YOTES
1592
Even though the evaluation of the coefficients requires some effort, their determination is straightforward. Furthermore, such a general relationship permits the convenient inclusion of as many terms as may be dcsired without the necessity of laborious integrations. The technique and the relationship reported in this work can be quite usefully employed in the considcration of many problems concerned with rotational averaging.
200
300 150
L 200
m'/o
AH (arqcr/cm')
100
Certain Aspects of the Interpretation of Immersiorial Heats of Gels
100
50
by X. Hackernian arid W. H. Wade Ilepartment of Chemiutry, The University of Texas, Austin, Texas (Received December 19. 1965) Vadr ( X I O '
I n a recent publication from this laboratory' it was denionstrated that disagreements between HarkinsJura and H.E.T. surface areas could be reconciled by measuring the B.E.T. areas of powdered samples preequilibrated with water vapor a t p / p o = 1.0. This study showed initial differences between the two techniques of approximately 1.5% for particulate, dispersed samples, whereas more marked disagreement was noted for both a condensed alumina gel and an artificially compressed sample which had been used in infrared studies. In the present study, the effect of a surface area varying during the process of pre-equilibration with various amounts of water vapor on the subsequent heats of immersion is discussed. The sample used was the gel reported ~ a r l i e r . ~ The ~ ~ heats of immersion in water were measured in a twin differential adiabatic calorimeter. Approximately 0.1-g. samples were pretreated a t various values of p / p o of water vapor on an outgassing apparatus. The volumes adsorbed per gram corresponding to the dosing pressures were picked from a previously measured isotherm.2 The heats of iinnicrsion, A l i i , for various values of V s d s are illustrated by the data points of Fig. 1. For these data points all A€Il values were normalized in the conventional manner to unit surface area (1 cm.2)on the basis of the surface area of the outgassed gel (221 m2/g.). Heats of iiiiinersiori were rncasurcd in duplicate with a usual agreenicnt of +2%. Surface arcas were obtairicd from 5 2 isothcrms mcasured over a relative prcssure range of 0.04 to 0.25 and consisted of six or The .lourlid of Physical Chemidrv
mOlO8/9)
Figure 1. Open circle d a t a points are heats of irnineruiori based on a constant surface area of 221 m.Z/g. Closed circle d a t a points are the surface areas for various amounts of water preadsorbed.
seven data points. All B.E.T. plots showed some slight curvature over this range and the surface areas of Fig. 1 represent the average values. The rnaximuni and minimum values were always within =t5% of the average. Duplicate isotherms were obtained for several points with an agreement of better than 1%. It is obvious that AHl does not approach the surface enthalpy of water (118.5 ergs/cm.2) asymptotically but actually drops below this value prior to nominal monolayer completion. (The B.E.T. analysis of the mole/g. water adsorption isotherm yields 25.7 X for V,,,.) It continues to drop to less than 10 ergs/cm.2 at the highest relative pressures. However, if the ineasurcd surface area of the gel sample with various values of Vads (Fig. 1) is applied point by point to the raw data, the resulting smoothed curve shown in Fig. 2 is obtained. These corrected data are certainly Inore understandable in that the heats of adsorption do approach the 118 ergs/cn~.2cxpccted of a hydrophilic surface. The resulting two sets of data represent different thermal phenomena. The uncorrected data, though (1)
W.H . Wade, J . P h y s . Chem., 6 8 ,
1029 (1064).
(2) R . L. Every, W . H. Wade, and N . Hackerman, iM., 6 5 , 937 (1961).
(3) W. H. Wade and N. Hackerman, ibid., 64, 1196 (1960). (4) A. C. Makrides and N. Hackerman. ibid., 63,594 (1959).
NOTES
1593
30C
5
20c
0 AH
AH
Kcol/molel~
(ergs/cm‘)
I oc
L-
IO
5
20
30 40 50 Vads (xto‘ moler/g)
60
70
Figure 2. Open circle data points are the corrected immersional h e a h . The Eiolid line is the differential heat of adsorption derived from the former.
normalized in the standard manner to unit area outgassed solid, might just as well, or even less ambiguously, have been normalized to unit weight of solid adsorbate (in units of ergs/g. or kcal./g.). These data are most closely related to the characteristics of the substrate. On the other hand, the corrected data are related to the surface of the adsorlbate film and its modification by the substrate. It is this latter quantity which is pertinent to surface area determinations by the Harkins-Jura absolute m e t h ~ d . ~ Harkins and JuraJ6 in discussing the immersion process, picture two contributions to the AHl: a heat corresponding t,o the heat of vapor phase adsorption referred to liquid water as reference, followed by subsequent exothermic destruction of a liquid water interface equal in area to t,hat of the sample’s surface area. The latter term is the surface enthalpy of the liquid (water in the present case). Harkins and Jura6 point out that the calculations of Razouke flor methyl alcohol adsorption on charcoal are in error because he did not subtract the surface enthalpy per c i x 2 of surface area from his heats of immersion per ~ i i i . ~Unfortunately, . Harkins and Jura themselves may also have been in error. The correct value to subtract is not the surface enthalpy of methyl a,lcohol per times the area of the outgassed sample, but rather the experimentally measured heat of immersion for the surface preequilibrated with methyl alcohol. This surface may
have a different area. For one sample of TiOz Harkins and Jura found the surface areas to be identical. The present authors have never found this to be the case, a t least, for any samples with surface areas greater than 3 m.2/g. Of course, the alumina gel discussed in this paper is an extreme example of surface area diminution during the adsorption process but nevertheless such diminution is an intimate part of the adsorption process. The heat contributions associated with such adsorption processes would be obtained from isosteric heat calculstions from adsorption isotherms, by adsorption calorimetry, or by immersion calorimetry. If gels are considered to be an assemblage of spherical particles, the variation of surface area in gel samples can generally be traced to three sources: (a) formation of a uniform adsorbate film with either a resulting increase or decrease in surface area depending on the coordination number (average number of spheres in contact with any given sphere) ; (b) reversible capillary condensstion in the contact zones of the spheres which always cause a diminution in surface area; and (e) irreversible filling of ((pores” formed by gel particles which do not touch but are in close proximity-this further reducing the surface area. The interpretation of isosteric heats where process (c) is operative is thermodynamically questionable, severely limiting interpretations at high relative pressures. Experimental surface area measurements for this gel sample show large (up to 40%) decreases a t submonolayer coverages where no hysteresis was noted in the adsorption isotherms,2 indicating that primarily mechanisms (a) and (b) are operative. Previous calculations’ showed that a small fraction of water held in contact zones is the major factor in decreasing the surface area. This is especially true for hydrophSlic gels with coordination numbers of 6-8 and indicates that there are sizeable contributions to the heat of adsorption from water molecules finding themselves in “force” fields of two gel pa,rticles surfaces, compounding the difficulties of interpretation of heats adsorption us. coverage. Moreover, the quantitative concept of “coverage” is meaningless. Large variations of surface area below monolayer coverage invalidates B.E.T. V , values and, of course, makes one wonder about the validity of the ?J2 areas themselves. Figure 2 illustrates that the differential heat of adsorption on the film, Le. , the slope of the corrected data, exhibits two small maxima. The positions of the maxima are close to the closure points jn the hysteresis loop of the adsorption-desorption isotherms for these (6) W. D. Harkins and G . J u r a , J . Am. Chem. Soc., 66, 919 (1944). (6) R. I. Razouk, J . P h y s . Chem., 45, 190 (1941).
Volume 68, Number 6 June, 1964
NOTES
1594
samples.2 Kiselev7 has noted such maxima for narrow pore distribution samples. I n conclusion, the simple immersion process as envisioned by Harkins and Jura must be re-examined, especially when applied to gel samples. The present authors,2 as well as others,Q-1° may have been wrong in arbitrarily subtracting the surface enthalpy of the immersion liquid from the experimental immersion heats of the outgassed surface in order to obtain the heat of adsorption.
Acknowledgment. The authors wish to express their appreciation to the American Petroleuni Institute and the Robert A. Welch Foundation for their support of this work. (7) A. V. Kiselev, “Proceedings, Second Congress on Surface Activity,’’ Vol. 2, Butterworth, London, 1957, p. 189. (8) B. Millard, E. G. Caswell, E. E. Leger, and D. R . Mills, J . Phys. Chem., 5 9 , 976 (1955). (9) G. J. Young and F. H. Healey, ibid., 5 8 , 881 (1954). (10) J. W. Whalen, ibid., 6 5 , 1676 (1961).
Proton Magnetic Resonance Spectrum of
4-Chloro-1,2-butadiene by Raymond C. Ferguson Contribution A’o. 142 f r o m Elastomer Chemicals Department> Experimental Station, E . I . d u Pant de Nemours and Company, Wilmington, Delaware (Received Decembe? 19, 196s)
The coupling constants of allenic systems are of interest because of measurable long range couplings’ and evidence that the relative sign of the JIB(H,H) coupling through the allenic bonds in 1-chloro-l,2butadiene2 is negative. The spectrum of an isomer of the latter compound, 4-chloro-1,2-butadiene (I), is reported in this note. 1
2
3
4
HZC=C=CH -CHZCl (BZ) (C) (-b)
I
Results The 60-Xc. proton resonance spectrum of I is a perturbed first-order pattern. Because of the geometry of the molecule and rapid rotation of the CH&l group, the =CH2 protons were magnetically equivalent, as were the CHZC1 protons. Thus, the spectrum was analyzed as an AZBZC case. The iterative computer method of Smalen and Reilly3 was employed to fit the spectrum. Each of the four The Journal of Physical Chemistry
possible combinations of relative signs of the coupling constants (JAB JBC f,and JCA A ) was processed through the complete iterative procedure to a “best” fit. Two of these were rejected because of significantly poorer frequency and intensity matches to the observed spectrum. However, an unequivocal choice between case 1 (all signs +) and case 2 (JBC -) could not be made on the same basis. The relative signs of the coupling constants were established by the field-sweep double resonance m e t h ~ d . The ~ definitive experiments involved observing the group A multiplet pattern while irradiating the group B multiplets. In the first-order approximation, the group A multiplet pattern is a doublet due to the coupling JCA = 7.7 c.P.s., each component of which is a triplet due to = 2.2 C.P.S. Similarly, the group B the coupling JAB multiplet pattern is a doublet due to JBC = 6.6 c.P.s., with each split into triplets with JAB = 2.2 c.p.s. Most of the lines in the observed spectrum are also split by second-order effects. For case 1, the high field “triplet” lines of group A (lines 1-8) have energy levels in common with the high field “triplet,” lines (lines 17-24) of group B; similarly, the low field “triplet” lines of group A (lines 9-16) and group B (lines 25-32) have energy levels in common. The reverse situation holds for case 2 : the high field “triplet” lines of group. A and low field “triplet” lines of group B have energy levels in common, as do the low field “triplet” lines of group A and high field “triplet” lines of group B. The conditions for spin decoupling groups A and B are that the second (irradiating) radiofrequency field should have an amplitude yHz/2n = JAB = 2.2 C.P.S. at a frequency Iv2 - ”11 approximately equal to the frequency separation between the group A and group B “triplets” having energy levels in ~ o m m o n . Thus, ~ the audio side-band (decoupling) frequencies differ for cases 1 and 2. For case 1, the decoupling frequency would have been approximately 52 C.P.S. Instead, the high field ‘%riplet” of group A collapsed a t a decoupling frequency of 58 c.P.s., and the low field “triplet” of group A collapsed a t 44 C.P.S. This was, in fact, the predicted behavior for case 2 and showed that the relative sign of JBC is negative. The n.ni.r. parameters for I are summarized in the
+,
(1) E. I. Snyder and J. D. Roberts, J . Am. Chem. Soc., 84, 1582
(1962). (2) S. L. Manatt and D. D. Elleman. ibid., 84, 1579 (1962). (3) J. D. Swalen and C. A. Reilly, J . Chem. Phys., 3 7 , 21 (1962). (4) W . A. Anderson and R . Freeman, ibid., 3 7 , 85 (1962).
