Abdullah M.
Abudayyeh‡
ac,
Michael S.
Bennington
ac,
Johan
Hamonnet
bc,
Aaron T.
Marshall
*bc and
Sally
Brooker
*ac
aDepartment of Chemistry, University of Otago, Dunedin, 9016, New Zealand. E-mail: sbrooker@chemistry.otago.ac.nz
bChemical and Process Engineering, University of Canterbury, Christchurch, 8041, New Zealand
cThe MacDiarmid Institute for Advanced Materials and Nanotechnology, New Zealand
First published on 11th March 2024
In aqueous pH 7 phosphate buffer, during controlled potential electrolysis (CPE) at −1.10 V vs. Ag|AgCl the literature square planar copper complex, [CuIILEt]BF4 (1), forms a heterogeneous deposit on the glassy carbon working electrode (GCWE) that is a stable and effective hydrogen evolution reaction (HER) electrocatalyst. Specifically, CPE for 20 hours using a small GCWE (A = 0.071 cm2) gave a turnover number (TON) of 364, with ongoing activity. During CPE the brownish-yellow colour of the working solution fades, and a deposit is observed on the small GCWE. Repeating this CPE experiment in a larger cell with a larger GCWE (A = 2.7 cm2), connected to a gas chromatograph, resulted in a TON of 2628 after 2.6 days, with FE = 93%, and with activity ongoing. After this CPE, the working solution had faded to nearly colourless, and visual inspection of the large GCWE showed a material had deposited on the surface. In a ‘rinse and repeat test’, this heterogeneous deposit was used for further CPE, in a freshly prepared working solution minus fresh catalyst, which resulted in similar ongoing HER activity to before, consistent with the surface deposited material being the active HER catalyst. EDS, PXRD and SEM analysis of this deposit shows that copper and oxygen are the main components present, most likely comprising copper and copper(I) oxide ((Cu2O)n) formed from 1. The use of 1 leads to a deposit that is more catalytically active than that formed when starting with a simple copper salt (control), likely due to it forming a more robustly attached deposit, which also enables the observed long-lived catalytic activity.
Hydrogen is an existing key industrial chemical (currently used to make NH3, CH3OH and direct reduced iron, as well as in refining fossil fuels) that is currently mostly made from fossil fuels which form so-called ‘grey’ or ‘brown’ hydrogen due to the associated carbon emissions.3 These processes need to be replaced by the production of ‘green’ hydrogen from water using renewable electricity, with no associated carbon emissions. ‘Blue’ hydrogen, formed from existing grey/brown processes but with carbon capture, is also being pursued, but green hydrogen is the ‘gold standard’, and is the focus herein. Indeed, the International Energy Agency (IEA)3 predicts that in the future, in addition to the growing use of green (and blue) hydrogen in industry, it will also have a growing role as an energy vector, as hydrogen has very high energy density by mass, albeit not by volume (so it is typically stored either at 200–700 bar or cryogenically as a liquid). Moreover, green hydrogen is a carbon-zero fuel, emitting only water when burnt, either by combustion to give heat or in a fuel cell to give electricity. Taken together, industrial and energy use, the IEA predicts that if we are to reach net-zero then globally hydrogen production will likely increase about five-fold by 2050, and by then will comprise mostly green and blue hydrogen.3
Green hydrogen is yet to become cost competitive with fossil fuel derived hydrogen,3 but this is likely imminent as global production is rapidly being scaled up and the costs of carbon-emissions are rising to encourage climate change mitigation measures. To assist in reaching that tipping point, there is ongoing interest in improving the efficiency of water splitting by renewable energy, with improved catalysts for the hydrogen evolution reaction (HER) being one area of focus.
Over the last couple of decades, earth-abundant metal molecular catalysts for hydrogen production, comprised mainly of cobalt,4–13 nickel14–19 and iron,20–22 have been developed and tested for photo- and electro-catalytic hydrogen production.21,23–29 Whilst many of them have been proven to be active for HER, the majority have been studied in organic media.
Very few molecular catalysts have been shown to be both stable and efficient for hydrogen production from neutral water.7,27 Of these, cobalt-based catalysts are more commonly reported than those based on other late-3d transition metal ions (Fe–Cu), despite the lower natural abundance of Co, possibly because the stability and relative ease of accessibility of the +1, +2 and +3 oxidation states of cobalt.9–13
Recently, copper-based complexes (Fig. 1)30–39 have attracted attention due in part to the higher natural abundance of copper, but it is rare for them to have been shown to be electroactive for HER in neutral aqueous solution. The highly cited example A reported by Wang, Sun and co-workers,36 plus three other examples B,32C,30D,31 and E33 are reported to be active as homogeneous HER electrocatalysts in water at a pH of either 2.5 or 7.
![]() | ||
Fig. 1 Some literature reports of copper-based molecular electrocatalysts for HER in aqueous buffer solutions, at pH 2.5 (A),36 or pH 7 (B,32C, D,30,31 and E33), as well as F,37G,38 and H39 which were used to deposit Cu2O/Cu(OH)2, Cu(0), or Cu(0)/Cu2O species, respectively, which were highly active electrocatalysts for HER, and (in red box) [CuLEt]BF4 (1,34,35 used in this work). |
Both A and B feature N5-donor tripodal ligands. Complex A, reported in 2014 by Wang, Sun and co-workers36 displays high activity for HER, with a turnover number (TON) = 3.0 × 104 and a turnover frequency (TOF) = 7.0 × 103 h−1, at pH 2.5, with an applied potential (Eapplied) of −0.90 V versus SHE, over 2 h, with faradaic efficiency (FE) of 96%. In 2017 Verani and co-workers reported that B was also an active catalyst, at a pH of either 2.5 (TON 3900) or 7 (TON 1670), with FE = 90%.32
Both C and D feature N2O2-donor ligands. The salen-like copper(II) complex, C,30 reported by Zhan and co-workers is an active HER catalyst at pH 7, with Eapplied = −1.23 V vs. SHE over 72 h, with TOF = 457 h−1 and FE = 92%.30 They also reported31 a related copper HER catalyst, D, which at pH 7 and Eapplied = −1.45 V vs. Ag|AgCl over 36 h, has TOFmax = 1332 h−1 and FE = 97%.31 No TONs were reported for either C or D.30,31
More recently, Padhi and co-workers showed that E, featuring an N3-donor terpyridine-based ligand, was active for HER in neutral aqueous solution, with Eapplied = −1.5 V vs. SCE.33
But it must be noted that distinguishing homogeneous from heterogeneous behaviour can be nontrivial,40,41 and some examples originally reported as active molecular catalysts for hydrogen production42 have subsequently been shown to act as precursors, “precatalysts”, to the catalytically active species.43 Indeed there are now many established examples of in situ reduction or decomposition of a molecular complex to form catalytically active nanoparticles and/or deposits on the surface of the working electrode.43–45
Molecular copper complexes have a particular propensity towards electrode deposition during electrocatalytic proton reduction reactions.39 In this context it should be noted that Cu2O and CuOH have previously been employed in photocathodes,46 for example in 2014 by Grätzel and Hu,47 and as photocatalysts,48 to promote hydrogen production.26 Copper nanoparticles have also been shown to be active HER electrocatalysts in water.38,49 The use of copper-based materials (not molecules) as HER electrocatalysts has recently been reviewed.50
Indeed, unlike copper complexes A–E which were reported to be homogeneous HER electrocatalysts, complexes F,37G38 and H39 (Fig. 1) were found to act as pre-catalysts.
Complex F was employed, in 2015 by Du and co-workers, to deposit a heterogeneous copper-based film (mainly Cu2O and Cu(OH)2) on an FTO electrode. The film was shown to be catalytically active for both water oxidation and water reduction at pH = 9.2.37 Interestingly, they showed that starting with just a simple copper salt CuCl2 instead of F, under the same conditions, resulted in a much less catalytically active deposit.37
Similarly, in 2016 Chen and co-workers38 reported that the Cu(II) oxime complex, G, could be used to electrodeposit a Cu(0) nanoparticle film which resulted in robust HER from neutral aqueous solution at Eapplied = −0.35 V vs. RHE for 8 hours.
Also in 2016, Siewert and co-workers reported complex H, with an N4-donor ligand and a coordinated triflate counterion, which under acidic conditions formed a heterogeneous Cu/Cu2O deposit that was active for HER.39 Unlike Du's work above, in this case there was little difference in the catalytic activity of the resulting deposit regardless of whether they started with H or a simple copper salt control, Cu(OTf)2.
The square planar copper(II) macrocyclic complex [CuIILEt]BF4 (1) was first reported by some of us in 2011 (Fig. 1),34 building on work by Black and Rothnie who made the analogous perchlorate complexes.51 Recently, some of us showed that testing 1 resulted in modest electrocatalytic and photocatalytic HER performance in dry acetonitrile:35,52 electrocatalysis carried out with acetic acid as the proton source resulted in TON = 12.5 after 6 h, with the activity ongoing.§ Building on those promising results, herein we report the results of testing 1 for electrocatalytic HER in neutral water.
![]() | ||
Fig. 2 Cyclic voltammetry, 0 → −1.5 → 0 V vs. Ag|AgCl, for 1 M neutral aqueous phosphate buffer in absence (blank, black) and presence of 0.30 mM of 1 (blue; including inset scan rate study 0 → −1.25 → 0 V) and simple copper salt (red, control), overlaid with CVs of complex 1 in MeCN/acetic acid (pink,35 note that the original data was collected vs. 0.01 AgNO3/Ag so it has been shifted to Ag|AgCl reference electrode by adding 0.37 V before being plotted here), using the small H-cell and GCWE (d = 3 mm, A = 0.071 cm2) with scan rate 100 mV s−1. |
Next, the CVs were scanned to more negative potentials, 0 → −1.50 → 0 V vs. Ag|AgCl, revealing a strong catalytic current (Fig. 2, main image in blue). Fig. 2 also shows the CVs of a 1 M neutral aqueous phosphate buffer solution recorded in the absence of 1 (blank, black) and in the presence of 0.30 mM copper(II) tetrafluoroborate hydrate (control, Cu salt, red). Both showed only a very weak current response (at −1.5 V, i = −0.15 and 0.27 mA for the blank and control respectively), whereas the same aqueous solution but containing 0.30 mM complex 1 displayed a sharp increase in current, starting at about −1.0 V and reaching i = −1.4 mA at −1.5 V (Fig. 2). Small bubbles were observed underneath the GCWE after running a CV with a freshly polished electrode (Fig. S6, ESI†). Such behaviour was also reported for the extremely active copper complex CuA in pH 2.5 aqueous solution (Fig. 1).36
Controlled potential electrolysis (CPE) experiments were initially carried out in the small H-cell at Eapplied = −1.1 V vs. Ag|AgCl for 2 hours (Fig. 3 and Fig. S7; this is approx. −1.47 V vs. 0.01 M AgNO3|Ag,53,54 see section S2.2†). The blank (black line, Fig. 3) showed minimal activity (0.05 C so TON2 h < 0.5), whilst the simple copper salt control (red curve, Fig. 3) showed some activity (2.1 C so TON2 h = 4.7) but the plot starts to plateau after the first hour, so the control has a short lifetime of activity (t1/2 = 40 min). In contrast, the solution containing complex 1 (dark blue line, Fig. 3) had activity nearly six-fold higher (12.6 C so TON2 h = 28) than that of the copper salt, and it remains active after 2 hours. To confirm this activity, another, fresh, CPE experiment was performed (light blue line, Fig. 3), which showed similar charge transfer (11.8 C so TON2 h = 26).
A rinse and repeat test (blue dots, Fig. 3), whereby the working electrode was taken out carefully and dipped a couple of times in fresh buffer electrolyte to rinse it, then put into a fresh solution without adding a new batch of catalyst 1, showed only a low charge transfer (0.75 C so TON2 h = 1.7). This rinse test indicates that any deposited material on the working electrode is not the active catalyst for HER – but this test can give false negatives,43 as we go on to show below.
To test if suspended nanoparticles or colloids were formed in the solution and are the catalytically active species, a mercury drop test36 was also carried out. Hence the CPE experiment on 1 was repeated, but this time in the presence of a drop of mercury (100 μL) in the bottom of the working chamber of the H-cell (magenta, Fig. 3). This has only a slightly negative effect on the catalytic activity (10.8 C so TON2 h = 24), compared with the two previous CPE runs in the absence of mercury (12.6/11.8 C; TON2 h = 28/26; dark and light blue lines, Fig. 3). The formation of catalytically active suspended nanoparticles under these experimental conditions is therefore unlikely.
Given HER activity is ongoing at 2 hours, an extended CPE experiment was conducted under the same conditions but for 20 hours. The charge/time profile again stays almost linear (after what appears to be an induction period at the start, see later) over the entire 20 hours (Fig. 4, for current-time plot see Fig. S8†). The charge transferred was 163 C which corresponds to TON20 h = 364 (assuming FE = 100%), activity ongoing.
Notably, the colour of the working solution faded significantly over this extended 20 h CPE run, to almost colourless at the conclusion, despite the ongoing catalytic HER activity observed. This provided a strong indication that, despite the test results described above (including: rinse and repeat; mercury drop), the catalytic material may not be molecular complex 1, but rather some kind of catalytically active species that has deposited on the GCWE during electrocatalytic HER.
To investigate the possibility that 1 forms a heterogeneous catalyst on the GCWE, a larger plate GCWE was employed (25 × 25 mm, with the O-ring used limiting the working diameter and area to 18.7 mm and 2.7 cm2, respectively) in a larger electrolysis cell (for more details see section S3 and Fig. S9 and S10 in the ESI†) to facilitate (a) the analysis of any surface products formed and (b) the use of a gas chromatograph to continuously monitor and quantify the hydrogen evolution during the CPE process, enabling the faradaic efficiency to be determined.
In this larger cell, CPE was carried out at Esetpoint = −1.1 V vs. Ag|AgCl with partial iR compensation using a positive feedback (PF) correction (see ESI section S3.3† for details).55,56 After a series of four consecutive CPE experiments on one working solution (dark blue trace in Fig. 5; see ESI for details, Fig. S13 and 14†), TON62 h = 2628, with activity ongoing, and mean FE = 92.6%. During the course of the CPE the solution faded from yellow to colourless (Fig. 6), with both of the bands in the visible, 458 nm (CT) and 674 nm (d–d), absent after the 62 h of CPE experiments. Visual inspection of the large plate GCWE after the CPE showed clear evidence of a brown deposit (Fig. S19†).
![]() | ||
Fig. 6 UV-VIS spectra of the electrochemical working solution of 1 before (blue) and after (orange) the CPE experiments shown in Fig. 5. The inset shows an expansion of the d–d band region. |
The large GCWE was rinsed in situ by keeping the cell intact but removing the electrolyte from the cell under argon flow, gently refilling and emptying the cell several times with Milli-Q water, then refilling and emptying the cell with fresh electrolyte solution, before refilling the cell with more fresh electrolyte solution (no additional 1). The current achieved by the rinsed large GCWE was similar to that of the original after the initial induction period, resulting in the same slope in the TON vs. t graph (dark gold trace, Fig. 5, Fig. S13 and 14†), and an additional 1466 TON (mean FE = 89%), for a cumulative total TON85 h = 4094 (mean FE = 92%).
In a repeat set of CPE experiments (light blue trace Fig. 5, and Fig. S16, S17†), after the initial 21 h of CPE the near-colourless working solution was collected, then the large GCWE was rinsed and tested as above, showing similar results (pale gold, Fig. 5, Fig. S16 and S17†). The near-colourless working solution was also tested, in a fresh cell with a freshly polished large GCWE, with no additional catalyst 1 added (see ESI†). In this case, the current and catalytic activity measured at −1.1 V vs. Ag|AgCl were drastically reduced (green trace, Fig. 5, Fig. S16 and S17†). These additional experiments clearly confirm that in neutral water the heterogeneous deposit, not molecular 1, is the active HER electrocatalyst.
The controls, using either Cu(OAc)2·H2O or Cu(BF4)2·6H2O, showed less but still moderate HER catalytic activity (black and grey traces Fig. 5 and Fig. S13–S15, and S18, ESI†). The deposits formed during CPE with these controls looked quite different, and were more delicate and detached readily (hindering any further characterisation of them), which is likely the cause of the lower activity (Fig. S19†).
Energy-dispersive X-ray spectroscopy (EDS) spectra (Fig. S20, S21 and Table S2, ESI†) and EDS elemental mapping (Fig. 8) of the large GCWE after CPE showed the presence of oxygen, carbon, potassium and copper. There is no significant nitrogen peak present in the EDS spectra (Fig. S21†), indicating that 1 is not intact on the surface, and the deposits may just be copper or copper oxide structures forming on the surface of the GCWE.
Indeed, powder X-ray diffraction (PXRD) spectra of the GCWE before and after CPE showed the formation of some crystalline material with peaks that appear to come from copper metal and copper(I) oxide ((Cu2O)n) (Fig. S22 and S23, ESI†).
Together, these SEM, EDS and PXRD analyses are consistent with a Cu/Cu2O deposit having formed on the large GCWE. Previous reports have shown similar copper-based deposits to be catalytically active for HER (both electrocatalytic37–39,46–48 and photocatalytic46–48) and this is also seen here (see previous section).
So in summary, it appears that the dissolved molecular [CuIILEt]BF4 (1) is not the active HER electrocatalyst, but rather that 1 has facilitated the formation of a robust and catalytically active Cu/Cu2O deposit on the GCWE. While the use of simple copper salts also formed catalytically active deposits, the deposits formed were much more delicate, tending to fall off the GCWE, hindering both the level and lifetime of the catalytic activity.
The initial CV studies in aqueous phosphate buffer at pH 7, with a small cell and GCWE (0.071 cm2), showed a sharp increase in current, starting at about −1.0 V and reaching icat = −1.40 mA at −1.5 V vs. Ag|AgCl, when 1 was present, but no such current enhancement was observed in the case of the blank or control (CuII(BF4)2·6H2O). Subsequently, CPE at −1.1 V vs. Ag|AgCl revealed a TON20 h of 364 (assuming FE = 100%) was achieved over 20 hours, but with ongoing activity observed.
Next the CPE was repeated under same conditions but using a larger cell and GCWE (A = 2.7 cm2) attached to a gas chromatograph, which also gave promising results: a TON62 h of 2628 over 62 h with mean FE = 93%, again with ongoing activity; indeed when that GCWE was rinsed and placed in fresh electrolyte solution the activity continued at a very similar rate, lifting these values to TON85 h = 4094 and mean FE = 92%, again with activity ongoing at 85 h (3.5 days). This large GCWE allowed surface characterisation of electrodeposited material. In addition to the result of the rinse test, the PXRD, SEM and EDS results confirm that a heterogenous electrocatalytically active deposit has formed, most likely comprised of copper and copper(I) oxide. This copper rich deposit is more robust when formed from 1 rather than from simple copper salts, likely accounting for the somewhat higher and ongoing HER electrocatalytic activity. Nevertheless, this is neither a viable synthetic route (6 steps then decomposition), nor is the resulting HER activity sufficient (Fig. S18,† typical average current is about 20 mA, which corresponds to a low current density of 7–8 mA cm−2) for industry applications. Hence our focus has moved onto the deposition of our molecular complexes onto a range of solid supports before testing them as heterogenous catalysts, as this should enhance their stability, catalytic activity and lifetime.
All electrochemical measurements were carried out on Ar saturated 1 M K2HPO3–KH2PO3 buffer electrolyte solution of pH 7. Cyclic voltammetry (CV) and controlled potential electrolysis (CPE) experiments were carried out using Ag|AgCl as the reference electrode with Pt mesh, plate, or wire as counter electrode and glassy carbon working electrodes (d = 3 mm, A = 0.071 cm2 for the small H-cell; d = 18.7 mm, A = 2.7 cm2 for the larger cell). For more details see ESI.† Turnover number (TON)57 was calculated as
TON = nH2/nCat |
Footnotes |
† Electronic supplementary information (ESI) available: Electrochemical instrumentation, cells and methodology, as well as additional cyclic voltammetry and controlled potential coulometry studies and working electrode surface analysis. See DOI: https://doi.org/10.1039/d4dt00224e |
‡ Present address: Institute of Condensed Matter and Nanosciences (IMCN), Molecular Chemistry, Materials and Catalysis (MOST), Place Louis Pasteur 1, 1348 Louvain-la-Neuve, Belgium. |
§ Note that whilst ‘rinse and repeat’ and ‘mercury drop’ tests were negative in that study, the possibility that the observed catalytic activity was due to a heterogeneous species formed from 1 during CPE in MeCN/acetic acid could not be, and was not, ruled out.35 |
This journal is © The Royal Society of Chemistry 2024 |