Fermi level equilibration in quantum dot-metal nanojunctions

Article abstract of DOI:10.1021/jp011576t

The process of Fermi level equilibration in 5 nm ZnO quantum dot-metal nanojunctions has been monitored using changes to both the surface plasmon band of the metal island and the sharp exciton band of the ZnO nanocrystals following photoinduced electron a

Full text of DOI:10.1021/jp011576t

8810
J. Phys. Chem. B 2001, 105, 8810-8815
Fermi Level Equilibration in Quantum Dot-Metal Nanojunctions†
Annabel Wood,‡ Michael Giersig,§ and Paul Mulvaney*,‡,|
School of Chemistry, UniVersity of Melbourne, ParkVille, Victoria 3010, Australia, Hahn-Meitner Institut,
Department Solar Energy Research, Glienickerstr.100, Berlin, 14109, Germany, and Max-Planck Institute for
Colloids and Surfaces, Potsdam-Golm, 14476, Germany
ReceiVed: April 26, 2001
The process of Fermi level equilibration in 5 nm ZnO quantum dot-metal nanojunctions has been monitored
using changes to both the surface plasmon band of the metal island and the sharp exciton band of the ZnO
nanocrystals following photoinduced electron accumulation. In the cases of silver, copper, and gold islands,
excess electrons reside on both the quantum dot and the metal, whereas for Pt islands, the excess electrons
reside exclusively on the Pt island. Electrons are transferred rapidly from Pt to the solvent ethanol, preventing
accumulation on the quantum dots. The combination of exciton bleaching and surface plasmon shifts provides
a simple way of probing the efficiency of small metal islands as redox catalysts on semiconductor particles.
Introduction
a metal island or adsorbed electron acceptor (for an electron
diffusion coefficient of ∼0.1 cm2 s-1 9 ), which is rapid enough
to compete with recombination processes.10-12
One of the fundamental challenges in quantum dot chemistry
is unravelling the various relaxation processes that occur
following the creation of electron-hole pairs or excitons within
the semiconductor particle. Generally, one is interested in two
extreme cases. In the first, maximum band edge luminescence
is desired; that is, all charge carriers should recombine radia-
tively, and no redox reactions should occur at the surface. At
the other extreme, one wants to extract chemical energy out of
the illuminated particle, by maximizing electron-hole separa-
tion. In artificial photosynthesis and photochemical energy
storage schemes, rapid, quantitative transfer of one of the charge
carriers to a solution-phase redox couple is then usually required.
One of the most common methods for facilitating electron
transfer from conduction band electrons to solution species has
been through the use of metal island catalysts deposited onto
the quantum dot (QD) surface. These metal islands mediate
multielectron redox processes with solution phase redox carriers,
and enhancements of both anodic and cathodic redox processes
have been observed at illuminated semiconductor electrodes that
have metal island deposits.1-8
For macroscopic n-type materials in contact with metals a
Schottky barrier (depletion layer) normally forms at the junction
between the two materials, which reduces the kinetics of electron
injection from the semiconductor conduction band into the metal.
Curran and Lamouche7 pointed out that because the diameter
of a QD is significantly smaller than that of the depletion layer,
there is no significant depletion layer within the QD to impede
electron transfer. Gerischer8 subsequently suggested that, for
hydrogen loaded metal islands at least, an accumulation layer
may in fact be created, which could accelerate electron transfer
to the metal island from the QD. Although it is not clear what
type of electronic barrier exists between the QD and metal
island, Gra¨tzel1 noted that free diffusion of charge carriers alone
will permit electron transfer within a time R2/π2 D ≈ 0.1 ps to
The question we were interested in probing was whether
photoelectrons in fact migrate efficiently to metal island catalysts
on an n-type semiconductor particle surface or whether some
electrons accumulate on the semiconductor. If excess electrons
remain on the semiconductor, then subsequent electron-hole
charge separation is likely to be very inefficient due to rapid
reaction of the nascent holes with the resident excess electrons
in the n-type material. However, to identify where the charges
reside at equilibrium, we need to be able to spectrophotometri-
cally monitor electrons on both sides of the junction separately.
One way to determine the presence of excess charge on the
QD is through exciton bleaching. For semiconductors with
strong exciton absorption bands in the visible part of the
spectrum such as ZnO, it is well established that excess
electrons13,14 induce a dramatic bleaching of the exciton band.
This is attributed to either the Mosse-Burstein effect (band
filling15 ), or to electric field modulated effects on the excitonic
levels by trapped carriers, the so-called Stark effect.16,17
Unfortunately, while titanium dioxide is the most commonly
used and efficient UV photocatalyst in use at present, its solid-
state properties are not particularly useful for such studies. Its
absorption spectrum is largely featureless, and excess photo-
electrons within the conduction band can only be observed from
the very weak blue color of the trapped electrons.1,18 Conversely,
ZnO is a photocatalytically active metal oxide, which exhibits
strong exciton absorption. It is also well-known that the exciton
absorption bands of both ZnO colloids and ZnO nanoparticle
films11,19,20 are strongly bleached by the presence of accumulated
conduction band electrons. This provides a simple optical
method for determining the presence of excess electrons on the
ZnO QDs.
Small metal islands of Ag, Au, and Cu can be deposited onto
the ZnO surface photochemically. If the metal islands are small
enough, then their characteristic surface plasmon bands will be
observed in the ZnO-metal absorption spectrum. The position
of the surface plasmon band position depends on the conduction
band electron density.21 Increases in electron density cause a
† Part of the special issue “Royce W. Murray Festschrift”.
* Corresponding author. Email: mulvaney@unimelb.edu.au.
‡ University of Melbourne.
§ Hahn-Meitner Institut.
| Max-Planck Institute for Colloids and Surfaces.
10.1021/jp011576t CCC: $20.00 © 2001 American Chemical Society
Published on Web 08/21/2001

Quantum Dot-Metal Nanojunctions
J. Phys. Chem. B, Vol. 105, No. 37, 2001 8811
blue shift in the band position. The changes in band position
can then be used to see whether the metal islands accumulate
charge efficiently from the illuminated semiconductor particles.
By combining both of these optical effects, it becomes
possible to see where the photoelectrons reside in an illuminated
quantum dot-metal nanojunction. The aim of this paper is thus
to demonstrate the equilibration of the quasi-Fermi level of an
illuminated quantum dot with the Fermi level of a contacted
metal island catalyst particle and to compare this process for
different metal catalysts. We will show that it is possible to
readily distinguish between the different metals as redox
catalysts, even through the use of quasi-steady-state photolysis
experiments.
Experimental Section
To prepare ZnO QDs (Q-ZnO), 100 mL of 20 mM analytical
grade zinc acetate dihydrate (Aldrich) dissolved in pure, dry
ethanol was placed in an Erlenmeyer flask equipped with as
septum and vigorously stirred. A 1.7 mL sample of tetramethyl-
ammonium hydroxide (TMAOH) (25 wt % in methanol from
Aldrich) was added at a rate of 100 µL/min, using a motor-
driven syringe to ensure continuous, reproducible addition. The
size of the quantum dots could be varied by heating or cooling
the zinc acetate solution to a desired temperature before addition
of TMAOH. For the experiments described here, a synthesis
temperature of 60 °C was used, which yielded Q-ZnO with a
mean diameter of approximately 5 nm. This procedure represents
an enormous simplification over the currently employed syn-
thesis, first described by Spanhel and Anderson,22 and yields at
-5 °C the smallest Q-ZnO to date, with an exciton peak at 270
nm. A medium-pressure mercury lamp (Bausch and Lomb) was
used as the excitation source in the photolysis experiments. The
lamp was fitted with a heat absorbing filter (Oriel #6127) and
a broad band-pass filter (Oriel #59152) so that only UV light
between 200 and 400 nm entered the photolysis cells. A 5 mL
sample of ZnO QDs was placed in a 1 cm path length quartz
cuvette, and ethanol or metal salts were added to bring the total
solution volume to 5.5 mL, with a Q-ZnO concentration of ∼
0.9 mM and a metal ion concentration of 25-250 µM. The
cell was fitted either with a fine capillary bubbler or a vacuum
line seal for degassing. The samples were gently degassed with
nitrogen for 15 min and then irradiated. Absorbance and
fluorescence spectra were taken at intervals through the
procedure above, using a Cary Bio-50 UV-vis spectrophotom-
eter and a Varian Eclipse fluorimeter. After photolysis, the QDs
were left to equilibrate for 1 h, which usually resulted in further
spectral changes. Air was then allowed back into the cell to
discharge excess electrons from the QDs. Transmission electron
microscopy (TEM) samples were prepared by placing a drop
of solution onto wax film and then placing a carbon-coated
copper grid carbon side down onto the drop. After 20 s, the
grid was placed on damp filter paper and left to dry in a
glovebox. TEM images were obtained using a Philips CM-12
TEM equipped with an EDS analyzer for determination of the
elemental composition of single particles. The images were
captured with a slow scan digital camera (Gatan) and then
filtered and processed using commercial image processing
software (NIH Image, Adobe Photoshop).
Figure 1. (a) Absorbance and fluorescence spectra of of zinc oxide
QDs made at 60 °C. λexc ) 280 nm. (b) Absorbance spectra of the
ZnO QDs, shown in Figure 1a (solid line), degassed with nitrogen and
irradiated for cumulative times of 15 min (dotted line) and 2 h (dotted
and dashed line), and 1 h after allowing air back into the cell (dashed
line).
spectra of the zinc oxide QD colloids used in this study. The
absorption spectrum has a well-defined exciton band at 321 nm.
The fluorescence spectrum has two peaks of interest, a very
weak ultraviolet emission at 359 nm, which is usually attributed
to direct exciton recombination,13,14,23 and a much broader,
visible emission centered at 520 nm, which is usually attributed
to the radiative recombination of free conduction electrons with
trapped holes. A variety of other mechanisms have been
postulated, which are reviewed by Hirschwald.24 However a
systematic review of the green emission by van Heusden et al.25
has recently concluded that the main process is indeed recom-
bination of a free electron and a trapped hole. The particles
formed via hydrolysis with TMAOH are stable for a few days
at -5 °C, but the particles slowly grow over a period of hours
at room temperature to about 5 nm diameter, at which point
they are stable for some months. Meulenkamp and co-workers
26,27
28
and Wong et al. have shown that this process is due to
Ostwald ripening.
In Figure 1b, the effects of UV irradiation of 5 nm diameter
ZnO QDs are shown. The solution was degassed with nitrogen
to remove dissolved oxygen. Degassing the solution had little
or no effect on the absorption spectrum. After several minutes
of UV irradiation, the absorption spectrum showed strong
bleaching around the exciton band, as observed by numerous
other groups for both ZnO colloids and ZnO electrodes.13,14,20
After 1 h of irradiation, the rate of formation and recombination
of excited electrons had reached a steady state, and little change
Results
Illumination of ZnO QDs. Room-temperature hydrolysis of
zinc acetate with TMAOH produces 5 nm ZnO particles which
are reasonably monodisperse. The QDs are crystalline and index
to wurtzite. Figure 1a shows the absorbance and fluorescence

8812 J. Phys. Chem. B, Vol. 105, No. 37, 2001
Wood et al.
occurred in either the absorbance or the fluorescence spectrum.
It is well established that the basic processes occurring in these
ZnO QDs are as follows:
hν8 ecb- + hvb
(1)
+
ZnO
9
Excess electrons are generated on the colloid particles due to
2hvb+ + C2H5OH f CH3CHO + 2H+
(2)
oxidation of the organic solvent. The process is very facile. The
resulting excess electrons may be trapped or mobile within the
conduction band. The liberated protons from reaction 2 are likely
to be adsorbed to the particle surface, due to the accumulation
of excess negative charge, as a means to maintain a constant
Helmholtz potential. Hole scavenging by the solvent results in
electron accumulation on the QD. Eventually, a steady state is
reached because holes created in the presence of existing free
conduction electrons are annihilated faster than they can be
scavenged by the solvent. These excess electrons can be titrated
by adding an acceptor such as methyl viologen (ꢀ397 ) 37 500
Figure 2. Absorption changes observed before (solid line) and after
photolysis of 1 mM deaerated 5 nm ZnO QDs in the presence of 0.1
mM AgClO4 in ethanol, for 15 min (dotted line) and 2 h (dotted and
dashed line), and then 1 h after allowing air into the cell (dashed line).
appeared as seen in Figure 2, and the solution turned a clear
yellow color. As irradiation continued, the silver plasmon band
increased in intensity and blue shifted, indicating a slow increase
in electron density on the silver particles.21,30 Throughout the
first 30 min of illumination during which reduction of silver
ions took place, the exciton band of the ZnO QDs remained
unbleached, indicating that no excess electrons were stored on
the QD while silver reduction was occurring. This is in stark
contrast to the case in the absence of silver, where bleaching
was observed within 5 min of illumination. After standing the
colloid for an hour, the ZnO-Ag exciton band was still
bleached, and the silver plasmon band was still blue shifted
from its original position, so the excess electrons on the Q-ZnO
and silver did not readily transfer to the solvent, ethanol. To
demonstrate that both the Ag and ZnO particles had accumulated
excess electrons, air was admitted back into the cell. After 2
min, the exciton peak had fully recovered, and the plasmon band
had red shifted by 20 nm due to the reduction of oxygen by
excess electrons on both the Ag island and the ZnO particle.
The solution at this point remained a transparent yellow-brown
color, and no particle coagulation was observed. Thus, both the
metal and QD had been strongly cathodically polarized from
the open-circuit potential in aerated ethanol.
M-1 cm-1 ; ꢀ602 ) 12 800 M-1 cm-1 29
necb- + nMV2+ (colorless) f nMV+ (blue)
)
(3)
Titration of irradiated particles, which displayed exciton bleach-
ing with MV2+ under careful exclusion of oxygen, demonstrated
that the ZnO particles had accumulated at least 10 electrons
each under steady state photolysis. This indicates that the ZnO
particles are effectively degenerate and that the Fermi level after
photolysis under nitrogen must lie close to the conduction band
edge. Thus, there is certainly exciton bleaching of degenerate
ZnO QDs. What fraction of these electrons is truly mobile and
what fraction is trapped in surface states remains a moot point.
The viologen titration demonstrates that the electrons transfer
rapidly to a suitable acceptor in solution, so they cannot be
deeply trapped in any case.
Upon standing the irradiated solution for 2-3 h, the excitonic
peak slowly began to recover. This is due to slow reduction of
the solvent by the excess electrons stored in the QD. These
results are similar to those obtained by other groups and indicate
that large numbers of electrons can be accumulated for several
hours on the ZnO particles in the absence of a labile acceptor
in solution or on the surface. The question now is how well
this accumulated charge can be scavenged by metal islands in
electrical contact with the QDs.
Gold Islands. The effects of KAu(CN)2 reduction on the
absorption spectrum of ZnO QDs are shown in Figure 3.
Colloidal gold displays a surface plasmon band at 520 nm, which
is ideal, since it is well removed from the region of the exciton
absorption in ZnO. In this case, conduction band electrons are
initially discharged via the reaction
Silver Islands. Figure 2 shows the effects of illumination of
a solution of ZnO QDs and containing 25 µM AgClO4. In the
presence of silver ions, the excess electrons generated through
reactions 1-2 now react with the metal ion to produce small
metal deposits on the QD surface.
ecb- + Au(CN)2- f Auo + 2CN-
(7)
After 5 min of irradiation, a peak at ∼545 nm began to form,
attributed to the plasmon band of gold nanoparticles.31 With
further illumination, this peak became more intense, and blue
shifted to an eventual position of ∼520 nm. The exciton peak
of Q-ZnO did not show bleaching until after 1 h of irradiation
time. This suggests that either more electrons are needed to raise
the Fermi energy in gold to the ZnO conduction band energy
level than is the case for silver, or the gold is a better catalyst
and was able to dissipate accumulated charge by electron transfer
to the solvent more rapidly than silver. After standing the
irradiated solution for less than 1 h, the exciton peak had fully
recovered, indicating a gradual shift of excess electrons from
the conduction band of zinc oxide onto the gold particles and,
ecb- + Ag+ads f Agsurf
(4)
Following the initial nucleation step (reaction 4), the nascent
Ag atoms scavenge further generated conduction band photo-
electrons. By analogy to the photographic process, this Ag island
growth is probably preceded by silver ion adsorption
Agsurf + Ag+ f Ag2
(5)
+
surf
After 5 min of irradiation, a surface plasmon band at 400 nm
Ag2 surf + e-cb f Ag2
(6)
+
+
surf

Quantum Dot-Metal Nanojunctions
J. Phys. Chem. B, Vol. 105, No. 37, 2001 8813
Figure 3. Absorption changes observed before (solid line) and after
photolysis of 1 mM deaerated 5 nm ZnO QDs in the presence of 0.1
mM KAu(CN)2 in ethanol, for 15 min (dotted line) and 2 h (dotted
and dashed line), and 1 h after allowing air into the cell (dashed line).
The plasmon band of Au has been expanded (right-hand axis) for clarity.
Figure 5. Absorption changes observed before (solid line) and after
photolysis of 1 mM deaerated 5 nm ZnO QDs in the presence of 0.1
-
mM HPtCl6 in ethanol, for 15 min (dotted line) and 2 h (dotted and
dashed line), and 1 h after allowing air into the cell (dashed line).
under nitrogen for several hours, indicating negligible rates of
electron transfer to the metal from the QD and from the metal
to the solvent. When air was admitted, the exciton peak
recovered instantly, and the copper band red-shifted to a final
position at 582 nm.
Pt Islands. The most potent hydrogenation catalysts are the
Pt group metals, Pt, Pd, and Ir.33 Unfortunately, unlike the
coinage metals, Ag, Au, and Cu, these and most other transition
metals do not display strong plasmon bands in the UV-visible
region. This is due to the damping effect of the d-d transitions
in these metals, which tend to wash out the free electron
contribution to the dielectric function.21 As nanoparticles, these
metals are normally a yellow-brown color due to the d-d
transitions, and their absorption spectra are largely featureless.
In Figure 5, the results of an experiment are shown in which
0.9 mM ZnO QDs were irradiated in the presence of 50 µM
H2PtCl6. After 5 min of UV irradiation, a ‘tail’ began to develop
in the ZnO spectrum due to the formation of platinum islands.
As irradiation continued, the platinum band increased in
intensity, but the exciton peak of Q-ZnO showed no bleaching.
In fact, even after extensive photolysis for several hours, no
bleaching of the exciton band was observed. Upon admission
of air, no significant change to the spectrum of the ZnO was
observed.
After photodeposition, small dark patches could be observed
on the ZnO particles by high-resolution electron microscopy.
It was difficult to resolve the islands well enough to get accurate
sizing. Figure 6 shows a high-resolution image of ZnO-Pt
particles. Despite some clumping of the particles, the two types
of crystal are readily resolved by electron diffraction. The inset
power spectrum shows spots belonging to Pt islands with the
typical fcc structure of bulk Pt. Also evident are spots which
can only be indexed to the ZnO (002) reflex. The identity of
both island and ZnO particles was confirmed by EDAX analysis.
Figure 4. Absorption changes observed before (solid line) and after
photolysis of 1 mM deaerated 5 nm ZnO QDs in the presence of 0.2
mM Cu(CH3COO)2 in ethanol, for 15 min (dotted line) and 2 h (dotted
and dashed line), and 1 h after allowing air into the cell (dashed line).
hence, into the solvent. This suggests that the main difference
between gold and silver islands is in fact the rate of catalyzed
ethanol reduction by the metal islands. The SP band was located
at 522 nm at this point. After allowing oxygen into the solution,
the position of the plasmon peak shifted immediately to 530
nm. The final color of the transparent, colloidally stable solution
was a strong purple-red, due to the presence of the gold
nanoparticles.
Copper Islands. The behavior of copper islands was similar
to that of silver and gold. Typical results are shown in Figure
4, where 0.2 mM Cu(CH3OO)2‚H2O was added to the Q-ZnO.
Addition of Cu2+ to Q-ZnO caused the appearance of a small
peak at 370 nm due to d-d transitions in the cupric ion.
Irradiation of the sample caused the appearance of a new peak
at 570 nm, attributed to the plasmon band of copper metal
particles. Cu nanoparticles display peaks at around 570 and 300
nm.31,32 As irradiation was continued, the 570 nm peak increased
in intensity but, unlike gold and silver, did not concurrently
blue shift during metal ion reduction. After 60 min of irradiation,
the Q-ZnO exciton peak showed slight bleaching, and the copper
plasmon band had blue shifted by a few nanometers. After 2 h,
the Q-ZnO exciton peak was fully bleached, and the copper
plasmon band was centered at 565 nm. No change was observed
to the exciton band nor to the metal plasmon band on standing
Discussion
The results presented here all indicate that during photoca-
talysis, a QD-metal nanojunction is not necessarily as efficient
a catalyst for electron discharge as may be initially envisaged.
Despite the fact that the empty conduction band energy levels
of all the metals lie at least 0.5-1.0 V below the ZnO
conduction band, sufficient charge accumulates on the metal
islands of Cu, Ag, and Au to ensure Fermi level equilibration
between the QD and metal. In fact, it takes some hours for

8814 J. Phys. Chem. B, Vol. 105, No. 37, 2001
Wood et al.
This is ignored in the left-hand side of Figure 7. The excess
electrons are slowly discharged by the solvent over hours once
photolysis and hole scavenging cease.
In the case of ZnO-metal nanojunctions, the situation is more
complex. The spectral changes to the surface plasmon band of
the ZnO-Ag, ZnO-Au, and ZnO-Cu particles indicate
substantial electron accumulation on the metal islands during
photolysis. The basic shifts that occur can be understood by
considering the case of silver. Once the silver ions are
completely reduced, the Fermi level in the Ag island still lies
well within the band gap, probably close to +0.15 V versus
NHE, its open-circuit value in water.30 As the UV-irradiation
continues, electronic charge flows to the Ag islands, causing a
slow increase in the Fermi energy (i.e., inducing a more negative
redox potential through an accumulation of excess electrons on
the silver islands). The conduction band of bulk ZnO (in water)
is located at least -0.8 V above the silver Fermi level.1,33 Yet
after continuous illumination for 2 h, the zinc oxide exciton
absorption band does bleach, indicating that electrons are being
stored on the QD and no longer being transferred to the metal
island. Weller and Hoyer found that the onset for exciton
bleaching of a ZnO electrode was around -0.6 V versus NHE.20
For the smaller ZnO particles used in this study, the redox
potential for electrons in degenerate particles will be about 200
mV more negative than this due to size quantization effects on
the energy levels. Electron transfer will cease once the silver
Fermi level coincides with the ZnO conduction band edge.
Hence, the Fermi level of the Ag island must have been
cathodically polarized by at least -0.8 V from its typical open-
circuit value in water. This condition is attained when both
exciton bleaching and a strong plasmon band blue shift are
observed. Consistent with this enormous polarization, the silver
plasmon band peak position was located at 374 nm, which is a
much shorter wavelength than normally encountered in aqueous
solution (378-385 nm) even under strongly reducing condi-
tions.21 This strong blue shift is only possible if silver is a very
poor catalyst for ethanol reduction.
Figure 6. High-resolution TEM image of the surface of coated ZnO-
Pt QDs. The Pt deposits are about 1.5-2 nm in size and are darker
than the ZnO particles due to the higher electron scattering cross section
of Pt. In the image, we can see two sets of lattice planes. In the power
spectrum (inset), we can distinguish two sets of points, which we assign
to Pt(111) and ZnO(002).
Figure 7. Scheme of the energy levels in photoirradiated ZnO QD-
metal nanojunctions. (left) UV irradiation of ZnO in ethanol produces
electron-hole pairs. The hole is rapidly scavenged by the solvent,
leading to excess electron accumulation on the particles and exciton
bleaching. The particles are degenerate after extensive photolysis in
the absence of air. (middle) In the presence of Ag, Au, or Cu islands,
electrons accumulate and pass to the islands where they are stored,
leading to the buildup of a large Helmholtz (or diffuse layer) potential
difference due to slow scavenging by the solvent. The Fermi level in
the metal is shifted cathodically until it merges with the QD conduction
band energy level. Almost all excess electrons are on the metal island.
Both exciton bleaching and a surface plasmon blue shift are observed.
(right) In the presence of Pt, rapid transfer of electrons from the Pt to
the solvent occurs. The metal Fermi energy remains close to the solution
redox level, and the QD Fermi energy remains in the mid-gap region.
No bleaching of the exciton during photolysis is observed, and no
electrons reside on the QD during photolysis.
The large potential difference, which drives the metal Fermi
level up until equilibration with the ZnO is achieved, is due to
charging of the metal Helmholtz and diffuse double layers (see
Figure 11, ref 4). The maximum Helmholtz potential difference
achievable is given approximately by
VH ) Ecb - Eredox
(8)
It is important to recognize that the excess electron density is
stored overwhelmingly on the metal islands because the
Helmholtz capacitance of the metal-solution interface is much
higher than the space charge capacity of the QD, even under
accumulation. Hence, the Fermi level within the QD will remain
near the bottom of the conduction band. Additionally, as long
as only a few electrons reside on the QD, there will not be a
significant shift in the position of the band edges of the QD.
Slow discharge of the metal electronic charge by the solution
redox couple once photolysis ceases gradually brings the Fermi
levels in the metal and in the QD back to their original positions.
For Ag, this takes some hours if oxygen is not present. Electron
scavenging by oxygen is rapid and depletes both sides of the
nanojunction in seconds.
Chen et al. previously carried out an investigation of the
ZnO-Ag colloid system.34,35 Our observations are similar to
theirs for this system. However, they concluded that the changes
in the silver SP band were due to the conversion of Ag2O islands
into Ag. We believe the effects are more easily explained in
terms of the Ag island electron density. The study by Ung et
electrons to discharge from ZnO-Ag nanoparticles in deaerated
ethanol. Conversely, for ZnO-Pt at least, the Fermi level under
identical photolysis conditions does not reach the conduction
band of ZnO QDs. This interpretation is qualitatively sum-
marized in Figure 7. The Fermi level in an open-circuit QD is
ill-defined but will be in the midgap region (left diagram), and
in principle, its position is dictated by the most facile redox
couple present in solution (dotted line). The QD will contain
no electrons in the conduction band. Illumination leads to
electron accumulation, and the Fermi level moves up toward
the conduction band energy level. Some movement of the band
edges with respect to the solvent may occur, depending on the
way counterions neutralize the electron density within the QD.

Quantum Dot-Metal Nanojunctions
J. Phys. Chem. B, Vol. 105, No. 37, 2001 8815
(2) Nosaka, Y.; Ishizuka, Y.; Miyama, H. Ber. Bunsen-Ges. Phys.
Chem. 1986, 90, 1199-1204.
(3) Chandrasekharan, N.; Kamat, P. V. J. Phys. Chem. B 2000, 104,
10851-10857.
(4) Nakato, Y.; Ueda, K.; Yano, H.; Tsubomura, H. J. Phys. Chem.
1988, 92, 2316-2324.
(5) Nakato, Y.; Shioji, M.; Tsubomura, H. Chem. Phys. Lett. 1982,
90, 453-456.
al. unambiguously demonstrated that the Fermi level in silver
colloids determines the position of the surface plasmon absorp-
tion band and that the shift can be quantitatively related to the
effect of the excess electron density on the plasma frequency.30
For Pt islands, electrons on the metal island are transferred
rapidly to the solvent and the metal Fermi level remains pinned
at close to Eredox (Figure 7 right diagram). This means there is
always a strong driving force for electron transfer to the Pt from
the ZnO, and few electrons accumulate during photolysis on
the QD. No exciton bleaching occurs.
(6) Hiesgen, R.; Meissner, D. J.Phys. Chem. B 1998, 102, 6549-6557.
(7) Curran, J. S.; Lamouche, D. J. Phys. Chem. 1983, 87, 5405-5411.
(8) Gerischer, H. J. Phys. Chem. 1984, 88, 6096-6097.
(9) Hilgendorf, M.; Spanhel, L.; Rothenhausler, C.; Muller, G. J.
Electrochem. Soc. 1998, 145, 3632.
(10) Rabani, J.; Behar, D. J. Phys. Chem. 1989, 93, 2559-2563.
(11) Kamat, P. V.; Patrick, B. J. Phys. Chem. 1992, 96, 6829-6834.
(12) Cavaleri, J. J.; Skinner, D. E.; Colombo, D. P., Jr.; Bowman, R.
M. J Chem Phys 1995, 103, 5378.
(13) Haase, M.; Weller, H.; Henglein, A. J. Phys. Chem. 1988, 92, 482-
487.
(14) Hoyer, P.; Weller, H. Chemical Physics Lett. 1994, 221, 379-
384.
(15) Burstein, E. Phys. ReV. 1954, 93, 632.
(16) Hilinski, E. F.; Lucas, P. A.; Wang, Y. J. Chem. Phys. 1988, 89,
3435-3441.
(17) Colvin, V. L.; Cunningham, K. L.; Alivisatos, A. P. J. Chem. Phys.
1994, 101, 7122-7138.
(18) Sarany, A.; Gao, R.; Rabani, J. J. Phys. Chem. B 2000, 104, 5848-
5853.
(19) Hoyer, P.; Eichberger, R.; Weller, H. Ber. Bunsen-Ges. Phys. Chem.
1993, 97, 630.
(20) Hoyer, P.; Weller, H. J. Phys. Chem. 1995, 99, 14096-14100.
(21) Mulvaney, P. Langmuir 1996, 12, 788-801.
The number of electrons needed to be accumulated to raise
the Fermi level is not easily measurable. Using spectroelectro-
chemistry, a double-layer capacitance of 80 µF/cm2 has been
measured for Ag colloids in aqueous solution.30 This value is
likely to be very sensitive to particle size, solvent polarity, and
dielectric perturbations due to the junction with the QD.
Assuming that the double-layer capacitance of the Ag island
attached to the ZnO is similar to that of the isolated Ag
nanoparticles in solution, the position of the plasmon band
position at 374 nm, when ZnO bleaching is seen, corresponds
to a redox potential of about -0.9 V versus NHE, and to the
storage of >50 electrons if the island radius is 1 nm. This
corresponds very closely to the estimated position of the ZnO
conduction band and confirms the fundamental result of this
study, that there is equilibration of the metal and QD Fermi
levels under photolysis. In fact, such blue-shifted surface
plasmon bands are rarely observed for silver particles in water,
and this further suggests that electron transfer from silver
particles to ethanol is intrinsically slower than in aqueous media.
(22) Spanhel, L.; Anderson, M. A. J. Am. Chem. Soc. 1991, 113, 2826-
2833.
Conclusions
(23) Bahnemann, D. W.; Kormann, C.; Hoffmann, M. R. J. Phys. Chem.
1987, 91, 3789-3798.
(24) Hirschwald, W.; Bonasewicz, P.; Ernst, L.; Grade, M.; Hofmann,
D.; Krebs, S.; Littbarski, R.; Neumann, G.; Grunze, M.; Kolb, D.; Schulz,
H. J. Zinc Oxide: Properties and behaViour of the bulk, the solid/Vacuum
and solid/gas interface; North-Holland Publishing Company: Amsterdam,
1981; Vol. 7.
(25) Vanheusden, K.; Seager, C. H.; Warren, W. L.; Tallant, D. R.;
Caruso, J.; Hampden-Smith, M. J.; Kodas, T. T. J. Lumin. 1997, 75, 11-
16.
(26) Meulenkamp, E. A. J. Phys. Chem. B 1998, 102, 5566-5572.
(27) Meulenkamp, E. A. J. Phys. Chem. B 1998, 102, 7764-7769.
(28) Wong, E. M.; Bonevich, J. E.; Searson, P. C. J. Phys. Chem. B
1998, 102, 7770-7775.
(29) Bird, C. L.; Kuhn, A. T. Chem. Soc. ReV. 1981, 10, 49-82.
(30) Ung, T.; Dunstan, D.; Giersig, M.; Mulvaney, P. Langmuir 1997,
13, 1773-1782.
The primary aim of this initial study was to demonstrate that
electron equilibration occurs in QD-metal colloid systems
during photolysis. It was shown that simple spectroscopic
methods enable one to monitor electron transfer processes at
metal-QD junctions relatively easily. It was also demonstrated
that during extensive photolysis, electron capture by metal
islands of Ag, Au, and Cu becomes inhibited as the redox
potential is cathodically shifted and approaches the ZnO
conduction band level. At this point, electrons reside for long
periods on both the QD and metal island. Photocatalysis is
consequently retarded by the subsequent, rapid hole capture in
the degenerate ZnO particle. Pt conversely appears to be an
efficient electron sink, and virtually no electrons reside on the
QD of a ZnO-Pt particle even during extended photolysis.
(31) Creighton, J. A.; Eadon, D. G. J. Chem. Soc., Faraday Trans. 1991,
87, 3881.
(32) Sosebee, T.; Giersig, M.; Holzwarth, A.; Mulvaney, P. Ber. Bunsen-
Ges. Phys. Chem. 1995, 99, 40-49.
(33) Bockris, J. O. M.; Khan, S. U. M. Surface Electrochemistry; Plenum
Press: New York, 1993.
Acknowledgment. The authors thank the ARC for support
of this research. A.W. is the recipient of a Melbourne Research
Scholarship. P.M. thanks the Humboldt Foundation for support
of part of this project.
(34) Chen, S.; Nickel, U. J. Chem. Soc., Chem. Commun. 1996, 133-
References and Notes
4.
(1) Gra¨tzel, M. Heterogeneous Photochemical Electron Transfer; CRC
Press: Boca Raton, FL, 1989.
(35) Chen, S.; Nickel, U.; Ren, X. J. Colloid Interface Sci. 1995, 176,
286-92.