A highly active nickel electrocatalyst shows excellent selectivity for CO2 reduction in acidic media

The development of selective electrocatalysts for CO2 reduction in water offers a sustainable route to carbon based fuels and feedstocks.


1.1
General Milli-Q water (18.2 MΩ) was used throughout. NaClO 4 , NaOH, HClO 4 , DClO 4 and D 2 O were used as received (Aldrich). Complexes 1 and 2 were synthesised as previously described. S1, 2 CO 2 , Ar and N 2 were purchased from BOC at CP grade or higher. Gas chromatography was performed using an Agilent 6890N employing N6 helium as the carrier gas (5 ml.min -1 ). A 5 Å molecular sieve column (ValcoPLOT, 30 m length, 0.53 mm ID) and a pulsed discharge detector (D-3-I-HP, Valco Vici) were employed. CO and H 2 peak areas were quantified with multiple calibrant gas injections and were re-calibrated daily. For FTIR experiments (Bruker Vertex) a CaF 2 IR cell (0.1 mm) was used.

1.2
Electrochemistry CVs were measured using a Palmsens 3 and a pear-shaped flask with a HMDE (geometric surface area = 0.023 cm 2 ), Pt coil counter electrode and Ag/AgCl (3 M NaCl) reference electrode (BASi). Experiments were purged with argon, nitrogen or CO 2 for 30 minutes prior to use. The pH was varied by adding HClO 4 (1 M) or NaOH (1 M) while keeping the purged solution under a blanket of the relevant gas to ensure a common value between Ar/N 2 and CO 2 experiments. Controlled potential electrolysis (CPE) used a custom mercury pool cell (4.15 cm 2 ) without stirring or gas bubbling. Prolonged CPE (7.5 hrs) was carried out under a constant CO 2 purge for the first 7 hours, with the faradaic efficiency being determined in the final 30 minutes when the cell is sealed allowing for build-up of products. The counter electrode was separated in a second compartment by a glass frit to minimise re-oxidation of products. Chronocoulometry experiments used a Whistonbrook DF63 potentiostat and Labview software (National Instruments), HMDE working electrode, platinum wire counter and a silver wire quasi-reference electrode. The cell was purged with argon (30 mins) before the experiments and kept under a blanket of argon throughout. A fresh mercury drop was used with every potential step. RDE measurements were performed using a Pine AFMSRCE and Bio-Logic SP-200 potentiostat. The gold electrode (Pine) was dipped in mercury for 1-2 mins, then excess mercury was removed by polishing and left to dry for 2 hours. Amalgam electrodes were found to be stable up to rotation rates of 900 rpm. In the main text we refer to the potential necessary for catalysis, which is defined as the potential when the current density equals 2 mA cm -2 . This current density is chosen as it is sufficiently higher than background currents (including currents arising from the initial adsorption of the catalyst).

Fig. S3: Experimental Pourbaix diagram for the Ni II/I couple of 2 (red) and 3 (black)
The potential of the Ni II/I couple is found to be largely pH independent for both 2 and 1. In Fig. 2b of the main text we note a correlation between the concentration of the protonated form of 2 and the measured catalytic current density under CO 2 . We therefore propose that the change in catalytic activity for 2 with pH is due to the presence of the local acid functionality.  The new feature at ca. -1 V NHE in Fig S5 is assigned to the presence of the protonated form of the carboxylic acid opening a different mechanism pathway, this peak in the CV corresponds to the feature seen to grow in the RDE in the main text at similar potentials. At pH 2 in the CV the peak at ca. -1 V NHE is heavily overlapped with a 2nd reduction at -1.2 V however it can still be identified by taking a second order derivative of the CV ( Fig. S5(b)).  The slight deviation from linearity in the plots is due to CO build up, which is known to deactivate [Ni(cyclam)] 2+ catalysts. Experiments carried out under a constant CO 2 purge at pH 2 using 2 over a 7.5 hour period show reasonable stability with FE for CO = 52 % and H 2 = 31 % being obtained in the final 0.5 of an hour.

Current vs. pH plots
A comparison of the current density recorded at -0.99 V NHE under CO 2 during the RDE measurement at 800 rpm for 1 and 2 with pH is presented in the main text (figure 2b). This potential was chosen as both catalysts are active towards CO 2 and the level of H 2 evolution is also relatively low. A similar trend in current density due to CO 2 reduction with pH is also observed at more negative potentials, Fig S10. Here we have plotted the difference in current density under CO 2 and under N 2 (conditions when only H 2 evolution can occur), to provide an estimate of the current density due to CO 2 reduction only. Experiments at pH 2 were not possible due to the restricted solvent window of the glassy carbon electrode.

Double-potential-step chronocoulometry
To evaluate the surface adsorption of 2 to the electrode surface as a function of the potential, double-potential-step chronocoulometry was carried out on 0.1 mM solutions of 2. The potential was kept at the value of interest for 30 seconds to ensure that the equilibrium quantity of [Ni I (cyclam-CO 2 H)] + was adsorbed, then stepped to 0 V for 100 ms and finally stepped back to the initial value. The charge passed during the first 10 ms of the first step was plotted vs. t 1/2 , while the charge passed during the first 10 ms of the second step was plotted vs. [τ 1/2 + (t-τ) 1/2 -t 1/2 ], where τ is the duration of the first step, and t is the time. The difference between the forward and reverse intercepts give an estimate of the surface coverage, according to the equation Q = nFAΓ, where Q is the charge, n the number of electrons, F the Faraday constant, A the area of the electrode and Γ the surface coverage in mol cm -2 . However as previously noted by Anson S3 whilst the reductive desorption is very rapid the time taken for [Ni(cyclam)] + (and its derivatives) to adsorb is significantly greater than would be expected solely for a diffusion controlled process hence the use of a long period to achieve equilibrium here (likely required due to a structural isomerisation). This is likely to be a greater cause of error in the application of the measured surface coverages to voltammetric sweep data than the estimated errors obtained from the fit to the intercept of charge vs. [τ 1/2 + (t-τ) 1/2 -t 1/2 ] which were typically on the order of 10%.

Rotating Disk Electrode
Analysis of RDE data allows for calculated k obs in the absence of diffusion control. Rotation rates of up to 900 rpm were achieved with no noticeable disruption of the amalgam coating.
The limiting currents are calculated from the intercept of the Koutecký-Levich plots, constructed using three independent measurements, a representative example of which is provided in Fig. S15. From these currents we calculate k obs using Eq 1. in the main text in the manner recently described elsewhere. S7,8 Here k obs describes the overall rate law for the system as the complete mechanism remains unproven. We note that the mechanism of CO 2 reduction by 1 and its derivatives has been proposed to involve initial reduction of Ni II followed by reaction with CO 2 , with a 2nd electron transfer and chemical step following. S4-6 We also propose that with 2 reduction to Ni I occurs prior to CO 2 binding as UV/Vis studies show no interaction between the Ni II and CO 2 , such a process would lead to change of coordination environment from square planar for 2, which would lead to a clear change in the UV/Vis spectrum S2 (Fig. S13)  V NHE k obs CO 2 (s -1 ) x10 2 Error ± (s -1 ) x10 2