Independent tuning of the pK_a or the E_1/2 in a family of ruthenium pyridine-imidazole complexes

Abstract

Two series of Ru^II(acac)₂(py-imH) complexes have been prepared, one with changes to the acac ligands and the other with substitutions to the imidazole. The PCET thermochemistry of the complexes has been studied in acetonitrile, revealing that the acac substitutions almost exclusively affect the redox potentials of the complex (|ΔE_1/2| >> |ΔpK_a|•0.059 V), while the changes to the imidazole primarily affect its acidity (|ΔpK_a|•0.059 V ≫ |ΔE_1/2|). This decoupling is supported by DFT calculations, which show that the acac substitutions primarily affect the Ru-centered t_2g orbitals, while changes to the py-imH ligand primarily affect the ligand-centered π-orbitals. More broadly, the decoupling stems from the physical separation of the electron and proton within the complex and highlights a clear design strategy to separately tune the redox and acid/basic properties of H atom donor/acceptor molecules.

Graphical Abstract

Two series of bis(acetylacetonate) ruthenium(II) pyridine-imidazole complexes have been prepared that show independent modulation of the imidazole pK_a and the ruthenium reduction potential E_1/2. This unusual decoupling of the pK_a and E_1/2 values are accomplished using different substitution patterns. The decoupling stems from differences in the orbitals affected by each substitution pattern and highlights a promising strategy to tune the individual components of PCET systems.

Introduction

Many chemical species react by proton-coupled electron transfer (PCET) and hydrogen atom transfer (HAT).^{1, 2} This includes free radical intermediates, metal complexes, and active sites in enzymes and on surfaces. Such reactions are important in a variety of societal, biological, and industrial processes, including combustion, oxidations in the atmosphere and natural waters, the trapping of reactive oxygen species in cells, and complex organic syntheses.^{1, 3–5}

The reactivity of a PCET reagent X/XH rests on its thermochemical properties, described by a square scheme (Scheme 1). The diagonal of the Scheme is the X-H bond dissociation free energy (BDFE), and the free energy of an HAT reaction is simply the difference between the reactant and product BDFEs (eq 1). Around the outside of the square are the thermochemistry of the proton transfer (PT) and electron transfer (ET) components, the pK_a and the reduction potential E°.

Scheme 1.

Square scheme for an XH/X PCET reagent and a drawing of the class of ruthenium complexes examined here.

$$\begin{gathered} \text{X}-\text{H}+\text{Y}\to \text{X}+\text{H}-\text{Y} \\ \Delta G^{\text{o}}{}_{\text{HAT}}=\text{BDFE}\left(\text{X}-\text{H}\right)-\text{BDFE}\left(\text{Y}-\text{H}\right) \end{gathered}$$ (1)

Across a series of compounds, it has long been known that the BDFE is much less sensitive to changes in substituents than the pK_a and E° values. For para-substituted toluenes, for example, the BDFEs are almost identical with electron-withdrawing or donating substituents. A p-cyano group increases the acidity of toluene by 10-11 units and renders the incipient anion ~660 mV harder to oxidize. Yet the BDFEs for PhCH₂–H and p-CNPhCH₂–H differ only by ~1 kcal mol⁻¹.⁶ For direct comparison, the 0.66 V variation in E° and the shift in pK_a both correspond to a change in ΔG° of 15 kcal mol⁻¹ (ΔΔG° = FΔE°, and ΔΔG° = 2.303RTΔpK_a).

This thermodynamic compensation between the pK_a and E° arises because an electron withdrawing group, for instance, makes a molecule more acidic (easier to lose H+) but also less reducing (harder to lose e⁻). Most systems are not as closely compensated as toluenes, as summarized by Tolman⁷ and discussed at greater length elsewhere.¹ Interestingly, the change in E° is nearly always greater than that in pK_a, perhaps due to the greater extent of delocalization of the electron that participates in the PCET reaction than that of the proton. This may also be because comparative studies across a series often place the substituents far from the reactive site, in this case the proton. For instance, the series of ‘substituted toluenes’ mentioned above does not include substitution at the α position, PhCH₂X.

While the equilibrium constants for PCET/HAT reactions such as eq 1 are determined by the BDFEs, the mechanisms are in large part determined by the pK_a and E° values. PCET reactions can involve concerted transfer of H⁺ + e⁻ (H•) or can proceed by stepwise PT-then-ET or ET-then-PT. A stepwise path is often followed when the relevant pK_a or E° make an initial PT or ET favorable, because the intrinsic barriers for moving a single particle are typically smaller than for PCET.

For concerted HAT reactions, the rate constants are affected by both the ΔG° and the nature of the XH and YH bonds. At the same driving force, HAT to or from carbon atoms is substantially slower than HAT involving just OH and NH bonds.⁸ Additionally, the polarity of H• donors and acceptors can significantly affect rate constants. An electrophilic H• acceptor, for example, will show selectivity towards more hydridic X–H bonds in the presence of acidic X–H bonds of similar or a bit greater strength. This has been coined the ‘polar-matching effect’ in the organic free-radical literature.^{9, 10}

In the last few years, there has been increasing interest in how rate constants and intrinsic barriers for HAT reactions could be affected by the mismatch of donor and acceptor pK_a and E°.^11–17 Such a mismatch has been suggested to lead to an ‘asynchrony’ or ‘imbalance’ in the H⁺ and e⁻ components of a concerted HAT reaction. In particular, the basicity of enzymatic and biomimetic metal oxo H• acceptors are suggested to play an important role in their selective cleavage of strong C–H bonds.^{11, 14, 18–20}

For all of these reasons, it would be valuable to be able to independently tune the acid/base, outer-sphere ET, and homolytic bond strengths of PCET reagents. This report develops an approach to that aim, placing different substituent patterns on a ruthenium complex active for PCET. A complementary approach is to combine separate redox and acid/base cofactors across a linker, as in the anilinium-substituted cobaltocene developed by the Peters group.^{21, 22}

We report here the synthesis, characterization, and PCET properties of five new derivatives of the ruthenium complex Ru(acac)₂(py-imH) (acac = 2,4-pentanedionato, py-imH = 2-(2’-pyridyl)imidazole, Scheme 1 and Figure 1). The parent compound (no substituents, 4) and the hexafluoro-acac derivative Ru(hfac)₂(py-imH) (3, hfac =1,1,1,5,5,5-hexafluoro-2,4-pentanedionato) were described by our laboratory some years ago.²³ The seven compounds form two series of complexes, one with substitutions to the acac ligands and the other with substitutions instead to the imidazole moiety. In short, substitutions to the py-imH ligand change predominantly the pK_a of the complex with minimal change to the redox potential, while the opposite is true for changes to the acac ligands. These series are a rare (if not the first) example of a system in which the pK_a and E° of a single molecule can be tuned with a large degree of independence.

Figure 1.

Drawings of Ru^II(acac)₂(py-imH) complexes (top row) and molecular structures of their oxidized, deprotonated forms Ru^III(acac)₂(py-im) from single-crystal X-ray diffraction (bottom row). These show the substitution patterns in acac-substituted (1-4) and py-imH-substituted (4-7) series. Complex 3 (not shown) is the hexafluoro-acac derivative Ru(hfac)₂(py-imH) (the same as 4 but with CF₃ instead of CH₃ groups).

Results

i). Synthesis

The substituted Ru^II(acac’)₂(py-imH) complexes (acac’ = acac, hfac, or the bis-^tBu or bis-Ph derivatives (in 1 and 2)) were synthesized similarly to the parent complex²³ with some modifications; see the Supporting Information (SI) for complete details. The appropriately substituted tris(acac’) complex was either purchased or synthesized from hydrated RuCl₃. [Ru(acac’)₂(MeCN)₂][OTf] (OTf = triflate)²⁴ was prepared from the tris(acac’) and serves as an excellent precursor to our complexes due to the relative lability of the MeCN ligands. The corresponding Ru(II) precursor, Ru(acac’)₂(MeCN)₂, was prepared directly from the tris(acac’) complex by heating in MeCN with activated zinc dust under inert conditions.²⁵ The Ru(acac’)₂(MeCN)₂ complexes are indefinitely air stable as solids and for hours to days in solution (except for ready decomposition in CH₂Cl₂ and CHCl₃). The [Ru(acac’)₂(MeCN)₂][OTf] complexes are even more stable, with decomposition in CH₂Cl₂ only occurring on the timescale of days to weeks.

The final Ru(acac’)₂(py-imH) complexes were prepared by reaction of the appropriately substituted py-imH ligand either with Ru(acac’)₂(MeCN)₂ or with [Ru(acac’)₂(MeCN)₂][OTf] followed by reduction over zinc dust. The ligand exchange reactions were performed at ca. 80 °C. The Ru^II(acac’)₂(py-imH) complexes were all air sensitive, attributed to their rapid oxidation by O₂ to form Ru(acac’)₂(py-im),²³ and were therefore handled and stored in an N₂-filled glovebox. The deprotonated Ru(III) analogs were also prepared and are air-stable (see SI for more detailed synthetic procedures).

The seven complexes studied here fall into two series (compounds in Figure 1 plus the hfac derivative 3). Compounds 1-4 differ in the groups on the 1 and 5 positions of the acac: ^tBu, Ph, CF₃, and CH₃ (parent). The series 4-7 all have acac ligands and differ in the N–N chelating ligand: 2-(2-pyridyl)imidazole (4), 2-(2-pyridyl)benzimidazole (5), and 2-(2-pyridyl)-4-trifluoromethylimidazole. The last ligand surprisingly formed two different isomers. Reaction with the Ru^III precursor forms only the complex with the CF₃ group on the imidazole carbon adjacent to the metal (6), while reaction with the Ru^II precursor places the CF₃ group adjacent to the NH bond (7). These isomers do not interconvert in solution at room temperature, neither in the Ru^IIimH nor the Ru^IIIim form. Ru^II and Ru^III complexes with the deprotonated form of the pyridyl 4,5-dinitroimidazole ligand, Ru(acac)₂(py-(NO₂)₂im) were prepared but decomposed rapidly upon attempted protonation of the distal imidazole nitrogen (see SI Figure S1).

ii). Crystallographic characterization

X-ray crystal structures were solved for the new complexes in their air-stable Ru^IIIim state (Figure 1). The structures of 6 and 7 confirm that they are isomers, with the CF₃ group proximal (6) or distal (7) to the ruthenium center. In 6, the Ru–N(im) distance is 0.03 Å longer than that in 7 and in all of the other complexes. The full crystallographic and metrical data for Ru^IIIim forms of 1, 2, 5-7, and Ru(acac)₂(py-(NO₂)₂im) are tabulated in the SI (the structure of 4-Ru^IIIim was previously reported, and 3-Ru^IIIim is unstable²³). All six structures share a very similar distorted octahedral geometry, with all trans angles >170°. The Ru–O bond lengths vary only a small amount across the series, with an overall range of 1.983(2)–2.029(3) Å and no single Ru–O bond varying more than 0.021 Å across the series. Similarly, the Ru–N distances do not vary very much, ranging from 2.039(6)–2.072(3) Å for Ru–N(py) and 1.994(6)–2.018(3) Å for Ru–N(im) (not including the 2.047(3) Å distance in 6 for the imidazole N next to a CF₃ or the minor disordered component in 5).

iii). Spectroscopic characterization

Each complex was characterized by ¹H NMR and UV-vis spectroscopies, and by mass spectrometry. The NMR spectra of the diamagnetic Ru^IIimH and paramagnetic Ru^IIIim complexes are all similar to the previously described spectra of 4 in these forms.²³ For instance, the ¹H chemical shifts in the Ru^III spectra range from 18.44 to −75.79 ppm, with sharper peaks for the protons farther from the metal. The Ru^II spectra are sharp when free of any of the Ru^III complexes, broadening apparently due to rapid exchange by electron transfer. The ¹H chemical shifts of the Ru^IIimH complexes are also sensitive to water content, shifting up to 0.2 ppm in its presence.

The isomeric CF₃-imidazole complexes 6 and 7 have close but distinct ¹⁹F NMR resonances in the Ru^IIimH forms (differing by 3 ppm). In contrast, the broad ¹⁹F resonances for paramagnetic Ru^IIIim congeners are separated by ~80 ppm.

UV-vis spectra of the Ru^II(imH) complexes in MeCN contain metal-to-ligand charge transfer (MLCT) bands in the visible region (Figure 2, Table 1) consistent with similar ruthenium(II) polypyridyl complexes.^26–29 The optical absorbances are considerably less intense for the corresponding Ru(III) complexes.

Figure 2.

Visible region of the optical spectra of the Ru^IIimH forms of 1-7, showing the characteristic MLCT bands.

Table 1.

MLCT energies and corresponding extinction coefficients for the major transitions of 1-7 in the visible region.^a

	RuII(imH)		RuIII(im)
	λmax,1 (ε)b	λmax,2 (ε)b	λmax (ε)b
1	583 (8,600)	426 (7,900)	495 (2,500)
2	568 (13,600)	425 (3,200)c	655 (600)c
3	519 (10,000)	481 (9,600)c	— d
4	568 (7,000)	428 (6,700)	486 (1,600)
5	625 (6,400)	423 (6,400)	507 (2,300)
6	577 (6,700)	408 (6,500)	519 (1,800)
7	593 (6,700)	420 (7,000)	517 (1,900)

^aListed are the main optical features in the visible region (see the SI for full spectra).

^bλ_max is ± 3 nm and extinction coefficient (ε) is ± 100 M⁻¹ cm⁻¹.

^cλ_max corresponds to a shoulder.

^d3 is unstable in the Ru^III(im) form.²³

iv). DFT Calculations: Structure, Orbital Character, and Spectroscopy

Density functional theory (DFT) calculations were performed to gain insight into the electronic changes across the series of complexes, specifically through calculation of the energies of the relevant molecular orbitals (MOs) and their localization on the Ru center, py-imH and acac ligands. The optimized structures were in good agreement with the available solid-state crystal structures, and they were similar across the family (see SI for computational details and coordinates). The calculated MO diagrams for 1-7 in the Ru^IIimH form are shown in Figure 3, in which each horizontal line indicates the energy of a molecular orbital, and the coloring shows the fractional contributions from Ru (black), py-imH (blue), and acac (red). The principal component of each orbital is identified in the labels at right.

Figure 3.

Computed Molecular orbital diagrams for 1-7. Percental contributions are shown by color: black (Ru), blue (py-imH), and red (acac).

The UV-vis spectra of the Ru^IIimH complexes were computed using TD-DFT, and the results are in excellent agreement with the experimental data (see Discussion). The lower-energy MLCT is primarily a R→py-imH transition, whereas the higher-energy MLCT is primarily Ru→acac. The substituents primarily move the manifold of orbitals together, so that the HOMO-LUMO gap is quite similar for almost all of the complexes, with the exception of 3. This is seen in the computed energies, the computed spectra, and in the experimental spectra. The most important transitions for each complex, their relative contributions from the Ru and ligands, and the corresponding electron difference density maps (EDDMs) are summarized in the SI (Table S8, Figure S87).

Significant changes in the acac substituents – compounds 2-4 – strongly affect the Ru-centered t_2g orbitals and the acac π-orbitals (π_acac). The hfac complex 3 is shown at the far left of Figure 3 because is it the most different. The four CF₃ groups strongly lower the energies of the acac π* and π orbitals (the latter fallen so much that they are below the bottom of the plot and not shown). The hfac ligands also lead to substantially lower Ru t_2g orbitals than most of the other compounds (by ca. ½ - ¾ eV). The diphenyl-acac complex 2 has similar effects but smaller in magnitude.

Substitutions to the py-imH ligand – compounds 4-7 – have little effect on the energies of the Ru or acac orbitals. The MO diagrams show differences primarily in the π and π* orbitals of the ligand. The benzimidazole complex 5 has lower π* orbitals due to its more extensive π system. The CF₃ substituent on the imidazole (proximal 6 and distal 7 isomers), also lowers the py-imH π and π* orbitals, without affecting the other orbitals significantly. These trends are paralleled in the oxidized deprotonated Ru^IIIim MO diagrams for these complexes (Figures S74–75) and are evident in the thermochemistry discussed in the next section.

v). Thermochemical measurements: pK_a, E_1/2, and BDFE

The pK_as of 1-7 in their Ru^II(py-imH) and [Ru^III(py-imH)]⁺ forms were determined in MeCN through titrations with an acid or base of similar pK_a and monitoring the equilibration by UV-vis spectroscopy (Figure 4; see SI). For complexes 1 and 4, the pK_a(Ru^IIimH) was not accessible due to the instability of the anion. Similarly, the instability of the oxidized forms of 3 prevented measurement of its pK_a(Ru^IIIimH⁺). The pK_a values of all titrants were obtained from self-consistent equilibration studies,³⁰ so the relative pK_a values are all ±0.1 unit. All the measured pK_a values are tabulated below in Table 2, and the individual titration data sets are provided in the SI.

Figure 4.

Overlay of dilution-corrected UV-vis spectra from the titration of 1-Ru^IIIimH⁺ with 0.05 (blue) to 100 (pink) equivalents of 2,4,6-collidine (pK_a = 15.00³⁰), and after the addition of excess strong base (dashed black line). Inset shows the linear plot of [Ru^IIIim][collidine–H⁺]/Ru^IIIimH⁺] vs. [collidine], with slope = K_eq.

Table 2.

Measured thermochemical parameters for the PCET of complexes 1-7.^a

	E 1/2 imH	E 1/2 im	pKared	pKaox	BDFE1 b	BDFE2 b	BDFECV c
1	−0.86	−1.23	23.0	16.8	55.8	—	55.6
2	−0.57	−0.89	22.3	16.2	61.6	62.6	61.6
3	0.29	−0.07	19.7	13.6	—	78.0	76.9
4	−0.64	−1.00	22.2	16.1	59.9	—	60.0
5	−0.60	−0.96	21.5	14.8	59.0	59.9	59.3
6	−0.57	−0.91	19.0	12.7	56.9	57.6	56.9
7	−0.55	−0.88	16.6	10.3	54.0	55.0	54.1

^aAll values measured in MeCN. E_1/2 values are in V vs Fc^+/0 and ± 0.01 V, pK_a values are ± 1.0, and BDFEs are ± 1.0 kcal mol⁻¹. Values in italics are calculated from the other three E_1/2/pK_a values using Hess’s Law.¹

^bBDFE₁ is calculated from E_1/2^imH and pK_a^ox, and BDFE₂ is calculated from E_1/2^im and pK_a^red, both using eq 2.¹

^cEach BDFE_CV is determined by CV in an appropriate buffer and using eqs 3 and 4.¹

Cyclic voltammograms (CVs) of 1-7 in MeCN containing 100 mM [n-Bu₄N][PF₆] all yield chemically reversible waves corresponding to the Ru^III/II redox couple. At a scan rate of 100 mV/s, anodic and cathodic peak currents were typically within 15% of each other and separated by about 60 mV, roughly the same as the ferrocene couple under the same conditions. pK_a data for 1 and CV data for 6 are shown in Figures 4 and 5 as representative examples (see SI for complete details). Upon in situ deprotonation of Ru^II-im with a strong base, the CV waves shifted negative by 330 - 370 mV, depending on the complex. The redox potentials for the deprotonated forms were confirmed by CV of the isolated Ru^III(py-im) complexes. The Ru^III/II E_1/2s of the protonated and deprotonated forms of each complex are given in Table 2.

Figure 5.

(A) CVs of 6-Ru^IIIimH⁺ before (blue) and after addition of excess triethylamine (pK_a = 18.83;³⁰ red) in MeCN solution containing ferrocene and [n-Bu₄N][PF₆]. (B) CVs of 6-Ru^IIIim in buffered 2,4,6-collidine/2,4,6-collidine–H⁺ (pK_a = 15.00³⁰) in varying ratios. (C) Plot of the E_1/2 vs Fc^+/0 of the 1e⁻/1H⁺ proton-coupled Ru^III/II redox couple vs logarithm of the buffer ratio, with slope of 59.0 ± 0.1 mV and y-intercept (dashed line) of −701 ± 5 mV. The colors of the points correspond to those of the corresponding CVs in part (B).

The BDFEs were calculated using the thermochemical square scheme in Scheme 1, using the appropriate pK_a and E_1/2 values in MeCN (eq 2). The constant C_G is the free energy required to reduce a proton to H^• (with ferrocene in MeCN).¹ Appropriate pK_a and E_1/2 values are the top and right sides of the square scheme, or the left and bottom sides—both lead from XH to X, the BDFE.

$$\mathit{\text{BDFE}}=1.37{\mathit{\text{pK}}}_a+\text{23.06}{E}^{°}+C_G$$ (2)

For each complex for which all four pK_a/E_1/2 values were measured, this analysis gives two independent measurements of the BDFE that agreed within 1.0 kcal mol⁻¹ of one another. This indicates the precision of the individual measurements.

The BDFEs were also measured electrochemically to provide more precise values (an example is shown in Figure 5). This was done by dissolving either the Ru^IIimH or Ru^IIIim form of the complex in MeCN containing electrolyte. Then to this solution was added a buffer with a pK_a in between the pK_as of the reduced and oxidized forms of the complex, such that oxidation/reduction of the complex was coupled to a proton transfer. Using the Nernst equation (eq 3) and the pK_a of the buffer, the standard state potential for this 1e⁻/1H⁺ (n = 1) reaction was measured and converted to the BDFE (eq 4).

$$E=E^{°}-\frac{2.303\mathit{\text{RT}}}{\mathit{\text{nF}}}\text{log}\left(\frac{\left[\mathit{\text{RuH}}\right]\left[A^{-}\right]}{\left[\mathit{\text{Ru}}\right]\left[\mathit{\text{HA}}\right]}\right)-\frac{2.303\mathit{\text{RT}}}{\mathit{\text{nF}}}{\mathit{\text{pK}}}_a\left(\mathit{\text{HA}}\right)$$ (3)

$$\mathit{\text{BDFE}}=23.06{\text{E}}^{°}{}_{\mathit{\text{PCET}}}+C_G$$ (4)

The E_1/2 shifted ~59 mV with the logarithm of the buffer ratio and was independent of the scan rate, indicative of a PCET process reversible on the CV timescale. Each BDFE obtained in this way was within error of the average value obtained through the individual pK_a and E_1/2 measurements. Since E_1/2 is usually measured more accurately then pK_as, these values are likely to be more accurate. This method of determining the BDFE is analogous to measuring the PCET open-circuit potential (OCP) but only applicable when the process is electrochemically reversible.³¹

BDFEs can often also be measured by equilibration with a compound with a known, similar BDFE.¹ However, the scarcity of 1e⁻/1H⁺ PCET reagents with BDFEs < 60 kcal mol⁻¹ limited this approach to the BDFE of 5, which was equilibrated with phenazine (BDFE = 58.7 kcal mol^{−1 31}). However, the uncertainty of this measurement was larger than those measurements detailed above (see SI).

The primary uncertainty in the electrochemically measured BDFEs is the uncertainty in in the C_G term, ± 1.0 kcal mol⁻¹.¹ Since these equations use the same C_G, the relative uncertainty in ΔBDFE values is much lower, estimated as ± 0.2 kcal mol⁻¹.

vi). DFT Calculations: E_1/2 and pK_a values

Relative E_1/2 and pK_a values were computed by DFT. These values were anchored to the parent complex (4) by shifting all the absolute values by a scalar, such that the average pK_a and E_1/2 of 4 match the experimental values. Doing so minimizes internal methodology errors and allows a more meaningful comparison of the thermochemical trends. Figure 6 pictorially shows the experimental and computed differences in these parameters from those for 4.

Figure 6.

Bar chart of the E_1/2 and pK_a changes from the parent complex 4 induced by substitutions at the acac (1-4) and py-imH (4-7) ligands, converted to free energies. Darker bars are experimental values and lighter bars are calculated.

The series with modifications to the py-imH ligand (4-7) shows excellent agreement between the experimental and calculated ΔpK_a and ΔE_1/2 values: all of the deviations are ≤1.0 kcal mol⁻¹. For the series with modifications to the acac ligands (1-4), a few computed values show larger deviations, up to 0.2 eV/5 kcal mol⁻¹ for the ΔE_1/2^im and ΔBDFE for compound 1.

Discussion

Across a series of related compounds, many PCET redox couples show small variations of BDFEs due to compensation of changes in the E_1/2 by changes in the pK_a. In contrast, the Ru(acac)₂(py-imH) system described here exhibits significant ‘decoupling’ of the pK_a and E_1/2. Substitutions on the acac ligands cause E_1/2 changes at least 5 times as large as the pK_a changes, when described in the same free energy units (|FΔE_1/2| = |ΔΔG°| = 2.303RT|ΔpK_a|). The most pronounced decoupling is seen in moving from 4 to 2. Introducing phenyl groups into the acac ligands makes the Ru^II(py-imH) complex 110 mV less reducing and the deprotonated Ru^II(py-im) analog 70 mV less reducing. However, the pK_a shift is only 0.1 (± 0.1) unit, equivalent to 6 ± 6 mV. Substituents on the py-imH ligand decouple in the opposite direction, with larger changes in pK_a than in E_1/2, to varying degrees. Moving from 4 to 7, the complex with the distal CF₃ group, lowers the pK_a values by 5.6 / 5.8 units (Ru^II / Ru^III complexes, resp.; ≡ 340 mV) while the reduction potentials shift only 90 / 120 mV (py-imH / py-im, resp.). On the other hand, in moving from 4 to the benzimidazole 5, the changes in E_1/2 values nearly compensate those in the pK_as. The overall trend of this decoupling is illustrated in Figure 6.

The DFT calculations provide insight into the origins of the E_1/2 / pK_a decoupling. The MO diagrams in Figure 3 show that the ruthenium d orbitals (t_2g orbitals) are significantly affected by changes to the acac ligands but negligibly by changes to the py-imH ligand. This explains why the E_1/2 values change much more across complexes 1-4 than across 4-7. The UV-vis spectra offer experimental support for this explanation. For complexes 1-4, the wavelength (λ_max) of the lowest-energy MLCT correlates with the E_1/2 of the complex (see Figure 7, λ_max,1). Because this transition is primarily R→py-imH in character, the transition energies reflect changes in the energies of the Ru-centered t_2g HOMOs. In contrast, across complexes 4-7 the changes in t_2g energies are small and the λ_max changes predominantly with the π*_py-imH orbital energies. Furthermore, the higher-energy MLCT around 420 nm (λ_max,2), which is primarily Ru→acac in character, shifts negligibly across the py-imH-tuned series (< 20 nm), suggesting that the energies of the t_2g (and π*_acac) orbitals remain unchanged for these complexes.

Figure 7.

The experimental and calculated λ_max of the two visible MLCT bands (left axis) and the E_1/2 (right axis) for each of Ru^IIimH complexes 1-7. The plot highlights the correlation of the E_1/2 with the energy of the lower-energy MLCT (purple) in complexes 1-4.

While substituents in the py-imH ligand do not strongly shift the reduction potentials, deprotonation to the anionic py-im ligand shifts the E_1/2 of all of the complexes by 345 ± 25 mV. The consistency of this shift is perhaps surprising in light of the differences in electronic structure between the compounds. The E_1/2 shift in part reflects the change in the overall charge of the complex upon deprotonation, and this shift is typical of imidazole-type complexes.^{1, 32}

The acidity of the imidazole N–H bond likely reflects the overall electron-rich or electron-deficient character of the ligand, which is reflected in the energies of the π_py-imH orbitals. These are the frontier orbitals most affected upon deprotonation (see Figure S76) and that shift significantly across the py-imH-tuned series. The π_py-imH orbitals are, however, negligibly affected by the acac substitutions, except for 3 (for which these changes are still dwarfed by the t_2g orbital energy shifts).

More broadly, the decoupling of the redox and acid/basic properties of these complexes is affected by the physical separation of the electron and proton within the molecule and the different distances from the substituent to the e⁻ or H⁺. In particular, the four acac substituents are quite distant from the acidic proton in the imidazole N–H bond. With the imidazole substituents, the substitutions in 5 and 7 are slightly closer to the acidic N–H bond, while the CF₃ group of 6 is three bonds away from both the N–H bond and the ruthenium center. This helps to explain why the decoupling is smaller in the py-imH-tuned series than in the acac-tuned series. However, the significant decoupling even in 6 underscores the primacy of the frontier orbitals in determining this decoupling.

While the data and computations in this report are all in acetonitrile solvent, the conclusions should be general in different media. The relative E° and pK_a values depend only on the differences in the free energies to transfer the relevant species from one solvent to another, and these differences should be small. For instance, the ΔG°_transfers should be very similar for Ru(acac)₂(py-imH)⁺ and Ru(acac)₂(py-CF₃imH)⁺, even between protic and aprotic solvents such as MeCN to H₂O, because the molecules have similar sizes, charges, and functionalities. Therefore, the difference between their E_1/2 values should not change significantly between the solvents. Solvent effects on PCET reactions have been discussed in our recent review.¹

The thermodynamic coupling of E_1/2 and pK_a in these Ru complexes can be compared with similar effects in organic molecules. The pK_as and redox potentials of unconjugated organic PCET reagents such as alkylamines and hydroxylamines are much more tightly coupled, their BDFEs being more substitution-independent.^{1, 7} In aromatic organic PCET reagents, such as phenols, indoles, arylamines, and toluenes, substitutions on the aromatic ring typically result in a larger change in redox potential than in pK_a. This likely stems from the greater delocalization of the electron into the ring than the proton—analogous to the HOMO in the Ru(acac)₂(py-imH) system being delocalized over the whole complex. Substitutions closer to the acidic proton in these organic systems might feasibly flip this decoupling in favor of changing the acidity. This seems true at least for substituted toluenes; for example, the pK_as of para-tolunitrile and 2-phenylacetonitrile in DMSO are 31 and 22,^{33, 34} while their gas phase ionization energies are very similar (NCPhCH₃, 9.32 eV; vs. PhCH₂CN, 9.39 eV).³⁵

Conclusions

In a newly synthesized series of Ru(acac)₂(py-imH) complexes, we have tuned the redox and acid/base properties independently of one another. The former is achieved through substitutions to the acac ligand and the latter through substitutions to the imidazole. DFT calculations reveal that this ‘decoupling’ of the E_1/2 and pK_a can be attributed to differences in the molecular orbitals that are affected by the different substitution patterns. More broadly, the decoupling arises from the placement of the substituents closer to the e⁻ (formally located on the ruthenium center) versus closer to the H⁺ (on the imidazole). The ability to separately tune the component properties of a PCET system is likely to be valuable in various future applications.