Accepted Manuscript Cobalt(III) tetraaza-macrocyclic complexes as efficient catalyst for photoinduced hydrogen production in water: theoretical investigation of the electronic structure of the reduced species and mechanistic insight Robin Gueret, Carmen E. Castillo, Mateusz Rebarz, Fabrice Thomas, AaronAlbert Hargrove, Jacques Pécaut, Michel Sliwa, Jérôme Fortage, Marie-Noëlle Collomb PII: DOI: Reference:

S1011-1344(15)00133-5 http://dx.doi.org/10.1016/j.jphotobiol.2015.04.010 JPB 10007

To appear in:

Journal of Photochemistry and Photobiology B: Biology

Received Date: Accepted Date:

14 January 2015 20 April 2015

Please cite this article as: R. Gueret, C.E. Castillo, M. Rebarz, F. Thomas, A-A. Hargrove, J. Pécaut, M. Sliwa, J. Fortage, M-N. Collomb, Cobalt(III) tetraaza-macrocyclic complexes as efficient catalyst for photoinduced hydrogen production in water: theoretical investigation of the electronic structure of the reduced species and mechanistic insight, Journal of Photochemistry and Photobiology B: Biology (2015), doi: http://dx.doi.org/10.1016/j.jphotobiol. 2015.04.010

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Cobalt(III) tetraaza-macrocyclic complexes as efficient catalyst for photoinduced hydrogen production in water: theoretical investigation of the electronic structure of the reduced species and mechanistic insight Robin Gueret,a Carmen E. Castillo,a Mateusz Rebarz,b Fabrice Thomas,a Aaron-Albert Hargrove,a Jacques Pécaut,c Michel Sliwa,b Jérôme Fortage,*a and Marie-Noëlle Collomb,*a a

Univ. Grenoble Alpes, DCM, F-38000 Grenoble, France CNRS, DCM, F-38000 Grenoble, France b

Laboratoire de Spectrochimie Infrarouge et Raman, UMR 8516 CNRS-Université Lille 1 Sciences et Technologies, 59655 Villeneuve d’Ascq Cedex, France c

Univ. Grenoble Alpes, INAC-SCIB, F-38000 Grenoble, France CEA, INAC-SCIB, Reconnaissance Ionique et Chimie de Coordination, F-38000 Grenoble, France *Corresponding authors. Tel. : +433 76 51 44 18 E-mail adresses : [email protected] (J. Fortage), [email protected] (M.-N. Collomb) Invited contribution to the Special Issue of J Photochem Photobiol B: Biology on Artificial Photosynthesis

Abstract We recently reported a very efficient homogeneous system for visible-light driven hydrogen production in water based on the cobalt(III) tetraaza-macrocyclic complex [Co(CR)Cl2]+ (1) (CR

=

2,12-dimethyl-3,7,11,17-tetra-azabicyclo(11.3.1)-heptadeca-1(17),2,11,13,15-

pentaene) as a noble metal-free catalyst, with [Ru II(bpy)3]2+ (Ru) as photosensitizer and ascorbate/ascorbic

acid

(HA-/H2A)

as

a

sacrificial

electron

donor

and

buffer

(PhysChemChemPhys 2013, 15, 17544). This catalyst presents the particularity to achieve very high turnover numbers (TONs) (up to 1000) at pH 4.0 at a relative high concentration (0.1 mM) generating a large amount of hydrogen and having a long term stability. A similar activity was observed for the aquo derivative [Co III(CR)(H2O)2]3+ (2) due to substitution of chloro ligands by water molecule in water. In this work, the geometry and electronic structures of 2 and its analog [ZnII(CR)Cl]+ (3) derivative containing the redox innocent Zn(II) metal ion have been investigated by DFT calculations under various oxidation states. We also further studied the photocatalytic activity of this system and evaluated the influence of varying the relative concentration of the different components on the H2-evolving activity. 1

Turnover numbers versus catalyst (TONCat) were found to be dependent on the catalyst concentration with the highest value of 1130 obtained at 0.05 mM. Interestingly, the analogous nickel derivative, [NiII(CR)Cl2] (4), when tested under the same experimental conditions was found to be fully inactive for H2 production. Nanosecond transient absorption spectroscopy measurements have revealed that the first electron-transfer steps of the photocatalytic H2-evolution mechanism with the Ru/cobalt tetraaza/HA-/H2A system involve a reductive quenching of the excited state of the photosensitizer by ascorbate (kq = 2.5 x 107 M-1 s-1) followed by an electron transfer of the reduced photosensitizer to the catalyst (ket = 1.4 x 10 9 M-1 s-1). The reduced catalyst can then enter into the cycle of hydrogen evolution.

Keywords: Hydrogen, Photocatalysis, Cobalt, Macrocyclic ligand, Water Chemistry.

Highlights: •

Efficient photoinduced H2 production in fully aqueous solution using cobalt tetraazamacrocyclic complexes harboring the redox non innocent pyridyldiimine moiety

•

Turnover numbers versus catalyst were found to be dependent on the catalyst concentration

•

Analogous zinc and nickel derivatives, tested under the same experimental conditions are fully inactive for H2 production.

•

The key initial steps of the photochemical cycle leading to the reduction of the cobalt catalyst were fully identified

•

DFT calculations show that both closed-shell and open-shell configurations of the cobalt(I) species, a key species for hydrogen evolution, can co-exist in solution

2

1. Introduction Hydrogen (H2) can be considered as a promising clean energy vector for the future that could represent a good substitute for fossil fuels.1 However, one of the main issues with H2 remains its production by sustainable ways. In this line, the production of H2 from water dissociation using sun light, also referred as artificial photosynthesis, has emerged as a very attractive approach.2-5 The two half reactions of water-splitting, the oxidation of water to O2 and the reduction of protons into H2, are often independently studied. The development of homogeneous photocatalytic systems using molecular compounds for protons reduction, the reductive part of the reaction, has experienced considerable interest over the last fifteen years.6-9 Numerous homogeneous systems have been reported combining a catalyst (Cat) based on rare or more earth abundant metals, a photosensitizer (PS) based on metallic complexes (Ru, Ir, Re, Os), porphyrins or organic dyes to a sacrificial electron donor (SD).1020

In some cases, the catalyst is chemically linked to the photosensitizer through a bridging

ligand.10-18,20-31 Some of these molecular homogeneous photocatalytic systems can operate very efficiently (in terms of turnover number, TON) in organic or mixed aqueous-organic solvents. Those reaching a turnover number versus catalyst above 100 in fully aqueous solution, an important conditions for further applications in photo-electrochemical watersplitting devices, were rare and restricted to rhodium32-36 and platinum37 based catalysts until very recently. Indeed, since 2012, several examples with catalysts based on more earth abundant metal as cobalt,38-51 iron52-55 and nickel56 were reported, demonstrating that such complexes can be also efficient catalysts in water.

Scheme 1. Structures of H2-evolving catalysts investigated in this study.

3

Among the cobalt systems active in water,38-51,57-59 we reported that a Co III complex of a tetraazamacrocyclic

ligand,

[CoIII(CR)Cl2]+

(CR

=

2,12-dimethyl-3,7,11,17-tetra-

azabicyclo(11.3.1)-heptadeca-1(17),2,11,13,15-pentaene) (1) (Scheme 1) is a very efficient H2-evolving catalyst when associated with [Ru(bpy)3 ]2+ (Ru)42 or CdTe35,60 quantum dots as photosensitizer, and ascorbate (HA-) as electron donor. Such [CoIII(CR)(X)2]n+ (X = halide, H2O or CH3CN) complexes were also investigated as electro- and photocatalysts for hydrogen production and for CO2 reduction in the eighties61,62 and more recently by the groups of Lau 63 and Peters.64,65 We have shown, in our photocatalytic studies, that under visible irradiation, the Ru/1 system gives up to 1000 turnovers at pH 4.0 versus the catalyst with a relatively low photosensitizer/catalyst ratio (10/1) and a high concentration of catalyst (1x10 -4 M), thus producing a significant amount of H2 (~ 12 mL for 5 mL of solution). This photocatalytic system also exhibits a remarkable long-term stability that exceeds 20 hours under our experimental conditions. The initial cobalt species involved in the catalysis is in fact [Co II(CR)(H2O)n]2+ (n = 1 or 2) generated in situ through a one-electron reduction of 1 by HAand exchange of chloride ligands with the solvent. A similar photocatalytic H2-evolving activity was thus obtained using the aquo derivative, [CoIII(CR)(H2O)2]3+ (2) (Scheme 1). We also demonstrated through comparative studies35 that 1 (or 2) at 1 x 10-4 M Cat, is about four time more active than [RhIII(dmbpy)2Cl2 ]+, previously reported to behave as the most efficient rhodium-based catalyst in purely aqueous solution.32,34,35 In addition, the cobalt diiminedioxime

[Co III{(DO)(DOH)pnBr2]

({(DO)(DOH)pn}

=

N2,N2’-propanediylbis(2,3-

butanedione 2-imine 3-oxime)), the most efficient cobaloxime derivative in water,38,66 produces more than one hundred time less hydrogen than 1. The electrochemical properties of [Co III(CR)(X)2]n+ in organic solvent are characterized by three reversible one-electron processes.42,63-65 If the first one is clearly a metal centered process Co III/II, the two following reversible reduction waves, formally assigned to “Co II/I” and “CoI/0” redox processes, can be either metal or ligand centered reduction since the redoxactive pyridyldiimine moiety of the ligand can be potentially reduced twice.67 We succeeded to electrogenerate and spectroscopically characterize in CH3CN the formally low-valent “CoI” form of 1 (or 2).42 This species was very recently isolated and crystallographically characterized by the group of Peters65 as [Co(CR)(CH3CN)]+, from a chemical reduction of [Co III(CR)(Br)2]Br in organic media. This experimental study supplemented by DFT calculations,65 suggests a description as low-spin Co II ions antiferromagnetically coupled to a ligand radical-anion (CR•-) at the solid state. A similar square planar geometry with a CH3CN molecule as axial ligand was found for the X-Ray structural characterization of the CoII 4

derivative.65 The high stability of the doubly reduced form, formally “Co I” can account for the high efficiency of the photocatalytic system with such cobalt(III) tetraazamacrocyclic catalysts. Indeed, the catalytic activity for protons reduction is triggered when the cobalt center is reduced to “Co I” or even at a further reduced state;42,64 such species are believed to be then protonated to generate hydride intermediates as key species for H2 evolution. Actually, electrochemical experiments in acid aqueous solutions42,63,64 have evidenced an intense catalytic current at a slightly more negative potential than the “CoII/I” couple, while the catalytic effect detected at the potential of the Co II/I wave is less intense. However, the interpretation of the electrochemical data for such Co tetraazamacrocyclic complexes in water is not trivial and from these studies it is not clear if the triply reduced ”Co 0” state is involved in electro- or photo-catalysis for hydrogen evolution in water since additional processes may occur.64 In this context, the aim of this article was to further investigate the properties of these cobalt complexes as they represent a very interesting class of H2-evolving catalysts in water. The electronic structure of the aquo derivative 2 under its initial and different reduced forms was evaluated by density functional theory (DFT). This theoretical study was complemented by calculations on the analog [ZnII(CR)Cl]+ (3) complex (Scheme 1), which harbors the redox innocent Zn(II) metal ion. For this new complex, which was crystallographycally characterized, the ligand-centered reduction process was unambiguously supported by a coupled electrochemical/spectroscopic study in CH3CN. In the present study we also further studied the H2-evolving activity of this class of cobalt catalysts in water in similar photocatalytic systems, i. e. in association with Ru and ascorbate, with the aim to evaluate the influence of varying the relative concentration of the different components on the H2-evolving activity. We have also explored the efficiency of the analogous nickel [NiII(CR)Cl2 ] (4) derivative in our photocatalytic conditions as [NiII(CR)]2+ was previously reported to act as an efficient electrocatalyst for proton reduction in acidic aqueous solution,68 and in parallel to our work, reinvestigated for photocatalytic H2-evolution in organic solvent.69 Finally, the key initial steps of the photocatalytic H2-evolution mechanism for the Ru/cobalt tetraaza/HA-/H2A system were identified from a photophysical study performed by nanosecond transient absorption spectroscopy.

5

2. Experimental 2.1. Materials and general. Acetonitrile (CH3CN, Fisher, HPLC grade), acetonitrile-d 3 (CD3CN, Euriso-top, 99.8 %D), ethanol (EtOH, Fisher, HPLC grade), tetra-n-butyl-ammonium perchlorate ([Bu 4N]ClO4, Fluka), bis(3-aminopropyl)amine (+99%, Aldrich), 2,6-diacetylpyridine (98%, Aldrich), CoCl2•6H2O (98%, Aldrich), NiCl2•6H2O (98 %, Aldrich), ZnCl2 (99 %, Aldrich), [Ru(bpy)3]Cl2 (Ru) (99%, Aldrich), L-ascorbic acid (H2A, 99%, Acros), sodium L-ascorbate (NaHA, 99%, Acros), and reference gas (1% and 5% H2 in N2, Air Liquide) were purchased from commercial suppliers. All reagents and solvents were used as received. Purification of water (15.0 MΩ.cm, 24°C) was performed with a milli-Q system (Purelab option, Elga). [Co III(CR)Cl2](ClO4)42 (1), [Co III(CR)(H2O)2](ClO4)342 (2), [NiII(CR)Cl2]70 (4) (CR = 2,12dimethyl-3,7,11,17-tetra-azabicyclo(11.3.1)-heptadeca-1(17),2,11,13,15-pentaene)

were

synthesized according to literature procedure. 1H NMR spectra were recorded on a Bruker 400 MHz spectrometer. Chemical shifts for 1H NMR spectra are referenced relative to residual protium in deuterated acetonitrile (CD3CN δ = 1.94 ppm). 1H NMR spectra were performed within the ICMG Chemistry Nanobio Platform, Grenoble.

2.2. Synthesis of [ZnII(CR)Cl](PF6) (3). 2,6-Diacetylpyridine (686 mg, 4.20 mmol) was dissolved in a 1:1 mixture of ethanol (60 mL) and water (60 mL) and stirred at 40°C until the complete dissolution of the starting material. The solution was stirred at 50°C for 10 minutes and an ethanol solution (40 mL) containing ZnCl2 (573 mg, 4.20 mmol) was added. Then bis(3-aminopropyl)amine (0.59 mL, 4.20 mmol) is added dropwise during 25 min. One drop of glacial acetic acid is added to the cloudy solution and the crude mixture is stirred at 70°C overnight under N2. After cooling to room temperature, the addition of one equivalent of NaPF6, has led to the formation of a white precipitate which was filtered off and washed with diethyl ether. Single crystals of [Zn(CR)Cl](PF6) suitable for X-Ray diffraction were obtained by slow diffusion of diisopropyl ether in an acetonitrile solution of the complex (2.25 g, 88.7 %). Anal. Calcd for C15H22ClF6N4PZn: C, 35.74; H, 4.40; N, 11.12. Found: C, 35.99; H, 4.31; N, 11.31.1H RMN (400 MHz, CD3CN) δ = 8.46 (t, J = 8.0 Hz, 1H), 8.25 (d, J = 8.0 Hz, 2H), 4.17 (dt, J = 13.2 and 3.2 Hz, 2H), 3.81 (t, J = 12.8 Hz, 2H), 3.27 (dd, J = 11.6 and 12.8 Hz, 2H), 3.09-3.01 (m, 3H), 2.512 (s, 3H), 2.508 (s, 3H), 2.21-2.14 (m,2H), 1.69-1.58 (m, 2H). IR (cm-1): 3449, 3240, 2939, 1649, 1586, 1437, 1380, 1361, 1273, 1261, 1243, 1203, 1079, 6

1057, 1038,905, 837, 795, 639, 558. ESI-MS: m/z (%) 356.9 (100) [M-PF6]+ (M = [ZnII(CR)Cl](PF6)), 161.1 (7) [M-Cl-PF6]2+.

2.3. X-ray structure determination. Single-crystal diffraction data were collected on an Oxford-Diffraction XCalibur S Kappa geometry diffractometer (MoKα radiation, graphite monochromator, λ 0.71073 Å). The CrysAlisPro program package (Agilent Technologies, Version 1.171.36.20) was used for data collection cell refinements and data reductions. An absorption correction (CrysAlisPro) was applied to the data. The molecular structure was solved by charge flipping method (superflip)71 and refined on F2 by full matrix least-squares techniques using SHELXTL package.72 For all complexes, all non-hydrogen atoms were refined anisotropically and hydrogen atoms were placed on ideal position and refined isotopically. CCDC 1060950 contain the supplementary crystallographic data for [Zn(CR)Cl](PF6).

2.4. Electrochemistry. The electrochemical measurements performed under argon atmosphere at room temperature. When CH3CN was used as solvent, the electrochemical experiments were performed in a dryglove box. Cyclic voltammetry experiments were performed using an EG&G model 173 potentiostat/galvanostat equipped with a PAR model universal programmer and a PAR model 179 digital coulometer. A standard three-electrode electrochemical cell was used. Potentials were referred to an Ag/0.01 M AgNO3 reference electrode in CH3CN + 0.1 M [Bu 4N]ClO4. Potentials referred to the Ag/AgNO3 system can be converted to the ferrocene/ferricinium couple by subtracting 87 mV, to the saturated calomel electrode (SCE) by adding 298 mV or to the normal hydrogen electrode (NHE) by adding 548 mV. The working electrode, polished with 2 µm diamond paste (Mecaprex Presi), was a carbon vitreous disk (3 mm in diameter) for cyclic voltammetry (Ep a, anodic peak potential; Epc, cathodic peak potential; E1/2 = (Epa + Epc)/2; ∆Ep = Epa - Ep c). Exhaustive electrolyse were carried out on a reticulated vitreous carbon electrode 60 PPI (the electrosynthesis Co. Inc.) three-dimensional meshes (1.8 cm x 0.7 cm x 0.4 cm). The auxiliary electrode was a Pt wire in CH3 CN + 0.1 M [Bu4N]ClO4. In all experiments the potential scan rate was 100 mVs-1. Progress of electrolysis was followed by the change in UV-Vis spectra with a MCS 501 UV-NIR (Carl Zeiss) spectrophotometer equipped with an automatic shutter. The light sources are halogen (CLH 500 20 W) and deuterium lamps (CLD 500) with optic fibers (041.002–UV SN 012105), 1 mm path-length cell.

7

2.5. DFT calculations. Theoretical calculations were performed with the ORCA program package.73 Geometry optimizations was performed by using the GGA functional BP8674-76 in combination with the TZVP 77 basis set for all atoms except the metal center (triply polarized core property basis set CP(PPP)). The RI approximation was used in combination with TZVP/J auxiliary basis set.78 with the appropriate Coulomb fitting sets.79 The integration grids were typically Grid4 in ORCA convention (increased grid (Grid5) was used in some instances), in combination with tight SCF convergence criteria. Solvent effects were accounted within the framework of the conductor like screening (COSMO) dielectric continuum approach (H2O for the Co(III) and Co(II) complexes, CH3CN for the reduced forms).80 The single-point energy calculations were carried out by using the B3LYP81 82 functional together with the TZV/P77 basis set. Optical properties were computed by using the hybrid functional B3LYP81,82 in combination with the TZV/P77 basis set. Electronic excitations were calculated by time-dependent DFT (TDDFT)83,84,85 by using the Tamm-Dancoff approximation.86,87 In order to optimize the computational cost the RIJCOSX approximation88 was used. 30 excited states were calculated in each case, with a maximal dimension of expansion space that is at least 8-times the number of roots.

2.6. General procedure for photocatalytic hydrogen generation. A homemade glass tubes with a diameter of 2 cm fused with a round bottom flask was charged with 5 mL of water, a magnetic stirrer (volume < 0.1 mL), the photosensitizer and the appropriate catalyst as defined for each set of experiments. The solution was stirred until the complete dissolution of the photosensitizer and the catalyst. The concentration of the photosensitizer was adjusted from UV-visible measurements. For the higher concentration in catalyst (1x10-4, 5x10-5 and 1x10-5 M), a glass tube of a total volume of 180 mL (head space volume = 175 mL) was used, while for experiments using the lower concentration of catalyst of 5x10 -6 M, a glass tube of 28.5 mL (head space volume of 23.5 mL) was used. Before irradiation experiments, the glass tube was covered with black foil to protect the solution from light. For an aqueous solution at pH 4.0, ascorbic acid (H2A) (484 mg, 0.55 M) and sodium ascorbate (NaHA) (544 mg, 0.55 M) were added to the solution to give a total concentration of 1.1 M. For solutions at pH 4.0 with a H2A/NaHA total concentration of 0.5 M, H2A (220 mg, 0.25 M) and NaHA (248 mg, 0.25 M) were added, and for 0.1 M, H2A (44 mg, 0.05 M) and NaHA (50 mg, 0.05 M) were added. The glass tube was sealed with a rubber septum and 8

the solution was degassed for about 45 minutes by nitrogen bubbling. Continuous irradiation was performed at 298 K under stirring with a xenon lamp (150 W, Hamamatsu L8253, type LC8-03) equipped with a 400–700 nm large band filter, which was placed 4 cm from the sample. The amount of hydrogen evolved was quantified from an analysis of the gas mixture in the headspace of the glass tube (sampling of 100 µL of gas) by gas chromatography (Perkin Elmer Autosystem XL Gas Chromatograph equipped with a 5 Å molecular sieve column (oven temperature = 303 K) and a thermal conductivity detector (TCD)), which uses argon as a carrier gas. Prior to each experiment, GC/TCD calibration was carried out by using two samples of the reference gas (1% and 5% H2 in N2).

2.7. Nanosecond transient absorption spectroscopy. Nanosecond transient absorption experiments were performed using a laser flash photolysis apparatus. Excitation pulses (460 nm, fwhm 4 ns, 1 mJ, 0.5 Hz) were provided by a 10-Hz Nd:YAG laser (Continuum Surelite II) coupled to an OPO (Continuum Panther EX OPO) and SH05 shutter (Thorlabs). The probe light was provided by a pulsed Xe lamp (XBO 150W/CR OFR, OSRAM). The transmitted light was dispersed by a monochromator (Horiba JobinYvon, iHR320) and analysed with a photomultiplier (R1477-06, Hamamatsu) coupled to a digital oscilloscope (LeCroy 454, 500 MHz). Synchronization of excitation pulses and acquisition time was secured with PCI-6602 8 Channel counter/timer (National Instruments). The experiment was controlled by a homemade software written in LabView environment. The recorded traces were averaged for several pulses and repeated for different wavelengths to reconstruct the spectra afterward. The deconvolutions of the individual decays with experimentally measured instrument response function (IRF) lead to 10 ns time resolution. Single wavelengths as well as global analyses of the transient absorption data were performed using Igor Pro 6.20. Samples were prepared in a glove box under argon atmosphere within a quartz cell (cell width = 10x10 mm2) in an aqueous solution at pH 4.0 (H2A 0.55 M and NaHA 0.55 M)) containing only [Ru(bpy)3]2+ (Ru) (1x10-4 M) or a mixture of Ru (1.3 x 10 -4 M) and 1 (2.4 x 10 -4 M).

3. Results and discussion 3.1. Synthesis and crystal structure of [ZnII(CR)Cl]+ (3). This new complex was synthesized in good yield as hexafluorophosphate salt by an adaptation of the literature procedure reported by Busch et al.89 for the synthesis of Zn(CR)I2. 9

The synthetic procedure for 3 is based on a condensation of 2,6-diacetylpyridine and bis(3aminopropyl)amine in presence of ZnCl2. The complex was precipitated by addition of NaPF6 salt. Single monocrystals of 3(PF6) were obtained by slow diffusion of diisopropylether into a concentrated CH3CN solution of the compound and the molecular structure has been determined by X-ray crystallography. The structure consists of two distinct [Zn(CR)Cl]+ cations, each associated with one PF6- as counter-anion. Selected bond distances and angles are listed in Tables S1-S2. Since both cations present similar geometry only one is discussed here. A view of the ORTEP of the cation is shown in Fig. 1. The Zn2+ is pentacoordinated to the four nitrogen atoms of the macrocyclic ligand and to a chloride ligand in axial position, in a distorted square pyramidal geometry. The zinc lies out of the plan of the macrocyclic ligand (0.62(4) Å), which is also distorted. The two Zn-Nimine bond lengths of 2.125(4) and 2.139(3) Å (Zn(1)-N(2), N(4), respectively) are longer than the Zn-Npyridine and Zn-Namine bond distances of 2.041(3) (Zn(1)-N(3)) and 2.049(4) Å (Zn(1)-N(1)), respectively. The Zn-Cl distance of 2.2418(12) Å is comparable to those of the Co-Cl bonds in axial positions (2.2439(10) and 2.2285(10) Å) found in the crystal structure of [Co III(CR)Cl2](ClO4)2 reported by Lau et al.63

3.2. Electrochemical properties of [ZnII(CR)Cl]+ (3) in CH3CN and comparison with the nickel and cobalt derivatives. The electrochemical properties of 3 have been studied in CH3CN. The cyclic voltammogram shows one reversible reduction wave at E1/2 = -1.58V (∆Ep = 64 mV) vs Ag/AgNO3 followed by a poorly reversible wave at E1/2 = -1.85V (∆Ep = 100 mV) (Fig. 2 and see Table S3 for the conversion of potentials vs SCE). The first reduction wave can be attributed to a ligandcentered process involving the reduction of one imine of the CR ligand (i.e. CR/CR•-). This assumption is strongly supported by UV-visible and EPR spectroscopy measurements (Figs. 3 and S1) in the course of an electrolysis of the solution at -1.58V (one electron consumed) performed at room temperature. Upon electrolysis, the initial colorless solution with an intense absorption band at 298 nm (shoulder at 307 nm) turns to pink with appearance of two transitions at 378 and 503 nm (Fig. 3(a)). The X-band EPR spectrum at 100 K of this solution exhibits a single isotropic signal centered at g = 1.99, characteristic of a radical species and thus consistent with a ZnII-stabilized ligand radical (Srad = ½) species (Fig. S1). Then after about 8 min the solution starts to turn to orange, and finally gradually fades after about 20 min when the electrolysis is almost finished. The color change to orange is characterized by a shift 10

of the band at 288 nm to 280 nm along with the appearance of a new band at 431 nm while the 378 and 503 nm absorption bands of the former [ZnII(CR•-)]+ disappear (Fig. 3(b)). The final evolution of the solution results in the substantial decrease of the 431 nm band while the band remains at 280 nm (Fig. 3(c)). At this stage the EPR signal has fully disappeared (Fig. S1). These changes underline the poor stability of the radical species over time and its high reactivity to form other diamagnetic species that were not further characterized. Accordingly, the resulting CV exhibits redox systems at E1/2 = -1.74 V (∆Ep = 100 mV) and at Epa = +0.69 V, in place of the initial ones (Fig. 2). The electrochemical properties of the cobalt and nickel derivatives 1, 2 and 4 are quite different to those of 3. Basically three reversible one-electron processes are observed (Table S3). If the more positive one is clearly assigned to metal centered process MIII/MII (M = Ni or Co), the two other reversible reduction waves can be assigned either to metal or ligand centered reduction process.42,62-65,67,69 Recent investigations by Wieghardt et al.67 on several nickel tetraaza-macrocyclic complexes by EPR spectroscopy coupled to DFT calculations, clearly established that for the sixcoordinate [NiII(CR)Cl2] (4) complex, the oxidation wave (E1/2 = +0.45 V vs Ag/AgNO3) in CH3CN is a metal centered process leading to [NiIII(CR)Cl2]+ whereas both reductions correspond to successive ligand reductions yielding respectively the square planar monocation [NiII(CR•-)]+ by release of two chloride and a square planar neutral species [NiII(CR2-)]0 (E1/2 = -1.03

and 2-

-1.49 V vs Ag/AgNO3; CR = doubly reduced bis(imino)pyridine species). Concerning the cobalt(III) tetraaza-macrocyclic compounds 1 and 2 and other derivatives of this family,42,63,64 the recent investigation of Peters65 suggests, as stated in the introduction, a ligand-centered process for the second reversible reduction process, “Co II/I”, at the solid state. The real nature of the triply reduced species is still not elucidated. For the aquo derivative 2, the three reversible redox processes in CH3CN are respectively located at E1/2 = +0.32, -0.79 and -1.79 V vs Ag/AgNO3. For 1, the first two processes at E1/2 = -0.41 and -0.86 V are shifted to more negative potentials in accordance with the donor ability of the chloro ligands, while the third one remains located at -1.79 V. The CoII and formally “CoI” forms were quantitatively generated in CH3CN by bulk electrolysis and fully characterized by UV-Visible (Fig. S2 for 2) and EPR spectroscopy.42 As expected, the CoIII and “Co I” solutions are EPR silent, while the initial Co II solution of 3 exhibits an EPR signal at 100 K typical of a low-spin d 7 Co II center.42 The CoII form has been also obtained in aqueous solution either by 11

electrochemical reduction of the initial CoIII complex or by reduction with ascorbate, resulting in similar spectroscopic features. In view to obtain spectroscopic characterization of the lower triply reduced state, formally “Co 0”, we have performed in this work a bulk electrolysis at E = -1.85 V of an electrogenerated solution of the aquo derivative in CH3CN at room temperature. Unfortunately, this reduced state was found to be too instable to allow for a UV-visible spectroscopic characterization. Attempts to chemically isolate this triply reduced compound by the group of Peters also failed, resulting in the isolation of the doubly reduced state with a deprotonated amine.65

3.3 TD-DFT investigations on the cobalt (2) and zinc (3) derivatives under their initial and reduced forms We investigated by DFT calculations the electronic structure of the aquo derivative 2 under its 3+/2+/+

forms (Tables 1-2, S4 and Figs. 4, 5, S3-S24). TD-DFT calculations were conducted at

the B3LYP level of theory in order to assign the UV-Vis-NIR transitions of the complex under each oxidation state. This study focuses on this single complex since in aqueous solution, i. e. under catalytic conditions, the chloro ligands of 1 are substituted by H2O molecules. This has been previously demonstrated by the similarity in the UV-visible spectra of both CoIII and Co II forms of 1 and 2 in water.42 DFT calculations were performed by considering water as solvent for the trication and the dication of 2, and acetonitrile for the monocation since the spectroscopic characteristic of this species was only obtained in this solvent. Geometry optimization on the low-spin complex [CoIII(CR)(H2O)2]3+ predicts a symmetrical coordination polyhedron, with Co-Nimine, Co-Npyridine and Co-Namine bond lengths of 1.971 Å, 1.870 Å and 1.994 Å, respectively (Table 1). These values are in good agreement with data reported by Lau et al.63 for the crystallized Co(III) complex [CoIII(CR)Cl2]+: Co-Nimine: 1.968(3) and 1.939(3) Å; Co-Npyridine : 1.847(3) Å and Co-Namine: 1.972(3) Å. The Cimine-Nimine bond distances are good markers of the oxidation state of the ligand, as pointed out by Wieghardt et al.67 The predicted Cimine-Nimine bond distances at 1.310 Å are consistent with the double bond character of the imine. These values are again in the range of those measured in the crystal structure of [Co III(CR)Cl2]+ (1.278(5) and 1.312(5) Å).63 TD-DFT calculations were performed on the geometry optimized of [Co III(CR)(H2O)2]3+. A visible electronic excitation was computed at 430 nm (f = 0.0039), which matches the experimental band detected at λexp = 421 nm (Figs. 4, S3 and Table 2). It corresponds to a mixture of LMCT and 12

d-d transitions. Regarding the dication, we considered both an octahedral [Co II(CR)(H2O)2]2+ and square pyramidal [Co II(CR)(H2O)]2+ geometry. The spin state was set to (S = ½) according to EPR data.42 The calculated Co-Nimine bond lengths are of 1.954 Å and 1.949 Å for [Co II(CR)(H2O)2]2+ and [CoII(CR)(H2O)]2+, respectively (Table 1). They are, as expected, smaller than those predicted in the trication and fall within the range of the experimental ones reported for the cobalt(II) complex [Co II(CR)(CH3CN)]2+ (1.957(1) and 1.971(1) Å).65 Furthermore, the Cimine-Nimine bond distances at 1.313 Å in both cases are similar to those calculated for [Co III(CR)(H2O)2]3+, consistent with an imine character of these groups. Several electronic excitations of moderate intensity are predicted in the range 450-600 nm for [Co II(CR)(H2O)2]2+ (Table 2 and Figs. S4, S14), which closely resemble the experimental features (Fig. 4(b)). As shown in Table 2 they are mainly assigned to charge transfer transitions. The [CoII(CR)(H2O)]2+ complex exhibits main absorptions at 406 and 417 nm (MLCT transitions). The multiline pattern observed experimentally is less reproduced by TDDFT calculations in the absence of the second water molecule (Figs. 4(c), S4), suggesting that the structure of the cobalt(II) complex is [Co II(CR)(H2O)2]2+. For the electronic structure of complex 2 under its mono-cationic form in CH3CN solution, three formulations were considered: singlet closed-shell [CoI(CR)(CH3 CN)]+, triplet [Co II(CR•-)(CH3CN)]+ and broken symmetry singlet [Co II(CR•-)(CH3CN)]+. Recently, this complex was isolated from CH3CN under a reduced form as single crystals.65 X-ray diffraction data showed that the metal ion is in a square pyramidal geometry, with an acetonitrile molecule coordinated in apical position. Most interestingly, the Cimine-Nimine bond are lengthened by ca. 0.025 Å after reduction (mean Cimine-Nimine bond distance = 1.299 Å in the dication that enlarges to 1.322 Å in the monocation). This supports a redox non-innocence of the ligand, which adopts an open-shell configuration in the monocation while the cobalt ion retains its formal (+II) oxidation state. In the present DFT investigation, we considered both a square pyramidal and a square planar geometry for the metal ion and three electronic configurations in each case. The broken symmetry singlet [CoII(CR•-)(CH3CN)]+ is lower in energy than the singlet closed-shell form [CoI(CR)(CH3CN)]+, as shown previously.65 The difference is however only 4.3 kcal/mol, indicating that the singlet closed-shell and open-shell are close in energy. When the fifth ligand is absent, the triplet [CoII(CR•-)]+ and closed-shell singlet [CoI(CR)]+ species are almost isoenergetic (the triplet state is slightly favored), with a triplet-singlet gap of 2.4 kcal/mol that is close to the error associated with calculations. In

13

each open-shell configuration the Cimine-Nimine bond is predicted to be larger (1.338-1.342 Å) than in the cobalt(II) and cobalt(III) complexes (1.310 and 1.313 Å, respectively) (Table 1). Reduction of the ligand indeed populates an orbital that is anti-bonding with respect to the imine (see ESI).90 Interestingly, an elongation of the Cimine-Nimine bond of the same magnitude is also predicted for the closed-shell singlet (1.333-1.328 Å). On the other hand, examination of the coordination bond distances reveals that the Co-Nimine (1.915-1.906 Å) and Co-Npyridine (1.809-1.804 Å) bond are significantly shorter in the closed-shell singlet than in the open shell configuration (1.938-1.929 and 1.853-1.831 Å, respectively). Thus, the coordination bonds may be better indicators for the oxidation state of the metal ion. With experimental CoN1imine, Co-N2imine and Co-Npyridine bond distances of 1.957, 1.972 and 1.848 Å, respectively, the oxidation state of the cobalt ion is clearly (+II) in the structure reported.65 Before discussing on the TD-DFT results, it is interesting to comment on the shape of the β-HOMO (lowest energy SOMO of the radical complexes) (Fig. S13). It is indeed expected to host the additional electron upon reduction of [CoII(CR•-)(CH3CN)]+ into the corresponding neutral species. The β-HOMO is the HOMO with the higher metal contribution, which extends over the ligand (see ESI). On this basis, both ligand- or metal-centered processes may be considered for the reduction of [CoII(CR•-)(CH3CN)]+.91 Since the energetic analysis points to a small energy gap between the closed-shell and openshell configurations, both valence isomers may be accessible in solution.92 To address this issue, the electronic spectra of each electronic configuration were computed by TD-DFT methods. Regarding the triplet radical [CoII(CR•-)(CH3CN)]+ two main electronic excitations are predicted in the Vis-NIR region: λcalc = 1439 nm (f = 0.0282) and 451 nm (f = 0.0362) (Table 2 and Fig. S23). Both are assigned to π-π* transitions involving the ligand radical; the NIR band corresponds to the αHOMO → αLUMO transition. While the latter reasonably matches the band detected at λexp = 432 nm, the former one does not reproduce satisfactorily the feature observed in Vis-NIR region: λexp = 680 nm and 808 nm (shoulder) (Fig. S23). When the broken symmetry singlet [Co II(CR•-)(CH3CN)]+ is considered, two main electronic excitations are predicted at shorter wavelengths, λcalc = 1177 nm (f = 0.0363) and 422 nm (f = 0.0177), which match slightly better the experimental features (Table 2 and Figs. 5(b)). In the absence of a fifth ligand, an intense band is again predicted at ca. 1150 nm, with oscillator strength f > 0.04 (Table 2). We alternatively considered a closed-shell electronic configuration for the complex. For [CoI(CR)(CH3CN)]+ three main Vis-NIR bands were computed at λcalc = 645 nm (f = 0.0474), 590 nm (f = 0.0278) and 470 nm (f = 0.0119) which more reasonably 14

match the experimental features (Figs. 5(a)). They mostly correspond to MLCT transitions. When the fifth ligand is not included the predicted spectrum is in poorer agreement with the experimental one (Table 2 and Fig. S24). Thus, while the band at λexp = 432 nm is well predicted by considering the broken symmetry singlet [CoII(CR•-)(CH3CN)]+ the energy of the NIR features is largely underestimated. Conversely, calculations on the singlet closed-shell [Co I(CR)(CH3CN)]+ slightly overestimate the energy of the NIR feature (590 and 645 vs. 680 and 808 nm) and reproduce satisfactorily the band at λexp = 432 nm. We additionally tested a GGA functional (BLYP) and a long-range-corrected hybrid (CAM-B3LYP) functionals (Table S4), which lead to essentially the same features. Both functionals however predict bands at slightly lower energy for the ligand radical species. Essentially the same spectrum was predicted for the closed shell form by using the CAM-B3LYP and B3LYP functionals, whereas the bands are predicted at slightly higher energy with the GGA functional. In order to get further insight on the signature of the radical complex [Co(CR•-)(CH3CN]+, we performed TD-TDFT calculations on its zinc analog, namely [ZnII(CR•-)(CH3CN)]+ (as well its chloride derivative [ZnII(CR•-)(Cl)]0 see ESI), wherein the ligand radical character was unambiguously established by EPR spectroscopy (see above). In both [ZnII(CR•-)(CH3CN)]+ and [ZnII(CR•-)(Cl)]0 two main electronic excitations were calculated in the visible region: 443 (f = 0.0706) and 1394 nm (f = 0.0419) for [ZnII(CR•-)(CH3CN)]+ and 450 (f = 0.0697) and 1451 (f = 0.0381) for [ZnII(CR•-)(Cl)]0 (Table 2 and Fig. S26). No band was predicted in the visible region for the precursors [ZnII(CR)(CH3CN)]2+ and [ZnII(CR)(Cl)]+, in agreement with experimental data. The two bands predicted for [ZnII(CR•-)(CH3CN)]+ correspond to π-π* transitions involving the radical moiety,67 similarly to the triplet [Co(CR•-)(CH3CN]+. They are calculated at energies remarkably similar to those predicted for triplet [Co(CR•-)(CH3CN]+, although the experimental spectra differ significantly between the two compounds. This behaviour suggests that the metal-centered alternative for the reduction of [Co(CR)(CH3CN]2+ remains viable pathway, confirming the energetic analysis. Overall, these DFT and TD-DFT calculations at the B3LYP level of theory indicate that the energy gap between the Co(I) and Co(II)-radical forms is rather small. The ligand-radical form is the most stable configuration, in agreement with X-Ray diffraction data on [Co II(CR•)(CH3CN)]+,65 although both configurations are nearly isoenergetic in the absence of axial ligation. It is therefore conceivable that both valence isomers of the reduced complex may exist in solution.

15

Table 1. Calculated bond distances (BP86) a Complex Co-Nimine [CoIII(CR)(H2O)2]3+ 1.971 [CoII(CR)(H2O)2]2+ 1.954 [CoII(CR)(H2O)]2+ 1.949 [CoI(CR)(CH3CN)]+ 1.915 [CoI(CR)]+ 1.906 [CoII(CR•-)(CH3CN)]+ 1.938 [CoII(CR•-)]+ 1.929 [ZnII(CR•-)(CH3CN)]+ 2.151 [ZnII(CR•-)Cl]0 2.172 [ZnII(CR•-)(CH3CN)]+ 2.122 [ZnII(CR•-)Cl]0 2.145 a

Co-Npyridine 1.870 1.843 1.843 1.809 1.804 1.853 1.831 2.036 2.058 1.986 2.006

Co-Namine 1.994 1.993 1.989 2.063 1.981 2.033 1.989 2.045 2.074 2.080 2.107

Cimine-Nimine 1.310 1.313 1.313 1.333 1.328 1.338 1.342 1.290 1.289 1.318 1.316

Bond distances in Å; The bond distances for the radical species are obtained from geometry optimization on the triplet forms.

Table 2. TD-DFT assignment of the electronic excitations of the cobalt and zinc complexes a Complex S f Assignment Orbitals λcalcd [CoIII(CR)(H2O)2]3+ 0 430 0.0039 LMCT/dd - [b] II 2+ [Co (CR)(H2O)] ½ 417 0.0029 MLCT - [b] 406 0.0022 MLCT αHOMO-2→αLUMO [CoII(CR)(H2O)2]2+ ½ 530 0.0013 MLCT - [b] 494 0.0038 MLCT αHOMO→αLUMO 421 0.0020 MLCT βHOMO-1→βLUMO+1 [CoI(CR)]+ 0 583 0.0367 MLCT HOMO-2→LUMO+1 393 0.0540 MLCT - [b] [CoI(CR)(CH3CN)]+ 0 645 0.0474 MLCT HOMO→LUMO 590 0.0278 MLCT - [b] 470 0.0119 MLCT HOMO-1→LUMO [CoII(CR•-)]+ 1 1149 0.0435 π- π* αHOMO→αLUMO 420 0.0284 π-π* αHOMO→αLUMO+2 [CoII(CR•-)(CH3CN)]+ 1 1439 0.0282 π-π* αHOMO→αLUMO 451 0.0362 π-π* αHOMO→αLUMO+2 428 0.0096 MLCT βHOMO-1→βLUMO+1 [CoII(CR•-)]+ 0 1180 0.0420 π-π* βHOMO→βLUMO 427 0.0286 π-π* βHOMO→βLUMO+2 [CoII(CR•-)(CH3CN)]+ 0 1177 0.0363 π- π* αHOMO→αLUMO 422 0.0177 - [b] π-π* [ZnII(CR•-)(CH3CN)]+ ½ 1394 0.0419 π- π* αHOMO→αLUMO 443 0.0706 π-π* αHOMO→αLUMO+3 [ZnII(CR•-)Cl]0 ½ 1451 0.0381 π- π* αHOMO→αLUMO 450 0.0697 π-π* αHOMO→αLUMO+1 a

Only the principal electronic excitations calculated above 390 nm are listed; plots of the calculated spectra, transition differences and MO are given in ESI. b The transition involves several orbitals with different contributions.

16

3.4.

Photocatalytic

activities

for

hydrogen

production

with

the

Ru/Cat/NaHA/H2A systems Photocatalytic hydrogen generation experiments were carried out under continuous visible light irradiation (400 - 700 nm) in 5 mL of water at pH 4.0 buffered with HA-/H2A and different relative concentrations of Ru/Cat. The pH of the aqueous solutions was adjusted by varying the relative concentration of HA- and H2A.35 At pH 4.0, the concentration NaHA and H2A are equimolar. Most of the photocatalytic experiments were performed with a total concentration of HA- and H2A of 1.1 M. The H2 produced was monitored in real time and quantified by GC analysis of the headspace gas mixture; this was then used to calculate the turnover number and the initial turnover frequency of catalyst (denoted TONCat and TOFCat, respectively).42 If control experiments, in the absence of either Ru or the HA-/H2A couple, produced no appreciable amount of hydrogen, a small amount of hydrogen was produced from solutions containing only Ru and HA-/H2A at pH 4.0, in agreement with previous observations (Table 3).7,35 This amount, stemming from Ru only, was systematically subtracted from the total H2 produced to calculate the corrected values of TONCat, TOFCat and VH2 values (denoted TONCat*, TOFCat* and VH2 *) in all experiments.93 The photocatalytic activities of all systems studied in terms of TONCat, TOFCat and VH2 are summarized in Table 3. All the experiments were repeated at least three times and reproducible values within +/- 5% were obtained. As shown in our previous studies,42 the maximum catalytic activity for H2 production with the photocatalytic Ru/1/HA- (0.55 M)/H2A (0.55 M) system is obtained at pH 4.0. The photocatalytic activity was evaluated at a relatively high concentration of catalyst (1x10-4 M) with different concentration of photosensitizer (1x10 -4 M, 5x10-4 M and 1x10-3 M) (Table 3). In such conditions, the system is very efficient and the H2 produced notably increases by increasing the concentration of Ru. For instance, TONsCat* as high as 820 and 1000 have been measured for 5x10-4 M and 1x10-3 M of Ru respectively, which correspond to large volumes of hydrogen (10 and 12.3 mL for 5 mL of solution) due to the relatively high concentration of catalyst (Table 3). The formation of cobalt colloids as possible active catalytic species, that can be generated by the catalyst decomposition in the course of the photocatalytic experiments, was ruled out by conducting mercury poisoning experiments.42 Additionally, if the cobalt catalyst is replaced by simple salt as CoCl2•6H2O, no significant H2 evolution was observed upon irradiation compared to the H2 produces by a Ru/HA-/H2A (1.1 M) solution (Table 3). Regarding the zinc derivative 3, as expected, no activity was found since the traces of hydrogen detected are produced by Ru/HA-/H2A (1.1 M) (Table 3). 17

Table 3. Photocatalytic activities of all systems studied towards hydrogen evolution, in terms of TONCat, TOFCat and VH2.a Experiments were carried out at 25°C in water (5 mL) at pH 4.0 under various relative concentration of components, catalyst (Cat), photosensitizer (Ru) and NaHA/H2A. TONCat are given after 21h of irradiation under visible-light (λ = 400 -700 nm). Ru[Ref] [mol L-1] 1 x 10-3, 42 5 x 10-4, 42 5 x 10-4, 42 5 x 10-4, 42 1 x 10-4, 42 5 x 10-4, this work 5 x 10-4,this work

Cat [mol L-1] -

NaHA/H2A [mol L-1]/pH 0.55/0.55/4.0 0.55/0.55/4.0 0.55/0.55/4.0 0.55/0.55/4.0 0.55/0.55/4.0 0.25/0.25/4.0 0.05/0.05/4.0d

Ru/Cat -

TONCata (TONCat*)b -

TOFCata (TOFcat*)b -

VH2a (VH2*)b mL 0.15 0.10 0.072 0.056 0.03 0.094 0.091

Irrad. timec 21h 21h 5h 3h 21h 30h 46h

CoCl2•6H2O 5 x 10-4,this work 0.55/0.55/4.0d 4 0.044 6h (1 x 10-4) 21h 1 x 10-3, 42 1 (1 x 10-4) 0.55/0.55/4.0 10/1 1014 (1002) 117 (108) 12.42 (12.27) -4, 42 -4 21h 5 x 10 1 (1 x 10 ) 0.55/0.55/4.0 5/1 828 (820) 106 (99) 10.15 (10.05) 21h 1 x 10-4, 42 1 (1 x 10-4) 0.55/0.55/4.0 1/1 345 (343) 41 (38) 4.23 (4.20) -4, this work -5 21h 5 x 10 1 (5 x 10 ) 0.55/0.55/4.0 10/1 1134 (1130) 268 (254) 6.95 (6.92) 5h 5 x 10-4, this work 1 (1 x 10-5) 0.55/0.55/4.0 50/1 780 (715) 583 (515) 0.95 (0.88) 3h 5 x 10-4, this work 1 (5 x 10-6) 0.55/0.55/4.0 100/1 721 (591) 1125 (988) 0.44 (0.38) -4, this work -4 30h 5 x 10 1 (1 x 10 ) 0.25/0.25/4.0 5/1 654 (647) 64 (62) 8.01 (7.92) 46h 5 x 10-4, this work 1 (1 x 10-4) 0.05/0.05/4.0d 5/1 457 (450) 17 (14) 5.60 (5.51) -4, this work -4 21h 5 x 10 3 (1 x 10 ) 0.55/0.55/4.0 5/1 5 0.06 -4, this work -4 21h 5 x 10 4 (1 x 10 ) 0.55/0.55/4.0 5/1 7 0.08 a TONCat and TOFCat are respectively the maximum turnover number and the initial turnover frequencies (TOFCat = TONCat h-1) per catalyst obtained by the system, and VH2 is the total volume of H2 produced by the system. b TONCat*, TOFCat* and VH2* are the corrected values of TONCat, TOFCat and VH2, respectively, obtained by subtracting the production of H2 stemming from Ru (at 5 x 10-4 M) without catalyst. c Irrad. Time: time of irradiation after which H2 production stopped. d In this experiment an acetate buffer was used at 1M.

We have further investigate the catalytic activity of 1 in water, by exploring the dependence of lower catalyst concentration (