J. Phys. Chem. 1985,89, 1291-1296
1291
Electrically Induced Changes in Latex Structure M. Tomita and T. G. M. van de Ven* Pulp and Paper Research Institute of Canada and Department of Chemistry, McGill University, Montreal, Quebec, Canada, H3A 2A7 (Received: June 12, 1984; In Final Form: November 29, 1984)
Electrically i n d u d changes in latex structure have been studied experimentally and theoretically. Bragg diffraction intensity curves obtained from ordered latices subjected to an electric field shifted to longer and shorter wavelengths depending on the polarity of the electric field. The magnitude of the shift was nearly proportional to the electric field strength and depended on volume fraction and ion concentration of the suspending medium. The electric field effect was larger for latices of lower volume fraction and lower ion concentration. A semiquantitative explanation has been given for the observed phenomena.
Introduction Previously' we showed that the optical properties of an ordered latex suspension changed upon applying an external electric field. These phenomena were attributed to a change in latex structure. In the experiments reported in a previous paper,' latex was sandwiched between two electrodes over which an electric field was applied, one of the electrodes being transparent. Optical properties such as diffraction color, a dark ring? and a ring-shaped halo2 were observed through the transparent electrode. When the transparent electrode was electrically negative, the diffraction color changed to a color of longer wavelength, the dark ring became larger, and the halo smaller. These changes in optical manifestations suggested that the latex changed its structure by increasing the spacing between layers of particles near the transparent electrode. Reversing the polarity resulted in exactly the opposite. From those observations a possible structure was proposed' in which the particles are more densely packed near the positive than near the negative electrode. Crandall and Williams3 and Furusawa and Tomotsu4 studied the effect of a gravity field on the latex structure by optical methods. They reported that particle packing gradually changes in the direction of gravity, particles being most densely packed at the bottom of the container. Gravity acts in a manner very similar to an electric field. In a gravity field, the field acts on the mass of the particles while in an electric field the field acts on the electric charge of the particles. In both cases the particle distribution in the direction of the external field reaches an equilibrium by a balance of forces due to the external field and due to double layer interactions. Observations of diffraction color and dark ring radius cannot be expected to provide accurate information on electricity induced changes in structure. Instead, the right angle Bragg diffraction technique5 was employed in the present study, which has proven to be a powerful method for detecting small changes in latex structure. In this paper we will present results obtained with this technique and we will propose a model to explain the dependence of volume fraction and ion concentration on electrically induced changes in structure. Theoretical Considerations In the right angle Bragg diffraction technique one measures the diffraction intensity as a function of wavelength. The peak wavelength corresponds to the spacing between layers of particles. According to the proposed structure in a previous paper,' the spacing gradually changes between the two electrodes. Therefore ~~
(1) M. Tomita and T. G. M. van de Ven, J . Opr. SOC.Am. A , 1, (1984). (2) M. Tomita and T. G. M. van de Ven, J. Colloid Interface Sci., 92, (1983). (3) R. Crandall and R. Williams, Science, 198, 293 (1977). (4) K. Furusawa and N. Tomotsu, J. Colloid Interface Sci., 93, (1983). ( 5 ) M. Tomita and T. G. M. van de Ven, J. Colloid Interface Sci., 49, (1984).
317
the simple conventional Bragg equation cannot be applied. However, it is possible to calculate the diffraction intensity as a function of the wavelength by a two-step procedure, i.e., first by calculating the particle distribution in a given electric field and subsequently by calculating the diffraction intensity from the calculated particle distribution. ( a ) Calculation of Particle Distribution in an Electric Field. In order to calculate the particle distribution in an electric field, we must solve the convective diffusion equation, i.e.
where c is the particle concentration, t the time, x the distance from the positive electrode, D the diffusion coefficient, and u the velocity due to specific or external forces acting on a particle. Taking into account Fat, the force due to the external electric field, and FwI,the force due to colloidal interactions, we can express u in eq 1 as Fext + Fwl = f u
(2)
where f is the friction coefficient. Substituting eq 2 into eq 1 and integrating once yields, assuming steady-state conditions (&/at = 0)
_ _ -- Fat + Fw1 Df dc c dx
(3)
In general, both D and f are functions of c. Because D and f vary with c in an opposite sense: the product Dfdepends only weakly on c and, for the sake of simplicity, we assume that the StokesEinstein relation holds, i.e. Df = kT
(4)
where k is the Boltzmann constant and T is the absolute temperature. With eq 4 and a relation which is derived by assuming that particles are locally close-packed, i.e.
a and 4 being the particles radius and local volume fraction, respectively, equation 3 can be rewritten as
To make the calculation feasible, we assume that the van der Waals force is negligible and that double-layer repulsive forces are additive, i.e.
367
where F', is the x component of the repulsive force between the 504 374
0022-3654/85/2089-1291$01,50/0
(6) M. M. Kops-Werkhoven and H. H. Fijnaut, J . Chem. Phys., 77,2242 (1982).
0 1985 American Chemical Society
1292 The Journal of Physical Chemistry, Vol. 89, No. 7, 1985
g
Tomita and van de Ven 1.1
0 0
1.o
+ d+
00 0 0 0
1.o
+X L Figure 1. Schematic representation of layers of particles in an electric field. Layers are parallel to the electrode surface. d is the spacing between layers. The particles are assumed to be locally close packed. Fd
is the net colloidal force due to the asymmetrical distribution of particles and F,,,is the force due to the external electric field. These forces and a random force due to diffusion are balanced at equilibrium (eq 3). ith pair of particles. Because of the asymmetrical distribution of particles, the summation of the forces between a reference particle and all the neighboring particles is not zero, but yields a net force in the x direction (see Figure 1). The repulsive force between a pair of particles, a distance h apart, is (according to the classical DLVO theory) given by F=2~c(~~ae-~~
(7)
where c is the permittivity of the medium, (the zeta potential of will, a particle, and K the reciprocal double-layer thickness. FWI in general, depend on the local particle distribution, i.e., it depends on the local spacing and on local variations in the spacing. Since xh = 7 ( h / a )is a function of d/a, hence a function of 4, FWl can be rewritten as
FWl= 2 ~ e ( %A7,4,d4/dx*)
(8)
where x* = x / a , 7 = Ka, and f is a function of T, 4, and d4/dx*. Although eq 7 is valid only for low ( potentials, due to other simplifying assumptions, more accurate expressions for FW1 are not warranted. F,,, in eq 5 can be estimated from the theory of electrophoresis, taking into account double-layer relaxation:
F,,, = 6 ~ a d Eg ( 7 J
(9)
where E is the electric field strength and g(T,() is a function of 7 and {describing the relaxation effect.’ For example, for a 1-1 electrolyte g(7,() E 0.52, for 7 = 1 and ( = 100 mV and g(7,() E 0.64 for 7 = 1 and ( = 50 mV. Figure 1 schematically shows the particle distribution in an electric field and the directions of Fco, and Fext. Using eq 8 and 9, we can rewrite eq 5 in the following nondimensional form:
Equation 10 is subjected to the normalization condition f l L I a 6 ( x * ) dx* = @o L o
where dois the volume fraction of the latex and L the distance (7)P. H.Wiersema, A. L. Loeb, and J. Th. G . Overbeek, J. Colloid Interface Sei., 22, 78 (1966).
: 0.9
0
0.5
1.0 0
0.5
1.0
WL
Figure 2. Particle distribution between the electrodes calculated for various nondimensionalvariables, expressed as the local distance between layers vs. the position betwecn the electrodes. The values of Ccx,are shown on each curve. (a) C,, = 68.3,T = 0.674,bo = 0.120; (b) C,, = 68.3,T = 1.51,& = 0.120; (c) Cd 68.3,T 0.674,& = 0.058;(d) C-1 = 17.1, T = 0.674,$0 = 0.120.
between the electrodes. Hence 4(x*) depends on five nondimensional constants Cool,T , Ccxt,a / L , and 40. More generally, the same equation applies to suspensions subjected to external fields other than electrical ones. For example, if the external field is a gravitational or centrifugal field, the particle distribution would also be described by eq 10 but with C,,, now given by
c,
=
mga 4?ra4Apg =-
kT
3kT
(14)
m is the mass, g the acceleration due to gravity, and Ap the density difference between particles and medium. Equation 10 can be solved numerically by assuming a value of 4(0). The initial guess is adjusted after integration until eq 13 is satisified. In this way 4(0) can be found by trial and error, and one can obtain a solution for 4(x*). Figure 2 (a-d) shows particle distributions calculated for various conditions. The particle distributions are shown as variations in d* = d/do as a function of x / L , where d is the spacing between layers of particles at x (see Figure 1) and do is the spacing in the absence of the electric field. The relation between d* and 4 is d* = (40/4)1/3.In the calculation of FWI, all the interactions with the nearest neighbors as well as the contribution of particles in a few subsequent shells were taken into account. As can be seen from Figure 2 (a-d), as the value of Ccx,increases, the particle distribution deviates from d* = 1; at the positive electrode, i.e., at x / L = 0, d* is smaller, while near the negative electrode it is larger than unity. Curves in Figure 2, a and b, are calculated for the same values of CWl,a / L , and +o but for different values of T . It can be seen from these curves that the lower the T vaiue, the higher the value of Cextrequired to change the particle distribution. This is because double-layer interactions are more pronounced at low T values. In other words, latex with a low 7 value is softer than that with high T values. Curves in Figure 2, a and c, are calculated for the same values of Cool,a / L , and 7 but for different values of c$~.By comparing these curves it can be seen that the higher the dovalue, the higher the values of C,,, required to change the particle distribution. Again this is because of stronger double-layer interactions in the latex with high @o values. Curves in Figure 2d are for conditions as in Figure 2a with the exception of C, which for Figure 2d is four times smaller. Comparing Figure 2d with 2a, one can seen that similar changes in d* occur for values of C, that are also four times smaller. This is because the contribution of the diffusion term is very small for the conditions assumed in the calculation of Figure 2. The number of layers also changes when the particles are subjected to an electric field. Since at high values of C,,,, d*
The Journal of Physical Chemistry, Vol. 89, No. 7, 1985
Electrically Induced Changes in Latex Structure increases quickly near x / L = 1, the number of layers is decreased a t high values of C,,,. But this change is small for the range of C, values considered. For example, the number of layers changes from 6265 to 6246 in Figure 2a and from 4918 to 4903 in Figure 2c. ( b ) Calculation of Diffraction Intensity. The calculation of diffraction intensities from unequally spaced layers is not straightforward. Light is scattered at each layer and, after passing a layer, the resultant light has a slightly different phase from the incident light.*s9 This is because the resultant amplitude is the sum of the incident amplitude and the amplitude of forward scattered light whose phase is 90' behind that of the incident light. Also, the phase changes by traveling the distance between layers. Apart from the phase change, the amplitude decreases gradually due to primary extin~tion.~ To calculate the diffraction intensity, we have to sum all the scattered amplitudes, taking into account their phases, although layers deep in the suspension do not contribute very much to the diffraction intensity because of the decrease in amplitude. To perform the calculations, we assume multiple reflections and negligible absorption of light by the materials (polystyrene and water). Let TI and SI be the incident and scattered amplitude from the j t h layer and dJ the distance between layer j - 1 and j. We want to calculate (S1/T1)2as a function of wavelength. SI, for example, consists of a part of TI which is reflected at the upper surface of thejth layer together with a part of SI+, which is transmitted through the j t h layer from below. Similarly, TI+, consists of a part of TI which is transmitted through thejth layer from above together with a part of SJ+,which is reflected from the lower side of thejth layer. We only take into account multiple reflections of this kind and neglect higher-order multiple reflections. If 0 is the incident angle, the phase change due to traveling the spacing d is w
= (2n-/X) d sin B
(15)
where X is the wavelength of the light in the suspending medium. From the above consideration the following equations apply:9
0.3
I
1
I
I
2.7
2.0
Coxt
"
2.5
2.6
A,
1293
a
id,
Figure 3. Calculated diffraction intensity curves as a function of wavelength. IP1l*is the ratio of diffracted intensity to the incident intensity. XA/do is the nondimensionalized wavelength in air. Values of C,, are shown on are each curve; Cml = 68.3, T = 0.674. (a) bo = 0.120; (b) bo = 0.058. The triangles are calculated points and the lines smooth fits.
= PTJ + qrJ+lSJ+1
(16)
= qrJ+lTJ + PrJ+12sJ+1
(17)
that of the medium. The absolute value of the reflection coefficient, q, must be slightly smaller than unity. We simply assume that
= e-'Y+I
(18)
q=B+p
(27r/X)dJ
(19)
wJ =
where p and q are the reflection and transmission coefficients, respectively. Now we assume that
Sn = PTn
(20)
where n is the number of layers. Therefore Snand T, are the reflection and incident amplitudes above the last layer. This assumption is justified when there is a transparent cell wall under the last layer. By using eq 16-20 we obtain the following equation 2
' W J + l q rIPJ+l
P,=p+
2
1 - ~rn-j+1 'Pn-j+ 1
(21)
where PI = S J / T J .We can now calculate PI from eq 21. Note that P, = p from eq 20. p can be calculated from light scattering theory.8J0 For 0 = 90': 3 ~ 4 dm2 - 1 p=-i-m2+1
(22)
where m is the ratio of the refractive index of latex particles to (8) M. Tomita and T. G. M. van de Ven, J. Colloid Interface Sci., 100, 112 (1982). (9) R. W.James, 'The Optical Principles of the Diffraction of X-rays", Bell, London, 1962. (10) P. A. Hiltner and I. M. Krieger, J . Phys. Chem.,73, 2386 (1969)..
(23)
n-lp1/4 is known as the coefficient of primary extin~tion.~ Using eq 18, 19, and 21-24, we can calculate the diffraction intensity which is proportional to IP112as a function of wavelength. Figure 3, a and b, shows thus calculated diffraction intensity curves as a function of X,/do, using the particle distributions shown in Figure 2. A, is the wavelength in air and is A, = 1.33X. Figure 4, a and b, shows thus obtained peak wavelengths as a function of C,,,. As can be seen from Figure 4, a and b, the electrically induced shift in the peak wavelength depends on the nondimensional variables introduced above. These predictions will be compared with experimental data in the next section.
Experimental Section To measure the spacing between layers of particles near the wall, the' right angle diffraction technique5 was employed. Figure 5 shows schematic details of the cell used and the experimental system. A latex sample LX was placed in cell C consisting of two parallel gold-coated glass plates which were separated by a rubber spacer, RS, of thickness 1.2 mm. The area of the glass plate in m2. B is a beam contact with the latex sample was 2.5 X splitter cube. R is a plastic rod whose top and bottom surfaces were polished. The beam splitter cube, plastic rod, and an upper glass plate of the cell were immersed in oil to avoid unwanted reflections from their interfaces. The incident light came from a monochromator and the intensity of the diffracted light was
Tomita and van de Ven
1294 The Journal of Physical Chemistry, Vol. 89, No. 7, 1985 I
I
I
I
500
520
540
I I
I..
I
I
1
u)
% 3 %
2 -5~10~~ 0
I ‘
5~10.~
-
5
I I
I
10
zL c 4 ”
2.8 0
hA/do
480
AACnm)
0
640
460 AAtnm) .
680
Figure 6. Examples of experimentally observed diffraction curves. I is the diffraction intensity in arbitrary units and A, is the wavelength in air: (a) $o = 0.122, c = M; (b) @,I = 0.058, c = M. Values of the electric current are shown.
2.6
-10-2
620
10-2 Cext
Figure 4. Calculated peak wavelengths as a function of C,,,. (a)
@I =
0.120; o: c,, = 68.3, T = 0.674; 8: c,,, = 17.1, T = 0.674; m: c, = 68.3, T = 1.51; 0 : C,, = 17.1, T = 0.674. (b) @I = 0.058; A: ,C , = 68.3, 7 0.674 A: ,C , = 17.1, T = 0.674.
incident light I
v 1
---
,\
’
R
I-
PO
diffracted light
I I‘//////////////~///~/ I \/
1
C
RS L X Figure 5. Schematicrepresentation of experimental system: C, cell; LX, latex; RS, rubber spacer; R, plastic rod; B, beam splitter cube; V, voltmeter; A , ammeter; PO, dc power supply.
measured with a photomultiplier. The output of the photomultiplier was connected to the Y axis of an X-Y recorder. The X of the X-Y recorder was driven by a voltage which was proportional to the wavelength of the incident light. For details, see ref 5. In Figure 5 , A and V are an ammeter and a voltmeter, respectively, and PO is a dc power supply with a polarity switch providing voltages in the range of -5 to 5 V. In the experiments accurate control of the electric current was essential. With this system the diffraction intensity was measured as a function of the wavelength of the light in air. Latices used were polystyrene latices, which were kindly supplied by Professor S. Hachisu. The particle radius of the latex was a = 64 nm determined by electron microscopy; the standard deviation was 1.3 nm. In order to study both the dependence of volume fraction and ion concentration, three samples were prepared, Le., (a) I$ = 0.058, c = M, (b) q5 = 0.122, c = M,(c) I$ = 0.1 18, c = 5 X 10” M. To prepare latex samples of particular ion concentrations, samples were dialyzed against KCl solutions. Volume fractions were determined by dried weight prior to each experiment. Results and Discussion Figure 6, a and b, shows examples of diffraction curves obtained for various electric currents, i. In Figure 6 , a and b, Z is the intensity of the diffracted light (in arbitrary units). As can be seen from Figure 6 , a and b, peak wavelengths shift to longer or shorter wavelengths depending on the polarity of the electric field. The range of the shift is about *4-7%. Diffraction peaks dis-
Figure 7. Experimentally obtained relation between electric current, i, and voltage, u: 0,@Io = 0.122, c = M; O,@Io = 0.058, c = lo-’ M, a = 64 nm. appeared for large negative electric currents. No gas bubbles were observed after each experiment, although gas generation can be expected to occur at electrode surfaces. This is probably because the electric current was small. Since the latex samples were still iridescent after each experiment, the samples were not contaminated seriously by ions from the electrodes. The height of the diffraction curves were not very reproducible, but the peak wavelengths were reproducible and did not depend on history or experimental sequence. Since, in our experimental system, the base line of the diffraction curves was not constant, and because of the poor reproducibility in peak height, no useful conclusions regarding the diffraction intensity can be drawn. Figure 7 shows the relation between the electric current, i, and voltage, v, for two latex samples. The values plotted in Figure 7 were not equilibrium values but were obtained as follows: the electric current was increased from 0 to a certain negative value and the voltage was read when the diffraction curve reached its equilibrium position on the X-Y recorder. The voltage still changed, probably because the electrode reaction was very slow compared with the relaxation time of the particle distribution process. Then the current was increased to a higher negative value. After the diffraction curves were obtained for negative electric currents, the electric current was decreased slowly to 0, then increased to positive values. This voltage-current relation will depend on the thickness of the spacer and probably also on the the material used in the coating at the glass. It can be seen from Figure 7 that the voltage required to induce changes in structure was less than *1,8 V. Figure 8 shows an example of transient behavior of diffraction curves. The experimental points in Figure 8 were obtained when the electric current was suddenly changed
The Journal of Physical Chemistry, Vol. 89, No. 7, 1985 1295
Electrically Induced Changes in Latex Structure
535
1 540
5
k
a
x
500
520
0
100
200
300
Time
I
400
(SI
Figure 8. An example of experimentally obtained transient peak shift. Values are obtained when electric current was changed from i = -1 pA to i = -2 pA. &, = 0.118, c = 5 X 10” M, and a = 64 nm.
from -1 to -2 PA, corresponding to the moment t = 0. It can be seen from Figure 8 that the peak wavelength shifts to a larger value as time goes on and asymptotically reaches its equilibrium position. From various experiments we performed we can conclude that polarization processes occurring at the electrodes strongly affect the relaxation time. This transient behavior warrants further study. Figure 9, a and b, shows comparisons of experimentally obtained peak wavelengths and theoretically predicted ones as a function of electric field strength. The electric field strength for the experimental points was estimated from the equation
E = i/AX,
I
I
I
-100
0
100
I
I
I
where i is the electric current, A is the area of the electrodes (2.5 X lo4 mZ),and X,is the specific conductivity of the suspending medium. Equation 25 assumes that electric conductivity of the latex suspension is the same as that of the suspending medium, and E is uniform throughout the suspension. The values of A, are 1.5 X lo4 and 7.5 X lo4 ohm-’ m-’ for c = 10” and 5 X 10” M KCl solutions, respectively. The peak wavelengths were read from diffraction curves, examples of which are shown in Figure 6 . The calculated peak wavelengths are obtained from the values of AA/do in Figure 4. The electric field strengths for theoretical points in Figure 4 are obtained by substituting the appropriate values of a, L, t, {, and T into eq 12. For the ( potential, a range from -50 to -100 mV was taken, which is typical for the latex used in our experiment^.^ Results thus calculated are plotted in Figure 9, a and b. The ranges between the two curves corresponding to ( = -50 and -100 mV are shaded. It can be seen from Figure 9 a and b, that the experimental data and the theoretical curves follow qualitatively the same trends, Le., the electric field effect is larger for lower ion concentrations and for lower volume fractions. However, quantitatively, the agreement between experimental and theoretical points is not very good, especially for high volume fractions (Figure 9 a). The reasons for the discrepancy must be found in our simplifying assumptions. We assumed pairwise additivity of the colloidal forces. This assumption may be too simple, especially for high volume fraction latex. Equation 25 may also be too simple. The electric conductivity of a suspension is usually lower than that of the suspending medium; therefore the actual electric field strength might be larger than the one estimated. This effect is also expected to be larger for high volume fractions. Equation 25 also assumes that the electric field is uniform throughout the suspension. Since the specific conductivity is affected by particle concentration, but the particle distribution is not uniform, the specific conductivity is not constant but a function of x. Hence the electric field is not uniform throughout the latex. The assumption made in the calculation of the diffraction intensity may also be too simple. The calculated wavelength from the calculated distribution of particles corresponds to the spacing at the electrodes. This means that the extinction is very quick and the light does not penetrate deeply
I
680 CI
i
-
Y
a
x
640
-10
(25)
I
0
10
E CWm) Figure 9. Experimentally obtained peak wavelength (symbols) and theoretically calculated ranges of peak wavelength (shaded regions) as a function of electric field strength, E: (a) 0 $o = 0.122, c = io-’ M, a = 64 nm; m, 6 = 0.1 18, c = 5 X lo-’ M, a = 64 nm; (b) A, &o = 0.058, c = 10” M, a = 64 nm. The shaded region corresponds to the range calculated for t = -50 to -100 mV.
into the suspension. The extinction assumed in eq 23 and 24 may be too large. Although our theory cannot explain experimental data precisely, it does explain both the ion concentration and the volume fraction dependence in a semiquantitative manner. The theory is most adequate for low volume fractions.
Concluding Remarks When ordered negatively charged latex is subjected to an electric field, the spacing between layers of particles parallel to the electrode increases at the negative electrode and decreases at the positive electrode.’ In the present paper, the electric field effect was studied quantitatively by using the right angle Bragg diffraction technique. With this technique the changes in spacing can be measured from a shift in the wavelength of the diffraction peak. It was found that the higher the electric field strength, the larger the shift in peak wavelength. The relation between peak wavelength and the electric field strength was found to be nearly linear. This electric field effect depends on the ion concentration of the suspending medium and on the volume fraction of the suspension. It was concluded that this electric field effect is larger for lower ion concentrations and for lower volume fractions. The effect of volume fraction and ion concentration was compared with theory and it was found that the theory can explain the experimental observations in a semiquantitative manner. The discrepancy between theoretical and experimental values was larger at higher volume fractions. In the theory the particle distribution was calculated by taking into account the electrical force due to the external electric field, the electrical repulsive force between particles due to double-layer interactions, and a random force due to diffraction of particles. The diffraction intensity was calculated from calculated particle distributions by taking into account multiple reflections. These calculations showed that, even without
1296
J. Phys. Chem. 1985, 89, 1296-1304 theories of colloidal interactions in concentrated dispersions.
absorption of light in the latex, the diffracted light has a peak wavelength that corresponds to the spacing near the electrode. This is because the particlelight interaction is so strong that the amplitude of the incident light diminishes rapidly in the latex and particles far from the viewing side do not contribute to the diffraction intensity. In principle, experiments of the kind described in this chapter could yield important information to test various
Acknowledgment. The authors thank Professor S. G. Mason for valuable encollragement. They ah0 are indebted to Professor s. Hachisu for kindly supplying latex samples and for valuable discussions. Registry No. Polystyrene (homopolymer), 9003-53-6.
A Lewis Basicity Scale for Nonprotogenlc Solvents: Enthalpies of Complex Formation with Boron Trlfluorlde In Dlchloromethane Pierre-Charles Maria and Jean-Franqois Gal* Laboratoire de Chimie Physique Organique, Universite de Nice, Parc Valrose-06034 Nice Cedex. France (Received: July 30, 1984; In Final Form: November 2. 1984)
A solvent Lewis basicity scale was established for 75 nonprotogenic solvents by measuring calorimetrically their enthalpies of complexation with boron trifluoride ( A P B F , ) in dichloromethane. Absence of side reactions was verified by calorimetry, spectroscopy, and by checking the stoichiometry of the adducts. Some enthalpies were also measured in nitrobenzene, showing that dichloromethanedoes not induee nonregular effects. Drawbacks of the Gutmann's DN scale are emphasized. Relationships between various Lewis and hydrogen bond basicity scales and - A W B F 1 are examined. A plot of Kamlet-Taft's fl vs. -AHoBF, shows a typical family dependence. A significant multilinear correlation of -AHoBFlagainst complexation enthalpies toward p-fluorophenol and iodine gives evidence that BF3, though stronger, exhibits an electrostatic-covalent acceptor character median between those of the two acids chosen as references. Attention is drawn to the BF, complexation sensitivity to steric hindrance. The - A P B F 3 scale appears as a useful tool for the rationalization of the Gibbs energies of transfer of alkali metal cations which depend mainly on the solvent Lewis basicity. In the correlation analysis of solvent effects the authors suggest the use of basicity parameters representative of the solute-solvent interaction under scrutiny.
Introduction Basicity is an essential solvent property often used to account for the influence of the solvent on chemical phenomena.' Basic solvents are usually classified as HBA (hydrogen bond acceptors)2 or EPD (electron pair donor^).^ Gutmann4 uses the term "donicity" as a measure of the ability to donate an electron pair and he has proposed the so-called donor number (DN) to express, in at least a semiquantitative manner, the donor strength of a solvent. Despite vigorous criticisms of either the concepts or the experimental values: DN is one of the most widely used empirical parameter of solvent basicity, probably by reason of the wide publicity made by the author.' In his excellent review: Jensen brings out the prominent features of the D N concept. We discuss below some failings of the DN scale and we propose another empirical parameter of solvent Lewis basicity defined as the enthalpy change for the reaction between (1) Benoit, R. L.; Louis, C. "The Chemistry of Nonaqueous Solvents", Lagowski, J. J., Ed.; Academic Press: New York, 1978; Vol. 5 , pp 63-1 19. (2) Kamlet, M.J.; Taft, R.W. J . Am. Chem. SOC.1976, 98, 377-383. (3) (a) Reichardt, C. 'Solvent Effects in Organic Chemistry"; Verlag Chemic: Weinheim, 1979. (b) Reichardt, C. Angew. Chem., I n f . Ed. Engl. 1979, 18, 98-1 10. (4) (a) Gutmann, V. 'Coordination Chemistry in Non Aqueous Solutions"; Springer Verlag: Wien, 1968. (b) Gutmann, V. 'The Donor-Acceptor Approach to Molecular Interactions"; Plenum Press: New York, 1978. ( 5 ) (a) Drago, R.S.Coord. Chem. Rev. 1980,33, 251-277. (b) Drago, R. S . Pure Appl. Chem. 1980, 52, 2261-2274. (6) (a) Taft, R. W.; Pienta, N. J.; Kamlet, M. J.; Arnett, E. M. J. Org. Chem. 1981,46,661-667. (b) Olofsson, G.; Olofsson, I. J. Am. Chem. SOC. 1973.95.7231-7233. (c) Lim, Y. Y.; Drago, R.S. Inorg. Chem. 1972,11, 202-204. (7) (a) Mayer, U.; Gutmann, V. Struct. Bonding (Berlin) 1972, 12, 113-140. (b) Gutmann, V.; Schmid, R. Coord. Chem. Rev. 1974, 12, 263-293. (c) Gutmann, V . Coord. Chem. Rev. 1976, 18, 225-255. (d) Gutmann, V. Electrochim. Acta 1976, 21, 661-670. (e) Gutmann, V. Chemfech1977, 7, 255-263. (0 Gutmann, V.; Resch, G. Comments Inorg. Chem. 1982, 1 , 265-218. (8) Jensen, W. B. Chem. Rev. 1978, 78, 1-22.
0022-3654/85/2089-1296$01 S O / O
gaseous boron trifluoride and the basic organic molecules, including those considered as "solvents", diluted in dichloromethane sol~tion:~ BF,(g)
+ :B(soln)
-
B:BF,(soln)
(1)
Experimental Section Chemicals. Boron trifluoride (Matheson, purity > 99.5%) is purified by slow passage through a trap cooled to about 160 K and then by a freeze-pumpthaw cycle. Every BF, transfer is conducted in a high vacuum apparatus, without contact with the atmosphere, and the gas is stored in a 1-L glass bulb linked to the vacuum line and closed by mean of a mercury valve. Dichloromethane (Merck Uvasol) is stored over a 4-A molecular sieve (Merck) in dark bottles under a blanket of ultrapure argon and used without further purification.1° The water content was measured by Karl Fisher titration and was found to be less than 20 ppm (by weight). Nitrobenzene (Fluka puriss, 99.5% min) was treated in the same way as dichloromethane. The majority of the compounds studied were of analytical grade (Aldrich or Ega, Merck, Fluka). All were analyzed by gas chromatography (GC). Those of purity greater than 99% were only dried by an appropriate reagent. The others were purified by distillation (on a spinning band column) or by preparative GC. Reagent grade sulfolane (Merck, 97%) was carefully crystallized three times and finally checked by G C on O.V. 17 (final purity > 99.9%). (9) (a) Preliminary communications: Elegant, L.; Fratini, G.; Gal, J. F.; Maria, P. C. Presented at the 'OnziEmes Journdes de Calorimttrie et d'Analyse Thermique", Barcelona, Spain, June 4-6, 1980 (Vol. 11, Abstract 3/21). (b) Maria, P. C.; Gal, J. F.; Elegant, L.; Azzaro, M. Presented at the 'Second Euchem Conference on Correlation Analysis in Organic Chemistry", Hull, England, July 18-23, 1982 (Abstracts p 39). (10) Spectroscopic grade dichloromethane is stabilized by 20 ppm of 2methyl-2-butene. Reagent grade CHIClz purified by treatment with concentrated H2SO4,I1distilled from P2O5l2and also stored in dark bottles over a 4 A molecular sieve gave identical calorimetric and spectroscopic results.
0 1985 American Chemical Society