4725
J. Phys. Chem. 1005, 87, 4725-4730
Double Potential Step Chronocoulometric Study of Adsorption of Trihalomercurate Ions on Mercury from Acetonitrile Marek Wojclechowskl,+ John J. O'Dea, and Janet Osteryoung State University of New York at Buffab, Deparhnent of Chemistry, Buffalo. New York 14214 (Received: September 7, 1982; I n Final Form: January 4, 1983)
The adsorption of HgX3- (X = C1, Br, or I) ions on mercury from acetonitrile solutions 0.1 M in tetraethylammonium perchlorate has been studied by double potential step chronocoulometry. The surface excesses of HgX3- ions correspond to the formation of a monolayer of HgClf or HgBr3- in a planar position and of Hg13in a mostly perpendicular position. The diffusion coefficients of halide and trihalomercurate ions calculated from chronocoulometric data are compared with the values obtained by polarographic methods. The values obtained from chronocoulometry for C1-, Br-, I-, HgC13-, HgBr3-, and HgIf are 2.6, 2.9, 3.6, 1.6, 1.7, and 1.9 x cm2/s, respectively.
Introduction The anodic oxidation of mercury in the presence of ligands which stabilize the oxidized form of mercury has been studied extensively. Most recently we have examined the anodic oxidation of mercury in acetonitrile solutions in the presence of chloride, bromide, and iodide.12 In the concentration range 0.01-1 mM halide, two anodic waves appear, the first of which is diffusion controlled in halide and due to formation of HgX,b2)-. At potentials sufficiently negative of the half-wave potential the higher complex, HgX?-, is formed, while on the diffusion plateau the product is HgXC. These conclusions are supported by voltammetry, coulometry, and conductance experiments and are in general accord with the findings of other investigators in similar systems.*5 The wave appearing at more negative potentials is complicated to some extent by adsorption of the product, the trihalomercurate ion:
The adsorption of product has been shown by reverse pulse polarography in solutions of X-, by normal pulse polarography in solutions of HgXC, and by electrocapillary experiments with X- and HgXC. These techniques provide excellent qualitative evidence for adsorption, but this evidence cannot be interpreted quantitatively. The diffusion current constant in the normal pulse and dc polarographic mode exhibits no trend over a range of concentrations (0.01-1 mM) and times (1-500 ms) the charge for which corresponds, as shown below, to monolayer coverages in the range 0.2-500%. This confirms that the composition of the adsorbed material is the same as that of the diffusing product, HgX3-. In this paper we present a quantitative study of adsorption of trihalomercurate ions on mercury in acetonitrile by means of double potential step chronocoulometry.6 Christie, Anson, and Osteryo~ng'~~ have shown that double potential step chronocoulometry can be applied successfully to the study of adsorption. Because the mathematical description is relatively model independent, the adsorption of reactant is the most clear-cut case for an investigation by this method. Assuming that the time necessary for reduction of adsorbed reactant is negligible compared with the total time scale of the experiment and that there are no slow chemical or electrochemical steps, 'Permanent address: Department of Chemistry, University of Warsaw, Pasteura 1,02-093Warszawa, Poland. 0022-365418312087-4725$07.50/0
the charge-time relationships of the forward ( t < 7) and reverse ( t > T ) steps are described by the following equations: Q(t7) - Q ( 7 ) = ~ ~ F A C , ~ ' ( D / T + )'/~ nFAI'[l - (2/7r) sin-' ( ~ / t ) ' + / ~Qdl ] (3) where 7 is the forward step duration, A is the electrode area, C,, D, and r are the bulk concentration, the diffusion coefficient, and the surface excess of reactant, respectively, Qdl is the charge required to charge the double layer, and 6' = ( t - 7)'12 71/2Analysis of the plot of Q(t 3 or t > 7 + 3 ms. As is common in chronocoulometric experiments, the potentiostat current limits at very short times. Under the most demanding conditions reported here, this condition persists only to 0.3 ms; at longer times the circuit is fully in control. This was checked carefully not only by observing directly the charge-time transient on an oscilloscope but also by using the well-studied test system of Cd(I1) in thiocyanate media and obtaining good agreement with previous reports.8 Thus, the unusual Q-t behavior displayed in, e.g., Figure 4, curves 14,cannot be attributed to instrumental artifact. This point can be amplified by considering the chronocoulometric curves of Figure 5 . These data were obtained under the same conditions as those of Figure 4, curves 4 and 4a, except that the step potential, El, was varied as noted. For E , = -1.0 V (as in Figure 4) the plot of Q vs. t'12 is strictly linear for t > 3 ms, but for less negative values of potential the curves are nonlinear to times as long as 30 ms. Chronoamperometric curves obtained under the same experimental conditions are completely consistent with the chronocoulometric curves of Figure 5. Furthermore, the current-time data at fixed potential are quantitatively the same as data generated by carrying out normal pulse voltammetry at a variety of pulse widths. For the conditions of Figure 5 the normal pulse voltammogram with pulse width of 5 ms has a "maximumn about 6 times the magnitude of the diffusion-controlled limiting current and about 300 mV wide at half-height. The solutions are not buffered in halide. Therefore, at short times halide produced by reduction of adsorbed material causes a shift in the solution equilibrium; although the predominant species in bulk solution is HgX,, near the electrode HgXd2-predominates. But a simple diffusion layer argument predicts shifts in half-wave potential due to this complication which are 10 times smaller than the peak widths observed. This type of maximum has been described in detail experimentally for chloride' and has been reported for the similar case of ethy1enediaminetetraacetate.l' Theoretical treatments of the effects of adsorption of reactants on normal pulse voltammograms have not predicted such very large
(10) Ridgeway, T. H.; Van Duyne,R. P.; Reilley, C. N. J.ElectroanaL Chem. 1972, 34, 267-82.
67-74.
3.0
1
-2.01
0
l
2
3
SORTLTIMEI
4
5
OR TWETEI,
6
?
6
3
SORTINS1
Figure 4. Q vs. t'/' (curves 1-6) and Q, vs. 8 (curves la-6a) for data obtained in solutions Containing 0.06 (curves 2, 2a, 4, 4a, 6, and sa) or 0.3 (curves 1, l a , 3, 3a, 5, and 5a) mM trihalomercurate ions. Curves 1, l a , 2, and 2a: HgC13-. Curves 3, 3a, 4, and 4a: HgBr,-. Curves 5, 5a, 6, and 6a: HgI,-. 7 = 70 ms. Potential parameters are as indicated in the Experimental Section.
(11) Stojek, Z.; Osteryoung, Janet.
J.Electroanal. Chem. 1981,127,
4720
The Journal of Physical Chemistty, Vol. 87,No. 23, 1983
Wojciechowski et at.
TABLE I : Values of nFr for Trihalomercurate Ions from Double Potential Step Chronocoulometry
I____
Cl -
Br-
I-
0.181 0.885 0.173 0.891 0.175 0.853
0.061 0.300 0.053 0.286 0.056 0.276
(0.82)f (0.83) (0.97) (0.96) (0.99) (0.97)
47.2 54.4 52.2 55.0 58.9 62.2
t -i
?-
t i 2
1.7 (6)e 1.1( 6 ) 1.1( 6 ) 2.2 ( 6 ) 1.1( 6 ) 3.3 (6)
44.4 52.8 49.4 53.9 56.1 61.7
i i
i ?-
i i
1.1( 4 ) 1.1( 4 ) 0.0 (6) 0.6 ( 6 ) 0.6 ( 6 ) 0.6 ( 6 )
47.8 I 54.4 i 52.8 57.8 i 57.2 F 66.1 2
*
1.7 ( 3 ) 0.6 (6) 1.1(3) 1.7 ( 6 ) 3.3 (3) 3.9 ( 6 )
46.5 53.9 51.5 55.6 57.4 63.3
a Solution 1. Solution 2. From Q ( f ~ 7vs. ) t"'. From Q ( ~ > T ) - Q ( 7 ) vs. 0 . e Mean value i SD (number of experiments). Formal concentration (CY,), where a , , the fraction of Hg(I1) present as HgX;, is calculated from data of ref 20.
peaks.12J3 This phenomenon may cause severe interferences in analytical applications. When understood, it might prove useful in studies of adsorption. The phenomena described above must be due to slow processes associated with the reduction of adsorbed species. It should be emphasized that there is no evidence for adsorption of halide at E,, for X- as described in the Experimental Section. Thus, considerable reorganization at the molecular level may be involved in the reduction process. As there is a substantial literature on adsorption of halides on mercury electrodes from organic solvents, this point is worth some elaboration. First, it should be noted that the experimental conditions here are unfavorable for halide adsorption. The initial potential in halide solutions is -800 (for C1-) or -1000 (Br-, I-) mV, the maximum concentration is 0.9 mM, and the concentration of perchlorate is 0.1 M. Thus, the potential is sufficiently negative, and the absolute and relative concentrations of halide are so small that one might not expect halide to be adsorbed. The second point is that previous reports of adsorption of halide form nonaqueous s ~ l u t i o n ' ~ -are '~ based on decrease of the differential double-layer capacity with increasing halide concentration, or with change in anion from, e.g., fluoride to a heavier halide, which appears to occur at potentia& at which mercury is oxidized to form halide complexes of Hg(I1). Matsui et al.17have discussed the adsorption of halides and mercuric halides in dimethylformamide in relation to voltammetric measurements, but their evidence is only suggestive. We have shown' that the potential region of suppressed surface tension in solutions of 0.1 M TEAP in acetonitrile containing 0.2 m M C1- or HgC1,- coincides with the potential region of stability of HgC13-, the range of the first anodic wave. Thus, significant adsorption of halides from nonaqueous solvents may occur only with significant faradaic reaction. In the present case, we find no experimental evidence for adsorption at the initial potential in solutions of X- (the final potential in solutions of HgX,-). This leads us to speculate that the kinetic phenomena that we see are due to slow desorption and solvation of halide ions produced by cathodic reduction of HgXT (ads). The surface excesses of HgC13-, HgBr,, and HgI,- ions are presented in Table I. Consistency of the results was checked three independent ways using experimental data obtained in two different solutions. The solutions con(12)Flanagan, J. B.; Takahashi, K.; Anson, F. C. J. Electroanal. Chem. 1977,85,257-66. (13)Van Leeuwen, H.P.:Sluvters-Rehbach, M.: Holub. K. J. Electroanal. Chem. 1982,135,13-24; (14)Payne, R. In 'Advances in Electrochemistry and Electrochemical Engineering";Delahay, P., Ed.;Wiley: New York, 1970;Vol. 7.pp 1-76. (15)Minc, S.;Jastrzebska, J.; Brzostowska, M. J.Electrochem. SOC. 1961, 108, 1160-3. (16)Levi, M. D.; Shlepakov, A. V.;Damaskin, B. B.; Bagotskaya, I. A. J. Electrochem. SOC.1982,138, 1-27. (17)Mataui, Y.;Kurosaki, Y.; Date, Y. Bull. Chem. Soc. Jpn. 1970,43, 1707-14.
TABLE 11: Potential Dependence of nFr for Trichloromercurate Ions potential, mV
E" E, CCl-ra
E, E"
-800 -200
CHgC1,
-800 --250
--lo00 -250
-800 -300
b -9
mM
mM
0.181 0.885
0.061 0.300
nFr ,C p C/cm 48.4 58.9
42.7 49.9
41.9 48.4
38.6 43.2
Solution 1. Solution 2; formal concentration. Mean values obtained as in Table I (last column).
tained the same concentration of halide ions, but only one contained mercury(I1) at the concentration of Cx-/3. In the first case we obtain n F r from the relation n F A r = OQ(HgX,-) - "Q(X-) (4) where "Q(HgX,-) is the forward intercept in the presence of HgX3- and "Q(X-) is the forward intercept in the presence of X-. This is the simplest and most reliable method. The two forward intercepts are well-defined experimentally, and the conditions for the two experiments are identical except for the presence or absence of Hg(I1) and the direction of potential change. Thus, the intercept "Q(X-) is the best approximation of the charge consumed for charging of the double layer in the experiment with HgX,-. Under the range of conditions employed here the total charge on the forward step in the presence of Xvaries from about O.6nFAr to 10 nFAr. Plots of Q vs. t1Iz are linear with no systematic trend in the slope. Thus, it appears, as confirmed below, that the double-layer charge does not depend on the coverage of HgX,-. This is not what one might expect, but it is in accord with experimental evidence on other systems.s A second method was to obtain n F r from the relation nFAr = "Q(HgX,-) - Qd{ (5) where Qd' is the charge required to charge the double layer for the conditions under which the intercept "Q(HgX,-) is determined, but in the absence of HgX3-. Double potential step chronocoulometric data for these experiments may also be analyzed by using eq 2 and 3 to give the value Q d l of the double-layer charge in the presence of HgX3-. The values of Qdl obtained in this way agreed with the corresponding values of Qd{ obtained in the supporting electrolyte solution within experimental error. The third way used to evaluate nt;T for HgX,- ions was to subtract "Q, values (from the plot of Q, vs. 6) for X- and HgX3-. This analysis was carried out only for the experiments where Q ( T ) >> nFAr = "Q,(X-) - "Q,(HgX,-). Values of surface excess obtained from reverse intercepts agree reasonably well with those obtained from forward intercepts. For these conditions only a small fraction of charge goes to reaction of absorbed material. This fraction
Adsorption of Trihalomercurate Ions on Mercury
The Journal of Physical Chem/stry, Vol. 87, No. 23, 1983 4729
TABLE 111: Surface Excess and Surface Area of Trihalomercurate Ions Adsorbed on Mercury from Acetonitrile r ,a mol cm-z S,b A Z calcd calcd
ion
HaC1; "
HgBr,Hg1,-
exptl
planarc
perpendiculard
2.4e 2.Bf 2.7e 2.gf 3.0e 3.3f
2.4. 3.0
5.3
2.1, 2.6
4.6
1.8, 2.2
3.1
exptl 6ge 59f 62e 57f 55e 50f
TABLE IV: Diffusion Coefficients cm2 s-l) of Halide Ions in 0.1 M TEAP/Acetonitrile Solution at 25.0 "C normal *Dulse -
double potential step ion
perpendicuplanarc lard
55. 70
31
chronocoulometry" SMDE
c1Br-
I-
a
Mean value
ref 1 and 2.
'
63, I 9
36
77, 95
45
a Surface excess. Surface area per ion. Numbers correspond respectively to a triangular or circular area occupied by the ion; 5 ' ~ - = (3)(.3''z)(3R + r)2/4;Scire 3 n ( r + 2R)'; R and r are the ionic radii of X- and Hgz+ions, respectively." S = ((2)(3llZ)+ 4)R2 + (2)(3'12)Rr,with the assumption of symmetrical structure. e From the data for 0.18 mM X - and for 0.06 mM HgX,- (Table I).
From the data for 0.9 mM X - and 0.3 mM HgX, (Table I).
is given by T ' / ~ I ' / ~ C ~ ( D and ~ )is' /less ~ than 0.3 for the data that wc analyzed this way. The complexity of the system notwithstanding, the linearity of Q, with 6 suggests this intuitive approach, and the reasonable results imply a more simple mathematical description than might otherwise be assumed. Standard deviations of the results, listed in Table I, affirm the good precision of the experiments. For larger concentrations of HgX, (Table I) the values of nI;T are slightly larger, but in this range there is not a strong dependence of n F r on concentration. The results listed in Table I1 show some potential dependence on the results obtained for trichloromercurate ions. Change of the more negative potential (on the plateau of the HgC13- reduction wave) does not affect the intercept "8,as long as the potential is sufficiently negative to achieve diffusion control. However, the more positive potential influences the intercept. This may be due to the sloped base line of the HgC1,- wave. When Eo < -0.25 V, traces of reduction of HgC1,- occur and the value of "Q decreases. For Eo > -0.25 V oxidation of Hg begins and the intercept is thus higher than for E,, = -0.25 V, where no current flows. Because of this it is impossible to decide whether the potential does or does not have an influence on the surface excess of HgClC ions on mercury. On the basis of the experimental results presented in Table I we have calculated the surface excesses (I') and effective surface areas (S) of the trihalomercurate ions at various concentrations. The surface area of the electrode occupied by one ion was estimated by S = l/NAr,where NA is Avogadro's number and r the surface excess in mol/A2. In Table I11 the experimental values are compared with the results of calculations based on assumptions regarding the shape and the position of the adsorbed ions on the electrode surface. We assume the ions are planar with equal X-Hg-X bond angles. In the "planar" configuration the plane of the molecule is parallel to the surface, and the specific area is based on closest packing of triangles or circles. In the "perpendicular" configuration the plane of the molecule is perpendicular to the surface, and the specific area is based on closet packing of rectangles. The experimentally determined surface excesses of the HgX3- ions are larger for the heavier halides. The calculated values of I' change in the opposite direction because the dimensions of trihalomercurate ions increase in the following order: HgC13-, HgBr3-, and Hg13-. The
voltammetryb
*
DME
2.6 f 0.3 ( 4 ) 2.4 2.3 2.9 .t 0.4 ( 4 ) 2.6 2.6 3.6 f 0.4 ( 4 ) 3.1 2.9 SD (number of experiments). From
TABLE V : Diffusion Coefficients cm2s - l ) of Trihalomercurate Ions in 0.1 M TEAP/Acetonitrile Solution at 25.0 "C
ion
HgC1,HgBr; HgI, a Mean value ref 1 and 2.
staircase (tastl voltammetry6 double potential step chronocoulometrya SMDE DME
*
1.6 i 0.2 ( 9 ) 1.9 2.0 1.7 .t 0.1 ( 1 2 ) 2.3 2.2 2.7 2.4 1.9 i. 0.1 ( 1 2 ) SD (number of experiments). From
experimental value of r for HgC1, agrees well with the value of r calculated assuming planar position of the adsorbed ions. The surface excess of HgBr3- ions is slightly higher than calculated for the planar position, while in the case of HgI, it is closer to the value with the perpendicular assumption. From the data shown in Table I11 it seems that trihalomercurate ions adsorb on mercury from acetonitrile solutions forming a monolayer. The HgClL ions within the monolayer are in the planar position. This is also true for HgBr,- ions, considering the precision of the experiments. The surface excess of HgI,- indicates a monolayer with the ions mostly in the perpendicular position. Of course, the value of the surface excess for Hg1,- could also be interpreted as being due to more than one layer in the planar position. But the surface excess in each case depends so little on concentration that we favor the hypothesis of monolayer formation. This description of the adsorbate as a simple monolayer of trihalomercurate ions is more attractive by virtues of coherence and economy than alternatives involving adsorbates with the required composition but different molecular formulation (e.g., HgX, X-). Diffusion Coefficients. Among other electrochemical methods double potential step chronocoulometry, as well as a double potential step chronoamperometry, appears to be a simple and rapid technique for evaluation of diffusion coefficients. It is particularly convenient and accurate when used with a computer-controlled system. The analysis of chronocoulometric data through a plot of Q m. t1izgives direct information regarding any deviation from diffusion transport of the electroactive species. According to eq 2 the slope (S,) of the Q ( ~ < T ) vs. t l / , line for a diffusion-controlled process can be expressed by the relation
+
S1 = 2 n F A C b ( D / ~ ) 1 / 2
From this the diffusion coefficient can be calculated. Tables IV and V contain the diffusion coefficientsof halide and trihalomercurate ions calculated from the double potential step chronocoulometric experiments described above. (Strictly speaking the experimental values for the Hg(I1) complexes represent averages since a3is slightly less than unity; cf. Table I.) In order to check the results we determined diffusion coefficientsby means of normal pulse
J. Phys. Chem. i003* 87,4730-4737
4730
(for X-) or staircase (for HgX3-) voltammetry using the same electrode (SMDE, dropping mode) and solutions, and using a conventional DME. The third columns in Tables IV and V contain values calculated from data included in the previous paper' and data which will be described elsewheree2 Since the normal pulse reduction wave of HgX3- does not have a well-developed limiting plateau (rather a maximum caused by slow processes associated with the reduction of adsorbed species), we had to utilize staircase voltammetry as the reference method for determination of D for the HgX3- ions. Except for the value of D for iodide ions obtained from the polarographic data (DME) and the value of D for triiodomercurate ions from the chronocoulometric data, all results are in good agreement. Notice that the changes of D within the halide ions and within the trihalomercurate ions are opposite to those which might be expected from the dimensions of the ions. The smallest halide ion, C1-, and the smallest trihalomercurate ion, HgC13-, have the smallest diffusion coefficients. A reasonable explanation is that the smaller ions, having a larger charge density, are more efficiently solvated by polar molecules of the solvent.
The diffusion coefficient of iodide ions in acetonitrile was determined by Macagano et a1.I8 using a platinum rotating disk electrode and by DePauli et al.I9 by means of chronopotentiometry. In both cases high concentrations of iodide (8-18 mM) were used and the supporting electrolyte contained Li+ and Na+ ions. In the previous paper' we reported evidence for the existence of Li+Cl- ion pairs in acetonitrile. The formation of ion pairs between I- and Li+ or Na+ might be responsible for lower (1.68 X cm2 (ref 18))values of D obtained in solutions containing lithium or sodium salts. Registry No. HgCl;, 14988-07-9; HgBr3-,21388-05-6; HgIi, 19964-11-5; C1-, 16887-00-6; Br-, 24959-67-9; I-, 20461-54-5; Hg, 7439-97-6. (18)Macagano, V.A.;Giordano, M. C.; Arvia, A. J. Electrochim. Acta . - - ,335-57. --(19)DePauli, C.; Iwasita, T.; Giordano, M. C. J.Electroanal. Chem. 1973,45, 233-45. (20) Coetzee, J. F.; Campion, J. J.; Liberman, D. R. Anal. Chen. 1973, 45.343-7. (21)'Handbook of Chemistry and Physics", 61st ed.; CRC Press: Boca Raton, FL, 1980. 1969.14. ~. ..
~
Effect of Chain Length on Mesomorphfsm of Steroid Esters of 4-(4-Alkylphenyl-X)benzoic Acids with X = CO, 0, S, and CH2 Mltsuhlro Koden, Tadao Yagyu, Shunsuke Takenaka, and Shlgekazu Kusabayashl Department of Applled Chemistfy, FacuRy of Engineerlng, Osaka University, Suita, Osaka 565, Japan (Received: October 20, 1982: In Final Form: March 8, 1983)
To examine the effect of bent shapes on mesomorphic properties a homologous series of steroid esters have been prepared: H(CHz),-4-C6H4-X-4-C6H4COOR, X = c o , 0, s, CHz,R = cholesteryl, P-sitosteryl, cholestanyl, stigmasteryl, ergosteryl; n = 0-15. The chain elongation results in an increase in not only the molecular length but also the breadth due to the angular linkage, X. The steroid portions are of primary importance for the mesomorphic properties of the present series, and the thermal stability of the mesophases is strongly dependent on the mesogenic power of the aryl portions, where the effective order is CO > 0 > S > CHz. The transition enthalpies and entropies for the smectic A-cholesteric and cholesteric-isotropic (Ch-I) transitions are almost independent of the chain length of the alkyl group, indicating that a long alkyl chain has no role from a thermodynamical point of view. Within the mesophases, the aryl and steroid cores are assumed to be piled up, interacting with each other, and the alkyl groups are apart from each other to avoid short-range interaction.
Introduction I t has been known that linearity, rigidity, and polarity are indispensable for displaying thermotropic liquid crystallinity.' Many mesogenic molecules hitherto reported have a hard-core portion consisting of aromatic rings and a relatively flexible terminal portion of alkyl chains R , ~ X - r - - - - b 2
where the central linkage, X, is usually -C=C-, -N= N-, -C=N-, --COO-, and so on. Some groups such as -CO-, U,-S,and -CHz- reduce the thermal stability of mesomorphic states due to their bent shapes. Therefore, these groups are called nonmesogenic linkages.2 Recently, (1)G.R. Luckhurst and G. W. Gray, "The Molecular Physics of Liquid Crystals", Academic Press, New York, 1979. 0022-3654/83/2087-4730$01.50/0
we reported that some cholesteryl esters of 4-arylbenzoic acids involving these linkages gave rise to highly stable mesophases, in spite of molecular di~tortion.~ Our interest in this paper is the effect of chain elongation on the themal stability of the compounds in Chart I. As can be seen from the structures, the mean axis of the alkyl chain at the aryl ring makes an angle of ca. 60° with respect to the longer axis involving the steroid core, and the chain elongation will lead to an increase in not only the molecular length but also the breadth, simultaneously. The molecular bend around the molecular center is assumed not to favor an anisotropic alignment of molecules. We will report the preparations and the thermal properties of the homologous (2) C. Destrade, N. H. Tinh, and H. Gasparoux, Mol. Cryst. Liq. Cryst., 59, 273 (1980);C. Destrade, F. Vinet, P. Maelstaf, and H. Gasparoux, ibid., 68,175 (1981). (3)M. Koden, S. Takenaka, and S. Kusabayashi, Mol. Cryst. Liq. Cryst., 88,137 (1982).
0 1983 American Chemical Society
