NOTES them to refer to some stable complex of bromine chloride with chloride i ~ n s . ~l8~ !An ' ~attempt has been made' t o determine the stoichiometric composition of this complex from the dependence of the electromotive force of BrCI solutions on the HC1 concentration. The composition is claimed to be BrCls6-. We cannot, however, quite agree with the authors' arguments and prefer to consider the composition still uncertain. In the same paper,' "neutral" BrCl in aqueous solution is reported to exhibit a weak and very flat absorption maximum at 343 mp. We should like to suggest that this maximum is due not to BrCl but to a small amount of BrCIZ(Z-l)- present in the neutral solution. This suggestion is based on the fact that the maximum at 343 mp becomes much more pronounced when chloride ion is added, continuing to increase' even when the chloride concentration has reached 5 M . This sharp peak at 343 mp (whose existence we have been able to confirm) can thus be confidently ascribed to the BrClz(z-l)-complex or complexes. Furthermore, whenever aqueous bromine chloride solutions are employed for preparative or analytical purposes, the solutions are stated to be stabilized by the addition of hydrochloric a ~ i d ~ , ' and ~ , ' ~are therefore in fact solutions of the chloride complex of BrC1. On the other hand, the present work seems to show conclusively that at sufficiently low chloride concentration the compound BrCl does exist in aqueous solution. This contention is based on the similarity between the spectrum obtained by us and those reported in the literature for BrCl in carbon tetrachloride and in the gas all exhibiting a maximum in the vicinity of 370 mp. For reasons outlined in the previous section, our results do not, however, lend themselves to a quantitative evaluation of the molar extinction coefficient of BrCI.
The Universal Nitrogen Isotherrn by Conway Pierce Department of Chemistry, University of California, Riverside, California 99609 (Received A'osember 30, 1967)
3673
a
vert Shull's t's to n's, 4.0 for Cranston and Inkley, and 3.54 B for Lippens. Only values starting at a relative pressure of 0.2 are included, since at lower pressures there is greater deviation among isotherms for various surfaces than above this pressure. Table I : Proposed n Values and Selected Isotherm Data
__--__ PIP0
0.20 0.25 0.30 0.35 0.4G 0.45 0.50 0.55 0.60 0.65 0.70 0.75 0.80 0 825 0.85 0.875 0.90 ~
--Isotherms-n values--------
Shull
Pierce
1.15 1.22 1.30 1.37 1.46 1.55 1.64 1.74 1.86 2.02 2.18 2.39 2.64 2.81 2.99 3.24 3.50
1.25 1.32 1.39 1.47 1.54 1.62 1.70 1.80 1.90 2.02 2.17 2.35 2.58 2.76 2.92 3.12 3.33
Cranston and Anatase, Lippens Inkley cc/g
1.23 1.31 1.42 1.51 1.61 1.72 1.83 1.95 2.08 2.23 2.42 2.66 2.98 3.22 3.46 3.82 4.22
1.20 1.27 1.34 1.42 1.52 1.61 1.70 1.79 1.90 2.00 2.13 2.30 2.52 2.66 2.80 2.95 3.14
3.8 4.1 4.4 4.6 4.9 5.2 5.5 5.8 6.2 6.6 7.0 7.6 8.5 9.0 9.5 10.2 11.0
h4Tg (cornposite), mg/g
2.70 3.00 3.40 3.80 4.07 4.35 4.58 4.85 5.10 5.40 5.80 6.30 6.90
Plots of the proposed n values, Figure 1, show that the ones given by Shull, Cranston, and Inkley, and Pierce are all in fairly close agreement with one another but that the Lippens values are quite different. Since all values are based on experimental measurements, the differences must be due either (1) to experimental error or (2) to differences in isotherms for different surfaces (which would preclude use of a single set of n values for all surfaces). Actually the differences are not important when the n or t values are used to compute wall film thicknesses in pore-size analyses, since at the higher relative pressures where the differences are greatest the wall film thickness is much smaller than the total pore radius. In other applications, however, the differences in various sets of n values may lead to wide disagreement in results, as discussed later. It is therefore important to establish which are the more accurate and t'heir dependence, if any, on the nature of the adsorbing surface. We have attempted to do this by comparing the various proposed n values with experimental isotherms for several nonporous surfaces, in order to ascertain which ones give the best fit.
Shulll noted that nitrogen adsorption in the multilayer region may be described by a single curve if plotted as the number of statistical layers (n = V/V,) instead of as the actual amount adsorbed. He used this empirical relation to develop a curve giving the thickness, t, of the nitrogen layer as a function of relative pressure. Later Cranston and Inkley, P i e r ~ e , ~ (1) C. G. Shull, J . A m e r . Chem. SOC.,70, 1405 (1948). and Lippens and associates4 published n or t curves for ( 2 ) R. \Ir. Cranston and F. A. Inkley, Advan. Catal., 9, 143 (1957). nitrogen adsorption, all differing somewhat from one (3) C. Pierce, J . Phys. Chem., 63, 1076 (1959). another. Lists of all these n valuesoare summarized in (4) B. C. Lippens, B. G. Linsen, and J. €I. de Boer, J . CataZ., 3, Table I. A layer thickness of 4.3 A was used to con32 (1964).
Volume 78, Number 10 October 1968
3674
NOTES
4-
3-
n 2-
1-
01
0.2
0.4
0.6
0.8
P/P. Figure 1. Comparison of proposed n values: S, Shull; C, Cranston and Inkley; L, Lippens; P, Pierce.
-
Figure 2.
2
1
2
n
3
V-n plots: (A) anatase, by the Lippens and Pierce
n’s; (B) anatase, using the Shull and the Cranston and Inkley
n’s; (C) graphitized carbon black MTg, by the Lippens and Pierce n’s for the composite isotherm of Table I; (D) Carbolac 1, by the Lippens and Pierce n’s for the de Boer isotherm.
The comparison is made by V-n plots such as those in Figure 2. In the absence of capillary condensation, the plot is a straight line whose slope is V / n or V,. When extrapolated to V = 0, this line should intersect the axes at the origin. The two conditions, that data points fit on a straight line and that this line extrapolate to the origin, combine to make a V-n or V-t plot a rigorous test of both the accuracy of the n values and the isotherm data. The Journal of Physical Chemistry
To test the proposed values, we sought as wide a variety as possible of isotherms extending to high relative pressures. Of those investigated, some were eliminated because the scatter of points shows that their accuracy is questionable. The ones finally selected for the tests were those of a n a t a ~ epotassium ,~ chloride,B silica powder prepared from silicon tetrachloridej7** quartz with hydrated ~ u r f a c e ,aluminum ~ foi1,’O and graphitized carbon blacks.8 After careful inspection of existing data for raw carbon blacks, it was concluded that there is too much variation in published isotherms for these to warrant their use with confidence, so none was included in the tests. All of the isotherms that were used appeared to be very accurate, as judged by the fit of points to a linear plot. Typical V - n plots from all four sets of n’s are shown in Figures 2A and B for the anatase isotherm. Plots for the other test isotherms were similar to these in all respects. From these we conclude that (1) the n values of Pierce, Shull, and Cranston and Inkley all fit the experimental isotherms better than Lippens’ n’s, ( 2 ) the Pierce values show the best over-all fit, giving a straight line that extrapolates to the origin in every case, (3) in the relative pressure range 0.2-0.8 the V-n plots appear to be completely independent of the nature of the solid surface, and (4)Lippens’ n’s do not give a V - n plot that extrapolates to the origin for any one of the test isotherms, but in every case the line intersects the V axis somewhat above zero. The facts that this behavior is so generally true and that the Lippens values differ so much from the other three sets lead to the conclusion that these nJsare not of the general applicability we find for the Pierce n’s. To date we have not been able to find any published isotherm (except from de Boer’s laboratory) that does give a V-n plot for Lippens’ n’s that extrapolates to the origin. We therefore feel that isotherm analyses based on these n’s are questionable. It is surprising to find that a single set of n’s can describe isotherms for such different types of surfaces as anatase, which is very nonuniform, and graphitized carbon blacks, which have the most uniform surfaces yet studied. It has been shown1‘ that isotherms for uniform graphite surfaces are stepped when measured a t sufficiently low reduced temperatures that po is of the order of 10 torr or less. Probably the nitrogen isotherm too would be stepped if we could achieve the (5) G. Jura and W. D. Harkins, J . Amer. Chem. SOC.,66, 1356 (1944). (6) A. G. Keenan and J. M. Holmes, J . Phys. Colloid Chem., 53, 1309 (1949). (7) G. J. Young, J . Colloid Sci., 13, 67 (1958). His isotherm (private communication) agrees closely with our own in ref 8. (8) C. Pierce and B. Ewing, J. Phys. Chem., 68, 2562 (1964). (9) A. V. Kiselev, Moscow State University, private communication. (10) R. Bowers, Phil. Mug., 44, 4676 (1953). (11) B. W.Davis and C. Pierce, ibid., 70, 1051 (1966).
NOTES proper temperature to give a po value this low, and the absence of steps at higher temperatures appears to be due to thermal agitation which smears out small differences in attractive forces in successive layers. The universality of the isotherm may then be attributed to the fact that it is measured at a relatively high reduced temperature. In view of the wide variety of surfaces represented in the test isotherms used, it seems likely that the same n values would apply to any nitrogen isotherm that has the sharp inflection point near V m that is associated with a high BET C value. On the other hand, we would not expect these n’s to apply for isotherms with a low C value, such as observed for nitrogen on Teflon.12 I n isotherms such as this it is unlikely that a complete monolayer is ever formed, since, as shown by Whalen, Wade, and Porter,13 the isotherm does not give a linear BET plot and attempts to compute the area from point B lead to low estimates.
Applications As reported earlier3 the n values of the universal nitrogen isotherm may be used (1) to compute V m from adsorption in the multilayer region, (2) to detect the beginning of capillary condensation in large pores or a t contact points between powder particles, (3) to detect the presence of micropores that fill at low relative pressures, and (4) to evaluate separately the amounts adsorbed in micropores and in a multilayer film on free surface not in such pores. Of the various methods that may be used for analysis of experimental isotherms by comparison with ideal values of the universal isotherm, we prefer V-n plots, such as those of Figure 2. We personally like this plot better than the equivalent V-t plot suggested by de Boer and associates14 because t values are themselves based on n values and involve an assumption about the thickness of an adsorbed layer. Also the slope of a V-n plot gives directly the value of Vm Recently de Boer and associates have reported14 several applications of V-t plots from the universal isotherm to some of the samples we had earlier investigated. Their results, if correct, discredit all of our own results for in every case they differ from ours. We have therefore sought to determine which are the more accurate in two important applications, (1) the determination of the area of graphitized carbon blacks or (2) the measurement of micropore adsorption in carbon blacks. Until this question is resolved, the existence of two different sets of values, particularly for graphite areas, can only lead to confusion. 1. Xurface Area of Graphitized Carbon Black. Nitrogen isotherms for carbon blacks are in agreement with the ideal free surface isotherm in the multilayer region but not in the vicinity of the Vmpoint. We have ~ h o w n ~ that * ~ J ~V , computed from the multilayer adsorption is about 1.19 times the apparent BET value and have postulated that the Nz molecules do not
3675 occupy the normal 16.2 A2/molecule at Vm but rather have a cross section which is 1.19 times this, or 19.3A2/ molecule. A hypothesis was proposed to explain this behavior. De Boer’s analysis agrees with our conclusion that Vmfrom the V / n ratio in the multilayer region is larger than the apparent BET value, but he finds a ratio of 1.10 for the two and proposes a different explanation than ours to account for this. A V-n analysis of the isotherm for XTg, Sterling NIT (3100”), is shown in Figure 2C. When our n’s are used, the points in the multilayer region a t relative pressures above 0.4 all fall on or very near a straight line that extrapolates to the origin. On the contrary, when we use Lippens’ n’s the plot has quite a different slope and does not intersect the axes a t the origin. The slope of the line using our n’s gives a V , value of 2.62 mg/g. The BET or point B V m is 2.19 mg/g. The respective areas computed from these are 9.1 and 7.65 m2/g, giving a ratio of 1.19, in agreement with our previous conclusions. The higher area, 9.1 m2/g, agrees with areas computed from adsorbates other than nitrogen. 11,16 de Boer’s V-t plot for Sterling FT (2700°), a sample whose isotherm has been shown to be identical with that for MTg when corrected for the difference in areas, gives a linear plot that does extrapolate to the origin when Lippens’ n’s are used. This shows that his isotherm is somewhat different from the one we use. Ours is a composite one constructed from our own data8 and the published data of Isirikyan and Kiselev,16 Holmes and Beebe,” and Polley, Schaeffer, and Smith.’* I n the range from 0.2710 to 0.871, all these isotherms lie very close together when reduced to the same area, and where there are differences an average curve is drawn. Values for this composite isotherm are given in the last column of Table I for use by others who may wish to make an independent check. We know of no other isotherm that is so well tested over so wide a pressure range, and the close agreement in the independent values attests to its accuracy. We conclude therefore that the isotherm used by de Boer must be incorrect. Since neither his isotherm nor the n values used for the V-t plot are in agreement with data of others, we cannot accept his conclusions regarding the area of graphitized carbon black. (12) D. Graham, Phil. Mug., 65, 1815 (1962). (13) J. W. Whalen, W.H. Wade, and J. J. Porter, J . Colloid Surjace Sci., 24, 379 (1967). (14) J. H. de Boer, B. G. Linsen, Th. van der Plas, and G. J. Zondervan, J . Cutal., 4, 649 (1965); J. H. de Boer, B. G. Linsen, and Th. J. Osinga, ibid., 4, 643 (1965); B. C. Lippens and J. H. de Boer, ibid., 4, 319 (1965). (15) C. Pierce and B. Ewing, J . A m e r . Chem. Soc., 84, 4070 (1962); J . Phys. Chem., 71, 3408 (1967). (16) A. A. Isirikyan and A. V. Kiselev, ibid., 65, 601 (1961). (17) J. M. Holmes and R. A. Beebe, ibid., 61, 1684 (1957). (18) M. H. Polley, W. D. Schaeffer, and W. R. Smith, ibid., 57, 469 (1953). Volume ?‘Z9Number 10 October 1968
NOTES
3676 6. Micropore Volume of Carbolac 1. The abnormal shape of isotherms for certain carbon blacks, such as Carbolac 1, suggests that they contain narrow micropores that fill at low relative pressure. When the volume adsorbed in such pores is included in the V , measurement of the surface area computation, the result is to give an apparent area that is much too large. I n 1959 we used the ideal isotherm to analyze adsorp&ion3by Carbolac 1, finding that the area of surface not in micropores is about 460 m2/g instead of the apparent value of near 1000 m2/g. Walker and Kotlensky19 have reported a similar analysis for Carbolac 1. Recently de Boer and a s s o c i a t e ~have ~ ~ also reported micropore analyses for Carbolac 1, based on V-t plots, but apparently they were unaware of the previous works, since no reference is made to them. Here too we find that use of our n’s gives quite different results than those with Lippens’ d s . An analysis for de Boer’s Carbolac 1 isotherm, using both sets of n’s, is shown in Figure 2D. If we assume that micropores fill by capillary condensation at relative pressures lower than those normally required to reach V m on a free surface, as demonstrated in studies of charcoals with very narrow pores,20it follows that the slope of the straight line portion gives the V , value for free surface not in pores and the intercept of this line on the Ti axis gives the amount adsorbed in the filled pores. Using our n’s for the V-n plot gives V,, the volume adsorbed in the filled micropores, as 126 cc (STP) and V,, the monolayer volume on free surface not in pores, as 105 cc (STP). With Lippens’ n’s we find V c = 155 cc (STP) and V , = 53 cc (STP). de Boer’s own analysis gives V , = 197 cc (STP) and V , = 64 cc (STP), but he has (incorrectly we think) assumed that the t values may be used to describe adsorption on the walls of very narrow pores. As noted above, such micropores appear to fill by capillary condensation which occurs at relative pressures even lower than those normally required to complete the first layer on a free surface. Consequently the t values based on free surface adsorption are not applicable. This point has also been noted by Sing.21 These Carbolac results, standing by themselves, do not provide any basis for a choice between our n’s and Lippens’. However, since his n’s do not fit the other isotherms discussed, it seems safe to conclude that the results with our n’s are the correct ones. Also we note that when our n’s are used to construct V-n plots for de Boer’s isotherms of two other carbon blacks, Spheron 6 and Elf 5, we find no indications of micropores in either whereas his own analysis does show such pores. To summarize we find no evidence in the recent work of de Boer and associates to discredit either our former n values or the results obtained in analyses of experimental isotherms by these. I n particular we find no reason to doubt our conclusions regarding the surface The Journal of Physical Chemistry
areas of uniform surface graphites and we still recombe used for the mend that a cross section of 19.3 nitrogen molecule at completion of the first layer on such surfaces.
Az
Acknowledgment. This research was supported in part by a grant from the Petroleum Research Fund administered by the American Chemical Society. Grateful acknowledgment is made to the donors of this fund. (19) P. L. Walker, Jr., and W. V. Kotlensky, Can. J. Chem., 40, 184 (1962). (20) C. Pierce, J. W. W h y , and R. N. Smith, J. Phys. Colloid Chem., 53, 669 (1949). (21) K. S. W. Sing, Chem. I n d . (London), 829 (1967).
Dielectric Relaxation in Pure Chloroform1 by Thirumalai V. Gopalan and Prasad K. Kadaba Department of Electrical Engineering, University of Kentucky, Lexington, Kentucky 40606 (Received February 88, 1968)
There is considerable discrepancy in the values of the loss factor e” of pure chloroform as reported by different workers. Conner and Smyth,2 for example, reported a value of 0.37 at 25” in the 10-cm region. The critical wavelength according to their measurements is 1.4 cm. F i ~ c h e ron , ~ the other hand, obtained a much smaller value of 0.0058 at 25” in the 5-cm region, where, according to the data of Conner and Smyth, one should obtain a value higher than 0.37 at 5 cm unless two peaks exist in the loss factor-frequency curve, one near the 1-cm region and the other above 10 cm. I n the present investigation, a systematic study of the microwave absorption of chloroform has been undertaken to verify the earlier results.
Experimental Section The real part E’ and the imaginary part e’’ of the dielectric constant of chloroform have been measured at 30” in the microwave region from 1.2 mm to 10 cm. A slight modification4 of the method developed by Surber6 has been used for the centimeter region. The measurements at 6.1 and 3.05 mm were made using the method of Pine, Zoellner, and RohrbaughU6 The (1) This work was supported in part by a contract from the U. E;. Atomic Energy Commission arid in part by a grant from the National Science Foundation. (2) W. P. Conner and C. P. Smyth, J. A m e r . Chem. SOC.,65, 382 (1943). (3) E. Fischer, Z . Naturforsch., Sa, 168 (1953). (4) P. K. Kadaba, J. Phys. Chem., 62, 887 (1958). (5) W. H. Surber, Jr., J. App2. Phgs., 19, 514 (1948). (6) C. Pine, W. G. Zoellner, and J. H. Rohrbaugh, J. Opt. SOC.Amer., 49, 1202 (1959).