The role of individual nickel surface species in the hydrogen evolution reaction on nickel in alkaline electrolytes

Analysis of the literature data

We begin by closely examining the literature data in Fig. 1a (see also Supplementary Table 1 and Supplementary Note 1), noting that the literature lacks a standardized HER reporting protocol, with activities reported variably as specific (j ECSA , mA cm−2), absolute (j geo , mA cm−2) or mass activities (j mass , mA mg−1). To establish a fair baseline for Ni-based catalysts, current (I) at a given potential (E) should be normalized to the electrochemically active surface area (ECSA, cm2), and uncompensated resistance (IR u ) should be corrected. After ECSA normalization and ohmic drop correction (for details, see Supplementary Table 1, Supplementary Note 2 and Supplementary Fig. 1), two trends emerge from Fig. 1b.

First, most data points have shifted to substantially lower activities (that is, higher overpotentials), placing most previously reported high-activity values below the activity of pure metallic Ni measured in this study (black dashed line, Fig. 1b). This suggests that most of the extraordinary activities reported in the literature originate from the high surface area of the catalyst samples rather than their intrinsic activity improvement. Second, the overall spread of the reported data clearly decreased. However, the remaining spread likely arises from the inherent complexity of Ni surface chemistry, where multiple Ni species—such as NiO, NiH x , β-Ni(OH) 2 , α-Ni(OH) 2 and NiOOH—can coexist on the surface. In what follows, we try to put the literature data into context by examining and determining the role of individual Ni species on well-defined surfaces and establish a foundation/baseline for the development of high-performance Ni-based catalysts.

Preparation and characterization of well-defined Ni surfaces

To deconvolute the complexity of Ni surface chemistry and its role in HER electrocatalysis, we developed specific methods/protocols that allowed us to prepare four types of surfaces (for more details, see Supplementary Note 3), each with a different surface chemistry and a range of coverages:

Ni, unmodified surface—prepared either in ultrahigh vacuum (UHV) or by reductive annealing in H 2 /Ar, followed by transfer with a protective water droplet.

Ni, modified with H x (surface and/or subsurface)—prepared electrochemically by holding unmodified Ni samples at potentials more negative of 0 V versus reversible hydrogen electrode (RHE). Coverage was controlled by adjusting the potential and duration of the cathodic hold.

Ni, modified with Ni(OH) 2 —prepared by cycling the unmodified Ni samples between −0.15 V and 0.55 V versus RHE. Different coverages were achieved by varying the number of consecutive cycles.

Ni, modified with NiO—prepared by oxidative annealing in O 2 /Ar. Coverage was controlled by the annealing duration.

In the following text, these surfaces are denoted as Ni0, Ni (H) , Ni (OH) and Ni (O) , respectively. To confirm their surface chemistry and morphology, the surfaces were characterized by cyclic voltammetry (CV), XPS, atomic force microscopy (AFM) and scanning electron microscopy (SEM), as shown in Fig. 2 and Supplementary Fig. 2.

Fig. 2: Surface characterization of individual Ni surfaces using AFM, CV and XPS measurements. Full size image a–c, AFM images (1 µm × 1 µm) with corresponding profile line scans (bottom of the panel) for Ni0 (a), Ni (OH) (b) and Ni (O) (c). The colour scales in the AFM images indicate the surface height (z-axis). Marked lines in a–c indicate the path used for the profile line scans. d–g, Representative cyclic voltammograms of metallic Ni0 (d, black), Ni (H) (e, blue), Ni (OH) (f, green) and Ni (O) (g, red). A representative surface coverage of ~30% is shown for all modified Ni surfaces. The cyclic voltammogram for Ni0 is included in e–g for reference. All experiments were carried out in Ar-saturated 0.1 M KOH solution at a scan rate of 50 mV s−1. h–j, High-resolution XPS Ni 2p 3/2 spectra for Ni0 (h), Ni (OH) (i) and Ni (O) (j) surfaces, respectively. Ni(OH) 2 is characterized by a single prominent Ni(II) peak (i), while NiO shows a distinctive double-headed peak with a shoulder at more positive binding energy (j) (details in Supplementary Fig. 7). The small, overlapping green peak in j corresponds to Ni(OH) 2 , consistent with the colour scheme used throughout the paper, and originates from air exposure during transfer to the XPS chamber. Source data

Ni0 surfaces were prepared in UHV by Ar+ ion sputtering and annealing. The SEM image of the sample shows an extended surface with large, 10–200 µm-sized grains (Supplementary Fig. 2). Within individual grains, the surface was close to atomically flat, as seen in the AFM image and corresponding profile line scan in Fig. 2a. The XPS analysis of the surface shows that essentially all oxygenated species (Supplementary Fig. 3a,b) and other possible impurities (Supplementary Fig. 3a) have been removed with our preparation procedure, as only the prominent Ni 2p 3/2 signal at 852.5 eV is present, indicative of metallic Ni (ref. 32) (Fig. 2h). The sample was then transferred under vacuum to a transfer vessel and moved to a N 2 -filled glove box for electrochemical characterization. A typical cyclic voltammogram recorded on this surface, between −0.15 V and 0.55 V versus RHE, shows an oxidation and a reduction peak, associated with the formation and reduction of Ni(OH) 2 , respectively, as described in equation (1) (refs. 6,33,34) (Fig. 2d). It served as our reference cyclic voltammogram for comparison with other surfaces (as explained in Supplementary Note 3 and Supplementary Fig. 4, the same surface can be prepared without the use of UHV).

$$\text{Ni}+2{\text{OH}}^{-}\to \text{Ni}{(\text{OH})}_{2}+2{\text{e}}^{-}$$ (1)

$$\frac{1}{x}\text{Ni}{\text{H}}_{x}+{\text{OH}}^{-}\to \frac{1}{x}\text{Ni}+{\text{H}}_{2}\text{O}+{\text{e}}^{-}$$ (2)

We point out that, as explained below, irreversible changes begin to occur on the surface within this potential range. To avoid these changes, the surface should be kept below 0.30 V (refs. 35,36).

Ni (H) surfaces were prepared by holding the potential of the Ni0 electrode between −0.05 V and −0.40 V versus RHE. Such cathodic potential holds result in an additional anodic charge in the subsequent cyclic voltammogram compared with the Ni0 surface (blue dashed line versus black dashed line in Fig. 2e). Historically, this extra charge has been attributed to the oxidation of Ni hydrides formed at negative potentials6,33, as described by equation (2); however, the exact nature and location of these species in alkaline electrolytes remain unresolved17. Owing to their high reactivity and low concentration, these species cannot be reliably analysed by available spectroscopy methods, in contrast to acidic environments where diffraction and spectroscopic evidence is available37. Therefore, we relied on voltammetry for quantitative surface characterization. Comparison of the anodic peaks at 0.40 V in the cyclic voltammograms of Ni0 (black dashed line in Fig. 2e) and Ni (H) (blue dashed line in Fig. 2e) reveals up to 74 µC cm−2 of additional charge, corresponding to the oxidation of 0.29 monolayer of H. To determine whether this charge originates from surface-adsorbed hydrogen (H ad ), we performed a CO displacement experiment (Supplementary Fig. 5 and Supplementary Note 3), which showed no current indicative of H ad desorption. This suggests that hydrogen resides beneath the surface. The subsurface H species can be removed by holding the electrode potential at 0.30 V versus RHE, without introducing irreversible changes to the Ni surface.

As noted above, exposing the Ni surface to potentials above 0.30 V induces irreversible changes, forming a Ni(OH) 2 -modified surface. The Ni(OH) 2 formed during the anodic sweep (equation (1)) is only partially reduced during the cathodic sweep into the HER region, yielding the Ni (OH) surface with a slightly smaller cathodic peak charge than its anodic counterpart. Consequently, each successive scan shows diminishing voltammetric peaks (Supplementary Fig. 6). Figure 2f (green dashed line) shows the 30th scan, where metallic Ni availability is substantially reduced (by ~30%) relative to Ni0. The AFM image (Fig. 2b) shows the surface populated with ~0.7 nm-high 3D islands, consistent with in situ scanning tunnelling microscopy observations by Marcus’ group36. The XPS Ni 2p 3/2 spectrum (Fig. 2i) confirms a surface composition dominated by metallic Ni and Ni(OH) 2 , indicated by intense spectral features at 852.5 eV and 856.0 eV, respectively (see fitting parameters in Supplementary Table 2). The O 1s spectra (Supplementary Fig. 3c and Supplementary Table 3) further support this, with a dominant hydroxide peak at 531.1 eV. Along with reduced peak areas in Fig. 2f, we observe a shift to more negative potentials for the anodic peak and a broad hump before Ni oxidation—features occasionally reported in the literature34,38,39 and increasingly pronounced with cycling. Overall, the CV indicates that the surface chemistry is substantially altered by the presence of irreversible Ni(OH) 2 .

Finally, Ni (O) surfaces are prepared by thermal oxidation of Ni0 surfaces at 300 °C in a stream of Ar/O 2 gas. The AFM image in Fig. 2c reveals that the surface is covered with oxidation-derived islands, and XPS (Fig. 2j) confirms that this surface predominantly consists of NiO and metallic Ni. Specifically, the high-resolution Ni 2p 3/2 spectrum shows a prominent metallic Ni peak at 852.5 eV, along with a characteristic double-headed fingerprint feature, comprising a main peak at 854.0 eV and a shoulder at 855.7 eV, both attributed to Ni(II) species originating from NiO (ref. 32). Note that this region is often misinterpreted, leading to erroneous conclusions11,15,22,23,26,27 (Supplementary Fig. 7 and Supplementary Note 3). Furthermore, Ni (O) surfaces are characterized by an intense peak at 529.3 eV in the O 1s spectra, which is characteristic of metal oxides (Supplementary Fig. 3d). Interestingly, the cyclic voltammogram for Ni (O) shows substantially different behaviour compared with Ni (OH) . Although the anodic and cathodic peaks in the cyclic voltammogram (red dashed line in Fig. 2g) are clearly diminished, we do not observe a pronounced shift in their peak potential. No additional cyclic voltammogram features were observed.

Having successfully prepared and characterized four model surfaces, we now move to linking the individual Ni species to the HER catalytic activity.

Correlation between individual Ni species and HER activity

The general rate equation (3) (see Supplementary Note 2 for definitions) captures two principal pathways by which surface or subsurface species can influence HER kinetics in alkaline media: through the \((1-{\Theta }_{{\rm{ad}}})\) term, representing the availability of active sites on the Ni electrode surface, and through the \({\Delta G}_{{{\rm{H}}}_{2}{\rm{O}}}^{\circ \ddagger }\) term, which reflects the standard free energy of the transition state for water dissociation.

$$I=-{nFA}{k}^{\circ }\left(1-{{\Theta }}_{\mathrm{ad}}({\Delta G}_{{\rm{s}}}^{\circ },E)\right){{\rm{e}}}^{-\alpha F(E-I{R}_{{\rm{u}}})/{RT}}\left({{c}_{{{\rm{H}}}_{2}{\rm{O}}}}_{(x=0)}{{\rm{e}}}^{-{(\Delta G}_{{{\rm{H}}}_{2}{\rm{O}}}^{\circ \ddagger }+r{{\Theta }}_{\mathrm{ad}})/{RT}}\right)$$ (3)

To deconvolute these two effects, that is, active site availability and transition state energetics, we first compared the HER performance of Ni0 with Ni (O) , Ni (H) and Ni (OH) surfaces, each prepared with ~30% coverage of the respective species (for details on the calculation of surface coverages, see Supplementary Note 4 and Supplementary Fig. 8 and Supplementary Note 7 and Supplementary Fig. 15). Electrochemical polarization curves (Fig. 3a), combined with DFT-calculated water dissociation barriers (Fig. 3b, Supplementary Table 5 and Supplementary Data 1), allowed us to evaluate how individual species change \({\Delta G}_{{{\rm{H}}}_{2}{\rm{O}}}^{\circ \ddagger }\) and influence HER kinetics beyond simple surface blocking. \({E}_{{\rm{a}}({{\rm{H}}}_{2}{\rm{O}})}^{\ddagger }\) was calculated as a proxy for \({\Delta G}_{{{\rm{H}}}_{2}{\rm{O}}}^{\circ \ddagger }\) under the assumption that the entropic terms are similar for all studied transition states and therefore will not markedly affect the observed trends. We then extend this analysis to explore how individual species coverage \((1-{\Theta }_{{\rm{ad}}})\) affects the overall activity, offering a quantitative framework for linking surface composition to HER performance.

Fig. 3: Role of individual Ni (sub)surface species in HER kinetics in alkaline solutions. Full size image a, Polarization curves showing the effect of different Ni species: Ni(OH) 2 (dashed green) promotes HER, NiO (dashed red) functions as a spectator (poisoning) species, and NiH x (dashed blue) inhibits HER relative to unmodified Ni0 (solid black). Measurements were performed in Ar-saturated 0.1 M KOH at 50 mV s−1. Curves are corrected for IR u drop, and the activities correspond to the surfaces presented in Fig. 2, each containing ~30% coverage/concentration of the individual species. HER activity is reported as the overpotential at −10 mA cm−2 or the current density at −0.30 V (Supplementary Fig. 9). Exchange current densities and Tafel slopes are provided in Supplementary Fig. 11 and Supplementary Table 4. b, Transition state structures for H 2 O → OH + H on atomistic models of Ni0, Ni (O) , Ni (H) (H + H owing to stoichiometry) and Ni (OH) . Green, red and white spheres represent Ni, O and H atoms, respectively. The red frame highlights the dissociating water molecule, and black arrows indicate subsurface hydrogen (hydride). c, Bar graph showing HER specific activities (j spec ) at −0.30 V with individual data points overlaid. Error bars represent mean ± s.d. The number of independent experiments/electrode preparations was n = 3, 3, 4 and 6 for Ni (H) , Ni (O) , Ni0 and Ni (OH) , respectively. d, Schematic activation-energy (Eₐ) landscape illustrating how individual Ni species modulate the energetic barrier for water dissociation on Ni. Path width reflects active site availability, whereas mountain-pass elevation represents intrinsic activity; higher elevations correspond to slower HER kinetics. The red path illustrates site blocking (\(1-{\Theta }_{{\rm{ad}}}\)), whereas the blue and green paths indicate modifications to Ni electronic structure that increase or decrease \({\Delta G}_{{{\rm{H}}}_{2}{\rm{O}}}^{\circ \ddagger }\), respectively. The bottom panel shows a 2D contour projection of the 3D energy landscape. Source data

Assuming the modifying species reside on the surface, this would correspond to 30% less Ni sites available for the reaction, that is, \((1-{\Theta }_{{\rm{ad}}})\) = 0.7. Interestingly, only Ni (O) surfaces exhibit a 30% decrease in activity (that is, geometric current density j geo ) compared with Ni0 (7 mA cm−2 versus 10 mA cm−2 at −0.30 V), as shown in Fig. 3a and Supplementary Fig. 9a,c. This indicates unchanged intrinsic (specific) activity (that is, j spec ) on Ni (O) surfaces, as normalizing the measured activity by \((1-{\Theta }_{{\rm{ad}}})\) = 0.7 yields the same value of 10 mA cm−2 as for Ni0 (Fig. 3c, Supplementary Fig. 9b,d and Supplementary Note 5). This aligns with DFT calculations (Fig. 3b, Supplementary Table 5 and Supplementary Fig. 12), which show a negligible change in the water dissociation barrier on NiO-modified Ni(111) (\({E}_{{\rm{a}}({{\rm{H}}}_{2}{\rm{O}})}^{\ddagger }\) = 1.04 eV versus 1.02 eV for Ni (O) versus Ni0). It is also consistent with our observation in cyclic voltammograms of Ni (O) surfaces (Fig. 2g), which show diminishing anodic and cathodic peaks with no pronounced changes in the peak position, signalling no apparent changes to the properties of the remaining available Ni active sites. Polarization curves, CV and DFT data collectively show that NiO behaves as a true spectator species, only affecting the availability of Ni active sites \((1-{\Theta }_{{\rm{ad}}})\) for the HER, while not affecting the \({\Delta G}_{{{\rm{H}}}_{2}{\rm{O}}}^{\circ \ddagger }\) of the remaining active sites. This is in stark contrast with a substantial body of literature11,23,29,40,41,42,43,44,45 that suggests NiO is a promoter of HER on Ni (a more extensive reference list is provided in Supplementary Note 7).

Although the reduced HER rate on the Ni (O) surface can be attributed solely to the loss of accessible active sites \((1-{\Theta }_{{\rm{ad}}})\), this is not the case for the Ni (H) surface, for which a 30% surface coverage by H cannot account for the observed 2.5-fold decrease in activity (4 mA cm−2 at −0.30 V), as shown in Fig. 3a and Supplementary Fig. 9a,c. As discussed above, hydrogen is absorbed at more negative potentials to form subsurface NiH x , placing H atoms beneath the surface rather than blocking it. This implies that active site availability remains unchanged (\((1-{\Theta }_{{\rm{ad}}})\) = 1), and the drop in performance reflects a true decrease in intrinsic activity (Fig. 3c and Supplementary Fig. 9b,d). DFT calculations support this, showing that adsorbed hydrogen atoms can possibly penetrate into the subsurface, forming an absorbed hydrogen layer. While such a layer does not seem to have much effect on water dissociation in the absence of HER, it has a slightly inhibiting effect (increasing \({E}_{{\rm{a}}({{\rm{H}}}_{2}{\rm{O}})}^{\ddagger }\) by 0.04 eV) on it at higher HER rates (that is, higher intermediate coverages), presumably by altering the electronic structure of Ni0 (Fig. 3b and Supplementary Table 5).

Finally, we take a closer look at the Ni (OH) surface. Despite the 30% reduction in available Ni sites, the HER activity (that is, j geo ) increased 11-fold (110 mA cm−2 at −0.30 V) (Fig. 3a and Supplementary Fig. 9a,c). Clearly, then, the irreducible Ni(OH) 2 , which remains stable even in the HER potential region, must play a promoting role in the dissociation of water by lowering \({\Delta G}_{{{\rm{H}}}_{2}{\rm{O}}}^{\circ \ddagger }\), in agreement with prior findings on Ni(OH) 2 /Pt and similar systems46,47,48.

To better understand the mechanistic intricacies of HER activation on Ni (OH) surfaces, we compared four distinct scenarios of water dissociation: on Ni adjacent to adsorbed OH ad (Supplementary Fig. 13a); on Ni adjacent to a single-layered Ni(OH) 2 partial film (Fig. 3b and Supplementary Fig. 13b); directly on a Ni(OH) 2 film (Supplementary Fig. 14a); and on Ni metal atoms located on a Ni(OH) 2 film (Supplementary Fig. 14b). By closely examining these four scenarios by DFT, we gained several noteworthy insights, summarized below (for more details, see Supplementary Note 6). First, water dissociation on Ni(OH) 2 alone is extremely unfavourable, both thermodynamically and kinetically, with reaction (\(\Delta{E}_{{\rm{rxn}}}\)) and activation (\({E}_{{\rm{a}}({{\rm{H}}}_{2}{\rm{O}})}^{\ddagger }\)) energies of +1.09 eV and +1.64 eV, respectively (Supplementary Table 5 and Supplementary Fig. 14a). In addition, Ni(OH) 2 films with adsorbed Ni atoms, Ni clusters or Ni layers, such as those reported elsewhere in the literature21,24,28, all exhibit prohibitive bandgaps (>1.5 eV) (Supplementary Fig. 14b). This is unsurprising, as Ni(OH) 2 is an insulator and should hinder electron transfer to metallic Ni sites, thus suppressing the HER.

Conversely, the experimentally observed promoting effect arises when OH ad or Ni(OH) 2 partial films are adsorbed directly on conductive Ni metal. In both cases, water dissociation occurs on Ni, with nearby OH ad or Ni(OH) 2 facilitating the reaction while simultaneously reducing the number of accessible active sites on the Ni surface (Fig. 3b and Supplementary Fig. 13). We find that both species notably reduce the \({E}_{{\rm{a}}({{\rm{H}}}_{2}{\rm{O}})}^{\ddagger }\)—OH ad by 0.06 eV and Ni(OH) 2 by 0.13 eV—thus explaining the observed increase in activity on Ni (OH) surfaces (Fig. 3a and Supplementary Table 5). We hypothesize that Ni(OH) 2 is more likely present than OH ad , as suggested by the topography observed in AFM (Fig. 2b). DFT calculations reveal that Ni(OH) 2 clusters on Ni create a new, ~100-fold more active HER site composed of two Ni atoms adjacent to the Ni(OH) 2 cluster. By contrast, Ni(OH) 2 species located within the cluster, and not in direct contact with adjacent metallic Ni, exhibits catalytic behaviour akin to the Ni(OH) 2 film discussed above (that is, unfavourable kinetics for HER) and behaves as a blocking species. Thus, the suboptimal spatial distribution of Ni(OH) 2 substantially reduces the number of active interfacial sites. In Supplementary Fig. 10 and Supplementary Note 5, we provide an estimate of the interfacial surface area, which shows that only 0.10–0.26 of the \({\Theta }_{{{\rm{Ni}}({\rm{OH}})}_{2}}\) is actually involved in electrocatalysis. This explains why the overall 15-fold improvement in HER specific activity of the Ni (OH) surface (Fig. 3c) remains far below the ~100-fold improvement predicted by DFT.

Taken together, the results on the four studied surfaces show that the HER specific activity follows the trend Ni (OH) ≫ Ni0 ≈ Ni (O) > Ni (H) (Fig. 3c), unambiguously demonstrating the distinct effects of each species at the specified coverage (~30%): the inhibiting effect of NiH x , the promoting effect of Ni(OH) 2 and the site-blocking (spectator) effect of NiO, creating an \({E}_{{\rm{a}}}\) landscape, where multiple types of active sites on Ni often work in parallel in transforming H 2 O to H 2 , as illustrated in Fig. 3d. However, the role of these species at only one coverage does not paint the whole picture of the complex Ni surface (electro)chemistry. To obtain a more quantitative insight, we performed a detailed analysis of the correlation between HER activity and the coverage of individual Ni species.

Coverage–activity relationship of modified Ni surfaces

We first demonstrate that increasing subsurface H concentration (coverage) leads to a monotonic decrease in HER activity (Supplementary Fig. 15a and Supplementary Note 7). This is achieved by extending the duration or magnitude of the cathodic potential hold, as shown in the cyclic voltammograms in Supplementary Fig. 15b. The activity drop originates from changes in the electronic environment induced by subsurface H, as corroborated by DFT calculations (Supplementary Note 6 and Supplementary Table 5).

Much more intriguing is the effect of surface coverage in the case of NiO and Ni(OH) 2 , and the interplay between these two species. To obtain information about the chemical identity and spatial resolution of the surface species and their surface coverage, we used a combination of CV, time-of-flight secondary ion mass spectrometry (ToF-SIMS) and XPS for surface analysis.

In the case of Ni (O) surfaces, the NiO coverage (Θ NiO ) was varied between 25% and 98%. These values were obtained from the cyclic voltammograms shown in Fig. 4a (details of the evaluation of ECSA are provided in Supplementary Note 4 and Supplementary Fig. 8). Both XPS and ToF-SIMS results align well with the cyclic voltammograms. The high-resolution XPS Ni 2p 3/2 spectra (Supplementary Fig. 16 and Supplementary Note 7) demonstrate the dominance of the two main species, that is, Ni and NiO. The abundance of metallic Ni and NiO at the interface can be best visualized through the ToF-SIMS 3D profiling shown for each coverage in Fig. 4d (for details, see Supplementary Fig. 17 and Supplementary Note 7). Note that the availability of metallic Ni sites in the surface layer, tracked via the Ni 3 − fragment signal (black colour), clearly shows a monotonic decrease in intensity, while the NiO− fragment (red colour) shows the inverse behaviour. The corresponding HER activity on these surfaces spans more than two orders of magnitude (Fig. 4b and Supplementary Fig. 18) and also shows a monotonic decrease in activity (that is, j geo ) with increasing NiO coverage (Fig. 4c), further confirming that NiO behaves as a mere spectator across the entire coverage range.

Fig. 4: Quantification of the blocking effect as a function of NiO surface coverage. Full size image a, Cyclic voltammograms for selected NiO coverages provide insight into the ECSA (see Supplementary Fig. 18 for full surface coverage data). b, Polarization curves, corrected for IR u drop, demonstrate a decrease in HER activity with increasing NiO coverage, confirming its role as a blocking species. The polarization curves correspond to the cyclic voltammograms shown in a. c, Logarithmic scale representation of the blocking trend as a function of NiO surface coverage, showing a spread in HER activity spanning more than two orders of magnitude. Bar graph shows HER activities at −0.30 V with data points overlaid. Error bars represent mean ± s.d. The number of independent experiments/electrode preparations was n = 4, 3, 3, 3, 3, 3 and 3 for Θ NiO values of 0%, 25%, 49%, 76%, 90%, 96% and 98%, respectively. d, ToF-SIMS 3D images of the Ni interface for different NiO coverages. The relevant fragments, NiO− and Ni 3 −, are marked with arrows. Note that 3D images serve as approximate visual representations of the interface; the corresponding depth profiles with full details are provided in Supplementary Fig. 17. All ToF-SIMS images are 300 µm × 300 µm in the lateral X–Y dimensions. Source data

By contrast, Ni (OH) surfaces exhibit a volcano relationship between HER activity and Ni(OH) 2 coverage (\({\Theta }_{{{\rm{Ni}}({\rm{OH}})}_{2}}\); Fig. 5a–c, Supplementary Fig. 19 and Supplementary Note 7), in agreement with observations by Liang et al.49 and Oshchepkov et al.40. Although this trend could be attributed to exceeding an optimal \({\Theta }_{{{\rm{Ni}}({\rm{OH}})}_{2}}\) threshold, as demonstrated for Ni(OH) 2 /Pt system48,50, the reality is more complex when considering the surface film’s composition and structure, as revealed by XPS and ToF-SIMS. At low coverages (0–28%), where we observe a monotonic increase in HER activity, both XPS and ToF-SIMS techniques indicate that the surface composition is dominated by Ni(OH) 2 on top of metallic Ni (Fig. 5d and Supplementary Figs. 20 and 21). However, the HER activity starts to diminish at surface coverages above 28%, resulting in activities similar to those of Ni (O) surfaces at coverages close to 97% (Fig. 5c). Interestingly, beyond 28% coverage, the further growth of the Ni(OH) 2 is accompanied by the emergence of a considerable amount of NiO, which becomes predominant when the total coverage by surface species reaches 60% (Fig. 5d and Supplementary Fig. 20). More importantly, ToF-SIMS 3D profiling (Fig. 5d) clearly reveals that NiO is positioned between the metallic Ni and the Ni(OH) 2 film (for details, see Supplementary Fig. 21 and Supplementary Note 7). The structure of the mixed oxide/hydroxide film is consistent with the previous valuable work of Hoppe and Strehblow51, as well as Medway et al.52, who proposed a two-layer Ni(OH) 2 /NiO structure. The key consequence of this structural arrangement is the disruption of the critical Ni(OH) 2 /metallic Ni interface, identified as the active HER site, leading to its deactivation. Since NiO acts purely as a blocking species, at high coverages the modified surface begins to resemble a NiO-covered one despite continued Ni(OH) 2 layer growth (Fig. 5d). Overall, it appears that none of the oxygen-containing Ni species exhibit a volcano-type relationship on Ni alone. At lower Ni(OH) 2 coverage (\({\Theta }_{{{\rm{Ni}}({\rm{OH}})}_{2}}\) ≤ 28%), HER activity is primarily governed by the \({\Delta G}_{{{\rm{H}}}_{2}{\rm{O}}}^{\circ \ddagger }\). However, once Θ exceeds ~28%, the \((1-{\Theta }_{{\rm{ad}}})\) term dominates owing to increasing NiO presence. Thus, maximizing HER activity requires balancing surface site availability (\(\left(1-{{\Theta }}_{\mathrm{ad}}\right)\) term) with the energetics of the (Ni–Ni(OH) 2 –H 2 O ad ) complex (\({\Delta G}_{{{\rm{H}}}_{2}{\rm{O}}}^{\circ \ddagger }\) term).

Fig. 5: Evolution of HER activity as a function of Ni(OH) 2 surface coverage: from promotion to blocking. Full size image a, Cyclic voltammograms for selected Ni(OH) 2 coverages provide insight into the ECSA (see Supplementary Fig. 19 for full surface coverage data). b, Polarization curves, corrected for IR u drop, initially show a monotonic increase in HER activity with increasing Ni(OH) 2 coverage, indicating the promoting effect. At higher surface coverages (above Θ ≈ 30%), the formation of the NiO underlayer begins to dominate, leading to a pronounced blocking effect. The curves correspond to the cyclic voltammograms shown in a. c, Logarithmic representation of the volcano-shaped activity trend as a function of overlayer Ni(OH) 2 and underlayer NiO surface coverage, showing a spread in HER activity spanning more than three orders of magnitude. Blue ribbons on the bars represent the HER activity of Ni (O) extracted from Fig. 4c. Dashed arrows on the bars for coverages ≥79% indicate the increase in HER activity of Ni (OH) compared with Ni (O) at the same or similar surface coverage. Bar graph shows HER activities at −0.30 V with data points overlaid. Error bars represent mean ± s.d. The number of independent experiments/electrode preparations was n = 4, 3, 4, 6, 4, 4, 3, 3 and 3 for \({\Theta }_{{{\rm{Ni}}({\rm{OH}})}_{2}}\) values of 0%, 6%, 19%, 28%, 38%, 79%, 92%, 96% and 97%, respectively. d, ToF-SIMS 3D images of the Ni interface for different coverages of the duplex Ni(OH) 2 /NiO layer. The relevant fragments, NiO 2 H−, NiO− and Ni 3 −, are marked with arrows. Note that the 3D images are approximate visual representations of the interface; the corresponding depth profiles with full details are provided in Supplementary Fig. 21. All ToF-SIMS images are 300 µm × 300 µm in the lateral X–Y dimensions. Source data

Towards a unified picture of HER on Ni surfaces

We return to the long-standing challenge highlighted in Fig. 1: the wide scatter in reported HER activities for Ni, which persists even after normalization to ECSA and correction for uncompensated resistance. As shown in this work, this variability stems not from experimental noise, but from the fundamental fact that Ni surfaces are chemically heterogeneous, and often not well controlled. The electrochemical behaviour of Ni is governed by the dynamic interplay of surface and subsurface NiO, Ni(OH) 2 and NiH x species, whose coverage, distribution and identity critically influence HER kinetics.

By systematically studying four well-defined Ni surfaces, we have mapped the HER landscape across relevant surface chemistries. Polarization curves and comparative analysis (Fig. 6) reveal that HER activity can easily vary by 3–4 orders of magnitude depending on surface composition, offering a plausible explanation for the majority of the HER activity values reported across the published literature (Fig. 1). The intrinsic HER activity of metallic Ni at −0.30 V versus RHE is ~10 mA cm−2 (black dashed line, Fig. 1), serving as a reference point. Deviations below this line (that is, towards lower overpotentials) typically reflect the presence of Ni(OH) 2 , which acts as a promoter at low coverages by forming highly active interfacial sites. Values above the line indicate rate-limiting contributions from NiO or subsurface NiH x . Specifically, NiO acts as a passive spectator, blocking access to active sites, while NiH x suppresses activity by increasing the water dissociation barrier. At higher Ni(OH) 2 coverages, a 3D hydroxide film forms with embedded NiO, limiting interfacial contact and leading to a volcano-type dependence of activity on coverage. As a result, HER activity for Ni/Ni(OH) 2 surfaces spans from the green dashed line and upward (that is, towards higher overpotentials) in Fig. 1.

Fig. 6: Origin of the full range of HER activities on Ni. Full size image a, Trends in HER activity as a function of the nature and surface coverage of different Ni species. All polarization curves were corrected for IR u drop and are presented as mean values with a shaded error band (see b for details). b, Bar graph with overlaid data points showing the HER activities of individual Ni surfaces at −0.30 V. Error bars represent mean ± s.d. The number of independent experiments/electrode preparations was n = 4, 6, 3, 3, 3 and 3 for Ni0, Ni (OH) (30%), Ni (O) (30%), Ni (H) (52%), Ni (O) (98%) and Ni (OH) (97%), respectively. Source data