Activation of H2O Tailored by Interfacial Electronic States at a Nanoscale Interface for Enhanced Electrocatalytic Hydrogen Evolution.

Despite the fundamental and practical significance of the hydrogen evolution reaction (HER), the reaction kinetics at the molecular level are not well-understood, especially in basic media. Here, with ZIF-67-derived Co-based carbon frameworks (Co/NCs) as model catalysts, we systematically investigated the effects of different reaction parameters on the HER kinetics and discovered that the HER activity was directly dependent not on the type of nitrogen in the carbon framework but on the relative content of surface hydroxyl and water (OH-/H2O) adsorbed on Co active sites embedded in carbon frameworks. When the ratio of the OH-/H2O was close to 1:1, the Co/NC nanocatalyst showed the best reaction performance under the condition of high-pH electrolytes, e.g., an overpotential of only 232 mV at a current density of 10 mA cm-2 in the 1 M KOH electrolyte. We unambiguously identified that the structural water molecules (SWs) in the form of hydrous hydroxyl complexes absorbed on metal centers {OHad·H2O@M+} were catalytic active sites for the enhanced HER, where M+ could be transition or alkaline metal cations. Different from the traditional hydrogen bonding of water, the hydroxyl (hydroxide) groups and water molecules in the SWs were mainly bonded together via the spatial interaction between the p orbitals of O atoms, exhibiting features of a delocalized π-bond with a metastable state. These newly formed surface bonds or transitory states could be new weak interactions that synergistically promote both interfacial electron transfer and the activation of water (dissociation of O-H bonds) at the electrode surface, i.e., the formation of activated H adducts (H*). The capture of new surface states not only explains pH-, cation-, and transition-metal-dependent hydrogen evolution kinetics but also provides completely new insights into the understanding of other electrocatalytic reductions involving other small molecules, including CO2, CO, and N2.

Introduction

The energy crisis and environmental issues have stimulated extensive research on alternative energy storage and conversion systems.[1,2] The hydrogen evolution reaction (HER) holds the key to the future hydrogen economy by controlling the efficient generation of hydrogen from water and harvesting clean energy in the reverse direction.[3,4] Despite their significance in both practical and fundamental aspects, the HER kinetics have been elusive and clouded by several long-standing puzzles, such as the anomalous hydrogen evolution behavior in high-pH environments[5−9] (generally, the emerging HER electrocatalysts achieve high activities in acidic electrolytes),[10−13] the extreme sensitivity of the surface electronic structure to the catalyst (for example, the physical origin of transition-metal- or oxide-doping and hydroxyl-adsorption-induced catalytic enhancement, Scheme a),[14] the promoter effects of alkali metal cations on the electrochemical reduction,[15−17] and even unexpected hydrogen evolution reaction activities on metal-free electrocatalysts.[8,11,18−23]

Scheme 1

(a) Surface chemisorption-bond- or noncovalent-interaction-dominated reaction kinetics of the HER. (b) Metal-centered d band theory, where the d-band center is tuned relative to the Fermi levels of the metals to optimize the electrocatalytic activity as a classical model. (c) Structural water molecule (SW) dominated p band intermediate (or transient) state model, which explains the reaction kinetics of the HER by providing alternative channels for proton and electron transfer. Herein, different from the traditional hydrogen bonding of water, the hydroxyl (hydroxide) groups and water molecules in structural water are mainly bonded through the spatial interaction between the p orbitals of O atoms, showing characteristics of a delocalized π-bond with a metastable feature. (d) When the metal is present, the ultra-fast switching between surface chemical bonds upon excitation can be written in a simple formula of 2HOM = HOO + HMM = 1, where H is the orbital hybridizations between surface atoms and the subscripts of OM, OO, and MM represent the chemical bond interactions between the surface metal and the O atom, O and the O atom in the SW, and the metal and the metal atom, respectively. (e) Energy level diagram of atomic orbital hybridization at the interface of the electrode. This scheme was taken from the modified version of Figure 5 in ref (85). At the nanoscale interface, on the one hand, SW as a bridging ligand can promote interfacial electron transfer by an inner sphere electron transfer mechanism; on the other hand, it can stabilize the transition states of water molecules during the dissociation of O–H bonds.

Recently, significant efforts have been focused on activity descriptors or design principles of catalysts by tuning the covalent and noncovalent interactions at the electrical double layer (EDL), as the surface electronic structure factor is a key parameter to elucidate the molecular-level origin of HER microkinetics[24] (Scheme a). In the models of covalent interactions, the d band center theory is widely applied by tuning the d band center relative to the Fermi levels of metals, which can control the binding energies of surface adsorbates such as adsorbed hydrogen (H*) or hydroxyl (OH*) and consequently optimize the electrocatalytic activity (Scheme b). Thus, several important reaction descriptors were proposed for the rational design and development of catalysts, including the hydrogen binding energy,[25,26] the OH–M bond strength (M denotes a transition-metal cation),[27−29] the eg occupancy,[30] etc. Even though case-to-case studies show the paramount importance of tuning the reaction kinetics using electronic factors, so far no descriptor can perfectly explain all the observed activity trends. Very recently, owing to the fast development of in situ operando spectroscopic and computational techniques for investigating interfacial water, more evidence has shown that the structure of interfacial water, which is dominated by weak noncovalent interactions, plays a key role in determining the reaction rate.[9,15,24,28,29,31−34] Additionally, some very important intermediate species in the EDL region were captured, such as the hydronium (H3O+) intermediate, even in high-pH electrolytes (its nature is not clear),[10] hydrogen-bonded and hydrated Na+ ion water, OHad–M+(H2O) clusters,[28] and hydroxyl–water–cation adducts.[15,35] However, it is still unclear how the interaction mode of the complexes formed by interfacial water, surface hydroxyl, or both with alkali metal ions or transition-metal ions and the formed electronic states of these surface structures affect the proton and electron transfer process at the nanoscale interface (Scheme a).

Recently, with the help of steady and transient absorption and emission spectra, we unambiguously confirmed that structural water molecules (SWs) adsorbed or confined at the nanoscale interface in the form of OH–·H2O could emit bright colors as nonconventional luminophors,[36−39] answering a century of debate over whether water is a colored or colorless liquid.[40,41] Differing from the traditional hydrogen bonding of water, the hydroxyl (hydroxide) groups and water molecules in structural water are mainly combined through the spatial interaction between the p orbitals of O atoms, which is a new kind of weak interaction with the characteristics of a conjugated π-bond and a metastable feature (Scheme c). Very interestingly, when water molecules are adsorbed onto the metal centers, due to the spatial interaction between the adjacent O atoms, a pair of chemisorption O–M bonds (M can be different metals) perpendicular to the metal surface can be very quickly switched to two interface states or bonds parallel to the metal surface upon excitation,[37] showing the feature of electron delocalization at the nanoscale interface (Scheme c–e). We call the interaction between two O atoms on the interface the p band intermediate or transient state (PBIS),[42] which has typical π-bond characteristics. The interaction between two metal atoms due to the redistribution of atomic orbitals is the well-known metallophilic interactions,[43−45] for example, the aurophilic interaction of Au–Au bonds.[46] It is important to note that these newly formed surface bonds or states have defined molecular energy levels, as evidenced by their adsorption, excitation, emission and ultrafast transient absorption spectra.[37,39] According to quantum mechanics, this dynamic process can be expressed in a simple formula of 2HOM = HOO + HMM = 1, where H is the orbital hybridization of surface atoms (Scheme d). If the interaction between two oxygen atoms (HOO) comes from the hydrous hydroxide or hydroxyl complex (or group), we call it SW. Obviously, the stability of SWs is strongly dependent on the microenvironment and the microstructure of catalyst, such as the strengths of acids and alkalis (pH), the surface hydrophobicity, and the size and composition of the active components of the catalytic active sites. In fact, using our newly developed PBIS model, we successfully explained the catalytic conversion mechanism of the selective hydrogenation of organic compounds containing unsaturated functional groups such as nitro and carbonyl groups, which involves multiple electron and proton transfers at the molecular level.[47−49] Since the electron and proton transfer process of traditional metal nanocatalysts for the selective reduction of organic compounds containing nitro groups is very similar to that of an electrocatalyst made of metal nanoparticles for the HER, the microkinetics of both reactions should follow the same working mechanism. The only difference is that in the HER the cathode is the supplier of electrons, while in the hydride catalytic reduction of nitro groups the adsorbed borohydride at the metal surface is an electron donor.[47,48] This exciting idea promotes us to study the reaction kinetics of the HER at metal nanocatalysts, since it is considered the simplest electrochemical reaction.

Herein, using Co-based nanocarbon frameworks (Co/NCs) derived from Co zeolitic imidazolate frameworks (ZIF-67) as the model catalysts,[50] we systematically study the effects of different reaction parameters on the HER kinetics, including the compositions of the nanocatalysts (the content of N and the degree of graphitization of carbon), the pH, the type and concentration the alkaline metal cation, the type of transition metal, etc. We surprisingly find that the HER activity is directly dependent not on the type of nitrogen in the carbon framework but on the relative contents of surface hydroxyl and water (OH–/H2O) adsorbed on Co active sites embedded in the carbon frameworks. Interestingly, when the OH–/H2O ratio is close to 1:1, the Co/NC nanocatalyst shows the best reaction performance under the condition of high-pH electrolytes, e.g., an overpotential of only 232 mV at a current density of 10 mA cm–2 in 1 M KOH electrolyte, even better than 5 wt % Pt/C. Furthermore, the characterization of the optical excitation and emission spectra together with 1H NMR spectroscopy evidence the presence of SWs in the alkali metal hydroxide solution, where the SWs construct unique surface states for bright color emission and proton activation even in the absence of Co/NC nanocatalysts. Using the SW-dominated surface intermediate state model within the double-layer region (Scheme ), the pH-, alkali-metal-cation-, and transition-metal-dependent reaction kinetics of the HER can be easily elucidated, where SWs as intermediates (or bridging ligands) greatly accelerate electron transfer via inner sphere electron transfer mechanics and concomitantly stabilize the transition state of the dissociation of the O–H bonds of water at the nanoscale interface. Our work provides a comprehensive understanding of the reaction kinetics of all redox reactions involving the electron and proton transfer at the nanoscale interface, not just the electrocatalysis reaction (and even enzyme catalysis), and thus provides opportunities to tune kinetics by subtly altering the interface states in addition to the catalyst’s electronic structure.

Scheme 2

Adsorption Mode of Hydrated Alkaline Metal Cations (AM+) of (a) Li+ and (b) Cs+ in the Electrical Double Layer (EDL) Region on the Electrode Surface

Under alkaline conditions, due to the adsorption of hydroxide groups (OH–), the negatively charged electrode surface strongly binds AM+ via electrostatic interactions in the EDL region. In the inner Helmholtz plane (IHP), SWs are strongly bound to the Co active sites (◊), showing the negatively charged surface; in the outer Helmholtz plane (OHP), Li+ and Cs+ are both densely adsorbed onto the electrode surface, while K+ cations with a medium hydration ability are loosely adsorbed onto the electrode surface (not shown). The thick dashed green line indicates the p orbital overlaps of O atoms in SWs, and the dashed blue line indicates H bonding; for clarity, the p orbital interactions between O atoms in the reactant water and O atoms in the SWs are not shown. At the nanoelectrode surface, SWs with the characteristics of π-bonds act as bridging ligands to accelerate interfacial electron transfer via the surface delocalization of SWs through the inner sphere electron transfer mechanism of Marcus theory; on the other hand, they concomitantly act as catalysts (or catalytic active sites) to activate the O–H bond of water via p orbital interactions between O atoms of the water reactant and SWs, promoting the breakage of the O–H bond by lowering the reaction barrier for water dissociation.

Results and Discussion

Preparation and Characterization of Co-Based Carbon Framework Catalysts Containing Nitrogen Atoms (Co/NC)

ZIF-67 were synthesized according to the reported method (SI).[8,50,51] The obtained ZIF-67 showed a well-defined polyhedron morphology with a size of ca. 740 nm, which was confirmed by the X-ray diffraction (XRD) pattern (Figure S1a) and the scanning electron microscopy (SEM) image (Figure S1b). Then, using high-temperature pyrolysis, the ZIF-67 nanoparticles were transferred to Co-based carbon framework catalysts (Figures a and b and S2), which were named Co/NC-T (T stands for the pyrolysis temperature). XRD patterns and TEM images in Figure S3 showed that the Co ions were self-reduced by carbonized organic linkers, yielding Co nanoparticles and atomically dispersed Co atoms that were anchored on nitrogen-doped porous carbon (Figure ).[50] Upon acid leaching, most of the Co NPs on the external surface of the C framework (Co/NC-T catalysts) were removed, but some very small Co NPs less than 10 nm could also be observed by XRD patterns and in the TEM images (Figures , S4, and S5). The close examination by TEM revealed more structural information on carbon frameworks as the pyrolysis temperature increased from 700 to 1000 °C (Figure S5). The size of the Co/NC NPs dramatically decreased to ca. 400 nm due to the shrinkage of the C frameworks (Figure S5). At the same time, with the great improvement of the reduction and graphitization, the content of Co in the C frameworks became lower, which was evidenced by the inductively coupled plasma atomic emission spectrometry (ICP-AES) analysis (Table S1) and the X-ray photoelectron spectroscopy (XPS) analysis (Figure S6 and Table S2). However, the primitive polyhedron morphology of the ZIF-67 NPs remained almost intact. In a word, using ZIF-67 as a carbon precursor, Co/NC-T catalysts were successfully synthesized through the high-temperature pyrolysis method (Figure ).

Figure 1

(a) SEM image (the inset is the particle size (PS) histogram), (b) TEM image, (c) EFTEM image, and (d–g) elemental mapping images of the Co/NC-800 catalyst.

Screening of the Best Co/NC-T Catalyst for the HER

We first investigated the effect of the pyrolysis temperature of the catalysts on the HER activity. We found that, under both acidic and base conditions, Co/NC pyrolyzed at 800 °C exhibited the best catalytic performance, even better than that of the commercial catalyst Pt/C (Figure ). It is important to note that even though Co/NC-800 was acid-leached, the Co NPs could not be completely removed, indicating that these Co NPs were embedded or sealed in the C framework (Figure a–c) and certainly could not be diffusively accessed by water molecules (or hydronium). In addition, without the acid-washing pretreatment, the Co NP Co/NC catalysts showed lower HER activities in both acidic and alkaline electrolytes (Figure S7), which clearly precluded the contributor role of metallic cobalt or cobalt oxide nanoparticles with larger particle sizes as active sites for the HER.[52−56] Thus, these embedded Co NPs or cobalt oxide NPs with larger particle sizes could not be active sites for the HER. In addition, if cobalt is the active site for the HER, the activity should be proportional to the final loading of Co in the carbon frameworks. However, this is not the case. We found that as the content of Co decreased, the reaction activity first increased and then decreased, reaching a peak at the medium pyrolysis temperature of 800 °C under both acidic and base conditions. This suggests again that the content of Co in the carbon framework is not the decisive factor that determines HER reaction activity, even though it is very important. Interestingly, the elemental mapping results showed that, besides the N and O atoms, the Co atoms were homogeneously and atomically distributed in the carbon framework component (Figure d–-g), confirming that Co–C–N was the possible active site in Co-based carbon catalysts for the HER.[6,57,58] However, the precise composition of the active sites and how it affected the reaction kinetics of the HER were yet clear, especially for such pH- and alkali-metal-cation-dependent effects. Since the Co/NC-800 catalyst was the best catalyst, we used it as a model catalyst to study the reaction kinetics of the HER.

Figure 2

HER polarization curves and Tafel plots, respectively, of Co/NC-T catalysts in (a and c) 1.0 M KOH and (b and d) 0.5 M H2SO4.

Identification of the Catalytic Active Site for the HER

In fact, it is really challenging to establish direct correlations between the electronic structure and the catalytic activity trend of Co/NC materials because of delicate changes of the microenvironment of Co/NC,[59] such as the graphitization degree of the carbon frameworks[60] and the valence states of N atoms.[52,61−64] Graphitized carbon, which was usually formed via catalysis over Co, Fe, etc., was vital for the electrical conductivity of the catalyst. As the pyrolysis temperature increased, the graphitization degree of the carbon frameworks also gradually increased, as evidenced by thermogravimetric (TG) analysis (Figure S8), C 1s XPS spectra (Figure S6i–l and Table S2), and XRD patterns (Figures S3a and Figure S4). However, as the graphitization degree increased , the reaction activity first increased and then decreased, reaching a peak at the medium pyrolysis temperature of 800 °C under both acidic and basic conditions (Figure ). Obviously, our observation showed that the graphitization degree of the carbon frameworks, although important, was not the decisive factor that determined the HER reaction activity.[53]

Recently, solid evidence showed that the HER activity of transition-metal (TM) based carbon catalysts was dependent on the type and valence state of N atoms in carbon frameworks, and the reaction activity increased monotonically with the content of N in a specific valence, such as pyridine-N, pyrrole-N and graphite-N.[8,65−68] XPS analysis of the N 1s peak allowed the identification of different concentrations of the N functionalities in the Co/NC-T catalysts. The high-resolution N 1s spectra of the Co/NC composite (Figure e–h) were deconvoluted into four peaks at 397.6, 398.5, 399.9, and 401.5 eV, corresponding to pyridinic-N, Co–N, pyrrolic-N, and graphitic-N. As the pyrolysis temperature increased from 700 to 1000 °C, the content of pyridinic-N gradually decreased, while the contents of pyrrolic-N and graphitic-N were increased monotonically as expected due to the transformation of pyridinic-N; the value of Co–N remained almost constant at 24.0% (Table S2).

Figure 3

O 1s and N 1s XPS spectra, respectively, of (a and e) Co/NC-700, (b and f) Co/NC-800, (c and g) Co/NC-900, and (d and h) Co/NC-1000.

Several research groups have reported that pyridinic-N is the active site that enhances the activity of N-doped carbon materials,[61,62] whereas others have suggested that more graphitic nitrogen atoms, rather than pyridinic ones, are important for the activity.[63,64] Hence, the exact catalytic role of each of the nitrogen forms in nanocarbon HER catalysts is still a matter of controversy.[69] Even though our results for N 1s XPS spectra of Co/NC-T did not build a direct correlation between the type and content of N atoms in carbon framework and the HER catalytic activity (Figure and Table S2), one must not neglect the influence of N on the catalytic activity, since it greatly affects the local pH of the catalytic active sites. Recently, several reports showed that the adsorption of hydroxyl groups onto the metal center significantly promoted the reaction kinetics of the HER.[9,10,15,28,29,31−33,35,70] There is a certain correlation between the type and amount of N and the number and bonding strength of surface hydroxyl or hydroxide groups, which ultimately determines the activity of the HER reaction. Thus, we shifted our concentration to analyzing oxygen-associated species using the O 1s XPS spectra. For Co/NC catalysts, the spectrum could be deconvoluted into four peaks centered at 530.0, 531.3, 532.5, and 534.3 eV (Figure a–d, respectively), which were assigned to Co–O, OH*, H2O*, and C–O, respectively.[71−73] The XPS O 1s spectra analyses verified the presence of a hydrous hydroxide complex on the Co metal centers of the Co/NC-T catalysts. Very interestingly, when the ratio of OH* to H2O was close to 1:1 (Table S2), Co/NC-800 catalyst exhibited the best catalytic performance for the HER (Figure a and b). This hydrous hydroxide complex was recently proposed to be SW, which could act as the intermediate species for surface electron transfer and stabilize the reaction transition states (Scheme d).[36−39,47−49] Thus, we hypothesized that the interfacial adsorbed SWs probably provide an alternative channel for surface electron transfer and concomitantly stabilize the transition state from the cleavage of the O–H bonds of water, promoting the dissociation (or activation) of water at the electrode and consequently accelerating the HER reaction kinetics. This also probably explained the important role of the doping N atoms (the type and content of N atoms in the carbon frameworks) in tuning the catalytic activity by changing the alkaline microenvironment of the Co coordination sphere.

In addition, the capture of SWs at the Co center also explained the higher activity of Co/NC in basic media (1.0 M KOH) compared to that in the acidic electrolyte (0.5 M H2SO4), which is in the linear sweep voltammetry (LSV) polarization curves (Figure a and b, respectively). Accordingly, electrocatalytic activity of Co/NC in 1.0 M KOH exhibited a smaller overpotential (232 mV at 10 mA/cm) and a lower Tafel slope (99 mV/decade) compared with that in 0.5 M H2SO4 (308 mV at 10 mA/cm and 215 mV/decade, respectively) (Figure ). When compared with Pt/C’s overpotential in basic media, interestingly, at applied potentials slightly higher than its onset potential (<0.25 V vs RHE), Pt/C actually gave a higher current density than the Co/NC catalyst. Otherwise, the Co/NC catalyst exhibited a smaller overpotential, which meant a greater electrocatalytic activity than even the commercially available Pt/C catalyst (Figure a). It is important to note that, in fact, our Co/NC-800 catalyst also exhibited a superior HER performance with a very small overpotential (308 mV) in 0.5 M H2SO4, which was very close to that of Pt/C catalyst (228 mV) at 10 mA/cm (Figure b). Our results were in strong contrast with previous reports of the activity being higher in an acidic electrolyte than in an alkaline electrolyte,[6,74] implying that the chemically adsorbed SWs at the electrode interface may play a decisive role in the HER. Although the performance of Co/NC is not the best among the reported HER catalysts, including both nonprecious and precious catalysts,[8,75−79] the concept of SWs as intermediate that regulate surface-coupled proton and electron transfer and stabilize the transition state of water dissociation provides completely new insights for understanding the mechanism of the HER at the molecular level.

The pH- and Alkali-Metal-Cation-Dependent Hydrogen Evolution Kinetics

To further verify the effect of adsorbed hydroxyl or hydroxide on the HER, the activity was tested in different concentrations of the KOH electrolyte. As can be seen in Figure a, the activities of the catalyst in different electrolytes followed the order of 1.0 M KOH ≈ 2.0 M KOH ≈ 3.0 M KOH > 0.1 M KOH ≫ 0.01 M KOH. This suggests the significant influence of the density of surface hydroxyl or hydroxide groups adsorbed on the metal surface on the reaction kinetics of HER as well as a saturated concentration, as the adsorption of a small amount of OH– was not conducive to the construction of interface state. On the contrary, the introduction of excessive OH– inhibited the formation of interfacial active H species (or adsorbed H species, Figure S12b). This indicates that neither small amounts or excessive amounts of hydroxyl or hydroxide groups are good for the formation of SWs (proper distance and configuration should be maintained between the two O atoms) at the metal nanoscale interface, as evidenced by the change of the Tafel slopes of the HER for the various concentrations of KOH solutions (Figure b). Tafel slopes in different electrolytes were compared in a similar range to avoid the influence from the high overpotential polarization and formed bubbles. It can be seen that the Tafel slope of Co/NC in the low [OH–] environment is greater than 120 mV/dec (Figure b), indicating that the RDS of the HER is a water dissociation step, i.e., the breakage of the O–H bond. However, when [OH–] increased to 1.0 M, the Tafel slope unexpectedly decreased to 99 mV/dec and even lower in higher-concentration KOH solutions (∼92 mV/dec in the 2.0 M KOH solution and ∼85 mV/dec in the 3.0 M KOH solution), suggesting that the RDS was either the proton recombination (Volmer–Tafel) step or the electrochemical desorption (Volmer–Heyrovsky) step on the basis of the conventional understanding of the reaction kinetics. Such base-concentration-dependent reaction kinetics were also observed in LiOH-based electrolytes (Figure S9), indicating the similar reaction mechanism in highly concentrated alkaline solutions. It is important to note that in the conventional understanding of the kinetics of the HER reaction, the breakage of O–H bonds, the electron transfer, and the subsequent activation of H atoms (H*) are chronologically isolated events instead of a synergetic process. Nevertheless, our results for the kinetics of the HER determined by Tafel slope calculations show that the boundaries of the three RDSs, i.e., the dissociation of the O–H bond of water, the formation of activated H*, and the desorption of H2, are not clear in alkaline conditions as they are cooperative processes (concerted electron and proton transfer process).

Figure 4

HER polarization curves and Tafel plots, respectively, of the Co/NC-800 catalyst under different concentrations of (a and b) KOH and (c and d) 1.0 M AMOH (AM+ = Li+, Na+, K+, Rb+, or Cs+).

In addition, it is well-known that the type of alkali metal cations significantly affects the kinetic of the HER. Thus, we further studied the HER performance in various AMOH solutions (AM+ = Li+, Na+, K+, Rb+, and Cs+). To our surprise, the trend of the HER activity did not follow the structure-making (or hydration) tendency of alkali metal cations (AM+) in alkaline solutions of 1.0 M AMOH (in the order of Cs+ < Rb+ < K+ < Na+ < Li+). Interestingly, our Co/NC-800 catalyst showed the highest activity in the 1.0 M KOH solution. Obviously, just considering the effect of hydration ability of AM+ on the catalytic activity is not enough to understand the microkinetics of the HER. Another overcrowding effect of the AM+ bound on the electrode surface, which blocks the active sites at the electrical double layer (EDL) region of the electrode, has to be also considered.[24]

Identification of SWs with a PBIS Using the Optical Spectrum

Combining the SW-dominated PBIS with the overcrowding effect of AM+ cations, we proposed a simple model to explain the alkali-metal-cation-dependent kinetics of the HER (Scheme ). In alkaline conditions, due to the adsorption of hydroxide groups (OH–), the negatively charged electrode surface strongly binds the AM+ through electrostatic interactions in the EDL region. In the inner Helmholtz plane (IHP), SWs are strongly bound to the Co active sites, showing the negatively charged surface. In the outer Helmholtz plane (OHP), Li+ and Cs+ are both densely adsorbed onto the electrode surface (Scheme a and b, respectively), while K+ cations with a medium hydration ability are loosely adsorbed on the electrode surface (not shown). Owing to the weak hydration capacity, the bulky Cs+ cations are strongly naked adsorbed with more free water at the electrode surface (a local hydrophobic surface is formed similar to the hydrophobic nanocavity in the metalloenzymes), which blocks the diffusion of water to the Co active site with SWs and consequently lowers the activity of the HER. In fact, if we just consider the electrostatic interaction, the electrode surface should be more densely covered (or packed) upon the introduction of Li+, which has the strongest structure-making tendency, due to the formation of a local “Water-in-salt” structure at the nanoscale interface[80] and thus should show the lowest chemical reactivity due to the blockage of the Co active site. However, this is not the case. There are two reasons to explain this abnormal observation of the HER activity. One reason is that even though the Li+ cation densely covers the electrode surface, its own strong hydration ability can continuously supply water to the Co active sites, leading to the high activity. If this is true, from the perspective of being more conducive to the mass transfer of the water reactant, its should show the best catalytic performance, higher even than that in the KOH solution. Obviously, it is not the case. The other reason is that, owing to its stronger hydration ability, Li+ in the alkaline solution can build an interface state of SWs by itself that is similar to the Co active site, with a common formula of {OHad·H2O@M+}; here, however, M+ is Li+ rather than Co transition metal. Since the catalyst in the KOH solution shows the best catalytic activity (Figure c), the latter explanation is obviously more reasonable, i.e., {OHad·H2O@Li+} is an active site for the HER due to the strong hydration capacity of Li+ but its activity is less than that of {OHad·H2O@Coδ+}.

Excitation and emission spectra of various alkali-metal-hydroxide DMSO solutions verified the above assumption that Li+ could form more stable SWs relative to those of other AM+ for photoluminescence (PL) emission due to its stronger hydration ability, and these SWs could act as another type of active site to active the O–H bond for water surface electron transfer and concomitantly provide an alternative channel for surface electron transfer due to the interfacial delocalization of SWs (Scheme c). As we reported earlier,[36−39] the PL emission of SWs was in fact independent of the type of adsorbed metal, herein the alkali metal cations, and under 370 nm UV irradiation all the AMOH solutions with DMSO as the solvent exhibited a very broad emission peak centered at ca. 450 nm with a shoulder at ca. 420 nm (Figure a, solid lines). This peak was accompanied by two excitation bands centered at ca. 295 nm and ca. 370 nm on the excitation spectrum (Figure a, dash lines), so-called PBISs, which are featured bands of SWs confined in the nanospaces or at the nanoscale interfaces.[36−39] Very interestingly, as the concentration of KOH increased, the intensity of the PL emission at ca. 420 and 450 nm first increased and then decreased (Figures b and S10), which completely followed the trend of the HER activity with the increase of the concentration of KOH (Figure a and b). This suggests the key role of SWs in tuning the kinetics of HER. However, differing from other AM+, the close inspection of the excitation–emission two-dimensional plots of the LiOH solution in DMSO solvent showed that as the LiOH concentration increased, the PL intensity ca. 420–440 nm increased significantly (Figure ). Most interestingly, the growth of a new excitation band centered at ca. 295 nm also corresponded to the PL at ca. 420 nm, indicating the formation of a new species for PL emission that had a higher energy level than those of of relatively free SWs in AMOH solutions, such as Na+, K+, Rb+, and Cs+ (Figures and S11). It is important to note that although free SWs could emit strong fluorescence, they did not contribute to the HER due to their intrinsic instability. Considering the superhydration capacity of Li+, which has the smallest ionic radius in the group of alkali-metal cations, we assigned the newly formed species with the excitation band at ca. 295 nm to Li+-bound SWs, {OHad·H2O@Li+}, where Li+ was an anchoring site for the SWs. Li+-bound SWs with high stabilities could act as catalytic active sites for the HER, but the catalytic activity was lower than that of Co-bound SWs, {OHad·H2O@Coδ+}. Furthermore, the Li+ bound SWs were more stable than free SWs, which was also evidenced by the lifetime measurement of the time-resolved luminescence decay profiles of the KOH (Figure c, 1.28 ns) and LiOH solutions (Figure d, 1.59 ns). Thus, the observation of the optical spectrum revealed that Co/NC-800 catalyst in KOH solution did in fact have the best catalytic performance for the HER, which was the result of a compromise between the electronic effect of the Co active centers and overcrowding effect of alkali-metal ions adsorbed on the electrode surface.

Figure 5

(a) Fluorescence spectra of different AMOH (AM+ = Li+, Na+, K+, Rb+, or Cs+) solutions with DMSO as the solvent and (b) those recorded with various concentrations of KOH. Time-resolved luminescence decay profiles of (c) KOH (c) and (d) LiOH solutions.

Figure 6

Two-dimensional excitation (EX)–emission (EM) plots of the solutions with various LiOH concentrations in the DMSO solvent showing the growth of the new excitation band centered at ca. 295 nm with increasing concentrations of LiOH.

Origin of HER Kinetics Dependent on the Binding Strength of the OH– Group to Transition Metals

In fact, the catalytic activity of the {OHad·H2O@M+} active sites is dependent on strength of the binding between the SWs and the metal cations at the electrode. The too strong or too weak binding of the OH– group to the metal is not conducive to the construction of the interface p band intermediate state (Scheme c–e), i.e., it is not conducive to the overlap of p orbitals of space oxygen atoms via the space interaction, which consequently slows the reaction kinetic of the HER due to the Sabatier principle.[81] This also explains why Co/NC-oPD catalyst showed the best catalytic activity for the HER in the group of first-row transition metals (Mn, Co, Ni), owing to the medium binding strength of OH– to the Co ions (Figure S12). These results thus shed new light on the rational design of electrocatalysts, as the HER (and the oxygen evolution reaction (OER)) activity can be attributed to differences in the stabilization of p band intermediate states, which are tailored by the nature of the cations in the electrolyte or the doping of multiple transition metals. These phenomena underscore the indispensable role of OHad in triggering the catalytic role of AM+, thus favoring the notion of the formation of {OHad·H2O@M+} adducts within the double-layer region, where M+ can be transition-metal or alkali-metal cations, demonstrating that transition metals are not necessary for the HER or the OER.[15,19,20,82,83]

SW-Dominated PBIS Model for Elucidating the Microkinetics of the HER

Based on the above electrochemical kinetic and optical spectrum analysis, a new SW-dominated PBIS model was proposed to explain the reaction microkinetics of the nanostructured Co/NC electrocatalysts for the HER. According to the PBIS theory (Scheme c–e), in an alkaline electrolyte, an ensemble of surface intermediate states could form through the space overlap of the p orbitals of O atoms of SWs at the electrode surface. SWs mainly play two roles to tune the reaction kinetics: on the one hand, they act as bridging ligands to accelerate interfacial electron transfer through surface delocalization via the inner sphere electron transfer mechanism of Marcus theory; on the other hand, they act concomitantly as catalysts (or catalytic active sites) to activate the O–H bond of water via the p orbital interaction of O atoms between the water reactant and SWs, which promotes the breakage (or stretching vibration at transition state of water activation by frequency resonance) of the O–H bond by lowering the reaction barrier for water dissociation (the partial p orbital of the O–H bond composed of sp hybridization is used to construct the PBIS with SWs, thus the bond strength of the O–H single bond is dramatically weakened due to the delicate change the microenvironment surrounding the catalytic active site). In the latter case, the coupling between the p orbitals of two O atoms in SWs may be weakened because the p orbital of the O atom of substrate water participates in the reconstruction of the surface intermediate state (PBIS). Thus, in the electrocatalysis HER system, there are three types of water molecules, free moving bulky water as reaction media, strongly adsorbed SWs on transition metals or alkali metal cation (Scheme ), and water reactant stabilized by SW (not shown in Scheme ), which could be as a transition state for water activation in HER wherein the bond strength of O–H of water reactant is significantly decreased.

More details on the concerted electron and proton transfer process can be described as follows. At high [OH–], large amounts of OH– are available in the electrolyte, and OH– can easily form the p band, strongly promoting the water dissociation process to form the active *H species, which was confirmed by 1H NMR (Figure S14). With more electrons being transferred on the surface, *H species combine with each other to form H2 gas. However, at a low concentration of OH– groups, the stabilized p band intermediate states cannot form. In this case, the Volmer step (H2O + e– → H* + OH–) becomes the RDS of the overall reaction, as indicated by the fact that the Tafel slope is greater than 120 mV/dec (Figure b). Under such a low [OH–] environment, higher overpotential is needed for the catalyst to interact with water to start water dissociation. In turn, this also explains why the Co/NC catalyst with a larger amount of *OH is the most active (Table S2). In addition, the cation-dependent activity can be explained by the formation of {OHad·H2O@AM+} adducts within the double-layer region (Scheme ). The nature of {OHad·H2O} adducts near the interface can significantly impact the activity of the HER, thus altering the kinetics of the HER by changing the unique interaction of SWs at the electrified interface. The hydration of the alkaline metal ion decreases in the order Li+ > Na+ > K+ > Rb+ > Cs+, and the interaction between AM+ and water molecules within the hydration shell also decreases from Li+ to Cs+. The strongest interaction between Li+ cations and their hydration water molecules can be demonstrated by the small Li+–O distance of ∼2 Å between the Li+ ion and water in its hydration shell, compared to ∼2.4 Å for Na+, ∼2.8 Å for K+, ∼3.0 Å for Rb+, and ∼3.1 Å for Cs+.[84] Therefore, interfacial OH– ions in the case of KOH form moderate interactions with surrounding water molecules compared to those in the case of LiOH, which form stronger interactions with water molecules at the interface, and those in the case of RbOH and CsOH, which form weaker interactions with water molecules at the interface (Scheme ). Thus, it is easy to understand that the activity trend in the order K+ > Na+ ∼ Li+ > Rb+ > Cs+ as a comprise between the electronic effect of the {OHad·H2O@M+} catalytic active sites and overcrowding effect of AM+. Thus, our work demonstrates how the kinetics of the water reduction reaction can be regulated by the SW-dominated PBIS model at the electrified interface (Scheme ). The unique interaction of oxygen atoms of {OHad·H2O} adducts at the interface can play a significant role in activating water molecules at the electrified interface, resulting in interesting pH- and cation-dependent effects on the HER kinetics. Such an understanding, which goes beyond tuning the binding energetics of the active site by chemical covalent bonding (Scheme a and b), is critical to the design of more active electrochemical interfaces.

Synergetic Surface Electron Transfer and Activation of the O–H Bonds of Water Promoted by SWs

Generally, the slow and sluggish kinetics of Pt NP catalysts under alkaline conditions are attributed to the high energy barrier for cleavage of the O–H bond of water as a rate-determining step (RDS) to form hydronium ions. However, more and more experimental evidence showed that, even in an induced high-pH environment, the local hydronium ions could be generated easily[10]; however, the direct capture of hydronium ions under alkaline conditions was never evidenced before. What is the difference for the proton, and how does it form? Our 1H NMR data (Figure S14) first provide direct evidence of the presence of local hydronium ions at electrode surface in alkaline conditions, and we find that their chemical shift is just between that of a proton (H+ in water, 4.5 ppm) and that of hydride (H–, −0.25 ppm) when they are trapped in the electron sea of the surface intermediate state (Scheme c). Using this new model, the low activity of Pt-based catalysts for the HER could be easily understood. For example, our reference Pt/C catalyst has an extremely low activity and long tails in high-pH conditions (Figure a), since the strong and saturated binding of the hydroxide group on Pt NPs prohibits the formation of SWs (or PBIS), thus blocking the surface electron transfer and prohibiting the activation of the O–H bond of water. In Scheme c and d, the strength of O–M (HOM) is a key parameter that regulates the formation of the surface p band transient state (HOO). This not only explains the volcano-type trend in catalytic activity for the HER depending on the binding strength of transition metals with hydroxide groups (Mn, Co, and Ni), as shown in Figure S12, but also explains why the Pt NP-based HER catalysts show particle-size- and oriented-crystal-facet-dependent catalytic activity, since the different particle sizes and crystal facets of metal NPs exhibit the different binding strengths and densities of hydroxide groups on metals.[29] The abnormally low activity of our reference Pt/C catalyst further proves the rationality of our PBIS model with the concept of SWs. In fact, very similar experimental phenomena have been reported in the literature, but the chemical reason was never explained.

It is interesting to note that even though the HER activity shows a strong pH- and alkali-metal-cation-dependent effect (Figures and , respectively), the change of the Tafel slopes is not as large as the previously reported value (less than 50 mV/dec) despite a clearly distinguished decreasing trend from 170 to 85 mV/dec with the increase of the KOH concentration from 0.01 to 3.0 M (Figure b). The same trend was found in different alkali metal electrolytes (Figure d). Based on the trend of pH- and alkali-metal-cation-dependent HER activity, our conclusion is that the conventional understanding of the microkinetics of the HER based on a Tafel analysis with a critical value of 120 mV/dec, where the cleavage of the O–H bond is the RDS when the Tafel slope is larger than 120 mV/dec and the proton recombination (Volmer–Tafel)step or the electrochemical desorption (Volmer–Heyrovsky) step is the RDS when the Tafel slope is less than 120 mV/dec, may not be reasonable. Indeed, these results suggested that these identified RDSs are isolated chemical events,[10] while our SW-dominated PBIS model provides solid evidence that interfacial electron transfer, the activation of water (the cleavage of the O–H bond), and the *H combination with proton are synergetic chemical process (or concerted electron and proton transfer process) in high alkaline conditions. Thus, the key to the HER with a high activity is to construct surface states to activate the O–H bonds of water and concomitantly promote concerted electron and proton transfer.[37,39,42,47,85] Using our model, the micro-kinetics of OER (the strong interaction between O atom of water reactant and SW, suggesting the consumption of surface lattice oxygen at the end of reaction), ORR (the weakening of O–O bond of O2 by SW), CO2RR (the breakage of C–O bonds of CO2 by SW), and NRR (the weakening of N–N bond of N2 by SW) can be easily elucidated.

Conclusion

Thus far, there is no unified model to explain the microkinetics of electrocatalytic hydrogen evolution, such as the pH-, alkaline-metal-ion-, and transition-metal-dependent reaction kinetics. Using ZIF-67-derived nanostructured carbon as a prototype electrocatalyst, we first provide solid evidence and confirmation that the interfacial SWs in the form of hydrous hydroxyl complexes absorbed on metals {OHad·H2O@M+} (M+ is a transition- or alkaline-metal cations) are active sites for water activation (or dissociation) and subsequent proton reduction. The space interaction of two O atoms of SWs could be a new type of weak interaction, with a strength between a hydrogen bond and a chemical covalent bond. Furthermore, due to the spatial overlap of the p orbitals of two O atoms in SWs, an ensemble of surface transient states is formed (Scheme c–e) that synergistically promotes the activation of water and interfacial electron and proton transfer at the nanoscale interface. It should be emphasized that these dynamic surface intermediate states (PBIS) are not stable, which explains the extremely sensitivity of the HER (even the OER and the oxygen reduction reaction) to microenvironments, such as pH- and cation-dependent effects. The concept of SW-dominated surface transient states that tune the microkinetics of the HER enables the optimization of the entire electrochemical interface to improve the activity as opposed to the optimization of only the catalyst structure, which is critical to the design of more active electrochemical interfaces for energy storage and conversion reactions involving small molecules (CO2, CO, N2, O2, and H2).

In addition, the physical basis of SW-dominated PBIS model is the dynamic redistribution of atomic orbitals at the metal–nanoscale interface (Scheme c and 1d) that concomitantly accompanies the nuclear quantum effect (atomic motion), which goes beyond the existing theoretical basis of density functional theory calculations (i.e., the Born–Oppenheimer approximation). The developed PBIS needs a more advanced theoretical calculation method to further prove the rationality of our model. Here we emphasize again that the concept of SWs is completely different from that of H-bonded water, as SWs, two adjacent water molecules, are mainly combined through the spatial overlap of the p orbitals of two O atoms to form a local chemical bond with π-bond characteristics, which provides an alternative channel for electron transfer by surface delocalization due to the spatial orbital overlap. The concept of SWs as alternative channels for concerted electron and proton transfer that concomitantly act active sites for the activation of the O–H bond of water may not only provide important guidance for the design and selection of catalysts or electrolytes for nanomaterial-catalyzed reactions in an aqueous environment, including CO2 or CO reduction[86−91] nitrogen reduction,[17,92,93] and other electrocatalytic reduction reactions,[94] but may also shed new light on nanoscale-range electron and proton transfer through the waterline “bridge” in the biological macromolecule system,[95−97] which could follow the inner sphere electron transfer model of Marcus theory with connected SWs as a ligand bridges (herein, the number of O atoms in “water line” should not be limited to two).[98−100]