Research ArticleChemistry

Water- and acid-stable self-passivated dihafnium sulfide electride and its persistent electrocatalytic reaction

See allHide authors and affiliations

Science Advances  05 Jun 2020:
Vol. 6, no. 23, eaba7416
DOI: 10.1126/sciadv.aba7416


Electrides have emerged as promising materials with exotic properties, such as extraordinary electron-donating ability. However, the inevitable instability of electrides, which is caused by inherent excess electrons, has hampered their widespread applications. We report that a self-passivated dihafnium sulfide electride ([Hf2S]2+∙2e) by double amorphous layers exhibits a strong oxidation resistance in water and acid solutions, enabling a persistent electrocatalytic hydrogen evolution reaction. The naturally formed amorphous Hf2S layer on the cleaved [Hf2S]2+∙2e surface reacts with oxygen to form an outermost amorphous HfO2 layer with ~10-nm thickness, passivating the [Hf2S]2+∙2e electride. The excess electrons in the [Hf2S]2+∙2e electride are transferred through the thin HfO2 passivation layer to water molecules under applied electric fields, demonstrating the first electrocatalytic reaction with excellent long-term sustainability and no degradation in performance. This self-passivation mechanism in reactive conditions can advance the development of stable electrides for energy-efficient applications.


Solid-state materials with excess electrons beyond solvated electrons are always attractive as an efficient electron reservoir that can reduce the energy barrier and improve the efficiency of catalytic reactions and electronic devices (1, 2). However, it is challenging to use their immense reactivity sustainably in active environments, such as in water, analogous to solvated electrons in aqueous solutions having a very short lifetime (3, 4). Electrides are exotic materials with inherent excess electrons acting as anions, which are trapped in structural cavities (59). One beneficial property of electrides is that when their surface is exposed to reactive atmospheres or is activated by an electric field, excess anionic electrons can be directly transferred or emitted into a surface reactant or vacuum with a small barrier, triggering chemical reactions or actuating electronic devices (1012). Nevertheless, their chemical instability that originated from excess anionic electrons that are loosely trapped in a large cavity space has been a great obstacle to a persistent and efficient solid-state electron source (13). Thus, the development of stable electrides in any active circumstances is the most urgent task.

Ambient-stable electrides are rare and require a particular surface structure that is made from chemically inert materials to passivate the reactive excess electrons. One representative is the room temperature stable [Ca24Al28O64]4+∙4e electride that is constructed from three-dimensionally connected subnanometer-sized cages, in which anionic electrons are trapped. At the cleaved surface of the [Ca24Al28O64]4+∙4e electride, the subnanometer-sized cages collapse into a chemically inert amorphous structure that is responsible for the stability in an ambient atmosphere and is revealed to have the electride character (14, 15). Recently, other stable electrides have been found in intermetallic compounds, which are constructed from a cage-linked chain structure, but their surface structures need to be investigated (16, 17). In particular, the Y5Si3 electride retains its catalytic capability for ammonia synthesis even after being dipped in water (16), which implies that the surface structure is most likely chemically inert. However, Y5Si3 electrides have not been applied to a catalytic reaction under wet conditions. An electride that is sustainable in reactive conditions such as water and acid and that can simultaneously transfer the excess anionic electrons has not been reported yet. For example, the amorphous layer at the [Ca24Al28O64]4+∙4e electride surface is easily soluble in water and alcohols, limiting its use as an electron-donating catalyst in such solutions (11). Thus, another approach is required for the development of novel electrides with long-term sustainability in reactive environments, especially in water- and acid-containing conditions. Meanwhile, considering that the performance of applications is exclusively dominated by transferring the excess anionic electrons, the transferred excess anionic electrons need to be compensated for keeping the charge neutrality and preventing the decomposition. For an ideal application, an electrochemical catalyst or electronic device that is sustainably activated by applying electric current is desirable but is not realized yet.

Here, we report a self-passivated electride with double amorphous layers that provides a rational design strategy for permanently sustainable electrides in strongly oxidative water and acid solutions. To demonstrate the availability of chemically stable electride as a catalyst, we also provide the first electrochemical catalytic reaction in water and acid solutions, which makes a breakthrough in the research field of electride materials. In the context of material design, we conceived that the designed electride needs to contain a metallic element, such as Hf, W, or Cr, which tends to form an amorphous oxide as a passivation layer. Several Hf-based binary compounds such as Hf2S and HfBr have been already synthesized (18, 19) and recently proposed as electride candidates by computational predictions (20, 21). Among the candidates, considering that chalcogens often formulate an amorphous material with transition metals (22), we investigated the Hf2S as a proof-of-concept material.


A two-dimensional electride, dihafnium sulfide

First, we elucidate the detailed electronic state of excess anionic electrons in Hf2S, which has not been considered in previous calculations (20). The calculated crystal structure of Hf2S in Fig. 1A shows that the excess anionic electrons occupy the two-dimensional interlayer spacing between the cationic layers (fig. S1). This feature is the hallmark of two-dimensional electrides with localized excess anionic electrons. From the density functional theory (DFT) calculations, it is noted that the anionic electrons confined in the vicinity of Hf atoms play a substantial role in the two-dimensional electronic structure. The projected band structure (Fig. 1B) and density of states (DOS) (Fig. 1C) of excess anionic electrons (e) near the Fermi level (EF) show a notable contribution as constituent Hf and S elements, which is analogous to those of carbide electrides (23, 24). The electron localization function (ELF) and charge density map (CDM) on the (21¯0) plane are shown in Fig. 1 (D and E, respectively). The local maxima of both ELF and CDM can be observed at the crystallographic positions of anionic electrons, indicating that there exist non-nuclear attractors (25). Each anionic electron has an atomic charge of 1.3e, which is determined by the integration of charge density within Bader’s basin (fig. S1), and forms a dumbbell shape with heads at the center of the Hf triangles of each two-dimensional array at the interlayer space (Fig. 1D and fig. S1). The partial CDM (PCDM) shows that the Hf2S has a two-dimensional conduction pathway through excess anionic electrons and Hf atoms (Fig. 1F and fig. S1). Moreover, the calculated Fermi surface has a prominent cylindrical shape of a typical two-dimensional electronic system (Fig. 1F). Therefore, all results obtained by DFT calculations strongly suggest that the Hf2S is a two-dimensional electride with strong localization of excess electrons in a dumbbell shape at interlayer space, which is distinct from the localized excess electrons of two-dimensional [Ca2N]+∙e and [Y2C]2+∙2e electrides (9, 24).

Fig. 1 Two-dimensional layered structure and electronic structure of Hf2S.

(A) Calculated crystal structure with a hexagonal unit cell (a = b = 3.38 Å, c = 11.85 Å, α = β = 90°, γ = 120°) represented by solid black lines. Dumbbell-shaped excess anionic electrons (e) are localized between [Hf2S]2+ layers. Maxima of excess anionic electrons are located at the Wyckoff position (0.333, 0.667, 0.57), which indicates strong localization at the position of e. (B) Total and projected band structures. Identical band dispersions imply the strong hybridization state of Hf and excess anionic electron. (C) Total and projected DOS per unit energy per formula unit (f.u.) on spheres located at Hf, S, and e positions. The DOS of e is filled green. (D) ELF, (E) CDM, and (F) PCDM on the (21¯0) plane. Positions of Hf, S, and e are denoted by white dashed circles. (G) Fermi surface in the first Brillouin zone.

The two-dimensional layered structure of Hf2S, which is an anti-TaS2–type crystal structure that belongs to the P63/mmc space group, was confirmed by single-crystal analysis (Fig. 2A), Rietveld refinement of the corresponding powder x-ray diffraction (XRD) pattern (Fig. 2B), and atomic-scale scanning transmission electron microscopy (STEM) observation combined with energy-dispersive x-ray spectroscopy (EDX) mapping (Fig. 2C and fig. S2). The layer unit consists of edge-sharing Hf6S trigonal prisms in the [Hf2S]2+ layer with a bonding length of 2.63 Å between Hf and S, which is larger than the sum (2.42 Å) of the ionic radii of Hf4+ (0.58 Å) and S2− (1.84 Å) (26). This discrepancy indicates that the Hf ion has a lower valence state than 4+. To ensure that the valence charges on Hf and S are close to 2+ and 2−, respectively, which were revealed by the Bader’s charge analysis on the charge density, we determined the chemical valence states of each element by x-ray photoemission spectroscopy (XPS). As shown in Fig. 2D, from the spectra of the single crystal after cleaving in an ultrahigh vacuum chamber (UHV-cleaved, top), the valence states of Hf and S were determined to be 2+ and 2−, respectively. Furthermore, we conducted a quantitative analysis with core level spectra to confirm the chemical composition, giving a good agreement with a Hf/S nominal atomic ratio of 2 (table S2). These results validate that the Hf2S is a two-dimensional electride with the chemical formula [Hf2+2S2−]2+∙2e with two excess anionic electrons per formula unit, which is consistent with the theoretical calculations. It should be noted that the chemical composition and ratio of components were preserved at the surface of the UHV-cleaved sample. However, for single crystals exposed to air (air-exposed) and water (water-exposed), we found the mixed-valence state of Hf, which is composed of major 4+ and minor 2+ states (Fig. 2D, middle and bottom). From the appearance of the O 1s peak in both air- and water-exposed samples, we verified that the Hf4+ state is attributed to the formation of hafnium dioxide at the surfaces.

Fig. 2 Characterizations of crystal structure and chemical states for [Hf2S]2+∙2e electride.

(A) XRD patterns for [Hf2S]2+∙2e single crystal. Out-of-plane 2θ scan (main) and in-plane ϕ scan (inset) for cleaved surface showing well-constructed hexagonal structure with exclusive (00l) diffraction and sixfold symmetry, respectively. (B) Rietveld refinement of powder XRD pattern. Inset shows a photograph of a cleaved shiny surface. Detailed crystal information from Rietveld refinement listed in table S1. (C) Atomic-scale high-angle annular dark-field STEM image and EDX elemental mapping result for the (21¯0) plane. The calculated structure is superimposed on the observed image and shows good agreement. (D) XPS spectra for Hf 4f, S 2p, and O 1s with UHV-cleaved, air-exposed, and water-exposed single crystals. All spectra are measured at the (00l) plane and calibrated from the spectrum of reference Au metal (fig. S3). Typical Hf4+, S2−, and O2− spectra are filled green, yellow, and red, respectively. The blue spectrum is expected to be Hf2+ because the peak of Hf 4f7/2 (15.5 eV) is located between the Hf metal (14.0 eV, 0 valence state) and HfO2 (17.9 eV, 4+ valence state). a.u., arbitrary units.

Chemically stable two-dimensional electride, [Hf2S]2+∙2e

The surface modification is a key ingredient for stabilizing the two-dimensional [Hf2S]2+∙2e electride. In contrast to previously reported two-dimensional electrides, including [Ca2N]+∙e and [Y2C]2+∙2e, which are extremely unstable in air due to the layered open structure that makes excess anionic electrons react with oxygen and water molecules within several minutes (24, 27), the [Hf2S]2+∙2e electride is extraordinarily stable in reactive atmospheres. Both air- and water-exposed powders exhibited no distinct change in the crystal structure, exhibiting identical XRD patterns to that of the pulverized powders of the UHV-cleaved sample (Fig. 3A). Further, the chemical stability in acid solutions was also confirmed (fig. S4). Most of all, it is notable that the two-dimensional layered structure remains unchanged even after soaking in water for 2 months and in acid for 1 month (fig. S4). Besides the crystal structure, the electrical properties also remain unaffected. Figure 3B shows the temperature dependence of the electrical resistivity (ρ) for [Hf2S]2+∙2e electrides that were exposed to different atmospheres, showing that metallic conduction is maintained with only a slight change in resistivity value, which arises from surface oxidation. As shown in the inset of Fig. 3B, both carrier concentration (NH) and electrical mobility (μ) also showed negligible changes. However, unlike the electrical transport properties, the surface state of the UHV-cleaved sample changed remarkably after exposure to air and water (Figs. 2D and 3C). From the ultraviolet photoemission spectroscopy (UPS) that reveals the work function (WF) values for each sample, it is observed that the metallic Fermi edge (indicated by the red arrow in the inset of Fig. 3C) of the UHV-cleaved sample with a WF value of 3.6 eV disappeared and the insulating states with an increased WF value of ~4.5 eV evolved upon exposure to air and water. Thus, we speculate that the entire surface is covered by insulating hafnium dioxide with a measured WF value of ~4.5 eV, which is slightly smaller than the reported value (28), most likely due to the contribution from the transferred excess anionic electrons, passivating the [Hf2S]2+∙2e electride and rendering the extraordinary chemical stability. To elucidate the surface modification in detail, we observed the surface structure by using TEM combined with EDX elemental mapping.

Fig. 3 Water-durable [Hf2S]2+∙2e electride passivated by double amorphous layers.

Comparison of (A) powder XRD patterns, (B) electrical transport properties, and (C) UPS spectra for UHV-cleaved (black), air-exposed (red), and water-exposed (blue) samples (Materials and Methods). XRD and UPS results indicate the increased resistivity of air- and water-exposed samples attributed to increased contact resistance between the electrode and insulating a-HfO2 at the [Hf2S]2+∙2e electride surface. Carrier concentration (NH) of 3.04 × 1022 cm−3 agrees well with the calculated value of 3.44 × 1022 cm−3 based on two electrons per formula unit with a chemical formula of [Hf2S]2+∙2e. Inset of (C) shows the metallic Fermi edge of the UHV-cleaved sample. (D) HRTEM images of surface structures for UHV-cleaved, air-exposed, and water-exposed [Hf2S]2+∙2e electrides. (E) STEM-EDX elemental mapping profile in the region of the white dashed box in (D). S- and O-dominant amorphous layers highlighted with different colors are distinguishable by a contrast difference (fig. S6). Oxygen signal underneath a-HfO2 layer ascribed to oxidation of top and bottom surface [perpendicular direction to the images in (D)] of the TEM sample. (F) STEM-EDX elemental-mapping images for the water-exposed [Hf2S]2+∙2e electride.

Double amorphous passivation layer-driven chemical stability under oxidizing condition

Figure 3D shows high-resolution TEM (HRTEM) images for samples prepared in three different atmospheres. In the UHV-cleaved sample, an amorphous layer (greenish region) was visible on the surface of crystalline [Hf2S]2+∙2e electride (Fig. 3D and fig. S5). A noticeable change is that the amorphous layer was expanded in thickness and divided into two different amorphous layers in the air- and water-exposed samples, forming a double-layered surface structure. From the EDX elemental mapping results for the water-exposed sample (Fig. 3, E and F), the inner (yellowish region) and outermost (reddish region) amorphous layers are revealed as Hf–S and Hf–O amorphous compounds, respectively. Considering together with the composition ratio obtained from XPS measurements (table S2), it is certain that the amorphous layer of the UHV-cleaved sample is amorphous dihafnium sulfide (a-Hf2S), and the outermost amorphous layer of the air- and water-exposed samples is amorphous hafnium dioxide (a-HfO2). This result thus concludes that the WF values of 3.6 and 4.5 eV (Fig. 3C) originate from the a-Hf2S and a-HfO2, respectively. Furthermore, it implies that the metallic a-Hf2S layer with a low WF value of 3.6 eV is unstable in air and water atmospheres, reacting with oxygen and forming a chemically stable a-HfO2 layer. This intriguing surface modification suggests that the a-Hf2S probably contains excess anionic electrons, maintaining the electride character, analogous to the amorphous [Ca24Al28O64]4+∙4e electride (14, 15). From the compositional distributions of double amorphous layers obtained by EDX elemental mapping [Fig. 3, E and F, see also electron energy-loss spectroscopy (EELS) results in fig. S5], it is verified that the Hf ions of the a-Hf2S layer undergo an outward diffusion to form the a-HfO2, resulting in a Hf-deficient a-Hf2−xS layer. Compared with the intensities of Hf and S for crystalline [Hf2S]2+∙2e, an increase in S/Hf ratio in the a-Hf2−xS layer is observed (the yellowish region in Fig. 3E). Once the a-Hf2S is exposed to air or water, the excess anionic electrons in a-Hf2S are transferred to the adsorbed oxygen molecules at the surface and bind with outward diffused Hf ions to form a-HfO2, leading to an a-HfO2/a-Hf2−xS/[Hf2S]2+∙2e structure denoted as a self-passivated [Hf2S]2+∙2e electride. When ~ 5 to 10 nm, further oxidation is prevented via blocking ion diffusion and electron transfer by a-HfO2, a well-known chemically inert compound even in some strong acids (29). Moreover, real-time TEM experiments with in situ heating show no significant changes in phases up to ~700°C, which shows that the thermal stability of the self-passivated [Hf2S]2+∙2e electride is also excellent (fig. S6 and movie S1). It should be noted that clarifying the unprecedented self-passivation mechanism by a double amorphous layer structure supports our material design strategy, the selection rule of constituent elements for the formation of amorphous oxide as a passivation layer, triggering the development of chemically stable electrides and exploring the surface structure, even for the existing stable electrides.

The first electrocatalytic reaction of electrides in acid solutions

Last, given the permanent stability of the self-passivated [Hf2S]2+∙2e electride under strongly oxidizing conditions, an electrocatalytic hydrogen evolution reaction (HER) was performed in a 0.5 M sulfuric acid solution (movie S2). Figure 4 (A and B) shows the HER polarization curves from linear-sweep voltammetry and Tafel plots for the self-passivated [Hf2S]2+∙2e electride, a-HfO2 thin films with different thickness (fig. S7), and HfO2 powder. To clarify the role of the excess electrons of the [Hf2S]2+∙2e electride on HER, a 10-nm-thick a-HfO2 thin film, which has a similar thickness and chemical states to the a-HfO2 layer on the [Hf2S]2+∙2e electride, was prepared on the gold electrode (fig. S7). Both the a-HfO2 thin film with 100-nm thickness and HfO2 powders showed no HER activity (Fig. 4A and fig. S7). For the a-HfO2 thin film with 10-nm thickness, the performance was also poor, displaying an overpotential to achieve the current density of 10 mA cm−2 and Tafel slope of 581 mV and 141 mV decade−1, respectively. It is noticeable that the self-passivated [Hf2S]2+∙2e electride exhibits a nearly identical Tafel slope value of 130 mV decade−1, suggesting that the dominant chemical reaction on the surface is unchanged; however, its electrocatalytic activity for HER is largely enhanced to an overpotential of 355 mV to reach 10 mA cm−2. From the performance difference between the 10-nm-thick a-HfO2 thin film and the self-passivated [Hf2S]2+∙2e electride, it is clear that the excess anionic electrons of the [Hf2S]2+∙2e electride have a critical role in the electrocatalytic HER. Based on the theoretical calculations and experimental results (Fig. 3C and fig. S8), the WF values of the [Hf2S]2+∙2e electride and a-Hf2S electride are of ~3.04 and 3.6 eV, respectively. These low WF values result in lower energy barrier (Eb) at the junction with a-HfO2 than that of Au/a-HfO2, as shown in Fig. 4C.

Fig. 4 Persistent electrocatalytic hydrogen evolution reaction of self-passivated [Hf2S]2+∙2e electride in acid solution.

(A) Polarization curves and (B) Tafel plots for self-passivated [Hf2S]2+∙2e electride, HfO2 powder, and thin films. The thick HfO2 powder (~45 μm) and film (~100 nm) are nonactive, whereas thin enough HfO2 film (~10 nm) shows HER activity with poor performance due to the tunneling current. HER performances of the self-passivated [Hf2S]2+∙2e electride maintained over 1 week (5000 cycles). The 5000th cycle of polarization curves selected to plot the Tafel slope (fig. S7). Plausible band diagrams of (C) self-passivated [Hf2S]2+∙2e electride and (D) HfO2 thin film on Au for electron transfer to evolve the hydrogen gas. The WF values are acquired from DFT calculations (for the [Hf2S]2+∙2e electride), UPS experiments (for the a-Hf2−xS and a-HfO2), and literature (37) for Au.

These results are remarkable from the fact that the self-passivated [Hf2S]2+∙2e electride does not react with water and acid solution, in contrast to the previous unstable electrides that are easily decomposed in water (fig. S4), but shows the electrocatalytic reaction under the applied electric field. In other words, the excess electrons of the self-passivated [Hf2S]2+∙2e electride were transferred through the thin a-HfO2 passivation layer under the applied electric field, facilitating the persistent electrocatalytic HER. However, the moderate HER performance of the self-passivated [Hf2S]2+∙2e electride, which is lower than that of state-of-the-art materials such as defect-rich ultrathin two-dimensional MoS2 (30), further requires the defect engineering and surface structure modification to enhance the active sites. The modification of passivated surface structure, such as metal nanoparticle loading, improves the electrocatalytic performances of the self-passivated [Hf2S]2+∙2e electride (fig. S7). On the basis of the above results, possible mechanisms for the electrocatalytic HER over the self-passivated [Hf2S]2+∙2e electride are summarized in Fig. 5. The three generally postulated mechanisms in the HER are the Volmer (forming adsorbed hydrogen atoms, Hads, originated in the reaction between proton, H+, and electron), the Heyrovsky (reaction of H+, Hads, and electron), and the Tafel (recombination of Hads) (31). Given the value of the Tafel slope for the self-passivated [Hf2S]2+∙2e electride, it is suggested that the Volmer step is the rate-determining step. The present HER performance and the crystal structure of the self-passivated [Hf2S]2+∙2e electride catalyst were persistently maintained over 1 week (5000 cycles; figs. S4 and S7), demonstrating that a sustainable electride catalyst even in strong acid-containing solutions is successfully developed.

Fig. 5 Schematic model for self-passivation and electrocatalytic reaction of [Hf2S]2+∙2e electride.

Structural model and possible pathway for HER over the self-passivated [Hf2S]2+∙2e electride as a working electrode. Volmer (H+ + e → Hads, ion discharge with the formation of adsorbed hydrogen, Hads), Heyrovsky (H+ + e + Hads → H2, subsequent desorption of Hads), and Tafel (2Hads → H2, recombination of Hads) reactions occur at the surface of a-HfO2 with transferring excess electrons from electride.


In summary, the development of the self-passivated [Hf2S]2+∙2e electride with persistent electrocatalytic characteristics presents a promising opportunity as a sustainable electride catalyst under reactive atmospheric conditions. From the successful demonstration of outstanding chemical stability originated from the double amorphous layer structure, which is the key factor of self-passivation mechanism in the Hf-based electride, the design strategy for a stable electride is readily extendable to multinary compounds composed of ubiquitous elements that have a functional passivation layer, allowing one to tune the surface properties. In addition to the unprecedented electrochemical catalytic reactions, we anticipate that our work will stimulate further research for exploiting electronic devices with long-term stability in self-passivated electrides.


Sample preparation

All synthetic steps were carried out in glove boxes filled with high-purity argon gas (Ar 99.999%) to minimize the oxidation of raw materials and synthesized samples. Stoichiometric polycrystalline [Hf2S]2+∙2e electrides were synthesized by the arc melting method under high-purity Ar atmosphere. We mixed and ground Hf powder (Alfa Aesar; 99.6%) and S trace metal basis (Sigma-Aldrich; 99.998%) in a 2:1 molar ratio. To minimize the vaporization of S via the arc melting method, mixed powder was pelletized and sintered at 500°C for 5 days in vacuum-sealed quartz tubes (a few millitorr) and then the quartz tubes were water-quenched down to room temperature, so that sintered pellets could be the mixture of several Hf-S intermediate compounds, such as HfS2, HfS, HfS3, and Hf2S. These intermediate compounds were melted in the Ar-filled arc melting chamber over 2000°C to fabricate polycrystalline coin- and rod-shaped [Hf2S]2+∙2e electride samples. To obtain a single phase and ensure the homogeneity of the samples, we repeated the melting process at least three times. Coin-shaped polycrystals were used to make powder for structural characterization and fabricate working electrodes for electrochemical experiments. Single-crystalline [Hf2S]2+∙2e electrides were grown by the floating zone melting method using polycrystalline rod-shaped [Hf2S]2+∙2e electrides and pieces of seed. Around 3 atm of pressure was applied into the chamber with high-purity Ar gas to prevent vaporization of S during the process. The growing speed was 1 to 3 mm/hour, and the rotation speed of feed and seed rods were 4 to 10 rpm because of the low-melt viscosity of the [Hf2S]2+∙2e electride.

Amorphous HfO2 thin films are prepared by the pulsed laser deposition method at room temperature. Using pure hafnium metal target, Hf films were deposited on a Cr/Au-coated SiO2 substrate under an Ar pressure of 10−6 torr by using KrF excimer laser (λ = 248 nm) with a fluence of 6.5 to 7.0 J cm−2 and a reputation rate of 5 Hz. Then, the Hf films were exposed to air to form a-HfO2, similar to the formation of an a-HfO2 layer on the [Hf2S]2+∙2e electride. The commercial HfO2 powders (Alfa Aesar; 99.95%, 325 mesh powder) were also used for the HER experiments.

Structural characterization by XRD

The crystal structure of Hf2S was investigated by XRD using a Rigaku SmartLab diffractometer with monochromatic Cu Kα radiation (8.04 keV) at room temperature. The well-ground powder samples and single crystals were prepared in glove boxes and measured in a plastic dome-type stage filled with Ar gas to avoid oxidation during measurements. Rietveld refinements of the powder XRD data were conducted with the GSAS-II software. To confirm the effect of oxidation to the crystal structure of the [Hf2S]2+∙2e electride, we prepared unoxidized powder handled in Ar-filled glove boxes and oxidized powder neglected in the air and water over 2 months. Then, the same XRD measurement procedures were conducted for each sample. The UHV-cleaved sample in Fig. 3A was prepared in glove boxes filled with high-purity Ar gas by cutting and pulverizing to make fresh surfaces.

DFT calculations

DFT calculations were performed using the generalized gradient approximation with the Perdew-Burke-Ernzerhof functional and the projector augmented plane-wave method implemented in the Vienna ab initio simulator program (VASP) code (3234). We used a unit cell that contains two formula units of Hf2S (see Fig. 1A and fig. S1). The wave function of electrons was expanded in a plane-wave basis set with a cutoff energy of 500 eV. Reciprocal space was sampled with a 17 × 17 × 4 Monkhorst-Pack k-point mesh for ELF (35), charge density, and partial charge density calculations. We performed geometric optimizations to relax the positions of ions and cell shape until the energy change between two successive ionic/cell relaxation steps was less than 10−5 eV. An empty sphere with a Wigner-Seitz radius of 1.207 Å was used to obtain the projected band structure and DOS on the position of the excess electron (“e” site). The local (atomic) charge of each ion including the excess anionic electron was computed by Bader’s charge decomposition method (36). The Bader volume for each site is computed as the volume containing a single charge density maximum and is separated from other volumes by a zero-flux surface of the gradients of the charge density.

Photoemission spectroscopy measurements and characterization

XPS and UPS measurements were performed to investigate chemical and electronic states of [Hf2S]2+∙2e electride surface. To obtain the O- and C-free fresh surface of [Hf2S]2+∙2e electrides, we cleaved the (00l) plane of the single-crystalline [Hf2S]2+∙2e electride (referred to as UHV-cleaved sample) under ultrahigh vacuum (UHV) conditions (~10−10 torr) with a post attached with Torr Seal before photoemission spectroscopy measurements. For oxidized samples (air- and water-exposed), the surface was cleaned up with 300°C heating in the chamber maintaining UHV condition to reduce the contamination of hydrocarbon, which is easily adsorbed to the sample surface under ambient atmosphere. Then, the measurement was carried out under a vacuum of ca. 3 × 10−10 torr. The photoemission results were obtained using a Scienta R4000 electron analyzer with an Al Kα x-ray source (1486.7 eV) and a discharge lamp that emitted an excitation line of He Iα (21.2 eV). Core spectra of Hf 4s and 4f, S 2p, and O 1s were measured for analysis, and all spectra were fitted by comparing the binding energies of the component peaks. Because the cleaved surface of the [Hf2S]2+∙2e electride is metallic, the Hf component peak was fitted with an asymmetric line shape, which arises from final-state effects. Every peak was calibrated by the spectrum of hydrocarbon on the sample and reference Au attached next to the sample. In the WF measurement with UPS, the negative bias voltage of 15 V was applied to clarify that the vacuum level of the sample exceeds that of an electron analyzer.

Electrical transport measurements and characterization

The resistivities and Hall effect measurements were performed to confirm the oxidation effects on the electrical properties. Three different samples, which are cleaved in the glove boxes (UHV-cleaved), exposed to air for 1 month (air-exposed), and immersed in the water for 1 month (water-exposed), were prepared and Au was deposited as electrodes with a thermal evaporator equipped in a glove box onto the surface of the samples to eliminate the noise from contact resistance with four-point and Hall bar configurations. After the contacts were made, Apiezon N grease was coated onto the sample surfaces to prevent further oxidation during measurements. The temperature-dependent resistivities and Hall effect measurements were performed by the physical property measurement system (PPMS DynaCool, Quantum Design). The carrier concentrations (NH) were derived from the Hall coefficient (RH) with the assumption that the electron mobility is much higher than the hole mobility. The equation of the Hall coefficient is given byRH=Nhμh2Neμe2e(Nhμh+Neμe)2

Here, Ne is the electron density, Nh is the hole concentration, μe is the electron mobility, μh is the hole mobility, and e is the elementary charge. Moreover, the Hall coefficient may become a simple formula, ∣RH ∣ = 1/Ne with the assumption above. The carrier concentration and mobility (μ) of the [Hf2S]2+∙2e electride in the inset of Fig. 3B is deduced from this simple formula. The UHV-cleaved sample in Fig. 3B was prepared in glove boxes filled with high-purity Ar gas by cutting and cleaving to make fresh surfaces.

TEM observation

TEM experiments were performed using a double-corrected TEM (JEM-ARM200F, JEOL) equipped with a cold field emission source, operating at 200 kV. To prevent oxidation of the TEM sample, a single-tilt vacuum transfer holder (Model VTST-4006, Gatan) was used for TEM experiments (UHV-cleaved samples). TEM samples for HRTEM and EELS were prepared as powder cleaved from a single crystal in the high-purity Ar-filled glove boxes. Dual-beam focused ion beam (AURIGA CrossBeam Workstation, Carl Zeiss) was also used to obtain atomic-scale STEM images and chemical EDX mapping data with JEOL-EDX in the STEM imaging mode. Dual-EDX detector was used to acquire atomic-scale EDX mapping data of the sample, and each detector has an effective detection area of 100 mm2. The sample drift during the acquisition was compensated by tracking the reference atom position assigned at the beginning of the acquisition. EELS experiments were performed using the same microscope equipped with a GIF system (Model 965 GIF Quantum ER). The spectrometer was set to an energy dispersion of 0.2 eV channel−1, and the energy resolution at the zero-loss peak is 0.9 eV.

In situ TEM heating experiments were conducted using a double-corrected TEM (JEM-ARM300F, JEOL) operated at 300 kV with Si micro-electromechanical systems–based in situ TEM heating holder (Wildfire, DENSsolutions). The specimen for the heating experiment was prepared by Focused Ion Beam (FIB) lift-out (Helios NanoLab 450 F1, FEI). The temperature was increased to 900°C in 100°C with a ramping rate of 1°C/s after stabilization for ~5 min at each temperature. TEM images were recorded with a complementary metal-oxide semiconductor camera with 4k resolution (OneView, Gatan).

Electrochemical characterization

The electrochemical testing took place in a three-electrode configuration. A graphite rod was used as a counter electrode to avoid Pt contamination effects, Ag/AgCl was used as a reference electrode, and the bulk self-passivated [Hf2S]2+∙2e electride disks (4 mm by 4 mm by 0.1 mm) were used as a working electrode. All tests were investigated in 50 ml of 0.5 M sulfuric acid (H2SO4) electrolyte. The applied potential was 0.05 to −1.0 V and the sweep rate was 5 mV/s. All conducting parts except for the reacting surface were covered with transparent nail polish. Commercial HfO2 powders with glassy carbon electrode and 10- and 100-nm-thick HfO2 film were also tested as a working electrode to compare HER activities with the self-passivated [Hf2S]2+∙2e electride.


Supplementary material for this article is available at

This is an open-access article distributed under the terms of the Creative Commons Attribution-NonCommercial license, which permits use, distribution, and reproduction in any medium, so long as the resultant use is not for commercial advantage and provided the original work is properly cited.


Acknowledgments: Funding: This work was supported by a National Research Foundation of Korea (NRF) grant funded by the Korean government (Ministry of Science & ICT) (no. 2015M3D1A1070639) and supported by the Institute for Basic Science (IBS-R011-D1). Author contributions: S.W.K. conceived the idea and organized the research. C.N.N., S.-G.K., J.B., K.L., Y.M., and S.W.K. carried out the computational studies. S.H.K. fabricated polycrystals and grew the single crystals. S.H.K. performed the XRD, XPS/UPS, and electrical transport measurements. S.H.K., K.H.L., K.L., S.-G.K., and S.W.K. analyzed the structural, electrical, and electronic properties. S.H.K., G.H., S.L., and Y.-M.K. carried out TEM experiments. S.H.K., S.L., S.H.O., Y.-M.K., and S.W.K. analyzed the TEM results. S.H.K. and K.C. performed electrochemical experiments and analyzed the data. All authors discussed the results and contributed to writing the manuscript. Competing interests: The authors declare that they have no competing interests. Data and materials availability: All data needed to evaluate the conclusions in the paper are present in the paper and/or the Supplementary Materials. Additional data related to this paper may be requested from the authors.
View Abstract

Stay Connected to Science Advances

Navigate This Article