C.-F.
Leung
*a,
S.-C.
Cheng
b,
Y.
Yang
a,
J.
Xiang
ac,
S.-M.
Yiu
b,
C.-C.
Ko
*b and
T.-C.
Lau
b
aDepartment of Science and Environmental Studies, The Education University of Hong Kong, 10 Lo Ping Road, Tai Po, Hong Kong, China. E-mail: cfleung@eduhk.hk
bDepartment of Chemistry, City University of Hong Kong, Tat Chee Avenue, Kowloon, Hong Kong, China. E-mail: vinccko@cityu.edu.hk
cDepartment of Chemical and Environmental Engineering, Yangtze University, Jingzhou, China
First published on 20th November 2017
The visible light-driven catalytic water reduction by a cobalt(II) tripodal iminopyridine complex [Co(tachpy3)](ClO4)2 (1) (tachpy3 = cis,cis-1,3,5-tris(pyridine-2-carboxaldimino)-cyclohexane) has been investigated in aqueous acetonitrile. The catalysis is found to be homogeneous and an impressive H2 TON of 16440 has been achieved, which is far better than that by a structurally related complex [Co(trenpy3)](ClO4)2 (2) (trenpy3 = tris-[4-(2-pyridyl)-3-azabut-3-enyl]amine) bearing identical donor groups. A comparison between the two complexes reveals that the catalytic properties are sensitive to the change in ligand rigidity and thus the coordination geometry, as reflected by their respective reaction potentials and H2 TONs. The catalysis of 1 proceeds via a penta-coordinate species formed by the protonation of the corresponding CoI state. The results in spectroscopic and electrochemical studies are supplemented by DFT computation.
There have been only a few studies on Co iminopyridine WRCs despite their fast catalytic rates.20–23 We report herein the photocatalytic water reduction by a cobalt(II) tripodal iminopyridine complex [Co(tachpy3)](ClO4)2 (1) (tachpy3 = cis,cis-1,3,5-tris(pyridine-2-carboxaldimino)-cyclohexane) in aqueous acetonitrile (Scheme 1).451 is found to be an active molecular WRC. To shed light on the factors affecting the reactivity, a structurally related complex [Co(trenpy3)](ClO4)2 (2) (trenpy3 = tris-[4-(2-pyridyl)-3-azabut-3-enyl]amine)46 which bears the same metal centre and donor groups is also investigated. The catalytic properties, e.g. stability and reaction potential, are compared using photo- and electrochemical methods supplemented with DFT computation. The coordination geometry and the more rigid ligand structure in 1 are suggested to contribute to its enhanced WRC reactivity.
No H2 could be detected in the absence of IrPS+, H2O or TEA (Fig. 1). Both the production rate and yield of H2 decrease, when the amount of H2O is lowered from 10% to 2.5% (Fig. S2†); increasing [H2O] to 15% results in no further improvement in the H2 yield. Cyclometallated Ir photosensitizers are known to decompose via the dissociation of the diimine ligand from the unstable reduced state.47,48 In an attempt to restore the H2 production, a fresh solution of IrPS+ (0.2 mM) was added to the reaction solutions after the first 16 h of photolysis. The activity of 1 was restored and a comparable amount of H2 (12.3 μmol) was produced after another 16 h of irradiation, indicating that the decomposition of IrPS+ is the major cause for the loss of activity. H2 production was maintained and a total H2 yield of 109 μmol was attained after 5 cycles of IrPS+ addition (overall 80 h of irradiation), whereas only 27.4 μmol H2 was recorded in the control experiment without the catalyst, leading to an effective TON of 16440 (vs. catalyst). The performance of several reported Co WRCs is shown in Table 1. Co catalysts bearing pentadentate polypyridyl ligands are, so far, the most efficient among all Co WRCs reported.50–52 A H2 TON of 11000 was reported for [CoII(aPPy)(Br)]Br (aPPy = bis-2,2′′-bipyridine-6-yl(pyridine-2-yl)methanol), using a ReI photosensitizer and ascorbic acid/ascorbate as the sacrificial donor under aqueous conditions,50 while a higher TON (30000) has been achieved when a secondary sacrificial electron donor was provided.51 A H2 TON of 20000 has recently been reported also in acetonitrile for Co(TMPA)Cl2 (TMPA = 1-(6-substituted-pyridin-2-yl)-N,N-bis(pyridin-2-ylmethyl)methane amine).52 The H2 TON for 1 is therefore amongst the highest reported for Co WRCs without the use of a secondary sacrificial donor. A similar yield (9.2 μmol) and response to H2O concentration have been observed for 2 in the first catalytic cycle (16 h) under the same conditions (Fig. 1 and S2†). However, the yield of H2 in the subsequent cycles is similar to that of the control experiment when additional IrPS+ was provided, reflecting that the catalyst has been deactivated after the first cycle. The change in the H2 yield during the first hour of photolysis at varied concentrations of 1 has been recorded (Fig. 2). At the selected concentrations, there was no observable lag phase and H2 yield increases linearly with time, indicating that the catalytic reaction is homogeneous.40 The first order rate constant (kinitial), obtained as the initial rate of H2 production in the first hour, exhibits linear dependence on [1]. A second order rate constant k2 of 0.25 h−1 was obtained accordingly.
Fig. 1 Hydrogen evolution of 1 and 2 (1 μM) in aqueous (10% v/v) acetonitrile at 25 °C. [IrPS+] = 0.2 mM and [TEA] = 0.2 M upon photoexcitation. |
Metal-based nanoparticles generated by electro- or photochemical deposition are found to be real active species for some noble and non-noble metal WRCs.40–44 To verify whether the observed catalytic activity is caused by the colloidal particle formation rather than by 1 itself, the photolysis was performed in the presence of Hg0 (Fig. S3†). The activity of 1 does not change significantly (<20%) in the presence of Hg0 (0.5 mL). The temporal change of the particle size during the photoreactions (first 8 h) has also been monitored using dynamic light scattering (DLS) measurement (Fig. S4†). It falls in line with the above result; particle formation and significant change in particle size were not observed during the photoreactions in comparison with a fresh reaction solution, indicating that the catalysis is homogeneous. Similar results are also observed for 2.
The change of the absorption spectrum of acetonitrile solution containing IrPS+ (0.25 mM) and TEA (6.25 mM) upon irradiation with visible light (λ > 420 nm) is shown in Fig. S5.† New absorption bands are observed at 463, 497 and 532 nm after 120 s, which are consistent with those reported for electrochemically generated IrPS0.44 The quenching rate constants (kq) for IrPS+ with TEA, 1 and 2 have been determined with Stern–Volmer plots (Fig. S6†) at varied concentrations of the quenchers (kq = 1.2 × 109, 4.8 × 108 and 1.3 × 108, respectively). The measured kq for IrPS+ with TEA is comparable to the recently reported value (1 × 109).44 Thus, the photoreaction is believed to proceed predominantly by the reductive quenching of photoexcited IrPS+ to produce IrPS0, which then reduces the Co catalysts (CoII to CoI) to mediate proton reduction. The quantum yields of H2 generation over the first 2 h for 1 and 2 are similar (φ = 0.56), as determined using a chemical actinometer.
To reveal the nature of active species and redox processes involved, cyclic voltammetry (CV) of 1 and 2 was performed in acetonitrile (Fig. 3 and Table 2). In the CV of 1 (Fig. 3a), quasi-reversible and reversible couples are observed at 0.47 and −0.68 V vs. SCE, tentatively assigned to the CoIII/II and CoII/I processes. The quasi-reversible couples at −1.50 and −1.80 V are assigned to the ligand-centred L0/˙− and CoI/0 processes. The CV of 2 exhibits similar CoIII/II and CoII/I couples at 0.45 V and −0.98 V vs. SCE (Fig. 3b, red line). An irreversible L0/˙− couple (Ep,c = −1.61 V vs. SCE) is observed on further cathodic scanning (Fig. 3b, black line). In the subsequent anodic scan, it is noted that CoIII/II and CoII/I couples became less reversible. The irreversible L0/˙− process is probably associated with the deactivation of 2 during the photocatalysis. A comparison between the redox potentials of the two complexes reveals a substantial difference of E(CoII/I) by 0.30 V, while E(CoIII/II) and E(L0/˙−) are less sensitive to the change in the coordination sphere. Thus, tachpy3 is suggested to better stabilize both the coordination environment and the CoI state than trenpy3.
Fig. 3 CVs of 1 mM (a) 1 and (b) 2 in 0.1 M nBu4NPF6 acetonitrile solution under an Ar atmosphere. Scan rate = 100 mV s−1. |
On addition of one mole equivalent of 4-bromoanilinium triflate (BAN), a new peak is observed in the CV of 1 at a slightly positive potential of the CoII/I couple (onset potential = −0.55 V). The peak current increases at higher BAN concentrations (1–6 mM), suggesting the catalytic reduction of protons via the CoI state (Fig. 4). A similar catalytic wave is also observed for 2 at a slightly positive potential of the corresponding CoII/I couple (onset potential = −0.76 V) in the presence of BAN. As [BAN] increases, the peak current of the wave increases, while the onset potential is anodically shifted. When aqueous citric acid (CA) was added instead of BAN, similar catalytic waves are also observed in the CVs at the same potentials (Fig. S7†). In the absence of the catalyst, both acids give rise to negligible reduction currents (Fig. S8†).
To characterize the intermediate involved in the protonation of the CoI state, controlled-potential electrolysis was performed (−1.0 V vs. SCE) with 0.1 M nBu4NPF6 acetonitrile solution containing 1 (1 mM). The solution was then diluted (×10) and analysed with ESI-MS (Fig. S9 and S10†). The observed single peak at m/z = 629.5 in the MS is assigned to the N2-adduct [Co(L) + H + PF6 + N2]+ (supported by the isotopic pattern) and suggests the formation of dicationic [CoI(η5-HL)(N2)]2+, which bears in the presence of solvent moisture a protonated tachpy3 ligand. Such a N2 adduct is not observed in the MS of the corresponding [CoII(L)]2+ (Fig. S10†). Under photochemical conditions, the more negative reductive IrPS0 state (−1.36 V vs. SCE)44 should in principle be sufficiently reducing to generate, in the presence of a proton source, similar five-coordinate [CoI(η5-HL)]2+. The formation of [CoI(η5-HL)]2+ upon protonation of [CoI(L)]+ is consistent with the observed new peak near the CoII/I couple and also with the DFT calculation (Fig. 6). CoIII hydride [CoIII(H)(η5-L)]2+ may then be formed via an intramolecular PCET process and further react to produce H2.15 The presence of a CoIII–H intermediate is supported by the CV of 1 (Fig. 4 and S7†). A new quasi-reversible couple is observed at −0.85 V vs. SCE, upon addition of 1 mole equivalent of the acids, and assigned to the [CoIII/II(H)(η5-L)]2+/+ process. A similar CoIII/II(H) couple was reported for a Co oxime WRC.15
X-ray crystal data obtained in this (Fig. S1, Table S1 and S2†) and previous work45 show that the CoII complexes 1 and 2 exhibit a trigonal prismatic and a distorted octahedral geometry, respectively. The energies of the possible spin states of the two complexes in acetonitrile were calculated by DFT methods.61–69 The high-spin states of 1 and 2 (with 3 unpaired d electrons) are predicted to be more stable than the low-spin states by 42.2 and 48.7 kJ mol−1, respectively. It is expected that only the high-spin state would be observed at room temperature. The measured effective magnetic moments (μeff) of 1 and 2 at room temperature are 5.11 and 4.54 BM, respectively, which are typical for a high-spin d7 configuration. The high-spin state nature of the two complexes at room temperature is also revealed in the UV-vis absorption spectra. By comparison of the absorption features, the experimental absorption spectra match with the calculated spectra of the high-spin states of the complexes, rather than the low-spin ones. The difference of the transition energies of the absorption features of the two complexes is reproduced by the calculation, but the energies are overestimated by 0.2–0.4 eV (Fig. S11†). Overestimation of absorption energy is commonly observed for other Co complexes.70,71
The optimized structures of 1 at the high-spin Co0, CoI and CoII states (Fig. 5 and S12, Tables S3 and S4†) exhibit a trigonal prismatic geometry. In contrast, an octahedral geometry is energetically favored for the low-spin CoII state. All optimized structures of 2 adopt a distorted octahedral geometry. The optimized structures for both complexes in the CoII state are therefore consistent with the experimental results from X-ray crystallography and spectrophotometry. A close look at the optimized structures of 1 and 2 gives possible explanations for the different coordination geometries. When comparing the degrees of freedom of the three imine N atoms, the tris(imino)cyclohexane group is less flexible compared to the tris(2-iminoethyl)amine moiety. Also, due to the partially filled Co dσ* orbitals, which are anti-bonding (Co–N) in nature, in the high spin state of 1 (in the oxidation state of Co0, CoI and CoII), their structures would have much longer Co–N bond lengths than the corresponding optimized structures in the low spin state. Thus, with longer Co–N in the high-spin state of 1, the C–N bonds between the imine and the cyclohexyl group will be strained if the complex adopted an (distorted) octahedral geometry. As a result, the distorted octahedral structure in the high spin state is much more unstable with respect to that in the trigonal prismatic geometry.
The reduction potentials of the two CoII/I redox couples were calculated using the differences of the Gibbs free energy of the optimized structures with CoII and CoI metal centers. The values of the calculated Co(II/I) reduction potentials for 1 and 2 are close to the experimental values and the large difference in reduction potentials observed in the experiment can be reproduced (Table 3). The calculated CoI/0 reduction potential for 1 is also consistent with the measured value. Given the fundamental difference in the coordination geometry for both the measured and optimized structures of the two complexes, the stabilized E(CoII/I) in 1 is attributed to its trigonal prismatic geometry.
Calculated E°(CoII/I)/V vs. SCE | Calculated E°(CoI/0)/V vs. SCE | |
---|---|---|
1 | −0.67 | −1.83 |
2 | −0.89 | — |
When the CoI state of 1 is optimized by the DFT method with an added proton, it is shown that the pyridine moieties of tachpy3 are more likely to be protonated. The protonation causes the pyridine moiety to dissociate from the CoI center, while the other five nitrogen-donor atoms are only slightly affected (Table S5†). Subsequently, a penta-coordinate structure [CoI(η5-HL)]2+ is obtained, in line with our experimental data (Fig. 6 and S10†). On the other hand, further reduction of the singly positively charged CoI state of 1 leads to an optimized structure which reveals the neutral doubly reduced [CoI(L˙−)]0 state and supports the assignment of ligand-centred reduction. The extra spin densities corresponding to the added electron would be mainly residing on the –NC(H)–C moiety of the two iminopyridine functions closer to the CoI centre, with a smaller amount also on the aromatic rings, (Fig. S13†), while much lower spin density would be located on the one further away.
The above results demonstrate that replacing the tris(2-iminoethyl)amine moiety in 2 with tris(imino)cyclohexane in 1 has significantly enhanced the catalyst stability and lowered the thermodynamic barrier for proton reduction. 1 is found to be an active homogeneous WRC with impressive catalytic performance. The Co–N bond in the singly reduced [CoI(L)]+ state of 1 is destabilized in the presence of a proton source to give the five-coordinate [CoI(η5-HL)]2+, which may then give [CoIII(H)(η5-L)]2+ to produce H2. Further investigation on the reaction mechanism is now in progress and will be published in due course.
The quantum yield for hydrogen generation was determined as follows. A reaction solution (2 mL) in a 1 cm quartz cuvette sealed with a rubber septum was degassed by bubbling argon for 20 min and then irradiated at 430 nm for 2 h using a 500 W mercury–xenon lamp equipped with a Newport monochromator (Model 77200). The hydrogen content in the headspace of the cuvette was then determined by GC-TCD. The incident light intensity at 430 nm was taken from the average values measured just before and after the photolysis experiment using ferrioxalate as a chemical actinometer and was corrected for the absorbance of the actinometer solution at the same wavelength.58 The quantum yield of hydrogen generation (ΦH2) of the catalytic system is defined by eqn (1) and calculated on the basis of the total amount of hydrogen evolved after 2 h of irradiation:59,60
ΦH2 = 2 × H2 produced (mole)/photon absorbed (Einstein) | (1) |
(2) |
(3) |
Footnote |
† Electronic supplementary information (ESI) available. CCDC 1565200 contains the supplementary crystallographic data for this paper. For ESI and crystallographic data in CIF or other electronic format see DOI: 10.1039/c7cy01524k |
This journal is © The Royal Society of Chemistry 2018 |