Journal ofMaterials Chemistry A

PAPER

Publ

ishe

d on

12

Oct

ober

201

5. D

ownl

oade

d by

Nat

iona

l Tai

wan

Uni

vers

ity o

n 15

/02/

2016

05:

35:4

9.

View Article OnlineView Journal | View Issue

An integrated co

aDepartment of Chemistry, National Taiwan

rsliu@ntu.edu.twbDepartment of Physics, National Taiwan N

E-mail: su.hu@ntnu.edu.twcNational Synchrotron Radiation Research CdProgram for Science and Technology of Acce

University, Hsinchu 30010, TaiwaneCollege of Sciences, Chongqing Univers

Chongqing 400065, ChinafInstitute of Nuclear Energy Research, AtomicgDepartment of Mechanical Engineering an

Technology, National Taipei University of ThDepartment of Physics, Tamkang University

† Electronic supplementary informa10.1039/c5ta06202k

Cite this: J. Mater. Chem. A, 2015, 3,23466

Received 7th August 2015Accepted 11th October 2015

DOI: 10.1039/c5ta06202k

www.rsc.org/MaterialsA

23466 | J. Mater. Chem. A, 2015, 3, 23

balt disulfide (CoS2) co-catalystpassivation layer on silicon microwires forphotoelectrochemical hydrogen evolution†

Chih-Jung Chen,a Po-Tzu Chen,b Mrinmoyee Basu,a Kai-Chih Yang,b Ying-Rui Lu,cd

Chung-Li Dong,h Chong-Geng Ma,e Chin-Chang Shen,f Shu-Fen Hu*b

and Ru-Shi Liu*ag

An integrated cobalt disulfide (CoS2) co-catalyst passivation layer on Si microwires (MWs) was used as

a photocathode for solar hydrogen evolution. Si MWs were prepared by photolithography and dry

etching techniques. The CoS2–Si photocathodes were subsequently prepared by chemical deposition

and thermal sulfidation of the Co(OH)2 outer shell. The optimized onset potential and photocurrent of

the CoS2–Si electrode were 0.248 V and �3.22 mA cm�2 (at 0 V), respectively. The best photocatalytic

activity of the CoS2–Si electrode resulted from lower charge transfer resistances among the

photoabsorber, co-catalyst, and redox couples in the electrolyte. X-ray absorption near edge structure

was conducted to investigate the unoccupied electronic states of the CoS2 layer. We propose that more

vacancies in the S-3p unoccupied states of the CoS2–Si electrode were present with a lower negative

charge of S22� to form weaker S–H bond strength, promoting water splitting efficiency. Moreover, the

CoS2 co-catalyst that completely covered underlying Si MWs served as a passivation layer to prevent

oxidation and reduce degradation during photoelectrochemical measurements. Therefore, the optimal

CoS2–Si electrode maintained the photocurrent at about �3 mA cm�2 (at 0 V) for 9 h, and its hydrogen

generation rate was approximately 0.833 mmol min�1.

Introduction

At present, the annual global energy consumption rate is about15 TW, and it is estimated to increase to 25–27 TW by 2050.1 In2011, approximately 81% of the worldwide energy requirementwas met utilizing fossil fuels. However, relying on fossil fuelsleads to 30.4 Gt carbon dioxide emission, resulting in seriousgreenhouse effect and global warming.2 Solar light, an inniteenergy source, is received by the Earth at an approximate rate of120 000 TW per year.1 Efficient harvesting of solar irradiation

University, Taipei 10617, Taiwan. E-mail:

ormal University, Taipei 11677, Taiwan.

enter, Hsinchu 30076, Taiwan

lerator Light Source, National Chiao Tung

ity of Posts and Telecommunications,

Energy Council, Taoyuan 32546, Taiwan

d Graduate Institute of Manufacturing

echnology, Taipei 10608, Taiwan

, Tamsui 25137, Taiwan

tion (ESI) available. See DOI:

466–23476

for energy generation is an optimized method for solving theenergy crisis. In 1972, Honda and Fujishima rst demonstratedthe conversion of solar energy into chemical fuels throughsplitting water.3 Various semiconductor materials have beenemployed to develop a highly efficient photocatalyst or photo-electrode. Silicon (Si) with a highly negative conduction bandedge (ca. �0.46 V vs. NHE) and small band gap (1.1 eV) hasfunctioned as the most promising photocathode material forsolar hydrogen production. However, low hydrogen evolutionreaction (HER) kinetics on Si surfaces have led to efforts ofmodifying noble Pt co-catalysts which are used as catalyticelectron collectors to decrease recombination.4–6 In our previousresearch, Ag–Si microowers were prepared as a photocathodefor the hydrogen generation reaction by a facile chemicalmethod.7 The optimized photocurrent of Ag–Si electrodes was�35 mA cm�2 at �1.0 V (vs. Ag/AgCl), and its hydrogen evolu-tion rate (HER) reached 8.3 mmol min�1 by the synergetic effectof the Ag co-catalyst and plasmonic assistance. However, thesehigh-cost metal materials limit the large-scale deployment andpractical applications because of their low abundance in theEarth.8

The research group of N. S. Lewis previously demonstratedthat the earth-abundant Ni–Mo alloy coated-Si microwires,which were prepared by the vapor–liquid–solid (VLS) method,show high efficiencies for solar hydrogen generation.9,10 In our

This journal is © The Royal Society of Chemistry 2015

Fig. 1 Schematic illustration of the CoS2 co-catalyst passivation layer-decorated Si microwire photocathode.

Fig. 2 Schematic illustration of the fabrication procedure of Simicrowire arrays.

Paper Journal of Materials Chemistry A

Publ

ishe

d on

12

Oct

ober

201

5. D

ownl

oade

d by

Nat

iona

l Tai

wan

Uni

vers

ity o

n 15

/02/

2016

05:

35:4

9.

View Article Online

previous study, marcasite CoSe2 nanorods were preparedthrough a hydrothermal reaction, and subsequently spin-coatedon Si microwire arrays as a high efficiency photocathode.11

However, CoSe2 nearly aggregated on top of Si microwires, andthe major restriction of using Si materials for solar watersplitting is easy oxidation to electrically insulating SiO2 aerelectrolyte exposure.12 The oxidized Si electrodes reduce themigration of photogenerated carriers to their surface and resultin serious degradation or deactivation. Therefore, methylfunctional groups were applied to protect Si from oxidationthrough a two-step halogenation/alkylation reaction. Mixedmethyl/allyl monolayer-modied Si showed comparablepassivation performance as CH3–Si while allowing facilesecondary surface functionalization. However, the durability ofthese organic materials for long periods remains unclear.13,14

The monolayer graphene, prepared by chemical vapor deposi-tion (CVD), was also employed on Si photocathodes to suppressoxidation. Besides, nitrogen-doped and defect abundant gra-phene, fabricated through N2 plasma treatment, decreased theoverpotential of Si electrodes to further enhance its perfor-mance. The photocurrent at 0 V versus RHE (pH¼ 0) of N-dopedgraphene-decorated Si still decayed by about 70% within 10 000s.15 Amorphous molybdenum sulde (MoSx) or titanium oxide(TiO2) was incorporated with a titanium-protected n+p-Siphotocathode for improved water splitting stability, whereasthe depth of titanium native oxide gradually increased withlong-term measurement.12,16 In addition, the titanium thin lmoxidized to thermodynamically stable Ti(aq)

3+ and Ti(aq)2+ ions at

voltages below�0.315 and�0.528 V versus NHE in acidic media(pH ¼ 0). Thus, an integrated MoS2 passivation and catalyticlayer was directly deposited on the n+p-Si electrode, carrying outsolar hydrogen generation at 0 V versus RHE for more than 5days without degradation.17 Recently, an amorphous MoSxClythin lm was deposited on a p-type Si photocathode to achievehigh cathodic current �20.6 mA cm�2 at 0 V (vs. RHE).18

Inspired from these previous studies, the photocatalytic dura-bility of the Si electrode can be improved by integrating elec-trochemically stable materials.

Recently, earth-abundant inorganic electrocatalysts play keyroles in the conversion of renewable energy.19 Among variousnon-noble electrocatalysts, cubic pyrite-phase transition metaldichalcogenides (with the general formula MX2, where M ¼ Fe,Co, or Ni and X ¼ S or Se) showed promising efficiencies forelectrolytic hydrogen evolution.20–24 Among numerous metaldisuldes, CoS2 particles revealed the lowest cathodic over-potential (�128 mV vs. RHE) to drive the HER at 1 mA cm�2.24

Besides, an amorphous cobalt sulde (Co–S) lm was preparedvia electrochemical deposition, and it had low catalytic onsetoverpotential and approximately 100% faradaic efficiency in pH7 phosphate buffer.25 In the present work, electrochemicallystable cobalt disulde (CoS2) was selected as a co-catalystpassivation layer to be decorated on Si microwires (MWs) asa photocathode for solar water splitting, as shown in Fig. 1. Theoptimized photocurrent of CoS2–Si MWs was �3.22 mA cm�2 at0 V (vs. RHE), and linear sweep voltammograms (LSV) weremeasured through 30 scans without variations. Moreover,chronoamperometry was carried out at 0 V (vs. RHE) for 9 h with

This journal is © The Royal Society of Chemistry 2015

no degradation, and its hydrogen generation rate was 0.833mmol min�1. These results showed that the CoS2 passivationlayer was extremely robust in protecting Si MWs from oxidation.

ExperimentalFabrication of Si MW arrays

Si MW arrays were fabricated through the photolithographytechnique and dry etching system, as shown in Fig. 2. Theboron-doped p-type (100)-oriented Si wafers (resistivity: 1–25 U

cm) were selected for the preparation of Si MWs. First, Si waferswere cleaned by the STDmethod and using diluted hydrouoricacid (HF) solution. The silicon dioxide (SiO2), which functionedas a hard mask, was grown on Si wafers through atmospheric-pressure chemical vapor deposition (APCVD). The photo-resistant layer with a thickness of 800 nm was deposited on theoxide layer by spin coating. Aer the exposure and developmentof the photoresist by using the photolithography technique, thepatterns were dened as square arrays with 1.7 mm pitch size.The revealing SiO2 hard mask was then removed under a mixedatmosphere of CHF3, CF4, and Ar by TE5000 dry etching. The O2

plasma was applied to eliminate the photoresist under a pres-sure of 1.1 torr at 250 �C byMattson AspenII Asher. Si MWs were

J. Mater. Chem. A, 2015, 3, 23466–23476 | 23467

Journal of Materials Chemistry A Paper

Publ

ishe

d on

12

Oct

ober

201

5. D

ownl

oade

d by

Nat

iona

l Tai

wan

Uni

vers

ity o

n 15

/02/

2016

05:

35:4

9.

View Article Online

subsequently prepared using inductively coupled plasma-reac-tive ion etching (ICP-RIE) with an etching rate of 2 mm min�1

under the SF6 and O2 atmosphere, and immersed in diluted HFacid solution for 60 s to remove residual SiO2 hard masks. Theback side of the Si MWs was polished until the total thicknesswas 350 mm. An e-gun evaporator system was then used todeposit the Al metal with 500 nm thickness, which functionedas a back electrode.

Fabrication of Co(OH)2–Si MW arrays

The Co(OH)2 synthesis method was performed followinga previous study.22 Co(OH)2–Si MW arrays were prepared bya facile chemical reaction. The Al back electrode of the Si MWarrays was covered with Teon tape, which functioned asa mask during the chemical reaction. Si MWs were then xed ona microscopic slide. Cobalt nitrate hexahydrate [Co(NO3)2-$6H2O] (2.91 g) was dissolved in 10mL of deionized (DI) water toprepare 10 mmol precursor solution. About 40 mL of concen-trated ammonia (NH4OH) solution (30–33%) was then slowlyadded dropwise into the precursor solution under vigorousstirring. Aer adding NH4OH(aq), the nal solution was furtherstirred for 30 min and then poured in a cap of a Petri dish. Themicroscopic slide with Si MWs was immersed upside down inthe mixed solution, and kept approximately 2 mm from thebottom of the Petri dish. Another cap of a Petri dish covered onits top was then heated to 85 �C by using an oven for the desiredduration. Aer the reaction, the Co(OH)2-deposited-Si MWswere rinsed with DI water and blown dry with a nitrogen gun.

Fabrication of CoS2–Si MW arrays

CoS2–Si MW arrays were prepared by thermal suldation ofCo(OH)2–Si MWs. The as-prepared Co(OH)2–Si MW arrays and 2g of sulfur powder were separately placed in different aluminacrucibles. The crucible with Co(OH)2–Si MWs was placed at themiddle of the alumina tube furnace, and another crucible withsulfur powder was placed at the upstream position. The cruci-bles were heated to 500 �C at a rate of 5 �C min�1 and held for 1h under an Ar atmosphere with a ow rate of 30–40 mL min�1.The tube furnace was then naturally cooled down to roomtemperature.

Fabrication of Pt–Si MW arrays

Pt–Si MW arrays were synthesized through chemical electrolessdeposition. The Cu wire was mounted on the Al back electrodeof Si MWs using Ag paste and subsequently covered with anelectronic insulating epoxy, which functioned as a mask toavoid the Al back electrode from being etched. About 80 mL ofHF acid (1 M) and H2PtCl6 (10 mM) mixed aqueous solution wasprepared in a Teon beaker. The Si MW electrode was thenimmersed in this reaction solution for 180 s. Aer electrolessdeposition of Pt NPs, the electrode was rinsed with DI water anddried with a nitrogen gun.

23468 | J. Mater. Chem. A, 2015, 3, 23466–23476

Photoelectrochemical measurement

The Al back electrode of the as-synthesized photocathode wasstuck to Cu wire through the Ag paste and then heated todryness at 60 �C in an oven. The electronic insulating epoxy wassubsequently coated on the photocathode to prevent theundesirable dark current. The photoelectrochemical cell wascomposed of a three-electrode system, and it was surroundedwith the circulation water to maintain the temperature at 25 �C.The as-prepared photocathode functioned as the working elec-trode. Moreover, the saturated Ag/AgCl electrode and Pt platewere the reference and counter electrodes, respectively. Beforeconducting the photoelectrochemical measurement, the Ptplate was cleaned with concentrated sulfuric acid to remove thesurface contamination. Sulfuric acid solution (0.5 M, 250 mL)was prepared as the electrolyte. All photoelectrochemical char-acterizations were carried out under the illumination of a xenonlamp equipped with an AM 1.5 lter, and the light intensity wasconstant at 100 mW cm�2. The linear sweep voltammograms(LSVs), incident photon-to-electron conversion efficiency(IPCE), chronoamperometry, and gas evolution data wererecorded using a potentiostat (Eco Chemie AUTOLAB, TheNetherlands) and General Purpose Electrochemical System(GPES) soware. LSV measurements were conducted from 0.5 Vto �0.45 V with a scan rate of 20 mV s�1. IPCE; chro-noamperometry, and gas evolution were carried out at 0 V.Electrochemical impedance spectroscopy (EIS) data weremeasured by using an electrochemical workstation (760D, CHInstruments) under solar simulation with an applied bias of 0.4V. The sweeping frequency was conducted from 560 kHz to 20Hz with a 10 mV amplitude.

Characterization of materials

Scanning electron microscopy (SEM) images were collected byutilizing a JEOL JSM-6700F eld-emission SEM to observe themorphologies of the photocathodes. The crystallization andabsorption range of the photocathodes were characterizedthrough a Bruker D2 PHASER X-ray diffraction (XRD) analyzerwith Cu Ka radiation (l¼ 1.54178 A) and a Thermo EVOLUTION220 spectrometer, respectively. Raman spectra were collectedusing a Thermo DXR microscope with a 532 nm laser. The X-rayphotoelectron spectroscopy (XPS) spectra were recorded usinga PHI Quantera surface analyzer with Al Ka radiation (l ¼8.3406 A). The X-ray absorption spectroscopy (XAS) experimentswere conducted at the 20A1 and 16A1 beamlines of the NationalSynchrotron Radiation Research Center (NSRRC) in HsinchuCity, Taiwan.

Theoretical computational calculations

The structural and electronic properties of the CoS2 bulkmaterials were calculated by the periodic ab initio CRYSTAL09code26 using the local Gaussian-type basis sets. To exhibit theopen shell character of 3d7 conguration of Co2+ ions in thecompound, we considered the spin-polarized solutions to thepresent DFT calculations, and the spin value of each Co2+ ionwas locked to 1/2 in the rst eight cycles of the self-consistent

This journal is © The Royal Society of Chemistry 2015

Paper Journal of Materials Chemistry A

Publ

ishe

d on

12

Oct

ober

201

5. D

ownl

oade

d by

Nat

iona

l Tai

wan

Uni

vers

ity o

n 15

/02/

2016

05:

35:4

9.

View Article Online

eld (SCF) calculations. The hybrid exchange–correlationfunctional WC1PBE27 was used in this work because of itspositive performance in our previous work.11 The all-electronbasis sets 86-411d41G and 86-311d1G were chosen for cobalt28

and sulfur29 atoms, respectively. The basis set optimization bythe improved “Billy” script30 was carried out to further decreasethe total energy of our studied system. In all the SCF calcula-tions, we adopted an 8 � 8 � 8 k-point mesh31 in the Brillouinzone and set the truncation criteria of the bielectronic integrals(Coulomb and HF exchange series) as 8, 8, 8, 8, and 16. A pre-dened “extra extra large” pruned DFT integration grid waschosen for high accuracy. The energy convergence threshold ofthe SCF iterations was set to 10�9 Hartree, and the Andersonmixing scheme with 50% mixing was used to improve theconvergence speed. The convergence thresholds of the root-mean-square of the gradient and nuclear displacement for thegeometry optimization were set to 0.00006 Hartree per Bohr and0.00012 Bohr, respectively.

Results and discussion

In this study, Si MWs were fabricated by using the photoli-thography technique and dry etching system, as shown in Fig. 2(see details in the Experimental section). The scanning electronmicroscopy (SEM) images revealed that the pitch size of the SiMW arrays was �2 mm, and the diameter and length of a singlewire were 1 and 10 mm, as shown in Fig. 3a and c, respectively.Compared with Si planar structures, the light absorption andsurface area of the Si MWs were both increased to enhance theirphotoresponse. Moreover, the one-dimensional structure of SiMWs provided shorter diffusion lengths for minority carriersand single conductive direction for majority carriers to reducerecombination. The CoS2–Si photocathodes were preparedthrough chemical deposition and thermal suldation ofCo(OH)2. For convenience, we abbreviated Si MWs withdifferent CoS2 loading amounts as “CoS2–Si-T,” in which T (nm)represents the thickness of CoS2. Fig. S1a† reveals that the

Fig. 3 (a and b) Top view and (c and d) cross-sectional SEM images ofbare Si and CoS2–Si-250 microwire arrays.

This journal is © The Royal Society of Chemistry 2015

diameters of CoS2–Si-200 increased to �1.4 mm compared withthe pristine Si MWs, and CoS2–Si-250 MWs had a diameter of�1.5 mm, as shown in Fig. 3b. However, further extending thethickness of the CoS2 layer covered the interspaces between SiMWs in Fig. S1b and c.† Fig. 3d also presents that this chemicaldeposition method fully coated all of the Si MW surfaces withthe CoS2 layer. Low-magnication top-view SEM images showedthat the CoS2 outer shell uniformly covered on Si MWs withoutby-product particles aggregated on top, as illustrated in Fig. S2.†Besides, the SEM elemental mapping of CoS2–Si-250 MWs inFig. S3† revealed that the CoS2 outer shell was uniformlydistributed on the Si inner cores.

X-ray diffraction (XRD) showed that CoS2–Si MWs wereconsistent with the pyrite CoS2 structure (JCPDS 41-1471) withspace group Pa�3, as shown in Fig. 4a. The diffraction peaks at 2theta values of 27.8�, 32.3�, 36.2�, 39.8�, 46.4�, and 54.9� cor-responded to the Miller indices of (111), (200), (210), (211),(220), and (311), respectively. The diffraction peak intensity of SiMWs at�62� was reduced with increasing thickness of the CoS2layer, indicating a thicker passivation layer fully covered onunderlying Si MWs. The crystal structure of the cubic pyrite-typeCoS2 was described by the Pa�3 (no. 205) space group, and its oneunit cell is shown in Fig. S4.† Each cobalt cation was clearlysurrounded by the six nearest sulfur anions to form a slightlydistorted octahedron, and an S–S covalent bond was presentbetween the two nearest-neighbor sulfur atoms. From an ionicperspective, CoS2 can be regarded as a Co2+ ion plus one doubly

Fig. 4 (a) XRD and (b) reflectance spectra of CoS2–Si microwire arrayswith various CoS2 thicknesses. All diffraction peaks matched thestandard pattern of pyrite-phase CoS2 (JCPDS no. 41-1471).

J. Mater. Chem. A, 2015, 3, 23466–23476 | 23469

Journal of Materials Chemistry A Paper

Publ

ishe

d on

12

Oct

ober

201

5. D

ownl

oade

d by

Nat

iona

l Tai

wan

Uni

vers

ity o

n 15

/02/

2016

05:

35:4

9.

View Article Online

charged S–S dimer. The calculated structural data are listed inTable S1† for comparison with the experimental results. Aergeometry optimization, the electronic band structure wascalculated and shown in Fig. S5.† A half-metallic character wasconcluded from the calculated position of the Fermi energylevel EF, which was in good agreement with other band structurecalculations previously reported. The traditional crystal eld(CF) theory can be used to explain such a half-metallic behavior.Given that the ground CF conguration for the six-coordinatedCo2+ cation should be t62ge

1g, an energy gap may be found

between the fully lled t2g bands and empty eg bands of “beta”electrons (with the down spin) because of the known t2g–eg CFsplitting of the 3d electrons, whereas the single eg electronoccupation for “alpha” electrons (with the up spin) couldobscure such a CF gap. The energy gap for beta electrons wascalculated as 2.7261 eV. The calculated magnetic moment ofeach Co cation in one unit cell was 1 mB, thereby showing goodagreement with both the experimental result (see Table S1†)and prediction of CF theory. Reectance spectroscopy revealedthat the visible and near-infrared illumination (400–1100 nm)was absorbed by the pristine Si MWs, but the shorter wave-length (<400 nm) irradiation was reected, as shown in Fig. 4b.However, decorating the CoS2 layer on Si MWs reduced thereectance of ultraviolet light. Notably, a higher thickness of theCoS2 outer shell slightly blocked the Si inner cores for theabsorption of incident light and decreased the amount ofphotogenerated carriers. Five Raman modes [Ag, Eg, Tg(1), Tg(2),and Tg(3)] of the CoS2 structure were observed in Fig. 5.32 Ag

mode was the in-phase stretching vibration of sulfur atoms in

Fig. 5 Raman spectra of CoS2–Si microwire arrays with various CoS2thicknesses, and schematic illustration of Raman vibration modes ofCoS2.

23470 | J. Mater. Chem. A, 2015, 3, 23466–23476

the dumbbells, but Tg(2) was the out-of-phase stretching mode.Eg was the librational vibration of sulfur dumbbells. However,Tg(1) and Tg(3) modes corresponded to different compositionsof stretching and librational vibration of sulfur atoms. TheRaman spectrum of CoS2–Si-200 MWs not only revealed thevibration modes of the CoS2 outer shell, but also the inner Sicores whose Raman shi was �525 cm�1, corresponding to thepristine Si MWs. This result indicated that the deposition of theCoS2 layer on Si MWs had no negative effects on their surface.Upon increasing the loading amount of the CoS2 layer byincreasing its thickness to 250 nm, the Raman peak of Si MWscould not be observed, which showed that the thicker CoS2outer shell had no pinholes to reveal underlying Si MWs andblocked the incident laser beam. This result was complemen-tary to the data of the XRD and reectance spectra.

The three-electrode photoelectrochemical cell was applied tomeasure LSV of CoS2–Si electrodes in 0.5 M H2SO4 aqueoussolution under solar illumination (100 mW cm�2), as shown inFig. 6a. The bare Si MWs showed no photoresponse at 0 V, butthe photocurrent of the CoS2–Si-200 electrode increased to�0.694 mA cm�2. The pristine Si MWs showed photocatalyticactivity until the applying bias was approximately �0.75 V, andits photocurrent at �0.85 V was �3.12 mA cm�2 in the inset ofFig. 6a. The photocurrent was enhanced to �3.22 mA cm�2 at0 V aer Si MWs were coated by a CoS2 layer with 250 nmthickness. However, further increasing the loading amount of

Fig. 6 (a) Linear-sweep voltammograms of CoS2–Si electrodes withvarious CoS2 thicknesses. The inset is the linear sweep voltammogramof the bare Si electrode under more negative applied bias. (b) Onsetpotential and photocurrent versus CoS2 thicknesses of CoS2–Sielectrodes.

This journal is © The Royal Society of Chemistry 2015

Paper Journal of Materials Chemistry A

Publ

ishe

d on

12

Oct

ober

201

5. D

ownl

oade

d by

Nat

iona

l Tai

wan

Uni

vers

ity o

n 15

/02/

2016

05:

35:4

9.

View Article Online

the CoS2 layer by prolonging the chemical deposition durationreduced the photocurrents of CoS2–Si-280 and CoS2–Si-310electrodes to �2.01 and �1.57 mA cm�2 at 0 V, respectively. Inorder to identify whether the CoS2 material is photoexcited, thecontrolled electrode was fabricated by synthesizing a CoS2 layeron uorine-doped tin oxide (FTO) glasses. The CoS2/FTO elec-trode was prepared through the same procedure as CoS2–Si-250MWs, and Fig. S6† shows no impurity in the XRD spectraobserved. In this study, the onset potential was dened as thevoltage of the photocurrent reached �1.0 mA cm�2. The onsetpotentials and current densities of the CoS2/FTO electrodeunder dark conditions and solar irradiation were exactly thesame, as shown in Fig. S7a.† This result shows that the CoS2layer only functioned as a co-catalyst, accelerating the transferphotogenerated carriers from Si MWs to the interface with theelectrolyte, instead of a photoabsorber. Aer iR correction fromthe Rs value of electrochemical impedance spectroscopy (EIS)characterization in Fig. S7b,† the onset potential of the CoS2/FTO electrode was �0.146 V, and its current reached �28.3 mAcm�2 at �0.234 V. Fig. S7c† reveals the Tafel plot of the CoS2/FTO electrode with iR correction. The classic two-electronreactions of the HER mechanism can be described by two stepsas follows: a discharging step of the Volmer reaction (H3O

+ + e�

/ Hads + H2O) followed by a desorption step of the Heyrovskyreaction (Hads + H3O

+ + e� / H2 + H2O); or a recombination ofthe Tafel reaction (Hads + Hads / H2), where Hads indicates thehydrogen atom adsorbed at the active site on the surface of thecatalyst. Moreover, the catalyst with a Tafel slope of 116, 38, or29 mV dec�1 showed that the rate determining step in the HERprocess was the Volmer, Heyrovsky, or Tafel reaction, respec-tively. If the hydrogen atoms were weakly bonded to the surfaceof the catalyst, the adsorption step would limit the overallreaction rate. By contrast, as the hydrogen atoms were stronglybonded to the surface of the catalyst, the desorption step wouldrestrict the overall reaction efficiency.33 The CoS2/FTO electrodewith a Tafel slope of 61.9 mV dec�1 revealed that the CoS2 layermight follow the Volmer–Heyrovsky mechanism, and rates ofthe discharge step and desorption step might be comparableduring the HER process.34

The transient currents of CoS2–Si electrodes with variousCoS2 thicknesses were measured at 0 V under chopped illumi-nation, as shown in Fig. S8a.† No photocurrent was generatedby the bare Si MWs, which was complementary to the LSVresult, as shown in the inset of Fig. S8a.† By contrast, all theCoS2–Si electrodes showed transient switch-on characteristicsunder pulsed illumination. Besides, no overshot-photocurrentwas observed in these CoS2–Si MWs as the incident light wasactivated. This result demonstrated that photogenerated elec-trons did not accumulate on the surface of the electrodes, butefficiently transferred to the electrolyte through the co-catalystassistance of the CoS2 layer. In this study, the CoS2–Si-250electrode was selected to carry out incident photon-to-electronconversion efficiency (IPCE) at 0 V under solar simulation (100mW cm�2) because of its optimized photocatalytic activity, asshown in Fig. S8b.† The maximum efficiency of 5.12% for CoS2–Si-250 MWs was achieved at a wavelength of 665 nm. Reec-tance spectra show that the visible and near-infrared

This journal is © The Royal Society of Chemistry 2015

illumination was absorbed and UV incident light was reectedby Si MWs in Fig. 4b. Besides, linear-sweep voltammogrammeasurements were conducted as shown in Fig. S7b† to revealthat the CoS2 layer functioned as a co-catalyst without contri-bution to photo-induced carriers. Therefore, the ICPE result ofthe CoS2–Si photocathode was detected between the wavelengthunder visible and near-infrared irradiation. Besides, previousstudies of the Si photoelectrode show the analogous IPCEresults.7,35,36

Fig. 6b shows variation in the onset potential and photo-current of CoS2–Si electrodes with various CoS2 thicknesses.The onset potential of pristine Si MWs was �0.810 V, but itpositively shied to �0.0844 V as the CoS2 thickness modiedon Si MWs was 200 nm. The improved-onset potential of CoS2–Si-250 was saturated at 0.248 V, but that of CoS2–Si-280 andCoS2–Si-310 electrodes were decreased to 0.167 and 0.127 V,respectively. These results demonstrated that a thinner CoS2layer inefficiently ameliorated HER kinetics on the surfaces of SiMWs, but the excess loading amount of the CoS2 outer shell alsoreduced the photocatalytic activity. The co-catalysts on photo-electrodes generally play three roles: passivation of surfacetraps, extraction of minority carriers and lowering the energybarrier for the redox reaction at the photoelectrode/electrolyteinterface.37 The surface of Si MWs retained high quality asshown in SEM images, because Si MWs were prepared byphotolithography and dry etching technique. The CoS2 layer onSi MWs improved simultaneously the photocurrent and on-setpotential, indicating that the CoS2 catalyst enhanced thekinetics of photo-induced carriers and reduced the possibility ofrecombination. Besides, the blocking absorption effect andcharge transfer efficiency were the factors contributing to theonset potential improvement of Si MWs by modifying the CoS2co-catalyst.38 The blocking absorption effect indicated thata thicker CoS2 outer shell reduced the absorption of theunderlying Si inner core and decreased the amount of photo-induced carriers to lower the photoresponse. However, thereectance spectra in Fig. 4b revealed a slight difference in theabsorption of CoS2–Si MWs with various decoration durations.This result indicated that the absorbing discrepancy of theCoS2–Si electrodes with different CoS2 loading amounts has nodominant role in photoelectrochemical performance. Bycontrast, the charge transfer efficiency, which might be themain factor in photoresponse, was related to the crystallizationof the CoS2 co-catalyst and resistances in the CoS2–Si electrode.The crystallization of the CoS2 outer shell inuenced therecombination possibility of carriers generated by Si MWs, butthe XRD spectra showed no evident difference between theCoS2–Si electrodes in Fig. 4a. Therefore, we proposed that themainspring of the photocatalytic discrepancy resulted from thevariation in charge transfer resistances between CoS2–Si elec-trodes with various deposition times and the electrolyte.

In order to conrm the as-described hypothesis, EIS wasconducted to analyse the charge transfer resistances among thephotoabsorber, co-catalyst, and redox couples in the electrolyte.EIS characterization was carried out under solar illuminationwith an applied bias of 0.4 V in Fig. 7a. The semicircle at highfrequency was the charge transfer resistance (Rct1) from

J. Mater. Chem. A, 2015, 3, 23466–23476 | 23471

Fig. 7 (a) Electrochemical impedance spectroscopy of CoS2–Sielectrodes at 0.4 V bias under solar illumination. (b) Schematic illus-tration of charge transfer resistances between the compositionmaterials of the photoelectrode (Rct1), and between the photo-electrode and redox couples in the electrolyte (Rct2). (c) Chargetransfer resistances versus CoS2 thicknesses of CoS2–Si electrodes.

Fig. 8 (a) Schematic illustration of Co L-edge and S K-edge X-ray

Journal of Materials Chemistry A Paper

Publ

ishe

d on

12

Oct

ober

201

5. D

ownl

oade

d by

Nat

iona

l Tai

wan

Uni

vers

ity o

n 15

/02/

2016

05:

35:4

9.

View Article Online

underlying Si inner cores to the CoS2 outer shell, but the lowfrequency one (Rct2) was from the CoS2 layer to redox couples inthe electrolyte, as shown in the schematic illustration in Fig. 7b.The charge transfer resistances versus various CoS2 thicknessesof CoS2–Si electrodes are summarized in Fig. 7c. Both chargetransfer resistances of CoS2–Si MWs were optimized as thethickness of the CoS2 layer reached 250 nm. The Rct2 value wascorrelated with the catalytic activity of the CoS2 outer shell.39

The Rct2 of CoS2–Si-200 was 369 U cm2 because of a lower HERkinetic improvement through a lower loading amount of theCoS2 layer. By increasing the CoS2 thickness to 250 nm, the Rct2value was dramatically reduced to 56.5 U cm2. However, furtherincreasing the loading amount of the CoS2 layer by prolongingthe chemical deposition time led to higher Rct2 values. Thisresult indicated that a highly thick CoS2 outer shell functionedas an electron-trapping center that enhanced longer diffusionlength of photoinduced carriers to react with redox couples inthe electrolyte and increased the probability of recombination.The Rct1 value was related to the interfacial quality and elec-tronic states between the CoS2 layer and Si MWs. The CoS2–Sielectrodes were prepared by identical Co(OH)2 deposition andthermal suldation methods, so we proposed that the interfa-cial quality was not the dominant affecting the parameter of theRct1 value. Besides, the previous study reveals that lower Rct1was due to more available electronic states in the co-catalyst.39

Therefore, the theoretical band structure calculation and X-rayabsorption near-edge structure (XANES) were applied to resolvethe electronic states of the CoS2 outer shell.

23472 | J. Mater. Chem. A, 2015, 3, 23466–23476

The diagrams for calculated partial and total densities ofstates (PDOS/DOS) of CoS2 are given in Fig. S9.† The known t2g–eg CF splitting could be found for the Co-3d states with both theup and down spins, which further conrmed our band structureanalysis. The t2g bands were considerably narrower than thevery dispersive eg bands. They were mainly located around �3.1and �1.9 eV for alpha and beta electrons, respectively, so theexchange splitting for the t2g states was estimated at 1.2 eV. Theupper eg band crossed the Fermi energy level with a width of 2.9eV, whereas the lower eg band was separated from the lower t2gband by an energy gap of 3 eV. The Co-3d states were stronglyhybridized with the S-3p states because of the strong overlapbetween each divalent cobalt ion and its nearest sulfur ligands,so a similar but more dispersive DOS pattern for the S-3p stateswas observed. The S-3d states were located at 6.8 eV and faraway from the S-3p states. The S-3s states ranged from�19 eV to�12 eV and corresponded to the outermost core band consist-ing of two sub-bands with an energy separation of 3 eV.

According to a previous study, the density functional theory(DFT) calculation was carried out to reveal that the sulfur atomsof the pyrite catalyst were HER active sites.40 Guo et al. prepareda MoS2–SiC hybrid electrocatalyst for hydrogen evolution.41 TheXPS spectra show the binding energy of C in the MoS2–SiCshied to a higher value in comparison with pristine SiC,indicating that surrounding electrons of C are transferred toother atoms. This result reveals that the electrons were acceptedfrom C and rapidly transferred to Si and S active edges. More-over, longer S–Hads bond length on the pyrite catalyst witha weaker bond strength reduced the kinetic energy barrier byeffectively promoting H–H bond formation. We proposed thatthe presence of more vacancies in the unoccupied states of CoS2resulted in more photogenerated electrons of Si MWs trans-ferring to CoS2 and enhancing the photocatalytic activity. Thetheoretical calculation of band structure shows that the unoc-cupied electronic states of the CoS2 layer were composed of Co-3d and S-3p orbitals. Therefore, Co L-edge and S K-edge XANESwere used to investigate the unoccupied electronic states of the

absorption spectra, and normalized XANES (b) Co L-edge and (c) S K-edge spectra.

This journal is © The Royal Society of Chemistry 2015

Fig. 9 (a) Band alignment diagram between CoS2 and Si materials.High-resolution Si 2p spectra of the as-prepared CoS2–Si-250 elec-trode (b) with increasing Ar ion etching time and (c) different coordi-nation conditions.

Paper Journal of Materials Chemistry A

Publ

ishe

d on

12

Oct

ober

201

5. D

ownl

oade

d by

Nat

iona

l Tai

wan

Uni

vers

ity o

n 15

/02/

2016

05:

35:4

9.

View Article Online

CoS2 layer because the Co L-edge and S K-edge absorption,respectively, involved the transition from the 2p to 3d states and1s to 3p states, as shown in Fig. 8a. The stronger XANESintensity indicates the presence of more vacancies in theunoccupied states of CoS2. In this study, CoS2–Si-200 and CoS2–Si-250 with the worst and optimized Rct1 values were selected toconduct XANES analysis, as shown in Fig. 8b and c. Two peaksin the Co L-edge spectrum of the CoS2 outer shell correspondedto L2 (�795 eV) and L3 (�780 eV) absorption. A shoulder at theL3 peak shied by approximately 4–8 eV, which indicated thattransitions from the Co-2p core level to empty Co-4s states werepresent.42 The Co L3 absorption intensity of the CoS2–Si-200electrode, stronger than that of CoS2–Si-250 MWs, showed thatmore vacancies were present in its unoccupied Co-3d states. Bycontrast, the S K-edge jump of the CoS2–Si-250 electrode washigher than that of CoS2–Si-200 MWs. This result revealed anopposite trend, which presented more empty states in the S-3porbital of the CoS2–Si-250 electrode. The theoretical calculationof the electronic states showed that the Co-3d and S-3p stateswere strongly hybridized. Therefore, more vacancies in theunoccupied states of the CoS2 layer normally simultaneouslyenhance both Co L-edge and S K-edge absorption. We proposedthat the reversed tendency of Co L-edge and S K-edge absorptionintensity, observed in CoS2–Si electrodes, indicated a chargetransfer between Co and S atoms of CoS2–Si-250 MWs. The Satoms of the CoS2–Si-250 electrode, donating electrons to emptystates in the Co-3d orbital, generated more vacancies in unoc-cupied states of the S-3p orbital. This result leads to morephotoinduced-electrons from Si MWs of the CoS2–Si-250 elec-trode that could be accepted through the S atoms and furthertransferred to the S active sites on the surface of CoS2 for thesolar hydrogen generation reaction. Moreover, Fig. 8c alsoshows that the S K-edge absorption of CoS2–Si-250 MWs slightlyshied to higher energy. This result indicated that the CoS2–Si-250 electrode presented a lower negative charge of S2

2� andformed a weaker S–Hads bond strength to promote water split-ting efficiency.

The previous work of Ding et al. revealed that there was anohmic contact between metallic MoS2 and Si.39 However, thework function of CoS2 was about 5.5 eV (vs. vacuum level),43 andthe Fermi level of p-Si (resistivity: 10 U cm) was 4.9 eV.44 Aerthe band alignment, Fig. 9a shows that there was a small barrierfor photo-generated carriers migrating to the surface of thephotocathode, leading to lower photocurrent and ll factor, ascompared with MoS2–Si photoelectrode materials. Moreover,based on the studies of MoS2–Si PEC devices,39,45–47 the resis-tivity of utilized-Si wafer was lower as compared to our research.The Si photocathode with lower resistance improved the diffu-sion length of photo-induced carriers and further reduced thepossibility of recombination. During the chemical depositionprocess of the Co(OH)2 layer, there was an unavoidable SiO2

layer formation on Si MWs, owing to the exposure of aqueousreaction solution. The X-ray photoelectron spectroscopy (XPS)spectra were applied to investigate the SiO2 layer between theCoS2 shell and Si MWs. To detect the underlying Si MWs ofCoS2–Si electrodes, Ar ion etching was applied to remove theCoS2 outer shell until the intensity of the Si 2p peak no longer

This journal is © The Royal Society of Chemistry 2015

increased in Fig. 9b. The Si/SiO2 ratio of the CoS2–Si-250photocathode was �3.30, as shown in Fig. 9c. The XPS spectrareveal that Si/SiO2 ratios of CoS2–Si-200 and CoS2–Si-310 MWs,prepared under the shortest and longest exposure duration toaqueous reaction solution, were respectively about 4.87 and2.07 in Fig. S10.† This result shows that the thickness of theSiO2 layer increased with longer chemical deposition time. Thethicker SiO2 layer of CoS2–Si-310 blocked photo-inducedcarriers from transferring to the surface of the photocathode.Although the oxide layer generating on CoS2–Si-200 wasthinner, the less loading amount of CoS2 was unable to accel-erate the kinetics of solar hydrogen evolution. According toabove discussions, the water splitting efficiency of the CoS2–Siphotocathode material can be further enhanced by using lowresistivity Si MWs and reducing the formation of the SiO2 layer.

In this study, LSVmeasurements were carried out 30 times toanalyze the stability of photocathodes. The scanned photocur-rent of pristine Si MWs was �3.12 mA cm�2 at �0.85 V, butgradually decayed to �0.0882 mA cm�2, which only retained�2.83% of the initial photocurrent, aer the 30th scanningmeasurement in Fig. S11a.† Besides, the onset potential of SiMWs reduced to �0.85 V aer the h scanning test, but theconsecutive scans must be applied more negative bias to attain�1.0 mA cm�2, as shown in Fig. S11c.† The CoS2–Si-250 elec-trode was selected for this characterization because of itsoptimal photoelectrochemical performance. Both the onsetpotential and photocurrent of the CoS2–Si-250 electroderevealed no degradation during the scanning test in Fig. 10a

J. Mater. Chem. A, 2015, 3, 23466–23476 | 23473

Fig. 10 (a) Linear-sweep voltammograms of the CoS2–Si-250 elec-trode scanned 30 times, and (b) onset potential and photocurrent ofthe CoS2–Si electrode versus scan times.

Fig. 11 Chronoamperometry of bare Si and CoS2–Si-250 electrodesunder solar irradiation at 0 V, and the inset is the gas evolution of the

Journal of Materials Chemistry A Paper

Publ

ishe

d on

12

Oct

ober

201

5. D

ownl

oade

d by

Nat

iona

l Tai

wan

Uni

vers

ity o

n 15

/02/

2016

05:

35:4

9.

View Article Online

and b. This result indicated that the CoS2 co-catalyst effectivelyfunctioned as a passivation layer to defend Si MWs fromoxidation in the electrolyte. Pt nanoparticle (NP)-modied SiMWs (Pt–Si) were also prepared through chemical electrolessdeposition for comparing the protective effects with the CoS2–Sielectrode. The onset potential and photocurrent of the Pt–Sielectrode were 0.0868 V and �3.53 mA cm�2 at 0 V, respectively,as shown in Fig. S11b.† The photocurrent of the Pt–Si electrodesreached saturation aer applying bias above �0.2 V, because ofthe limited ion diffusion rate of the electrolyte or electronconducting rate of the Si MWs. The h scanned onset poten-tial and photocurrent of the Pt–Si electrode decreased to 0.0666V and �3.03 mA cm�2 at 0 V. Aer scanning for 30 times, theonset potential of Pt–Si MWs was maintained at 0.0364 V, andthe retention of the initial onset potential was only approxi-mately 41.9% in Fig. S11d.† Besides, the 30th scanned photo-current gradually reduced to �1.99 mA cm�2 at 0 V. This resultdemonstrated that Pt NPs, which only functioned as a co-cata-lyst, efficiently transferred the photoinduced carriers to theinterface and reduced recombination. However, Pt NPs wereunable to prevent the Si MWs from durability degradationbecause of oxidation. Further increasing the loading amount ofPt NPs to fully cover the surface of Si MWs using the prolongedelectroless deposition duration possibly improved its photo-catalytic stability. However, the high expense of Pt NPs limitedtheir practical utilization, and excess Pt NPs also blocked theabsorption of underlying Si MWs to reduce the photo-electrochemical performance. Besides, high loading amounts of

23474 | J. Mater. Chem. A, 2015, 3, 23466–23476

Pt NPs on Si MWs might also serve as trapping centers toincrease the probability of recombination.

XPS spectra were recorded not only to investigate the stabilityof the CoS2 passivation layer, but also to characterize whetherthe oxide layer formation on Si MWs leads to photocatalyticdecay aer the LSV measurement. Fig. S12a† shows the XPSspectra of the CoS2–Si electrodes with various CoS2 thicknesses.Notably, no Si peaks were observed in the spectra. This resultindicated that the CoS2 layer was completely coated on theentire surface of Si MWs without pinholes or cracks. High-resolution Co 2p XPS spectra in Fig. S12b† revealed two peaksoriginated from Co-2p3/2 and 2p1/2 of the CoS2 layer at 778.8 and794.0 eV. There was a broad band of the CoS2 layer at�162 eV inS 2p XPS spectra. Besides, Fig. S12b† revealed no satellite peaksat 780.9 and 797.8 eV in the Co 2p spectra, as well as the sulfatepeak at 168.7 eV in the S 2p spectra, indicating that the CoS2layer was stable without oxidation in air or even aer 30 LSVscans (purple curves in Fig. S12b†).22 The XPS Si 2p region of thebare Si and Pt–Si electrode showed two peaks with bindingenergies of 99.5 and 103.2 eV, separately ascribed from the SiMWs and the SiO2 oxide layer, respectively, as shown inFig. S13.† The Si/SiO2 ratios of the bare Si and Pt–Si MWs wereapproximately 0.708 and 1.33, respectively, but the ratiosdecreased to 0.511 and 0.567, respectively, aer 30 LSV scancharacterization. This result revealed that Si MWs oxidized tothe insulating SiO2 layer to enhance the possibility of recom-bination and result in serious photoelectrochemical degrada-tion as shown in Fig. S11.† Notably, the SiO2 peak of the CoS2–Si-250 electrode did not enhance aer 30 LSV scans, as shown inFig. 9c and S14.† Thus, the CoS2 passivation layer was extremelyrobust to stand up to the electrolyte from self-oxidation.Besides, the CoS2 outer shell fully covered the surface of the Siinner core to prevent oxide formation during the photo-electrochemical reaction. These factors predominantlycontributed to the considerably higher photocatalytic stabilityof CoS2–Si electrodes compared with pristine Si and Pt–Sielectrodes.

CoS2–Si-250 electrode under solar irradiation at 0 V.

This journal is © The Royal Society of Chemistry 2015

Paper Journal of Materials Chemistry A

Publ

ishe

d on

12

Oct

ober

201

5. D

ownl

oade

d by

Nat

iona

l Tai

wan

Uni

vers

ity o

n 15

/02/

2016

05:

35:4

9.

View Article Online

Chronoamperometry was conducted at 0 V under solar illu-mination. The photocurrent of the CoS2–Si-250 electrode wasstably retained at �3 mA cm�2 for 9 h in Fig. 11. This ndingindicated that the CoS2 passivation layer was not only extremelystable to overcome self-oxidation, but also robust to protectunderlying Si MWs from oxidation leading to serious recombi-nation of photoinduced carriers. Moreover, no photoresponsewas observed from the bare Si electrode in Fig. 11, whichrevealed that the current of CoS2–Si-250 MWs was not contrib-uted from the dark current or irreversible side reaction duringthe 9 h chronoamperometric measurement. The CoS2 layerefficiently transferred minority carriers of Si MWs to the inter-faces with the electrolyte and reacted with protons to generatehydrogen gas through the co-catalyst effect. The gas evolution ofthe CoS2–Si-250 electrode (the area was �1 cm2) was charac-terized at 0 V under solar illumination, as shown in the inset ofFig. 11. The hydrogen generation rate was 0.833 mmol min�1. Inaddition, the faradic efficiency of hydrogen was 83.2%.

Conclusions

In summary, we developed a CoS2 co-catalyst passivation layer-modied Si MW photocathode for water splitting. The opti-mized onset potential and photocurrent of the CoS2–Si elec-trode reached 0.248 V and �3.22 mA cm�2 (at 0 V), respectively,as the thickness of CoS2 was 250 nm. The low charge transferresistances of CoS2–Si-250 MWs among the photoabsorber, co-catalyst, and redox couples in the electrolyte led to high pho-toelectrochemical activity. Moreover, the XANES spectrarevealed that more vacancies in the unoccupied S-3p states ofthe CoS2–Si-250 electrode were present, which functioned asactive sites of water splitting, accepting more photogeneratedelectrons from Si MWs. The lower negatively charged S2

2� ofCoS2–Si-250 MWs also reduced the bond strength of S–Hads andpromoted its water splitting efficiency. The CoS2 outer shell onSi MWs showed a stable characteristic without self-oxidationaer the photocatalytic measurement, which was observedthrough the XPS spectra. Moreover, the CoS2 co-catalyst func-tioned as a robust passivation layer and inhibited the oxidationof Si MWs to further prevent the degradation of photocatalyticactivity. The CoS2–Si-250 MWsmaintained a photocurrent at�3mA cm�2 (at 0 V) for 9 h, and its hydrogen generation rate wasapproximately 0.833 mmol min�1.

Acknowledgements

The authors are grateful for the nancial supports of theMinistry of Science, Technology of Taiwan (Contract Nos. MOST103-2112-M-003-009-MY3 and MOST 104-2113-M-002-012-MY3),Academia Sinica (Contract No. AS-103-TP-A06), and NationalTaiwan University (104R7563-3). C.-G. Ma appreciates thenancial support from the National Natural Science Foundationof China (Grant No. 11204393) and Natural Science Foundationof Chongqing (Grant No. CSTC2014JCYJA50034).

This journal is © The Royal Society of Chemistry 2015

Notes and references

1 R. E. Blankenship, D. M. Tiede, J. Barber, G. W. Brudvig,G. Fleming, M. Ghirardi, M. R. Gunner, W. Junge,D. M. Kramer, A. Melis, T. A. Moore, C. C. Moser,D. G. Nocera, A. J. Nozik, D. R. Ort, W. W. Parson,R. C. Prince and R. T. Sayre, Science, 2011, 332, 805–809.

2 S. N. Habisreutinger, L. Schmidt-Mende and J. K. Stolarczyk,Angew. Chem., Int. Ed., 2013, 52, 7372–7408.

3 A. Fujishima and K. Honda, Nature, 1972, 238, 37–38.4 S. W. Boettcher, E. L. Warren, M. C. Putnam, E. A. Santori,D. Turner-Evans, M. D. Kelzenberg, M. G. Walter,J. R. McKone, B. S. Brunschwig, H. A. Atwater andN. S. Lewis, J. Am. Chem. Soc., 2011, 133, 1216–1219.

5 I. Oh, J. Kye and S. Hwang, Nano Lett., 2012, 12, 298–302.6 P. Dai, J. Xie, M. T. Mayer, X. Yang, J. Zhan and D. Wang,Angew. Chem., Int. Ed., 2013, 52, 11119–11123.

7 C. J. Chen, M. G. Chen, C. K. Chen, P. C. Wu, P. T. Chen,M. Basu, S. F. Hu, D. P. Tsai and R. S. Liu, Chem.Commun., 2015, 51, 549–552.

8 J. R. McKone, N. S. Lewis and H. B. Gray, Chem. Mater., 2014,26, 407–414.

9 J. R. McKone, E. L. Warren, M. J. Bierman, S. W. Boettcher,B. S. Brunschwig, N. S. Lewis and H. B. Gray, EnergyEnviron. Sci., 2011, 4, 3573–3583.

10 E. L. Warren, J. R. McKone, H. A. Atwater, H. B. Gray andN. S. Lewis, Energy Environ. Sci., 2012, 5, 9653–9661.

11 M. Basu, Z. W. Zhang, C. J. Chen, P. T. Chen, K. C. Yang,C.-G. Ma, C. C. Lin, S. F. Hu and R. S. Liu, Angew. Chem.,Int. Ed., 2015, 54, 6211–6216.

12 B. Seger, T. Pedersen, A. B. Laursen, P. C. K. Vesborg,O. Hansen and I. Chorkendorff, J. Am. Chem. Soc., 2013,135, 1057–1064.

13 H. Haick, P. T. Hurley, A. I. Hochbaum, P. Yang andN. S. Lewis, J. Am. Chem. Soc., 2006, 128, 8990–8991.

14 L. E. O'Leary, E. Johansson, B. S. Brunschwig and N. S. Lewis,J. Phys. Chem. B, 2010, 114, 14298–14302.

15 U. Sim, T.-Y. Yang, J. Moon, J. An, J. Hwang, J.-H. Seo, J. Lee,K. Y. Kim, J. Lee, S. Han, B. H. Hong and K. T. Nam, EnergyEnviron. Sci., 2013, 6, 3658–3664.

16 B. Seger, A. B. Laursen, P. C. K. Vesborg, T. Pedersen,O. Hansen, S. Dahl and I. Chorkendorff, Angew. Chem., Int.Ed., 2012, 51, 9128–9131.

17 A. B. Laursen, T. Pedersen, P. Malacrida, B. Seger,O. Hansen, P. C. K. Vesborg and I. Chorkendorff, Phys.Chem. Chem. Phys., 2013, 15, 20000–20004.

18 X. Zhang, F. Meng, S. Mao, Q. Ding, M. J. Shearer,M. S. Faber, J. Chen, R. J. Hamers and S. Jin, EnergyEnviron. Sci., 2015, 8, 862–868.

19 M. S. Faber and S. Jin, Energy Environ. Sci., 2014, 7, 3519–3542.

20 D. Kong, J. J. Cha, H. Wang, H. R. Lee and Y. Cui, EnergyEnviron. Sci., 2013, 6, 3553–3558.

21 A. Ivanovskaya, N. Singh, R.-F. Liu, H. Kreutzer,J. Baltrusaitis, N. Trung van, H. Metiu and E. McFarland,Langmuir, 2013, 29, 480–492.

J. Mater. Chem. A, 2015, 3, 23466–23476 | 23475

Journal of Materials Chemistry A Paper

Publ

ishe

d on

12

Oct

ober

201

5. D

ownl

oade

d by

Nat

iona

l Tai

wan

Uni

vers

ity o

n 15

/02/

2016

05:

35:4

9.

View Article Online

22 M. S. Faber, R. Dziedzic, M. A. Lukowski, N. S. Kaiser,Q. Ding and S. Jin, J. Am. Chem. Soc., 2014, 136, 10053–10061.

23 D. Kong, H. Wang, Z. Lu and Y. Cui, J. Am. Chem. Soc., 2014,136, 4897–4900.

24 M. S. Faber, M. A. Lukowski, Q. Ding, N. S. Kaiser and S. Jin,J. Phys. Chem. C, 2014, 118, 21347–21356.

25 Y. Sun, C. Liu, D. C. Grauer, J. Yano, J. R. Long, P. Yang andC. J. Chang, J. Am. Chem. Soc., 2013, 135, 17699–17702.

26 R. Dovesi, V. R. Saunders, C. Roetti, R. Orlando,C. M. Zicovich-Wilson, F. Pascale, B. Civalleri, K. Doll,N. M. Harrison, I. J. Bush, P. D'Arco and M. Llunell,CRYSTAL09 User's Manual, University of Torino, Torino,2009.

27 R. Demichelis, B. Civalleri, M. Ferrabone and R. Dovesi, Int.J. Quantum Chem., 2010, 110, 406–415.

28 R. Dovesi, F. F. Fava, C. Roetti and V. R. Saunders, FaradayDiscuss., 1997, 106, 173–187.

29 T. Homann, U. Hotje, M. Binnewies, A. Borger, K. D. Beckerand T. Bredow, Solid State Sci., 2006, 8, 44–49.

30 C. G. Ma and M. G. Brik, Comput. Mater. Sci., 2012, 58, 101–112.

31 H. J. Monkhorst and J. D. Pack, Phys. Rev. B: Solid State, 1976,13, 5188–5192.

32 S. G. Lyapin, A. N. Utyuzh, A. E. Petrova, A. P. Novikov,T. A. Lograsso and S. M. Stishov, J. Phys.: Condens. Matter,2014, 26, 396001.

33 J. D. Benck, T. R. Hellstern, J. Kibsgaard, P. Chakthranontand T. F. Jaramillo, ACS Catal., 2014, 4, 3957–3971.

34 Z. Huang, Z. Chen, Z. Chen, C. Lv, H. Meng and C. Zhang,ACS Nano, 2014, 8, 8121–8129.

35 A. Kargar, K. Sun, Y. Jing, C. Choi, H. Jeong, G. Y. Jung, S. Jinand D. Wang, ACS Nano, 2013, 7, 9407–9415.

23476 | J. Mater. Chem. A, 2015, 3, 23466–23476

36 A. Kargar, K. Sun, Y. Jing, C. Choi, H. Jeong, Y. Zhou,K. Madsen, P. Naughton, S. Jin, G. Y. Jung and D. Wang,Nano Lett., 2013, 13, 3017–3022.

37 J. Li and N. Wu, Catal. Sci. Technol., 2015, 5, 1360–1384.38 H. M. Chen, C. K. Chen, C. J. Chen, L. C. Cheng, P. C. Wu,

B. H. Cheng, Y. Z. Ho, M. L. Tseng, Y. Y. Hsu, T. S. Chan,J. F. Lee, R. S. Liu and D. P. Tsai, ACS Nano, 2012, 6, 7362–7372.

39 Q. Ding, F. Meng, C. R. English, M. Caban-Acevedo,M. J. Shearer, D. Liang, A. S. Daniel, R. J. Hamers andS. Jin, J. Am. Chem. Soc., 2014, 136, 8504–8507.

40 D.-Y. Wang, M. Gong, H.-L. Chou, C.-J. Pan, H.-A. Chen,Y. Wu, M.-C. Lin, M. Guan, J. Yang, C.-W. Chen,Y.-L. Wang, B.-J. Hwang, C.-C. Chen and H. Dai, J. Am.Chem. Soc., 2015, 137, 1587–1592.

41 X. Guo, X. Tong, Y. Wang, C. Chen, G. Jin and X.-Y. Guo, J.Mater. Chem. A, 2013, 1, 4657–4661.

42 V. N. Antonov, O. V. Andryushchenko, A. P. Shpak,A. N. Yaresko and O. Jepsen, Phys. Rev. B, 2008, 78, 094409.

43 Y. Xu and M. A. A. Schoonen, Am. Mineral., 2000, 85, 543–556.

44 I. Kentaro, A. Tomoki and K. Hikaru, Mater. Res. Express,2015, 2, 075901.

45 B. Seger, A. B. Laursen, P. C. K. Vesborg, T. Pedersen,O. Hansen, S. Dahl and I. Chorkendorff, Angew. Chem., Int.Ed., 2012, 51, 9128–9131.

46 A. B. Laursen, T. Pedersen, P. Malacrida, B. Seger,O. Hansen, P. C. K. Vesborg and I. Chorkendorff, Phys.Chem. Chem. Phys., 2013, 15, 20000–20004.

47 L. Zhang, C. Liu, A. Wong, J. Resasco and P. Yang, Nano Res.,2015, 8, 281–287.

This journal is © The Royal Society of Chemistry 2015