Probing Photocatalytic Hydrogen Evolution of Cobalt Complexes:Experimental and Theoretical Methods①

WUHi-Su MIAO Ti-Fng②SHIHi-Xi XU Yun②FUXin-LingQIAN Li

a (School of Chemistry and Materials Science, Huaibei Normal Uniνersity, Huaibei 235000, China)

b (College of Pharmacy, Youjiang Medical College for Nationalities, Baise 533000, China)

ABSTRACT This work reports on the synthesis and characterization of a series of Schiff-base cobalt(Ⅲ)complexes 1～4 that exhibit an obvious catalytic activity for hydrogen production in aqueous solution using fluorescein (FL) and triethanolamine (TEOA) as photosensitizer and electron donor, respectively. The complexes display the capability of splitting of water for H2 evolution. Under optimized conditions, complex 3 shows better properties for photocatalysis, 25 mg of which can release 152.3 μmol of H2 after irradiation for 3 h. The mechanism for light-driven H2 production was explored by experiments and density functional theory (DFT). Meanwhile, the reason of releasing hydrogen was explained theoretically in detail. The research results will help to understand the interaction of cobalt complexes with the photosensitizer and design new photocatalysis for the future.

Keywords: hydrogen production, cobalt complex, photocatalysis, theoretical calculation;

1 INTRODUCTION

Hydrogen (H2), a source of clean and renewable fuel, has attracted great interest for its potential to store solar energy and to reduce the current dependence on fossil fuels[1,2].Reduction of water to H2with visible light as trigger force has been a subject of attractive attention, and great efforts have been devoted to the design of metal complexes for proton reduction[3-5]. Over the past decades, for reasons of price and practicality, a considerable number of photocatalysts for H2generation based on non-precious metal complexes such as Fe, Co, Ni, Cu, and Mn have been reported[6-16]. Recently, Li and co-workers have described the photocatalytic proton reduction catalyzed by a cobalt(II)-salen complex with a remarkable turnover number (TON) of＞ 319 in a MeOH-H2O solution[17]. Although there has been significant progress in designing molecular catalysts for H2evolution, the search of efficient, robust, and earth-abundant metal photocatalysts that can work in aqueous solution still require further development.

With good performance and low cost, Co-complexes supported by different organic ligands have been reported to catalyze the reduction or oxidation of water in organic solution[3,18-26]. However, the hydrogen production performance in pure water was hardly reported. Moreover,the current hydrogen-evolution mechanism for Co complexes is still vague, adding difficulty for us to design catalysts and to select photosensitizers. To find out the mechanism of hydrogen released, Co(Ⅲ) complexes with Schiff base were synthesized, and their structure diagrams are shown in Fig. 1.Compared with the reported hydrogen-evolution system, our system was tested in aqueous solution, which was composed of Co complex, fluorescein (FL) and triethanolamine (TEOA)as a catalyst, a photosensitizer and an electron donor,respectively. Here, Co-complexes provided active sites by obtaining electrons from FL under light irradiation,facilitating the proton reduction. Moreover, the best catalytic environment was found by changing the pH and the concentration of photosensitizer continually. Compared with the experimental method, density functional theory (DFT)provides us a convenient way to understand the relationship between structure and properties, and helps us to explore the hydrogen-evolution mechanism of Co complexes[27,28]. In this work, the interaction properties between photosensitizers and catalysts were calculated, and the hydrogen-evolution reason of Co complexes was explored. The results will be favorable for us to understand the hydrogen-evolution mechanism, design the catalyst and select the photosensitizer.

Fig. 1. Structure diagrams of complexes 1～4 and FL

2 EXPERIMENTAL AND THEORETICAL SECTION

2. 1 Experimental

2. 1. 1 Preparation of samples

Cobalt(II) acetate tetrahydrate (C4H6CoO4·4H2O),salicylaldehyde (C7H6O2), ethylenediamine (N2H4),1,2-diaminocyclohexane (C6H14N2), 2,3-diaminopyridine(C5H7N3),o-phenylenediamine (C6H8N2), ethanol,dichloromethane (CH2Cl2), acetic acid, acetonitrile (MeCN),fluorescein (FL), and triethylamine (TEOA) were purchased from Macklin Reagent (Shanghai, China) and used without any further purification.

2. 1. 2 General procedure for the synthesisof Schiff base ligands

The ligands were synthesized according to reported literatures modified slightly[29-34]. A solution of salicylaldehyde (4 mmol) and methanol (15 mL) was added dropwise to a solution of ethylenediamine (2 mmol) and methanol (10 mL). The obtained mixture was refluxed for 2 h,which caused a precipitate to form. Then, the precipitate was separated by filtration, washed with cold methanol and dried in vacuum. Finally, it was recrystallized from dichloromethane to yield as a pure Schiff base ligand.

(R,R)-N,N′-bis-salicylidene-1,2-cyclohexanediamine Golden yellow solid, yield 0.264 g, 82%; m.p. 115～117 ℃.1H NMR (600 MHz,d6-CDCl3):δ13.34 (s, 2H), 8.28 (s, 2H),7.26 (t,J= 7.8 Hz, 2H), 7.17 (dd,J= 7.9, 1.8 Hz, 2H), 6.91(d,J= 8.3 Hz, 2H), 6.81 (t,J= 7.4 Hz, 2H), 3.36～3.31 (m,2H), 1.99～1.94 (m, 2H), 1.92～1.87 (m, 2H), 1.78～1.72 (m,2H), 1.52～1.48 (m, 2H).13C NMR (600 MHz,d6-CDCl3):δ164.77, 164.21, 161.31, 161.03, 132.30, 132.22, 131.55,131.52, 118.90, 118.72, 118.65, 118.57, 117.07, 116.82,72.67, 69.50, 33.15, 30.75, 24.23, 22.55. HRMS (ESI): calcd.for [C20H18N2O2+H]+323.1754, found 323.1756.

(R,R)-N,N′-bis-salicylidene-o-phenylenediamine Yellow solid, yield 0.269 g, 85%; m.p. 165～166 ℃.1H NMR (600 MHz,d6-DMSO):δ12.97 (s, 2H), 8.92 (s, 2H),7.66 (d,J= 7.7 Hz, 2H), 7.50～7.36 (m, 6H), 6.96 (d,J= 7.9 Hz, 4H).13C NMR (600 MHz,d6-DMSO)δ164.05, 160.42,142.27, 133.47, 132.50, 127.84, 119.76, 119.50, 119.11,116.70. HRMS (ESI): calcd. for [C20H18N2O2+H]+317.1285,found 317.1286.

(R,R)-N,N′-bis-salicylidene-2,3-diaminopyridine

Yellow solid, yield 0.257 g, 80%;1H NMR (600 MHz,d

6-DMSO):δ12.88 (d,J= 310.0 Hz, 2H), 9.56 (s, 1H), 8.98(s, 1H), 8.49～8.41 (m, 1H), 7.92 (dd,J= 7.8, 1.5 Hz, 1H),7.82～7.78 (m, 1H), 7.71 (dd,J= 7.6, 1.7 Hz, 1H), 7.51～7.43 (m, 3H), 7.04～6.96 (m, 4H).13C NMR (600 MHz,d6-DMSO):δ165.52, 164.15, 161.12, 160.29, 150.57, 146.57,139.07, 134.21, 133.81, 133.15, 132.22, 128.98, 123.92,119.55, 119.23, 116.78, 116.73. HRMS (ESI): calcd. for[C19H16N3O2+H]+3 18.1 23 7, fo und 3 18.12 68.

2. 1. 3 General procedure for the synthesis of metal complexes

Co(II)-salen complexes were synthesized from salen ligands and cobalt(II) acetate in ethanol according to the procedure described[35-37]. Co(III)-salen complexes were then purified according to the reported literatures[37,38]. That is,0.45 mL acetic acid was added to the Co(II)-salen complex(0.20 g) in dichloromethane (12 mL). After stirring the mixture for 50 min, the solvent of dichloromethane was removed with a rotavapor and the excess acetic acid was also removed under vacuum[39-44]. Since the complexes were too insoluble to allow solution NMR and polarimetry studies,their infrared spectrum data were measured.

[N,N′-bis(salicylidene)-trans-1,2-cyclohexanediamine]cobalt(II). Red solid, yield 62%, m.p. 300～302 ℃; IR (KBr,cm-1)ν(C=N) 1603,νs(Co=O) 627,ν(Co=N) 415; HRMS-ESI m/z[Co]+calcd. for C20H20CoN2O2: 379.0857, found: 379.0851.

Fig. 2. Infrared spectra of complexes 1～4

[N,N′-bis(salicylidene)-o-phenylenediamine] cobalt(II)Brown solid, yield 64%, m.p. ＞ 300 ℃; IR (KBr, cm-1)ν(C=N) 1610; HRMS-ESI m/z[Co]+calcd. for C20H14CoN2O2:373.0387, found: 373.0383.

[N,N′-bis(salicylidene)-2,3-diaminopyridyl] cobalt(II)Brown solid, yield 58%, m.p. ＞ 300 ℃; IR (KBr, cm-1)ν(C=N) 1615,ν(Co=N) 550,ν(Co=O) 468; HRMS-ESI m/z[Co]+calcd for C20H14CoN2O2: 374.0340, found:374.0346.

2. 1. 4 FT-IR analysis

The peak around 3030 cm-1was assumed to be the C-H stretching vibration on the phenyl group (Fig. 2). There is no peak around 2560 cm-1. It is believed that the ligand and the central metal are coordinated together[45]. The four complexes have strong vibration peaks around 1290 cm-1, which is caused by the stretching vibration of the C-O bond. Similar ligands show peaks[46-48]in the range of 1280～1340 cm-1.The stretching vibration of C=N bond has moved to a lower frequency[49,50]. Since the C=N is coordinated with the central metal Co,Vas(Co-O) andVs(Co-O) can be detected in all complexes.Vas(Co-O) appears around 750 cm-1, andVs(Co-O) peaks around 650 cm-1. The peaks at 1710 cm-1are assigned to the carbonyl peak of carboxylate. It can be inferred that the last carboxyl group has been formed by the product of the previous step. The peak at 1540 cm-1is from the phenyl breathing mode, and that at 415 cm-1is assigned to the Co-N stretching vibration[50,51].

2. 1. 5 Characterizations

Proton NMR spectra were recorded on a Bruker DRX 600 spectrophotometer in deuterated reagent by the residual solvent signal as an internal calibration. High resolution mass spectroscopic (HRMS) data of the products were collected on an Agilent Technologies 6540 UHD Accurate-Mass Q-TOF LC/MS using ESI. Hitachi UV-3600 UV-vis spectra were used to measure the UV-vis spectra with a 1 cm quartz cell.Photoluminescence (PL) emission spectra were measured on a fluorescence spectrometer (JASCO FP-6500) at 20 ℃.FT-IR spectra were recorded on a NEXUS470 spectrophotometer, and KBr was used as the substrate. Cyclic voltammetry (CV) was measured in an electrochemical workstation equipped with a Pt wire as the counter electrode, a glassy carbon as the working electrode, and a Ag/AgCl as the reference electrode at a scan rate of 100 mV/s in 0.1 M Bu4NPF6DMF solution. The pH value was determined by a PHS-25 pH meter.

2. 1. 6 Photocatalytic evaluation

Photocatalytic HER was tested in a commercial reaction cell (Lab Solar II, Perfect Light Co.,T= 10 ℃). The generated H2was measured by an online gas chromatograph(GC7900, TCD as detector, Ar as carrier gas). A Xe lamp equipped with a cut-off filter (PLS-SEX 300,λ≥ 400 nm,300 W) was used as the light source. The visible light intensity was about 100 mW/cm2measured by a PL-MW2000 spectroradiometer. In a typical run, a threecomponent system consisting of 10 mg cobalt-salen molecules (as photocatalyst), a certain amount of FL (as a photosensitizer), and 10 mL TEOA (as sacrificial reagent)was subjected to H2evolution experiments in 90 mL H2O without another organic reagent added. The pH of the system was adjusted by the addition of HCl and NaOH solutions.The mixed system was stirred in the dark for 30 min before turning on the lamp. During the HER test, the evolved H2was measured every 30 min by online GC. The apparent quantum efficiency (AQE) of H2evolution was measured with a band pass filter (λ= 400 nm) by equation (1). The irradiation area was about 3.14 cm2and the intensity of the irradiation was about 18.8 mW/cm2.

2. 2 Computational method

2. 2. 1 Geometric optimizations

Our previous research[52]shows that the solvent effect is very important for the calculation of cobalt complexes. Hence,the ground-state complexes 1～4 were subjected to full geometry optimization at the B3LYP/6-31G(d) level in aqueous solution by the CPCM model[53,54]. Optimized minima were verified with vibrational calculations. To study the interaction of Co complexes with FL, the FL was also optimized in aqueous solution by the CPCM model at the same level of theory, and its structural diagram is also shown in Fig. 1.

2. 2. 2 Molecular binding and optimization

The optimized complexes were bound to FL by the Dock6.0 program[55,56]. The box size, grid space, energy cut of distance, and maximum orientation were set as 30, 0.3,9999, and 150,000, respectively. Other binding parameters were set to default values, and thus we obtained binding geometries of the complexes with FL, which were further optimized in aqueous solution by the same method. To study the excited properties of the binding geometries, the lowest singlet states of binding geometries were optimized in aqueous solution at the same level of theory. Meanwhile, the optimized minima were verified with the vibrational calculations.

2. 2. 3 Calculation of electron-transfer rate constants

The non-adiabatic electron-transfer (ET) process is usually tackled by the theory of Marcus semi-classical model[57,58].Hence, we can obtain the rate constant by the following equation:

Here,λis the reorganization energy,hthe Planck’s constant,HADthe ET matrix element,Tthe temperature, andΔG°the normal Gibbs free energy change of the ET reaction.

2. 2. 4 Calculation of the reorganization energy λ

Cobalt complexes binding to FL can be excited by the light irradiation, leading to ET from FL to complex at a moving speed ofket. Take Co-complex and FL as an example, this process can be expressed as follows:

After electron transfer,λcan be computed by the following equation.

Here,λ1is the reorganization energy of FL andλ2is the reorganization energy of cobalt complex.

To obtain the accurate reorganization energy, we replaced the Co-complex and FL in reaction (3) with the complex and FL in the optimized binding-geometry in the ground state (S0),respectively. Similarly, we replaced the Co-complex-and FL+in reaction (3) with the complex and FL in the optimized binding-geometry in the lowest singlet state (S1), respectively.We first obtained the total energy ofE(FL) at its geometry in the optimized binding-geometry inS0, and the total energy ofE(FL+) at its geometry in the optimized binding-geometry in theS1. Besides, we also calculated the energyE′(FL) at the geometry of FL+and the energyE′(FL+) at the geometry of FL. Thus, we obtainedλ1by

The total reorganization energy was calculatedνiaEqs. 4, 5 and 6.

2. 2. 5 Calculations of HDA

The ET matrix elementHDAis usually estimated by the generalized Mulliken-Hush (GMH) theory, which can be calculated by the following equation[59]:where ΔEijis the energy gap between the ground state and the excited one, Δuijis the dipole moment difference between statesiandj, andmijdenotes the transition dipole moment connecting the two states, which were the optimized binding-geometries inS0andS1.

2. 2. 6 Calculations of ΔG◦

The binding-geometry of cobalt complex with FL can be excited by the light irradiation. The process can be expressed as follows:

Where Co-complex-FL and ′Co-complex-FL indicate the binding-geometries inS0andS1, respectively. The standard Gibbs free energy change of the ET reaction (8) was calculated by:

WhereGCo-complex-FLandG′Co-complex-FLare the energies of the optimized binding-geometries inS0andS1, respectively.Meanwhile, the correction to Gibbs free energy was added toGandG′.

The calculations above were performed with Gaussian16 program-package[60].

3 RESULTS AND DISCUSSION

3. 1 Optical properties

The UV-vis absorption spectra of complexes 1～4 were measured at room temperature in MeCN:H2O = 1:1 (see Fig.3). All complexes exhibited a strong-energy peak at about 260 nm and a poor light absorption in the visible region.Interestingly, complexes 3 and 4 have weaker absorption peaks near 325 and 330 nm, respectively, whereas 1 and 2 have not. Besides, the FL demonstrated a strong visible light absorption observed at about 480 nm, which ensured the possibility to trigger the visible-light-induced water splitting.

Fig. 3. Absorption spectra of complexes 1～4 and FL at room temperature in MeCN:H2O = 1:1

3. 2 PL detection

The PL spectra of complexes 1～4 were measured and shown in Fig. 4. Generally speaking, weaker fluorescence signals show weaker photogenerated carrier recombination[61,62].According to the viewpoint, complex 3 should have strong ability to promote the separation of photogenerated charge carriers relative to complexes 1 and 2. This suggests that the recombination of photogenerated charge carriers can be delayed effectively for 3, resulting in a lot of H2generated.Although the fluorescence performance of 4 was weaker than that of 3 in Fig. 4, the hydrogen-evolution amount of 4 was less than that of 3. The reason may be that the N atom in the pyridine ring on the equatorial plane inhibits the separation of photogenerated charge carriers and the performance of hydrogen production. In general, the peak shapes of PL spectra for the four complexes were similar, and the photocatalytic performances of hydrogen production were more or less the same (see the following ‘Photocatalytic H2production’ section). This indicates that modifying different groups for Schiff base complexes have a little impact on the hydrogen-production performance[63,64].

Fig. 4. PL spectra of complexes 1～4 excited by 340 nm light at room temperature

3. 3 Photocatalytic H2 production

The photocatalytic performance of complexes for hydrogen production was evaluated. The time-course of H2amount is shown in Fig. 5. It indicates that, after irradiation for 180 min, about 24.07, 48.24, 152.3, and 63.73μmol of hydrogen are released over complexes 1, 2, 3, and 4,respectively. Obviously, the catalytic hydrogen-production performance of complex 3 is significantly better than that of the other complexes in aqueous solution. It also indicates that the hydrogen-production performance of complex 3 in aqueous solution is better than that of complex 1. In addition,the hydrogen-production performance of complex 1 was also tested in the mixed solution of ethanol and water as a try,showing that the H2amount in the mixed solution is less than that in aqueous solution. Such a result gives us a little insight,i.e., the hydrogen-production performance of Co complexes can be improved by substituting the aqueous solution for the organic solvent. Meanwhile, this practice can effectively reduce the environmental pollution and the cost.

Fig. 5. Comparison of photocatalytic performance of H2 evolution for complexes 1～4. Reaction conditions: 25 mg photocatalyst,12 mg FL, 90 mL water, and 10 mL TEOA; 300 W xenon lamp (λ ≥ 400 nm); pH = 7

The photosensitizer concentration and the pH of medium affect the HER performance, so they were optimized with complex 3 as the catalyst. As shown in Fig. 6, when the FL amount is less than 12 mg, therH2of complex 3 is almost proportional to the FL amount. Further increasing the amount of FL to 14 mg can rapidly reducerH2from a maximum of 152.3 to 92.30 μmol, indicating that the ratio of PS to catalyst is a key factor to influence the evolution of hydrogen in the system, which may be due to the saturation of electrons provided by FL when the concentration was too high. The presence of sufficient FL greatly increases the chances of the catalyst obtaining electrons, thereby improving the catalytic efficiency of the system. However, excessive FL is detrimental to its dispersion and the absorption of visible light, thus decreasing the H2amount. An analogous phenomenon was reported in the literature[65].

Fig. 6. Effect of FL content on hydrogen generation. Reaction conditions: 25 mg complex 3, different contents of PS,90 mL water, and 10 mL TEOA; 300 W xenon lamp (λ ≥ 400 nm); pH = 7

Moreover, the pH can significantly affect the interaction among photosensitizer, catalyst and TEOA, and further affects the evolution of H2in the photocatalytic system[66].Fig. 7 illustrates the variation of H2amount with the initial pH value in the range from 5 to 13. The maximal amount of H2(ca. 152.3 μmol) can be obtained at pH = 7, whereas the hydrogen production drops sharply when the pH of the system is greater or less than 7. Fig. 7 also shows that the amount of hydrogen was only 23.52μmol after three hours of light at pH = 13. When pH = 5, the detected amount of hydrogen is 31.42μmol. When the alkalinity is too high, the proton concentration in the system is significantly reduced,making difficult protonation for the intermediates, which affects proton reduction and the evolution of H2. Strongly acidic solution can cause TEOA protons, resulting in a poor supply capacity. This may be the main reason of the decrease in hydrogen generation under lower pH.

Fig. 7. Effect of the pH value of the medium on hydrogen generation. Reaction conditions: 25 mg complex 3,12 mg FL, 90 mL water, and 10 mL TEOA; 300 W xenon lamp (λ ≥ 400 nm)

3. 4 Cyclic voltammetry

The cyclic voltammograms of complexes were tested in a 0.1 M Bu4NPF6DMF solution. The redox potentials were obtained through the Ag/AgCl pair, and were adjusted to a standard hydrogen electrode. As shown in Fig. 8a, the reversible redox potentials are at -0.625, -0.602, -0.661 and-0.670 V (νsNHE), corresponding to the CoIII/CoIIredox couples of complexes 1～4. Moreover, complexes 1～4 display redox potentials at -0.785, -0.762, -0.735 and-0.754 V (vs NHE), respectively, tentatively assigned as the CoII/CoIcouple. As shown in Fig. 8b, after the addition of FL,there is a significant cathodic shift for each catalyst. The CoIII/CoIIredox couples of complexes 1～4 change into-0.954, -0.967, -0.862 and -0.811 V. Redox potentials of complexes 1～4 are at -1.130, -1.223, -1.055 and -0.912 V,respectively, corresponding to the redox pairs of CoII/CoI.This is mainly attributed to the electron donation of FL. Such a result indicates that the main influence on the catalytic performance of complexes is not the potential, but effective ET and photogenerated carrier recombination.

Fig. 8. CVs of (a) 1.0 mM complexes and (b) the existence of 1.0 mM FL measured in 0.1 M Bu4NPF6 DMF solution.The scanning rate was 100 mV/s. Electrode setup: glassy carbon working electrode,Pt wire counter electrode, Ag/AgCl reference electrode

3. 5 Role of electronic sacrificial agents

The essence of photocatalytic hydrogen production is that electrons in the photosensitizer are excited creating photogenerated electrons, which are quickly transferred to the active site of the catalyst. The rapid ET makes the concentration of electrons on the active site of the catalyst expand rapidly, and the hydrogen protons in the solution get these electrons easily, leading to hydrogen gas generated.Research finds that many catalysts usually cause the photo-generated electrons to recombine due to some defects.Fortunately, the electronic sacrificial agent can suppress the recombination, improve the quantum conversion efficiency and increase the hydrogen-evolution amount. As shown in Fig. 9, under optimal conditions, the hydrogen production results of complex 3 were studied with or without electron sacrificial agent. We can see that, after 3 h of irradiation, the hydrogen-evolution amount with the electron sacrificial agent was 38 times that without the electron sacrificial agent.

Fig. 9. Effect of electron sacrificial agent on hydrogen evolution. Reaction conditions: 25 mg complex 3, 12 mg FL,90 mL water, and additions of 10 mL TEOA or nothing; 300 W xenon lamp (λ ≥ 400 nm); pH = 7

3. 6 Stability of the photocatalytic system

The stability of the catalytic system is critical. In Fig. 10,the hydrogen evolution rate of the reaction system is decreased with irradiation time maybe due to the decomposition of the catalyst or photosensitizer[67-69]. To investigate the specific reasons for the inactivation of the system, 6 mg of photosensitizer was added to the photocatalytic system after 6 hours of irradiation. The results show that the hydrogen-production performance was significantly improved than before. Therefore, the loss of hydrogen evolution activity in the reaction system is mainly attributed to the decomposition and consumption of the photosensitizer.

Fig. 10. Time courses of photocatalytic H2 evolution with a supplement of complex 3 and FL during reaction. Reaction conditions: 25 mg complex 3, 12 mg FL, 90 mL water, and 10 mL TEOA; 300 W xenon lamp (λ ≥ 400 nm); pH = 7

3. 7 Main hydrogen bond between complex and FL

The optimized results show that complexes bind to FL mainly by the hydrogen bond OH between one oxygen atom on the carbonyl group of complexes and one hydrogen atom on the benzene ring of FL. Take complex 3 and FL as an example, the optimized binding geometry in theS0is shown in Fig. 11. The calculated lengths of OH bond between complexes 1～4 and FL are 0.2501, 0.1686, 0.2428 and 0.2433 nm in theS0, and 0.2442, 0.1645, 0.1657 and 0.2237 nm in theS1, respectively. Corresponding gaps are 0.0059,0.0041, 0.0771 and 0.0196 nm, respectively. Such a result shows that the lengths of OH bond inS1shorten significantly relative toS0. It is advantageous for ET from FL to complexes by the OH bond. The result also shows that the OH bond length between complex 3 and FL is shortened greatly when electrons are excited. This implies that the excited electrons can easily transfer from FL to complex 3 by the hydrogen bond, and thus 3 may obtain the most electrons from FL. Based on the idea, complex 3 should have strong hydrogen-evolution abilities relative to complexes 1, 2 and 4.Instead, the lengths of OH bond between complexes 1, 2 and FL shorten a little when electrons are excited, showing that the excited electrons transfer from FL to complexes 1 and 2 difficultly by the hydrogen bonds. Therefore, 1 and 2 should release hydrogen difficultly, in accordance with the above experimental result.

Fig. 11. Optimized binding geometry of complex 3 with FL in the S0 and the main hydrogen bond OH

3. 8 Computed ET rate constants

Table 1 gives the computedλ,HDAandket. We can see that complex 3 shows the greatestket. This indicates that the speed of ET between complex 3 and FL is much faster than those between the other complexes and FL. Such a process may lead to a great number of electrons staying on 3 by the light irradiation. Large quantities of H+in the aqueous solution around complex 3 can then capture the electrons, easily causing H2released. We predict that the H2amount by 3 should be the most. Meanwhile, we predict that the order of H2amount is 3 ＞ 4 ＞ 2 ＞ 1, in accordance with our experimental result.

Table 1. Calculated λ1 (Complex), λ2 (FL), λtot, HDA and ket for Complexes 1～4

3. 9 Natural charge populations of complexes

Complexes 1～4 with FL can be excited by the light irradiation. To explore the change of charges on the complexes, the natural charge populations of complexes inS0andS1were computed and shown in Table 2. The net charges on complexes 1～4 in the binding geometries are 0.0059 |e|,0.0687 |e|, 0.0009 |e| and 0.0176 |e| in theS0, and 0.0071 |e|,0.0792 |e|, 0.0795 |e| and 0.0297 |e| in theS1, respectively.Their gaps (ε) are 0.0012 |e|, 0.0105 |e|, 0.0786 |e| and 0.0121|e|, respectively. This shows that complex 3 gained more electrons from FL than other complexes under light irradiation. Hence, large quantities of H+around complex 3 can capture electrons easily, resulting in more H2released relative to complexes 1, 2 and 4. Based on the charge viewpoint, we predict that the order of H2amount should be 3 ＞ 4 ＞ 2 ＞ 1, in accordance with the above experimental result.

Table 2. Natural Charge Populations (in |e|) of Complexes 1～4 in the Binding Geometries in the S0 and S1 and Gaps (ε)

3. 10 Mechanism discussion

Based on the described experimental and theoretical results above, the mechanism for light-driven H2production with complex 3 is proposed in Scheme 1. By the visible light, the excited FL created photogenerated electrons, which can transfer from FL to complex 3 rapidly by the hydrogen bond between them. The hydrogen-bond shortening in the excited state helps to ET. Meanwhile, the rapid ET is also favorable for a great number of electrons to stay on complex 3. Hence,large quantities of H+in the solution around 3 can capture the electrons easily, resulting in more H2released. This may be the primary reason for complex 3 with better hydrogenevolution properties. Therefore, the ET velocity and the number of ET are the key factors to the evolution of H2.

Scheme 1. Proposed mechanism for H2 evolution in the photocatalytic system

4 CONCLUSION

As photocatalytic hydrogen-evolution catalysts, cobalt(Ⅲ)complexes 1～4 with Schiff base were synthesized and a hydrogen-evolution system composed of Co(Ⅲ) complexes in aqueous solution, together with FL and TEOA as photosensitizer and sacrificial donor was formed. For complex 3, a maximum H2amount of 152.3 μmol was recorded after 3 h of visible light irradiation under optimal conditions. Using DFT theory, the reason of releasing hydrogen was explained by the calculated ET rate constants,the number of ET and the change of lengths for hydrogen bonds. Besides, the mechanism for light-driven H2production was explored by experimental and theoretical methods. This work provides a theoretical method to discover excellent catalysts, saving a lot of time and energy.