Electrolyte Viscosities in Associated Solvents1 - ACS Publications

JOHN P. BARE AND JAMES F. SKINNER transfer from water to a solvent of lower dielectric con- stant, and the decrease in pK, for (CH&NH+ from. 9.76 to 8...
1 downloads 0 Views 901KB Size
434

JOHNP. BAREAND JAMES F. SKINNER

transfer from water to a solvent of lower dielectric constant, and the decrease in pK, for (CH&NH+ from 9.76 to 8.6 on transfer from HzOto solvent S is as anticipated for an acid of type AH+. The increase of 106 in the second-order rate constant for reaction of KO- with p-nitrophenyl acetate is attributable to an increase in the activity of the solvated ion compared to that of the

transition state. Both the low mobility and enhanced nucleophilicity of the hydroxyl ion in solvent S may be related, a t least in part, to a structure as I. Contribution to the high activity of I in solvent S arises from the decreased activity of HzO in solvent S. The reaction can be written as in (2). For (CH3),N thie magnitude

HO-(H?O)a

H

h /

---f

[HO-X]+

+ 3Hz0

(2)

of rate enhancement is not seen indicating little alteration of the relative activity of ground and transition state on solvent transfer.

H H---d --H

H-0

+X

Acknowledgment. This work was supported by a grant from the National Institutes of Health. Appre ciation is expressed to an unknown referee who suggested we consider structure I.

A--0 HI

I

Electrolyte Viscosities in Associated Solvents1 by John P. Bare and James F. Skinner* Department of Chemistry, William8 College, Williamstown, Massachusetts 01.967

(Received May 6 , 1071)

Publication costs assisted by Williams College

Jones-Dole viscosity B coefficients have been determined at 25’ for NaI, KI, and CsI in a number of hydrogenbonded solvents. B(Na1) increases from 0.357 to 1.135 for the series of solvents: glycerol, 1,3-propanediol1 l12-propanediol,methanol, 1-propanol, !&propanol,1-butanol, 1-pentanol, and 1-hexanol. This increase has been attributed to a corresponding decrease in the hydrogen-bonded association of the solvents. In the two dihydric alcohols, B decreases with increasing crystallographic cation radius, B(Cs1) being negative. The are similar to the values in the monohydric values of B (NaI) in 2-aminoethanol and 2,2’,2’’-nitrilotriethanol alcohols, while B(Cs1) assumes small but positive values in these solvents. B(K1) and B(Cs1) are both more negative in DzOthan in HzO correlating with the previously reported enhanced association in the former solvent.

Introduction The concentration dependence of the viscosity of a q ~ e o u s , ~n-o~n a q ~ e o u s , ~and - ~ mixed s o l ~ e n telec~,~ trolytic solutions has been interpreted in terms of the semiempirical Jones-Dole equationlo 9/90 =

1

+ AC’I2+ BC

(1)

where 9 and qo are the solution and solvent viscosities, C is the molarity, and A and B are adjustable paramet,srs. The square-root term represents the contribution to the viscosity from the ion-ion coulombic interactions, taken into account by Falkenhagenl’ in terms of limiting equivalent conductances and solvent properties. For monatomic ions in aqueous solution, the B coefficient has been interpreted in terms of specific ionsolvent interactions ? positive values being attributed The Journal of Phgsical Chemistry, Vol. 76, N o . 9, 1978

to an enhancement of the solvent association by the electrolyte and negative values being attributed to a (1) This work is based in part on the honors thesis of J. B., 1970. (2) H. S. Frank and W.-Y. Wen, Discuss. Faraday Soc., 24, 133 (1957). (3) R. H. Stokes and R . Mills, “Viscosity of Electrolytes and Related Properties,” Pergarnon Press, Elrnsford, N. Y., 1965. (4) B. R. Breslau and I. F. Miller, J. Phys. Chem., 74, 1056 (1970). (5) G. Jones and H. J. Fornwalt, J . Amer. Chem. Soc., 57, 2041 (1935). (6) E. M. Mukherjee, J. Phys. Chem., 74, 1942 (1970). (7) R. Gopal and P. P. Rastogi, Z. Phys. Chem. (Frankfurt am Main), 69, 1 (1970). (8) D. Feakins, D . J. Freernantle, and K. G. Lawrence, J . Chem. Soc. D,970 (1968). (9) D. Singh, V. 6. Yadav, and B. K. Goel, Z . Phys. Chem. (Fyankfu?-t am Main), 68, 242 (1969). (10) G. Jones and M . Dole, J. Amer. Chem. Soc., 51, 2950 (1929). (11) 1%.Falkenhagen and M. Dole, Phys. Z . , 30, 611 (1929).

435

ELECTROLYTE VISCOSITIES IN ASSOCIATED SOLVENTS weakening of the hydrogen bonding in the solvent. I n a previous communication from this laboratory,12 the first negative B values in nonaqueous media were reported. The present investigation represents an extension of our study of electrolyte viscosities to a variety of alcoholic solvents. While hydrogen-bonded association would be expected to be a significant factor in determining the behavior of all of these solvents and solutions therein, an effort has been made to select solvents in which the relative importance of this phenomenon could be systematically evaluated through viscosity measurements. I n studying the properties of the alkali halides in nonaqueous solvents, one is restricted in the choice of solute by the extreme hygroscopicity of the lower formula weight compounds and the low solubility of the higher formula weight compounds below a solvent dielectric constant of about 30. For these ieasons, the number of alkali halides whose specific viscosities have been measured is limited. The purposes of the present study were fourfold. First, importance of the cation in the electrolyte B coefficient in highly structured solvents was elucidated through a study of NaI, KI, and CsI in 1,2-propanediol ( 6 2 0 = 32.0) and 1,3-propanediol (ep0 = 35.0) and NaX in glycerol (1,2,3-propanetriol) to complement the previous workI2 on K I and CsI in the last solvent. Second, B(Na1) was determined in six monohydric alcohols-methanol, 1-propanol, 2-propanol, 1-butanol, 1-pentanol, and 1-hexanol-to illustrate the effect of increased alkyl chain length in the solvent. Third, viscosities of 2-aminoethanol and 2,2”2’’-nitrilotriethanol (triethanolamine) solutions of NaI and CsI were studied which, in conjunction with B coefficients for these electrolytes in comparable hydroxylic solvents, enabled an estimation of the relative importance of the amino and hydroxyl groups in ion-solvent interactions as exhibited in electrolytic viscosities. Finally, it has been suggested that heavy water (DzO) is more associated than ordinary water (HzO), and as this should bt: reflected in the B coefficients of alkali halides, the viscosities of CsI and KI have been studied in these two solvents.

Experimental Section Cesium iodide (A. D. Mackay Inc., 99.9%), potassium iodide (J. T. Baker, analyzed reagent) and sodium iodide (Alfa Inorganics, Ultrapure) were used as received, after drying at 110’. While the sodium salt was adequately soluble in the monohydric alcohols of three or more carbons, neither of the other salts was sufficiently soluble for viscosity measurements. The DzO (Mallinckrodt, 99.8% isotopic purity) was handled in a drybox and the HzO was twice distilled. The (fiqallinckrodt, anhydrous reagent) refluxed Over magnesium filings under nitrogen for 3 days and fractionally distilled: boiling

range of fraction collected, 64.5-64.7 (lit.la bp 64.509) ; density, 0.78652 g/ml (lit.14value 0.78655). The remaining solvents, 1-propanol (A), 2-propanol (B), 1-butanol (C), 1-pentanol (D), 1-hexanol (E), 2-aminoethanol (F), 2,2’,2’’-nitrilotriethanol(G), 1,3propanediol (H), 1,2-propanediol (I), and glycerol (J), were dried in the dark for 3 weeks over one of the following dehydrating agents : Na2S0~-K2C03(I), CaC12-KzC03 (11), Na2S04 (XII). The solvent was then fractionated under dry nitrogen either at 1 atm (heated 75-cm column of glass helices) or a t reduced pressure (heated 40-cm Vigreux column), a middle fraction being collected. The purified solvent was stored in the dark under nitrogen. It was necessary to perform two distillations on the 1-hexanol to give a solvent which did not discolor on the solution of the sodium iodide. The details of the individual purifications are shown in Table I.16--23 All solutions were prepared by weight in flasks previously flushed with dry nitrogen and were stirred overnight in the dark prior to use. All measurements were made at 25 =k 0.81”. Densities were measured in Sprengel-type pycnometers (22 ml) calibrated with distilled water. Viscosities were measured in a Cannon-U bbelohde viscometer for the water, heavy water, and methanol solutions and in Cannon-Fenske routine viscometers for the other systems. The viscometers were modified for use in a closed, dry atmosphere. Flow times (400-600 sec) were reproducible to +0.1 sec. Specific viscosities, vsp, were determined from vsp

=

vr

- 1=

tp/topo

(2)

where p, pot t, and to are the densities and flow times for the solution and solvent, respectively. The A and B coefficients were determined from the plots of qSp/ C’/z against C’/z (Figures 1-4). The linear dependence of solution density on electrolyte molarity was evaluated from (12) K. Crickard and J. F. Skinner, J. Phys. Chem., 73, 2060 (1969). (13) “Organic Solvents, Physical Properties and Methods of Purification,” John A. Itiddick and Emory E. Toops, Jr., Ed., Intersct ence, New York, N. Y., 1955. (14) M. A. Coplan and R. M. Fuoss, J . Phys. Chem., 68, 1177 (1964). (15) J. Timmermans, “Physioo-Chemical Constants of Pure Organic Compounds,” Elsevier, Amsterdam, 1950. (16) “Handbook of Physics and Chemistry,” Robert C. Woast, Ed., 49th ed, Chemical Rubber Publishing Co., Cleveland, Ohio, 1968. (17) F. Hovorka, H. P. Lankelma, and S. C. Stanford, J . Amer. Chem. Soc., 60, 820 (1938). (18) J. N. Pearce and L. F. Berhenke, J . P h w Chem., 39, 1005 (1935). (19) P. W. Brewster, F. C. Schmidt, and W. 3.Schaap, J. Amer. Chem. SOC., 81, 5532 (1959). (20) “Dictionary of Organic Compounds,” Eyre and Spottiswoodc Ltd.,

lgB5.

(21) A. F. Gallaugher and H. Kibbert, J. Amer. Chem. sOc.3 59,2514 (1937). (22) T. T. Puck and H. Wise, J . Phys. Chem., 50, 329 (1946). (23) H. T. Briscoe and W. T.Rinehart, ibid., 46, 387 (1942). The Journal of Physieal Chemistry, VoE. 76,No* 8,PQYB

JOHNP. BAREAND JAMESF. SKINNER

436 Table I : Purification of Solvents" Drying agent

Sol-

vent

Supplier

A B C D E

Fisher Aldrich (QQOJ,, anh.) Eastman (White Label) Eastman Eastman Eastman (White Label) Eastman (White Label) Aldrich Aldrich Eastman

F G H

I

J a

I1

Boiling range,

96.8-97.1 (97.15,13) 81.0-82.0 (82.40,13) 117.3-117.5 (117.73,13) 138.0-138.2 (138,06,13) 157.4-157.6 (157.47,13) 63.8-64.O(32mm) (65.0(5mm), 18) 168.3-168.8(1mm)(175.0(2mm), 18) 115.8-115.9 (12.5 mm) 97.3-97.5 (17 mm) 139.2-140.1 (1 mm)

I I I11 I11 I11 I11

I I11 I11

Densities and refractive indexes a t 25" unless otherwise noted.

-1

=

Density, po

"C

Refractive index

0.79998 (0.79950,13) 0.78093 (0.78095,13) 0.80581 (0.80572,15) 0.81096 (0.81104,16) 0.81583 (0.81556,17) 1.0117 (1.0117,19) 1.12081 (1.1242,20", 16) 1.04892 (I.050,21) 1.03267 (1,0328,22) 1.25824 (1.2583,23)

1.3829 (1.3835,13) 1.3747(1.3747,13)

...

... ...

1,4533(1.4639, 20°, 13) 1.4832(1.4852,20",20) 1.4380(1.4396,20', 13)

... ...

Literature values and references are in parentheses.

kC

(3) Values of A , B , and IC, computed by the method of least squares, are given in Table II.24 p/po

Table I1 : Solution Parameters at 25' Elec-

trolyte

Solvent

k

A

0.028 i 0.005 0.077 f 0,005 0.094 i 0.004 0.0156 f 0,0005 0.016 & 0.002

NaI

Methanol

0,1736

NaI

1-Propanol

0.1580

NaI

2-Propanol

0.1651

NaI

1,3-Propanediol

0.1079

NaI

1,2-Propanediol

0.1049

NaI

1-Butanol

0,1539

0,004 0.085 i 0.1522 0.004 0.032 =k 0.1475 0.004 0.07796 0.003 f 0.001 0.010 fi 0.1136 0,001 0,09567 0.043 f

NaI

1-Pentanol

NaI

1-Hexanol

NaI

Glycerol

NaI

2-Aminoethanol

NaI KI

2,2',2''-Nitrilotriet hanol 1,2-Propanediol

KI

1,3-Propanediol

KI CSI

Water 0,1227 Deuterium oxide 0,1005 1,2-Propanediol 0.1938

CSI

1,3-Propanediol

0.1937

CSI

2-Aminoethanol

0.2021

CSI

2,2',2"-Nitrilo0.1768 triethanol Water 0.2017 Deuterium oxide 0.1798

KI

CSI CSI

0.099 =t

o.od5 0.1166 0.1125

0.037 i 0.005 0.024 =k 0,005

... ...

0.043 i 0.005 0,009 i 0.001 0.035 f 0.001 0.024 f 0.005

... ...

The Journal of Physical Chemistry, Vol. 76,No. 3,1978

B

0.652 f 0.010 0.826 f 0,010 0.842 f

0.009 0.415 f 0.007 0.475 i 0.003 0.883 f 0.007 0.989 i 0,007 1.135 & 0,008 0.357 i 0.005 0.843 f 0.007 0.977 i 0.007 0.069 f 0.002 0.053 ZIZ 0.002 0.0719 -0.096 -0,106 i 0.005 -0.116 f 0.005 0.385 f 0,003 0.204 f 0.010 -0.120 -0.134

-

I

0.2

I

I

0.4 c+ 0.6

oh

Figure 1. Viscosity B coefficient plot: A, NaI in 1,2-propahediol; B, NaI in 1,3-propanediol; C, NaI in glycerol; D, KI in 1,2-propanediol; E, K I in 1,3-propanediol; F, CSI in 1,2-propanediol; G, CsI in 1,3-propanediol.

Discussion The presence in one water molecule of two hydroxyl protons and two pairs of nonbonded oxygen electrons results in the unusually high degree of intermolecular association in liquid water at room temperature. A single water molecule can participate in hydrogen bonding simultaneously with four other water molecules. The search for a generally acceptable model on which may be based an explanation of the behavior of liquid water continues in many laboratories. The fact that in dilute aqueous solution, alkali halides of high surface charge density give solutions more viscous than the solvent while those of low surface charge density give (24) Listings of concentration, density, and specific viscosity will appear immediately following this article in the microfilm edition of this volume of the journal. Single copies may be obtained from the Business Operations Office, Books and Journals Division, American Chemical Society, 1155 Sixteenth Street, N.W., Washington, D. C. 20036. Remit check or money order for $3.00 for photocopy or $2.00 for microfiche.

ELECTROLYTE VISCOSITIESIN ASSOCIATED SOLVENTS

437

1

I

0.1

, CF

I

I

0.2

03

Figure 4. Viscosity B coefficient p l o t : i , KI in H z O ; ~ ,] K I in D20; , CsI in HzO; , CsI in DzO.

0

Table 111: B Coefficients in Polyhydric AlcoAols a t 25' Figure 2. Viscosity B coefficient plot: N a I in 1-hexanol (A), 1-pentanol (B), 1-butanol (C), 2-propanol (D), 1-propanol (E), and methanol (F). The vertical scale has been lifted by 0.05 and 0.10 for 1-pentanol and 1-hexanol, respectively, for clarity.

N e1

Glycerol 1,2-Ethanediol 1,2-Propanediol 1,3-Propanediol

0.357

...

0.475 0.415

KI

-0.185" 0.033" 0.069 0.053

CSI

- 0.408" -0.080" -0,106 -0.116

Reference 12.

// o

0.0'

I

I

0.2

I

ci

I

I

I

0.4

Viscosity B coefficient plot: A, N a I in 2,2',2"-nitrilotriethanol; B, N a I in 2-aminoethanol; C, CsI in Saminoethanol; D, CsI in 2,2',2"-nitrilotriethanol.

Figure 3.

"negative viscosities" or solution viscosities less than the solvent has been explained in terms of the effect of the electrolyte on the quasicrystalline structure of water. A polyhydric alcohol molecule should also afford the opportunity for extensive intermolecular association, where each hydroxyl group could, in principle, hydrogen bond to two other molecules, resulting in a threedimensional solvent structure similar to that in water. Table I11 summarizes the B coefficients for three alkali halides in four polyhydric alcohols and Figure 1 illustrates the validity of eq 1 for these systems. The

linearity of the qsp/C'/z against C'/' plot for NaI in glycerol to concentrations in excess of 0.6 M corroborates our previous findings12in this solvent. Two points can be made from these results. First, the B coefficient becomes less positive with decreasing surface charge density of the cation in agreement with behavior found in water. The cesium salt is the only one to exhibit negative B values in all four solvents, although B(K1) is negative in glycerol and close to zero in the other solvents. Second, the B coefficients in Table Ill are less positive than those for these electrolytes in any other solvent investigated to date, including those in this study. This suggests that in these four solvents the hydrodynamic effect of ion-solvent interaction is different from that in unassociated or weakly associated solvents, but similar to that exhibited in water. A search for further solvents in which this phenomenon will be present will probably be hampered by decreasing solubility of the lower surface charge density electrolytes with decreasing dielectric constant. Formamide (ez5 = 109.5) might, however, prove to be an interesting solvent from the point of view of negative or very small B coefficients as several of its properties are similar to those of water. From a recent study of proton relaxation rates in aqueous and nonaqueous electrolytic solutions, Engel and Hertz25 reported findings in good agreement with (25) G. Engel and H. G . Hertl;, Ber. Bzmsenges. Phys. Chern., 7 2 , 808

(1968). The Journal of Physical Chemistry, Vol. 76, N o . 3,1972

JOHN P. BAREAND JAMESF. SKINNER

438 previous interpretations of positive and negative Jones-Dole B coefficients. Jn water, 1,2-ethanediol1 and glycerol, those electrolytes which exhibited negative viscosity B coefficients also gave solvent-proton relaxation rates faster in solution than in the pure solvents. I n methanol, ethanol, formamide, and Nmethylformamide, the other solvents studied in which hydrogen bonding might be expected to be significant, evidence for a weakening of solvent structure was not found in the relaxation rate studies. Intermolecular hydrogen bonding in monohydric alcohols will lead to dimeric or polymeric associationz6 depending on the nature of the alkyl group. I n neither case, would there result anything resembling the ex-

H //‘

\

\

//’

R-0

0-R

H

R

\0-H

R

R

_____\ -.O-H

- _ _ _ 0-H __

__..__

\

R

\O-H.-----. ______

tensive three-dimensional solvent structure described above. These substances would be expected to be intermediate between the polyhydric alcohols and liquids such as dimethyl sulfoxide, propylene carbonate, and acetone where intermolecular interactions would be minimal. There have been a number of studies of electrolyte viscosities in methanol and ethanol. I n three of these s t u d i e ~ the , ~ ~data ~ ~ are ~ either too scattered or a t too high concentration to permit calculation of B coefficients. Jones and FornwaW reported values of E of 0.6747, 0.7396, and 0.7635 for KI, KBr, and KCI, respectively, in methanol at 25”. Their plot of q8p/C’’z against C”z for K I shows a distinct break at approximately 0.02 M and then a further linear portion with a smaller slope of 0.42. Cox and WolfendenZ8reported a value of B of 1.15 for XaI in ethanol at 18”. A recalculation of B, excluding two points of large deviation, gives a value closer to 1.05. The authors are unaware of any viscosity studies on the alkali halides in any of the longer chain alcohols, Table IV gives B values for Nar, chosen because of its suitable solubility, in seven monohydric alcohols. These results represent the first study of the JonesDole B coefficient in a homologous series of solvents. With one exception, the increase in B parallels the (see Figure 2)* increase in the length Of the As the size of the solvent molecule increases, the polar group remaining the same, the volume of the hydrodynamic entity consisting of the ion and the oriented solvent molecules will be expected to increase. Even degree Of Orientation Of the in ‘-hexan’’ a hydroxyl groups about the cation must be present in The Journal of P h ~ s i c a lChemistry, Vol. 76, No. 3, 197.9

~

Table IV:

B Coefficients of NaI in Monohydric Alcohols Methanol Ethanol 1-Propanol 2-Propanol 1-Butanol I-Pentanol 1-Hexanol

a

0.652

1. 15a ( 1 . 0 5 ) 0.826 0.842 0.883

0,989 1.135

Reference 28, 18”. Recalculation gives 1.05.

order that the electrolyte dissolve approciably. Janz29 reported viscosity data for NaI in acetonitrile, but again the concentrations were too great to permit an evaluation of B for comparison with the present work. Feakins30reported values of 0.608 and 0.573 for NaCl and NaBr in N-methylformamide, from which a value of 0.53 could be approximated for NaI. The B(Na1) values in the monohydric alcohols are, therefore, more positive than any values reported in other solvents. As part of our investigation of associated solvents, it was thought that the ethanolamines would present an interesting comparison with simple alcoholic solvents. These solvents have received little attention. Pearce and Berhenkels reported the dipole moments of the mono-, di-, and triethanolamines. Briscoeal reported some conductance results in monoethanolamine which were later found to be in error by Brewster’g who made more careful correction for solvent conductance. Diethanolamine is a solid at 25” and therefore could not be included in the present study. The B coefficients (see Table 11) show the same strong dependence on cation, B(Na1) - B(Cs1) being 0.458 and 0.773 in 2-aminoethanol and 2,2’,2’’-nitrilotriethanol, as has been found in the highly associated polyhydric alcohols, but in all cases the B values are positive. Even in 2,2’,2’’-nitrilotriethanol,a solvent similar to glycerol in its physical properties, B(Cs1) is small but positive (0.204). The B(Cs1) values of 0.385 and 0.204 in these two associated solvents are significantly below the value of 0.68 in dimethyl s~lfoxide,3~ a solvent in which hydrogen bonding would not be expected t o be present. (26) (a) H. Eyring, M . 5. Jhon, J. Grosh, and E. R. Van Artsdalen, J . Chem. Phys., 47, 2231 (1967); (b) M. Saunders and J. B. Hyne, ibid., 29, 1319 (1958); (c) G. E. McDuffie, Jr., and T. A. Litovitz, ibid., 37,1699 (1962) ; (d) L. H. Thomas and R.Meatyard, J . Chem. soc., 1986 (1963). (27) (a) F. H. Getman, J . Am. Chem. Soc., 30, 1077 (1908); (b) F. K. Ewart and H. R. Raikes, J. Chem. Soc., 1907 (1926). (28) W. M. Cox and J , F, Wolfenden, proc. Roy. Sot, Ser. A , 145,475 (1934). (29) R. P. T. Tomkins, E. Andalaft, and G. J. Jans, Trans. Faraday Soc., 65, 1906 (1965). (30) D. Feakins and K. G. Lawrence, J . Chem. Soe. A , 212 (1906). (31) 13. T. Briscoe and T. P. Dirkse, J . Phys. Chem., 44, 388 (1940). (32) M. D. Archer and R. P. H. Gasser, Trans. Faraday Sac., 62,3451 (1966).

ELECTROLYTE VISCOSITIES IN ASSOCIATED SOLVENTS The B(Na1) values of 0.826, 0.843, and 0.415 in 1-propanol, 2-anninoethano1, and 1,3-propanediol, respectively, suggest that from the point of view of viscosity, the effects of the methyl and amino groups in the 2 position are indistinguishable, while the second hydroxyl group produces a much greater degree of solvent structure, disrupted by the electrolyte, giving the smaller value of 0.415. suggests that A substantial amount of at room temperature, the intermolecular hydrogen bonding is more extensive in heavy water (DzO) than in ordinary water (H20). For example, the viscosity (25’) of D 2 0 (1.096 cP) is about 20% greater than that of HzO (0.890 cP). The temperatures of maximum density33 of DzO and HzO are 11.23 and 3.98, respectively, suggesting that the quasi-crystalline solvent structure is more thermally stable in the deuterated solvent than in the protonated solvent. A number of investigations of the properties of electrolytes in DzO have been made.36-38 The viscosities of DzO solutions of various alkali metal salts have recently been rep0rted.393~0 However, in both cases, the concentrations used were greater than those for which the Jones-Dole equation is thought t o be valid. The B coefficients for the present investigation, together with values for several quaternary ammonium halides in H z 0 3and D20137are given in Table V. As far as the authors know, a comparison of the latter values has not been made. Although the viscosities of H2O solutions of Cs141and K142have been reported, it was decided to repeat the measurements in this laboratory for consistent comparison with the DzO results. Table V : B Coefficients in HzO and D20 (25”) Electrolyte

KI CSI (CH8hNBr (CzHshNBr (CzH7hNBr (C4Ha)aNBr a

Reference 42.

HzO

-0.0719 - 0.0755“ -0.120 -O.lMb 0.0855 0.349 1.06 1.36

DzO

-0.096 -0.134

0.08 0.31 0.79 1.26

Reference 41.

The present values of the B coefficients in HzO are in satisfactory agreement with the literature values. The values in DzO for the potassium and cesium salts are approximately 30 and lo%, respectively, more negative than the corresponding H 2 0 values. This suggests that these electrolytes destabilize the solvent association t o a greater extent in D20. Figure 5 compares the present results on KI in DzO with those of S e l e ~ k i . I~t~is interesting t o note that the two most

439

0.0‘

%$,

. . \ . e

0

e

?O‘b

-0%-

I

0

09

..e.

0

1.0

I

2.0

CT

Figure 5 . Viscosity B coefficient plot: 0, KI in DzO; below C’/z = 0.5, this work; above C’Ia = 0.5, ref 40; 8 , CsCl in HzO, ref 39; 0 , CsCl in DaO, ref 39.

dilute points in Selecki’s work are approximately collinear with the present results. O ~ t r o f f ’ sresults ~~ on CsCl in H 2 0 and DzO are also shown in Figure 5 . The Falkenhagen expression11 would predict very similar values of A , the ordinate intercept on this plot, for CsCl in the two solvents, so i t appears probable that work at lower concentrations would show a more negative B(CsC1) in D2O than in H 2 0 . The minimum and subsequent increase above about 1 M illustrate the need in the Jones-Dole equation for terms of higher order in concentration. The B coefficients for the quaternary ammonium salts are less positive in DzO than in HzOin agreement with the alkali halide results. When the present results are viewed in the light of the literature on viscosity B coefficients for alkali halides in aqueous and nonaqueous media, some tentative generalizations can be made. The interpretation of aqueous B coefficients presented by Frank and Wen2 in terms of “structure makers” and “structure breakers” appears to be sound and has been widely accepted. Stokes3 has utilized the Frank-Wen model for ionsolvent interactions in suggesting an explanation of electrolytic viscosities based on three separate contributions i o the B coefficient: qE, the viscosity increment resulting from the size and shape of the ions; qA, an increase in viscosity resulting from ordering of (33) G. Nemethy and H. A. Scheraga, J . Chem. Phys., 41,680 (1964). (34) M. Falk and T. A . Ford, Can. J . Chem., 44, 1699 (1966). (35) H. A. Risk and Y. M. Girgis, 2. Phys. Chem. (Frankfurt a m M a i n ) , 65, 269 (1909). (36) J. Greyson, J . Phys. Chem., 66, 2218 (1962); 71, 2210 (1967). (37) R. J. Kay and D. F. Evans, ibid., 69,4216 (1965). (38) C. G. Swain and D. E’. Evans, J. Amor. Chem. SOC.,88, 383 (1966). (39) A. G. Ostroff, B. S. Snowden, Jr., and D. E. Woessner, J.Phys. Chem., 73, 2784 (1909). (40) A. Selecki, B. Tyminski, and A. G. Chmielewski, J . Chem. Eng. Data, 15, 127 (1970). (41) G. Jones and H. J. Fornwalt, J . Amer. Chem. Soc., 58, 619 (1936). (42) M. Kaminsky, Z. Phus. Chem. (Frankfurt a m M a i n ) , 5 , 154 (1955). The Journal of Physical Chemistry, Vol. 76, No. 3,1972

440 the solvent molecules in the immediate vicinity of the ions as a result of the ionic field; and finally, qD, a decrease in the solvent viscosity as a result of the disruptive presence of the ions. For alkali halides, the magnitude of qE will reflect the extent of the solvation and the size of the solvent molecules in the cosphere of the ions. The relative sizes of the three terms resulting from ion-solvent interaction will determine the size and sign of the B coefficient. I n water, the B coefficients are all numerically small, ranging from B(LiC1) of 0.140 t o B(Cs1) of -0.120, and because the contribution to B from the qE term will be small, the viscosity will be determined by the relative sizes of the qA and q” terms in the Stokes expression. If the structure enhancement term predominates, the B value will be positive while if structure disruption predominates, B will be negative. Hence, the sign of B has been used to distinguish the two classes of electrolytes. It should be noted, however, that this approach may require certain modifications in discussion of nonaqueous solutions. Very little has been said about the applicability of the Frank-Wen model and the Stokes partition of the B coefficient for nonaqueous solutions. I n these cases, the polar solvent molecules are larger than the water molecule and the ionic cosphere will be larger, giving more importance t o the qE term. Even if qD is larger than qA, the overall sign of B may be positive. The authors would suggest that in nonaqueous solvents, it is not possible to say that only when B is negative is the electrolyte participating in a net structure-breaking interaction with the solvent. For example, the small positive values of B(Cs1) of 0.385 and 0.204, relative t o the B(Na1) values, in 2-aminoethanol and 2,2‘,2“nitrilotriethanol should probably be interpreted as evidence for predominance of the qD term over the qA term with the qE term fairly large (and positive) for both electrolytes. I n the interest of brevity, a complete tabulation of all the B values considered will not be made, but the relevant literature will be cited. First, fairly extensive work on the electrolytes Nal, KI, and CSI shows the wide range of values which the Jones-Dole coefficient can assume for a given electrolyte. B(K1) varies from -0.0719 in water to 1.30 in N-methylpropionamide7 while B(Cs1) varies from -0.408 in glyceroll2 t o 0.68 in dimethyl s ~ l f o x i d e . ~ This ~ variation points t o the complexity of the hydrodynamic phenomena for which an explanation is sought. Second, there is a fairly clear distinction between solvents in which the B coefficient shows a dependence on the alkali halide in question and those solvents in which the alkali halide B coefficients are all rather close together. The authors have interpreted this as evidence for the relative importance of intermolecular solvent association. I n water, the polyhydric alcohols, and the ethanolamines, the results t o date show that B always becomes The Journal of .Physical Chemistry, Vol. 78, N o . 3, 197.9

JOHN P. BAREAND JAMES F. SKTNNER less positive with increase in cation or anion size. I n water, glycerol, l,Zpropanediol, and 2,2’,2”nitrilotriethanol, the difference B(Na1) - B(Cs1) is 0.128, 0.765, 0.581, and 0.773, respectively. Only in these solvents where B appears t o depend on both the solvent and the electrolyte are negative values of B found. The dependence of B on electrolyte in dirnethylforma~ide~ (LiC1, 0.59; KI, 1-10; 25”) is difficult to explain in terms of solvent association because hydrogen bonding should not be possible in this case. I n the second group of solvents, where solvent association would be less pronounced, are dimethyl sulfoxide, the monohydric alcohols, N-methylformamide, N-methyla~etamide,~ N-methylproprionamide, and propylene carbonatee6 I n methanol, B values of 0.652, 0.7635, 0.7396, and 0.6747 have been reported5 for NaI, KC1, KBr, and KI, respectively. I n N-methylformamide values of 0.59, 0.608, 0.573, 0.634, 0.590, 0.56, and 0.60 have been reported30~~~ for LiC1, NaC1, NaBr, KC1, KBr, KI, and CsI while in N-methylproprionamide values of 1.25, 1.30, and 1.37 have been 1eported7,~~ for LiC1, KI, and KC1. I n contrast t o what was observed in the highly associated solvents, the B coefficients for the different alkali halides, in a given solvent, show only a small variation with surface charge density. Where comparison is possible in methanol, N-methylformamide, and Nmethylproprionamide, B increases slightly with increasing cation size, in marked opposition to the trend in water, the polyhydric alcohols and the ethanolamines where the larger cations are the better “structure breakers.” The hydrodynamic effect of cation-solvent interaction is quite different in this second group of solvents. The B coefficients decrease slightly in these solvents with increasing anion size. It should be noted that in both groups of solvents, a given electrolyte will show considerable variation in the B coefficient in different solvents. A simple explanation of the large variation in B with solvent for a given electrolyte and, secondly, of the classification of solvents into two groups depending on the sensitivity of B t o the nature of the electrolyte, is not possible. One would predict that both qE and qA would become less positive and qD would become more negative as the ion size increased with the corresponding decrease in coulombic field at the surface of the ion. These changes would all be expected to reduce B and would explain the sensitivity of B to the nature of the electrolyte in the highly associated solvents. I n the unassociated solvents, qA and qD would be negligible and the magnitude of B would reflect primarily the size of the solvated ions. For a given solvent, one would expect the larger ions to be less (43) P. P. Rastogi, Bull. Chem. Soc. Jap., 43, 2442 (1970). (44) 1’.B. Hoover, J . Phys. Chem., 68, 576 (1064).

ELECTROtYTE

VISCOSITIES I N ASSOCIATED SOLVENTS

solvated and therefore to exhibit smaller B values. For a given electrolyte, the size of the qE contribution could be interpreted in terms of the relative sizes of the solvent molecules in the solvation shell. As pointed out already, the B(Na1) values in the monohydric alcohols, with one exception, increase with increasing size of the solvent molecules. The values’p43 of B(K1) of 0.56, 1.01, and 1.30 in N-methylformamide, N-methylacetamide, and N-methylproprionamide also appear to reflect the size of the solvent molecule. Some hydrogen-bonded association would be present in the amides, and it would probably be most pronounced in the least sterically hindered N-methylformamide. A case could be made for explanation of the increase in B in the amides as a result of decreasing importance of the vD term, relative to the qA term, in going from N-methylformamide to N-methylproprionamide. Finally, the values of B(Na1) of 0.977, 0.826, 0.415, and 0.357 in 2,2’,2’’-nitrilotriethanol, l-propanol, 1,&propanediol, and glycerol present an interesting series. The monohydric alcohol presumably represents a case where qA and qD are not important and B reflects the size of the solvated shell of the ion through the qE contribution. I n 1,3-propanediol, the ions would still be expected to interact with only one end of the solvent molecule and the size of the ionic cosphere would be similar to that in l-propanol. The decrease in B(Na1) of 0.411 reflects the dominance of the qD term over the vA term, what Frank and Wen have referred to in water as net “structure breaking.”2 It

441 would be unwarranted to predict the nature of the solvation in glycerol where the ion is exposed to a highly polar, strongly associated medium. It, is unlikely that the presence of any ion could increase the degree of solvent association ih a solvent already so extensively hydrogen bonded, making the vA term negligible. Although qD would be expected to be more important in glycerol than in 1,3-propanediol, it is not possible to estimate the relative importance of changes in vE and qD in bringing about the further small decrease of 0.058 in B(Na1). B(Na1) in 2,2”2’’-nitrilotriethanol (0.977) is an interesting example. The solvent molecules are large with four polar sites possible for ion-solvent interaction. The difference, B(Na1) B(CsI), of 0.773 suggests that specific ion-solvent interactions are important. However, it again seems unlikely that an ion could increase the hydrogen bonding in a solvent already so associated. The value of B will be determined by the relative magnitudes of the vE and vD contributions, the former being more important for NaI giving B = 0.977 while the two terms achieve more equal weight for CsI giving a B value of 0.204. The speculative nature of these suggestions emphasizes the need for much more investigation of ionsolvent interactions for a wider variety of electrolytes in nonaqueous media.

Acknowledgment. The authors gratefully acknowledge funds made available as part of a grant to Williams College by the Alfred P. Sloan Foundation.

The Journal of Physical Chemistry, Vol. 76, No. 3, 1972