VICTORIt. D ~ I TAND ~ ; NOELH. TURNBR
2718
BH, with trimethylamine is probably equal, within experimental error, to the high-pressure limiting value. The comparison of the rates for the associat,ion of trimethylamine with BF3 and with BH3 is informative about the role of boranes as electron pair acceptors. Any ordering of so-called “acceptor strength” based on simple structural correlations is apt to be misleading. Although BHa is estimated to be much less stabilized in its planar geometry than BF3,8,git is clearly an even less efficient acceptor toward trimethylamine than BF3. A similar conclusion can be made with respect to the reference base pyridine on the basis of thermochemical result^.^^^^ It is evident that the efficiency of BH3 as an acceptor species in the presence of a typical amine donors is not enhanced by virtue of its more flexible structure compared with BF3. Apparently, in the reaction of BFa with amine donors, even as highly substituted as trimethylamine, any additional energy required to overcome planar stabilization in BFs is readily supplied by the reaction mechanism. This observation complies qualitatively with Rhlliken’s charge-transfer model if it is also postulated that a change from planar to pyramidal geometry is energetically favored during the ap-
proach of an electron to either BFaor BH3. Apparently this gain in energy is also sufficient to effect charge transfer a t donor-acceptor separations large enough to make any difference in the relative siaes of BFa and BHI an unimportant factor in the reaction rates. Finally, as the intermolecular distance at which electron transfer takes place is reflected to some extent in the collision efficiency with which these reactions take place, our results indicate that the interaction distance for BF3 is larger than for BH3. This in turn suggests a larger electron affinity for BF3 than for BH3,which is quite reasonable. Under our conditions, BH3 reacts less efficiently with PF3 than with (CH&N. However, if one were able to take into account the falloff behavior of the PF3system, the rate constants might well turn out to be comparable a t the high-pressure limit.
Acknowledgments. The assistance of G. W. Mappes in setting up the apparatus is gratefully acknowledged. (31) H. C. Brown and L. Domash, J . Amer. Chem. SOC.,7 8 , 5384 (1956).
The Dynamic Adsorption of Water Vapor by a Fiber Drawn from a Melt of Vycor by Victor R. Deitz* and Noel H. Turner Naval Research Laboratory, Washington, D . C. 20390 (Received December 91,1970) Publication costs assisted by the Naval Research Laboratory
The adsorption of water vapor was measured volumetrically on pristine nonporous fiber drawn from a melt of Vycor rods a t 2050’. The fiber was pristine in that it had never been in contact with aqueous media or high relative humidities prior to the adsorption measurements. The krypton surface area of the fiber was 0.073 m2/g. One sample of fiber was used for all reported measurements, and it was heated under vacuum to either 300, 500, or 800’ before each exposure to water vapor. The water vapor adsorption was then followed a t either 60, 80, 100, or 120’ a t p / p o of 0.04 and below, where p is pressure and po is the vapor pressure of water a t the temperature of adsorption. The amount of water vapor adsorbed increased with time, up to contact times of 4000 min. The hydration of a silica network, containing 3% boric oxide, was a dynamic process with a positive temperature coefficient over the experimental range employed. The data were fitted by an isochronal adsorption model where each isochrone followed a linear behavior N , , , = kt,,p i,,,, where N , , , is the amount adsorbed, k t , , the slope, and i,,,the intercept on the N , < ,axis; the subscripts t and T designate the time and temperature, respectively. The significance of the constants k , , , and i,,, are discussed. Desorption measurements indicated that the water vapor was adsorbed tightly. The similarity of the results after fiber pretreatment a t 300, 500, or 800’ indicated that the interior of the fiber could be dehydrated with relatively little change in the surface sites. This was indicated by a similar magnitude in the interaction with water vapor measured a t 120’. The first stage of the surface hydration process was the dissociative adsorption of a water molecule and the hydration of a siloxane bond. The second stage, in rapid succession, was the further adsorption of water molecules through hydrogen bonding to the surface products formed.
+
Introduction The boundary of a glass fiber at the gas-solid interface can gain or lose wat,er by (a) adsorption or desorption process via the gas phase and (b) migration from and to The Journal of Physical Chemiatry, Vol. 76, N o . 18, 2971
the bulk glass by some solid transport mechanism. At sufficiently high temperatures and partial pressures of water vapor, both processes can contribute and the overall gain or loss may be written
2719
DYNAMIC AD~ORPTION OF WATERVAPOR HzO(g)
I- S,iw
J_ Sgi*ae (HzOL
Vgme
(HaO),
where Sglasa and Vglassrepresent the surface and volume, respectively. Although, as indicated, these overall reactions can be reversed, it has been very difficult to establish the extent to which a steady state might have been approached. At temperatures below 200”, the water vapor flux at the glass surface appears to be dominated by a gas-adsorption process. Previous studies of the water vapor-silica reaction have been concerned with the amount of initial surface water and the nature of the surface bonds. This study is concerned with the interaction of water vapor with a fiber drawn from a melt of nonporous Vycor.l Many previous investigations have been concerned with a large variety of silica solids formed by p r e c i p i t a t i ~ n ~or, ~as a product of gas reaction^,^^^ for example, Cabosil. The dehydration of these materials by heat treatments in vacuo form the adsorption sites for water vapor in the boundary surface. The present study uses a pristine silica surface, i.e., one which had never been in contact with a high relative humidity or with aqueous media. The latter are known to modify silica surfaces to varying extentsa since contact with aqueous systems expedites the removal and/or deposition of hydrolysis products. I n the present study the adsorption process has been followed at rather low partial pressures in order to avoid any appeal to a capillary condensation model. The highest relative pressures employed at the temperatures studied were 0.04 at 60”, 0.02 a t 80°, 0.008 a t loo”, and 0.004 at 120”. A low surface coverage should result at these relative pressures if a physical adsorption process alone were present, and, accordingly, attention could be focused on the first step of the hydration process. Actually, the observations have shown that the hydration of a silica network is a dynamic process with a positive temperature coefficient over the experimental range employed.
Experimental Section
Formation of Fiber. The fiber was drawn from a melt of Vycor rod (Corning No. 7913) in a single-bushing tungsten crucible at 2050” with the kind cooperation of the Whittaker Corps7 The operations were conducted in purified argon. The fiber, accumulated on a rotating drum also mounted within the argon atmosphere, was transferred to the reaction vessel A of fused quartz shown in Figure 1. The evacuated tube D was inserted to diminish the dead space, and the closure was made using ConFlat flanges with a copper gasket. The upper Pyrex tubulation, equipped with a seal-off bubble E, was attached to the vacuum system. Both the Vycor rod and the resulting fiber have been analyzed and the Bz03contents were 3.02 and 2.96%, respectively.8 This result agrees closely with the analyses published by Nordberg in 1944.9 The sample
VACUUM To
C B
Figure 1. Diagrammatic sketch of reaction vessel.
of fiber (58.731 g) had a BET krypton area of 4.3 m2 per sample, using 20.4 per adsorbed Kr atom.’O The geometrical area calculated from the weight, length, and density of the fiber was 4.2 mz. Hence, the fiber
A2
(1) V. R. Deitz and N. H. Turner, J . Phys. Chem., 74, 3832 (1970). (2) R. K. Iler, “The Colloid Chemistry of Silica and Silicates,” Cornell University Press, Ithaca, N. Y., 1955. (3) W. H . Wade, R . L. Every, and N. Hackerman, J . Phys. Chem., 64, 355 (1960). (4) R. S.McDonald, ibid., 62, 1168 (1958). (5) J. J . Fripiat and J. Uytterhoven, ibid., 66, 800 (1962). (6) L. Holland, “The Properties of Glass Surfaces,” Chapman and Hall, London, 1964, p 546. (7) One of the authors (V. R. D.) wishes to acknowledge Mr. Stanley Freske of the Narmco Research and Development Division for his complete cooperation in overcoming the problems in drawing the fiber. (8) The authors are indebted to 0. R. Gates and 9. H. Cress of the Analytical Chemistry Branch of this Laboratory for these analyses. (9) M. E. Nordberg, J. Amer. Ceram. Soc., 27, 299 (1944). (10) V. R . Deitz and N. H. Turner, “Cross-Sectional Area of Adsorbed Gases on Pristine E-Glass Fibre of Known Area,” Surface Area Determination (Supplement to Pure and Applied Chemistry), International Symposium on Surface Area Determination, Butterworths, London, 1970, pp 43-54.
The Journal of Physical Chemistry, Vol. 76, No. 18, 1971
2720
VICTORR. DEITZAND NOELH. TURNER
was essentially nonporous and smooth, The distribution of the boria in the sample is unknown; it is conceivable that a selective surface adsorption may have taken place in drawing the fiber, However, the high viscosity of the melt and the rapid cooling of the fiber make appreciable migration unlikely. All of the experimental work was carried out on the above sample of fiber weighing 58.731 g. The adsorption measurements were made by a volumetric procedure already described.l' A capacitance manometer was used with readout via an integrating digital voltmeter. The calibration was made against a precision McLeod gauge and checked at frequent intervals. The capacitance manometer readings C and the pressures P were correlated by a series of relationships of the type
C
P = kl
+ kzC
A computer program was written to facilitate the calculations. The water vapor was quantitatively introduced in known amounts from a special device1' in multiple doses of 9.769 pmol. It was stored in a cold finger cooled with liquid nitrogen. Time zero of the water reaction was counted from the moment when the liquid nitrogen was removed and the finger heated rapidly t o ambient temperature with a small hot air source. The importance of the correction for the adsorption of water vapor by the walls of the apparatus has already been discussed.'l I t was necessary to make a series of measurements using an empty sample container with the temperature gradients of the experimental procedure. I n this manner, the corrections for wall adsorption can be made to include those for thermal transpiration which become significant when the absolute pressures are below 1 Torr. The temperature gradient between the sample space and the controlled rack temperature (26.9") is located in the annular region defined by two concentric cylinders C and D. The corrections for thermal transpiration12+1a have not been calculated for this geometry, which makes calibration necessary. The reaction vessel and the sample were outgassed at temperatures of 300, 500, or 800". A furnace with the necessary temperature control was raised into position for this purpose. For the water vapor adsorption measurements the reaction vessel was surrounded by a jacket through which a stream of thermostated oil ( i0.10") was circulated. The outgassing process of a solid is a very important operation. According to the mechanism of outgassing mentioned at the beginning of this paper, the first heating may release the molecular species held on the surface and the additional heatings to higher temperatures can then evolve species that diffuse from the interior. It appeared feasible with the pristine fiber from Vycor to remove the adsorbed gases preferentially The Journal of Physical Chemistry, Vol. 7 6 , N o . 19, I971
by heating in a vacuum at 300" for periods of 1-3 days. The outgassing was interrupted at regular intervals by isolating the reaction vessel and measuring the pressure buildup during 30 or 60 min. This procedure was repeated until the final pressure change decreased to a rate corresponding to 1 nmol/hr m2 of the sample; the rate included the outgassing of the walls of the adsorption system. The behavior upon dehydrating the sample at 500 and 800" will be discussed in the Results section following the presentation of the adsorption data for the 300"-treated sample.
Results Frequent measurements of krypton adsorption a t 77.4"K were made during the investigation. The surface area during the series with 300" outgassing averaged 4.3 f 0.2 m2 per sample, or 0.073 m2/g. After the several pretreatments at 500" the average area was 4.9 f: 0.1 m2 and after the 800" treatment 4.7 f 0.1 m2. When the evacuated and dehydrated fiber was exposed to water vapor, the pressure decreased continuously and the changes were measurable over a 3-4-day period. Figure 2 shows the behavior when 78.15 pmol of water vapor was introduced at 100". Since the pressure decrease was in part due to a small adsorption on the walls of the system, the kinetic interpretation to be given will be based on the corrected adsorption by the sample. Figure 2 also gives the amount of water vapor adsorbed by the fiber alone, i.e., after correcting for the wall adsorption The sequence in the magnitudes of water introduced to the outgassed sample was varied without influence on the good reproducibility that was realized in all cases. All together, four different additions of water vapor were studied at each of four temperatures (60, 80, 100, and 120"). The above data were based on the outgassing at 300" ; these were followed by two groups of measurements made on the sample after outgassing at 500" and at 800", respectively. After outgassing the sample at 300", the system was isolated from the pumps and the heating continued at 500" with liquid nitrogen on the cold finger. The residual gas at 77.4"K, a small fraction of the total, was determined. Periodically, the cold finger was isolated from the sample and the condensable gases determined from the total pressure after warming. Figure 3 is concerned with the summation of the total gas evolved upon heating at 500". The interval designated A (72 hr) in Figure 3 covers the initial outgassing in this range. At that point the sample was cooled to 120" and water vapor (97.69 pmol) was introduced for an adsorption experiment. (11) V. R. Deitz and K.H. Turner, J. Vac. Sei. Tech., 7, 577 (1970). (12) F. C. Tompkins and D. E. Wheeler, Trans. Faraday Soc., 29, 1248 (1933). (13) T . Takaishi and Y . Sensui, ibid., 59, 2503 (1963).
2721
DYNAMIC ADSORPTION OF WATERVAPOR 60
50
g
"1 -2
zw
30s
20
10
800
400
1200 Boo TIME, mlrwtes
2000
2400
2800
Figure 2. Time dependence of pressure decrease and of the adsorption of water vapor (78.15 pmol introduced) a t 100' by the fiber alone corrected for wall adsorption.
I
I
I
I
I
100-
initial heating at 500"; B, after 97.69 pmol of H20 introduced a t 120'; C, after 48.8 pmol of HzO introduced at 120'; D, after 78.1 pmol of H20introduced at 120".
Thereafter, the sample was heated again a t 500" and the difference between that introduced and that evolved was plotted as point B. Subsequent additions behaved as shown in Figure 3. Two conclusions seem warranted from these data. First, both time and temperature are important factors in outgassing the fiber at 500", and the persistent evolution of gas is direct evidence of a diffusion of products from the interior of the fiber. This technique shows that the fiber exhibited a memory of the time spent at 500". Todd14reported in 1955 the temperature dependence in the outgassing of a number of glasses including Vycor brand 96% silica glass, KO. 7910. The dotted line in
Figure 3 was calculated from his results for behavior a t 500". Todd interpreted his results in terms of diffusion from a semiinfinite body which gave linear plots of the The present results for Vycor amount evolved vs.
dt.
Table I is a summary of the outgassing products expressed as the residual gas at 77.4"Kand the condensable gas determined after warming the cold finger. Using short heating times (60 min) at 500", an amount of condensable gas evolved approximately equivalent to the water vapor introduced, butj for long heating times (1000 and 4300 min) at the same temperature of 500", additional quantities of gas, both condensable and noncondensable at 77.4"1>) ~1 and & ( t , ~ ) p
