bulk. Energy-dispersive X-ray measurements show that the composition of the ternary material
can be varied throughout the entire composition range from HgS to HgSe. In all cases, the linear absorption
of these binary and ternary species is narrow with well-resolved transitions at both the band edge and at
higher energies. Complementary band edge emission is also observed with no apparent deep trap emission.
Size- and composition (x)-dependent optical properties of these clusters are investigated using photolumi-
nescence (PL) and photoluminescence excitation (PLE) spectra. In the case of HgSe clusters, the size-dependent
behavior of up to four excited states is followed. For HgSe1-xSx clusters, where x varies from 0 to 1, a
size/composition-dependent progression of up to five excited states is observed.
Introduction
Cd32S50).9 This work was also extended to make ZnS, ZnSe,
and ZnTe clusters10,11 as well as homologous CdSe clusters12
(Cd4Se6, Cd8Se13, Cd10Se16, Cd17Se28, and Cd32Se50), providing
further examples of the trailing molecular limit in these
materials. Banin has since investigated the evolution of the CdSe
band gap within this molecular limit and has also studied the
nature of the emitting state, showing the importance of the
surface in small semiconductor clusters.13
Over the last twenty years, a great deal of work has gone
into understanding the properties of matter at the nanometer
scale. One class of mesoscopic system receiving much attention
are colloidal semiconductor quantum dots (QDs) or nanocrystals
(NCs).1 Progress in this area has benefited from the dynamic
interplay between advances in synthesis, optical characterization,
and theory.2 Despite this synergy, a major bottleneck in the field
has traditionally been the availability of high quality, crystalline
samples with narrow size distributions, high photoluminescence
efficiencies, and controlled surface chemistries. II-VI materials
in particular have led to new opportunities for better understand-
ing and utilizing the size-dependent optical and electrical
behavior of QDs3,4 as a result of the discovery in the early 90’s
by Murray and others that one could make high-quality QDs
through the thermolysis of pyrophoric organometallic reagents.5,6
Prior to these advances, however, analogous examples of high
quality II-VI nanocrystalline compounds could already be
found in small clusters of zinc and cadmium chalcogenides of
precise molecular weight. These clusters represent the molecular
limit of QDs and are also some of the first examples of colloidal
nanoparticles. They contain less than 100 atoms and typically
lie in the size range between 1 and 2 nm. For example, in 1982
Dance and co-workers were able to isolate discrete clusters of
CdS and ZnS of the stoichiometry M10S20 (M ) Zn, Cd).7 The
core of such Cd and Zn compounds was found to be congruent
with the unit cell of the cubic sphalerite structure, making them
nanometer-sized (tetrahedral) fragments of the bulk. Larger CdS
clusters were subsequently isolated by Wang and Herron
(Cd32S50)8 as well as by Weller and co-workers (Cd17S30,
In the case of II-VI mercury chalcogenides, very little work
exists to date on either clusters or QDs, although interest in
these materials abounds. Bulk mercury chalcogenides are
semimetals widely used in infrared sensing applications. Alloys
of Hg compounds (Hg1-xCdxTe, for example) are particularly
important systems because of the wide range of tunable optical,
electrical, and magnetic properties, achieved through compo-
sitional tuning (x) of this material, from pure semimetallic Hg-
(S,Se,Te) to narrow gap Hg1-xCdx(S,Se,Te) and ultimately to
wider gap semiconductor Cd(S,Se,Te). At the same time, such
material tunability can also be explored by making QDs or
clusters of these compounds, exploiting well-known quantum
confinement effects that occur when the physical size of the
NC is smaller than the bulk exciton Bohr radius. Furthermore,
the ease of compositional tuning suggests that a similar
parameter space can be investigated, not only through size, but
also through composition, with the creation of ternary mercury
chalcogenide QDs/clusters leading to tunable absorbing and
emitting species spanning anywhere from the visible into the
far-infrared. Such compounds have potential use as visible/
infrared fluorescent tags, lasing/amplifying elements at 1.3 and
1.55 µm, and in remote sensing applications.14
While a great deal of work has been done to explore
organometallic precursor routes to QD synthesis in other II-
VI and III-V materials,15-21 analogous preparations for II-VI
mercury chalcogenide QDs or clusters have not been reported.
This likely stems from the toxicity and dangers associated with
corresponding organometallic Hg(II) precursors in NC prepara-
* Corresponding author. Current address: University of Notre Dame,
Department of Chemistry and Biochemistry. E-mail: mkuno@nd.edu.
† NRL Optical Sciences Division.
‡ NRL Materials Physics Division.
§ George Washington University, Washington, DC.
| NRL Electronic Science and Technology Division.
Current Address: Triton Systems Inc., Chelmsford, MA.
10.1021/jp0274684 CCC: $25.00 © 2003 American Chemical Society
Published on Web 05/17/2003

Binary and Ternary Mercury Chalcogenides
J. Phys. Chem. B, Vol. 107, No. 24, 2003 5759
tions that involve the high-temperature decomposition of
pyrophoric reagents in a mildly coordinating solvent. Further-
more, significant differences exist in the reactivity, as well as
chemistry, of mercury relative to other group II elements,22
making such approaches potentially problematic. To illustrate,
in the synthesis of mercury chalcogenide NCs, reactions between
metal and chalcogen precursors are commonly observed to
progress very quickly at room temperature, often bypassing the
nanocrystal stage, to yield bulk or bulklike material. At best,
NCs with broad size distributions and poorly resolved optical
transitions are obtained. This has been seen previously23 and
more recently in our attempts to make mercury chalcogenide
QDs.24 By contrast, reactions of Cd and Zn compounds with
corresponding chalcogen precursors are generally well con-
trolled, even at temperatures >200 °C, as evidenced by a variety
of routes now available for making high-quality zinc- and
cadmium-based QDs.5,6,18-21
quality QDs (5% size distributions, >10% quantum yields,
highly crystalline) are produced, comparable to those obtained
from more traditional organometallic methods. Such preparations
represent alternative approaches to making high-quality NCs
and possess distinct benefits over previous syntheses.5,6 Ad-
vantages include tunable reactivity achieved by varying the
overall ligand concentration, use of cheaper and safer metal
precursors, and, most importantly, the improved reproducibility
of the preparations.35-37
In this vein, we have explored the use of many Hg(II)
coordinating ligands covering a vast range of stability constants
(K).40 Ligands with log K values between 1 and 50 were
investigated. Those with stability constants less than 1017 such
as fatty acids (log K 6-9) and amines (log K 10-18) resulted
in uncontrolled reactions between Hg(II) and common chalcogen
precursors, leading to the precipitation of bulk or bulklike
material. At the other limit where very strongly coordinating
ligands, such as thiols (log K > 30), phosphines (log K 30 to
37), or phosphine oxides (log K > 30), are used, crystal
nucleation is severely retarded, requiring elevated temperatures
to help dissociate the metal-ligand complex. Unfortunately,
often associated with this combination of strong ligands and
high temperatures is the reduction of Hg(II) to Hg(0), which
competes with the desired reaction. Such dissociation pathways
have been documented for Hg(II)-thiolate complexes41 and
suggest that reduction is a favored decomposition pathway
because of mercury’s high reduction potential.42
What mercury chalcogenide NC syntheses have been reported
largely deal with solvothermal or solvent-based approaches to
making these compounds. These approaches entail sonochem-
istry,25 microwave-assisted heating,26,27 and the thermolysis of
common mercury precursors.28,29 Other methods make use of
metal ion impregnated Langmuir Blodgett films30 or other
restricted environments such as vesicles31 or sol gel networks32
to grow HgS QDs. Most, if not all, of these preparations lack
the controlled growth kinetics and narrow size distributions
characteristic of more developed organometallic syntheses such
as those for CdSe, InP, and InAs.5,6,15-17 To date, the most
successful syntheses for mercury chalcogenide nanomaterials
appear to be those developed by Weller and Fenske for QDs
and clusters, respectively. Weller et al. have synthesized HgTe
nanocrystals using an aqueous preparation wherein mercury salts
are exposed to hydrogen telluride gas bubbled through the
reaction mixture.33 HgTe QDs between 3 and 6 nm diameter
are obtained, emitting with high integrated yields broadly
between 800 and 1400 nm. HgSe clusters (Hg6Se12, Hg10Se16,
Hg32Se50) have been isolated by the Fenske group and in
particular, the Hg32Se50 species was found to have a 2.4 nm
edge length with structured absorption beginning at about 546
nm (2.27 eV).34 Both preparations, however, suffer from the
same problems which generally plague the mercury chalco-
genides, namely, the instability, high reactivity, and tendency
for uncontrolled reactions leading to the parent bulk, rather than
nanocrystalline, material. For example, continued growth of
HgTe QDs was observed, evidenced by a continuous red-shift
of the absorption over a period of days.33a Likewise, Hg32Se50
clusters in THF or toluene, when warmed to room temperature,
undergo rapid growth leading to a black precipitate,34b which
is likely bulk HgSe or large NCs with broad size distributions,
similar to those observed by Brennan.23
To circumvent these problems, we have found alternative
routes for making high-quality mercury chalcogenide QDs by
employing not only strong Hg(II) coordinating ligands, but also
by segregating the metal and chalcogen precursors in different
phases of the reaction mixture. In our previous work,24 we have
used a common micellar medium (bis(2-ethylhexyl)sulfosucci-
nate (AOT) in cyclohexane) to achieve this phase separation of
both mercury and chalcogen precursors. The general micellar
approach entails the use of a surfactant, which has both
hydrophobic and hydrophilic regions and organizes in alkanes
into water-bearing and water-excluding regions. Within the
water-bearing regions, aqueous metal precursors can be seques-
tered, making the micelles a compartmentalized source of metal
ions.43 Using this approach, we have been able to make high-
quality HgS and HgSe QDs controllably in the presence of
thioglycerol, a strong Hg(II) coordinating ligand. In the case of
HgS, small QDs between 1 and 5 nm in diameter are obtained,
which absorb and emit in the visible region of the spectrum.
When passivated with Cd or Zn, the emission quantum yield
(QY) increases dramatically and there is no further growth
observed over a period of weeks.24
In this paper, we report a similar synthesis of small binary
and ternary mercury chalcogenide clusters, which approach the
molecular limit of these materials. The synthesis is a generaliza-
tion of the above precursor phase separation and strong ligand-
mediated approach. Rather than traditional micellar media,
however, a nonionic surfactant in an organic phosphine is used.
Wide-angle X-ray diffraction studies of the resulting particles
show that they possess the cubic zinc blende phase of the parent
bulk material and range in size from 2 to 3 nm in diameter.
Energy-dispersive X-ray measurements show that the stoichi-
ometry of the ternary material (x, HgSe1-xSx) can be varied
throughout the entire composition range from x ) 0 to x ) 1.
Preliminary characterization of the surface of these particles is
conducted through 1H and 31P nuclear magnetic resonance
(NMR) experiments. The optical properties of these clusters are
subsequently investigated using photoluminescence (PL), pho-
As part of a search for new infrared emitting NCs, we have
recently explored possible “ligand-mediated” routes to making
mercury chalcogenide QDs using approaches recently brought
to attention by Peng,35-37 Murray,38 and others39 for making
high-quality CdS, CdSe, CdTe, and PbSe nanocrystals. In their
studies, these groups have illustrated that through the choice of
an appropriate metal binding ligand, such as long chain fatty
acids or alkylphosphonic acids, crystal nucleation and growth
can be controlled by the ligand, which modulates the concentra-
tion of reactive species in the immediate vicinity of the growing
QD. Control over the particle growth rate is achieved by
retarding the heterogeneous reaction rate to various degrees with
more (or less) strongly binding metal ligands or by increasing
(or decreasing) their steric hindrance. In such a manner, high-

5760 J. Phys. Chem. B, Vol. 107, No. 24, 2003
Kuno et al.
toluminescence excitation (PLE), and low-temperature time-
resolved PL decay experiments. The evolution of states as a
function of cluster size and/or composition is investigated. In
the case of HgSe, the size-dependent behavior of up to four
excited states is followed; whereas for HgSe1-xSx clusters, where
x varies from 0 to 1, a size/composition-dependent progression
of up to five excited states is observed.
room temperature with no significant precautions to prevent
exposing the solution to air. However, in general, a nitrogen
atmosphere can be used to prevent any possible combustion of
TOP or TBP.
To isolate HgSe1-xSx clusters, excess acetone is added to the
growth solution. The amount of acetone required to remove most
(all) of the material from solution depends on the sulfur content
(x); samples having more sulfur require more acetone to
precipitate from solution. In practice, 30-40 mL of acetone is
sufficient to recover nearly all the material in a given prepara-
tion. The suspensions are then centrifuged and a colored
precipitate is recovered. This precipitate can then be re-dispersed
in toluene or chloroform, yielding a strongly colored solution
and residual insoluble material. Solutions can subsequently be
centrifuged or passed through glass wool to yield a homoge-
neous solution. Samples with x < 0.4 are much less soluble in
toluene than their counterparts with x > 0.5. It should also be
noted that samples washed more than once with acetone or
methanol or those initially exposed to too much acetone
generally become insoluble in any solvent.
Experimental Section
The following chemicals were purchased from Aldrich and
were used without further purification: mercury(II) acetate
(HgAc2), bis(2-ethylhexyl)sulfosuccinate (AOT), thioglycerol,
and polyoxyethylene 4-lauryl ether (Brij30). Bis(trimethylsilyl)-
sulfide ((TMS)2S) and bis(trimethylsilyl)selenide ((TMS)2Se)
were purchased from Acros and used as received. Technical
grade trioctylphosphine (TOP) and tributylphosphine (TBP)
were purchased from Fluka and likewise used without further
purification. Deionized water was obtained from a Millipore
MilliQ unit. Care should be exercised to avoid exposure to
mercury compounds. In addition, TBP is pyrophoric, and
prolonged exposure to air should be avoided.
A stock solution of the mercury precursor is prepared by
dissolving 100 mg of HgAc2 (3 × 10-4 mol) in 3 mL of
deionized water, and then adding 27 µL (3 × 10-4 mol) of
thioglycerol. Thioglycerol should be added last, because it reacts
immediately with HgAc2 to produce an insoluble black pre-
cipitate. The stock solution keeps for several weeks with
eventual discoloration and precipitation of a gray/black powder
determined to be HgS from energy-dispersive X-ray spectros-
copy (EDXS). In the current synthesis, the equimolar amount
of thioglycerol prevents the hydrolysis of Hg(II) and eventual
precipitation of HgO that normally occurs in aqueous solutions.
It should be noted that polar solvents other than water can also
be used to create the above mercury stock solution. For example,
dimethylformamide (DMF) works equally well with no obvious
difference in the overall behavior of the synthesis.
Chalcogen stock solutions are prepared by dissolving 2 µL
(1 × 10-5 mol) of (TMS)2S in 10 mL hexanes and likewise 2.6
µL (1 × 10-5 mol) of (TMS)2Se into 10 mL of hexanes. In
either case, the actual amount of hexane used does not appear
to be critical and only serves to dilute the stock solution
sufficiently such that subsequent reactions between the metal
and chalcogen are controlled. To prepare binary HgSe clusters,
only the above (TMS)2Se solution is used in the synthesis. For
the preparation of ternary compounds, appropriate amounts of
each chalcogen solution are combined to achieve an effective
S/Se solution with the desired sulfur content (x, HgSe1-xSx).
For example, to target x ) 0.3, 3 mL of the (TMS)2S stock
solution is mixed with 7 mL of the (TMS)2Se stock solution.
In a typical preparation, 8 mL of TOP is mixed with 2 mL
of Brij30 in a three-neck flask. Once clear, 0.11 mL (1 × 10-5
mol Hg) of the mercury stock solution is introduced. The
solution is stirred until clear. The actual amount of TOP used
does not appear to be critical to the synthesis, provided it
remains in excess. Likewise, the actual amount of Brij30 also
does not appear to be critical, but amounts greater than 1 mL
should be used to better solubilize the aqueous Hg(II) solution.
In addition, Brij30 can also be replaced by 3.4 g (0.01 mol) of
AOT. However, we found this to be inconvenient as moderately
high temperatures (∼100 °C) were needed in order to solubilize
AOT in TOP. In all cases, when the Hg(II)-containing micro
emulsion is clear, the chalcogen solution is brought out of the
glovebox and is introduced into the flask at a rate of ∼1 mL/
min using a syringe pump. All preparations are conducted at
NMR samples are prepared by precipitating the clusters with
acetone, centrifuging the suspension to recover the precipitate
and immediately redissolving the recovered material in deuter-
ated toluene. Proton (1H) and phosphorus (31P) NMR spectra
1
are acquired on a Bruker AC300 spectrometer (300 MHz H;
121 MHz 31P) equipped with a multinuclear probe. EDXS
samples are prepared by washing the recovered precipitate
multiple times with methanol (or acetone) to remove any excess
surfactant. The precipitate is then dried under vacuum and
dispersed onto conductive carbon tape mounted to a scanning
electron microscope (SEM) sample holder. EDXS measurements
are performed with an EDAX energy-dispersive X-ray spec-
trometer attached to a Leo 1550 SEM operating at an accelerat-
ing voltage of 10 kV. Wide-angle X-ray (WAXS) samples are
prepared in the same manner and are occasionally sonicated
during the washing procedure to assist in removing any excess
surfactant. Diffraction experiments are conducted using Cu KR
radiation derived from a 12 kW rotating anode generator
(Rigaku) on samples mounted to silicon (001) substrates. Peak
positions and their full width at half-maximum were obtained
using a commercial peak fitting program.
Linear absorption experiments are carried out with a diode
array UV-visible spectrometer (Hewlett-Packard, HP 8453),
while photoluminescence (PL) and photoluminescence excitation
(PLE) experiments are conducted at room temperature using a
Spex Fluorolog fluorimeter. During PL experiments, samples
are typically excited at 410 nm (∼3 eV) with light from a 400
W Xe arc lamp passed through a double monochromator.
Emitted light is detected with a second double monochromator
and a Hammamatsu R928 photomultiplier tube. A long pass
colored glass filter is often used to suppress unwanted solvent
Raman contributions to the emission spectra. PLE experiments
are conducted by monitoring the intensity of the room temper-
ature emission at three fixed wavelengths, generally at positions
red of the emission maximum, at 2/3 height, 1/2 height,and on
the shoulder of a second feature accompanying the main
emission peak, while scanning the excitation wavelength to the
blue of this position.
Time-dependent fluorescence lifetimes are measured using
a time-correlated single-photon counting (TCSPC) apparatus.
The excitation source for TCSPC measurements is a mode-
locked Nd:YLF laser, operating at 100 MHz. The second
harmonic (527 nm) of the Nd:YLF synchronously pumps a
cavity-dumped dye laser using Rhodamine 590 as the lasing

Binary and Ternary Mercury Chalcogenides
J. Phys. Chem. B, Vol. 107, No. 24, 2003 5761
medium. A single plate birefringent filter is inserted within the
dye laser to generate laser oscillation at 600 nm. The variable
repetition rate (20-30 kHz) cavity-dumped output is then
frequency-doubled (300 nm; 4.13 eV) using a 3 mm KDP
crystal. Samples are inserted into a He flow through cryostat
(Janis) and cooled to 20 K. Emission from the sample is
spectrally filtered with a monochromator and detected with a
Hamamatsu R2809U-11, water-cooled, micro channel plate. The
instrument response is measured by detecting scatter off of a 1
cm path length cell and has a full width at half-maximum of
∼100 ps.
Results and Discussion
As described in the Introduction, previous attempts to
synthesize mercury chalcogenide nanocrystals have suffered
from problems associated with the high reactivity of Hg(II) and
ensuing reactions to produce nanocrystals have tended toward
broad size distributions and/or bulklike material. Our previous
work has focused on exploring possible ligand-mediated prepa-
rations for HgX (X ) S, Se) NCs and has led to the development
of small HgS and HgSe QDs emitting in the visible region of
the spectrum. The strategy used to make such QDs entails not
only strong Hg(II)-coordinating ligands but also the phase
separation of both metal and chalcogen precursors. In this
manner, reactions between Hg(II) and S2-/Se2- are controlled,
allowing the synthesis of mercury chalcogenide QDs with
improved size distributions. Here we describe a generalization
of the above approach using a surfactant in trialkylphosphine
mixture to make small binary and ternary mercury chalcogenide
NCs or clusters, which also absorb and emit in the visible region
of the spectrum.
Figure 1. Linear absorption (solid line) and complementary band edge
emission (dashed line) of HgSe1-xSx clusters with 0.1 < x < 0.8.
reported spectra are for the equilibrium structure of each
HgSe1-xSx cluster, taken after the complete, stoichiometric
addition of both chalcogen precursors. Prior to this, the
absorption spectrum of each sample is nearly identical but less
intense, suggesting a smaller concentration of the final equi-
librium species. Although it is possible that the species formed
at nonstoichiometric chalcogen additions have different sizes,
the small (1-2 nm) shift of the absorption, which occurs during
addition, and the much more dramatic rise of the sample optical
density suggest an increase in concentration rather than particle
growth.
The emission of these ternary species is weak at room
temperature and is red-shifted relative to the band edge
absorption. It is often obscured by solvent Raman contributions,
making it necessary to measure the spectrum under concentrated
conditions, achieved by isolating the sample from the growth
solution and re-dispersing it in toluene.44 The band edge
emission has an asymmetric shape with a shoulder suggesting
either a phonon progression toward lower energies or emission
from a surface-related trap state. In Figure 1, the energy
difference or “global” Stokes shift between the peak of the band
edge absorption and band edge emission appears to be composi-
tion (x)-dependent, becoming smaller with decreasing overall
sulfur content. Values of the global shift in Figure 1 range from
100 to 220 meV (x ) 0.19 to x ) 0.74) and represent an upper
limit to the true shift given the potential influence of emission
reabsorption.
Figure 2a shows the absorption and emission of HgSe clusters
made in the same manner. As with HgSe1-xSx clusters, the band
edge absorption is well-resolved with structure at higher
energies. In particular, one sees a second state followed by a
shoulder to the blue, consisting of several unresolved transitions.
The complementary band edge emission is also narrow and red-
shifted relative to the absorption. A value of the global Stokes
shift is 87 meV and is again an upper limit due to potential
reabsorption effects.
The spectra in Figure 2a are for stable HgSe clusters with a
band edge absorption maximum at 595 nm (2.08 eV). This 2.08
eV species is unusually stable, and evidence for its thermody-
namic stability lies in the observation that no further growth is
In this preparation, Hg(II) ions passivated by thioglycerol are
phase-segregated into water-bearing regions of a polyoxyeth-
ylene ether surfactant (Brij30; alternatively called C12E4 where
E ) OCH2CH2). The remainder of the growth solution consists
of either trioctylphosphine (TOP) or tributylphosphine (TBP),
which acts as a coordinating solvent for the growing mercury
chalcogenide clusters. To initiate growth, common chalcogen
precursors ((TMS)2S, (TMS)2Se) are introduced into the organic
phosphine phase with a syringe pump. Upon introduction of
the selenium precursor, slow color changes occur over the course
of minutes, with a gradual evolution of the solution’s color from
light yellow to burgundy. When mixtures of sulfur and selenium
precursors are added to make ternary clusters, very little of this
color progression is evident. Instead the optical density of the
reaction solution increases with only a small spectral shift of
the sample absorption spectrum. The reaction to make binary/
ternary mercury chalcogenide clusters likely stems from the
decomposition of (TMS)2S and (TMS)2Se into corresponding
S
2- and Se2- ions, which then migrate through the phase barrier
to react with sequestered Hg(II) ions. Both the phase barrier
and Hg(II) coordinating species, TOP (or TBP) and thioglycerol,
modulate subsequent reactions leading to the controlled growth
of both HgSe and HgSe1-xSx clusters.
Evidence for the vastly improved growth kinetics can be seen
in the UV-visible absorption spectra of the resulting particles.
Figure 1 shows both the band edge absorption and emission of
ternary HgSe1-xS clusters, isolated from the growth solution
x
with acetone and re-dispersed in toluene. The stoichiometry of
the introduced chalcogen solution is varied to achieve values
of x between 0.1 and 0.8, spanning nearly the entire composition
range of the alloy. Sharp structured absorption is observed in
each case with a clearly resolved band edge transition followed
by a weaker “second” state and finally a prominent shoulder
consisting of unresolved transitions at higher energies. The

5762 J. Phys. Chem. B, Vol. 107, No. 24, 2003
Kuno et al.
Figure 3. Wide-angle X-ray diffraction spectra of HgSe1-xSx clusters.
(a) HgSe0.7S0.3, (b) HgSe0.4S0.6. Stick spectra for bulk HgSe and â-HgS
are provided both above and below the experimental patterns for
comparison purposes.
energies. As more chalcogen is added, both the second and third
states diminish in intensity while the first state gains strength,
becoming the dominant feature in the spectrum. The similarity
of spectrum 1 (Figure 2b) to the absorption spectrum of Hg32-
Se50 clusters34 as well as their joint dissimilarity to the 2.08 eV
species (Figure 2a) supports the notion that different sized HgSe
clusters are being generated during the nonstoichiometric phase
of the current synthesis.
Figure 2. (a) Linear absorption (solid line) and complementary band
edge emission (dashed line) of stable 2.08 eV HgSe clusters. (b)
Progression of HgSe cluster linear absorption during growth, numbered
from bottom to top.
Mercury chalcogenide clusters can be isolated from the
growth solution by adding excess acetone. They can subse-
quently be re-dispersed in a suitable solvent such as toluene or
chloroform. In the case of HgSe1-xSx clusters, samples in toluene
or chloroform without excess TOP show evidence of continued
growth over the course of several days to a week with eventual
darkening of the solution and precipitation of a black/brown
residue. In the case of HgSe clusters, only the stable 2.08 eV
clusters species can be isolated from solution with acetone and
re-dispersed in toluene or choloroform. They exhibit comparable
instability toward growth over long periods of time in the
absence of excess TOP. Smaller HgSe clusters that are seen at
nonstoichiometric chalcogen additions (Spectra 2-8, Figure 2b)
cannot be isolated in this fashion, however. Although the
addition of large amounts of acetone does cause partial
precipitation of the smaller species, what is often recovered has
generally evolved into the 2.08 eV species. The inability to
isolate small HgSe clusters therefore represents a problem for
studying their photophysics because their room-temperature
emission is weak and often obscured by the overwhelming
Raman signal of TOP.44
Wide-angle X-ray scattering (WAXS) spectra of isolated
HgSe1-xSx clusters confirm the presence of a ternary solid
solution, showing a single pattern for the alloy rather than two
discrete, superimposed, patterns for HgS and HgSe. Slow
precursor addition rates and high chalcogen dilutions likely
compensate for any major differences in reactivity of (TMS)2S
or (TMS)2Se toward Hg(II) that would otherwise allow the
nucleation and growth of two compositionally distinct cluster
species. Figure 3 shows WAXS patterns for HgSe1-xSx samples
with predicted x values of approximately (a) 0.3 and (b) 0.6
(see EDXS discussion below). Three distinct peaks with Miller
indices of (111), (220), and (311) are identified for each sample.
Based on the positions of the three assigned reflections, extracted
lattice parameters are a ) 6.0142 ( 0.0255 Å (Figure 3a) and
a ) 5.9461 ( 0.0115 Å (Figure 3b). For comparison, the lattice
parameters for â-HgS and HgSe are 5.8517 and 6.085 Å,
observed beyond this “magic” size. Addition of more mercury
and/or chalcogen precursor only increases the concentration of
the 2.08 eV species. Similarly, rapid injection of (TMS)2Se,
rather than its slow addition through a syringe pump, simply
results in the immediate formation of this stable compound.
Temperature jumps such as heating the reaction mixture from
room temperature up to 100 °C may be used to partially
overcome this bottleneck, and push the growth to a second
“magic” size with a band edge absorption close to 628 nm (1.97
eV). This is generally accompanied by a dramatic color change
of the solution from red to green. Growth to the 1.97 eV species,
however, is not complete and a bimodal absorption spectrum
is typically observed with a fraction of clusters at 1.97 eV and
the rest at 2.08 eV. Alternatively, injection of the chalcogen
precursor into the reaction mixture at elevated temperatures
(<100 °C) promotes growth beyond 2.08 eV, but again with
the same apparent bimodal distribution of resulting species. In
addition, increasing the temperature of the reaction mixture often
causes mercury chalcogenide clusters to precipitate from solu-
tion, as do repeated precipitation/resuspension steps, making it
problematic to push the growth beyond 2.08 eV in this manner.
Figure 2b shows the absorption spectrum of binary HgSe
clusters during addition of the chalcogen precursor. Unlike
HgSe1-xSx clusters, an evolution of the absorption spectrum
(numbered from bottom to top) is observed as they converge
to the 2.08 eV magic size and complete (1:1) addition of the
selenium precursor. At the earliest measured time (Spectrum 1
in Figure 2b), the absorption spectrum shows three well-resolved
transitions. The energy of these three states (2.23, 2.40, and
2.61 eV) nearly coincide with those seen in the absorption
spectrum of Hg32Se50 clusters (2.27 eV, 2.48 eV, and ∼2.68
eV).34 However, the third state (2.61 eV) is markedly pro-
nounced whereas in the Hg32Se50 absorption spectrum it is weak
and rapidly evolves into a continuum of states toward higher

Binary and Ternary Mercury Chalcogenides
J. Phys. Chem. B, Vol. 107, No. 24, 2003 5763
Figure 4. Composition of HgSe1-xSx clusters as determined by energy-
dispersive X-ray and WAXS measurements. The solid black line is
the theoretical 1:1 line representing uniform introduction of the added
sulfur. WAXS results are shown by the asterisks.
respectively. By interpolating between the lattice parameters of
either binary compound (â-HgS or HgSe), the ternary composi-
tion is estimated to be (a) HgSe0.7S0.3 and (b) HgSe0.4S0.6 using
Vegard’s Law, consistent with the composition measured via
EDXS as described below. An average crystallite size is
determined to be 3 and 2.5 nm, respectively, using the Scherrer
equation, indicating relatively small changes in cluster size with
large changes in composition.
To confirm the composition of HgSe1-xSx and HgSe clusters,
energy-dispersive X-ray (EDXS) measurements were conducted
on a series of particles. Samples were probed at multiple
locations to generate an average stoichiometry for the ensemble.
Figure 4 shows the results of the measurements for HgSe1-xSx
samples with values of x between 0 and 1. Results obtained
from WAXS experiments described in Figure 3 are denoted by
asterisks, indicating good agreement between both techniques.
In the figure, the solid black line represents the ideal 1:1 sulfur
added/incorporated case. The composition of the ternary clusters
is sulfur-rich over most reaction conditions, particularly for
situations with low overall sulfur content (x < 0.5). This
suggests that S2- competes more aggressively than Se2- in
reactions with Hg(II). Still, the entire composition range of the
HgSe1-xSx solid solution can be spanned using the above
synthesis. In the case of HgSe clusters, EDXS analysis shows
the stoichiometry of the samples to be nearly 1:1 in good
agreement with the initial concentration of Hg(II) and Se2-
precursors.
Nuclear magnetic resonance (NMR) experiments are con-
ducted to investigate the surface of these clusters and in
particular to determine the identity of the passivating ligands.
For ternary clusters, both proton (1H) and phosphorus (31P) NMR
spectra are shown in Figure 5a,b. While thioglycerol may act
to passivate resulting HgSe1-xSx clusters, both TOP and Brij30
are in excess by factors greater than 103 and 102, respectively,
suggesting that these latter two compounds are the dominant
surface coordinating agents. It should be noted that TOP oxidizes
to trioctylphosphine oxide (TOPO) under ambient conditions
and this must also be considered as a (potential) passivating
agent. Furthermore, technical grade TOP (90%) contains a
number of impurities, which may act to passivate the clusters
as well.
Figure 5. (a) 1H and (b) 31P NMR data of HgSe1-xSx clusters in
deuterated toluene. The inset in (b) is the proton decoupled spectrum
of the doublet.
0.92 ppm relative to TMS) is problematic due to the overlapping
contributions from TOP/TOPO/TOP impurities. The Brij30
contribution to these overlapping multiplets, however, can be
estimated through its downfield (3.4 ppm) component. Using
this, we quantify its contribution to the experimental HgSe1-xSx
1H spectrum and determine that the ratio of phosphines/
phosphine oxides to Brij30 is 15:1. This shows that TOP, TOPO,
and/or TOP impurities are in excess by at least an order of
magnitude, suggesting that these phosphine/phosphine oxide
species predominantly passivate the surface of these clusters.
To further identify these ligands, 31P NMR experiments are
performed on solutions of HgSe1-xSx clusters. Chemical shifts
are measured relative to an internal standard of H3PO4 (0 PPM),
and spectra are acquired with and without broadband proton
decoupling. The phosphorus spectrum in Figure 5b shows the
presence of a number of peaks both upfield and downfield
relative to H3PO4. In particular, a peak at -32 ppm can be
identified as free, unbound TOP. Likewise, a peak at 45 ppm
can be assigned to free, unbound TOPO with a peak at 55 ppm
arising from an impurity in the technical grade TOP. Of special
relevance are the two peaks at 33 and 29 ppm of nearly equal
intensity. To see whether the peaks are actually a doublet, 31P
NMR spectra were proton decoupled. The results of broadband
proton decoupling show that the two peaks do indeed converge
into a single peak centered at 31 ppm (inset) and the extracted
coupling constant is 440 Hz, in excellent agreement with
Proton NMR spectra of isolated HgSe1-xSx clusters in
deuterated toluene show the presence of Brij30 as a multiplet
at 3.4 ppm, which is assigned to the ethylene protons in the
repeat unit of the polyoxyethylene ether. Identification of Brij30
through protons further upfield (multiplets at 1.52, 1.28, and
1
45
reported values for JPH
.
This, along with the appearance of
the same proton-coupled doublet in the 31P NMR of technical
grade TOP (described below), provides direct evidence that at

5764 J. Phys. Chem. B, Vol. 107, No. 24, 2003
Kuno et al.
least one of the impurities in TOP has a phosphorus-hydrogen
bond. Possible species include primary or secondary phosphines
(R2PH, RPH2) or phosphinic acids.
This proton-coupled species exists as a minority impurity in
technical grade TOP with a 31P resonance at 29 ppm. Majority
impurities exhibit peaks at 54 and 133 ppm. In the 31P NMR of
HgSe1-xSx clusters, however, the doublet is shifted slightly
downfield to 31 ppm and, most significantly, appears as a sizable
contribution to the overall spectrum. This enhancement can be
rationalized if primary or secondary phosphines or phosphinic
acids have significantly stronger binding affinities to HgSe1-xSx
clusters than TOP, causing the 31P spectrum of isolated clusters
to be enriched in this proton-coupled impurity. The slight
downfield shift can be explained by an increase in nuclear
shielding upon coordination with mercury. It should be pointed
out that phosphine-based impurites have been shown to be
critical ingredients in mediating the synthesis of CdSe QDs,35-37
and has led to the recent development of ligand-mediated
nanocrystal syntheses.
No broad phosphorus resonance are seen in Figure 5b, unlike
similar experiments conducted on TOP/TOPO passivated CdSe
QDs.46 All 31P NMR transitions for the mercury chalcogenide
clusters are narrow and have chemical shifts similar to that
observed for each compound free in solution. This suggests that
all passivating ligands are weakly bound to the surface of the
clusters, exchanging readily with the solution such that the net
chemical environment is relatively homogeneous. This hypoth-
esis is supported by the high sensitivity of the cluster solubility
to repeated washings as well as by the observation of long-
term instability in cluster suspensions, which do not contain
excess TOP/TOPO or TOP impurity. An alternative explanation
for the narrow NMR transitions is the presence of well-defined
binding sites on the surface of the small clusters, unlike the
inhomogeneous surface of larger 2-6 nm diameter CdSe QDs.
However, there do not appear to be significant differences in
the chemical shifts for each ligand, particularly for the impurity
doublet, “free” versus “bound” positions, making this scenario
unlikely.
Figure 6. Photoluminescence (dashed line) and photoluminescence
excitation spectra (solid line) of HgSe1-xSx clusters where (a) x ) 0.19,
(b) x ) 0.35, (c) x ) 0.51, and (d) x ) 0.58. In all cases, the emission
intensity as a function of excitation energy was monitored at the
positions denoted by the vertical lines. Spectra are offset for clarity.
blue of the first transition is resolved. This state can also be
seen in samples with x < 0.5 and in Figure 6a,b appears as a
small shoulder on the blue edge of the first peak. The spectra
in Figure 6 are representative of all samples investigated and,
in particular, those in Figure 6a,b (Figure 6c,d) are characteristic
of the line shapes/intensities seen in samples with x < 0.5 (x >
0.5).
Apart from improving the spectral resolution, the PLE spectra
in Figure 6 establish that the observed luminescence is directly
related to HgSe1-xSx clusters rather than to other extrinsic
sources of emission. Previous PLE experiments on CdSe
clusters13 have shown that emission from such particles is
occasionally complicated by an admixture of discrete sizes, for
example Cd17Se30 and Cd8Se13, within the same sample. In the
case of ternary mercury chalcogenide clusters, such mixtures
could arise from the presence of compositionally distinct species
(e.g., HgS or HgSe) and/or distinct sizes within a given sample.
However, no substantial intensity changes occur when the
emission energy being monitored is altered, suggesting that a
single HgSe1-xSx species is the source of the emission. This is
corroborated by the WAXS data, which indicate the successful
preparation of the ternary alloy.
A primary goal of the current investigation is to learn more
about the optical properties of mercury chalcogenide clusters
and/or QDs. To this end, photoluminescence excitation (PLE)
experiments are conducted on binary and ternary mercury
chalcogenide clusters. This technique is a “size-selective” optical
method in the sense that a small fraction of the size-broadened
distribution of emitting states is probed; it has previously been
used to reveal states in the linear absorption of various NCs,
which were otherwise obscured by the inhomogeneous size
distribution present in most samples. Figure 6 shows the
resulting PLE spectra of four HgSe1-xSx samples with x varying
from 0.19 to 0.58. In all cases, PLE spectra of each sample
were taken by monitoring the emission intensity at three different
energies (denoted by vertical lines in the accompanying emission
spectrum) while scanning the excitation wavelength to the blue
of this position. These energies generally correspond to locations
at 2/3 height, 1/2 height, and on the broad, low-energy shoulder
of the main peak, effectively sampling the entire spectral profile
of the emission.
Noticeable differences and improvements in the PLE spectra
of all samples are apparent relative to the absorption spectra in
Figure 1. Most prominently, the third shoulder in Figure 1,
consisting of several overlapping transitions, is clearly resolved
into at least two distinct transitions shown by the states centered
near 3.5 eV in Figure 6a,b. Furthermore, in HgSe1-xSx samples
with x > 0.5 (Figure 6c,d), a small state immediately to the
PLE experiments were also conducted on stable 2.08 eV HgSe
clusters. Figure 7 shows data taken by monitoring the emission
intensity at three energies corresponding to positions at 2/3
height, 1/2 height, and on the shoulder of a second feature
accompanying the main emission band. Figure 7 shows subtle
improvements over the ensemble absorption spectrum (Figure
2a) similar to that observed for HgSe1-xSx clusters, although

Binary and Ternary Mercury Chalcogenides
J. Phys. Chem. B, Vol. 107, No. 24, 2003 5765
Figure 7. Photoluminescence (dashed line) and photoluminescence
excitation spectra (solid line) of stable 2.08 eV HgSe clusters. In all
cases, the emission intensity as a function of excitation energy was
monitored at positions denoted by the solid vertical black lines. A small
sharp feature immediately to the blue of the first excited state is a
solvent Raman artifact. Spectra are offset for clarity.
the effect is less dramatic. In particular, the region near 3.4 eV,
which was present as an unresolved shoulder in the absorption
data, is now better resolved. The PLE data also indicate that
the observed emission originates from stable 2.08 eV HgSe
clusters. As with HgSe1-xSx clusters, no dramatic intensity
changes occur in any of the PLE spectra when the observation
energy is tuned over the entire emission profile, suggesting that
different size fractions do not exist in HgSe samples.
Figure 8. Fitting results to PLE and absorption spectra of HgSe1-xSx
and HgSe clusters. Experimental data (solid line) and fit (dotted line)
are superimposed and offset from the extracted states, which are shown
below and labeled by Greek letters. (a) HgSe0.81S0.19, (b) HgSe0.42S0.58
(c) small HgSe clusters, and (d) stable 2.08 eV HgSe clusters.
,
To analyze PLE experiments conducted on HgSe1-xSx clusters
with x between 0.1 and 0.8, spectra were fit using a nonlinear
least squares Levenberg-Marquardt algorithm. In all cases,
experimental PLE data were fit to a sum of Gaussians from
which peak positions of underlying states could be determined.
The results of such a fitting procedure for two representative
HgSe1-xSx samples with low (x ) 0.19) and high (x ) 0.58)
sulfur content are shown in Figure 8a,b. The fit (dotted line)
and experimental spectrum (solid line) are superimposed and
offset from the fitting results. In certain cases a polynomial or
Gaussian background function is used to improve the quality
of the fits and is shown by the broad dashed lines in Figure
8a,b.
Up to five excited states are identified by the fitting procedure.
These states are labeled by Greek letters in order of their
progression from lower to higher energies. Because of the broad
background used to account for unresolved transitions, no
oscillator strengths are reported. An interesting observation,
immediately apparent from this analysis, is the behavior of the
two lowest excited states (denoted R and â). In Figure 8b, â is
resolved whereas in Figure 8a it has merged with R to create a
single asymmetric peak. The gradual evolution of the spacing
between R and â appears to be a function of x and/or size. This
observation, that up to two states coexist under what appears
to be a single peak, explains why in the ensemble absorption
spectrum of ternary clusters (Figure 1), the first peak appears
broad in samples with large values of x and progressively
narrows with decreasing sulfur content.
Because intermediate HgSe clusters other than the 2.08 eV
species cannot be isolated, we are currently unable to conduct
PLE experiments on these compounds. However, the evolution
of their ensemble absorption spectrum (Spectra 1-9, Figure 2b)
can be studied using the same nonlinear least-squares fitting
procedure. Results of the analysis for “small” (Spectrum 1,
Figure 2b) and “large” (Spectrum 9, Figure 2b) HgSe clusters
are shown in Figure 8c,d. As before, the fit (dotted line) and
experimental data (solid line) are superimposed and offset from
the fitting results. Up to four excited states are identified and
denoted by primed Greek letters in ascending order, from lower
to higher energies. A broad Gaussian or polynomial background
function representing unresolved transitions is also occasionally
used to improve the fit, and consequently, no oscillator strengths
for the various transitions are reported. Furthermore, the
absorption spectrum of the stable 2.08 eV HgSe species
(Spectrum 9, Figure 2b) closely matches the PLE data (Figure
7), showing the self-consistency of the data and the approach.
Figure 9a,b summarizes the results of the fitting procedure
for both HgSe1-xSx and HgSe clusters. Positions of extracted
states are denoted by energy offsets relative to the ground
excited state. In the case of HgSe1-xSx clusters, the progression
of up to five excited states is seen (denoted by alternating open
and closed symbols in the figure). Solid lines are a linear fit
through the data and also serve as a guide to the eye. Whether
the trends seen in Figure 9a arise solely from changes in
composition (x) or from differences in size (or both) is difficult

5766 J. Phys. Chem. B, Vol. 107, No. 24, 2003
Kuno et al.
Figure 10. Time-resolved emission decay of HgSe1-xSx clusters taken
at 20 K. Fluorescence decays at four emission wavelengths were
monitored.
excited at 4.13 eV (300 nm) using the doubled output of a dye
laser. Emission decays were collected at four energies, at the
peak and to the red or blue of the main emission band. Figure
10 shows HgSe1-xSx fluorescence lifetime plots taken at four
different emission wavelengths. Each curve can be fit by a sum
of three exponentials, using similar parameters. In particular, a
long, microsecond component of the data is apparent with an
average (standard error) lifetime of 1.7 (4) µs. This microsecond
component accounts for 40 (10)% of the decay and suggests
that the emission from HgSe1-xSx clusters is forbidden. The
fastest component has an average (standard error) lifetime of
110 (30) ns, accounting for 7 (4)% of the decay. The remaining
intermediate component has an average lifetime of 500 (100)
ns and accounts for 50 (10)% of the decay. Such long decays
have previously been seen in CdSe clusters, where emission
was argued to be surface mediated.13 In HgSe1-xSx clusters, this
may also be the case but further work is needed to verify this
and to exclude any recombination through internal core states.
We additionally note that no significant changes of the decay
profile occur over the range of energies investigated, which
encompasses the broad, low-energy shoulder of the emission
(Figure 1). This suggests that rather than multiple sources of
emission (that is, from some combination of core states, deep
trap states, or “shallow” surface states), only one type of emitting
state is present.
Figure 9. Summary of excited-state energies in (a) HgSe1-xSx and (b)
HgSe clusters. In both cases, energies are denoted as offsets relative to
the ground excited-state energy. Extracted states from the fitting
procedure are labeled by lowercase Greek letters in ascending order,
from lower to higher energies.
to ascertain due to complications in systematically sizing the
ternary clusters. However, the preliminary WAXS data indicate
that the size variation is small over the composition range,
suggesting that stoichiometry (x) is primarily responsible for
the progression of excited states. What is clear, though, is that
with increasing x (or size), the spacing between R and â
decreases while γ appears to be independent of either. States
labeled δ and ꢀ (where resolvable) also diverge with decreasing
x and energy of the first excited state.
A similar behavior is observed with HgSe clusters. The
second state labeled â′ converges toward R′ with decreasing
energy of the first excited state. The offset of the third state, γ′,
however, is nearly constant, which is analogous to the behavior
seen for γ in HgSe1-xSx clusters. Although δ′ in Figure 9b can
only be extracted in three samples, it also shows a similar trend
as δ (Figure 9a) diverging with decreasing energy of the first
excited state.
A compelling issue is the remarkable similarity in the
progression of excited states for both HgSe and HgSe1-xSx
clusters. Despite their differences in composition and oscillator
strength (compare Figure 1 to Figure 2), the spacings between
states not only follow the same divergent or convergent trends
relative to the lowest excited state, but also have energy offsets
of nearly equal magnitude. It is not yet understood whether such
trends are expected nor is it yet possible to assign the various
interband transitions. Furthermore, the divergent trend of δ (δ′)
and ꢀ (ꢀ′) with decreasing energy of the first excited state is
opposite to trends typically observed in other quantum dot
systems.47,48 Another point of interest is the relative contribution
of each valence band to the different transitions, whether the
lowest excited-state involves an excitation from Γ8 or Γ6 to the
conduction band or, more interestingly, if and when a reversal
of Γ8 and Γ6 (from an inverted band structure to a more typical
zinc blende ordering) occurs and how this affects the observed
transitions.49
Conclusion
The work presented herein describes the synthesis, charac-
terization, and preliminary optical studies of binary and ternary
mercury chalcogenide clusters. The synthesis used to create such
clusters is a generalization of a strong Hg(II) ligand and
precursor phase separation scheme that we have found necessary
to control the high reactivity of Hg(II). It has the novelty of
using a water/trialkylphosphine mixture as the growth medium,
bridging previous work on NC synthesis using micellar media
and more recent ligand-mediated and/or organometallic prepara-
tions conducted in both coordinating and noncoordinating
solvents.
Characterization with WAXS and EDXS shows that both
binary (HgSe) and ternary (HgSe1-xSx) mercury chalcogenide
clusters can be made. In particular, EDXS results reveal that
the entire compositional range of the ternary solid solution can
be spanned with the above synthesis. Such clusters appear to
be passivated by TOP/TOPO, and NMR experiments further
indicate that an important role is played by proton-coupled
impurities in technical grade TOP. Preliminary optical experi-
Low-temperature fluorescence decay experiments were con-
ducted on HgSe1-xSx clusters to provide a lower limit for their
radiative lifetime. Isolated clusters in toluene were cooled to
20 K using a He flow-through cryostat and were subsequently

Binary and Ternary Mercury Chalcogenides
J. Phys. Chem. B, Vol. 107, No. 24, 2003 5767
(12) (a) Behrens, S.; Bettenhausen, M. Deveson, A. C.; Eichhofer, A.;
Fenske, D.; Lohde, A.; Woggon, U. Angew. Chem., Int. Ed. Engl. 1996,
35, 2215. (b) Behrens, S.; Fenske, D. Ber. Bunsen-Ges. Phys. Chem. 1997,
101, 1588. (c) Behrens, S.; Bettenhausen, M.; Eichhofer, A.; Fenske, D.
Angew. Chem., Int. Ed. Engl. 1997, 36, 2797.
(13) (a) Soloviev, V. N.; Eichhofer, A.; Fenske, D.; Banin, U. J. Am.
Chem. Soc. 2000, 122, 2673. (b) Soloviev, V. N.; Eichhofer, A.; Fenske,
D.; Banin, U. J. Am. Chem. Soc. 2001, 123, 2354. (c) Soloviev, V. N.;
Eichhofer, A.; Fenske, D.; Banin, U. Phys. Status Solidi 2001, 1, 285.
(14) Kershaw, S. V.; Harrison, M.; Rogach, A. L.; Kornowski, A. IEEE
J. Selected Top. Quantum Electron. 2000, 6, 534.
(15) Micic, O. I.; Curtis, C. J.; Jones, K. M.; Sprague, J. R.; Nozik, A.
J. J. Phys. Chem. 1994, 98, 4966.
(16) Guzelian, A. A.; Katari, J. E. B.; Kadavanich, A. V.; Banin, U.;
Hamad, K.; Juban, E.; Alivisatos, A. P.; Wolters, R. H.; Arnold, C. C.;
Heath, J. R. J. Phys. Chem. 1996, 100, 7212.
ments to investigate the linear absorption of HgSe and HgSe1-xSx
clusters reveal an interesting progression of states dependent
on size in the case of HgSe and on a combination of size and
composition for the ternary compounds. Time-resolved emission
experiments at low temperatures complement work on the linear
absorption by probing the emitting state and show that the
emission, despite being weak with significant nonradiative
channels, is relatively slow. Decays are multiexponential and
occur over the course of hundreds of nanoseconds to several
microseconds, suggesting that the emission is forbidden in
nature. However, whether it occurs through internal triplet-like
core states or through surface-related states is not yet established.
The current success in the development of high-quality
mercury chalcogenide clusters, brings to light a number of
compelling issues which will benefit from further study. From
a synthetic perspective, better management of the high reactivity
of Hg(II) and improved stabilization of larger clusters or QDs
is necessary to push the growth beyond 2.08 eV and out of the
visible region of the spectrum. One possible means of doing
this is to use 2.08 eV HgSe clusters as “seeds” for growth in a
more suitable medium, as described in recent work by Strouse.19
In addition, the present work presents a challenge for the
development of a theoretical construct which can explain the
unusual similarity between excited-state energy spacings of
HgSe and HgSe1-xSx clusters, their dissimilarity in oscillator
strengths, the identity of long-lived emitting states in these
materials, as well as predict the precise point where a semimetal-
to-semiconductor transition occurs. Similarly, ab initio structural
calculations, which can help determine the precise stoichiometry
and shape of stable mercury chalcogenide clusters, such as the
2.08 eV HgSe species, are needed to better understand these
materials.
(17) Guzelian, A. A.; Banin, U.; Kadavanich, A. V.; Peng, X.; Alivisatos,
A. P. Appl. Phys. Lett. 1996, 69, 1432.
(18) Hines, M. A.; Guyot-Sionnest, P. J. Phys. Chem. B 1998, 102, 3655.
(19) Cumberland, S. L.; Hanif, K. M.; Javier, A.; Khitrov, G. A.; Strouse,
G. F.; Woessner, S. M.; Yun, C. S. Chem. Mater. 2002, 14, 1576.
(20) Torimoto, T.; Kontani, H.; Shibutani, Y.; Kuwabata, S.; Sakata,
T.; Mori, H.; Yoneyama, H. J. Phys. Chem. B 2001, 105, 6838.
(21) Talapin, D. V.; Haubold, S.; Rogach, A. L.; Kornowski, A.; Haase,
M.; Weller, H. J. Phys. Chem. B 2001, 105, 2260.
(22) Cotton, F. A.; Wilkinson, G. In AdVanced Inorganic Chemistry;
Wiley: New York, 1988.
(23) Brennan, J. G.; Siegrist, T.; Carroll, P. J.; Stuczynski, S. M.;
Reynders, P.; Brus, L. E.; Steigerwald, M. L. Chem. Mater. 1990, 2, 403.
(24) (a) Higginson, K. A.; Kuno, M.; Bonevich, J.; Qadri, S. B.; Yousuf,
M.; Mattoussi, H. J. Phys. Chem. B 2002, 106, 9982. (b) Kuno, M.;
Higginson, K. A.; Bonevich, J.; Qadri, S. B.; Yousuf, M.; Mattoussi, H.
SPIE Proc. 2002, 146, 4808.
(25) Zhu, J.; Liu, S.; Palchik, O.; Koltypin, Y.; Gedanken, A. J. Solid
State Chem. 2000, 153, 342.
(26) Wang, H.; Zhang, J.-R.; Zhu, J.-J. J. Cryst. Growth 2001, 233,
829.
(27) Liao, X.-H.; Zhu, J.-J.; Chen, H.-Y. Mater. Sci. Eng. 2001, B85,
85.
(28) Li, Y.; Ding, Y.; Liao, H.; Qian, Y. J. Phys. Chem. Solids 1999,
60, 965.
(29) Zeng, J.-H.; Yang, J.; Qian, Y.-T. Mater. Res. Bull. 2001, 36, 343.
(30) Elliot, D. J.; Furlong, D. N.; Grieser, F. Colloids Surf. A 1999,
155, 101.
(31) Korgel, B. A.; Monbouquette, H. G. Langmuir 2000, 16, 3588.
(32) Guglielmi, M.; Martucci, A.; Fick, J.; Vitrant, G. J. Sol-Gel Sci.
Technol. 1998, 11, 229.
(33) (a) Rogach, A.; Kershaw, S.; Burt, M.; Harrison, M.; Kornowski,
A.; Eychmuller, A.; Weller, H. AdV. Mater. 1999, 11, 552. (b) Harrison,
M. T.; Kershaw, S. V.; Rogach, A. L.; Kornowski, A.; Eychmuller, A.;
Weller, H. AdV. Mater. 2000, 12, 123. (c) Harrison, M. T.; Kershaw, S. V.;
Burt, M. G.; Eychmuller, A.; Weller, H.; Rogach, A. L. Mater. Sci. Eng.
2000, B69-70, 355.
(34) (a) Bettenhausen, M.; Fenske, D. Z. Anorg. Allg. Chem. 1998, 624,
1245. (b) Eichhofer, A.; Troster, E. Eur. J. Inorg. Chem. 2002, 2253.
(35) Qu, L.; Peng, A.; Peng, X. Nano Lett. 2001, 1, 333.
(36) Yu, W. W.; Peng, X. Angew. Chem., Int. Ed. 2002, 41, 2368.
(37) Battaglia, D.; Peng, X. Nano Lett. 2002, 2, 1027.
(38) Murray, C. B.; Sun, S.; Gaschler, W.; Doyle, H.; Betley, T. A.;
Kagan, C. R. IBM Res. DeV. 2001, 45, 47.
(39) Wehrenberg, B. L.; Wang, C.; Guyot-Sionnest, P. J. Phys. Chem.
B (in press).
(40) Martell, A. E.; Smith, R. M. In Critical Stability Constants; Plenum
Acknowledgment. We thank the Office of Naval Research
for financial support. K.A.H. thanks the National Research
Council for a postdoctoral fellowship. We also thank F. V.
Mikulec and C. D. Merritt for helpful discussions during the
course of this work.
Supporting Information Available: Energy dispersive
X-ray spectra of binary (HgSe) and ternary (HgSe1-xSx) mercury
chalcogenide NCs. This material is available free of charge via
the Internet at http://pubs.acs.org.
References and Notes
(1) An excellent discussion can be found in the following book.
Gaponenko, S. V. In Optical properties of semiconductor nanocrystals;
Cambridge University Press: New York, 1998.
(2) Yoffe, A. D. Adv. Phys. 2002, 51, 799.
(3) Klein, D. L.; Roth, R.; Lim, A. K. L.; Alivisatos, A. P.; McEuen,
P. L. Nature 1997, 389, 699.
(4) Huynh, W. U.; Dittmer, J. J.; Alivisatos, A. P. Science 2002, 295,
2425.
(5) Murray, C. B.; Norris, D. J.; Bawendi, M. G. J. Am. Chem. Soc.
1993, 115, 8706.
(6) Bowen Katari, J. E.; Colvin, V. L.; Alivisatos, A. P. J. Phys. Chem.
1994, 98, 4109.
(7) (a) Choy, A.; Craig, D.; Dance, I.; Scudder, M. J. Chem. Soc., Chem.
Commun. 1982, 1246. (b) Dance, I. G.; Choy, A.; Scudder, M. L. J. Am.
Chem. Soc. 1984, 106, 6285. (c) Lee, G. S. H.; Craig, D. C.; Ma, I.; Scudder,
M. L.; Bailey, T. D.; Dance, I. G. J. Am. Chem. Soc. 1988, 110, 4863. (d)
Lee, G. S. H.; Fisher, K. J.; Craig, D. C.; Scudder, M. L.; Dance, I. G. J.
Am. Chem. Soc. 1990, 112, 6435.
Press: New York, 1974.
(41) Hoffman, G. G.; Brockner, W.; Steinfatt, I. Inorg. Chem. 2001,
40, 977.
(42) E° (M2+ + 2e f M); Zn (-0.762 V), Cd (-0.402 V), Hg (+0.854
V) from Cotton and Wilkinson, ref 22.
(43) Pileni, M. P. In Structure and ReactiVity in ReVerse Micelles; Pileni,
M. P., Ed.; Elsevier: Amsterdam, 1989.
(44) Raman contributions from TOP and toluene arise between 2800
(8) Herron, N.; Calabrese, J. C.; Farneth, W. E.; Wang, Y. Science
1993, 259, 1426.
and 3100 cm-1
.
(45) In Nuclear Magnetic Resonance (NMR) Data; Gupta, R. R.,
Lechner, M. G., Eds.; Landolt-Bornstein Numerical Data and Functional
Relationships in Science and Technology, Group III, 1997; Vol. 35.
(46) Kuno, M.; Lee, J. K.; Dabbousi, B. O.; Mikulec, F. V.; Bawendi,
M. G. J. Chem. Phys. 1997, 106, 9869.
(47) Norris, D. J.; Bawendi, M. G. Phys. ReV. B 1996, 53, 16338.
(48) Banin, U.; et al. J. Chem. Phys. 1998, 109, 2306.
(9) (a) Vossmeyer, T.; Reck, G.; Schulz, B.; Katsikas, L.; Weller, H.
J. Am. Chem. Soc. 1995, 117, 12881. (b) Vossmeyer, Y.; Reck, G.; Katsikas,
L.; Haupt, E. T. K.; Schulz, B.; Weller, H. Science 1995, 267, 1476.
(10) (a) Ali, B.; Dance, I. G.; Craig, D. C.; Scudder, M. L. J. Chem.
Soc., Dalton Trans. 1998, 1661. (b) Nyman, M. D.; Hampden-Smith, M.
J.; Duesler, E. N. Inorg. Chem. 1996, 35, 802.
(11) (a) Eichhofer, A.; Fenske, D.; Pfistner, H.; Wunder, M. Z. Anorg.
Allg. Chem. 1998, 624, 1909. (b) Pfistner, H.; Fenske, D. Z. Anorg. Allg.
Chem. 2001, 627, 575.
(49) Roth, W. L. In Physics and Chemistry of II-VI Compounds; Aven,
M., Prener, J. S., Eds.; North-Holland: Amsterdam, 1967.