CHAPTER 6 The Surface Chemistry of Silica - cntq.gob.vecntq.gob.ve/cdb/documentos/quimica/224/6.pdf · Barby (I 3) contributed a broad survey of ... Various aspects of the surface

CHAPTER 6

The Surface Chemistry of Silica

The properties of amorphous silicas of high specific surface area, from the smallest colloidal particles to macroscopic gels, depends largely on the chemistry of the surface of the solid phase. This is of practical importance i n the technology of cracking catal j sts, mineral processing, ceramics, and adsorbents. I t is also directly involved in the manufacture and use of siliceous fillers and thickening agents in organic systems, including paints, inks, elastomers, and lubricating greases. Within the last half- centry, a much clearer picture of the nature of siliceous surfaces has been obtained, and new products based on chemical modifications of these surfaces have been developed.

Haber ( I ) expressed the view that an atom a t the surface of a solid is only partly saturated on its inner side and therefore possesses “residual valences” on its outer side. Langmuir ( 2 ) extended the idea that these “residual valences” were responsible for the adsorption of foreign atoms or molecules on the surface and formulated the following equation for adsorption equilibrium on the assumption that the law of mass action holds:

I P x / m ab b

- + - P - =

where p is the vapor pressure of adsorbate. x / m is the amount of adsorbate adsorbed per gram of adsorbent, a is a constant, and b is proportional to the maximum number of adsorbate molecules which can be accommodated per unit area of surface. This assumes that when the surface is covered by one layer of adsorbed molecules no second layer is adsorbed.

Where there is a strong chemical interaction between the adsorbed molecule and the atoms in the surface, a complete monomolecular layer is formed even when there is only a relatively low concentration of adsorbate in the liquid or gas phase adjacent to the surface. Further adsorption of a second layer can occur only through the interaction of weaker secondary forces extending from the surface beyond the first adsorbed layer. I n such cases, the formation of the first layer is termed “chemisorp-

622

Reviews and Summaries 623

tion” and follows the above Langmuir adsorption isotherm, and the formation of a second layer, i f it occurs, is termed “physical adsorption.” However, i f there is only weak interaction between the atoms in the surface and the adsorbate molecules in the first layer, a second layer may begin to form even before the first is completed; all the adsorption in this case is termed “physical adsorption.”

The key characteristic of the siloxane (SiOSi) surface of SiO, is that the so-called “residual valences” react with water so that at ordinary temperature the surface becomes covered with silanol (SiOH) groups. This chapter deals with chemisorption of molecules on the siloxane or silanol surfaces of silica so that a single layer of molecules is adsorbed involving covalent or ionic binding with the surface atoms. Hydrogen bonds are also formed between polar atoms, such as an oxygen atom of an ether, and the hydrogen atom of silanol groups on the surface. In all cases there is a binding force between a specific atom of the adsorbate and an atom on the surface so that once the silica surface becomes covered, no second layer is adsorbed.

NOTE. Where silicon, Si, is not shown with a valence of four, such as SiOH or Si,OH, it should be understood that the silicon is on the surface of silica and is linked through three siloxanes, SiOSi, to three adjacent silicon atoms.

REVIEWS AND SUMMARIES

Surveys of the surface chemistry of silica have continued to appear since 1950. Iler (3) in 1955 emphasized the behavior of the silanol groups on the surface, especially their dehydration and chemical interactions including esterification and chemisorption of organic molecules.

As early as 1936, Kiselev proposed that the active surface of silica gel was covered by O H groups bound to the SiO, skeleton (4), and in 1940 Carrnan (5) recognized that on the surface of anhydrous SiO,, water adds to the oxide to create silanol groups. In 1959, Belyakova et al. ( 6 ) reviewed their contributions during the previous 10 years showing that the surface silanol groups are the points at which water molecules are adsorbed, and that many organic molecules with polar groups are adsorbed and held to the S iOH groups through hydrogen bonding. Furthermore, they presented data showing that when the adsorption isotherms of water are expressed as micromoles per square meter, widely different forms of amorphous silica with hydrated surfaces adsorbed the same amount of water at the same low partial pressure. Thus the nature of the underlying silica solid phase makes very little difference; all the silica surfaces are alike provided no rnicropores are present. The concentration of SiOH groups on the silica surface seemed to be I I f I micromoles m-z, or 6.6 O H groups nrn-*. An extensive bibliography up to 1959 was included.

In 1966 Boehm (7, 8) published two extensive surveys on surface groups on different solids including silica. Data on the number of OH groups per square nanometer by a variety of methods were compared and esterification and other surface ractions were reviewed. In 1967, Hair (9) described the chemistry of silica sur-

624 The Surface Chemistry of Silica

faces including surface hydration and reactions, particularly from the standpoint of infrared absorption studies. In 1968, Snyder ( I O ) discussed the structure of the surface of silica with emphasis on the relation to adsorption characteristics and use of porous silicas in chromatography. I n 1970, Okkerse ( 1 I ) in an extended review of the preparation and properties of porous silica described the hydroxylated silica surface and its structure, with the view that at the surface there is probably only one OH per silicon atom, and reviewed data on dehydration. I n 1972, Snoeyink and Weber (12 ) covered the same ground as Boehm, adding newer references. I n 1976, Barby ( I 3) contributed a broad survey of “silicas”, that is, gels and powders, with a detailed discussion of the nature of the surface silanol groups.

A general survey of the structure and chemistry of oxide surfaces in relation to adsorption was published by Kiselev i n 1971 (14). Various aspects of the surface chemistry of silica in contact wi th water have been surveyed in a general article on the colloid-water interface by Wiese et al. (15).

With these numerous reviews, the question can be logically asked whether still another review here is justified. However, it is evident that most of the emphasis has been on the nature of the hydration of the surface and in no case have the different aspects of the surface chemistry, including behavior i n contact with water and organic compounds, been made available in one convenient place.

N A T U R E OF T H E SILICA S U R F A C E

Many diverse chemical and physical methods for examining the composition and structure of the surfaces of solids are now available. These have been described in some detail by Kane and Larrabee (16). Many of these are more suitably applied to macroscopic surfaces than to powders and colloids, but among them are some that have not yet been used in studying silica. In this chapter, emphasis is placed on what has been learned about the surface chemistry rather than on the methods used, since these are available in cited references.

A very effective summary of the nature of the silica surface as known in 1965 was presented by Hockey (17). A more recent survey is that of Boksanyi, Liardon, and Kovats (18) with regard to attaching organosilyl groups to the surface.

Structure of the Underlying Silica

Though it has been found that different forms of silica appear to act alike in regard to the adsorption of water (6), it would certainly be surprising i f the SiOH groups on clean surfaces of all the different crystalline and amorphous forms would behave exactly the same. Indeed Stober (19) has shown that the SiOH groups on the rare, extremely dense form of silica, stishovite, i n which each silicon is coordinated with six oxygen atoms, act as those on alumina do as far as hydrogen bonding is concerned. Stishovite adsorbed polyvinylpyridineN-oxide only very weakly whereas the other forms of silica surfaces adsorbed it strongly. Since there are some density

Nature of the Silica Surface 625

differences, for example, between quartz and amorphous silicas, we would expect some small differences to exist. However, work by Meyer and Hackerman (20) seems to show that particles size or radius of curvature of the surface may be a more important variable than differences between the amorphous and ordinary crystalline states of silica.

Definition of Surface

The term surface will be understood to mean the boundary of the nonporous solid phase. By custom, the “surface” usually means the boundary that is impervious to nitrogen, the adsorbate most commonly used to measure the surface area. However, there are micropores into which water but not nitrogen can penetrate. Since the area in micropores is difficult t o define, the “surface” will generally be understood to mean that which is measured by the usual BET method of nitrogen adsorption.

Even a small amount of impurity, i f located on the surface, can greatly modify surface properties. For example, if 100 ppm of sodium impurity in a silica gel having a specific surface area of 200 m3g-’ were all located on the surface, there would be one sodium atom per 100 nm2. The atoms would be only I O A apart if evenly distributed. Fowkes and Burgess (21) showed that the surface of silica traps sodium atoms. Even the purest quartz collects I O l 3 c m - 2 atoms within 100 A of the surface. This creates a negative oxide ion charge a t the surface. If the surface layer is etched off with H F it becomes uncharged, but after some weeks a t room temperature more sodium comes from the interior. A bulk concentration of 2 ppm Na can result in 2% sodium on the surface after annealing a t 1000°C. Since silica containing 2 ppm is considered extremely pure, it can be seen that in most gels and precipitates, sodium may play an unrecognized role.

The Hydroxylated Surface

Since the silicon atoms on the surface of amorphous silica are, by definition, not in an exactly regular geometrical arrangement, it is obvious that the hydroxyl groups attached to these silicon atoms will not be exactly equidistant from each other (Figure 6.1). They are therefore not all equivalent either in their behavior in adsorption or in chemical reactions.

Also, in aqueous solution it can be imagined that extra monosilicic acid molecules might condense with the surface to give attached silicon atoms with two or even three attached hydroxyl groups, a s shown in Figure 6.1 a t F and G . Although these groups have been postulated to explain certain data, it is probable that they condense further leaving only S i O H groups on the dried surface.

On silica dried from water, the hydrogen-bonded water molecules as in Figure 6. I come off in vacuum a t ordinary temperature or at 150°C in the atmosphere.

The silanol number, that is, the number of hydroxyl groups per unit surface area, has been a matter of much research and discussion. It was observed by Belyakova et

62 6 The Surface Chemistry of Silica

' 1'0 0' I 'C " 0 0 0

Figure 6.1. Postulated types of hydroxyl groups on the surface of amorphous silica. A , vicinal hydrated; B , vicinal anhydrous; C, siloxane-dehydrated; D, hydroxylated surface; E , isolated; F , geminal; G. vicinal, hydrogen bonded, Nore: F and G probably do not actually exist on a dried surface.

al. (6) that the silanol number was about the same on different types of silica. On the other hand, quite diverse results were reported by other workers. As pointed out by Dubinin. Bering, and Serpinskii ( 2 2 ) in 1964, in the surface of different samples of amorphous silica made in different ways, the packing of the tetrahedra of oxygen a toms, each containing a silicon atom, is not regular and can vary. When silica particles a re formed in water the number of O H groups can be affected in several ways as shown in Figure 6.2:

"Buried" uncondensed S i O H groups below the surface of the siloxane network give rise to distortions a t the surface and when deeper can be removed only at high temperature. Some S i O H groups can be located just within the surface and increase the average packing density of OH groups per unit surface area as a t B. On particles with small radii, the curved surface with a positive radius of character tends to hold S i O H groups farther apart so they can form fewer hydrogen bonds between them, making them more readily removable at higher temperature, as at C, as compared with D.


A

D

Figure 6.2. Factors affecting the number of hydroxyl (silanol) groups i n and on a silica particle formed i n water. A , buried hydroxyl distorts the surface, facilitating presence of geminal silanol groups; B , hydroxyl group j u s t below surface is more stable and increases number per uni t area; C, hydroxyl groups on very small (under 100 nm) particles are less crowded per u n i t area; D, hydroxyl groups on a larger particle are closer together and can form more stable hydrogen-bonded pairs.

4. Conversely, in crevices a t points of contact between particles where there is a negative radius of curvature, for example, in micrbpores, the surface O H groups are brought closer together by the curvature and dehydration is more difficult.

5 . In particles formed at low temperature and low pH by aggregation of still smaller particles there are micropores or submicropores as described in Chapter 4, with hydroxylated surfaces in which water is strongly retained. Such surfaces are not measured by the BET method because nitrogen cannot penetrate so that the concentration of OH per square nanometer appears abnormally high.

State of Water at the Hydroxylated Surface

In a detailed study of the behavior of water at the silica surface, dehydrated at 7OO0C, Klier and Zettlemoyer (23) have shown that water sits “oxygen down” on the S iOH groups, at least a t the beginning stages of adsorption. Then hydrogen- bonded clusters of H,O molecules may be formed even before all the SiOH groups have adsorbed H,O molecules to form Si ,OH:OH, groups. This means that on isolated SiOH groups water is less strongly absorbed than on top of other water molecules already adsorbed.

Si,OH + H,O = %,OH : OH, Si ,OH:OH, + x H,O = Si,OH:OH,(OH,)

6.0 kcal mole-’ 10.5 kcal mole-’


where Si, represents a silicon atom at the surface. Water is almost a s stable in a cluster of five or six molecules as in bulk.

On a surface that is fully hydroxylated, the H,O molecules first cover all SiOH groups because of multiple hydrogen bonding. Also, when colloidal particles are in water and fully hydroxylated. there is evidence from viscosity measurements by Dalton and ller (24) that there is a monolayer of water molecules more or less immobilized at the uncharged S iOH surface by hydrogen bonding.

The idea that the immobilization extends for several molecular diameters into the aqueous phase has little basis in fact. Otherwise, the viscosity would be greater than calculated for an adsorbed monolayer. Also, Lyklema (25) concluded from studies of AgI dispersions that the electrokinetic potential coincided with the outer Helmholtz plane potential and "there are no thick stagnant water layers at the i n - terface."

Experiments by Peschel and Aldfinger ( 2 6 ) , in which the viscosity of water was measured between spherical and flat fused silica surfaces. detected an increase in viscosity when the surfaces were as far apart as 160 nm. However, the water was not at pH 2. the isoelectric point of silica, and the silica surface even in pure water must develop a negative charge and ionic double layer, which was not taken into account.

Within the first t w o molecular layers of water at the surface, viscosity is no doubt quite high because the first layer of water is largely in an equilibrium hydrogen- bonded state with Si,OH groups. Furthermore, this viscosity is probably shear-sensi- tive because of the time factor involved in making and breaking the hydrogen bonds.

The surface of crystalline quartz is quite different from amorphous silica. I t shows certain peculiarities such a s adsorbing monosilicic acid (Chapter I ) . Such unusual adsorption on a surface may be due to special geometric fitting of adsorbate to adsorbent and does not necessarily relate to the ordering of water molecules. Nevertheless, there does seem to be considerable evidence that some types of surfaces may cause adjacent water molecules to assume some degree of order. If the solid surface is crystalline and the surface atoms are in an orderly arrangement one can then imagine an orderly arrangement of adsorbed water molecules which, through hydrogen bonding. could cause an extension of such order (ice-like structure) for some distance into the solution. Drost-Hansen ( 2 7 ) summarized the evidence for such a phenomenon. However, there is no convincing evidence that such ordered structuring of water can be induced by the surface of amorphous silica, on which the SiOH groups have no regularity beyond two or three silanol sites.

The state of water adsorbed i n a film only a few molecules thick may represent a different situation. Plooster and Gitlin (28) measured the heat capacity of such films of water adsorbed on silica. In this situation there is also a liquid-vapor interface with a surface tension that is absent when the silica surface is immersed in bulk water. With thin films only 50-75 A thick, a heat capacity maximum is observed below 0°C and the ice-water transition is suppressed. In such thin films lying between solid-liquid and liquid-vapor interfaces, the silanol surface seems to have an effect on water structure as far as several molecular diameters from the surface ut around 0°C where srrong hydrogen bonding is known to occur.

It would be of interest to study the thermal expansion of water a t around 0°C i n a 30-40% sol of 100 A silica particles at the isoelectric point a t p H 2 where only


hydrogen bonding is involved. In this case all the water molecules are within 100 A of the silanol surface. If the influence of the silanol surface extends very far, the maximum density should no longer be a t 4°C. Silica has essentially a zero expansion coefficient and so can be ignored. Studies of the properties of water i n the adsorbed state on the silica surface include nmr effects (29, 30), heat content (31), entropy measurements (32), and vapor pressure (33).

Electrical Conductivity of the Surface

In water suspension free from soluble electrolytes, the surface of silica acts as a con- ductor to the degree that the surface is charged and counterions are available to provide charge transfer. Obviously, an aluminosilicate-coated surface with H i counterions would provide highest conductivity similar to a cation-exchange resin in H' form. However, the conductivity is not due solely to the mobile ions in the diffuse double layer, as once proposed by Bikerman (34). Kittaka (35) concluded that conductivity was due to ion transfer not only in the diffuse double layer but also in the "fixed" double layer. However, since the pH was not measured and since the H + ion has a far greater mobility than any other ion, there is some possibility that i t played an unrecognized role. It so happened i n Kittaka's tests that highest conductivity was noted i n solutions providing the most H + ions, namely, AICI, and especially Th(NO,),, which give a lower pH.

On the dry silica surface an interesting observation of the electrical conductivity was made by Muroya and Kondo (36). They prepared a strong wafer of silica gel I mm thick with an area on one side of 3 cmz. This was coated with gold electrodes on each side and the conductivity measured from 150 to 650°C. The surface area of the pores was 650 m z g - ' and porosity 0.33 ml g- ' . Conductivity was clearly due to the protons of the S O H groups. The conductivity based on the area within the pores ranged from about ohm- ' a t 25-160°C on the dried SiOH surface. The sample a t 600°C had become surface dehydrated, and conductivity was IO-'* o h m - ' but on cooling decreased to a much lower value than previously. The extrapolated conductivity at 25°C then appeared to be 10-30-10-'0 ohm-' . The silanol surface is obviously the zone of conductivity.

t o

Distinguishing Adsorbed Water from Silanol Groups

Determining the number of silanol groups on the surface requires that a distinction be made between S iOH groups and physically bound or hydrogen-bonded water. Conceivably, some water molecules might be so strongly held that a t the temperature required for desorption, some loss of silanol groups may also occur through dehydration. Probably this occurs a t least in micropores where the energy of adsorption of water is very high and the silanol groups are most readily dehydrated.

De Boer and Vleeskens (37) found that silica dried in air a t 120°C has lost all physically adsorbed water but a t llO°C still retains water if the air is humid. This assumes, of course, that no micropores are present; otherwise adsorbed water can be retained in micropores up to 180°C even though the surface hydroxyl groups begin to be lost (38). They found that on surfaces which have been heated or autoclaved,


or heated and rewetted repeatedly, the number of O H groups per square nanometer tended toward a value of 4.5-5.0 after drying at 120°C. However, with “virginal” silica gels and hydrated silicas from minerals which had never been heated, much higher hydroxyl contents per unit area were calculated. Others have also observed that even on wide-pored gels or colloidal silica that has been dried at up to 150°C but never further heated, the total water content equals 6-8 O H

Noll. Damm, and Fauss (44) reported that the Karl Fischer reagent reacts with water but not with silanol groups of silica gel. By subtracting water determined in this way from the total evolved a t 1100°C from silica gel they concluded the surface contained 5 . 2 O H nm-z . ller has observed that on samples of silica gel dried at 170°C, containing total water equal to 8 O H about 2.3 reacted rapidly with the Karl Fischer reagent leaving 5 . 7 O H nm-* by difference, after which there was a long slow reaction. Kellum et al. (45) reported the same phenomenon and extrapolated the later. linear part of the curve to zero time to calculate the water equivalent. T o determine S iOH groups, Kellum and associates developed a titration procedure using lithium aluminum di-n-butylamide as the base in dimethoxyethane and using N-phenyl-p-aminoazobenzene as indicator while titrating in 1 : 1 tetrahydrofuran and pyridine.

Reaction of SiOH groups with diborane is another approach. Van Tongelen, Uyt- terhoeven, and Fripiat (46) compared this method with the Grignard reagent and found the latter much faster. Fripiat et al. (47) used the Grignard reagent, measuring methane evolved from the SiOH groups, and found 4.6 OH nm-z. This was confirmed by Morimoto and Naono (48). who found good agreement with ignition loss, indicating the reagent reacts with all O H groups.

The prevailing view is that low temperature drying under vacuum is the only way to removed adsorbed water without the possibility of disturbing hydroxyl groups. Bermudez (49) has shown by nmr that at llO°C in 6 hr not only is the water removed but also some silanol groups from the surface, a s well a s some internal water.

Wirzing (50) used the 5265 cm-I band to identify water and clearly distinguish i t from various types of hydroxyl groups. This band appears when H,O is bonded to an S iOH groups and so also permits the determination of OH groups when the silica is permitted to adsorb water. Adsorbed water was distinguished from S iOH groups using infrared adsorption also by Baverez and Bastick (51).

As will be discussed later, the main problem is not distinguishing between water molecules and S iOH groups that are on the surface, but rather between these surface species and water of S iOH groups held within the silica particle that do not come out except a t high temperature.

Lange ( 5 2 ) claims that on hydroxylated silica there are two kinds of adsorbed water, one that is desorbed during drying at 25-105°C and the other that is evolved a t 105-180°C. H e calls the first type “physically adsorbed” and the second type “hydrogen bonded.” The physically adsorbed water requires an activation energy of 6.6-8.2 kcal mole-’ for removal; the second requires IO kcal mole-’. Upon read- sorption the hydrogen-bonded water follows a Langmuir isotherm; the physically adsorbed then follows a type I 1 isotherm with monolayer coverage at p/po = 0.18.

(Table 6.1).

Table 6.1. Hydroxyl Croups per q u a r e Nanometer by Different Methods

Specific Drying Method Surface

Area Method of Ref. Type of SiO, "C Atm. Time (mZ g-') Determination OH nm-*

Shapiro and Weiss (39) Gel 155 - - 300 Diborane 7.9 Lowen and Broge (40) Deionized colloid 110 Air 16 hr 182 Ignition, I I00"C 10.0 Iler (41) Deionized colloid I20 Air 4days 160 Ignition, I 100°C 8.0 ller (41) 169 Ignition. 1IOO"C

150 Air 2 hr 160 Ignition, 1 I00"C Deionized colloid 120 Air 4days

Erkelens and Linsen Mallinckrodt I20

Mallinckrodt, heated 120 to 800°C. soaked in water, 5x

(42) I7 hr 59 1 Ignition, 1200°C

17 hr 166 Ignition. 1200°C

Infrared

Infrared

8. I 6.3 9.6 8.6 5.2 4.0

Davydov, Kiselev, Rehydrated gels, aero- I50 - - 39-750 Combination methods 5.0-5.7 and Zhuravlev (43) gels, Aerosils

E e

63 2 The Surface Chemistry o f Silica

I t is probable that the hydrogen-bonded water represents the adsorption of water only on those silanol groups that are not strongly hydrogen bonded to adjacent groups.

Nmr has been used by Mank, Baran. and Yankovskaya to distinguish labile protons and O H groups on the silica surface and OH groups within the particles (53a). The spacing of SiOH groups was estimated to be 5.0-5.4 A, which would correspond to about 3.9 OH nm - * i f distributed on a hexagonal grid. or 3.7 OH nm - z i f on a square grid. Kvlividze (53b) in 1964 listed prior references and published nmr studies on wide-pored gel dehydrated in vacuum a t 200°C. The adsorption of water was followed and measurements made a t low temperature from 83 to 273'K.

Internal H.rdro.vj,l Groiqr and Trapped Water

As investigations continued it became clear that i n almost all kinds of amorphous silicas there are silanol groups not on11 on the surface but throughout the particle structure. These trapped silanol groups arise i n different ways depending on the process by Nhich the particles were formed.

In pyrogenic silicas, as discussed in Chapter 5. the spherical particles of the order of 10-20 nm in diameter were formed by aggregation of elementar! particles only 1-2 nm in size condensed from the flame. These contain some surface SiOH groups which become locked within the final particle which sinters nearly to theoretical densit) before it cools.

In silica gel made from sodium silicate the primar) polysilicic acid particles are usuall! 2 - 3 nm i n size when they aggregate and i n the course of "aging" the gel some SiOH groups become trapped as particles coalesce. Cornier. Baverez, and Bastick (54) showed that surface groups can indeed be trapped when silica gel is highl! compressed.

In the case of particles grown by autoclaving in water at high temperature i t might be thought that the silica network structure would densify and rid itself of internal S iOH groups. Houever. Davqdov and Kiselev (55) showed that, i n fact. water remains within the structure. When driven out i t may leave micropores as discussed i n detail b! Kiselev (14).

When colloidal particles are formed in solution by the "buildup" process as described i n Chapter 5, the pH is about 9 and a certain small number of sodium ions are adsorbed at the same time that silica is deposited on the growing particles. These are associated with u a t r r and even after they are removed by prolonged acid treatment and ion exchange, 3 few strongly held water molecules or silanol groups remain deep in the structure,

Still another source of internal S iOH groups is diffusion of H,O into the solid SiO, structure. Doremus (56) states that just as H,O diffuses into vitreous silica at high temperature. i t can penetrate amorphous SiO, to a distance of 150 A where a small amount can exist as S iOH pairs.

I t was rather unexpected to find that pyrogenic silicas which had condensed from a flame still contained micropores. Thompson ( 5 7 ) showed by infrared studies that there is a small but significant volume of micropores ranging up to 0.01 cm3 g I


which hold water very strongly. Water was removed from the outer surface of silica particles by pumping under high vacuum at 26"C, but not from capillaries where water is probably strongly hydrogen bonded to opposite sides of capillary walls. This water continues to come off up t o 400°C but above this temperature the capillaries become closed by formation of Si-0-Si bonds and are not subsequently opened up by rehydration.

Some data by Uytterhoeven, Hellinckx, and Fripiat (58) revealed that on pyrogenic Aerosil at IOO"C, only 35% of the SiOH groups were on the surface, and on a xerogel only 61 %. O n the other hand, when silica particles are grown gradually by the deposition of soluble silica in hot alkaline solution it would be expected that fewer hydroxyl groups would be trapped in the structure. However, no real evidence has been advanced to show that this is so. It could possibly be shown by comparing total hydroxyl groups versus surface area for particles of different sizes, all made in the same way. The O H content should extrapolate to the value for the internal concentration with very large particles where the surface becomes negligible.

When silica is autoclaved in water a t high temperature water diffuses into the solid phase to reach an equilibrium concentration. According to Chertov et al. (59), after being autoclaved silicas with surface areas of 21-600 mz g - * contain 1 meq of internal water per gram of SiO,, which would be about 0.9%. During autoclaving the surface area was reduced but the surface hydroxyl concentration remained at 5.6 O H nm-,. However, the bulk water content of0.9% was constant (60).

From this it is obvious that in calculating the number and behavior of silanol groups on the surface, the internal groups must not be included. Unfortunately this has not been done in many investigations. It was only after it was realized that the internal bound water could be driven out at high temperature and the resulting micropores closed that silica samples were obtained on which dehydration and rehydration studies could be carried out reproducibly. Even then for unequivocal results it is necessary to measure the surface h)droxyl groups by a method that responds only to the SiOH groups on the surface.

The presence of internal H,O (or S iOH groups which probably occur i n pairs as H,O hydrolyzes internal Si-0-Si linkages) also has confused the infrared studies. Both internal and external SiOH groups give absorption bands. In particular, the assumption of geminal hydroxyl groups [=Si(OH),] on the surface to explain the high liquid water content of previously unheated silica is now debatable. Internal uncondensed silanol groups appears to be a more likely explanation.

Hydroxyl Groups per Square Nanometer

It now seems generally agreed that on the smooth, nonporous heat-stabilized amorphous silica surface that is fully hydroxylated there are 4-5 SiOH groups nm-, (100 A,) which remain when the sample is dried at 120-150°C.

This value is cited in most of the summary papers (4-13). Such a surface is obtained after heating the powder or gel t o elevated temperature to drive off internally held water and O H groups and close the pores, then rehydrating in water.

On rehydrated pyrogenic silica Armistead et al. (61) showed that out of 4.6 O H


nm-2 on the surface, I .4 i 0. I were free hydroxyl groups not hydrogen bonded to neighbors, and 3.2 * 0.1 were mutually hydrogen-bonded pairs. In addition, there were 1.6 O H n m - ? internal S iOH groups which were irreversibly removed upon annealing.

Type of Group Identifying

Absorption Band (cm - I )

Single OH on surface 3750 Paired OH’s on surface 3 540 Internal OH’s. hydrogen-bonded 3650 Molecular H,O 3400, 1627

Deuterium o f heavy water exchanges only with the hydrogen atoms of hydroxyl groups on the surface but not with those in the interior of silica. Zhuravlev et al . (62) compared the weight loss on ignition with S iOH groups on the surface as revealed by deuterium exchange on silica with area of 340 m 2 g - l and I I O 8, pore diameter. They found more total water than was bound a s surface hydroxyl groups, which amounted to 5.2 OH n m 2 .

Results by deuterium exchange as well as those by chemical methods using LiCH, or MgCHJ have been summarized by Davydov, Kiselev, and Zhuravlev (43) and are included i n Figure 6.3 and Table 6.1 . Later Agzamkhodzhaev et al . (63) used this method on a series of hydroxylated gels, aerogels, and other silicas of different pore sizes and found the surface concentration ranged from 4 . 2 to 5.7 O H nm-,. Using a similar method Madeley and Richmond (64) found that on four silicas with surface areas of 374-701 mz g-I the range of O H nm-* was only 4.27-4.63, with no trend noted. This appears t o be the method of choice since exchange is apparently limited to the surface. The surface area is best determined by krypton adsorption using the BET method since results are less affected by the nature of the surface than when nitrogen is used. The exchanged H,O-D20 mixture is analyzed by mass spec- trometry.

Wirzing (50, 65) showed that the number of O H groups can also be determined by adsorbing water on the OH groups and measuring the resulting unique infrared absorption band a t 5265 c m - I . A well-sintered sample of silica powder (Mallinckrodt) that had been repeatedly heated to 800”C, cooled, and rehydrated eight times, with an area of 166 m 2 g - l and obviously free from micropores, was examined. After drying at 120°C i t had 5.2 OH nm- , by ignition and 4.0 by this infrared method. On an aerogel the value was 4 .4 .

O n silica that has nor been heat treated, according to De Boer and Vleeskens (66), there were 6 .2 OH nm - 2 but after repeated wetting, heating, and drying, the surface became smoother on an atomic scale so that the value 4 . 6 0 .2 OH n m - 2 was reached. On the dry, fu l ly hydroxylated surface about half of the silanol groups are present a s hydrogen-bonded pairs. In such a random distribution of surface groups it is evident that many SiOH groups will remain single and unpaired (67).

The hydroxylated surface, with adsorbed water, has two types of H,O and two of S iOH, according to Anderson and Wickersham (68). A monomeric H,O molecule


6

5

N 4

5 z % 3

r 0

2

I 1

0 100 300 500 700

TEMP "C Figure 6.3. Dehydroxylation of the silica surface versus temperature: Shaded area: range of data on a variety of types of silica by Davydov, Kiselev, and Zhuravlev (43). Dehydration of annealed (7OOOC) and rehydrated silica. A , in air; B, in vacuum. [Based on Taylor, Hockey, and Pethica (84).] Broken line C, data on silica, not annealed. [From Uytterhoeven et al. (83). Fripiat et al. (47), Zhuravlev and Kiselev (85) . and Taylor, Hockey, and Pethica (84,86).]

may be bonded to a single S O H . A cluster of H,O molecules may form a hydrogen- bonded network. There can be nonbonded SiOH groups and S O H groups bonded to H,O molecules.

Proton nuclear magnetic resonance (pmr) was used by Bermudez (69) to determine the number of surface silanol groups even when the surface was covered with up to three monolayers of adsorbed water. The gel had a surface area of 800 m2 g - ' and the silanol concentration was found to be 7 x I O - 8 mole m-2 or 4.2 O H nm-2.

As will be discussed in connection with dehydration and rehydration of the silica surface, much of the past data showing more than 4.6 OH nm-z is probably due to the presence of internal SiOH groups which were assumed to be on the surface.

Determining the number of O H groups on the surface by chemical methods has given inconsistent results, according to Boehm (8). He points out that by most chemical reactions, for example, esterification or chlorination, only about half of the number of O H groups appear to react, out of the 5 O H nm-, known to be present by determination of active hydrogen or water loss.

It may be that these substituent groups are too large to be accommodated at every former SiOH site. When the substituent group such as ethoxy or chlorine is slightly larger than OH, only about every other O H group can be substituted since the posi- tions of the underlying silicon atoms are fixed. However, the methyl group is small enough to fit in. As reported elsewhere, between four and five methoxy groups per


square nanometer can be attached to the surface. This is also w h y the use of CH,Li and CH,Mgl by Fripiat et al. (47) has been successful. Another methylating agent is the dimethylzinc complex with tetrahydrofuran (ZnMe, . ZTHF) used by Hanke (70), who measured the OH concentration on the surface of an aerosil by measuring the volume of CH, evolved.

The only other negative substituent small enough to replace all the O H groups is the fluorine atom, but the S iF surface is not easy to develop without attacking the SiO, structure.

Triisobutglaluminum i n heptane was used by Lieflander and Stober (71) to react with the surface S iOH. After thorough washing with heptane to remove excess reagent. the surface retained 5.7 micromoles AI on the surface after hydrolysis:

- R H t H,O Si,OH + AIR, -Si,OAIR; X3i,OAI(OH), + 2 R H

Theoretical Concentralion of Surface Hydroxyl Groups

Calculation of the silanol number of the silica surface can be approached i n several ways. The simplest is that proposed by ller (3) based purely on geometric considera- tions and the density of amorphous SiO,. This has been described in Chapter 1 and leads to the conclusion that there should be 7.8 silicon atoms nm-* a t or very near the surface. He assumed originally that there should thus be 7.8 SiOH groups nm- , of surface. However, Boehm (8) pointed out that since all the silicon atoms cannot be exactly at the boundar!. some must be above and some below, and hence only half of the silicon atoms would bear O H groups so that there would be only 3.9 O H n m -,.

ller also suggested that since the density and refractive index of amorphous silica is close to that of cristobalite and tridymite the concentration of surface hydroxyls might be estimated from the crystal structures. The 1 .O.O crystal plane of beta cristobalite that was selected led to the conclusion that on each 50.2 A2 of surface there were two lower level silicon atoms bearing no hydroxyls and two upper levels each bearing two hydroxyl groups, giving 8 OH nm-,. A similar calculation for the tridymite surface gave 4 .6 O H nrn-,. This has been further considered by Peri and Hensley (67). who pointed out that if the 1.0.0 of cristobalite is considered each silicon atom would hold two hydroxyls, making about 8 O H nm-*, but if these were removed in pairs randomly by a Monte Carlo method there would remain 4.56 O H nm-, as either geminal or vicinal pairs.

Since the early experimental data then available (Table 6.1) indicated a value of about 8 O H nm-2, i t was assumed that the higher figure was more probable. This was given some support by Belyakova, Dzhigit, and Kiselev (72) and Zhdanov and Kiselev ( 7 3 ) . who found a s much as I 2 micromoles OH nm- , or 7.2 OH nrn-,. However, De Boer and Vleeskens (74, 75) then pointed out that since beta cristobalite crystallizes in octahedra, the hydroxyl concentration should be calculated from the I . I . I plane of the octahedral face. This surface is represented in Figure 6.4 . The calculated hydroxyl group concentration is 4 .55 O H nm-,. Experimental data confirmed this result.


Figure 6.4. Atomic arrangement of the ( I . I . I ) octahedral face of cristobalite: 4.55 SiOH n m - z . Large circles, oxygen atoms; small circles, silicon atoms at surface; dashed circles, posi- tion of hydroxyl groups on the surface, attached to the underlying silicon atoms. The atomic sizes are not to scale.

From subsequent data it now appears that the heat-stabilized surface of amorphous silica does closely resemble the octahedral face and 4.6 OH nm-* is a commonly observed value for rehydrated silicas that have been “smoothed” or annealed a t high temperature, then fully rehydrated and dried.

Dehydration and Rehydration

As the temperature is raised, hydroxyl groups condense to form siloxane bonds and water is evolved. The reduction in hydroxyl groups on the surface with increasing temperature has been widely studied and the results have been summarized by Barby (13), Okkerse ( I I ) , and Strelko (76), who furnished 168 references.

The phenomenon can be studied by starting with a thermally stable, fully hydroxylated surface and following the removal of hydroxyl groups as water is evolved in a stream of air or nitrogen or a vacuum. Since water vapor strongly catalyzes rearrangement of siloxane bonds, it is not surprising that dehydration data a re difficult to reproduce unless the water vapor is removed by vacuum or unless the sample is small and spread in a thin film to permit rapid escape of vapor.

An opposite approach would be to start with a pristine silica surface created in an inert gas or vacuum. Fused quartz has been pulverized in a vacuum by Antonini and Hochstrasser (77), who used electron spin resonance (esr) to study the broken bonds. Si. sites were created and they reacted with CO, to give a Si+CO,- complex. The surface loses its activity as it picks up water or other impurities from within the


sample. Zhdanov (78 ) reported that hydration of the surface was much faster than in the case of surface dehydrated at high temperature.

Kautsky and Michel (79) apparently prepared a form of silica. the surface of which is free from silanol groups, by the oxidation of siloxane, Si,O,H,. Unlike ordinary hydrated silicas. this form of SiO, did not absorb rhodamine dye to give a fluorescent product.

Probably because of . the difficulty of obtaining completely hydrogen-free, finely divided silica of sufficient uniform and high specific surface area. few hydration studies appear to have been made. Such silica might be obtained by vaporizing and condensing SiO, in a dry oxygen atmosphere.

Pristine fibers of fused quartz containing 3 % B,O, were used by Dietz and Turner (80) for such studies. although the presence of boron in the surface may confuse the picture. The most significant observation was that rehydroxylation of the anhydrous surface is followed rapidlj by adsorption of water molecules on the newly formed silanol surface. Also, when the surface is dehydroxylated at 8OO0C, rehydration is much slower than when dehydroxylated at a lower temperature. but can still proceed. “Nucleation sites” for rehydration are postulated and these become fewer at higher temperature.

De hydra tion

Probably hundreds of studies of dehydration of silica have been made but few permit accurate assessment of the degree of dehydroxylation of the silica surface as the temperature is increased. Data are not especially consistent but that should not be surprising since silicas vary so enormously in structure, and also in trace impurities. For example, the presence of some K - ions on the silica surface can lower the dehydroxylation temperature by 100-2OO0C (81).

Since the dehydration is a nonequilibrium process, the rote a t which water is lost at any stage is a function of temperature and the concentration of remaining silanol groups. Thus to obtain a significant value for OH nm- , a t a given temperature, the t ime of heating or final minimum rate of water loss should be defined. However, there is a relatively rapid loss of water for the first 6-10 hr after the temperature is raised to a new level and then a much slower loss that can go on for days. Usually dehydration is carried on for a fixed time after which, experience has shown, only a “slight” further change occurs.

Different types of silica may have different packing densities of the S iOH groups on the surface and different thermal stabilities of these groups, as pointed out by Imelik and co-workers (82) . However, the number of silanol groups per square nanometer on a wide variety of silicas. measured so as to include only those groups on the surface and not in the interior, after being heated to a given elevated temperature, is shown in Figure 6.3.

In the higher temperature range the concentration of residual SiOH groups was measured by Curthoys et al. (87) as follows:


Temperature ( “ C ) O H

700 I . 2 800 0.9 900 0.65

IO00 0.4

As a general rule, starting with a fully hydroxylated surface, there are about 5 O H nm-* at 150°C, and at 800°C only about 1 O H n r 2 . Silanol groups start to condense and evolve water extensively above about 170°C. At 400°C somewhat less than half of the hydroxyl groups have been removed so that most of the hydroxyl groups are still adjacent to a t least one other. These can adsorb water and thus the surface is readily rehydrated. Above 40O-45OoC, more hydroxyl groups are removed, leaving larger silaxane areas that are less readily rehydrated (88). At about 75OOC only free, unpaired SiOH groups are present a t a concentration of 1.3 OH nm-* a s shown by Thorp (89) by dielectric isotherms and other methods.

The possibility of thermally dehydrating the silica surface without loss of area was shown by Dzisko, Vishnevskaya, and Cheslova (90). They were probably the first to correctly conclude that physically adsorbed water was removed a t I 1 5 O C and that most of the remaining “bound” water was present as a layer of hydroxyl groups on the surface because the percent of “bound” water was proportional to the specific surface area. They also concluded that between I15 and 600°C an increasing number of hydroxyl groups condensed with liberation of water and minimum loss of surface area. They estimated that each OH group occupied 15 A. so that there should be 6.7 OH nm-z .

According to Boehrn (8), since distances between OH groups differ on the hydroxylated surface, some OH groups are closer together and stronger hydrogen bonds are formed between them. These show infrared absorption at 3520 cm-I. Others are farther apart and are more weakly hydrogen bonded, and show absorption at 3660 cm-l . When heated, the O H groups that are nor so strongly hydrogen bonded to each other come off first: that is, the 3660 peak recedes first (91-94).

The absorption peaks have been identified as follows:

Type of O H Group Peak (cm - l )

Isolated, single S iOH or 3745-3750 “free” hydroxyl groups

S iOH groups (vicinal) mutual hydrogen bonded

groups with hydrogens bonded to each other

on the above

Isolated pairs of adjacent 3650-3660

Adjacent pairs of S iOH 3540-3550

Water molecule adsorbed 3400-3500


Hair (9) summarized the investigations of the silica surface by infrared absorption techniques and described how the various bonds came to be identified.

Kondo and Muroya (95) have observed individual stages in the dehydration process using differential thermal analysis combined with thermogravimetr ic analysis and infrared absorption measurements. These steps a re accompanied by disappearance of an infrared absorption band:

Temperature ("C) Disappearing Band (cm-')

300 400 500

3230 3460 3620

The band for isolated SiOH groups a t 3750 c m - ' remains u p to higher temperature. In view of the undoubted presence of internal hydroxyl groups it is not possible to

determine what changes these steps represent. It would be very interesting to repeat the study with a sodium-free aerogel which has been heat stabilized a t 1000°C and rehydrated by boiling in water for a day or so. Such a silica with much less internal hydration should have a surface area of 100-200 m z g - ' which might still be sufficient for the above types of measurements.

Kunawicz, Jones, and Hockey (96) and Bermudez (49, 69) indentified the different types of silanol groups by their reactions with BCI,. Other reactions by which silanol groups have been studied are summarized in a later section.

De Boer and Vleeskens (37) found that when the silica is heated to 650°C and cooled and rehydrated i n liquid water at 90°C and again dried a t 12OoC, the O H n m - z value is somewhat lower. I f the process is repeated, it reaches 4.6 OH nm-,. I f the original SiO, is heated only once to 890°C and rehydrated and dried the value is 4.6 OH nm-*.

In this way it was demonstrated that a smooth annealed silica surface, when fully hydrated, had one OH per surface Si atom and a concentration of 4.6 OH n n r 2 . I t also showed that while the surface was being annealed, the internal water and/or O H groups were being removed and the structure densified so that less water reentered during the rehydration of the surface.

It was concluded by Okkerse ( 1 1 ) after reviewing the available information that there is no real evidence f o r the existence o f the postulated geminal =S i (OH) , groups. and certainl}' not f o r -Si(OH), on any dried silica surface. If a silica has an apparent hydroxyl content of more than 4.6 OH n m - 2 , the excess water must be present as inrernallj' trapped water of S O H groups.

The internal water and hydroxyl groups has been shown by Chertov et al. (60) to be removed starting at 200°C; the maximum rate per degree temperature rise is at about 500°C for all silicas. However, for more nearly complete removal, the temperature must be around 1000°C. At this temperature, the surface also is largely dehydrated to the siloxane condition.

The mechanism of dehydroxylation has been considered by Hockey and Pethica (97). They point out that i t is most likely that there is a migration of protons rather than hydroxyl groups. They propose that protons migrate via strained oxygen


bridges. It would be interesting to determine whether dehydration will proceed more rapidly in the presence of a low concentration of water vapor than in a vacuum.

There is some evidence by Kondo, Fujiwara, and Muroya (98) that when the sodium ion content of a gel is increased from 5 to 60 ppm, dehydration is somewhat facilitated. The most obvious effect is lowering the sintering temperature at which rapid loss of surface aea occurs, but in all cases, there is little sintering below 700°C. In the case of the lowest sodium content the concentration of hydroxyl groups as OH nm-* followed the upper limit of the shaded area of Figure 6.3.

For absolute dehydration of porous silica, heating in dry chlorine at 600-IO00"C is very effective. This appears to be necessary to obtain glass completely free of silanol groups for optical purposes (99). Another approach is to react the silica such as Aerosil with SiCI, a t 400°C and then heat to 700°C. The resulting surface is essentially free of O H groups (100).

Dehydration of the surface of crystalline quartz might be expected to be more uniform since all points on the surface should be chemically alike (except for edges). Stober (101) studied this phenomenon using quartz particles of different sizes and concluded that a t least on the surface of quartz, dehydration occurs in the stages shown in Figure 6.5. Stober found that even after thorough outgassing at IOO"C, one molecule of extremely tightly adsorbed water is retained for each two silanol groups on the quartz surface. This appears to be quite different from the behavior on amorphous silica. It suggests a powerful hydrogen-bonding capacity of the SiOH groups on the surface and is possibly related to the peculiar power of this crystal surface to adsorb multilayers of monosilicic acid from solution as shown by Baumann and described in Chapter 1, Ref. 151.

Another indication of a major difference between the surface of amorphous silica and quartz is furnished by Young and Bursh (IOZa), who found that the heat of

Figure 6.5. Stages in the dehydration of the surface of crystalline quartz. A , water irreversibly adsorbed on the quartz surface; E , quartz surface immediately after thermal dehydration: C, anhydrous siloxane surface after thermal dehydration. [After Stober (IOI).]


adsorption of water on anhydrous quartz surface is about twice that on amorphous silica, that is. 285 versus 120-190 ergs c w 2 .

Egorov and associates (IO2b) presented an interesting outline of the nature of the hydration and dehydration of the surface, and observations relating these processes and the specific surface area of silica and also its absorptive properties.

EFFECT OF P A R T I C L E S I Z E AND P O R E D I A M E T E R . According to Brunauer, Kantro, and Weise (103) and confirmed by Whalen (104), small particles such as a particle diameter of 3.7 nm are surface dehydrated a t lower temperature than larger ones such as 6.4 nm. Whalen found that silica with an area of 330 m 2 g- ’ held 7 OH at I IO”C, and one with 650 m2 g - I held 5.5 OH nm-’.

Gels with pore diameters of I O , 20, and 27 A were dehydrated a t a series of temperatures by Dzis’ko, Vishnevskaya, and Chesalova (105) and the remaining water contents measured. The ones with smaller pores retained the most water as shown i n Table 6.2. These results are explainable on the basis that when the surface hydrox) Is are hydrogen-bonded to each other, they are less readily removed a t high temperature. I t is probable that the dehydration mechanism involves thermal dissociation of a proton which migrates and combines with a hydroxyl group forming water. Such dissociation would require more energy i f the proton, that is, hydrogen, was shared between adjacent oxygen atoms.

The effects of particle size where the radius of curvature is positive and of pore diameter where it is negative are shown in Figure 6.6. I t seems logical that on the inside of a small pore, the OH groups should be closer together and thus stabilized by hydrogen bonding: other forces can probably also be invoked.

As silica gel is dehydrated, the relative ratio of single (free) hydroxyl groups to adjacent hydrogen-bonded (reactive) hydroxyls is closely related to pore diameter, as recognized by Snyder ( I O ) . H e stated that the surface of a wide pore is more like that of “crystalline” silica and thus will contain more free hydroxyls. It is not a matter of crystallinity but o f geometry. Crystal faces are flat and in this regard are similar to the surface in a large pore which approaches the “flat” condition when viewed on an atomic scale. In small pores with diameter less than 100 8, the negative

Table 6.2. Hydroxyl Group Retention in Gels

Average Pore Diameter

IO A 2 0 A 27 A -

Dehydration Area Bound OH Area Bound OH Area Bound OH ( “C) ( m l g - l ) H,O(%) nm-z (m’g-’) H,O(%) nm-z (m’g-l) H,O(%) nm-2

I I5 400 6.5 10.8 540 6.1 7 . 5 450 3.8 5 .6 3 00 480 4.4 6.1 500 4.3 5 . 7 500 4.0 5 . 3 6 0 0 375 3.4 6.0 400 2.9 4.8 420 2.3 3.6 700 280 1.5 3.6 340 1.7 3 . 3 210 1 . 1 3 . 5

Source Dzis’ko, Vishnevskaya, and Chesalova (105)


B Figure 6.6. Effect of curvature of the silica surface on dehydroxylation. A , small posiiive radius of curvature (small particles) fewer hydrogen bonds, most easily dehydroxylated; B, large radius of curvature (large particles. flat surface) more hydrogen bonds, less readily dehydroxylated; C, small negative radius of curvature ( in small pores), most hydrogen bonding, most difficult

C . . . . . . to dehydroxylate.

curvature brings the silanol groups closer together so there can be more mutual hydrogen bonding and greater stability toward removal a t high temperature.

Snyder and Ward (106) presented data relating S,/S,, the fraction of total hydroxyl groups that are highly reactive, to pore diameier. The pore diameter has little effect until the diameter is Less than 100 A, and then in pores from 100 to 50 A diameter the reactive fraction increases from around 0.05 to 0.8. Since the silicon atoms on the surface, at a concentration of 4.6 O H nm-I, must be separated by an average distance of about 4.7 A, it can be visualized that the OH groups would not be brought appreciably closer t o each other until the pore diameter becomes less than 100 A (Figure 6.6). In a 50 A diameter pore the radius is only about five times that of the Si-Si distance and the O H groups must be much closer together than on a flat surface.

After dehydration a t high temperature there are, therefore, more hydrogen- bonded pairs of O H groups within such small pores than in larger ones.

Snyder and Ward propose that such a pair of OH groups is much more reactive than a single OH with a reactant such as (CH,)$iCI. In effect, the one oxygen can begin to coordinate with the incoming silicon atom while the CI is interacting with the hydrogen of the second OH:


This concerted effect is not possible with an isolated SiOH group and reaction requires a higher temperature.

I n contrast with the above, the influence o f surface curvature on hydroxyl groups has been calculated by Bakyrdzhiev (107) and data were obtained on four silica gels dehydrated at 300-600°C. He reported that dehydration occurs at lower tempera- lure in narrower pores. This seems opposite t o the above observations by Dzis’ko, Vishnevskaya. and Chesalova.

However, there is room for ambiguity here. The silica gel with narrower pores may have been formed from smaller silica particles. As shown in Figure 6 . 7 , if the gel was not aged, the surface within the pores would consist mostly of small spherical particles with a posit iw radius of curvature. In this case the hydroxyl groups would be less strongly hydrogen bonded and therefore easier to remove as water. I t is thus possible that on the same starting gel, an aging treatment can convert most of the inner silica surface from a positive to negative curvature, as in B . I n such an aged gel the smaller the pores, the more stable the O H groups.

Rehydration

Studies by Young and Bursh (92. IO2a) have clearly shown that water molecules are adsorbed only on the hydroxylated silica surface and not on the siloxane surface which is essentially hydrophobic. However, rehydration must involve adsorption of water as the first step, so that hydration probably occurs on siloxane oxygen sites

.--7-

A

B Figure 6.7. Change from positive to negative curvature of the surface as silica gel is aged. A , unaged gel. Most of area in pores has a positive radius of curvature: B , aged gel. Most of the silica surface in pores has a negative radius of curvature. Slightly heavier lines show the areas with positive radius.


next to a silanol site. It was found that surfaces dehydrated up to 400-425°C were readily rehydrated, whereas above this temperature rehydration became progressively slower. At 425°C the surface retains about 2 O H nm-2 or 40-45% of the original 4.6 OH n m Z . The dehydrated siloxane sites are probably isolated, At this point the heat of adsorption of water per unit area of silica is at a maximum,

Rehydration of the partly dehydroxylated surface is catalyzed by alkali (108). However, high pH also promotes loss of surface area during prolonged contact with water. Addition of a little NH,OH provides a safe pH range; certainly rehydration is slow below pH 5.

Snyder (IO, 109) reports that rehydration of the partly dehydrated siloxane surface even in liquid water is exceedingly slow a t 25°C. The silica must be heated in water at 95°C for several hours. However, this can be done only with heat-stabilized silicas of low surface area; high surface area powders would undergo drastic changes i n structure and reduction in surface area under these conditions.

Rehydration of a highly dehydrated surface probably begins next to a free hydroxyl group since its characteristic infrared band becomes weaker (91). The idea, often stated, that the silica surface dehydrated at 1000°C cannot again be fully hydrated is not true, as shown by Agzamkhodzhaev et al. (1 IO). They found that the more completely the surface was dehydrated, the longer the time required for rehydration. A silica aerogel that was heated a t 1 I00"C for 10 hr, having a surface of 144 m2 g-l , retained only 0.06 O H nm-*. After being boiled in water for 60 hr the surface area was still 108 m2 g-I and rehydration of the surface was complete with 4.5 OH n m 2 . A surface dehydrated to 0.66 O H n m 2 a t 900°C for IO hr required several years in water a t ordinary temperature to become rehydrated.

It is likely that rehydration occurs only next to a hydroxyl group and that the hydroxylated area grows in patches as hydration proceeds along the boundary between hydroxylated and siloxane areas. I t will therefore initially appear to be autocatalytic. Data by Volkov, Kiselev. and Lygin ( I I I ) seem to support this idea since on a highly dehydrated surface prepared at 900-11OO0C, water was first chemisorbed by opening up strained siloxane bonds, forming silanol groups and water, then was adsorbed on these rather than on the siloxane surface. Further opening of siloxane bonds could occur only adjacent to these spots.

Surface Energies

Brunauer, Kantro, and Weise (103) reported the following values for the total surface energy (enthalpies) determined from the heat of dissolution in HN0, -HF mixtures, of silica samples of different known surface areas with different states of dehydroxylation, all a t 23°C:

Energy (ergs cm -z)

259 * 3 129 =t 8 130 i 7

Pure siloxane surface Pure silanol surface Heat of hydration


The heat of hydroxylation of siloxane surface to silanol at 23°C is 259 * 13 cal g - l

of water. or 4660 * 230 cal mole- ' . Kiselev ( I I ? ) reported that when silica gels of different surface areas were heated

at 300°C the hydroxylation of the surface decreased with increasing surface area. At the same time the surface energy of the silica increased. The effects were linear between the following approximate values:

Specific Surface Area S iOH Surface Energy (m 'g - ' ) (micromoles m -z) (ergs cm - 2 )

100 700

5.0 170 3.3 200

As pointed out by Brunauer, Kantro, and Weise (103) and predicted by Iler (3), the total energy of the silanol surface is only slightly higher than the total surface energy of liquid water, 118.5 ergs

The development and status of the theory of surface free energy of solids and interfaces have been reviewed by Good ( I 13). The theories relate to physical adsorption, wetting, and phase separations, but not to irreversible processes such as chem isorption.

at 25°C.

Heat of Wetting of Silica Surface

I t is obvious that i f studies are made with microporous silicas with pores of different sizes. very little can be concluded about the nature of the surfaces by measuring the heat of wetting. Only nonmicroporous silicas can be considered.

Based on early data on the heat of wetting of various silica powders with different degrees of surface hydration, as measured by Patrick ( I 14), Iler (3) extrapolated the data to 0 and 100% surface coverage and obtained the following values.

Heat of Wetting (ergs cm -*)

Silanol surface 190 Siloxane surface I30 Difference 60

However, there was no assurance that some micropores were not present. Taylor, Hockey, and Pethica (84) have obtained data on the heat of immersion in

water of an annealed, rehydrated silica surface at various stages of dehydroxylation at temperatures where no sintering or change i n surface area occurs. For the fully hydroxylated material i t was 160 + 3 ergs and this was independent of the area of the silica from 8 to 150 m' g-I.


The heat of immersion of silicas dehydrated at various temperatures is very con- fusing unless the silica is first annealed at 700°C to eliminate micropores or surface Si,OH groups. and then rehydrated. Otherwise the heat of immersion differs at different temperatures (27 vs. 45'C) a s shown by Taylor, Hockey, and Pethica (84). This is a t least partly due to the different arrangements of SiOH groups on the surface which can undergo change upon being wetted. On the other hand, the annealed silica, rehydrated, has a stable surface that has the same heat of wetting a t 27 and 45°C. and this increases with the number of hydroxyl groups on the surface:

Heat of Immersion (Approximate)

O H n m - 2 (ergs c w 2 )

0 117 (extrapolated) 2 122 3 I30 4.7 160

This value of 117 ergs for the heat of wetting of the siloxane surface is more dependable and somewhat lower than the value of 130 ergs c m - z calculated by Iler. Other higher values obtained by other workers can probably be explained by the fact that there were micropores present or the silica surface had not been stabilized by annealing. Taylor, Hockey, and Pethica show that the heat of wetting of the surface of unannealed silica a t maximum hydroxylation of 4.7 O H nm - 2 may approach 200 ergs cm-2 .

There a re many factors affecting the heat of immersion of nonannealed hydroxylated silicas that remain to be investigated. For example, Taylor, Hockey, and Pethica found that dehydration in vacuo resulted i n a higher heat of wetting than heating in air-probably because in the latter case the steam, which was not immediately drawn off, promoted rearrangements of surface SiOH groups.

A differential wetting calorimeter was developed for measuring heats of immersion of solids by Tyler et al., ( 1 15). Well characterized pyrogenic silicas (Aerosil) were used for heat of wetting experiments in water and benzene.

The nonwetting or hydrophobic character shown by the siloxane surface when contacted with liquid water led Laskowski and Kitchener (116) to the conclusion that the work of adhesion of water to a solid surface consists of three terms:

I . Dispersion forces (van der Waals). 2. Hydrations of nonionic polar sites. that is. bonding with SiOH groups. 3. Ionization.

On silica, the second factor controls wettability. The contact angle and zeta potential between water and vitreous silica was measured on the surface, which had been made hydrophobic with (CH,),SiCI. After a t ime the original hydrophobic sur-


face became wetted (contact angle zero) even though methyl groups were still on the surface. After removing physically adsorbed water the surface was again hydrophobic. (It is likely the surface was not fully covered with methyl groups.) When covered with close-packed hydrocarbon groups, the surface shows no such reversible phenomenon; this has been proved in the case of estersils, where a surface covered with closely packed butyl groups remains hydrophobic for months in water.

The heat of wetting by polar liquids ( I 17, 118) decreases as the silica surface is dehydroxylated. With a nonpolar liquid like heptane the heat of wetting increases. Even i n the case of methanol the heat of wetting on the completely dehydroxylated surface is 7 5 ergs cm -, as compared to 50 for water. I t appears that the heat of wetting depends on the association of the polar groups in the organic molecule with the polar silanol surface and the association of the hydrocarbon portions of the molecule with the hydrophobic siloxane surface. A fully hydroxylated silica surface immersed in water shows no hydrophobic characteristics.

PHYSICAL ADSORPTION OF NONIONIC LOW MOLECULAR WEIGHT COMPOUNDS

Physical adsorption studies on silica during the past half-century must number 1000 or more. In this section only typical examples of nonionic physical adsorption are discussed. Ionic adsorption is dealt with later.

The methods and techniques for studying adsorption such as infrared absorption, nuclear magnetic resonance, and dielectric measurements are not described here but may be learned through references.

Hydrogen bond formation between electronegative atoms or pi electrons of adsorbate molecules and the hydrogen atoms of the silanol groups on the silica surface plays a major role i n adsorption of molecules from the vapor state and from nonaqueous solution. It also is a major factor in adsorption from aqueous solution of nonionizing types of molecules. This role of the hydrogen bond has been reviewed by Hair (9 ) , Little ( I 19), and Kiselev and Lygin (120).

Adsorption of Vapors

Reversible adsorption implies that only intermolecular physical forces are involved and that desorption has a very low activation energy. This includes molecules which are hydrogen bonded to the surface. However, there may be cases where there are several points of hydrogen-bonded attachment to the surface, such as with C,H,OC,H,OC,H,OC,H,, where desorption may appear essentially irreversible, or at least the adsorption follows a Langmuir-type isotherm. Some multiply-bonded surface complexes may be so stable that they might be considered as essentially chemisorbed.

Adsorption of gases and vapors on gels and powders has been studied mainly as a method of characterizing the solids. Also, such data are essential to evaluating the

Nonionic Low Molecular Weight Compounds 649

practical merits of gel adsorbents. The extensive review of the physical adsorption of gases and vapors in the first part of Chapter 5 should be referred to a t this point, Only a few aspects of the subject are included here and these are restricted generally to silica surfaces free from micropores.

Numerous investigations make it clear that adsorption of polar molecules or aromatic compounds through pi bonds occurs most strongly on silanol groups that are not already hydrogen-bonded with neighbors. Thus on silica dehydrated at 500°C on which there remain about 2.7 O H nm-*, one C,H,NH, is adsorbed per O H group at room temperature (121). Ammonia is similarly adsorbed stoichiometrically on S iOH groups according to Bastik (122).

For a brief summary of the phenomenon of adsorption on various surfaces, especially silica, reference should be made to the short monograph by Kiselev (123). The role of hydrogen bonding in adsorption of different classes of compounds was revealed by infrared absorption studies, for example, diethyl ether (124) pyrrole ( I 25), and t-BuOH ( 1 26).

The strength of the hydrogen bond is indicated by the shift in the infrared adsorption band of the O H due to stretching. Sempels and Rouxhet (127) compared the electron donor characteristics of a series of weak hydrogen bonding agents in this way. They showed that the hydrogen bonds are the same in solution as on the solid surface.

The effect of adsorption of a molecule with different configurations such as CIC,H,CI was to increase the cis-trans ratio from 1.0 in CHCI, solution to 1.9 on the silica surface (128). With acetylacetone i t was the enol form that was involved in the hydrogen bond. This observation suggests that the molecular form favored at the surface is the one that provides the greatest number of hydrogen bridges:

CH3 CH3 CH3 CH3 I I I I

I I I I Si Si Si Si Si

Steric hindrance can prevent hydrogen bonding and adsorption on the surface as in the case of 2-chloropyridine, where the large chlorine atom apparently prevents close approach (129).

Activated AI,O,-SiO, gel does not adsorb pyridine vapor as strongly as silica does. The nature of adsorption on silica-alumina has been well reviewed by Hair (9), who points out that on an activated silica-alumina gel the only hydroxyl groups are SiOH. These must therefore be fewer per unit area than on hydroxylated silica and thus adsorb less pyridine.

The adsorption of ammonia on the silica surface is not a simple process according to Bliznakov and Polikarova (130). An increasing amount of irreversibly bound ammonia remains below 70°C. According to Boyle and C a w (131) the adsorption is


partly chemical and partly physical. Pyridine also presents a complex picture (132) which was somewhat clarified by infrared studies of adsorbed substituted pyridines ( I 33).

Adsorption of amine vapors has been well studied. Because of their basicity they are strongly adsorbed on S iOH sites, the ones that are weaker bases by hydrogen bonding (Si,OH-:NH,_,R,) and those that are stronger bases by salt formation (Si,O- *NH,+,R,). Bartell and Dobay (134) found that diethyl and mono- and dibutylamines formed monolayers on the silica surface a t p / p o of only 0.01, then further pore filling occurred a t 0.5-0.8. The following areas were covered by each amine molecule.

Amine Area (A2)

43 57 33

A study of the adsorption of methyl bromide by Van Cauwelaert, Van Assche, and Uytterhoven (135) suggests that the hydroxyl groups on pyrogenic silica are distributed heterogeneously and that they are not located in pairs a s commonly thought.

Chessick and Zettlemoyer found that in measurements of the integral heats of formation of monolayers of water and organic vapors on silica, the -AH values, increased linearly with - TAS (136).

An automatic adsorption isotherm recorder using an electrobalance has been developed that plots adsorption and desorption isotherms from 760 to lo-’ torr and from - I96 to 500OC for any adsorbate noncorrosive to steel ( 1 37).

Effeci of Dehj~droxylarion on Adsorprion

WATER. Shapiro and Kolthoff (138) found that silica gel loses some of its capacity to adsorb methyl red from benzene solution as the gel is dehydrated, even a t temperatures where loss in surface area through sintering does not occur. Thus methyl red must be adsorbed on S iOH groups on the surface, but not on the dehydrated Si-0-Si areas. They also found that the amount of water which is adsorbed when the heated gel is cooled and exposed to various humidities parallels the adsorption of methyl red. When the vapor pressure of water is lower than that of the liquid phase, water i s not adsorbed on the dehydrated oxide surface, but only on the S iOH groups. Since water vapor is not adsorbed on the dehydrated areas, rehydration is slow i f the silica is exposed only to water vapor.

The freshly formed, anhydrous surface of glass also apparently rehydrates only slowly. For example, Brunauer (139) points out that the difference in the heats of adsorption of water on silica gel and on virgin glass surfaces can be explained by the fact that the gel has a layer of chemisorbed water on the surface which attracts water, so that adsorption proceeds readily, while freshly formed glass surface may be “devoid of water nuclei” upon which adsorption could originate. In the latter

Nonionic Low Molecular Weight Compounds 65 1

case, higher vapor pressure must be reached before the first layer of water is established. Brunauer also cited Trouton (140), who reported that glass wool dried 70 hr at 160°C over P,O, adsorbed water with great difficulty, but when less completely dried, adsorbed water readily. All of this suggests that the dehydrated surface is actually hydrophobic. The electrons of the strained siloxane oxygen atoms are not free to enter into hydrogen bond formation.

The dehydration of the silica surface has been shown by Zhdanov (141) to cause a hysteresis loop in the H,O adsorption-desorption isotherm the first time it is run on a dehydrated silica gel. [Zhdanov may possibly have discovered the dehydration of the silica surface before Dzis'ko, Vishnevskaya, and Cheslova (90) carried out their work.] Zhdanov concluded that H,O is adsorbed mainly at the OH groups, but that the oxide bridges produced by dehydration can be rehydrated to O H groups by adsorbed water. The evidence that H,O is adsorbed on O H groups, according to Zhdanov, is as follows. When silica gel was heated to 500°C instead of 3OO0C, there was an additional loss of 2 .2 millimoles of O H groups per gram of silica. When the adsorption isotherms after heating at 300 and 5OO0C, respectively, were compared at p / p o = 0.1, the silica which had been heated at 500°C adsorbed 1.95 millimoles of O H groups per gram less than the silica heated at 300°C. At p / p o = 0.2, the difference was 2.7 millimoles of O H groups per gram. Thus for every O H group lost from the surface, about one less molecule of water was adsorbed at low pressure.

One can therefore visualize that the first adsorbed water on a partly dehydrated surface may be represented as follows:

H H H H 0 0

H H H H H H H H 0 0 0 H H H 0- 0 0 /O\ 0 0

-Si-0-Si -0-Si-0-Si-0-Si-0-Si-0-Si-

Adsorption of water only on isolated, hydrogen-bonded O H pairs was studied by Hertl and Hair (142) with unusual results. The adsorption isotherm showed definite steps at H,O pressures of about 9.5 and 16 torr. First the O H pair adsorbed one HzO, then at the second step two more H,O molecules were adsorbed, and in an indistinct third step, a total of six H,O were probably adsorbed. All the other O H groups on the surface surrounding each isolated pair of SiOH groups had been eliminated by reaction with (CH,),SiCI which did not react with the pair. Thus these hydrophilic spots adsorbed H,O to form a cluster as follows:

/H

H\< lH' P- H- - .O \

H

H' 'H H

\ ,0- H - - -0

\

/

/

0 0-Si-0-H

0 0-Si-0-H'

0

',\\ / \ \ \ / 0 \ ,,/\ ,f' \

H H


The mechanism of interaction of water with the'xilica surface was further considered in detail by Prigogine and Fripiat (143). It was made clear that it is H,O molecules adsorbed on isohred SiOH groups that react with the adjacent siloxane surface to form two new SiOH groups. Also, vicinal groups adsorb water, which then hydrates the siloxane surface. It is postulated that this strongly hydrogen- bonded water becomes more highly dissociated and thus able to attack the siloxane bond and open it.

The specificity of the adsorption of H,O on the SiOH groups at low vapor pressure was proposed as a way of characterizing the state of hydroxylation of the surface (144a). Dubinin (144b) emphasized the importance of the SiOH group in adsorption of any material where there is a donor-acceptor component in the interaction, that is, hydrogen bonding.

The behavior of water on hydroxylated and dehydroxylated silicas which contain micropores is complex, a s shown by Baker and Sing (145). The nonporous dehydroxylated silica surface is only partially covered at the stage that would nor- mally correspond to monolayer coverage by the BET method of analysis. However, when micropores are present, even though the silica has been dehydrated at the same high temperature, the water molecules are more strongly adsorbed and form close to a BET monolayer. There is the probability that in micropores dehydration is less complete and the results of Baker and Sing seems to support this idea. A few residual hydroxyl groups could act as nuclei for adsorption in such small pores.

I N E R T G A S E S . It has already been pointed out that the heat of adsorption of nitrogen on silica decreases as the surface is dehydrated (40). Aristov et al. (146) found that argon was insensitive to the state of dehydration of the surface and thus is better suited to surfaces of different chemical compositions. This was confirmed by Bassett, Boucher, and Zettlemoyer (147), who used the area 16.6 A* for the argon molecule.

ORGANIC MOLECULES AND OTHERS. For polar molecules the SiOH groups is the site a t which adsorption occurs. Hence for maximum adsorption the silica surface should be free of adsorbed water and have a maximum concentration of SiOH groups. For this reason a fully hydroxylated silica that has been dried at about 175°C is optimum. Also, according to Snyder (IO) for polyunsaturated hydrocarbons or multipolar molecules, the fully hydroxylated surface is better in providing multiple adsorption sites that can fit the adsorbate. It is on the isolated, freely vibrating hydroxyl groups on the thermally dehydrated silica surface that strongest adsorption of organic compounds occurs, according to Hair and Hertl (148) and Clark-Monks and Ellis (149). Aromatic hydrocarbon adsorbates interact with these hydroxyls in a 1 : 1 ratio. On mutually hydrogen-bonded adjacent SiOH pairs, there is little tendency for

hydrogen bonding except with strong electron donor atoms such as water or methanol. In fact, water reacts preferentially a t these sites. Once water is adsorbed, the site then has a free hydrogen active for adsorption through hydrogen bonding. This might be represented as

Nonionic Low Molecular Weight Compounds

H 1 .o,

653

Ether oxygens can also bond to the self-bonded O H pairs, destroying their mutual hydrogen bonds (124).

Curthoys et al. (87) used adsorption isotherms and spectral and calorimetric methods for studying adsorption of a variety of types of organic molecules on fully hydroxylated and highly dehydroxylated silica surfaces. It was concluded that the hydrogen-bonding strength of the molecules on the surface was the same as in solution. The silicas used were prepared with nonporous uniform surfaces, and heats of adsorption and infrared spectra over the whole range of surface coverages were measured on both the hydroxylated and dehydroxylated surfaces. This has given clear and unequivocal data that are much easier to interpret than those of previous investigators who did not use such distinctly different types of surfaces.

The strength of the hydrogen bond was evaluated approximately as the difference i n the heats of adsorption on the hydroxylated surface at 50% coverage and on the dehydroxylated surface where the value was least. Good agreement was obtained between the bond energies calculated from the heats of adsorption and from spectral data for which equations were developed. It was shown that the heat of adsorption was the sum of the heat of condensation plus the heat of hydrogen-bond formation.

The data show that the molecules tested fall approximately into three classes in regard to strength of interaction with the hydroxylated surface:

Low. Medium.

Strong. Pyridine, (C,H,),N.

When t-butyl alcohol is adsorbed on silica it is easy to detect the hydrogen bonds that are formed. Davydov, Kiselev, and Lygin (150) found that the alcoholic O H hydrogen bonds only with free SiOH groups and not with the groups that were close enough together to bond with each other. The tertiary alcohol does not react further to form the ester linkage as primary and secondary alcohols do. Dzhigit et al. (151) found that the heat of adsorption was 18 kcal mole-', but this was drastically reduced when the silica surface was dehydrated, reducing the amount of hydrogen bonding. As the surface is progressively dehydrated, the adsorption of methanol is not reduced nearly as much as is the adsorption of water (152). I t is possible that the methyl group becomes associated with the hydrophobic siloxane surface adjacent to the silanol group where it is adsorbed.

Ar, N,, CCI,, n-C6Hl,, cyclohexane. CH3N02, CeH,; CH,COOC,H,, CH3CN, tetrahydrofuran, dioxane,

acetone, (C,H,),O.

654 The Surface Chemistry o f Silica

I n the case of amine vapors it is a question of degree whether the bond is a hydrogen bond or an ionic attraction:

S iOH + NR, = SiOH - : N R ,

S iOH + NR, = SiO- + H N R ;

Obviously the answer lies somewhere between, depending on the reactive basicity of the nitrogen atom and the acidity of the silanol hydrogen atom. However, in the case of the quaternary ammonium ion; for example, (CH,),N+, one must assume only ionic attraction can exist.

Bauer and Stober (153) observed first the formation of a monolayer of mono-, di-, or triethylamine on silica, both hydroxylated and dehydroxylated surfaces, and measured the heats of adsorption on two reversibly adsorbed molecular layers.

lsosteric heats of adsorption of 23 compounds on the 3750 cm -' surface hydroxyl group of silica were measured by Hertl and Hair (154). The isosteric heat of adsorption usually is an average over a total surface which may not be homogeneous, for example, the partly dehydrated silica surface. Hair (155) developed a spectroscopic method for determining heats of adsorption on specific kinds of sites.

Polydimethylsiloxane oil is adsorbed on the surface o f silica, whether hydroxylated or dehydrated. With the oil of molecular weight 350,000, one molecule covers 6 x IO' A*, with a thickness of IO A . This is as flat as the strained molecular chain can lie according to Kiselev. Novikova. and El'tekov (156).

Sulfur dioxide is adsorbed on silica gel with initial heat of adsorption of 25-30 kcal mole- ' for the first 0.01 micromole m - z adsorbed, dropping to 6-7 kcal mole- ' on dehydroxylated and I2 kcal mole- ' on hydroxylated silica. Isotherms were determined on different types of gels by Glass and Ross (157). Adsorption of CH,NH, was also studied.

Adsorption from Solution-Nonionic

Nonaqueous Solutions

Physical adsorption from solution is here arbitrarily defined as not including adsorption of ions, which is covered in a following section. There does not appear to be any general rule based on structure or physical properties of organic molecules to predict the relative affinity for the silanol surface i n specific binary or more complex mixtures. Since the pK, of the silanol surface is similar t o that of water and the interfacial energy is so small i t is likely that intermolecular interactions with water will parallel those with the silanol surface.

In solutions or liquid mixtures it would seem that the forces that result in concentrating a given molecule at the liquid-silanol boundary will involve the following:

1. The tendency toward phase separation which is reflected by the excess concentration of the molecule at the liquid-vapor boundary.

Nonionic Low Molecular Weight Compounds 655

2 . The energy of interaction of the S iOH as an electron donor or acceptor with respect to the subject molecule.

Robert (158) found that ethyl alcohol, adsorbed from heplane on silica gel (S, =

400). occupied an area of 18 A* and concluded that the molecules were adsorbed per- pendicularly to the surface with the hydroxyl groups hydrogen bonded to the O H groups of the silanol surface. Benzene covered 55 A and thus must lie flat. Acetic acid appeared to form a double layer, since the area covered per molecule was only 9.1 A. Jones and Outridge (159) studied the adsorption of mixtures of butanol and benzene on silica gel, both by liquid and vapor phase methods. The butanol and benzene compete for the silica surface, but the benzene is replaced by butanol as the butanol concentration in the liquid phase is increased to 100%. Similar studies were reported by Matayo and Wightman (160) on systems involving ethanol, cyclohexane, and benzene.

The role of hydrogen bonding was long recognized, and studies are typified by the infrared study of Griffiths, Marshall, and Rochester (161), who showed that the oxygen atoms in diethyl ether and in acetone, when adsorbed on silica from CCI,, are hydrogen bonded to SiOH. They also noted that nitrogen atoms of substituted pyridines vary in base strength and this causes different degrees of stretching of the OH bond which results in shifts in the wavelength of the infrared adsorption bands.

C,H,N: H-OSi

An apparatus was developed for measuring infrared spectra of molecules a t the liquid-solid interface using matched sample cells i n a closed circulation system that can be evacuated and attached to purification and storage systems for the liquids ( I 62).

The adsorption of alcohols from carbon tetrachloride onto the surface of silicas that were dehydroxylated to various degrees has been measured by Ganichenko, Kiselev, and Krasil’nikov (163). The alcohol molecules are oriented with the hydroxyl groups toward the S iOH surface groups. With increasing chain length the alcohols acted more nearly alike on surfaces hydrated to different degrees and approached the same surface coverage of about 3 micromoles m-’ for n-hexyl alcohol. Longer-chain alcohols are adsorbed to form a monolayer even on a partly dehydroxylated surface. The effect of pore size on adsorption of a series of alcohols from carbon tetrachloride was examined by Bonetzkaya and Krasil’nikov (164).

Cholesterol adsorbed from benzene lies flat on the hydroxylated surface, each molecule covering 94 A2 (165).

Adsorption of carboxylic acids from nonpolar solvents is best explained on the assumption that the carboxyl groups form hydrogen bonds with the silica surface, according to Elder and Springer (166). The sorption of acids from nonpolar solvents decreases with the number of carbon atoms in the molecule in the following order: acetic, propionic, crotonic, benzoic, and palmitic. Also, sorption decreases with change of solvent in the order carbon tetrachloride, toluene, nitrobenzene, dioxane, and water. The relative extents of hydrogen bonding between solute, solvent, and gel play an important role.


Adsorption of n-jiifry acids from benzene and hexane were compared. Benzene interacts with S O H groups through pi bonding whereas hexane does not. Hence from benzene the C,.,, acids reached a limiting coverage of only 0.5 mole nm-l , whereas in hexane i t was 1.85 down to 1.4 for C,.16, according to Armistead, Tyler, and Hockey (167). The reason for the low coverage in benzene is that in this solvent the C,.,, acids are dimerized by hydrogen bonding and are in equilibrium with a low concentration of monomer. It i s only the monomer that is adsorbed.

On silica that is partly surface esterified with n-BuOH, lauric acid from hepprane is adsorbed only on the nonesterified areas (165). Adsorption of C1, and C,, acids and oleic acid from decane-acetic acid mixtures was compared by Rakhlevskaya et al . ( 168).

The C,, dioic acid was much less strongly adsorbed than C,, alcohol or Cla amine on pyrogenic silica. although some unsaturated chains were apparently broken ( 1 69).

Oleic and linoleic acids are adsorbed on hydroxylated silica gel from carbon tetrachloride according to Marshall and Rochester ( 170). Infrared absorption shows that hydrogen bonds are formed with adjacent silanol groups, perhaps a s follows:

When the silica with adsorbed linoleic acid was dried and heated, chemisorbed linoleate groups remained and were stable to 280°C. There was indication of interaction of the silanol surface with double bonds in the chain.

Adsorption of aromatic hydrocarbons from mixtures with saturated hydrocarbons has been given considerable attention, no doubt in connection with practical methods of separation in mind. Adsorption data indicate that the aromatic molecules are preferentially adsorbed on silanol groups and that best separation may be attained with small-pored gels, provided the pores are large enough to admit the aromatic molecule, for example, about 20 A diameter for benzene or 45-70 8, for trimethyl- or alkylbenzenes (171-173). Interaction of benzene with the S iOH surface and mean jump length of the molecules was examined by Winkler et al. (174) using proton magnetic relaxation. The tendency of hydrocarbons to adsorb on the S O H surface parallels the interfacial tension between the hydrocarbons and water ( 1 75).

Phenols are preferentially adsorbed on the silanol surface from nonpolar solvents but can be used to measure the solvent-silica interaction. This system was studied by Davis, Deuchar, and Ibbitson (176), who noted the degree to which the solvent was co-adsorbed with the phenol. Solvents interacted with the S iOH surface increasingly in the order hexane, cyclohexane, CCI,, CH,C,H,, and C,H,.

Nitroalkanes from methane to butane and nitrobenzene were adsorbed from benzene. alcohol, acetone, and also water mixtures. Equations were found by Jones and Mill (177) to correlate results. Adsorption of drugs on silica is of possible importance in pharmaceutical formulations and analyses. Rupprecht and Kindl ( 1 78) showed that alkaloids are adsorbed by hydrogen bonding between S iOH and basic nitrogen atoms

Next Page

Documents

CHAPTER 6 The Surface Chemistry of Silica - cntq.gob.vecntq.gob.ve/cdb/documentos/quimica/224/6.pdf · Barby (I 3) contributed a broad survey of ... Various aspects of the surface