Organic & Biomolecular Chemistry

Cyclodextrin and modified cyclodextrin complexes of E-4-tert-butylphenyl-4-oxyazobenzene: UV-visible, ¹H NMR and ab initio studies

Bruce L. May^a, Jacobus Gerber^a, Philip Clements^a, Mark A. Buntine^a, David R. B. Brittain^a, Stephen F. Lincoln*^a and Christopher J. Easton^b
aDepartment of Chemistry, University of Adelaide, Adelaide, SA 5005, Australia. E-mail: Stephen.Lincoln@adelaide.edu.au
^bResearch School of Chemistry, Australian National University, Canberra, ACT 0200, Australia

-Cyclodextrin, -cyclodextrin, N-(6^A-deoxy--cyclodextrin-6^A-yl)-N-(6^A-deoxy--cyclodextrin-6^A-yl)urea and N,N-bis(6^A-deoxy--cyclodextrin-6^A-yl)urea (CD, CD, 1 and 2) form inclusion complexes with E-4-tert-butylphenyl-4-oxyazobenzene, E-3^–. In aqueous solution at pH 10.0, 298.2 K and I = 0.10 mol dm^–3 (NaClO₄) spectrophotometric UV-visible studies yield the sequential formation constants: K₁₁ = (2.83 ± 0.28) × 10⁵ dm³ mol^–1 for CD·E-3^–, K₂₁ = (6.93 ± 0.06) × 10³ dm³ mol^–1 for (CD)₂·E-3^–, K₁₁ = (1.24 ± 0.12) × 10⁵ dm³ mol^–1 for CD·E-3^–, K₂₁ = (1.22 ± 0.06) × 10⁴ dm³ mol^–1 for (CD)₂·E-3^–, K₁₁ = (3.08 ± 0.03) × 10⁵ dm³ mol^–1 for 1·E-3^–, K₁₁ = (8.05 ± 0.63) × 10⁴ dm³ mol^–1 for 2·E-3^– and K₁₂ = (2.42 ± 0.53) × 10⁴ dm³ mol^–1 for 2·(E-3^–)₂. ¹H ROESY NMR studies show that complexation of E-3^– in the annuli of CD, CD, 1 and 2 occurs. A variable-temperature ¹H NMR study yields k(298 K) = 6.7 ± 0.5 and 5.7 ± 0.5 s^–1, H = 61.7 ± 2.7 and 88.1 ± 4.2 kJ mol^–1 and S = –22.2 ± 8.7 and 65 ± 13 J K^–1 mol^–1 for the interconversion of the dominant includomers (complexes with different orientations of CD) of CD·E-3^– and (CD)₂·E-3^–, respectively. The existence of E-3^– as the sole isomer was investigated through an ab initio study.

The aim of this study is to improve insight into the structural and mechanistic factors controlling the formation of cyclodextrin (CD) and linked CD dimer inclusion complexes. The significance of the linked CD dimers is that CDs are linked together so that their simultaneous complexation of a guest species may be varied as the sizes of the CD annuli are changed. We have chosen to study the interactions of CD and CD separately and as components of N-(6^A-deoxy--cyclodextrin-6^A-yl)-N-(6^A-deoxy--cyclodextrin-6^A-yl)urea¹ (1) and N,N-bis(6^A-deoxy--cyclodextrin-6^A-yl)urea² (2) with E-4-tert-butyl-4-oxyazobenzene (E-3^–) (Scheme 1) through the complexes CD·E-3^–, (CD)₂·E-3^–, CD·E-3^–, (CD)₂·E-3^–, 1·E-3^–, 2·E-3^– and 2·(E-3^–)₂. While the effects of the difference in annular size of CD and CD on complex stoichiometry and stability have been widely studied^3,4 their impact on complexation by linked CDs, exemplified by 1 and 2, is less explored,^1,5,6 as are their effects on rate processes exemplified by the exchange between the includomers (complexes with different orientations of CD) of CD·E-3^– and (CD)₂·E-3^– reported here. The choice of E-3^– is based on it possessing a large hydrophobic 4-tert-butylphenyl group which discriminates between the annular sizes of CD and CD and a smaller and less hydrophobic 4-oxybenzene group which discriminates to a lesser extent. In contrast to the azobenzenes which exist in thermally and photochemically controlled equilibria of their E and Z isomers,^7–10 only E-3^– was observed under the conditions of this study. Ab initio modelling¹¹ has been undertaken to gain insight into the existence of E-3^– as the sole isomer.

	Scheme 1

The parent guest species E-4-tert-butylphenyl-4-hydroxyazobenzene, E-3H, and its CD complexes are insufficiently water soluble for convenient study. As a consequence the much more soluble E-4-tert-butylphenyl-4-oxyazobenzene, E-3^–, and its CD and CD dimer complexes were studied at 298.2 K in aqueous 0.05 mol dm^–3 borate buffer at pH 10 and I = 0.10 mol dm^–3 (NaClO₄). The variations of the molar absorbance of E-3^– with CD, CD, 1 and 2 concentrations are shown in Figs. 1–3 and Figs. S1–S5, ESI, and the derived complexation constants appear in Table 1. Complexation by CD causes a much greater molar absorbance change than does CD consistent with the smaller size of the CD annulus producing a greater change in the environment of E-3^– (Figs. 1 and 2, Figs. S1 and S2, ESI). Both CD·E-3^– and (CD)₂·E-3^– are formed as are their CD analogues. In addition to the statistical effect on the relative magnitudes of sequential complexation constants, the differences in hydrophobicity and size between the two ends of E-3^– accentuate the difference in the magnitudes of K₁₁ and K₂₁ (Table 1).

Complex	K11/dm3 mol–1	K21/dm3 mol–1	K12/dm3 mol–1
CD·E-3–	(2.83 ± 0.28) × 105
(CD)2·E-3–		(6.93 ± 0.06) × 103
CD·E-3–	(1.24 ± 0.12) × 105
(CD)2·E-3–		(1.22 ± 0.06) × 104
1·E-3–	(3.08 ± 0.03) × 105
2·E-3–	(8.05 ± 0.63) × 104
2·(E-3–)2			(2.42 ± 0.53) × 104

	Fig. 1 The change in the UV-visible molar absorbance of E-3–(indicated by arrows) with increase in [CD]total in aqueous borate buffer solution at pH 10.0, 298.2 K and I= 0.10 mol dm–3(NaClO4); [E-3–]total= 1.92 × 10–5 mol dm–3 and [CD]total= 1.20 × 10–6–1.20 × 10–3 mol dm–3.

	Fig. 2 The change in the UV-visible absorbance of E-3–(indicated by arrows) with increase in [CD]total in aqueous borate buffer solution at pH 10.0, 298.2 K and I= 0.10 mol dm–3(NaClO4): [E-3–]total= 1.86 × 10–5 mol dm–3 and [CD]total= 1.22 × 10–6–1.22 × 10–3 mol dm–3.

The two isosbestic points seen in the molar absorbance variation of E-3^– with [1]_total (Fig. 3 and Fig. S3, ESI) are consistent with E-3^– and 1·E-3^– being the dominant species in solution whereas the spectral variation of E-3^– with [2]_total (Figs. S4 and S5, ESI) is consistent with the formation of 2·E-3^– and 2·(E-3^–)₂ as shown in Scheme 2 wherein CD and CD are shown as truncated cones where the narrow ends represent the primary hydroxy ends of the annuli. Despite the linking of CD and CD in 1 and 2 there is little cooperativity between them in complexing E-3^– in 1·E-3^– and 2·E-3^– as shown by the similarity of their K₁₁ values to those of CD·E-3^– and CD·E-3^–. (This contrasts with the complexation of the Methyl Orange anion by CD and 2 for which K₁₁ = 2.16 × 10³ and 1.05 × 10⁵ dm³ mol^–1, respectively, despite the structural similarity of this anion and E-3^–as shown in Scheme 1.⁵) The four systems show a variation of only a factor of 3.8 in K₁₁ for CD·E-3^–, CD·E-3^–, 1·E-3^– and 2·E-3^–. The complexation of a second E-3^– in 2·(E-3^–)₂ is characterised by K₁₂ which is only a factor of 3.3 less than K₁₁ for 2·E-3^–. It appears that the hydrophobic 4-tert-butylphenyl groups of the two E-3^– are complexed in each CD component annulus of 2. This is consistent with the analogous 1·(E-3^–)₂ complex not forming at detectable levels probably because the CD annulus is too small to accommodate the 4-tert-butylphenyl group of E-3^– and complexation of the charged 4-oxybenzene group is weaker such that the formation of 1·(E-3^–)₂ does not compete effectively with the formation of 1·E-3^–. This infers significant stereochemical differences in the complexing of E-3^– by either CD or CD alone or as components of 1 and 2, as is supported by the ¹H NMR studies discussed below. Nevertheless, all of the complexes exhibit high stabilities within the general range of CD complex stabilities.

	Fig. 3 The change in the UV-visible absorbance of E-3–(indicated by arrows) with increase in [1]total in aqueous borate buffer solution at pH 10.0, 298.2 K and I= 0.10 mol dm–3(NaClO4); [E-3–]total= 1.92 × 10–5 mol dm–3 and [1]total= 2.28 × 10–7–2.28 × 10–4 mol dm–3. Isosbestic points occur at 394 and 430 nm.

	Scheme 2

The ¹H NMR spectra of E-3^– and its CD complexes were run at higher concentrations than those used in the UV-vis studies. To achieve these concentrations it was necessary to prepare all solutions in D₂O from E-3H and the chosen CD or CD dimer in 0.10 mol dm^–3 NaOD such that pD 12. This may indicate that it is necessary to deprotonate a CD hydroxy group (which is expected to have a pK_a 12 on the basis that the pK_as of OH(2) and OH(3) are 12.33 for CD⁴) to achieve the required solubility. (In the absence of a CD, E-3^– has a very low solubility under these conditions.) Despite this, the stoichiometry of the CD complexes formed appeared to correspond to those detected in the UV-vis studies.

In the ¹H NMR spectrum of an equimolar solution of E-3^– and CD the tert-butyl resonance appears as two singlets and the E-3^– aromatic H1–4 doublets arising from AABB spin–spin splitting are duplicated, consistent with the formation of CD·E-3^– as two includomers, A and B (Scheme 3). The ¹H ROESY NMR spectrum of CD·E-3^– (Fig. 4) shows strong cross-peaks between the H1–3 resonances of E-3^– and those of the CD 3H, 5H and 6H protons of the annular interior. No cross-peaks are observed for the tert-butyl protons and the CD protons consistent with the tert-butyl group residing in the vicinity of either the primary or secondary hydroxy ring of CD. These observations suggest that in both CD·E-3^– includomers CD is positioned over the diazo bond of E-3^– such that protons H1–3 are closest to the interior of the CD annulus. Although the tert-butyl protons are insufficiently close to the interior of the CD annulus to generate cross-peaks, the magnetic environments of the includomers A and B are sufficiently different to produce two tert-butyl resonances. These observations may indicate that CD is sterically hindered from interacting with the tert-butyl group of E-3^– in CD·E-3^–. This interpretation is supported by the observation of strong tert-butyl and H1 and H2 E-3^– cross-peaks in the ¹H ROESY NMR spectrum of CD·E-3^– (Fig. 5) which indicates that the tert-butyl protons are within 4 Å of the CD annular H3, 5 and 6 protons consistent with the larger CD annulus more readily accommodating the 4-tert-butylphenyl group. The tert-butyl and H1–H4 E-3^– resonances of CD·E-3^– appear as a sharp singlet and well resolved doublets, respectively, consistent with CD·E-3^– existing as either a single includomer or two includomers in fast exchange. This deduction is supported by the CD resonances being much better resolved for CD·E-3^– than is the case for CD in CD·E-3^–.

	Fig. 4 1H 600 MHz ROESY NMR spectrum of 0.01 mol dm–3CD and E-3– which exist dominantly as CD·E-3– in 0.10 mol dm–3 NaOD at 298 K. The cross-peaks enclosed in the rectangles correspond to NOE interactions between the protons indicated on the F1 and F2 axes.

	Fig. 5 1H 600 MHz ROESY NMR spectrum of 0.01 mol dm–3CD and E-3–, which exist dominantly as CD·E-3– in 0.10 mol dm–3 NaOD at 298 K. The cross-peaks enclosed in the rectangles correspond to NOE interactions between the protons indicated on the F1 and F2 axes.

	Scheme 3

The two tert-butyl ¹H resonances of CD·E-3^– appear in the area ratio 2.86 1 at 298 K consistent with includomers A and B of CD·E-3^– existing in the same ratio and H⁰ = 11.4 ± 2.6 kJ mol^–1 and S⁰ = 47.0 ± 8.4 J K^–1 mol^–1 derived from the kinetic parameters characterizing the interconversion of the two includomers (Fig. 6). The same ratio is also shown by the duplicated aromatic doublet resonances of CD·E-3^– (Fig. S6, ESI). The assignment of these resonances to specific includomers can not be made with certainty. As temperature increases, the pair of tert-butyl resonances broaden and coalesce and complete lineshape analysis^12,13 in the range 298–318 K yields the kinetic parameters in Table 2, where k_A and k_B are the decomplexation rate constants for includomers A and B, respectively, and k_AX_A = k_BX_B where X_A and X_B are the corresponding mole fractions. Interconversion probably occurs through decomplexation of the CD·E-3^– includomer A to CD and E-3^– followed by the formation of CD·E-3^– includomer B. Under these circumstances, k_A*/k_A = K₁₁ and k_B*/k_B = K₁₁ where k_A* and k_B* are the second-order rate constants for the formation of includomers A and B, respectively. In principle, the addition of a second CD to form (CD)₂·E-3^– followed by the dissociation of an CD to give CD·E-3^– provides an alternative path for interconversion. However, as it appears that the CD annulus is too small to readily pass over the 4-tert-butylphenyl group of E-3^–, this second mechanism seems less likely. If it is assumed that the 4-oxybenzene group of E-3^– more easily enters the wide end of the CD annulus than it does the narrow end, the dominant includomer is A. The negative sign of S_A for the decomplexation of includomer A reflects an increase in order as the transition state is reached whereas the opposite applies for includomer B as is discussed below.

Complex	k(298 K)/s–1	H/kJ mol–1	S/J K–1 mol–1
CD·E-3– includomer A	6.7 ± 0.5 (kA)	61.7 ± 2.7	–22.2 ± 8.7
CD·E-3– includomer B	19.2 ± 1.5 (kB)	73.1 ± 2.7	24.8 ± 8.7
(CD)2·E-3– includomer C	5.7 ± 0.5 (kC)	88.1 ± 4.2	65 ± 13
(CD)2·E-3– includomer D	14.7 ± 0.5 (kD)	79.2 ± 4.2	43 ± 13
CD·4–a	13.3	44.6	–73.9
CD·4–a	0.22	47.5	–98.3
CD·52–b	4.4	49.7	–67
(CD)2·52–b	0.032	60.1	–76
a Ref. 14.  b Ref. 16.

	Fig. 6 Representative variable-temperature 1H 600 MHz NMR spectra of the tert-butyl protons of E-3– in CD·E-3– showing kA at each temperature (kA= 10.4 s–1 at 303 K and 23.7 at 313 K; kB= 18.7, 32.2, 51.3, 84.6 and 127 s–1 at 298, 303, 308, 313 and 318 K, respectively.) The spectra are not plotted to a constant vertical scale. The solution is 0.01 mol dm–3 in CD and E-3– in 0.10 mol dm–3 NaOD.

The ¹H ROESY NMR spectrum of a solution in which the mole ratio of E-3^– and CD was 1 3 shows cross-peaks between the H1–3 and tert-butyl protons of E-3^– and the H3, H5 and H6 protons of CD but no analogous cross-peaks for H4 of E-3^– (Fig. 7). This is consistent with the formation of (CD)₂·E-3^– (Scheme 3) in which one CD envelopes the tert-butyl group of E-3^– and part of the adjacent phenyl group while the second CD is positioned partly over the diazo bond and part of both phenyl rings with H4 of E-3^– being too distant to generate cross-peaks with CD H3, H5 and H6. This disposition of the two CD in (CD)₂·E-3^– is probably aided by the 4-tert-butylphenyl group of E-3^– being strongly hydrophobic whereas the 4-oxybenzene group is less so. The tert-butyl resonance is split into two singlets and each of the E-3^– H3, H5 and H6 doublets are split into two doublets consistent with (CD)₂·E-3^– existing as two includomers. However, the observations on the CD·E-3^– includomers are consistent with the 4-tert-butylphenyl group of E-3^– only being enveloped by the wider end of CD under the influence of increased steric crowding as a second CD complexes E-3^– in (CD)₂·E-3^–. Thus, the structures of the two includomers of (CD)₂·E-3^– (C and D) are probably as shown in Scheme 2, and may only form from the CD·E-3^– A includomer. Consequently, interconversion between the C and D includomers of (CD)₂·E-3^– occurs through decomplexation of the CD adjacent to the 4-oxybenzene group of E-3^– followed by complexation of another CD in the opposite orientation. By this reasoning, the E and F includomers (CD)₂·E-3^– (Scheme 3) are likely to be relatively unstable because the narrow end of the CD annulus envelopes less of the 4-tert-butylphenyl group of E-3^– in the E and F includomers than is the case for the C and D includomers, and the second CD is thereby more hindered in its complexation of E-3^–. The ¹H ROESY NMR spectrum of (CD)₂·E-3^– shows cross-peaks between the H1–3 and tert-butyl protons of E-3^– and the H3, H5 and H6 protons of CD but no cross-peaks for H4 of E-3^–. This spectrum is similar to that observed for CD·E-3^– in that both the CD and E-3^– resonances are much better resolved than are the CD and E-3^– resonances of (CD)₂·E-3^– (Fig. S7, ESI).

	Fig. 7 1H 600 MHz ROESY NMR spectrum of 0.06 mol dm–3CD and 0.02 mol dm–3E-3–, which exist dominantly as (CD)2·E-3– in 0.10 mol dm–3 NaOD at 298 K. The cross-peaks enclosed in the rectangles correspond to NOE interactions between the protons indicated on the F1 and F2 axes.

The two singlets observed for the tert-butyl group and pairs of doublets H2, H3 and H4 of E-3^– in (CD)₂·E-3^– exist in the area ratio 2.46 1 at 305 K consistent with two includomers (C and D) of (CD)₂·E-3^– existing in the same ratio (Fig. 8). This ratio diminishes with increasing temperature and H⁰ = 8.90 ± 0.43 kJ mol^–1 and S⁰ = 21.9 ± 1.4 J K^–1 mol^–1. The pairs of resonances broaden and coalesce with increase in temperature and complete lineshape analysis of the tert-butyl resonances in the range 303–323 K yields the kinetic parameters in Table 2, where the decomplexation rate constant, k_C, refers to includomer C (Scheme 3) and is related to k_D for includomer D through the relationship: k_CX_C = k_DX_D where X_C and X_D are the mole fractions of includomers C and D, respectively. Thus, k_C/k_C* = K₂₁ and k_D/k_D* = K₂₁ where k_C* and k_D* are the second-order formation rate constants for includomers C and D, respectively. The aromatic E-3^– protons show three pairs of doublet resonances assigned to H2–4 at 280 K and a doublet (with twice the intensity of any one of the pairs of doublets), which is attributed to the two H1 doublets possessing coincident chemical shifts in the includomers C and D (Fig. S8, ESI). Increasing temperature causes the pairs of E-3^– doublets assigned to H2 and H3 to coalesce to broad singlets at 323 K while that assigned to H4 coalesces to a doublet as the interconversion rate increases with temperature. This difference results from the greater chemical shift differences between the doublet pairs for H2 and H3 (0.079 and 0.077 ppm, respectively) and that for H4 (0.056 ppm). The coincident E-3^– doublets assigned to H1 change little over the entire temperature range. These observations indicate that H1 experiences no significant difference in magnetic environment in the (CD)₂·E-3^– includomers while the difference in magnetic environment changes in the sequence H2 H3 > H4 for the other protons.

	Fig. 8 Representative variable-temperature 1H NMR (600 MHz) spectra of the tert-butyl protons of E-3– in (CD)2·E-3– showing the kC at each temperature (kC= 19.0 s–1 at 308 K and kD= 24.6, 44.2, 75.0, 119 and 184 s–1 at 303, 308, 313, 318 and 323 K, respectively). The spectra are not plotted to a constant vertical scale. The solution is 0.06 mol dm–3 in CD and 0.02 mol dm–3 in E-3– in 0.010 mol dm–3 NaOD.

It has been calculated that if all motions of the CD host and the guest species ceased on going from the free state to the complexed state an entropy change of –209 to –251 J K^–1 mol^–1 would result.¹⁴ On this basis the decomplexation process should have a positive entropy change of similar magnitude. This represents an upper magnitude limit for S for the A and B includomers of CD·E-3^– and the C and D includomers of (CD)₂·E-3^– as it is unlikely that a complete cessation of motion occurs in the transition states for interconversion. The accompanying hydration changes in both the CD and E-3^– complex components occurring as the transition state is approached, particularly for the entry of water into the CD annulus as it is partly or completely vacated by E-3^–, are likely to make negative entropic contributions. Changes in hydration of the E-3^– 4-oxybenzene group may also occur but their entropic contributions are not readily estimated.

Comparison may be made between the CD·E-3^– data and those for the decomplexation of CD·4^–, where 4^– (Scheme 1) is 3,5-dimethyl-4-hydroxy-4-sulfonatoazobenzene which is similar in size and shape to E-3^– but forms a much less stable and single includomer (K₁₁ = 1.02 × 10³ dm³ mol^–1).¹⁵ Both NMR and modelling studies indicate that the CD annulus in CD·4^– is orientated with its narrow end toward the sulfonato group of 4^–. Decomplexation occurs through a fast step followed by a slower step characterised by the parameters in Table 2. The decomplexation of the includomers of CD·E-3^– may also proceed in two steps where the slower step is that kinetically characterised by ¹H NMR in this study. By comparison with the slower decomplexation step for CD·4^–, k_A(298 K) for includomer A of CD·E-3^– is 30 times greater, H is larger by a factor of 1.3 and S is of the same sign but 0.23 the size. For includomer B of CD·E-3^– k_A(298 K) is 87 times larger, H is larger by a factor of 1.5 and S is of the opposite sign and 0.25 the size S of that for CD·4^–. The most obvious differences between the 4^– and E-3^– complexes are that E-3^– incorporates the 4-tert-butylphenyl group where 4^– has a more sterically hindering 3,5-dimethyl-4-hydroxyphenyl group, and E-3^– has a 4-oxybenzene group where 4^– has a 4-sulfonatoazobenzene group, which is likely to be more extensively hydrated. Thus, S for CD·4^– may involve greater negative entropic contributions due to the hydration changes accompanying its decomplexation, which would explain the less negative S and the positive S characterising the A and B includomers of CD·E-3^–, respectively. The different signs of S for includomers A and B of CD·E-3^– probably reflect the opposite orientations of CD in the includomers but further interpretation is infeasible with the present data. Greater participation of water in the approach to the transition state may offset some of the enthalpy change required to disrupt secondary bonding between CD and 4^– as CD·4^– decomplexes and thereby lower the overall H by comparison with those for the interconversion of the CD·E-3^– includomers.

Similar reasoning may be applied to the larger Mordant Orange 10 dianion, 5^2– (Scheme 1), which forms both CD·5^2– and (CD)₂·5^2– (characterised by K₁₁ = 1.26 × 10⁴ dm³ mol^–1 and K₂₁ = 8.8 × 10³ dm³ mol^–1, respectively) which are 22 times less stable than CD·E-3^– and of similar stability to (CD)₂·E-3^–, respectively.¹⁶ The decomplexation of CD·5^2– and the decomplexation of the first CD from (CD)₂·5^2– are characterised by the parameters in Table 2. By comparison with the parameters for (CD)₂·5^2–, k_C(298 K) and k_D(298 K) for (CD)₂·E-3^– are 180 and 460 times greater, H_C and H_D are larger by factors of 1.5 and 1.3 and S_C and S_D are smaller by factors of 0.86 and 0.57 and are opposite in sign.

The ¹H ROESY NMR spectrum of a solution equimolar in 1 and E-3^– at 298 K shows cross-peaks between the CD and CD 3H, 5H and 6H protons and the E-3^– H1–4 and tert-butyl protons consistent with the formation of 1·E-3^– (Fig. 9 and Scheme 2). The appearance of a cross-peak arising from the H4 proton of E-3^– suggests that the urea linker between the CD and CD of 1 imposes a structure on 1·E-3^– which envelopes E-3^– to a greater extent than is the case in (CD)₂·E-3^– and (CD)₂·E-3^–. The resonances of the H1–4 protons of 1·E-3^– are greatly broadened, consistent with an exchange process occurring between different magnetic environments at an intermediate rate on the NMR timescale, whereas that of the tert-butyl protons is not. A possible explanation is that because of the urea linker and the homochirality of CD and CD each glycopyranose unit is unique such that each orientation of E-3^– (with respect to the axis passing approximately through the centres of the CD and CD annuli of 1) represents a rotomer of 1·E-3^– in each of which E-3^– experiences different magnetic environments. Thus, rotation between the rotomer orientations results in broadening of the H1–4 resonances of the E-3^– protons of 1·E-3^–. The lack of broadening of the tert-butyl resonance is attributed to rapid C–C bond rotation within the tert-butyl group and also about the tert-butyl to phenyl bond resulting in a faster environmental averaging than that between the rotomers. Alternatively, a shuttling motion of E-3^– along the complex axis could be the source of the resonance broadening. At 323 K the E-3^– aromatic resonances of 1·E-3^– are much narrower consistent with an acceleration of the rate process.

	Fig. 9 1H 600 MHz ROESY NMR spectrum of 0.01 mol dm–31 and E-3– which exist dominantly as 1·E-3– in 0.10 mol dm–3 NaOD at 298 K. The cross-peaks enclosed in the rectangles correspond to NOE interactions between the protons indicated on the F1 and F2 axes.

The ¹H ROESY NMR spectrum of an equimolar solution of 2 and E-3^– shows well-resolved doublets arising from H1-4 of E-3^– with the first thee doublets showing moderately strong cross-peaks and the H4 doublet showing weaker cross-peaks with the H3, H5 and H6 protons of the CD of 2 consistent with the formation of 2·E-3^– (Fig. 10). Again, the tert-butyl protons show much stronger cross-peaks commensurate with their higher population. In contrast to 1·E-3^–, none of the E-3^– resonances of 2·E-3^– is broadened consistent with a faster rate of either rotomerization or shuttling permitted by the looser fit of the two CD annuli of 2 to E-3^–.

	Fig. 10 1H 600 MHz ROESY NMR spectrum of 0.01 mol dm–32 and E-3– which exist dominantly as 2·E-3– in 0.10 mol dm–3 NaOD at 298 K. The cross-peaks enclosed in the rectangles correspond to NOE interactions between the protons indicated on the F1 and F2 axes.

In the presence of one equivalent of adamantane-1-carboxylate (6^– in Scheme 1) new cross-peaks arising from the adamantyl protons and H3, H5 and H6 of the CD components appear and the cross-peaks remain for the protons of E-3^– (Fig. S9, ESI). This is consistent with the dominant complex in solution being 2·E-3^–·6^– with the 4-tert-butylphenyl group of E-3^– occupying one CD annulus and 6^– the other. This suggests that the complexation of the 4-tert-butylphenyl group of E-3^– by one CD is a major stabilising force in 2·E-3^– with the 4-oxybenzene group being less strongly complexed by the second CD. As 6^– complexation by CD¹⁷ is characterised by K₁₁ = 1.8 × 10⁴ dm³ mol^–1 it follows that the internal complexation of the 4-oxybenzene group in 2·E-3^– is characterised by a K₁₁ less than this. (Displacement of the 4-oxybenzene group by 6^– may involve rotation of one CD component about the N–C(6) bond of 2 so that it is no longer approximately collinear with the second CD component.) In contrast, a solution 0.01 mol dm^–3 in each of 1·E-3^– and 6^– shows no cross-peaks arising from interaction of the 6^– protons with H3, H5 and H6 of the CD of 1. This is consistent with the complexation of the 4-tert-butylphenyl group of E-3^– by the CD component of 1 in 1·E-3^– and with 6^– being too large to compete effectively with the 4-oxybenzene group of E-3^– for occupancy of the annulus of the CD component of 1. (K₁₁ = 1.4 × 10² dm³ mol^–1 for the complexation of 6^– by CD.¹⁷).

The ¹H ROESY NMR spectrum of a solution 0.01 mol dm^–3 in 2 and 0.02 mol dm^–3 in E-3^–, which exists dominantly as 2·(E-3^–)₂, as indicated by the UV-vis studies discussed above, shows cross-peaks very similar to those observed for 2·E-3^– (Fig. S10, ESI).

Theoretical investigations using the Gaussian 98 suite of programs¹¹ and the Pople-type 6-31g(d,p) basis set were undertaken to gain insight into the failure to observe E to Z (trans to cis) isomerization in 3^–. Two possibilities for the lack of such isomerization were investigated: (i) while ground-state E-3^– has a large isomerization barrier, that for ground-state Z-3^– is very small so that Z-3^– rapidly reverts to E-3^–, which is the only detected isomer, and as a consequence should E-3^– photoisomerize to Z-3^– it will rapidly revert to E-3^– thermally. (ii) The first and second excited singlet electronic states of 3^–, S₁ and S₂, respectively (corresponding to n–* and –* transitions) have a preferred transoid conformation. Calculations at the HF/6-31g(d,p) theory level show the isomerization of E-1^– to Z-1^– and vice versa to be characterised by barriers of 183.1 and 105.9 kJ mol^–1, respectively. This indicates that the isomerization of E-1^– to Z-1 would be slow and that possibility (i) can not account for the absence of Z-1^–. In general terms, calculations show that the first and second excited singlet electronic states of 3^–, S₁ and S₂, respectively (corresponding to n–* and –* transitions) have a preferred transoid conformation which indicates that possibility (ii) is the more likely explanation for the absence of Z-1^–. Further discussion and details of the calculations appear as ESI.

Aqueous solutions were prepared with water purified with a Waters Milli-Q system to give a specific resistance of >15 M cm which was then boiled for 30 min to remove CO₂ and allowed to cool in a container fitted with a soda lime guard tube. Sodium perchlorate (Fluka) was twice recrystallised from water, and the anhydrous salts was obtained by drying to constant weight over P₂O₅ under vacuum prior to use (CAUTION: Anhydrous perchlorate salts are potentially explosive and should be handled with care). UV-vis spectra of E-3^– and either CD, CD, 1 or 2 in 0.050 mol dm^–3 borate buffer (total buffer concentration at pH 10.0 prepared from boric acid and NaOH at I = 0.10 mol dm^–3 adjusted with NaClO₄) were run at 298.2 ± 0.1 K in matched quartz cuvettes of 1 cm path length against a reference containing the same buffer and supporting electrolyte. Absorbance data were collected at 0.5 nm intervals with a Cary 300 Bio double beam spectrophotometer.

Complex stoichiometry and complexation constants were determined through non-linear least squares fitting of algorithms for the formation of 1 1, 1 1 and 2 1, 1 2, and 2 1 complexes to the absorbance variation of E-3^– with concentration of CD, CD, 1 and 2 at 1 nm intervals in the range 350–500 nm by using Method 5 of Pitha and Jones,¹⁸ through a least-squares regression routine DATAFIT¹⁹ using the MATLAB formalism.²⁰ Using the CD/E-3^– system as an example, the observed absorbance, A, is related to the molar absorbances of the species in solution, , and their concentrations through:

A = CD[CD] + E-3[E-3–] + CD·E-3[CD·E-3–] + (CD)2·E-3[(CD)2·E-3–]

(1)

_CD

¹H (600 MHz) and ¹³C (75.5 MHz) NMR spectra were run on Inova 600 and Varian Gemini 300 spectrometers, respectively. Solutions of E-3^– and either CD, CD, 1 or 2 were prepared from E-3H in 0.10 mol dm^–3 NaOD in D₂O and had a pD 12. Chemical shifts were referenced against external trimethylsilylpropiosulfonic acid. Elemental analyses were performed by the Microanalytical Service of the Chemistry Department, University of Otago, Dunedin, New Zealand. CD and CD (Nihon Shokhuin Kako Co.) were dried by heating at 100 °C under vacuum for 18 h.

N-(6^A-Deoxy--cyclodextrin-6^A-yl)-N-(6^A-deoxy--cyclodextrin-6^A-yl)urea (1) and N,N-bis(6^A-deoxy--cyclodextrin-6^A-yl)urea (2) were prepared as in the literature and gave spectroscopic data in good agreement with those reported.^1,2

A cold solution of sodium nitrite (BDH, 560 mg, 8.11 mmol) in water (5 cm³) was added dropwise to a vigorously stirred solution of 4-tert-butylaniline (Aldrich, 800 mg, 5.37 mmol) in concentrated aqueous hydrochloric acid (Ajax Univar, 3 cm³) and water (3 cm³) cooled in an ice-salt-bath, at such a rate that the temperature of the mixture did not exceed 0 °C. The resultant solution was stirred for a further 5 min at 0 °C and then added dropwise to a stirred solution of phenol (Aldrich, 860 mg, 9.15 mmol) in 10% aqueous sodium hydroxide (Ajax Univar, 10 cm³) cooled to 0 °C. The reaction mixture was left to stir at 0 °C for 30 min and the crude product was collected by vacuum filtration and washed with water. The product was recrystallised from aqueous ethanol as yellow–orange needles (1.18 g, 87%); mp 124–126 °C. Elemental analysis for C₁₆H₁₈N₂O·0.5H₂O: C, 72.98; H, 7.27; N, 10.64. Found C, 73.38; H, 7.32; N, 10.85%. _H (CDCl₃) 8.83 (d, J = 8.4 Hz, 2H), 7.80 (d, J = 8.4 Hz, 2H), 7.51 (d, J = 8.4 Hz, 2H), 6.90 (d, J = 8.4 Hz, 2H), 1.36 (s, 9H). _C (CDCl₃) 158.1, 154.0, 150.6, 147.2, 126.0, 124.8, 122.2, 115.8, 31.2.

Footnote

Electronic supplementary information (ESI) available: UV-vis spectra and data, NMR spectra and ab initio figures and discussion. See http://www.rsc.org/suppdata/ob/b4/b415594g/