Transition metal dichalcogenides as catalysts for the hydrogen evolution reaction: The emblematic case of “inert” ZrSe2 as catalyst for electrolyzers

Abstract

The development of earth-abundant electrocatalysts (ECs) operating at high current densities in water splitting electrolyzers is pivotal for the widespread use of the current green hydrogen production plants. Transition metal dichalcogenides (TMDs) have emerged as promising alternatives to the most efficient noble metal ECs, leading to a wealth of research. Some strategies based on material nanostructuring and hybridization, introduction of defects and chemical/physical modifications appeared as universal approaches to provide catalytic properties to TMDs, regardless of the specific material. In this work, we show that even a theoretically poorly catalytic (and poorly studied) TMD, namely zirconium diselenide (ZrSe₂), can act as an efficient EC for the hydrogen evolution reaction (HER) when exfoliated in the form of two-dimensional (2D) few-layer flakes. We critically show the difficulties of explaining the catalytic mechanisms of the resulting ECs in the presence of complex structural and chemical modifications, which are nevertheless evaluated extensively. By doing so, we also highlight the easiness of transforming 2D TMDs into effective HER-ECs. To strengthen our message in practical environments, we report ZrSe₂-based acidic (proton exchange membrane [PEM]) and alkaline water electrolyzers operating at 400 mA cm^–2 at a voltage of 1.88 and 1.92 V, respectively, thus competing with commercial technologies.

INTRODUCTION

The scale-up of the production of green hydrogen from renewable energies represents a cornerstone of a shift away from reliance on fossil fuels to reach carbon neutrality,^[1] that is, the achievement of net-zero CO₂ emissions,^[2] or even contribute to negative CO₂ emissions.^[3] As a striking example, photovoltaic (PV)-driven water splitting (WS) is a sustainable and viable technology that transforms sunlight into “green hydrogen energy.”^[4] For example, PV-driven WS can theoretically reach a solar-to-hydrogen energy conversion efficiency (η_STH) even higher than 28% with an efficient bipolar alkaline electrolyzers coupled with two-junction PV cells with an active material band gap combination of 1.60 eV/0.95 eV, (such as perovskite-Si tandem PV cells).^[5,6] Thanks to the establishment of mature and commercially available PV technologies with power conversion efficiencies above 20%^[7] and limited system installed expenses (< $1.2/W_dc for utility-scale PV system),^[8] the costs of green hydrogen production, compression and distribution are now lower than $2.90/Kg.^[9] Not surprisingly, green hydrogen refueling stations have been proposed and are under commercial operation worldwide.^[4] In this context, the development of earth-abundant and efficient electrocatalysts (ECs) that can operate at high current densities in WS electrolyzers, such as proton exchange membrane (PEM) ones,^[10] is pivotal for a worldwide replication of the current green hydrogen production plants, towards the realization of the so-called hydrogen economy.^[11] Therefore, intensive research is focused on discovering new ECs for WS reactions,^[12–16] and nanostructuring their surface to increase the electrochemically active surface area.^[17–19] The class of layered transition metal dichalcogenides (TMDs) has represented a workhorse in the research field of EC for the hydrogen evolution reaction (HER), as testified by the wide literature available.^[20–25] Experimental studies have been coupled with theoretical calculations to identify the location of the catalytic sites (e.g., defective edges^[26–28] and basal planes^[29–31]), while clarifying the activity of their different phases (e.g., 1T,^[32,33] 1T’,^[34–36] 2H,^[34] 3R^[37,38] and 6R^[39]), as well as the effects of strain engineering^[40–43] and chemical doping.^{[40,41,44,45]}

By rationalizing these results, some strategies like material nanostructuring and hybridization, introduction of defects, as well as chemical/physical modifications appear almost universal approaches to attain highly catalytic TMDs, regardless of the specific material. One may start wondering whether there is any reason to claim that one TMD is better than another one. Currently, to the best of our knowledge, despite the impressive catalytic activity claimed and experimentally demonstrated for the TMDs, there are no commercial water electrolyzers based on these materials. In fact, only some proof-of-concept electrolyzers have been reported using, for example, MoS₂,^[46–49] RuTe₂,^[50] FeS₂^[51] and NbS₂^[52] as cathode materials. Such aspects indicate the need to systematically validate the use of TMDs as HER-EC into practical electrolyzers to advance the technologies for the electrochemical hydrogen production.

In this work, we show that even a theoretically catalytically inactive TMD, namely 1T- ZrSe₂,^[53,54] can be nanostructured in the form of two-dimensional (2D) few-layer flakes and treated through microwave (MW) irradiation and NiCl₂-induced chemical modification to act as efficient HER-EC. To the best of our knowledge, only one experimental study evaluated ZrSe₂ (and other group-4 TMDs) for the HER,^[55] however, clarifying the effects of chemical and physical modifications on its catalytic activity. Meanwhile, recent theoretical studies indicated that the doping of ZrSe₂ with 3d transition metals could be a promising strategy to improve the catalytic activity of this material, which are otherwise inert for the HER.^[53] Moreover, Zr-based TMDs have been also proposed as promising building blocks to create TMD heterostructures with high HER-activity.^[54] Lastly, ab initio computational screening suggested that the engineering of vacancy density in ZrSe₂ can tune the material free energy for the hydrogen adsorption (ΔG_Hads) to almost zero,^[56] as for the Pt case. By an in-depth structural and chemical characterization of our ECs, we evidence the presence of defects, mixed crystalline and amorphous regions, and significant surface oxidation. We also show that the exfoliated flakes play an appealing role in hosting metallic catalytic dopants, namely Ni, which can display high specific activity for the HER. Undoubtedly, the multiple physical and chemical features complicate the explanation of the catalytic mechanisms, even though our results show the easiness of transforming 2D TMDs into effective HER-ECs. To validate our results in an industrially relevant environment, we finally demonstrate ZrSe₂-based acidic (PEM) and alkaline water electrolyzers that potentially compete with commercial technologies based on well-established ECs.

RESULTS AND DISCUSSIONS Material synthesis, exfoliation and characterization

Amongst TMDs, ZrSe₂ is a group-IV TMD that has been poorly investigated experimentally for WS applications. First-principles density functional theory (DFT) calculations can be preliminary used to predict the theoretical activity of an EC.^[57,58] In particular, the free energy for the hydrogen adsorption (ΔG_Hads) can be used as an indicator for the HER-activity of an EC.^[59] As described by the empirical Sabatier principle exemplified by the volcano-shaped overpotential (η_HER) versus ΔG_Hads plot,^[59] a candidate HER-EC must show a ∆G_Hads close to 0 eV. For the particular case of ZrSe₂, its stoichiometric and thermodynamically favorable trigonal phase (1T, space group: P$$\bar{3}$$m1) exhibit ∆G_Hads higher than 0.7 eV for both basal planes and edges (including metal and chalcogen terminations).^[54,57] This means that no relevant HER-activity is expected for ZrSe₂, unless other “second-order” factors, such as strain,^[60,61] support interaction^[62] and chemical modifications (e.g., doping^[44,63] and surface oxidation^[64]) drastically change the electronic structures of this material, and thus its catalytic properties. In fact, the energy of the adsorption of reactants on the surfaces of an EC strongly depends on certain characteristics of the surface electronic structure of the EC. Hereafter, we will show that these “secondary” factors, namely chemical and structural modifications, can transform the inert 2D 1T-ZrSe₂ into an effective HER-EC.

First, 1T-ZrSe₂ crystals (hereafter named bulk-ZrSe₂) were grown by direct reaction of the composing atomic elements, using Zr and Se powders with a stoichiometry of 1:2, following protocols previously described in literature (additional details are reported in Supporting Information).^[55] The as-synthesized crystals, hereafter named bulk-ZrSe₂, were initially characterized by means of scanning electron microscopy (SEM) coupled with energy-dispersive X-ray spectroscopy (EDS). Figure S1a shows the SEM image of the crystal powders, consisting of disc-shaped platelets with lateral sizes ranging from 5 to 40 μm. Figure S1b and c show the corresponding EDS maps for Zr and Se, respectively, whose analysis revealed a Zr:Se atomic ratio of 0.57 ± 0.02. The discrepancy of the chemical composition of the crystals with respect to an ideal stoichiometry is ascribed to the surface oxidation of 1T-ZrSe₂ crystals under ambient exposure.^[55]

As facilitated by the layered structure of the 1T-ZrSe₂, bulk-ZrSe₂ sample was exfoliated to obtain 2D flakes through prototypical ultrasonication-assisted liquid-phase exfoliation (LPE) in anhydrous isopropyl alcohol (IPA). Noteworthy, other industrial LPE methods, for example wet-jet milling exfoliation,^[65–68] can be prospectively used for the exfoliation step of the bulk-ZrSe₂, as demonstrated for other layered materials.^[65–68] The un-exfoliated crystals were discarded from the resulting dispersion by means of sedimentation-based separation via ultracentrifugation,^[69,70] obtaining the first set of samples named ex-ZrSe₂. A second set of samples was obtained by adding a solution of Nickel(II) chloride (NiCl₂) salt into the LPE-produced ex-ZrSe₂, with a ZrSe₂:Ni weight ratio of 5:1, followed by 1 hour of ultrasonication. These samples are hereafter named ex-ZrSe₂:Ni. By considering a mechanism proposed for group-6 TMDs,^[63,71,72] as well as the solubility limit of NiCl₂ in polar alcohol (> 1 g L^–1),^[73] NiCl₂ can react with ex-ZrSe₂ through the following cascade reaction:

Step 1: Electron transfer from ex-ZrSe₂ to Ni²⁺

• ex-ZrSe₂ + Ni²⁺ + 2Cl^– $$\rightleftarrows$$ ex-ZrSe₂²⁺ + Ni⁰↓ + 2Cl^– and/or

• ex-ZrSe₂ + 2Ni²⁺ + 4Cl^– $$\rightleftarrows$$ ex-ZrSe₂²⁺ + Ni⁰↓ + NiCl₄^2–

in which ↓ indicates a precipitate state;

Step 2: Neutralization of charged species and complex formation

• ex-ZrSe₂²⁺ + 2Cl^– $$\rightleftarrows$$ ex-ZrSe₂:2Cl and/or

• ex-ZrSe₂²⁺ + NiCl₄^2– $$\rightleftarrows$$ ex-ZrSe₂:NiCl₄

Beyond the formation of ex-ZrSe₂:2Cl and ex-ZrSe₂:NiCl₄ complexes, Ni⁰↓ may react with O₂ or H₂O after ambient exposure, forming NiO or Ni(OH)₂ on the surface of the ex-ZrSe₂. Moreover, other reactions, such as the anchoring of Ni atom by defective ex-ZrSe₂, as well as the reaction between NiCl₂ with by-product issued by the oxidation of ZrSe₂, may also occur. To produce two more sets of samples, the nanoflake dispersions (ex-ZrSe₂ and ex-ZrSe₂:Ni) were subsequently treated by 2.45 GHz MW in a home-made reactor, obtaining the samples hereafter named MW-ex-ZrSe₂ and MW-ex-ZrSe₂:Ni. This liquid-phase MW treatment was recently proposed by our group for exfoliated 6R-TaS₂ to modify the structure and morphology of the nanoflakes by reducing the size of their crystal grain (mono-to-polycrystalline domain conversion) and their lateral size (fragmentation).^[39] Briefly, MW can be strongly absorbed by ex-ZrSe₂ through “dipole polarizations” of structural defects and/or edges,^[74,75] as well as “Maxwell-Wagner-Sillars polarizations”^[76] and MW-absorbing dipoles formed at the solvent/ZrSe₂ interfaces.^[39] Noteworthy, the MW irradiation can strengthen or even trigger a chemical interaction between ZrSe₂ and Ni ions. Although it is not the scope of this work, DFT simulations have shown that TMDs can interact with metallic species, anchoring catalytic metal atoms (e.g., Ni, Pd and Pt) in defective sites.^[39,77–79] In particular, recent studies proved that a low concentration of metal impurities could drastically enhance the catalytic performance of TMDs.^[39,78,79] Hereafter, the effect of the LPE, MW irradiation and NiCl₂ doping on the structural, chemical and catalytic properties of the ZrSe₂ crystals will be experimentally evaluated. Additional details of the material exfoliation and treatments are reported in the Supporting Information. The structural properties of the investigated ZrSe₂ crystals were evaluated through X-ray diffraction (XRD) and Raman spectroscopy. The XRD patterns of the samples (Figure 1A) match one of the 1T-ZrSe₂ (ICSD card: 652247), whose CdI₂ layered structure preferentially orients their (001) plane parallel to the substrate.^[80–83] After exfoliation, and in particular after MW treatment, an additional peak emerges at 23.5°, which may correspond to monoclinic ZrO₂ (space group: P2₁/c). No XRD peaks attributable to the presence of Ni were observed, indicating the absence of detectable crystalline domains of Ni-based compounds. Therefore, Raman spectroscopy analysis focused on bulk-ZrSe₂, ex-ZrSe₂ and MW-ex-ZrSe₂ (Figure 1B). The main Raman signature of bulk-ZrSe₂ is the out-of-plane Raman A_1g mode (195.5 cm^–1), even though the weak in-plane E_g mode (145.7 cm^–1) is also observed.^[84–86] Other weak peaks around 135 cm^–1 and between 230 and 280 cm^–1 have been ascribed to the density of two-phonon states of 1T-ZrSe₂.^[85,86] After exfoliation and MW treatments, both ex-ZrSe₂ and MW-ex-ZrSe₂ still show the Raman signatures of the bulk material, without any relevant shift of the peak positions, in agreement with previous literature.^[84] The additional peak centered at 301 cm^–1 corresponds to the second-order Raman peak of Si substrate,^[87] whose contribution becomes more significant with decreasing the thickness of the investigated crystals.^[84] Size and morphology of the exfoliated materials (ex-ZrSe₂ and MW-ex-ZrSe₂) were evaluated through bright-field transmission electron microscopy (BF-TEM) and atomic force microscopy (AFM) analyses.

FIGURE 1. A, XRD patterns of the investigated samples: bulk-ZrSe2, ex-ZrSe2, ex-ZrSe2:Ni, MW-ex-ZrSe2 and MW-ex-ZrSe2:Ni. B, Raman spectra of the bulk-ZrSe2, ex-ZrSe2, and MW-ex-ZrSe2. C, BF-TEM image of representative flakes of the ex-ZrSe2 sample and (D) the statistical analysis of their lateral dimension data. E, AFM image of representative flakes of the ex-ZrSe2 sample and (F) the statistical analysis of their thickness data. G, BF-TEM and (H) AFM images of representative nanoflakes of the MW-ex-ZrSe2 sample.

Figure 1C shows a BF-TEM image of the ex-ZrSe₂, which mainly consists of flakes with irregular shapes but sharp edges. The data of the lateral size of the flakes are fitted by a log-normal distribution peaked at 25.1 nm (Figure 1D). The AFM analysis (Figure 1E and F) indicates that the flakes have a nanometric thickness, fitted by a log-normal distribution peaked at 5.3 nm. Being the c lattice constant of the ZrSe₂ crystal equal to 6.1282 Å,^[80–83] the thickness data indicate the prevalence of few (≤10)-layer flakes in the sample. After MW irradiation (MW-ex-ZrSe₂ sample), the exfoliated flakes evolve towards milled irregular nanostructures, some of them resembling zero-dimensional quantum dots (Figure 1 G and H). These nanoflakes tend to agglomerate, impeding an accurate size statistical analysis. The exfoliated samples were analysed by high-resolution TEM (HR-TEM) imaging and energy-dispersive X-ray spectroscopic analysis and mapping coupled with scanning TEM imaging (EDS-STEM), after the reaction with NiCl₂. Figure 2A and B present the HRTEM images of portions of a representative ex-ZrSe₂:Ni flake, which exhibits extended regions with single-crystal ZrSe₂ structure (ICSD card: 652247, see fast Fourier transform (FFT) in the inset of Figure 2A). After MW treatment, the MW-ex-ZrSe₂:Ni sample consists of flakes with extended amorphous regions incorporating nm-size crystals (Figure 2C and D), indicating a substantial structural evolution compared to the non-MW treated sample. The precise monitoring of the structure of the flakes using time-resolved atomic-scale imaging coupled with atomistic simulations is beyond the scope of this current work. Nevertheless, it is reasonable that the MWs absorbed by the ZrSe₂ flakes can provide sufficient thermal energy to activate structural modifications resulting from the migration of adatoms, interstitial atoms and metal/chalcogen vacancies, as reported for other TMDs (e.g., TaS₂,^[39] MoS₂,^[88–90] MoSe₂^[91] and WS₂^[92]). In fact, the dislocation coalescence initiates dynamic grain boundaries and line defects, whose sliding and strain fields lead to the fragmentation of single-crystal domains of the flakes. Consequently, the latter decrease in lateral size, as shown in Figure 1 G and H. Figure 2E–I report the STEM-EDS analysis of the MW-ex-ZrSe₂:Ni samples. The quantitative elemental analysis indicates a Ni:Se atomic ratio of 0.3, which is significantly higher than the one obtained for ex-ZrSe₂:Ni (i.e., 0.16, as revealed by STEM-EDS analysis on the flake reported in Figure S2).

FIGURE 2. A, HR-TEM image of a portion of a flake of ex-ZrSe2:Ni, with fast Fourier transform (FFT) of the framed region (inset panel), matching [212]-oriented 1T-ZrSe2 and (B) magnified, frequency-filtered portion of (A). C, HR-TEM image of a portion of a flake of MW-ex-ZrSe2:Ni, exhibiting nm-sized crystallites within an amorphous matrix, with FFT of the framed region (inset panel), matching with randomly-oriented 1T-ZrSe2 and (D) magnified, frequency-filtered portion of (C). E, High-angle annular dark field-STEM (HAADF-STEM) image of a flake of MW-ex-ZrSe2:Ni and (F-I) corresponding STEM-EDS maps for Zr (Kα), Se (Kα), Ni (Kα) and O (K), respectively.

These results suggest the interaction between exfoliated MW-ex-ZrSe₂ and Ni species is stronger than the one between ex-ZrSe₂ and Ni species, as further evaluated through X-ray photoelectron spectroscopy (XPS) analysis hereafter. Noteworthy, STEM-EDS data also reveal an oxidation of the flakes both in ex-ZrSe₂:Ni and MW-ex-ZrSe₂:Ni samples (Figures S3 and S4). In addition, the flakes in MW-ex-ZrSe₂:Ni are more oxidized than in ex-ZrSe₂:Ni (O:Se atomic ratio of 1.37 and 1.80, respectively), probably because of the increase of the surface area of the sample after the fragmentation induced by the MW treatment. The same specimens were analyzed by XPS measurements to further evaluate their surface chemical composition. As shown in Figure S5, the doublets associated with the Zr⁴⁺ states in ZrSe₂ and ZrO₂ with energy binding peaks between 182 and 186 eV primary contribute to the Zr 3d spectra, in agreement with previous literature.^[55,93–95] These peaks are clearly distinguishable only in bulk-ZrSe₂, while they overlap in the exfoliated samples due to the marked broadening, which is likely attributed to the presence of intermediate states. In addition, Se LMM Auger peaks are also observed at 174.6 and 179.1 eV.^[96] The Se 3d spectra of the samples (Figure S6) exhibit a pronounced doublet with peaks between 53.4 and 55.1 eV associated with the Se^2– states in ZrSe₂.^{[55,93,94,97]} At higher energies, additional doublets are attributed to Se^–(2-x) states in metal-deficient ZrSe₂ sites or elemental Se (Se⁰).^[55,93,94] In ex-ZrSe₂:Ni and MW-ex-ZrSe₂:Ni, the additional band between 55 and 62 eV is ascribed to Se²⁺ state in SeO₂ (whose Se 3d doublet peaks do not show pronounced spin orbit splitting and is therefore fitted with only one peak),^[93,98] whose formation may be therefore induced by the presence of Ni species. Noteworthy, the surface oxidation of the ZrSe₂ crystals can strongly affect their catalytic properties, whose origin elucidation is therefore intricated. We remark that a detailed analysis of surface oxidation of novel TMD ECs is often disregarded, as we previously criticized for relevant recent cases, for example, group-5 TMDs,^[39,64] for which relevant literature reported record-high mass activities^[99,100] even comparable to Pt-based ECs (e.g., > 1000 A g^–1 at ∼50 mV overpotentials).^[99] Lastly, the Ni 2p spectra of the ex-ZrSe₂:Ni and MW-ex-ZrSe₂:Ni (Figure S7) reveal doublets associated with Ni²⁺ states of both NiO and Ni(OH)₂,^[100–102] as well as metallic Ni⁰.^[100–102] Importantly, the catalytic activity of metallic Ni in alkaline media can be significantly improved in the presence of low-valence-state oxides.^[103–106] In fact, the dissociative adsorption of water can be enhanced at the metal/Ni oxide interface, as proposed by previous studies.^[107–109] Overall, all these data evidence the presence of several morphological/structural and chemical features that can synergistically determine the catalytic properties of our materials, as we will experimentally show hereafter.

Electrochemical characterization of ZrSe₂-based electrodes

The electrodes were produced by depositing the as-produced samples (i.e., ex-ZrSe₂ and MW-ex-ZrSe₂, ex-ZrSe₂:Ni and MW-ex-ZrSe₂:Ni) onto buckypaper current collectors via vacuum filtration, as described in previous works.^[36,63,110] The as-produced electrodes are hereafter named with the same name as the active materials. Platinum on carbon (Pt/C) and buckypaper were also investigated as reference electrodes, together with an electrode obtained by depositing a NiCl₂ dispersion onto a buckypaper (electrode hereafter named Ni). The mass loading of the Zr and Ni on the electrodes was precisely measured through inductively coupled plasma-optical emission spectrometry (ICP-OES) measurements to determine the specific activity of our ECs accurately. Thanks to their 2D morphology, the exfoliated ZrSe₂ crystals are effectively trapped by the buckypaper, and their mass loading is approximately dictated by the amount of the corresponding dispersion that are deposited onto the buckypaper. On the contrary, Ni²⁺ and Cl^– ions may pass through the whole filter without totally interacting with ex-ZrSe₂. Thus, the final amount of Ni depends on the specific chemical interaction of NiCl₂ salt with ZrSe₂ crystals, which, in turn, can show distinctive structural and chemical characteristics depending on their physical and chemical treatments. Nevertheless, ICP-OES data revealed a nearly equal Zr:Ni atomic ratio (∼2.01) for both ex-ZrSe₂:Ni and MW-ex-ZrSe₂:Ni electrodes, indicating an effective Ni trapping by the electrodes, as confirmed by the negligible presence of Ni in the dispersion downstream of the filtering system. Top-view SEM images of the electrodes (Figure S8) indicate a uniform coverage of the ECs over the buckypaper current collectors. Figure 3A and B show the linear sweep voltammetry (LSV) scans measured for the investigated electrodes in both acidic (0.5 M H₂SO₄) and alkaline (1 M KOH) media. These data indicate that both MW treatment and the presence of Ni species increase the HER-activity of ex-ZrSe₂. Their combination allows MW-ex-ZrSe₂:Ni to achieve an overpotential at a cathodic current density of 10 mA cm^–2 (ƞ₁₀) of 0.082 V in 0.5 M H₂SO₄ and 0.081 V in 1 M KOH, approaching the values measured for Pt/C benchmark. By comparing the LSV curve of the investigated electrodes, the HER-activity of the ex-ZrSe₂:Ni in 0.5 M H₂SO₄ is primarily attributed to the exfoliated materials, since its LSV curve almost overlaps the one of the corresponding Ni-free electrode (i.e., ex-ZrSe₂). In addition, in 0.5 M H₂SO₄, the reference electrode (Ni) shows ƞ₁₀ (0.303 V), significantly higher than those of ex-ZrSe₂ (0.163 V); and ex-ZrSe₂:Ni (0.159 V). As described by its Pourbaix diagram,^[111] Ni is found to be stable in acidic media in its metallic state (Ni⁰) only at overpotentials higher than 0.246 V in 0.5 M H₂SO₄, being the standard reduction potential of the Ni²⁺ + 2e^– $$\rightleftarrows$$ Ni reaction equal to—0.249 V.^[112] In absence of any interaction with the support, Ni species, as those detected in our materials, will dissolve in the acidic medium, without affecting the catalytic activity of our electrode at overpotentials lower than 0.246 V. Nevertheless, Ni atoms can chemically interact with surrounding materials, as shown for various type of nanomaterials, including graphene,^[113] MXenes,^[114] as well as TMDs.^[77] This last effect may explain the significant increase of the catalytic activity of MW-ex-ZrSe₂:Ni, in which Ni may anchor to defective sites introduced by the MW treatments. Contrary to acidic media, low-valence-state oxide of Ni can be found stable under the standard condition in alkaline media.^[103,104] Therefore, in 1 M KOH, the presence of Ni species increases the HER-activity of both Ni-containing electrodes compared to the Ni-free counterparts. Nevertheless, the significantly superior HER-activity of the MW-ex-ZrSe₂:Ni compared to the other tested electrodes indicates the presence of synergistic effects given by the combination of both Ni and Zr species, including metallic Ni, oxides (e.g., NiO_x and ZrO₂) and ZrSe₂, as well as structural defects of the latter. For example, previous studies demonstrated that the catalytic activity of metallic Ni in alkaline media is significantly enhanced when Ni is covered by low-valence-state oxides.^[103–106] According to ground-breaking works,^[107,108] Ni oxides tend to adsorb OH^– due to their strong electrostatic affinity of locally positive Ni²⁺ species and unfilled d orbitals in Ni²⁺.^{[109,115,116]} Meanwhile, metallic Ni sites, effectively adsorb H⁺, thus initiating the Volmer process^{[104,106,117]} without being de-activated by the adsorption and/or inefficient release of OH^–.^{[104,106,117]} Clearly, Zr oxides and ZrSe₂ may also play a role in such synergistic effects, resulting in an effortless increase of their HER-activity, as well as the one of metallic Ni.

FIGURE 3. A-B LSV curves measured for the investigated electrodes (ex-ZrSe2, MW-ex-ZrSe2, ex-ZrSe2:Ni, MW-ex-ZrSe2:Ni, Ni (salt), Pt/C and bucky-paper) in 0.5 M H2SO4 and 1 M KOH, respectively. C-D, MA versus potential plots of ex-ZrSe2, MW-ex-ZrSe2, ex-ZrSe2:Ni, MW-ex-ZrSe2:Ni, Pt/C in 0.5 M H2SO4 and 1 M KOH, respectively. The MA are referred to the following metallic component: Zr, Ni (determined by ICP-OES measurements) and Pt. E, Chronoamperometry stability tests for MW-ex-ZrSe2:Ni in 0.5 M H2SO4 and 1 M KOH

Undeniably (again), detailed mechanistic insights into electrocatalysis of the investigated materials are currently limited due to difficulties in theoretical modeling of HER processes involving many material species (i.e., metallic Ni, Ni and Zr oxides, ZrSe₂ and their complex or doped-ZrSe₂), which may also form in situ during HER operation and cannot be identified by trivial ex-situ analysis. To ensure the effectiveness of our most performant ECs, namely MW-ex-ZrSe₂:Ni, gas chromatography (GC) measurements were performed to exclude the presence of parasitic reactions that can lead to erroneous interpretations of the real catalytic properties for the HER. The GC analysis of evolved gases revealed a nearly 100% faradaic efficiency for the HER in both 0.5 M H₂SO₄ and 1 M KOH. Even though η₁₀ is a commonly accepted metric used to compare and characterize various ECs, it depends on the mass loading of the same ECs, which is also dictated by the morphology (porous or flat) of the current collector. Being strongly dependent on the mass loading of the ECs, which can significantly differ among different reports, η₁₀ is not a suitable metric for evaluating the intrinsic activity of an EC.^[118] As we will show later, the overpotentials at defined current densities are instead crucial parameters for the design and engineering of the electrodes in industrial water electrolyzers.^[119] By following the recommendation reported in literature,^[118–120] the mass activity (MA) of our ECs, defined as the ratio between the HER-cathodic current and the mass of the ECs, was evaluated to assess the specific (intrinsic) catalytic activity of our ECs. This analysis has been recommended as a universal approach that can be used even in presence of porous substrates, such as our bucky-papers. In fact, even though the electrochemically accessible/active surface area (ECSA)-normalized activity (j_ECSA) is also often used as a metric for the evaluation of the intrinsic catalytic activity of a catalyst, the presence of porous substrates can lead to significant ECSA overestimation when using traditional measurement method,^[119] for example, double-layer capacitance measurement by means of cyclic voltammetry measurements at different scan rates in a non-faradaic region.^[121] For this reason, we do not refer to j_ECSA for the specific case of our electrodes. Figure 3C and D show the MA versus overpotential plots for the investigated electrodes in 0.5 M H₂SO₄ and 1 M KOH, respectively. In these plots, MA was referred to as the as-deposited mass of the Zr, Ni (determined by ICP-EOS measurements) and Pt metallic components. Since possible losses occurring during electrochemical tests were neglected, the real MA might have been even underestimated in this work. Noteworthy, the most performant electrodes, that is, MW-ex-ZrSe₂:Ni, reach a MA of 500 A g^–1 at overpotential as low as 0.153 V, which is only ∼0.1 V higher than the Pt MA measured for the Pt/C electrode. In addition, MW-ex-ZrSe₂:Ni also exhibited a stable catalytic activity over an hour time-scale, retaining 96.8% and 94.5% of their initial current density (50 mA cm^–2) after 5 hours of operation in 0.5 M H₂SO₄ and 1 M KOH, respectively (Figure 3E). It is worth noticing that Ni, according to its Pourbaix diagram,^[111] is not stable in acidic media, as previously discussed. However, several works proved the Ni, once incorporated into proper material scaffolds, can stably operate as catalytically active sites for the HER in acidic media.^{[113,122–125]} In particular, Ni species can stably adsorb on the basal plane of TMDs,^[113] as well they can also be coordinated to the chalcogen edges of the TMDs.^[113]

Evaluation of ZrSe₂ ECs in water electrolyzers

The electrochemical activity of our electrodes proved to us that 1T-ZrSe₂ crystals, even though they are theoretically predicted to be poor HER-ECs, can be exfoliated, and treated through MW irradiation and chemical NiCl₂ doping to reach MA for HER of interest for practical water electrolyzers. As discussed before, the catalytic activity for the HER of our best ECs, namely MW-ex-ZrSe₂:Ni, can be likely attributed to various structural and chemical features, such as the presence of defective edges, surface oxidation, the existence of Ni species (metallic Ni and Ni (suboxides) onto the surface of the exfoliated materials, as well as specific chemical interaction between such Ni species and the exfoliated ZrSe₂. Clearly, we realize that a definitive explanation of the origin of the catalytic activity cannot be easily provided, at the current stage. From a certain point of view, it is neither of interest for the message of this work, since we are convinced that our approach can be applied even to other ECs, including TMDs, of greater interest for the HER compared to ZrSe₂, which is indicated theoretically as poor catalytic material. On the one hand, it is implicitly recommended to maintain a critical attitude during claiming the discovery of novel catalytic nanomaterials, such as TMDs, whose plethora of structural and chemical modifications risks jeopardizing a real advancement in the rational search and design of effective ECs. On the other hand, our results remark the simplicity of transforming nanomaterials, namely TMDs, into effective ECs for the HER. To strength the validity of this message, water electrolyzers were produced using MW-ex-ZrSe₂:Ni as cathode (EC mass loading = 1.9 mg cm^–2) in membrane electrode assembly (MEA) configuration, as sketched in Figure 4A. Both acidic (PEM) and alkaline electrolyzers were investigated to evaluate our cathode performance. Nafion 117 and Zirfon Perl UTP 500+ were used as the PEM membrane and diaphragm for the PEM and alkaline electrolyzers, respectively. For the anode, IrO₂ (deposited on Ti mesh, mass loading = 3.6 mg cm^–2) and a stainless steel (SS) mesh were used as ECs for the oxygen evolution reaction (OER) for PEM and alkaline electrolyzers, respectively. Thus, our alkaline electrolyzer is noble metal-free, as targeted by alkaline systems.^[115,126] Additional details regarding MEA preparation are reported in Supporting Information. During the operation of PEM electrolyzers, the deionized water is supplied to the anode side via peristaltic pumps. Once the water is split through the OER to produce protons, the latter move to the cathode through the PEM until being reduced to H₂ by MW-ex-ZrSe₂:Ni. As a comparison, an acid-fed configuration pumping a 0.5 M H₂SO₄ solution in both anode and cathode sides with separate feeds to reduce gas mixing was also tested. For the alkaline electrolyzer, an alkaline solution of 30 wt% KOH was supplied with separate feeds to both anodic and cathodic sides. At the cathode, the HER process split the water into H₂ gas and free anions, which can move through the diaphragm to the anode side, where the OER occurs. Figure 4B shows the LSV curves recorded for the electrolyzers at 80°C (without iR-compensation). The anode distilled water-fed PEM electrolyzer delivered a current density of 400 mA cm^–2 at 2.00 V, while the acid-fed configuration operated at 1.95 V at the same current density. The alkaline electrolyzers delivered a current density of 400 mA cm^–2 at 1.92 V. Being the best performing electrolyzer, the alkaline one was evaluated for an extended current density range, up to a maximum value of 1.2 A cm^–2, for which it operated at a voltage of 2.05 V (Figure 4C).

FIGURE 4. A, Sketch of the water electrolyzers using MEA configurations based on the designed MW-ex-ZrSe2:Ni cathode. The PEM electrolyzers use RuO2 OER-EC deposited onto Ti mesh gas diffusion layers, while the alkaline electrolyzers use SS mesh-based anodes which also act as gas diffusion layers. B, LSV curves recorded for: PEM electrolyzers based on MW-ex-ZrSe2:Ni as the HER-EC for the cathode and IrO2 as the OER-EC for the anode (MW-ex-ZrSe2:Ni//IrO2) (two PEM configurations using distilled water feed into anode and 0.5 M H2SO4 feed in both cathode and anode, respectively, are reported); alkaline electrolyzer based on MW-ex-ZrSe2:Ni as the HER-EC for the cathode and SS mesh as the anode (MW-ex-ZrSe2:Ni//SS) and using a 30 wt% KOH electrolyte feed into both cathode and anode. C, Polarization curves recorded for the alkaline electrolyzer up to 1.2 A cm–2

Noteworthy, our electrolyzers outperformed prototypical noble-metal-free systems based on Ni and SS electrode pairs, which typically show current density of 200 mA cm^–2 at voltage equal or superior to 2 V.^[117,126] Table S1 reports the comparison between our electrolyzers and those reported in the relevant literature, mainly focusing on those based on TMDs. The proposed electrolyzers exhibit superior performance compared to most of the ones previously developed with TMDs. Our alkaline electrolyzers are also competitive with commercially available alkaline systems,^{[115,116,127,128]} typically operating at voltage higher than 1.8 V for a current density of 300 mA cm^–2 and 1 atm pressure. Moreover, our anode distilled water PEM electrolyzers and alkaline electrolyzers exhibited nearly stable voltage at 400 mA cm^–2 (Figure S9a), which is consistent with the stability observed for our MW-ex-ZrSe₂:Ni as cathodes (Figure 3E). Clearly, for the message of this work, it was not our focus to perform more extended stability tests in such systems. Differently, the acid-fed PEM electrolyzer degraded over subsequent cycles (Figure S9b), likely because of the instability of IrO₂ anode in acidic media.^[129]

CONCLUSION

In summary, we have reported the bulk synthesis, the exfoliation in 2D form, as well as the physical and chemical treatment of 1T-ZrSe₂ crystals to be used as ECs for HER in both acidic (0.5 M H₂SO₄) and alkaline (1 M KOH) media. Despite the theoretical catalytic inertness of the 1T-ZrSe₂ (i.e., reaction free energy for hydrogen adsorption (ΔG_Hads) ≥ 0.7 eV), we show that both structural and chemical modifications induced by microwave irradiation and NiCl₂-based chemical treatments are feasible strategies to transform exfoliated 2D samples into effective HER-ECs. By in-depth characterizing our exfoliated materials, we discuss how it is hard to robustly explain the origin of the catalytic activity of the resulting materials. This is because of the concomitant occurrence of significant surface oxidation of 1T-ZrSe₂, structural modifications (i.e., presence of edge and defective sites), complex species (i.e., doped 1T-ZrSe₂ flakes), the formation of catalytically active Ni species (metallic Ni and low-valence state Ni oxides), as well as the possible EC reduction during HER-operation. However, our results remark how it is straightforward to get HER-active 2D TMDs, regardless of the specific material. To practically validate this message, we have produced and characterized water electrolyzers using MEA configurations and ZrSe₂-based cathodes. Our optimized proton-exchange membrane and anion-exchange membrane water electrolyzers operated at 300 mA cm^–2 at voltage of 1.88 and 1.92 V, respectively, thus competing with commercially available systems, as well as noble-metal free electrolyzers reported in literature. Overall, the emblematic case of the 2D ZrSe₂ crystals that became effective HER-ECs up to their implementation into practical electrolyzers aims to spur a critical attitude in proposing novel materials as water splitting ECs, as well as the establishment of laboratory technologies up to an industrial level.

ACKNOWLEDGMENTS

This project has received funding from the European Union's Horizon 2020 Research and Innovation programme under grant agreement no. 987273 (SENSIBAT) and grant agreement no. 881603-GrapheneCore3, the MSCA-ITN ULTIMATE project under grant agreement no. 813036, and the Bilateral project GINSENG between NSFC (China) and MAECI (Italy) (2018–2020) by the Natural Science Foundation of Shandong Province (ZR2019QEM009). This project was supported by the Czech Science Foundation (GACR No. 20–16124J). P.M. was supported by specific university research grant no. A2_FCHT_2020_055. Marilena I. Zappia received funding from PON Research and Innovation 2014–2020 (CUP H25D18000230006) by the Italian Ministry of University and Research. Kseniia Mosina was supported by specific university research (MSMT No. 20-SVV/2021). The authors thank the Materials Characterization Facility—Istituto Italiano di Tecnologia—for the support in XRD data acquisition/analysis and Electron Microscopy facility—Istituto Italiano di Tecnologia—for the support in TEM data acquisition/analysis.

CONFLICT OF INTEREST

The authors declare no conflict of interest.

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available from the corresponding author upon reasonable request.