Super-resolution mapping of photogenerated electron and hole separation in single metal-semiconductor nanocatalysts

Article abstract of DOI:10.1021/ja409011y

Metal-semiconductor heterostructures are promising visible light photocatalysts for many chemical reactions. Here, we use high-resolution superlocalization imaging to reveal the nature and photocatalytic properties of the surface reactive sites on single Au-CdS hybrid nanocatalysts. We experimentally reveal two distinct, incident energy-dependent charge separation mechanisms that result in completely opposite photogenerated reactive sites (e^- and h⁺) and divergent energy flows on the hybrid nanocatalysts. We find that plasmon-induced hot electrons in Au are injected into the conduction band of the CdS semiconductor nanorod. The specifically designed Au-tipped CdS heterostructures with a unique geometry (two Au nanoparticles at both ends of each CdS nanorod) provide more convincing high-resolution single-turnover mapping results and clearly prove the two charge separation mechanisms. Engineering the direction of energy flow at the nanoscale can provide an efficient way to overcome important challenges in photocatalysis, such as controlling catalytic activity and selectivity. These results bear enormous potential impact on the development of better visible light photocatalysts for solar-to-chemical energy conversion.

Full text of DOI:10.1021/ja409011y

Article
pubs.acs.org/JACS
Super-Resolution Mapping of Photogenerated Electron and Hole
Separation in Single Metal−Semiconductor Nanocatalysts
Ji Won Ha, T. Purnima A. Ruberu, Rui Han, Bin Dong, Javier Vela,* and Ning Fang*
Ames Laboratory, U.S. Department of Energy, and Department of Chemistry, Iowa State University, Ames, Iowa 50011, United States
*
S Supporting Information
ABSTRACT: Metal−semiconductor heterostructures are
promising visible light photocatalysts for many chemical
reactions. Here, we use high-resolution superlocalization
imaging to reveal the nature and photocatalytic properties of
the surface reactive sites on single Au−CdS hybrid nano-
catalysts. We experimentally reveal two distinct, incident
energy-dependent charge separation mechanisms that result
−
in completely opposite photogenerated reactive sites (e and
+
h ) and divergent energy flows on the hybrid nanocatalysts.
We find that plasmon-induced hot electrons in Au are injected
into the conduction band of the CdS semiconductor nanorod.
The specifically designed Au-tipped CdS heterostructures with a unique geometry (two Au nanoparticles at both ends of each
CdS nanorod) provide more convincing high-resolution single-turnover mapping results and clearly prove the two charge
separation mechanisms. Engineering the direction of energy flow at the nanoscale can provide an efficient way to overcome
important challenges in photocatalysis, such as controlling catalytic activity and selectivity. These results bear enormous potential
impact on the development of better visible light photocatalysts for solar-to-chemical energy conversion.
INTRODUCTION
their separate isolated components because one can conven-
iently tune the catalytic efficiency of the heterostructure by
controlling the size, distribution, and loading of metal particles.
In this context, it is important to elucidate the fundamental
photocatalytic mechanisms operating on metal−semiconductor
hybrid heterostructures to design and develop better (more
active, selective, and stable) photocatalysts.
■
Anisotropic semiconducting nanoparticles with high surface-to-
volume ratios have gained much attention over the past decade
as photocatalysts for many chemical reactions including
1
−6
photoelectrochemical hydrogen production
and photo-
7
chemical degradation of organic pollutants. Most semi-
conductor photocatalysts are currently based on wide-gap
materials that have limited applicability outside of the UV
range. However, UV light constitutes less than 5% of the solar
spectrum; therefore, it is highly desirable to investigate
photocatalysts with tunable activity that can be widely applied
under visible light.
The photocatalytic behavior of metal−semiconductor
heterostructures active in the visible has been studied mainly
14−21
at the ensemble level.
For example, Amirav and Alivisatos
demonstrated photocatalytic hydrogen production using Pt-
18
tipped CdS nanorod heterostructures; notably, the catalytic
properties were considerably influenced by the individual
particle sizes, structures, and so forth. To surmount the
challenge arising from the intrinsic heterogeneity associated
from ensemble-averaged measurements, it is highly desirable
and necessary to employ photocatalytic measurements at the
single-particle level.
Metal-modified semiconductor nanorods are promising
visible-active heterostructured photocatalysts for the following
reasons: First, semiconductor nanorods, as compared to
spherical semiconductors, provide the advantage of multiple
exciton generation and more efficient photoinduced charge
8
,9
separation.
Second, the physical spatial separation of
Recently, single-particle catalysis using methods such as
photogenerated electrons and holes in metal−semiconductor
heterostructures can suppress recombination and enhance
2
2
electrochemical detection, surface plasmon spectrosco-
23−25
26−33
10
py,
and single-molecule fluorescence microscopy
overall photocatalytic efficiency. Third, recent advances in
solution-phase synthesis enable the highly predictable and
reliable fabrication of metal−semiconductor heterostructures
with controlled size, shape, and composition tailored for
has been demonstrated. Novo et al. reported the direct
observation of oxidation reactions on single Au nanocrystals
23
using dark-field microscopy. Xu and co-workers presented the
11−13
catalytic properties of Au nanoparticles at the single-particle
specific applications.
Importantly, this unique ability to
2
9,30
level using single-molecule fluorescence microscopy.
selectively build metal−semiconductor heterostructures with
several morphologies and spatial relationships between their
individual components can be used to engineer and direct
energy flows at the nanoscale. This is a great advantage over
Received: August 30, 2013
Published: January 12, 2014
©
2014 American Chemical Society
1398
dx.doi.org/10.1021/ja409011y | J. Am. Chem. Soc. 2014, 136, 1398−1408

Journal of the American Chemical Society
Article
Figure 1. Photocatalytic Au−CdS hybrid heterostructures reveal two distinct photoinduced charge separation mechanisms. (a) TEM image of single
high Au−CdS heterostructures (10.8 wt %). The CdS nanorods are 186(±56) nm long and 6.0(±0.9) nm wide. The diameter of Au nanoparticles
ranges from 2 to 7 nm. (b) UV−vis absorption spectra of pure CdS nanorods (blue curve), Au nanoparticles (red curve), and high Au−CdS
heterostructures (10.8 wt %, yellow curve). The green- and purple-dotted lines indicate the excitation source of a 532 nm laser and a 405 nm laser,
respectively. (c) Two charge separation mechanisms in these hybrid heterostructures. Mechanism A starts by excitation at the metal at 532 nm, thus
−
+
forming hot electrons (e ) and holes (h ) in the metal. The hot electrons are then transferred to the semiconductor’s conduction band. Mechanism
B starts with excitation at the semiconductor at 405 nm, thus forming electron−hole pairs at the semiconductor. Photogenerated electrons are then
trapped by the gold metal.
Roeffaers et al. used single-molecule fluorescence microscopy to
structures, and to clarify the factors that control catalytic
activity and selectivity in metal−semiconductor heterostruc-
tures.
monitor single turnover reactions catalyzed by crystals of a
3
3
layered double hydroxide (LDH). Furthermore, high-
resolution superlocalization fluorescence imaging in single-
particle catalysis has also been carried out by several
To address the aforementioned challenges, we synthesized
CdS semiconductor nanorods decorated with Au nanoparticles
and studied the real-time redox photocatalysis of these hybrid
heterostructures at the single-particle level with millisecond
time resolution. We chose Au−CdS hybrid heterostructures as
photocatalysts because they provide efficient light absorption
3
2,34−36
groups.
Zhou et al. employed superlocalization fluo-
rescence imaging to unveil the surface reactivity of Au nanorod
catalysts, and found that surface defects play important roles in
3
2
catalytic activity. De Cremer et al. reported high-resolution
reactivity mapping of epoxidation reactions catalyzed by
37
and photocatalytic activity in the visible range. We employ
high-resolution superlocalization fluorescence imaging to
resolve the nature and photocatalytic properties of th−e
photogenerated carriers and redox reactive sites (electrons, e
35
mesoporous titanosilicates.
More recently, superlocalization fluorescence imaging of
single metal−semiconductor heterostructures has been dem-
onstrated by Tachikawa et al.; in this report, single-molecule
super-resolution imaging of reactive sites on single Au-TiO2
hybrid particles were studied using an oxidation−reduction
+
and holes, h ) in Au−CdS hybrid heterostructures, using single-
molecule detection following the oxidation of nonfluorescent
amplex red to highly fluorescent resorufin. Moreover, we verify
two distinct, incident energy-dependent charge separation
mechanisms using specifically designed Au-tipped CdS
heterostructures. We further provide a new insight into the
working mechanism of photocatalysis (driven by light in Au
and CdS) and surface catalysis (driven by Au surface) in the
Au−CdS heterostructures during the fluorogenic oxidation
reaction.
34
(
redox)-responsive fluorescent dye. Despite the recent study
of photocatalysis by Au−TiO hybrid particles at the single-
2
particle level, the proposed mechanisms of action have not been
clearly verified experimentally, and it is still necessary to
develop metal−semiconductor heterostructures that can be
fully operated with tunable activity under visible light. Further,
many photocatalytic properties such as selectivity in metal−
semiconductor heterostructures remain largely unanswered,
and our understanding of the identity of the surface reactive
sites and factors affecting the catalytic efficiency in these hybrid
materials is still very limited. Therefore, the remaining grand
challenges in this area are to experimentally verify mechanisms
of action in metal−semiconductor heterostructures, to
investigate and develop highly active metal−semiconductor
heterostructures with tunable activity under visible light, to
resolve the single surface reactive sites where individual
photoinduced redox reactions occur on these photocatalytic
RESULTS AND DISCUSSION
■
Two Charge Separation Mechanisms in Au−CdS
Heterostructures. Au−CdS hybrid heterostructures used
here were synthesized according to modified literature
38,39
procedures (see details in the Supporting Information).
Figure 1a shows a transmission electron microscopy (TEM)
image of single CdS nanorods densely decorated with Au
nanoparticles (“high” surface-Au density). The diameter of the
Au nanoparticles ranged from 2 to 7 nm, and the Au loading on
1
399
dx.doi.org/10.1021/ja409011y | J. Am. Chem. Soc. 2014, 136, 1398−1408

Journal of the American Chemical Society
Article
Figure 2. Single-molecule superlocalization fluorescence imaging over single Au−CdS heterostructures during a photocatalytic oxidation reaction.
a) Experimental scheme using total internal reflection fluorescence microscopy to probe the fluorogenic oxidation reaction of nonfluorescent
(
amplex red (A) to highly fluorescent resorufin (R) by a Au−CdS nanocatalyst. (b) Typical image of fluorescent products at localized spots during
the catalytic reaction under 532 nm illumination. (c) Segment of a typical fluorescence intensity trajectory from the fluorescent spot squared in red in
b with 1 μM amplex red, 20 mM H O , and 10 mM phosphate buffer (pH 7.5). Temporal resolution is 50 ms. (d) Fluorescence image of a single
2
2
resorufin molecule during one burst circled in green in (c). (e) Three-dimensional representation of the image in (d). The center position of the
fluorescence image can be determined with nanometer accuracy (±5 nm) and is marked as a red cross in (d).
the CdS nanorod was calculated to be about 10.8 wt %. The
mean length and diameter of CdS nanorods used in this study
were 186(±56) nm and 6.0(±0.9) nm, respectively (Figures S1
and S2).
energy barrier (Schottky junction) at the metal−semiconductor
interface or tunnel through it to become conduction electrons
3
4,47,48
in the semiconductor.
This results in a spatially separated
electron−hole pair where the electron resides in the semi-
conductor (CdS), whereas the hole remains in the metal (Au).
Mechanism B starts with 405 nm absorption by the
semiconductor, followed by charge transfer of photoexcited
electrons to the metal. The electrons transferred to the Au
metal increase the electron density within the Au nanoparticles,
In Au−CdS hybrid heterostructures, the optical properties of
40
the original components are essentially retained. Hence, the
optical absorption spectra of Au−CdS heterostructures exhibit
both the characteristic excitonic and continuous absorption of
the CdS nanorods below 475 nm (2.61 eV), together with an
additional broad absorption centered around 532 nm (2.33 eV)
due to the plasmonic Au nanoparticles (Figure 1b). This can be
ascribed to the separate optical excitations of the semi-
conductor and metal components in the hybrid material. This
is advantageous because it enables to separately probe each one
of the two heterostructure components using two different
incident light wavelengths.
49
thus shifting the Fermi level toward more negative potentials.
The transfer of electrons to the metal continues until the Fermi
level equilibrates with the conduction band edge of the
49
semiconductor. In contrast to the previous mechanism A,
mechanism B (405 nm excitation) charges the gold negatively
and the CdS semiconductor positively. In other words,
mechanisms A and B result in two different charge-separated
Au−CdS species of completely opposite polarity. So far,
mechanism B is well established and widely used in
Because of the separation of the plasmonic and excitonic
features in Au−CdS hybrid heterostructures, two distinct
photoinduced charge separation mechanisms are expected to
exist depending on the incident excitation energy as shown in
Figure 1c. Mechanism A starts with 532 nm absorption by the
gold metal, followed by the formation of d-band electron−hole
pairs excited above the Fermi level upon decay of surface
plasmons in the metal. Excited plasmonic nanoparticles that are
much smaller than their plasmon resonance wavelength can be
3
,15
photocatalytic chemistry.
However, mechanism A is still
poorly understood in nanocatalysis. Therefore, in the present
study, we first try to experimentally verify the existence and
catalytic activity of mechanism A in these Au−CdS hybrid
heterostructures using 532 nm excitation, and further
investigate the capability to optically control the direction of
energy flow together with the catalytic activity and selectivity
offered by these two mechanisms as depicted in Figure 1c.
Catalytic Activity of Au−CdS Heterostructures in
Fluorogenic Oxidation Reactions. To test and study the
two aforementioned charge separation mechanisms in Au−CdS
hybrid heterostructures, we used the fluorogenic oxidation
reaction of nonfluorescent amplex red (10-acetyl-3,7-dihydrox-
−
+
41,42
an efficient source of hot electrons (e ) and holes (h ).
The
hot electrons have energies between vacuum level and the work
4
1,43
function ϕ of gold metal (ϕ = 4.83 eV).
These energetic
4
1,42,44−46
electrons could be potentially used in photochemistry.
For example, Mukherjee et al. recently demonstrated that hot
electrons formed on Au nanoparticles can be used for the room
temperature dissociation of H , which requires a dissociation
enthalpy of 436 kJ/mol (4.51 eV). A significant fraction of the
yphenoxazine) to produce highly fluorescent resorufin (λ
=
2
exc
41
31,32
563 nm; λem = 587 nm, at pH 7.5).
Experimentally, we first
energetic electrons in Au nanoparticles can either overcome the
tested if high Au−CdS heterostructures (Figure 1a) were active
1
400
dx.doi.org/10.1021/ja409011y | J. Am. Chem. Soc. 2014, 136, 1398−1408

Journal of the American Chemical Society
Article
in catalyzing the oxidation of amplex red to resorufin by
ensemble-averaged measurements. An aliquot of a solution
containing Au−CdS heterostructures was injected into a
cuvette containing a solution of 1 μM amplex red, 20 mM
H O , and 10 mM phosphate buffer with pH 7.5. The cuvette
was illuminated at 532 nm (±25 nm band-pass filter), and the
fluorescence intensity was measured every 30 s of continuous
illumination. As shown in Figure S3, the fluorescence intensity
arising from resorufin increases as a function of time in the
presence of high Au−CdS heterostructures, but remains
constant in the absence of high Au−CdS heterostructures
under the same conditions. This indicates that the Au−CdS
heterostructures are active for the fluorogenic oxidation
reaction under mechanism A at 532 nm. We estimate about
To further confirm that each fluorescence burst comes from
a single turnover catalytic reaction of amplex red over Au−CdS
heterostructures, we carried out further control experiments at
the single-particle level. We observed no stochastic fluorescence
bursts over Au−CdS heterostructures in the absence of amplex
red reactant or over pure CdS nanorods while the rest of the
experimental conditions were kept constant. In addition, no
digital single-molecule fluorescence bursts were observed over
Au−CdS heterostructures with resorufin product in solution
under the same conditions. There are three important findings
from these control experiments. First, a 532 nm green laser
does not excite isolated metal-free CdS nanorods, which means
that the intensity signal observed during catalysis is only
contributed from Au-modified heterostructures under mecha-
nism A at 532 nm. Second, our Au−CdS heterostructures do
not exhibit any time-dependent emission fluctuations (or
blinking) that could affect fluorescence bursts or our data
analysis (see more detailed discussion on blinking in the
Supporting Information). This is further verified by our
previous report to show no fluorescence emissions of CdS
nanorods with aspect ratios greater than 3 due to exciton
2
2
39% of resorufin product is produced in 15 min under the
specified conditions (Figure S3).
To better understand the photocatalytic properties of the
Au−CdS hybrid heterostructures, we performed the same
fluorogenic oxidation reaction at the single-particle level. A
sample was prepared by spin-casting Au−CdS heterostructure
solution on a positively functionalized quartz slide. The
concentration of the Au−CdS heterostructures immobilized
50
quenching by surface defects. Third, our results indicate that
diffusion of free resorufin in solution does not yield any
significant fluorescence bursts through its binding to the
nanoparticles under our experimental conditions (see more
details in the Supporting Information). This is also verified by
previous reports carried out under similar experimental
−2
on the quartz slide was controlled to be ∼1 μm for single-
particle catalysis and was checked using a differential
interference contrast (DIC) microscope (Figure S4). The
sample slide was then measured under a prism-type dual-color
total internal reflection fluorescence (TIRF) microscope
29,31
conditions.
Furthermore, every product formation (or
(Figure S5). A 532-nm green laser was used to excite both
dissociation) event in each single-molecule fluorescence
trajectory (Figure 2c) appears as a sudden increase (or
decrease) in intensity. In contrast, Xu et al. demonstrated
that photobleaching of single resorufin molecules happens on
much longer time scales (∼25 s) under similar 532 nm laser
the metal in Au−CdS heterostructures and the fluorescent
resorufin product. Before initiating the fluorogenic oxidation
reaction, we first shined the 532-nm laser beam onto the
sample for 20 min to remove possible fluorescent dusts or
impurities. We then introduced the reactant-containing solution
29
intensities. This strongly indicates that each observed sudden
(
1 μM amplex red, 20 mM H O , 10 mM pH 7.5 phosphate
2 2
decrease in intensity is attributed to the dissociation of
adsorbed resorufin from the reactive site where it is produced.
Therefore, the fluorescence bursts observed in this study are
attributed to the oxidation of amplex red to fluorescent
resorufin at reactive sites on single Au−CdS heterostructures,
and each fluorescence burst corresponds to a single catalytic
turnover that results in the formation of one resorufin molecule
at a reactive site, followed by its subsequent dissociation away
from that reactive site on a Au−CdS heterostructure.
buffer) over the Au−CdS heterostructures within a flow
chamber. A highly fluorescent resorufin product was formed
at one of many possible reactive sites on a single Au−CdS
heterostructure (Figure 2a), and was detected by an electron-
multiplying charge-coupled device (EMCCD) camera (Figure
S5b). The resorufin product (pK = 5.8) is deprotonated and
a
negatively charged in the solution with pH = 7.5. We recorded
movies of stochastic fluorescence bursts at many localized spots
on the quartz surface with a time resolution of 50 ms (Figure
Photocatalysis and Surface Catalysis Responsible for
Activity of Au−CdS Heterostructures. We showed that the
Au−CdS heterostructures are active for the fluorogenic
oxidation reaction under mechanism A at 532 nm. However,
Zhou et al. have demonstrated that isolated, unmodified (CdS-
free) Au nanoparticles mediate the oxidation of amplex red to
2b). Figure 2c shows a segment of a typical fluorescence
intensity trajectory from one spot squared in red in Figure 2b
containing stochastic fluorescence ON−OFF signals. If these
signals were produced from multiple resorufin molecules we
would expect a distribution of variable intensity fluorescent
signals, but the consistent height of the two-state ON-level in
Figure 2c indicates that each fluorescence burst comes from a
single resorufin molecule. Figure 2d is a fluorescence image of a
single resorufin molecule during one burst circled in green in
Figure 2c and the fluorescence intensity of a single resorufin
molecule spread over a few pixels as a point spread function
32
resorufin by surface catalysis. Therefore, it is important to
ensure that photocatalytic mechanism A, involving excited
surface plasmons on Au followed by charge separation into the
CdS semiconductor, was actually operating and was responsible
for activity under 532 nm illumination. For this purpose, we
performed several ensemble experiments (see more details in
Supporting Information). Among key results from these
experiments, we found that the absorbance due to resorufin
product builds up about 4 times faster under continuous 532
nm illumination than it does in the dark (only surface catalysis)
as shown in Figure S8. We also corroborated that 532 nm green
light cannot excite pure (metal-free) CdS nanorods; this
supports a null (zero) contribution to activity from the CdS
nanorods alone via mechanism A at 532 nm in the presence of
(
PSF). The center position of this PSF can be determined with
nanometer accuracy by two-dimensional (2D) Gaussian fitting
of its fluorescence profile (Figure 2e). This enables us to
localize the position of every resorufin molecule on a single
Au−CdS heterostructure during the catalytic reactions. In this
study, red-fluorescent beads with a diameter of 100 nm were
used as position markers to correctly localize the position of
each resorufin molecule (Figures S6 and S7).
1
401
dx.doi.org/10.1021/ja409011y | J. Am. Chem. Soc. 2014, 136, 1398−1408

Journal of the American Chemical Society
Article
Figure 3. Photocatalysis and surface catalysis in Au−CdS heterostuctures under mechanism A at 532 nm. (a) Approximate energy band positions of
the Fermi level (E ) of bulk Au and the conduction band (CB) of CdS. The redox potentials of some of molecules are shown on the right side. The
F
energy scale is indicated using the normal hydrogen electrode (NHE). A more detailed energy level diagram is provided in Figure S11. (b)
Schematics depicting the formation of resorufin product (R) from amplex red (A) in photocatalysis (top) and surface catalysis (bottom).
the heterostructures (Figures S9 and S10). These results clearly
indicate that the surface-bound Au nanoparticles become
excited and make the Au−CdS heterostructures distinguishably
more active under 532 nm illumination.
Next, it is important to clarify how resorufin molecules are
produced under photocatalytic mechanism A. For a wide band
performed further ensemble-averaged experiments using 5 nm
Au nanoparticles (no charge separation) under 532 nm
illumination (Figures S12 and S13). As shown in Figure S13,
the fluorescence intensity was increased by 0.085 in Au−CdS
heterostructures with charge separation when compared to Au
nanoparticles without charge separation under 532 nm
illumination. The increase by 0.085 is ascribed to the charge
separation in Au−CdS heterostructures. Furthermore, the
fluorescence intensity was found to be increased by 0.025
from Au−CdS heterostructures under dark conditions to Au
nanoparticles under 532 nm illumination. The fluorescence
increase by 0.025 can be ascribed to the photocatalysis
gap titanium dioxide (TiO ) in an aqueous environment,
2
−
photogenerated electrons (e ) are known to reductively react
with adsorbed oxygen (O ) as the primary electron acceptor to
2
•
−
51−54
generate surface-bound superoxide anion radicals (O2 ).
In addition, photogenerated holes (h ) oxidatively react with
+
adsorbed water (H O) to generate surface-bound hydroxyl
2
•
52,53
•−
radicals (HO ).
Both of these two surface-bound oxygen
involving O2 radicals and heating effect by Au. Therefore,
radicals are known to be very reactive toward nonfluorescent
amplex red, catalyzing its fluorogenic oxidation to fluorescent
we found that the charge separation have considerable impact
on the activity of Au−CdS heterostructures, while photo-
induced heating effect by Au nanoparticles does not have as
significant an impact on the activity under 532 nm illumination
(see details in the Supporting Information).
5
4,55
resorufin.
Therefore, we checked if these surface-bound
oxygen radicals were responsible for the photocatalytic activity
of Au−CdS heterostructures under mechanism A at 532 nm
(
Figure S8). Plasmon-induced hot electrons in Au upon 532
So far, we verified the existence and activity of photocatalytic
mechanism A in Au−CdS heterostructures under 532 nm
excitation. However, it is also important to take a closer look at
the aforementioned surface catalysis (Figure S8c, dark) that
also contributes to the activity of Au−CdS heterostructures
under mechanism A. Although surface catalysis on Au was
nm excitation can be transferred into the conduction band
4
7,48,56
(CB) of CdS.
As shown in Figures 3a and S11, the CB
energy E for CdS is sufficiently negative to reduce O to
CB
2
•−
57
•−
O2
. This means that O2 radicals can be produced in Au−
CdS heterostructures under 532-nm illumination. However,
32
photogenerated holes with the Fermi energy E in Au do not
reported by Zhou et al., our understanding of the detailed
mechanism is still limited. We therefore performed ensemble-
averaged experiments to better understand the working
mechanism behind surface catalysis on Au nanoparticle. Amplex
red used in this study has been widely used as sensitive
F
•
have sufficient energy to oxidize H O to HO because the E of
2
F
•
Au is more negative than the energy for HO /H O reaction
2
57
(
∼
Figures 3a and S11). Therefore, the increase of activity by
4 times of Au−CdS heterostructures under 532 nm
illumination in Figure S8d can be ascribed to the formation
of reactive O2 radicals followed by their reaction with amplex
red to form resorufin product (Figure 3b, photocatalysis).
Furthermore, we also need to consider the effect of charge
separation (e in CdS and h in Au) on the activity of Au−CdS
heterostructures. Recently, Wu et al. reported that the charge-
separated state is relatively short-lived for short Au−CdS
heterostructures. However, our long Au−CdS heterostruc-
tures (186 nm × 6 nm on average) may slow down the charge
recombination through the formation of Au−CdS Schottky
barrier that expels electrons away from the interface. In this
respect, it is expected that the charge-separated state will be
relatively longer-lived in our Au−CdS heterostructures under
mechanism A, and the charge separation can have significant
influence on their catalytic activity. To clearly verify this, we
fluorogenic probes to detect horseradish peroxidase (HRP) and
•
−
58,59
H O in many immunoassays.
Amplex red, in the presence
2
2
of HRP, reacts with H O to produce highly fluorescent
2
2
resorufin. More specifically, HRP is used as a catalyst to convert
−
+
•
H O into HO radicals, followed by their reaction with amplex
2
2
54
red to form resorufin. Similarly, it is reported that Au
nanoparticles used as a catalyst in this study can also
5
6
•
60
decompose H O2 into HO radicals on their surfaces.
2
Therefore, it is expected that resorufin product could be
•
formed by the reaction of amplex red with HO radicals
56
generated from H O in the presence of Au as described in
2
2
Figure 3b (bottom). From our ensemble experiments, we found
that H O is directly related to the activity of surface catalysis
2
2
on Au (Figure S14). This result suggests that surface catalysis
•
on Au happens in the presence of H O and that reactive HO
2
2
1
402
dx.doi.org/10.1021/ja409011y | J. Am. Chem. Soc. 2014, 136, 1398−1408

Journal of the American Chemical Society
Article
Figure 4. Heterogeneous and distinct photocatalytic activity of reactive sites on Au−CdS heterostructures. (a) Two distinct microenvironments
−
+
around two different reactive sites (e and h ) under 532 nm excitation. The reactive sites around photogenerated holes have a positively charged
microenvironment, while the reactive sites around photogenerated electrons have a negatively charged microenvironment. In this study, dimethyl
•
•
sulfoxide (DMSO) is used to quench the HO radicals (or suppress the surface catalysis on Au). DMSO is oxidized by the HO radicals to
−
1
methanesulfinic acid (MSIA, CH O S). (b) Histogram of the rate of product formation τ
for the fluorescence bursts obtained from over 100
4
2
off
individual nanoparticles in the absence of DMSO (blue curve) or in the presence of DMSO (red curve). (c) Histogram of the rate of product
−1
dissociation τon for the fluorescence bursts obtained in the absence of DMSO (blue curve) or in the presence of DMSO (red curve). The distinct
dissociation kinetics of the two basic reactive sites are observed because of the differently charged microenvironments as depicted in (a).
radicals may be involved as an intermediate in the oxidation
To clearly verify the aforementioned reactive sites and the
formation of HO radicals in surface catalysis on Au, we used
•
reaction. Further verification and discussion on the formation
•
of HO radicals in surface catalysis is described in the following
dimethyl sulfoxide (DMSO, 20 mM) known to get oxidized by
•
section.
HO radicals to form the stable compound methanesulfinic acid
61,62
•
Two Distinct Dissociation Kinetics in Single-Molecule
Experiments. In the fluorescence intensity trajectory, photo-
catalytic events have two important characteristic durations, τoff
and τon (Figure 2c); τoff is the characteristic time before the
formation of a fluorescent product on the Au−CdS
(MSIA, CH
O
4
S).
2
If HO radicals are actually involved in
surface catalysis on Au, the surface catalysis will be suppressed
in the presence of DMSO under mechanism A at 532 nm, and−
it is expected that fluorescence bursts should come from the e
reactive sites only.
Figure S15a shows a segment of a fluorescence intensity
trajectory over one Au−CdS heterostructure in the absence of
DMSO and the time profile for the two rates for each
fluorescence burst. We observe that the rates of product
formation and product dissociation over the same hetero-
structure vary at different times. Figure S15b shows a segment
of a fluorescence intensity trajectory over one Au−CdS
heterostructure in the presence of DMSO and the time profile
for the two rates for each fluorescence burst. From these
heterostructure, and τ is the characteristic time related to
on
the dissociation of the product molecule from the nanocatalyst
surface. The inverse of the values of τoff and τon represent the
single-reactive site rates of product formation and of product
dissociation, respectively. Therefore, a higher value of τon−
indicates faster dissociation of the resorufin product, while a
1
−1
higher value of τoff indicates faster formation of the resorufin
product.
It is important to note that some fluorescence bursts show
long τon values of 2−6 s, while other fluorescence bursts show
short τon values of 0.3−1.2 s (Figure 2c). We believe that these
two distinct τon values can be ascribed to two charge-separated
species containing photogenerated carriers at different locations
and microenvironments (Figures 1c and 4a). In this experiment
experiments, it is clear there is heterogeneity in catalytic activity
among the e− reactive sites on the same Au−CdS
heterostructure. This previously unobserved trait is always
masked in the ensemble-averaged measurements. More
importantly, as compared to the fluorescence bursts in the
absence of DMSO (Figure S15a), all the fluorescence bursts in
(
mechanism A at 532 nm), the reactive sites around
−1
the presence of DMSO (Figure S15b) show τ
values higher
on
−
photogenerated electrons (the e reactive sites) have a
negatively charged microenvironment which can promote
quicker dissociation of negatively charged resorufin molecule
due to the repulsive electrostatic force (Figure 4a). In
comparison, the reactive sites around photogenerated holes (or
−1
than 0.6 s , where the averaged rate of product dissociation
−1
−1
⟨
τ ⟩ is 1.18 s over a single Au−CdS heterostructure. This
1
on
−
−1
suggests that the lower τon value of around ∼0.25 s in
31
+
Figure S15a could be ascribed to the h reactive sites around a
positively charged microenvironment in Au. More importantly,
+
•
the h reactive sites) have a positively charged microenviron-
the result further suggests that HO radicals are involved in the
ment which can enhance the adsorption of the negatively
charged resorufin product (Figure 4a).
fluorogenic oxidation reaction as described in Figure 3b
(bottom). Nevertheless, the individual τon and τoff values
31
−1
−1
1
403
dx.doi.org/10.1021/ja409011y | J. Am. Chem. Soc. 2014, 136, 1398−1408

Journal of the American Chemical Society
Article
Figure 5. Superlocalization of single surface reactive sites on a high Au−CdS heterostructure during photocatalysis. (a) TEM image of typical single
high Au−CdS heterostructures (top). Enlarged TEM image (bottom) shows high density of Au nanoparticles. (b) Super-resolution mapping of
individual reactive sites on a high metal loading Au−CdS heterostructure at 532 nm. Each one of the circled cross or minus signs corresponds to one
+
resorufin product molecule. The circled blue crosses show a product molecule caused by the h reactive sites on Au, while the circled red minuses
−
shows a product molecule from the e reactive sites. The catalytic reaction was carried out with 1 μM amplex red and 20 mM H O in 10 mM pH 7.5
2
2
phosphate buffer.
in single-molecule events are stochastic, and hence, their
statistical properties, such as average values and distributions,
are required to draw a more concrete and meaningful
conclusion about the distinct dissociation kinetics in single-
molecule experiments.
% Au) under 532 nm laser excitation for 60 min through
localizing the position of every resorufin molecule with
nanometer accuracy (Figure 5; see more details on super-
localization of the center position in the Supporting
Information). Figure 5a shows TEM images of typical single
Au−CdS heterostructures with high metal loading, showing
that the Au nanoparticles are deposited and distributed along
the length of the CdS nanorod. Figure 5b is a super-resolution
image mapping single surface reactive sites during the
photocatalytic oxidation reaction on a Au−CdS hybrid
heterostructure with a similarly high metal loading. Plasmon-
induced energetic electrons (encircled-marked as red minuses),
initially created in the metal particles, are injected into the
conduction band of the CdS semiconductor nanorod. There-
fore, our experimental results are consistent with the existence
of mechanism A involving photoinduced charge transfer from
the gold metal to the CdS semiconductor under 532 nm light.
To the best of our knowledge, this is the first report resolving
and mapping individual reactive sites on single Au−CdS hybrid
heterostructured photocatalysts using a high-resolution super-
localization imaging technique.
Photocatalytic Mechanism B in Au−CdS Heterostruc-
tures. It is also important to check the photocatalytic activity
of CdS nanorods under mechanism B (Figure 1c). We
therefore tried to verify the existence of mechanism B involving
photogenerated electrons and holes in CdS nanorods at the
ensemble level. We checked if pure (metal-free) CdS nanorods
can catalyze the oxidation reaction of amplex red to resorufin,
which is necessary to support the existence of mechanism B
(see more details in the Supporting Information). In the
presence of CdS nanorods, the resorufin product’s absorption
maximum at 573 nm increased with an interval of 30 s, while it
remained constant in the absence of CdS nanorods under
identical conditions (Figure S17). This result clearly shows that
the resorufin product is also formed through photocatalysis
under mechanism B (Figures S17d and S18).
To more clearly elucidate the distinct dissociation kinetics
between the two different reactive sites, we further measured
many more Au−CdS hybrid heterostructures for both cases
(
with and without DMSO). In each case, we collected many
fluorescence bursts from over 100 individual nanoparticles for a
statistical analysis. Figure 4b shows a histogram of the rate of
product formation for all the fluorescence bursts. We found
−
1
that the averaged rate of product formation ⟨τ ⟩ is 0.0168
off
−1
−1
s
in the absence of DMSO, while ⟨τ ⟩ in the presence of
off
−1
DMSO is 0.00403 s . This difference in the averaged rate of
product formation can be explained by the effect of DMSO
competing with, and hindering amplex red near the nano-
catalyst surface, thus suppressing the surface catalysis on Au.
This DMSO effect is further verified by ensemble experiments
(Figure S16). In addition, Figure 4c shows a histogram of the
rate of product dissociation for all the fluorescence bursts1.
Interestingly, we find that the rate of product dissociation τon−
obtained without DMSO has two distinct distributions (or
peaks) for individual fluorescence bursts (Figure 4c), which is
not observed in the presence of DMSO. Therefore, Figure 4c
−
1
verifies that the fluorescence bursts yielding the lower τ
of
on
−1
around 0.25 s on the left side of the distribution come from
+
the h reactive sites, while the fluorescence bursts yielding the
higher τon− around 1.2 s on the right side of the distribution
1
−1
−
arise from the e reactive sites. The distinct dissociation kinetics
for the two different reactive sites are further supported by a
3
1
recent report.
High-Resolution Mapping of Single Turnover Events
on Au−CdS Heterostructures. The finding on the two
distinct dissociation kinetics is important because we can
directly distinguish catalytic fluorescence bursts caused by the
+
reactive sites from those by the e− reactive sites in a
h
We further elucidate the formation of reactive radicals
fluorescence intensity trajectory. More importantly, this allows
for resolving individual electron and hole transfer events
following photoinduced charge separation in single Au−CdS
hybrid heterostructures. We thus tried to map the individual
reactive sites on a single high Au−CdS heterostructure (10.8 wt
responsible for photocatalytic activity under mechanism B in
CdS nanorods. As shown in Figure S11, O2 radicals can be
produced in CdS nanorods under mechanism B, while
•
−
photogenerated holes in CdS do not have enough energy to
•
react with H O to produce HO radicals (see details in the
2
1
404
dx.doi.org/10.1021/ja409011y | J. Am. Chem. Soc. 2014, 136, 1398−1408

Journal of the American Chemical Society
Article
Figure 6. Engineering energy flows on single Au-tipped CdS nanorod heterostructures. (a) TEM image of typical single Au-tipped CdS nanorod
heterostructures (top). Enlarged TEM image (bottom) shows a Au nanoparticle at the tip. (b) Schematic illustrating two distinct photocatalysis
mechanisms with the opposite direction of energy flow (opposite polarity after photoinduced charge-separation). In mechanism A at 532 nm, the
photogenerated energetic electrons in the gold metal are injected to the conduction band (CB) of the semiconductor. In mechanism B at 405 nm,
the photogenerated electrons in the CB of the semiconductor are rapidly trapped by the gold metal. (c) Super-resolution mapping of single reactive
sites on a Au-tipped CdS nanorod heterostructure during the oxidation reaction at 532 nm (mechanism A). (d) Super-resolution mapping of single
reactive sites on a Au-tipped CdS nanorod heterostructure during the same oxidation reaction after turning on the 405 nm laser (in addition to the
532 nm laser, needed to excite the resorufin product) (mechanism B).
Supporting Information). However, we observed that pure
metal-free) CdS nanorods are still active in the presence of
nanoscale particle. However, in our study, the relatively long
distance between the two Au nanoparticles (186 nm on
average) allows us to avoid this pitfall, and the unique geometry
enables us to clearly prove the two charge separation
mechanisms with the opposite direction of energy flow.
(
•−
superoxide dismutase (SOD) to quench O2 radicals (Figure
S19). This strongly indicates that another radical species are
produced from holes generated in CdS. We found that
photogenerated holes in CdS have enough energy to react
An absorption spectrum of Au-tipped CdS heterostructures
is provided in Figure S20. These Au−CdS heterostructures
differed from those mentioned above in that they have a very
low metal loading (0.54 wt % Au on CdS). As depicted in
Figure 6b, we can selectively excite either the gold metal
domains at 532 nm to turn ON mechanism A, or the CdS
semiconductor nanorod domains at 405 nm to turn ON
mechanism B. We note however, that under single-molecule
experiments presented here, a 532-nm laser was always needed
to excite resorufin product, and both 405 and 532 nm lasers
were used in our study of mechanism B. Figure 6c shows a
super-resolution image mapping the reactive sites during the
same amplex red to resorufin oxidation reaction taken at 532
2
‑
−
−
with some of the ligands (S (or HS ) and RCOO ) present in
high concentrations on the surface of the CdS nanorod
5
7,63−66
(
Figures 3a and S11).
These reactions can easily
•
‑
•
generate highly reactive radical species such as S (or HS )
sulfanyl) and RCOO (carboxyl),
•
63−65
(
each of which can in
turn be responsible for the oxidation of amplex red to the
6
6−68
fluorescent resorufin product.
Therefore, under mecha-
•
−
nism B, O
radicals can be generated from electrons, while
S /HS or RCOO radicals are expected to be produced from
holes.
2
•
‑
•
•
Verification of Two Charge Separation Mechanisms
on Au-tipped CdS Heterostructures. To clearly demon-
strate the existence of two distinct photocatalysis mechanisms
having the opposite direction of energy flow (i.e., having the
opposite polarity after photoinduced charge separation) at the
single-particle level, we further synthesized Au-tipped CdS
heterostructures (Figures 6a and S1a). The specifically designed
Au-tipped CdS heterostructures are visible-active photocatalysts
with a unique geometry: two Au nanoparticles at both ends of
each CdS nanorod. The 5−15 nm superlocalization accuracy
typically found in single-molecule superlocalization studies can
make it very challenging to pinpoint the turnover events on a
+
nm over 1 h. Interestingly, we find that the h reactive sites
(circled blue-cross) are positioned at the gold tips on both ends
−
of the heterostructures, while the e reactive sites (circled red-
minus) are located along the inside length of the CdS nanorods
within a distance of a few tens of nanometers from the gold
tips. Therefore, this result reveals that at 532 nm we turned ON
mechanism A, which proceeds by charge (electron) transfer
from the gold metal to the CdS semiconductor nanorod. This is
consistent with our observations described above for the CdS
nanorods with high Au loading (Figure 5b). In addition, Figure
1
405
dx.doi.org/10.1021/ja409011y | J. Am. Chem. Soc. 2014, 136, 1398−1408

Journal of the American Chemical Society
d shows a super-resolution image mapping the reactive sites
Article
6
Further Discussion on Photocatalytic Properties of
Au−CdS Heterostructures. Besides the capability of tuning
their photocatalytic selectivity, another advantage of hybrid
metal−semiconductor heterostructures such as Au−CdS over
its separate components is that we can control the photo-
catalytic activity by two methods. First, the photocatalytic
activity in these hybrid heterostructures can be controlled by
varying the metal (Au) loading on the semiconductor (CdS)
nanorod under mechanism A at 532 nm. Specifically, we find
that the rate of resorufin product formation rises fast when
increasing the Au loading on the CdS nanorods at 532 nm
(Figure S24). Mukherjee et al. have also demonstrated a steady
obtained after turning on the 405 nm laser in addition to the
32 nm laser (the latter needed to excite the resorufin
5
product). We ensured that CdS nanorods in this study are
excited at 405 nm under a dual-color TIRF microscope (Figure
+
S21). Critically, in this case the h reactive sites (encircled blue-
crosses) are distributed along the inside length of the CdS
−
nanorod, while the e reactive sites (encircled red minuses) are
located at both ends. Note that Fermi level equilibration can
significantly raise the energy the level of small Au nanoparticles
compared to bulk Au, closer to the conduction band of the CdS
4
9
semiconductor. This supports that the electrons transferred
to the Au metal from the CdS semiconductor can have enough
increase of photocatalytic rate of HD molecule production with
•−
41
energy to react with O to produce O radicals (Figure S11).
increasing Au loading on TiO
.
2
Second, the photocatalytic
2
2
Efficient charge separation by the CdS nanorods could lead to a
heterogeneous carrier distribution and compact location along
the tips. However, the photogenerated carriers can travel along
the CdS semiconductor nanorods, followed by their reaction to
create and localize surface-bound radicals. It should be noted
that although both mechanisms A and B are turned ON when
using both 405 and 532 nm lasers, the super-resolution image
activity can also be controlled by varying the wavelength of
incident light, as supported by the spectra of Au−CdS hybrid
heterostructures (Figure 1b, Figures S20 and S22). Therefore,
it should be noted that Au−CdS hybrid heterostructures enable
an efficient way to study and overcome important challenges in
photocatalysis, such as providing synthetic and optical handles
to control activity and perhaps, selectivity, in nanocatalysis.
The stability of Au−CdS heterostructures over our
experimental time and conditions needs to be discussed. We
recently demonstrated that metallic nanoparticles decorated on
the CdS semiconductor surface significantly enhance activity
and also greatly stabilize the CdS semiconductor nanorods
against photoinduced degradation. To clearly elucidate the
stability of our Au−CdS heterostructures, we carried out
ensemble measurements in the photocatalytic oxidation of
amplex red to resorufin (see more details in the Supporting
Information). We used a halogen lamp (150 W) to excite the
Au−CdS heterostructures, and this lamp produces a continuous
spectrum of light from 360 to 2000 nm. Therefore, under this
illumination, both mechanisms are turned ON and operating in
these heterostructures. We found that Au−CdS heterostruc-
tures are stable under our experimental conditions (Figures S25
and S26).
Because sunlight contains a wide range of wavelengths
between the ultraviolet, visible and near-infrared spectral
regions, we expect both mechanisms (A and B) to be active
under direct sunlight illumination. Distinguishing between
these two mechanisms and separating their individual
contributions to overall photocatalytic activity (turnover) and
selectivity is thus critical to fully understand, improve, and
devise new solar-to-chemical energy conversion technologies.
We expect these results will have an enormous positive impact
in the development of better photocatalytic structures for solar-
to-chemical energy conversion.
(Figure 6d) clearly differs from the super-resolution image
when using the 532 nm laser only (Figure 6c), revealing the
existence of mechanism B: Charge (electron) transfer from the
CdS semiconductor nanorod to the gold metal at the tip. We
note that mechanism B dominates when both lasers are used.
Given the similar power intensities, this observation indicates
that photogenerated electrons and holes appear to be more
active and longer-lived when generated at CdS than when
14
56
generated at Au (see below). Therefore, we have disclosed
the existence of two distinct photocatalysis mechanisms in Au−
CdS heterostructures. The photoinduced redox activity of these
Au−CdS heterostructures can be selectively turned ON by two
different mechanisms using two different wavelengths of
incident light.
We also compared the rates of product formation and
dissociation for mechanism A (532 nm) and mechanism B (405
and 532 nm) in Au-tipped CdS heterostructures during the
photocatalytic oxidation reaction. We found that the averaged
−1
rate of product formation ⟨τoff⟩ for mechanism B was about 6
times higher than that of mechanism A (Figure S22). In other
words, this indicates that mechanism B produces resorufin
molecules faster than mechanism A and that more resorufin
molecules are produced in a certain time under mechanism B.
This could be ascribed to the formation of more active and
long-lived photocarriers primarily in CdS nanorod under
56
mechanism B at both 405 and 532 nm. In addition, we
found two distinct dissociation kinetics in Au-tipped CdS
heterostructures for both mechanisms A and B (Figure S23).
−
1
−
CONCLUSIONS
In conclusion, we report on the first direct measurements to
resolve the nature and characteristic activities of two
The averaged rate of product dissociation ⟨τ ⟩ for the e
■
on
−1
−1
reactive sites in mechanism A was 1.46 s , while ⟨τ ⟩ in
mechanism B it was 1.57 s . The averaged rate of dissociation
τ ⟩ for the h reactive sites in mechanism A was 0.41 s ,
on
−1
−
1
+
−1
−
+
⟨
fundamental surface reactive sites (e and h ) generated by
photoinduced charge separation in single Au−CdS hybrid
heterostructures during photocatalytic oxidations as models for
other solar-to-chemical energy conversion reactions. Single-
molecule photocatalysis with high-resolution superlocalization
imaging allows us to reveal two distinct, incident energy-
dependent charge separation mechanisms that result in
completely opposite energy flows and polarities on single
Au−CdS heterostructures. This finding is very important for
the following reasons. First, it can help us design and develop
better metal−semiconductor heterostructures that are highly
on
−
1
−1
while ⟨τon⟩ in mechanism B it was 0.32 s .
There is an important point that needs to be further
discussed in Figure 6. Recently, Majima et al. proposed
mechanism A in Au−TiO hybrid particles to explain their
2
34
finding from visible-light-induced redox reactions. The results
obtained here on the specifically designed Au-tipped CdS
heterostructures directly verify the existence of two fundamen-
tally distinct charge separation mechanisms together with the
capability to optically control the direction of charge transfer on
the heterostructures at the nanoscale.
1
406
dx.doi.org/10.1021/ja409011y | J. Am. Chem. Soc. 2014, 136, 1398−1408

Journal of the American Chemical Society
Article
active for photocatalytic reactions under visible light. Second, it
enables a better understanding of the nature and catalytic
properties of single catalyst reactive sites in these hetero-
structures. Third, it permits us to potentially engineer the
direction of energy flows on the heterostructured nanomaterials
at the nanoscale. We therefore expect that our results have an
enormous potential impact on the development of better
photocatalyst structures.
(9) Wu, K.; Zhu, H.; Liu, Z.; Rodriguez-Cordoba, W.; Lian, T. J. Am.
Chem. Soc. 2012, 134, 10337.
(
10) Bao, N. Z.; Shen, L. M.; Takata, T.; Lu, D. L.; Domen, K. Chem.
Lett. 2006, 35, 318.
11) Peng, X. G.; Manna, L.; Yang, W. D.; Wickham, J.; Scher, E.;
Kadavanich, A.; Alivisatos, A. P. Nature 2000, 404, 59.
12) Mokari, T.; Rothenberg, E.; Popov, I.; Costi, R.; Banin, U.
Science 2004, 304, 1787.
13) Alemseghed, M. G.; Ruberu, T. P. A.; Vela, J. Chem. Mater.
011, 23, 3571.
(
(
(
2
EXPERIMENTAL SECTION
The Supporting Information contains more detailed experimental
methods.
(14) Ruberu, T. P. A.; Nelson, N. C.; Slowing, I. I.; Vela, J. J. Phys.
Chem. Lett. 2012, 3, 2798.
■
(15) Berr, M.; Vaneski, A.; Susha, A. S.; Rodriguez-Fernandez, J.;
Doblinger, M.; Jackel, F.; Rogach, A. L.; Feldmann, J. App. Phys. Lett.
2010, 97, 093108.
(16) Janet, C. M.; Viswanath, R. P. Nanotechnology 2006, 17, 5271.
(17) Menagen, G.; Macdonald, J. E.; Shemesh, Y.; Popov, I.; Banin,
U. J. Am. Chem. Soc. 2009, 131, 17406.
(18) Amirav, L.; Alivisatos, A. P. J. Phys. Chem. Lett. 2010, 1, 1051.
(19) Dukovic, G.; Merkle, M. G.; Nelson, J. H.; Hughes, S. M.;
Alivisatos, A. P. Adv. Mater. 2008, 20, 4306.
(20) Murdoch, M.; Waterhouse, G. I. N.; Nadeem, M. A.; Metson, J.
B.; Keane, M. A.; Howe, R. F.; Llorca, J.; Idriss, H. Nat. Chem. 2011, 3,
489.
(21) Costi, R.; Saunders, A. E.; Elmalem, E.; Salant, A.; Banin, U.
Nano Lett. 2008, 8, 637.
(22) Xiao, X.; Pan, S.; Jang, J. S.; Fan, F.-R. F.; Bard, A. J. J. Phys.
Chem. C 2009, 113, 14978.
(23) Novo, C.; Funston, A. M.; Mulvaney, P. Nat. Nanotechnol. 2008,
Single-Molecule Fluorescence Measurements. Single-mole-
cule fluorescence experiments were carried out on a prism-type dual-
color total internal reflection fluorescence (TIRF) microscope. A 532
nm laser beam (10 mW) was focused onto the sample to directly
excite both Au−CdS hybrid heterostructures and the resorufin product
molecules. Besides the 532 nm laser, a 405 nm laser beam (5 mW) was
also used to excite the Au−CdS heterostructures in this study. The
fluorescence from resorufin product during the oxidation reactions was
collected via a Nikon Plan Fluor 100× oil iris objective (NA = 1.4)
through a filter (532 LP, Chroma Technology Corp). All the
fluorescence images were captured with an Andor iXonEM+ 897
CCD camera (Belfast, Northern Ireland). The collected movies and
images were analyzed using MATLAB and NIH ImageJ (http://
rsbweb.nih.gov/ij/). The fluorescence intensity trajectories were
extracted from localized fluorescence spots individually across the
entire movie. The intensity of each bright spot in an image was
2
obtained by integrating the signal counts over an area of ∼1 μm .
3
, 598.
24) Tang, M. L.; Liu, N.; Dionne, J. A.; Alivisatos, A. P. J. Am. Chem.
Soc. 2011, 133, 13220.
25) Liu, N.; Tang, M. L.; Hentschel, M.; Giessen, H.; Alivisatos, A.
P. Nat. Mater. 2011, 10, 631.
26) De Cremer, G.; Sels, B. F.; De Vos, D. E.; Hofkens, J.; Roeffaers,
M. B. J. Chem. Soc. Rev. 2010, 39, 4703.
27) Chen, P.; Zhou, X.; Shen, H.; Andoy, N. M.; Choudhary, E.;
Han, K.-S.; Liu, G.; Meng, W. Chem. Soc. Rev. 2010, 39, 4560.
28) Tachikawa, T.; Majima, T. Chem. Soc. Rev. 2010, 39, 4802.
(
ASSOCIATED CONTENT
Supporting Information
■
(
*
S
Experimental details, additional results and discussion, and
supporting figures. This material is available free of charge via
the Internet at http://pubs.acs.org.
(
(
AUTHOR INFORMATION
Corresponding Authors
nfang@iastate.edu
(
■
(29) Xu, W.; Kong, J. S.; Yeh, Y.-T.; Chen, P. Nat. Mater. 2008, 7,
9
92.
(
30) Xu, W.; Kong, J. S.; Chen, P. Phys. Chem. Chem. Phys. 2009, 11,
767.
31) Xu, W.; Jain, P. K.; Beberwyck, B. J.; Alivisatos, A. P. J. Am.
Chem. Soc. 2012, 134, 3946.
32) Zhou, X.; Andoy, N. M.; Liu, G.; Choudhary, E.; Han, K.-S.;
Shen, H.; Chen, P. Nat. Nanotechnol. 2012, 7, 237.
33) Roeffaers, M. B. J.; Sels, B. F.; Uji-i, H.; De Schryver, F. C.;
Jacobs, P. A.; De Vos, D. E.; Hofkens, J. Nature 2006, 439, 572.
34) Tachikawa, T.; Yonezawa, T.; Majima, T. ACS Nano 2012, 7,
63.
35) De Cremer, G.; Roeffaers, M. B. J.; Bartholomeeusen, E.; Lin,
vela@iastate.edu
2
(
Notes
The authors declare no competing financial interest.
(
ACKNOWLEDGMENTS
■
(
This work was supported by the Laboratory Directed Research
and Development Program of the Ames Laboratory, U.S.
Department of Energy. The Ames Laboratory is operated for
the U.S. Department of Energy by Iowa State University under
contract no. DE-AC02-07CH11358.
(
2
(
K.; Dedecker, P.; Pescarmona, P. P.; Jacobs, P. A.; De Vos, D. E.;
Hofkens, J.; Sels, B. F. Angew. Chem., Int. Ed. 2010, 49, 908.
REFERENCES
(36) Roeffaers, M. B. J.; De Cremer, G.; Libeert, J.; Ameloot, R.;
■
(
(
(
1) Fujishima, A.; Honda, K. Nature 1972, 238, 37.
2) Gratzel, M. Nature 2001, 414, 338.
Dedecker, P.; Bons, A.-J.; Buckins, M.; Martens, J. A.; Sels, B. F.; De
Vos, D. E.; Hofkens, J. Angew. Chem., Int. Ed. 2009, 48, 9285.
(37) Shaviv, E.; Schubert, O.; Alves-Santos, M.; Goldoni, G.; Di
̈
3) Cho, I. S.; Chen, Z.; Forman, A. J.; Kim, D. R.; Rao, P. M.;
Jaramillo, T. F.; Zheng, X. Nano Lett. 2011, 11, 4978.
4) Shankar, K.; Basham, J. I.; Allam, N. K.; Varghese, O. K.; Mor, G.
Felice, R.; Vallee
Nano 2011, 5, 4712.
́
̈
, F.; Del Fatti, N.; Banin, U.; Sonnichsen, C. ACS
(
K.; Feng, X.; Paulose, M.; Seabold, J. A.; Choi, K.-S.; Grimes, C. A. J.
Phys. Chem. C 2009, 113, 6327.
(38) Robinson, R. D.; Sadtler, B.; Demchenko, D. O.; Erdonmez, C.
K.; Wang, L.-W.; Alivisatos, A. P. Science 2007, 317, 355.
(39) Saunders, A. E.; Popov, I.; Banin, U. J. Phys. Chem. B 2006, 110,
25421.
(
5) Walter, M. G.; Warren, E. L.; McKone, J. R.; Boettcher, S. W.; Mi,
Q.; Santori, E. A.; Lewis, N. S. Chem. Rev. 2010, 110, 6446.
(
6) Lewis, N. S. Science 2007, 315, 798.
(40) Mongin, D.; Shaviv, E.; Maioli, P.; Crut, A.; Banin, U.; Fatti, N.
(
7) Sunada, K.; Kikuchi, Y.; Hashimoto, K.; Fujishima, A. Environ. Sci.
D.; Vallee, F. ACS Nano 2012, 6, 7034.
Technol. 1998, 32, 726.
8) Kumar, S.; Jones, M.; Lo, S. S.; Scholes, G. D. Small 2007, 3,
633.
(41) Mukherjee, S.; Libisch, F.; Large, N.; Neumann, O.; Brown, L.
V.; Cheng, J.; Lassiter, J. B.; Carter, E. A.; Nordlander, P.; Halas, N. J.
Nano Lett. 2012, 13, 240.
(
1
1
407
dx.doi.org/10.1021/ja409011y | J. Am. Chem. Soc. 2014, 136, 1398−1408

Journal of the American Chemical Society
Article
(
(
(
(
(
42) Moskovits, M. Science 2011, 332, 676.
43) Anderson, P. A. Phys. Rev. 1959, 115, 553.
44) Frischkorn, C.; Wolf, M. Chem. Rev. 2006, 106, 4207.
45) Christopher, P.; Xin, H.; Linic, S. Nat. Chem. 2011, 3, 467.
46) Mubeen, S.; Lee, J.; Singh, N.; Kramer, S.; Stucky, G. D.;
Moskovits, M. Nat. Nanotechnol. 2013, 8, 247.
47) Mubeen, S.; Hernandez-Sosa, G.; Moses, D.; Lee, J.; Moskovits,
M. Nano Lett. 2011, 11, 5548.
48) Lee, J.; Mubeen, S.; Ji, X.; Stucky, G. D.; Moskovits, M. Nano
Lett. 2012, 12, 5014.
(
(
(
(
(
49) Jakob, M.; Levanon, H.; Kamat, P. V. Nano Lett. 2003, 3, 353.
50) Ruberu, T. P. A.; Vela, J. ACS Nano 2011, 5, 5775.
51) Ishibashi, K.-I.; Fujishima, A.; Watanabe, T.; Hashimoto, K. J.
Phys. Chem. B 2000, 104, 4934.
52) Hoffmann, M. R.; Martin, S. T.; Choi, W.; Bahnemann, D. W.
Chem. Rev. 1995, 95, 69.
(
(
(
53) Thompson, T. L.; Yates, J. T. Chem. Rev. 2006, 106, 4428.
54) Chih, T.; Jao, H.-J.; Wang, C. M. J. Electroanal. Chem. 2005, 581,
159.
(
(
55) Gorris, H. H.; Walt, D. R. J. Am. Chem. Soc. 2009, 131, 6277.
56) Wu, K.; Rodríguez-Cordoba, W. E.; Yang, Y.; Lian, T. Nano Lett.
́
2
(
5
(
013, 13, 5255.
57) Fujishima, A.; Zhang, X.; Tryk, D. A. Surf. Sci. Rep. 2008, 63,
15.
58) Chong, R.; Rho, J.-E. R.; Yoon, H. J.; Rho, T.-H. D.; Park, P. S.;
Kim, Y.-H.; Lee, J. H. Biosens. Bioelectron. 2012, 32, 19.
59) Chong, R.; Rho, J.-E. R.; Yoon, H.-J.; Park, P. S.; Rho, T.-H. D.;
Park, J. Y.; Park, L.; Kim, Y.-H.; Lee, J. H. Talanta 2013, 116, 403.
60) Thetford, A.; Hutchings, G. J.; Taylor, S. H.; Willock, D. J. Proc.
Roy. Soc. A 2011, 467, 1885.
61) Baptista, L.; Clemente da Silva, E.; Arbilla, G. Phys. Chem. Chem.
Phys. 2008, 10, 6867.
(
(
(
(
(
62) Popham, P.; Novacky, A. Plant Physiol. 1991, 96, 1157.
63) Giovanelli, D.; Lawrence, N. S.; Jiang, L.; Jones, T. G. J.;
Compton, R. G. Sens. Actuators, B 2003, 88, 320.
64) Lawrence, N. S.; Davis, J.; Jiang, L.; Jones, T. G. J.; Davis, S. N.;
Compton, R. G. Analyst 2000, 125, 661.
65) Costentin, C.; Robert, M.; Saveant, J.-M. J. Am. Chem. Soc. 2006,
28, 8726.
(
(
́
1
(
66) Burkitbaev, M. High Energy Chem. 2003, 37, 216.
(
67) Tiee, J. J.; Wampler, F. B.; Oldenborg, R. C.; Rice, W. W. Chem.
Phys. Lett. 1981, 82, 80.
68) Fang, H.-J.; Dong, W.-B.; Zhang, R.-X.; Hou, H.-Q. Acta Phys.
Chim. Sin. 2006, 22, 761.
(
1
408
dx.doi.org/10.1021/ja409011y | J. Am. Chem. Soc. 2014, 136, 1398−1408