Supporting Information

Copyright WILEY-VCH Verlag GmbH & Co. KGaA, 69469 Weinheim, Germany, 2015.

Supporting Information for Adv. Mater., DOI: 10.1002/adma.201503906

Selenide-Based Electrocatalysts and Scaffolds for Water Oxidation Applications Chuan Xia, Qiu Jiang, Chao Zhao, Mohamed N. Hedhili, and Husam N. Alshareef*

DOI: 10.1002/adma.201503906 Article type: Communication

Selenide-Based Electrocatalysts and Scaffolds for Water Oxidation Applications Chuan Xia, Qiu Jiang, Chao Zhao, Mohamed N. Hedhili, Husam N. Alshareef*

C. Xia, Q. Jiang, Dr. C. Zhao, Dr. M. N. Hedhili, Prof. H. N. Alshareef Materials Science and Engineering, King Abdullah University of Science and Technology (KAUST), Thuwal 23955-6900, Saudi Arabia *E-mail: [email protected]

Keywords: selenides; oxygen evolution; metallic scaffolds, hybrid catalyst

Experimental details Cobalt precursor synthesis. Cobalt carbonate hydroxide (cobalt precursor) nanoarrays were synthesized by the hydrothermal method. Typically, 8 mmol of Co(NO3)2·6H2O, 16 mmol NH4F and 15 mmol of urea were dissolved in 80 ml deionized water to form a clear pink solution. This solution was subsequently transferred into a Teflon-lined autoclave with a piece of carbon fabric (CFC) immersed into the solution and kept at 120 ℃ for 6 h. After cooling down to the room temperature, the precursor coated carbon fabric was obtained and washed with deionized water and ethanol. Nickel Cobalt precursor synthesis. Nickel-cobalt carbonate hydroxide (nickel cobalt precursor) nanoarrays were synthesized by the hydrothermal method. Typically, 8 mmol of Co(NO3)2·6H2O, 4 mmol Ni(NO3)2·6H2O and 15 mmol of urea were dissolved in 80 ml deionized water to form a clear pink solution. This solution was subsequently transferred into a

Teflon-lined autoclave with a piece of carbon fabric (CFC) immersed into the solution and kept at 120 ℃ for 4 h. After cooling down to the room temperature, the precursor coated carbon fabric was obtained and washed with deionized water and ethanol. Preparation of fresh 1 M NaHSe. Firstly, NaBH4 (0.16 g, ~4 mmol) was dissolved in 2 mL Argon or N2 statured deionized water in a glass bottle. Then, Se powder (0.16 g, ~2 mmol) was added into the glass bottle with protective Argon or N2 atmosphere. The glass bottle was gently shaken until the black Se powder was completely dissolved and a clear white solution was obtained according to the following equation: 4NaBH4 + 2Se + 7H2O = 2NaHSe + Na2B4O7 + 14H2. Co0.85Se and (Ni, Co)0.85Se synthesis. In a typical procedure, one piece of as-prepared precursor (Co-precursor or NiCo-precursor on CFC), 38 mL Argon or N2 statured deionized water and 2 mL of fresh prepared 1 M NaHSe solution were loaded into a Teflon lined stainless steel autoclave. The autoclave was heated at 160 ℃ for 24 h, and then cooled to room temperature naturally. The final product (~4.3 mg cm-2 for Co0.85Se, ~5 mg cm-2 (Ni, Co)0.85Se) was washed and dried in vacuum at 60 ℃. (Ni, Co)0.85Se@NiCo-LDH synthesis. The (Ni, Co)0.85Se@NiCo-LDH (~6 mg cm-2) coaxial nanoarrays were prepared by electrodeposition. The obtained (Ni, Co)0.85Se was used as the working electrode and placed in an electrochemical cell in a standard 3-electrode configuration, by using Pt wire as the counter electrode and a saturated calomel electrode (SCE) as the reference electrode.[1] The potentiostatic deposition was carried out at -1.0 V for 60 seconds. And, the electrolyte was prepared by dissolving 7.5 mmol CoCl2·6H2O and 7.5 mmol Ni(NO3)2·6H2O into 50 mL deionized water.

Structural characterization. X-ray diffraction (XRD) spectra were collected by a Bruker diffractometer (D8 Advance) with Cu Kα radiation, λ = 1.5406 Å. X-ray photoelectron spectroscopy was obtained from Kratos AXIS Ultra DLD. The morphology and microstructure of the samples were characterized by SEM (Nova Nano 630, FEI) and HRTEM (Titan 80-300 kV). Energy-dispersive spectroscopy (EDS) elemental line scans were performed on the same instrument in scanning TEM (STEM) mode. The spherical-aberration-corrected HRTEM images were obtained by FEI Titan 80–300 Cubed Super-Twin transmission electron microscope. Atomic force microscopy (AFM) images were recorded using a using a Digital Instrument MultiMode AFM with a Nanoscope 4 controller operating in tapping mode. ICP-OES test was performed on Varian 720-ES. The contact angle was imaged using Theta/Attension optical tensiometer. Electrochemical measurements. All Electrochemical tests were carried out at room temperature in three-electrode (half-cell) configurations and recorded using Bio-Logic VMP3. O2-satured 1 M KOH was used as electrolyte for all measurements. The as-prepared Co0.85Se, (Ni, Co)0.85Se and (Ni, Co)0.85Se@NiCo-LDH were directly used as working electrodes. A Pt wire and a saturated Ag/AgCl was used as counter electrode and reference electrode, respectively. Before electrochemical measurements, all the samples are firstly stabilized at 0.55 V vs. saturated Ag/AgCl for 10 minutes. Linear sweep voltammograms (LSV) were measured from 0 V to 0.8 V vs. saturated Ag/AgCl at a scan rate of 0.5 mV s-1. Electrical impedance spectroscopy was recorded under the following conditions: ac voltage amplitude 5 mV, frequency ranges 10 5 to 0.1 Hz and open circuit. Chronopotentiometric measurments were performed at current density of 10 mA cm-2. The potentials in this work were converted to a reversible hydrogen

electrode (RHE) scale according to the Nernst equation (ERHE = EAg/AgCl + 0.059 pH + 0.197); the overpotential (η) was calculated according to the following formula: η (V) = ERHE -1.23 V.

Supplementary notes 1. IR correction of LSV data To fairly and accurately compare the catalytic activity of different nanostructured metal selenides, the polarization curves were corrected for all ohmic losses throughout the system. Firstly, the equivalent series resistance (Rs) can be obtained from the EIS Nyquist plot as the first intercept of the main arc with the real axis. Afterwards, the ohmic losses can be calculated and extracted from the raw polarization curves using Ohm’s Law, further giving the IR-corrected data. The values of Rs in the various measurements are low and very similar to one another, which helps ensure that no one data set has been over- or under-corrected relative to others.[2, 3]

LSV polarization curves of NiCo-LDH decorated nickel cobalt selenides nanoarrays with or without IR correction.

2. Determination of mass activity The values of mass activity (A g-1) were calculated from the catalyst loading m (mg cm-2) and the measured current density J (mA cm-2) at η=300 mV vs RHE. Mass activity = J/m

3. Electrical transport property measurement of as-prepared selenides Firstly, the selenides powders were prepared by the same procedure as described above without the addition of CFC substrate into the hydrothermal system. Then, the (Ni, Co)0.85Se and Co0.85Se powders were cold-pressed (hydraulically) into pellets of 3×10×0.5 mm size with a pressure of 1161 MPa for transport measurements by a physical property measurement system (PPMS-9, Quantum Design). In order to gain a more stable resistance signal for the metallic samples, the standard ac lock-in method was used. 4. Estimation of electrochemical active surface area (ECSA) and Roughness factor (Rf) The ECSA was determined by measuring the capacitive current associated with doublelayer charging from the scan-rate dependence of cyclic voltammetry (CV). By plotting the difference in current density (J) between anodic and cathodic sweeps (ΔJ) at a fixed potential against the scan rate, a linear trend is observed. The fitting slope is twice of the double-layer capacitance (Cdl), which is linearly promotional to the ECSA. These values of Cdl permit comparison of relative surface activity of different electrodes particularly in same electrolyte. To determine Cdl for various electrodes, the potential window of CVs was 0.05 V-0.15V vs. saturated Ag/AgCl with the scan rates from 10 mV/s - 120 mV/s. The corresponding Rf are calculated from the following equation[4]: Rf = Cdl/60 µF cm-2

CV curves and linear fit of current densities at 0.1 V versus scan rates of (A, B) pristine carbon cloth, (C, D) cobalt selenides nanoarrays, (E, F) nickel cobalt selenides nanoarrays and (G, H) Ni-Co LDH decorated nickel cobalt selenides nanoarrays.

5. Typical XRD patterns of as-prepared Co0.85Se and the (Ni, Co)0.85Se@NiCo-LDH nanostructures

The representative XRD pattern for as-prepared (A) (Ni, Co)0.85Se nanotube arrays on CFC, and (B) hybrid (Ni, Co)0.85Se@NiCo-LDH nanostructure.

6. The elements mapping of the (Ni, Co)0.85Se nanostructures As we mentioned in the manuscript, this (Ni, Co)0.85Se was prepared via in situ selenization of our NiCo-precursor. The advantage of the one-step direct transformation insures elements are distributed evenly. To check this, EDS (energy-dispersive spectroscopy) and EFTEM (energy-filtered transmission electron microscopy) analysis were employed. As shown below, the EDS data (A-D) clearly demonstrate the uniform distribution of all the elements on large-scale. Moreover, even down to nanoscale (E-H), individual (Ni, Co)0.85Se nanotube still shows even contribution of Ni, Co and Se. Thus, we can conclude here that uniformly-doped and well-crystallized ternary (Ni, Co)0.85Se nanotube arrays have been successfully prepared.

EDS and EFTEM elemental analysis of as-prepared (Ni, Co)0.85Se.

7. SEM images of the catalysts after the electrolysis process As

a

proof-of-concept,

the

morphologies

of

the

(Ni,Co)0.85Se

and

the

(Ni,Co)0.85Se@NiCo-LDH nanostructures were further examined after 24h continuous electrocatalysis. While some microstructural degradation of our dense (Ni, Co)0.85Se nanoarrys has taken place, the original morphology of our selenides still can be distinguished (A-B). In contrast to (Ni, Co)0.85Se, the hybrid (Ni,Co)0.85Se@NiCo–LDH electrode shows a much better capability to retain its original morphology (C-D) in spite of the partial pulverization of some nanotubes. This is believed to come from the outer NiCo-LDH shell. During the vigorous O2 evolution reaction, such an ultrathin shell could be a buffer layer to prevent (at least slow down) the degradation of the active selenides. Also, this NiCo-LDH could play the role of a “glue” to make individual nanotubes more physically stable. From our observation, we believe that our electrodes show very reasonable electrochemical and physical stability at this stage of development. Moreover, the (Ni, Co)0.85Se nanoarrays act as good scaffolds for other materials to achieve more physically stable nanostructures and better electrochemical performance.

SEM images of (A-B) (Ni, Co)0.85Se and (C-D) hybrid (Ni, Co)0.85Se@NiCo-LDH nanostructures after 24h cycling in electrolyte.

8. Estimation of TOF Here, the TOF (turnover frequency) was calculated by assuming that every metal atom is involved in the catalysis (lower TOF limits were estimated) by the following equation: TOF =

𝐽∗𝐴 4∗𝐹∗𝑛

Where J is the measured current density at η=300 mV, A is the surface area of our electrode, F is the Faraday constant, and n is the moles of the metal atom on the electrode. The calculated TOFs for Co0.85Se and (Ni, Co)0.85Se are 0.0012 s-1 and 0.003 s-1, respectively. Obviously, the Ni doped sample demonstrates a significantly enhanced performance. While the estimated TOFs of our selenides electrode are not very high, such a TOF number are still comparable with previously reported Ni or Co based electrocatalysts (ref. 5). Note that our calculations are based on the high mass-loading electrodes and assuming that every metal atom is involved in the catalysis. Hence, we believe that the “actual TOF” should be much higher than this lower limit value. 9. Digital photographs of operating electrode

Digital photographs of operating nickel cobalt selenide electrode evolving O2(g) at (A) onset point and (B) η =400 mV, respectively.

Supporting figures

Figure S1. Microstructure of nickel cobalt selenide. SEM and TEM images of the morphology of (Ni, Co)0.85Se nanotube arrays.

Figure S2. Electrochemical performance of catalysts. The LSV and electrochemical impedance spectroscopy (EIS) plots of Co0.85Se, (Ni, Co)0.85Se and (Ni, Co)0.85Se@NiCo-LDH.

Figure S3. Characterization of surface properties. Contact angle of CFC and (Ni, Co)0.85Se.

In order to fairly compare the different surface nature of our selenides and CFC substrate, we performed the following experiment. We, firstly, choose a large piece of CFC and only grow the selenides on half of that CFC. As we can see above, the darker region is (Ni, Co)0.85Se coated CFC. Next, we directly measure the contact angle at the left region (without selenides) and right region (with selenides) to identify their different surface characters. Apparently, the left part (CFC) shows a superhydrophobic nature with a contact angle of 161°, whereas the selenides coated substrate demonstrates a superhydrophilic surface. More specifically, the right part can easily absorb the droplet. More details can be found at the supporting videos.

Figure S4. Structural analysis of (Ni, Co)0.85Se. HRTEM image of the surface of (Ni, Co)0.85Se nanotube, indicating the formation small angle grain boundaries while the large nanocrystal appears essentially flat.

Figure S5. TEM of bare Co0.85Se. The aberration-corrected HRTEM of Co0.85Se at or near the surface

These images clearly show our Co0.85Se own an analogous surface with (Ni, Co)0.85Se, consisting of nanocrystals with amorphous boundaries. While the defects and stepped surface are still observed in Co0.85Se, it is more difficult to find some obviously faulted grain, especially at the surface. The above typical images demonstrate a relatively more perfect surface compared with (Ni, Co)0.85Se. Hence, we believe, from our observations, the Ni-doped sample exhibits much higher bulk defects than the pristine Co0.85Se.

Figure S6. XPS analysis of nickel cobalt selenide. The XPS survey spectra of (Ni, Co)0.85Se.

Figure S7. Structural characterization and EDS analysis of (Ni, Co)0.85Se. (A-B) TEM and (C-D) high angle annular dark field (HAADF) STEM images of (Ni, Co)0.85Se nanotube with corresponding (E) {Yang, 2014 #315} Se and O.

In the detailed structural study at the surface of (Ni, Co)0.85Se, sometimes we noticed there was an amorphous thin layer (~1 nm) contaminated at the surface of our selenides nanotube. From the HAADF-STEM elemental line scan analysis, we observed that the Se and O were dominated at such a thin region, indicating the possibility of formation of amorphous SeO2 (or SeOx) during the hydrothermal process. Of note, from the HAADF-STEM image, the nanotube morphology could be further confirmed. Combination of the XRD and XPS data, here we can conclude that the formation of amorphous SeO2 (or SeOx) is indeed happened.

Figure S8. XPS analysis of Co0.85Se. High-resolution XPS spectra for (A) Co 2p and (B) Se 3d and peak fitting analysis of Co0.85Se.

Co0.85Se shows a similar Co 2p spectra with (Ni, Co)0.85Se, but the Se 3d behavior is completely different. To be specific, the Se 3d spectra of Co0.85Se can be well fitted into two peaks, namely 3d3/2 and 3d5/2. Here, the doublet separation energy (3d3/2 - 3d5/2) and the corresponding peak area ratio (3d3/2 / 3d5/2) of Co0.85Se are around 0.9 eV and 0.6667, respectively. Such a result perfectly matches with the standard conclusion[5], suggesting the formation of only one kind of metal-Se bond (Co-Se). However, in (Ni, Co)0.85Se (Figure 4C), the Se 3d need to be deconvoluted into four sub-bands in order to obtain a reasonable fitting analysis. That result indicates two kinds of metal-Se bond are presenting in (Ni, Co)0.85Se, strongly supporting the introduction of Ni into Co0.85Se lattice.

Figure S9. Characterization of (Ni, Co)0.85Se@NiCo-LDH. TEM images of (Ni, Co)0.85Se@NiCo-LDH and NiCo-LDH; and the tapping-mode AFM topographical images of NiCo-LDH with corresponding height profiles deposited on a silicon substrate.

From the TEM images, we can confirm that the (Ni, Co)0.85Se was decorated by the ultrathin curved mesoporus nanosheets. The HRTEM image suggested that the layered nanosheets owned an interspacing of ~0.5 nm, agreement with the reported interlayer distance of NiCo-LDH (0.5-0.6 nm).[2, 6] Atomic force microscopy (AFM) was performed to determine the height of the nanosheets. Apparent, the AFM measurements reveal a various height of single

nanosheet, originating from its curved morphology. The flat part shows a thickness of 2-4 nm of the NiCo-LDH, confirming that such a layered nanosheet is made of only few atomic layers (4-8 layers).

Supporting Table Table S1. OER activities of some benchmark catalysts in alkaline solution

1 M KOH 1 M KOH

η @ 10 mA cm-2 (mV) 324 255

j @ 300 mV (mA cm-2) 5.3 36.4

Tafel slope (mV dec-1) 85 79

1 M KOH

216

97.5

77

This work

1 M KOH 0.1 M KOH 1 M KOH

427 450 258

49 49 33.6

S[7] 39 S[8]

1 M NaOH

285

35

S[9]

1 M KOH

/

0.064 ~2 ~30 80 ( 315 mV) 100 (315 mV)

54

S[10]

0.1 M KOH

460

/

65

11

0.1 M KOH

520

/

/

S[11]

1 M KOH

460

/

90

29

1 M KOH

320

~5

51

35

1 M KOH

367

0.683

40

38

0.1 M KOH

400

~4

156

16

1 M KOH

565

/

292

25

1 M KOH 0.1 M KOH

840 331

/ /

312 42

25 6

1 M KOH

310

~9.5

40

2

Material

Electrolyte

Co0.85Se (Ni, Co)0.85Se (Ni,Co)0.85Se @NiCo-LDH IrOx Mn3O4/CoSe2 De-LNiFeP/rGO Iron-Doped Nickel Oxide Core-ring NiCo2O4 Ni@NiCohydroxides 3D NF/PC/AN NiCo2O4 nanoarrays ZnxCo3-xO nanoarrays Exfoliated NiCoLDH Graphene/NiCo2O4 hybrid paper NiCo2O4 nanoneedles CoPi on ITO α-Ni(OH)2 Exfoliated NiFeNs

Reference This work This work

Supporting references [1] M. Shao, Z. Li, R. Zhang, F. Ning, M. Wei, D. G. Evans, X. Duan, Small 2015. [2] H. Liang, F. Meng, M. Cabán-Acevedo, L. Li, A. Forticaux, L. Xiu, Z. Wang, S. Jin, Nano Lett 2015, 15, 1421. [3] M. S. Faber, R. Dziedzic, M. A. Lukowski, N. S. Kaiser, Q. Ding, S. Jin, J Am Chem Soc 2014, 136, 10053. [4] Y. Li, P. Hasin, Y. Wu, Adv Mater 2010, 22, 1926. [5] W.-z. Chen, A. Kahn, P. Soukiassian, P. Mangat, J. Gaines, C. Ponzoni, D. Olego, Phys Rev B 1994, 49, 10790; J. Moulder, W. Stickle, P. Sobol, K. Bomben, There is no corresponding record for this reference. [6] F. Song, X. Hu, Nat Commun 2014, 5. [7] L. Trotochaud, J. K. Ranney, K. N. Williams, S. W. Boettcher, J Am Chem Soc 2012, 134, 17253. [8] Y. Liu, H. Wang, D. Lin, C. Liu, P.-C. Hsu, W. Liu, W. Chen, Y. Cui, Energ Environ Sci 2015. [9] E. L. Miller, R. E. Rocheleau, J Electrochem Soc 1997, 144, 3072. [10] B. Cui, H. Lin, J. B. Li, X. Li, J. Yang, J. Tao, Adv Funct Mater 2008, 18, 1440. [11] J. Wang, H. x. Zhong, Y. l. Qin, X. b. Zhang, Angewandte Chemie International Edition 2013, 52, 5248.