Leandro Rezende
Franco
*ab,
Kalil Cristhian Figueiredo
Toledo
c,
Tiago Araujo
Matias
c,
C. Moyses
Araujo
bd,
Koiti
Araki
c and
Kaline
Coutinho
a
aInstituto de Fisica, Universidade de Sao Paulo, Cidade Universitaria, 05508-090 Sao Paulo, SP, Brazil
bDepartment of Engineering and Physics, Karlstad University, 65188 Karlstad, Sweden. E-mail: leandro.franco@kau.se
cInstituto de Quimica, Universidade de Sao Paulo, Av. Lineu Prestes 748, Butanta, 05508-000 Sao Paulo, SP, Brazil
dMaterials Theory Division, Department of Physics and Astronomy, Ångström Laboratory, Uppsala University, 75120 Uppsala, Sweden
First published on 22nd August 2023
Mononuclear polypyridine ruthenium (Ru) complexes can catalyze various reactions, including water splitting, and can also serve as photosensitizers in solar cells. Despite recent progress in their synthesis, accurately modeling their physicochemical properties, particularly in solution, remains challenging. Herein, we conduct a theoretical investigation of the structural and electronic properties of a mononuclear Ru-aqua polypyridine complex in aqueous solution, considering five of its possible oxidation/protonation states species: [RuII(H2O)(py)(bpy)2]2+, [RuII(OH)(py)(bpy)2]+, [RuIII(H2O)(py)(bpy)2]3+, [RuIII(OH)(py)(bpy)2]2+ and [RuIV(O)(py)(bpy)2]2+, where py = pyridine and bpy = 2,2′-bipyridine. At first, we investigate the impact of proton-coupled and non-coupled electron transfer reactions on the geometry and electronic structure of the complexes in vacuum and in solution, using an implicit solvent model. Then, using a sequential multiscale approach that combines quantum mechanics and molecular mechanics (S-QM/MM), we examine the explicit solvent effects on the electronic excitations of the complexes, and compare them with the experimental results. The complexes were synthesized, and their absorption spectra measured in aqueous solution. To accurately describe the QM interactions between the metal center and the aqueous ligand in the MM simulations, we developed new force field parameters for the Ru atom. We analyze the solvent structure around the complexes and account for its explicit influence on the polarization and electronic excitations of the complexes. Notably, accounting for the explicit solvent polarization effects of the first solvation shells is essential to correctly describe the energy of the electronic transitions, and the explicit treatment of the hydrogen bonds at the QM level in the excitation calculations improves the accuracy of the description of the metal-to-ligand charge-transfer bands. Transition density matrix analysis is used to characterize all electronic transitions in the visible and ultraviolet ranges according to their charge-transfer (CT) character. This study elucidates the electronic structure of those ruthenium polypyridyl complexes in aqueous solution and underscores the importance of precisely describing solvent effects, which can be achieved employing the S-QM/MM method.
In 1982, It was reported the first complex capable of carrying out the oxidation of water, a water oxidation catalyst (WOC).17 This complex, known as Blue Dimer, cis,cis-[(bpy)2(H2O)Ru(μ-O)Ru(H2O)(bpy)2]4+ where bpy = 2,2′-bipyridine, is an oxo-ligand bridged di-nuclear Ru-complex. This catalyst inspired the next developments as demonstrated by the hegemony of WOCs with two metal centers until 2005, when Zong and Thummel demonstrated that mononuclear complexes can exhibit higher electrocatalytic activity,18 whose mechanism was elucidated in 2008 using the [Ru(H2O)(bpm)(tpy)]2+ (bpm = 2,2′-bipyrimidine and tpy = 2,2′;6′,2′′-terpyridine) complex as catalyst.19 The catalytic active species was demonstrated to be the high valence [RuV(O)(bpm)(tpy)]3+ species, generated by reaction with a CeIV oxidizing agent (cerium ammonium nitrate, CAN), where the O–O bond was formed by water nucleophilic attack (WNA) to the RuV = O moiety.20 However, following studies carried out with the [Ru(pic)(bda)] (bda2− = 2,2′-bipyridine-6,6′-dicarboxylate and pic = 4-picoline) complex family indicated that the water oxidation reaction mechanism activated complex was in fact binuclear. In other words, a catalytic active intermediate should be formed in situ by the interaction of two M-O units (I2M).21 In addition, bulky axial ligands such as isoquinoline and phthalazine were very effective in enhancing the high catalytic activity of some of those compounds.21 Currently, there are several mononuclear catalysts able to mediate the four-electron oxidation of water to dioxygen.22,23
Considered one of the precursors in the application of transition metal complexes in artificial photosynthesis, the cis-[RuII(H2O)(py)(bpy)2]2+ complex24 (where py = pyridine, see Fig. 1), prepared in the early 1980s, was one of the first polypyridine complexes capable of two sequential proton-coupled electron-transfer processes generating a high oxidation state species, at a potential as low as 0.95 V.24 More recently, in 2019, this complex was adopted as a electrocatalyst model, in a study by X-ray photoelectron spectroscopy (XPS) in connection with the liquid-jet approach, aiming the characterization of the relevant intermediate species involved in the electrocatalytic water splitting process,25 with minimal interference on the reaction mechanism. Herein, we focus on the investigation of the electronic and optoelectronic properties of five different oxidation/protonation species of this complex, namely: [RuII(H2O)(py)(bpy)2]2+, [RuII(OH)(py)(bpy)2]+, [RuIII(H2O)(py)(bpy)2]3+, [RuIII(OH)(py)(bpy)2]2+ and [RuIV(O)(py)(bpy)2]2+.
In the last decades, the development of new reliable and efficient Ru compounds and the understanding of the mechanism of the electrocatalytic water cleavage reaction have undoubtedly been benefited by classical molecular dynamics,26–28 but the recent advances in computational chemistry also, such as quantum molecular dynamics simulations with the Density Functional Theory (DFT),29–36 and hybrid simulations combining molecular dynamics and DFT.37–42 However, implementing the classical atomistic simulations, that should be a powerful tool for researchers to access the structure and dynamics of Ru-complexes, still is a challenging task. In fact, the presence of a transition metal such as Ru still represents a challenge to the parametrization of the bonded and non-bonded interactions in the complexes, although it is well established that classical molecular simulation is able to deal with large organic systems in the microsecond scale.43–45 All parameters in a force field are highly sensitive to the oxidation and spin state of the metal, to the nature of the ligands coordinated to the metal center, and to the effects of microsolvation in potentially proton donor or acceptor regions; etc.46–50 Thus, defining the force field of the metal and the complex represent the major challenges given the high variability on the charge-transfer and degree of delocalization dependent on solvent effects, as well as the covalent and ionic character of the metal-ligand interactions.
In fact, a fine theoretical description of such systems has been precluded by the difficulties in parameterization and modeling due to the strong intramolecular covalent interactions as well as solvent effects, especially in the case of aqua complexes prone to strong solvent effects and proton-coupled electron-transfer reactions in protic solvents, especially in aqueous solution. In this work, we present a theoretical/experimental study about the solvent effects on the structural and electronic properties of the five different species obtained by successive proton-coupled redox processes of [RuII(H2O)(py)(bpy)2]2+ complex in aqueous solution. The effects of the solvent and oxidation/deprotonation reactions (namely ET = electron-transfer, PT = proton-transfer, and PCET = proton-coupled electron-transfer reactions) on the structural conformation, electronic density distribution and the frontier molecular orbitals, as well as their electronic transitions in the UV-visible range, were investigated. Such study was carried out adjusting the force field (FF) parameters of the metal center using QM-based calculations of the complexes’ interaction potential, making possible a more quantitative evaluation of the solvent structure around and the key role of the intermolecular hydrogen bonds formed by the coordinated water ligand derived species (OH− and O2−) and free water molecules for a more precise description of their structural, electronic and optical properties.
Quantum mechanics (QM) calculations were carried out within the Density Functional Theory (DFT)56,57 framework using the hybrid B3LYP functional58,59 in the restricted Kohn–Sham formalism for the closed-shell systems, [RuII(H2O)(py)(bpy)2]2+ and [RuII(OH)(py)(bpy)2]+, and in the unrestricted Kohn–Sham formalism for the open-shell doublets [RuIII(H2O)(py)(bpy)2]3+ and [RuIII(OH)(py)(bpy)2]2+, and triplet [RuIV(O)(py)(bpy)2]2+ systems. Other spin densities were addressed before as being less energetically stable.25 The basis set aug-cc-pVDZ60,61 was adopted for C, O, N and H atoms and the basis set with pseudopotential aug-cc-pVDZ-PP62 was adopted for Ru. This basis set combination was denominated (aug)-cc-pVDZ-(PP-Ru). The molecular structures of the complexes were obtained from geometry optimization processes and their dynamical stability confirmed by checking the vibrational frequencies calculated at the same level of theory.
The solvent distribution around the complexes was simulated using the Monte Carlo method considering a Ru-complex surrounded by 4000 water molecules in a cubic box, with an initial side length of about 50 Å. During the simulations, the intramolecular conformation of the solute was held fixed in its optimized geometry, except for the aqua ligand derived species interacting with a water molecule, which was simulated as any other solvent molecule. The isothermal–isobaric (NPT) ensemble was employed at room temperature (298.15 K) and pressure (1 atm). The simulations consisted of a thermalization stage of 8 × 108 Monte Carlo steps, followed by an equilibrium sampling of 16 × 108 Monte Carlo steps.
The interatomic interactions were modeled by the Lennard-Jones (LJ) and Coulomb potentials. The LJ parameters for N, C and H atoms were extracted from the OPLS-AA force field63 using the pyridine molecule and the TIP3P model64 was adopted for water molecules. The Ru LJ parameters were tuned to correctly describe the QM energy profile of Ru and the coordinated water molecule in the aqua complexes, prioritizing the equilibrium distance and the binding energy. The potential energy profile of the aqua complexes was obtained from a QM energy scan calculation varying the distance of the coordinated water molecule and the Ru ion around the equilibrium position, in 15 equally separated distances from 1.8 to 3.3 Å. The Counter Poise Correction65,66 was adopted to correct any basis set superposition errors. Herein, the QM calculations were realized at the same theory level adopted in the geometry optimizations of all Ru-complex species, as described above. The use of the restricted or unrestricted Kohn–Sham formalisms to calculate the dissociation of the coordinated water molecule in RuIIH2O was investigated (see Tables S5 and S6, ESI†), and very similar energy profiles were obtained with both formalisms.
The atomic charges of the Coulomb term were obtained using the fit of the QM electrostatic potential of the Ru-complex with the CHELPG (Charges from the Electrostatic Potential on a Grid) procedure.67 The polarization of the Ru-complex due to the aqueous solution was included in the QM calculation by the Polarizable Continuum Model (PCM).68 This procedure for evaluating the atomic charges of the polarized solute in aqueous solution has been shown to be better than the standard procedure based on atomic charges calculations of the solute in vacuum at the QM level suggested by the common force fields, i.e. HF/6-31G(d).69–72 Standard procedures such as Metropolis sampling technique, image method, periodic boundary condition, cut-off-radius, and long-range correction73 were adopted as before.74 All interactions were explicitly computed considering a cut-off-radius about 22 Å.
Furthermore, a sequential multi-scale approach combining quantum mechanics and molecular mechanics (S-QM/MM)51,75,76 was utilized to obtain the theoretical electronic spectra of all five ruthenium complex species in solution. After performing the Monte Carlo simulation, a total of 100 statistically uncorrelated Monte Carlo configurations were selected. Then, the Time Dependent Density Functional Theory (TD-DFT) was used to calculate the first 50 electronic transitions of the complexes in each Monte Carlo configuration, with solvent effects treated at different levels of approximation. The solvent effects were accounted under three different approximations: (i) as an electrostatic embedding composed of the atomic point charges of the 1000 closest solvent molecules (X + PC); (ii), and as before but considering the water molecules that makes hydrogen bonds with the complex as part of the QM region (X + HB + PC); and (iii) treating the solvent as a continuous dielectric with the Polarizable Continuum Model (X + PCM).
The QM calculations were performed using the Gaussian 09 package.77 The Monte Carlo simulations were performed with the DICE program.78 The electronic excitations were assigned based on the analyzes of the corresponding transition density matrices, using the Theoretical Density, Orbital Relaxation, and Exciton analysis package (TheoDORE).79–82 In order to improve clarity and facilitate the reading, now on a simplified nomenclature will be adopted for the complexes. The closed-shell systems will be identified as [RuII(H2O)]2+ and [RuII(OH)]+, and the open-shell systems as [RuIII(H2O)]3+, [RuIII(OH)]2+ and [RuIV(O)]2+.
In order to compare the calculated spectra with the experimental results, all Ru-complex species were generated upon oxidation and/or deprotonation reactions of the [RuII(H2O)]2+ complex, and the Ultraviolet-Visible spectra were measured in the laboratory, as described in the ESI.†
Polypyridine ligands confer characteristic spectroscopic and redox properties to the Ru-complex species, that can also be influenced by environment conditions. From vacuum to aqueous solution, the bond length changes induced by the implicit solvent are not so significant. For example, we observed the reduction of all Ru–N bond lengths of all Ru-complex species and reduction (increase) of the Ru–O bond length in those with a coordinated water (hydroxide or oxide) ligand, however the changes are smaller than 0.03 Å. Dihedral angles involving the pyridine or any of the water derived ligands and the metal ion are the most affected by the solvent, undergoing changes from 8 to 53 degrees. In contrast, dihedral angles involving the bipyridine and the metal ion changed at most 3 degrees, thus demonstrating their higher structural stability and low sensitivity to solvent effects (Detailed information is available in ESI†). Fig. 2 shows the superposition of the optimized geometries of all Ru-complex species in vacuum and in aqueous solution, as well as the root mean square deviation (RMSD) values of the vacuum (VAC) and solution (SOL) geometries. The RMSDVAC/SOL is 0.16, 0.14, 0.11, 0.09 and 0.01 for [RuIII(H2O)]3+, [RuII(OH)]+, [RuIV(O)]2+, [RuIII(OH)]2+ and [RuII(H2O)]2+, respectively. Therefore, on average, the [RuII(H2O)]2+ complex undergoes the smallest conformational changes due to solvent effects, while the species with the highest total charge, [RuIII(H2O)]3+, undergoes the largest conformational changes. On the other hand, the less charged species [RuII(OH)]+ is almost as sensitive to the solvent as the most charged one, being a counterintuitive result considering the major contribution of the solute-solvent electrostatic interactions. This demonstrates that the extent of structural changes induced by solvent effects may not be only correlated to their total charges, but also to specific solute/solvent interactions, as addressed later on.
The oxidation of the metal ion always increases its electron withdrawing character inducing an increase of the acidity of coordinated water or hydroxo ligand leading to their deprotonation. Accordingly, the oxidized/deprotonated Ru-complex species can be generated by reaction with a strong enough oxidizing agent by mechanisms limited by an electron-transfer step, that can be followed or preceded by acid–base equilibrium, more specifically a proton transfer (PT), or a proton-coupled electron-transfer (PCET) mechanism. Such reactions cause structural changes in the Ru-complex species, but the most relevant ones take place in the coordination sphere of the metal complex. Among the oxidized/deprotonated species, the bond lengths of Ru with the three bipyridine N-atoms (N3, N19 and N25) varied from 2.060 to 2.134 Å in vacuum and from 2.058 to 2.106 Å in aqueous solution. The Ru to pyridine N-atom (N45) bond length is slightly longer, and varied from 2.150 to 2.163 Å in vacuum and from 2.140 to 2.159 Å in aqueous solution. The Ru–N11 and Ru–O bonds are the most affected by the oxidation/deprotonation reactions but exhibiting opposite behavior. Whether in vacuum or in solution, the Ru–O bond is longer in [RuII(H2O)]2+ and gradually was shortened along the sequence [RuIII(H2O)]3+, [RuII(OH)]+, [RuIII(OH)]2+ and [RuIV(O)]2+. On the other hand, following this trend, the Ru–N11 bond length was gradually increased (see Fig. 3) as expected for a significant Trans Influence.87
Additionally, the angles and dihedrals associated with the bipyridine ligands and the metal center are less sensitive to PT, ET, or PCET reactions, demonstrating once again the structural stability conferred to the Ru-complex species by these ligands. On the other hand, the pyridine and the water derived ligands are very sensitive to those types of reactions, that can induce dihedral rotation changes of up to 50 degrees. Some of such changes can be seen in Fig. 2, and in more details in the ESI.†
Effects of PT, ET, and PCET reactions can also be observed on the charge distribution of the Ru-complex species, see Fig. 4. The mono-oxidation of [RuII(H2O)]2+ in solution causes variations of +0.36 a.u. in the metal center, +0.61 a.u. in the polypyridine ligands (bpy1 + bpy2 + py) and +0.03 a.u. in the ligand water molecule electronic density. For [RuII(OH)]+ the variations are +0.39, +0.37 and +0.24 a.u., respectively. Therefore, less than 40% of the electronic density lost in the monoelectronic oxidation of these Ru-complex species comes from their metal center. On the other hand, the mono-deprotonation of [RuII(H2O)]2+ (or [RuIII(H2O)]3+) in solution causes variations of −0.61 (−0.58), +0.35 (+0.11) and −0.74 (−0.53) a.u. in the charge densities on the metal center, polypyridine (bpy1 + bpy2 + py) and water derived ligands, respectively. Therefore, the loss of 1 proton leads to changes in the electronic distribution which depend on the oxidation state of the Ru-complex species, where the greatest variations are observed in the metal center and in the water derived ligands. Notably, when the [RuII(H2O)]2+ (or [RuIII(H2O)]3+) complex is deprotonated in solution, the Ru charge density decreases from 0.94 to 0.33 a.u. (1.30 to 0.72 a.u.).
Furthermore, the charge variations due to the oxidation, or deprotonation, reactions can also drastically depend on the solvent effects. For example, the mono-deprotonation of [RuIII(H2O)]3+ in vacuum causes a variation of −0.59 a.u. in the total charge density on the polypyridine ligands, which represents the largest shift in the opposite direction (by +0.11 a.u.) as compared with that observed in aqueous solution. This is a direct consequence of the electron density shift due to polarization effect of the medium, since the polypyridine ligands behave as electron density acceptor or donor in the aqua and hydroxo complex species, respectively. In PCET reactions a compensatory movement of the electronic density is observed, and the atomic charge variations are less abrupt. For example, the mono-oxidation/deprotonation of [RuIII(H2O)]3+ to [RuIV(O)]2+ species causes variations of 0.29, −0.09, and −0.20 a.u. in the metal center, polypyridine ligands (bpy1 + bpy2 + py) and water derived ligands charge densities, respectively. Other possible PCET reactions can be seen in Fig. 4.
The deprotonation of the water molecule coordinated to the metal center leads to a remarkable reduction in the length of the Ru–O bond. This is expected since the loss of a proton enhance the electronic density on the oxygen atom and its basicity, as well as its covalent interaction with the electron deficient metal center. Similarly, the oxidation of the Ru also entails a reduction in the Ru–O distance since it leads to a decrease in the ionic radius and an increase in the positive charge on Ru, thus enhancing the ionic and the covalent interactions. Consequently, the combination of both effects (deprotonation and oxidation) induce a dramatic Ru–O bond length decrease from 2.24 Å in [RuIIH2O]2+ to 1.92 Å in the [RuIIIOH]2+ complex. This means that the associated potential/bond energy is much more negative (and the bond much stronger) upon oxidation coupled with deprotonation. This is perfectly consistent with the fact that the atomic charge densities, adjusted to represent the electrostatic potential of Ru and O atoms, becomes more separated after any of these processes. The calculated atomic charges for all Ru-complex species, in vacuum and in solution are available in ESI.†
As general trend, the FMOs energies are remarkably destabilized (it becomes less negative) by the solvent, and the magnitude of this destabilization is proportional to the total charge of the complexes: 2.32 eV for the complex with charge 1+; ranging from 5.15 to 5.31 eV for the complexes with charge 2+, and 7.96 eV for the complex with charge 3+. Such destabilizing orbitals effect can happen, for example, when the FMOs dipole moments do not have the same direction as the molecular dipole moment, thus inducing a destabilizing effect.89 Conversely, the H(or S)OMO and LUMO orbitals stability are also connected to the total charges of the Ru-complex species. In solution, the H(or S)OMO (and LUMO) energies are −5.02 (−2.44), −6.11 (−2.73), −7.12 (−2.96), −7.05 (−3.03), −7.92 (−3.37) eV for [RuII(OH)]+, [RuII(H2O)]2+, [RuIII(OH)]2+, [RuIV(O)]2+ and [RuIII(H2O)]3+, respectively. Thus, the higher the total charge of the complex, more negative tend to be the FMOs. Besides, the H(or S)OMO–LUMO energy differences in vacuum (and in solution) are 2.39 (2.59), 3.46 (3.38), 3.95 (4.02), 4.02 (4.16), and 4.48 (4.55) for [RuII(OH)]+, [RuII(H2O)]2+, [RuIV(O)]2+, [RuIII(OH)]2+ and [RuIII(H2O)]3+, respectively. Therefore, the solvent effects cause the reduction of the H(or S)OMO–LUMO energy gap of [RuII(H2O)]2+ and its increase for the other species. Similarly to H(or S)OMO and LUMO values, the energy gap also follows the total charge trend of the complexes.
Specially for closed-shell systems, the HOMO–LUMO energy gap can be associated to the system electronic excitability, resistance to charge transfer, and other properties. The chemical potential (μ), for example, characterizes the tendency of electrons to escape from the valence band and the chemical hardness (η) measures the internal charge transfer resistance of the system.90 According to the Maximum Hardness Principle, molecular systems in equilibrium must be in a state of maximum hardness90 and transition states must be in a state of minimum hardness.91 According to the definitions given by Parr and Pearson92–94 and the Koopmans theorem, μ = (εL + εH)/2 and η = (εL − εH)/2, where εL and εH are the energy of LUMO and HOMO respectively.89 For [RuII(H2O)]2+ (and [RuII(OH)]+), μ is worth −9.70 (−6.15) and −4.42 (−3.73) eV in vacuum and in solution, respectively. It means that the solvation or the abstraction of a proton makes the chemical potential less negative, facilitating the removal of electrons from the complexes, thus making easier any chemical process. These are expected, since the loss of a proton decreases the electrostatic attraction between nucleus and electron in the complexes thus decreasing the resistance to electron transfer, since electrons will need less energy to escape. Otherwise, the chemical hardness (η) for [RuII(H2O)]2+ (and [RuII(OH)]+) is 1.73 (1.69) and 1.19 (1.29) eV in vacuum and in solution, respectively. Therefore, the solvation makes the chemical hardness a little less positive for both closed-shell Ru-complex species, indicating that they have lower resistance to charge transfer processes in aqueous solution. Moreover, the increase/decrease of η upon mono-deprotonation of [RuII(H2O)]2+ in vacuum/solution demonstrates a predominant role of the solvent in determining the charge transfer resistance of the complex under PT reactions. A table containing the values of the FMOs for all Ru-complex species is available in ESI.†
Reference | ε (kcal mol−1) | σ (Å) | q (a.u.) |
---|---|---|---|
Allinger et al. (for RuII)95 from MM3 force field | 0.438 | 2.08 | — |
Adlhart et al. (for RuII)96 from sybyl/tripos 5.2 force field | 0.560 | 2.64 | — |
Rappé et al. (for RuII)97 from UFF force field | 0.056 | 2.64 | — |
Sebesta et al. (for RuII)46 | 0.418 | 2.618 | 0.872 |
Sebesta et al. (for RuIII)46 | 1.644 | 2.523 | 0.790 |
Our (for RuII) | 0.39 | 1.81 | 0.94 |
Our (for RuIII) | 0.39 | 1.99 | 1.30 |
Our (for RuIV) | 0.39 | 1.81 | 1.01 |
The work of Allinger et al.95 is quite general and comprises more than 100 atoms of the periodic table, including the Ru atom. The authors presented a broad set of LJ parameters developed for the MM3 force field. Adlhart et al.96 presented LJ parameters for Ru from a modified version of the sybyl/tripos 5.298 force field. As mentioned by the authors, the LJ parameters were developed to be used in a QM/MM study of olefin-metathesis reaction catalyzed by Ru carbene complexes. Rappé et al.97 developed the famous Universal Force Field (UFF) where the authors presented a full periodic table force field for molecular mechanic simulations, that includes LJ parameters for Ru. Sebesta et al.,46 in turn, brings up the first parameterization study for Ru considering two different oxidation states (RuII and RuIII). The authors employed the supramolecular approximation99 to adjust the LJC parameters for various transition metals considering a neutral metal complex and a “test” molecule. [M(NH3)xCly] and [M(H2O)x(SH)y] type structures were adopted as neutral complexes, where M is a transition metal and x or y indicate a variable number of ligands. As test molecules, they adopted NH3, CH4 and H2O. The non-bonded parameters were then obtained to provide the best description of the metal-ligand quantum interaction energies.
The coordination of the metal with a water molecule, a hydroxide or an oxide ion can completely modify the electronic structure of the complex. For metal complexes with a coordinated aqua ligand, it is mandatory also consider the interaction of the metal center with the water molecules in solution to properly describe the intra and intermolecular non-bonded interactions of the complexes. In this work, we propose a set of non-bonded parameters for Ru in the oxidation states II, III and IV (“Our” in Table 1), based on the accurate description of the QM energy profile of Ru and the coordinated water molecule in the polypyridyl coordination environment of the studied closed-shell Ru-complex species. Fig. 6 shows the LJC energy profile of the water ligand in the [RuII(H2O)]2+ and [RuIIIH2O]3+ complexes obtained with the LJC parameters presented in Table 1 in comparison with the QM energy profile calculated with B3LYP/(aug)-cc-pVDZ(PP-Ru). From the literature, the set of parameters proposed by Allinger et al. is the one that provides the LJC energy profiles closest to the QM ones, but with less negative minimum energy (Emin), by 4.7 kcal mol−1 for RuII, by 1.8 kcal mol−1 for RuIII; and distorted r0 by 0.2 Å for RuII, and by 0.1 Å for RuIII. Moreover, the set of parameters proposed by Sebesta et al. is the one that provides the LJC energy profiles furthermost to the QM ones, with even less negative E0 (1.9 kcal mol−1 for RuII, and 17.9 kcal mol−1 for RuIII) and even more distorted r0 (0.5 Å for RuII, and 0.7 Å for RuIII). LJC energies profiles intermediary to the LJC profiles of Allinger et al. and Sebesta et al. were obtained when the parameterizations proposed by Rappé et al. and Adlhart et al. was considered.
![]() | ||
Fig. 6 Energy profile for the Ru-complex and interaction of the aqua and hydroxo ligands in [RuII(H2O)]2+ (a) and [RuIII(H2O)]3+ (b) complexes. |
Our set of LJC parameters, obtained with the precision of two digits, gave the best agreement of the LJC and QM energy profiles, with an excellent description of the minimum energy (E0) and equilibrium distance r0 for both RuII and RuIII, reaching a deviation of only 1.4 and 0.8 kcal mol−1 for the QM minimum energies of the RuII and RuIII complexes, respectively. Also, we fitted exactly the QM minimum distance in both oxidation states. Equal ε values were obtained for RuII, RuIII and RuIV, and different σ values were obtained for RuII, RuIII and RuIV. The fitted σ value for RuIII is 0.2 Å larger than for RuII and RuIV complexes. Moreover, a more positive charge for RuIII and RuIV than RuII were obtained by fitting the electrostatic potential of the Ru-complex species using the CHELPG method. Even though the charge of RuII is expected to be more positive after a mono-oxidation, as we have obtained, the parameters proposed by Sebesta et al. do not follow this tendency, since a charge of RuIII smaller than RuII is proposed by them. Last, but not less important, as long as we have adopted B3LYP without long range dispersion corrections, the LJ parameters adjusted for Ru accurately describe only the region near the Ru–O equilibrium distance.
![]() | ||
Fig. 7 (a) Minimum density distribution function (MDDF) of the Ru-complex species in aqueous solution. (b) Snapshot of the Monte Carlo simulation showing the [RuII(H2O)]2+ complex surrounded by the solvent water molecules in the first solvation shell. (c–h) Radial distribution functions between Ru, O, and the oxygen (Ow) and hydrogen (Hw) atoms of the solvent molecules for [RuII(H2O)]2+ (c) and (d), [RuII(OH)]+ (e) and (f) and [RuIV(O)]2+(g) and (h). The characteristic distances indicating the beginning of each RDF, the position of the first peak and the position of the first valley are indicated in the graphs by the labels r0, r1 and r2, respectively. The RDFs of Ru-complex species in the 3+ oxidation state are shown in ESI.† |
In the aqua complexes, the integral of any RDFs involving Ru, Ow, and Hw, up to a distance of 3.6 Å, is exactly equal to 1, which demonstrates that during the Monte Carlo simulations only the coordinated water molecule interacts directly with the metallic center. The average distance, during the simulations, between the oxygen of the coordinated water molecule and the Ru atom was 2.3 ± 0.1 Å (r1 in g(r)Ru–Ow), for the species in the oxidation states II or III, that is slightly larger than the distances obtained from the optimized geometries in solution (2.214 Å for [RuII(H2O)]2+ and 2.148 Å for [RuIII(H2O)]3+. This is a consequence of the explicit treatment of the specific interactions in the simulations, which compete with the metallic center increasing the Ru–O bond.
In the hydroxo or oxo complexes, these moieties were treated as part of the solute (fixed in the optimized geometry relative positions), and therefore, their atoms were not counted in g(r)Ru–Ow and g(r)Ru–Hw. Concerning these complexes, the intensity of the first peak in g(r)O–Hw is smaller for [RuIII(OH)]2+, indicating that in the 3+ oxidation state, the hydroxide complex attracts less hydrogens of water molecules in solution. In other words, compared to [RuIII(OH)]2+, the [RuII(OH)]+ complex can stablish a larger number of hydrogen bonds as a proton acceptor. Lastly, in the [RuIV(O)]2+ complex, the position of the first peak in g(r)O–Hw (r1 = 1.8 Å) is prior to the position of the first peak in g(r)O–Ow (r1 = 2.8 Å). This is an expected result since the oxide ion has a negative charge that necessarily attracts protons from the solution, making hydrogen bonds as a proton acceptor. In any complex, hydrogen bonds are stablished only between the coordinated aqua, hydroxo or oxo ligand and water molecules in solution, never involving the metallic center directly.
To account for the number of hydrogen bonds (NHB) formed by the Ru-complex species, the criteria r < 3.5 Å and Θ < 40° were adopted for the maximum distance between electronegative atoms X and Y participating in the hydrogen bond, and the XYH angle of formation. Additionally, an energetic criterion was adopted according to which only hydrogen bonds with an interaction energy (EHB) of at least −0.01 kcal mol−1 are computed. Two types of hydrogen bonds were observed, the O–H⋯Ow type, where the Ru-complex donates a proton to a water molecule in solution and the O⋯Hw–Ow type, where the Ru-complex accept a proton from a water molecule in solution, see Fig. 8. The average number (N), distance (r), angle (θ) and energy (EHB) of the accounted hydrogen bonds are listed in Table 2. Histograms showing the distributions of r, θ and EHB are presented in ESI.†
O–H⋯Ow | O⋯Hw–Ow | |||||||
---|---|---|---|---|---|---|---|---|
N | r (Å) | θ (degree) | E (kcal mol−1) | N | r (Å) | θ (degree) | E (kcal mol−1) | |
[RuII(H2O)]2+ | 2 | 2.9 ± 0.2 | 14.9 ± 8.8 | −9.2 ± 3.0 | 0.03 | 3.1 ± 0.2 | 20.5 ± 11.1 | −0.9 ± 0.7 |
[RuIII(H2O)]3+ | 2.1 | 2.8 ± 0.2 | 14 ± 8.8 | −13.5 ± 3.4 | 0.01 | 3.3 ± 0.1 | 33.9 ± 5.3 | −2.5 ± 1.7 |
[RuII(OH)]+ | 0.4 | 3.2 ± 0.2 | 15.2 ± 9.1 | −1.9 ± 1.2 | 2 | 2.8 ± 0.2 | 11.6 ± 6.9 | −7.9 ± 2.1 |
[RuIII(OH)]2+ | 1 | 2.9 ± 0.2 | 10.4 ± 6.3 | −8.0 ± 2.3 | 1.3 | 2.9 ± 0.2 | 13.8 ± 8.0 | −4.3 ± 1.9 |
[RuIV(O)]2+ | 0 | 0 | 0 | 0 | 1.1 | 2.8 ± 0.2 | 15.7 ± 9.0 | −3.7 ± 2.2 |
In the Ru-complex species with a coordinated water molecule, the O–H⋯Ow hydrogen bonds are preponderant, whereas hydrogen bonds, in which the Ru-complex acts as proton acceptor, is a very rare event (NHB < 0.05). On the other hand, Ru-complex species with a coordinated hydroxide can make hydrogen bonds both as proton donor and acceptor as well. The [RuII(OH)]+ complex makes on average 0.4 hydrogen bonds as a donor and 2 hydrogen bonds as an acceptor. The [RuIII(OH)]2+ complex makes on average 1 hydrogen bond as a donor and another 1.3 hydrogen bonds as an acceptor. Therefore, the O⋯Hw–Ow hydrogen bonds are preferred by the hydroxo complexes. Finally, the [RuIV(O)]2+ complex makes on average 1.1 hydrogen bonds only as an acceptor.
The distribution of hydrogen bond distances is generally unimodal (see histograms in ESI†), ranging from approximately 2.4 to 3.5 Å, with mean values in the 2.8 and 2.9 Å range. The distribution of bond angles also has unimodal character, ranging from 0 to 40 degrees and with average values in the range of 10.4 and 33.9 degrees. The energy distributions are unimodal for the [RuII(H2O)]2+, [RuIII(H2O)]3+ and [RuIV(O)]2+ complexes, but bimodal for [RuII(OH)]+ and [RuIII(OH)]2+ species. This behavior can be assigned to the two types of hydrogen bonds (O–H⋯Ow, and O⋯Hw–Ow) realized by these Ru-complex species, which have different energy distributions. The [RuIII(H2O)]3+ complex establishes the most intense hydrogen bonds with the solvent, with an average energy of −13.5 kcal mol−1, followed by [RuII(H2O)]2+, with an average energy of −9.2 kcal mol−1. The [RuII(OH)]+ and [RuIII(OH)]2+ complexes establish hydrogen bonds with an average energy of −1.9 and −8.0 kcal mol−1 (OH⋯Ow) and −7.9 and −4.3 kcal mol−1 (O⋯Hw–Ow), respectively. The weakest hydrogen bonds are realized by the [RuIV(O)]2+ complex, with an average energy of −3.7 kcal mol−1. All values with respective standard deviations are listed in Table 2.
In general, the Coulomb interaction plays a fundamental role in the formation of hydrogen bonds in solution. Small shifts of the atomic charges in a molecule can completely change the characteristics of the hydrogen bonds formed.101,102 The H-bonds made by [RuII(H2O)]2+ are on average 70% less intense than that realized by the [RuIII(H2O)]3+ species. This behavior is justified by the higher total charge and smaller size of the complex in the 3+ oxidation state. Considering hydroxo complexes, for example, the differences in the number and average energy of OH⋯Ow and O⋯Hw–Ow type hydrogen bonds are mainly related to the atomic charge densities on the oxygen and hydrogen atoms of the hydroxo ligand. The atomic charges on the oxygen atom are −0.89 and −0.76, and on the hydrogen atom are 0.29 and 0.39, for [RuII(OH)]+ and [RuIII(OH)]2+, respectively. Notably, the charge on the O-atom is more negative in the 2+ oxidation state thus attracting more protons from solution, and the charge on the H-atom is larger in the 3+ oxidation state, thus attracting more strongly the O-atom of the water molecules in solution. Furthermore, the number of H-bonds and the modulus of the O⋯Hw–Ow interaction energy by the hydroxo and oxo complexes increases as a function of the O-atom negative charge, which is −0.57, −0.76, and −0.89 for [RuIV(O)]2+, [RuIII(OH)]2+, and [RuII(OH)]+, respectively. Furthermore, we will see in the next section that H-bonds play an important role in defining the electronic excitation energies of the complexes.
In order to assign the measured UV-Vis absorption spectra, the electronic excitations of the Ru-complex species were calculated using the TD-DFT theory with the B3LYP functional and the (aug)-cc-pVDZ(PP-Ru) basis sets. In preliminary tests (ESI†) it was demonstrated that this combination of methods and basis sets provides the best theoretical/experimental spectral matching with one of the lowest computational cost. In addition to B3LYP, the following methods were also tested: single-excitation CI, M06-2X, BHANDH, CAM-B3LYP, LC-ωPBE, and PBE0. Among these, the B3LYP and PBE0 methods showed the best performance, but with PBE0 energy values slightly overestimated. The following basis set types were considered: (i) all-electron basis sets for non-metallic atoms, combined with basis sets with effective core potential for Ru; (ii) all-electron basis sets for metallic and non-metallic atoms; and (iii) all-electron basis sets with Douglas-Kroll-Hess type zero-order relativistic correction. Interestingly, the inclusion of relativistic effects via effective potential was shown to be adequate to calculate the electronic transition energies of our Ru-complex species. Furthermore, as shown in Fig. 9, the solvent effects were investigated using both the traditional Polarizable Continuum Model (PCM)68 and the S-QM/MM approach. Two approximations have been considered in the last case, namely: (i) treating the solvent molecules as point charges (PC approximation) and (ii) treating the solvent molecules that make hydrogen bonds with the Ru-complex species by quantum mechanics and the remaining as point charges (HB + PC approximation). The spectra of [RuII(H2O)]2+ and [RuII(OH)]+ calculated in vacuum and in solution, using the three solvent effects approximation levels (see ESI† for other complexes), are shown in Fig. 9:
The energies and intensities of the calculated electronic excitations in solution are in good agreement with the experimental data, where the transitions in the visible range showed to be more sensitive to the solvent effects. From the convolution of the calculated electronic excitations with Lorentzian functions with 0.25 eV of width at half-height, the following theoretical values were obtained for the longer wavelength absorption peak of [RuII(H2O)]2+ (and [RuII(OH)]+) species: 432 (520) nm in vacuum and 444 (510), 442 (482) and 456 (491) nm in solution using the PCM, PC and HB + PC approximations, respectively. Comparing with the experimental values of this absorption band (472 nm for [RuII(H2O)]2+ and 506 nm for [RuII(OH)]+), the inclusion of solvent effects and the improvement of their description by considering explicitly the solvent molecules as point charges (PC approximation) and additional inclusion of a quantum mechanical treatment of the solvent molecules that realize hydrogen bonds with the complexes (HB + PC), contributed to a more accurate description of the UV-Vis spectra of these complexes.
Previous studies have shown the importance of the vibrational motion of ligands and solvent for understanding and modelling the optical properties and dynamics of Ru-complex species with bpy ligands.104–109 The necessity of including at least the first solvation shell in the QM calculations has also been emphasized for the calculation of electronic properties of Ru complexes.106,110 In our work, especially for [RuII(H2O)]2+, we have also investigated the impact of including a larger number of explicit solvent molecules, going beyond hydrogen bonds. We took into account the complete first solvation shell when calculating the first five electronic excitations, albeit considering only five Monte Carlo configurations. Fig. S10 (ESI†) illustrates that the incorporation of the first solvation shell does not significantly improve the agreement with experimental results in terms of the energy of the excitations. Additionally, regardless of whether we employed the B3LYP or PBE0 functionals, our conclusions remained consistent. However, it is worth noting that the B3LYP functional exhibited a closer alignment with experimental observations. Therefore, treating only the hydrogen bonds at the quantum mechanical (QM) level and employing the electrostatic embedding approach to account for the remaining solvent molecules represents a reasonable compromise to describe and characterize the electronic spectra of the studied Ru-aqua complexes in aqueous solution.
In order to advance in the understanding of these UV-Vis absorption spectra, the assignment of the calculated electronic excitations was realized according to their spatial distribution on the molecular structure of the Ru-complex species (Fig. 10f–j). Further quantitative assessment was obtained by analyzing the transition density matrices from the TD-DFT calculations using a fragment-based analysis, as implemented in the Theoretical Density, Orbital Relaxation, and Exciton analysis (TheoDORE) package.79–82 The complexes were fragmented into 5 groups of atoms (bpy1, bpy2, py, aqueous group and the Ru atom) and the transitions were quantitatively assigned based on their MLCT, MC, LMCT, LLCT, and LC contributions. See Fig. 10(f)–(j).
For [RuII(H2O)]2+ and [RuII(OH)]+, the absorption bands with peaks at 472, 336 nm and 506, 366 nm, respectively, are mainly composed by MLCT-type transitions (about 60%), and in lesser degree by LLCT and LC transitions (∼30%). Also, there is an increase of the LLCT and LC contributions in the UV region below 300 nm, that goes up to 56%, and a reduction of the MLCT transitions to 30%, indicating that the UV band around 290 nm is mainly composed by transitions centered in the pyridyl ligands. The oxidation of the [RuII(H2O)]2+ species to [RuIII(H2O)]3+ causes a drastic reduction of the contributions of the MLCT transitions down to 16% since the Ru(III) ion is a much stronger electron acceptor than Ru(II), thus increasing the LMCT contribution to 32%, considering the first 30 transitions. The MLCT contributions dropped to 18% around 300 nm while the LC and LLCT contributions were enhanced to almost 70%.
In [RuIII(OH)]2+, the MCLT contributions (34%) play an important role until up to 300 nm, competing with the LC and LLCT contributions (up to 35%), and the LMCT contributions (15%). In the 300 nm region, the MLCT contributions were estimated in 24% while the LC, LLCT and LMCT contributions added up to 71%. In the case of the [RuIV(O)]2+ complex, the LC and LLCT contributions corresponded to 47% of the first 30 transitions, and this number rose to 69% when the last 20 transitions were considered. In contrast, the MLCTs contributed with 25% for the first 30 transitions and only 21% for the last 20 transitions. The LMCT contributions corresponded to 17% of the contributions in the first set and only 7% in the second set of electronic transitions. Thus, the [RuIV(O)]2+ complex has the second highest percentage of LMCT contributions among the first 30 transitions, second only to the [RuIII(H2O)]3+ complex. In general, the contribution of MC transitions is relatively low among all type of transitions in any of those Ru-complex species, being less than 16% of the total.
In more detailed contribution maps (see ESI†), it is possible to observe that the MLCT contributions involve predominantly the transfer of electron density from the metal to the bipyridine ligands (bpy1 and bpy2), and to a lesser extent, from the metal to the pyridine ligand. In contrast, in the LMCT contributions, the electronic density is transferred predominantly from the bipyridine ligands to the ruthenium ion. There are also LMCT contributions involving the pyridine, the bipyridines and the water derived ligands in a minor extent. On the other hand, the bipyridine ligands concentrate almost all of the LLCT contributions. Specifically, in the [RuII(OH)]+ and [RuIV(O)]2+ complexes, it is also observed that the water derived ligands donates a small portion of its electronic density to the other ligands. Over the entire calculated spectrum, the OH-LCT and O-LCT contributions were both computed to be around 20%, as detailed in the ESI.†
Our findings reveal that the solvation, oxidation, and deprotonation of the [Ru–OH2]2+ species lead to structural changes primarily in the Ru coordination sphere, with the distances and angles between the metal center and the water derived ligands or pyridine ligand atoms being the most sensitive. Global geometric changes resulting from solvation do not correlate with the total charge of the ruthenium complex species, as evidenced by the shortening of the Ru–N and Ru–O bond distances in the aqua complexes and lengthening in hydroxo and oxo complexes. Conversely, the oxidation of the Ru(II) complexes increases the Ru–N bond lengths, especially the bond opposite to the O-atom, whereas shortening the Ru–O bond length, which is more pronounced in hydroxo complexes. In contrast, deprotonation of the aqua/hydroxo ligand consistently leads to a shortening of the Ru–O bond and elongation/contraction of the opposite Ru–N bond, with the pyridine ligand being less sensitive to such effects than to the oxidation of the metal center. Notably, the bipyridine ligands remained insensitive to electron or proton-transfer processes.
Alongside geometric changes, oxidation and deprotonation processes have opposite effects on the electron density at the metal site. The conversion of the aqua into the hydroxo ligand and then into the oxo ligand lead to the transfer of electronic density and decrease of the positive charge on Ru-atom, whereas its oxidation tends to increase it. The polarization of the Ru-complex species due to the presence of the aqueous solution leads to larger variations in the electronic density on the metal center and water-derived ligands. Among all of the atoms in the [Ru(OH2)]2+ complex, the Ru atom is the most sensitive to oxidation and deprotonation processes, undergoing the largest changes in the electronic density distribution.
The frontier molecular orbitals of the complexes, HOMO or SOMO and LUMO have predominantly d(metal) and π*(bipyridines) orbital character, respectively. This implies that the electronic transitions between these orbitals have metal-to-ligand charge-transfer (MLCT) character. Notably, the solvent tend to destabilize the frontier molecular orbitals relative to vacuum as the total charge of the complex increases. Furthermore, analysis of the effect of solvation, oxidation, and deprotonation processes on the HOMO–LUMO energy gap suggests that the migration of electronic density is lower in the more highly charged complexes, but tends to increase when the aqua ligand is successively deprotonated to the hydroxo and oxo ligands.
The solvation shell surrounding the complexes in solution displays similarity, with differences only found in the microsolvation region where intermolecular hydrogen bonding plays a significant role. In the aqua complexes, the Ru–O distance is primarily influenced by hydrogen bonding rather than the oxidation state of the complex. Meanwhile, the hydroxo complexes form a higher number of hydrogen bonds compared to the other species, with the strongest hydrogen bond being present in the complex with the highest total charge.
The UV-Vis absorption spectra of the [Ru(OH2)]2+ complex and its oxidation/deprotonation species were simulated using the B3LYP/(aug)-cc-pVDZ/(PP-Ru) theory level, which provided the best description at the lowest computational cost. The UV-Vis spectra of the Ru-complex species were found to be more sensitive to solvent effects in the visible region, where closed-shell species exhibit broad MLCT absorption bands and solution pH-sensitive electronic transitions, wich lose intensity or disappear when the complex is oxidized, while LC and LLCT bands prevail in the ultraviolet region. The MLCT bands in the visible have a predominant contribution of metal-to-bipyridine charge-transfer transitions and are best described when the complex/solvent hydrogen bonds are treated at the QM level and the other water molecules are treated as point charges. The theoretical and experimental UV spectra were found to be in excellent agreement, validating the LJC parameters developed for the Ru-atom and the S-QM/MM protocol employed in this work.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d3cp02154h |
This journal is © the Owner Societies 2023 |