University of Birmingham H-bond acceptor parameters for anions

: UV/vis absorption titrations have been used to investigate the formation of H-bonded complexes between anionic H-bond acceptors (HBAs) and neutral H-bond donors (HBDs) in organic solvents. Complexes formed by three di ﬀ erent HBDs with 15 di ﬀ erent anions were studied in chloroform and in acetonitrile. The data were used to determine self-consistent HBA parameters ( β ) for chloride, bromide, iodide, phosphate diester, acetate, benzoate, perrhenate, nitrate, tri ﬂ imide, perchlorate, hexa ﬂ uorophosphate, hydrogen sulfate, methyl sulfonate, tri ﬂ ate, and per ﬂ uorobutyl sulfonate. The results demonstrate the transferability of H-bond parameters for anions between di ﬀ erent solvents and di ﬀ erent HBD partners, allowing reliable prediction of anion recognition properties in other scenarios. Carboxylates are the strongest HBAs studied, with β parameters ( ≈ 15) that are signi ﬁ cantly higher than those of neutral organic HBAs, and the non-coordinating anion hexa ﬂ uorophosphate is the weakest acceptor, with a β parameter comparable to that of pyridine. The e ﬀ ects of ion pairing with the counter-cation were found to be negligible, provided small polar cations were avoided in the less polar solvent (chloroform). There is no correlation between the H-bonding properties of the anions and the p K a values of the conjugate acids. 22 dissolved 2 host of this solution successively added the cuvette, the UV/vis absorption spectrum addition. The UV/vis absorption spectra analyzed t 1:1 binding isotherm optimizing association


■ INTRODUCTION
Molecular recognition events involving anions are fundamental to a wide range of biological processes. 1 The regulation of noncovalent interactions formed in a specific but transient manner between biomolecules is widely exploited within the body to perform essential tasks and achieve function. 2 Synthetic systems have extensively employed the formation of H-bonding interactions to anions as a key molecular recognition motif in supramolecular chemistry, 3,4 finding applications in a wide range of processes including catalysis, 5 responsive materials, 6 ion extraction, 7 transportation, 8 and sensing. 9 However, there are numerous factors that influence the properties of noncovalent interactions, complicating the analysis of the operational basis of complex systems. In order to gain a deeper insight into the factors which govern molecular recognition, a quantitative approach to non-covalent interactions is required. Despite the importance of H-bonding to anions, the development of a quantitative analysis of the behavior of charged species in solution has been slow to emerge.
Abraham developed quantitative scales for the H-bond acceptor (HBA) and H-bond donor (HBD) strengths of neutral organic functional groups. 10 These scales were established through analysis of experimentally determined association constants (K) for formation of H-bonded complexes in carbon tetrachloride solution. 10−12 The Hbonding properties of neutral molecules were described using eq 1.
where c 1 and c 2 are constants related to the solvent, and α 2 H and β 2 H represent the HBD and HBA properties of the solutes. H-bond descriptors α 2 H and β 2 H have been measured for a wide range of different compounds 10−12 and have been used to construct linear free energy relationships (LFERs) to predict the behavior of neutral solutes in a number of biological and physicochemical processes. 13 We have extended this approach to explicitly include the influence of solvent on solution-phase equilibria between H-bonded solutes. The solvent competition model illustrated in Figure 1 is based on a comparison of the relative stabilities of pairwise contacts between solvent and solute. 14 Provided HBD and HBA parameters are available for the solutes (α and β) and the solvent (α s and β s ), eq 2 can be used to predict the Gibbs free energy change (ΔG°) for formation of a H-bonded complex in any solvent. 15 where the adverse free energy associated with formation of a bimolecular complex in solution has been experimentally determined to be 6 kJ mol −1 in carbon tetrachloride and is assumed to be a constant in other solvents. It is also possible to use eq 2 to determine H-bond parameters for solutes or solvents on the basis of experimentally measured association constants for H-bond formation. 14−19 For example, if the α, α s , and β s parameters for a particular system are known, then experimental determination of the association constant (K) for formation of a complex between the HBD and any HBA can be used to determine the β value through rearrangement of eq 2 to give eq 3.
This approach has been used to quantify the HBA strength of a wide range of neutral organic functional groups. 19−22 To date, trialkyl amine oxide and trialkyl phosphine oxide are two of the strongest neutral organic HBAs on the H-bond acceptor scale, with β values of 11.6 and 10.7, respectively. 22,23 The HBA properties of neutral organometallic compounds have been determined in the same way, and a β value of 12.1 was measured for trans-[Ni(F)(2-C 5 NF 4 )(PEt 3 ) 2 ]. 23 Here, this approach is used to add anionic species to the β H-bond scale. Marcus and co-workers have utilized the Gibbs free energy of transfer (ΔG°t r ) and enthalpy (ΔH°t r ) of ions from water to other solvents and correlated the values with properties of the solvents using multiple linear regression analysis. 24 Linear solvation energy relationships (LSERs) constructed from these data were employed to elucidate information about interactions of the ions with the first solvation shell in aqueous and nonaqueous solvents. The LSER approach has also been used to derive H-bond descriptors for the solvation properties of ionic species. 25 Several groups have reported association constants for Hbonded complexes between quaternary ammonium halide salts and HBDs (phenols, trihaloalkanes, and alcohols), 26−29 and Abraham and co-workers used these results to estimate β 2 H values for halide anions. 30 As the H-bond scales of Hunter and Abraham can be interconverted, 23 estimates of β values for halides can be obtained from these results: 11.3−12.7 (Cl − ), 10.2−10.7 (Br − ), and 8.7−9.9 (I − ). Here, we describe measurement of the HBA parameters (β) of a diverse series of monovalent anions using experiments in three different solvents and titrations with three different HBDs. The influence of ion pair formation on the HBA properties of the anions has also been quantified through systematic variation of the counter-cation.

■ RESULTS
Three donors of differing HBD strength4-nitro-3-trifluoromethylphenol (1 α = 5.1), 4-nitrophenol (2 α = 4.7), and 4phenylazophenol (3 α = 4.3)and 18 HBAs (5−22, Scheme 1) were selected to study the formation of H-bonded complexes between neutral donors and charged acceptor species (Scheme 2). The three HBD all have an absorption maximum in the UV/vis region that changes significantly on formation of a H-bond, facilitating measurement of association constants. In addition, the HBD parameters span a range of values that are all near the top of the α scale, and so stable complexes are formed even with less polar anions in competitive solvents. Phosphine oxide 4 was additionally employed as a HBA to align with previous studies undertaken for neutral solutes. 20, 21 The 15 distinct anions selected as HBAs include a series of halides and sulfonates, acetate, benzoate, phosphate diester, perrhenate, nitrate, and perchlorate (5−19, Scheme 1). Titration experiments were also carried out with the TBA salts of succinimide, tetrafluoroborate, cyanate, and nitrite, but problems relating to deprotonation of the HBD, overlapping signals, or lack of appreciable binding precluded use of these data for determination of β parameters for these anions.
To permit direct comparison of the HBA properties of the anions in Scheme 1, the same tetrabutylammonium counterion was used in 5−19. The influence of the counterion on Hbonding interactions with Br − and I − was investigated through variation of the alkyl group of the quaternary ammonium cation: butyl (11), octyl (20), and ethyl (21) for Br − ; butyl (16) and hexyl (22) for I − (Scheme 1).
Two different solvents were used for the titration experiments: acetonitrile and chloroform. These solvents have quite different H-bonding and dielectric properties: chloroform has a dielectric constant of 4.8 and is primarily a HBD, whereas acetonitrile has a dielectric constant of 37.5 and is primarily a HBA. 15 The tetraalkylammonium salts used as guests in the titration experiments have good solubility in both solvents. The

Journal of the American Chemical Society
Article neutral H-bond scales were originally developed on the basis of measurements made in carbon tetrachloride and trichloroethane, 10−12 so to test interchangeability with the more polar solvents used here, UV/vis absorption titration experiments were also carried out for the neutral HBA 4 in carbon tetrachloride, chloroform, and acetonitrile. The effect of the water content of the solvent was also investigated.
Association Constants. Compounds 1−3 all have a UV/ vis absorption band that is sensitive to formation of H-bonds, providing a convenient method to monitor binding. 15,17,20 Accordingly, UV/vis absorption titration experiments were carried out by adding each of the HBAs, 4−22, to each of the HBDs, 1−3, in acetonitrile and in chloroform. Representative UV/vis absorption spectra from the titrations are shown in Figure 2. Upon the addition of increasing quantities of 4−22 to 1−3, bathochromatic shifts of the UV/vis bands were seen: typically 284−306 nm for 1, 300−322 nm for 2, and 344−358 nm for 3 in chloroform ( Figure 2).
In general, the data for the anion titrations fit well to a 1:1 binding isotherm, and the resulting association constants are shown in Table 1. In some cases (15, 16, and 22), UV/vis absorption of the guest prevented acquisition of suitable data due to overlap with the spectrum of the host, 1 or 2 (see Supporting Information). However, 3 absorbs at significantly longer wavelengths than 1 and 2 ( Figure 2), so it was possible to measure association constants with this HBD. In acetonitrile at high concentrations of guest (9 and 11), proton transfer was observed in some cases. Formation of phenolate anions is associated with the appearance of absorption bands at 405 nm for 1, 420 nm for 2, or 476 nm for 3 (see Supporting Information). 20c,21 However, it was possible to obtain a reasonable degree of saturation (55−70%) before deprotonation became significant, and so association constants could be determined. For the neutral HBA 4, the titration data were fit to an isotherm that allowed for a second weaker binding interaction, as described previously. 15 The association constants measured for the complexes span 3 orders of magnitude (Table 1). The relative polarities of the solvents and solutes determine the stabilities of the H-bonded complexes (Table 2). Consequently, the largest association constants are observed in the carbon tetrachloride, followed by chloroform, with the lowest association constants in acetonitrile. For example, the stability of the 2·4 complex goes from 99 000 M −1 in carbon tetrachloride to 1400 M −1 in chloroform and 110 M −1 in acetonitrile. For the weaker HBAs (13−19 and 22), the association constants were too low to be measured in acetonitrile.
Salts are likely to be strongly solvated by adventitious water present in nonpolar solvents. The water content of the acetonitrile used in these experiments was determined to be 0.02%. 31 In order to determine whether this quantity of water is sufficient to perturb the association constants reported in Table  1, water was deliberately added to the acetonitrile stock solvent, and the titrations were repeated. The results are illustrated in Figure 3.
In the presence of increasing quantities of water, the association constant of the 1·TBABr complex falls, but relatively large amounts of water are required to have a significant effect (similar behavior was observed for the 1· TOABr and 1·TEABr complexes, see Supporting Information). It is clear from Figure 3 that small amounts of water have a negligible effect on the association constant, and so the results in Table 1 are not perturbed by water content.
The variation in association constant with HBD follows the same trend for all HBAs. The complex formed with 1 is more stable than the corresponding complex formed with 2, which is more stable than the 3 complex, in agreement with HBD parameters (α = 5.1, 4.7, and 4.3, respectively). The differences in association constants are most marked in chloroform, where the 1 complexes are up to 2 orders of magnitude more stable than the corresponding 3 complexes.
The variation in association constant with HBA follows the same trend for all HBDs: The carboxylates, BzO − and AcO − , form the most stable complexes with all of the HBD, and the complexes formed with PF 6 − have the lowest association constants, which is consistent with the use of PF 6 − as a non-interacting anion. The ranking of the anions is the same as the Hofmeister series (Cl − > NO 3 − ∼ Br − > I − > ClO 4 − ), which orders anions by their ability to induce the precipitation of proteins from aqueous solution. 32−34 For the halide series, Cl − , Br − , and I − , the stabilities of the complexes decrease on moving down the group, in agreement with the literature. 26,27 For the sulfonate ion series, the stabilities of the complexes follow the inductive effect of the substituent. The methyl group in MeSO 3 − increases the association constants compared with HSO 4 − , which has a hydroxyl group. The more electron-withdrawing fluoroalkyl groups in CF 3 SO 3 − and CF 3 (CF 2 ) 3 SO 3 − have the opposite effect.
H-Bond Acceptor Parameters. To determine whether a self-consistent set of H-bond parameters can be obtained in the

Journal of the American Chemical Society
Article three different solvents, the association constants for the neutral HBA, 4, were analyzed using eq 2. Figure 4 shows that the association constants measured for 4 are consistent with eq 2 if the H-bond parameters in Table 2 are used. The values in Table 2 were obtained by optimizing to fit the experimental data, but in most cases they are identical or very close to values deduced from the literature. 35 The exception is the β value for chloroform, where the theoretical value of 0.8, which was obtained from a semiempirical calculation of the molecular electrostatic potential surface, is slightly too low to account for the experimental association constants in Table 1. 14 The results for the neutral HBA in Figure 4 suggest that the association constants in Table 1 can be used the H-bond parameters from Table 2 to determine β values for the anions in 5−22. Table 3 shows the values of β derived from the association constants in Table 1 using eq 3. Remarkably good agreement is found for β values measured with different HBDs in different solvents. For example, the six independent measurements of β for 8 (TBACl) all fall between 12.0 and 12.3. Thus, the data for each HBA can be combined to obtain a representative average value that can be used with different HBDs in different solvent environments. In addition, we can be reasonably confident that the values obtained for HBAs, where  a Average of at least two titrations. Errors are quoted at the 95% confidence limit. In all cases, greater than 55% saturation of the binding isotherm was achieved. b Deprotonation of the H-bond donor was observed upon addition of guest. c The salt was not sufficiently soluble to obtain a saturation binding isotherm. d The association constant was too low to be measured. e The absorption of the solute obscured the spectrum. f Association constant too high to be measured using UV/vis spectroscopy.

Journal of the American Chemical Society
Article multiple experimental measurements were not possible, are likely to be transferable to different systems. Figure 5 shows the agreement between the experimentally measured free energies of complexation (ΔG°) and the values calculated using eq 2 with the average β values from Table 3.
Influence of the Counter-cation. Both the nature of the counterion and the polarity of the solvent are important factors in determining the type of ion pair formed in a salt solution. 36−38 Bjerrum described the existence of three different types of ion pairs in solution: (i) contact ion pairs, which involve direct interaction between oppositely charged species; (ii) solvent-shared ion pairs, wherein oppositely charged species are separated by one layer of solvent molecules; and (iii) solvent-separated ion pairs, which have more than one layer of solvent separating the oppositely charged species. 37 The nature of the ionic species present in solution can be further complicated through the formation of higher aggregates of ions (triplets and quadruplets) at high salt concentrations. 38 To establish whether ion pairing of the salt has a significant effect on the H-bond parameters reported in Table 3, three different counter-cations were used for the bromide anion: tetraoctylammonium (20), tetrabutylammonium (11), and tetraethylammonium (21). A value of β was determined for each of these salts with each of the three HBDs in both chloroform and acetonitrile. The β values for all 18 measurements fall within the range 10.2−11.0, but there are some clear trends. In particular, the values measured for TEABr in chloroform are consistently lower than those of the other systems. Figure 6 shows the average β values for the three different HBD complexes as a function of solvent and counterion. There is a clear outlier with a β value about 0.5 lower than those of the other five systems. We ascribe this difference to stronger ion pairing of the TEABr salt in chloroform, which lowers the apparent β value by competing with H-bond formation.
The dielectric constant of a solvent is inversely proportional to the association constant for ion pairing of the salt. 39 Acetonitrile has a high dielectric constant (37.5), 40 so loose ion   The red data points were measured in chloroform and the blue data in acetonitrile.

Journal of the American Chemical Society
Article pairs are most likely to be present in this solvent, i.e., solventseparated or solvent-shared. 37 Chloroform has a low dielectric constant (4.8), 40 so contact ion pairs and higher aggregates are likely to predominate in this solvent. 36,38 The counter-cation is therefore more likely to affect H-bonding interactions with anions in chloroform. It appears that the larger tetraalkylammonium cations do not interact sufficiently strongly with the anion to affect the measured association constants in either solvent, but the smaller, more polar tetraethylammonium cation forms a tighter ion pair with the bromide anion in chloroform, and this interaction competes with H-bond formation. For more polar counter-cations, the effects of ion pairing are likely to be more important, but it appears that most of the measurements presented in Table 3 are not perturbed by such effects. The only other system where ion pairing plays a role is in the complexes formed by the TBMA salt 7. The methylammonium headgroup is relatively polar, and the complexes formed in chloroform result in significantly lower β values than the complexes in acetonitrile. The data obtained in chloroform were therefore not used in determining the average values of β for 21 or for 7 in Table 3.

■ DISCUSSION
The HBA properties of the anions are illustrated graphically in Figure 7. BzO − and AcO − are the strongest HBAs studied (β = 15.1 and 15.0, respectively), and PF 6 − is the weakest (β = 7.0). The β values measured for the halides fall within the ranges estimated from the literature (see Introduction). The HBA properties of Br − are similar to those of phosphine oxide 4, while I − is comparable to a sulfoxide HBA. 23 Of the four sulfonate ions involved in this study, MeSO 3 − has the largest β value (11.3), while HSO 4 − has a β value of 10.4, and CF 3 (CF 2 ) 3 SO 3 − and CF 3 SO 3 − have lower β values (9.4). The substantial differences in β for these ions highlight the influence that the electronic nature of substituents can have on HBA properties. Nitrate has a β value of 10.7, which is similar to those of Br − and 4. The β value of ReO4 − (9.4) is comparable to that of the weakest halide, I − , and lower β values of 8.4 and 7.3 were obtained for ClO 4 − and N(SO 2 CF 3 ) 2 − , respectively. The HBA capability of the perchlorate anion is similar to that of a urea functionality, while N(SO 2 CF 3 ) 2 − is comparable to pyridine. 14 The weakly coordinating anion PF 6 − has a β value of 7.0, which is significantly lower than that of the perchlorate ion on the HBA scale. Figure 8 compares the HBA parameters measured for the anions with the pK a values of the conjugate acids in water. 41 There is no relationship between the covalent interaction of an anion with a proton and the non-covalent interaction with a HBD, which means that pK a cannot be used for predicting HBA properties. There are some trends for anions within the same class, i.e., halides (black data points) and oxygen acceptors (red data points), but the correlations are weak. For example, CF 3 SO 3 − and ReO 4 − have comparable HBA properties (β = 9.4) but very different pK a values (−5.9 and −1.3).

■ CONCLUSION
UV/vis absorption titrations have been used to characterize the H-bonded complexes formed between 15 different anions and three different H-bond donors in chloroform and in acetonitrile. The data are successfully described by the solvent competition model, allowing the determination of the H-bond acceptor parameter (β) for each of the anions. The transferability of the HBA parameters means that they can be used for estimating the properties of non-covalent complexes formed between anions and any HBD in any solvent environment.
Carboxylates are the strongest H-bond acceptors studied, with a HBA strength that surpasses those of all previously reported neutral species, and PF 6 − is the weakest charged HBA studied. 14 The variation in β values measured for a series of sulfonate anions shows that the electronic effects of substituents are similar to those observed for neutral HBA. The influence of ion pairing on the measured β values was investigated through systematic variation of the counter-cation. Ion pairing only competes with H-bond formation for small polar countercations (tetraethylammonium) in less polar solvents (chloroform), and the β values reported here are not perturbed by the effects of ion pairing. There is no relationship between the Hbonding properties of the anions and the pK a values of the conjugate acids.
We anticipate that the quantification of HBA parameters for anions will be valuable in disentangling the roles of solvent and receptor in anion recognition and in facilitating the future design of complex supramolecular architectures.

Journal of the American Chemical Society
Article mM in MeCN (2), 0.022 and 0.043 mM in MeCN (3); 0.12 and 0.14 mM in CHCl 3 (1), 0.070 and 0.12 mM in CHCl 3 (2), and 0.035 and 0.049 mM in CHCl 3 (3)]. A 2 mL portion of this solution was removed and added to a quartz cuvette, and the UV/vis spectrum was recorded. The guest (4−22) was dissolved in 1−2 mL of the host solution. Aliquots of this solution were successively added to the cuvette, and the UV/vis absorption spectrum was recorded after each addition. The UV/vis absorption spectra were analyzed using a Microsoft Excel spreadsheet to fit the changes in absorption at fixed wavelengths to a 1:1 binding isotherm by optimizing the association constant and absorption of the free and bound host using purposewritten VBA macros.