NOTES
3832
occurring during this period. During this extended period of substantial ethylene hydrogenation on the WOa, there was no evidence of a change in color from the original yellow characteristic of unreduced W03. Similar behavior was observed in several other runs of shorter duration. Some difficulty was experienced in obtaining the same level of catalytic activity on different samples of WOa. However, in none of the experiments, conducted at temperatures ranging from 125 to 250”, was a change in color of the W03detected as a result of the ethylene hydrogenation reaction. Data on the dependence of the hydrogenation rate on hydrogen pressure showed the reaction to be first order in hydrogen. Limited data on the dependence of the rate on ethylene pressure indicate that the order in ethylene is between zero and one, which implies an intermediate degree of coverage of the active sites by ethylene. The surface coverage would, of course, be expected to vary with temperature, being lowest at the highest temperatures studied. It has been demonstrated clearly by Melville and Robbl that the “blueing” of WO3 or Moo3 is a very sensitive method for detecting hydrogen atoms in a system in which a gas phase reaction between ethylene and hydrogen atoms occurs. This indicates that the “blueing” reaction competes satisfactorily with the addition of hydrogen atoms to ethylene. One might then reasonably expect that the formation of tungsten blue would have been observed in the present work if hydrogen atoms were involved in the reaction. The very rapid formation of tungsten blue from yellow W03 has been demonstrated recently by several investigators who employed various forms of platinum to activate molecular hydrogen in experiments with mixtures of platinum and W03.3-5 I n considering the experimental conditions of Melville and Robb, it is t o be noted that the ethylene partial pressures were lower by two orders of magnitude than those employed in the present work. However, the h!telville and Robb experiments were done a t room temperature, which is much lower than the temperatures employed in the current investigation. Consequently, the coverage of the W03 surface by ethylene may not have been greatly different in the two investigations, since the effect of temperature should largely compensate for the effect of ethylene pressure on surface coverage. As a consequence of the present work, it is suggested that the ethylene hydrogenation reaction on WO3 does not involve adsorbed hydrogen atoms as intermediates, based on the absence of a color change from yellow to blue a t the surface of the oxide, It is therefore proposed that the hydrogenation of ethylene on wo3involves hydrogen rather than adsorbed hydrogen atoms. The simplest reaction scheme would appear to be one in the reaction occurs via collision Of hydrogen molecules from the gas phase with adsorbed ethylene, involving a direct addition of a hydrogen I
The Journal of Physical Chemistry, Vol. 74, No. 21, 1970
molecule to the ethylene double bond. The first-order dependence of the reaction rate on hydrogen pressure is consistent with such a mechanism. The mechanism of ethylene hydrogenation on wo3 suggested in the present work is very simple. It is not expected that the proposed mechanism will apply to all types of hydrogenation catalysts. I n the case of metal catalysts, for example, adsorbed hydrogen atoms are in all probability involved in the hydrogenation. I n any case, the simple mechanism envisioned in the present work would not readily account for all the varied phenomena encountered in ethylene hydrogenation on metals, as exemplified by the many results of experiments with de~teriurn.~,’ (3) S. Khoobiar, J . Phys. Chem., 68,411 (1964). (4) H.W.Kohn and M. Boudart, Science, 145,149 (1964). (5) J. E. Benson, H. W. Kohn, and M. Boudart, J . Catal., 5 , 307 (1966). (6) J. Turkevich, D. 0. Schissler, and P. Irsa, J . Phys. Colloid Chem., 55,1078 (1951). (7) C.Kemball, J. C h e m SOC.(London),735 (1956).
Time-Dependent Adsorption of Water Vapor on Pristine Vycor Fiber
by Victor R. Deitz and Noel H. Turner Naval Research Laboratory, Washington, D. C. 20590 (Received J u n e 16, 1970)
The dehydration and rehydration of many silica powders have been studied and the results have many similarities and some difference^.'-^ In all of these cases the samples had been subjected to specified pretreatments entailing liquid water.4 This note is concerned with the interaction of water vapor with a pristine fiber drawn from a melt of nonporous Vycor in a tungsten crucible at 2050”. The preparation of the fiber (conducted in an argon atmosphere) is similar to that reported for E - g l a ~ s . ~The rates of water vapor adsorption were determined at surface coverages of less than a monolayer and a t contact times up to 4000 min. The term “pristine” is used to define one very important aspect in the pretreatment of the Vycor fiber, namely, that it had never been exposed to an aqueous media. (I) R. K. Iler, “The Colloid Chemistry of Silica and Silicates,” Cornelluniversity Press, Ithaca, N. Y.,1955. (2) A. V. Kiselev, 10th Colston sot. Symp., D. H. Everett and F. Stone, Butterworths, London, 1958,p 236. (3) M . Folman and D. J. C. Yates, Trans. Faraday Soc., 54, 429 (1958). (4) J. H. DeBoer and J. M . Vleeskens, Proc. K o n . Ned. A k a d . Wetensch., Ser. B., 61,2 (1958). (5) V, R , Deita, and N , H, Turner, “Symposium of Surface Area Determination,” University of Bristol, ~ d 16-18,1969. y
3833
NOTES
\
'
1
2
3 4 5 PRESSURE, torr
6
7
Figure 2. Adsorption of water vapor plotted as isochrones (designations in minutes) using the d a t a of Figure 1.
TIME (minutes)
Figure 1 . Pressure decrements in the adsorption of water vapor on Vycor fiber at 100' for the indicated number of doses, each being 9.769 pmol of H20.
The latter is known to modify silica surfaces to varying extents.' The sample of Vycor fiber (58.731) had a BET krypton area of 4.3 m2 per sample (20.4 A2 per adsorbed Kr atom5). The geometrical area calculated from the weight, length, and density of the Vycor fiber was 4.2 m2. Hence, the Vycor fiber is essentially nonporous and not rough. After outgassing of the sample a t 300" before each experiment for a period of at least 16 hr, the adsorption of water vapor was determined at 60, 80, 100, and 120". Figure 1illustrates the pressure decrease for water vapor adsorption a t 100" in a fixed volume as a function of .time. A cold finger, containing a precisely known quantity of water vapor, was heated rapidly from - 196" to room temperature to start the reaction. The four curves in Figure 1 (16, 8, 6, and 4) correspond to the introduction of the indicated number of doses of water vapor, each dose being 9.769 pmol of HzO. Good reproducibility was obtained and there was no dependence of the number of doses; this indicates that the outgassing a t 300" before each introduction of water vapor
was adequate for dehydrating the surface of the Vycor fiber. Independent experiments have shown that outgassing at 500" gave the same adsorption data a t 120" as the data obtained after a 300" outgassing. The strong dependence on contact time is the new feature of these observations. Obviously, the reproducibility of water vapor studies among different investigators should depend upon this new parameter. The reaction is indeed a memory process, where time zero is the initial contact of the fiber with the water vapor. The data of Figure 1 permit the determination of pressure at any specified time interval when a smooth curve is drawn between the observed points. The corresponding adsorption, corrected for wall adsorption, may then be presented by a series of isochrones. Figure 2 gives the isochronal adsorption isotherms (10, 100, 300, 500, 1000, 2000, and 3000 min) of water vapor on the Vycor fiber a t 100". Each isochronal isotherm fol-
nc,T=
~
T
+ G,T P
(1)
lows a linear behavior where nt,T is the amount adsorbed, p the pressure, k t , T the slope and &,T the intercept on the nl,T axis; the subscripts 1 and T designate the time and temperature, respectively. The magnitudes of temperature and pressure were selected in order to realize a fractional surface coverage by water molecules. Since the Vycor formulation is mainly silica, it is proposed that the elementary mechanism consists of two consecutive reactions: (I) a chemisorption process to break the Si-0-Si bonds in The Journal of Physical Chemistry, Vol. 74, No. 21, 1970
3834 the strained surface networks, and to form SiOH groups; (11) the subsequent physical adsorption of water vapor to the reaction products on the surface, mainly the -Sifor the isoOH groups. The increasing values of it,T chrones of Figure 2 are ascribed to the chemisorption process. In order to attain 100% geminal formation, =SI(OH)z, in an octohedral face of 8-cristobalite, it is necessary to chemisorb 8.04 pmol of water vapor/m2 or 34.5 pmol per above sample of Vycor fiber. Adsorption above this magnitude is attributed to the monolayer forming on the geminal silanol groups. The limiting value for i has not quite been reached by the 3000-min isochrone at 120°, the most severe condition so far investigated. The linear isochronal adsorption isotherms of Figure 2 suggest that a simplified behavior is valid for the pressure-dependent adsorption that takes place on top of the chemisorbed water. The summation of the energy requirements to break one -Si-O-linkage of the silica network, to break one H-0 bond in a water molecule, and to form two surface Si-OH groups is not large. An estimate, calculated from the enthalpies of formation of definite crystalline hydrated silicas and that of silica and water vapor, is in the range -3.5 to -5.3 kcal per Si-OH group. The calculated isosteric isochronal heats of adsorption are of this magnitude. An indication of time dependency in the water vapor silica reaction can be seen in the work of Hockey and Pethicaa6 These investigators found different adsorp-
COMMUNICATIONS TO THE EDITOR tion after 15 min and after 1 hr; however, these contact times were too short to bring out the new behavior described in this note. The importance of the experimental procedure employed in the dehydration of silica for infrared studies is well e ~ t a b l i s h e d . ~It~is~ now equally essential to keep a log of the contact time, since the subsequent hydration has now been shown to be a memory process. Previous investigations also have been with dehydrated “silica-gel” type of surface boundaries. A variable amount of surface hydration has been shown to be introduced in the procedure for sample preparat i ~ n , ~ This ~ ~ Ocomplication to an understanding of the silica-water vapor interaction has been brought out by careful heats of immersion measurements. Pristine Vycor fiber, and also pristine E-glass fiber, are not subject to this disturbance. Some processes that add the finish to the glass fiber in reinforced composite materials are concerned with pristine fiber. Thus, the influence of contact time in the adsorption of water vapor may be a critical parameter in the adhesion of polymers in glass reinforced composite materials. (6) J. A. Hockey and B. A. Pethica, Trans. Faraday Soc., 57, 2247 (1961). (7) L. H. Little, “Infrared Spectra of Adsorbed Species,” Academic Press, New York, N. Y., 1966. (8) C. Morterra and M. J. D. Low, J . Phys. Chem., 73,321 (1969). (9) W. H. Wade, R. L. Every, and N. Hackerman, ibid., 64, 355 (1960). (10) A. C.Makrides and N. Hackerman, ibid., 63,594 (1959).
C O M M U N I C A T I O N S T O THE E D I T O R
Electron Spin Resonance Spectra of Radicals Formed from Nitrogen Dioxide and Olefins S i r : Some time ago, Schaafsma and Kommandeur obtained esr spectra from a range of organic compounds, including simple olefins, by reaction with KOZ, all of which were assigned to charge-transfer complexes. Subsequently, we showed that in the particular case of methylmethacrylate, reaction occurred to give a nitroxide radical,2 which was identified by comparing the liquid- and solid-state esr spectra with those of authentic nitroxides. The liquid-phase spectrum taken alone ~70uldnot have provided a convincing identification, but together with that .from the solid, which is a far better “fingerprint” of a nitroxide radical, the identification was thought to be sound. The Journal of Physical Chemistry, Vol. 7 4 , N o . 41, 1070
This was part of a more wide-ranging study of reactions between NO and NOz and organic materialss in which it was shown that both iminoxy radicals and nitroxides could be formed. In many cases, identifications were supported by analysis of detailed hyperfine patterns and by comparison with authentic samples. Solid-state spectra again supported these identifications. We therefore thought it probable that several of the species detected by Schaafsma and Kommandeur’ were similar radicals and not charge-transfer complexes, but Bielski and Gebicki nevertheless used the charge(1) T.J. Schaafsma and J. Kommandeur, J . Chem. Phys., 42, 438 (1965). (2) J. A. McRae and M. C. R. Symons, Nature (London),210, 1259 (1966). (3) W. M.Fox, J. A. McRae, and M. C. R. Symons, J . Chem. SOC.A , 1773 (1967).
