D.M. Clementz, T. J. Pinnavaia, and M. M . Mortland
196
tion number of the iron is undetermined. For anhydrous iron(II1) chloride in IlMPA a band was observed (Figure 7) at about 25,000 cm-l for a 0.001M concentration which was not observed when the Concentration was 0.0001 or 0.01 M . The observed baind is similar to the one reported by Meek and Drag037 for iron(II1) chloride in triethylphosphate. As in their work, it wig5 found that the more dilute solutions were yellow in color, while at higher concentrations of the metal salt, they became red. Upon dilution, the red solutions became yellow. The existence of multiple equilibria in solutions of anhydrous iron(II1) chloride in nonaqueous solvents has received much a t t e n t i ~ n . a ~Apparently, -~~ the tetrachloroferrate(Il1) con is formed at low concentrations of the metal halides but its spectrum becomes masked at higher concentrations by charge-transfer transition originating with o t h v species present in solution. The electronic spectra of anhydrous copper(I1) chloride and bromide are similar to that reported for Cu(HMPA)4(C104)2 i,n the high-energy region but differ considerably in the low-energy region. It is not clear from the elec-
tronic spectra which species are present in solution although tetrahedral complexes are indicated. Conductivity measurements for anhydrous copper(I1) chloride in HMPA show the solution to be a nonconductor. This suggests the presence of bis complexes in solution analogous to that found for cobalt(I1) chloride. Linearity of the pmr shift plots indicates dominance of one species in solution. Acknowledgments. The authors are indebted to Professor Paul G. Sears for use of his conductance apparatus, Dr. Paul L. Corio for the HMPA employed in this study, and Mr. Claude Dungan for checking the nmr rneasurements with the HAGO-IL spectrometer. This work was partially supported by the University of Kentucky Research Foundation. (37) D, W. Meek and R. S. Drago, J. Amer. Chem. SOC.,83, 4322 (1961 ) . (38) R. S. Drago, D. M. Hart, and R. L. Carison, J , Amer. Chem. SOC., 87, 1900 (1965). (39) J. Fajer and H. Linschitz, J. Inorg. Nucl. Chern., 30, 2259 (1968).
ry of Hydrated Copper(l1) Ions on the Interlamellar Surfaces of Layer Silicates. An Electron Spin Resonance Study D. U. Clementz, Thomas J. Pinnavaia,” and M. M. Mortland Ilepdrtmcnts of Crop and Soil Science, Geology, and Chemistry, Michigan State Unwersity, East Lansing, Michlgan 48823 iReceived August 23, 7971)
‘Fhe stereochemistry of hydrated Cu(LI) ions on the interlamellar surfaces of microcrystalline layer silicates has been investigated by observing the anisotropic components of the g factor in the esr spectra of oriented film samples at room temperature. When a monolayer of water occupies the interlamellar regions the ion has axial symmetry and the symmetry axis is perpendicular to the silicate layers. The Cu(U) ion most likely is coordinated to four water molecules in the xy plane and to two silicate oxygens dong the z axis. Under conditions where two layers of water occupy the interlamellar regions, the ion is in an axially elongated tetragonal field of six water molecules and the symmetry axis is inclined with respect to the silicate layers at an angle near 45”. If several layers of water molecules occupy the interlamellar regions, the Cu(Hz0)e2f ion tumbles rapidly and gives rise to a single, isotropic esr signal analogous to that normally observed for the ion a t temperatures above 50°K.
htrodu6tion Recent investigations have shown that Cu(I1) ions can form r complexes with various molecules when present at the interlamellar cation exchange sites of certain layer silicate minerals, known as montmorillonites, or better, smectites.1-6 The silicate layers undoubtedly play an important role io stabilizing the complexes since Cu(I1) in other environments, including homogeneous solution as well as the so1,id state. is not known to form arene complexes. In view of this oibieivation, information concerning the nature of Cu(][I) ions on layer silicate surfaces is of interest. The objective of the present work was to investigate the stereochemistry of exchangeable hydrated Cu(I1) ions in the interlamellar regions of various layer silicates by The Journal of Physical Ckemhtry, VO/. 77, No. 2, 7973
means of esr spectroscopy. Although the minerals are microcrystalline, they are potentially well suited for such studies because highly ordered films can be prepared in which the crystallites are oriented with their planes of hydrated metal ions parallel to each other. Thus it should be possible to deduce the orientations of the anisotropic components of g with respect to the anionic silicate surface. Although esr spectroscopy has been used previously to study hydrated Cu(I1) and other paramagnetic ions in (1) H. E. Doner and M. M. Mortland, Science, 166,1406 (1969)~ (2) M. M. Mortiand and T. J. Pinnavaia, Nature !London), Phys. Sci., 229,75 (1971). (3) T. J. Pinnavaia and M. M. Mortland, J. Phys. Chem., 75, 3957 (1971). (4) D. El. Fenn and M. M. Mortland, Proceedings of the International Clay Conference, Madrid, June 1972. (5) J. P. Rupert, personal communication.
Interla.mellarSurfaces ol Layer Silicates amorphous resins,6 isotropic zeolites,? and layer silicates,s the special Utility of the technique when applied to oriented samples of the latter types of compounds has been only recently recogni7,1d.8c All of the layer silicates investigated in the present study are ireluted in that the silicate layers consist of two silica sheets that enclose an octahedral layer which is occupied Iby nonexchangeable cations such as Al(III), Fe(III), Mg(II), and Li(1). The negative charge on the infinite two-dimensional silicate framework originates from positive charge deficiencies in the octahedral layer or by the replacemont of Si(IV) by a trivalent ion (e.g , Al(II1)) in the tetrahedral silica sheets. Minerals of both types were investigated in order to assess the effect of the site of positive charge deficiency in the silicate framework on the stereochemistry of the hydrated Cu(I1) ions in the interlamellar regions.
erimental Section The following naturally occurring layer silicates of known unit cell composition were used: hectorite (Hector, Calif.), montmoril~on~te (from Upton, Wyo., Chambers, Ariz., and Otay, Calif.), saponite (Scotland), and vermiculite (from Llano, Tex. and Libby, Mont.). The Cu(I1) exchange forms were prepared by slurrying for several hours ca. 2.0 g of the < 2 - p fraction of the mineral in 500 ml of an aqueous or methanol solution of 1.0 N CuCl2, centrifuging, and discarding the supernatant liquid. The procedure wag repeated three times, and then the excess chloride was wnoved by washing with water or methanol until a negative chloride ion test with AgNOB was obtained. Each sample was given a final wash with methanol, dried in air, ant3 then allowed to stand several days at 100% relative humidity to displace any adsorbed methanol with water The sodium exchange forms were prepared in an analagous fashion, except that the exchange reactions were carried out in aqueous solution, and the products were isolated from aqueous suspension by freeze drying. The esr spectra of randomly oriented powder samples of the sodium exchange forms were recorded to identify the resonances due to nonexchangeable paramagnetic ions in the silicate layers. Only the vermiculite from Libby, Mont., exhibited a very broad (AH = 320 G) resonance with a g valuie near 2.0. This resonance is believed to be due to iron(IEl) ions which OCCUPY octahedral positions in the silicate layers as similar spectra have been observed previouslyQ fur hydromicas in which Fe(1II) ions occupy octahedral environments. Although the sodium exchange form of each -Jf the other silicates gave resonance signals near g = 4.0, which may arise from Fe(II1) in tetrahedral sites,1° none showed resonances near g = 2.0 which were sufficiently intmse to obscure Cu(I1) signals in X-band spectra. In the case of the Libby vermiculite, the Cu(1I) signals could be remlved from the iron signal in the Qband spectrum at 77°K. Some dipolar interactions probably occur between Iron and copper, because the width of the copper resonances increased with increasing iron concentration in the silicate layer. In general, highly ordered self-supporting films of the Cu(P1) exchanKe forms of the layer silicates were prepared by evaporating at room temperature an aqueous suspension of the rnrneral on a flat polyethylene or Teflon surface and the2 peeling the films away. Since the crystallites are oriented with thek silicate layers parallel to the film surface, narrow strips of film (ca. 3 X 10 mm) placed vertically irr ta 4-mm quartz glass esr tube or on a Teflon
197
200 Gauss
A
E3
Hspectra (first derivative curves) for Cu(ll) hectorite. Spectra A and B, respectively, are for randomly oriented powder samples at 300 and 77'K. Spectrum C is for an oriented film sample at 300'K with the silicate layers positioned parallel to H. In spectrum D the layers are positioned perpendicular to H.
Figure 1. Esr
holder could be positioned in the cavity of an esr spectrometer with the silicate layers at a known angle t o the external magnetic field. Films of the montmorillonite sample from Otay, Calif., exhibited poor mechanical strength, suggesting that only partial orientation of the crystallites occurs upon evaporation of the suspension. In this case, however, an esr spectrum for a partially oriented sample was obtained by evaporating the suspension on a thin Teflon strip and placing the entire strip in an esr tube. Esr X-band spectra were obtained with a Varian E-4 spectrometer; the &-band spectrum for Cu(1I) was recorded on a Varian Model V4503 Q-band spectrometer. A Phillips X-ray diffractometer with copper radiation and a nickel filter was used to determine the (001) spacings of the copper(I1) exchange forms of the layer silicates. Results and Discussion Hectorite is an example of a layer s-ilicate in which the negative charge on the silicate layers originates exclusively from a positive charge deficiency in the octahedral positions. When allowed to equilibrate in air under ambient conditions, the Cu(I1) exchange form exhibits a 001 spacing of 12.4 A and a water to copper ratio of about 8:1. Since the silicate lattice c dimension is 9.6 A, the thick(6) R. Cohen and C. Heitner-Wirguin. Inorg. Chim. Acta., 3, 647 (1969); R . J. Faber and M. T. Rogers, J. Amer. Chem. Soc., 81, 1849 (1959); C. Heitner-Wirguin and R. Cohen, J. Phys. Chem., 71, 2556 (1967): K. Urnezawa and I,Yamabe. 6ulI. Chem. SOC. Jaa.. 45, 56'(1972). (7) J. T. Richardson, J. Catal., 9, 178 (1967): J. Turkevich, Y. Qno, and J. Soria, ibid., 25, 44 (1972). ( 8 ) (a) A. Furuhata and K. Kuwata, Nendo Kagaku. 9, 19 (1969); Chem. Absfr.. 72. 105744: Ib) Y. i. Taracevich and F. D. Ovcharenko, Proceedings of the' Intkrnational Clay Conference, Madrid, June 1972; (c) 6. R. Angel and P. I. Hall, Proceedings of the International Clay Conference, Madrid, June 1972. (9) I. V. Matyash, Geochem. Int., 6, 676 (1967). (IO) R. C. Kernp,J. Phys., 4, 111 (1871). The Journal of Physical Chemistry, Vol. 77, No. 2, 1973
198
D. M. Clementz, T. J. Pinnavaia, and M. M . Mortland
TABLE I: Esr Uala tor Hydrated Cu(ll) Ions on the Interlamellar Surfaces of Various
Layer Silicates
Layer silicate
Unit cell formulaa
-----
Temp,"K
g1
Gauss
A. I nterl_amellarRegions Occupied by a Monolayer of Waterd C ~ O . ~ I [ M ~ ~ . ~ Z L ~ O . ~ ~ A ~ ~ . ~ ~ ~ ( S ~ ~ . O300 O ) O Z ~2.08 ( F ~ O H 100 )~ 77 2.08 100
Hectorite Montmorillonite (Upton, Wyo.)~ Montmorillonite (Chambers,Ariz ) Montmorillonite (Otay, Calif) Saponite
g,,
104~~, cm-' gave
____--
~~o.sn[A~3.a6~eo.32Mgo.66](A~o.iaSi7.~o)0za(OH)4
77
2.09
158
2.34 2.33 2.34
C=UO .4dA12 . M F ~ o . ~ ~ M ,851 s o(AI0 .22Si7.7~) 0 2 0 (OH) 4
77
2.09
140
2.33
175
2.16
~Ua.63[A1~.69Fea.ii M ~ I . Z O ~ ( A ~ O . O ~ S ~ ~ ~ . ~ S ) O77 ~ ~ ( O H2.09 )~
120
2.33
190
2.17
90 125
2.35 2.35 -2.3
145 180
2.17 2.16
2.40
115
%.si
[Mgs.azMna .a1 Ala.04Fe0.o~l ( A l l .ooSi7.OO)OZO (OH) 4
Vermiculite (Llano,Tex.)
cui .oo[Alo.soFeo.azMg~.661 (A1z.zeSi5. 7 2 ) 0 2 0 (OH) 4
Vermiculite (Llano,Tex.) Vermiculite.
(see above)
B.
300 77 300
2.08 2.08 2.10
-80
165 175 175
2.17 2.16 2.17
e
Interlamellar Regions Occupied by Two Layers of Waterf 300
2.10
45
2.20
GU0.95[AIa .I 4Fei .a6Mg4.38](A12.16Si5.do20(OH) 4 77 2.16 320 2.38 145 2.23 (Libby, Vex.) a Water of hydration is omitted from the unit cell formula; cations enclosed in brackets fill octahedral positions in the silicate layer, whereas those
enclosed in parentheses fill tetrahedral positions. Sources of the unit cell formulae are as follows: hectorite, American Petroleum Institute, Research Project 49; Upton montmorillonite, G. J. Ross and M. M. Mortland, Soil Sci. Soc. Amer. Proc., 30, 337 (1966); Otay and Chambers montmorillonites, L. G. Schultz, Clays Clay IWin., 37, 115(1969); saponite, R. C. Mackenzie, Min. Mag., 31, 672(1957): Llano and Libby vermiculites, M. D. Foster, Clays Clay Min., 10, 70(1'961). Maximum line width of the perpendicular component, Calculated average value of g, Except for Cu(ll) vermiculite, which was dried over P205, monolayers of water were obtained by allowing the samples to dry in air at room temperature. e A l l not resolved, due to presence of paramagnetic impurity. f Two layers of water were obtained by allowing the samples to dry in air.
--,--IC zfr
Silicate
Laver
0
/------Silicate --- Layer .I_-
I I _ _ _ _ _ _
Silicate
A
Layer
B
Schematic representation of the stereochemistry of hydrated Cu(ll) undler conditions where (A) one layer and (B) .two layers of water occ~ipythe interlamellar regions. Figure 2.
ness of the interlainellar region (2.8 A) indicates that the Cu(1I) ions are hydrated by a monolayer of water molecules. Esr spectra of the Cu(1I) ions under these conditions are illustrated in Figure 1. Spectra A and B for randomly oriented powder samples at room temperature and a t 77"K, respectively, conaiw of clearly defined g, and 8, components as expected for Cu(I1) with axial symmetry and (g,, - gL) PH > hA. Although hyperfine splitting due to B3Cu and "Cu ( I = 3/2) irj well resolved for the parallel component, none was observed for the perpendicular component. When the spectrum of an oriented film sample is recorded with the silicate layers parallel to the magnetic field direction (cf. spectrum C) only g, is observed. On the other hand, when the film is oriented with the silicate layers perpendicular to R,only the A,, components of g,, are visible. Thus the symmetry axis of the hydrated ion is positioned perpendirwiar to the silicate layers. The esr results, together with the fact that a monolayer of water occupiet the interlamellar region, are consistent The Journal of P h y s m l Chemistry, Voi. 77, No. 2, 7973
only with an environment in which copper(I1) is coordinated to water molecules in the xy plane and to surface oxygens of the silicate lattice along the z axis. Most likely, four water molecules are bound to copper as shown schematically in Figure 2A. The remaining water molecules must occupy outer spheres of coordination. If most of this latter water is removed by heating the silicate under vacuum at 110", the esr spectral features of an oriented film sample remain unchanged. Thus the amount of outer sphere water, which has been recently described by Farmer and RusselP as forming a dielectric link between the exchangeable cation and the silicate surface, does not alter the basic stereochemistry of the ion. The esr spectral parameters for air-dried Cu(I1) hectorite are presented in Table I (part A) along with those for related Cu(I1) layer silicates under conditions where a monolayer of water occupies the interlamellar regions. In each case an anisotropic spectrum with ,g and g,, components was observed at room temperature but the hyperfine splitting of the parallel component was better resolved at 77°K. Among the three montmorillonite samples shown in the table, 77-95% of the net negative charge on the silicate layers originates from cationic charge deficiencies in the octahedral sheet of the silicate framework, but the number of Cu(I1) ions per unit cell and the nature of the ions occupying octahedral position differ from those in hectorite. Since the surface area of one face of the unit cell in all of these silicates is approximately 50 A2,the average distance between Cu(I1) ions is ca. 16 A in hectorite and ca. 9.0-12.5 A in the montmorillonites. In saponite, and the two vermiculites, where all of the negative charge on the silicate layers is due to positive charge deficiencies in the tetrahedral sheets, the average distance between @u(II)ions are estimated to be 10 and 7 A, respectively. (11)
V. C. Farmer and J. D. Russell, Trans. Faraday Soc., 67, 2737 (1971).
lnteriamellar Surfaces of Laver Silicates Despite thekle differences in Cu(I1)-Cu(I1) distances, position of positive charge deficiency in the silicate layers, and the nature of the metal ions in the octahedral and tetrahedral sheets of the silicate framework, oriented film samples of each layer silicate showed the same esr spectral changes as those described for Cu(I1) hectorite when the film is positioned parallel and perpendicular to the applied magntltic field. Moreover, in each case the magnitude of g is greater than gL (cf. Table I). This result is consistent with Cu(I1) ion in an axially elongated tetragonal crystal field and the unpaired electron occupying a d(x2 - y 2 ) 0rbital.l' Axial elongation is also indicated by the magnitudes of the interlamellar thicknesses. In the case of Cu(T1) hectorite, where the interlamellar thickness is 2.8 A, if the radius of a silicate oxygen atom is taken to be 1.4 A,13 then the copper(II) silicate oxygen bond length is ca. 2.8 A. The expected distance of the Cu-OH2 bonds is ca. 2.0 A An example of the esr spectra obtained for an oriented film of Gu(I1) montmorillonite is provided in Figure 3A. We turn now to the deduction of the stereochemistry of hydrated Cu(1I) ions when two layers of water molecules occupy the interlamellar region. Earlier X-ray diffraction studies of laye). silicates14 indicate that when two layers of water are present, divalent metal ions should be octahedrally coordinated to six water molecules. However, in the case of Cu(I1) the octahedron should be distorted, as required by the Jahn-Teller theorem. Vermiculite is especially well suited for obtaining two layers of watel- rnoleeules in the interlamellar regions as the high surface charge density permits a maximum of only two layers even when the silicate is fully hydrated. Samples of thti other layer silicates with all interlamellar regions uniformly occupied by two layers of water molecules are more difficult to achieve as they can be swelled beyond two layers of water molecules and thus tend to give interstratified systems. The presence of two layers of water molecules in the interlarneila r regions of air-dried oriented films of Llano Cu(II) vermiculite was verified by observing X-ray reflections of several rational orders that corresponded to a 001 spacing of 14.2 A. The esr spectrum of the film at room temperature consisted of g and gl components when the silicate layers were positioned both parallel and perpendicular to the external magnetic field as shown in Figure 3B. The lack of any appreciable change in the relative intensities of g, and gl upon altering the position of the film in the magnetic field indicates that the symmetry axis of the tetragonal ion 1s inclined with respect to the silicate surface at an angle near 45". Also, the Cu-OH2 bonds along the symmetry itxis are longer than those in the xy plane, as g > g,. 14 richematic representation of the stereochemistry of the ion i s shown in Figure 2B. Anisotropy in the g factor of Cu(H20)e2+ is rarely observed at room temperature. One previous example was reported by Fujiwasa and coworkers in a study of cupric sulfate rjolutrona confined in the molecular space of poly(vinyl alcohol) gels.1:) ut isotropic thermal motions are normally sufficiently rapid above 50°K to give a single esr line.16 One suggested motion involves the rapid exchange of the ion betw ?en three equivalent Jahn-Teller distorted pond to axial elongation along the three O-Cu-H20 axes.17 However in the absence of rapid tumbling this motion will not lead to averaging o f g and g I whe s sorbed on a surface in the manner illustrated 111
199
A
,
V
c
I
Figure 3. First derivative esr lines (300°K) Far Cu(ll) in oriented film samples of layer silicates. In each case the top spectrum is for the silicate layers positioned parallel to H , and the bottom
spectrum is for the layers positioned perpendicular to H : (A) Cu(ll) montmorillonite (Upton) when a monolayer of water occupies the interlamellar regions; (B) Cu(l I ) vermiculite (Llano) with two layers of water in the interlamellar region (the line marked x is assigned to a paramagnetic ion present either in the silicate framework or as an impurity since the line was also observed In the Na+ exchange form); (C) interstratified Cu(il) montmorillonite (Chambers) with varying layers of water in the interlamellar regions. In each case the magnetic field increases from left to right, the field sweep is 1000 G , and the vertical line indicates the resonance of a standard pitch sample with g = 2.002%.
Esr spectral parameters for two Cu(1I) vermiculites with two layers of water in the interlamellar regions are presented in Table I (part B). The g values are somewhat larger than those found for the planar aquo complex, and the calculated average values of g are in good agreement with the observed values of g,, for Cu(II) in aqueous solutionls and for CU(Hz0)62+ at the exchange sites of resins.6 The exceptionally large line width observed for the Libby Cu(1I) vermiculite is undoubtedly due to magnetic interactions between Cu(I1) and Fe(II1) which is present in large amounts in the silicate framework (cf. Experimental Section). Attempts to observe the stereochemistry of Cu(B) ions hydrated by more than two water layers were complicated by interstratification of the silicate as indicated earlier. An example of the esr spectra obtained for an interstratified svstem is given in Figure 3C. The high-field line in figure is d i e to the g i component of cu(II) ions hyJ. E. Wertz and J. R. Botton, "Electron Spin Resonance," McGrawHill, New York. N. Y., 1972, p 287. L. Paulina. "The Nature of the Chemical Bond,' Cornell Universitv Press, Ith>ca, N. Y., 1960. G. F. Walker, Clays C/ayMin., 4, 101 (1956). S. Fujiwara, S. Katsumata. and T. Seki, J. Phys. Chem., 71, 116 (1967). B. R. 'McGarvey in "Transition Metal Chemistry," '401. 3. R. L. Carlin, Ed., Marcel Dekker, New York, N. Y . , 1968. A. Hudson, Mol. Phys.. 10, 575 (1966). S.Fujiwara and H. Hayashi,J. Chem. Phys., 431,23 (1965). The Journalof Physicai Chemistry, Voi. 77, No. 2, 1973
200
R. H.Cox, L. W. Harrison, and W. K. Austin
drated by a monolayer, whereas the lower field line, which is orientation independent, is due to Cu(II) hydrated by two or more 'layers of water. It was possible, however, to investigate the nature of the hydrated copper(1l) ions when the interlamellar regions of Cu(I1) hectorite were fully expanded by soaking the silicate in water for 48 hr. Under these conditions the 001 X-ray reflection corresponded to an interlamellar thickness of about 10 A and the esr spectrum of an oriented film sample consisted of a single isotropic line with g =
2.192, independent of its position with respect to H. Thus when several layers of water are present the Cu(l1) ion tumbles rapidly, averaging the g , a n d g l components. Acknowledgment. The support of the National Science Foundation through Grant No. GP-33878 is gratefully acknowledged. The authors wish to thank Dr. Roger V. Lloyd for many helpful discussions and for his assistance in obtaining the Q-band spectrum of the Libby Cu(II) vermiculite.
on A s ~ ~ c ~ffects a t on ~ ~ the~ Nuclear Magnetic Resonance Parameters of Aromatic ions. 111. Cyc oiionatetraenyl Anion, Cyclooctatetraene Dianion, and Tropyliium Cation' ~i~~~~~ H. Cox,* Lester W. Harrison, and Walter K. Austin, Jr. Repnrtment of Chemistry, University ofGeorgia, Athens, Georgia 30601 (Received August 21, 1972) Publica'jon costs assrsted by the Petroleum Research Fund
The effect of counterion, solvent, and temperature on the proton nmr spectrum of the cyclononatetraenyl anion, cyclooctatetraene dianion, and tropylium cation have been investigated. In DME and THF, the lithium and sodium salts of the cyclononatetraenyl anion exist as an equilibrium mixture of contact and solvent-separated ion pairs whereas the potassium, rubidium, and cesium salts exist only as contact ion pairs. Salts of cyclooctatetrene dianion exist as contact ion pairs in all solvents examined. Lithium-7 nmr shifts support this view. The nmr spectrum of several tropylium salts shows the chemical shift of the tropylium cation to be independent of counterion but solvent dependent. These results suggest that the tropylium salts exist in solution as free solvated ions and solvent-separated ion pairs.
Introduction Several recent investigations have led to a better understanding of the physical and chemical properties of organoalkali metal salts in ether solution.2-10 Of particular importance has been the finding that some carbanions exist in solution ab ia rapidly equilibrating mixture of two types of ion pairs (contact and solvent separated), whose ratio depends upon the counterion, solvent, and temperat ~ r e This . ~ concept has been firmly established experimentally and has bean used to explain a wide variety of previous data on carbanion solution^.^^-^^ Further work bas shown that this ion pairing scheme must be expanded to include at least two different types of contact ion pairs in equilibrium with the solvent separated ion pair.sJO These two types of contact ion pairs are thought to exist in different solvataon states. Other things being equal, the major factor determining the fraction of solvent-separated ion pairs formed in a series of aromatic anions appears to be the area over which the negative charge may be delocalized. Since it has been previously reported that the cyclopentadienyl anion forms only contact ion pairs with alkali metal cations in ether solvents,14 it appeared of interest to determine the ring size necessi3ry in an annulene anion in order to have formation of solven1,-separated ion pairs. Furthermore, from previous investigations, it appeared that the effect of solvent1 and concentration1* on the chemical shifts of ion The Journai of Physical Chemistry, Vol. 77, No. 2, 1973
pairs might be different for contact and solvent-separated ion pairs. Since all protons are equivalent in the annulene anions, investigation of these ions permits a larger concentration range to be examined than with previous examples. We report here the results of our proton and lithium-7 nmr investigation of the ion pairing of the cyclononatetraenyl anion (CNT-) and the'dianion of cyclooctatetraene (COT2-) in various ether solvents. The results suggest that it is possible to have both contact and solvent-separated ion pairs present in solutions of CNTwith the proper choice of solvent and metal ion. Salts of (1) For part I i of this series see R. H. Cox, Can. J. Chem., 49, 1377 (1971). (2) T. E. Hogen-Esch and J. Srnid, J. Amer. Chem. Soc., 88, 307, 316 (1966). (3) L. I. Chan and J. Srnid, J. Amer. Chem. Soc., 90, 4654 (1966). (4) T. Eilingsen and J. Smid, J. Phys. Chem., 73,2712 (1969). (5) T. E. Hogen-Esch andJ. Smid, J. Amer. Chern. Soc., 87, 669 (1965). (6) R. V. Slates and M. Szwarc, J. Amer. Chem. SOC., 89, 6043 (1967). (7) D. Nlchoils, C. Sutphen, and M. Szwarc, J . Phys. Chem., 72, 1021 (1968). (8) N. Hirota, J. Amer. Chem. SOC., 90, 3603 (1966). (9) N. Hirota, R. Carraway, and W. Schook, J. Amer. Chem. SOC., 90, 3611 (1968). (10) J. W. Burieyand R. N. Y0ung.J. Chem. Soc. €3, 1016 (1971). (11) M. Szwarc, "Carbanions, Living Polymers and Electron Transfer Processes," Interscience, New York, N. Y., 1968. '(12) M. Sswarc, Accounts Chem. Res., 2, 87 (1989). (13) J. F. Garst, "Solute-Solvent interactions," J. F, Coetzee and C. D. Ritchie, Ed., Marcel Deckker, New York, N. Y.. 1969, p 539. (14) J. E. Grutzner, J. M. Lawlor, and L. M. Jackrnan, J. Amer. Chem. SOC.,94, 2306 (1972).