1178
R. H. DIRTIUS, ill. T. EMERSON, AKD G. R. CHOPPIX
sions were found to obey K e d s law after the field value was corrected for the electrode potential. This was done by measuring the value of the current a t the end of the current trace on the oscilloscope and multiplying it by the resistance determined independently with an impedance bridge. The predominance of permanent dipole orientation in weak fields was also confirmed in this manner. Summary and Conclusions Birefringence measurements of monodisperse bentonite suspensions prepared by careful centrifugation and redispersion on a supercentrifuge were carried out utilizing monochromatic light and a Babinet-Solei1 compensator. It was obseJved that in the range of particle sizes 2500 to 7000 A. all bentonite micelles exhibit negative birefringence in weak d.c. fields and birefringence reversal with increasing field strength. The magnitude of negative birefringence, as well as the value of birefringence reversal voltage, decreases as the ratio of the length of semimajor axis to the thickness of the particles increases. An examination of specific birefringence Anlc a t various concentrations of the suspensions Phowed that although the negative birefringence of bentonite increases with increasing concentration, suspensions exhibit birefringence reversal at infinite dilution. It appears, therefore, that an intrinsic property of the bentonite macromolecules is responsible for their abnormal behavior. Measurements with sinusoidal and square wave fields
Vol. 67
of various frequencies indicate that the particles have a permanent dipole along their symmetry axis. This dipole brings about an orientation of the disk-shaped particles in a direction perpendicular to the alignment a t higher voltages, thereby giving rise to negative birefringence in weak fields. I n stronger fields the particles orient under opposing torques due to permanent and induced dipoles. At some particular field value characteristic for a particle size, the two orientation torques are equalized, and zero net birefringence ig obtained. In saturating fields the induced polarization orientation is predominant. Interactions in suspensions of higher concentrations give rise to an augmentation of orientation along the symmetry axis, although the interaction effect is relatively smaller on induced orientation. Finally, the positive birefringence in weak d.c. fields shows obedience to K e d s law, and saturation data under a.c. as well as d.c. fields confirm predominance of induced polarization orientation in strong fields. Acknowledgments.-The authors take pleasure in acknoxvledging the helpful suggestions of Professor C. T. O’Konski of the University of California during this investigation. The authors are also grateful to a number of colleagues in Advanced Technology (Development Laboratory, IBM Corp., San Jose, Calif .) for their cooperation in this project, to Mr. R, K. Van T’alkenberg for preparing the electron micrographs, and to Mr. Terrance Kelly for assistance with numerical computations.
NGCLEAR MAGNETIC RESONAECE STUDY OF ION-EXCHANGE RESINS. I. HYDRATED DOTVEX-50 RESINS BY R. H. DINIUS,M. T. EMERSON,~ AKD G. R. CHOPPIN Department of Chemistry, Florida State University, Tallahassee, Florida Receitled July 2, 1962 The proton magnetic resonance spectra for hydrated ion-exchange resin samples have been measured. The effect on the chemical shift and line width was observed as a function of resin phase hydration, resin cross linkage, and temperature for Dowex-50. The effects are interpret,ed in terms of changes in water structure and rapid proton exchange in the resin phase.
Introduction Bauman and EichhornZ first suggested that ionexchange resins could be considered as concentrated solutions of strong electrolytes. In this model, a Donnan membrane equilibrium is established between the resin phase and the external solution phase. Solvent and mobile ions are free to exchange between the phases, but the fixed resinate ions are restrained in the resin phase. When this model was modified to take into account the strain energy (“swelling pressure”) resulting from the volume expansion of the cross-linked resin on sorption of solvent, reasonable success was achieved in accounting for niany properties of ionexchange resins. 3--6 This thermodynamic model has been criticized by (1) Department of Chemistry, Wayne State University, Detroit, Miohigan. (2) W. C. Bauman and J. Eichhorn, J . Am. Chem. Soe., 69, 2832 (1947). (3) ( a ) H. P. Gregor, ibid., ‘73, 3537 (1951); (b) E. Glueckauf and G. P. Kitt, Proc. Roy. Soc. (London), A228, 322 (1955). (4) J. F. Duncan, Australtan J . Chem., 8 , 293 (1955). (5) G. E. Myers and G. E. Boyd, J . Phys. Chen., 60, 521 (1956).
Rice and Harris6 as not being an accurate physical representation of the ion-exchange resins. These authors offer a model of a resin as a cross-linked polyelectrolyte gel in which ion pair formation accounts for the specific interaction between the fixed exhange groups and the exchangeable ions. A comparison of the heats, free energies, and entropies of exchange for resins and for concentrated chloride solutions showed no detailed agreement, although the free energies were of the same order in the two systems.’ These data were interpreted as indicating that ion pair formation is responsible for the specific interactions of the resin. It was further postulated that the majority of cations (for polystyrenesulfonic acid resin) not in ion pairs are distributed in a charged layer around the resin chains. The formations of ion pairs is further supported by the work of Kotin and Nagasawa.8 ( 6 ) S. A. Rice and F. E. Harris, Z. Physzk. Chem. (Frankfurt), 8 , 207 (1956). (7) E. H. Cruiokshank and P. Meares, Trans. Paraday Soc., 63,1299 (1957). (8) L. Kotin and &.I.Nagasawa, J . Chena. Pkys., 86, 873 (1962).
June, 1963
NUCLEAR hdAGNETIC
I:ESONASCE
Glueckaufg has recently postulated that the ionexchange resin is a gel-like structure within which there is a large variation in the degree of cross linking. He also pghtulates fissures inside the gel with widths up to 2000 A,, these fissures being connected by many small channels t,hrough which solvent and electrolyte may diffuse. Using this model he has been able to explain the deviations from Donnan equilibrium which was found earlier.1° An investigation of the proton magnetic resonance spectra of hydrated ion-exchange resins was undertaken in an attempt to obtain further information on the physical state of the internal resin phase. Data on the chemical shifts and line widths for HnO in -the resin phase could provide a basis for accepting or rejecting the ion pair formation model as a correct physical picture. Furthermore, it is possible in principle to obtain rates of exchange and rates of diffusion as well as equilibrium constants for processes in the resin. Previously, we have made a preliminary report1' on the chemical shift as a function of water content. I n an independent note and paper, Gordon12 has reported somewhat similar studies on resins immersed in water and other solvents. Experimental The hydrogen form of the cation-exchange resin, Dowex-50, used in this work., was obtained from Bio-Rad Laboratories Co. as 200-300 mesh spherical beads. Special care was taken to remove contaminating metal ions by carrying each resin sample through a washing process described previously.ls Spark spectroscopic analysis indicated that the metal ion concentrations in the purified resin were less than 15 parts per million. The washed re& was stored wet. Prior to use, the resin was dried (for about 2 hr.) to constant weight a t 80" under a pressure of 0.3 mm. Initially, the water content of the dried resin was determined by a Karl Fischer titration in methanol. After a number of such determinations had indicated the absence of water, constancy of weight on drying was assumed a sufficient criterion for complete removal of the water. Samples of resin having a definite water content were made by equilibrating the samples with water vapor and measuring the increase in weight due to the added water. The resin sampleai were sealed in 5 mm. n.m.r. sample tukles to ensure against gain or loss of water. Samples used to study the transition from a one- to a two-phase system were prepared by weighing increments of water into weighed samples of dried resin. The nuclear magnetic resonance spectra were taken on a Varian HR 60 spectrometer. An external reference of cyclohexane was sealed in a 1 mm. capillary and inserted inside the resin sample tube. All chemtcsl shiftg were measured with respect t o cyclohexane and converted to water as a reference by subtracting the water-cyclohexane shift. Negative shifts mean shifts to lower field. The bracketing side-band technique was used in calibrating the spectra. The audio side-bands were produced by audio modulation of the static magnetic field using a Ilewlett-Packard 200 CD audio oscillator, the audiofrequency being measured by a Hewlett-Paekard 623 B frequency meter. The effect of temperature was studied rather qualitatively, since a thermostated probe was not available at the time, by warming or cooling the samples before running the spectrum. The spectrum was then run as rapidly as possible while the temperature drifted back t o room temperature. For this work starting temperatures of 2-3" and 90" were used. A11 other spectra were taken a t 25 f I". E. Glueokrtuf, Proc. Roy. Soc. (London), 8268, 350 (1962). (10) E. Glueckauf and R. E. Watts, ibid., A261, 339 (1962). (11) R. H. Dinius and G. 11. Choppin, J . Phys. Chem., 66, 268 (1962). (12) J. E. Gordon, Claern. 2nd. (London), 267 (1962); J . Phys. Chem., 66, 1150 (1962). (13) G . R. Choppin and R. H. Dinius, Inorg. Chem., 1, 140 (1962). (9)
STCDY OF
ION-EXCHASGE RESXKS
2 MOLES H,O/Eg.
1179
RESIN
-+ @ - 5.0
178MOLES
HO ,
200MOLES
-
HzO/Eg. RESIN
-0.52
-079
Fig. l.-X.m.r.
Eg RESIN
00
INCREASING Ho + spectra for Dowex-50 (4% DVB) resin samples of varying degrees of hydration.
7.0. 6.0.
/ 4%DVB
Fig. 2.-Observed chemical shift for hydrated Dowex-50 resin as a function of hydration, where P is the mole fraction of protons on hydronium ions.
Results and Discussion The n.m.r. spectra of three forms of Dowex-50 ionexchange resin having 4, 8, and 16% divinylbenzene (DVB) content were measured at different degrees of hydration. A single rather broad liiie due to the water protons is observed a t low water content. A second line appears as soon as the water begins to form a layer on the outside of resin bands. Typical spectra for the 4% DVB resin are shown in Fig. 1. It is noted that the width of the single resonance line a t low water content increases as the water content decreases. The position of the line shifts markedly to lower field with decreased water content. Plots of the observed shift of the internal water line as &,function of the amount of hydration are shown in Fig. 2 for the three forms of resin. These chemical shifts are not corrected for magnetic susceptibility differefices. The samples are composed of a collection of near y pbrfect spheres. The magnetic field inside a spheri a1 sample is independent of the bulk susceptibilities. l2 The observed chemical shift variations can be explained if there is a rapid exchange of protons between the acid sites of the resin and the water protons. The resulting shift would be given by the weighted average of the chemical shifts of the acid, water, and hydronium ion protons
d
R. H. DIXIGS,31. T. EMERSOS, AKD G. R . CHOPPIN
1180
which are in equilibrium. Hood and Reillyl*have shown that the chemical shift can be used to obtain acid dissociation constants of strong acids. We have reported similar work on p-toluenesulfonic acid. l1 It was originally hoped that a similar treatment could be used to obtain acid dissociation constants for the ion-exchange resins. This was not possible since one cannot sufficiently saturate the resins with water to obtain the needed dilute solution shifts. However, it can be shown that in concentrated solutions of strong acidsll 6obsd
GlffP
where 61 is the intrinsic hydronium ion shift, a the degree of acid dissociation, and P the mole fraction of protons of hydronium ions. Since 61 is assumed constant, by comparing the slopes of the plots of 6obsd us. P of several acids, a crude estimate of the magnitude of a can be obtained. This, then, provides a measure of the relative acid strength. Approximate acid strengths were obtained by comparing the slopes in Fig. 2 with similar data for strong a ~ i d ~ . ~The ~ s indications ~ ~ J ~ are that the 4% DVB resin is a stronger acid than HC1 or p-toluenesulfonic acid and probably similar in strength to HC104. The 16% resin s e e m to be weaker and more on the order of nitric acid. The 8% falls somewhere in between. It is reasonable to postulate that the order of acid strengths results from the increase in the more extensive organic cross linking for the higher DVB resins, which causes a decrease in the effective dielectric constant in the vicinity of the sulfonate group.ll The lower dielectric constant in higher DVB resins would promote ion pair association, and it seems likely that this is the explanation of the different apparent acidities rather than any difference in the intrinsic acid dissociation constants. The proton shift of the water inside the resin may very well be dependent upon factors not present in the solutions for which the above treatment was worked out. For instance, in the derivation of the chemical shift, the shift for the water protons is assumed constant and set equal to zero (using pure water as the reference). However, it is known that the shift of pure water is dependent on the degree of hydrogen bonding, as has been shown by the variation of its shift with temperature.l6 Any interactioiis which would change the degree of hydrogen bonding inside the resin from that in pure water would give rise to a variation in the observed shifts not attributable to the acid dissociation process. The non-polar and, hence, hydrophobic character of the resin matrix would be expected to discourage a uniform distribution of water throughout the resin. Instead, for unsaturated resins, we would expect the water molecules to cluster around the highly polar acid sites forming domains of water. The size and shape of these domains would be limited by the geometry of the surrounding resin matrix; the more highly cross-linked the resin, the smaller the size. It is reasonable to expect that the presence of highly polar sites in the resin will affect the structure of the (14) G. C . Hood and C . A. Reilly, J. Chem. Phus., 27, 1126 (1957). (15) J. A. Pople, W. G. Schneider, and H. J. Bernstein, "High Resolution Nuclear Magnetic Resonance," McGraw-Hill Book Co., Kew York, N. Y., 1960, p. 443. (16) W. G. Schneidsr, Ha ili B@mstein,and J. A. Rople, J A Ghem, Phye., 88, 801 cla6s)i
Vol. 67
water by changing the degree of hydrogen bonding. The extent of the disruption of structure is dependent on the amount, of water present. The greater change in the water structure would be expected to occur a t lorn water concentrations. If there is only one water molecule at a site, it could bond only with the site itself. One would expect that the protons of this molecule would have a different chemical shift than pure liquid water. It is known that the resonance line moves upfield as hydrogen bonds in water are broken,16 and hence it is reasonable to expect that at lower water concentrations there would be an upfield shift contribution to the water line position for the resin. This would cause a decrease in slope a t higher P in Fig. 2. From a measurement of the swelling of the resin as water is added, we know the resin matrix expands up to a point near saturation, a t which point it can expand no farther. It is known that large osmotic pressures are present near the point of s a t u r a t i ~ n . ~It, ~is~reasonable to expect that the large forces produced would affect the internal water structure by attempting to crowd the molecules closer together, thereby enhancing the hydrogen bonding and producing an increase in the negative chemical shift a t high P. The presence of aromatic rings in the form of divinylbenzene units might also affect the observed shift through possible n-bonding. Also, the aromatic ring currents would tend to shift the resonance to higher fields1*and thus tend to decrease the slopes. The magnitude of the shifts would be dependent on the number of benzene rings present and hence should increase in going from the 4% to the 16% DVB resin. The upfield shift should also increase with decrease in distance from the benzene ring to the water protons. Thus the shift should increase somewhat with water content. This last effect would be progressively smaller for the 8 and 16% DVB resins. The effect of the resin on the water structure is apparent in Fig. 2 by causing non-linear variations in the plots. The water in the 8% DVB resin is not able to assume a normal water structure until a fairly high degree of hydration is reached and never becomes as complete as the 4q/, DVB resin. The water in the 16% DVB resin never reaches a high degree of structure before the matrix forces take over and causes a decrease in slope. The conclusions regarding the resin effect on the water structure do not, however, invalidate the previously proposed acid strengths, since association between the sulfonate site and the proton would increase with acid strength and polarity. The width of the internal water line is much larger (20 to 100 c.P.s.) than that of pure water (0.5 c.P.s.) but much less than in ice, indicating that the water molecules are in rapid, though probably somewhat restricted, motion. The results of the line width measurements on l6Y0 DVB resin are shown in Fig. 3. Although extensive data on the 4 and 8% resins were not taken, the indication was that the same trends are present except for a shift of the minimum to progressively lower P from 4 to 16% DVB. The peak in the plot occurs as a layer of water begins to form on the exterior of the bead. A t this point the exchange of protons with the external phase causes the line to (17) E. Glueckauf,Proc. Roy. SOC.(London), 8214, 207 (1952). (18) J, 5 , Waugh and R. W. Feasendeni J , A m . C h a m Boo.$ '79, 846 (1967).
June, 1963
SUCLEAR MAGXETIC REBONASCE STCDY O F IOS-EXCHBNGE
broaden and the spectrum appears as shown in Fig. 1. The shift in the peak position to lower water content with increased resin cross linking reflects the decrease in water capacity of the more highly cross-linked resins. For resin samples with less than 2 moles of water per equivalent of resin the lines become too broad to observe by high resolution techniques. The matrix resinate proton n.m.r. line-width of 8 gauss has been reportedl9and therefore is not observable by high resolution techniques. It is postulated that a large part of the increase in line width of the water inside the resin bead as compared to that in a concentrated electrolyte solution is caused by magnetic field inhomogeneity inside the resin. Above, we have pictured the hydrated resin beads as containing domains of water surrounded by the resin matrix. This structure can be approximated on a semimicro scale by a sample of organic resin beads with water filling the interstitial space. Since a collection of beads does not form a solid mass the water is free to diffuse between interstitial domains. To determine the broadening of this model system, we filled a sample tube with 50 to 100 mesh beads and water. The measured line width was about 10 C.P.S.and was practically independent of the amount of water present. Similar findings have been made by Gordon.12 A second experiment which shows the dependence of the line width on the size of the water domains for domain sizes approaching microscopic dimensions was performed with flowers of sulfur in water. A freshly shaken sample of sulfur dispersed uniformly through the mater phase exhibited a water line comparable in width to pure water. However, as the sulfur settled, the line width increased to a width of approximately 100 c.p.s. This increase is obviously related to the dlecrease in interstitial space between sulfur particles as the sulfur settles, L e . , domain size. This parallels the decrease in water domain size which would accompany the shrinkage of a hydrated resin as the water of hydration is removed, and therefore we would expect an increase in line width as the amount of water in the resin decreases. Other mechanisms of broadening such as restriction of motion of the water molecules and exchange with the acid protons are also probably responsible for some of the observed line broadening. One would expect that as the water content is reduced the remaining water molecules wou1.d become more tightly bound to the resin and thus cause ti decrease in transverse and longitudinal relaxation times T2 and 2'1. Proton exchange in acid-water systems is known to be much too fast to measure by n.m.r. We would, therefore, expect this exchange to be very rapid even in the resin phase since they are strong and easily ionizable acids. The change in line position also shows we are looking at a fast exchange process and this mode would not contribute to the line broadening. A crude measurement of T I by the progressive saturationz0method indicated that the T I for water in the saturated resin was the same as that of pure water and decreased by about an order of magnitude a t the lowest water content. These results should only be taken as an order of magnitude indication. They do (19) L. V, Holroyd, R. S. Codrington, B. A. Mrowca, and E. Guth, J, A p p l . Phvs., 22, 696 (1951). (201 E. L. Hahn and DAEi Maxwell, Physi Reo,, 84, 1246 (1961),
q, 1
, 2
,
RESINS
,
,
,
3 4 5 6 WATER CONTENT.
1181
,
,
7
0
1
Fig. 3.-The variation of the full width at half maximum for hydrated Dowex-50 (16% DVB) resin aEi a function of wat,er content in moles of water per equivalent of resin.
A 47cps
90°C.
ocps
Increasing Ho Fig. 4.4V.m.r. spectra for hydrated Dowex-50 (4% DVB) resin at different temperatures.
indicate that the restriction of motion of the water molecules is not very great. Another possible cause of the increase in line widths is that the line may be composed of a number of close lines arising from water protons in different parts of the resin which are not 'exchanging with each other. At lower degrees of hydration one would expect to find that some acid sites were more highly hydrated than others, and thus, as discussed above, the protons at the different sites need not have the same chemical shift. The experimental spectrum, in this case, would be a statistical distribution of lines forming, apparently, a single broader resalnance line. A critical evaluation of these possibilities awaits the measurements of T I and 2'2 by a more accurate method. On adding excess water t Q the resin, a second line
S.K. ALLEY,JR.,A N D R. L. SCOTT
1182
appears. This line corresponds to the water external to the resin and, for the 47, DVB resin, appears at about, 0.79 p.p.m. upfield from the internal mater peak (see Fig. 1, d and e ) . As the external water is reduced, the two lilies coalesce into a single very broad peak. This behavior is indicative of exchange of resin phase water protons with the external water. The effect of temperature mas studied to confirm this hypothesis. The resulting spectra a t three temperatures are shown in Fig. 4. At approximately 0' the two lines are sharpened, and the minimum between the peaks is greatly reduced. At approximately 90' the valley is much shallower, with the height of the minimum being about 90% of that of the peak (see Fig. 4a). As the samples return to rooin temperature the spectra return to their original shape a t 2 5 O , indicating complete reversibility of this process. The merging of the two lilies as the temperature is raised is consistent with an increase in the rate of exchange with increase in temperature. The measurement of the diffusion rate of water from the external phase into the resin phase has been made, using DzO. il sample of the 4% DVB resin (100 mesh) was saturated with HzO and centrifuged to remove excess water. To measure the diffusion rate, the H20 hydrated resin was immersed in DzO and the time dependent spectrum recorded for the transformation of the single resin water line into the double peak spectrum of the two phase resin-external water system. The reverse process of adding deuterium-form resin saturated with D20 to external HzOalso was studied. I n a third experiment, the water-saturated resin was added to DzO with stirring in a therniostated container. Samples of liquid were withdrawn a t 3 to 5 sec. intervals and assayed for proton content by n.m.r. This latter experiment was similar 'to that of Boyd and Soldano, who used O18-labeled water.21 The data of all three experiments were treated in the manner of Boyd and Soldaiio,22and the value of the diffusion cm1.2 coefficient of the 4% DVR resin was 800 X set.-' corresponding to an exchange lifetime of 13 see., (21) G. E. Boyd and B. A. Soldano, J . Am. Chem. Soe., 76, 6105 (1953). (22) We wish to thank Dr. G. E. Boyd for supplying the theoretical curves
necessary to make these calculations.
Vol. 67
which is within the experimental error of the earlier measurements. *O Conclusions Our observations on the variation of the n.m.r. spectra of the hydrated Dowex-50 resins are consistent with the polyeloctrolyte gel model of the ion-exchange resin proposed by Glueckauf. The change in line position and line width with the decrease in degree of hydration indicate the internal watcr protons are exchanging rapidly enough to average out the internal field inhomogeneities and the chemical shifts of any individual species present. The behavior of the chemical shift of the resin phase water protons is explainable by the strong acid character of the Dowex-50 resin. The shift and line width data lead to a picture of the hydrated resin in which the water molecules form domains of water around the polar acid sites. The physical structure of the water is changed by the polar sites and the organic matrix. The chemical shift data can be interpreted to indicate that the apparent acid strength increases with decrease in resin cross linking. However, other interactions can also affect the measured line position. One such interaction is that of ion pair formation. In this study we are not able to determine the contributions of these other possibilities. I n the presence of excess water the internal and external water protons are easily distinguishable by n.m.r. The temperature dependence of the spectra of this system indicates a rather slow rate of proton transfer between the resin and liquid phases. This result is consistent with the Donnaii membrane model. A quantitative measurement of the rate of proton exchange was carried out using D20. The rate of proton exchange was found to be very nearly the same as the rate of 0l8exchange measured earlier by Boyd. Acknowledgment.-This research was supported by contracts with the U. S. Atomic Phergy Commission and the Office of Xaval Research. We also wish to thank Dr. Ernest Grunwald for his interest and encouragement and both Dr. Grunwald and Dr. Robert Kromhout for helpful discussions.
N.M.R. STUDIES OF HYDROGEN BONDING I N HYDROFLUOROCARBON SOLUTIONS BY S.K. ALLEY,JR.,AND R. L. SCOTT Department of Chemistry, University of California, Los Angeles 24, California Received October 16, 1962
Suclear magnetic resonance spectroscopy has been used as a tool for obtaining information regarding the strengths of hydrogen bonds formed between CiF,,H and electron donor solvents such as ketones and amines. Concentration and temperature dependences of chemical shifts of the involved data were used to obtain free energies and enthalpies of hydrogen bonding. The strengths of the hydrogen bonds between CiF16H and ketones or amines were found to be in the range of 2-5 kcal./niole. These specific interactions substantially change the solubility of C?FlIHin electron donor solvents as compared with that of C ~ F M .
Introduction Fluorinated carbon compounds contailling terminal protons show considerably different solubility behavior than do those which are fully fluorinated or fully protonated. These terminal protons are quite electrophilic
and are capable of forming hydrogen bonds with appropriate Lewis bases such as ketones and amines. The magnitude Of these interactions is difficult to measure because of overlapping effects such as differences in intermolecular attractions between fluorine-hydrogen
