Aenm202103750 Sup 0001 Suppmat

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 28

Supporting Information

for Adv. Energy Mater., DOI: 10.1002/aenm.202103750

Hydrogen Evolution Linked to Selective Oxidation of


Glycerol over CoMoO4—A Theoretically Predicted
Catalyst
Xiaowen Yu,* Rafael B. Araujo, Zhen Qiu, Egon Campos
dos Santos, Athira Anil, Ann Cornell, Lars G. M.
Pettersson, and Mats Johnsson*
Supporting Information

Hydrogen evolution linked to selective oxidation of glycerol over CoMoO4 – a theoretically


predicted catalyst

Xiaowen Yu,*+ Rafael B. Araujo,+ Zhen Qiu, Egon Campos dos Santos, Athira Anil, Ann

Cornell, Lars G. M. Pettersson, and Mats Johnsson*

1
Computational details

Density functional theory (DFT) calculations were carried out using the Projected Augmented

Wave (PAW) method to solve the Kohn-Shan equations as implemented in the Vienna Ab initio

Simulation Package (VASP).[S1] The spin polarized generalized gradient approximation has

been employed with the Perdew, Burke, and Ernzerhof (PBE) parametrization to describe the

exchange and correlation term of the Kohn-Sham Hamiltonian.[S2] Plane waves were expanded

to an energy cutoff of 520 eV, while Brillouin sampling was performed in a reciprocal grid with

a space resolution of 2π x 0.04 Å–1. Heyd−Scuseria−Ernzerhof (HSE06)[S3] hybrid functional

calculations were performed with 25% of exact exchange. Structural relaxations were

conducted with energy criteria and atoms force convergence set to 10–6 eV and 0.01 eV Å–1.

Moreover, ferromagnetic ordering (FM) has been employed to model all catalysts.

Band centers (BCs) are calculated as the centroid of projected density of states (PDOS) on

O 2p orbitals and transition metal (TM) d orbitals relative to the Fermi level (herein shifted to

0 eV) and read as:


𝑒
∫𝑒 𝑚𝑎𝑥 eD(e)de
𝑚𝑖𝑛
BC = 𝑒 (1)
∫𝑒 𝑚𝑎𝑥 D(e)de
𝑚𝑖𝑛

Here, e stands for the energy of the electronic bands while D(e) is the density of states for such

energy. Therefore, the case of the O 2p band center D(e) refers to the density of states projected

on p states of oxygens, while for the d BCs, D(e) refers to the projected density of states on d

states of TM. The lower energy bound (𝑒𝑚𝑖𝑛 in eq. 1) is chosen to be –15 eV. O 2p and TM d

states usually lie above –15 eV relative to the Fermi level, thus, motivating such a choice of

𝑒𝑚𝑖𝑛 . The upper limit, 𝑒𝑚𝑎𝑥 , is vaguer. Jacobs et al.[S4] has investigated the O 2p band center

as a descriptor for properties like oxygen surface exchange rates and aqueous oxygen evolution

reaction current densities in perovskite catalysts. Better correlations were obtained when BCs

were calculated over occupied and unoccupied electronic states of the catalysts. Dickens et

al.[S5] have studied O 2p centers as electronic structure descriptors for oxygen reactivity. Lower
2
model errors were also obtained when unoccupied states were accounted to get the BCs (𝑒𝑚𝑎𝑥

= 2 eV relative to the Fermi level). Both investigations argue that unoccupied states might be

chemically relevant during the oxidative/reductive reaction since such transformations involve

electrons being transferred from/to the catalysts. Yet, it is not clear a priory what total number

of states is needed to properly describe bond formation/break during an oxidative/reductive

reaction. Here, we have set 𝑒𝑚𝑎𝑥 = 3 eV for the computation of O 2p descriptors and 𝑒𝑚𝑎𝑥 = 0

eV for (O 2p–TM d) descriptor. Moreover, a weighted averaged d band center is computed for

catalysts having more than one TM.

Catalytic Descriptors Towards GOR

This subsection discusses: i) the electronic structure of the spinel catalysts MCo2O4 (M = Mn,

Fe, Co, Ni, Cu, and Zn) (Figure S1); ii) the existing correlations between the O 2p BCs and the

TM d BCs with the intrinsic GOR catalytic activity, as reported by Han et al.[S6]; iii) how this

correlation can be used as a tool for a rational selection of oxide catalysts delivering high

activity towards GOR.

Figure S1. Crystal structure of the spinel oxides. Red depicts O and blue octahedrally and
tetrahedrally coordinated Co atoms, respectively; golden balls are octahedral sites (named M)
populated by Mn, Fe, Co, Ni, Cu, and Zn.

3
Spinel cobaltite usually crystallizes in two main phases, distinguished by their cation

distribution in their lattice structure.[S7] Liu et al. have investigated the energetics of MCo2O4

(X = Mn, Ni, Cu) employing DFT calculations,[S8] which point to the inverse spinel phase is

thermodynamically favorable compared to the normal spinel phase. Herein, the inversed spinel,

existing in a cubic structure with Fd-3m space group, will be used as the appropriate model.

The transition metal cations are tetrahedrally and octahedrally coordinated by oxygen links one

tetrahedral and three octahedral cations, see Figure S1. Due to the crystal field effect, TM d

states sited at the octahedral sites split into three low lying eg states and two t2g high lying

orbitals. Meanwhile, tetrahedral sites split d states into three high energy orbitals and two low

lying energy orbitals. Finally, the TM d orbitals interact with O 2p orbitals to form 𝜎 and 𝜋

bonding and antibonding states.

The PDOS of the family MCo2O4 (M = Mn, Fe, Co, Ni, Cu, and Zn) are calculated together

with O 2p BCs and TM d BCs at the HSE06 level of theory (Figure S2) to highlight the

transformations underwent by their electronic structure under the exchange of the TM centers.

Generally, TM d states yielded the most dominant contributions to the density of states close to

the Fermi level for the compounds MnCo2O4, FeCo2O4, and CoCo2O4. This, however, changes

for NiCo2O4, CuCo2O4, and ZnCo2O4 where O 2p states systematically increase influence on

the density of states close to the Fermi level. This directly impacts on the catalyst activity once

states close to the Fermi level might interact with intermediates during the reaction cycles.

Moreover, the gap between O 2p BCs and TM d BCs become smaller with the series MCo2O4

(M = Mn, Fe, Co, Ni, Cu, and Zn), where for Cu the value of (O 2p – TM d) gets close to 0 eV,

while for Zn becomes positive (O 2p BC is higher in energy than TM d BC). These gaps (O 2p

– TM d) are associated with the covalence of the TM–O bond (covalence can be quantified as

the distance between the O 2p BC and TM d BC where higher values indicate less covalence)[9]

that also controls the bond strength between O interacting intermediates and the metal sites of

the catalysts.
4
Figure S2. Projected density of states (PDOS) of MCo2O4 (M = Mn, Fe, Co, Ni, Cu and Zn)
calculated with HSE06 level of theory. Here, black background stands for the summed d states
of Co centers, blue background stands for the summed d states of the M metals and green lines
are the summed O 2p states. Moreover, the black line at 0 eV highlights the Fermi level, while
the red and green lines are the averaged M d band center and the O 2p band center, respectively.

A volcano shape is obtained when the intrinsic GOR catalytic activity, reported by Han et

al.[S6], is plotted as a function of (O 2p – TM d) at the HSE06 level of theory (see Figure S3).

The linear regression fit for the catalysts presenting O 2p BC smaller than the TM d BC revealed

a R2 of 0.80 and, hence, confirming the existing correlation (Figure S3b). On the very right side

of the volcano, the number of measurements does not allow us to perform a linear fitting, but

the catalytic performance decreases with the (O 2p – TM d) upwards shift.

5
Figure S3. Intrinsic GOR catalytic activity vs. (a) O 2p band centers and (b) O 2p – TM d band
centers. O 2p band centers presented in (a) are obtained by integrating electronic states lower
than 3 eV in energy. (b) Calculated (O 2p – TM d) descriptors considering only occupied
electronic states (until 0 eV, once Fermi level is set to 0 eV in the PDOS).

Despite the existing correlation between the HSE06 (O 2p – TM d) and activity, this

approach holds the drawback of being computationally time-consuming and, therefore, not

suitable for an extensive screening investigation. One strategy to circumvent the application of

the HSE06 approach is to use O 2p BCs instead of (O 2p – TM d) BCs as a GOR activity

descriptor. These states suffer minor influence from the self-interaction error inherent of GGAs

than TM d states, hence, enabling the usage of a functional like PBE without losing accuracy.

A volcano-like shape is also obtained when plotting the GOR catalytic activity vs. O 2p BCs of

the spinel oxide catalysts (Figure S3a). The obtained R2 from the liner fitting performed on the

more left side catalysts of the volcano is 0.99 and, thus, confirms that O 2p BC also acts as a

descriptor of the GOR catalytic activity for the spinel oxide catalysts.

As a design parameter, the O 2p BC was firstly proposed by Grimaud et al.[S10] where oxide

catalysts with better performance for EOA were proposed having O 2p BC positioning not too

close to and not too far from the Fermi level. They showed that, on one side, O 2p BC too far

from the Fermi level yields strong interactions between the catalytic surface and intermediates,

hence, regulating quantities such as overpotential. On the other side, O 2p BC too close to the

Fermi level decreases the stability of oxide catalysts. It has been shown that O 2p BC scales

with the formation of oxygen defects.[S11] Subsequently, Jacobs et al.[S4] have also assessed the

6
effectiveness of O 2p BC descriptors at predicting properties like oxygen reduction surface

exchange rates etc. by employing different DFT levels of theory for a set of perovskites oxide

catalysts. Surprisingly, various measures of catalytic activity for oxygen evolution reaction

(OER) and oxygen reduction reaction (ORR) were best correlated with PBE – despite the well-

known problem of the over delocalization of the TM d states.

The (O 2p – TM d) descriptor has also been employed in several investigations, e.g., Sun et

al.[S12] have demonstrated that the water oxidation mechanism is regulated by the (O 2p – TM

d) descriptor of the spinel oxide catalysts. They claimed that high catalytic activity might be

obtained for materials presenting an intermedia value of (O 2p –TM d). Very low values of (O

2p – TM d) lead to a strong TM–O bond, hence, preventing the formation of active sites on the

catalytic surface. In other words, the active cation centers are coved by O atoms at the catalytic

surface. On the contrary, high values of (O 2p – TM d) results in polar TM–O bonds; thus, bond

breakage leads to ionic species that hardly contribute to the oxidative/reductive reaction. When

TM–O is not so polar and not so covalent, bond breakage would exist, and it might expose

metals with unpaired electrons that are more prone to form active sites for the oxidative cycle.

Wang et al.[S13] have investigated the spinel catalysts ZnNixCo2-xO4 (x = 0–0.8) for methane

oxidation. The distance between O 2p and TM d states of the octahedrally coordinated sites is

also the driven descriptor for the selected oxidation mechanism towards the methane oxidation.

They argued that for low energy lying O 2p BC (farther from the Fermi level than TM d states),

the higher metal character of the states on the Fermi level indicates a mechanism with the metal

cations as the adsorption center. In contrast, O 2p BC, closer to the Fermi level (TM d states

lying lower energy than the O 2p states), leads to an enhanced oxygen lattice participation in

the reaction mechanism. Moreover, for the latter cases, the thermodynamic stability of the

catalytic surface is destabilized with possible surface reorganization due to a higher probability

of oxygen defect formation.[S14]

7
Here, a volcano shaped plot is obtained when the intrinsic GOR catalytic activity is plotted

as a function of the (O2p – TM d) and/or O 2p descriptors, hence, indicating that the learned

relations discussed above for OER and ORR could serve as inspiration to understand the case

of GOR on oxide catalysts where: i) for catalysts with lowest (O 2p – TM d) or O 2p values,

the strong bond strength between intermediates interacting with the catalytic surface via an O

atom is expected due to that the high covalence in TM–O bond might produce higher

thermodynamic barriers in the oxidation/reduction reaction, and thus affecting the delivered

activity by enhancing the reaction overpotential and current. Moreover, stronger TM–O bond

strength leads to fewer TM centers exposed at the catalytic surface, hence, lowering the number

of active sites for the oxidative reaction. ii) For catalysts with higher (O 2p – TM d) or O 2p

values, it is likely that the catalytic surface undergoes a structural rearrangement because of

their lower thermodynamic stability as compared to the cases with lower (O 2p – TM d) values,

and thus changing the activity of the relevant catalyst. Moreover, similarly to the case of

methane oxidation,[S13] the GOR reaction mechanism could be affected by lattice oxygen

participating processes, therefore hindering the catalyst’s delivered activity.

An effective strategy to seek efficient oxide catalysts is to compute O 2p BCs and further

use it to filter the values that are not too low or too high. The limits of “low” and “high” of O

2p BCs can be selected based on the results from the above spinel oxide catalysts. Moreover,

the O 2p BCs can be computed using the PBE approach since O 2p states are less affected by

the self-interaction error than TM d metal states. This ensures the screening’s effectiveness

compared to a computationally time-consuming hybrid functional approach (e.g., HSE06).

Subsequently, the selected catalysts are tested for the relation (O 2p – TM d), aiming to confirm

that their TM d BCs are closer to the Fermi level than their O 2p BCs; where for these cases,

the HSE06 approach is employed. Especially, one could select cases where the calculated TM

d BCs are closer to the Fermi level – a negative value for (O 2p – TM d), but still not too

negative to ensure a balanced level of covalence.


8
Chemicals and materials

Cobalt chloride hexahydrate (CoCl2•6H2O, Alfa Aesar, 98%), nickel chloride hexahydrate

(NiCl2•6H2O, Alfa Aesar, 99.95%), manganese chloride, anhydrous (MnCl2, Alfa Aesar, 97%),

sodium molybdate dihydrate (NaMoO4•2H2O, Sigma-Aldrich, ≥ 99.5%), hydrochloric acid

(HCl, Honeywell, ACS reagent, ≥ 99.5%), ethanol (absolute, VWR chemicals, ≥ 99.5%),

glycerol (bidistilled, VWR chemicals, ≥ 99.5%), dihydroxyacetone (DHA, AmBeed, ≥ 98%),

DL-glyceraldehyde (Sigma-Aldrich, ≥ 90%, GC grade), DL-glyceric acid (TCI, 20% in water,

2 M), sodium mesoxalate monohydrate (Sigma-Aldrich, ≥ 98%), glyoxylic acid monohydrate

(Sigma-Aldrich, 98%), formaldehyde solution (Sigma-Aldrich, 36.5–38%% in H2O), tartronic

acid (Sigma-Aldrich, ≥ 97%), glycolic acid (Sigma-Aldrich, ≥ 99%), oxalic acid (Sigma-

Aldrich, ≥ 98%), formic acid (Sigma-Aldrich, reagent grade, ≥ 95%), potassium hydroxide

(KOH, VWR chemicals) were purchased and used without further purification. The nickel foam

was cleaned ultrasonically in diluted HCl (3 M) for 15 min, then washed several times in water,

and finally allowed to dry naturally before use. Ultrapure water (resistivity ˃ 18.2 MΩ cm at

25 oC) from Millipore (Simplicity water purification system) was used to prepare the

electrolytes.

Characterization

Scanning electron microscopy (SEM) images were obtained using a SEM (JEOL JSM-7000F)

equipped with an energy dispersive spectrometer (EDS). Transmission electron microscopy

(TEM) images were recorded on a JEOL JEM-2100F TEM. Powder X-ray diffraction (PXRD)

was carried out on a PANalytical X’Pert Pro diffractometer using Cu Kα radiation. X-ray

photoelectron spectra were performed on a Physical Electronics Quantera II Scanning XPS

Microprobe instrument with Al Kα as the radiation source, 100 μm spot size and an operating

power of 100 W. The base pressure of the instrumental chamber was maintained at 7×10–7 bar.

Survey spectra were obtained with a pass energy of 224 eV and step size of 0.1 eV, whereas

9
high-resolution XPS spectra were carried out with a pass energy of 55 eV and resolution of

0.05. The XPS peaks were fitted by using the XPS-PEAK software. Raman spectra were

recorded on the LabRAM HR 800 Raman spectrometer with a laser wavelength of 532 nm.

Experimental Section

Preparation of CoMoO4 catalyst on nickel foam: The CoMoO4 nanorod catalyst was grown in

situ on the nickel foam by a facile hydrothermal reaction followed by annealing at 500 oC. In a

typical procedure, a piece of nickel foam (2 × 3 cm) was cleaned by a diluted HCl solution (3

M) under ultrasonication for 15 min to remove the surface oxide layer. Then the nickel foam

was washed by deionized water several times until the pH of detergent is 7. The cleaned nickel

foam was finally dried in a vacuum oven for further use. Two solutions were prepared with

dissolving 0.75 mmol CoCl2•6H2O and Na2MoO4•2H2O in 7.5 mL ultrapure water.

Successively, the Na2MoO4•2H2O solution was added dropwise to the CoCl2•6H2O solution

and kept stirring for 30 min. The mixture was then transferred into a 20 mL Teflon lining with

a cleaned nickel foam placed against the wall, which was finally sealed in a stainless steel

autoclave and heated at 160 oC for 12 h. After the autoclave was cooled down naturally to room

temperature, the resultant CoMoO4•xH2O/NF was ultrasonically washed by deionized water

three times to remove residual reactants and then dried in a freeze dryer overnight. Successively,

the above CoMoO4•xH2O/NF was calcinated in a furnace at 500 oC for 2 h; the heating and

cooling rate was 2 and 3 oC min–1, respectively. The product finally obtained is CoMoO4/NF.

The analogous catalyst compounds NiMoO4 and MnMoO4 on nickel foams were prepared by a

similar method as CoMoO4, except for changing the CoCl2•6H2O precursor to NiCl2•6H2O and

MnCl2, respectively. The powders obtained at the same condition without placing nickel foam

in the hydrothermal reaction were also collected for PXRD, SEM, and TEM characterization.

Electrochemical measurement: All the GOR tests were conducted at a three-electrode system

by using an SP-50 potentiostat (Biologic, France). A divided cell was adopted to avoid influence

10
of the cathode reaction. The anode chamber included a CoMoO4/NF electrode (1 × 1 cm2) and

a Hg/HgO electrode as the working and reference electrode respectively, which were immersed

in a 10 mL electrolyte of 1.0 M KOH and 0.1 M glycerol with Ar bubbling during the test. The

cathode chamber included a graphite rod electrode as the counter electrode with a 10 mL

electrolyte of 1.0 M KOH. The anode and cathode chambers are separated by Nafion 117

membrane and fixed by a stainless steel clip. All electrodes were first applied for 100 CV cycles

at 100 mV s–1 in pure 1.0 M KOH solution to reach a stable state before transferring to solutions

with glycerol. The scan rate of LSV curves for GOR is 1 mV s–1 to minimize the capacitive

current. The current density was normalized by geometric surface area (GEO) and

electrochemical surface area (ECSA), respectively. ECSA was calculated by evaluating the

double layer capacitance in a non-faradaic region. Products of GOR were analyzed by HPLC

(Agilent 1260 Infinity II isocratic pump, multisampler and multicolumn thermostat with a 1290

Infinity II refractive index detector). The analytical columns, in series, included a Bio-Rad

guard column with a standard cartridge holder with Micro-Guard cation H+ cartridge (4.6 × 30

mm), a Bio-Rad Aminex HPX-87H column (7.8 × 300 mm), and Shodex Sugar SH1011 column

(8 × 300 mm). All columns in series were kept at 30 oC and the eluent was 8 mM H2SO4 at a

flow rate of 0.25 mL min–1. Typically, 100 μL of the product solution was collected and

neutralized with 100 μL of 0.5 M H2SO4 for HPLC analysis. OER tests were conducted at the

same system as GOR except for the anode chamber fed with 1.0 M KOH without glycerol. All

potentials recorded from the electrochemical tests were calibrated in relation to the reversible

hydrogen electrode (ERHE) by using the equation: ERHE = EHg/HgO + 0.933 V.

In situ Raman measurement: All Raman spectra were conducted on a confocal Raman

microscope (RM H 1800). In situ Raman experiments were performed in a custom-made quartz

electrochemical cell with a three-electrode setup controlled by an SP-50 potentiostat (Biologic,

France). The nickel foam supported catalyst, a graphite rod, and a Hg/HgO (1.0 M KOH)

electrode were used as the working electrode, counter electrode, and reference electrode,
11
respectively. In situ Raman spectra in Figure 6b-d were collected by applying

chronoamperometry at selected potentials with focusing on the working electrode surface by a

long-distance 10× objective (Olympus, equipment with a cage right-angle mount). The

acquisition time for each spectrum was 10 s with 30 sweeps from 200 to 1800 cm–1. The 532

nm laser with an ND filler of 100 % (laser power 56 mW) was used to generate the Raman

spectra. Raman spectra of the electrolytes plotted in Figure S17 were collected by focusing on

the electrolyte and followed the same acquisition conditions as described above. The schematic

diagram of the in situ Raman setup is shown in Figure 6a.

Product analysis:

i) The selectivity of a specific product was described as the following equation:

𝑐𝑜𝑛𝑐𝑒𝑛𝑡𝑟𝑎𝑡𝑖𝑜𝑛 𝑜𝑓 𝑡ℎ𝑒 𝑠𝑝𝑒𝑐𝑖𝑓𝑖𝑐 𝑝𝑟𝑜𝑑𝑢𝑐𝑡


𝑠𝑒𝑙𝑒𝑐𝑡𝑖𝑣𝑖𝑡𝑦 (%) = × 100%
𝑡𝑜𝑡𝑎𝑙 𝑐𝑜𝑛𝑐𝑒𝑛𝑡𝑟𝑎𝑡𝑖𝑜𝑛 𝑜𝑓 𝑎𝑙𝑙 𝑝𝑟𝑜𝑑𝑢𝑐𝑡𝑠

ii) Carbon balance is an index to evaluate the difference between the amount of carbon

produced as products and carbon consumed from glycerol, which is calculated by the following

equation:

2 1
𝐶3𝑝𝑟𝑜𝑑𝑢𝑐𝑡 + 3 × 𝐶2𝑝𝑟𝑜𝑑𝑢𝑐𝑡 + 3 × 𝐶1𝑝𝑟𝑜𝑑𝑢𝑐𝑡
𝐶𝐵 (%) = × 100%
𝐶𝑖𝐺𝑙𝑦 − 𝐶𝑓𝐺𝑙𝑦

Where, CiGly and CfGly indicated the concentration of initial glycerol and concentration of final

glycerol, respectively; C3product represented the concentration of three-carbon product

(including glyceraldehyde, dihydroxyacetone, glycerate, and tartronate); similarly, C2product

represented the concentration of two-carbon product (including glycolate, oxalate, and

glyoxylate); C1product represented the concentration of one-carbon product (including formate

and formaldehyde).

12
iii) Faradaic efficiency of GOR

The half-reaction formula for each product from GOR is listed as follows:

Products Anode reaction formula e– (mol)

Glycerate CH2OH-CHOH-CH2OH + 5OH– → CH2OH-CHOH-COO– + 4H2O + 4e– 4

Tartronate CH2OH-CHOH-CH2OH + 10OH– → –OOC-CHOH-COO– + 8H2O + 8e– 8

Glyoxylate CH2OH-CHOH-CH2OH + 19/2OH– → 3/2CHO-COO– + 8H2O + 8e– 16/3

Glycolate CH2OH-CHOH-CH2OH + 13/2OH– → 3/2CH2OH-COO– + 5H2O + 5e– 10/3

Oxalate CH2OH-CHOH-CH2OH + 14OH– → 3/2–OOC-COO– + 11H2O + 11e– 22/3

Formate CH2OH-CHOH-CH2OH + 11OH– → 3CHOO– + 8H2O + 8e– 8/3

Formaldedyde CH2OH-CHOH-CH2OH + 2OH– → 3CH2O + 2H2O + 2e– 2/3

Faradaic efficiency (FE) of GOR is calculated based on the following equation:

10 16 22 5
4 × 𝐶𝑔𝑙𝑦𝑐𝑒 + 8 × 𝐶𝑡𝑎𝑟𝑡 + × 𝐶𝑔𝑙𝑦𝑐𝑜 + × 𝐶𝑔𝑙𝑦𝑜 + × 𝐶𝑜𝑥𝑎 + 8 × 𝐶𝑓𝑜𝑟𝑚𝑎𝑡𝑒 + × 𝐶𝑓𝑜𝑟𝑚𝑎𝑙𝑑𝑒ℎ𝑦𝑑𝑒
𝐹𝐸(%) = 3 3 3 3 × 𝑉 × 𝐹 × 100%
𝑄

Where, Cglyce, Ctart, Cglyco, Cglyo, Coxa, Cformate, and Cformaldehyde are the concentration (mol L–1) of

glycerate, tartronate, glycolate, glyoxylate, oxalate, formate, and formaldehyde; V is the volume

of electrolyte (10 × 10–3 L); F is the Faradaic constant (96485 C mol–1); Q is the total charge

(C) passed during electrochemical reaction.

iv) Faradaic efficiency of the corresponding HER

The full reaction formula for each product from GOR is listed as follows:

Products Full reaction formula H2 (mol)

Glycerate CH2OH-CHOH-CH2OH + OH– → CH2OH-CHOH-COO– + 2H2 2

Tartronate CH2OH-CHOH-CH2OH + 2OH– → –OOC-CHOH-COO– + 4H2 4

Glyoxylate CH2OH-CHOH-CH2OH + 3/2OH– → 3/2CHO-COO– + 4H2 8/3

Glycolate CH2OH-CHOH-CH2OH + 3/2OH– → 3/2CH2OH-COO– + 5/2H2 5/3

13
Oxalate CH2OH-CHOH-CH2OH + 3OH– → 3/2–OOC-COO– + 11/2H2 11/3

Formate CH2OH-CHOH-CH2OH + 3OH– → 3CHOO– + 4H2 4/3

Formaldedyde CH2OH-CHOH-CH2OH → 3CHO + 5/2H2 5/6

FE of the corresponding HER is calculated based on the following equation:

𝑃𝑉𝐻2
𝑅𝑇 × 𝑉𝑒𝑙𝑒𝑐𝑡𝑟𝑜𝑙𝑦𝑡𝑒
𝐹𝐸(%) = × 100%
5 8 11 4 5
2 × 𝐶𝑔𝑙𝑦𝑐𝑒 + 4 × 𝐶𝑡𝑎𝑟𝑡 + × 𝐶𝑔𝑙𝑦𝑐𝑜 + × 𝐶𝑔𝑙𝑦𝑜 + × 𝐶𝑜𝑥𝑎 + × 𝐶𝑓𝑜𝑟𝑚𝑎𝑡𝑒 + × 𝐶𝑓𝑜𝑟𝑚𝑎𝑙𝑑𝑒ℎ𝑦𝑑𝑒
3 3 3 3 6

Where, P is the pressure (101 × 103 Pa  1 atm); VH2 is the measured H2 gas produced from the

cathode chamber, which is evaluated by gas displacement (see Figure 7a); R is the ideal gas

constant (8.314 Pa m3 K–1 mol–1); T is the temperature (298 K at room temperature); Velectrolyte

is the volume of electrolyte solution (10 × 10–3 L); Cglyce, Ctart, Cglyco, Cglyo, Coxa, Cformate, and

Cformaldehyde are the concentration (mol L–1) of glycerate, tartronate, glycolate, glyoxylate,

oxalate, formate, and formaldehyde.

Reference

[S1] a) G. Kresse, J. Furthmüller, Phys. Rev. B 1996, 54, 11169; b) G. Kresse, D. Joubert,
Phys. Rev. B 1999, 59, 1758.
[S2] J. P. Perdew, K. Burke, M. Ernzerhof, Phys. Rev. Lett. 1996, 77, 3865.
[S3] a) J. Heyd, G. E. Scuseria, M. Ernzerhof, J. Chem. Phys. 2003, 118, 8207; b) J. Heyd,
G. E. Scuseria, J. Chem. Phys. 2004, 121, 1187.
[S4] R. Jacobs, J. Hwang, Y. Shao-Horn, D. Morgan, Chem. Mater. 2019, 31, 785.
[S5] C. F. Dickens, J. H. Montoya, A. R. Kulkarni, M. Bajdich, J. K. Nørskov, Surf. Sci.
2019, 681, 122.
[S6] X. Han, H. Sheng, C. Yu, T. W. Walker, G. W. Huber, J. Qiu, S. Jin, ACS Catal. 2020,
10, 6741.
[S7] G. Wu, J. Wang, W. Ding, Y. Nie, L. Li, X. Qi, S. Chen, Z. Wei, Angew Chem Int Ed
Engl 2016, 55, 1340.
[S8] S. Liu, D. Ni, H.-F. Li, K. N. Hui, C.-Y. Ouyang, S. C. Jun, J. Mater. Chem. 2018, 6,
10674.
[S9] a) J. Suntivich, W. T. Hong, Y. L. Lee, J. M. Rondinelli, W. L. Yang, J. B. Goodenough,
B. Dabrowski, J. W. Freeland, Y. Shao-Horn, J. Phys. Chem. C 2014, 118, 1856; b) W.
T. Hong, K. A. Stoerzinger, B. Moritz, T. P. Devereaux, W. L. Yang, Y. Shao-Horn, J.
Phys. Chem. C 2015, 119, 2063; c) W. T. Hong, K. A. Stoerzinger, Y. L. Lee, L.
Giordano, A. Grimaud, A. M. Johnson, J. Hwang, E. J. Crumlin, W. L. Yang, Y. Shao-
Horn, Energy Environ. Sci. 2017, 10, 2190.
[S10] A. Grimaud, K. J. May, C. E. Carlton, Y. L. Lee, M. Risch, W. T. Hong, J. G. Zhou, Y.
Shao-Horn, Nat. Commun. 2013, 4, 2439.

14
[S11] Y. L. Lee, J. Kleis, J. Rossmeisl, Y. Shao-Horn, D. Morgan, Energy Environ. Sci. 2011,
4, 3966.
[S12] Y. M. Sun, H. B. Liao, J. R. Wang, B. Chen, S. N. Sun, S. J. H. Ong, S. B. Xi, C. Z.
Diao, Y. H. Du, J. O. Wang, M. B. H. Breese, S. Z. Li, H. Zhang, Z. C. J. Xu, Nat. Catal.
2020, 3, 554.
[S13] T. Wang, J. Y. Wang, Y. M. Sun, Y. Duan, S. N. Sun, X. Hu, S. B. Xi, Y. H. Du, C.
Wang, Z. C. J. Xu, Appl. Catal. B: Environ. 2019, 256, 117844.
[S14] A. Grimaud, A. Demortiere, M. Saubanere, W. Dachraoui, M. Duchamp, M. L. Doublet,
J. M. Tarascon, Nat. Energy 2017, 2, 16189.

Supporting Tables

Table S1. Selected materials with O 2p BCs between –3.65 (eV) and –3.55 (eV).

Id on Materials N of atoms in Material Band Gap O 2p BCs Stability

Project the unit cell (eV) (eV)

mp-559159 30 Cu4As2O9 0.00 –3.59 not stable

mp-1105271 20 MnBiO3 0.00 –3.57 not stable

mp-608595 20 NaBrO3 0.08 –3.56 not stable

mp-550070 8 NbBr2O 0.93 –3.58 stable[a]

mp-22649 20 VCdO3 0.37 –3.60 not stable

mp-1190590 24 NClO4 0.00 –3.58 not stable

mp-1192710 28 Ni(ClO2)2 0.00 –3.57 not stable

mp-29203 28 La3Co3O8 0.00 –3.62 not stable

mp-17289 24 CoMoO4 0.00 –3.55 stable

mp-31513 30 Nb2Co4O9 0.00 –3.61 not stable

mp-19424 20 TiCoO3 0.00 –3.63 stable

mp-510281 8 FeCuO2 0.85 –3.56 stable[a]

mp-9600 14 Cu2GeO4 0.00 –3.61 not stable

mp-554874 20 CuGeO3 0.00 –3.62 not stable

mp-1193545 28 Cu(SbO2)2 0.50 –3.62 not stable

mp-541147 34 Cu4Se3O10 0.00 –3.61 stable[a]

mp-562987 5 VCuO3 0.86 –3.59 not stable

15
mp-1079306 20 Sr3Fe2O5 0.00 –3.57 not stable

mp-19218 10 Sr3Fe2O5 0.74 –3.58 not stable

mp-1077874 7 Tm(FeO2)2 0.00 –3.59 not stable

mp-1080477 9 I(NO3)2 0.00 –3.56 not stable

mp-1193141 30 MnInO3 0.94 –3.60 not stable

mp-1193780 30 MnInO3 0.84 –3.57 not stable

mp-21874 28 La2NiO4 0.03 –3.59 not stable

mp-27510 14 MgMn2O4 0.00 –3.60 stable

mp-1189476 20 NdMnO3 0.00 –3.61 not stable

mp-7961 5 Sr3SnO 0.00 –3.61 stable[a]

mp-18717 5 SrVO3 0.24 –3.56 not stable

mp-21303 17 Sr4V3O10 0.00 –3.61 not stable

mp-22391 12 Sr3V2O7 0.58 –3.61 stable[a]

mp-1198168 58 Fe5(BO3)6 0.00 –3.61 not stable

mp-1201286 58 La2(SeO8)3 0.05 –3.56 not stable

[a] TMs are not tetrahedrally or octahedrally coordinated.

Table S2. BC values for the selected three materials during screening.

Material O 2p (PBE) O 2p (HSE06) TM 3d O 2p – TM d

(eV) (eV) (HSE06) (eV) (HSE06) (eV)

α-CoMoO4 –3.55 –4.10 –3.92 –0.18

β-CoMoO4 –3.55 –4.00 –3.80 –0.20

MgMn2O4 –3.58 –4.05 –3.87 –0.18

TiCoO3 –3.62 –3.99 –3.86 –0.13

Table S3. Comparison of the CoMoO4/NF catalyst with the recently reported non-noble metal

catalysts for electrochemical oxidation of small organic molecules assisted H2 production.

16
Catalyst Substratea Electrolyteb Temp. E(10)c ΔEd Ref.
CoMoO4 Ni foam 1.0 M KOH + 60 oC 1.105 V 314 mV
0.1 M Glycerol This work
CoMoO4 Ni foam 1.0 M KOH + 25 oC 1.239 V 294 mV
0.1 M Glycerol

CuCo2O4 CFP 0.1 M KOH + 25 oC 1.26 V 290 mV ACS Catal. 2020,


0.1 M Glycerol 10, 6741
NiCo2O4 CFP 0.1 M KOH + 25 oC 1.30 V 270 mV ACS Catal. 2020,
0.1 M Glycerol 10, 6741

Ni-Mo-N CC 1.0 M KOH + 25 oC 1.30 V 270 mV Nat. Comm. 2019,


0.1 M Glycerol 10, 5335

NC@CuCo2Nx CFP 1.0 M KOH + 25 oC 1.25 V 210 mV Adv. Funct. Mater.


15 mM benzyl 2017, 27, 1704169
alcohol

Ni2P Ni foam 40 mL 1.0 M 25 oC 1.32 V 160 mV Angew.Chem. Int.


KOH + 0.5 (onset) (onset) Ed. 2019, 58,
mmol THIQs 12014

NiFeOx Ni foam 1.0 M KOH + 25 oC 1.33 V 190 mV Nat. Comm. 2020,


100 mM glucose (@50 (@50 11, 265
mA cm-2) mA cm-2)

Ni2P-UNMs Ni foam 1.0 M KOH + 25 oC 1.34 V 180 mV Appl. Catal. B:


0.125 M Environ. 2020, 268,
benzylamine 118393

Mo–Ni alloy Ni foam 1.0 M KOH + 25 oC 1.345 V 145 mV J. Mater. Chem. A,


nanoparticle 10 mM benzyl 2019, 7, 16501
alcohol

hp-Ni Ni foam 1.0 M KOH + 25 oC 1.35 V 160 mV ACS Catal. 2017,


10 mM benzyl (onset) (onset) 7, 4564
alcohol

NiSe@NiOx Ni foam 1.0 M KOH + 25 oC 1.35 V 200 mV Appl. Catal. B:


10 mM HMF (onset) (onset) Environ. 2020, 261,
118235

MoO2-FeP@C Ni foam 1.0 M KOH + 25 oC 1.359 V 115 mV Adv. Mater. 2020,


10 mM HMF 32, 2000455

Ni2P NPA Ni foam 1.0 M KOH + 25 oC 1.35 V 150 mV Angew.Chem.Int.


10 mM HMF (onset) (onset) Ed. 2016, 55, 9913

Ni3S2 Ni foam 1.0 M KOH + 25 oC 1.35 V 150 mV J. Am. Chem. Soc.


10 mM HMF (onset) (onset) 2016, 138, 13639

Co−P Cu foam 1.0 M KOH + 25 oC 1.38 V 150 mV ACS Energy Lett.


50 mM HMF (@20 (@20 2016, 1, 386
mA cm-2) mA cm-2)

17
Co(OH)2@HOS CFP 1.0 M KOH + 25 oC 1.385 V 186 mV Adv. Funct. Mater.
3.0 M methanol 2020, 30, 1909610

Co3O4 CFP 1.0 M KOH + 25 oC 1.445 V 55 mV ACS Cent. Sci.


nanosheets 1.0 M ethanol 2016, 2, 538

NiSe/NF Ni foam 0.1 M KOH + 25 oC 1.32 V 160 mV Angew. Chem. Int.


1.0 mmol Ed. 2018, 57,
benzylamine 13163
a) ‘CFP’ indicates ‘carbon fiber paper’; ‘CC’ represents ‘carbon cloth’. b) ‘THIQs’ indicates

‘tetrahydroisoquinolines’; ‘HMF’ indicates ‘5-hydroxymethylfurfural’. c) ‘E(10)’ represents the

potential to reach 10 mA cm–2 unless otherwise marked. d) ‘ΔE’ represents the potential difference

between the organic molecule’s oxidation and oxygen evolution reaction’.

Table S4. Summary of the product formation with passed charges.

Product concentration (mM)

Charge Oxal Tartr Glyoxyl Glyce Glycol For Form Glycero Carbon FE(GOR)
(C) ate onate ate rate ate mald ate l(mM) balance (%)
ehyd (%)
e

0 0 0 0 0 0 0 0 100.732 --- ---

30 0.024 0 0 0.068 0.436 0.690 9.008 92.090 95.459 87.824

60 0.106 0 0 0.075 0.909 1.61 17.417 89.500 96.956 86.466

90 0.170 0 0.103 0.166 1.424 2.096 26.685 83.155 94.749 88.507

120 0.267 0 0.172 0.249 1.965 2.373 35.588 79.669 95.059 88.497

150 0.316 0 0.275 0.419 2.370 2.531 44.543 74.345 93.432 88.253

180 0.495 0.005 0.438 0.486 2.883 2.808 55.222 71.553 95.174 91.759

210 0.597 0.021 0.471 0.586 3.366 2.906 65.009 65.947 93.334 91.452

18
240 0.570 0.029 0.603 0.657 3.823 2.980 72.865 61.880 92.487 89.757

300 0.847 0.045 0.842 0.891 5.125 2.761 95.882 52.709 92.234 92.664

Ave. 94.3 89.5

19
Supporting Figures

Figure S4. (a-b) SEM images of the CoMoO4•xH2O nanorods on nickel foam. (c) SEM image
of the CoMoO4•xH2O nanorod powder and the corresponding EDS spectrum. (d) Co K, (e) Mo
K, and (f) O K EDS mapping images of panel c.

Figure S5. (a) SEM image of CoMoO4 powder. (b) Corresponding EDS spectra of panel a.
(c) Co K, (d) Mo K, and (e) O K EDS mapping images of panel a.

20
Figure S6. (a-c) SEM images of NiMoO4/NF electrode. (d-e) TEM images of NiMoO4 nanorod.
(f) PXRD pattern of NiMoO4 powder. (g) Ni 2p, (h) Mo 3d, and (i) O 1s XPS spectra of the
NiMoO4/NF electrode.

21
Figure S7. (a-c) SEM images of the MnMoO4/NF electrode. (d-e) TEM images of MnMoO4
nanosheet. (f) PXRD pattern of MnMoO4 powder. (g) Mn 2p, (h) Mo 3d, and (i) O 1s XPS
spectra of NiMoO4/NF electrode.

Figure S8. Scan-rate dependent CV profiles of (a) MnMoO4/NF electrode, (b) NiMoO4
electrode, and (c) CoMoO4 electrode.

22
Figure S9. The concentration of products using the CoMoO4/NF electrode applied at different
potentials based on HPLC analysis.

Figure S10. The concentration of products using the CoMoO4/NF electrode applied at a
constant potential of 1.28 V vs. RHE as a function of reaction time based on HPLC analysis.

Figure S11. The concentration of products using the CoMoO4/NF electrode applied at a
constant potential of 0.30 V vs. Hg/HgO in relation to reaction temperature based on HPLC
analysis.

23
Figure S12. LSV profiles of all possible intermediates and products (50 mM) from GOR
dissolved in 1.0 M KOH electrolyte using the CoMoO4/NF electrode.

24
Figure S13. HPLC spectra of 50 mM (a) mesoxalate, (b) glyoxylate, (c) formaldehyde, (d)
glyceraldehyde, (e) DHA, (f) glycerate, (g) tartronate, (h) glycolate, (i) oxalate, (j) formate
dissolved in 1.0 M KOH before and after electrolysis at 1.28 V vs. RHE for 1.0 h using the
CoMoO4/NF electrodes. All solutions were neutralized before HPLC analysis. Impurities from
alkali and column are marked by a star symbol (*), unidentified peaks are labelled by a question
mark (?).

25
Figure S14. Raman spectra of pristine CoMoO4/NF in air.

Figure S15. Raman spectra of the CoMoO4/NF electrode in 1.0 M KOH electrolyte.

Figure S16. Magnified LSV profiles of the CoMoO4/NF electrode in 1.0 M KOH with and
without 0.1 M glycerol.

26
Figure S17. Raman spectra focused on the electrolyte of 1.0 M KOH (black), 0.1 M glycerol
in 1.0 M KOH (blue), 0.1 M formic acid in 1.0 M KOH (red).

Figure S18. CoMoO4 nanorod after GOR. (a) HRTEM image, (b) high-angle annular dark-field
(HAADF) image, (c) corresponding EDS mapping images of the nanorod.

Figure S19. (a) Chronopotentiometry profile of the CoMoO4/NF electrode applied at a constant
current of 100 mA. (b) The corresponding charge as a function of time curve according to panel
a. (c) The corresponding HPLC spectra of products from the anode compartment collected
every 30 C passed.

27

You might also like