Spectral observation of conversion between ionized vs. non-ionized proton-coupled electron transfer interfaces

Abstract

Two-point hydrogen bonding between acid and base functionalities provides a convenient method for the modular assembly of proton-coupled electron transfer (PCET) networks, especially when that interface comprises an amidinium and two-point anionic partner; a system is presented that permits the proton configuration within the interface to be determined when pK_a values of the conjugate acids are known.

Proton-coupled electron transfer (PCET) may be established within donor–acceptor dyads assembled by a hydrogen bonded interface.^1,2 PCET networks assembled from Watson–Crick base pairs³ and dicarboxylic acid dimers⁴ do not support proton transfer within the interface thus limiting the extent to which PCET may be investigated. For this reason, the asymmetric proton interface of an amidinium bound to two-point, anionic partners, especially carboxylate, has come to the fore.^5–10 The two tautomers shown in Chart 1 are possible: an ionized amidinium–carboxylate or a non-ionized amidine–carboxylic acid, both of which can support proton transfer along a PCET coordinate.

Chart 1.

The amidinium–carboxylate interface combines the dipole of an electrostatic ion pair interaction, a two-point hydrogen bond, and secondary electrostatic interactions to produce a highly stable proton interface.^11–15 These factors along with the relative pK_a values of amidinium and carboxylate, conspire to determine the extent to which the interface configuration is ionized.¹⁶ The electrostatic interactions within the interface stabilize charge accumulation, leading in some cases to an ionized interface even when a simple ΔpK_a calculation predicts formation of a non-ionized interface in a solution of a given dielectric constant.¹⁰ We now show the ΔpK_a value necessary to achieve the ionized vs. the non-ionized interface by stepping through a series of binding moieties (BMs) with different acidity constants. The transition from a non-ionized to ionized interface occurs as the acidity of the BM increases.

Efficient binding of an amidinium (amH⁺) to carboxylate and sulfonate anions occurs in aprotic solvents of low dielectric constant. A series of BMs of varying pK_a values in ACN was chosen as summarized in Table 1.^17,18 The trend in the dependence of the amH⁺:BM⁻ association and interface configuration on the dielectric constant of the solvent can be probed along the series tetrahydrofuran (THF, ɛ = 7.58), dichloromethane (DCM, ɛ = 8.93) and acetonitrile (ACN, ɛ = 37.5).¹⁹

Table 1.

$\mathrm{\Delta }{G}_{\mathrm{p}K\mathrm{a}}$ and interface configuration for 1:BM complexes

Binding moiety		pKaa	$\mathrm{\Delta }{G}_{\mathrm{p}{K}_{\mathrm{a}}}/\text{eV}$ (1:BM)b	Ionized interface?
				THF	DCM	ACN
1	Purpurin-amH+	9.55
2	3,5-Dinitrobenzoate	17	0.44	✓	✓	✓
3	Phenylsulfonate	7	−0.15	✓	✓	✓
4	Acetate	21	0.68
5	Chloroacetate	18	0.50		✓
6	Dichloroacetate	14	0.26	✓	✓	✓

^aIn CH₃CN, from ref. 17.

^bAt 298 K, calculated from pK_a values recorded in ACN, $\mathrm{\Delta }{G}_{\mathrm{p}{K}_{\mathrm{a}}}$ = −RT[pK_a(BM)/pK_a(l-am⁺)].

Interrogation of the proton configuration of the amidinium interface was accomplished using purpurin 1, in which conjugation is maintained between the amidinium and porphyrin via an unsaturated five-membered ring. We have shown that the absorption spectrum of 1 is sensitive to its protonation state.¹⁰ Reduction of the five-membered isocyclic ring of 1 yields the corresponding chlorin homologue. Unlike 1, the absorption spectrum of the chlorin–amidinium is invariant with pH (see Fig. S1, ESI^†).

Fig. 1 shows the absorption spectrum of 1 before (1-amH+) and after (1-am) addition of the base 4-dimethylaminopyridine (DMAP). Deprotonation of the amidinium functionality in ACN is indicated by a shift in the absorption maximum of the Soret band from 428 to 423 nm (Fig. 1(c)). Similar spectral trends are observed in THF and DCM (Fig. 1(a) and (b), respectively). Absorption spectra (Fig. S2–S4, ESI^†) demonstrate analogous spectral shifts with each BM and exhibit well-anchored isosbestic points during the titration of the BM with 1-amH⁺. A 1: 1 binding motif is signified from a Benesi–Hildebrand analysis of the titration data (1/ΔA vs. 1/[BM]) (Fig. S2–S4, ESI^†).²⁰ The binding constant, Kassoc, for 1:2 is found to be ~4.5 × 10⁴ M⁻¹ and for 1:3 is ~1.5 × 10² M⁻¹ in ACN. The Kassoc for 4, 5 and 6 are all the same within error and are measured to be ~9.9 × 10³ M⁻¹.

Fig. 1.

UV-visible absorption spectra of 1-amH⁺, solid line; 1-am formed by deprotonated of 1-amH⁺ with DMAP, dotted line. (a) λ_max(1-amH⁺) = 430 nm shifts to λ_max(1-am) = 426 nm in THF, 2000 eq. DMAP, (b) λ_max(1-amH⁺) = 436 nm shifts to λ_max(1-am) = 426 nm in DCM, 550 eq. DMAP and (c) λ_max(l-amH⁺) = 428 nm shifts to λ_max(1-am) = 423 nm in ACN, 950 eq. DMAP. Inset: Soret peak position of the purpurin vs. $\mathrm{\Delta }{G}_{\mathrm{p}{K}_{\mathrm{a}}}$ for each of the associated 1:BM complexes. The highest and lowest peak maxima (solid circles) represent 1-amH⁺, solid line; 1-am, respectively.

The endpoint spectrum of each titration reveals the configuration of the proton interface as indicated by the final position of the Soret (S₂ ← S₀ transition at B430 nm) or Q-band (S₁ ← S₀ transition at ~650 nm) maximum. Conversion of 1-amH⁺ to 1-am in the non-ionized interface is indicated by a shift of the peak maxima to wavelengths nearly coincident with that observed for 1-am upon deprotonation of 1-amH⁺ with DMAP. We interpret the small 1–2 nm shifts of the Soret band of 1-amH⁺ in the presence of the BM to the establishment of a non-ionized amidine–carboxylic acid interface.¹⁰

Fig. 1 (inset) shows the final Soret peak position of 1 in each solvent for all BMs studied. The solid data points at each end represent the peak position of 1-amH⁺ (highest wavelength) and 1-am (lowest wavelength). The dotted line in each plot shows an approximate boundary for the transition between the ionized to non-ionized interface as determined by the final Soret position, and augmented by spectral analysis of the initial and final spectra (see ESI^† for details of spectral analysis). The unavailability of additional BMs with precise pK_a values limits the observation of a clearly identifiable pK_a transition between the two tautomers.

Experimental observation of the proton interface configuration for each BM and determination of K_assoc for a given interface enables the construction of the thermodynamic cycle shown in Scheme 2. Since the interface under consideration is formed from amH⁺ and ⁻O₂C as the reactants (Scheme 2 top, left), either an amH⁺ … ⁻O₂C or an am…HO₂C interface is produced. The driving force for formation of the former (ΔG₁) is measured directly from the measurement of K_assoc for the ionized 1-amH^+:−O₂C complex, while the driving force for the latter (ΔG₃) is determined for the non-ionized 1-am…HO₂C complex (Scheme 2, inset). The driving force for formation of am…HO2C from the neutral analogues (ΔG₂) can be determined by the difference between the ΔG₃ and $\mathrm{\Delta }{G}_{\mathrm{p}{K}_{\mathrm{a}}}$ as shown by the inset in Scheme 2,

$$\mathrm{\Delta }{G}_2=\mathrm{\Delta }{G}_3-\mathrm{\Delta }{G}_{\mathrm{p}{K}_{\mathrm{a}}}$$ (1)

Accordingly, the stabilization imparted by proton transfer within the interface (ΔG_interface) can be determined from the free energies of the square scheme and the pK_as of the constituent moieties ( $\mathrm{\Delta }{G}_{\mathrm{p}{K}_{\mathrm{a}}}$).

Scheme 2.

Square scheme representation of free energies involved in determining the tautomeric form of the interface.

Although the binding of the various BMs of Table 1 to 1-amH⁺ was carried out in ACN, DCM and THF, the thermodynamic description and stabilization energy of the interface outlined above can only be rigorously determined for ACN since pK_a values for the BMs used in this study have been determined in this solvent. Inspection of Fig. 1(c) reveals that conversion between the tautomeric forms of the interface occurs between BMs 4 and 5. Thus, Scheme 2 is analyzed for these two BMs as they most closely define the razor’s edge between formation of the ionized and neutral hydrogen bonding interface. Accordingly, these two points allow for the determination of the minimum interface stabilization necessary to obtain an ionized interface as per eqn (2):

$$\mathrm{\Delta }{G}_{\text{interface}}=(\mathrm{\Delta }{G}_2+\mathrm{\Delta }{G}_{\mathrm{p}{K}_{\mathrm{a}}})-\mathrm{\Delta }{G}_1$$ (2)

The am…HO₂C tautomer is formed upon association of 1-amH⁺ with 4, allowing for the determination of ΔG₂ via eqn (1). ΔG₁ is obtained from K_assoc(1:5), since association of 1-amH⁺ with 5 generates the ionized amH^+:−O₂C complex. That the ionized interface for 1:5 persists when the pK_a difference dictates formation of the am…HO₂C interface reveals the importance of electrostatic contributions to the interface configuration. The electrostatic stabilization of the interface imparted by formation of the ionized amH^+:−O₂C complex (ΔG_interface) can be determined using eqn (2); it is minimally 0.18 eV in ACN. This electrostatic interaction energy can be calculated for point charges, q₁ and q₂, according to V = (q₁q₂)/(4πɛr) where ɛ is the dielectric constant of the medium. An electrostatic stabilization energy of 0.10 eV is calculated for the amidinium-carboxylate salt bridge at a distance of r = 3.9 Å in THF. This distance is measured between the central carbon atoms of the amidinium and carboxylate groups from the energy optimized structure.⁸ The experimental results reported here agree with a simple ion solvation model. The discrepancy likely arises from the approximation that the electrostatic charge within the interface is confined to a point-dipole (Table 2).

Table 2.

Calculated free energy values for thermodynamic considerations for assignment of DGInterface in CAN.

Binding moiety (BM)		pKaa	ΔG1b/eV	ΔG2/eV	ΔG3/eVb	$\mathrm{\Delta }{G}_{\mathrm{p}{K}_{\mathrm{a}}}$c/eV	ΔGinterface/eV
1	Purpurin-amH+	9.55	–	–	–	–	–
2	3,5-Dinitrobenzoate	17	−0.28	–	–	−0.44	–
3	Phenylsulfonate	7	−0.13	–	–	0.15	–
4	Acetate	21	–	0.44	−0.24	−0.68	0.18
5	Chloroacetate	18	−0.24	–	–	−0.50
6	Dichloroacetate	14	−0.24	–	–	−0.26	–

^aIn CH₃CN, from ref. 17.

^bΔG = −RTlnK_assoc(1:BM).

^cAt 298 K, calculated from pK_a values recorded in ACN, $\mathrm{\Delta }{G}_{\mathrm{p}}{K}_{\mathrm{a}}$ = −RT[pK_a(BM)/pK_a(1-amH⁺)].

While the dearth of known pK_a values in DCM and THF preclude a similar thermodynamic analysis in those solvents, a qualitative understanding of interface behaviour is inferred from experimental results. In the dielectric environment of DCM, the proton interface maintains the ionized configuration at the same ΔpK_a difference as ACN, however in the lower dielectric environment of THF conversion to the ionized tautomer shifts to lower ΔpK_a. THF does not tolerate charge buildup to the extent that ACN and DCM do, and thereby favours a non-ionized interface configuration at a lower free energy.

The energetics of the amidinium-carboxylate interface are essential to assigning a general interface stabilization energy with the goal of accurately projecting the tautomeric state of the proton interface to PCET systems. For example, the hydrogen bonded donor-acceptor, system D−[H⁺]−A (D = amidinium appended Zn(II)TMP, A = naphthalene diimide carboxylic acid) reveals a PCET rate of 9.50 × 10 s⁻¹ with a kinetic isotope effect (KIE) of 1.06 at room temperature. In this case, the interface is the non-ionized amidine-carboxylic acid tautomer. When this D−[H⁺]−A system is subtly altered to produce an ionized amidinium-sulfonate interface (A = naphthalene diimide sulfonic acid) of similar driving force, (Table 1, represented by the 1:3 interface), the change in interface configuration has a clear effect on the PCET rate and solvent dependent ET parameters (|V|, λ, E_act). The PCET rate roughly doubles to 19 × 10 s⁻¹ with a KIE of 1.17 and more strikingly ET parameters show a clear solvent dependence through the ionized interface, which is absent in the first system. The enhancement of PCET rate and solvent dependent ET parameters for the same D–A pair bearing different tautomeric forms of the interface, illustrates the pronounced effect that proton position exerts on photoinduced charge transfer.

Scheme 1.