J. Phys.
4988
Chem. 1982, 86, 4988-4992
Potentiometric and Hydrogen-I Nuclear Magnetic Resonance Studies on the Solvation of Cryptand (2,2,2) and Its Complexes with K+ and Ag+ in Acetonitrile 4- Water Mixtures B. 0. Cox, Department of Chemistry, University of Stiriing, Stirllng, FK9 4LA Scotknd
P. Flrman, D. Gudlln, and H. Schneider' h4ax-Planck-Institut fijr blophyskaiische Chemie, 03400 GGttingen, West Germany (Recelved: June 30, 198 1; I n Final Form: August 4, 1982)
The stability constants of the complex between cryptand (2,2,2) and K+ in this binary solvent system have been determined by potentiometric titrations and used to calculate the solvation energies of the cryptate complex. K(2,2,2)+is more strongly solvated in acetonitrile + water mixtures than in pure water and the Gibbs free energies of transfer of K(2,2,2)+are even more negative than the corresponding data for Ag(2,2,2)+.In order to explain more specifically the cryptate ion-solvent interactions, we have made 'H NMR measurements and the chemical shifts, 6, of the various protons of (2,2,2) and the two metal ion cryptates were determined as a function of mole fraction, XAN, of acetonitrile. At low XAN a rapid change of 6 for the methylene protons bonded to the oxygen atoms indicates that in water the cryptates are in an unfavorable conformation state with respect to the complex in the mixture.
Introduction In order to investigate further the solvation of cryptate complexes, we have extended earlier work4to include the Inclusion complexes between macrobicyclic diazapolystabilities and solvation energies of the corresponding ethers (cryptands)' and metal cations have recently atpotassium (2,2,2) cryptate in acetonitrile + water mixtures. tracted much interest2 because of the unusually large It is shown that the striking differences in the Gibbs free stabilities of the complexes (cryptates) and the high seenergies of transfer of K+ (ref 5) and Ag+ (ref 6 and 7) as lectivity of the cryptands.' In previous studies on cryptate a function of mole fraction are considerably reduced by formation with alkali-metal ions in several solvents it has the cryptate formation process. However, the Gibbs free been shown that there is a large variation with solvent of energies of transfer of the two cryptates are by no means the stability constants, but qualitatively the same selecidentical. 'H NMR chemical shift experiments have also tivity pattern is obtained in all solvent^.^ been performed on the free ligand (2,2,2),on K(2,2,2)+,and The cryptand (2,2,2)-N((CH2CH20)2CH2CH2]3Non Ag(2,2,2)+and the results are discussed in relation to forms the most stable complex among the alkali-metal cryptand-ion, cryptand-solvent, and cryptate-solvent incations with the potassium ion. For this ligand, Ag+ and teractions. K+, which have almost equal ionic radii, are particularly appropriate for the study of differences in interactions with Experimental and Results the binding sites of the cryptand and of the solvation of Materials. Cryptand (2,2,2) was a commercial sample the corresponding cryptate complexes. The cavity of (Merck) and was used without further purification. The (2,2,2)is of the appropriate size to wrap around the cations inorganic salts were the same as those previously used: and to shield them from direct interaction with the solvent. KC104 (Merck anhydrous), KNO, (Merck anhydrous), However, the Gibbs free energy of solvation of Ag(2,2,2)+ AgN03 (Merck anhydrous), Et4NC104 (Fluka, purum). is significantly dependent upon the solvent composition The solvent mixtures were prepared by weight from acein mixtures of acetonitrile and water, although the preftonitrile (Merck, p.a.) and bidistilled water or acetoerential solvation is less pronounced than that of the unnitrile-d3 (Merck, 99%) and D20 (Merck, 99.9%). complexed silver Cyclohexylamine (Merck) was dissolved in D20 to form a basic solution for 'H NMR measurements with un(1)J.M.Lehn,Struct. Bonding (Berlin),16,l (1973);Acc. Chem. Res., protonated (2,2,2). tert-Butyl alcohol (Merck p.a.) was 11, 49 (1978). used as an internal reference at a concentration of 0.01 M. (2)(a) J. M.Lehn and J. P. Sauvage, J . Am. Chem. SOC.,97,6700 Stability Constants of K(2,2,2)+.The complex forma(1975);(b) E. Kauffmann, J. M. Lehn, and J. P. Sauvage, Helu. Chim. Acta, 59, 1099 (1976); (c) J. M.Lehn and F. Montavon, ibid., 61,67 tion of K+ with (2,2,2) in mixtures of acetonitrile and water (1978); (d) G. Anderegg, ibid., 58, 1218 (1975); (e) B. Spiess, F. Arat 25 "C was studied by potentiometric titration of a Ag+ naud-Neu, and M. J. Schwing-Weill, ibid., 62,1531 (1979); (f) F. Arsolution with a solution of K+ and (2,2,2).2j9k The connaud-Neu, B. Spiess, and M. J. Schwing-Weill, ibid., 60,2633(1977);(g) R. Greaser, D. W. Boyd, A. M. Albrecht-Gary,and J. P. Schwing, J.Am. centration of free Ag+ was determined by the potential of Chem. SOC.,102,651 (1980); (h) E. L. Yee, 0. A. Gansow, and M. J. Weaver, J . Am. Chem. SOC.,102, 2278 (1980); (i) A. I. Popov, in 'Stereodynamics of Molecular Systems", Proceedings of a Symposium at the State University of New York at Albany, April 23-24, 1979,R. H. Sarma, Ed., Pergamon Press, Elmsford, NY, 1979,pp 197-207; G) B. G. Cox,H. Schneider, and J. Stroka, J. Am. Chem. SOC.,100,4746 (1978); (k) J. Gutknecht, H. Schneider, and J. Stroka, Inorg. Chem., 17,3326 (1978). (3)B. G. Cox,J. Garcia-Rosas,and H. Schneider, J. Am. Chem. SOC., 103,1384 (1981). 0022-3654/82/2086-4988$01.25/0
(4)B.G. Cox,C. Guminski, and H.Schneider,J. Am. Chem. SOC.,104, 27R9 (1982). ~
(5) B. G. Cox, R. Natarajan, and W. E. Waghorne, J. Chem. SOC. Faraday Trans. 1, 75,86 (1979). (6)H. M. Koepp, H. Wendt, and H. Strehlow, Ber. Bunsenges. Phys. Chem., 64,483 (i960). (7)B.G. Cox,A. J. Parker, and W. E. Warrhorne, J . Phys. Chem., 78,
1731 (1974).
0 1982 American
Chemical Society
The Journal of Physical Chemistry, Vol. 86,No. 25, 7982 4989
Solvation of Cryptand (2,2,2) and Its Complexes
TABLE I: Stability Constants (log K,)of K(2,2,2)+ and Gibbs Free Energies of Transfer (AG,,) of KClO,, (2,2,2), and K(2.2.2W10. . , , , * from Water to Acetonitrile + Water Mixtures a t 2 5 "C
g
XAN= 1- XH,O
log KSa
A Gtr( KCIO,)bsc
0.0 0.05 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 0.95 1.0
5.6. 5.58.e 5.4f 6.0 6.5 7.1 7.7 8.1 8.6 8.9 9.2 9.7 10.3 10.9h 11.4, 10.71g
0.00 -0.40 0.17 -1.26 t 0.17 -1.71 0.17 -1.43 t 0.17 -0.91h -0.06 0.17 t 1.54h t 3 . 5 4 t 0.17 t 5.77h +8.74 t 0.23 + 10.92 f 0.23 + 1 2 . 7 r 0.30
a K , in M - I , A log K,) = k O . 1 . a G t r in kJ mol-I. From ref 2k. Interpolated values.
r,
AG,,( 2,2,2)b9C AG,,(K( 2 , 2 , 2 ) C 1 0 , ) * ~ ~
0.0
* *
0.5 0.4
* 0.3
*
* *
0.6 0.4 0.4 0.5
1.3 kJ mol-I. e From ref 2d.
From ref 4.
a silver electrode measured with respect to a Ag/Ag+ reference electrode and was used to calculate the equilibrium constant of eq 1. This equilibrium constant is the K+ Ag(2,2,2)+ s Ag+ K(2,2,2)+ (1)
+
0.0 -4.8 -9.2 -13.6 - 16.3 -17.4 -18.3 -17.9 -16.8 -16.0 - 15.1 -15.5 -15.5
* 0.4 * 0.5
-2.1 -2.8 -3.3 + -2.9 t -2.2h -1.1 -0.6h t0.2 i t 1.6h t3.0 t3.8 +4.9 f
l
From ref 2a.
f
a
+
ratio of the stability constants for K(2,2,2)+and Ag(2,2,2)+. The measured equilibrium constant of eq 1,together with previously determined stability constants of Ag(2,2,2)+,4 have been combined to provide the stability constants for K(2,2,2)+,which are given in Table I. The titrations were done at constant ionic strength, I = 0.1 M, using Et4NC104. The concentrations of the reacting species were -1.5 X lo-, M in (2,2,2), -1 X lo-, M in Ag+, and -3 X in K+. The equilibrium constant of eq 1is independent of activity coefficients which cancel as long as the ionic strength varies insignificantly during the titration process. Where the stability constant of the potassium complex was larger than that of the silver complex (at higher mole fractions of acetonitrile), the arrangement in the potentiometric titration was changed and a solution of Ag+ was added to a solution containing K+ and (2,2,2). Gibbs Free Energies of Transfer, AGtr. The solubilities of KC104,and hence AGJKC1O4), in acetonitrile + water mixtures were available from previous investigation^.^ These were combined with AGtr(2,2,2)4to determine the changes of solvation of K(2,2,2)C104with mole fraction, according to eq 2, where K,(S) and K,(H20) represent the AGtr(K(2,2,2)C10J = -RT In (K,(S)/K,(H20))+ AGt,(KC104) + AGt,(2,2,2) (2) stability constants of K(2,2,2)+in solvent S and water. The Gibbs free energy of transfer of K(2,2,2)C104refers to transfer from water to solvent S and values calculated by using eq 2 are listed in Table I, together with AGk(KC1O4) and AG,(2,2,2) values: The perchlorate ion has been used as a counterion because its free energy of transfer between water and acetonitrile is small: so that the variation of AGt,(K(2,2,2)C104)with mole fraction reflects predominantly the behavior of K(2,2,2)+. Apart from this, differences in properties of the various perchlorate salts represent of course differences between the properties of the cations involved (K+,Ag+, K(2,2,2)+, and Ag(2,2,2)+). ' H NMR Experiments. The 'H NMR experiments were carried out by using the rapid scan mode at 90 MHz on a Bruker HFX-90 spectrometer equipped with an Aspect 2000 computer for data accumulation and transformation. The 19FNMR resonance of C6F6was used as external lock. All measurements were made at 25 f 1 "C in 5-mm 0.d. Wilmad spinning tubes equipped with 2-mm 0.d. capil-
6
Li2,-+-1 ,
3 l ~ P m l2
1
1
A
A',
6 3
0
Ipp"
-/1/ i
, i
6
Figure 1. 'ti NMR spectra of (a) (2,2,2) and (b) K(2,2,2)+ and Ag(2,2,2)+ in acetonitrile (AN) water mixtures. Reference: teff-butyl alcohol. (a) (2,2,2) at X , = 0.00(A), 0.20 (B), 0.36 (C), 0.61 (D), 0.77 (E), and 1.OO (F). The spectra are shifted successively by 0.170 ppm to the right. (b) K(2,2,2)+ at X,, = 0.00 (A), 1.00 (B); Ag(2,2,2)+ at X,, = 0.00(C), 1.00 (D). The spectra B and D are shifted to the right by 0.525 ppm.
+
laries filled with C6FP Measurements were carried out on (2,2,2), Ag(2,2,2)+, and K(2,2,2)+ in several mixtures of acetonitrile water as well as in the pure solvents. All proton chemical shifts are reported with respect to the methyl resonance of tert-butyl alcohol (0.01 M). The concentrations of (2,2,2), K(2,2,2)N03,and Ag(2,2,2)N03 were 0.05 M in all cases. For measurements on the cryptates, a molar excess of 1%of KNO, or AgNO, over (2,2,2) was maintained to reduce the proportion of (2,2,2) in the free state to a negligible value. Although the cryptand exchange is slow on the NMR time scale, so that in principle resonances due to free and complexed (2,2,2) could be studied in the same solution, this arrangement of concentration simplifies the observed spectra. Since cryptands are quite strongly basic, the protonation of (2,2,2) in water and aqueous acetonitrile solutions had to be taken into consideration. It can be readily shown by using the measured pK, value in water (pK, = 9.86l8 for the first protonation equilibrium that, for a total cryptand concentration of 0.05 M, the proportion of protonated cryptand amounts to less than 4 % . However, to ensure that the (2,2,2)H+present had no measurable effect on the chemical shift determination, we recorded spectra of (2,2,2) in the appropriate solvent mixture alone, and
+
(8) B. G. Cox, D. Knop, and H. Schneider, J. Am. Chem. SOC.,100,
3268 (1980).
4990
The Journal of Physical Chemistry, Vol. 86, No. 25, 1982
IkJ moi'l
3 .CN - H,O 25%
i
I0
_ _ ~ C
02
L
04
-
06
XAN"1-XH
G8
"+
C
free energies of transfer of (2,2,2), KCIO,, and the corresponding cryptate salts. Flgure 2. Gibbs
AgCIO,,
with added cyclohexylamine (0.05 M, pK, = 10.68). Cyclohexylamine forms basic solutions by hydrolysis, and an additional important factor is that its proton spectrum is at low magnetic field, far from the resonance lines of the cryptand. The spectra of (2,2,2) in acetonitrile + water mixtures and the spectra of K(2,2,2)+and Ag(2,2,2)+in the pure solvents are given in Figure 1, a and b. The individual records in Figure la, A-E, are shifted successively by 0.170 ppm to improve clarity, and in Figure l b spectra B and D are shifted to the right by 0.525 ppm.
Discussion Gibbs Free Energies of Transfer. The two univalent cations K+ and Ag+, despite being of almost the same size, behave very differently in the solvent mixtures. Both are preferentially solvated in the acetonitrile + water system but by different components. The potassium ion, like other alkali-metal ions, undergoes primarily electrostatic interactions with the solvent and has a positive Gibbs free energy of transfer from water to acetonitrile. Data for KC104in the mixtures (Table I and Figure 2) show a slight initial decrease in free energy, but a steady increase from XAN= 0.2 as the water content of the solvent decreases. The initial decrease in free energy, the magnitude of which is very small compared with the increase in AG,, (KClO,) toward Xu = 1, could reflect the increase in the basicity of water molecules when basic polar solvent molecules are dissolved in it.9 In sharp contrast, Ag+ is preferentially solvated by acetonitrile and the very negative Gibbs free energies of transfer for AgC104 reflect the very strong interaction of the univalent dl0 cations with nitrilic solv e n t ~ .The ~ ~ results ~ for AgC104, particularly the large decrease in free energy on addition of relatively small amounts of acetonitrile to water (Figure 2), are not unexpected in view of the strength of the Ag(CH3CN)2+ complex in ~ a t e r . ~ , ~ J O The large difference between the Gibbs free energies of transfer of KClO, and AgC104, which increases with increasing X m to a value of 30 kJ mol-' in pure acetonitrile, (9) D. Feakins, K. H. Khoo, J. P. Lorimer, D. A. O'Shaughnessy, and P. J. Voice, J. Chem. SOC.Faraday Trans. 1, 72, 2661 (1976). (10) S. E. Manahan and R. T. Iwamoto, J. Electroanal. Chem., 14, 213 (1967).
Cox
et al.
is reduced drastically to no more than 6 kJ mol-' on complexation of the two cations by cryptand (2,2,2) and, in addition, the order is reversed. Two features of the results for the cryptate salts are noteworthy: first, the quite strongly negative values of the free energies of transfer of K(2,2,2)C104and Ag(2,2,2)C104and, secondly, the obvious difference between their AG,, values at a given mole fraction. The decrease in Gibbs free energies of transfer for the cryptate salts (Table I, Figure 2) is in accord with the data for tetraalkylammonium salts. With increasing length of the alkyl groups R, the Gibbs free energies of transfer of R4NC104from water to acetonitrile get less positive and then increasingly more negative in the sequence Me4NC104 (+1.7 kJ mol-'), Et4NC104(-3.8 kJ mol-'), Pr,NC1O4 (-13.4 kJ mol-'), and Bu4NCL04(-28.5 kJ-l).11J2 The cryptate ions thus show the behavior of large hydrophobic organic cations, although their decrease in free energy is smaller than might be expected on the basis of their ionic radii of around 5.4 A, which is the sum of the ionic crystallographic radius of the cation plus the thickness of the ligand.' Thus, Gibbs free energies of transfer of the two cryptates are considerably less negative than the corresponding energy of Bu4NC104,although its cationic radius of 4.9 A13 is clearly smaller than that of the cryptate ions. This suggests that a contribution of the type of solvation found for uncomplexed (2,2,2) has to be taken into account. The free ligand shows positive AG, values, attributable to a stronger interaction with water of the bonding groups of the ligand, which are free to rotate toward and interact with the solvent. Similarly the differences between the two cryptates could be explained in terms of differences in the interactions of the ligand donor atoms with the solvent, depending upon the conformational flexibility of the complexes. The potassium cryptate behaves as a more hydrophobic ion than that of the silver cation, The difference is unlikely to be the result of the very small differences in radii of the cations, although in more extreme cases (e.g., K+ to Li+ or K+ to Cs+) changes in size of the cryptated cation result in large variations in AG,, values of cryptates from water to nonaqueous media, including a~etonitrile.~ The most obvious difference between the Ag+ and K+ cryptates lies in the relative strengths of the interaction of the metal ions with the nitrogen atoms of (2,2,2),which strongly influences the M+-N distance. For Ag+ this distance is only 2.48 A,l a value considerably smaller than either that of 2.78 A for the K+ complex14or the sum of the crystallographic radius of Ag+ and the van der Waals radius of nitrogen (2.77 A). The observed free energies of transfer suggest that one effect of the strong Ag+-N interaction is to allow the oxygen atoms of the bridging groups greater freedom to interact with the solvent. ' H N M R Spectra. The proton NMR spectra of cryptand (2,2,2) and of its 1:l complexes with metal ions have been reported first by Dietrich, Lehn, and Sauvage.15 The spectra consist of a singlet for the OCH2CH20protons, which overlaps with the AA'XX' "triplet" spectrum for the NCH2CHz0protons and, shifted upfield, a "triplet" for the NCH2 protons. Knochel et a1.16 have studied extensively the effects of metal ion complexation and changing (11)B. G. Cox and W. E. Waghorne, Chem. SOC.Rev., 9, 381 (1980). (12) B. G. Cox, Annu. Rep. Prog. Chem., Sect. A, 70, 249 (1973). (13) D. F. Evans and M. A. Matesich, J. Solution Chem., 2, 193 (1973). (14) D. Moras, B. Metz, and R. Weiss, Acta Crystallogr., Sect. B, 29, 383 (1973). (15) B. Dietrich, J. M. Lehn, and J. P. Sauvage, Tetrahedron, 29, 1647 (1973). (16) A. Knochel, J. Oehler, G. Rudolph, and V. Sinnwell, Tetrahedron, 33, 119 (1977).
The Journal of Physical Chemistry, Vol. 86, No. 25, 1982 4991
Solvation of Cryptand (2,2,2) and Its Complexes
-
X A \ Z +
a
a
b
Figure 3. 'H NMR chemical shift of CH,OCH,CH,OCH, protons of (2,2,2), K(2,2,2)+. and Ag(2,2,2)+ as a function of mole fraction. Shifts are given in ppm, downfield from (CH,),COH as internal standard (0.01 M). In part b the center of the multiplet is plotted. (X) 0.05 M (2,2,2); (+) 0.05 M (2,2,2) with cyclohexylamine; (0) 0.05 M K(2,2,2)N03; (A) 0.05 M Ag(2,2,2)N03.
solvent on the 'H NMR (270 MHz) of cryptands. They observed a strong solvent effect on the resonance lines of (2,2,2) in water and acetonitrile, which they attributed to specific interactions between the cryptand and solvent molecules. We have studied the dependence on mole fraction of the IH NMR spectra of (2,2,2)and the cryptates K(2,2,2)+and water solvent system. Ag(2,2,2)+ in the acetonitrile Considering first the spectra of (2,2,2) itself in the mixtures, it may be seen that the singlet for the OCH2CH20 protons, which is superimposed on the resonance of the NCHzCHzO protons, shifts from a relatively central position within the triplet in water to completely overlap the low-field triplet line in acetonitrile (Figure la). For this reason it is difficult to define precisely the position of the central peak of the triplet and so, in graphical representations of the peak positions (Figure 3b), we have used the arithmetic mean of the two sharper outer lines. This procedure has also been applied to all other multiplets, in both the free ligand and cryptate complexes (e.g., Figure 4a), as the central resonance of the triplets is often broad and structured. Where this value could be compared directly with that of the position of the highest peak of the central line, it differed by no more than *0.005 ppm. An obvious feature of ligand (2,2,2) is that the chemical shifts of its various methylene protons are affected quite differently by the change in solvent (Figures 3, a and b, and 4a). Thus, on going from water to acetonitrile the resonance of the NCHz protons is shifted upfield in a simple monotonic manner by 0.25 ppm (Figure 4a), those of the NCHzCHzOprotons similarly upfield but by only 0.10 ppm (Figure 3b), and the singlet due to OCH2CH20 is almost unaffected by the solvent (Figure 3a), showing a slight sigmoidal variation covering only about 0.015 ppm. The monotonic variation of the signals due to the two methylene groups in NCH2CH20,the magnitude of which is much larger for the CH2group a to nitrogen, presumably results from relatively strong hydration of the nitrogen atoms. As the nitrogen lone pairs are in rapid equilibrium between exo and endo conformations,' direct interaction with water could occur in an exo conformation, and the proportion of time spent in an exo conformation would be expected to be higher in more highly aqueous media. It
+
V'C.
%,-
-x*3
b
Flgure 4. 'H NMR chemical shift of NCH, protons of (2,2,2), K(2,2,2)+, and Ag(2,2,2)+ and of protons of dioxane and CH3CNas a function of mole fraction. Shifts are given in ppm down field from (CH,),COH as internal standard (0.01 M). In part a the center of the multiplet is plotted. In part be the central quintet line of CHD2CNis taken in 0.05 M solutions of (X) (2,2,2), (+) (2,2,2) with cyclohexylamine, (0) K(2,2,2)N03, (A)Ag(2,2,2)N03, and (0)1,4dioxane.
should be noted that the shifts do not reflect, for example, a varying contribution from N-protonated forms. The positions of the resonance lines are unaffected by the presence or absence of cyclohexylamine, as may be seen from the results shown in Figures 3 and 4 obtained in acetonitrile + water mixtures with and without equimolar concentrations of cyclohexylamine. The hydrogen bonds between water molecules and nitrogen atoms of (2,2,2) deshield the protons of the methylene group bonded directly to the nitrogen atoms more strongly than the protons of the methylene groups one bond length further away. The relative lack of solvent dependence of the chemical shift of the OCH2CH20 singlet is surprising. These methylene groups of (2,2,2) have the greatest conformational freedom and the ether oxygen atoms are expected to interact with water molecules, although less strongly than the nitrogen atoms. To gain more information relevant to this question, we have also determined the chemical shift of 1,4-dioxane (0.05 M) in acetonitrile + water mixtures (Figure 4b), because this molecule consists of OCH2CH20groups. In contrast, the resonance line of 1,4-dioxane is shifted monotonically with increasing acetonitrile content of the solvent by 0.08 ppm to higher field, similarly to the NCH2CH20signals. In view of this we suggest that the lack of solvent dependence of the (2,2,2) singlet results from the time-averaged position of the methylene protons of OCHzCH20being in the interior of the ligand in highly aqueous media, forming a hydrophobic cavity within the ligand and allowing interaction of the ether oxygen atoms with water as a hydrophilic surface of the cryptand.I6 As the acetonitrile content of the solvent increases, such a conformation would become less favorable and the methylene protons would tend to rotate out and come into contact with the solvent. The whole process is clearly complicated by the influence of interactions with the nitrogen atoms but could explain the differences in the behavior of the OCHzCHzOprotons in (2,2,2) and of dioxane and the very small dependence of the OCHzCH20 singlet upon solvent composition. The interpretation given above for the solvent dependence of the (2,2,2) resonances is in accord with the increase in the Gibbs free energies of transfer of (2,2,2) with increasing mole fraction of acetonitrile (Figure 2). The effect of cryptate formation is to shift the resonances of K(2,2,2)+to higher magnetic fields and those of Ag(2,2,2)+ to lower magnetic fields, in both cases with
4992
Cox et ai.
The Journal of Physical Chemistty, Vol. 86,NO. 25, 1982
respect to the corresponding resonance lines of (2,2,2) (Figures 3 and 4). This is true in all cases except for the resonances due to the NCHz protons of Ag(2,2,2)+in highly aqueous media, which occur at higher magnetic fields than those of (2,2,2). This latter effect may be attributed to specific hydration of (2,2,2) which is lost on formation of the complex, in which the endc-endo form of the nitrogen bridgeheads are favored and the oxygen atoms will be directed predominantly toward the metal ion. If we compare the results in pure acetonitrile, it can be seen that complexation with Ag+ results in a considerable deshielding of the protons, the effect progressively decreasing as the distance of the methylene groups from the bridgeheads increases. This results from the partially covalent Ag+-N interaction,”J8 with the consequent transfer of electron density away from the nitrogen, ultimately reducing the electron density around the protons. The variation of the chemical shifts of the cryptates with solvent composition provides additional information on the solvation of the cryptates. The resonance lines of OCHzCHzOand NCHzCH20are shifted to higher magnetic fields on addition of only small amounts of acetonitrile (XAN5 0.2, Figure 3). Thereafter, the chemical shifts remain constant. This suggests that in water the very unfavorable hydrophobic interactions are reduced somewhat by the adoption of cryptate conformations allowing interaction of the solvent with oxygen atoms of the ligand. Such conformations would be expected to be unstable with respect to the complex itself. Once sufficient acetonitrile has been added (ca. 0.2 in XAN),the cryptates can adopt an energetically more favorable conformation, with donor atoms directed toward the metal and solvation involving acetonitrile molecules. Once this situation has been achieved, little further change of the chemical shifts of the resonance lines should occur. This interpretation is in agreement with the measured Gibbs free energies of (17)G.Schwarzenbach, Chimia, 27, 1 (1973). (18)L. E.Orgel, J. Chem. Soc., 4186 (1958).
transfer of K(2,2,2)+ and Ag(2,2,2)+which indicate that both are more stable in acetonitrile water mixtures than in pure water (Figure 2 and earlier discussion). An alternative explanation of the chemical shift variation in terms of a strong specific interaction of the cryptates with acetonitrile seems improbable in view of (a) the results for dioxane in the solvent mixtures (Figure 4), (b) the lack of any sharp change in the chemical shift of the acetonitrile quintet with solvent a t low XAN(Figure 4),and (c) the aprotic nature of acetonitrile. It is noticeable that the chemical shifts of the NCHz protons are almost independent of solvent compositions, for both Ag(2,2,2)+and K(2,2,2)+. This suggests that the cation-nitrogen interaction is sufficiently strong not only to restrict the conformational freedom of the nitrogen but to cancel the influence of the more mobile chain segments of the cryptate on the NCHz protons. With regard to the differences in the Gibbs free energies of transfer for K(2,2,2)+and Ag(2,2,2)+(Figure 2) it would seem that these are too small to cause effects on the NMR spectra which can be explained, within the experimental accuracy of the chemical shift determinations, by differences in the 6 vs. X A N plots for the OCHzCH20and NCHzCH20protons (Figure 3). Finally, as in any NMR experiments involving the measurement of the chemical shifts of protons not involved in strong H-bonding equilibria and strong association equilibria, it is necessary to comment on the choice of a suitable reference signal. We have taken the methyl resonance of tert-butyl alcohol as its chemical shift varies only slightly (