Tung H. To*ab,
Dang B. Tran
bc,
Vu Thi Thu Had and
Phong D. Tran
*b
aGraduate University of Science and Technology, Vietnam Academy of Science and Technology, 18 Hoang Quoc Viet, Hanoi, Vietnam. E-mail: to-hai.tung@usth.edu.vn
bUniversity of Science and Technology of Hanoi, Vietnam Academy of Science and Technology, 18 Hoang Quoc Viet, Hanoi, Vietnam. E-mail: tran-dinh.phong@usth.edu.vn
cHo Chi Minh City University of Education, 280 An Duong Vuong, Ho Chi Minh City, Vietnam
dInstitute of Chemistry, Vietnam Academy of Science and Technology, 18 Hoang Quoc Viet, Hanoi, Vietnam
First published on 16th September 2022
We report herein on the use of two binuclear cobalt complexes with the N,N′-bis(salicylidene)-phenylmethanediamine ligand as catalysts for the H2 evolution in DMF solution with acetic acid as proton source. Both experimental analyses (electrochemical analysis, spectroscopy analysis) and theoretical analysis (foot-of-the wave analysis) were employed. These catalysts required an overpotential of ca. 470 mV to catalyze the H2 evolution and generated H2 gas with a faradaic efficiency of 85–95% as calculated on the basis of after 5 hour bulk electrolysis. The kinetic investigation showed the maximal TOF value of 50 s−1 on the basis of an ECEC mechanism. Two cobalt centers, standing at a long distance of 4.175 Å, operated independently during catalysis without a synergetic effect or cooperation capability.
In 2009, Peters and colleagues reported on the preparation of dicobalt pyridazine-bridged complexes that functioned as efficient H2 evolution catalysts in acetonitrile electrolyte solution with 2,6-dichloroanilinium tetrafluoroborate as the proton source.10 A series of binuclear cobalt catalysts containing polypyridine ligands and a pyrazine-bridging linkage was reported by Long and colleagues.11 These catalysts have been evaluated for both electrocatalytic H2 evolution and photocatalytic H2 evolution in coupling with a [Ru(bpy)3]2+ photosensitizer. Gray and colleagues reported on a dicobalt coordination compound with dimethylglyoxime as the ligand and one [BO4] linkage.12 This complex was found to be active for catalyzing the proton reduction in acetonitrile solution using protonated dimethylformamide as the proton source with an overpotential of 954 mV; TON of 3.3, and the faradaic yield of H2 production of 88%. Another dinuclear cobalt HER catalyst with a tetrakis-Schiff base ligand was reported by Dinolfo.13 The catalyst required a relatively high overpotential of 880 mV and 310 mV when using 55 mM of trifluoroacetic acid and 55 mM acetic acid as the proton sources, respectively. The bulk electrolysis at −1.88 V vs. Fc+/0 (corresponding to 530 mV of overpotential) in the acetic acid solution generated H2 with a faradaic efficiency of 72–94%.
The H2 evaluation mechanism occurred on mononuclear cobalt complexes catalysts has been intensively investigated. It has been accepted that the protonation of CoI species generating the CoIII–H hydride represents the key elemental step of catalysis. Subsequently, H2 can be produced by two pathways, namely (i) heterolytic pathway where the CoIII–H species or its reduced state CoII–H is protonated releasing a H2 molecule; (ii) homolytic pathway where two CoIII–H species combine releasing a H2 molecule via a reductive elimination.14,15 Hence, development of binuclear Co molecular catalysts for the H2 evolution could be of great interest given that the homolytic pathway which is more energetic favorable compared with the heterolytic pathway could occur in a unimolecular or bimolecular way. In 2013, Fukuzumi and colleagues reported on detailed kinetics and discussed on the H2 evolution mechanism on dinuclear cobalt complexes containing bispyridino pyrazine and terpyrdine ligands.16 This study confirmed the formation of CoIII–H hydride complex and the heterolytic cleavage to produce H2. In 2018, a bis(thiosemicarbazone) cobalt complex prepared by Artero and colleagues showed an attractive catalytic activity for H2 evolution.17 On the basis of DFT calculations, the H2 evolution was proposed to follow the ligand-assisted metal-centered pathway which involved the ligand reduction and the formation of CoIII–H species. The work reported by Verani and colleagues demonstrated that the distance between two cobalt atoms was decisive to the formation of CoII–(H−)–CoII hydride, e.g. via the two-electron donation from each CoI moiety to the proton.18 This bridging-hydride intermediate is reasonable with most known Co2(μ-H) have Co–Co distance in the range of 2.2–2.3 Å.19,20
In this work, we report on the synthesis and structural characterization of two binuclear cobalt complexes as well as their electrocatalytic properties for hydrogen evolution in an organic medium.
![]() | ||
Fig. 1 Molecular structure of complex 2. Hydrogen atoms are omitted for clarity. Symmetry operations used to generate equivalent atoms: i1 −x, y,1/2 − z. |
Actually, the X-ray diffraction analysis on single crystals obtained from the slow diffusion of n-hexane into a dichloromethane solution of complex 2 provided the similar structure to that reported previously.23
The Co(2) atom shows the tetrahedral geometry which is created by the coordination with two oxygen atoms and two imine nitrogen atoms of two ligand molecules. Because of +3 oxidation state, the Co(1) atom displays the regular octahedral geometry with two additional pyridine ligands. We note that for the cases of dicobalt complexes aforementioned,10,11,13,18 the Co(III) center has always octahedral geometry but geometry of the Co(II) center may varies. The complexes reported by Peters,10 Dinolfo13 and Long11 showed Co(II) center with an octahedral coordination geometry while that reported by Verani18 showed a tetrahedral geometry for the Co(II) center. Herein, our current complexes 1 and 2 show Co(II) center with tetrahedral geometry. These complexes 1, 2 show common bond length usually found for the Co–N or Co–O coordination bonds within cobalt complexes, e.g. insignificant difference of only less than 0.3 Å was determined (Tables S1 and S2†).10–13,16–18 The distance between two Co atoms is determined to be 4.175 Å which is rather too long for expecting the formation of a hydro-bridging linkage. Indeed, with Co–Co distances of 3.7–4.3 Å, formation of a hydride-bridging linkage is not favorable.18–20 Therefore, these complexes 1 and 2 are expected to use the two Co centers as two independent active sites during the catalytic H2 evolution without a strong electronic communication between these two centers.
![]() | ||
Fig. 2 Cyclic voltammograms of 1 (red trace) and 2 (blue trace) recorded at a stationary glassy carbon electrode in DMF with a potential scan rate of 50 mV s−1. |
These two events were also found for the complex 2 (Fig. 2, blue trace) including the CoIICoII/CoICoI redox at E1/2 of −1.87 V vs. Fc+/0 (with peak-to-peak ΔEp of 82 mV). However, due to the presence of counter anion triiodide I3− in the structure, the complex 2 showed also a redox event of I−/I couple having Epc of −1.14 V vs. Fc+/0 and Epa at 0.12 V vs. Fc+/0 (Fig. S1†). The broad peak of this couple overlapped the peak at Epc = −0.70 V vs. Fc+/0 assigned to the reduction of CoIIICoII/CoIICoII couple (Fig. 2).17 A clear observation of CoIIICoII/CoIICoII reduction peak could be confirmed when excess amount of acetic acid was added (Fig. S2†). This phenomenon was explained by the reaction of acetic acid with iodine molecule leading to a decrease in the amount of iodine participating in the reduction process, thereby reducing the peak overlap.
To ensure that the proton reduction is catalyzed by homogeneous catalysts 1 and 2 but not their eventual deposited products, “rinse test” control experiments were carried out. The glassy carbon electrode, used for recording CVs of a solution containing 1 mM catalyst 1 and 60 mM AA, was taken out and rinsed intensively with DMF solvent. Subsequently, the electrode was polarized when being immersed in a DMF solution containing 60 mM AA but being free of any catalyst 1 or 2. Negligible cathodic current was recorded, demonstrating the non-catalytic character of this electrode. In other word, we can conclude no catalytic active deposits have been produced from the homogeneous solution of catalysts 1 and 2 (Fig. S3†). This is in contrast to some previous studies, which described about the formation of deposited cobalt nanoparticle from cobalt complexes as heterogeneous catalysts. Aukauloo and colleagues24 and Artero and colleagues25 showed that the actual active species for H2 evolution consisted of Co-based nanoparticles (mentioned as CoOx) formed on the electrode surface due to the degradation of the pyridine–oxime cobalt complex and diamine–dioxime cobalt complexes, respectively. We note that, the cobalt complex decomposition causing the formation of cobalt based deposits acting as the actual catalyst for the water oxidation reaction has been also demonstrated by Hetterscheid and colleagues.26
We then aimed to probe the structural change of complexes 1, 2 during their catalytic operation by using the UV-Vis spectroscopy analysis. Prior to use for the catalysis, complexes 1 and 2 in DMF exhibited two absorption bands: the first band at λmax of 266 nm (for 1) and 276 nm (for 2) and the second band at λmax of 378 nm (for 1) and 400 nm (for 2) (Fig. S4†). After each potential polarization cycle for assaying the electrochemical catalytic activity of complexes with the increase concentration of acetic acid, the UV-Vis spectra of catalyst solutions were collected. It can be seen that the absorption bands remained at the same wavelengths but their absorption intensity decreased progressively (Fig. 4). This observation suggests a partial degradation of complexes 1 and 2 during the H2 evolution catalysis. However, the degradation is likely different to a simple decomplexation as the characteristic absorption band at 325 nm of the free ligand was not observed (Fig. 4 and S4).†
![]() | ||
Fig. 4 UV-Vis spectra of complexes 1 (a) and 2 (b) after evaluating electrochemical activity with different concentration of acetic acid. |
Bulk electrolysis was conducted at a constant potential of −1.93 V vs. Fc+/0 using a DMF electrolyte solution constituted of 1 mM catalyst 1 (or 2) and 60 mM AA. The amount of H2 gas produced was quantified by GC analysis. On the basis of 5 hours bulk electrolysis experiments, the average faradaic efficiency was deduced to be 95% for the catalyst 1 (namely CoIICoII complex) and 86% for the catalyst 2 (namely CoIIICoII complex) (Fig. S5†). After 5 hours of bulk electrolysis, the electrochemical properties of the remained catalyst solutions were examined using a new fresh glassy carbon electrode. It was found that catalytic activities remained almost unchanged for both catalysts 1 and 2 (Fig. S6 and S7†).
Indeed, overpotential (η) is a key descriptor to appreciate a H2-evolution catalyst and compare one to others. It can be deduced from the onset potential, being the potential where the H2 evolution starts to emerge,27 the potential required to sustain a given catalytic current density,27 or the half-wave potential of the catalytic wave.28 Herein, the overpotential required for the H2 evolution when using both catalysts 1 and 2 in DMF solution is determined at the half-way potential in the presence of 60 mM concentration of acetic acid. Under such an operation condition, the thermodynamic reduction potential of the acetic acid in DMF can be estimated by the eqn (1) following Artero and colleagues,29 providing the
value of −1.46 V vs. Fc+/0 (see ESI†). Thus, the overpotentials for the H2 evolution on complex 1 and 2 are 460 and 470 mV, respectively (Fig. S8†).
![]() | (1) |
In comparison with Dinolfo's complex13 having the comparable structure (based on shift-base ligand), our complexes requires more than 150 mV overpotential for catalytic operation using the same acetic acid as proton source. We note that two types of complexes were investigated in two different solvents (Dinolfo's in Acetonitrile and ours in DMF) so this difference is possible. When comparing with dicobalt–glyoxime complexes reported by Gray and colleagues,12 our current dinuclear cobalt complexes showed better catalytic activity with requirement of lower onset overpotential (460 mV vs. 954 mV) although the former complexes were evaluated the activity for HER with stronger acid (protonated DMF pKa 6.1 in acetonitrile4). The detailed comparison of these two catalysts with other published dicobalt complexes is presented in the Table S6†.
In this case, the concentration of AA is largely excess, thus the acid concentration tends to be constant during the catalysis. Fig. 5 shows the evolution of maximum catalytic current at different potential scan rates in function of catalyst concentration. Linear dependence can be seen for both catalysts, from which the first order can be deduced.
![]() | ||
Fig. 5 The line graphs of catalytic current versus concentration of complexes 1 (a) and 2 (b) with different scan rate in the presence of 60 mM acetic acid solution. |
We then carried out the foot-of-the-wave analysis (FOWA) to determine the catalytic rate constant and predict the maximum turnover frequency. In comparison with the plateau current analysis, FOWA can reduce the effect of side phenomenon such as the depletion of proton at the surface of electrode or the deactivation of catalyst after potential scanning cycles, e.g. deactivation due to the formation new complexes of conjugated base-catalyst or prohibition of active sites by hydrogen evolved. Few FOWA models have been developed by Costentin and Saveant which could be applied for different H2 evolution mechanisms.30,31 Actually, a find prediction of catalysis mechanism is required prior to applying the FOWA.32 As aforementioned, the H2 evolution is initiated by the one-electron reduction of CoII center generating CoI species (denoted hereafter as E for electron transfer step) which is then protonated providing CoIII–H-hydride (denoted here after as C for chemical step). This hydride complex underwent the second EC process resulting in the evolution of H2 molecule. In other words, the H2 evolution occurred on catalysts 1 and 2 via a ECEC mechanism. We note this ECEC mechanism was usually found for cobalt complexes such as cobaloximes,14,33,34 diimine–dioxime cobalt complexes35 or pendant amines cobalt complexes36,37 (Scheme 2).
![]() | ||
Scheme 2 The proposedly heterolytic mechanism of 1 or 2 catalyzing HER (where [CoI], [CoII] and [CoIII] represented the active sites). |
For such a ECEC mechanism, the ratio of the catalytic current over the peak current is a function of overpotential as described by eqn (2) wherein icat is the catalytic current (mA), ip is the cathodic peak current (mA), R = 8.314 J mol−1 K−1, T = 299 K, F = 96485C mol−1,31,38 ν is scan rate (V s−1), kobs = kcat[H+] (s−1), E is the potential at the foot of the wave (V) and Ecat/2 is the half-wave potential of the catalysis (V).
![]() | (2) |
The plots of icat/ip in function of exp[−F/RT(E − Ecat/2)] is depicted in Fig. 6.
![]() | ||
Fig. 6 The plot of icat/ip as a function of exp [−F/RT(E − Ecat/2)] for 1 mM catalysts 1 (a) and 2 (b) in various concentration of acetic acid. |
The values of the observed rate constants kobs and the catalytic rate constants kcat were extracted from the slope of linear plots and showed in Table S5.† Using these rate constants, the maximal turnover frequency (TOFmax) at the foot of the catalytic wave was deduced (Fig. 7). In the case of complex 1, the catalytic rate constant increased gradually when increasing AA concentration from 39 mM to 159 mM before it reached a plateau at higher acid concentration. The TOFmax value therefore approximates kcat[AA] and can be extrapolated to be 50 s−1 for AA concentration of 1 M. In the shape contrast, no obvious evolution of the catalytic rate kcat was found for the complex 2. The kcat value fluctuated around 800 M−1 s−1 when the concentration of AA increased from 39 mM to 159 mM. For the higher acid concentration, the slope became steeper, showing the faster rate of catalysis. The [H+]-dependent rate constant showed the catalysis had not come into the saturated region of acetic acid. Thus, complex 2 was expected to afford the TOFmax value of over 1600 s−1 with 1 M AA. It could be explained because the diffusion coefficient of 2 was about one order of magnitude higher than that of 1 (see ESI†), which can foster the rate of electron transfer.
![]() | ||
Fig. 7 The kcat extracted from FOWA in various concentration of AA with 1 mM catalyst 1 (a) and 2 (b). |
Footnote |
† Electronic supplementary information (ESI) available. See https://doi.org/10.1039/d2ra05109e |
This journal is © The Royal Society of Chemistry 2022 |