- Email: [email protected]
Thermodesorption of water from silicate surfaces
Thermodesorption of Water from Silicate Surfaces G A R Y M. N I S H I O K A AND J A N E T A. S C H R A M K E Research and Development Division, Owens-Corning Fiberglas Corporation, Technical Center, Granville, Ohio 43023
Received March 19, 1984; accepted October 4, 1984 A sensitive electrolytic technique suitable for quantitative measurement of the different states of water bound to low-area substrates (0.1 mZ/g) is described. Water is thermally desorbed from a substrate and measured by means of a P205 electrolytic cell. The method is sensitive enough to detect adsorbed water on solids with specific areas as small as 0.1 m2/g, This technique was used to study water on pure silica and silicas doped with metal cations. Strongly bound molecular water and water resulting from dehydroxylation reactions were detected and quantitatively measured. Results were consistent with previous studies of porous silicas. The presence of cation-aluminate sites on the silica surface greatly increased the amount of strongly bound water. Cation-aluminate sites are proposed as being the primary adsorption sites for water on glass. Silane coupling agents, often used to promote water resistance, therefore do not reduce water adsorption by the replacement of polar silanols with less polar organosilanes. Rather than reducing the thermodynamic driving force for adsorption, silanes appear to reduce the rate at which water is adsorbed to the glass surface. © 1985AcademicPress,Inc. posites immersed in water occurs at the glass-resin interface (8, 9). Silane coupling agents are normally coated onto glass fibers to reduce this moisture-induced degradation o f composites; the efficacy o f the coupling agent coating is inversely related to its water wettability (10). Electrical conductivity is a third property affected by water. The conductivity o f the glass surface increases with the a m o u n t o f adsorbed water (1 1, 12). Electrical conductivity presumably increases due to an aqueous layer containing dissolved surface cations. T h e same effect causes the degradation o f electrical resistance o f glass fiber-reinforced circuit boards when exposed to moisture (13, 14). M o d e r n understanding o f the interaction o f water with glass is to a great extent based on the m a n y studies o f water adsorbed on porous silica (15-18). Surface silanol groups are the primary sites for water adsorption onto silica (19). There are four to five silanol groups per square n a n o m e t e r on a smooth, nonporous, heat-stabilized, fully hydroxylated
INTRODUCTION
N u m e r o u s studies indicate that adsorbed water controls significant properties o f glass. F o r example, the strength o f E-glass fibers decreases with increasing h u m i d i t y in the testing e n v i r o n m e n t (1-3). Static and dyn a m i c fatigue o f glass also increase with increasing testing humidity (4-6). These stress corrosion effects are thought to be caused by the reaction of water with the silicate network at crack tips. This reaction causes b o n d rupture at the tip, exposing new bonds, which rupture after further water adsorption occurs. T h e process continues until the crack grows to the critical size required for spontaneous failure. Adhesive failure at the polymer-glass interface is also induced by water attack. The strength o f plastics reinforced with inorganic fillers is greatly reduced after immersion in water. Indeed, a standard test o f glass-reinforced composites is to measure shear strength o f a test composite before and after immersion in water (7). T h e failure m o d e o f corn102
0021-9797/85 $3.00 Copyright © 1985 by Academic Press, Inc. All rights of reproduction in any form reserved.
Journal of Colloid and Interface Science, Vol. 105, No. 1, May 1985
THERMODESORPTION OF WATER
silica surface. Two types of silanols exist, "isolated" and "paired." Water molecules adsorb with oxygen atoms next to silanols; the adsorption energy of a single watersilanol hydrogen bond (25 kJ/mole) is less than the adsorption energy of water on a silanol-water complex (44 kJ/mole) (20). This means that water is less strongly adsorbed on a single silanol group than on other, already adsorbed water molecules. A silica surface therefore contains clusters of water molecules hydrogen-bonded to each other and to one or several silanol groups on the surface. Although this picture is often invoked in studies of glass surfaces, there are obvious reasons why the nature of water adsorbed on glass may differ from the foregoing description. The most obvious reason is that glasses are chemically different from pure silica. The surface of a glass contains not only silanol groups, but metal hydroxy groups. Metal hydroxy groups may be expected to be highenergy sites for water adsorption, with properties different from silanols. A second difference is that commonly studied silica powders have surface roughness on the scale of hundreds of Angstroms, whereas most glass surfaces are smooth on this scale. Regions of high positive curvature on a rough surface have a greater separation of surface silanols. Conversely, on regions of high negative curvature surface silanols are closer together. These effects of curvature probably influence the adsorption of water on porous substrates but are not important factors on smooth, low-area surfaces. A third difference exists for glass fibers. Glass fibers are formed under extremely rapid cooling conditions (> 10 6 ° C / s ) . Because of this rapid cooling, the density and refractive index of glass fibers are significantly less than those of bulk glass samples of identical composition. The open structure of the glass fiber silicate network is probably reflected in the structure of the surface, and no doubt influences the interaction of glass fibers with water.
103
In spite of these obvious differences between silica and glass, the water interaction is far less studied with the glass surface than with the silica surface, for two reasons: (i) the one-component silica is simpler than a multicomponent glass, and (ii) commonly studied silica powders are high-surface-area substrates--generally, about 200 m2/g. Most commercial glasses have low specific areas. Glass fibers, for example, have specific surface areas of 0.2 m2/g. In particular, the second difficulty hinders studies of adsorbed water on glass. All forms of infrared spectroscopy are inadequate for the study of low-surface area substrates owing to insufficient sensitivity. All but the most sophisticated gravimetric systems (sensitive to +__1 ppm of weight) lack the requisite sensitivity for measurement of adsorbed water. The inadequacy of standard techniques prompted our development of a thermodesorption method with a sensitive electrolytic detector for the study of water on low-area substrates. This thermodesorption method can detect 0.1 ~g of water desorbed from a 1-g sample, which is sufficient sensitivity to detect water thermodesorbed from samples with specific areas as small as 0.1 mZ/g. Figure 1 is representative of the types and amount of water desorbed from glass fibers as measured by the thermodesorption technique. The rate of water evolved with change in temperature, normalized for area, is plotted against temperature. The quantity of water evolved within any temperature region is measured by the area under the curve between selected temperatures. Evidently water does not desorb uniformly with increasing temperature, but rather in specific temperature zones, each peak presumably corresponding to a different type of adsorbed water. The structure of the water represented by the peaks shown in Fig. 1 is not clear. Presented herein is an interpretation of some of these peaks, obtained by examination of the thermodesorption of water from wellcharacterized high-area substrates. Also preJournal of Colloid and Interface Science, Vol. 105, No. 1, May 1985
104
NISHIOKA
AND
SCHRAMKE
oc) \
~.4-
i-
~.2-
121 w m 0.1 bA
e.0
'
1 400
200
'
1
600
TEMPERATURE
'
1 800
1008
( oC)
FIG. 1. Representative desorption data of E-glass fibers. Each peak corresponds to a separate state of water associated with E,glass.
sented is a study of the effect of adsorbed cations on the interaction of water with the silica surface, and a proposed mechanism for how silane coupling agents affect the interaction between water and glass. APPARATUS
AND
PROCEDURE
.
Figure 2 is a block diagram of the thermodesorption apparatus. The analysis is as follows: a weighed sample is placed in a small tube furnace; the furnace temperature is raised at a programmed rate, usually 4°C/ min thereby desorbing water from the sample; nitrogen carrier gas flows through a sieve
dryer and into the tube, transporting the desorbed water to the electrolytic cell. The electrolytic cell, manufactured by DuPont Instruments, consists of a thin film of phosphorus pentoxide (P205), deposited between two helically wound electrodes. Water is adsorbed by the P205, and electrolyzed by current flowing between the electrodes. Electrolysis changes the water to hydrogen and oxygen which discharge through a vent with the carrier gas, coulometrically regenerating P205. The charge required to regenerate P205 is related to the quantity of desorbed water, which can therefore be calculated by moniELECTROLYTZC
CARRIERGASt FLOWMETER ~,
DRYER ~
OVEN ~
cELL t VENT
I
CONTROLLER] " COUNTER
COMPUTERl hO. 2. A block diagramof the thermodesorptionsystem.This apparatusmeasuresdesorbedwater of g r a m s a m p l e s w i t h a p r e c i s i o n to a few t e n t h s o f a m i c r o g r a m , Journal of Colloid and Interface Science, Vol. 105, No. I, May 1985
THERMODESORPTION
toting the current through the cell. The cumulative amount of desorbed water, oven temperature, and time are continuously registered by a computer. Weighed samples of powdered silicas were analyzed with this instrument. The specific surface area of each sample was determined by the one-point BET method. The amount of water desorbing from the sample holder was measured in a blank run, and subtracted from all measurements. This correction is small; values are reduced by a few percents. Areas of peaks were obtained by direct measurement of the cumulative water recorded between two temperatures. All data were normalized on the basis of surface area. The contact angle of water on glass fibers was measured on a Wilhelmy wetting balance (21). A single filament is suspended from a microbalance and is partially immersed in water. The beaker containing the water sits on a slowly ascending elevator, and the force on the filament is measured. The force measured is equal to the filament weight minus buoyancy plus the vertical component of the surface tension acting on the perimeter of the fiber. Since the filament weight, perimeter, buoyancy, and the surface tension of water are known, the advancing contact angle may be calculated.
OF
WATER
105
exchange resin; warmed to 50°C, and gelled. It is mixed with an equal volume of n-propyl alcohol, filtered, washed with alcohol, airdried, and dried in an oven at 150°C. The surface area of our sample, measured by the BET method, is 124 m2/g. A1. Aluminate-modified silica. Precipitated silica prepared in "A" is slurried in water. Sodium aluminate solution (containing NaA102 and NaOH) is slowly added to the silica suspension, stirred for an hour at 25°C, and the pH reduced to 10 with acetic acid. The amount of sodium aluminate solution added is controlled so that two aluminate groups per square nanometer of surface would be produced.
B. Sodium Form of Silica Surface Ten grams of precipitated silica "A" is slurried in 200 ml H20 and titrated with 0.1 N NaOH to pH 8.
C. Calcium Form of Silica Surface Ten grams of precipitated silica "A" is slurried in 200 ml H20 and 35 ml of 0.28 M calcium formate is added. The pH increases from 3.7 to 7.5. The pH is raised from 7.5 to 8 by adding 8 ml of 0.1 N N a O H and the silica is then filtered, washed, and dried at 125°C.
MATERIALS
E-glass fibers (Owens-Corning Fiberglas) are from commercial yarns used to reinforce electrical circuit boards. The fumed silica studied was Cab-O-Sil M5, obtained from Cabot Corporation and used as received. The precipitated and doped silicas were supplied by R. K. Iler, prepared by the following procedures:
A. Precipitated Silica An amorphous silica powder is prepared from Ludox AS40 (DuPont). Ludox AS40 is a silica sol containing particles Of 22 nm diameter stabilized with NH4OH to pH 9. The Ludox is deionized with mixed ion-
D. Sodium Form of AluminateModified Silica Half of the sample prepared in " A I " is washed to reduce the pH to 8.5, filtered, and dried at 125°C in air.
E. Calcium Form of AluminateModified Silica Half of sample "Al" is converted to the calcium form by ion exchange with an excess of calcium formate solution: 0.28 M calcium formate solution is added to the aluminatemodified silica slurry. The slurry is stirred 1 h, filtered, and washed until free from excess calcium ions. Just before filtering the solution, Journal of ColloM and Interface Science, Vol. 105, No. 1, May 1985
106
NISHIOKA AND SCHRAMKE
0.1 N N a O H is added to raise the pH from
7.2 to 8.1. The product is air-dried at 125°C. RESULTS AND DISCUSSION Cab-O-Sil M5 (fumed silica) is a powdered silica formed by oxidation of SIC14. The following properties of Cab-O-Sil M5 have been previously established: (1) a specific surface area of 200 _+ 20 mZ/g and (2) a surface concentration of 3.5 to 4.5 silanols/ n m 2. Because Cab-O-Sil M5 has been extensively studied, it was chosen as the substrate for initial study. Figure 3 contains thermodesorption data of Cab-O-Sil M5 taken under different conditions of measurement. The dot-dash curve represents an ambient sample of Cab-O-Sil placed in the oven and the instrument started immediately. The first peak at the lowest temperature appears as soon as the sample is inserted, and can only be derived from physically adsorbed water. The dashed curve results if the sample is placed in the oven and equilibrated at room temperature in the dry nitrogen gas until the initial peak disappears. Only after this first
peak disappears, usually after 2 h, is the instrument started and data recorded. Although most of the physically adsorbed water is desorbed by this treatment, a small amount of water strongly bound to highenergy sites remains. This water is represented by the peak at 100°C. The amount of water desorbing above 150°C is identical in these two curves, and can be wholly accounted for as the product of the dehydroxylation of surface silanols. The measured value of 1.7 H 2 0 / n m 2, equivalent to 3.4 silanols/nm 2, is in excellent agreement with reported values of clustered surface silanol concentration of 3/nm 2 (22). An interesting effect, not previously reported, is the desorption of water in two temperature zones. The boundary between the two desorption regions is at 450°C, the reported transition between reversible and irreversible dehydroxylation. These data indicate approximately equal amounts of each type of clustered group. The solid curve is a desorption curve of Cab-O-Sil which had first been immersed for 24 h in boiling water. It was then placed in the sample chamber, and the instrument was
0.88
,o 0j I :E v 8.86\
.J o T o E (J O. 0 4 -
v
I-
w m
0.02-
"\ 0.00 280
408
TEMPERATURE
600
800
(°C)
FIG. 3. Water desorbed from Cab-O-Sil M5. (--) Cab-O-Sil immersedin boilingwater 24 h, equilibrated 2 h in dry nitrogen 2 h before measurement. (---) Cab-O-Silequilibratedin dry nitrogen 2 h prior to the measurement. (.... ) Instrument started immediatelyafter Cab-O-Sil inserted. Journal of Colloid and Interface Science, Vol. 105, No. 1, M a y 1985
THERMODESORPTION
107
OF WATER
e.5
2 o
o
~-
0,4-
w
E
1~.3-
0.2---
I
o
N
•
e,e
l 8
~ _ _ . ~ _ ,
'
208
t
'
48{) TEMPERATURE
aa
i
J
600
880
(°C)
FIG. 4. The desorption of water from doped sificas. (A) Precipitated silica, (B) silica + sodium, (C) silica + ca|cium, (D) silica + a|uminate + sodium, (E) silica + aluminate + calcium. For clarity these curves have been offset.
started immediately. Immersing Cab-O-Sil in boiling water is known to create additional silanol groups (23). The measured value of 4.7 silanols/nm 2 following this treatment is consistent with a fully hydroxylated, annealed, silica surface (18). How certain cations, such as are present on glass surfaces, determine the amount of adsorbed water is revealed in Fig. 4. Figure 4 compares the water desorbed from a silica precipitated from a Ludox colloid (Sample A), and the same silica doped with various cations (Samples B-E). For all silicas, about five molecules of water per square nanometer are desorbed above 200°C. This is equivalent to 10 OH groups per square nanometer,
agreeing with previous studies (18) of nonheat-treated porous silicas. The water comes not only from surface-dehydroxylation reactions, but also from dehydroxylation occurring just beneath the silica surface. Table I reports the approximate atomic densities of sodium, calcium, and aluminum on the surface of the powdered silicas. These values were obtained by extraction of a known weight of each silica in a known volume of O. 1 N HC1, and solution analysis by atomic absorption. Table I reveals that only minor amounts of sodium or calcium are adsorbed by silanol sites. Indeed, the amount of cationic impurity present on the pure silica surface is about equal to the number of
TABLE I Surface C o n c e n t r a t i o n s
Sample (A) (B) (C) (D) (E)
SiO2 SiO2 SiO2 SiO2 SiO2
+ + + +
Na ÷ C a 2÷ A10- + Na ÷ A I O - + C a 2+
Na (atoms/nm2)
Ca (atoms/nm2)
0.28 0.12 0.01 2.00 0.31
0.14 0.03 0.37 0.04 1.02
AI (atoms/nm2) 0.006 0.004 0.004 i 2.16 1.73
H20 (molecules/nm 2) 1.5 1.7 1.9 3.9 4.6
Journal of Colloid and Interface Science, Vol. 105, No. 1, May 1985
108
NISHIOKA AND SCHRAMKE
cations present on the washed silicas. The amount of adsorbed sodium or calcium increases if aluminum is present on the silica surface. The ratio of sodium or calcium to aluminum is roughly 1:1 and 1:2, respectively. The measured concentration of aluminum is in good agreement with the predicted value of 2 atoms/nm 2 (see Materials). Also listed in Table I is the amount of water responsible for the desorption peak at 100°C in Fig. 4. The quantity of water desorbed is listed as an equivalent surface concentration. Surface aluminum is seen to greatly increase the amount of this strongly bound water. The data in Table I and results of previous investigators (18) lead us to the following picture of the surface of these silicas: All the silicas contain surface silanol (SiOH) groups. Surface silanol groups are only weakly acidic (18). Therefore, only a small fraction of silanols exist as the conjugate base, SiO-. It is onto SiO- groups that Na ÷ or Ca 2+ adsorb; Table I indicates that few of the available silanols exist as SiO- groups. In Samples D and E aluminum is present on the surface as aluminol (A1OH) or its conjugate base aluminate (A10-). Since A1OH is known to be strongly acidic, aluminum exists mainly as A10- groups. Therefore, each sodium cation is held by one aluminate site, each calcium cation is held by two aluminate sites. Water is adsorbed by SiOH, SiONa, and (SiO)2Ca groups. SiONa and (SiO)2Ca groups would probably take up more water; however, the surface concentration of these groups is low. Table I reveals that surface A1ONa or (A10)2Ca groups attract water to a much greater extent than surface silanol groups. Typical glass formulations contain oxides of silicon, aluminum, calcium, magnesium, and sodium. A reasonable deduction is, therefore, that the major sites on glass for adsorption of water are the cation-aluminate complexes, and not surface silanols. Organosilanes react with glass surfaces to confer improvements in surface-controlled properties. For example, silanes increase wetJournal of ColloM and Interface Science, Vol. 105, No. 1, May 1985
strength retention of a glass-reinforced plastic or reduce surface electrical conductivity of glass. These improvements are thought to be due in part to the reduction of surface energy because of the replacement of polar surface silanols with the organosilane molecule. Adsorption of water by the glass surface is thus reduced; and reduced water adsorption confers better glass properties. The results listed in Table I and Fig. 4 imply a different role for silanes. Replacement of silanols by organosilanes does little to alter the number of high-energy surface sites, since these are predominantly cation-aluminate complexes. Therefore, the thermodynamics of water adsorption are not appreciably altered by reaction of organosilanes with the glass surface. However, since organosilanes are relatively large molecules, they can cover high-energy surface sites. The function of organosilanes may therefore be to retard the rate of water adsorption and not to prevent water adsorption onto the glass surface. An alternate explanation is that silanes react with aluminate sites. In this case silanes would confer thermodynamic protection against water adsorption. The data presented in Fig. 5 support the first hypothesis. Curve A shows the amount of water desorbed from heat-cleaned E-glass fibers. These fibers were heat-cleaned at 250°C for 48 h to remove organic coatings, and stored at 20% RH for 6 months. Curve B shows the amount of water desorbed from similar heat-cleaned fibers that were further dip-coated in an aqueous organosilane solution (0.01 M a-glycidoxypropyltrimethoxysilane), cured at 165°C for 5 min, and likewise stored at 20% RH for 6 months. Presumably each of these glass fibers held their equilibrium amount of adsorbed water after 6 months. The treatment given these samples simulates the commercial treatment of glass cloth used in the reinforcement of epoxy resins. Silane-coated cloths remain significantly hydrophobic after the storage period, indicating that little degradation of the coating occurs. That there was an appreciable
THERMODESORPTION OF WATER
109
0.S o oj t vz
B.4-
b,J
8 (~ (:z (J
0.3-
B t-
0.2-
0.1-
0.0 180
280 TEMPERATURE
380
400
( o C)
FIG. 5. Water desorbed from (A) heat-cleaned glass fibers and (B) silane-treated glass fibers.
quantity of silane on Sample B was demonstrated by the wetting experiments listed in Table II. The water/glass contact angles were measured after the 6-month storage. Figure 5 reveals that the silane-treated glass desorbs as m u c h water at 100°C as the uncoated glass. These data confirm our hypothesis that organosilanes do not react with primary adsorption sites on the glass surface, and therefore act only to retard rather than to inhibit water adsorption. A searching examination of the role of organosilanes in water adsorption kinetics is clearly necessary in order to understand their role in composites. The heat of desorption of water desorbing at 100°C can be measured by heating rate experiments (24, 25). Assuming a linear heating rate and freely occurring readsorption,
TABLE II Advancing Contact Angle, Water on Glass Sample
Advancing angle
Heat-cleaned glass filaments Silane-treated glass filaments
40° 62°
the m a x i m u m temperature Tm at the maxim u m of the desorption peak is related to the heating rate (b) by 2 In T m - In b - -A- H + In (1
RTm
-
0m)2VsAH
FR exp( A S/R)
[11
for first-order desorption. Here A H = desorption heat, O m = fractional surface coverage at Tm, Vs = volume of solid phase, F = flow rate of cartier gas, AS = entropy of adsorption, and R = gas constant. For second-order desorption with freely occurring readsorption: 2 In Tm - In b AH -
RTm
+
In
(1
-- 0 m ) 2 V s l ) m A H
2OmFR exp( AS]R) '
[21
where Vm = a m o u n t of adsorbed water at full coverage per unit volume of solid, and Om = coverage at Tin. Therefore, the slope of a plot of 2 In Tm - In b vs 1~Tin will provide the adsorption enthalpy at constant flow rate (F) and initial surface coverage for both firstand second-order desorption. This analysis assumes Om not to vary with heating rate. Journal of Colloid and Interface Science, Vnl. 105, No. i, May 1985
1 10
NISHIOKA AND SCHRAMKE TABLE IIl
this manner both physically adsorbed and chemisorbed water can be quickly measured.
Desorption Heats
SUMMARY Sample
Desorption heat (kJ/mole)
Silica Silica + A10- + Na+ Silica + A10- + Ca2+
27 22 20
Table III lists the results of this analysis for the Ludox silica, and silicas doped with sodium aluminate, and calcium aluminate. The peak maximum was measured at 10 different heating rates, from 4 to 40°C/min. Slopes were calculated from a least-squares analysis of the data. A representative plot for silica doped with sodium aluminate is shown in Fig. 6. A good linear fit was displayed for all samples. The desorption heat of about 25 kJ/mole resulting from this analysis is small and demonstrates that water desorbing at 100°C is physically adsorbed. Although most of the physically adsorbed water desorbs immediately in the dry nitrogen gas at room temperature, some water remains adsorbed onto high-energy sites and does not desorb below 100°C. A bond energy of 25 kJ/mole is equivalent to a 5-A separation between a water molecule and a single positive charge; alternatively, it is the energy of a single hydrogen bond between water and a surface silanol (26). An improved method of analysis would be to equilibrate each sample in nitrogen containing a known pressure of water vapor. Upon switching to dry nitrogen, the quantity of water desorbing at room temperature could then be measured, yielding one point on the adsorption isotherm. Variation of the water/ nitrogen ratio in the pretreatment would result in a fairly rapid method to obtain a water adsorption isotherm. Analysis of the shape of the desorption curve with time could also yield additional information about the adsorption energies at lower coverage. In Journal of Colloid and Interface Science, Vol. 105, No. 1, May 1985
Well-characterized powdered silicas were analyzed by an electrolytic thermodesorption technique. Water desorbing at room temperature is physically adsorbed and was not quantitatively measured for any samples, although a procedure is suggested that permits its measurement. Water desorbing at 100°C is attributed to molecular water strongly bound to silanols or cation-aluminate sites. Water desorbing above 150°C is considered as the product of the dehydroxylation of surface hydroxyl groups. Organosilanes, used in surface treatments of glass, appear to retard the rate of water adsorption and not to reduce the driving force for adsorption. The thermodesorption technique offers a rapid, convenient, and sensitive means for the measurement of the different species of water associated with various substrates. Because of its sensitivity, it is uniquely qualified for the study of low-area substrates. The results reported here provide the basis for future studies of low-area glasses. 11 . 0
10.5-
10.0-
I
9.5-
% ~z
9.0-
8.5-
/
6.0
'
0.20
I
0.22
'
I
0.26
'
I
O. 2 7
IO01T M
FIG. 6. Heat ofdesorpfionplot obtainedat 10 different heating rates for silica doped with sodium aluminate. These points are obtained from the peak occurring at 100°C.
THERMODESORPTION OF WATER ACKNOWLEDGMENTS The authors gratefully acknowledge the samples supplied by Dr. Ralph K. ller. The authors are also grateful to Charles K. Holloway for modifying the thermodesorption apparatus, and to Sarah Singer for measuring the wetting properties of glass fibers. REFERENCES 1. Burgman, J. A., and Hunia, E. M., Glass Technol. 11, 147 (1970). 2. McKinnis, C. L., in "Fracture Mechanics of Ceramics" (Bradt, Hasselman, and Lange, Eds.), Vol. 4. Plenum, New York, 1978. 3. Cameron, N. M., Ph.D. thesis. University of Illinois, 1965. 4. Hollinger, D. L., and Plant, H. T., Proceedings, 19th Annual Conference, SPI llA, 1 (1964). 5. Schmitz, G. K., and Metcalfe, A. G., I&EC Prod. Res. Dev. 5 (1966). 6. Metcalfe, A. G., and Schmitz, G. K., Glass TechnoL 13, 5 (1972). 7. Brelant, S., in "Treatise on Adhesion and Adhesives" (R. L. Patrick, Ed.), Vol. 2, Chap. 8. Dekker, New York, 1969. 8. Hojo H., Tsuda, K., and Koyama, M., lnt. Conf. Org. Coatings Sci. Technol. 1, 221 (1979). 9. Schrader, M. E., and Block, A., J. Polym. Sci. Part C34, 281 (1971).
111
10. Wesson, S. P., and Jen, J. S., Abstr. 5th Annu. Meeting Adhesion Soc. 9 (1982). 11. Tomozawa, M., and Takata, M., J. Non-Cryst. Solids 45, 141 (1981). 12. Pak, V. N., and Ventou, N. G., Soy. J. Glass Phys. Chem. 6, 223 (1980). 13. Yang, H., and Tyler, A., Ind. Eng. Chem., Prod. Res. Dev. 16, 252 (1977). 14. Continaud, M., Bonniau, P., and Bunsell, A. R., J. Mater. Sci. 17, 867 (1982). 15. Klier, K., et al., J. Phys. Chem. 77, 1458 (1973). 16. Hair, M. L., "Infrared Spectroscopy in Surface Chemistry," p. 80. Dekker, New York, 1967. 17. Little, L. H., "Infrared Spectra of Adsorbed Species," p. 259. Academic Press, New York, 1966. 18. Iler, R. K., "The Chemistry of Silica." Wiley, New York, 1979. 19. Young, G. J., J. Colloid Sci. 13, 67 (1958). 20. Klier, K., and Zettlemoyer, A. C., J. Colloid Interface Sci. 58, 216 (1977). 21. Miller, B., in "Surface Characteristics of Fibers and Textiles" (M. J. Schick, Ed.). Dekker, New York, 1977. 22. Armistead, C. G., et al., J. Phys. Chem. 73, 3947 (1969). 23. Bassett, D. R., et al., J. Colloid Interface Sci. 27, 649 (1968). 24. Egashira, M., et aL, Bull. Chem. Soc. Jpn. 51, 3144 (1978). 25. Egashira, M., et al., J. Phys. Chem. 85, 4125 (1981). 26. Zettlemoyer, A. C., and McCafferty, E., Croat. Chem. Acta 45, 173 (1973).
Journal of Colloid and Interface Science, Vol. 105, No. 1, May 1985
Received March 19, 1984; accepted October 4, 1984 A sensitive electrolytic technique suitable for quantitative measurement of the different states of water bound to low-area substrates (0.1 mZ/g) is described. Water is thermally desorbed from a substrate and measured by means of a P205 electrolytic cell. The method is sensitive enough to detect adsorbed water on solids with specific areas as small as 0.1 m2/g, This technique was used to study water on pure silica and silicas doped with metal cations. Strongly bound molecular water and water resulting from dehydroxylation reactions were detected and quantitatively measured. Results were consistent with previous studies of porous silicas. The presence of cation-aluminate sites on the silica surface greatly increased the amount of strongly bound water. Cation-aluminate sites are proposed as being the primary adsorption sites for water on glass. Silane coupling agents, often used to promote water resistance, therefore do not reduce water adsorption by the replacement of polar silanols with less polar organosilanes. Rather than reducing the thermodynamic driving force for adsorption, silanes appear to reduce the rate at which water is adsorbed to the glass surface. © 1985AcademicPress,Inc. posites immersed in water occurs at the glass-resin interface (8, 9). Silane coupling agents are normally coated onto glass fibers to reduce this moisture-induced degradation o f composites; the efficacy o f the coupling agent coating is inversely related to its water wettability (10). Electrical conductivity is a third property affected by water. The conductivity o f the glass surface increases with the a m o u n t o f adsorbed water (1 1, 12). Electrical conductivity presumably increases due to an aqueous layer containing dissolved surface cations. T h e same effect causes the degradation o f electrical resistance o f glass fiber-reinforced circuit boards when exposed to moisture (13, 14). M o d e r n understanding o f the interaction o f water with glass is to a great extent based on the m a n y studies o f water adsorbed on porous silica (15-18). Surface silanol groups are the primary sites for water adsorption onto silica (19). There are four to five silanol groups per square n a n o m e t e r on a smooth, nonporous, heat-stabilized, fully hydroxylated
INTRODUCTION
N u m e r o u s studies indicate that adsorbed water controls significant properties o f glass. F o r example, the strength o f E-glass fibers decreases with increasing h u m i d i t y in the testing e n v i r o n m e n t (1-3). Static and dyn a m i c fatigue o f glass also increase with increasing testing humidity (4-6). These stress corrosion effects are thought to be caused by the reaction of water with the silicate network at crack tips. This reaction causes b o n d rupture at the tip, exposing new bonds, which rupture after further water adsorption occurs. T h e process continues until the crack grows to the critical size required for spontaneous failure. Adhesive failure at the polymer-glass interface is also induced by water attack. The strength o f plastics reinforced with inorganic fillers is greatly reduced after immersion in water. Indeed, a standard test o f glass-reinforced composites is to measure shear strength o f a test composite before and after immersion in water (7). T h e failure m o d e o f corn102
0021-9797/85 $3.00 Copyright © 1985 by Academic Press, Inc. All rights of reproduction in any form reserved.
Journal of Colloid and Interface Science, Vol. 105, No. 1, May 1985
THERMODESORPTION OF WATER
silica surface. Two types of silanols exist, "isolated" and "paired." Water molecules adsorb with oxygen atoms next to silanols; the adsorption energy of a single watersilanol hydrogen bond (25 kJ/mole) is less than the adsorption energy of water on a silanol-water complex (44 kJ/mole) (20). This means that water is less strongly adsorbed on a single silanol group than on other, already adsorbed water molecules. A silica surface therefore contains clusters of water molecules hydrogen-bonded to each other and to one or several silanol groups on the surface. Although this picture is often invoked in studies of glass surfaces, there are obvious reasons why the nature of water adsorbed on glass may differ from the foregoing description. The most obvious reason is that glasses are chemically different from pure silica. The surface of a glass contains not only silanol groups, but metal hydroxy groups. Metal hydroxy groups may be expected to be highenergy sites for water adsorption, with properties different from silanols. A second difference is that commonly studied silica powders have surface roughness on the scale of hundreds of Angstroms, whereas most glass surfaces are smooth on this scale. Regions of high positive curvature on a rough surface have a greater separation of surface silanols. Conversely, on regions of high negative curvature surface silanols are closer together. These effects of curvature probably influence the adsorption of water on porous substrates but are not important factors on smooth, low-area surfaces. A third difference exists for glass fibers. Glass fibers are formed under extremely rapid cooling conditions (> 10 6 ° C / s ) . Because of this rapid cooling, the density and refractive index of glass fibers are significantly less than those of bulk glass samples of identical composition. The open structure of the glass fiber silicate network is probably reflected in the structure of the surface, and no doubt influences the interaction of glass fibers with water.
103
In spite of these obvious differences between silica and glass, the water interaction is far less studied with the glass surface than with the silica surface, for two reasons: (i) the one-component silica is simpler than a multicomponent glass, and (ii) commonly studied silica powders are high-surface-area substrates--generally, about 200 m2/g. Most commercial glasses have low specific areas. Glass fibers, for example, have specific surface areas of 0.2 m2/g. In particular, the second difficulty hinders studies of adsorbed water on glass. All forms of infrared spectroscopy are inadequate for the study of low-surface area substrates owing to insufficient sensitivity. All but the most sophisticated gravimetric systems (sensitive to +__1 ppm of weight) lack the requisite sensitivity for measurement of adsorbed water. The inadequacy of standard techniques prompted our development of a thermodesorption method with a sensitive electrolytic detector for the study of water on low-area substrates. This thermodesorption method can detect 0.1 ~g of water desorbed from a 1-g sample, which is sufficient sensitivity to detect water thermodesorbed from samples with specific areas as small as 0.1 mZ/g. Figure 1 is representative of the types and amount of water desorbed from glass fibers as measured by the thermodesorption technique. The rate of water evolved with change in temperature, normalized for area, is plotted against temperature. The quantity of water evolved within any temperature region is measured by the area under the curve between selected temperatures. Evidently water does not desorb uniformly with increasing temperature, but rather in specific temperature zones, each peak presumably corresponding to a different type of adsorbed water. The structure of the water represented by the peaks shown in Fig. 1 is not clear. Presented herein is an interpretation of some of these peaks, obtained by examination of the thermodesorption of water from wellcharacterized high-area substrates. Also preJournal of Colloid and Interface Science, Vol. 105, No. 1, May 1985
104
NISHIOKA
AND
SCHRAMKE
oc) \
~.4-
i-
~.2-
121 w m 0.1 bA
e.0
'
1 400
200
'
1
600
TEMPERATURE
'
1 800
1008
( oC)
FIG. 1. Representative desorption data of E-glass fibers. Each peak corresponds to a separate state of water associated with E,glass.
sented is a study of the effect of adsorbed cations on the interaction of water with the silica surface, and a proposed mechanism for how silane coupling agents affect the interaction between water and glass. APPARATUS
AND
PROCEDURE
.
Figure 2 is a block diagram of the thermodesorption apparatus. The analysis is as follows: a weighed sample is placed in a small tube furnace; the furnace temperature is raised at a programmed rate, usually 4°C/ min thereby desorbing water from the sample; nitrogen carrier gas flows through a sieve
dryer and into the tube, transporting the desorbed water to the electrolytic cell. The electrolytic cell, manufactured by DuPont Instruments, consists of a thin film of phosphorus pentoxide (P205), deposited between two helically wound electrodes. Water is adsorbed by the P205, and electrolyzed by current flowing between the electrodes. Electrolysis changes the water to hydrogen and oxygen which discharge through a vent with the carrier gas, coulometrically regenerating P205. The charge required to regenerate P205 is related to the quantity of desorbed water, which can therefore be calculated by moniELECTROLYTZC
CARRIERGASt FLOWMETER ~,
DRYER ~
OVEN ~
cELL t VENT
I
CONTROLLER] " COUNTER
COMPUTERl hO. 2. A block diagramof the thermodesorptionsystem.This apparatusmeasuresdesorbedwater of g r a m s a m p l e s w i t h a p r e c i s i o n to a few t e n t h s o f a m i c r o g r a m , Journal of Colloid and Interface Science, Vol. 105, No. I, May 1985
THERMODESORPTION
toting the current through the cell. The cumulative amount of desorbed water, oven temperature, and time are continuously registered by a computer. Weighed samples of powdered silicas were analyzed with this instrument. The specific surface area of each sample was determined by the one-point BET method. The amount of water desorbing from the sample holder was measured in a blank run, and subtracted from all measurements. This correction is small; values are reduced by a few percents. Areas of peaks were obtained by direct measurement of the cumulative water recorded between two temperatures. All data were normalized on the basis of surface area. The contact angle of water on glass fibers was measured on a Wilhelmy wetting balance (21). A single filament is suspended from a microbalance and is partially immersed in water. The beaker containing the water sits on a slowly ascending elevator, and the force on the filament is measured. The force measured is equal to the filament weight minus buoyancy plus the vertical component of the surface tension acting on the perimeter of the fiber. Since the filament weight, perimeter, buoyancy, and the surface tension of water are known, the advancing contact angle may be calculated.
OF
WATER
105
exchange resin; warmed to 50°C, and gelled. It is mixed with an equal volume of n-propyl alcohol, filtered, washed with alcohol, airdried, and dried in an oven at 150°C. The surface area of our sample, measured by the BET method, is 124 m2/g. A1. Aluminate-modified silica. Precipitated silica prepared in "A" is slurried in water. Sodium aluminate solution (containing NaA102 and NaOH) is slowly added to the silica suspension, stirred for an hour at 25°C, and the pH reduced to 10 with acetic acid. The amount of sodium aluminate solution added is controlled so that two aluminate groups per square nanometer of surface would be produced.
B. Sodium Form of Silica Surface Ten grams of precipitated silica "A" is slurried in 200 ml H20 and titrated with 0.1 N NaOH to pH 8.
C. Calcium Form of Silica Surface Ten grams of precipitated silica "A" is slurried in 200 ml H20 and 35 ml of 0.28 M calcium formate is added. The pH increases from 3.7 to 7.5. The pH is raised from 7.5 to 8 by adding 8 ml of 0.1 N N a O H and the silica is then filtered, washed, and dried at 125°C.
MATERIALS
E-glass fibers (Owens-Corning Fiberglas) are from commercial yarns used to reinforce electrical circuit boards. The fumed silica studied was Cab-O-Sil M5, obtained from Cabot Corporation and used as received. The precipitated and doped silicas were supplied by R. K. Iler, prepared by the following procedures:
A. Precipitated Silica An amorphous silica powder is prepared from Ludox AS40 (DuPont). Ludox AS40 is a silica sol containing particles Of 22 nm diameter stabilized with NH4OH to pH 9. The Ludox is deionized with mixed ion-
D. Sodium Form of AluminateModified Silica Half of the sample prepared in " A I " is washed to reduce the pH to 8.5, filtered, and dried at 125°C in air.
E. Calcium Form of AluminateModified Silica Half of sample "Al" is converted to the calcium form by ion exchange with an excess of calcium formate solution: 0.28 M calcium formate solution is added to the aluminatemodified silica slurry. The slurry is stirred 1 h, filtered, and washed until free from excess calcium ions. Just before filtering the solution, Journal of ColloM and Interface Science, Vol. 105, No. 1, May 1985
106
NISHIOKA AND SCHRAMKE
0.1 N N a O H is added to raise the pH from
7.2 to 8.1. The product is air-dried at 125°C. RESULTS AND DISCUSSION Cab-O-Sil M5 (fumed silica) is a powdered silica formed by oxidation of SIC14. The following properties of Cab-O-Sil M5 have been previously established: (1) a specific surface area of 200 _+ 20 mZ/g and (2) a surface concentration of 3.5 to 4.5 silanols/ n m 2. Because Cab-O-Sil M5 has been extensively studied, it was chosen as the substrate for initial study. Figure 3 contains thermodesorption data of Cab-O-Sil M5 taken under different conditions of measurement. The dot-dash curve represents an ambient sample of Cab-O-Sil placed in the oven and the instrument started immediately. The first peak at the lowest temperature appears as soon as the sample is inserted, and can only be derived from physically adsorbed water. The dashed curve results if the sample is placed in the oven and equilibrated at room temperature in the dry nitrogen gas until the initial peak disappears. Only after this first
peak disappears, usually after 2 h, is the instrument started and data recorded. Although most of the physically adsorbed water is desorbed by this treatment, a small amount of water strongly bound to highenergy sites remains. This water is represented by the peak at 100°C. The amount of water desorbing above 150°C is identical in these two curves, and can be wholly accounted for as the product of the dehydroxylation of surface silanols. The measured value of 1.7 H 2 0 / n m 2, equivalent to 3.4 silanols/nm 2, is in excellent agreement with reported values of clustered surface silanol concentration of 3/nm 2 (22). An interesting effect, not previously reported, is the desorption of water in two temperature zones. The boundary between the two desorption regions is at 450°C, the reported transition between reversible and irreversible dehydroxylation. These data indicate approximately equal amounts of each type of clustered group. The solid curve is a desorption curve of Cab-O-Sil which had first been immersed for 24 h in boiling water. It was then placed in the sample chamber, and the instrument was
0.88
,o 0j I :E v 8.86\
.J o T o E (J O. 0 4 -
v
I-
w m
0.02-
"\ 0.00 280
408
TEMPERATURE
600
800
(°C)
FIG. 3. Water desorbed from Cab-O-Sil M5. (--) Cab-O-Sil immersedin boilingwater 24 h, equilibrated 2 h in dry nitrogen 2 h before measurement. (---) Cab-O-Silequilibratedin dry nitrogen 2 h prior to the measurement. (.... ) Instrument started immediatelyafter Cab-O-Sil inserted. Journal of Colloid and Interface Science, Vol. 105, No. 1, M a y 1985
THERMODESORPTION
107
OF WATER
e.5
2 o
o
~-
0,4-
w
E
1~.3-
0.2---
I
o
N
•
e,e
l 8
~ _ _ . ~ _ ,
'
208
t
'
48{) TEMPERATURE
aa
i
J
600
880
(°C)
FIG. 4. The desorption of water from doped sificas. (A) Precipitated silica, (B) silica + sodium, (C) silica + ca|cium, (D) silica + a|uminate + sodium, (E) silica + aluminate + calcium. For clarity these curves have been offset.
started immediately. Immersing Cab-O-Sil in boiling water is known to create additional silanol groups (23). The measured value of 4.7 silanols/nm 2 following this treatment is consistent with a fully hydroxylated, annealed, silica surface (18). How certain cations, such as are present on glass surfaces, determine the amount of adsorbed water is revealed in Fig. 4. Figure 4 compares the water desorbed from a silica precipitated from a Ludox colloid (Sample A), and the same silica doped with various cations (Samples B-E). For all silicas, about five molecules of water per square nanometer are desorbed above 200°C. This is equivalent to 10 OH groups per square nanometer,
agreeing with previous studies (18) of nonheat-treated porous silicas. The water comes not only from surface-dehydroxylation reactions, but also from dehydroxylation occurring just beneath the silica surface. Table I reports the approximate atomic densities of sodium, calcium, and aluminum on the surface of the powdered silicas. These values were obtained by extraction of a known weight of each silica in a known volume of O. 1 N HC1, and solution analysis by atomic absorption. Table I reveals that only minor amounts of sodium or calcium are adsorbed by silanol sites. Indeed, the amount of cationic impurity present on the pure silica surface is about equal to the number of
TABLE I Surface C o n c e n t r a t i o n s
Sample (A) (B) (C) (D) (E)
SiO2 SiO2 SiO2 SiO2 SiO2
+ + + +
Na ÷ C a 2÷ A10- + Na ÷ A I O - + C a 2+
Na (atoms/nm2)
Ca (atoms/nm2)
0.28 0.12 0.01 2.00 0.31
0.14 0.03 0.37 0.04 1.02
AI (atoms/nm2) 0.006 0.004 0.004 i 2.16 1.73
H20 (molecules/nm 2) 1.5 1.7 1.9 3.9 4.6
Journal of Colloid and Interface Science, Vol. 105, No. 1, May 1985
108
NISHIOKA AND SCHRAMKE
cations present on the washed silicas. The amount of adsorbed sodium or calcium increases if aluminum is present on the silica surface. The ratio of sodium or calcium to aluminum is roughly 1:1 and 1:2, respectively. The measured concentration of aluminum is in good agreement with the predicted value of 2 atoms/nm 2 (see Materials). Also listed in Table I is the amount of water responsible for the desorption peak at 100°C in Fig. 4. The quantity of water desorbed is listed as an equivalent surface concentration. Surface aluminum is seen to greatly increase the amount of this strongly bound water. The data in Table I and results of previous investigators (18) lead us to the following picture of the surface of these silicas: All the silicas contain surface silanol (SiOH) groups. Surface silanol groups are only weakly acidic (18). Therefore, only a small fraction of silanols exist as the conjugate base, SiO-. It is onto SiO- groups that Na ÷ or Ca 2+ adsorb; Table I indicates that few of the available silanols exist as SiO- groups. In Samples D and E aluminum is present on the surface as aluminol (A1OH) or its conjugate base aluminate (A10-). Since A1OH is known to be strongly acidic, aluminum exists mainly as A10- groups. Therefore, each sodium cation is held by one aluminate site, each calcium cation is held by two aluminate sites. Water is adsorbed by SiOH, SiONa, and (SiO)2Ca groups. SiONa and (SiO)2Ca groups would probably take up more water; however, the surface concentration of these groups is low. Table I reveals that surface A1ONa or (A10)2Ca groups attract water to a much greater extent than surface silanol groups. Typical glass formulations contain oxides of silicon, aluminum, calcium, magnesium, and sodium. A reasonable deduction is, therefore, that the major sites on glass for adsorption of water are the cation-aluminate complexes, and not surface silanols. Organosilanes react with glass surfaces to confer improvements in surface-controlled properties. For example, silanes increase wetJournal of ColloM and Interface Science, Vol. 105, No. 1, May 1985
strength retention of a glass-reinforced plastic or reduce surface electrical conductivity of glass. These improvements are thought to be due in part to the reduction of surface energy because of the replacement of polar surface silanols with the organosilane molecule. Adsorption of water by the glass surface is thus reduced; and reduced water adsorption confers better glass properties. The results listed in Table I and Fig. 4 imply a different role for silanes. Replacement of silanols by organosilanes does little to alter the number of high-energy surface sites, since these are predominantly cation-aluminate complexes. Therefore, the thermodynamics of water adsorption are not appreciably altered by reaction of organosilanes with the glass surface. However, since organosilanes are relatively large molecules, they can cover high-energy surface sites. The function of organosilanes may therefore be to retard the rate of water adsorption and not to prevent water adsorption onto the glass surface. An alternate explanation is that silanes react with aluminate sites. In this case silanes would confer thermodynamic protection against water adsorption. The data presented in Fig. 5 support the first hypothesis. Curve A shows the amount of water desorbed from heat-cleaned E-glass fibers. These fibers were heat-cleaned at 250°C for 48 h to remove organic coatings, and stored at 20% RH for 6 months. Curve B shows the amount of water desorbed from similar heat-cleaned fibers that were further dip-coated in an aqueous organosilane solution (0.01 M a-glycidoxypropyltrimethoxysilane), cured at 165°C for 5 min, and likewise stored at 20% RH for 6 months. Presumably each of these glass fibers held their equilibrium amount of adsorbed water after 6 months. The treatment given these samples simulates the commercial treatment of glass cloth used in the reinforcement of epoxy resins. Silane-coated cloths remain significantly hydrophobic after the storage period, indicating that little degradation of the coating occurs. That there was an appreciable
THERMODESORPTION OF WATER
109
0.S o oj t vz
B.4-
b,J
8 (~ (:z (J
0.3-
B t-
0.2-
0.1-
0.0 180
280 TEMPERATURE
380
400
( o C)
FIG. 5. Water desorbed from (A) heat-cleaned glass fibers and (B) silane-treated glass fibers.
quantity of silane on Sample B was demonstrated by the wetting experiments listed in Table II. The water/glass contact angles were measured after the 6-month storage. Figure 5 reveals that the silane-treated glass desorbs as m u c h water at 100°C as the uncoated glass. These data confirm our hypothesis that organosilanes do not react with primary adsorption sites on the glass surface, and therefore act only to retard rather than to inhibit water adsorption. A searching examination of the role of organosilanes in water adsorption kinetics is clearly necessary in order to understand their role in composites. The heat of desorption of water desorbing at 100°C can be measured by heating rate experiments (24, 25). Assuming a linear heating rate and freely occurring readsorption,
TABLE II Advancing Contact Angle, Water on Glass Sample
Advancing angle
Heat-cleaned glass filaments Silane-treated glass filaments
40° 62°
the m a x i m u m temperature Tm at the maxim u m of the desorption peak is related to the heating rate (b) by 2 In T m - In b - -A- H + In (1
RTm
-
0m)2VsAH
FR exp( A S/R)
[11
for first-order desorption. Here A H = desorption heat, O m = fractional surface coverage at Tm, Vs = volume of solid phase, F = flow rate of cartier gas, AS = entropy of adsorption, and R = gas constant. For second-order desorption with freely occurring readsorption: 2 In Tm - In b AH -
RTm
+
In
(1
-- 0 m ) 2 V s l ) m A H
2OmFR exp( AS]R) '
[21
where Vm = a m o u n t of adsorbed water at full coverage per unit volume of solid, and Om = coverage at Tin. Therefore, the slope of a plot of 2 In Tm - In b vs 1~Tin will provide the adsorption enthalpy at constant flow rate (F) and initial surface coverage for both firstand second-order desorption. This analysis assumes Om not to vary with heating rate. Journal of Colloid and Interface Science, Vnl. 105, No. i, May 1985
1 10
NISHIOKA AND SCHRAMKE TABLE IIl
this manner both physically adsorbed and chemisorbed water can be quickly measured.
Desorption Heats
SUMMARY Sample
Desorption heat (kJ/mole)
Silica Silica + A10- + Na+ Silica + A10- + Ca2+
27 22 20
Table III lists the results of this analysis for the Ludox silica, and silicas doped with sodium aluminate, and calcium aluminate. The peak maximum was measured at 10 different heating rates, from 4 to 40°C/min. Slopes were calculated from a least-squares analysis of the data. A representative plot for silica doped with sodium aluminate is shown in Fig. 6. A good linear fit was displayed for all samples. The desorption heat of about 25 kJ/mole resulting from this analysis is small and demonstrates that water desorbing at 100°C is physically adsorbed. Although most of the physically adsorbed water desorbs immediately in the dry nitrogen gas at room temperature, some water remains adsorbed onto high-energy sites and does not desorb below 100°C. A bond energy of 25 kJ/mole is equivalent to a 5-A separation between a water molecule and a single positive charge; alternatively, it is the energy of a single hydrogen bond between water and a surface silanol (26). An improved method of analysis would be to equilibrate each sample in nitrogen containing a known pressure of water vapor. Upon switching to dry nitrogen, the quantity of water desorbing at room temperature could then be measured, yielding one point on the adsorption isotherm. Variation of the water/ nitrogen ratio in the pretreatment would result in a fairly rapid method to obtain a water adsorption isotherm. Analysis of the shape of the desorption curve with time could also yield additional information about the adsorption energies at lower coverage. In Journal of Colloid and Interface Science, Vol. 105, No. 1, May 1985
Well-characterized powdered silicas were analyzed by an electrolytic thermodesorption technique. Water desorbing at room temperature is physically adsorbed and was not quantitatively measured for any samples, although a procedure is suggested that permits its measurement. Water desorbing at 100°C is attributed to molecular water strongly bound to silanols or cation-aluminate sites. Water desorbing above 150°C is considered as the product of the dehydroxylation of surface hydroxyl groups. Organosilanes, used in surface treatments of glass, appear to retard the rate of water adsorption and not to reduce the driving force for adsorption. The thermodesorption technique offers a rapid, convenient, and sensitive means for the measurement of the different species of water associated with various substrates. Because of its sensitivity, it is uniquely qualified for the study of low-area substrates. The results reported here provide the basis for future studies of low-area glasses. 11 . 0
10.5-
10.0-
I
9.5-
% ~z
9.0-
8.5-
/
6.0
'
0.20
I
0.22
'
I
0.26
'
I
O. 2 7
IO01T M
FIG. 6. Heat ofdesorpfionplot obtainedat 10 different heating rates for silica doped with sodium aluminate. These points are obtained from the peak occurring at 100°C.
THERMODESORPTION OF WATER ACKNOWLEDGMENTS The authors gratefully acknowledge the samples supplied by Dr. Ralph K. ller. The authors are also grateful to Charles K. Holloway for modifying the thermodesorption apparatus, and to Sarah Singer for measuring the wetting properties of glass fibers. REFERENCES 1. Burgman, J. A., and Hunia, E. M., Glass Technol. 11, 147 (1970). 2. McKinnis, C. L., in "Fracture Mechanics of Ceramics" (Bradt, Hasselman, and Lange, Eds.), Vol. 4. Plenum, New York, 1978. 3. Cameron, N. M., Ph.D. thesis. University of Illinois, 1965. 4. Hollinger, D. L., and Plant, H. T., Proceedings, 19th Annual Conference, SPI llA, 1 (1964). 5. Schmitz, G. K., and Metcalfe, A. G., I&EC Prod. Res. Dev. 5 (1966). 6. Metcalfe, A. G., and Schmitz, G. K., Glass TechnoL 13, 5 (1972). 7. Brelant, S., in "Treatise on Adhesion and Adhesives" (R. L. Patrick, Ed.), Vol. 2, Chap. 8. Dekker, New York, 1969. 8. Hojo H., Tsuda, K., and Koyama, M., lnt. Conf. Org. Coatings Sci. Technol. 1, 221 (1979). 9. Schrader, M. E., and Block, A., J. Polym. Sci. Part C34, 281 (1971).
111
10. Wesson, S. P., and Jen, J. S., Abstr. 5th Annu. Meeting Adhesion Soc. 9 (1982). 11. Tomozawa, M., and Takata, M., J. Non-Cryst. Solids 45, 141 (1981). 12. Pak, V. N., and Ventou, N. G., Soy. J. Glass Phys. Chem. 6, 223 (1980). 13. Yang, H., and Tyler, A., Ind. Eng. Chem., Prod. Res. Dev. 16, 252 (1977). 14. Continaud, M., Bonniau, P., and Bunsell, A. R., J. Mater. Sci. 17, 867 (1982). 15. Klier, K., et al., J. Phys. Chem. 77, 1458 (1973). 16. Hair, M. L., "Infrared Spectroscopy in Surface Chemistry," p. 80. Dekker, New York, 1967. 17. Little, L. H., "Infrared Spectra of Adsorbed Species," p. 259. Academic Press, New York, 1966. 18. Iler, R. K., "The Chemistry of Silica." Wiley, New York, 1979. 19. Young, G. J., J. Colloid Sci. 13, 67 (1958). 20. Klier, K., and Zettlemoyer, A. C., J. Colloid Interface Sci. 58, 216 (1977). 21. Miller, B., in "Surface Characteristics of Fibers and Textiles" (M. J. Schick, Ed.). Dekker, New York, 1977. 22. Armistead, C. G., et al., J. Phys. Chem. 73, 3947 (1969). 23. Bassett, D. R., et al., J. Colloid Interface Sci. 27, 649 (1968). 24. Egashira, M., et aL, Bull. Chem. Soc. Jpn. 51, 3144 (1978). 25. Egashira, M., et al., J. Phys. Chem. 85, 4125 (1981). 26. Zettlemoyer, A. C., and McCafferty, E., Croat. Chem. Acta 45, 173 (1973).
Journal of Colloid and Interface Science, Vol. 105, No. 1, May 1985