Photocatalytic CO2 to CO Reduction with Cyclometalated Pt(II) Complexes of N^C^N Pincer Dipyridylbenzene Ligands: A DFT study

1. Introduction

The problem of global overheating dates back long before the industrial revolution. The excessive use of fossil fuels and the massive release of greenhouse gases into the atmosphere have turned the term "global warming" into an uncomfortable reality of modern life. CO₂, although not a primary greenhouse gas, nevertheless has a significant contribution to global warming. In 2012, global CO₂ emissions, which come mainly from the burning of fossil fuels, reached a total of 34.5 billion tons [1].

The steadily increasing concentration of CO₂ in the atmosphere is now a real problem. CO₂ emissions from the massive consumption of fossil fuels are largely responsible for global climate change: an increase in the concentration of CO₂ in the atmosphere causes global warming as well as ocean acidification from the uptake of atmospheric CO₂ [2].

Man's growing needs for energy lead to the ominous conclusion that CO₂ emissions will not remain stagnant, but will increase even more. This results in the scientific community focusing on how the abundance of CO₂ will be used to produce materials of commercial interest. Only 1‰ of the total abundance of CO₂ is currently used for chemical synthesis, which is mainly due to its chemical inertness (it is highly oxidized in nature) but also to the fact that capturing and storing CO₂ is an expensive process [3].

Many studies have been focused on how to deal with the problem of the continuous increase in atmospheric CO₂ while at the same time there are ongoing efforts and studies not only to capture CO₂ but at the same time to catalytically convert it into "fine chemicals" such as CO, CH₃OH and HCOOH [3].

Among various strategies to achieve this goal, the biggest and most attractive challenge is the efficient conversion of CO₂ into useful compounds using sunlight as an energy source [4]. Studies on this issue have shown the photochemical and electrochemical reduction of CO₂ to CO, to CH₃OH, but also to HCOOH, using transition metal electrodes, metal complexes, semiconductors as well as some organic molecules [5,6,7,8].

Much research has focused on how to sequester CO₂ from coal plant emissions, which account for 76% of all global emissions [9]. Recent advances in catalyst design combined with the reduced cost of clean energy has made the use of CO₂ more attractive as a feedstock for chemical production. Particular interest has focused on the conversion of CO₂ to CO as a step in the petrochemical production process [10].

Mainly there are two approaches for the CO₂ reduction namely either via heterogeneous or homogeneous catalysis. Both approaches could be achieved either electrocatalytically or photocatalytically. The heterogeneous electrocatalytic CO₂ reduction is done with electrolysis with the aid for example of nanostructured materials [11] while heterogeneous photocatalytic CO₂ reduction utilizes mainly various types of semiconductors as photocatalysts [12]. On the other hand, the homogeneous electrochemical/photochemical CO₂ reduction is mainly based on transition metal complexes [13]. So far, coordination complexes of numerous metals have been studied as catalysts such as the first row Mn, Fe, Co, Ni and Cu, also the second row Ru, Rh, Pd, as well as the third row W, Re, Os, Ir [13].

Up to now, the Ru and Re complexes have received the major focus and are the most widely studied. Probably, the interest for these complexes stems from the inspiring, seminal works of Lehn et al. [14,15,16], who studied the electro/photocatalytic CO₂ reduction by Re(I) and Ru(II) complexes. Since then, many other Re(I) and Ru(II) complexes have been studied for the electro/photocatalytic CO₂ reduction [2,17,18].

The reduction of CO₂ to either CO or HCOOH proceeds via the following chemical equations:

CO₂ + 2e⁻ + 2H⁺ → CO + H₂O

(1)

CO₂ + 2e⁻ + 2H⁺ → HCOOH

(2)

The reduction potentials of the above 2e^-/2H⁺ reductions are -0.52 and -0.61 V vs SHE [19] and the e^- are provided either from the electrodes in the case of electrocatalysis or by a ‘sacrificial donor’ e.g. TEOA in the case of photocatalysis. At least for Re(I) and Ru(II) complexes, the proposed mechanism involves, in almost all cases, the formation of an coordinatively unsaturated, 17e^- very reactive intermediate [2]. Thus, for example, in the reduction of CO₂ with cis-[Ru(bpy)₂(CO)X]ⁿ⁺ a mixture of CO/HCOO^- is obtained [19]. Two mechanisms are proposed to operate in conjunction, having in common the very reactive, five-coordinated, cis-[Ru(bpy)₂(CO)]⁰ intermediate. The product selectivity depends upon the reactivity of this intermediate towards CO₂ or H⁺ [2]. The so called η¹-CO₂ complex mechanism operates upon CO₂ capture by the cis-[Ru(bpy)₂(CO)]⁰ intermediate, forming the cis-[Ru(bpy)₂(CO)(η¹-CO₂)] adduct which upon protonations yields CO. On the other hand, in the so called hydride mechanism, operating simultaneously with η¹-CO₂ complex mechanism, the cis-[Ru(bpy)₂(CO)]⁰ intermediate captures H⁺ forming a hydrido intermediate which upon reduction and CO₂ insertion into the Ru-H bond yields HCOO^-. However, it should be noticed that the unsaturated five-coordinated [Ru(bpy)₂(CO)]⁰ intermediate has not yet been detected [2]. Accordingly, there have been proposed two other mechanisms for the reduction of CO₂ [18].

Ishitani et al. [20], using various spectroscopic techniques, demonstrated the existence of the fac-[Re^I(bpy)(CO)₃(OCH₂CH₂NR₂)] formed upon coordination of TEOA^- anion to the metal centre of the 17e^- five-coordinated fac-[Re^I(bpy)(CO)₃]⁺ intermediate. They also demonstrated the existence of the fac-[Re^I(bpy)(CO)₃(O(CO)OCH₂CH₂NR₂)] produced upon insertion of CO₂ into the Re-O bond of the fac-[Re^I(bpy)(CO)₃(OCH₂CH₂NR₂)] intermediate. A similar, CO₂ insertion product has been observed also for the [Ru(dmb)₂(CO)₂]²⁺ complex in DFM-TEOA solution [21]. The CO₂ to CO reduction upon coordination of TEOA^- anion and subsequent CO₂ insertion has also been scrutinized for Re(I) complexes with the aid of DFT calculations [22].

Finally, based upon both experimental [23,24] as well as theoretical [25] studies there is also another proposed mechanism for the CO₂ reduction by transition metal complexes thought to operate via dimerization of the five-coordinated, very reactive intermediate. Hence, Muckerman et al. [25] employing DFT electronic structure calculations, showed that CO₂ is inserted between two five-coordinated [Re(dmb)(CO)₃]^· (dmb = 4,4′-dimethyl-bpy), radical intermediates bridging their metal centers. Through second CO₂ addition to this dimeric species, and via a transition state, CO is produced.

Although, so far complexes of a large number of transition metals have been studied for the electro/photocatalytic CO₂ reduction, Pt complexes have attracted very little attention. Ceballos et al. [26], demonstrated the electrocatalytic CO₂ reduction to formate using the [Pt(dmpe)₂](PF₆)₂ complex with high Faradaic efficiency and low overpotential. The vast majority of studies on CO₂ reduction, involving Pt metal have been devoted to heterogeneous catalysis. For example, Zhang et al. [27], dealt with the photocatalytic reduction of CO₂ to CH₄, under the influence of ultraviolet radiation, with water as solvent on a TiO₂ catalyst with Pt. Quite numerous other studies have been also appeared in the literature [28,29,30,31,32,33,34] concerning the heterogeneous CO₂ transformation.

To the best of our knowledge, neither experimental nor theoretical studies have been done so far, for the photocatalytic CO₂ to CO reduction employing Pt(II) square planar complexes as catalysts. Accordingly, we instigated to scrutinize, by means of electronic structure calculations, the photocatalytic reduction of CO₂ to CO using square planar Pt(II) cyclometalated complexes bearing N^C^N coordinating substituted dipyridylbenzene ligands of the general formula [Pt(5-R-dpb)Cl] (dpb = 1,3-di(2-pyridyl)benzene anion, R = H, N,N-dimethylaniline, thiophene, bodipy). Our targets are: 1) To study the photophysical properties of these complexes in order to investigate if they could be used as both photosensitizers as well as catalysts, 2) to explore the mechanism of the catalytic reaction and 3) to probe how the nature of the N^C^N substituted bipyridyl ligand affects both the photophysical properties and mechanism of action of the Pt(II) complexes under study.

2. Results and Discussion

The molecular structure of the square planar Pt(II) complexes employed for the CO₂ to CO photocatalytic reduction in this study is depicted in Scheme 1.

The CO₂ to CO photocatalytic reduction is thought to proceed via two steps: 1) The photoexcitation/reductive quenching step and 2) the actual catalytic cycle. The two steps are depicted in Figure 1 along with the molecular structures of all species participating in the overall process. Let us start examining first the initial step which henceforth we shall call it the Photoexcitation/Reductive Quenching step and next we will deal with the second step which we shall call it the Catalytic Cycle step.

2.1. Photoexcitation/Reductive Quenching step

In the first step, the Pt(II) catalyst, upon irradiation, is excited to its first singlet excited state, S₁. For a third row transition metal complex like those under study, it is expected the transition to occur from the S₁ state to the first triplet excited state, T₁ via Intersystem Crossing, ISC due to Spin-Orbit Coupling, SOC. Then, if the lifetime, τ of the T₁ state of the complex is large enough, a reductive quenching occurs where the complex upon accepting an e^- from a donor, D e.g. TEOA or TEA, yields the One Electron Reduced, OER complex (Figure 1). Notice that, the experimentally determined τ of the T₁ state of 1 has been found to be 7.2 μs, in CH₂Cl₂ solvent at 298ºC [35] which is large enough to permit reductive quenching. The optimized geometries of 1 – 4 in their S₀ and T₁ states as well the geometries of the respective OER species, along with selected structural parameters calculated at the PBE0-GD3BJ/Def2-TZVP level, in DCM solvent, are given in Figure 2 as well as in Figures S1–S3 of the Supplementary Material. The calculated structural parameters of 1 in its S0 ground state are in excellent agreement with the respective structural parameters derived from the X-ray structural analysis of this complex [36]. Thus, for example, the difference between the calculated and X-ray derived bond lengths around the coordination sphere is in the range of only 0.001 – 0.016 Å while the calculated bond angles differ by only 0.3 – 0.5º from the respective X-ray experimental values. Upon excitation to the T1 excited state, no significant differences in the structural parameters could be observed (Figure 1b and Figures S1–S3). Accordingly, the changes of the Pt-Cl and Pt-C coordination bonds are in the ranges 0.03 – 0.001 Å and 0.009 – 0.019 Å respectively while the changes of the Pt-N coordination bond lie within the range 0.0 – 0.015 Å. On the other hand, the bond angles around the coordination sphere change slightly by about 1º in all cases, 1 - 4. Finally, the same holds also for the organic framework of the N^C^N pincer ligand, where the structural changes upon the S₀ → T₁ excitation are negligible. Next, the one electron reduction of the T₁ state causes more significant structural changes. Thus, there are more obvious structural differences between the optimized geometries of the T₁ state and the OER species (Figure 1c and Figures S1–S3). The most striking structural change is observed in the Pt-Cl coordination bond which is elongated by 0.007 - 0.032 Å in the OER species as compared to the T₁ state species. The same holds also, if we compare the OER species with the S₀ state species Also the Pt-C coordination bond is elongated by 0.035 Å in contrast to the two Pt-N coordination bonds which are shortened.

Finally, the bond angles around the coordination sphere do not change significantly upon one electron reduction.

2.1.1. Absorption Spectra

Since excitation of the precursor complexes, 1 - 4 is a prerequisite in the photocatalytic CO₂ conversion, we set out to study their absorption spectra in DCM by means of TDDFT calculations. The simulated absorption spectra of 1 – 4 in DCM at the TDDFT/CAM-B3LYP/Def2-TZVP level are depicted in Figure 3. Inspection of Figure 3 reveals that the simulated absorption spectra of 1 – 4 exhibit two bands while 1, exhibits also a shoulder. The strongest band, of highest energy for 1, appears around 225 nm accompanied by a shoulder at 243 nm. Finally, the lowest energy band, of medium intensity, appears at 344 nm. In terms of Natural Transition Orbitals (NTOs), the band peaking at 225 nm, arises mainly from an electronic transition at 225 nm which is due to two electron excitations between the NTO pairs depicted in Figure 4. Based upon the shape of these NTOs the band could be assigned as MLCT/IL/LL´CT. Next, the shoulder around 243 nm, arises mainly from an electronic transition at 253 nm due to two electronic excitations (Figure 4) and is assigned as MC/LC/MLCT. Finally, the lowest energy band at 344 nm, arise from an electronic transition which due to one electronic excitation involving an NTO pair depicted in Figure 4. This band could be assigned as MLCT/LC. Upon substitution of one H of the N^C^N pincer ligand with strong donor groups such as PhNMe₂ in complex 2 or thiophene in complex 3, the simulated absorption spectra change significantly. Accordingly, for complexes 2 and 3 we observe a weak band, of highest energy, peaking at 229 and 224 nm respectively. Next, there is the strongest band, of lower energy, at 276 nm for both 2 and 3. Finally, there is a shoulder, of lowest energy, in the region 370 – 380 nm. The NTO pairs, relevant to the most significant bands appearing in the absorption spectra of 2 and 3, are depicted in Figures S4 and S5 respectively (See Supplementary Materials). Thus, the bands in the absorption spectra of 2 at 229 nm and of 3 at 224 nm arise mainly from electronic transitions at 227 and 224 nm respectively. Each of these two electronic transitions are due to two electronic excitations between the NTO pairs depicted in Figures S4 and S5. Based upon the shapes of these NTO pairs, the highest energy bands in the absorption spectra of 2 and 3 are assigned as MLCT/IL. Next, the strongest bands in the absorption spectra of 2 and 3, peaking at 276nm, are due to mainly an electronic transition at 263 and 280 nm respectively. The latter, are due to two electronic excitations each and based upon the shape of the participating NTO pairs, we could assign the strongest bands as MLCT/IL. Finally, the shoulder, of lowest energy, appearing in the spectra of 2 and 3 in the region 370 – 380 nm is due to mainly an electronic transition at 376 and 366 nm respectively and based upon the shape of the participating NTOs are assigned as MLCT/IL.

Upon introduction of the diazaborinine group in the N^C^N pincer ligand, the simulated absorption spectrum of 4 changes dramatically with respect to the other complexes under study. Accordingly, the spectrum is red shifted, exhibiting two main bands, one in the UV region, peaking at 260 nm, and the other in the visible region, peaking around 400 nm. The former, arises mainly from an electronic transition at 257 nm while the latter arises from an electronic transition at 422 nm. The electronic transition at 257 nm is due to two electronic excitations and based upon the shapes of the participating NTO pairs (Figure S6) could be assigned as MLCT/IL. On the other hand, the electronic transition at 422 nm is almost solely due to one electronic excitation between NTOs located at the diazaborinine substituent (Figure S6) and therefore the band in the visible region is assigned as IL.

2.1.2. Excited State Electrochemistry

One of the most important steps in the photocatalytic CO₂ conversion by transition metal catalysts is the reductive quenching of their T₁ state upon receiving an e from a sacrificial donor e.g. TEOA. Therefore, we instigated to study the T₁ excited state reduction potentials by means of DFT calculations. The ground state redox potentials could be calculated employing the Born – Haber cycle [37] depicted schematically in Scheme 2.

The standard absolute ground-state reduction potential, E⁰_red is calculated by the following equation:

E⁰_red = −ΔG°(soln. redox)/ZF,

(1)

where F is the Faraday constant (23.061 kcal per volt gram equivalent) and Z is unity for one-electron redox processes. ΔG°(soln., redox) is obtained from the following equation:

ΔG°(soln., redox) = ΔG°(gas, redox) + ΔG°(solv.,[Pt(5-R-dpb)Cl]^•-) - ΔG°(solv.,[Pt(5-R-dpb)Cl], S₀)

(2)

In Table 1 are tabulated the values of the reduction potentials calculated for complexes 1 – 4.

2.2. Catalytic Cycle Step

The initial photoexcitation/reductive quenching step ends with the formation of the OER species. The latter, upon losing a chloride ligand, yields a three coordinated intermediate, ImA which starts the actual catalytic cycle (Figure 1). ImA could be considered as the ‘true’ catalyst while the initial Pt(II) complex could be considered as a ‘precatalyst’. Based upon earlier studies [2], the Catalytic Cycle Step could proceed upon one electron reduction of ImA with the aid of an electron donor, D like for example TEOA yielding ImB. Next, there is one of the most important steps i.e. the CO₂ addition to the metal center of ImB. After two subsequent protonations (e.g. with TEOA which can act as proton source as well [2]) we arrive, through transition state TS, to the carbonyl intermediate, ImF. A second electron reduction of the latter yields ImG from which we obtain CO and the initial ‘true’ catalyst, ImA.

2.2.1. The η¹-CO₂ complex

Since the CO₂ capture/activation step, upon coordination to the metal center of the catalyst, is considered to be of paramount importance in the electro/photocatalytic CO₂ conversion [38,39,40], we set out to study in depth the structural, bonding and electronic properties of ImC.

Structural Properties

The optimized geometry of 1_ImC calculated at the PBE0-D3/Def2-TZVP level, in DCM solvent, is depicted in Figure 5a while those of 2_ImC, 3_ImC, and 4_ImC are given in Figures S5–S7 of the Supporting Information.

Perusal of Figure 5a and Figures S5–S7 reveals that, in all complexes, CO₂ is coordinated to the Pt metal center via the C atom in a η¹ bonding mode. The complex formed via the η¹-CO₂ bonding mode is considered to be of pivotal importance for the majority of the electron transfer reactions for the photo/electrocatalytic CO₂ conversion [2,40]. The Pt-CO₂ bond length is estimated to be 2.132 - 2.135 Å indicative of a bonding interaction. Upon coordination to the Pt metal center, the CO₂ molecule is activated and from linear becomes bended with a <O-C-O bond angle equal to 120.8º while the C-O bonds are equal to 1.270 Å, elongated by about 0.110 Å as compared to the ‘free’ CO₂ molecule (1.160 Å).

Bonding and Electronic Properties

The covalent nature of the Pt-CO₂ bond is reflected in the shape of HOMO calculated for 1_ImC and depicted schematically in Figure 5b (the HOMOs of 2_ImC, 3_ImC and 4_ImC). The HOMO is a bonding MO mainly constructed by the in-phase combination of the Pt d_z2 AO with the π* MO of CO₂ in line with previous studies [38]. The bond dissociation energy, D₀ of the Pt-CO₂ bond was found to be -36.8, -36.9, -35.1 and -10.3 kcal/mol for 1_ImC, 2_ImC, 3_ImC and 4_ImC respectively. Obviously, substitution in position 5 of the N^C^N pincer ligand has no significant impact on D₀(Pt-CO₂). Exception is complex 4, where the introduction of the diazaborinine group reduces D₀(Pt-CO₂) to almost more than a third as compared to the rest of the complexes. Nevertheless, based upon the magnitude of D₀(Pt-CO₂), the interaction between ImB and CO₂ is expected to be a relatively favorable interaction.

To further analyze the Pt-CO₂ bond in ImC, we employed the Atoms in Molecules, AIM method. In Figure 5c are depicted the Bond Critical Points, BCPs found with AIM method performed for 1_ImC (similar BCPs are detected also for intermediate ImC formed by complexes 2 – 4). According to Bader’s [41,42] theory, the presence of a BCP between two atoms indicates bond formation. Inspection of Figure 5c reveals the existence of a BCP between the Pt metal center and C atom of the CO₂ ligand, indicating a Pt-CO₂ bonding interaction. It has been proposed that [43,44] that based upon the values of certain properties of BCPs, the bonding interactions could be classified into three categories: (a) Pure closed-shell interactions (e.g., ionic bonds, hydrogen bonds and van der Waals interactions) characterized by |V_BCP|/G_BCP < 1 (∇²ρ_BCP > 0 and H_BCP > 0); (b) pure open-shell (covalent) interactions characterized by |V_BCP|/G_BCP > 2 (∇²ρ_BCP < 0 and H_BCP < 0); and (c) intermediate bonds with 1 < |V_BCP|/G_BCP < 2 (i.e., ∇²ρ_BCP > 0 and H_BCP < 0) were V_BCP is the potential energy density at BCP, G_BCP is the kinetic energy density at BCP, ∇²ρ_BCP is the Laplacian of electron density, ρ at BCP and finally H_BCP is the energy density at BCP. Accordingly, the calculated values of these properties for the BCP found between Pt and CO₂ in intermediates 1_ImC – 4_ImC are given in Table 2.

Inspection of Table 1 reveals that the AIM parameters calculated for the Pt-CO₂ BCP, do not vary significantly between the ImC intermediates of 1 – 4. Based upon the values of AIM parameters given in Table 1, we can assume that Pt-CO₂ interaction falls into the third category i.e. intermediate nature arising from an interplay of covalent, electrostatic, charge transfer and probably weak dispersion forces components. However, the latter should be excluded for the Pt-CO₂ bond, in terms of the Reduced Density Gradient, RDG defined by the following equation [45]:

$$\mathrm{RDG}\left(\mathrm{r}\right)=|\nabla \rho (\mathrm{r}\left)\right|/2{\left({3\pi }^2\right)}^{1/3}\rho {\left(\mathrm{r}\right)}^{4/3}$$

(3)

where, ρ(r) is the electron density ρ at point r. The 3D isosurface map of RDG, depicted in Figure 5d, reveals absence of any weak interactions between the Pt metal centre and CO₂ (notice that the green isosurface represents weak interactions regions). Similar RDG maps are observed for complexes 2 – 4 as well. In addition, inspection of Figure 5c reveals existence of two BCPs located between the O atoms of CO₂ and the H atoms of the N^C^N pincer ligand. Also, the RDG function (Figure 5d) reveals non covalent interactions between those atoms. Therefore, these H bond interactions, further contribute to the overall interaction of CO₂ with Im_B intermediate.

Another, useful method for analysing the chemical bond, is the NBO method, which is a partitioning scheme for the electronic charge distribution in a molecule. Thus, we employed NBO calculations to analyse the Pt-CO₂ bond from this point of view. The results of the NBO analysis are given in Table 3.

The Natural Charges obtained from NBO analysis, are found to be 0.235 – 0.260 for Pt and 0.494 – 0.497 for C atom of the CO₂ ligand. This, results in a repulsive electrostatic interaction between Pt and CO₂ in line with previous findings [38,46]. NBO analysis reveals also the existence of a bonding, BD, σ NBO, which is depicted in Figure 6a for 1_ImC (similar BD NBOs are observed also for the rest respective intermediates).

The linear combinations of the BD[σ(Pt-CO₂)] NBOs found for 1_ImC – 4_ImC are given in Table 2. Thus, for example, the σ(Pt-C) bonding NBO of 1_ImC has an occupation number equal to 1.857 |e|, arising from the interaction of the sp^2.23d^1.08 hybrid orbital of Pt (23.15% s, 51.66% p and 25.03% d character) with the sp^1.88 hybrid orbital of the C donor atom of CO₂ (34.72% s and 65.25% p character), and is described as σ(Pt–C) = 0.558h_Pt + 0.830h_C. In Figure 6b are also depicted the shapes of the NBOs participating in donor – acceptor, hyperconjugative interactions related to the Pt-CO₂ bond in 1_ImC. According to NBO analysis [47], there is a stabilization of a system due to charge transfer interactions between specific donor-acceptor NBOs. The stalization energy, ΔE(2) arising from these hyperconjugative (stereoelectronic) interactions is given by the following equation:

ΔE(2) = q_iF_ij/(ε_i – ε_j)

(4)

where q_i is the occupancy of the donor NBO, F_ij are the off-diagonal Fock-matrix elements, ε_i and ε_j is the energy of the donor and acceptor NBO respectively. The two hyperconjugative interactions, related to Pt-CO₂ bond, given in Figure 6b involve σ(Pt–C) and σ*(Pt–C) NBOs and stabilizing the system by overall 86.1 kcal/mol. The overall stabilization due to hyperconjugative interactions for 2_ImC, 3_ImC and 4_ImC is 84.5, 85.5and 80.5 kcal/mol respectively.

Another, method for analyzing a chemical bond is the Charge Decomposition Analysis, CDA [48]. We employed the latter to study the Pt-CO₂ interaction for 1_ImC representative case. The net charge transfer from the Pt fragment towards CO₂ is calculated to be marginal, amounting to only 0.048 |e| (the donation, d and backdonation, b terms were found to be -0.023 and -0.071 |e| respectively and are somewhat balanced). Finally, the charge polarization term, r has a negative value equal to -0.350 indicating electronic charge removal from the overlapping into the non-overlapping regions upon formation of 1_ImC. The orbital interaction diagram for the formation of 1_ImC from fragments [Pt(dpb)]^- (Fragm. 1) and CO₂ (Fragm. 2) is depicted in Scheme 3.

Perusal of Scheme 3 reveals that the bonding HOMO of the complex 1_ImC, reflecting the covalent Pt-CO₂ interaction, is composed by 34% of LUFO+1 of Fragm. 1 and 55% of LUFO of Fragm, 2. In other words, the HOMO of 1_ImC, is constructed by the in-phase interaction of the, mainly Pt d_z2 character, LUFO+1 of the [Pt(dpb)]^- fragment with a π* MO of the CO₂ fragment.

2.2.2. Energetic Reaction Profiles

The free energy reaction profiles calculated for 1 - 4, corresponding to the respective catalytic cycle step (Figure 1) are given in Figure 7. Inspection of the energetic profiles reveals that the CO₂ to CO conversion by the ‘real’ catalyst, ImA is a strongly exergonic reaction. Upon reduction, ImA yields the reactive intermediate ImB which subsequently captures a CO₂ molecule to yield the η¹-CO₂ complex, ImC. Successive protonations of the latter result in the formation of the, immensely stabilized, intermediate ImE. Next, ImE is converted to the carbonyl intermediate, ImF through a concerted mechanism and via a transition state, TS, surmounting an activation barrier of around 38 kcal/mol. Finally, upon second electron reduction we obtain CO while regenerating the ‘real’ catalyst, ImA.

The optimized geometries along with selected structural parameters of all intermediates and TSs involved in the Catalytic Cycle are given in Figures S7–S10. Perusal of the later reveals that the substitution of one H atom of 1 by an R substituent (Scheme 1) has no significant impact on the structural parameters of the [Pt(5-R-dpb)Cl] complexes.

3. Computational Methods

Full geometry optimization has been performed for all species under study, without symmetry constraints, using the 1997 hybrid functional of Perdew, Burke and Ernzerhof [49,50,51,52,53,54] as implemented in the program Gaussian16W [55]. This functional uses 25% exchange and 75% weighting correlation and is denoted as PBE0. Dispersion interactions were accounted by using the D3 version of Grimme dispersion with Becke−Johnson damping [56]. The Def2-TZVP basis set for all atoms was used for the geometry optimizations. The computational protocol will be hereafter denoted as PBE0-GD3BJ/Def2-TZVP. All stationary points have been identified as minima (number of imaginary frequencies, NImag = 0). Natural bond orbital (NBO) population analysis was done employing the methodology by Weinhold [47] as implemented in G16W software. The atoms in molecules (AIM) of Bader [41], the reduced gradient density (RDG) method [57] and the CDA method [48] were used as implemented in Multiwfn software [58]. The Gibbs free energy was calculated to be 298.15 K and 1 atm pressure. Solvent effects were calculated via the Polarizable Continuum Model (PCM) using the integral equation formalism variant (IEF-PCM), which is the default method of G16W (self-consistent reaction field (SCRF)) [59] while Dichloromethane, DCM was used as the solvent. Time-dependent density functional theory (TD-DFT) calculations [60,61,62] were performed on the ground-state, S₀ equilibrium geometries in DCM solvent using the CAM-B3LYP/Def2-TZVP/PCM computational protocol, taking account the first 30 excited states.

4. Conclusions

The photocatalytic reduction of CO₂ to CO by [Pt(5-R-dpb)Cl] (dpb = 1,3-di(2-pyridyl)benzene anion, R = H, N,N-dimethylaniline, thiophene, bodipy), Pt(II) square planar complexes, has been studied by means of DFT electronic structure calculations. The overall process is thought to proceed via two main steps: (i) Photoexcitation/Reduction of the initial complexes and (ii) the main catalytic cycle, producing CO. The TDDFT simulated absorption spectra showed that these complexes absorb mainly in the UV region. However, a small change in the N^C^N pincer ligand i.e. substitution in position 5 with diazaborinine shifts the absorption spectrum to the red, exhibiting a strong band at 400 nm in the visible region. Thus, it is anticipated that changes in the pincer ligand could provide complexes that absorb in the visible, making them ideal for use as photocatalysts for CO₂ to CO conversion. Next, the excited state reduction potentials, E^0*_red, dictating the reductive quenching of the Pt(II) complexes in their T₁ state, have been calculated by DFT. Based upon E^0*_red, values we conclude that the oxidizing ability of the Pt(II) complexes in their T₁ state follows the series 1 > 3 > 2 meaning that 1 is expected to be more easily reduced amongst all complexes under study. The second step of the photocatalytic CO₂ to CO conversion starts with a three coordinated Pt(II), ImA intermediate which upon one e- reduction captures CO₂ and after successive protonations and reductio yields CO. The Pt-CO₂ bond of ImC intermediate, being crucial for the hole process, has been scrutinized. It is revealed that this bond is relatively strong with D₀ in the range -36.9—-10.3 kcal/mol, exhibiting a complex nature comprising mainly covalent and hyperconjugative interaction which compensate the repulsive electrostatic interactions. The Pt-CO₂ bond involves also H bonding with the N^C^N pincer ligand. Overall, the catalytic cycle is estimated to be strongly exergonic process. Taking into account that 1 exhibits a very high T₁ excited state lifetime, τ, a whole new series of Pt(II) complexes could be synthesized bearing suitable pincer ligands in order to absorb in the visible, making them ideal for the CO₂ to CO conversion using sunlight.