nature catalysis

Article https://doi.org/10.1038/s41929-024-01126-3

A strongly coupled Ru–CrOx cluster–cluster

heterostructure for efficient alkaline
hydrogen electrocatalysis

Received: 10 January 2023 Bingxing Zhang1,2, Jianmei Wang1, Guimei Liu2, Catherine M. Weiss3,
Danqing Liu1, Yaping Chen1, Lixue Xia4, Peng Zhou 5, Mingxia Gao1,
Accepted: 9 February 2024
Yongfeng Liu 1, Jian Chen6, Yushan Yan 3, Minhua Shao 2,7,
Published online: 11 March 2024 Hongge Pan 1,6 & Wenping Sun 1,8

Check for updates

Constructing well-defined heterostructure interfaces in catalysts is
an efficient strategy to break the so-called scaling relationships and to
accelerate the reactions involving multiple intermediates. Here a cluster–
cluster heterostructure catalyst composed of crystalline ruthenium
cluster and amorphous chromium oxide cluster is designed to realize
high-performance alkaline hydrogen electrocatalysis. The strongly coupled
cluster–cluster heterostructure interface induces a unique interfacial
interpenetration effect, which simultaneously optimizes the adsorption of
intermediates on each cluster. The resulting catalyst exhibits impressive
catalytic activities for the hydrogen oxidation reaction (exchange current
density of 2.8 A mg−1Ru) and the hydrogen evolution reaction (mass
activity of 23.0 A mg−1Ru at the overpotential of 100 mV) in alkaline media.
The hydroxide exchange membrane fuel cell delivers a mass activity of
22.4 A mg−1Ru at 0.65 V and outstanding durability with no voltage loss
over 105 h operation at 500 mA cm−2. The present work demonstrates
the superiority of cluster–cluster heterostructure interface towards the
development of advanced catalysts.

The renewable hydrogen cycle is recognized as the foundation of an platinum group metal (PGM)-free electrocatalysts5–9. However, the
eventual hydrogen society1–4. Compared with the proton exchange alkaline media also brings great challenges for HEMELs and HEMFCs
membrane-based elctrolysers and fuel cells, hydroxide exchange mem- because the reaction kinetics of alkaline hydrogen evolution reaction
brane electrolysers (HEMELs) and hydroxide exchange membrane fuel (HER) and hydrogen oxidation reaction (HOR) is around two orders of
cells (HEMFCs) are attractive alternative technologies towards the magnitude lower than that in the acidic environment10–13, resulting in
renewable hydrogen cycle. The working conditions of HEMELs and undesirable performance and high loading of precious PGM-based elec-
HEMFCs are less corrosive, which allows for using more cost-effective trocatalysts. Under this circumstance, exploring more cost-effective

School of Materials Science and Engineering, Zhejiang University, Hangzhou, P. R. China. 2Department of Chemical and Biological Engineering, The Hong
1

Kong University of Science and Technology, Kowloon, China. 3Department of Chemical and Biomolecular Engineering, University of Delaware, Newark,
DE, USA. 4State Key Laboratory of Silicate Materials for Architectures, International School of Materials Science and Engineering, Wuhan University
of Technology, Wuhan, P. R. China. 5School of Environment and Energy, Peking University Shenzhen Graduate School, Shenzhen, China. 6Institute of
Science and Technology for New Energy, Xi’an Technological University, Xi’an, P. R. China. 7Energy Institute, and Chinese National Engineering Research
Center for Control and Treatment of Heavy Metal Pollution, The Hong Kong University of Science and Technology, Kowloon, China. 8State Key Laboratory
of Clean Energy Utilization, Zhejiang University, Hangzhou, P. R. China. e-mail: [email protected]; [email protected]

Nature Catalysis | Volume 7 | April 2024 | 441–451 441

Article https://doi.org/10.1038/s41929-024-01126-3

non-platinum hydrogen electrocatalysts (for example, Ru) with high

a b
activity would play a crucial role in constructing high-performance
and practical hydrogen energy devices14–17.
Recently, heterostructured electrocatalysts regulated by interface
engineering have drawn intense attention in improving the catalytic
performance for various reactions18–22. Finely tailoring the interface
chemistry is conducive to modulate the geometric structure, elec-
tronic structure, interfacial bonding and/or lattice strain, and to create Typical supported catalysts
synergistic effect between different components at the interface23–27,
which have been demonstrated to substantially promote the reac-
tion kinetics and even durability. However, interface engineering is
still mainly limited to the typical supported heterostructured cata- Ru
lysts (Fig. 1a), in which the active metal species, ranging from single Cr
atoms to clusters to nanoparticles, are supported and modulated via
O
the metal–support interaction by the large-sized catalyst supports H
(for example, metal oxides)18,28–32. It has been widely demonstrated
that supports could play a vital role as co-catalysts for a variety of reac-
Interfacial penetration Cluster–cluster heterostructure
tions33,34, while the size-sensitive properties of the supports are rarely
investigated in typical supported catalysts, in which the large-sized
Fig. 1 | Schematic illustration. a, A common supported catalyst. b, The cluster–
support cannot be regulated much due to the relatively bulk nature.
cluster heterostructured catalyst.
Therefore, it would provide more possibilities for heterostructured
catalysts if the size of the co-catalysts is reduced to be comparable
with that of the metal species, generating cluster–cluster or particle–
particle heterostructures, where the exact role of the co-catalyst in High-angle annular dark-field scanning transmission electron
a reaction would be fully disclosed. More importantly, the strongly microscopy (HAADF–STEM) images show that most nanoclusters
coupled cluster–cluster interface may be capable of simultaneously in Ru–CrOx@CN present with obvious cluster–cluster interface
regulating the two clusters in a unique way, eventually forming abun- (Fig. 2a–c), which link Ru and CrOx clusters to form a Janus cluster–
dant highly active interfacial regions. cluster heterostructure (~2–3 nm, Supplementary Fig. 5). The d-spacing
In this Article, we propose to develop cluster–cluster hetero- of 2.1 Å can be assigned to the (002) crystal planes of Ru (Fig. 2b),
structure catalysts for accelerating the reactions involving multiple whereas the amorphous counterpart is deduced to be CrOx. The results
intermediates. As a proof-of-concept demonstration, a Janus clus- are consistent with the X-ray diffraction (XRD) patterns, where the
ter–cluster heterostructure catalyst is developed for catalysing the Ru–CrOx@CN and Ru@CN samples only display the weak diffraction
HER and HOR in alkaline media (Fig. 1b). The cluster–cluster hetero- peaks of Ru, but no Cr-based crystalline phases can be observed in both
structure is composed of crystalline Ru (c-Ru) cluster and amorphous Ru–CrOx@CN and CrOx@CN (Supplementary Fig. 7). The bright-field
CrOx (a-CrOx) cluster on nitrogen-doped carbon nanosheets (CN), high-resolution TEM (HR-TEM) images also reveal the cluster–cluster
which is denoted as Ru–CrOx@CN. Importantly, the Ru atoms from Ru heterostructure in Ru–CrOx@CN (Fig. 2d,e).
cluster can penetrate across the interface into CrOx due to the strong In addition, the energy-dispersive X-ray spectroscopy (EDS) ele-
cluster–cluster interactions (highlighted in the yellow circle of Fig. 1b). ment mapping images (Fig. 2f and Supplementary Fig. 8) show that Ru
The unique interpenetrated interface largely improves the interfacial and Cr species are concomitantly connected on the CN, confirming the
charge redistribution, resulting in more optimized adsorption energies presence of Ru–CrOx heterostructure. Electron energy loss spectros-
of the reaction intermediates on each cluster, and thus significantly copy (EELS) mapping images (Fig. 2g) clearly show that the crystalline
promotes the rate-determining Volmer step for alkaline hydrogen Ru cluster is bonded with the amorphous CrOx cluster. To further verify
electrocatalysis. This work could inspire the studies on the rational the structure, the line-scan EELS spectra were collected across the
design of advanced heterostructured catalysts for simultaneously cluster–cluster interface (Fig. 2h). The Ru signal intensity decreases
achieving high activity and stability in hydrogen electrocatalysis and when crossing from the crystalline cluster to the amorphous region
other catalysis applications. over the interface, while the Cr signal appears and increases accord-
ingly, suggesting the formation of strong cluster–cluster interactions at
Results the interface with Ru penetration into CrOx. At the cluster–cluster inter-
Materials synthesis and characterization face (Fig. 2i–k), the Ru M-edge spectroscopy shifts to higher energy,
The Ru–CrOx@CN catalyst with strong cluster–cluster interactions suggesting the increased oxidation state of Ru35, while the Cr species
was synthesized by pyrolysing Ru3+−Cr3+-impregnated CN (Supple- show a decreased oxidation state according to the red-shift of the nar-
mentary Fig. 1) at 400 °C in 10% H2/Ar atmosphere as schematically rowed Cr L-edge spectroscopy at the interface36. In addition, no obvious
illustrated in Supplementary Fig. 2, where the two adsorbed ions were O signal is detected in the metallic Ru cluster (Fig. 2h), while intense O
converted to Ru clusters and CrOx clusters owing to their largely dif- signal emerges at the interface (Fig. 2j). Notably, this synthesis method
ferent anti-reduction ability. The content of Ru and Cr in Ru–CrOx@ was successfully extended to construct some other metal oxide-based
CN is determined to be 5.5 and 3.8 wt%, respectively, by the inductively cluster–cluster heterostructures, which show a combination of crystal-
coupled plasma (ICP) atomic emission spectrometer (Supplementary line Ru cluster and crystalline MnOx or ZnO cluster (Ru–MnOx@CN or
Table 1 and Note 1), which keep consistent with those in the precursor Ru–ZnO@CN, Supplementary Figs. 9 and 10).
(Supplementary Fig. 1). Raman spectra show that Ru–CrOx@CN presents two much weaker
The Ru@CN and CrOx@CN samples were also synthesized using peaks at 260–360 cm−1 and ~530 cm−1 compared with those of Cr2O3
the same method without addition of the other metal precursor (Sup- (Fig. 3a), which is a typical character of amorphous CrOx cluster37. In
plementary Figs. 3–5). Notably, the synthetic conditions in terms of addition, the intensity ratio of the D band and the G band (ID/IG) of Ru–
synthesis temperature and Ru/Cr ratio were also optimized, and the Ru– CrOx@CN is 1.08 and is higher than that of Ru@CN (0.85), suggesting
CrOx@CN synthesized at 400 °C with Ru/Cr ratio of 1/1 was confirmed the formation of more distorted structure in CN that might be catalysed
to have the optimal morphology and structure (Supplementary Fig. 6). by Ru or CrOx during the pyrolysis process.

Nature Catalysis | Volume 7 | April 2024 | 441–451 442

Article https://doi.org/10.1038/s41929-024-01126-3

a b c d
0.21 nm

2 nm

e 2.5 nm

101 101

002
2 nm 0.5 nm 1 nm
[010]

f g h
3 2 1

Ru M2,3 OK Cr L2,3

Intensity
1 nm

CrOx cluster

Ru cluster

5 nm 400 450 500 550 600

Energy loss (eV)

i j OK
k Cr L
Ru M
Intensity

Intensity
Intensity

1 1 3

2 2 2
450 460 470 480 490 500 520 540 560 560 570 580 590

Energy loss (eV) Energy loss (eV) Energy loss (eV)

Fig. 2 | Structural characterizations of Ru–CrOx@CN. a,b, Representative images. The Ru clusters and CrOx clusters are circled by yellow and red
(a) and enlarged (b) HAADF–STEM images. The red circles highlight the colours, respectively. g, EELS mapping images. The different colours depict the
representavive Ru–CrOx clusters. Inset in b is the fast Fourier transform (FFT) following: green, Ru; red, Cr. h–k, The line scanning EELS spectra (h) and the
pattern with the [010] zone axis. The dashed line indicates the interface enlarged EELS spectra of Ru M-edge (i), O K-edge (j) and Cr L-edge (k) for
between crystalline Ru cluster and amorphous CrOx cluster. c, Inverse FFT different positions on one Ru–CrOx cluster. Insert in h is the scanning path in
image obtained by selecting the FFT patterns in b. d,e, Representative (d) and the STEM–HAADF image for EELS profile spectra extraction, with the labelled
enlarged (e) HR-TEM images. The Ru clusters and CrOx clusters are circled positions in i–k.
by blue and red colours, respectively. f, STEM–EDS elemental mapping

To distinguish the local structure of CrOx cluster from the Cr2O3 structure (XANES) spectra (Fig. 3d) further indicate that the Cr atoms
reference, soft X-ray absorption spectroscopy with total electron yield show an oxidation state of lower than +3 in Ru–CrOx@CN. Fourier
(TEY) mode was carried out. The Cr L-edge near-edge X-ray absorption transforms of the extended X-ray absorption fine-structure (FT–EXAFS)
fine-structure (NEXAFS) spectroscopy of Ru–CrOx@CN shows lower spectra (Fig. 3e) display that the second-shell Cr–O–Cr coordination
intensity and broader spectral feature compared to Cr2O3 (Fig. 3b), at 2.58 Å of Ru–CrOx@CN upshifts compared with that of CrOx@CN
indicating the higher electron density of Cr originated from electron (2.54 Å), which may be caused by the presence of more disordered
donation by Ru and the more disordered local structure of CrOx cluster, CrOx and Cr–O–Ru coordination with longer interatomic distance at
which is consistent with the STEM and EELS results. As for O K-edge the cluster–cluster interface. The detailed coordination structure for
NEXAFS spectra (Fig. 3c), compared with that of Cr2O3, the weaker Cr–O, Cr–O–Cr and Cr–O–Ru paths is obtained by the least-squares
peak at 531.4 eV and the pronounced shoulder feature at 529.4 eV for EXAFS curve fitting (Fig. 3f), which shows coordination numbers
Ru–CrOx@CN reflect the electron-rich O sites and highly amorphous of 4.0, 2.8 and 0.6, respectively (Supplementary Table 2), reveal-
character of CrOx cluster38, respectively. To further verify this result, ing a typical low-coordinated CrOx cluster structure with interfacial
the X-ray photoelectron spectroscopy (XPS) was performed and Cr–O–Ru interaction.
clearly reveals the strong cluster–cluster interactions induce electron For Ru K-edge XANES, the Ru adsorption edge position of Ru–
transfer from Ru cluster to CrOx cluster in Ru–CrOx@CN, resulting in CrOx@CN upshifts compared with Ru foil and Ru@CN (Fig. 3g), indicat-
electron-rich Cr and O sites (Supplementary Figs. 11–13). ing that the valence of Ru is higher than that of Ru@CN, which probably
To deeply explore the unique electronic structure and cluster– results from electron transfer from Ru to CrOx as evidenced by the NEX-
cluster interactions of Ru–CrOx@CN catalyst, X-ray absorption spec- AFS spectra and XPS results (Fig. 3d and Supplementary Figs. 11–13). In
troscopy was collected. The Cr K-edge X-ray absorption near-edge the Ru K-edge FT–EXAFS spectra, the Ru–Ru coordination at 2.45 Å is

Nature Catalysis | Volume 7 | April 2024 | 441–451 443

Article https://doi.org/10.1038/s41929-024-01126-3

a b c
Ru–CrOx@CN Ru-CrOx@CN 1s-3d
D G Cr L3 Cr2O3 1s-4sp

Ru@CN Cr L2

Intensity
Intensity

Intensity
Cr2O3

Cr(OH)3

570 580 590 520 530 540 550

0

00

0
0

00
20

00
40

60

Photon energy (eV) Photon energy (eV)

1, 6
1, 2

2,
Raman shift (cm–1)

d e f
1.2 Ru–CrOx@CN Raw
Fit
Normalized µ(E)

CrOx@CN

FT (k3x(k))
FT (k3x(k))
0.8 Cr foil
Cr2O3

0.4

0 5,996 5,998 6,000

5,980 6,000 6,020 6,040 0 2 4 6 0 2 4 6

Energy (eV) R (Å) R (Å)

g h i
1.2 Raw
Fit
Normalized µ(E)

FT (k3x(k))

FT (k3x(k))
0.8
Ru–CrOx@CN
Ru@CN
0.4 Ru foil
RuO2

0
22,100 22,150 22,200 0 2 4 6 0 2 4 6
Energy (eV) R (Å) R (Å)

j 2.0 2.0 10 2.5

Ru-CrOx@CN Ru@CN Ru foil RuO2
5 5 5 5
8 2.0
1.5 1.5
4 4 4 4
R + α (Å)

R + α (Å)

R + α (Å)

R + α (Å)
6 1.5
3 1.0 3 1.0 3 3 Ru–Ru
Ru–Ru Ru–Ru Ru–Ru 4 1.0
2 2 2 2
Ru–O 0.5 Ru–O 0.5 Ru–O
2 0.5
1 1 1 1
0 0 0 0
2 4 6 8 2 4 6 8 2 4 6 8 2 4 6 8

k (Å–1) k (Å–1) k (Å–1) k (Å–1)

Fig. 3 | Fine structural characterizations. a, Raman spectra. The red dashed- EXAFS fitting curves of Ru–CrOx@CN in R space. g, Ru K-edge XANES spectra.
lines highlight the similar Raman peak positions bewteen Ru-CrOx@CN and h, k3-weighted Ru K-edge FT–EXAFS spectra in R space. i, Ru K-edge EXAFS fitting
Cr2O3. b,c, Cr L-edge (b) and O K-edge (c) NEXAFS spectra. d, Cr K-edge XANES curves of Ru–CrOx@CN in R space. j, WT for the k3-weighted EXAFS of different
spectra. e, k3-weighted Cr K-edge FT–EXAFS spectra in R space. f, Cr K-edge Ru samples.

greatly weakened, but the Ru–O/N coordination at 1.56 Å is enhanced Electrochemical HOR and HEMFCs performance
compared with that of Ru@CN (Fig. 3h). The above results indicate that, The catalytic performance of the electrocatalysts for alkaline HOR
in addition to the interfacial interaction between Ru cluster and CN was firstly investigated using the rotating disk electrode method with
presented in Ru@CN (Supplementary Figs. 14–16), abundant Ru atoms a standard three-electrode system. Figure 4a shows the HOR polari-
penetrate into the CrOx cluster, forming a strongly coupled Ru–CrOx zation curves of different catalysts in H2-saturated 0.1 M KOH solu-
cluster–cluster interface in Ru–CrOx@CN, which is consistent with the tion. The anodic current increases much faster with increasing the
above EELS results. As compared with Ru@CN, the larger R value for overpotential on Ru–CrOx@CN than those on Pt/C and Ru@CN. The
Ru–Ru coordination in Ru–CrOx@CN may arise from the contribution calculated half-wave potential (E1/2) for Ru–CrOx@CN is 10 mV, which is
of Ru–O–Cr coordination. The detailed coordination number and bond much lower than Ru@CN (14 mV) and Pt/C (18 mV). The HOR activity of
length along with other EXAFS fitting results are shown in Supplemen- Ru–CrOx@CN is also much higher than those of other control samples
tary Table. 3, and additional Ru–O–Cr with coordination number of (Supplementary Fig. 17). These results indicate that the introduction
0.7 can also be fitted from the Ru K-edge EXAFS results. of appropriate and well-defined Ru–CrOx cluster–cluster interface
The wavelet transform (WT) of EXAFS (Fig. 3j) shows that Ru– can remarkably boost the HOR activity. The electrochemical active
CrOx@CN exhibits one intensity maximum value for Ru–Ru coordina- surface area (ECSA)-normalized current densities (Supplementary
tion at ~7.5 Å−1 in k space and 2.3 Å in R space, which are smaller and Fig. 18) reveal that Ru–CrOx@CN also exhibits the highest intrinsic
larger than those (~8.5 Å−1 and 2.2 Å) of Ru@CN, further suggesting HOR activity compared with the other two samples.
the Cr–O–Ru contribution at the cluster–cluster interface (Supple- The kinetic current density (jk) and exchange current density (j0)
mentary Note 2). of HOR/HER are then calculated by the Koutecky–Levich equation and

Nature Catalysis | Volume 7 | April 2024 | 441–451 444

Article https://doi.org/10.1038/s41929-024-01126-3

a b c 2,500 rpm
3
10 3
1,600 rpm
2 900 rpm
2

jk (mA cm–2)
1
j (mA cm–2)

j (mA cm–2)
Ru-CrOx@CN 400 rpm
1 0.8
Ru@CN 0.1 Slope: 4.61 cm2 mA–1 s–1/2

j (cm mA )
1

–1
Pt/C 0.6
0

2
CrOx@CN 0.01
0.4

–1
–1
0
0.001 0.06 0.09 0.12
–1/2 1/2
0.15
ω (s )
–2
0 0.05 0.10 0.15 –0.01 0 0.01 0 0.05 0.10 0.15

E (V versus RHE) E (V versus RHE) E (V versus RHE)

d jk,s (mA cm–2)

e f
6 1.2 1.00 25 This work

Mass activity (A mg–1metal)

H2/O2

Power density (W cm–2)

4 jk,m (A mg–1) 1.0 20
0.75
O2
E1/2 (mV)

Voltage (V)
16 0.8
12 Air 15
8 8
12 4 0.50 Ru/meso C
16 Ru 7Ni3/C
20 0.6 10
0.4 Pd–CeO2/C PtRu/C
Pt–RuO2
0.25 5 PtRu/N–C
1 0.4 PdIrRu/C PtRu/Mo2C–TaC
0.8 NiW
Pd–CeO2/C NiMo/KB NiCrNi–H2–NH3
2 0 CeOx–Pd/C Ni52Mo13Nb35
1.2 0.2 0
3 0 1 2 3 0.1 1 10
j0,m (A mg–1) j0,s (mA cm–2) Current density (A cm ) –2
Anode metal loadings (mg cm–2)

g h
20 1.0
Specific PPD (W mg–1metal)

H2/O2 –2
This work @500 mA cm
0.8
15
Voltage (V)

Ru/meso C Ru 7Ni3/C 0.6

10
Pd–CeO2/C
Pt–RuO2 PtRu/C 0.4
5 PdIrRu/C
PtRu/N–C
Pd–CeO2/C Ni–H2–NH3
0.2
CeOx–Pd/C NiMo/KB NiCr NiW
0 PtRu/Mo2C–TaC Ni52Mo13Nb35
0
0.1 1 10 0 25 50 75 100
–2
Anode metal loadings (mg cm ) Time (h)

Fig. 4 | Electrochemical HOR performance and HEMFCs performance. a, HOR at 90 °C and cathode humidifier temperature at 97 °C, back pressures were
polarization curves of different catalysts in H2-saturated KOH solution (0.1 M) symmetric at 200 kPag, H2 flow rate at 1.0 l min−1 and O2/CO2-free air flow rate at
with a scanning rate of 5 mV s−1 at 1,600 rpm. b, Calculated HOR kinetic current 2.0 l min−1. f,g, Comparisons of the mass activity (f) at 0.65 V and specific peak
densities (jk) versus potential plots recorded from a. c, HOR polarization curves power densities, with PPD, (g) normalized by the mass of anode metal of HEMFCs
of Ru–CrOx@CN at different rotating speeds (inset is Koutechy–Levich plot). for H2-O2 as feeding gas (detail shown in Supplementary Table 6). h, Long-term
d, Compasions of E1/2, mass- and ECSA-normalized activities of Ru–CrOx@ stability tests of the H2-O2 fuel cell at 500 mA cm−2 using the Ru–CrOx@CN
CN (red), Ru@CN (green) and Pt/C (black). e, Polarization and power density (0.055 mgRu cm−2) as the anode catalyst. Test conditions: cell temperature at
curves of HEMFCs with Ru–CrOx@CN (0.055 mgRu cm−2) in anode and Pt/C 80 °C, H2 flow rate at 0.5 l min−1 and O2 flow rate at 0.5 l min−1, symmetric back
(0.4 mgPt cm−2) in cathode. Test conditions: anode humidifier temperature pressures at 150 kPag.

Butler–Volmer fitting, indicating Ru–CrOx@CN shows the highest its much stronger capacity for hydroxyl adsorption (Supplementary
kinetic activity among the presented catalysts (Fig. 4b). The Koutecky– Fig. 20). Furthermore, Ru–CrOx@CN displays excellent durability
Levich plot for Ru–CrOx@CN is derived from the polarization curves without obvious decay for 40 h operation (Supplementary Fig. 21). The
collected at different rotating speeds (Fig. 4c), which shows a slope of HAADF–STEM, EDS mapping images, XPS spectra and EXAFS fitting
4.61 cm2 mA−1 s−1/2, close to the theoretical value (4.87 cm2 mA−1 s−1/2) results show that, after durability evaluation, Ru–CrOx@CN can main-
for a two-electron transfer process39. After normalizing jk and j0 tain the cluster–cluster heterostructure with the interfacial penetration
based on the Ru mass loading on the electrode (Fig. 4d), the mass and interaction nearly unchanged, except for slight agglomeration and
activity (13.76 A mg−1Ru) and j0,m (2.8 A mg−1Ru) of Ru–CrOx@CN are detachment (Supplementary Figs. 22–24).
14.5 and 3.6 times as high as those of Ru@CN, respectively (Supple- The high HOR mass activity of Ru–CrOx@CN is beneficial to deliver
mentary Table 4). The ECSA-normalized jk,s and j0,s of Ru–CrOx@CN high-performance HEMFCs with low loading of Ru. To this end, we
are 5.1 and 1.0 mA cm−2, which are 16 and 3.9 times those of Ru@CN, assembled a membrane electrode assembly (MEA) using commer-
respectively. Notably, the HOR activity of Ru–CrOx@CN is among the cial Pt/C as the cathodic oxygen reduction catalyst (0.4 mgPt cm−2)
best reported HOR catalysts evaluated under the similar conditions and Ru–CrOx@CN as the anodic HOR catalyst with a low loading of
(Supplementary Table 5). 0.055 mgRu cm−2. Figure 4e shows the polarization and power density
The durability and CO-tolerance capability of Ru–CrOx@CN were curves of the HEMFCs measured under both H2/O2 and H2/air (CO2-free)
further explored. The HOR current density of Ru–CrOx@CN only conditions. The cell delivers a peak power density (PPD) of 0.88 and
shows a small decay of 10% at 50 mV during 2,000 s in the presence of 0.68 W cm−2 with O2 and air as the cathodic feeding gas, respectively.
1,000 ppm CO, much lower than Pt/C (decay by 40%, Supplementary The cell performance is comparable with those of the state-of-the-art
Fig. 19). The notable CO-tolerance of Ru–CrOx@CN is attributed to HEMFCs40,41, even though the cell has the lowest loading of anode

Nature Catalysis | Volume 7 | April 2024 | 441–451 445

Article https://doi.org/10.1038/s41929-024-01126-3

a b 0 c 30
0

MA (A mg–1precious metal)
–4
30

Overpotential (V)
–100

Ru-CrOx@CN
–8

MA (A mg–1)
j (mA cm–2) 20

–12 20
–200

Ru@CN
Ru–CrOx@CN
–16

Pt/C
10
Ru@CN 10
–300
Pt/C –20
CrOx@CN
–24 0 0
–400
–0.2 –0.1 0 0.1 –0.16 –0.08 0
E (V versus RHE) E (V versus RHE)

d e 60 f
Ru–CrOx@CN
0.04 o-CoSe2|P

Tafel slope (mV dec–1)

Ru@CN Ru@CN
50 Ru@CQDs RuP 2–CPM 10
Overpotential (V)

–1 Ru@NC SLNP L-RP/C

57.0 mV dec Pt/C
0.03 Ru–Ni@Ni 2P–NRs
Ru 56Pd44 /C CoP/Ni5P4/CoP
40 1 Ru@C2N

TOF (s–1)
–1
46.0 mV dec Ru/OMSNNC Ru@NC MoNi4/MoO3-x
0.02 Co-NiS 2 NSs
This work Ru@MWCNT
0.1 P-Fe 3O4/IF
30 Ru@MWCNT
RuCo@NC Ru@GnP
0.01
Ru–Mo 2C@CNT 2DPC–RuMo
30.1 mV dec
–1
20 0.01 N,P-Mo 2C@C
RuCo ANSs
0 Ru 2P/WO3@NPC
10 0.001
0.6 0.8 1.0 1.2 1.4 –0.20 –0.15 –0.10 –0.05 0
0 10 20 30 40
–2
log [j(mA cm )] Overpotential (V) E (V versus RHE)

Fig. 5 | Electrochemical HER performance. a, HER polarization curves of deviation of three independent measurements. d, Tafel plots. e, Comparison
different catalysts in a nitrogen-saturated 1 M KOH electrolyte with a rotating of overpotential and tafel slope of different Ru-based catalysts for HER
speed of 1,600 rpm. b, Mass-normolized HER polarization curves of different shown in Supplementary Table 7. f, TOF values of different catalysts shown in
catalysts. c, Comparison of overpotential at 10 mA cm−2 and mass activity (MA) Supplementary Table 8.
at 100 mV of different catalysts for HER. The error bars represent the standard

catalyst. The cell shows a current density of 1.23 A cm−2 at the typical remarkable specific and mass activity, which are significantly supe-
operating potential of 0.65 V for automotive applications with O2 rior to those of Ru@CN and Pt/C catalysts (Fig. 5a,b). Specifically, the
as the cathodic feeding gas, which is comparable with the cell with overpotential to reach 10 mA cm−2 is only 7.0 mV for Ru–CrOx@CN
high-loading state-of-the-art PtRu/C anode catalyst (0.2 mgPtRu cm−2), (Fig. 5c), much lower than those for Ru@CN (24 mV) and Pt/C (34 mV).
and the performance even shows much superiority at higher oper- The mass activity of Ru–CrOx@CN (23.0 A mg−1) at the overpotential
ating potentials (Supplementary Fig. 25). The cell with low-loading of 100 mV is 2.1 and 23.0 times as high as that of Ru@CN (10.7 A mg−1)
PtRu/C anode catalyst (0.055 mg cm−2) was also tested, and shows and Pt/C (1.0 A mg−1), respectively. The Tafel slopes are determined to
much inferior power density to the one with Ru–CrOx@CN anode be 30.1, 46.0 and 57.0 mV dec−1 for Ru–CrOx@CN, Ru@CN, and Pt/C,
catalyst (Supplementary Fig. 26), further confirming the intrinsically respectively (Fig. 5d), which suggests the rate-determining step might
high activity of Ru–CrOx@CN for the fuel cell applications. Addition- switch from the sluggish Volmer step to Tafel step for Ru–CrOx@CN16.
ally, the performance of the cell with the Ru–CrOx@CN anode catalyst As can also be seen from Fig. 5e and Supplementary Table 7, it is clear
is less affected by the gas flow rate as well (Supplementary Fig. 27). that the alkaline HER activity of Ru–CrOx@CN is among the best of the
Furthermore, the activities and specific PPDs normalized by the PGM-based catalysts.
PGM mass are 22.4 A mg−1Ru and 16.1 W mg−1Ru under H2/O2 condition The intrinsic HER activity was finally evaluated by calculating the
(Fig. 4f,g), respectively, and 16.4 A mg−1Ru and 12.3 W mg−1Ru under H2/air turnover frequency (TOF) values according to the estimated number
(CO2-free) condition (Supplementary Fig. 27), respectively. The mass of Ru active sites. As shown in Fig. 5f, the Ru–CrOx@CN exhibits a
activities and specific PPDs are much higher than those of PtRu/C high TOF of 20.7 s−1 at an overpotential of 100 mV, which is higher
(6.15 A mg−1PtRu and 5.9 W mg−1PtRu under H2/O2 condition; 5.0 A mg−1PtRu than the other two catalysts and most reported HER catalysts (Sup-
and 4.6 W mg−1PtRu under H2/air (CO2-free) condition) and other cata- plementary Table 8). Additionally, negligible potential decay can be
lyst systems in reported HEMFCs (Fig. 4f,g, Supplementary Fig. 28 observed at the cathodic current density of 100 mA cm−2 for Ru–CrOx@
and Supplementary Table 6). Furthermore, in addition to the decent CN during 20 h chronopotentiometry test (Supplementary Fig. 29),
electrochemical performance, the Ru–CrOx@CN catalyst is also much confirming the robust durability for alkaline HER. The HAADF–STEM,
more cost-effective due to the absence of high-cost Pt element. EDS and XPS results of Ru–CrOx@CN after the durability test further
Notably, the cell with such a low loading of anode catalyst validate its stable structural integrity and chemical composition dur-
(0.055 mgRu cm−2) nearly shows no voltage loss after a 105 h operation ing the electrocatalysis processes (Supplementary Figs. 30 and 31).
at a constant current density of 500 mA cm−2 under the H2/O2 testing Overall, the high intrinsic HOR/HER activity and durability endow
condition (Fig. 4h and Supplementary Table 6), which is the most excel- Ru–CrOx@CN as a very promising candidate for practical HEMFCs
lent durability for HEMFCs reported to date (note: the reported best and HEMELs.
catalyst shows 4.4% voltage loss in 100 h with a loading of 0.2 mgRu cm−2
under similar operation conditions14). The exceptional durability of Mechanism investigation
the cell confirms the superior structural stability of the Ru–CrOx@ To reveal the intrinsic mechanism for the exceptional activity of Ru–
CN catalyst. CrOx@CN towards hydrogen electrocatalysis, comprehensive experi-
ments and theoretical simulations were performed. Firstly, the RuOx@
Electrochemical HER performance CN and RuOx–CrOx@CN samples were synthesized by annealing Ru@
The electrocatalytic performance of Ru–CrOx@CN for HER was inves- CN and Ru–CrOx@CN at 250 °C for 3 h in air (Supplementary Figs. 32
tigated in N2-saturated 1 M KOH solution. Ru–CrOx@CN exhibits and 33). The RuOx@CN and RuOx–CrOx@CN samples show much lower

Nature Catalysis | Volume 7 | April 2024 | 441–451 446

Article https://doi.org/10.1038/s41929-024-01126-3

a 8 b c 3.0
HOR 80 mV
60 mV
HER HOR 2.5
–3

Overpotential (mV)
6 .1 m 40 mV

Phase angle (°)

Vd 20 mV

log (R2 (Ω))

ec –1 0 mV 2.0
–2.
5m
4 Vd –20 mV
ec –1 –40 mV 1.5
–1.6 m HOR
V dec –1 –60 mV
2 –100 mV 1.0
HER –140 mV
–1.3 mV dec–1 HER
12.5 13.0 13.5 14.0 –1 0 1 2 3 4 5 –150 –100 –50 0 50 100
pH log (frequency (Hz)) Overpotential (mV)

d 1.0 e f

0.8

PDOS (eV−1)
0.6
jOD/jOH

HER HOR
0.4

Ru
0.2
Cr
O
0
–0.008 –0.004 0.004 0.008 –20 –15 –10 –5 0 5
Ru Cr O
E (V versus RHE) Energy (eV)

g h
0.2 0.2 *
OH adsorption energy (eV)
H adsorption energy (eV)

0
H*+OH*
0 0 2H*
Free energy (eV)

2H* + OH* 0.65 eV

–0.2 –0.2 –0.4

–0.4 –0.4
–0.8
Ru cluster 0.11 eV
Ru cluster H* + H2O
Ru/Ru1CrOx *+H2O
–0.6 Ru/Ru1CrOx –0.6
Ru/Ru2CrOx
Ru/Ru2CrOx 0.19 eV
–0.8 –0.8 –1.2
Reaction path

Fig. 6 | Mechanism investigation. a, The linear plots of overpotential at KOH and 0.1 M KOD solutions. e, The PDOS of Cr, Ru and O atoms of Ru/Ru2CrOx.
0.5 mA cm−2 versus pH for Ru–CrOx@CN (red) and Ru@CN (green). The error f, The differential charge density distributions between Ru cluster and Ru2CrOx
bars represent the standard deviation of three independent measurements. cluster. Colours indicate: grey, Ru; blue, Cr; and red, O. Yellow and olive represent
b, Bode phase plots of the in situ EIS on Ru–CrOx@CN. c, EIS-derived Tafel plots increased and decreased charge distributions, respectively. g, H adsorption
of the Ru–CrOx@CN (red) and Ru@CN (green) catalysts obtained from the energy on Ru and OH adsorption energy on CrOx for different models.
hydrogen adsorption resistance R2. d, The kinetic isotope effect value versus h, Calculated energy profile for hydrogen oxidation into H2O on Ru, Ru/Ru1CrOx
potential of Ru–CrOx@CN (red) and Ru@CN (green) catalysts in aqueous 0.1 M and Ru/Ru2CrOx. Asterisk (*) denotes the active site.

catalytic activities than Ru–CrOx@CN, suggesting the metallic Ru clus- indicate that the hydrogen adsorption involved in the low-frequency
ter is crucial for the Ru–CrOx cluster–cluster heterostructure towards region shows an obviously different behaviour in the HOR/HER region
fast hydrogen electrocatalysis30. between Ru–CrOx@CN and Ru@CN44. This process can be quantita-
The critical hydrogen adsorption and hydroxyl adsorption behav- tively analysed by fitting the Nyquist plots (Supplementary Fig. 38 and
iours involved in the Volmer step of HOR/HER were experimentally Tables 9 and 10), in which the second parallel component R2 reflects the
explored as well. The onset potential of CO stripping for Ru–CrOx@ hydrogen adsorption behaviour45. The EIS-derived Tafel plots of logR2
CN (0.27 V) is substantially lower than that of Ru@CN (0.45 V) (Supple- versus overpotential (Fig. 6c) show that Ru–CrOx@CN displays a much
mentary Fig. 34a), indicating the stronger binding ability with hydroxyl lower hydrogen adsorption resistance than Ru@CN in the whole over-
for Ru–CrOx@CN42. Hydrogen underpotential deposition (HUPD) potential region, implying its faster hydrogen adsorption and transfer
test shows that the hydrogen desorption peak position (0.187 V) for kinetics. Furthermore, the HOR/HER activity of Ru–CrOx@CN has an
Ru–CrOx@CN is lower than that of Ru@CN (0.197 V), revealing that H/D kinetic isotope effect of 0.65–0.85 in the micropolarization region
the Ru–H bonding is weakened in Ru–CrOx@CN (Supplementary (Fig. 6d), whereas Ru@CN possesses an H/D isotope effect of 0.25–0.35
Fig. 34b)43. (Supplementary Fig. 39), implying faster water dissociation and forma-
The Ru–CrOx@CN catalyst exhibits a slower change of HOR/HER tion kinetics on the Ru–CrOx interface for an accelerated Volmer step46.
catalytic activity than that of Ru@CN with increasing the pH value To further illustrate the mechanism, density functional theory
(Fig. 6a and Supplementary Fig. 35), suggesting Ru–CrOx shows faster (DFT) calculations were performed on the cluster–cluster hetero-
hydroxyl adsorption and transfer kinetics in alkaline media, which can structure model that was built on the basis of the experimental char-
mitigate the OH− concentration effect on the sluggish Volmer step. The acterizations. The cluster–cluster heterostructure consisting of a
operando electrochemical impedance spectroscopy (EIS) was carried ~0.8 nm Ru cluster and a ~0.6 nm CrOx cluster was built with increased
out to analyse the hydrogen adsorption kinetics at different potentials. penetration interface from Ru/Ru1CrOx to Ru/Ru2CrOx. During build-
The Bode phase plots (Fig. 6b and Supplementary Figs. 36 and 37) ing the cluster–cluster heterostructure model, the ab initio molecular

Nature Catalysis | Volume 7 | April 2024 | 441–451 447

Article https://doi.org/10.1038/s41929-024-01126-3

dynamics calculation was performed to achieve the stable models (Sup- 10.0 mg CN in a beaker (50 ml), followed by ultrasonic dissolution.
plementary Fig. 40). The projected density of states (PDOS) calculation The suspension was stirred continuously in a water bath at 70 °C with
reveals a strong orbital overlap among Ru 4d, Cr 2p and O 2p orbitals a magnetic stirring bar until the solvent was completely evaporated.
for Ru/Ru2CrOx (Fig. 6e), confirming the strongly coupled interaction Finally, the above precursor was calcined at 400 °C for 3 h with a heating
between Ru and CrOx clusters. The charge density difference mapping rate of 5 °C min−1 under 10% H2/Ar atmosphere, eventually obtaining
of Ru/Ru2CrOx (Fig. 6f and Supplementary Fig. 41) verifies the electron the final Ru–CrOx@CN sample.
transfer from Ru to Cr via Ru–O–Cr as well as the electron transfer from
Ru to O, consistent with the EELS, XPS, NEXAFS and XANES analyses. Synthesis of Ru@CN and CrOx@CN
The adsorption behaviours of H and OH on different sites are The control samples of Ru@CN and CrOx@CN were also synthesized
presented in Fig. 6g and Supplementary Figs. 42 and 43. Compared using the same method, and the metal precursors are RuCl3·xH2O
with the Ru cluster, the interfacial interaction can significantly reduce (2.7 mg) for Ru@CN and CrCl 3·6H 2O (2.7 mg) for CrO x@CN,
the hydrogen adsorption on Ru sites and increase OH adsorption on respectively.
CrOx clusters. In addition, Ru penetration across the Ru–CrOx interface
leads to more optimized hydrogen adsorption and hydroxyl adsorp- Synthesis of RuOx@CN and RuO2–CrOx@CN
tion on Ru cluster and on CrOx cluster, respectively, tremendously RuOx@CN and RuO2–CrOx@CN were obtained by calcining Ru@CN
facilitating the alkaline hydrogen electrocatalysis kinetics. The whole and Ru–CrOx@CN samples at 250 °C for 3 h in air, respectively.
reaction pathway from H2 to H2O is shown in Fig. 6h, according to the
most common Tafel–Volmer mechanism for alkaline HOR47,48. The Synthesis of Ru–MnOx@CN and Ru–ZnO@CN
rate-determining step is calculated to be the Volmer step as confirmed The Ru–MnOx@CN and Ru–ZnO@CN were synthesized using the same
by the experimental results (Fig. 6a–d). The reaction barriers are method as that for Ru–CrOx@CN synthesis. The metal precursors
calculated to be 0.65, 0.19 and 0.11 eV on Ru cluster, Ru/Ru1CrOx and are RuCl3·xH2O (0.01 mmol) and MnCl2 (0.01 mmol) for Ru synthesis
Ru/Ru2CrOx, respectively. The results clearly indicate that the hetero­ of MnOx@CN, and are RuCl3·xH2O (0.01 mmol) and Zn(NO3)2·6H2O
structure with increased Ru penetration could largely reduce the reac- (0.01 mmol) for synthesis of Ru–ZnO@CN.
tion barrier of the Volmer step in HOR.
Overall, the above results uncover that the strongly coupled clus- Characterization
ter–cluster interaction with a unique interfacial penetration effect is Powder XRD pattern was characterized by a RigakuDmax-rc X-ray dif-
capable of simultaneously optimizing the hydrogen adsorption on fractometer with Ni-filtered Cu Kα (λ = 1.5418 Å) radiation. TEM was
Ru and hydroxyl adsorption on CrOx, which is beneficial to break the carried out on a FEI Tecnai G2 F20 S-TWIN field emission microscope
scaling relationship and accelerate the elemental reaction kinetics. at 200 kV accelerating voltages. HAADF–STEM was obtained by a FEI
Consequently, the Volmer step is substantially accelerated, and acidic Titan G2 60–300 scanning/transmission electron microscope oper-
hydrogen electrocatalysis-like behaviours were obtained in alkaline ated under 300 kV with a probe spherical aberration corrector. XPS
medium, showing ultrahigh kinetic current density in HOR and low spectra were collected by VG Scientific ESCALab220i-XL spectrometer
Tafel slope in HER. It can be expected that the concept of cluster–cluster using Al Kα radiation equipped with 500 μm X-ray spot, in which the
heterostructured catalysts can be further extended to promote a vari- base pressure was controlled at 3 × 10−10 mbar. Raman spectra were
ety of catalytic reactions in both activity and stability, especially those carried out on a laser confocal fibre Raman spectrometer (Renishaw
involving multiple elemental reactions and reaction intermediates. inVia-Reflex). ICP mass spectometry analysis of the catalysts was car-
ried out on ICP optical emission spectrometry, Shimadzu ICPE-9800.
Conclusions NEXAFS spectra were collected at the photoemission end-station at
In summary, we developed a highly efficient cluster–cluster hetero- beamline BL10B in the National Synchrotron Radiation Laboratory
structured electrocatalyst composed of c-Ru cluster and a-CrOx cluster (NSRL) in Hefei, China. The XAFS experiments were performed on
for alkaline HOR and HER. The strong cluster–cluster interaction along beamline BL11B and BL14W1 at the Shanghai Synchrotron Radiation
with the well-defined interface plays a key role in substantially boost- Facility. Data analyses of XANES and EXAFS were performed by the
ing the alkaline hydrogen electrocatalysis performance. Experimental Athena and Artemis modules in the IFEFFIT software package49. The
results and DFT calculations demonstrate that the interfacial Ru–O–Cr spectra were corrected according to the first and largest peak of the
bonds regulated by Ru penetration effect largely enhance the HOR and first derivative of XANES, then normalized and background removed
HER activity. The unique interface could significantly reduce the energy with a k3-weighting and a Rbkg value of 1.0. The quantitative structural
barrier for breaking and forming the H–OH bond to accelerate water parameters were obtained by a least-squares curve parameter fitting
dissociation and formation for HER and HOR, respectively. The optimal for the EXAFS spectra by ARTEMIS module.
Ru–CrOx@CN is demonstrated to be among the best reported electro-
catalysts for alkaline hydrogen electrocatalysis, which also enables Electrochemical measurements
HEMFCs to deliver exceptional mass-normalized peak power density A standard three-electrode cell system coupled with electrochemical
and durability. This work paves a way for developing high-performance station (Chi 760E) was used for evaluating the electrocatalytic perfor-
cluster-scale heterostructured catalysts with broadened interface mance. Graphite rod, Hg/HgO electrode and glassy carbon electrode
chemistry towards various energy applications. (GCE) with a diameter of 5 mm, were used as the counter electrode,
reference electrode and working electrode, respectively. The GCE was
Methods used after consecutively polishing with Al2O3 slurry with particle size of
Synthesis of Ru–CrOx@CN 50 nm. The electrocatalyst ink was prepared by dispersing 2 mg catalysts
The CN support was synthesized by pyrolysing the mixture of urea in a 0.5 ml ethanol/H2O (3/1) solution containing of 5% Nafion solution.
(10.0 g) and citric acid (0.67 g). The pyrolysis process was performed Then, 10 µl of the electrocatalyst ink was dropped onto the GCE surface.
at 550 °C for 2 h and then 900 °C for 1 h with a heating rate of 3 °C min−1 HOR tests were conducted in a H2-saturated 0.1 M KOH aqueous solu-
under Ar atmosphere. For the preparation of Ru–CrOx@CN, 7.0 ml tion at a rotating speed of 1,600 rpm with a scanning rate of 5 mV s−1
ethanol solution containing 2.7 mg ruthenium(III) chloride hydrate and an 95% iR correction (i, current; R, solution resistance). HER tests
(0.01 mmol Ru, RuCl3·xH2O, 99.99%, 37 wt% Ru, Sinopharm Chemical were performed in a 1 M KOH aqueous solution with a rotating speed
Reagent) and 2.7 mg chromium(III) chloride hexahydrate (0.01 mmol of 1,600 rpm. The EIS was measured at different overpotentials with an
Cr, CrCl3·6H2O, 99.99%, Sinopharm Chemical Reagent) was mixed with amplitude voltage of 5 mV in a frequency range of 100 kHz to 100 mHz.

Nature Catalysis | Volume 7 | April 2024 | 441–451 448

Article https://doi.org/10.1038/s41929-024-01126-3

The kinetic current (jK) was calculated according to the Koutecky– 0.055 mgcat cm−2 with an ionomer to carbon ratio of 0.4 for the cathode
Levich equation: and anode, respectively.
The catalyst-coated membranes were then pretreated in 50 ml of
1/j = 1/jK + 1/jD (1)
1 M NaHCO3 for 1 h, changing the solution after 30 min, and then rinsed
in 100 ml of DIW for 5 min before cell assembly. The cell was assembled
where j is the measured current and jD represents the diffusion current. in 5 cm2 active area hardware from ElectroChem Inc. with Sigracet
The exchange current density (j0) was acquired by fitting jK with 22BB gas diffusion layers and 140 µm FEP gasket for both anode and
the Butler–Volmer equation: cathode. Once assembled, the cells were heated to 95 °C with an anode
humidifier temperature of 90 °C and a cathode humidifier temperature
jK = j0 [exp(αFη/RT ) − exp[−(1 − α)Fη/RT ]] (2)
of 97 °C with symmetric back pressure of 200 kPag. The fuel cells were
broken in under H2 and CO2-free air by increasing the current from 0 to
where R is the universal gas constant (8.314 J mol−1 K−1), F is the Faraday 1.4 A cm−2 in increments of 0.1 A cm−2 holding for 5 min. Once the cell
constant (96,485 C mol−1), α is the transfer coefficient and T is the was broken in, polarization curves were measured at four different
temperature (298.15 K). conditions (Supplementary Table 11). Three forward and reverse scans
were measured in increments of 0.1 A cm−2, holding for 10 s and using
Calculation of ECSA 0.35 V as the minimum voltage. The second forward and reverse scans
The Cu–UPD method was used to determine the ECSA. Cyclic volta­ were averaged for the overall cell performance.
metry (CV) curves were acquired in 0.5 M H2SO4 solution in the absence For the durability test, the cell was assembled in 5 cm2 active area
and presence of 5 mM CuCl2. To obtain monolayered Cu, the Pt/C and Fuel Cell Technologies hardware with single serpentine graphite flow
Ru-based catalysts were polarized at a deposition potential of 0.012 fields. Both anode and cathode gas diffusion electrodes were Sigracet
and −0.008 V (versus saturated calomel electrode) for 100 s, respec- 22BB with 140 µm FEP gaskets. Once assembled, the cell and humidi-
tively. All the curves were acquired via the CV technique at a scan rate fiers were heated to 80 °C with a symmetric back pressure of 150 kPa(g).
of 20 mV s−1. The ECSA was evaluated from the integral area of Cu–UPD The anode humidifier temperature was dropped to 79 °C while the cell
peaks (QCu) with the subtraction of the double layer in the linear sweep and cathode humidifier temperature remained 80 °C. The durability
voltammetry curves and a charge density of 420 μC cm−2 (Qs): test was performed at a constant current of 0.5 A cm−2 until the cell
voltage reached 0.4 V, with anode and cathode flows of 0.5 l min−1 H2
ECSA = QCu /Qs (3)
and 0.5 l min−1 O2, respectively.

Computational details for calculations

Calculation of the number of active sites (n) All the spin-polarized DFT calculations on this system were per-
The n was calculated by: formed with Vienna ab initio simulation package (VASP)50. The
projector-augmented wave method and Perdew–Burke–Ernzerhof51,52
n = QCu /2F (4)
functional within the generalized gradient approximation were applied
to describe ionic cores and exchange-correlation effects, respectively.
where F is the Faraday constant of 96,485 C mol−1 and 2 represents the A cut-off energy of 450 eV was adopted and the DFT-D3 method was
fact that two electrons are transferred. used to calculate van der Waals interactions51. The convergence crite-
ria for residual force and energy were set to 0.02 eV Å−1 and 1 × 10−5 eV,
Calculation of TOF respectively. The Brillouin zone was sampled at Gamma point for
The TOF (s−1) was determined according to: geometry optimization. The Ru/CrOx heterostructure consisting of
a ~0.8 nm Ru cluster (38 Ru atoms) and a ~0.6 nm CrOx cluster (14 Cr
TOF = I/(2nF ) = I/QCu (5)
atoms and 18 O atoms) was built. Ru/Ru1CrOx and Ru/Ru2CrOx were
built by replacing one or three Cr atoms with Ru atoms, respectively.
where I is the current (A) of the linear sweep voltammetry curve. Before performing free energy calculations, the models of Ru cluster,
Ru/Ru1CrOx and Ru/Ru2CrOx were built from the ab initio molecular
CO stripping test dynamics (AIMD) calculations at 300 K. The Andersen thermostat53
CO adsorption was firstly performed with amperometric i-t test at and NVE ensemble were adopted in the AIMD calculations. The time
0.05 V versus reversible hydrogen electrode (RHE) for 10 min in a step in each AIMD simulation was set as 2 fs, and 2,000 steps were
CO-saturated 0.1 M KOH aqueous solution. Then the electrode was adopted to achieve an equilibrated system. The system energy and
transferred into a N2-saturated 0.1 M KOH aqueous solution. The CO models before and after AIMD calculations were shown in Supple-
stripping current was obtained by CV in the potential range of 0–1.0 V mentary Fig. 39. The free energy calculation was performed by the
versus RHE at a scan rate of 20 mV s−1. computational hydrogen electrode model proposed by Nørskov and
co-workers54. The free energy of the proton–electron pair is equal to
HEMFCs test that of 1/2 H2(g) − 0.0592 × pH. The calculation is performed at the
Fuel cells were fabricated using 5 cm2 active area catalyst-coated mem- potential of 0 V (versus RHE). The free energy change was acquired by:
branes. The electrode inks were prepared in deionized water (DIW) and
ΔG = ΔE + EZPE − TΔS (6)
isopropyl alcohol (IPA), sonicated in an ice bath for 1 h and sprayed
onto a 17 µm poly(aryl piperidinium) polymer (PAP-TP-85) (Versogen,
Inc.) using an Iwata-Medea airbrush. The cathode ink was prepared where ΔE is the difference of electronic energy. ΔS is the entropy (S)
by mixing 10 mg of 40 wt% Pt/C (40 wt% Pt on Vulcan XC-72R from the variation, ΔEZPE is the change of zero-point energy (ZPE), T is the tem-
Fuel Cell Store), 0.05 ml of DIW, 72 mg of 5 wt% PAP-TP-85 dissolved perature (T = 298.15 K). The ZPE and S of the molecular in gas phase
in ethanol, and 1 ml of IPA. The anode ink consisting of 10 mg of cata- were taken from NIST55 database.
lyst and carbon, 0.05 ml DIW, 76 mg of 5 wt% PAP-TP-85 dissolved in
ethanol, and 1 ml of IPA. The inks were then sprayed onto the PAP-TP85 Data availability
membrane (ca. 17 µm) on either side. The resulting catalyst layers The data supporting the findings of this study are available within the
consist of 0.4 mgPt cm−2 with an ionomer to carbon ratio of 0.6 and article and its Supplementary Information files. All data is available

Nature Catalysis | Volume 7 | April 2024 | 441–451 449

Article https://doi.org/10.1038/s41929-024-01126-3

from the authors upon reasonable request. The atomic coordinates of 21. Feng, J.-X., Ye, S.-H., Xu, H., Tong, Y.-X. & Li, G.-R. Design
the computational models are given in Supplementary Data 1. and synthesis of FeOOH/CeO2 heterolayered nanotube
electrocatalysts for the oxygen evolution reaction. Adv. Mater. 28,
References 4698–4703 (2016).
1. Glenk, G. & Reichelstein, S. Economics of converting renewable 22. Zheng, X. et al. Multifunctional active-center-transferable
power to hydrogen. Nat. Energy 4, 216–222 (2019). platinum/lithium cobalt oxide heterostructured electrocatalysts
2. Staffell, I. et al. The role of hydrogen and fuel cells in the global towards superior water splitting. Angew. Chem. Int. Ed. 59,
energy system. Energy Environ. Sci. 12, 463–491 (2019). 14533–14540 (2020).
3. Stamenkovic, V. R., Strmcnik, D., Lopes, P. P. & Markovic, N. M. 23. Lao, M. et al. Platinum/nickel bicarbonate heterostructures
Energy and fuels from electrochemical interfaces. Nat. Mater. 16, towards accelerated hydrogen evolution under alkaline
57–69 (2017). conditions. Angew. Chem. Int. Ed. 58, 5432–5437 (2019).
4. Hu, Y. et al. Coplanar Pt/C nanomeshes with ultrastable oxygen 24. Zhang, J., Zhang, Q. & Feng, X. Support and interface
reduction performance in fuel cells. Angew. Chem. Int. Ed. 60, effects in water-splitting electrocatalysts. Adv. Mater. 31,
6533–6538 (2021). 1808167 (2019).
5. Firouzjaie, H. A. & Mustain, W. E. Catalytic advantages, 25. Wu, G. et al. In-plane strain engineering in ultrathin noble metal
challenges, and priorities in alkaline membrane fuel cells. nanosheets boosts the intrinsic electrocatalytic hydrogen
ACS Catal. 10, 225–234 (2020). evolution activity. Nat. Commun. 13, 4200 (2022).
6. Varcoe, J. R. et al. Anion-exchange membranes in electrochemical 26. Yang, Y. et al. Enhanced electrocatalytic hydrogen oxidation on
energy systems. Energy Environ. Sci. 7, 3135–3191 (2014). Ni/NiO/C derived from a nickel-based metal–organic framework.
7. Shi, L., Zhao, Y., Matz, S., Gottesfeld, S., Setzler, B. P. & Yan, Y. Angew. Chem. Int. Ed. 58, 10644–10649 (2019).
A shorted membrane electrochemical cell powered by hydrogen 27. Gerber, I. C. & Serp, P. A theory/experience description of
to remove CO2 from the air feed of hydroxide exchange support effects in carbon-supported catalysts. Chem. Rev. 120,
membrane fuel cells. Nat. Energy 7, 238–247 (2022). 1250–1349 (2020).
8. Li, C. & Baek, J.-B. The promise of hydrogen production from 28. Rong, H., Ji, S. & Zhang, J. et al. Synthetic strategies of supported
alkaline anion exchange membrane electrolyzers. Nano Energy atomic clusters for heterogeneous catalysis. Nat. Commun. 11,
87, 106162 (2021). 5884 (2020).
9. Meyer, Q., Zeng, Y. & Zhao, C. In situ and operando 29. Yang, F. et al. Boosting hydrogen oxidation activity of Ni in alkaline
characterization of proton exchange membrane fuel cells. media through oxygen-vacancy-rich CeO2/Ni heterostructures.
Adv. Mater. 31, 1901900 (2019). Angew. Chem. Int. Ed. 58, 14179–14183 (2019).
10. Kweon, D. H. et al. Ruthenium anchored on carbon nanotube 30. Zhou, Y. et al. Lattice-confined Ru clusters with high CO tolerance
electrocatalyst for hydrogen production with enhanced Faradaic and activity for the hydrogen oxidation reaction. Nat. Catal. 3,
efficiency. Nat. Commun. 11, 1278 (2020). 454–462 (2020).
11. Cong, Y., Yi, B. & Song, Y. Hydrogen oxidation reaction in alkaline 31. Zheng, X. B., Li, B. B., Wang, Q. S., Wang, D. S. & Li, Y. D. Emerging
media: From mechanism to recent electrocatalysts. Nano Energy low-nuclearity supported metal catalysts with atomic level
44, 288–303 (2018). precision for efficient heterogeneous catalysis. Nano Res. 15,
12. Sheng, W., Gasteiger, H. A. & Shao-Horn, Y. Hydrogen oxidation 7806–7839 (2022).
and evolution reaction kinetics on platinum: acid vs alkaline 32. Lin, J., Wang, W. & Li, G. Modulating surface/interface structure
electrolytes. J. Electrochem. Soc. 157, B1529 (2010). of emerging InGaN nanowires for efficient photoelectrochemical
13. Mahmood, J. et al. An efficient and pH-universal ruthenium-based water splitting. Adv. Funct. Mater. 30, 2005677 (2020).
catalyst for the hydrogen evolution reaction. Nat. Nano. 12, 33. Zhang, B., Chen, Y., Wang, J., Pan, H. & Sun, W. Supported
441–446 (2017). sub-nanometer clusters for electrocatalysis applications.
14. Xue, Y. et al. A highly-active, stable and low-cost platinum-free Adv. Funct. Mater. 32, 2202227 (2022).
anode catalyst based on RuNi for hydroxide exchange membrane 34. Zhang, B. et al. Atomically dispersed chromium coordinated
fuel cells. Nat. Commun. 11, 5651 (2020). with hydroxyl clusters enabling efficient hydrogen oxidation on
15. Zhang, J. et al. Engineering the near-surface of PtRu3 ruthenium. Nat. Commun. 13, 5894 (2022).
nanoparticles to improve hydrogen oxidation activity in alkaline 35. Lu, B. Z. et al. Ruthenium atomically dispersed in carbon
electrolyte. Small 17, e2006698 (2021). outperforms platinum toward hydrogen evolution in alkaline
16. Wan, C., Zhang, Z. & Dong, J. et al. Amorphous nickel hydroxide media. Nat. Commun. 10, 631 (2019).
shell tailors local chemical environment on platinum surface for 36. Cui, W. W. et al. Cr(III) Adsorption by cluster formation on
alkaline hydrogen evolution reaction. Nat. Mater. 22, 1022–1029 boehmite nanoplates in highly alkaline solution. Environ. Sci.
(2023). Technol. 53, 11043–11055 (2019).
17. Sheng, W., Zhuang, Z., Gao, M., Zheng, J., Chen, J. G. & Yan, Y. 37. Jagminas, A., Niaura, G. & Žalnėravičius, R. et al. Laser light
Correlating hydrogen oxidation and evolution activity on induced transformation of molybdenum disulphide-based
platinum at different pH with measured hydrogen binding energy. nanoplatelet arrays. Sci. Rep. 6, 37514 (2016).
Nat. Commun. 6, 5848 (2015). 38. Gago, R., Vinnichenko, M., Hübner, R. & Redondo-Cubero, A.
18. Zhao, G., Rui, K., Dou, S. X. & Sun, W. Heterostructures for Bonding structure and morphology of chromium oxide films
electrochemical hydrogen evolution reaction: a review. grown by pulsed-DC reactive magnetron sputter deposition.
Adv. Funct. Mater. 28, 1803291 (2018). J. Alloy. Compd 672, 529–535 (2016).
19. Zhao, G., Li, P., Cheng, N., Dou, S. X. & Sun, W. An Ir/Ni(OH)2 39. Zhuang, Z. et al. Nickel supported on nitrogen-doped carbon
heterostructured electrocatalyst for the oxygen evolution nanotubes as hydrogen oxidation reaction catalyst in alkaline
reaction: breaking the scaling relation, stabilizing iridium(V), and electrolyte. Nat. Commun. 7, 10141 (2016).
beyond. Adv. Mater. 32, 2000872 (2020). 40. Singh, R. K. et al. Synthesis of CeOx‐decorated Pd/C catalysts
20. Liang, Q. et al. Interfacing epitaxial dinickel phosphide to 2D by controlled surface reactions for hydrogen oxidation in
nickel thiophosphate nanosheets for boosting electrocatalytic anion exchange membrane fuel cells. Adv. Funct. Mater. 30,
water splitting. ACS Nano 13, 7975–7984 (2019). 2002087 (2020).

Nature Catalysis | Volume 7 | April 2024 | 441–451 450

Article https://doi.org/10.1038/s41929-024-01126-3

41. Cong, Y., Chai, C., Zhao, X., Yi, B. & Song, Y. Pt0.25Ru0.75/N–C (grant no. LZ22B030006), the Research Grant Council of the Hong
as highly active and durable electrocatalysts toward alkaline Kong Special Administrative Region (C6011-20G), the Innovation and
hydrogen oxidation reaction. Adv. Mater. Inter. 7, 2000310 (2020). Technology Commission of the Hong Kong Special Administrative
42. Shen, L.-f et al. Does the oxophilic effect serve the same role Region (no. ITC-CNERC14EG03). B.Z. acknowledges support
for hydrogen evolution/oxidation reaction in alkaline media? from China Postdoctoral Science Foundation (no. 2021M692757),
Nano Energy 62, 601–609 (2019). the National Postdoctoral Program for Innovative Talents
43. Ramaswamy, N., Ghoshal, S., Bates, M. K., Jia, Q., Li, J. & (no. BX2021251) and Zhejiang Provincial Natural Science Foundation
Mukerjee, S. Hydrogen oxidation reaction in alkaline media: (no. LQ23E010005). The authors thank beamline BL11B and BL14W1
relationship between electrocatalysis and electrochemical of the Shanghai Synchrotron Radiation Facility (SSRF) for providing
double-layer structure. Nano Energy 41, 765–771 (2017). the XAFS beamtime, and Photoemission End-Station (BL10B)
44. Lu, Y. et al. Tailoring competitive adsorption sites by at the National Synchrotron Radiation Laboratory (NSRL) for
oxygen-vacancy on cobalt oxides to enhance the electrooxidation NEXAFS beamtime.
of biomass. Adv. Mater. 34, 2107185 (2022).
45. Li, J. et al. A fundamental viewpoint on the hydrogen spillover Author contributions
phenomenon of electrocatalytic hydrogen evolution. B.Z. performed the catalyst development, characterizations
Nat. Commun. 12, 3502 (2021). and performance evaluation. J.W., D.L., Y.C. and J.C. helped in
46. Zhang, T., Yuan, B., Wang, W., He, J. & Xiang, X. Tailoring *H electrocatalysis experiments and material characterizations.
intermediate coverage on the CuAl2O4/CuO catalyst for enhanced G.L. and C.M.W. helped in fuel cell test. L.X. and P. Z. helped in DFT
electrocatalytic CO2 reduction to ethanol. Angew. Chem. Int. Ed. calculations. HP., M.G., Y.L., Y.Y., M.S., W.S. and B.Z. analysed the data.
62, e202302096 (2023). All the authors contributed to the overall scientific discussions and
47. Zhao, G., Jiang, Y., Dou, S.-X., Sun, W. & Pan, H. Interface edited the manuscript. W.S., H.P. and B.Z. conceived the idea and
engineering of heterostructured electrocatalysts towards co-wrote the paper.
efficient alkaline hydrogen electrocatalysis. Sci. Bull. 66,
85–96 (2021). Competing interests
48. Zhu, S., Qin, X. & Xiao, F. et al. The role of ruthenium in improving The authors declare no competing interests.
the kinetics of hydrogen oxidation and evolution reactions of
platinum. Nat. Catal. 4, 711–718 (2021). Additional information
49. Ravel, B. & Newville, M. ATHENA, ARTEMIS, HEPHAESTUS: data Supplementary information The online version contains supplementary
analysis for X-ray absorption spectroscopy using IFEFFIT. material available at https://doi.org/10.1038/s41929-024-01126-3.
J. Synchrotron Radiat. 12, 537–541 (2005).
50. Kresse, G. & Furthmüller, J. Efficient iterative schemes for ab initio Correspondence and requests for materials should be addressed
total-energy calculations using a plane-wave basis set. Phys. Rev. B to Hongge Pan or Wenping Sun.
54, 11169–11186 (1996).
51. Perdew, J. P., Burke, K. & Ernzerhof, M. Generalized gradient Peer review information Nature Catalysis thanks Ligang Feng,
approximation made simple. Phys. Rev. Lett. 77, 3865–3868 (1996). Marion Giraud and the other, anonymous, reviewer(s) for their
52. Grimme, S. Semiempirical GGA-type density functional contribution to the peer review of this work.
constructed with a long-range dispersion correction. J. Comput.
Chem. 27, 1787–1799 (2006). Reprints and permissions information is available at
53. Andersen, H. C. Molecular dynamics simulations at constant www.nature.com/reprints.
pressure and/or temperature. J. Chem. Phys. 72, 2384 (1980).
54. Nørskov, J. K. et al. Origin of the overpotential for oxygen Publisher’s note Springer Nature remains neutral with regard to
reduction at a fuel-cell cathode. J. Phys. Chem. B 108, jurisdictional claims in published maps and institutional affiliations.
17886–17892 (2004).
55. Atkins, P., De Paula, J. & Keeler, J. Atkins’ Physical Chemistry Springer Nature or its licensor (e.g. a society or other partner) holds
(Oxford University Press, 2017). exclusive rights to this article under a publishing agreement with
the author(s) or other rightsholder(s); author self-archiving of the
Acknowledgements accepted manuscript version of this article is solely governed by the
This work was financially supported by the National Key Research terms of such publishing agreement and applicable law.
and Development Program of China (no. 2022YFB4002503), the
National Natural Science Foundation of China (no. 92261119 and © The Author(s), under exclusive licence to Springer Nature Limited
52171224), the Natural Science Foundation of Zhejiang Province 2024

Nature Catalysis | Volume 7 | April 2024 | 441–451 451