Cooperative anion binding and electrochemical sensing by modular podands

Abstract

A series of podands based on two or three hydrogen bonding “arms” situated in mutually ortho, meta, or para relationships about an aryl core have been prepared, and their affinities for simple inorganic anions were measured. Of the two-arm hosts the meta compound and to a lesser extent the ortho host exhibit a cooperative anion binding effect. The two arms function essentially independently in the para derivative. The mutually meta three-arm host shows dramatically enhanced cooperative binding. Conformational changes within the meta two-arm host result in significantly enhanced electrochemical anion sensing compared with the more conformationally rigid three-arm host.

The study of the noncovalent binding (1–13), sensing (14–16), extraction (17), and reactivity (18) of anionic species has become an area of remarkable importance and topicality over the past decade. Very real intrinsic challenges (2, 6, 12, 19) in anion binding chemistry (in comparison to the analogous chemistry of cations) have spawned a plethora of imaginative and sophisticated hosts, some exhibiting significant interanion discrimination in terms of either reactivity or especially binding affinity. For anion-sensing applications, in particular, the challenge faced by chemists goes beyond the achievement of strong and highly selective anion binding and becomes one of detecting and amplifying an anion binding event to produce a measurable output (8, 9, 14, 20–22). The host system therefore must be capable of selective binding, translation of the binding event into a signal, signal transduction via a spacer or linker to a signaling unit, and readout (14). Commonly, electrochemical anion sensing has been achieved potentiometrically (23) by either neutral ferrocenyl species or cationic cobaltocinium or ruthenium π-arene derivatives (1, 9, 20, 22, 24–28). Observed anion-induced changes in metallocene redox potential are generally in the region of 50–150 mV, although ΔE values of up to 240 mV have been reported for cobaltocinium-based sensors (9, 20). These significant electrochemical responses may be attributed to through bond and through space effects arising from the second sphere coordination of the target anion to the metallocene. It has been shown by extensive work by Beer et al. (9, 20, 22), amongst others, that amide spacer units give good coupling between the binding and readout sites. In luminescent sensors an alternative method of signal transduction from binding to readout involves a binding-induced conformational change that can bring two lumophores into close proximity to one another, triggering excimer or exciplex emission (16, 29–31). We now report generation of electrochemical systems that adopt such a conformational change-based approach to anion sensing. A small part of this work has appeared in a preliminary communication (32).

Methods

Mass spectra were run at King's College London on a Jeol AX505W spectrometer in fast atom bombardment (FAB) mode in a thioglycerol or nitrobenzyl alcohol matrix. NMR spectra were recorded on a Bruker AV-400 spectrometer operating at 400 MHz. IR spectra were recorded on a Perkin–Elmer Paragon 100 Fourier transform IR spectrometer as Nujol mulls. Microanalyses were performed at the University of North London. All reactions were carried out under nitrogen, although the products showed no oxygen or moisture sensitivity. Many crystalline solid products readily lost enclathrated solvent. NMR titrations were carried out by using either a Bruker AV-400 or AV-500 spectrometer operating at 400 and 500 MHz, respectively, at room temperature. All chemical shifts are reported in ppm relative to tetramethylsilane.

3-Pyridylferrocenylmethylamine (1).

Ferrocenecarboxaldhyde (1.00 g, 4.67 mmol) and 3-aminopyridine (0.440 g, 4.68 mmol) were dissolved in CH₂Cl₂ (50 ml) and stirred for 6 h at reflux. Sodium borohydride (1.77 g, 46.8 mmol) was added, and stirring continued for 1 h further. After this period the excess borohydride was destroyed by the dropwise addition of HCl (2M) until effervescence was complete and the solution was slightly acidic. NaOH (2 M) then was added until the solution was slightly basic. The mixture was extracted into CH₂Cl₂, washed with water, and dried over MgSO₄. Evaporation to dryness gave the pure product as an orange solid (1.25 g, 4.34 mmol, 93%). ¹H NMR (CHCl₃-d, J/Hz, δ/ppm): δ 3.85 (s, br, 1H, NH); 3.92 (d, 2H, J = 4.7, CH₂); 4.08–4.12 (m, 7H, Fc); 4.17 (s, 2H, CH₂); 6.85 (ddd, 1H, J = 8.3, 3.0, 1.4, ArH); 7.03 (dd, 1H, J = 8.3, 4.7 Hz, ArH); 7.92 (dd, 1H J = 4.7, 1.4, ArH); 8.01 (d, 1H J = 3.0, ArH). FAB-MS: m/z = 292 [M⁺]. IR (ν/cm⁻¹): 3248 s (NH). Anal. calcd for C₁₆H₁₆N₂Fe: C, 65.78; H, 5.52; N, 9.59%. Found: C, 65.81; H, 5.45; N, 9.52%.

α,α′-Bis(3-pyridylferrocenemethylamine)-ortho-xylene Hexafluorophosphate (3a).

α,α′-Dibromo-o-xylene (2a, 0.5 g, 1.89 mmol) and 1 (1.10 g, 3.79 mmol) were dissolved in ethyl acetate (50 ml) and refluxed for 6 h. After this time, a light orange solid was observed in the flask, and the solution had turned almost colorless. The solution was filtered, and the solid was washed with ethyl acetate and dried to yield 3a as the bromide salt (0.989 g, 1.17 mmol, 62%). ¹H NMR (CHCl₃-d, J/Hz, δ/ppm): δ 4.00 (m, 4H, CH₂); 4.07 (s, 4H, Fc); 4.15 (s, 10H, Fc); 4.24 (s, 4H, Fc); 6.18 (s, 4H, CH₂); 7.00 (s, br, 2H, NH); 7.15 (m, 2H, ArH); 7.25 (m, 2H, ArH); 7.35 (m, 2H, ArH); 7.49 (s, 2H, ArH); 8.27 (s, 2H, ArH); 8.70 (s, 2H, ArH). FAB-MS: m/z = 767 [M⁺-Br]. The above bromide salt (0.989 g, 1.17 mmol) was dissolved in dichloromethane (50 ml) and added to a solution of ammonium hexafluorophosphate (20 equivalents, 3.81 g, 23.4 mmol) in water. The mixture was stirred vigorously for 6 h to ensure thorough mixing of the layers. The mixture was separated, and the organic layer was evaporated to dryness to yield 3a (0.869 g, 0.889 mmol, 76%). ¹H NMR (MeCN-d₃, J/Hz, δ/ppm): δ 4.06 (d, 4H, J = 5.8, CH₂); 4.14 (s, 4H, Fc); 4.20 (s, 10H, Fc); 4.21 (s, 4H, Fc); 5.58 (s, 4H, CH₂); 5.88 (s, br, 2H, NH); 7.31–7.38 (m, 2H, ArH); 7.51–7.56 (m, 2H, ArH); 7.57–7.70 (m, 8H, ArH). FAB-MS: m/z = 833 [M⁺-PF₆]. Anal. calcd for C₄₀H₄₀Fe₂N₄P₂F₁₂: C, 49.10; H, 4.12; N, 5.73%. Found: C, 48.97; H, 4.17; N, 5.68%.

α,α′-Bis(3-pyridylferrocenemethylamine)-meta-xylene Hexafluorophosphate (3b).

α,α′-Dibromo-meta-xylene (2b, 0.51 g, 1.91 mmol) and 1 (1.13 g, 3.87 mmol) were dissolved in dichloromethane (50 ml) and refluxed for 18 h. After this time the resulting dark orange solution was evaporated to dryness to leave an orange powder of 3b as the bromide salt (1.47 g, 1.73 mmol, 91%). ¹H NMR (CHCl₃-d, J/Hz, δ/ppm): δ 4.02–4.08 (m, 8H, Fc and CH₂); 4.15 (s, 10H, Fc); 4.23 (s, 4H, Fc); 5.70 (s, 4H, CH₂); 7.18–7.53 (m, 7H, ArH); 7.62 (d, 2H, J = 7.3, ArH); 8.07 (s, 1H, ArH); 8.78 (d, 2H, J = 5.0, ArH); 9.40 (s, 2H, ArH).

The above bromide salt (1.47 g, 1.73 mmol) was dissolved in dichloromethane and added to a solution of ammonium hexafluorophosphate (6.37 g, 39.08 mmol) in water. The mixture was stirred vigorously for 4 h. The layers were separated, and the organic layer was washed with water before being dried over MgSO₄ and filtered. The solution was evaporated under reduced pressure to yield a dark orange/brown crystalline solid (1.04 g, 1.06 mmol, 61%). ¹H NMR (MeCN-d₃, J/Hz, δ/ppm): δ 4.12 (m, 4H, CH₂); 4.16 (m, 4H, Fc); 4.20–4.24 (m, 14H, Fc); 5.53 (s, 4H, CH₂); 5.89 (s, br, 2H, NH); 7.30–7.59 (m, 4H, ArH); 7.64 (m, 4H, ArH); 7.89 (m, 4H, ArH). FAB-MS: m/z = 977 [M⁺-H], 833 [M⁺-PF₆]. Anal. calcd for C₄₀H₄₀Fe₂N₄P₂F₁₂: C, 49.10; H, 4.12; N, 5.73%. Found: C, 48.99; H, 4.25; N, 5.64%.

α,α′-Bis(3-pyridylferrocenemethylamine)-para-xylene Hexafluorophosphate (3c).

α,α′-Dibromo-para-xylene (2c, 0.500 g, 1.91 mmol) and compound 1 (1.11 g, 3.78 mmol) were dissolved in dichloromethane (50 ml) and refluxed for 2 h. During this time a dark orange precipitate formed in the reaction flask. The flask was cooled, and the mixture was filtered to obtain 3c as the bromide salt (0.95 g, 1.11 mmol, 58%). ¹H NMR (MeOH-d₄, J/Hz, δ/ppm): δ 4.12 (m, 4H, Fc); 4.18–4.22 (s, 10H, Fc); 4.25 (m, 4H, Fc); 4.63 (s, 4H, CH₂); 5.69 (s, 4H, CH₂); 7.56 (s, 4H, ArH); 7.68–7.71 (m, 4H, ArH); 8.12–8.2 (m, 4H, ArH).

The bromide salt was converted to the hexafluorophosphate salt by a counterion metathesis. The above product (0.95 g, 1.11 mmol) was dissolved in methanol (100 ml) with ammonium hexafluorophosphate (20 equivalents, 3.64 g, 22.32 mmol). The solution was heated slightly (≈35°) to increase the solubility of the bromide salt and stirred for 1 h. After cooling a yellow precipitate formed. This precipitate was filtered and washed with methanol to leave the hexafluorophosphate salt (0.61 g, 0.62 mmol, 56%). ¹H NMR (MeCN-d₃, J/Hz, δ/ppm): δ 4.12–4.14 (m, 8H, Fc and CH₂); 4.22 (s, br, 14H, Fc); 5.54 (s, 4H, CH₂); 5.91 (s, 2H, NH) disappears after D₂O shake; 7.74 (m, 4H, ArH); 7.62–7.66 (m, 4H, ArH); 7.90–7.95 (m, 4H, ArH). FAB-MS: m/z = 978 [M⁺], 833 [M⁺-PF₆]. Anal. calcd for C₄₀H₄₀Fe₂N₄P₂F₁₂: C, 49.10; H, 4.12; N, 5.73%. Found: C, 48.95; H, 4.06; N, 5.61%.

1,3,5-Tris(3-pyridiniumferrocenylmethylamine)-2,4,6-triethyl Benzene Hexafluorophosphate (4).

Compound 1 (1.00 g, 3.47 mmol) and 1,3,5-tri(bromomethyl)triethyl benzene (1.53 g, 3.47 mmol) was stirred for 12 h in CH₂Cl₂ (100 ml). After this time, the solution was evaporated to dryness to give the desired product as the tribromide salt. ¹H NMR (CH₃CN-d₃, J/Hz, δ/ppm): δ 0.99 (t, 9H J = 7.3, CH₃), 2.42 (q, 6H J = 7.3, CH₂); 4.01 (s, br, 6H CH₂); 4.11 (m, 27H, Fc); 5.65 (s, 6H, CH₂-py⁺); 6.95 (s, br, 3H, NH); 7.45 (m, 3H, pyH); 7.62 (m, 3H, pyH); 8.10 (s, br, 3H, pyH); 8.59 (s, 3H, pyH). FAB-MS: m/z = 1,235 [M⁺-Br]. Anal. calcd for C₆₃H₆₉Br₃N₆Fe₃⋅2H₂O: C, 55.90; H, 5.44; N, 6.21%. Found: C, 55.65; H, 5.73; N, 6.49%. Counterion metathesis to give the hexafluorophosphate salt was achieved by stirring a solution of the tribromide salt (1.00 g, 0.77 mmol) in CH₂Cl₂ (50 ml) with a solution of NaPF₆ (3.86 g, 23 mmol) in water (50 ml). After 1 h the organic layer was separated, washed with water, and dried over MgSO₄. Evaporation to dryness gave the pure host as the hexafluorophosphate salt (1.03 g, 89%). ¹H NMR (CH₃CN-d₃, J/Hz, δ/ppm): δ 0.88 (t, 9H J = 7.1, CH₃); 2.56 (q, 6H J = 7.3, CH₂); 4.13 (d, 6H J = 5.0, CH₂); 4.17 (s, 6H, Fc); 4.22 (s, 15H, Fc); 4.24 (s, 6H, Fc); 5.68 (s, 6H, CH₂); 5.90 (s br, 3H NH) disappears after D₂O shake; 7.56–7.64 (m, 9H ArH); 7.84 (s br, 3H, ArH). FAB-MS: m/z = 1,367 [M⁺-PF₆]. IR (ν/cm⁻¹): 3,415 s (NH), 842 s PF₆. Anal. calcd for C₆₃H₆₉F₁₈N₆Fe₃P₃: C, 50.02; H, 4.60; N, 5.56%. Found: C, 50.12; H, 4.52; N, 5.55%.

X-Ray Crystallography.

Crystal data for 3c(PF₆)₂⋅H₂O: C₄₀H₄₂F₁₂Fe₂N₄OP₂, M 996.40 g⋅mol⁻¹; triclinic, space group P₁, a = 9.792(2); b = 10.648(2); c = 10.664(2) Å; α = 74.750(3); β = 69.810(3); γ = 85.170(3)°. U = 1,006.8(3) ų; Z = 1; μ = 0.895 mm⁻¹; T = 100 K. Reflections measured, 3,696; unique data, 2,847; parameters, 272, R₁ [F² > 2σ(F²)] 0.2634, wR₂ (all data) 0.5817. Crystal data for 3c(PF₆)₂⋅2Me₂CO: C₄₆H₅₂F₁₂Fe₂N₄O₂P₂, M 1,094.56 g⋅mol⁻¹; triclinic, space group P₁; a = 8.2620(17); b = 11.672(2); c = 12.490(3) Å; α = 81.325(5); β = 77.986(5); γ = 87.710(5)°. U = 1164.6(4) ų; Z = 1; μ = 0.783 mm⁻¹; T = 120 K. Reflections measured, 6,134; unique data, 3,907; parameters, 313, R₁ [F² > 2σ(F²)] 0.1411, wR₂ (all data) 0.2622. In both cases crystals proved to be extremely difficult to obtain, and the best samples were extremely weakly diffracting and showed evidence of twinning. As a result the overall precision of the structures is extremely poor; however, the locations of the anions and the overall molecular conformation are unambiguous. For a description of data collection and refinement in our laboratory see ref. 33.

Molecular Modeling.

Simulations were carried out by using PC SPARTAN PRO, version 1.05 (Wavefunction, Irvine, CA). Conformational analysis of tris(pyridinium) compound 4 used the MONTE CARLO search engine starting from the geometry of the crystal structure (J.W.S. and W.J.B., unpublished work) to generate starting points from which cone and partial cone models for 4 were constructed. Molecular mechanics using the Merck molecular force field were used to determine the optimum host and host–guest complex geometries. Interaction energies (8) were calculated by using: E (host–guest complex) − E (host) − ΣE (guests). The interaction energies of the 1:1 complexes (with 4 in the “3-up” conformation) were calculated to be −231 kcal⋅mol⁻¹ for Cl⁻ but only −99 kcal⋅mol⁻¹ for PF. No consideration was given to energies associated with desolvation of anions before ligand binding.

Electrochemistry.

Electrochemical experiments were performed in a low-volume three-electrode cell in dry, degassed acetonitrile, analyte concentration of 5 × 10⁻⁴ M. The background electrolyte was NBu₄PF₆ (0.1 M). The working electrode was a Pt disk of 2-mm diameter, and the auxiliary electrode was a Pt wire. Potentials are reported relative to Ag/AgO in background electrolyte, against which ferrocene is oxidized at 0.512 V. To prevent fouling, the working electrode was polished before each addition of anion.

Results and Discussion

A series of bis(ferrocenyl) molecular clips were designed and synthesized according to the straightforward, modular route outlined in Scheme S1. Yields ranged from good to excellent, and the approach is highly amenable to variation. The design of hosts 3a–c comprises an aryl “core” with two flexible ferrocenyl “arms” situated in a mutually ortho, meta, or para geometry. Each arm contains a cationic pyridinium moiety exerting an electrostatic attraction for anions. Selectivity is introduced by the presence of a directional hydrogen-bonding NH functionality as well as the weaker directionality imparted by pyridinium CH⋅⋅⋅anion interactions. The NH primary binding site is coupled to the ferrocenyl signaling moiety solely via a saturated CH₂ linker, which should greatly reduce through-bond effects of anion binding on the potential of the Fe(II)/Fe(III) redox couple. It is anticipated that compounds 3a–c will adopt a transoid conformation in the presence of weakly coordinating anions in which the two NH groups are far apart on opposite sides of the molecule. The addition of one equivalent of more tightly bound anions should result in a conformational change to a cisoid form resulting from chelation of the anion guest by the two NH moieties accompanied by electrochemical changes resulting from the increased proximity of the two ferrocenyl units (e.g., Scheme S2). The degree to which this conformational change occurs should depend on the anion in question and the substitution pattern at the aryl core.

Scheme 1.

A modular approach to flexible and preorganized cationic hosts for anions.

Scheme 2.

Proposed anion-induced conformational switching in molecular clips of type 3.

The transoid noninteracting disposition of the arms in the hexafluorophosphate salt of para isomer 3c was confirmed in the solid state by two single crystal x-ray structure determinations of 3c⋅H₂O (crystallized from acetonitrile; Fig. 1a) and 3c⋅2Me₂CO (crystallized from acetone; Fig. 1b; see Methods for details). In the former case the single crystallographically independent NH group forms a bifurcated hydrogen-bonded interaction to two fluoro groups of the PF anion. The host–anion interaction is supported further by a pyridyl CH⋅⋅⋅F hydrogen bond. The weakly coordinating nature of the PF anion is highlighted by the structure of the acetone solvate 3c⋅2Me₂CO. In this case the NH and adjacent CH groups interact with the oxygen atom of the included acetone guest, whereas the PF anions are held solely by a series of weak CH⋅⋅⋅F hydrogen bonds (34–40). In both cases it is evident that each PF is bound weakly by a single arm in a nonchelating fashion, as expected.

Figure 1.

(a) X-ray crystal structure of podand 3c⋅H₂O showing noncooperative anion binding. (b) Structure of 3c⋅2Me₂CO showing hydrogen bonds to acetone solvent molecules. Hexafluorophosphate anions are bound by weak CH⋅⋅⋅F interactions.

Hosts 3a–c were titrated with a variety of anions (as the NBu salts) in acetonitrile solution, and the resultant changes to the ¹H NMR spectra were monitored. The resulting binding constants calculated by using HYPNMR 2000 (41, 42) are shown in Table 1. Significant problems were encountered with precipitation in some cases; however, K₁₁ values were obtained for all three two-arm hosts with NO. Within the series, there is considerable variation with the meta host, 3b, binding nitrate five times more strongly than the para compound 3c despite the identical composition of the binding sites. This result suggests that although the anion binding arms in 3c act independently (as in the x-ray structural results), the arms in the meta species and to some extent the ortho compound display a marked chelate effect resulting in dramatically enhanced binding. The meta host 3b also proved effective at binding a range of other anions including halides and particularly acetate despite the host's flexible (and hence nonpreorganized) nature. Interanion selectivity apparently is governed by negative charge density considerations, with only the relatively diffuse perrhenate and triflate not being bound at all (in preference to the PF counter ion). For comparison, the one-arm model compound 5 was also prepared, and its affinity for various anions was measured. No data could be obtained for the majority of anions because of solubility constraints; however, 5 binds much more weakly to acetate in acetonitrile, which is consistent with the absence of an anion chelate effect. This result demonstrates the existence of cooperativity between the two arms in compounds 3a and 3b in particular.

Table 1.

Binding constants (K₁₁, M⁻¹) determined by ¹H NMR titration for the interaction of the new hosts with various anions

Anion	Host K11, M−1
	3a	3b	3c	4	5
Cl−	*	1,340	*	17,380	*
Br−	*	*	*	2,950	*
I−	*	282	*	1,860	*
NO	462	1,233	263	1,410	*
CH3CO	—	4,515	*	3,680	126
ReO	—	≈0	—	4	—
CF3SO	—	≈0	—	—	—

Anions as n-Bu₄N⁺ salts in MeCN-d₃, host o-pyridyl, NH, and CH₂ protons were monitored. Binding constants were determined with the aid of HYPNMR 2000 software with a 1:2 or 1:3 host/guest model assumed for compounds 3 or 4, respectively. Errors in K are ≈5–15% (41, 42). 

^*Precipitation makes binding constant determination impossible. 

The much more conformationally rigid three-arm host 4 also was prepared in 89% yield from 1,3,5-tri(bromomethyl)-2,4,6-triethyl benzene (43) in the same way as compound 3. The presence of the three ethyl groups has been shown to impart a significant energetic preference (10–15 kJ⋅mol⁻¹; refs. 44–46) for alternation about the aryl core, resulting in a preorganized 3-up, 3-down conformation for 4 that should prove much more effective and selective for binding smaller anions such as halides but will not undergo the kind of conformational change shown in Scheme S2 after anion binding.

NMR titration results indicate that 4 is an extremely effective host for halide anions with a very high binding constant for chloride. The marked selectivity sequence Cl⁻ > Br⁻ > I⁻ is consistent with the size match of the anions for the host conical cavity and negative charge density considerations (Table 1). Halide complexation-induced chemical shift changes (Δδ) were up to 0.85 ppm for the pyridyl CH singlet protons, H(s), and 1.59 ppm for the NH protons. Most spectral changes were almost complete after the addition of one equivalent of anion, suggesting that binding occurs first at the central cavity via a sixfold array of C-H⋅⋅⋅X⁻ and N-H⋅⋅⋅X⁻ hydrogen bonds. Interestingly, host 4 is much better able to discriminate between the spherical halides and the trigonal acetate and nitrate compared with two-arm hosts of type 3. Although nitrate is bound approximately equally effectively by the comparable meta compounds 3b and 4, chloride binding by 4 is an order of magnitude more effective than 3b.

The acetate anion gives remarkably large chemical shift changes attributed to its high basicity (Δδ 3.52 ppm for the NH protons in 4) but is bound slightly less effectively by 4 than 3b. Close examination of the behavior of various nuclei as a function of chemical shift suggests that acetate binding by 4 may occur by a different mechanism to the binding of halides (Fig. 2). The very gradual change in the chemical shift behavior of the pyridyl singlet resonance, pyH(s), after addition of acetate is in marked contrast to the near saturation of this shift at one equivalent of halide anion and suggests that this hydrogen atom is not involved directly in binding the first anion, which in turn implies that the first acetate anion is bound in a chelating fashion by two NH moieties from adjacent arms. Given the bulk of the acetate anion, it is likely that this binding occurs in a 2-up, 1-down conformation. Binding of the second acetate anion to the remaining free arm then involves both the NH and pyridyl CH protons as it does for halides. We have shown that this effect is even more pronounced in the tris(anthracenyl) analogue of 4 in which the greater bulk of the anthracene moieties destabilizes the 3-up conformation (32).

Figure 2.

NMR titration curves for compound 4 with MeCO₂⁻ (a) and Br⁻ (b). The behavior of the pyridyl singlet H(s) [PyH(s)] is markedly different in the two compounds.

The involvement of the 3-up cone conformation in halide binding is supported by molecular mechanics calculations (see Methods), which indicate that that chloride may be bound within the molecular cavity in a sixfold array of weak and strong hydrogen bonds to NH and pyridyl CH protons (consistent with observed ¹H NMR chemical shift changes). The calculated interaction energy proved significantly greater for Cl⁻ than PF. The cone (3-up) conformer is favored over the partial cone conformer with one ferrocenyl group on the opposite side of the molecule to the other two (2-up, 1-down; Fig. 3).

Figure 3.

Model of the cation in 4 showing a bound Cl⁻ anion enveloped in a sixfold array of C-H⋅⋅⋅Cl⁻ and N-H⋅⋅⋅Cl⁻ hydrogen bonds (calculated NH⋅⋅⋅Cl distances: 2.56, 2.65, and 2.87 Å). The existence of these interactions is confirmed by NMR titration and the x-ray crystal structure of 3c. Molecular mechanics results show that the interaction energies are far smaller for the analogous hexafluorophosphate complex.

The electrochemical response of hosts 3a–c and 4 (as the hexafluorophosphate salts) as a function of added (NBu₄)X concentration (X = Cl, Br, I, NO₃) was examined by using cyclic voltammetry. Hosts 3a–c all displayed a single reversible redox couple centered on 0.58 V vs. Ag/AgO in acetonitrile solution. After titration with up to 5 equivalents of Cl⁻, a linear cathodic shift in the E_1/2 value was observed, reaching 64 mV for 3b. Analogous titration with Br⁻ gave a shift of ≈40 mV response for 3a and 3b but a smaller shift for 3c (28 mV). With NO, 3a gave the largest response of 37 mV, whereas 3b and 3c both shifted by ≈20 mV. The addition of iodide gave almost no change in the redox potential. More interestingly, the highly preorganized 4, which exhibits a much larger binding constant for halides (Cl⁻ in particular, Table 1), shows a much more modest electrochemical response. Compound 4 displays a single oxidation wave: E_1/2, 0.56 V. Titration with up to 5 equivalents of NBu₄X (aliquots of 1 equivalent at a time) gave small total shifts ranging from 36 (NO) to 29 (Cl⁻), 22 (Br⁻), and 10 mV (I⁻). The magnitudes of the shifts for compounds 3 and 4 are unremarkable when compared with related host species (8, 9, 20, 25) and suggest limited through-bond coupling between the binding and signaling moieties. However, the selectivity pattern Cl⁻ > Br⁻ > I⁻ is in good agreement with that obtained by NMR and supports the model of effective, chelate anion binding of all the anions studied. The large electrochemical response of 3b compared with 4 strongly suggests a significant conformational change in 3b after halide binding (cf. Scheme S2) that is not found in the preorganized 4. Conversely, nitrate binding seems to alter the conformation of 4, perhaps because of its steric bulk. Thus, the flexible two-arm host 3b exhibits more effective binding signal transduction for Cl⁻ and Br⁻ despite its intrinsically weaker binding. This surprising result points the way for the design of efficient signaling in molecular electrochemical sensors.

In conclusion, a series of redox-responsive anion hosts have been developed displaying a marked anion chelate effect and well defined structural selectivity. The flexible, modular synthetic approach adopted means that a wide range of hosts of varying structure may be prepared readily in high yield. The more flexible, less preorganized host species show poorer binding and selectivity, as might be expected. However, binding-induced conformational change in these species results in efficient electrochemical signaling.

Scheme 3.

Abbreviation

FAB

fast atom bombardment