Suppression of band edge migration at the p-GaInP2/H2O interface under illumination via catalysis

Article abstract of DOI:10.1021/jp000387s

The band edges of p-GaInP₂ are observed to migrate toward negative potentials during current flow under illumination in solutions with pH ranging from 1 to 14.5. The migration is not caused by a change in the pH of the semiconductor microenvironment but is a result of accumulation of photogenerated electrons at the p-GaInP₂/water interface due to poor interfacial kinetics. This less than optimal interfacial charge-transfer rate can be catalyzed by treating the surface with transition-metal ions (e.g., Ru^III, Rh^III, Co^III, Os^III) which results in a suppression of band edge migration. As compared to an unmodified p-GaInP₂ surface, the metal-ion treatment does not induce any appreciable band edge shift in the dark but effectively suppresses the band edge migration under illumination. Ru^III and Rh^III are found to act as better hydrogen-evolution catalysts than electrodeposited Pt.

Full text of DOI:10.1021/jp000387s

J. Phys. Chem. B 2000, 104, 6591-6598
6591
Suppression of Band Edge Migration at the p-GaInP2/H2O Interface under Illumination via
Catalysis
Ashish Bansal† and John A. Turner*
National Renewable Energy Laboratory, 1617 Cole BouleVard, Golden, Colorado 80401-3393
ReceiVed: January 28, 2000; In Final Form: April 9, 2000
The band edges of p-GaInP2 are observed to migrate toward negative potentials during current flow under
illumination in solutions with pH ranging from 1 to 14.5. The migration is not caused by a change in the pH
of the semiconductor microenvironment but is a result of accumulation of photogenerated electrons at the
p-GaInP2/water interface due to poor interfacial kinetics. This less than optimal interfacial charge-transfer
rate can be catalyzed by treating the surface with transition-metal ions (e.g., RuIII, RhIII, CoIII, OsIII) which
results in a suppression of band edge migration. As compared to an unmodified p-GaInP2 surface, the metal-
ion treatment does not induce any appreciable band edge shift in the dark but effectively suppresses the band
edge migration under illumination. RuIII and RhIII are found to act as better hydrogen-evolution catalysts than
electrodeposited Pt.
Introduction
At present, however, there is no information on the energetics
(position) of the band edges or on the kinetics of charge transfer
at the p-GaInP2/water interface under illumination, i.e., under
conditions of H2 evolution. Several impedance measurements
in the literature have previously reported that band edges of
semiconductors are not fixed with respect to the reference in
the solution and are prone to migration under illumination.5-15
These changes in flatband potential under illumination have been
interpreted to indicate evidence of surface effects such as charge
trapping, adsorption, or changes in the surface chemistry. The
band edge positions are also known to be strongly affected by
interfacial kinetics under illumination.9,11 Increased band edge
migration as a result of illumination has been observed for
interfaces with poor interfacial charge-transfer ability, and this
migration has been suppressed by treating the semiconductor
surfaces with transition-metal catalysts.10,16
In this paper we describe our work on evaluating the effect
of interfacial kinetics on the energetics of the p-GaInP2/water
interface by determining the band edge positions of bare
p-GaInP2 surfaces under illumination, i.e., under conditions of
H2 evolution. Capacitance-voltage (C-V) and current-voltage
(I-V) measurements were done under various levels of il-
lumination, in aqueous solutions of various pH to evaluate the
extent of band edge migration. In addition, we describe the
results of C-V and I-V measurements on p-GaInP2 surfaces
modified with transition-metal ions to evaluate their effective-
ness as reagents for inducing band edge shifts in the dark and
as electrocatalysts for suppressing band edge migration under
illumination. In this study, the flow of photogenerated cathodic
current at the semiconductor surface under illumination corre-
sponds to generation of hydrogen from photoelectrolysis of
water.
Photoelectrochemical splitting of water to produce hydrogen
and oxygen using sunlight is a promising application of
semiconductor electrochemistry, with increasing relevance in
the present times of heightened environmental consciousness.
To obtain a stable and efficient practical semiconductor based
system, four criteria must be simultaneously satisfied. The band
gap of the semiconductor must be greater than 1.7 eV, the band
edges of the semiconductor must overlap the H2/H2O and O2/
H2O redox potentials under H2/O2 eVolution condition, and
charge transfer across the semiconductor-liquid interface should
be fast (Figure 1). In addition, the semiconductor surface should
be chemically stable in the aqueous media.
Although p-type GaInP2 with a band gap of 1.8-1.9 eV has
been identified as a promising photocathode for water splitting,
its band edges are 0.2-0.4 V too negative to effect photoelectroly-
sis.1-3 Electrochemical investigations of GaInP2 in aqueous
media have found that the semiconductor surface is unstable
and susceptible to corrosion.4 Furthermore, it is not known if
the charge-transfer rate from p-GaInP2 to water under illumina-
tion is sufficiently high or if this rate can be further catalyzed.
Previously, Kocha et al. carried out surface modification of
GaInP2 to shift the semiconductor band edges positive.2,3
Capacitance-voltage (C-V) measurements in the dark and
current-voltage (I-V) measurements under illumination indi-
cated that the GaInP2 band edges could be shifted up to 0.3 V
positive by adsorbing 8-quinolinol to the GaInP2 surface.
However, their attempts to split water with the quinolinol-
modified surface proved unsuccessful. It was speculated that
further treatment with a catalyst was necessary to lower the
hydrogen and oxygen evolution overpotentials before photo-
electrolysis of water could be achieved. That work implicitly
assumed that during hydrogen evolution under illumination, the
band edges of GaInP2 remained in the same energetic position
as in the dark.
Note: A semiconductor surface treatment, such as the
adsorption of a transition-metal ion, can have two distinct effects
on the energetics of the interface as described below. Because
of change of net charge in the Helmholtz layer, adsorption of a
transition metal as a charged species (ion) on the semiconductor
surface can change the band edge position relative to the band
edge position for an underivatized surfacesthe band edge
* To whom correspondence should be addressed. Fax: (303) 384-6655.
E-mail: john_turner@nrel.gov.
† E-mail: ashish_bansal@nrel.gov.
10.1021/jp000387s CCC: $19.00 © 2000 American Chemical Society
Published on Web 06/22/2000

6592 J. Phys. Chem. B, Vol. 104, No. 28, 2000
Bansal and Turner
illumination is compared with the band edge position for the
same surface in the dark. Although both of these effects have
been individually described in the literature as “shifts” of band
edges,5-15 we have used the terms “shift” and “migration” in
this paper to distinguish between the two.
Experimental Section
Materials. The zinc-doped 3 µm thick p-type Ga0.51In0.49
P
epilayers, henceforth referred to as the p-GaInP2 epilayers, were
grown by atmospheric-pressure organometallic vapor-phase
epitaxy on approximately 350 µm thick Zn-doped GaAs
substrates misoriented from the (100) surface by 2° toward
(110). A growth temperature of 700 °C, growth rate of 4 µm/h,
and phosphine pressure of 4.5 Torr were used.17 The carrier
concentration in the GaInP2 layer was (6-12) × 1016 cm-3
.
Ohmic contacts were made by electrochemically depositing gold
onto the unpolished backsides of the substrate. The wafers were
then cut into small pieces and were attached to a wire using
electrically conductive silver epoxy (H-31, Epoxy Technology
Inc.). After curing the Ag epoxy in an oven at 80 °C for a few
hours, the backside and edges of the electrode were sealed with
a nonconducting epoxy (Hysol 9462, Dexter Corp.), leaving only
the polished surface (0.1-0.2 cm2) of the crystal exposed. All
electrodes were etched in concentrated H2SO4 for durations
between 5 and 60 s before electrochemical measurements.
Chemicals. KOH (J. T. Baker), concentrated H2SO4 (J. T.
Baker), and K2SO4 (Aldrich) were used as received. Phthalate
(pH 4), phosphate (pH 7), and carbonate (pH 10) buffer solutions
were obtained from Beckman. OsCl3, RuCl3, RhCl3, Co(NH3)6-
Cl3, and K2PtCl6 were obtained from Strem and used as received.
All metal-ion solutions were made with house deionized water.
Metal-Ion Treatment. For metal-ion chemisorption, 0.010
M solutions of OsCl3, RuCl3, and RhCl3 were made from dilute
HCl (pH 1.5), while a 0.010 M solution of Co(NH3)6Cl3 was
made in 0.010 M KOH(aq) (pH 12.0). A pH 1.5 solution was
made of 7.7 mM K2PtCl6 using H2SO4. In a typical metal-ion
(RuIII, RhIII, OsIII, CoIII) treatment, a p-GaInP2 electrode was
etched in concentrated H2SO4 for 60 s, dried in a stream of
nitrogen, and was immediately immersed in a metal-ion solution.
After 60 s, the electrode was emmersed, rinsed with house-
distilled water, and blown dry under a stream of nitrogen. Upon
immersion of these electrodes in 3 M KOH solution for
electrochemical measurements, some metal-ion compound was
observed to desorb off the epoxy and into the solution. Fresh 3
M KOH solution was used for each experiment to avoid any
trace metal cross-contamination. Pt was electrodeposited by
electrochemically biasing the p-GaInP2 electrode in the K2PtCl6
solution at -1.0 V vs SCE for approximately 15 s under
illumination, corresponding to a reductive current density of
∼1 mA/cm2. No hydrogen bubbles were seen to evolve during
Pt deposition. The p-GaInP2 surfaces, which were only etched
and were not immersed in any metal-ion solution, are referred
to as “untreated” surfaces.
Figure 1. (a) Band gap criteria: The band gap of the semiconductor
must be greater than 1.7 eV. ηa and ηc indicate anodic and cathodic
overpotentials for electrode reactions, respectively. (b) Band edge
overlap criteria: The band edges of the semiconductor must overlap
the H2/H2O and O2/H2O redox potentials under H2/O2 evolution
condition. (c) Fast kinetics criteria: Charge transfer across the
semiconductor-liquid interface must be fast. For a successful hydrogen-
producing PEC system, all three criteria must be simultaneously
satisfied.
positions for both the derivatized and the underivatized surfaces
being compared vs a solution reference in the dark. This effect
is referred to as band edge “shift” or flatband “shift” in this
paper. On the other hand, the measured band edge position vs
a solution reference can also move as a result of illumination
due to accumulation of photogenerated charges at the interface.
This movement is not due to a change in the charge in the
Helmholtz layer. Its extent is determined by the level of
illumination at the semiconductor surface and the interfacial
charge-transfer kinetics. This movement can be suppressed if
the transition-metal species acts as an electrocatalyst and
catalyzes charge transfer across the interface, thereby reducing
the accumulation of charge at the interface. In this paper, the
movement of band edges due to photoelectron accumulation is
referred to as “migration”. In measuring the “migration” effect,
the band edge position of a surface under a given level of
Electrochemical Measurements. Capacitance-voltage (C-
V) and current-voltage (I-V) measurements were performed
in a conventional three-electrode setup to deconvolute the effects
of counter electrode kinetics from those of the semiconductor
electrode. The setup consisted of a p-GaInP2 working electrode
(∼0.08-0.13 cm2), Pt gauze (∼2 cm2) counter electrode, and a
saturated calomel electrode (SCE) as the reference electrode.
The samples were illuminated with a fiber optic illuminator
housing an EKE type 150 W tungsten-halogen lamp. Data were
collected using a Solartron 1286 Electrochemical Interface
connected to a Solartron SI 1260 Impedance/Gain-Phase

Band Edge Migration at the p-GaInP2/H2O Interface
J. Phys. Chem. B, Vol. 104, No. 28, 2000 6593
SCHEME 1
Analyzer. Simultaneous C-V and I-V measurements were
made at a frequency of 10 kHz with a 20 mV p-p ac amplitude
and a scan rate of 20 mV/s. For measurements in the dark, data
were usually collected in a potential window of ∼1.3 V. The
limits of this window with respect to SCE depended upon the
pH of the solution and ranged from -1.7 to +0.3 V vs SCE.
The negative end of the scan window was extended up to -2.5
V, as necessary, for collecting data under illumination. Since
the instrument measured the ac and the dc current signals across
the same resistor, the current sensitivity of the setup had to be
adjusted as necessary for higher light intensities, and therefore
the C-V data for higher light intensities are less accurate than
those at lower light intensities or in the dark.
Impedance Data Analysis. A detailed impedance analysis
of the p-GaInP2/water system in the dark has been previously
reported, in which capacitance data in the frequency range 500
< f < 105 Hz, were attributed to the space charge layer of the
semiconductor.3 For the purpose of this study the series RC
circuit (Scheme 1), where Rs is the series resistance of the circuit
and Csc is the capacitance of the space charge region, was
deemed as the simplest circuit that could adequately describe
the trends in the movement of the flatband potential under
illumination.18 The slope and x-axis intercept of the Mott-
Schottky plots were used to determine dopant density and the
flatband potentials, respectively.
Results
Flatband Potentials of Etched p-GaInP2/Water Interfaces.
Figure 2 shows three representative Mott-Schottky plots for
etched p-GaInP2 electrodes immersed in 0.20 M K2SO4 solutions
of pH 1, 7, and 13. Data collected at each pH in the dark and
under illumination show a negative migration of the flatband
potential with increasing levels of light intensity. Figure 3a
shows the corresponding flatband potentials calculated from
similar Mott-Schottky plots for five different pH values. The
curves for pH 1 and 13 were measured in unbuffered solutions
of 0.20 M K2SO4, which were made acidic or basic using the
necessary amount of H2SO4 or KOH, respectively. The curves
for pH 4, 7, and 10 were measured in commercial phthalate
(pH 4), phosphate (pH 7), and carbonate (pH 10) buffer
solutions.
Figure 3b shows the same data as the net migration of the
flatband potentials measured vs the values observed in the dark.
It is seen that even at low levels of illumination, e.g., ∼0.1
mA/cm2, the GaInP2 flatband potentials (band edges) move up
to 0.3 V negative as compared to their positions in the dark.
As the level of illumination is increased, the negative migration
of the band edges becomes more pronounced, with values
reaching almost 1.0 V for current densities around 10 mA/cm2.
Effect of Transition-Metal Ions on the Flatband Potential
of the p-GaInP2/Water Interface in the Dark. Figure 4 shows
the flatband potentials of p-GaInP2 electrodes, measured in 3
M KOH solutions in the dark, which have been exposed to four
different transition-metal-ion solutions before immersion in the
KOH solution. In addition, data are also included for photo-
electrochemically deposited Pt for comparison. No significant
shifts in measured flatband potentials were observed in the dark
as a result of treatment with the five transition metals. The
maximum shift exhibited by the metal-ion-treated surface is
∼0.05 V.
Figure 2. Mott-Schottky plots of p-GaInP2 electrodes immersed in
pH 1, 7, and 13, measured in the dark and under varying levels of
illumination.
Effect of Transition-Metal Ions on the Flatband Potential
of the p-GaInP2/Water Interface under Illumination. Figure
5 shows Mott-Schottky plots for an untreated and a RuIII-treated
p-GaInP2 surface, measured in the dark and under illumination
in 3 M KOH solution. Figure 6a shows the corresponding
flatband potentials calculated from similar Mott-Schottky plots
of etched and metal-ion-treated p-GaInP2 electrodes in the 3 M
KOH solution. In addition, the figure also shows flatband
potentials measured with a GaInP2 electrode modified with
electrodeposited platinum. Figure 6b shows the same data as
migration of the flatband potentials measured vs the values
observed in the dark. While the etched only (untreated) surface
exhibits significant migration of the flatband potential under
illumination, the migration is significantly suppressed for the
metal-ion-treated surfaces. Whereas, corresponding to a hydro-
gen evolution current density of 0.7-1.0 mA/cm2, the underiva-
tized surface showed a -0.34 V migration in flatband potential
with respect to its position in the dark and the Pt-coated electrode
shows a -0.24 mV migration, the RhIII- and RuIII-treated
surfaces showed only 0.08-0.09 V migration under similar
conditions. Table 1 lists the flatband potentials (Vfb) observed
for the underivatized and the metal-ion-treated surfaces in the
dark and under illumination. The “light” data reported corre-
sponds to a current density of 0.7-1.0 mA/cm2.
Effect of Transition-Metal Ions on the I-V Properties of
the p-GaInP2/Water Interface. Figure 7 shows the effect of
metal-ion treatment on the current-voltage properties of the

6594 J. Phys. Chem. B, Vol. 104, No. 28, 2000
Bansal and Turner
Figure 5. Mott-Schottky plots for (a) an untreated and (b) a RuIII-
treated p-GaInP2 surface, measured in the dark and under illumination
in 3 M KOH solution.
Discussion
Figure 3. (a) Flatband potentials of etched p-GaInP2 electrodes,
measured vs SCE, at five different pH values for varying levels of
illumination. (b) Migration of flatband potentials vs the Vfb values
observed in the dark (Vfb,light - Vfb,dark) in solutions of five different
pH.
A practical single semiconductor photoelectrochemical system
for water splitting requires that the three conditions of band
gap, band edge overlap, and fast kinetics must be simultaneously
satisfied (Figure 1). The first two conditions are determined,
respectively, by the energetics of the semiconductor and the
semiconductor/liquid junction, while the third condition is
determined by the kinetics of the reaction of interest at the
interface. In the present case, the reaction of interest is
generation of hydrogen from photoelectrolysis of water. From
the above description it would seem that in a p-type semicon-
ductor/liquid junction cell, the first two conditions would be
the only necessary and sufficient conditions for H2 evolution at
the semiconductor interface. However, we have shown here that
fast interfacial kinetics is also a necessary condition for any
significant amount of hydrogen evolution. The kinetics of
interfacial charge transfer strongly affects the energetics of a
semiconductor/liquid junction. Furthermore, we have shown (as
also have others7-15,19-21) that it is erroneous to assume that
the band edge positions of the semiconductors remain in the
same position during current flow under illumination as in the
dark, except perhaps for interfaces with extremely fast interfacial
charge-transfer rates. This realization is significant for GaInP2
because it implies that any surface modification that is applied
to shift semiconductor band edges to achieve band edge overlap
must also not compromise interfacial kinetics or else any band
edge shifts could be nullified by band edge migration under
illumination.2,3
Figure 4. Flatband potentials of untreated (etched only) and metal-
ion-treated p-GaInP2 electrodes in the dark in 3 M KOH solution.
untreated (etched only) and metal-ion-treated p-GaInP2 elec-
trodes in 3 M KOH solution. The metal-ion-treated surfaces
exhibit improved fill factors and decreased hysteresis as
compared to the untreated surface. The open circuit voltages
(Voc) observed for the metal-ion-treated surfaces were generally
more positive than the untreated surface. Table 1 lists the Voc
values observed for the underivatized and the metal-ion-treated
surfaces for the positive and the negative going scans and
corresponds to a current density of 0.7-1.0 mA/cm2.
Mott-Schottky measurements described above show that the
band edges (flatband potentials) of p-GaInP2 migrate toward
negative potentials, under illumination. We attribute this migra-
tion to the accumulation of photogenerated electrons at the
p-GaInP2/water interface due to poor interfacial charge-transfer

Band Edge Migration at the p-GaInP2/H2O Interface
J. Phys. Chem. B, Vol. 104, No. 28, 2000 6595
H2O + e- f 1/2H2 + OH-(base)
H+ + e- f 1/2H2(acid)
(1)
(2)
Increased basicity (decreased acidity) of the semiconductor
electrode microenvironment can manifest itself as a negative
movement of flatband potentials when measured against a pH
insensitive electrode like the SCE.8,22
Both electron accumulation and OH- formation (H+ con-
sumption) can cause a negative movement in the position of
the semiconductor band edges and this movement is exacerbated
for both processes by increasing light intensity (Figure 8).
Fortunately, the two effects can be distinguished. As shown in
Figure 8a, a pH change displaces the semiconductor band edges
and the H2/H2O and O2/H2O redox potentials in the same
direction and roughly by the same amount for a pH sensitive
semiconductor surface like p-GaInP2.1 In contrast, as shown in
Figure 8b, electron accumulation at the surface displaces only
the semiconductor band edges negative and does not move the
solution redox potentials. In other words, while a pH change
does not move the band edge position with respect to the
hydrogen and oxygen eVolution potentials, accumulation of
photogenerated electrons does. Differentiating between the two
is important because if, as a result of surface modification, the
“band edge overlap condition” can be achieved in the dark, then
it will not be affected when the interfacial pH changes as a
result of illumination (Figure 8a), whereas it will be affected
by electron accumulation (due to poor interfacial charge-transfer
kinetics) under illumination (Figure 8b).
To minimize the effects of pH change of the semiconductor
microenvironment in our study, the experiments were done
either in highly acidic or basic solutions or in buffered solutions.
It is expected that at extreme pH values, the change in pH due
to generation of base via reaction 1 or consumption of acid via
reaction 2 would be minimal due to the high buffering capacity
arising from high concentrations of OH- and H+ ions in the
solution, respectively. Hence, at these extreme pH values, the
displacement of the band edge due to illumination can be
attributed primarily to poor rate of charge transfer across the
GaInP2/water interface. For the intermediate pH region, both
phenomena can be significant. Therefore, for intermediate pH
values buffered solutions were used to maintain the pH of the
electrode microenvironment at values similar to the pH of the
bulk solution.
Figures 2 and 3 show that the band edge positions of p-GaInP2
in water are highly dependent on the level of illumination of
the electrode irrespective of the pH of the solution. Even at
low short circuit current densities (Jsc) corresponding to low
light intensities, where the change in the pH of the electrode
microenvironment can be considered negligible, a negative
migration of the flatband potential of up to 0.3 V is observed
as compared to its value in the dark. This migration is further
enhanced by as much as 1.0 V at higher Jsc values, suggesting
strong accumulation of photogenerated electrons at the inter-
face.9,11,13-15,19-21 Another point of note is that the migration
of flatband potential is greater for pH 13 than for pH 1 (Figure
3b). This difference in the extent of migration is not unexpected,
as the charge-transfer reaction here corresponds to hydrogen
production and it is well-known that hydrogen production is
more facile in acidic solutions than in basic solutions of similar
concentrations, especially at an uncatalyzed surface.
Figure 6. (a) Flatband potentials of untreated (etched only) and metal-
ion-treated p-GaInP2 electrodes under illumination in the 3 M KOH
solution. (b) Migration of flatband potentials vs the Vfb values observed
in the dark (Vfb,light - Vfb,dark) in 3 M KOH solution.
TABLE 1: Capacitance-Voltage and Current-Voltage
Measurements on Untreated and Metal-Treated p-GaInP2
Surfaces in 3 M KOH Solution
Vfb,dark, V Vfb,light,a V
Voc,p,c V Voc,n,c V
vs SCE
vs SCE
∆Vfb,b V vs SCE vs SCE
untreated
-0.19
-0.15
-0.15
-0.14
-0.15
-0.18
-0.53
-0.39
-0.24
-0.44
-0.38
-0.26
-0.34
-0.24
-0.09
-0.30
-0.23
-0.08
-0.88
-0.67
-0.67
-0.88
-0.77
-0.64
-0.52
-0.40
-0.41
-0.37
-0.43
-0.54
Pt treated
RuIII treated
OsIII treated
CoIII treated
RhIII treated
a Vfb values measured at an illumination level corresponding to a
current density of 0.7-1.0 mA/cm2. b Band edge migration under
c
illumination (Vfb,light - Vfb,dark). Voc values observed for positive (p)
and negative (n) going scans corresponding to a current density of 0.7-
1.0 mA/cm2.
kinetics.9,11,13-15,19-21 At higher light intensities, as more current
is forced through the interface, a greater equilibrium concentra-
tion of electrons is required to maintain that current and, hence,
the band edge migration increases with light intensity.
Alternatively, at least some of the negative movement of the
flatband potential observed under illumination could also be
attributed to a change in the pH of the microenvironment of
the semiconductor electrode surface.22 At a p-type semiconduc-
tor, hydrogen evolution at the interface under illumination is
accompanied by the generation of OH- (or consumption of H+).
The reaction at the interface can be represented as
Further proof that these displacements of flatband potentials
are not due to changes in the pH of the electrode microenvi-
ronment comes from data in Figures 5 and 6. Negative

6596 J. Phys. Chem. B, Vol. 104, No. 28, 2000
Bansal and Turner
Figure 7. Current-voltage (I-V) curves observed for untreated and metal-ion-treated p-GaInP2 electrodes in 3 M KOH solution.
OH- production via reaction 1.22 The suppressed migration
observed on ruthenium- and rhodium-treated surfaces (Figure
6) further supports the argument that the primary cause of band
edge migration here is electron accumulation and not a change
in the interfacial pH.10 Note that in addition to the arguments
presented above, when data from the metal-ion-treated surfaces
are compared with the data obtained from the untreated surface
in the same solution and at the same photocurrent density, any
residual effects of surface pH change are further nullified. This
truly allows us to attribute any band edge migration effects to
poor interfacial charge-transfer kinetics. These results clearly
indicate that the charge-transfer rate from etched p-GaInP2 to
water is significantly less than optimal and must be further
catalyzed. In fact, these results underscore the necessity of
catalyzing charge transfer across the p-GaInP2/water interface
in order to suppress the undesired migration of the band edges
under illumination.
It is well-known that metal ions deposited onto semiconductor
electrodes act as effective catalysts for transferring charges
across a semiconductor/liquid interface.10,16,23-39 Transition-
metal-ion treatment has been shown to increase the charge-
transfer rate at the InP/water interface,16,24-27 CdTe/Sn2+/4+
interface,28-31 GaP/water interface,32-34 and the GaAs/Se-/2-
-
(aq) interface.10,35-39 Figures 5-7 show that this catalytic
behavior is applicable to the p-GaInP2/water interface also. In
fact, all four metal ions used in this study were found capable
of electrocatalysis, albeit to different extents, as evident by the
decrease in the band edge migration under illumination.
Since electrodeposited platinum has been widely used as a
hydrogen evolution catalyst in the literature,16,26,34,40-44 we
compared its effectiveness with the four transition-metal ions
evaluated in this study. It was observed (Figure 6) that ruthenium
and rhodium deposited in an electroless manner exhibited the
greatest catalytic effect, even greater than electrochemically
deposited platinum. The catalytic effect of the metal-ion
treatment was also apparent as the improvement in the fill factors
and a decrease in hysteresis of the I-V curves in Figure 7.10,37,39
The flat-band potential and, therefore, the energetics of the
band edges of any semiconductor/electrolyte system are con-
trolled by the charge in the Helmholtz inner layer. Any change
Figure 8. (a) A pH change displaces the semiconductor band edges
and the H2/H2O and O2/H2O redox potentials in the same direction and
roughly by the same amount for a pH sensitive semiconductor surface
(e.g., p-GaInP2). (b) Electron accumulation at the surface moves only
the semiconductor band edges negative and does not move the solution
redox potentials. Both sides of each figure are shown at the same band
bending (capacitance). The dashed line indicates the potential of a pH
insensitive reference electrode (e.g., SCE).
migrations of flatband potential are also seen for the etched
p-GaInP2 surface in the highly basic 3 M KOH solutions (Figure
6, open circles), where one would not expect any further
basification of the electrode microenvironment as a result of

Band Edge Migration at the p-GaInP2/H2O Interface
J. Phys. Chem. B, Vol. 104, No. 28, 2000 6597
in the charge in that layer changes the measured flat-band
potential and subsequently the energetics of the band edges.
The variation of the semiconductor flat-band potential with pH
is due to the change in the adsorbed charge (H+ and OH-) on
the electrode surface. Adding negative charge (such as OH-
ions) shifts the bands to more negative potentials, whereas
adsorption of a positive charge (such as H+ ions) shifts the bands
positive.
and small positive increases in the Voc. Light-limited current
levels are reached at lower applied voltages. RuIII and RhIII
treatments are found to suppress band edge migration better
than electrochemically deposited platinum treatment, which is
widely used as a hydrogen evolution catalyst in the literature.
The observation that even the best catalysts of this study (RuIII
and RhIII) show ∼0.25 V band edge migration at a current
density of ∼10 mA/cm2 suggests that other catalysts can be
identified that can support higher currents through the interface
without causing appreciable band edge migration. Details of
such work will be reported shortly.
Although the metal-ion treatment does not shift the band
edges of p-GaInP2 positive enough to affect unassisted photo-
electrolysis, nevertheless, it can be applied to tandem cell
designs where it can reduce the voltage assistance required by
the p-GaInP2 top layer to split water, thereby increasing the
efficiency (by decreasing the energy input required) of assisted
hydrogen production.44 Work in this area is currently in progress.
Transition-metal ions can also shift the flatband potential and,
therefore, the band edge positions by adsorption of ionic charges
at the semiconductor surface (within the Helmholtz layer).10 In
contrast to significant catalytic effects (vide supra), the metal-
ion treatment in our experiments did not exhibit any significant
band edge shifts in our study. The maximum shift exhibited by
the metal-ion-treated surface in the dark was ∼0.05 V, which
is significantly less than the 0.2-0.4 V required for unassisted
photoelectrolysis (Figure 4, Table 1). The ∼0.05 V shift is
within the experimental error and indicates that the metal-ion
treatment described here cannot be used to affect band edge
shift at the GaInP2 electrodes to achieve the “band edge overlap”
condition required for unassisted photoelectrolysis (Figure 1b).
The nature and amount of the catalyst at a semiconductor/
electrolyte interface is very important in determining the activity
of the electrocatalyst. The coverage should be high enough to
affect a high rate of hydrogen production and yet should not be
so high as to impede light absorption by the semiconductor.
Although the exact nature and the amount of the transition-
metal species adsorbed on the p-GaInP2 surface is presently
unknown, a previous study on a similar substrate (GaAs)
suggests that the metal ions may be deposited as complexes, in
submonolayer to a few monolayers amounts, via a redox reaction
with the surface semiconductor atoms.38 In any case, the lack
of flatband shifts in the dark upon metal-ion treatment suggests
that the metal complexes are present as overall neutral species
on the GaInP2 surface. If precious metals are to be used as
catalysts for photoelectrochemical hydrogen evolution, it is
important that the absolute minimum amount required for the
highest catalysis be determined. Although the amount of catalyst
deposited on the semiconductor surface has not been controlled
or optimized in this study, it is clear that significant catalysis
can be obtained even with near monolayer amounts of transition-
metal ions.
Acknowledgment. The authors wish to thank Sarah Kurtz,
Jerry Olsen, and Dan Friedman for the p-GaInP2 samples and
Andreas Maier, Jao van de Laagemat, and Holm Wiesner for
the Mott-Schottky data collection and analysis programs. The
work was supported by the Hydrogen Program of the U.S.
Department of Energy.
References and Notes
(1) Kocha, S. S.; Turner, J. A.; Nozik, A. J. J. Electroanal. Chem.
1994, 367, 27.
(2) Kocha, S. S.; Turner, J. A. J. Electrochem. Soc. 1995, 142, 2625.
(3) Kocha, S. S.; Turner, J. A. Electrochim. Acta 1996, 41, 1295.
(4) Khaselev, O.; Turner, J. A. J. Electrochem. Soc. 1998, 145, 3335.
(5) Allongue, P.; Blonkowski, S. J. Electroanal. Chem. 1991, 316, 57.
(6) Allongue, P.; Blonkowski, S.; Lincot, D. J. Electroanal. Chem.
1991, 300, 261.
(7) Yoneyama, H.; Sakamoto, H.; Tamura, H. Electrochim. Acta 1975,
20, 341.
(8) Nozik, A. J.; Memming, R. J. Phys. Chem. 1996, 100, 13061.
(9) Allongue, P.; Cachet, H.; Horowitz, G. J. Electrochem. Soc. 1983,
130, 2352.
(10) Allongue, P.; Cachet, H. J. Electrochem. Soc. 1984, 131, 2861.
(11) Watts, D. K.; Koval, C. A. J. Phys. Chem. 1996, 100, 5509.
(12) Uhlendorf, I.; Reineke-Koch, R.; Memming, R. Ber. Bunsen-Ges.
Phys. Chem. 1995, 99, 1082.
(13) Allongue, P.; Blonkowski, S.; Souteyrand, E. Electrochim. Acta
1992, 37, 781.
(14) Kelly, J. J.; Memming, R. J. Electrochem. Soc. 1982, 129, 730.
(15) Koval, C. A.; Segar, P. R. J. Am. Chem. Soc. 1989, 111, 2004.
(16) Kobayashi, H.; Mizuno, F.; Nakato, Y. Jpn. J. Appl. Phys. 1994,
33, 6065.
(17) Kurtz, S. R.; Olson, J. M.; Kibbler, A. E.; Bertness, K. A. J. Cryst.
Growth 1992, 124, 463.
(18) A preliminary analysis of impedance data using the series-parallel
circuit between 40 Hz and 60 kHz indicated that the parallel resistor Rp
was always greater than 30 kΩ. Moreover, Mott-Shottky (1/C2 vs V) plots
made with capacitances calculated using both the series-parallel and the
series circuits gave similar trends in migration of flatband potentials with
illumination. The Mott-Shottky plots were made using capacitance values
in the region where the current was light limited and was invariant with
applied potential. In this potential range, the Mott-Schottky plots were
also found to be linear. Deviations from linearity occurred in potential
regions (closer to Vfb) where the current changed with potential, and
capacitance data were not evaluated in this region.
(19) Uhlendorf, I.; Reineke-Koch, R.; Memming, R. J. Phys. Chem.
1996, 100, 4930.
Conclusions
This work shows that the p-GaInP2 band edges migrate
negatively under illumination in solutions with pH ranging from
1 to 14.5. The displacement is not due to a change in the pH of
the semiconductor microenvironment but is caused by the
accumulation of photogenerated electrons at the interface due
to poor interfacial charge-transfer kinetics. Differentiating
between the two is important for developing appropriate surface
modification strategies to realize an effective water-photoelec-
trolysis system. From the experiments described above, it
becomes imperative that any surface modification scheme used
to achieve the “band edge overlap condition” must not com-
promise the charge-transfer kinetics at the interface. Our
observations indicate a less than optimal rate of charge transfer
across the p-GaInP2/water interface and underscore the necessity
of catalysis at the p-GaInP2/water interface.
(20) Fantini, M. C. A.; Shen, W. M.; Gambino, J. P.; Tomkeiwicz, M.
J. Appl. Phys. 1989, 65.
(21) Mishra, K. K.; Osseo-Asare, K. J. Electrochem. Soc. 1992, 139,
749.
Treatment with transition-metal ions is found to effectively
electrocatalyze transfer of photogenerated electrons from the
semiconductor to the liquid and to arrest the undesired band
edge migration under illumination. p-GaInP2 electrodes treated
with RuIII, RhIII, CoIII, and OsIII ions exhibit improved fill factors
(22) Erne´, B. H.; Maroun, F.; Ozanam, F.; Chazalviel, J. N. Electrochem.
Solid State Lett. 1999, 2, 231.
(23) Bockris, J. O. M.; Szklarczyk, M.; Contractor, A. Q.; Khan, S. U.
M. Int. J. Hydrogen Energy 1984, 9, 741.
(24) Aharon-Shalom, E.; Heller, A. J. Electrochem. Soc. 1982, 129,
2865.

6598 J. Phys. Chem. B, Vol. 104, No. 28, 2000
Bansal and Turner
(25) Heller, A.; Vadimsky, R. G. Phys. ReV. Lett. 1981, 46, 1153.
(26) Heller, A.; Aharon-Shalom, E.; Bonner, W. A.; Miller, B. J. Am.
Chem. Soc. 1982, 104, 6942.
(36) Menezes, S.; Heller, A.; Miller, B. A. J. Electrochem. Soc. 1980,
127, 1268.
(37) Tufts, B. J.; Abrahams, I. L.; Casagrande, L. G.; Lewis, N. S. J.
Phys. Chem. 1989, 93, 3260.
(27) Heller, A. Science 1984, 223, 1141.
(28) Mandal, K. C.; Basu, S.; Bose, D. N. Solar Cells 1986, 18, 25.
(29) Bose, D. N.; Hedge, M. S.; Basu, S.; Mandal, K. C. Semicond.
Sci. Technol. 1989, 4, 866.
(30) Bose, D. N.; Basu, S.; Mandal, K. C. Thin Solid Films 1988, 164,
13.
(31) Mandal, K. C.; Basu, S.; Bose, D. N. J. Solid State Chem. 1987,
71, 559.
(32) Butler, M. A.; Ginley, D. S. Appl. Phys. Lett. 1983, 42, 582.
(33) Nakato, Y.; Abe, K.; Tsubomura, H. Ber. Bunsen-Ges. Phys. Chem.
1976, 80, 1002.
(34) Nakato, Y.; Tonomura, S.; Tsubomura, H. Ber. Bunsen-Ges. Phys.
Chem. 1976, 80, 1289.
(35) Parkinson, B. A.; Heller, A.; Miller, B. J. Electrochem. Soc. 1979,
126, 954.
(38) Tufts, B. J.; Abrahams, I. L.; Caley, C. E.; Lunt, S. R.; Miskelly,
G.; Sailor, M. J.; Santangelo, P. G.; Lewis, N. S.; Hedman, B. M.; Roe, A.
L.; Hodgson, K. O. J. Am. Chem. Soc. 1990, 112, 5123.
(39) Bansal, A.; Tan, M. X.; Tufts, B. J.; Lewis, N. S. J. Phys. Chem.
1993, 97, 7309.
(40) Frank, A. J.; Honda, K. J. Photochem. 1985, 29, 195.
(41) Khan, S. U. M.; Majumder, S. A. Int. J. Hydrogen Energy 1989,
14, 653.
(42) Khader, M. M.; Hannout, M. M.; El-Dessouki, M. S. Int. J.
Hydrogen Energy 1996, 21, 547.
(43) Gao, X.; Kocha, S.; Frank, A. J.; Turner, J. A. Int. J. Hydrogen
Energy 1999, 24, 319.
(44) Khaselev, O.; Turner, J. A. Science 1998, 280, 425.