The interface between water and the In-rich InP(001) surface is studied by density functional theory with water coverage ranging from single molecules to multiple overlayers. Single molecules attach preferably to three-fold coordinated surface In atoms. Water dissociation is energetically favorable but hindered by an energy barrier that decreases with increasing water coverage. There is an attractive interaction between InP adsorbed water molecules that leads to the formation of molecular clusters and complete water films for water-rich preparation conditions. Water films on InP are stabilized by anchoring to surface-bonded hydroxyl groups. With increasing thickness, the water films resemble the structural properties of ice Ih. The oxygen and hydrogen evolution reactions on InP are characterized by overpotentials of the order of 1.7-1.8 and 0.2-0.3 eV, respectively. While the calculated bulk positions of the InP band edges are outside the range of the redox potentials for oxygen and hydrogen evolution within local DFT, the situation is different at the actual interface: Here, the interface dipole lifts the InP valence band maximum above the redox potential for oxygen evolution and favors hydrogen evolution.
The interface between water and the In-rich InP(001) surface is studied by density functional theory with water coverage ranging from single molecules to multiple overlayers. Single molecules attach preferably to three-fold coordinated surface In atoms. Water dissociation is energetically favorable but hindered by an energy barrier that decreases with increasing water coverage. There is an attractive interaction between InP adsorbed water molecules that leads to the formation of molecular clusters and complete water films for water-rich preparation conditions. Water films on InP are stabilized by anchoring to surface-bonded hydroxyl groups. With increasing thickness, the water films resemble the structural properties of ice Ih. The oxygen and hydrogen evolution reactions on InP are characterized by overpotentials of the order of 1.7-1.8 and 0.2-0.3 eV, respectively. While the calculated bulk positions of the InP band edges are outside the range of the redox potentials for oxygen and hydrogen evolution within local DFT, the situation is different at the actual interface: Here, the interface dipole lifts the InP valence band maximum above the redox potential for oxygen evolution and favors hydrogen evolution.
The applications of III–V compound
semiconductors range
from the area of nanotechnology, where they are used for quantum wells,[1] quantum dots,[2] and
nanowires[3] to optoelectronics, multilayered
high-efficiency solar cells,[4] and artificial
leaves for solar-to-hydrogen conversion.[5] Group-III phosphides show promising results for solar powered water
splitting, due to high solar to energy conversion. In particular,
the combination of InP with other materials is efficiently used in
this context.[6−9] The rapid InP corrosion under operating conditions is one of the
problems that still needs to be solved, however.While the clean
InP surface reconstructions are well understood,[10−13] relatively little is known about
the microscopic structure and electronic
properties of InP in contact with oxygen and water.[14−17] May and co-workers interpret
their optical spectroscopy data[18] in terms
of a dissociative adsorption of water on InP(001), resulting in the
formation of In–O–P bonds. No hydroxyl signatures were
found by photoelectron spectroscopy, indicating that the dissociative
adsorption of water releases both hydrogen atoms of the water molecule.
On the other hand, a computational study by Wood et al.[19] finds the dissociative adsorption of water favorable
for GaP surfaces but not for InP. For both substrates, a dissociation
barrier of about 0.8 eV was calculated. Photoelectron spectroscopy
measurements also detected a dissociative adsorption of water on Ga-rich
GaAs(001) surfaces, leading to surface adsorbed OH and H species.[20] Further deprotonation results in the formation
of Ga–O, Ga–OH, and As–H bonds. High water pressures
cause water physisorption directly on the GaAs surface or to be anchored
to Ga–OH bonds. In case of GaP, it was noted that the water
surface interaction depends on the surface reconstruction. Ga-rich
surfaces show a higher reactivity than P-rich surfaces. Photoelectron
and optical spectroscopy[16] suggest the
presence of both water and hydroxyl groups on Ga-rich GaP surfaces.
This is backed by a theoretical study on the reaction pathways.[21]The present study aims at a thorough understanding
of the interaction
between water and the In-rich InP(001) surface. The adsorption and
surface reactions of single water molecules, water monolayers, and
water multilayers are studied by density functional theory (DFT) calculations.
In addition, we explore pathways for oxygen and hydrogen evolution
reactions (OER and HER).
Methodology
The present DFT calculations
are performed with the Vienna Ab Initio
Simulation Package (VASP).[22] The electron
exchange and correlation interaction are modeled within the generalized
gradient approximation (GGA) using the PBE functional.[23] The electron–ion interaction is described
by the projector-augmented wave (PAW) scheme.[24,25] An energy cutoff of 500 eV limits the plane-wave basis used to expand
the wave functions. The calculations are performed for periodic supercells
that contain 12 atomic layers, the adsorbed water, and a vacuum region
of about 15 Å. A dipole correction scheme is used to minimize
spurious interactions across the vacuum region. The slab bottom In
layer is passivated with pseudohydrogen with Z =
1.25. Surface periodicities of 2 × 4 (unit cell of the In-rich
(001) surface) and 2 × 12 (for better commensurability with thick
water overlayers) were used for the water–InP interface calculations.
Adsorption energy convergence tests were performed using 4 ×
4 surface unit cells. The corresponding surface Brillouin zones were
sampled with Γ-centered 6 × 3 × 1, 12 × 2 ×
1, and 3 × 3 × 1 k-point meshes, respectively.All
interfaces are structurally relaxed until the forces acting
on the atoms are lower than 0.02 eV/Å. Potential energy surface
(PES) calculations were performed to determine the most favorable
adsorption sites for low coverages. The PES were sampled at 50 equidistant
mesh points. At each mesh point, two and nine different starting configurations
are probed for hydroxyl group and water molecule adsorption, respectively.
In the calculations, the oxygen ion is laterally fixed. Its vertical
distance to the surface as well as all other degrees of freedom are
not constrained.The starting geometries for water overlayers
that equal or exceed
monolayer coverage were derived from both the ice I structure as well as randomized water molecule positions.
To determine the ground state of the water overlayers, the total-energy
minimization is complemented by simulated annealing. To that end we
performed ab initio molecular dynamics (MD) calculations in the NVT
(canonical) ensemble using a temperature of 400 K to reproduce ambient
liquid water. Temperatures were maintained with a Nosé Hoover
Thermostat.[26]The adsorption energy
is calculated aswhere Etotal and Eclean are the total energy of the adsorbed system
and the clean surface, respectively. In case of water adsorption, EA corresponds to the total energy of gas-phase
water molecules. In case of hydroxyl group and oxygen adsorption,
we define EA asandwhile hydrogen adsorption is referenced to
gas-phase H2. These definitions allow for direct comparison
between dissociative and molecular adsorption energies. Adsorption
energies for specific low coverage structures were corrected for the
zero-point energies (ZPE).In order to compare energetically
structures that correspond to
different water coverages, one needs to consider the grand canonical
potentialwhere F(n) is the total free energy of the surface with n water molecules adsorbed. It is approximated here by the
DFT total
energy, assuming similar entropy contributions for different adsorption
configurations. The chemical potential μH accounts for the availability of water. Water rich-conditions are
defined by the μH value for bulk water,
which we approximate by calculations for the ice I phase. In the following, ΔμH refers to the difference of the water chemical
from that of ice I.
Its dependence on temperature and pressure can be calculated in the
approximation of a polyatomic ideal gas[27] aswhere kB is the
Boltzmann constant, λ the de Broglie thermal wavelength of the
water moleculem is the mass of the water
molecule, andare its rotational
and vibrational partition
functions, respectively. We used experimental values for the moments
of inertia I and vibrational
frequencies ωα of a water molecule.[28] The geometrical parameter of a water molecule
H2O is given by σ = 2 (equal-sided triangle).Surface chemical reactions, in particular oxygen and hydrogen evolution
reactions, are characterized in the present work by the Gibbs free
energy G(p,T) differences
of products and educts. For the calculation of the Gibbs free energy
differences, we follow Nørskov et al.[29] and approximatewhere ΔG = −e·U and ΔGpH = −kBT ln(10)·pH, account for the reaction
energetics dependence on the potential U and the
pH value,[30] respectively. For specific
low-coverage structures, we accounted additionally for entropic and
ZPE corrections. The reaction energy barriers are calculated in the
nudged elastic band (NEB) approximation.[31]
Results and Discussion
In-rich InP(001) surfaces form a
(2 × 4) surface reconstruction
featuring a single In–P heterodimer in the uppermost atomic
layer for a wide range of preparation conditions.[32] This so-called mixed-dimer structure, see Figure , is the starting point for
the present calculations.
Figure 1
Calculated potential energy surfaces for single
water molecules
(a), an OH group (b), and a H atom (c) on the InP (001)(2 × 4)
mixed-dimer surface. Blue and orange spheres indicate In and P, respectively.
Calculated potential energy surfaces for single
water molecules
(a), an OH group (b), and a H atom (c) on the InP (001)(2 × 4)
mixed-dimer surface. Blue and orange spheres indicate In and P, respectively.Previous experimental and theoretical studies[18,19] suggested the sites A–D indicated in Figure a to be the most favorable water adsorption
positions. Potential energy surface calculations provide a systematic
approach to the most favored adsorption geometry for single molecules.
The PES calculated here, see Figure a, indicates that 3-fold coordinated surface In atoms
provide the most favorable bonding sites for water, corresponding
to the sites A, B, and C identified already in ref (19). In combination with the
PES calculated for the adsorption of OH and H (Figure b,c, respectively), the most relevant configurations
for the low-coverage stage of the water–InP interface can be
identified. The corresponding adsorption energies are listed in Table . From convergence
tests using 2 × 4 and 4 × 4 surface unit cells, we conclude
on an interaction energy of about 5 meV affecting the single molecule
adsorption energies. The influence of the electronic entropy at room
temperature is about 2 orders of magnitude smaller. Therefore, it
is neglected in the following. The ZPE is more important. As seen
in Table , it typically
affects the adsorption energies by about 0.1 eV, in some cases even
by 0.22 eV per adsorbate. However, the ZPE does not alter the energy
trends derived from the DFT total energies.
Table 1
Adsorption
Energies (in eV) and Bond
Configurations for H2O, O, and OH at Specific Sites of
the InP Mixed-Dimer Surface, cf. Figure
site
configuration
Eadsa
Eadsa,b
B
H2O–In
–0.408
–0.324
A
H2O–In
–0.384
–0.302
A′
In–O–P
0.387
0.160
D
OH–P
0.665
0.591
A
In–OH
0.646
0.540
E
In–OH–In
–0.073
–0.158
A
In–OH
and P–Hc
–0.274
–0.291
E
In–OH–In and P–Hc
–0.508
–0.524
B and B′
H2O–In
–0.989
–0.791
D and D′
In–O–P
–0.175
–0.621
E and E′
In–OH–In
–0.200
–0.369
Positive energies characterize configurations
unfavorable with respect to an intact adsorbed water molecule.
Includes ZPE corrections.
P–H bond forms on In–P
mixed dimer.
Positive energies characterize configurations
unfavorable with respect to an intact adsorbed water molecule.Includes ZPE corrections.P–H bond forms on In–P
mixed dimer.The most favorable
water adsorption configuration at the site B
along with the adsorption-induced charge density redistribution is
shown in Figure .
The water molecule donates electrons of its 1b1 orbital
into the empty In dangling bond of p character.[10] This results in a weak O–In bond with
length 2.5 Å formed at the surface. The substrate geometry is
modified only slightly upon adsorption. No water-induced electronic
states appear in the energy region of the InP band gap.
Figure 2
Structural
configuration and water-induced charge redistribution
for water adsorbed on InP (001) on site B. Red and blue isosurfaces
indicate regions of charge accumulation and depletion, respectively.
Structural
configuration and water-induced charge redistribution
for water adsorbed on InP (001) on site B. Red and blue isosurfaces
indicate regions of charge accumulation and depletion, respectively.Experimentally, water adsorption was reported to
be dissociative,[18] with oxygen bound near
the phosphorus atom of
the mixed-dimer (sites D and D′ in Figure ) while previous theoretical results[19] found molecular adsorption to be favored over
dissociation. Here it must be said, however, that in ref (19) the possibility of water
dissociation was studied only for adsorption on the mixed-dimer. The
present calculations support dissociative adsorption, resulting in
a configuration different from the one proposed in ref (18), however. Rather, the
hydroxyl group is attached to an In–In bond (site E in Figure ) and the hydrogen
adsorbs at the mixed-dimer P atom, as shown in Figure . This configuration is 0.1 eV (0.2 eV with
ZPE corrections) lower in energy than the most favorable molecular
adsorption at site B. The dissociative adsorption leads to a marked
distortion of the InP(001) surface geometry, induced in particular
by the insertion of the hydroxyl group oxygen into the second layer
In–In bond. As seen by the charge-density redistribution in Figure , strong In–O–In
and P–H bonds form. Nevertheless, similarly to the intact molecule
adsorption, no water-induced electronic states appear in the energy
region of the InP band gap.
Figure 3
Structural configuration and water-induced charge
redistribution
for water dissociatively adsorbed on InP (001), with OH on site E
and H on the top P. Red and blue isosurfaces indicate regions of charge
accumulation and depletion, respectively.
Structural configuration and water-induced charge
redistribution
for water dissociatively adsorbed on InP (001), with OH on site E
and H on the top P. Red and blue isosurfaces indicate regions of charge
accumulation and depletion, respectively.The energy barrier for water dissociation is calculated here using
the nudged elastic band method. Molecular adsorption at site A is
assumed to be the starting configuration. Six images at equidistant
points between the starting configuration and the dissociatively adsorbed
OH group at site E and hydrogen at the topmost P atom are considered.
The energy path is shown in Figure . The transition state corresponds to the hydroxyl
group transfer from position A to position E. We obtain an energy
barrier of 0.87 eV. This of the same order of magnitude as determined
earlier for dissociation on the mixed dimer.[19]
Figure 4
Minimum
energy path for dissociative adsorption of single water
molecules on InP(001). The starting configuration is the water molecule
adsorbed at site A and the final configuration corresponds to the
one shown in Figure .
Minimum
energy path for dissociative adsorption of single water
molecules on InP(001). The starting configuration is the water molecule
adsorbed at site A and the final configuration corresponds to the
one shown in Figure .To study a slightly increased
water coverage, we consider the adsorption
of two molecules per surface unit cell. The minimum energy configuration
for molecular adsorption corresponds to the molecules bonded at sites
B and B′, cf. Figure a. Compared to single molecules adsorbed at these sites, an
energy reduction of 0.09 eV (0.07 eV with ZPE corrections) per molecule
is found. This demonstrates an attractive interaction and suggests
the study of higher water coverages. In the following, we increase
the water coverage to up to 56 molecules per (2 × 4) surface
unit cell. Given that the influence of the zero-point corrections
does not alter the energy trends derived for low-coverage water adsorption,
ZPE corrections are neglected for high coverages.The phase
diagram in Figure compares the energetics of various coverages of molecularly
adsorbed water. It can be seen that the one-monolayer configuration
(Figure a) is more
favored than small molecular clusters comprising up to five molecules
(see inset in Figure ). Even more favorable, however, is the water bilayer (Figure b). According to the present
calculations, for intermediate values of the water chemical potential
(ΔμH ≈
0.5 eV, corresponding roughly to standard pressure at room temperature)
there will be a direct transition from the clean InP surface to the
water bilayer covered surface. For more water-rich conditions, even
thicker water layers will form. For computational reasons, the coverage
in the present study is limited to three monolayers (Figure c).
Figure 5
Phase diagram of the
water-adsorbed InP(001) surface in dependence
on the water chemical potential ΔμH. Here adsorption configurations comprising 1,
2, and 5 molecules (m) are compared with the formation of 1, 2, and
3 layers (l) and the clean surface (zero). The pressure and temperature
dependence of ΔμH refer to eq .
Figure 6
Energetically favorable In–P (001)(2 × 4)
structures
with one (a), two (b), and three (c) adsorbed water layers.
Phase diagram of the
water-adsorbed InP(001) surface in dependence
on the water chemical potential ΔμH. Here adsorption configurations comprising 1,
2, and 5 molecules (m) are compared with the formation of 1, 2, and
3 layers (l) and the clean surface (zero). The pressure and temperature
dependence of ΔμH refer to eq .Energetically favorable In–P (001)(2 × 4)
structures
with one (a), two (b), and three (c) adsorbed water layers.Structurally, the one, two and three water layer
configurations
bear some similarity to ice I, see Figure . Because of the
incommensurability of the ice I basal plane with the InP (2 × 4) surface unit cell, as
well as due to the InP surface morphology, however, some disorder
arises. In particular, irregular pentagons and hexagons form. The
mono- and bilayer structures are characterized by water O bonded to
three-fold coordinated In atoms, resembling the single molecule bonding
to sites A, B, and C in Figure . The number of these bonds reduces upon formation of the
water bilayer. They are completely quenched for the three-layer structure,
which is strongly reminiscent of ice I. The InP surface phase diagram in dependence on
temperature and water partial pressure is shown in Figure . It should be noted that at
the transition between the clean InP surface and the water covered
substrate some variety of adsorption configurations can be expected:
As shown by the inset in Figure , in a narrow range of the water chemical potential
many competing structures occur, that is, a strong influence of kinetic
effects on the actual interface configuration will lead to a variety
of interface structures.
Figure 7
Phases diagram of the InP (001) mixed dimer
surface as a function
of the temperature and pressure. The values of the water chemical
potential μH for which the phase transitions
occur are marked by the solid lines.
Phases diagram of the InP (001) mixed dimer
surface as a function
of the temperature and pressure. The values of the water chemical
potential μH for which the phase transitions
occur are marked by the solid lines.A variety of stable dissociated water adsorption configurations
can be identified in case of the water-layer covered InP(001) surfaces.
The most favorable one for the monolayer covered surface is shown
in Figure . It corresponds
to hydroxyl attached to second layer In–In bonds (bonding site
E) and hydrogen attached to the mixed-dimer P atom. This configuration
is stable also upon hydrogen transfer from the mixed-dimer to a H2 reservoir. This contrasts with similar adsorption configurations
without the water monolayer, which are not stable upon H desorption.
The stability of the dissociated configurations in the presence of
the hydrate shell can be explained by the fact that the surface bonded
OH group acts as an anchor for the water layer above. Similar findings
are reported for the GaAs(100) surfaces.[20] The hydration shell also affects the barrier for water dissociation,
as shown the NEB calculations in Figure . Here the minimum energy path for a molecule
from the water layer that dissociates and adsorbs at the surface is
shown. The molecule, originally in a local minimum at the interface
between the water layer and the InP surface, needs to overcome a barrier
of ∼0.3 eV for dissociation. However, the entire energy path
during the dissociation is below the energy of the starting configuration.
The hydrate shell also affects the energy barriers involved in the
formation of surface adsorbed hydroxyl groups. For example, we calculate
a nearly vanishing energy barrier of 7 × 10–3 eV for the H2O–In + In–O–In →
OH–In + In–OH–In process. Oxygen bonding close
to the topmost P atom, as concluded from experimental data,[18] however, remains unfavorable also for the water
layer covered surface.
Figure 8
Most stable water dissociation configuration identified
here for
the monolayer covered InP (001) surface. The surface adsorbed OH group
is highlighted.
Figure 9
Minimum energy path for dissociative adsorption
of single water
molecules embedded at the water–InP interface.
Most stable water dissociation configuration identified
here for
the monolayer covered InP (001) surface. The surface adsorbed OH group
is highlighted.Minimum energy path for dissociative adsorption
of single water
molecules embedded at the water–InP interface.Finally, we explore the energetics of water splitting. The
corresponding
redox reaction consists of the oxidation 2H2O → O2 + 4H+ + 4e– and the reduction 4H+ + 4e– →2H2, known as oxygen evolution reaction (OER) and hydrogen evolution
reaction (HER), respectively. They are studied here by calculating
the Gibbs free energy G(p,T) changes according to eq . We assume the reaction H2 ⇌ 2(H+ + e–) to be in equilibrium, that is, the reference potential is given
by the “numerical” hydrogen electrode, corresponding
to a vanishing Gibbs free energy difference ΔG = 0. Consequently, the energy of proton–electron pairs in
the reaction pathways is given by . The overpotential η is
defined as
the difference between the largest energy difference ΔG and the corresponding potential at zero pH. The water to oxygen
oxidation potential for the OER is 1.23 eV[33] while that for HER vanishes by definition.The are many possible
pathways for the OER. Here, we restrict ourselves
to the associative reaction mechanism for water splitting.[29,34] This mechanism is assumed to have a peroxide intermediate OOH adsorbed
on the surface. An alternative mechanism would be for two water molecules
to separately dissociate into two surface-bonded O atoms. The activation
barrier associated for the recombination of two adjacent oxygen atoms
into O2 is rather large.[35] Here
we calculate an energy barrier of ∼4 eV for the recombination
process within the NEB method, rendering this process less likely.
The steps of the associative chemical reactions readwhere X*
denotes a species X adsorbed at the
surface. The present calculations of the respective reaction energetics
are based on the most favored adsorption configurations determined
from the potential energy surfaces discussed above. On the basis of
the respective energy differences, the data in Figure are obtained. Here, results neglecting
and including ZPE corrections are compared. The reaction A is exergonic
for neutral conditions, and there is only a very small energy increase
observed for reaction B. With ZPE corrections both reactions, A and
B, are exergonic. This could possibly explain the experimental findings
in ref (18), where
no persistent hydroxyl groups, but the release of gaseous hydrogen,
were observed. There is a considerable energy required for reaction
C, necessitating an external bias voltage. Applying an external potential
of 3 eV results in the orange solid curve shown in Figure . Generally, we find that
the inclusion of ZPE corrections tends to reduce the OER energy differences.
Fortuitously, however, the change in the overpotential due to ZPE
corrections is small. We calculate an overpotential η of about
1.7 eV. The reactions depend on the acidity of the electrolyte. Decreasing
pH, that is, increasing acidity favors H-adsorbed surfaces and requires
larger potentials for hydrogen desorption. Reaction A becomes endergonic
for pH lower than 5.5 (2.8 with ZPE corrections). At zero pH, the
energy difference amounts to 0.33 eV, and around half of that with
ZPE corrections, 0.17 eV. Reaction B becomes exergonic for pH larger
than 7.8 (5.3 with ZPE corrections), allowing a water molecule to
release surface adsorbed oxygen without an external potential. The
reaction energies for extreme acidic and alkaline electrolytes are
shown by dashed lines in Figure . It should be kept in mind, however, that the reactions
itself will modify the pH value and that for alkaline conditions different
OER processes might occur.[30,36]
Figure 10
Comparison of the Gibbs
free energy for the OER for different potentials
and pH values. Bottom and top figures with and without ZPE energy
term, respectively. Numbers represent the energy barriers for 0 pH
and no potential.
Comparison of the Gibbs
free energy for the OER for different potentials
and pH values. Bottom and top figures with and without ZPE energy
term, respectively. Numbers represent the energy barriers for 0 pH
and no potential.The HER process involves
the reduction of the protons to molecular
hydrogen and is modeled here aswhere we employ the
mixed dimer as most favored
adsorption site for both atomic and molecular hydrogen. The corresponding
Gibbs free energy differences are shown in Figure . A potential of 0.68 and 0.6 eV is required
for reaction E in neutral electrolytes with and without ZPE corrections,
respectively, as shown by the black solid curve. Increasing the pH
increases this value, as less hydrogen ions are available for surface
adsorption. In contrast, the energy required for both reactions E
and F is lowered for acidic conditions. At pH levels below 3.6 reaction
F is exergonic, that is, occurs spontaneously. At extreme acidic conditions,
corresponding to zero pH, the energy difference for reaction E is
0.19 eV (0.26 eV with ZPE corrections), which also defines the overpotential,
see dashed line in Figure . The ZPE corrections increase the energy required for the
reaction E, in contrast to the reduction of the OER energies and in
contrast to reaction F. Even if ZPE corrections slightly (by less
than 0.1 eV) increase the HER overpotential, it is still about an
order of magnitude lower than that for the OER. In addition to the
favorable reaction energetics found here, the possibility of long-range
surface hydrogen transport at the InP(001)–water interface
pointed out in ref (37) may enhance hydrogen evolution on the InP surface.
Figure 11
Comparison of the Gibbs
free energy for the HER for different potentials
and pH values. Bottom and top figures with and without ZPE energy
term, respectively. Numbers represent the energy barriers for 0 pH
and no potential.
Comparison of the Gibbs
free energy for the HER for different potentials
and pH values. Bottom and top figures with and without ZPE energy
term, respectively. Numbers represent the energy barriers for 0 pH
and no potential.However, the OER and
HER energetics discussed so far is limited
to the thermodynamics and does not consider the kinetics of the underlying
processes. The latter are determined by the reaction barriers. In
the reaction A, for example, the dissociation of a water molecule
into a surface adsorbed hydroxyl group and a desorbing proton occurs.
For the dissociation of surface adsorbed water into surface adsorbed
OH and H on clean InP an energy barrier of about 0.8 eV needs to be
overcome, see Figure . In addition, there could be an additional kinetic barrier for hydrogen
desorption. Using the NEB method and assuming desorption from the
adsorbed water molecule configuration, we calculate that an energy
of about 2.7 eV is required for hydrogen desorption, see Figure . This energy matches
in fact the difference in the chemical potentials of molecular and
atomic hydrogen complemented by the Gibbs free energy difference of
reaction A. Thus, as one might expect, the water dissociation rather
than the hydrogen desorption presents an additional barrier hindering
the OER.
Figure 12
Minimum energy paths of the desorption of a hydrogen on the surface.
The starting point is the OH adsorbed at site E and H on the top P,
corresponding to the final configuration of the water dissociation
shown in Figure .
Minimum energy paths of the desorption of a hydrogen on the surface.
The starting point is the OH adsorbed at site E and H on the top P,
corresponding to the final configuration of the water dissociation
shown in Figure .Such barriers, however, can be expected to be modified
by the presence
of additional water molecules forming a hydration shell. In Figure , we show the OER
energetics assuming the reactions to take place in the presence of
one monolayer water adsorbed. The energies shown in Figure correspond to the average
of the six most favorable water overlayer configurations. Because
the large number of molecules involved, ZPE corrections were neglected.
It can be seen that reactions B and D are the ones most affected by
the hydration shell. They become favorable. Reaction C, on the other
hand, requires slightly more energy, resulting in an increase of the
overpotential to η = 1.83 eV. The HER energetics in the presence
of a water monolayer is shown in Figure . Reaction E is hindered by the additional
water layer. This results in an increase in the overpotential to η
= 0.31 eV. Thus, the overpotentials for both OER and HER are slightly
increased by the presence of additional water. We cannot exclude,
however, that reaction mechanisms different from the ones consider
here may change this picture.
Figure 13
Comparison of the Gibbs free energy for
the OER in the presence
of one water monolayer for different potentials and pH values. Numbers
represent the energy barriers for 0 pH and no potential.
Figure 14
Comparison of the Gibbs free energy for the HER in the presence
of one water monolayer for different potentials and pH values. Numbers
represent the energy barriers for 0 pH and no potential.
Comparison of the Gibbs free energy for
the OER in the presence
of one water monolayer for different potentials and pH values. Numbers
represent the energy barriers for 0 pH and no potential.Comparison of the Gibbs free energy for the HER in the presence
of one water monolayer for different potentials and pH values. Numbers
represent the energy barriers for 0 pH and no potential.The absolute energy positions of valence and conduction bands
are
crucial for water splitting: The conduction band minimum should be
above the redox potential for hydrogen evolution, and the valence
band maximum needs to be below the redox potential for oxygen evolution. Figure compares the band
alignments of InP with the water oxidation and hydrogen reduction
potentials, 1.23 and 0.00 eV, respectively. On the physical scale,
they correspond to −5.67 and −4.44 eV. To locate the
InP band edges we calculate the work function ΔΦ, that is, the energy necessary to transfer an electron from the
valence band to the vacuum level, and assume a band gap of 1.35 eV
for the bulk. Our calculations for InP bulk result in ΔΦ = 5.79. This leads to OER and HER potentials that are both inside
the band gap, albeit close to the edges, in agreement with the data
reported in ref (38). This would suggest InP as a candidate for unbiased water splitting.
However, the electronic structure obtained on the DFT-GGA level of
energy suffers from an inaccurate description of the electron exchange
and correlation effects. In order assess the influence of the GGA
on the energy alignment, hybrid DFT calculations using the HSE functional[39] were performed. They increase the work function
to ΔΦ = 6.01, which seems to indicate
that InP is suitable for OER rather than HER. However, this is still
not the complete picture. The bulk calculations do not necessarily
describe the electron energies at the surface correctly. They may
be affected by an electric dipole layer arising from surface relaxation
and reconstruction. In addition, surface states and adsorbates can
be expected to modify the work function. In case of Ga-rich GaP surfaces,
a work function reduction upon water adsorption was assumed as a result
of a dipole layer formed by water molecules that bond with hydrogen
in down position.[40] Depending on the surface
polarity, a work function increase or decrease was found upon water
adsorption on lithium niobate.[41] In case
of InP(001) it was found earlier that the work function decreases
gradually with increasing In coverage.[10] Here a work function of ΔΦ = 4.49 and
a band gap of 0.85 eV are calculated for the mixed-dimer InP(001)(2
× 4) surface on the DFT-GGA level of theory. This lifts the valence
band edge above the oxidation potential, see Figure . This result is corroborated by hybrid
DFT. It predicts a work function of ΔΦ = 4.75 and a band gap of 1.46 eV for the mixed-dimer reconstructed
surface. The adsorption of hydrogen and hydroxyl groups, single water
molecules as well as water monolayer formation reduces the work function
further, below the value of the clean InP surface. Thus, the present
calculations suggest the InP surface primarily for hydrogen evolution.
Figure 15
Band
alignment of InP bulk and surface (calculated on the DFT-GGA
level of theory) in relation to the water splitting redox potentials.
Red/black lines depict CBs/VBs.
Band
alignment of InP bulk and surface (calculated on the DFT-GGA
level of theory) in relation to the water splitting redox potentials.
Red/black lines depict CBs/VBs.
Conclusions
In summary, we studied the interaction of water with In-rich InP(001)
surfaces using density functional theory. Zero-point energy corrections
are found to influence molecular adsorption energies by up to 0.2
eV as well as to slightly modify the OER and HER energetics.Single adsorbed water molecules are found to bond preferentially
to three-fold coordinated In surface atoms. Dissociative adsorption
is found to be more favorable than molecular adsorption but hindered
by a sizable energy barrier. Upon dissociation, the hydroxyl group
preferentially bonds to In–In dimers, while hydrogen adsorbs
at the mixed-dimer P atom.Upon increasing the water coverage,
an attractive interaction between
the water molecules governs the interface morphology. There is a small
range of preparation conditions corresponding to intermediate values
of the water chemical potential where a variety of structures are
nearly degenerate, and single molecules, molecular clusters, and completely
water-covered surface patches coexist. For water-rich conditions,
water layers form. This leads to a distinct reduction of the reaction
barrier for water dissociation and may explain experimental findings
stating dissociative adsorption.While the morphology of the
monolayer water film is strongly influence
by the InP surface structure and characterized by numerous surface
In–water oxygen bonds, multilayer structures resemble the molecular
arrangement in ice I. Surface-bonded hydroxyl groups formed upon dissociation of water
act as anchor points for the water overlayer, similar to observations
for other III–V materials.The overpotentials for OER
and HER are calculated here to be of
the order of 1.7–1.8 and 0.2–0.3 eV, respectively. These
values can be lowered upon variation of the electrolyte pH. While
DFT calculations of bulk InP suggest the material for unbiased water
splitting, the picture changes if InP surface calculations are performed.
They lift the valence-band maximum above the oxidation potential.
This finding is corroborated by hybrid DFT. Both the calculated overpotentials
as well as the energy position of the InP band edges thus suggest
InP surfaces for hydrogen evolution. The HER may be additionally assisted
by In–O–In bonds that result from InP surface oxidation
and provide additional hydrogen adsorption sites.
Authors: Tianshuo Zhao; Nuri Oh; Davit Jishkariani; Mingliang Zhang; Han Wang; Na Li; Jennifer D Lee; Chenjie Zeng; Manisha Muduli; Hak-Jong Choi; Dong Su; Christopher B Murray; Cherie R Kagan Journal: J Am Chem Soc Date: 2019-09-16 Impact factor: 15.419
Authors: Matthias M May; Hans-Joachim Lewerenz; David Lackner; Frank Dimroth; Thomas Hannappel Journal: Nat Commun Date: 2015-09-15 Impact factor: 14.919