The Surface Chemistry of Amorphous Silica. Zhuravlev Model

Colloids and Surfaces

A: Physicochemical and Engineering Aspects 173 (2000) 138

The surface chemistry of amorphous silica. Zhuravlev model

L.T. ZhuravlevInstitute of Physical Chemistry, Russian Academy of Sciences, Leninsky Prospect 31, Moscow 117915, Russia

Received 14 January 1999; accepted 21 February 2000

Abstract

A review article is presented of the research results obtained by the author on the properties of amorphous silicasurface. It has been shown that in any description of the surface silica the hydroxylation of the surface is of criticalimportance. An analysis was made of the processes of dehydration (the removal of physically adsorbed water),dehydroxylation (the removal of silanol groups from the silica surface), and rehydroxylation (the restoration of thehydroxyl covering). For each of these processes a probable mechanism is suggested. The results of experimental andtheoretical studies permitted to construct the original model (Zhuravlev model-1 and model-2) for describing thesurface chemistry of amorphous silica. The main advantage of this physico-chemical model lies in the possibility todetermine the concentration and the distribution of different types of silanol and siloxane groups and to characterizethe energetic heterogeneity of the silica surface as a function of the pretreatment temperature of SiO2 samples. Themodel makes it possible to determine the kind of the chemisorption of water (rapid, weakly activated or slow, stronglyactivated) under the restoration of the hydroxyl covering and also to assess of OH groups inside the SiO2 skeleton.The magnitude of the silanol number, that is, the number of OH groups per unit surface area, aOH, when the surfaceis hydroxylated to the maximum degree, is considered to be a physico-chemical constant. This constant has anumerical value: aOH,AVER4.6 (least-squares method) and aOH,AVER4.9 OH nm

2 (arithmetical mean) and isknown in literature as the KiselevZhuravlev constant. It has been established that adsorption and other surfaceproperties per unit surface area of silica are identical (except for very fine pores). On the basis of data published inthe literature, this model has been found to be useful in solving various applied and theoretical problems in the fieldof adsorption, catalysis, chromatography, chemical modification, etc. It has been shown that the BrunauerEmmettTeller (BET) method is the correct method and gives the opportunity to measure the real physical magnitude of thespecific surface area, SKr (by using low temperature adsorption of krypton), for silicas and other oxide dispersedsolids. 2000 Elsevier Science B.V. All rights reserved.

Keywords: Amorphous silica; Surface characterization; Silanol and siloxane groups; Internal silanols; Physico-chemical model;Physico-chemical constant; BET method

www.elsevier.nl:locate:colsurfa

1. Introduction

Research into the silica-water system is impor-tant both for elucidating the theoretical aspects of

the problems involved and for practical applica-tions [1]. In this connection an investigation of theso-called combined, structurally bound water [25] in dispersed amorphous silica is of interest.

0927-7757:00:$ - see front matter 2000 Elsevier Science B.V. All rights reserved.

PII: S0927 -7757 (00 )00556 -2

L.T. Zhura6le6 : Colloids and Surfaces A: Physicochem. Eng. Aspects 173 (2000) 1382

This term describes OH groups that are bound viathe valence bond with Si atoms on the silicasurface (hydroxyl coverage), and in some caseswith Si atoms inside the particles of silica.

In the 1930s, studies of the condensation pro-cesses of silicic acids, carried out by Hofmann,Endell and Wilm [6], Rideal [7] and Kiselev [8],and slightly later by Carman [9], showed thathydroxyl (silanol) groups, SiOH, should bepresent on the surface of silicates and silicas. Onthe basis of measurements of the heat of wettingand a comparison of the adsorption data with thedata from chemical analysis and the correspond-ing results reported in the literature, Kiselev sug-gested that the water evolved during calcinationof silica gel, besides physically adsorbed water, isformed from OH groups that are chemically heldon the silica surface. This suggestion led to anunderstanding of the dehydroxylation mechanism[8].

Yaroslavsky and Terenin [1113], by using aninfrared spectroscopy method, proved for the firsttime the existence of hydroxyl groups on the silicasurface (porous glass). This fact was soon confi-rmed by Kurbatov and Neuymin [14]. Now nu-merous spectral and chemical dataunambiguously confirm the presence of the OHgroups on such SiO2 surface.

Silanol groups are formed on the surface by twomain processes [1,36]. First, such groups areformed in the course of silica synthesis, e.g. duringthe condensation polymerization of Si(OH)4 (Fig.1a). Here, the supersaturated solution of the acidbecomes converted into its polymeric form, whichthen changes into spherical colloidal particles con-taining SiOH groups on the surface. Upondrying, the hydrogel yields xerogel, the finalproduct, which retains some or all of the silanolgroups on its surface. Secondly, surface OHgroups can form as a result of rehydroxylation ofdehydroxylated silica when it is treated with wateror aqueous solutions. The surface silicon atomstend to have a complete tetrahedral configuration,and in an aqueous medium their free valencebecomes saturated with hydroxyl groups (Fig. 1b).

The surface properties of amorphous silica,which is considered to be an oxide adsorbent, inmany cases depend on the presence of silanolgroups. At a sufficient concentration these groupsmake such a surface hydrophilic. The OH groupsact as the centers of molecular adsorption duringtheir specific interaction with adsorbates capableof forming a hydrogen bond with the OH groups,or, more generally, of undergoing donoracceptorinteraction. The removal of the hydroxyl groupsfrom the surface of silica leads to a de-

Fig. 1. The formation of silanol groups on the silica surface: (a) Condensation polymerization; (b) Rehydroxylation.

L.T. Zhura6le6 : Colloids and Surfaces A: Physicochem. Eng. Aspects 173 (2000) 138 3

Fig. 2. Types of silanol groups and siloxane bridges on thesurface of amorphous silica, and internal OH groups (see text).Qn-terminology is used in NMR, where n indicates the numberof bridging bonds (OSi) tied to the central Si atom: Q4,surface siloxanes; Q3, single silanols; Q2, geminal silanols(silanediols).

Various problems related to silica surface char-acteristics are encountered in different areas ofscience and technology: physics, chemistry andphysical chemistry, agriculture, soil science, biol-ogy and medicine, electrical energetics, the oilprocessing industry, the metallurgical and miningindustries, some fields of geology, etc. Traditionalbuilding and other materials based on silica, suchas cement, concrete, firebrick, silicate glasses,rough and fine ceramics, and enamels, occupy asignificant place in human life. Different types ofsilica are widely used as efficient adsorbents andselective absorbents, active phase carriers in catal-ysis, fillers for polymeric systems, adsorbents andsupports for gas and liquid chromatography,thickeners for dispersion mediums, binding agentsfor molding materials, reinforcing fibres, and soforth. Chemical modification of the surface ofdispersion silica has received a large amount ofinterest; this process allows researchers to regulateand change adsorption properties and technologi-cal characteristics of composite materials. Of latethe use of SiO2 is on the rise in the manufactureof modern high-quality materials (microelectron-ics, optics, fiber optics, liquid crystals, differentcomposites including biocomposites, orderednanostructured silica materials, monodispersedcolloids, etc.).

In the past 50 years, many reviews have ap-peared on the subject of surface chemistry ofsilica. And we refer readers to these numerousmonographs, reviews, and the most interestingand significant, in our opinion, articles [1,11300].

In view of this, is there need for yet anotherreview in this field? The author of the presentreview article has carried out numerous experi-mental studies on the subject in question, system-ized such important characteristics as theconcentration and the distribution of differenttypes of silanol groups, established the energeticheterogeneity of the surface in a wide temperaturerange of the pretreatment, and investigated thecharacteristics of bound water inside SiO2 parti-cles. Besides, it was made careful study of thestructure characteristics of many different silicasamples. On the basis of these researches theauthor was able to construct an original physico-

crease in the adsorption, and the surface acquiresmore and more hydrophobic properties [1,36].

Surface OH groups are subdivided as following(Fig. 2): (i) isolated free (single silanols), SiOH;(ii) geminal free (geminal silanols or silanediols),Si(OH)2; (iii) vicinal, or bridged, or OH groupsbound through the hydrogen bond (H-bondedsingle silanols, H-bonded geminals, and their H-bonded combinations). On the SiO2 surface therealso exist surface siloxane groups or SiOSibridges with oxygen atoms on the surface. At last,there is structurally bound water inside the silicaskeleton and very fine ultramicropores, dB1 nm(d is the pore diameter), i.e. internal silanolgroups.

The last number of decades saw a rapid growthin those fields of science and technology that dealwith production and utilization of various colloidand microheterogeneous forms of silica with de-veloped surfaces, such as sols, gels, and powders.The properties of a pure silica, as an oxide adsor-bent, are determined in the first place by: (i) thechemical activity of the surface this activitydepends on the concentration and the distributionof different types of OH groups, and on thepresence of siloxane bridges; and (ii) the porousstructure of the silica.


chemical model [301303], describing the surfaceproperties of amorphous silica (referred in litera-ture as the Zhuravlev model). Therefore, it wouldbe pertinent to compare in this review article theresults obtained by the author with those reportedin literature.

In this model for the amorphous silica surfacethe determining factor is the presence of silanolgroups and siloxane bridges. The concentration ofthese groups depends on the conditions of thermaltreatment of the SiO2 sample in vacuo (or onother types of pretreatment). It is necessary totake into account possible changes occurringsimultaneously in the degree of surface coveragewith adsorbed water molecules or with differentsurface groups and in the energetics of dehydra-tion (the removal of physically adsorbed water),dehydroxylation (the removal of silanol groups)and rehydroxylation (the restoration of the hy-droxylated covering) processes. Also it ought tofollow some changes in structure of the surfaceand the skeleton of the silica matrix. These factorsdetermine the starting conditions that are neces-sary for working out the model for the amor-phous silica surface. To avoid the introduction ofsuch complicating factors as the possible effect onthe silica surface properties of any impurities,structural defects, other functional groups andactive sites, etc. they are not considered at thisstage.

2. Experimental section

The versions of the method of deuterium ex-change (DE method) developed by the author[304306,311,312,321,322,328] have been used fordetermining the concentration of the hydroxylgroups on the surface of dispersed oxide adsor-bents. The advantage of this method is that, un-der certain conditions, deuterium exchange islimited to the surface and does not involve struc-turally bound water inside silica. According tothis DE method an isotopic exchange takes placebetween a known quantity of heavy water D2Oand an unknown number of hydroxyl groups onthe surface of the sample. The concentration ofthe original OH groups or the silanol number,

that is, the number of OH groups per unit surfacearea, aOH (OH nm2), is determined from theknown ratio of the concentrations of the isotopes,[H]:[D], in the water vapor phase, after comple-tion of the deuterium exchange, and from theknown weight of the sample and its specific sur-face area:

aOHdOH(S) NA 1021:S (1)

or

aOHK dOH(S) :S (1%)

where dOH(S) (mmol OH groups g1 of SiO2) is theconcentration of OH groups on the surface ofSiO2 per unit mass of the sample as obtained fromthe DE data; S (m2 g1), the specific surface areaof the sample as determined by the BET method[10] by using low temperature adsorption of kryp-ton (the area occupied by one Kr atom in amonolayer, vm,Kr0.215 nm2) [50]; NA, theAvogadro number; and K602.214 is constant.

The determination of the isotopic compositionof water vapor following the DE reaction wascarried out by using mass spectrum measurementsand other methods of the isotopic analysis. One ofthe variants of this DE method [311,322] made itpossible to determine the silanol number andsimultaneously to study the kinetics of water ad-sorption and isotopic exchange between D2O andthe surface OH groups. The minimal measuredamount of water on the SiO2 surface formed from2 104 to 2 106 moles H2O with a rela-tive error of 915% (the different versions).

The DE method has some drawback: it cannotbe used for determining separately the molecu-larly adsorbed water and different kinds of silanolgroups. For such a determination we used themass spectrometric thermal analysis in conjunc-tion with the temperature-programmed desorp-tion (the MTA-TPD method), the infraredspectroscopic method, and some others.

The process of the removal of physically ad-sorbed water and hydroxyl groups from the sur-face of the silica sample has been investigatedusing the MTA-TPD method. In general, theversions of the MTA-TPD method, worked outby the author [302,303,332], make it possible toobtain the following: thermal desorption curves


(or mass thermograms); the spectra of volatilecomponents under the condition of linear heatingof the sample (in the temperature range 251000C); the energy of activation of thermal des-orption; and other kinetic parameters. The kineticparameters have been estimated by known meth-ods [341346]. A characteristic feature of theMTA-TPD method is a high sensitivity due to thedesign of a pyrolysis chamber. An open-type mi-crocrucible for holding the SiO2 sample is simulta-neously as a thermojoint of the thermocouple andlocates near the ion source of the mass spectrome-ter. The weight of the SiO2 sample was less than10 mg. The precision of determining of the sampletemperature at any given moment of time andthroughout the entire temperature range was 91.5C.

In our work we used pure amorphous silicasamples of different origins: silica gels, aerosils(pyrogenic silicas), aerosilogels and porousglasses. In all about 150 samples were investi-gated; for 100 of these determinations were madeof the silanol number when the surface was hy-droxylated to the maximum degree [301303]. Alarge number of different samples had to be usedin order to obtain a reliable physico-chemicalconstant (see below).

The following varieties (ai) of amorphous sil-ica were investigated (Table 5):a. laboratory-made and industrial silica gels syn-

thesized by the acidic method (26 samples);b. laboratory-made silica gels obtained by the

acidic method with the use of hydrothermaltreatment at the hydrogel or xerogel stage(nine samples);

c. laboratory-made silica gel obtained by hydrol-ysis of tetraethoxysilane (one sample);

d. laboratory-made and industrial silica gels ob-tained by the Bard ion exchange method [339]from alkali and acid sols (14 samples);

e. laboratory-made aerosilogels obtained from anaqueous suspension of aerosils (20 samples);

f. laboratory-made porous glasses obtained byleaching sodium borosilicate glass (tensamples);

g. rehydroxylated industrial silica gels (12samples);

h. rehydroxylated laboratory-made and indus-trial aerosilogels (five samples);

i. rehydroxylated laboratory-made porousglasses (three samples).

The skeleton structure of the silica samples was ofa globular- [1,36] or sponge-form [320]. The sam-ples differed strongly in the following characteris-tics: specific surface area, types of pore,distribution of pores according to their size, andthe density of packing of the particles. The studyof such structures proved to be an importantstage in our work [301303,305311,317320,323327]. To determine the structure of dif-ferent samples, besides the adsorption of Kr andN2, we measured the total isotherms of adsorptionand desorption including capillary condensa-tion, of water vapor, methanol and benzene, weused the method of apparent density, the mer-cury-porosimetric method, and the electron-mi-croscopic and the kinetic methods. The silicasamples had specific surface areas SKr varyingfrom 9.5 to 945 m2 g1 [301303].

When measuring aOH one of the most impor-tant factor is the specific surface area, SKr, asdetermined by BET method based on low temper-ature adsorption of krypton [50]. This inert sub-stance Kr was chosen because the adsorption ofkrypton on the silica surface as determined by thedispersion interaction is non-specific in nature, i.e.it is insensitive to changes in the degree of hydrox-ylation of the sample. Thus, the silanol numberaOH was determined on the surface of the pores,which were accessible to Kr atoms. According tothe classification (IUPAC) by Dubinin [340] suchpores include: macropores, d\200400 nm;mesopores, 3.03.2 nmBdB200400 nm; andsupermicropores, 1.21.4 nmBdB3.03.2 nm (dis the pore diameter). In those cases where thebiporous SiO2 samples contained very narrowpores (ultramicropores, dB1 nm) together withwide pores (mesopores), the samples were consid-ered to be wide-pores ones, and the very narrowpores were excluded from consideration when de-termining aOH. The diameter of ultramicropores iscomparable with that of water molecules, andtherefore only water molecules can penetratethem. The OH groups in these very narrow poreswere classified not as surface silanol groups but asbound water inside the silica particles.


Determinations were made of the structuralcharacteristics of the silica samples both beforethermal treatment and after treatment at tempera-tures from 180200 up to 10001100C.

To study surface dehydration and dehydroxyla-tion of amorphous silica by the MTA-TPDmethod, without side undesirable effects, it wasnecessary to prepare a standard SiO2 sample. Thissample had to meet the following requirements:1. the surface of the starting sample should be

completely hydroxylated;2. the diameter of the pores should be much

greater than the size of the water molecule sothat the effects of diffusive retardation and thereadsorption inside pores are suppressed to amaximum extent;

3. there should be no structurally bound waterinside the silica skeleton so that the side effectdue to the evolution of water from the bulk ofthe sample at higher temperatures was absentat all;

4. the silica sample should be free of extraneousimpurities on its surface and inside it; and

5. the structure of the SiO2 skeleton should bestabilized.

The method of synthesizing such the standardsample has been reported [302,303,332]. It wasfound that the most suitable type of SiO2 whichmeets the above-mentioned requirements was anaerosilogel, amorphous silica containing uni-formly wide pores (mesopores) [142]. It was pre-pared from an aqueous suspension of very purepyrogenic silica (aerosil, S180 m2 g1). Thestabilized structure of the silica skeleton was ob-tained by multiple heating (in the atmosphere orin the water vapor) at temperatures up to 940Cwith following cooling of the sample each time.The distribution of the pores in terms of theirdiameter, d, lay within a narrow range with amaximum at 51 nm. The specific surface area ofthe sample was S79 m2 g1 as determined bythe BET method from the low temperature ad-sorption of Kr, and this standard sample wasnamed S-79.

The comparison of the results obtained by theauthor [301303,332,334338] with those re-ported in literature covers the following topics: (i)limiting temperature for removing physically ad-

sorbed water from the hydroxylated surface ofamorphous silica; (ii) completely hydroxylatedstate of the surface; (iii) structurally bound waterinside the particles of amorphous silica; (iv) dehy-droxylation and rehydroxylation of the silica sur-face; (v) energetic heterogeneity of the surface;and (vi) physico-chemical model of amorphoussilica surface: main stages and distribution ofvarious types of surface groups.

3. Results and discussion

3.1. Boundary temperature for remo6ingphysically adsorbed water from the hydroxylatedsurface of amorphous silica

To elucidate the nature of hydroxyl coverageand to quantitatively determine the concentrationof OH groups on the silica surface it is necessaryto distinguish these groups from the molecularlyadsorbed water. Although many investigationshave been carried out by different researchersusing various methods for determining the valueof the boundary temperature, TB, for removingthe physically adsorbed water, there is no agree-ment between the values obtained so far. Thefollowing examples of the results reported in liter-ature show a wide discrepancy of TB values.

Iler, after reviewing numerous reported results,concluded in his monograph [1], that on silicadried from water, the hydrogen-bounded watermolecules came off at room temperature in vac-uum or at 150C in the atmosphere. The prevail-ing view is that drying in vacuum at lowtemperatures is the only way to remove the ad-sorbed water without disturbing the OH groups.Also, this is the conclusion reached by Gregg andSing in their monograph [158].

De Boer et al. [3739] found that silica dried inair at 120C loses all physically adsorbed waterbut at 110C still retains water if the air is humid.According to Okkerse [104] the removal of allphysically adsorbed water at 120C is possibleonly if the silica sample is free of micropores.When silica contains micropores the adsorbedwater can be retained in the micropores at tem-peratures up to 180C even though the surface of


wide pores begins to be free of OH groups.Bermudez [106] in his investigations using theNMR method has established that after exposurefor 6 h at 110C all water is removed from thesilica surface including some silanol groups, aswell as some internal water. Taylor, Hockey et al.used 115C during subjecting silica to pretreat-ment either in air or in vacuum [73,82].

Young and Bursh investigated the interactionof water vapor with the surface of amorphoussilica using different methods (thermal analysis,heat of immersion in water and others) [41,51].On the basis of these results the authors con-cluded that the dexydroxylation of surface OHgroups begins at 180C.

The removal of physically adsorbed water fromthe silica surface has been studied by many re-searches using thermogravimetric, thermodesorp-tion and other thermographic methods. As anexample, in Fig. 3 is presented the TGA (ther-mogravimetric analysis) thermogram of a silica

sample, obtained by Vansant, Van Der Voort andVrancken [263]. This thermogram distinctly showsa sharp DTG (differential thermogravimetricanalysis) peak, attributed to the loss of physicallyadsorbed water from the surface of the poroussample. The profiles of curves 1 and 2 (DTG andTGA, respectively) indicate that desorption ofsuch physisorbed water is completed at 150Cand is followed by a broad region of weight loss,due to dehydroxylation process.

It is necessary to note, that in numerous ther-mographic investigations, fulfilled by different au-thors, an endothermic effect, i.e. a rather sharpDTA peak, in the range about 120200C is alsodirectly related to the removal of physically ad-sorbed water from the SiO2 surface.

A number of researchers gave the limiting tem-perature TB\200C. Thus, Baverez and Bastick[66] on the basis of IR spectroscopic analysisfound that TB$240C. Fripiat and Uytterhoeven,using the same method, came to the conclusion[58] that physically adsorbed water is completelyremoved only at 300C.

Some Russian and Ukrainian researchers sug-gested a concept according to a part of Si atomson the silica surface are coordinately unsaturated(so-called nonhydroxyl centers of the II-type).Thus, Kiselev et al. [83,102] assumed that thesewater molecules are completely removed in vac-uum only at 400C. Chuiko et al. [150,188]considered that the adsorbed water molecule (withthe coordinate bound) near the II-type site can belocated either over the face of the siliconoxygentetrahedron (Scheme 1(a)) or under the Si atom ofthe silanol group (Scheme 1(b)). The centers ofthe II-type ensure a strong retention of the coor-dinately bound water molecules on the SiO2 sur-face up to 650C, with the energy of theinteraction of one H2O molecule with such acenter being up to 62 kcal mol1 (260 kJmol1) [150].

It could be continued an enumeration of simi-lar examples, because many data are given inliterature that show a broad scatter of TB values.

In order to find the limiting temperature,TB, we have carried out a series of investigat-ions by employing the DE and MTA-TPDmethods together with some others [301

Fig. 3. Thermogravimetric analysis of a hydroxylated silicawith physically adsorbed water on the surface (the mesoporoussilica gel, S400 m2 g1): (1) DTG curve, (2) TGA curve(From Vansant et al. [263]).

Scheme 1. Chuikos scheme (a) and (b).


Fig. 4. (1) TGA curve, or total loss of water during thermaltreatment of hydrated silica gel: each point on the curve wasobtained by summing up the amount of water that came off infixed temperature intervals and the amount of water measuredby the DE method; (2) DTG curve, or rate of the water loss(for each 1C increase in temperature), obtained by graphicdifferentiation of curve 1.

S-79 by the MTA-TPD method [302,303,332]. Aseries of curves representing the reduced massthermograms of water obtained at the same rateof linear heating (b5.8 grad min1) are shownin Fig. 5. The intensity, I, of the spectral peak ofwater, m:Z18 (as determined according to theordinate axis), characterizes the rate of formationor the rate of thermal desorption of water. Priorto measurements by the MTA-TPD method thesesamples S-79 were subjected to treatment in vac-uum under different conditions of heating (Table1). It is necessary to note that a mass spectralanalysis of the escaping volatile products duringthe heating (251000C) of the standard silicasample S-79 has shown that water is practicallythe only substance to be identified.

An examination of the ascending branches ofcurves 17 and curve 8 up to the maximum (pointA in Fig. 5) shows the following.

Fig. 5. Mass thermograms of water for the standard silicasample S-79 (rate of heating b5.8 grad min1); T, tempera-ture (K); I, normalized intensity of the peak due to water ionm:Z18 (arbitrary units): yellow zone corresponding to thefree water; light blue zone corresponding to the adsorbed H2Omultilayers, region I; dark blue zone corresponding to theadsorbed H2O monolayer, region I; red zone corresponding tothe condensation of vicinal OH groups, subregion IIa; crimsonzone corresponding to the condensation of free OH groups,subregion IIb. Point A indicates the maximum of kineticcurves 18; points Ai, the maxima of curves 913 and 1517(iI, II, III,VIII); curves 8 and 14 are boundary curves (seeTable 1 and text).

303,306,311,329,332]. In our complex thermo-graphic investigations of several different silicasamples, we have detected endothermic peak at150C on a DTA curve [329], and this effect, asit noted before, is related to the removal of phys-ically (molecularly) adsorbed water.

Using a method which combines thermogravi-metric measurements, deuterium exchange, andmass spectral analysis, we have investigated [306]large-pore silica gel (S320 m2 g1). In Fig. 4are shown the total loss of water, i.e. the loss ofphysically adsorbed water, hydroxyl coverage,and structural water inside globules of the SiO2sample, as a function of temperature, or the TGAcurve 1. The rate of the water loss (for each 1Cincrease in temperature) is shown in the DTGcurve 2. From the shape of curves 1 and 2, it canbe seen that under step-wise heating conditions ina vacuum from room temperature to 150200Cthere came off mainly physically adsorbed water.At higher temperatures, there was desorption ofthe surface OH groups and the removal of struc-tural water from the bulk of the SiO2 sample also.

We have carried out studies of dehydration anddehydroxylation using the silica standard sample


Table 1Conditions for pretreatment in vacuum of standard samples S-79, conditions of studies by MTA-TPD method (at b5.8 degmin1) and temperature of mass thermogram maximum (K)

Conditions for pretreatment and conditions of studies by MTA-TPD methodThermogram no. Temperature (K)

Suspension of SiO2 in an excess of water frozen at 196C; evacuation in vacuum for1 6001 h at 196C; gradual warming up from 196 to 0C with the recording of massthermogram; heating under linear condition from 0 to 1000C with the recording ofthe thermogramLoading of the wet sample; evacuation for 66 h at 25C; heating under linear2 600condition from 25 to 1000C with the recording of the thermogram

600Ditto, with evacuation for 87 h at 25C34 600Ditto, with evacuation for 306 h at 25C

600Ditto, with evacuation for 14 h at 107C5Ditto, with evacuation for 2 h at 136C6 600Ditto, with evacuation for 1 h at 169C7 600

600Dotted line (curve 8)-boundary curve (see text)8Loading of the wet sample; evacuation for 14 h at 203C; heating under linear9 621condition from 25 to 1000C with the recording of the thermogram

10 688Ditto, with evacuation for 12 h at 240C11 711Ditto, with evacuation for 12 h at 276C

718Ditto, with evacuation for 14 h at 286C1213 Ditto, with evacuation for 16 h at 335C 769

875Dotted line (curve 14)-boundary curve (see text)14Loading of the wet sample; evacuation for 13 h at 438C; heating under linear15 916condition from 25 to 1000C with the recording of the thermogramDitto, with evacuation for 14 h at 495C16 972

1037Ditto, with evacuation for 14 h at 588C17

Curve 1 reflects the behavior of the ther-mogram for an SiO2 sample introduced into apyrolyzer as a suspension with a large excess ofliquid water. The experiment was carried out at atemperature range starting from 196C (thesample was in a frozen state) and up to 1000C.As can be seen, a very intense maximum appearsin the range of 270350 K. This effect is directlyrelated to the evaporation of the excess of waterin the liquid phase (free water yellow colour inFig. 5), which was in suspension, and is notrelated to the desorption of water. Upon furtherheating, thermogram 1 passes through a secondmaximum, which is weaker but clearly defined, atthe characteristic point A.

Curves 2, 3 and 4 (ascending regions) corre-spond to samples which have been first subjectedto treatment in vacuo at room temperature for 66,87 and 306 h, respectively; curves 5, 6, 7 and 9correspond to samples treated at 107, 136, 169and 203C, respectively. Curve 8 (as well as curve14, which is considered below) was not registered

experimentally; it was constructed by interpola-tion with account taken of the behavior of theascending regions of the neighboring ther-mograms 7 and 9. An important conclusion canbe drawn from the experiments: the maxima of allthe kinetic curves 17 occurred at the same tem-perature TA (600 K, point A in Fig. 5 and Table1). It should be noted that for curves 17 thetemperature of the prelimenary treatment of thecorresponding samples of S-79 did not exceed200C. When the temperature of pretreatmentwas above 200C, the maxima of the correspond-ing thermograms were shifted into the region oftemperatures greater than TA (to the right ofpoint A). Such a shift was the greater, the higherthe temperature of the preliminary treatment ofSiO2 (curves 917, Fig. 5 and Table 1). Curve 8lies at the boundary since it is the last (on theright) which still passes through the stationarypoint A.

An analysis of region I to the left of theboundary curve 8 shows that the region represents


numerous ascending branches of the mass ther-mograms each of which passes through a com-mon point A. It is reasonable to assume thatregion I corresponds to the state when, in addi-tion to the surface coverage consisting of hy-droxyl groups, the surface contains physicallyadsorbed water. The removal of such water(molecularly adsorbed water within the limits ofmultilayers and a monolayer, curves 18) at dif-ferent initial degrees of coverage has no any ef-fect on the position of the maximum (point A).This means that there coexist simultaneouslytwo independent types of bound water on theSiO2 surface: physically adsorbed water (region Ito the left of the boundary curve 8 light blueand dark blue colours) and chemisorbed wateror hydroxyl coverage (region II to the right ofthe boundary curve 8 red and crimsoncolours). Any change in the state of molecularlyadsorbed water (which depends on the condi-tions of pretreatment of the S-79 sample invacuo at temperatures B200C, Table 1) in noway affects the hydroxyl coverage. Thus, thecharacteristic point A (at TA600 K), whichcorresponds to the maximal rate of thermal des-orption of water from the SiO2 surface, is anindicator of the dehydroxylation process but notof the dehydration process.

In order to determine the kinetic parameterscharacteristic of dehydration (region I, Fig. 5)

we resorted to graphic construction of the sub-tractive differential thermokinetic curves, whichwere then processed according to a known pro-cedure [332,342,343]. The results, which aresummarized in Table 2, show that the activationenergy of desorption ED increases from 6 to 10kcal mol1 (2644 kJ mol1) as the extent ofthe silica surface covered with physically ad-sorbed water decreases to uH2O:0. Data incolumn 2 of Table 2 show that the experimen-tally determined kinetic order of the thermaldesorption reaction n is close to unity. Thisconfirms that the water adsorbed on the surface,which is in a molecular form, is removed inregion I (Figs. 11 and 12, see below).

On the basis of the ratios between the areasbelonging to regions I and II we can concludethat under the experimental conditions employedthe amount of physically adsorbed water (forsamples, which were pretreated at temperaturesmore 25C, Table 1), i.e. the area limited byascending curves 4 and 8 (dark blue colour, Fig.5), is less than a single monolayer of water onthe silica surface.

Thus, the main mass of adsorbed water, in-cluding the region corresponding to polymolecu-lar adsorption (between curves 1 and 4 lightblue colour), is removed in vacuum at roomtemperature. However, a small amount of physi-cally adsorbed water, within the limits of amonolayer (between curves 4 and 8), remains onthe hydroxylated surface of silica up to approxi-mately 200C.

The increase in the values of ED with a decreasein the degree of surface coverage by adsorbedwater in region I (Table 2 and Fig. 12, see below)is close to the observed changes in the values ofpure differential heat of adsorption of water va-por (QAL) (where QA is the differential heat ofadsorption, and L is the heat of condensation) ata low degree of coverage by adsorbed water onthe surface of silica, as shown by Dzhigit et al.[55]. As may be expected, the approximate equal-ity between ED and (QAL) indicates that thedirect process of physical adsorption of watervapor on the hydroxylated silica surface is non-

Table 2Dehydration: determination of the kinetic parametersa

Activation energy ofReaction order nSubtractivethermokinetic desorption EDcurve

kJ mol1kcal mol1

0.85(23) 6.2 25.91.25(25) 7.3 30.61.34(26) 7.3 30.70.97 6.4(36) 26.90.85 8.9(58) 37.1

(68) 0.84 9.4 39.30.91(69) 10.4 44.2

a Note: (ij )* The ordinate of such a curve at any point ata fixed temperature equals the difference between the ordinatesof the corresponding ascending branches of reduced ther-mograms i and j (Fig. 5).


Table 3Heating rate (b) and the temperature corresponding to themaximum (point A)

12.3b (deg min1) 25.05.8

TA(K) (the average of several 599.8 630.9 659.0determination)

steeper while still passing through the commonpoint A. On the basis of the shape of thethermograms of the subregions in region I,one can probably speak of the presence oftwo types of physically adsorbed water at alow degree of coverage. This accords with ourdata obtained by the method of molecular dy-namics [303,330,333] (Fig. 18, see below). Theactivation energy of desorption of the two typesof adsorbed water (within the limits of a mono-layer) lies approximately in the range ED68and 810 kcal mol1 (Table 2, Fig. 12, see be-low).

Now we shall determine the exact value of thelimiting temperature, TB, which separates thesetwo processes: dehydration and dehydroxylation.

Let us consider the ascending sectors inboundary thermogram 8 and in the neighboringthermogram 9 (Figs. 5 and 10, see below). Inorder to determine the activation energy ED inthe neighborhood of point A we used the non-isothermal method [341,346]. Experimentally ob-tained data on the position of point A fordifferent rates of linear heating of the sampleare summarized in Table 3. The graphic repre-sentation of the linear dependence of the differ-ence (2log TA log b) as a function of thereciprocal of the maximum temperature 1:TAyields 16.5 kcal mol1 for the activation energyED.

To determine the reaction order n (for a givenED), we used the method suggested bySmolyaninov et al. [344]. The reaction order forcurve 8 was found to be n2.00 (Table 4). Thekinetic parameters for thermograms 8 and 9 ob-tained by known methods [344346] are alsoshown in Table 4.

An interpolation based on the data for ther-mograms 7 and 9 (Fig. 5) yields the temperatureof the preliminary treatment of SiO2 in the casewhere the ascending branch of the kinetic curvemust exactly follow the boundary thermogram8. The threshold temperature corresponding tothe completion of dehydration and the begin-ning of dehydroxylation was found to be TB190910C [302,303,332].

Table 4Kinetic parameters determined for thermal desorption on thebasis of the shape of the ascending branches of the boundarythermogram 8 and the thermogram 9 (between regions I and IIin Fig. 5)a

Activation energy ofReactionThermokineticcurve order n desorption ED

kJ mol1kcal mol1

8 16.5 69.12.0071.18 (2.0)* 17.0

1.709 18.8 78.69 80.619.2(2.0)*

a Note: * Parentheses indicate that n2 is taken as a knownvalue.

activated, and EA$0 [70]. The rate of physicaladsorption depends only on the rate at whichthe water molecules approach the surface ofporous SiO2 sample.

Mass thermograms 24 (ascending regions,Fig. 5, and Table 1) relate to samples whichhave undergone preliminary treatment in vacuoat room temperature, followed by increasinglyprolonged periods of treatment. It can be seenthat for the subregion between curves 2 and 5the slope of the ascending sections of the ther-mograms decreases and there is a weak maxi-mum (curves 3 and 4) at approximately400420 K. For the next subregion betweencurves 5 and 8, with an increase in the tempera-ture of preliminary treatment from about 100 to200C (Table 1), the ascending sections become


Thus, it was established that at TB (for amor-phous silica subjected to pretreatment in vacuo at190C) sharp changes of the parameters takeplace. These changes correspond to the change inthe activation energy of desorption ED approxi-mately from 10 to 17 kcal mol1 (and thus tochanges in a number of other thermodynamicfunctions), and to the change in the kinetic orderof the limiting stage of thermal desorption n from1 to 2 (Tables 2 and 4, Figs. 11 and 12, seebelow).

Rebinder [44] examined earlier theoreticallyhow the water layer is bound to the dispersedmaterials in the course of drying. The isothermicfree energy (or characteristic binding strength ofwater on the surface) of the free water equalszero:

DFART ln(p0:p) at pp0 (2)

where R is the gas constant, and T is the temper-ature. In the region of removing of physicallyadsorbed water, the magnitude of A increasescontinuously (at po\p). In the case of chemicallybound water (OH groups on the SiO2 surface), themagnitude of A will increase abruptly. Thus, theleapwise increase in A was predicted by Rebinder[44] and experimentally confirmed by us (thesharp increase in the kinetic parameters, ED andn, at TB190C, Figs. 11 and 12) [302,303,332].

A critical comparison of various data in litera-ture regarding the threshold temperature at whichthe physically adsorbed water comes off from thehydroxylated surface of amorphous silica revealsthe following. The value of TB190C practicallycoincides with TB180C obtained earlier byYoung and Bursh [41,51]. In our opinion thisvalue of TB190910C holds true for variousamorphous silica samples having different struc-tural characteristics.

3.2. Completely hydroxylated state of the silicasurface. Structurally bound water inside theparticles of amorphous silica

To develop a model describing the silica surfaceit is necessary first of all to have reliable quantita-tive data on the concentration of OH groups as afunction of the preliminary thermal treatment in

vacuum of SiO2 samples. This is particularly im-portant for the initial fully hydroxylated state ofthe silica surface.

At the same time it is well known that differenttypes of amorphous dispersed silica contain notonly OH groups on the surface, but also struc-turally bound water within the silica skeleton andinside the ultramicropores of the sample. It isnecessary to make a distinction between internalwater, which has no effect on the surface pro-cesses, and the hydroxyl coverage. The latter de-termines all variety of interactions of differentsubstances with the active OH sites on the SiO2surface.

Let us consider the results obtained by theauthor, using the DE method [301303]. In Table5 are summarized values S, dOH(S) and aOHdOH(S) :Sthat were obtained for 100 different samples ofamorphous silica (the varieties ai, see above).Each numerical value of dOH(S) and, respectively,aOH is an average of two or more measurements(altogether 231 measurements were carried out).The surface of these samples was subjected to themaximum degree of hydroxylation, i.e. the hy-drated samples were thermally pretreated in vacuoat 180200C (see above TB190910C).

In those cases where the biporous samples ofSiO2 contained very narrow pores (ultramicropo-res, dB1 nm) in addition to wide pores (meso-pores) [307,311,318320,331], the samples wereconsidered only to be wide-pore ones, and thevery narrow pores were excluded from consider-ation in determining aOH (see above).

At first, it would be pertinent to examine thevalue for the concentration of OH groups on thesurface of SiO2 per unit mass of the sample, dOH(S) ,as a function of S (Fig. 6). It can be seen that theexperimental values dOH(S) are bounded by twoslightly diverging straight (broken) lines passingthrough the origin of coordinates. There is a highvalue of the correlation coefficient of the linearregression equation between dOH(S) and S, r0.99,that was calculated for the numbers (in 100points) of these two independent each from otherphysical magnitudes, changing within very wideranges: so, the surface SKr of such silica samples ischanged within range of their values from 9.5 to945 m2 g1 (Table 5), and the diameters, d, of


Tab

le5

Spec

ific

surf

ace

area

,S

(m2

g1),

and

conc

entr

atio

nof

surf

ace

OH

grou

ps,d

OH

(S)

(mm

ole

OH

g1)

anda

OH

(OH

nm

2),

for

100

diff

eren

tsa

mpl

esof

amor

phou

ssi

lica

who

sesu

rfac

ew

assu

bjec

ted

toth

em

axim

umde

gree

ofhy

drox

ylat

iona

6N

o.sa

mpl

e7

18

910

1112

1314

1516

1718

1920

23

45

945

710

734

270

212

328

316

297

274

294

340

384

S36

374

466

081

590

585

571

780

56.

415.

545.

732.

561.

902.

782.

942.

712.

462.

346.

552.

946.

613.

513.

385.

706.

77d

OH

(S)

5.56

5.12

6.39

4.9

4.5

4.1

4.7

4.7

5.7

5.4

5.1

5.6

5.5

5.4

4.8

5.2

5.5

5.6

5.2

5.0

4.4

4.5

4.3

aO

H

26N

o.sa

mpl

e27

2128

2930

3132

3334

3536

3738

3940

2223

2425

750

3896

3912

162

4839

209.

554

549

812

012

915

516

116

140

0S

558

670

4.71

4.17

6.23

0.32

0.83

0.29

1.01

0.57

0.40

0.34

0.18

0.07

3.39

0.92

1.13

1.15

1.31

4.90

1.14

3.26

dO

H(S

)

5.2

4.5

5.0

5.1

5.2

4.5

5.0

5.5

5.0

5.3

5.4

4.3

4.1

4.3

4.4

4.3

4.9

4.4

5.7

4.9

aO

H

4647

4849

5051

5253

5444

5545

5657

5859

6043

4241

No.

sam

ple

241

245

307

350

424

4311

212

913

314

216

316

816

816

8S

170

168

195

196

199

205

1.92

1.87

2.55

3.31

3.03

0.39

0.80

0.86

0.91

1.06

1.43

1.08

1.39

1.26

1.28

1.26

1.44

dO

H(S

)1.

421.

361.

374.

25.

14.

84.

65.

05.

74.

35.

44.

34.

04.

14.

54.

04.

54.

64.

55.

14.

34.

24.

1a

OH

6667

6869

7071

7273

74N

o.sa

mpl

e75

6176

7778

7980

6263

6465

180

273

290

311

320

312

256

238

102

7718

070

170

6427

1211

S17

517

517

01.

441.

211.

472.

272.

022.

432.

662.

382.

001.

900.

970.

730.

600.

590.

270.

110.

101.

411.

451.

22d

OH

(S)

4.9

aO

H5.

04.

34.

24.

75.

04.

64.

74.

85.

75.

75.

25.

56.

05.

75.

45.

05.

04.

24.

8

8687

8889

9091

9293

9484

9582

9697

9899

100

8385

No.

sam

ple

81 262

270

241

250

262

262

283

298

305

4710

813

314

214

511

4245

270

270

S27

0d

OH

(S)

1.96

2.31

1.95

2.09

2.18

2.26

2.47

2.38

0.35

0.83

0.99

1.18

1.08

0.11

0.40

0.46

2.11

2.24

2.38

2.42

4.9

4.7

4.8

5.0

4.8

5.0

4.7

4.5

4.6

4.5

5.0

4.5

5.4

5.8

aO

H5.

76.

15.

34.

75.

05.

3

aN

ote:

nos.

126

ofSi

O2

sam

ples

are

the

grou

p(a

);no

s.27

35,

grou

p(b

);no

.36

,gr

oup

(c);

nos.

375

0,gr

oup

(d);

nos.

517

0,gr

oup

(e);

nos.

718

0,gr

oup

(f);

nos.

819

2,gr

oup

(g);

nos.

939

7,gr

oup

(h);

nos.

981

00,

grou

p(i

)(s

eete

xtan

dF

ig.

7).


Fig. 6. Dependence of concentration of surface OH groups perunit mass of the sample SiO2 determined by DE method, dOH

(S)

(mmol OH g1), on the surface area of silica, S (m2 g1)(Table 5). Straight (solid) lines passing through the origin ofcoordinates are average values of the silanol number (100 SiO2samples): aOH,AVER (dOH

(S) :S)AVER4.6 OH nm2 (least-

squares method) and aOH,AVER4.9 OH nm2 (arithmetical

mean).

cal mean). The absolute error (the standardsquare divergence) of the silanol number aOHforms value DaOHD(dOH(S) :S)90.5 OH nm2which characterizes a scatter of points concerningtheir mean magnitude. Besides, the high value ofthe coefficient of correlation (r0.99) confirmsthat the BrunauerEmmettTeller (BET) method[10] is the correct one and gives the opportunityto measure the real physical magnitude of thespecific surface area SKr for the dispersed silica(and other oxide dispersed solids).

The values of the silanol number for 100 SiO2samples aOH, depending on their specific surfacearea S, are shown in Fig. 7 [301303]. The shadedhorizontal band is the range of experimental data,and for a completely hydroxylated surface thesilanol number aOH is mainly between 4.2 and 5.7OH nm2 (Table 5). But, as already mentioned,the samples of amorphous silicas were preparedby different methods and had different structuralcharacteristics. The horizontal (broken) lines areaverage values of the silanol number: aOH,AVER4.6 OH nm2 (least-squares method) andaOH,AVER4.9 OH nm2 (arithmetical mean)(see also Fig. 6). Thus, the silanol number aOH isindependent of the origin and structural charac-teristics of amorphous silica.

Our measurements carried out by DE and othermethods showed that different types of amor-phous dispersed silica contain not only surfaceOH groups (dOH(S) ), but also structurally boundwater inside the silica skeleton and inside the veryfine ultramicropores (dOH(V) ) [301303,305

pores are changed within range from about 1.0 to1000 nm and higher. This shows that the concen-tration of hydroxyl groups dOH(S) is directly propor-tional to the specific surface area S of theamorphous silica samples under investigation. Itmeans that we have indeed determined only thesurface concentration of OH groups, dOH(S) .Straight (solid) lines passing through the origin ofcoordinates are average values of the silanol num-ber (Fig. 6): aOH,AVER (dOH(S) :S)AVER4.61$4.690.5 OH nm2 (least-squares method) andaOH,AVER4.89$4.990.5 OH nm2 (arithmeti-

Fig. 7. Concentration of the surface hydroxyl groups (the silanol number) aOH for silicas having different specific surface areas S,when the surface has been hydroxylated to a maximum degree: symbols ai indicate different types of amorphous silica (see Table5 and text); the shaded area is the range of experimental data (100 samples of SiO2 with different SKr values from 9.5 to 945 m

2

g1); broken lines are average values of the silanol number: aOH,AVER4.6 OH nm2 (least-squares method) and aOH,AVER4.9

OH nm2 (arithmetical mean) (see Fig. 6).


311,313,314,316319]. Thus, the dOH(S) value is onlya part of the total content of OH groups dOH insilica; the other part of OH groups, dOH(V) dOHdOH

(S) , is contained within the bulk. According toour IR spectral measurements (together withDavydov and Kiselev) the latter consists of silanolgroups inside the silica sample (the absorptionband of stretching vibrations is about 3650 cm1)[309]. The distribution of OH groups between thesurface and the volume of the sample depends ona number of factors, but mainly on the method ofpreparation of the silica sample and its subse-quent treatment.

The total concentration of OH groups, gOH (i.e.the number of OH groups), for one arbitrary SiO2particle can be represented as the sum [303,316]:

gOHk16k2s (3)

where 6 and s are the volume and surface of theparticle, respectively, k1 and k2 are the propor-tionality coefficients, within the volume and sur-face, respectively, for a certain standardmaximum-hydroxylated state of silica both insidethe bulk of the particle and on its surface. Themagnitude of gOH is related to the total concentra-tion of the structural hydroxyl groups dOH perunit mass of silica by:

gOHdOH6g (4)

where g is the density of the silica skeleton andm6g is the mass of this SiO2 particle. Moreover,s is expressed through the specific surface area Sas follows:

sS6g (5)

By using Eqs. (4) and (5) we can determine theratio dOH:S. This ratio describes the concentrationof all OH groups within the bulk and on thesurface of silica per unit surface area:

dOH:Sk1:Sgk2 (6)

If coefficients k1 and k2aOH are constant andare independent of the structure of silica, i.e. ofthe specific surface area S, then Eq. (6), within thecoordinates dOH:S and S, stands for the equationof an equilateral hyperbola relative to the asymp-totes dOH:S and k2aOH. If we assume that thestructural hydroxyl groups are located only on the

surface, then k10 and dOH:Sk2aOHconstant.

Therefore within the coordinates dOH:S and Swe obtain two boundary curves (Fig. 8(a)): ahyperbola B (dOH:Sk1:SgaOH), which ex-presses the dependence of the dOH:S ratio on thespecific surface area S for a maximum-hydroxy-lated state both inside the bulk of the sample andon its surface, and a straight line A (dOH:SaOH), which describes only the surface concentra-tion of hydroxyl groups, with no account beingtaken of the structurally bound water inside thesilica skeleton.

An analysis of the experimental results describ-ing the concentration of structurally bound waterwithin the coordinates dOH:S and S, obtained bymany authors for the maximum-hydroxylated sili-cas [1,16,2027,3134,3739,4549,5358,6567,71 83,88,92,97,99 102,106 109,130,159 161,194,214,215,233,234,253,254,280], including ourdata [301311,313320,323327,334338], showsthe following (Fig. 8(b)). First, the values of k2aOH, obtained in our work using the DE methodfor 100 different samples of amorphous silica(Fig. 8(a) and (b), open circles within a red band,cf. also Fig. 7), lie longitudinally with respect tothe lower boundary straight line A, and k2aOH4.6 (4.9) OH nm2. Secondly, the experi-mental points reported by other researchers andby us using different methods, for more than 170different samples of SiO2 with S in the range4.3960 m2 g1, are located throughout the spacebetween the upper hyperbola B and the lowerstraight line A and represented the total contentof OH groups (Fig. 8(a) and (b), black triangleswithin a green zone and a red band).

Thus the region of experimental points betweenthe boundary curves A and B (Fig. 8(a)) shouldbe considered to be the region of internal struc-turally water, i.e. of internal hydroxyl groups.Therefore, the experimental data reported in theliterature, which are usually based on the loss ofmass during calcination of the sample at hightemperatures, describe the total concentration ofOH groups in the sample, both within the bulkand on the surface (the ordinate of the pointsexceeds or equals k2aOH, Fig. 8(a) and (b)). It isto be expected that with a decrease in S, i.e. with


Fig. 8. (a) Theoretical dependence of dOH:S on the specific surface area S for a model sample of SiO2 in a maximum-hydroxylatedstate both inside the bulk of the sample and on its surface [303,316]: (i) equilateral hyperbola B, the samples contain OH groups bothin the bulk and on the surface of the sample (the upper boundary); (ii) straight line A, the samples contain only surface OH groups(the lower boundary). (b) dOH:S as function of the specific surface area S obtained by different researchers for different samples witha maximum-hydroxylated surface (see text): (i) a green zone, and also a red zone, corresponding to experimental data (blacktriangles, the total content of OH groups) obtained by different authors and by us (for more than 170 samples of SiO2 with S inthe range 4.3960 m2 g1); (ii) only a red zone (band) corresponding to experimental data (open circles, the surface concentrationof OH groups) obtained by us by the DE method (100 samples with S in the range 9.5945 m2 g1, Fig. 7 and Table 5).

an increase in the size of the silica particles, therelative concentration of internal structurallybound water for freshly prepared samples canincrease. This explains the shape of curve B forthe maximum-hydroxylated sample. The con-stancy of k2aOH, which is independent of S,shows that the silanol number for the maximum-hydroxylated state of surface is represented by areproducible physico-chemical constant (Figs. 7and 8(a) and (b)).

Next we shall examine the data reported inliterature on the theoretical and crystallochemicalestimates of the concentration of OH groups fordifferent types of silica hydroxylated to a maxi-mum degree. In 1950s Iler and De Boer andVleeskens developed two basic models.

Based on the geometry of spherical SiO2 parti-cles and the density of amorphous silica (g2.20g cm3) Iler [1,23] estimated the number of Siatoms on the silica surface. The assumption wasthat for each surface Si atom there is one OHgroup, and Iler obtained the silanol numbers of

aOH7.85 OH nm2 [23] and 7.8 OH nm2 [1]by using two calculation variants. However, asBoehm noted [80], only half of the free valency Siatoms are capable of holding OH groups. There-fore, Boehm gives aOH3.93 OH nm2.

Another model was proposed by De Boer andVleeskens [38]. It is based on a concept that thereis a similarity between the density and the refrac-tive index of amorphous and crystalline modifica-tions of silica cristobalite and that oftridymite. The authors pointed out that sinceb-cristobalite crystallizes in octahedra, the silanolnumber should be calculated for the octahedralface {111}. The value for aOH was found to be4.55 OH nm2. The density of a- and b-cristo-balite and that of a- and b-tridymite lie within anarrow range: from 2.20 to 2.34 g cm3. Byconsidering the most probable cleft surface ofthese SiO2 modifications the authors concludedthat the silanol number, aOH, lies in the rangefrom 4.55 to 4.85 OH nm2. The minimal silanolnumber for completely hydroxylated SiO2 surface


(both, for crystalline and amorphous modifica-tions) is aOH4.690.2 OH nm2. The averagearea for a single Si atom on the surface holdingone OH group is 0.21790.010 nm2 [38]. Be-sides, De Boer et al. [37,38], in their ther-mogravimetric investigations of wide-poreamorphous SiO2 it is subjected to pretreatmentat 600C and then rehydroxylated in an auto-clave, followed by the removal at 120C of thephysically adsorbed water, found that aOH4.55.0 OH nm2. This experimental value ispractically the same as the theoretical one. So,De Boer and Vleeskens were the first to obtaina reliable value for aOH. However, owing to thenon-correct thermogravimetric method they used(for determination of the concentration of OHgroups on the SiO2 surface), De Boer et al.[37,38] reached the erroneous conclusion that forthe starting hydrated samples of amorphous sil-ica which were not subjected to pretreatment at600C and then rehydroxylated, aOH (68)\aOH,MIN4.6 OH nm2. It is obvious thatsuch a conclusion fails to take into account thepresence of structural water inside the silica par-ticles and attributes the total loss of mass of thesample after high-temperature annealing only tothe loss of OH groups on the surface.

Zhuravlev, Kiselev et al. in their early worksusing the DE method [305311] determined ex-perimentally for the first time the value of aOHin the small interval 4.85.2 OH nm2 on thesurface for the different initial, fully hydroxy-lated samples of the wide-pore silicas. Thesesamples contained OH groups on the surface aswell as structural water inside SiO2 particles,and the authors observed that dOH:S\aOH oreven dOH:SaOH, i.e. the total content of OHgroups was much more than the content ofsilanols only on the surface.

The theoretical values were reported in litera-ture for the concentration of OH groups on thesurface of different types of silica [1,23,25,32,38,80,92,159,160,265,285]. These data show thatat present b-cristobalite model of De Boer andVleeskens is accepted as the correct one. Thus,for instance, Branda et al. [265], Sindorf and

Maciel [159,160] and Chuang and Maciel [285]when working out their models to describethe surface of amorphous silica, resorted to theconcept developed by De Boer and Vleeskens,but in these models they also took into ac-count the existence on the silica surface of acertain part of geminal silanols (besides the{111} face they considered the {100} face ofb-cristobalite).

As can be seen from Figs. 6 and 7, our exper-imentally obtained averaged values aOH,AVER4.6 (4.9) OH nm2 show that atmaximum-hydroxylated state of the amorphoussilica surface, following the activation of SiO2 invacuum at 180200C, each Si atom holds ap-proximately one or in some cases two OHgroups (geminal silanols).

To sum up, the magnitude of the silanol num-ber, which is independent of the origin andstructural characteristics of amorphous silica, isconsidered to be a physico-chemical constant[301303]. The results fully confirmed the ideapredicted earlier by Kiselev and co-workers[31,32,48] on the constansy of the silanol num-ber for a completely hydroxylated silica surface.This constant now has a numerical value:aOH,AVER 4.6 (4.9) OH nm2 (two calculationmethods, see above) and is known in literatureas the KiselevZhuravlev constant.

The constant aOH,AVER4.6 OH nm2 has apractical application: it can be used for deter-mining the specific surface area S (m2 g1) ofamorphous dispersed silica with a maximum-hydroxylated surface [316,317] (it is necessaryto remember about the high value of thecorrelation coefficient, r$1, between dOH(S) andS, see above). From Eq. (1%) we have the magni-tude S :

SK dOH(S) :aOH,AVERK % dOH(S) (7)

where K602.214 and K %130.916 are con-stants, and dOH(S) (mmol OH groups g1 SiO2) isthe concentration of the silanol groups on thesilica surface referred to unit mass of SiO2, asdetermined by the DE method or by some otherindependent and correct method.


3.3. Dehydroxylation and rehydroxylation of thesilica surface. Distribution of 6arious types of sur-face groups. Energetic heteregeneity of the silicasurface

Many papers have been published on the differ-ent aspects of the subject mentioned in the title.This is understandable since the chemistry of thesilica surface as determined mainly by the concen-tration, the distribution and the reactivity of thesurface silanol groups is of theoretical and practi-cal importance.

Let us consider the dehydroxylation process ofthe silica surface. The DE method was used todetermine the average value of the silanol num-ber, depending on the temperature of pretreat-ment in vacuo. The experimental results obtainedfor 16 samples of amorphous silica are shown inFig. 9. The samples differed from one another inthe method of their synthesis and their structuralcharacteristics: the specific surface area of thesamples varied from 11 to 905 m2 g1, and theirporosity also varied within a wide range. Despitethese differences the value of aOH at a giventemperature of treatment is similar for all thesamples, and the decrease in the value of aOHunder similar heating conditions also follows ap-proximately the same pattern. In Table 6 (column

2) are presented the most probable values ofaOHaOH,T (concentration of the total OHgroups averaged according to data in Fig. 9),beginning with the state of maximum hydroxyla-tion (180200C, the first line), followed by thevalues for the degree of surface coverage with thetotal hydroxyl groups, uOHuOH,T (column 8)[301303]. As can be seen from data in Fig. 9 andTable 6, the total values of aOH decrease consider-ably in the range from 200 to about 400C; be-tween 400 and 1100C decrease becomes notablysmaller. Correspondingly the total value of uOH inthe first steep section of the plot decreases from 1to about 0.5, and in the second, more flat sectionit drops from 0.5 to very small values approach-ing zero.

Our results [301303,305311,313315,317,334338] accord well with the data on the dehy-droxylation of the amorphous silica surface ob-tained by Fripiat, Uytterhoeven et al. [58,74].These authors determined the silanol number aOHon the basis of the reaction of OH groups withorganometallic reagents CH3Li and CH3MgI.Similarly, there is good agreement between ourresults and the data obtained by Taylor, Hockeyet al. [73,82]. Here aOH was determined by theweight loss upon the heating of the SiO2 samplesfrom 115 to 1100C. Also, the data of Unger[112,146] derived from the reaction of OH groupswith CH3Li and the heavy water HTO accord wellwith our results. There is a qualitative agreementbetween our results and the data as reported byPapee [24], Gokcek and Boehm [80,81] and sev-eral other authors. The b-cristobalite models ofdehydroxylation process as developed by Brandaet al. [265], Sindorf and Maciel [159,160] andChuang and Maciel [285] support the evidence ofa two-stage temperature dependence of the silanolnumber aOH.

Based on data in Table 6 (columns 2 and 8) itis possible to assess the most probable values ofaOH and uOH within a wide range of temperaturesfor the treatment of SiO2 samples. These valuesare independent of the degree of silica dispersive-ness if the starting samples have been obtainedunder the condition of a completely hydroxylatedsurface. Many researchers use in their studies ourfunctionality of aOH f(TC) and uOH f(TC)

Fig. 9. Silanol number aOH as a function of the temperature ofpretreatment in vacuo for different samples of SiO2. Thebroken lines delimit the range of experimental data (16 sam-ples with different S from 11 to 905 m2 g1). The subregionsof dehydroxylation are: IIa from 200 to 400C, and IIbfrom 400 to 1100C (see text).


Tab

le6

Surf

ace

conc

entr

atio

nof

the

diff

eren

tty

pes

ofO

Hgr

oups

,Si

atom

san

dSi

OSi

brid

ges

(whi

char

efr

eeof

OH

grou

ps)

and

degr

eeof

cove

rage

ofth

esi

lica

surf

ace

wit

hth

ese

grou

psre

spec

tive

lya

asa

func

tion

ofpr

etre

atm

ent

tem

pera

ture

inva

cuo,

wit

hth

ein

itia

lst

ate

corr

espo

ndin

gto

the

max

imum

degr

eeof

surf

ace

hydr

oxyl

atio

n(fi

rst

line)

Isol

ated

OH

Tot

alO

HG

emin

alO

HV

icin

alO

HSi

atom

s,a

Si

SiO

Sibr

idge

s,u

OH

,Tu

OH

,Iu

OH

,Gb

uO

H,V

uS

iT

empe

ratu

reof

vacu

umgr

oups

,a

OH

,Vgr

oups

,a

OH

,T(S

inm

2)

grou

ps,a

OH

,Ia

SiO

Si

grou

ps,a

OH

,Gb

(OH

nm

2)

pret

reat

men

t,T

(SiO

Sinm

2)

(OH

nm

2)

(OH

nm

2)

(OH

nm

2)

(C

)

018

020

01.

004.

600.

260.

130.

610

1.20

0.60

2.80

00.

530.

770.

360.

110.

301.

050.

231.

4030

03.

551.

650.

501.

130.

510.

450.

060

400

0.49

2.35

2.05

0.30

02.

251.

400.

390.

340.

050

2.80

0.61

500

00.

251.

551.

800

3.10

1.55

0.33

0.29

0.04

00.

6760

01.

301.

500.

201.

730.

250.

200.

050

3.45

0.75

700

00.

250.

901.

151.

950.

150.

130.

020

800

0.85

0.70

0.60

0.10

03.

902.

100.

090.

090

04.

200.

9190

00

00.

400.

402.

1810

000.

050.

250.

050

00.

950.

250

04.

352.

230.

030.

030

04.

450.

970

1100

0.15

0.15

02.

3012

000

00

00

1.00

00

04.

60

aIn

dexe

s;T

,to

tal

OH

grou

ps;

I,is

olat

edO

Hgr

oups

;G

,ge

min

alO

Hgr

oups

;V

,vi

cina

lO

Hgr

oups

.b

Dat

ain

colu

mns

4an

d10

are

the

corr

ecti

ons

for

gem

inal

OH

grou

ps[1

59,1

60].


Fig. 10. Total mass thermogram 9 (rate of heating b5.8grad min1): I, intensity of the H2O peak (arbitrary units); IB,intensity of the background signal (a.u.); T, temperature (K);Tm, characteristic point [temperature (K)] corresponding to themaximal rate of desorption; A I, maximum of the differentialcurve; B, dip in the descending branch A IBC (see Fig. 5 andtext).

In Fig. 10 is shown the overall mass ther-mogram 9 obtained by the MTA-TPD methodand recorded by an electronic potentiometer. Thisdifferential thermal desorption curve, owing to itsnearness to the boundary curve 8 (Fig. 5), showspractically the complete spectrum of the hydroxylgroups removed from the silica surface. At firstthis surface corresponds to the maximum level ofhydroxylation, while at the end of the heating athigh temperature it becomes strongly dehydroxy-lated. In its appearance the thermogram (curve 9)is similar to the thermal kinetic curve obtained byYoung [41]. Its asymmetric form also indicates theenergetic non-uniformity of the surface.

It should be noted that both in the case of thethermogram curve 9 and in the case of otherkinetic curves 17 and 1013 (Figs. 5 and 10)there is a characteristic dip B on the descendingABC section. The differential curve 14 (Fig. 5)with a maximum in the region of the dip B wasdrawn by extrapolation, taking into account theform of the ascending section of the neighboringthermogram (curve 15). The beginning of thisthermogram (curve 14) corresponds to about400C and coincides with the temperature corre-sponding to a change in the slope of two descend-ing and approximately rectilinear sections of thetemperature dependence of the silanol numberaOH (Fig. 9). Therefore both the dip B itself andthe ascending section of curve 14 which passesthrough such a dip B practically delimit the differ-ent kinds of silanol groups on the silica surface(see below). The kinetic parameters for differentthermograms 917 are summarized in Tables 4and 8, and as can be seen from Fig. 11, theexperimentally determined kinetic order of thethermal desorption reaction n$2 for the wholeregion II.

The dependence of the activation energy ofdesorption ED on the temperature (Fig. 12) inregion II is characterized by two approximatelyrectilinear sections: in the range from 190 toabout 400C (subregion IIa) and above 400C(subregion IIb), with a notable change in the slopeas one goes from the first to the second section.Here too, the change in the slope at 400Ccorresponds to the change in the slope at the sametemperature between two sections of the tempera-ture dependence of aOH (Fig. 9).

as a set of physico-chemical constants (at fixedtemperatures).

Let us consider in detail the data obtained bythe MTA-TPD method (Fig. 5). These data coverthe entire region II which is confined to the left ofthe ascending section of the curve 8 and encom-passes thermograms 917. After preliminary ther-mal treatment in vacuo under the given condition(Table 1) the standard silica samples S-79 werefirst subjected to cooling up to room temperatureand then to linear heating in the 251000C tem-perature range. It should be noted that the ap-pearance of each separate mass thermogramwithin 90.5C corresponded to the temperatureof the preliminary treatment for a given sample.We have here a clear manifestation of the charac-teristic memory effect of the silica surface and itshydroxyl coverage with respect to the temperatureof preliminary thermal treatment in vacuo. Also,as noted above, with an increase in temperature ofpreliminary treatment of the S-79 sample there isan increase in temperature in region II whichcorresponds to the maximum of each individualmass thermogram (Fig. 5 and Table 1). All thisconfirms the fact that the silica surface which hasbeen subjected to different degrees of hydroxyla-tion is energetically non-uniform.


Table 7Values of the average area occupied by one OH group and the average distance between two adjacent OH groups followingpretreatment in vacuo of amorphous silica at different temperatures

Average value of the area occupied byTemperature of vacuum pretreatment, Average distance between two adjacent OHT (C) groups, L (nm)one OH group, vOH (nm

2)

Isolated silanolVicinal silanol Between vicinal Between isolatedsilanolssilanols

180200 0.450.068 0.295 0.760.42 0.2950.068 0.73300

400 0.43 0.740.55500 0.840.67 0.92600

700 0.87 1.051.43800 1.352.50 1.79900

1000 4.00 2.261100 6.67 2.91

From the known temperature dependences ofthe activation energy and the silanol number inregion II we can obtain the dependence of theactivation energy of dehydroxylation, ED, on theconcentration of OH groups, aOH, or the depen-dence of ED on the extent to which silica iscovered with hydroxyl groups, uOH, as shown inFig. 13 [302,303,332].

The second order of the reaction (Tables 4 and8 and Fig. 11) observed for the entire region II isdirect confirmation of associative desorption,which proceeds as a result of the reaction betweenthe surface silanol groups (condensation) leadingto the formation of siloxane bonds and molecularwater:

(SiOH) (SiOH)

(SiOSi )H2O (8)

Thus, the condensation, Eq. (8), is characteris-tic of both subregions (IIa and IIb) despite thefact that the activation energies, ED, for thesesections differ significantly (Figs. 12 and 13).

The components of the silanol number thesilanol number of free isolated OH groups, aOH,I,and the silanol number of vicinal OH groupsbound via the hydrogen bonds, aOH,V, on thesurface of SiO2 were determined (together withAgzamkhodzhaev and Galkin) by the DE methodand IR spectroscopic measurements, depending

on the temperature of the preliminary treatmentin vacuo (Fig. 14) [302,303,315]. The sample con-sisted of compressed aerosilogel (S330 m2 g1),which was free of narrow pores and having aninitially fully hydroxylated surface. It has beenshown that the vicinal hydroxyl groups can beremoved from the SiO2 surface by treatment inthe temperature range of 200 to 400C. How-ever, the intensity of the absorption band of freeisolated hydroxyl groups increases in the tempera-ture interval from 200 to about 400C, while

Table 8Dehydroxylation: determination of the kinetic parameters

ReactionThermokinetic Activation energy oforder ncurve desorption ED

kcal mol1 kJ mol1

(2.0)*10 20.9 87.3(2.0)*11 21.1 88.5(2.0)*12 22.7 94.9

13 (2.0)* 24.2 101.92.00 104.715 25.0

116.927.915 (2.0)*16 2.27 37.7 158.016 (2.0)* 44.8 187.4

1.9717 49.1 205.450.32.2617 210.7

(2.0)*17 49.6 207.6

* Parentheses indicate that n2 is taken as a known value.


Fig. 11. Reaction order of the thermal desorption of water, n,as a function of temperature T (C) of the pretreatment ofsilica in vacuo: region I, dehydration; region II, dehydroxyla-tion; (Tables 2, 4 and 8 and Fig. 12); the shaded bands are therange of experimental data.

also includes the values of aOH) of OH groups onthe surface of the sample, depending on the tem-perature of the pretreatment in vacuo. The con-

Fig. 13. Activation energy of water desorption, ED (in regionII): (i) as a function of the surface concentration of OHgroups, aOH; (ii) as a function of the surface coverage of SiO2with OH groups, uOH (see text).

Fig. 12. Activation energy of water desorption, ED, as afunction of temperature T (C) of the pretreatment of silica invacuo: (i) region I, dehydration; (ii) IIa, IIb, subregions ofregion II, dehydroxylation (Tables 2, 4 and 8, see text).

Fig. 14. Distribution of the surface groups as a function of thetemperature of pretreatment in vacuo (Zhuravlev model-1):curve 1, average concentration of the total OH groups, aOH,T;curve 2, average concentration of the free isolated OH groups,aOH,I; curve 3, average concentration of vicinal OH groupsbound through the hydrogen bonds, aOH,V; curve 4, averageconcentration of surface Si atoms that are part of the siloxanebridges and free of OH groups, aSi. The arrow indicates thecombined data obtained by the DE method and the IRspectroscopic method.

above 400C this intensity decreases. The correla-tion of data according to the silanol number at400C (Table 6, column 2) and the intensity of theIR absorption band of hydroxyl groups as shownin Fig. 14 is based on the fact that for the samplescalcined in vacuo at 400C there are practicallyonly free isolated hydroxyl groups [315]. Thismeans that the total concentration of silanolgroups on the silica surface at 400C, aOH,T, corre-sponds to the concentration of free isolated hy-droxyl groups, aOH,I. The concentration of freeisolated OH groups on the silica surface wascalculated from the absorption coefficient (which


Fig. 15. Distribution of free isolated OH groups: survey of thedifferent models (from Vansant et al. [263]): (1) Zhuravlevsmodel as a reference one (DE method) with an error region of90.5 OH nm2 (the shaded area); (2) Gillis-DHamers(pyridine); (3) Haukka (1H NMR); (4) Gillis-DHamers (H2Odesorption); (5) Gillis-DHamers (IR band shift); (6) Van DerVoort (IR integration); (7) Fink (IR deconvolution) (see text).

approximately constant up to 400C, and thenincrease considerably. Thus, for example, at1100C the area occupied by one isolated OHgroup is vOH$6.7 nm2 and the distance betweentwo free isolated adjacent OH groups is L$2.9nm (Table 7, columns 3 and 5).

It is interesting to compare the shape of thedistribution curve of free isolated OH groups,aOH,I, as a function of the pretreatment tempera-ture of SiO2, which we obtained by the DEmethod, with data reported in literature. Such acomparison (Fig. 15) has been made by Vansant,Van Der Voort and Vrancken in their monograph[263], where they considered seven different mod-els: (1) the model of Zhuravlev [302,303] as areference one with an error region of 90.5 OHnm2 (the shaded band, Fig. 15); (2) the model ofGillis-DHamers et al. [238], which is based onpyridine desorption from the silica surface; (3) themodel of Haukka et al. [253,254], which is basedon 1H NMR data for isolated and vicinal silanols;(4) the model of Gillis-DHamers et al. [263],which is based on the water desorption from thesilica surface; (5) the model of Gillis-DHamers etal. [238], which is based on the IR band shift offree isolated OH groups as a function of tempera-ture; (6) the model of Van Der Voort et al.[237,263], which is based on the integration of IRbands due to different types of OH groups; and(7) the model of Fink et al. [190], which is basedon the deconvolution of the summed-up IR bandinto constituent components belonging to differ-ent types of silanol groups. It can be seen fromFig. 15 that almost all models 17 are situated inthe error region. This shows good agreement be-tween the data obtained by different researchers.Furthermore, it confirms the existence of a realtemperature distribution of free isolated silanols,with a maximum at about 400C. It is necessaryto note that Yaroslavsky, by using the IR spec-troscopy method, discovered for the first time [11]a qualitative temperature course of free isolatedOH groups with the maximum at about 500C.

Thus, our results (Figs. 9 and 1214) indicatethat the dehydroxylation of the silica surface pro-ceeds via two stages: subregion IIa and subregionIIb.

centration of vicinal OH groups, aOH,V, at 200400C (Fig. 14, curve 3) and the concentration ofsurface Si atoms, aSi, that are part of the siloxanebridges SiOSi and free of hydroxylgroups, throughout the range of temperatures(Fig. 14, curve 4) can be determined from thetotal concentration of OH groups, aOH,T, and theconcentration of free isolated OH groups, aOH,I.

On the basis of our data one can calculate theaverage value of the area occupied by one OHgroup, vOH (nm2), as well as the average distancebetween these groups, L (nm), at temperaturerange of 2001100C (Table 7). Our IR spectro-scopic investigations (together with Davydov andKiselev) [309] showed that the difference instretching vibrations between the band of free OHgroups (3750 cm1) and the band due to vicinalOH groups (maximum at 3550 cm1) is Dn:200 cm1. According to Lippincott andSchroeder [347], the value for Dn corresponds to aprobable OHO distance of 0.295 nm. Thisgives vOH$0,068 nm2 per one vicinal OH group.It is not difficult to calculate the other values ofvOH and L. Since these OH groups are distributedat random, we can assume that their distributionis uniform throughout the silica surface. In thiscase, the average areas per one group, vOH, andthe average distances, L, between them remain


Let us consider the case where the degree ofcoverage of the silica surface with OH groups ishigh, 1]uOH\0.5, which corresponds to subre-gion IIa (Fig. 13). This range is characterized bythe linear energetic non-uniformity on the surface,which is described by the following empiricalrelationship:

ED31.412.3uOH (9)

where ED is the energy of activation in the range1925 kcal mol1 (the values ED for the initialnon-linear section in Fig. 13 are from 16.5 to 19kcal mol1). This activation energy ED is almostindependent of the silanol concentration and isdetermined mainly by a set of perturbations dueto vicinal OH groups. These perturbations ceasewith the disappearance of the vicinal groups at400C (at uOH$0.5). Thus, the subregion IIa ischaracterized by the presence of lateral interac-tions (hydrogen bonds) between the neighboringOH groups.

Let us consider the coverage of the silica sur-face with hydroxyl groups uOHB0.5, which corre-sponds to subregion IIb (Fig. 13). In this case, themain role is played by free isolated hydroxylgroups and siloxane bridges. This subregion ED isstrongly dependent on the concentration of hy-droxyl groups, i.e. the activation energy of des-orption is in the range from 25 to 50 kcalmol1 (and probably higher), and increases witha decrease in the concentration of the OH groups.When there are only free isolated silanols (and,perhaps, free geminals, see below) surrounded bysiloxane SiOSi bridges, the latter bridges can ac-quire a relatively large area following high tem-perature treatment of SiO2. Under theseconditions the main mechanism describing thetransfer of OH groups corresponds to condensa-tion Eq. (8) via disordered migration of protonson the surface (a process of the activated diffusionof OH groups). At the final stage, water isevolved, owing to the interaction of two OHgroups that accidentally approach each other to adistance of about 0.3 nm (a characteristic lengthof H-bond). The mechanism describing the migra-tion of protons is not entirely clear. It probablyinvolves the interaction of the protons at elevatedtemperatures with O atoms of the neighboring

SiOSi bridges, resulting in the formation of newsurface OH groups, which are displaced relativeto their initial position. In other words, this mech-anism can be represented as the transition fromone local minimum of the potential energy intoanother by means of jumps between the neigh-boring siloxane bridges. Obviously, at a low con-centration of OH groups such a diffusion ofprotons along the silica surface will limit thecondensation process Eq. (8).

In the monograph by Vansant et al. [263] arepresented experimental results of Gillis-DHamerset al., obtained by the TPD method. These resultspertain to the dehydration and the dehydroxyla-tion of water from the surface of silica gel samplewith a specific area of 400 m2 g1. The obtaineddata on the dependence of the activation energyof water desorption, ED, on the concentration ofOH groups, aOH, on the silica surface are in goodquality agreement with our previously data[302,303,332], although the absolute values of EDexceed our numerical values. Gillis-DHamers etal. have reported the appearance of three approx-imately linear sections. At high values of aOH(from 4.5 to 5.0 OH nm2) a sharp decline in EDis attributed to the transition from dehydration todehydroxylation (cf. section in the region of190C in Fig. 12). At values of aOH from approx-imately 4.5 to 2.0 OH nm2 there appears aweakly sloping section of ED which is due to thecondensation of bridged vicinal OH groups. Thiscorresponds to the subregion IIa in Figs. 12 and13. Finally, at silanol values of aOHB2.0 OHnm2 the sharply ascending section of ED is dueto the condensation of free hydroxyl groups onthe SiO2 surface, which correspond to the subre-gion IIb in Figs. 12 and 13 of our work.

In the last two decades, owing first of all to theinvestigations carried out by Sindorf and Maciel[159,160] and others based on 29Si NMR spectro-scopic measurements, it became possible to differ-entiate between single silanols and geminalsilanols on the silica surface. These results confi-rmed the hypothesis proposed earlier by Peri andHensley [79,92].

Therefore, it was necessary to correct our re-sults on the distribution of silanols with accounttaken of the presence of geminal silanols on the


SiO2 surface. Data on the content of geminalsilanols we took from the works

Documents

The Surface Chemistry of Amorphous Silica. Zhuravlev Model