Articles

https://doi.org/10.1038/s41560-019-0462-7

Decoupled hydrogen and oxygen evolution by

a two-step electrochemical–chemical cycle for
efficient overall water splitting
Hen Dotan1,5, Avigail Landman2,5, Stafford W. Sheehan   1,3, Kirtiman Deo Malviya   1,
Gennady E. Shter4, Daniel A. Grave1, Ziv Arzi   1, Nachshon Yehudai4, Manar Halabi4, Netta Gal1,
Noam Hadari1, Coral Cohen1, Avner Rothschild   1* and Gideon S. Grader   4*

Electrolytic hydrogen production faces technological challenges to improve its efficiency, economic value and potential for
global integration. In conventional water electrolysis, the water oxidation and reduction reactions are coupled in both time and
space, as they occur simultaneously at an anode and a cathode in the same cell. This introduces challenges, such as product
separation, and sets strict constraints on material selection and process conditions. Here, we decouple these reactions by divid-
ing the process into two steps: an electrochemical step that reduces water at the cathode and oxidizes the anode, followed by
a spontaneous chemical step that is driven faster at higher temperature, which reduces the anode back to its initial state by
oxidizing water. This enables overall water splitting at average cell voltages of 1.44–1.60 V with nominal current densities of
10–200 mA cm−2 in a membrane-free, two-electrode cell. This allows us to produce hydrogen at low voltages in a simple, cyclic
process with high efficiency, robustness, safety and scale-up potential.

C
lean energy utilization and its integration with an infrastruc- a method of decoupled water splitting that overcomes a substan-
ture built for energy storage and transportation requires tial barrier to implementation; namely, the energy conversion effi-
effective fuel production. This helps to overcome the inter- ciency. In doing so, we also provide more degrees of freedom in our
mittency of solar and wind power and provides a high-energy- scheme, enabling optimization of process parameters beyond con-
density fuel for transportation. To create such fuels in a sustainable ventional electrolysers8. We achieve this by dividing the water oxi-
manner, water oxidation is a critical step, utilized by the same photo- dation reaction into two steps: an electrochemical step that oxidizes
synthetic organisms that we derive our fossil fuels from today1. This the anode, followed by a spontaneous chemical step that reduces the
process oxidizes water to oxygen and liberates protons and electrons anode back to its initial state by oxidizing water.
for fuel generation2,3. These can be used to produce carbon-based
fuels by reducing carbon dioxide, or hydrogen by a more kineti- Electrochemical–chemical water splitting
cally facile proton reduction. Beyond its ease of synthesis, hydrogen The process we have developed is illustrated in Fig. 1, which com-
has further advantages over carbon-based fuels in its high energy pares alkaline water electrolysis (Fig. 1a) and our proposed process
density per unit mass, benign combustion by-products (H2O) and (Fig. 1b). The fuel formation reaction (hydrogen evolution in the
potential to create synthetic feedstock chemicals downstream by case of water splitting) is unchanged in this scheme, since it occurs
proven and widely implemented technologies such as the Haber– at the cathode, whereas our process only modifies the water oxi-
Bosch process for ammonia production4, or CO2 hydrogenation to dation reaction at the anode. Accordingly, in our process, water is
methane, methanol and other higher-order alkanes and alcohols5. reduced to hydrogen gas at the cathode, liberating OH– ions. The
In renewable processes for producing hydrogen, carbon-based four-electron oxygen evolution reaction (OER), which takes place at
fuels and ammonia, the reagent protons and electrons originate the anode in conventional electrolysis (Fig. 1a), is divided into two
from the oxidation of water, with oxygen as a by-product3. In a consecutive steps comprising four one-electron oxidation reactions
purely electrolytic scheme, the water oxidation and reduction reac- of a nickel hydroxide (Ni(OH)2) anode12,13, followed by spontaneous
tions are tightly coupled in both time and space, as they occur oxygen evolution and anode regeneration in a thermally activated
simultaneously at two electrodes—an anode and a cathode—placed chemical step14,15, as illustrated in Fig. 1b (see also Supplementary
together in the same cell6. This coupling introduces operational Note 1 and Supplementary Fig. 1 for more details).
challenges, such as H2/O2 crossover at low current densities, which Thus, in our two-step electrochemical–thermally activated
hampers operation under variable renewable energy sources such chemical (E-TAC) cycle, the O−O bond is formed in the sec-
as solar and wind7,8, and sets strict constraints on material selection ond step16, when the oxidized anode (now nickel oxyhydroxide
and process conditions. (NiOOH)) is reduced to Ni(OH)2 in a spontaneous, exergonic
Following on previous work exploring different paths to decou- chemical reaction that produces O2 and regenerates the anode17,
ple the water oxidation and reduction reactions7–11, here we propose completing the overall water-splitting reaction. Similarly to the

1
Department of Materials Science and Engineering, Technion—Israel Institute of Technology, Haifa, Israel. 2The Nancy and Stephen Grand Technion
Energy Program, Technion—Israel Institute of Technology, Haifa, Israel. 3Catalytic Innovations, Fall River, MA, USA. 4Department of Chemical Engineering,
Technion—Israel Institute of Technology, Haifa, Israel. 5These authors contributed equally: Hen Dotan, Avigail Landman.
*e-mail: avner@mt.technion.ac.il; grader@technion.ac.il

786 Nature Energy | VOL 4 | SEPTEMBER 2019 | 786–795 | www.nature.com/natureenergy

Nature Energy Articles
a b Step 1: hydrogen evolution Step 2: oxygen evolution
e– e–
e– e–
H2 Power O2 H2 Power O2

NiOOH → Ni(OH)2
Ni(OH)2 → NiOOH
Separator

Cathode
Cathode

Anode

Cathode
OH – OH– 25 °C 95 °C

HER: 4H2O + 4e– → 4OH – + 2H2 HER: 4H2O + 4e – → 4OH – + 2H2 4NiOOH + 2H2O → 4Ni(OH)2 + O2
– –
–
OER: 4OH → 2H2O + O2 + 4e – (Ni(OH)2 + OH → NiOOH + H2O + e ) × 4

Fig. 1 | Schematic of alkaline water electrolysis and the E-TAC water-splitting process. a, In alkaline water electrolysis, which typically takes place at
elevated temperatures (50–80 °C)6, the OER and HER are coupled in both time and space, as they occur simultaneously at an anode and a cathode,
which are placed together in the same cell. A diaphragm or anion exchange membrane separates the anode and cathode compartments and prevents
O2/H2 crossover. b, E-TAC water splitting proceeds in two consecutive steps. An electrochemical step (left) reduces water by the conventional HER at
the cathode, liberating hydroxide ions (OH–) that oxidize a nickel hydroxide (Ni(OH)2) anode into nickel oxyhydroxide (NiOOH). This step is followed by
a chemical step (right), wherein the NiOOH anode reacts with water to spontaneously produce oxygen. The first (electrochemical) reaction occurs at
ambient temperature (~25 °C), whereas the second (chemical) reaction proceeds at elevated temperatures (~95 °C) for the optimum rate of reaction. The
first and second reactions sum up to the overall water-splitting reaction, 2H2O → 2H2 + O2.

oxygen-evolving complex in natural systems18, our scheme uses four that may occur on overcharging21, additives that catalyse the
metal centres to store oxidative equivalents, which helps to level the Ni(OH)2/NiOOH reaction and inhibit oxygen evolution are used22–26.
potential of the elementary oxidation reactions, thereby reducing One of the most commonly used additives is cobalt. Cobalt
the overall reaction overpotential19. Compared with previous work improves the electron and proton conductivity of the anode and
in this emerging field8, our approach enables decoupled water split- shifts the Ni(OH)2/NiOOH redox potential cathodically, expand-
ting with enhanced efficiency and versatility, as shown below. ing the potential window between charging and overcharging27–30.
Therefore, Ni1-xCox(OH)2 anodes can be charged at a lower poten-
Electrode design tial than their undoped Ni(OH)2 counterparts, to a higher state of
In brief, our E-TAC water-splitting process replaces the conven- charge (SOC), without evolving oxygen.
tional water oxidation reaction in alkaline electrolysis with a two- Unlike battery anodes that should avoid oxygen generation, our
step cycle in which the anode is first charged (electrochemically) anodes must generate oxygen in the second step of the E-TAC cycle.
and then regenerated (chemically). We emphasize that the hydrogen Additionally, it is essential to supress the spontaneous regenera-
evolution reaction (HER) in the first step of our process remains tion reaction, which is accompanied by oxygen evolution, during
the same as in alkaline electrolysis, except that it occurs at ambient the hydrogen generation (first) step. Therefore, our anodes and
temperature instead of elevated temperature (typically 50–80 °C)6. operating conditions are tailored to suppress oxygen evolution in
Therefore, the same cathode materials used in alkaline electrolysis, the first step and enhance oxygen generation in the second step
such as Raney nickel6 or other HER catalysts, can be used in our (see Supplementary Notes 2–4 and Supplementary Figs. 2–13 for
process. In this work, we use platinized nickel-coated stainless-steel details). In this study, we use Ni0.9Co0.1(OH)2 anodes, prepared by
mesh cathodes as a standardized benchmark HER cathode for our electrochemical impregnation (ECI) of nickel foam substrates (see
proof-of-concept experiments. Methods and Supplementary Figs. 2 and 3). ECI electrodes have a
The anode, which in our process functions in a completely dif- higher surface area than sintered or pasted electrodes, and can there-
ferent manner from conventional alkaline electrolysis, requires fore sustain higher current densities at lower overpotentials27–30.
careful material selection and optimization. The key selection cri- Furthermore, the regeneration rate is much higher for the ECI
teria for the anode include: stability in alkaline solutions; cyclability anodes compared with their pasted counterparts (Supplementary
of the metal hydroxide and oxyhydroxide phases; a redox potential Note 3). Optical and scanning electron microscope (SEM) micro-
higher than the reversible OER potential (1.23 VRHE) and lower than graphs, as well as X-ray diffractograms of the anodes are presented
the potential at which oxygen evolves on the anode material (see in Supplementary Figs. 14–24. These show conformal coating of
Supplementary Note 1 for details); high capacity; fast charging and a the nickel foam substrate by a roughly 9-μm-thick layer of β-nickel
fast rate of regeneration. These considerations all point to Ni(OH)2/ hydroxide. The coating layer composition was determined by
NiOOH anodes, which are commonly used in alkaline secondary energy-dispersive spectroscopy, yielding an Ni:Co ratio of 6.7:1.0
batteries due to their cyclability and high energy density20. Ni(OH)2 (where x ≅ 0.1; that is, Ni0.9Co0.1(OH)2) (Supplementary Table 1).
reversibly oxidizes to NiOOH according to the following reac-
tion12,13: Ni(OH)2 + OH– ↔ NiOOH + H2O + e–. Electrochemical characterization
Although the Ni(OH)2/NiOOH standard redox potential is The open-circuit potential (OCP) as a function of the SOC27,31 and
higher by 190 mV than the reversible OER potential, the OER is the cyclic voltammogram of a Ni0.9Co0.1(OH)2 anode were mea-
suppressed kinetically by its large overpotential. This enables alka- sured and compared with those of an undoped Ni(OH)2 anode
line batteries to operate thousands of cycles without parasitic oxy- (Supplementary Figs. 25 and 26). The OCP ranges from 1.31–
gen evolution that leads to swelling and malfunction, as long as 1.41 VRHE for the Ni0.9Co0.1(OH)2 anode, and from 1.37–1.42 VRHE for
they are not overcharged. To suppress parasitic oxygen evolution the undoped Ni(OH)2 anode, over a SOC ranging from 10–90%.

Nature Energy | VOL 4 | SEPTEMBER 2019 | 786–795 | www.nature.com/natureenergy 787

Articles Nature Energy

The addition of cobalt shifts the potential cathodically by approxi- a

mately 50 mV, suppressing oxygen evolution at the anode during 900
the hydrogen generation step. The results reported hereafter were
obtained with Ni0.9Co0.1(OH)2 anodes. 600

Current density (mA cm−2)

The electrochemical properties of our Ni0.9Co0.1(OH)2 anode
were investigated, and compared with benchmark water oxidation 300
catalyst NiFe layered double hydroxide (LDH). Cyclic voltammo-
grams of the Ni0.9Co0.1(OH)2 and NiFe LDH anodes are presented 0
in Fig. 2a. The Ni0.9Co0.1(OH)2 anode displays a large pseudo-
capacitive oxidation wave of Ni+2 to Ni+3, beginning at 1.3 VRHE –300
during the anodic sweep, preceding the water oxidation currents of
both anodes. This shows the possibility to surpass the electrochemi- –600
cal efficiency of the NiFe LDH water oxidation catalyst.
To distinguish between the pseudo-capacitive anode oxidation –900
0.7 0.9 1.1 1.3 1.5 1.7 1.9
current and the steady-state water oxidation current, chronoam- Potential (VRHE)
perometry measurements were carried out at constant potentials
between 1.3 and 1.7 VRHE (Supplementary Fig. 27), as shown in b 200
Fig. 2b for 1.48 VRHE. The asymptotic values obtained at different
potentials yield the Tafel plot of the steady-state water oxidation
current of both anodes, presented in Fig. 2c and overlaid in Fig. 2a 150

Current density (mA cm−2)

(dotted curves). For the Ni0.9Co0.1(OH)2 anode, the steady-state
water oxidation current is much smaller than the pseudo-capacitive
current measured in the cyclic voltammogram, indicating that the 100
dominant process at these potentials is the pseudo-capacitive anode
charging reaction. In contrast, when comparing steady-state water
oxidation and pseudo-capacitive currents for the NiFe LDH anode,
50
most of the net current arises from water oxidation.

Faradaic efficiency and H2 gas purity

To determine the Faradaic efficiency of the Ni0.9Co0.1(OH)2 anode 0
0 100 200 300 400 500 600
charging reaction, we first compared the capacitive charge with Time (s)
the total charge that flows during chronoamperometric charging.
On charging a fully discharged anode at a potential of 1.48 VRHE, c 3.0
the Faradaic efficiency was larger than 99.7% for more than 2,000 s,
during which the anode was charged by more than 50 C cm−2 2.5
log[current density (mA cm−2)]

(Supplementary Note 4 and Supplementary Fig. 27c). To verify

2.0
that no oxygen was produced, the dissolved oxygen concentration
in the electrolyte was monitored while charging a fully discharged 1.5
anode at a constant potential of 1.48 VRHE. No rise in the dissolved
oxygen concentration was observed up to a charge of 55 C cm−2 1.0
(Supplementary Fig. 28).
The dissolved oxygen concentration was also monitored under 0.5
conditions as similar as possible to the hydrogen generation step of
the E-TAC process (see Methods) (that is, while charging the anode 0
at a constant current). In addition, the dissolved oxygen concen-
tration was also measured under chronoamperometric charging of –0.5
1.2 1.3 1.4 1.5 1.6 1.7 1.8
a regenerated Ni0.9Co0.1(OH)2 anode at 1.48 VRHE. In both cases, no Potential (VRHE)
rise in the dissolved oxygen concentration above the background
level was observed up to 3 C cm−2 (Supplementary Figs. 29 and 30), Fig. 2 | Electrochemical properties of Ni0.9Co0.1(OH)2 and NiFe LDH
confirming that no oxygen was produced while charging the anode anodes. a, Cyclic voltammograms measured at a scan rate of 50 mV s−1.
under these conditions. Finally, gas chromatography measurements b, Chronoamperometric current traces at 1.48 VRHE. c, Tafel plots of the
displayed a rise in hydrogen concentration with no rise in the oxy- steady-state water oxidation current as a function of the potential,
gen concentration in the headspace of a sealed cell on anode charg- from 10-min amperograms, such as those presented in b. The dotted
ing for 100 s at a constant current density of 50 mA cm−2 (that is, curves in a represent the steady-state water oxidation current extracted
up to a charge of 5 C cm−2) (Supplementary Fig. 31). These results from c. For all graphs, the Ni0.9Co0.1(OH)2 anode is shown in blue and the
confirm that no parasitic side reactions such as oxygen evolution NiFe LDH anode is shown in red. All tests were carried out in 5 M KOH at
take place while charging the anode during the hydrogen generation ambient temperature.
step of the E-TAC process.

Proof of concept
In our proof-of-concept experiment, hydrogen was produced heating or cooling) for 100 s at a current density of 50 mA cm−2.
at the platinized nickel-coated stainless-steel cathode while the We limited the applied cell voltage to not exceed the thermoneu-
Ni0.9Co0.1(OH)2 anode was oxidized to Ni0.9Co0.1OOH in 5 M KOH tral voltage (1.48 V) by more than 120 mV. This helped to minimize
aqueous solution (pH 15)32,33. A champion Ni0.9Co0.1(OH)2 anode heat loss, and prevented oxygen generation and anode overcharg-
with a regenerated capacity of 5 C cm−2 was used. The hydrogen ing that may have transformed NiOOH from the β-phase to the γ-p
generation steps were carried out at ambient temperature (without hase20,22,34,35. On the basis of the dissolved oxygen measurements

788 Nature Energy | VOL 4 | SEPTEMBER 2019 | 786–795 | www.nature.com/natureenergy

Nature Energy Articles
a 1.60 b 1.7
Cycle 1
Cycle 2
Cycle 3
Cycle 4
Cycle 5 V cell
1.55 Cycle 6

Potential (VRHE)/voltage (V)

Cycle 7 1.6

Anode potential (V RHE)

Cycle 8
Cycle 9
Cycle 10
1.50

1.5
E anode
1.45

1.4
1.40

1.35 1.3
0 20 40 60 80 100 0 1 2 3 4 5 6 7 8 9 10 11
Time (s) Cycle number

c 1.60 d
6

1.55 5

Hydrogen produced (ml)

Cell voltage (V)

4
Cycle 1
1.50 Cycle 2
3 Cycle 3
Cycle 4
Cycle 5
2 Cycle 6
1.45 Cycle 7
Cycle 8
1 Cycle 9
Cycle 10

1.40 0
0 1 2 3 4 5 6 7 8 9 10 0 2 4 6 8 10 12 14 16 18 20 22
Cycle number Energy consumption (mWh)

Fig. 3 | E-TAC water splitting in alkaline solution. a, Anode potential (Eanode, full lines) and cell voltage (Vcell, dashed lines) during the hydrogen generation
steps of ten consecutive 100 s/100 s E-TAC cycles at a nominal current density of 50 mA cm−2 in 5 M KOH solution. b, Corresponding average anode
potential (Eanode, data points) and range (bars) in each cycle. c, Average cell voltage (Vcell, data points) and range (bars) in each cycle. d, Cumulative
hydrogen production (at 1 atm and 25 °C) as a function of electrical energy consumption during the E-TAC cycles. The blue and red dotted lines in
b represent the potential of the OER reaction extracted from Fig. 2c at a steady-state current of 50 mA cm−2 of the Ni0.9Co0.1(OH)2 anode and NiFe
LDH anode, respectively. The cycle-to-cycle variations in the cell voltage in a are attributed to the cycling procedure that involved disconnecting and
reconnecting the anode manually between the steps, leading to small variations in the contact and series resistances.

(Supplementary Fig. 30), the Faradaic efficiency of the anode charg- blue and red dotted lines in Fig. 3b, respectively). At these poten-
ing reaction was 99.2% ± 0.1% for this amount of charge. The cell tials, oxygen evolution is negligible (Supplementary Figs. 27 and
voltage, Vcell, and anode potential, Eanode, were recorded during the 30). The overall cell voltage varied from 1.44–1.56 V (Fig. 3a), with
hydrogen generation (charging) steps and are presented in Fig. 3a. an average of 1.5 V for all ten cycles (Fig. 3c). The E-TAC process’
After each of the 100-s hydrogen generation steps, the charged voltage efficiency36, Vth/Vcell, where Vth is the thermoneutral voltage
anode was taken out of the cold cell and placed in a hot (T = 95 °C) of the water-splitting reaction (1.48 V under standard conditions),
electrolyte (5 M KOH) for 100 s to accelerate spontaneous oxy- gave an average of 98.7% for all ten cycles. From the current × time
gen release and regeneration of its initial state. Significant bubble product (that is, charge), one can calculate that 6.4 ml hydrogen (at
(O2) formation was observed during the anode regeneration step, 1 atm and 25 °C) was generated during the ten cycles presented in
whereas during the hydrogen generation step, bubbles (H2) were Fig. 3, consuming 20.8 mWh based on the current and cell volt-
formed only on the cathode (Supplementary Videos 1 and 2). age measured during the cycles, as shown in Fig. 3d. Therefore, the
The cell voltage and anode potential increased during each E-TAC process achieved a low electrical energy consumption of
of the charging steps in a consistent manner, displaying little only 3.24 kWh m−3, or 39.9 kWh kg−1 hydrogen.
variability from cycle to cycle with no steady drift (Fig. 3a). This In addition to the proof-of-concept experiment presented in
electrochemical behaviour indicates that the process is cyclable Fig. 3, further tests were carried out to examine the efficiency and
and proceeds without degradation; at the end of each cycle, the cyclability over a range of operational current densities and a large
anode’s SOC returns to its initial state with no measurable change. number of cycles. First, similar tests were carried out at different
To confirm this, the anode cyclability was verified by ex  situ nominal current densities ranging from 10–200 mA cm−2. The
microstructural (SEM) and crystallinity (X-ray powder diffraction) average cell voltage was found to increase from 1.44 V at 10 mA cm−2
analyses before, during and after the E-TAC cycles (Supplementary to 1.60 V at 200 mA cm−2 (Supplementary Figs. 32 and 33). The
Figs. 14–21). anode potential in these tests was lower than that of the NiFe
During the hydrogen generation steps, the anode potential var- LDH benchmark anode (Supplementary Fig. 33), showing
ied from 1.37–1.45 VRHE (Fig. 3a)—well below the OER potential of the higher electrochemical efficiency of our E-TAC approach
both the Ni0.9Co0.1(OH)2 and NiFe LDH anodes (presented by the compared with alkaline electrolysis with a state-of-the-art water

Nature Energy | VOL 4 | SEPTEMBER 2019 | 786–795 | www.nature.com/natureenergy 789

Articles Nature Energy

a O2 H2 *R
b

Multi-cell system Single-cell system

Power
Wash

Water
inlet
O2 H2

Cell
Cell C–Z
Cell B Wash
A

Cold fluid/H2 production

Hot fluid/O2 production
Intermediate-temperature fluid/
removing residual gases

Fig. 4 | Schematic of the multi-cell E-TAC process. a, General electrolyte flow in the process, showing the cold electrolyte circulating in cell A, generating
H2 gas that is phase-separated in the middle (blue) tank on the left, while the hot electrolyte circulates in cell B, generating O2 gas that is phase-separated
in the red tank on the left. The intermediate-temperature wash fluid in the grey tank on the left is circulated through a bypass most of the time. This fluid
is used to ‘push’ the cold and hot electrolyte solutions to their respective tanks at the end of the H2 generation or O2 generation steps. b, Single-cell cycle,
showing the regenerative operation in each cell, where H2 is generated under an electric bias at ambient temperature, and O2 is generated at a hotter
temperature without applying electric bias (open circuit). *R, residual gases.

oxidation catalyst. Second, 100 cycles were carried out at the same a low-temperature electrolyte circulates through cell A, moving
nominal current density as shown in Fig. 3 (50 mA cm−2), providing the hydrogen bubbles to the hydrogen separator (shown in blue).
evidence for stable, extended operation in alkaline solutions Simultaneously, a hot electrolyte flows through cell B to regener-
(Supplementary Fig. 34). ate the (previously charged) anode, producing oxygen and moving
the oxygen bubbles to the oxygen separator (shown in red). There
E-TAC water splitting in carbonate-buffered solutions are no moving electrodes in the entire system, simply hot and cold
The proof-of-concept experiments presented so far were carried moving electrolyte solutions.
out in concentrated alkaline solution, which may be unsuitable In addition to the hot and cold electrolytes, an intermediate-
for certain applications where highly caustic conditions cannot be temperature fluid will be used to displace the cold (hot) electrolyte
tolerated. The option to decrease the solution pH to less hazard- within each cell into the proper separation tank to avoid mixing of
ous levels is useful for low-cost implementation, due to the reduced the hydrogen-saturated (oxygen-saturated) electrolyte with the hot
balance of system costs that come with operation at extreme pH (cold) electrolyte during the switch between cycle steps. The cycle
values and the potential for the use of low-purity water sources37. that each single cell undergoes is illustrated in Fig. 4b, showing that
We thus demonstrate the versatility of our approach by also operat- between the cold and hot steps, the intermediate displacement fluid
ing in a carbonate–bicarbonate (3:2 ratio of K2CO3:KHCO3) buffer fills the cell. In addition, as shown in Fig. 1b, Fig. 4b shows that the
electrolyte with a pH of 10.6. In this test, the anode potential var- electrodes are connected to a power source only during the hydro-
ied from 1.43–1.50 VRHE at a nominal current density of 25 mA cm−2 gen production step. The system will be hermetically sealed, so that
(Supplementary Fig. 35). This shows that the E-TAC water-splitting high-purity hydrogen and oxygen streams are expected to be pro-
process is not limited to alkaline solutions, with the potential to be duced. In addition, the temporal separation of hydrogen and oxygen
applied more broadly in the electrochemical industry. production in each cell adds a measure of safety to the proposed
concept by reducing the potential for gas mixing. Finally, as shown
Proposal for scale-up and deployment in Fig. 4a, the system is envisioned to contain multiple cells, such
The results of our proof-of-concept experiment (Fig. 3) and the that at any given point in time, some cells will be producing hydro-
complementary tests presented in Supplementary Figs. 27–35 show gen, while the others produce oxygen.
the potential of the E-TAC water-splitting process to yield high
electrochemical efficiency at a wide range of current densities Energy and heat balance
with versatile operational conditions, while ensuring safe opera- The results presented in Fig. 3 show only the electrical energy con-
tion with no H2/O2 crossover. To transform these laboratory tests sumption during the first step of the E-TAC cycles, which is car-
into an operational prototype system that produces pure hydrogen ried out at ambient temperature without heating. To accelerate the
and oxygen gas streams, we envision a multi-cell system, as illus- anode regeneration reaction on the second step, the anode is heated
trated in Fig. 4a. Unlike the proof-of-concept tests presented above, to 95 °C. This introduces some heat loss to the overall energy bal-
this system contains stationary anodes and cathodes in each cell, ance of our process, as discussed in detail in Supplementary Notes
operating in a swing configuration. During hydrogen production, 5 and 6. However, since the first step in the E-TAC process is endo-

790 Nature Energy | VOL 4 | SEPTEMBER 2019 | 786–795 | www.nature.com/natureenergy

Nature Energy Articles
Table 1 | Comparison of different water-splitting schemes
Method Operation mode Membrane/ Current density Average cell Technology readiness level Reference
diaphragm (mA cm−2) voltage (V)
Decoupled watter splitting
 E-TAC Swing Not required 10 1.44 Research and development This work
50 1.50
100 1.56
200 1.60
Solid-state redox Swap/swing Not required 20 2.00 Research and development 10
mediator 5 2.10 Research and development 11
Soluble electron-coupled Swing PEM 100 2.94 Research and development 7
proton buffers Continuous 50 2.37 Research and development 9
Electrodeposition and Swing Not required 40 2.47 Commercial (Ergosup) 53
dissolution 60 2.75
Conventional (coupled) water splitting
Alkaline electrolysis Continuous Diaphragma 10 1.52 Research and development 54
(NiFe LDH WOC)
Alkaline electrolysis Diaphragma 50 1.464 44
(NiFe OOH WOC) 500 1.586
Alkaline electrolysis AEM 200 1.5 Commercial (Hydrogenics) 55
(zero gap)
Alkaline electrolysis Diaphragm 200–450 1.8–2.4 Commercial 6
(conventional)
PEM electrolysis PEM 500–2,500 1.7–2.1 Commercial 6
(zero gap)
a
No diaphragm was employed in the reported tests. AEM, anion exchange membrane; WOC, water oxidation catalyst. Operating parameters and electrochemical efficiency (average cell voltage) are shown
for landmark reports on decoupled water splitting, alkaline electrolysis with champion catalysts, commercial alkaline electrolysers and PEM electrolysers.

thermic for Vcell < 1.56 V (Supplementary Note 6), whereas the sec- steps, we are able to reduce the anode potential to a level competi-
ond step is exothermic (Supplementary Table 2), the heat loss that tive with the best reported water oxidation catalysts31,41–45. It is fur-
arises from swinging between the cold and hot steps can be mini- ther noted that the efficiency of the E-TAC process is considerably
mized by careful heat management and thermal insulation of the higher than that of previous reports using the Ni(OH)2/NiOOH
hot components, and increasing the capacity of the anode so that redox couple as an auxiliary electrode to decouple the water oxida-
the heat released during O2 generation compensates for any other tion and reduction reactions10,11. This is because, in our approach,
minor heat loss. We believe that this heat loss will add no more than oxygen is generated spontaneously by reacting Ni(OH)2/NiOOH
2 kWh kg−1 H2 (Supplementary Note 6). and water at our modified anodes, without the need for any elec-
trode polarization from an external power source. In contrast, in
Advantages of the E-TAC approach previous reports10,11 oxygen was generated electrochemically at a
Our E-TAC process splits water in a membrane-free cell, generat- standard anode by polarizing the anode while the Ni(OH)2/NiOOH
ing hydrogen and oxygen in temporally separate steps. This yields auxiliary electrode discharged. Our anode design enables water per-
several generic advantages of membraneless electrolysis compared meation to NiOOH sites, and the inclusion of cobalt improves elec-
with conventional water electrolysis, as reported elsewhere38. These tron and proton conductivity while cathodically shifting the Ni+2/
include lower material and assembly costs, derived from reduced Ni+3 redox potential, which enables this chemistry.
device complexity (Supplementary Note 7); lower maintenance A similar approach that combines electrochemical and chemi-
costs; high tolerance to impurities in the electrolyte and adverse cal reactions was presented previously7. It uses a molecular redox
operating conditions, which typically foul membranes; and the mediator that serves a similar role to the Ni(OH)2/NiOOH anode
opportunity to operate in non-hazardous buffer solutions, which in our scheme. Unlike the previous approach, which requires the
also raises the operational safety while enabling the use of low-cost use of an ion exchange membrane to prevent self-discharge, our
catalyst and/or construction materials. The problem of H2/O2 cross- use of a solid redox mediator (the Ni(OH)2/NiOOH anode) pre-
over across the membrane/separator in proton exchange membrane vents charged species from crossing between anode and cathode,
(PEM)/alkaline electrolysis6,39,40 is avoided in our process, paving thus eliminating the need for a membrane. Furthermore, the sol-
the path for partial-load water splitting7,8 and high-pressure hydro- uble redox mediator reported in ref. 7 operates in acidic solution,
gen production, as in the Ergosup hydrogen generator (https:// and therefore requires rare-Earth catalysts, such as platinum. In
www.ergosup.com/en/electrolysis-under-pressure). contrast, our solid redox mediator operates in alkaline solutions,
Another key merit of the E-TAC process is its high efficiency. enabling the use of low-cost, Earth-abundant catalysts. Lastly, our
By tackling the four-electron OER using single-electron oxidation approach replaces the conventional water oxidation reaction, which
events of nickel hydroxide, and dividing water oxidation into sepa- has a higher overpotential than proton reduction, thereby yielding
rate one-electron electrochemical and thermally activated chemical higher efficiency than in previous studies7,9. Table 1 compares the

Nature Energy | VOL 4 | SEPTEMBER 2019 | 786–795 | www.nature.com/natureenergy 791

Articles Nature Energy

electrochemical efficiency (that is, the cell voltage) of our E-TAC anode with a regenerated charge of 5 C cm−2. All other experiments were carried
process with other landmark reports on decoupled water splitting, out with standard Ni0.9Co0.1(OH)2 anodes with a regenerated charge of 3 C cm−2.
In addition, for comparison of the Ni0.9Co0.1(OH)2 anodes with undoped Ni(OH)2
as well as with alkaline electrolysis using champion catalysts, and (Supplementary Figs. 25 and 26), we prepared an undoped Ni(OH)2 anode by the
commercial alkaline electrolysers and PEM electrolysers. same process without cobalt nitrate.
Currently, our anodes offer nominal current densities of up to
200 mA cm−2, at the low end of commercial alkaline water elec- NiFe LDH anode fabrication. The same nickel foam substrate described in
the previous section was used for the fabrication of the NiFe LDH anode. The
trolysers6,40. Achieving higher current densities requires further
deposition solution comprised 0.095 M Ni(NO3)2 and 0.05 M FeCl2 in deionized
optimization using high-surface-area substrates and optimizing water. Before deposition, the solution was purged with N2 gas for 20 min. The same
the anode fabrication process. It is noteworthy that indirect com- cell configuration described in the previous section was also employed for NiFe
parison between our anodes and electrolysis anodes is difficult LDH anode fabrication. Deposition was carried out by applying a pulsed cathodic
because the E-TAC process occurs within the volume of the active current of 10 mA for 20 cycles of 2 s on/off pulses, following the methods described
elsewhere49. Following deposition, the anode was washed in deionized water.
Ni0.9Co0.1(OH)2 layer, whereas in electrolysis, only the surface of the
water oxidation catalyst participates in the reaction. Direct com- Ni0.9Co0.1(OH)2 anode structural characterization. The microstructure, phase
parison with a benchmark water oxidation catalyst (NiFe LDH), and elemental composition evolution of our Ni0.9Co0.1(OH)2 anodes were
using the same substrate and measurement conditions, shows that examined by optical microscopy (using an Olympus BX51 optical microscope),
the operation potential of our anode is lower than that of the water scanning electron microscopy (using an FEI ESEM Quanta 200), energy-
dispersive X-ray spectroscopy (using an Oxford Instruments EDS system) and
oxidation catalyst (Fig. 3b). X-ray diffraction (using a Rigaku SmartLab diffractometer), following different
stages in the fabrication and operation processes (Supplementary Figs. 14–21 and
Conclusion Supplementary Table 1). To reduce cracking due to vacuum drying and shrinkage
With growing utilization of intermittent renewable electricity under the high-vacuum conditions (1.4 × 10−2 torr) in the SEM, the wet anodes
around the globe, efficient and viable ways of storing excess energy were frozen before placing them inside the SEM. However, during SEM operation,
the ice sublimed and cracks were formed. To show that the cracks observed in the
will become increasingly important. Our E-TAC process shows that SEM images of our anodes are drying shrinkage cracks that resulted from the high-
water splitting at nearly thermoneutral conditions is indeed possi- vacuum conditions inside the SEM, optical microscopy imaging of wet anodes was
ble, and we emphasize the many avenues that can be explored to fur- carried out before, during and after SEM imaging (Supplementary Figs. 22–24).
ther optimize this process. Electrochemical properties and diffusion
kinetics can be improved by design of optimized anode materials, as Electrochemical characterization. The reversible potentials for anodes
prepared with and without cobalt (Ni0.9Co0.1(OH)2 and Ni(OH)2, respectively)
is shown here by the addition of cobalt. These anodes can be further were measured as a function of their SOC. In this and all of the following
engineered to tolerate high pressures for pressurized hydrogen pro- electrochemical tests, each anode was connected as the working electrode to a
duction. This leads to applications for electromobility in transporta- potentiostat (Ivium-n-Stat multichannel Potentiostat/Galvanostat), with a nickel
tion refuelling stations, chemical synthesis and energy storage when foil/foam counter electrode and an Hg/HgO/1 M NaOH reference electrode in
coupled with a fuel cell system. We also show that our approach is 5 M KOH electrolyte at ambient temperature. Here, each anode was charged at
a constant current density of 20 mA cm−2 to SOCs between 10 and 90%. At each
tolerant to carbonate-containing electrolytes at moderate pH val- SOC, the current was interrupted and the OCP was measured for 10,000 s, allowing
ues, which are significantly less hazardous than concentrated KOH adequate time for the interface between the anode and electrolyte to equilibrate.
solution. This, furthermore, leads to the potential for CO2 reduc- Then, the anode was discharged at the same current to SOCs between 90 and 10%,
tion coupled with anode oxidation in a similar membrane-free cell, and at each point the current was interrupted and the OCP was measured again.
The reversible potential at each SOC was calculated from OCP versus log[t] plots,
with an added benefit of enhanced product separation by spatially
based on the method described in refs. 50,51 (Supplementary Fig. 25). Following this
and temporally separating CO2 reduction from oxygen generation46. experiment, the Ni0.9Co0.1(OH)2 anode was charged using constant applied current
Approaches like these open the door to electrolytic chemical pro- densities of 10, 20, 50 and 100 mA cm−2 in the same cell to the same SOCs between
duction with efficient energy utilization8. 10 and 90%, and the polarization potentials were measured as a function of the
SOC (Supplementary Fig. 25). Two additional similar anodes, with and without
cobalt, were used to preform cyclic voltammetry sweeps at a scan rate of 50 mV s−1
Methods (Supplementary Fig. 26).
Chemicals. Potassium hydroxide solutions were prepared using potassium
hydroxide flakes (AR; Bio-Lab) in deionized water. Carbonate buffer solutions were
prepared using potassium carbonate (minimum 99%; Spectrum) and potassium Parameters affecting the anode regeneration reaction. A commercial nickel
carbonate (99.5–101%; Sigma–Aldrich). Nitrate solutions for anode ECI were hydroxide battery anode (Batterix) was used to establish an efficient and reliable
prepared using nickel nitrate powder (99.9+% Ni; Strem Chemicals), cobalt nitrate charge regeneration measurement technique. These tests were carried out in 1 M
powder (≥98%; Sigma–Aldrich) and sodium nitrite powder (99.0%; Merck) in NaOH. The anode was charged at a constant current density of 10 mA cm−2 to a
deionized water and ethanol (Ethanol ABS AR 99.9%; Gadot). The NiFe solutions total charge of 5 mAh cm−2, and its final polarization potential was recorded. Then,
were prepared using nickel nitrate powder and ferrous chloride (iron(ii) chloride, the anode was discharged at a constant current density of 5 mA cm−2 and recharged
98%; Sigma–Aldrich). at 10 mA cm−2 to a total charge of 10 mAh cm−2, and its polarization potential was
recorded again. This was repeated in charging intervals of 5 mAh cm−2 up to a
Ni0.9Co0.1(OH)2 anode fabrication. All Ni0.9Co0.1(OH)2 anodes were prepared by total charge of 25 mAh cm−2. Next, the anode was discharged to the same charging
ECI27,47,48. Nickel foam substrates (Ni-4753; RECEMAT; 3 mm thickness; 400-µm points, and its discharging polarization potential was recorded at the end of each
pores; 95.2% porosity; 5,400 m2 m−3 specific surface) were cut into 1 cm × 2 cm discharging step. This experiment was repeated three times (Supplementary Fig.
pieces (Supplementary Fig. 2a). The substrates were dipped in acetone and ethanol, 4). Then, two methods for evaluating the regenerated charge density, Qregen, were
cleaned ultrasonically in isopropanol for 15 min, and then washed with deionized examined (Supplementary Fig. 5). In the first method (method 1), the anode
water. Before deposition, the substrates were vacuum immersed in the deposition underwent cycles of charging → regeneration → discharging steps. First, the anode
solution for 30 min by placing the substrates in a vacuum chamber (Struers was discharged at a constant current density of 5 mA cm−2 to a cutoff potential
Epovac) and flooding them with the deposition solution. The deposition solution of 1 VRHE. Then, the anode was charged at a current density (jcharge) of 10 mA cm−2
was composed of nickel nitrate, sodium nitrite and cobalt nitrate in an aqueous to an accumulated charge density, Qcharge = jcharge × t of 20 mAh cm−2. Next, the
mixture of ethanol in water, titrated to an initial pH of 4 with nitric acid. A portion anode was regenerated by immersing it in 1 M NaOH aqueous solution at 80 °C
of 1 cm × 1 cm of each nickel foam substrate was dipped in the deposition solution, for 32 min. Finally, the anode was discharged again at a constant current density
and the remaining portion was used as the electrical contact (Supplementary (jdischarge) of 5 mA cm−2 to the same cutoff potential of 1 VRHE while monitoring
Fig. 2b). Each substrate was connected as the cathode to a potentiostat (SP- the transferred charge density, Qdischarge = jdischarge × t. This cycle was repeated
150 potentiostat (BioLogic Science Instruments) or Ivium-n-Stat multichannel three times. In this method, the regenerated charge density was calculated as
Potentiostat/Galvanostat (Ivium Technologies)) and placed between two nickel Qregen = Qcharge − Qdischarge. In the second method (method 2), the anode was initially
foam pieces acting as anodes (Supplementary Fig. 3). Deposition was carried out discharged under the same conditions, to the same cutoff potential of 1 VRHE. Then,
at a constant temperature of 80 °C, by applying a cathodic current of 60 mA. After the anode was charged to a cumulated charge density of 20 mAh cm−2, reaching
deposition, each electrode was washed thoroughly with deionized water, then a final polarization potential of 1.52 VRHE. The final potential, Ef = 1.52 VRHE, was
aged in a 5 M KOH aqueous solution at 90 °C for 30 min. The proof-of-concept thus correlated with the final charge density of 20 mAh cm−2. Next, the anode was
experiment presented in Fig. 3 was carried out using a champion Ni0.9Co0.1(OH)2 regenerated under the same conditions as in method 1, but unlike method 1, the

792 Nature Energy | VOL 4 | SEPTEMBER 2019 | 786–795 | www.nature.com/natureenergy

Nature Energy Articles
anode was not discharged electrochemically following regeneration. This was and dried in a vacuum furnace at 40 °C until its weight stabilized. The anode’s
repeated three times. In this method, the regenerated charge density was evaluated weight gain from the Ni0.9Co0.1(OH)2 coating was 0.26 g. To evaluate the anode’s
as: Qregen = Qcharge. capacity, it was subjected to a series of charge/discharge cycles until its discharge
The same commercial nickel hydroxide battery anode was used to investigate capacity stabilized, yielding a capacity of 25 mAh cm−2 (90 C cm−2). Before the
the effects of regeneration conditions on the regeneration process. The anode E-TAC experiment presented in Fig. 3, the anode was completely discharged
was charged at a constant current density of 5 mA cm−2 to a cutoff potential of electrochemically at a current density of 5 mA cm−2.
1.42 VRHE, then charged again at a constant potential of 1.42 VRHE until a cutoff For the ten E-TAC cycles presented in Fig. 3, the anode was connected as the
current density of 2.5 mA cm−2 was reached. Following this step, the anode working electrode to a potentiostat (VersaSTAT 3; Princeton Applied Research)
was disconnected and placed in 5 M KOH electrolyte at different temperatures in a three-electrode configuration, with a 1 cm2 platinized nickel-coated stainless-
(T = 60, 70, 80 and 90 °C) for different time durations (t = 2, 4, 8, 16 and 32 min) steel mesh counter electrode and an Hg/HgO/1 M NaOH reference electrode in
to regenerate the active material. After each regeneration step, the anode was 5 M KOH electrolyte at ambient temperature (without active heating or cooling).
reconnected to the electrochemical cell and charged again. Qregen was calculated Additionally, an electrometer (Keithley 6514 System Electrometer) was used to
according to method 2. This procedure was repeated five times at each data point measure the cell voltage (that is, the potential difference between the anode and
(Supplementary Fig. 6). the cathode). The cell was operated at a constant current density of 50 mA cm−2
A comparative performance assessment between a pasted commercial nickel (for both the anode and the cathode) for 100 s, charging the anode by 5 C cm−2.
hydroxide battery anode (Batterix) and our electrochemically impregnated Then, the charged anode was disconnected from the first cell and placed in another
Ni0.9Co0.1(OH)2 anode was carried out in 1 M NaOH by charging the anodes at (second) cell filled with hot 5 M KOH electrolyte kept at a temperature of 95 °C
current densities of 20, 50 and 100 mA cm−2 until cutoff potentials of 1.44, 1.46 and by a thermostatic bath for 100 s. Then, it was taken out of the hot (second) cell
1.52 VRHE, respectively, and then discharging them at a current density of 5 mA cm−2 and reconnected as the working electrode in the cold (first) electrochemical cell.
until a cutoff potential of 0.92 VRHE. This was repeated three times for each current Finally, the cell was operated again at 50 mA cm−2 for 100 s. This cycle was then
density (Supplementary Fig. 12). repeated ten times, moving the anode from the cold cell to the hot cell and back to
the cold cell.
Capacitive and Faradaic current measurements. A Ni0.9Co0.1(OH)2 anode and a The effect of the current density on the E-TAC process was examined by a
NiFe LDH anode were compared according to the following. First, the anodes were series of E-TAC cycles carried out at different currents densities of 10, 25, 50, 100
subjected to cyclic voltammetry scans from 0.80–1.75 VRHE at a scan rate of 50 mV s−1 and 200 mA cm−2 (Supplementary Fig. 32). The cell voltage and the potential of
(Fig. 2a). Then, the anodes were subjected to a series of chronoamperometric a Ni0.9Co0.1(OH)2 anode relative to an Hg/HgO/1 M NaOH reference electrode
polarizations at potentials of 1.3–1.7 VRHE at 100 mV intervals. Each polarization test were measured simultaneously, using a standard anode with a regenerated charge
was carried out for 600 s while monitoring the anode’s current (Supplementary Fig. 27). of 3 C cm−2. Before cycling, the cell was operated at a constant current density of
The final current density was plotted against the polarization potential, and these 50 mA cm−2 until a cutoff potential of 1.43 VRHE was reached. During each hydrogen
Tafel plots are presented in Fig. 2c. The Ni0.9Co0.1(OH)2 anode was charged before the generation step, a charge of 3 C was transferred to the anode. Then, the charged
chronoamperometry tests to shorten the time required to reach the steady-state OER anode was disconnected and placed in a hot (95 °C) 5 M KOH electrolyte for
current. Finally, a fully discharged Ni0.9Co0.1(OH)2 anode was polarized at a constant 120 s for the regeneration step. This E-TAC cycle was repeated five times for each
potential of 1.48 VRHE for 2,300 s to evaluate the Faradaic efficiency of the anode current density. The test at a current density of 200 mA cm−2 was carried out under
charging reaction at this potential (Supplementary Fig. 27). slightly different conditions from the other tests. During the hydrogen generation
step, the temperature was kept at 34 ± 2 °C, while the temperature of the oxygen
Dissolved oxygen and gas chromatography measurements. For the dissolved generation step was elevated to 102 ± 2 °C.
oxygen measurements, a completely discharged Ni0.9Co0.1(OH)2 anode with The conversion to VRHE from the raw data measured
a nominal active area of 10 cm2 was placed in a hermetically sealed cell. The in VHg/HgO was done according to ref. 52: E°Hg/HgO(T) = E°Hg/
HgO(298 K) − 1.120 × 10  × (T − 298) – 3.388 × 10  × (T − 298) , where E is the
−3 −6 2 o
dissolved oxygen measurements were carried out in 0.5 M KOH solution
(200 ml) at ambient temperature. A dissolved oxygen meter (Hanna Instruments; reversible potential.
HI5421 DO/BOD/OUR/SOUR/Temperature Bench Meter) was submerged
in the electrolyte. The pH was kept below 14 to avoid damaging the dissolved Stability and cyclability tests. A (standard) Ni0.9Co0.1(OH)2 anode was initially
oxygen probe. The cell was purged with nitrogen gas (N2) for 30 min, followed by charged at a constant current density of 50 mA cm−2 to a cutoff potential of
chronoamperometric polarization of the anode at a constant potential of 1.48 VRHE. 1.54 VRHE. Then, the anode was disconnected from the power source and placed
The current and dissolved oxygen concentration were monitored during charging in a hot (95 °C) 5 M KOH solution for 60 s for the TAC step. After 60 s, the anode
to 600 C (60 C cm−2) with stirring (Supplementary Fig. 28). was reconnected to the potentiostat, and the cell was operated at a current density
To verify that no oxygen was produced during the hydrogen production step of 50 mA cm−2 at the anode for 60 s (3 C cm−2), or until the same cutoff potential of
of the E-TAC cycle, the anode was subjected to E-TAC cycles with charging steps 1.54 VRHE was reached. This cycle was repeated 100 times (Supplementary Fig. 34).
carried out at a constant current of 100 mA (10 mA cm−2) until a cutoff charge of
30 C (3 C cm−2), and the regeneration steps were carried out for 120 s in hot (95 °C) E-TAC water splitting in carbonate–bicarbonate electrolyte. A (standard)
5 M KOH solution (without the probe, so as not to damage it). The cycles were Ni0.9Co0.1(OH)2 anode was connected as the working electrode in a carbonate
repeated until stabilization was reached. Then, the electrochemical cell was sealed buffer solution containing 0.6 M K2CO3 and 0.4 M KHCO3 (pH measured as 10.6)
and purged with nitrogen gas (N2) for 30 min with stirring. Finally, the anode under ambient conditions. The cell was operated at a constant current density
was charged again at 100 mA for 30 C while the dissolved oxygen concentration of 25 mA cm−2 for 100 s (2.5 C cm−2) or until a cutoff potential of 1.53 VRHE was
was measured (Supplementary Fig. 29). It is noted that since the electrolyte reached. Then, the charged anode was disconnected and placed in a hot buffer
solution was less concentrated (0.5 M KOH) than in the E-TAC tests (5 M KOH), solution containing 0.6 M K2CO3 and 0.4 M KHCO3 kept at a temperature of
the overpotential for the charging reaction was significantly higher than in 85–90 °C by a thermostatic bath for 100 s. At the end of the second step, the anode
the E-TAC tests. Therefore, the current density was set to 10 mA cm−2 to avoid was taken out of the electrolyte and reconnected as the working electrode in the
reaching high potentials (>1.6 VRHE). As a control test, the same measurement electrochemical cell, where it was operated again at a constant current density of
was repeated with an uncoated nickel foam anode to perform water electrolysis 25 mA cm−2 for 100 s (2.5 C cm−2). This procedure was repeated ten times to show
at 100 mA (Supplementary Fig. 29). Following this test, the Ni0.9Co0.1(OH)2 anode stability during cycling (Supplementary Fig. 35).
was regenerated again in 5 M KOH solution at 95 °C for 120 s, then reconnected
to the electrochemical cell. The cell was purged with N2 for 30 min, followed by Data availability
chronoamperometric polarization of the anode at a constant potential of 1.48 VRHE. The data that support the plots within this paper and other findings of this study
The current and dissolved oxygen concentration were measured for a total charge are available from the corresponding authors upon reasonable request.
of 80 C (8 C cm−2) (Supplementary Fig. 30).
For the gas chromatography measurements, an Ni0.9Co0.1(OH)2 anode with a
nominal active area of 1 cm2 was placed in a hermetically sealed cell. Before the
Received: 19 January 2019; Accepted: 5 August 2019;
measurement, the electrolyte was purged with N2 for 60 min, and a sample of 1 ml Published online: 13 September 2019
was extracted from the cell with a syringe and injected into sealed 20 ml vials that
were pre-washed with N2. Then, a constant current of 50 mA (50 mA cm–2) was References
applied between the electrodes for 100 s, as in the proof-of-concept test presented 1. Barber, J. Photosystem II: the water splitting enzyme of photosynthesis and
in Fig. 3, and another 1 ml sample was extracted from the headspace of the sealed the origin of oxygen in our atmosphere. Q. Rev. Biophys. 49, e14 (2016).
cell and injected into a vial. A sample of 0.6 ml was taken from each vial and 2. Hunter, B. M., Gray, H. B. & Muller, A. M. Earth-abundant heterogeneous
analysed by gas chromatography using an Agilent 7890A gas chromatography water oxidation catalysts. Chem. Rev. 116, 14120–14136 (2016).
system (Supplementary Fig. 31). 3. Seh, Z. W. et al. Combining theory and experiment in electrocatalysis:
insights into materials design. Science 355, eaad4998 (2017).
E-TAC water-splitting tests in alkaline solution. A champion Ni0.9Co0.1(OH)2 4. Galloway, J. N. et al. Transformation of the nitrogen cycle: recent trends,
anode with a regenerated charge of 5 C cm−2 was washed with deionized water questions, and potential solutions. Science 320, 889–892 (2008).

Nature Energy | VOL 4 | SEPTEMBER 2019 | 786–795 | www.nature.com/natureenergy 793

Articles Nature Energy
5. Sheehan, S. W. et al. Commercializing solar fuels within today’s markets. 35. Chen, J., Bradhurst, D. H., Dou, S. X. & Liu, H. K. Nickel hydroxide as an
Chem 3, 3–7 (2017). active material for the positive electrode in rechargeable alkaline batteries. J.
6. Smolinka, T., Ojong, E. T. & Garche, J. in Electrochemical Energy Storage for Electrochem. Soc. 146, 3606–3612 (1999).
Renewable Sources and Grid Balancing 103–128 (Elsevier, 2015). 36. LeRoy, R. L. & Bowen, C. T. The thermodynamics of aqueous water
7. Rausch, B., Symes, M. D., Chisholm, G. & Cronin, L. Decoupled catalytic electrolysis. J. Electrochem. Soc. 127, 1954–1962 (1980).
hydrogen evolution from a molecular metal oxide redox mediator in water 37. Esswein, A. J., Surendranath, Y., Reece, S. Y. & Nocera, D. G.
splitting. Science 345, 1326–1330 (2014). Highly active cobalt phosphate and borate based oxygen evolving
8. Wallace, A. G. & Symes, M. D. Decoupling strategies in electrochemical water catalysts operating in neutral and natural waters. Energy Environ. Sci. 4,
splitting and beyond. Joule 2, 1390–1395 (2018). 499–504 (2011).
9. Symes, M. D. & Cronin, L. Decoupling hydrogen and oxygen evolution 38. Esposito, D. V. Membraneless electrolyzers for low-cost hydrogen production
during electrolytic water splitting using an electron-coupled-proton buffer. in a renewable energy future. Joule 1, 651–658 (2017).
Nat. Chem. 5, 403–409 (2013). 39. Millet, P. in Hydrogen Production: Electrolysis (ed. Godula-Jopek, A.) 63–116
10. Chen, L., Dong, X., Wang, Y. & Xia, Y. Separating hydrogen and oxygen (Wiley‐VCH, 2015).
evolution in alkaline water electrolysis using nickel hydroxide. Nat. Commun. 40. Guillet, N. & Millet, P. in Hydrogen Production: Electrolysis (ed. Godula-
7, 11741 (2016). Jopek, A.) 117–167 (Wiley‐VCH, 2015).
11. Landman, A. et al. Photoelectrochemical water splitting in separate oxygen 41. Seitz, L. C. et al. A highly active and stable IrOx/SrIrO3 catalyst for the
and hydrogen cells. Nat. Mater. 16, 646–651 (2017). oxygen evolution reaction. Science 353, 1011–1014 (2016).
12. Bode, H., Witte, J. & Dehmelt, K. Zur kenntnis der 42. Zhou, H. et al. Highly active catalyst derived from a 3D foam of Fe(PO3)2/
nickelhydroxidelektrode—I. Uber das nickel (ii)-hydroxidhydrat. Electrochim. Ni2P for extremely efficient water oxidation. Proc. Natl Acad. Sci. USA 114,
Acta 11, 1079–1087 (1966). 5607–5611 (2017).
13. Bode, H., Witte, J. & Dehmelt, K. Zur kenntnis der nickelhydroxidelektrode. 43. Dong, B. et al. Two-step synthesis of binary Ni–Fe sulfides supported on
II. Über die oxydationsprodukte von nickel(ii)‐hydroxiden. Z. Anorg. Allg. nickel foam as highly efficient electrocatalysts for the oxygen evolution
Chem. 366, 1–21 (1969). reaction. J. Mater. Chem. A 4, 13499–13508 (2016).
14. Mao, Z. & White, R. E. The self-discharge of the NiOOH/Ni(OH)2 electrode 44. Zhou, H. et al. Water splitting by electrolysis at high current densities under
constant potential study. J. Electrochem. Soc. 139, 1282–1289 (1992). 1.6 volts. Energy Environ. Sci. 11, 2858–2864 (2018).
15. Winsel, A. & Fischer, C. New apparatus for the measurement of the 45. Han, L., Dong, S. & Wang, E. Transition-metal (Co, Ni, and Fe)-based
self-discharge of the nickel hydroxide electrode. J. Power Sources 34, 331–338 electrocatalysts for the water oxidation reaction. Adv. Mater. 28, 9266–9291
(1991). (2016).
16. Diaz-Morales, O., Ferrus-Suspedra, D. & Koper, M. T. M. The importance of 46. Chen, C., Khosrowabadi Kotyk, J. F. & Sheehan, S. W. Progress toward
nickel oxyhydroxide deprotonation on its activity towards electrochemical commercial application of electrochemical carbon dioxide reduction. Chem 4,
water oxidation. Chem. Sci. 7, 2639–2645 (2016). 2571–2586 (2018).
17. Macdonald, D. D. & Challingsworth, M. L. Thermodynamics of nickel- 47. Subbaiah, T., Mallick, S. C., Mishra, K. G., Sanjay, K. & Das, R. P.
cadmium and nickel-hydrogen batteries. J. Electrochem. Soc. 140, 606–609 Electrochemical precipitation of nickel hydroxide. J. Power Sources 112,
(1993). 562–569 (2002).
18. Cox, N. et al. Electronic structure of the oxygen-evolving complex in 48. Zhu, W., Zhang, D. & Ke, J. Electrochemical impregnation and performance
photosystem II prior to O–O bond formation. Science 345, of nickel hydroxide electrodes with porous plaques of hollow nickel fibres. J.
804–808 (2014). Power Sources 56, 157–164 (1995).
19. Dau, H. et al. The mechanism of water oxidation: from 49. Batchellor, A. S. & Boettcher, S. W. Pulse-electrodeposited Ni–Fe (oxy)
electrolysis via homogeneous to biological catalysis. ChemCatChem 2, hydroxide oxygen evolution electrocatalysts with high geometric and intrinsic
724–761 (2010). activities at large mass loadings. ACS Catal. 5, 6680–6689 (2015).
20. Mcbreen, J. in Handbook of Battery Materials (eds Daniel, C. & Besenhard, J. 50. Conway, B. E. & Bourgault, P. L. The electrochemical behavior of the
O.) 149–168 (Wiley-VCH, 2011). nickel–nickel oxide electrode: part I. kinetics of self-discharge. Can. J. Chem.
21. Snook, Ga, Duffy, N. W. & Pandolfo, A. G. Evaluation of the effects of oxygen 37, 292–307 (1959).
evolution on the capacity and cycle life of nickel hydroxide electrode 51. Barnard, R., Randell, C. F. & Tye, F. L. Studies concerning charged nickel
materials. J. Power Sources 168, 513–521 (2007). hydroxide electrodes. I. Measurement of reversible potentials. J. Appl.
22. Oliva, P., Laurent, J. F. & Leonardi, J. Review of the structure and the Electrochem. 10, 109–125 (1980).
electrochemistry of nickel hydroxides and oxy-hydroxides. J. Power Sources 8, 52. Nickell, R. A., Zhu, W. H., Payne, R. U., Cahela, D. R. & Tatarchuk, B. J. Hg/
229–255 (1982). HgO electrode and hydrogen evolution potentials in aqueous sodium
23. Armstrong, R. D., Briggs, G. W. D. & Charles, E. A. Some effects of the hydroxide. J. Power Sources 161, 1217–1224 (2006).
addition of cobalt to the nickel hydroxide electrode. J. Appl. Electrochem. 18, 53. Bienvenu, G. Method for co-generation of electric energy and hydrogen. US
215–219 (1988). patent 8,617,766 (2013).
24. Ortiz, M. G., Castro, E. B. & Real, S. G. Effect of cobalt electroless 54. Chen, G. et al. Accelerated hydrogen evolution kinetics on NiFe‐layered
deposition on nickel hydroxide electrodes. Int. J. Hydrog. Energy 39, double hydroxide electrocatalysts by tailoring water dissociation active sites.
6006–6012 (2014). Adv. Mater. 30, 1706279 (2018).
25. Wang, X. et al. Surface modification and electrochemical studies of spherical 55. Pletcher, D. & Li, X. Prospects for alkaline zero gap water
nickel hydroxide. J. Power Sources 72, 221–225 (1998). electrolysers for hydrogen production. Int. J. Hydrog. Energy 36,
26. Wang, X., Luo, H., Yang, H., Sebastian, P. J. & Gamboa, S. A. Oxygen catalytic 15089–15104 (2011).
evolution reaction on nickel hydroxide electrode modified by electroless
cobalt coating. Int. J. Hydrog. Energy 29, 967–972 (2004).
27. Portemer, A., Delahaye-Vidal, A. & Figlarz, M. Characterization of active
Acknowledgements
The research leading to these results has received funding from the Israeli Ministry of
material deposited at the nickel hydroxide electrode by electrochemical
National Infrastructure, Energy and Water Resources, as well as the European Research
impregnation. J. Electrochem. Soc. 139, 671–678 (1992).
Council under the European Union’s Horizon 2020 research and innovation programme
28. Johnson, B. A., Ferro, R. E., Swain, G. M. & Tatarchuk, B. T. High surface
(grant agreement number 727606). The results were obtained using central facilities at
area, low-weight composite nickel fiber electrodes. J. Power Sources 47,
Technion’s Hydrogen Technologies Research Laboratory, created and supported by the
251–259 (1994).
29. Ng, P. K. & Schneider, E. W. Distribution of nickel hydroxide in sintered Nancy and Stephen Grand Technion Energy Program and the Adelis Foundation, and at
nickel plaques measured by radiotracer method during electroimpregnation. the N2–H2 alternative fuel laboratory donated by Ed Satell. G.S.G. also acknowledges the
J. Electrochem. Soc. 133, 17–21 (1986). support of the Arturo Gruenbaum Chair in Material Engineering. The authors thank
30. Becker, M. D., Garaventta, G. N. & Visintin, A. Pulse-current Y. Ein-Eli for reading the manuscript and providing useful suggestions for improving it,
electrodeposition for loading active material on nickel electrodes for and N. Haimovich for some of the heat balance calculations.
rechargeable batteries. ISRN Electrochem. 2013, 732815 (2013).
31. Jamesh, M. I. & Sun, X. Recent progress on Earth abundant electrocatalysts Author contributions
for oxygen evolution reaction (OER) in alkaline medium to achieve efficient H.D. conceived the conceptual idea. H.D. and A.L. designed the experiments and
water splitting—a review. J. Power Sources 400, 31–68 (2018). analysed the data. A.L. performed the main experiment presented in Fig. 4 of the
32. Meites, L. Handbook of Analytical Chemistry (McGraw-Hill, 1963). article, and most of the experiments presented in the Supplementary Information.
33. Licht, S. pH measurement in concentrated alkaline solutions. Anal. Chem. 57, S.W.S. designed and performed the experiments presented in Fig. 3 of the article, and
514–519 (1985). tested alternative electrolyte solutions. K.D.M., A.L., M.H., N.Y. and C.C. prepared
34. Oshitani, M., Takayama, T., Takashima, K. & Tsuji, S. A study on the electrodes. K.D.M., H.D. and D.A.G. performed the material characterizations.
the swelling of a sintered nickel hydroxide electrode. J. Appl. Electrochem. 16, Z.A., N.Y., N.G. and N.H. performed the complementary experiments presented in the
403–412 (1986). Supplementary Information. G.E.S. assisted in the experimental and system design. A.R.,

794 Nature Energy | VOL 4 | SEPTEMBER 2019 | 786–795 | www.nature.com/natureenergy

Nature Energy Articles
G.S.G., A.L., H.D., S.W.S. and D.A.G. wrote the manuscript. A.R. and G.S.G. supervised Additional information
and guided the entire project. Supplementary information is available for this paper at https://doi.org/10.1038/
s41560-019-0462-7.
Competing interests
H.D., A.L., G.E.S., A.R. and G.S.G. hold the following patent applications in relation to Reprints and permissions information is available at www.nature.com/reprints.
the E-TAC process and systems: Patent Cooperation Treaty international applications Correspondence and requests for materials should be addressed to A.R. or G.S.G.
EP3221493A1, WO2016079746A1, US20170306510A1 and JP2017526564A (2015),
and Israeli Patent Application number 258252 (2018). The above authors hold shares in Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in
H2Pro—a start-up company that aims to develop hydrogen generators based on the E- published maps and institutional affiliations.
TAC process. © The Author(s), under exclusive licence to Springer Nature Limited 2019

Nature Energy | VOL 4 | SEPTEMBER 2019 | 786–795 | www.nature.com/natureenergy 795