First-Principles Study of the Band Diagrams and ... - ACS Publications

Subscriber access provided by Fudan University

Article

First-principles study of the band diagrams and Schottky-type barrier heights of aqueous TaN interfaces 3

5

Eriko Watanabe, Hiroshi Ushiyama, and Koichi Yamashita ACS Appl. Mater. Interfaces, Just Accepted Manuscript • DOI: 10.1021/acsami.6b12261 • Publication Date (Web): 02 Mar 2017 Downloaded from http://pubs.acs.org on March 9, 2017

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

ACS Applied Materials & Interfaces is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 30

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Applied Materials & Interfaces

First-principles study of the band diagrams and Schottky-type barrier heights of aqueous Ta3N5 interfaces

Eriko Watanabe*, Hiroshi Ushiyama*, Koichi Yamashita Department of Chemical System Engineering, School of Engineering, The University of Tokyo, Bunkyo-ku, Tokyo 113-8656, Japan E-mail: [email protected], [email protected]

KEYWORDS Tantalum nitride, Photoelectrocatalyst, Schottky barrier, Aqueous interface, Band edge position, Density functional theory

ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ABSTRACT

The photo(electro)chemical production of hydrogen by water splitting is an efficient and sustainable method for the utilization of solar energy. To improve photo(electro)catalytic activity, a Schottky-type barrier is typically useful to separate excited charge carriers in semiconductor electrodes. Here, we focused on studying the band diagrams and the Schottky-type barrier heights of Ta3N5, which is one of the most promising materials as a photoanode for water splitting. The band alignments of the undoped and n-type Ta3N5 with adsorbents in vacuum were examined to determine how impurities and adsorbents affect the band positions and Fermi energies. The band edge positions as well as the density of surface states clearly depended on the density of ON impurities in the bulk and surface regions. Finally, the band diagrams of the n-type Ta3N5/water interfaces were calculated with an improved interfacial model to include the effect of electrode potential with explicit water molecules. We observed partial Fermi level pinning in our calculations at the Ta3N5/water interface, which affects the driving force for charge separation.


Page 2 of 30

Page 3 of 30

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


1. Introduction Photoelectrochemical (PEC) production of hydrogen by water splitting is an efficient and sustainable method for the utilization of solar energy. Water oxidation (R1) is an anodic reaction that has a non-zero overpotential even on highly active electrocatalysts such as RuO2 and IrO2.1,2 2H2O + 4 h+ → O2 + 4H+

(R1)

A thermodynamic requirement for water oxidation is that the position of the valence band (VB) of a semiconductor must be situated below the redox potential of O2/H2O. Compared to photocatalytic water splitting, a PEC reaction has the advantages of separating the reaction sites of hydrogen and oxygen evolution and the capability of controlling the Fermi energy by the electrode potential. Ta3N5 is one of the most promising materials for use as a photoanode for water splitting. It has a bandgap of 2.1 eV, suitable for visible light absorption,3,4 and a band-edge position straddling the redox potential of H+/H2 and O2/H2O.5 To lower the overpotential of the water oxidation reaction, in addition to lowering the activation energy of the reaction itself, it is important to suppress the recombination of excited charge carriers (electrons and holes). A Schottky-type barrier (SB) is typically important for separating excited charge carriers

in

semiconductor

electrodes.

For

example,

the

band

bending

of

a

semiconductor/co-catalyst heterojunction, such as RuO2/ZnO or RuO2/TiO2, has been examined experimentally to show that the existence of an SB at the interface enhances photocatalytic activity.6,7 Moreover, some heterogeneous characteristics at the interface, such as a mixed phase



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

of TiO2 or a mixed domain of Ta3N5, improves the activity due to the interfacial band diagrams accelerating charge separation.8,9 In the case of a semiconductor/electrolyte system, band bending is also known to form, and the thermodynamic factors regarding the magnitude of the band bending can shift the onset potential of water oxidation.10 For the utilization of Ta3N5 as a photoanode, efforts to improve the activity have mainly focused on increasing its photoabsorption properties,11,12 improving its crystallinity,13 modifying its surface and improving its contact with a co-catalyst.14,15 The surface reactions of oxygen adsorption,16 water adsorption,17 and oxygen reduction18 have also been reported. However, less attention has been paid to the semiconductor/electrolyte interface, and only flat-band (FB) potential measured by Mott–Schottky plots has been addressed as a property of the Ta3N5/water interface.5,12,14 The examined FB potential has varied between different studies. One study measured the FB potential to be −0.5 V vs. RHE together with conduction band (CB) and VB positions of −0.5 and 1.6 V vs. RHE.12 Other examinations have reported FB potentials of −0.5 V vs. RHE and 0.0 V vs. RHE.5,14 Interestingly, Nurlaela et al. noted that the modification of only the topmost layer of Ta3N5 changes the FB potential.14 These studies confirm the significance of obtaining the energy diagram of the interface. In our study, band diagrams and the SB height of the Ta3N5/water interface under electrochemical conditions is presented. In particular, the occurrence of band-edge pinning (BEP) or Fermi level pinning (FLP) is examined by first-principles calculations combined with a double reference method.19 Ta3N5 is, in general, an n-type semiconductor with an O-enriched and N-deficient structure.20 From thermodynamic analysis, the formation of sites where an N atom is substituted by an O atom (denoted ON) is favorable under the synthetic conditions,21 and the aqueous stability of Ta3N5 is so poor that Ta2O5 is formed in the potential range between


Page 4 of 30

Page 5 of 30

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


H+/H2 and O2/H2O at pH 0–14.22 In addition, vacancies at N sites (denoted VN) are formed as a result of self-oxidation in photocatalytic processes.23 In our study, electrochemical conditions are assumed to reveal the interfacial properties under equilibrium conditions. Thereby, bulk ON or surface ON is introduced as a model of n-type Ta3N5. In addition, band alignment is easily affected by adsorbents such as O and OH. Therefore, the n-type Ta3N5/water interfacial structure is first examined, and then, a band diagram with the stable interfacial structure is determined.

We note that several methods to study solid/electrolyte interfaces under applied potential have been proposed so far.24 The electrode potential of a given structure can be derived using the work function of the system, as shown in the following equation: =

(1)

where Φ is the work function of the system and ΦSHE is the standard hydrogen electrode (SHE) potential relative to vacuum. One representative method to shift the electrode potential involves adding excess charge with a homogeneous background counter charge.19 The excess charge attracts counter charge at the solid/electrolyte interface similar to a capacitor. The electrode potential can be controlled by changing the amount of excess charge due to dipoles formed at the interface changing the work function of the system. This method can be used in the framework of standard DFT calculations; however, one should take care of undesired interactions between excess charge and the homogeneous background charge. Another method to vary the charge distribution is to introduce neutral atoms with high or low electronegativity that can be easily ionized.25 The advantage of this method is that it avoids the introduction of the artificial homogeneous background charge; however, undesired interactions from the introduced atoms are unavoidable.



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

2. Method DFT calculations were performed using the GPAW26,27 code with the ASE simulation package.28 The GGA/PBE29 functional and PAW method30 were employed for all calculations, and the gpaw-setups-0.9.11271 was used for all elements. The size of the optimized unit cell was a = 3.90 Å, b = 10.32 Å, and c = 10.35 Å, which is in good agreement with previously reported experimental values20 (a = 3.89 Å, b = 10.22 Å, and c = 10.27 Å). We confirmed that the difference between the optimized and experimental lattice constants has little influence on the band diagrams (see Figure S1). To model the surface, the (100) surface was selected because it is the stoichiometric phase and has the lowest surface energy.16 We note that the band positions differ according to the surface orientation. Surface-oriented behavior is also an important subject; however, it is beyond the scope of this study. The size of the slab model was varied depending on the purpose of the study. As the minimum surface model (used in section 3.1), an asymmetric (3 × 1 × 1) supercell consisting of six layers (96 atoms) with at least a 7 Å vacuum region on each side was used, as shown in Figure 1. During the geometry optimization, two bottom layers with 32 atoms were kept fixed. For the examination of the band position relative to vacuum level (section 3.2), an asymmetric (6 × 1 × 1) supercell was used due to the band position showing relatively slow convergence with the size of the slab model (Figure S2). For the examination of the interfacial SB height under a given electrode potential (section 3.3), a symmetric (6.5 × 1 × 1) supercell was employed with explicitly introduced ice-like water molecules. The modeling of ice-like water is explained in the SI. To account for the electrode potential, a double reference method was employed.19 We confirmed that the symmetric (6.5 × 1 × 1) supercell and the symmetric (9.5 × 1 × 1) supercell exhibited nearly the same diagram


Page 6 of 30

Page 7 of 30

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


under neutral conditions in an inner 8 Å area from the Ta3N5 surface (see Figure S3). All atomic structures were visualized using the XCrySDen software.31

Figure 1 The (3 × 1 × 1) supercell model for a Ta3N5(100) surface from (a) [100], (b) [010], and (c) [001] directions. Gray, blue, and red balls indicate Ta, N, and O atoms, respectively.

The electronic structures were calculated by the finite difference method with a grid parameter of approximately 0.18 Å and with a k-point sampling on a (1 × 3 × 3) grid for the slab in a vacuum. For the calculation of the slab in aqueous conditions, a grid parameter of 0.2 Å and a k-point sampling on a (1 × 2 × 2) grid were employed to reduce the computational cost. The systems with and without water were optimized until the maximum force on the atoms was smaller than 0.1 eV/Å and 0.05 eV/Å, respectively. A dipole correction was applied in asymmetric systems.32 The Hubbard U parameter was added for calculating the band alignments and diagrams.33 The U parameter was determined to be 2.0 eV for Ta atoms by the linear response method that removes the self-interaction error.34 The evaluation of the U parameter is discussed in the SI. Although we use a U = 5.0 eV, which represents the value that can approach a band gap similar to the experimental one, we confirmed that the position



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 8 of 30

of the VB does not change because it consists mainly of N 2p orbitals.16 Thus, although a U value of 2.0 eV underestimates the band gap to be 1.68 eV (exp. 2.1 eV),4 the position of the VB is little influenced by the band gap error. Here, we studied the solid/electrolyte interface with the double-reference method that included the effect of the electrode potential of the system using a solid/electrolyte/vacuum system as a “reference”.19 The method allows us to vary the electrode potential by changing the excess charge in the system and has been successfully applied to the Pt/water interface. Calculating the semiconductor/electrolyte interface with the surface-charging method is difficult because the potential distribution is not localized at the interface; it occurs in a region far away from the surface that requires a large slab model in a simulation. As such, the application of this method to semiconductor/electrolyte interfaces are limited to highly doped semiconductors in which the potential redistribution by the excess charge is limited to the interface region. We carefully determined the size of the slab model and varied the excess charge so that our semiconductor/electrolyte system satisfied the requirements of the surface-charging method.

The position of the Fermi energy must be determined by DFT calculations with careful considerations. The Fermi energy in a semiconductor is the charge-neutral level, which is in a one-on-one correspondence with the electrode potential. The Fermi energy in an intrinsic semiconductor is determined by the positions and the density of states (DOS) of the CB and VB,35 as shown in the following equation:

= + + ln

(2)

where and are the energy of VB maximum (VBM) and CB minimum (CBM), respectively, and and are the effective density of states of the VB and CB, respectively.5 The effective density of states can be calculated as the following:


Page 9 of 30

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


=

# $#

( ! " %& ' )

*

! = + / =

(4) #$#

) ! " %&' (

! = + /-

(3)

*

(5) (6)

where + + ) is the density of states of VB (CB) and - is the volume of the unit cell. To avoid contributions from the fixed atoms in the slab, only the density of states projected on the relaxed part of the slab was used. For an n-type semiconductor, the Fermi energy is also determined to satisfy charge neutrality. In this case, the formula for the Fermi energy is classified into two parts according to the position of the donor level induced by dopants. When the Fermi level is located at a surface state or above the CBM by at least ≈ 0.025 eV (T = 300 K), the Fermi level is determined by the directly calculated value. On the other hand, if the positions of the CBM and the donor level are very close (i.e., within an order of ), the Fermi energy should be calculated as the following:

= + ln

6

(7)

where 7 is the dopant density.35 However, such small doping densities cannot be captured by our calculations due to the limited size of the unit cell. Thus, we did not use (7) in our calculations. The calculation of the Fermi level depends on the DOS (eq. (2)), i.e., it depends on the amount of surface states relative to bulk states. When we used the local density of states (LDOS) for only the relaxed part of the slab model, its surface area corresponded to 87.7 m2 g-1 on the undoped surface. An experimentally synthesized Ta3N5 nanocrystal has a surface area of approximately 60 m2 g-1.14, 36 Thus, the weight of the surface state from our model is reasonable.



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 30

3. Results and discussion 3.1. Surface structures of n-type Ta3N5 A. Surface-ON Ta3N5 model As a low dopant model, a model for a surface with ON impurities was employed. Here, we assumed that the introduction of ON impurities proceeds from the surface under aqueous conditions associated with the electrochemical production of nitrogen or ammonia, as shown in the following reactions: Ta3N5 + xH2O → Ta3N5–xOx + 2xH+ + 2xe– + x/2 N2 (N2 production) (R2) Ta3N5 + xH2O + xH+ + xe– → Ta3N5–xOx + xNH3 (NH3 production)

(R3).

The stability of the ON surface under electrochemical conditions was evaluated by calculating the reaction energies as follows:

∆GN2 = G(Ta3N5-xOx) + x/2 G(N2) + 2x89 : $ – G(Ta3N5) – xG(H2O)

(8)

∆GNH3 = G(Ta3N5-xOx) + xG(NH3) – G(Ta3N5) – xG(H2O) – x89: $

.(9)

Here, a computational hydrogen electrode was used to evaluate 89: $ .37 For the ON surface models, between one and four ON impurities per unit cell were introduced, denoted as “nON surface” (n = 1 – 4). These models were equivalent to a doping concentration of 9.4 × 1013 e/cm–2, 1.9 × 1014 e/cm–2, 2.8 × 1014 e/cm–2, and 3.8×1014 e/cm–2 for n = 1, 2, 3, and 4, respectively. The positions of the ON impurities were determined to be most stable on the 1ON and 2ON surface models. For the 3ON surface, a stable structure was found by adding one ON impurity onto the most stable 2ON surface. The 4ON surface was determined similarly. The


Page 11 of 30

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


structures of the nON surfaces (n = 1 – 4) are shown in Figure S4. Figure 2 shows the reaction energy as a function of electrode potential. Generally, ammonia was produced in the negative potential region, and nitrogen was produced in the positive potential region. Thermodynamically, an increased number of ON impurities enhanced the stability. This result is somewhat consistent with previous work showing that bulk Ta2O5 is thermodynamically more stable than bulk Ta3N5 in water.22 In reality, the creation of ON impurities is not only controlled by the reaction energy but also by reaction kinetics. With the intention of demonstrating interfacial structures and band diagrams of n-type Ta3N5, 2ON and 4ON surfaces were selected for further investigations.

Figure 2. Reaction energies for N2 (solid line) and NH3 (dotted line) production to form nON (n = 1–4) surface structures. The purple lines located near the bottom indicate that the 4ON surface is the most stable.

Next, the surface geometry of the adsorbents was examined because the surface area is considered to be covered by the stable adsorbents OH and O. Here, we denote adsorbents with “*”. We assume that such adsorbents are generated by reactions with water, as shown in the following reactions: * + H2O → OH* + H+ + e−


(R4)


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 12 of 30

∆GOH = G(OH*) + 89 : $ – G(*) – G(H2O)

(10)

* + H2O → O* + 2H+ + 2e−

(R5)

∆GO = G(O*) + 289 : $ – G(*) – G(H2O).

(11)

The coverage of O* and OH* was limited to 0.94–1.87 molecule nm−2 in our models. The Gibbs free energy of each reaction was also dependent on the electrode potential. The resulting surface phase diagrams are shown in Figure 3. On the undoped surface (Figure 3(a)), the surfaces with OH*, O*+OH*, and 2O* were most stable below −0.5 V vs. RHE, between −0.5 and 0.6 V vs. RHE, and above 0.6 V vs. RHE, respectively. On the other hand, on the 2ON surface (Figure 3(b)), the surface with 2OH* was most stable at all potentials. On the 4ON surface (Figure 3(c)), the surface structure switches from the one with 2OH* to 2O* at −0.3 V vs. RHE. In the OER potential region, i.e., 1.5 V vs. RHE (1.23 V + overpotential), surfaces with 2O*, 2OH*, and 2O* are most stable on the undoped, 2ON, and 4ON surfaces, respectively. The reason for the high stability 2O* on the undoped surface is explained as follows. On the undoped surface, O2 molecules cannot adsorb, and the O-adsorbed structure is unstable without surface corruption or reconstruction.18 Thus, on the undoped surface with 2O*, the binding of two O atoms causes the displacement of surface N atoms. As a result, one stable N-N pair per unit cell is formed at the surface with a bond length of 1.25 Å (Figure S5). On the other hand, the stability trend of the ON surfaces can be explained by the binding energies of an oxygen molecule and OH. From previous calculations,18 the binding energies of both O2 and OH increase with the number of ON impurities. The increase in the OH binding energy is smaller than that in the O2 binding energy; thus, surfaces with O* are more stable than those with OH* as the ON impurities increase.


Page 13 of 30

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 3. Reaction energies of OH and O adsorption on the (a) undoped surface, (b) 2ON surface, and (c) 4ON surface. Colors indicate adsorbents on the surface.

B. Bulk-ON Ta3N5 model Next, the bulk-ON Ta3N5 model was considered as a high dopant model. Here, bulk ON impurities are introduced to match the experimentally reported ratio of O and N; replacement of one N atom with one O atom in the unit cell gives rise to a Ta3N4.75O0.25 composition, which is similar to the experimentally obtained Ta3N4.80O0.20 composition.20 The Ta3N5 has three distinct N atoms in its unit cell. The position of the ON impurity was determined thermodynamically, and we obtained a structure with a bulk ON at the 3-coordinate N site (Figure S6). The lattice constants of the bulk-ON model were a = 3.91 Å, b = 10.28 Å, and c = 10.36 Å, which are very similar to those of the undoped Ta3N5, except for a 0.4% reduction in the b-axis. The surface based on this composition is denoted as the “bulk-ON surface”. The position of the ON impurity and the reduction of the b-axis lattice constant are also consistent with previous experimental results.20 Similar to the presentation of the surface-ON models, the equilibrium structure of the surface adsorbents in an electrochemical environment was determined and is depicted in Figure 4. From the calculations, the most-stable surface adsorbents switched from 2OH* to O*+OH* at



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

1.24 V vs. RHE. This potential is almost the same as the redox potential of O2/H2O. Thus, both 2OH* and O*+OH* were employed for further investigations.

Figure 4. Reaction energies of OH and O adsorption on the bulk-ON surface. 3.2. Band alignments of n-type Ta3N5 as a function of the surface structure As discussed above, under the applied potential, ON impurities and O and OH adsorbents are induced at the interface. Here, we discuss how ON impurities and adsorbents affect the band diagram of Ta3N5 in vacuum. To get the band diagram in real space, the LDOS was calculated on the undoped, 2ON, and 4ON surfaces, as shown in Figure 5. On the undoped surface, the surface-dangling bond state lied above the VBM and the Fermi level lied between the top of the dangling bond level and the CBM (Figure 5(a)). The undoped surface was an intrinsic semiconductor with the position of the VBM in the bulk region almost matching the redox potential of O2/H2O. On the other hand, the 2ON and 4ON surfaces were n-type semiconductors with the Fermi energies located in the CBs. The positions of the VB and CB were almost the same in the 2ON and 4ON surfaces (only a 0.1 eV shift was observed, Figures 5(b) and (c)). In addition, the surface-dangling bond states disappeared because they were saturated by excess electrons donated by the ON impurities.17


Page 14 of 30

Page 15 of 30

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 5. Band diagrams of the (a) undoped surface, (b) 2ON surface, and (c) 4ON surface in vacuum with geometries from the [001] direction. Only atoms near the surface are shown in the geometry. The averaged z-coordinate of the atoms in the topmost layer is set to be zero in the x-axis. The vacuum level is set to be zero in the y-axis. The upper yellow line at 4.44 eV is the redox potential of H+/H2, and the lower line at 5.67 eV is the redox potential of O2/H2O. The red line indicates the Fermi energy in each figure.

In Figure 6, the band alignments obtained by LDOS calculations of the surfaces without adsorbents, the surfaces with OH*, the surfaces with O*, and the surfaces detected at 1.23 V vs. RHE are shown. Surface states only existed on the undoped surface regardless of the adsorbents. Generally, the adsorptions of OH and O induce a down-shift of the band positions because the negatively charged OH and O form dipoles at the interface, whereas the reduction by OH is relatively small or negligible. The position of the surface states varied according to the surface



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

adsorbent. The undoped surfaces both without any adsorbent and with OH* had surface states just above the VB, implying that these surfaces have relatively high stabilities. The surface states on the undoped surface with O* and 2O* came from N–N bonding or anti-bonding states, as shown in Figure S5. The adsorption of O atoms with negative charges resulted in the donation of holes to Ta3N5. Thus, the undoped surface with adsorbed O atoms showed some properties of a p-type semiconductor. The holes were distributed at the surface N atoms and strengthened the N–N bonding. This situation is similar to the self-oxidation process of photocatalytic water oxidation (shown in the below reaction). 2N3– + 6 h+ → N2. (R6) The band alignments of the 2ON surfaces with 2OH* and with O* and of the 4ON surface with 2O* also showed similar behavior. Because the formal valences of the adsorbed OH and O are −1 and −2, respectively, the intrinsic nature of the semiconductor was recovered on these surfaces. Although the densities of the ON impurities in our models were much less than those of the experimentally reported values, our results indicate that a reduction of the doping density is induced by the adsorbents.


Page 16 of 30

Page 17 of 30

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 6. Band alignments of the undoped, 2ON, and 4ON surfaces with the different adsorbents. Yellow dashed lines indicate the redox potentials of H+/H2 and O2/H2O. Original LDOS data are shown in Figure S7.

The band alignments of the bulk-ON surface with adsorbents were also calculated and are shown in Figure 7. The surfaces with 2OH* and O*+OH* are also shown because these were stable at approximately 1.23 V vs. RHE. On the bulk-ON surfaces, a similar trend was observed; the adsorptions of OH and O induced the down-shift of the band positions. Clearly, the position of the VB in the bulk-ON was lower than that in the undoped surface and was always situated below the redox potential of O2/H2O regardless of the adsorbents, which indicates that Ta3N4.75O0.25 is available for water oxidation. We note that the previous DFT calculations on 0.83–2.50% O-enriched Ta3N5−xOx or on O-enriched and Ta-vacant Ta2.91N4.58O0.41 showed a similar down-shift of the band position,12,21 whereas calculations on Ta3N4.83O0.25 showed an up-shift of the band position.12 Because the band positions are sensitive to a dipole, only a slight change from normal stoichiometry results in either an up- or down-shift of the positions.



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Figure 7. Band alignments of the bulk-ON surface with adsorbents. Yellow dashed lines indicate the redox potentials of H+/H2 and O2/H2O. Original LDOS data are shown in Figure S8.

3.3 SB at the n-type Ta3N5/water interface in under electrochemical conditions The band diagrams of the n-type Ta3N5/water interface as a function of electrode potential were investigated to reveal the behavior of the SB, i.e., whether BEP or FLP occurs. The bulk-ON surface model with water molecules was used in this subsection. The systems with excess charge were optimized by relaxing the adsorbents and water molecules except for the “reference water” near the boundary. In our calculations, an SB is observed due to the charge transfer from the semiconductor to the water. Here, we can define four properties from the LDOS calculations on the Ta3N5/water interfacial systems: the energy of the VB in bulk (EV), the energy of the VB at the surface (EVS), the semiconductor (Schottky-type) barrier height (VSC), and the Fermi energy. An example of a band diagram with an SB is shown in Figure 8. Adding positive charge into the system resulted in a lowering of the bulk valence level because of an increased dipole at the interface. States in z > 3 Å came from the molecular orbital of water in the electrolyte, and states between z = 1 and 3 Å originated from the adsorbents.


Page 18 of 30

Page 19 of 30

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 8 Band diagram and geometry of the bulk-ON surface with 2OH* with excess charge of 2.5e in the calculated cell. Only atoms near the interface are shown in the geometry. The averaged z-coordinate of the atoms in the topmost layer is set to be zero in the x-axis. The vacuum level is set to be a reference. The upper yellow line at 4.44 eV is the redox potential of H+/H2, and the lower line at 5.67 eV is the redox potential of O2/H2O. The red line indicates the Fermi energy.

We note that we limited our calculations in the potential region from -0.86 to 0.73 V vs. RHE because the double-reference method was applied to the doped semiconductor/electrolyte systems. Above 0.73 V vs. RHE, the Fermi levels were located outside of the CB, indicating that the doping density was fully reduced by the excess charges. The band diagrams calculated in the potential region from -0.86 to 2.08 V vs. RHE are shown in Figure S9. We can discuss the



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

behavior of the Ta3N5/water interfaces according to the shifts of EV, EVS, and VSC. In the potential region from -0.86 to 0.73 V vs. RHE, the EV shifted downward as the electrode potential increased with a slope of 0.83. Accordingly, the EVS also shifted downward as the electrode potential increased; however, its slope (0.41) was smaller than that of Ev. The slope of 0.41 indicates that the diagram at the interface showed intermediate behavior between BEP and FLP; BEP behavior is indicated by a slope of 0.0, and FLP behavior is indicated by a slope of 1.0. The FB potential of −0.8 V vs. RHE was obtained to satisfy VSC = 0 by fitting VSC with a linear function, which agrees well with the experimental values between −0.5 V and 0.0 V vs. RHE.5,12,14

Figure 9 Band diagrams as a function of electrode potential. The position of EV and EVS are shown as red and blue lines, respectively, referring to the left y-axis, and VSC values are shown as yellow lines referring to the right y-axis. Original LDOS data are shown in Figure S10.

From our calculations, partial FLP is considered to prevent the formation of a high SB; the partial FLP at the photoanode/water interface was unfavorable because the driving force for charge separation cannot be fully gained by the potential increase.10 To clearly show the origin of FLP, changes in the electron density and electrostatic potential between −0.86 V vs. RHE and


Page 20 of 30

Page 21 of 30

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


0.73 V vs. RHE were calculated. In Figure 10(b), the positive red area represents an electron accumulation region, and the negative blue area represents an electron reduction region. As the electrode potential was increased, electron reduction occurred in a wide area inside Ta3N5 (−12 Å < z < 0 Å in Figure 10) and the local interfacial region (1 Å < z < 3 Å in Figure 10). Accordingly, the change in the electrostatic potential varied in the same region (Figure 10(c)). Full BEP is the limit where all the changes in the electron density and electrostatic potential occur in the Ta3N5 bulk region, whereas full FLP is the limit where all changes occur at the interface. Therefore, partial redistribution of electrons at the interface can be the origin of FLP at the Ta3N5/water interface. Considering the traditional Bardeen’s model, which states that the surface states attribute to a constant SB height,38 it is probable that such charge redistribution is related to the existence of surface states and molecular orbitals of the adsorbents. However, these states or orbitals cannot be completely removed, implying an intrinsic partial FLP nature at the Ta3N5/water interface. .

Figure 10 Changes in the electron density and electrostatic potential between −0.86 V vs. RHE



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

and 0.73 V vs. RHE. (a) Geometry of the bulk-ON surface with 2OH* in an aqueous solution. (b) Change in the electron density (∆n = n(0.73 V vs. RHE)−n(−0.86 V vs. RHE)). The positive area represents an electron accumulation region, and the negative blue area represents an electron reduction region. (c) Change in the electrostatic potential (∆V = V(0.73 V vs. RHE) −V(−0.86 V vs. RHE)). An averaged z-coordinate of atoms in the Ta3N5 topmost layer is set to be zero in the x-axis.

4 Conclusions

We have presented a band diagram of the Ta3N5/water interface using DFT combined with a double-reference method. Surface ON or bulk ON impurities were introduced as n-type Ta3N5 models. We first found the stable structure of each surface model based on surface adsorbents. Then, the effect of the adsorbents on the band alignment was discussed. The adsorptions of O* and OH* negatively shifted the band position and changed the doping density of Ta3N5. Finally, band diagrams of the n-type Ta3N5/water interface were obtained. Based on our calculations, the SB height is dependent on the electrode potential, and unfavorable partial FLP was found. In our study, it was clearly shown that the band edge positions and the density of surface states depend on the density of ON impurities in the bulk and at the surface. In a real system, the density of ON impurities is not homogeneous, and other impurities such as N vacancies and Ta vacancies exist. Calculations of the SB height on such systems will provide more realistic charge carrier behavior of the Ta3N5/water interface in PEC conditions. In addition, the water-splitting reaction is very complex, and it consists of several types of bulk systems and interfaces. Future work should also focus on investigations of semiconductor/co-catalyst and/or co-catalyst/electrolyte interfaces.


Page 22 of 30

Page 23 of 30

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


ASSOCIATED CONTENT

Supporting Information. Explanations of the structure of ice-like water and evaluation of the U parameter. Band edge and surface state positions of Ta3N5 with optimized and experimental lattice constants (Figure S1); band edge and surface states positions using (6 × 1 × 1) and (9 × 1 × 1) supercells (Figure S2); band diagrams of neutral (6.5 × 1 × 1) and (9.5 × 1 × 1) supercells (Figure S3); structures of nON surfaces (n=1 – 4) (Figure S4); surface structures of the undoped surfaces with one or two O* (Figure S5); positions of ON in bulk-ON structures (Figure S6); LDOS of the undoped, 2ON and 4ON surfaces (Figure S7); LDOS of bulk-ON surfaces (Figure S8); band diagrams as a function of electrode potential (Figure S9); LDOS of bulk-ON surfaces as a function of electrode potential (Figure S10).

AUTHOR INFORMATION

Corresponding author

*Email: [email protected], [email protected]



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Notes

The authors declare no competing financial interest.

ACKNOWLEDGMENTS

E. W. was supported by the Japan Society for the Promotion of Science through Program for Leading Graduate Schools (MERIT). The computations were performed at the Research Center for Computational Science, Okazaki, Japan, and the Supercomputer Center, the Institute for Solid State Physics, the University of Tokyo, Japan.

REFERENCES (1) McCrory, C. C. L.; Jung, S.; Peters, J. C.; Jaramillo, T. F. Benchmarking Heterogeneous Electrocatalysts for the Oxygen Evolution Reaction. J. Am. Chem. Soc. 2013, 135, 16977– 16987. (2) Lyons, M. E. G.; Floquet, S. Mechanism of Oxygen Reactions at Porous Oxide Electrodes. Part 2--Oxygen Evolution at RuO2, IrO2 and IrxRu1-xO2 Electrodes in Aqueous Acid and Alkaline Solution. Phys. Chem. Chem. Phys. 2011, 13, 5314–5335. (3) Hara, M.; Hitoki, G.; Takata, T.; Kondo, J. N.; Kobayashi, H.; Domen, K. TaON and Ta3N5 as New Visible Light Driven Photocatalysts. Catal. Today , 2003, 78, 555–560. (4) Hitoki, G.; Ishikawa, A.; Takata, T.; Kondo, J. N.; Hara, M.; Domen, K. Ta3N5 as a Novel


Page 24 of 30

Page 25 of 30

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Visible Light-Driven Photocatalyst (λ

First-Principles Study of the Band Diagrams and ... - ACS Publications

Recommend Documents