Cooperative Effect of Cations and Catalyst Structure in Tuning Alkaline Hydrogen Evolution on Pt Electrodes

The kinetics of hydrogen evolution reaction (HER) in alkaline media, a reaction central to alkaline water electrolyzers, is not accurately captured by traditional adsorption-based activity descriptors. As a result, the exact mechanism and the main driving force for the water reduction or HER rate remain hotly debated. Here, we perform extensive kinetic measurements on the pH- and cation-dependent HER rate on Pt single-crystal electrodes in alkaline conditions. We find that cations interacting with Pt step sites control the HER activity, while they interact only weakly with Pt(111) and Pt(100) terraces and, therefore, cations do not affect HER kinetics on terrace sites. This is reflected by divergent activity trends as a function of pH as well as cation concentration on stepped Pt surfaces vs Pt surfaces that do not feature steps, such as Pt(111). We show that HER activity can be optimized by rationally tuning these step–cation interactions via selective adatom deposition at the steps and by choosing an optimal electrolyte composition. Our work shows that the catalyst and the electrolyte must be tailored in conjunction to achieve the highest possible HER activity.


Single Crystal Preparation
Prior to each experiment, the surface of a given Pt single crystal was first prepared by flame annealing (30 sec for a disk and 10 sec for a bead) and subsequent cooling (ca.120 sec) in 1:2 Ar/CO atmosphere.Then the crystal was transferred to the working cell with a protective water droplet.CO atmosphere was used for cooling the Pt single crystals for two reasons: (i) strongly adsorbing gas like CO forms a protective adlayer at the surface which prevents any surface contamination or unwanted surface roughening due to ambient oxygen (ii) it has been shown previously that Pt single crystals prepared in CO atmosphere generally give well-ordered and unreconstructed surfaces.(1-3)More specifically, it has been shown both by Attard et al. and by Kolb et al. that CO atmosphere leads to the formation of unreconstructed Pt(110)-(1x1) surface.(1,3) Moreover, Kolb et al. have shown both with in-situ STM measurements as well as with electrochemical measurements that only the Pt(100) surface prepared under CO atmosphere forms a flat, unreconstructed surface with large terraces free of islands or holes.In all other atmospheres, Pt(100) forms the hexagonally reconstructed structure with islands and defects.(1) Prior to each measurement, the CO adlayer was electro-oxidatively desorbed from the Pt single crystal and thereafter the blank CVs and HER CVs were recorded.We note that based on the integrated charge of the CO stripping peak (Fig. S2), the total amount of CO 2 formed is quite small (ca. 10 -12 -10 -13 gCO 2 ) and hence, we do not expect any concomitant carbonate contamination to affect our activity measurements, especially under HER relevant conditions.For calculating the current densities, the geometric surface area of the respective single crystal was used.

Polycrystalline Pt Preparation
Before each experiment, Pt polycrystalline disk (diameter = 5mm, Pine instruments) was mechanically polished on Buehler micro-polishing cloth (8 inches) with decreasing sizes of diamond polishing suspension, namely, 3µm, 1 µm and 0.25µm.Next, the disk was sonicated in ultrapure water and acetone for ca. 10 min.to remove any inorganic/organic impurities and mounted on the RDE tip (E6/E5 ChangeDisk tips in a PEEK shroud; Pine Research).Thereafter, the surface was electrochemically polished in 0.1 M H 2 SO 4 electrolyte by cycling between 1.8 V RHE to -0.5 V RHE for 200 cycles at a scan rate of 1 Vs -1 .The representative CV of polycrystalline Pt was recorded by cycling the electrode in the same electrolyte between 0.05 V RHE to 1.2 V RHE at a scan rate of 50 mVs -1 .In order to calculate the current densities, the electrochemically active surface area (ECSA) of the electrode was determined by calculating the charge under the hydrogen desorption peak and dividing it by the specific charge associated with the oxidation of a monolayer of adsorbed hydrogen on Pt (Q H ox,ML,Pt = 230 Ccm -2 ).( 4)

Electrochemical Measurements
The electrochemical measurements in alkaline media were carried out in a home-made PTFE cell while the measurements in acidic media were conducted in a homemade borosilicate glass cell.The reference electrode was separated from the working compartment with the help of a Luggin capillary and the counter electrode was a Pt wire (99.99% purity), unless otherwise stated.with Ar to remove any dissolved oxygen form the electrolyte.Moreover, during the measurement, Ar was bubbled over the headspace of the electrochemical cell, in order to eliminate any interference from the ambient oxygen.A homemade reversible hydrogen electrode (RHE) was used as the reference electrode in all the experiments.All the electrochemical measurements were carried out using a Biologic (VSP-300) potentiostat.For all the measurements done with Pt single crystals a hanging meniscus configuration was used and for the measurements done with polycrystalline Pt, an MSR rotator (Pine Research) was used which was rotated at 2500 rpm.For all the CVs taken, 85% Ohmic drop compensation was performed and for all the steady-state potentiostatic measurements 100% Ohmic drop compensation was applied.Before measuring the HER activity, the blank CV of all the Pt surfaces were recorded (scan rate: 10 mV s -1 ) in the working cell to check the quality of surface preparation.All the studies for the pH dependence of HER were done in Ar sat.0.1 M electrolytes (xM NaOH + yM NaClO 4 such that x+y = 0.1).All the studies for the bulk cation concentration dependence of HER were done in Ar sat.electrolytes either at pH 11 (10 -3 M NaOH/CsOH) or at pH 13 (0.1 M NaOH/CsOH) where the cation concentration was varied by adjusting the concentration of NaClO 4 or CsClO 4 in the electrolyte, such that the total concentration of perchlorate anions was similar at different pHs.The CVs for HER activity were taken in the potential window of 0 V RHE to -0.1 V RHE (iR corrected) at a scan rate of 10 mVs -1 .Additionally, to obtain the Tafel data, chronoamperometry was performed in the potential window of the CVs (0 V RHE to -0.1 V RHE ; iR corrected) at 10mV potential steps.
Generally, 20 seconds per potential step were enough to reach the steady state.EIS measurements were performed at pH 11 (10 -3 M NaOH) either with 0 mM NaClO 4 or with 50 mM NaClO 4 containing electrolytes.The electrochemical impedance spectroscopy (EIS) measurements were done in the potential window of 0.1 V RHE to 0 V RHE at 20 mV steps, with frequencies ranging from 30 KHz to 1 Hz and a peak to peak amplitude of 5 mV.Moreover, a 10 F shunt capacitor bridge was added between a secondary counter electrode (Pt wire) and the reference electrode in order to eliminate any artefacts caused by the non-ideal behavior of the potentiostat at high frequencies.(5) The impedance data was fit with an appropriate equivalent electrochemical circuit (EEC; shown in Fig. S14 or Fig. S15) with the help of EIS Zfit (part of Biologic's EC-Lab software).

Adatom Electrodeposition at Pt(553) Steps
Metal adatoms were selectively electrodeposited on Pt(553) steps by following the procedure outlined previously.(6,7) Detailed characterization of these surfaces can be found there, and in Figs.S21-25.In brief, the as-prepared Pt(553) single crystal was transferred to an electrochemical cell containing 0.1 M HClO 4 plus 10 -6 M salt of the respective adatom (i.e.AgClO 4 or RuCl 3 or NH 4 ReO 4 ) and cycled for 10-200 cycles between 0.05 V RHE to 0.35 V RHE .Afterwards, the adatom modified Pt(553) surface was ready for electrochemical measurements (as outlined in the last subsection).To remove the adatoms from the surface, the electrode was subjected to repeated cycles of chemical etching (in conc.HNO 3 ) and flame annealing.
The adatom coverage at the step edge of Pt(553) was calculated by measuring the change in charge under the step-associated peak of Pt(553) in the blank cyclic voltammograms (0.05 V RHE to 0.35 V RHE on Re* Pt(553) and 0.05 V RHE to 0.8V RHE on all the other surfaces; see Fig. S21-23).For Re* Pt(553), potentials more anodic than 0.35 V RHE were avoided to prevent oxidation of the adatoms.For correctly comparing the HER activity of different adatom modified surfaces, the current densities obtained for HER were normalized for the adatom coverage at the surface.(6)

Supplementary Text
Step Density Determination The theoretical step density ( step atom theo. ) of Pt(S)-[n(111) x (110)] surfaces, namely, Pt(111), ᴦ Pt (15,15,14), Pt(554), Pt(553), Pt(331) and Pt(110) can be calculated as step atom theo.= 1/(nᴦ 2/3); (8) where n is the width of the terrace.In theory, the width of Pt(111) terrace is infinite which is reduced to 30 atoms in Pt(151514), 10 atoms in Pt(554), five atoms in Pt(553), 3 atoms in Pt(331) and 0 atoms in Pt(110).However, in practice, Pt(110) shows (1x2) missing row reconstruction which leads to the formation of two atom long (111) terraces and reduces its step density by half.Hence, the real step density of a surface can differ from the theoretical value, either due to the presence of any imperfections at the surface or due to reconstruction (such as in Pt(110)).That is why it is better to use the experimental step density ( step atom exp. ) or the ᴦ surface concentration of step atoms (nmol cm -2 ).The experimental step density of a surface can be estimated from the experimentally obtained charge for hydrogen desorption peak (Q H ox,Pt(S)-[n(111) x (110)] ) as follows: step atom exp.= ; where F is Faraday's constant and A is ᴦ . the geometric area of the electrode.(9) EIS data fitting and interpretation The fitting of impedance data in the H upd region is generally done by using the EEC shown in Fig. S14 (EEC1).(10,11) In this case, only the adsorption of hydrogen is considered at the interface, and its kinetics is captured by R ad and C ad .In acidic media, the kinetics of H adsorption becomes too fast to measure (i.e.R ad ≈ 0) and this circuit reduces to a simple RC circuit where no real distinction can be made between the double layer capacitance (C dl ) and the pseudocapacitance due to hydrogen adsorption (C ad ).However, in alkaline media, the kinetics of H adsorption is quite slow and consequently, double layer charging and H adsorption have two distinct time constants, such that C dl and C ad can be reliably separated from one another.We note that under ideal conditions, i.e. in the absence of any specific adsorption, the double layer capacitance (C dl ) is further represented by two capacitances that are in series with each other, C i , the inner layer capacitance, and C diff , the diffuse layer capacitance , where at any point, C dl is dominated by the smaller of the two capacitances.(12) According to Gouy-Chapman-Stern theory, C i represents the capacitance due to the charged species that are located in the Outer Helmholtz plane and should be independent of the electrolyte.On the other hand, C diff represents the capacitance due to any charged species that are truly diffuse and it varies both with potential and with the identity/composition of the electrolyte.In theory, in dilute enough electrolytes, C diff will dominate the overall double layer capacitance and it will be directly proportional to the square root of electrolyte concentration in the diffuse layer (C diff ∝ ).However, a recent study by our group has shown that even in very dilute . electrolytes C dl differs from the theoretical values expected from Gouy-Chapman theory (C G- C ).( 13) This is due to the presence of attractive ion-surface interactions, which are not accounted for in the original GCS theory.Hence, even though some proportionality is expected between the C dl and the electrolyte concentration in the diffuse layer, it will be more complicated than what is suggested by the simplified formula for C diff .Nevertheless, based on the C dl values we obtain by using EEC1 (Table S1, Fig. S14-S17), we can see that C dl and hence the near surface concentration of cations clearly increases with increasing step density of Pt.However, while the data at high frequencies shows a good fit with EEC1, at low frequencies a good fit is not obtained with EEC1, especially on stepped Pt single crystals (Fig. S14b).We note that because C dl is mainly obtained due to the perturbations at high frequency (due to its small time constant), the poor quality of the fit at low frequencies should not have an impact on the correct estimation of the C dl values.Nevertheless, we will briefly discuss what can be the possible reason for these discrepancies and what can be done to rectify this.One possible reason can be the co-adsorption of other species during the H upd process.It is well-known that the stepassociated peak on Pt arises due to the replacement of H adsorbed with OH adsorbed .(14) Hence, in this potential range, co-adsorption of H and OH has to be considered, and a parallel arm has to be added to EEC1 to account for the kinetics of OH adsorption, as shown in Fig. S15 (EEC2).Moreover, this view has been further refined by our group recently, where we showed that alkali metal cations also co-adsorb along the steps and weaken OH adsorption, thus resulting in the non-Nernstian shift of this step associated peak.In view of this, perhaps an additional arm would be required to truly fit the impedance data around the potential window of the stepassociated peak.We note that in our EIS measurements we stay at potentials more negative than the step-associated peak.Nevertheless, it has been shown with DFT calculations that in alkaline electrolytes (pH 14), cation co-adsorption is favorable on Pt single crystals, at potentials as negative as 0 V RHE .(15,16) Hence, the fact that our EIS data fits better with EEC2 (Fig. S15b) could suggest that there is in-fact cation specific adsorption taking place at the surface which competes with H adsorption.This can have interesting implications both with regards to the true surface composition of Pt during HER and with regards to exact nature in which the cations interact with dissociating water molecules at the interface.However, a detailed analysis of this requires a separate study unto itself and is therefore a topic of future work.Importantly, the value of C dl obtained from the fits, remains largely unchanged regardless of the circuit employed (Table S1).This is understandable since the different circuits only account for the changes in the low frequency window of the measurements i.e. changes which are related to the adsorption phenomenon at the interface, while C dl , as previously mentioned, presents itself in at higher frequencies.Regardless, in our paper we use EEC2 to plot the C dl values as this circuit gives better overall fits for the EIS data.533) and Pt polycrystalline as recorded in pH 11, pH 12 and pH 13 electrolytes at a scan rate of 10 mVs -1 .For the preparation of single crystals, the procedure outlined in Materials and Methods subsection 2 was followed and for the preparation of polycrystalline Pt the procedure outlined in Materials and Methods subsection 3 was followed.

Fig. S1 .
Fig. S1.Blank CVs ofPt(111), Pt(100), Pt(110), Pt(553), Pt(533) and Pt polycrystalline as recorded in pH 11, pH 12 and pH 13 electrolytes at a scan rate of 10 mVs -1 .For the preparation of single crystals, the procedure outlined in Materials and Methods subsection 2 was followed and for the preparation of polycrystalline Pt the procedure outlined in Materials and Methods subsection 3 was followed.

Fig. S5 .
Fig. S5.Tafel data for HER (potential vs log of current density) as obtained from chronoamperometry measurements at different Pt surfaces in 0.1 M NaOH (pH 13), 0.01 M NaOH +0.09 M NaClO4 (pH 12) and 0.001 M NaOH + 0.099 M NaClO4 (pH 11) recorded a t 20 mV potential steps.All the measurements on Pt single crystals were done in stationary hanging meniscus configuration and the measurement at the polycrystalline Pt rotating disk electrode was done at 2500 rpm.The Tafel slopes (mVdec -1 ) at each pH are indicated next to the plots.Note that the Tafel slope is not constant, though at the most negative potentials is consistently close to 120 mV/dec.
Fig. S7.(upper row) Blank CV (left) and HER activity measurement (right) on Pt(110) at pH 11 at different concentrations of NaClO 4 (increasing concentration of NaClO 4 in going from light to dark data points).(middle row) Blank CV (left) and HER activity measurement (right) on Pt(110) at pH 13 at different concentrations of NaClO 4 (increasing concentration of NaClO4 in going from light to dark data points).(lower row) Effect of Cs + vs. Na + on Pt(110) in pH=11 NaOH + 5 mM of either NaClO 4 or CsClO 4 .

Fig
Fig. S8.(A) Blank CV (left) and HER activity measurement (right) on Pt(111) at pH 11 at different concentrations of NaClO 4 (increasing concentration of NaClO 4 in going from light to dark data points).(B) Blank CV (left) and HER activity measurement (right) on Pt(111) at pH 13 at different concentrations of NaClO 4 (increasing concentration of NaClO 4 in going from light to dark data points).

Fig. S9 .
Fig. S9.Tafel data for HER (potential vs log of current density) as obtained from chronoamperometry measurements at Pt(111) (left) and Pt(110) (right) at different concentrations of NaClO 4 (increasing concentration of NaClO 4 in going from light to dark data points) in (A) pH 11 and (B) pH 13 electrolytes.

Fig
Fig. S10.(A) Blank CV (left) and HER activity measurement (right) on Pt(111) at pH 11 at different concentrations of CsClO 4 (increasing concentration of CsClO 4 in going from light to dark data points).(B) Blank CV (left) and HER activity measurement (right) on Pt(111) at pH 13 at different concentrations of CsClO 4 (increasing concentration of CsClO 4 in going from light to dark data points).

Fig. S11 .
Fig. S11.(A) Blank CV (left) and HER activity measurement (right) on Pt(110) at pH 11 at different concentrations of CsClO 4 (increasing concentration of CsClO 4 in going from light to dark data points).(B) Blank CV (left) and HER activity measurement (right) on Pt(110) at pH 13 at different concentrations of CsClO 4 (increasing concentration of CsClO 4 in going from light to dark data points).

Fig
Fig. S14.(A) EEC1 used to fit the data in the H upd region; it features R Ohmic which represents the solution resistance, C dl which represents the double layer capacitance and R ad , C ad which together represent the kinetics of H adsorption at the surface.(10)(B) Nyquist admittance plots with the corresponding fits obtained with EEC1 at 0.04 V RHE at different Pt single crystals in 0.001 M NaOH (pH 11).

Fig. S15 .
Fig. S15.(A) EEC2 which is a modification of EEC1and features an additional branch that takes into account any coadsorption phenomenon in the H upd region;(9) it features R Ohmic which represents the solution resistance, C dl which represents the double layer capacitance, R ad1 , C ad1 which together represent the kinetics of H adsorption at the surface and R ad2 , C ad2 which together represent the kinetics of co-adsorption of a secondary species at the surface.(10)(B)Nyquist admittance plots with the corresponding fits obtained with EEC2 at 0.04 V RHE at different Pt single crystals in 0.001 M NaOH (pH 11).

Fig. S16 .
Fig. S16.Specific double layer capacitance (C dl ; µFcm -2 ) as derived by using the equivalent electric circuit shown in Supplementary Fig. S15 on different Pt single crystals at different potentials (vs RHE) in 0.001 M NaOH (pH 11) as a function of experimentally derived step density ( step atom exp.; nmol cm -2 ).The experimental step density ( step atom exp. ) was calculated by ᴦ ᴦ integrating the charge for the hydrogen adsorption peak as obtained from the bank cyclic voltammograms shown in Fig. S11.

Fig. S17 .
Fig. S17.Specific double layer capacitance (C dl ; µFcm -2 ) as derived by using the equivalent electric circuit shown in Supplementary Fig. S15 on different Pt single crystals at different potentials (vs RHE) in 0.001 M NaOH (pH 11) as a function of theoretical step density ( step atom theo. ) which was calculated by using the procedure discussed in Supplementary Text ᴦ subsection 1.

Fig. S18 .
Fig. S18.double layer capacitance (C dl ; µFcm -2 ) as derived by using the equivalent electric circuit shown in Supplementary Fig. S15 on different Pt single crystals at different potentials (vs RHE) in 0.001 M NaOH (pH 11) plus 50 mM NaClO 4 as a function of experimentally derived step density ( step atom exp.; nmolcm -2 ).The experimental step density ( step atom ᴦ ᴦ exp. ) was calculated by integrating the charge for the hydrogen adsorption peak as obtained from the bank cyclic voltammograms shown in Fig. S11.
The components of the electrochemical cell were cleaned prior to each experiment by boiling them five times in ultrapure water.When not in use, they were stored in 1 g/L solution of KMnO 4 .Before boiling, any traces of KMnO 4 and MnO 2 were removed by submerging them in a diluted solution of acidified H 2 O 2 (few drops of conc.H 2 SO 4 and 10-15 mL H 2 O 2 in excess water) for half an hour.Before every experiment, the electrolytes were purged for ca.30 min.

Table S1 .
The values of specific double layer capacitance (C dl ; µFcm -2 ) as derived from EEC1 (shown in Fig.S14) and EEC2 (as shown in Fig.S15) at 0.04 V RHE on different Pt single crystals in 0.001 M NaOH (pH 11).