Processing of low-density silica gel by critical point drying or ambient pressure drying

Processing of low-density silica gel by critical point drying orambient pressure dryingValerie nd,Thomas M.Harris *,Dale C.TeetersDepartment of Chemistry and Biochemistry,University of Tulsa,600S.College Ave.,Tulsa,OK74104-3189,USAReceived 5July 2000;received in revised form 20February 2001AbstractVarious methods for the production of low-density silica from silica gel were studied.The silica gel was prepared by the `two-step'sol±gel method.The `DSB process'(developed by Deshpande,Smith and Brinker),which takes the gel through solvent exchange,reaction with trimethylchlorosilane (TMCS)and ambient pressure drying (APD),was thenapplied.This processing provided a greater total pore volume,and more mesopores with diameters >50 A,than critical point drying (CPD),the conventional method for producing an aerogel.The high porosity was found to be due pri-marily to the solvent exchange step;in fact,the reaction with TMCS reduced the porosity.Reaction of the gel with trimethylmethoxysilane (TMMS)in the mother liquor (ethanol/water)provided higher pore volumes than the DSB process.This extra porosity may be attributed to poor wetting of the polar solvent in contact with the surface made hydrophobic through reaction with the TMMS;thus,the capillary forces that cause the gel to shrink are reduced.Finally,it was discovered that some loss of porosity occurs when CPD is conducted with carbon dioxide;speci®cally,the pressure pulse that occurs when the CPD cell is initially ¯ooded with this liquid can damage the structure of the silica network.Ó2001Elsevier Science B.V.All rights reserved.1.IntroductionAerogel materials exhibit very high surface area and signi®cant nanoporosity.A number of appli-cations for these materials have been explored [1].The use of silica aerogels as high e ciency thermal insulation has received the greatest attention [2,3].The conventional method of producing an aerogel,discovered by Kistler [4]in 1931,involves removal of the solvent from a `wet gel'under supercritical conditions.This technique,referred to as critical point drying (CPD),avoids the build up of tensilestresses that cause the polymeric network to col-lapse as the vapor±liquid interface recedes into the gel.Since CPD requires the use of an autoclave,the technological utilization of aerogels has lagged far behind their development in the laboratory.However,a discovery by Deshpande,Smith and Brinker [5]involving silica aerogels may ®nally allow the full commercial potential of these ma-terials to be realized.In the `Deshpande,Smith and Brinker (DSB)process',the wet gel is washed with an aprotic solvent,reacted with trimethyl-chlorosilane (TMCS)and then dried at ambient pressure.This treatment minimizes shrinkage of the gel,through a reduction in the surface tension of the solvent and the contact angle between the solvent and the surface of the silica network[6].Journal of Non-Crystalline Solids 283(2001)11±17/locate/jnoncrysol*Corresponding author.Tel.:+1-9186313090;fax:+1-9186313404.E-mail address:thomas-harris@ (T.M.Harris).0022-3093/01/$-see front matter Ó2001Elsevier Science B.V.All rights reserved.PII:S 0022-3093(01)00485-9In addition,as the last trace of solvent evaporates from the gel the shrunken silica network springs back to its original dimensions.This behavior in-dicates that the TMCS eliminates residual silanol groups which are otherwise capable of forming siloxane bridges across the pores and locking the network into the shrunken con®guration[7].In the DSB process,the wet gel initially pos-sesses an ethanol/water mixture as the solvent(the mother liquor).Since TMCS is chemically unsta-ble in this solvent,the DSB process includes a solvent exchange step in which the gel is washed with several portions of an inert solvent.In the present study the feasibility of utilizing azeotropic distillation to exchange the solvent,and substi-tuting a more stable silylating reagent for TMCS, were considered.In the course of this work fur-ther insight into the production of aerogels through ambient pressure drying(APD)has been gained.2.Experimental methodologyThe wet gel utilized throughout this study was prepared by the`two-step process'developed by Brinker and co-workers[8,9].The®rst step in-volves acid(hydrochloric)catalysis of the hydro-lysis of tetraethyl orthosilicate(TEOS).The second step involves condensation polymerization of the silanol groups resulting from the hydrolysis reactions.The condensation reaction is inhibited by the acid added in the®rst step;thus,the second step includes the addition of ammonium hydroxide to neutralize the acid.The®rst step was conducted in a reaction vessel (with water-cooled condenser)placed on top of a stirrer/hot plate.The reaction mixture,which in-cluded244ml of TEOS(Aldrich,98%pure),244 ml of absolute ethanol(Pharmco),19.6ml of de-ionized water and0.8ml of1.0M hydrochloric acid(Fisher Scienti®c),was heated at60°Cfor2h. The resultant`acidic sol'was then stored in a polyethylene bottle in a freezer(À20°C).In the second step of the process20ml of the acidic sol were combined with2.0ml of0.50M ammonium hydroxide solution(Fisher Scienti®c).The mixture was then placed in a sealed polyethylene container and aged for24h at50°Cin a water bath.Gela-tion occurs within minutes at this temperature. To exchange the solvent in the gel by`washing', the block of gel was contacted with three50ml portions of ethanol,followed by two50ml por-tions of an aprotic solvent(typically heptane).The ethanol washes were conducted at50°Cfor1h, while the aprotic solvent washes were for2h at 50°C.To exchange the solvent by azeotropic dis-tillation,the gel was broken up and transferred to a distillation apparatus(equipped with a Dean-Stark trap and addition funnel)containing75ml of toluene.The contents of the¯ask were then heated slowly to boiling.During the distillation, 25ml of toluene was added from the addition funnel each time25ml of distillate were recovered. Complete removal of the ethanol and water was indicated by a constant temperature of110°C,the boiling point of toluene.The®nal step of the DSB process involves reaction of the silica gel with TMCS.Additional inert solvent was added to the22ml of solvent-exchanged gel,and then3.0ml of TMCS(Aldrich, 98%)was added;this quantity corresponds to %0.55moles per mole of silicon in the silica network.The reaction was allowed to proceed for20h at50°C.An alternative reagent, trimethylmethoxysilane(TMMS)(Aldrich,98%) was also evaluated in this study.Since this com-pound is relatively stable in ethanol and water, solvent exchange did not need to be carried out prior to its reaction with the silica gel.Two di erent drying procedures were em-ployed.In both cases,the gel was®rst broken up into small chunks.For APD,the chunks were simply placed in a fume hood overnight.For CPD, which was conducted in a Polaron Jumbo cell, %100ml of solvent were added to the cell before the gel,to prevent inadvertent APD prior to the CPD process.The cell was then closed and cooled to18°C.Next,liquid carbon dioxide was allowed to®ll the remainder of the cell,and then displace the solvent through a valve at the bottom,while the cell was cooled to18°C.The liquid CO2supply was then shut o and a small vapor space was allowed to form at the top of the cell.Next,the cell was heated to38°C(in increments of2°C,at8min intervals),which converts the carbon dioxide tond et al./Journal of Non-Crystalline Solids283(2001)11±17the supercritical state.Finally the cell was vented slowly.Under these conditions the CO2escapes from the gel without generating a vapor±liquid interface.After APD or CPD,the gels were calcined at 500°Cfor15min.As indicated by Fourier-trans-form infrared spectroscopic analysis,this heat treatment completely evaporated the residual sol-vent and combusted all organic functional groups from the silica network.The surface area and porosity of the resultant materials were then de-termined through nitrogen physisorption analysis, using an Autosorb-1(Quantachrome).The5-point BET calculation was used to determine the surface area.The BJH model and the desorption isothermdata to calculate the cumulative pore volume as a function of pore diameter[10].Smith et al.[7]note that the surface tension of liquid nitrogen is not zero,and thus some shrinkage of an aerogel may occur during measurement of the desorption iso-therm.However,the calcination treatment em-ployed in this study should strengthen the silica network,and thus minimize this source of error. Several of the samples described in this report were replicated in order to assess the repeatability of the processing(as well as that of the surface area and pore volume determination).For exam-ple,three samples that were reacted with TMMS and then subjected to APD provided surface area values with a relative standard deviation(RSD)of 5.6%,and total pore volume values with an RSD of6.2%.3.ResultsPorosity data for silica gel resulting from the two-step process and subjected to APD or CPD are presented in Fig.1.APD of this gel yields a material with signi®cant porosity(0.68cm3/g of total pore volume).However,this porosity corre-sponds primarily to micropores(pore diameter <20 A)and small mesopores(20±50 A).When the gel is subjected to CPD the total pore volume is not substantially higher,but nearly half of this volume corresponds to intermediate mesopores (50±100 A)and large mesopores(>100 A).Porosity data for material resulting from the DSB process are presented in Fig.2.In this pro-cess the gel is subjected to washing with an aprotic solvent,reaction with TMCS and then APD. Hexane was employed in the washing step for the ®rst sample,while heptane was used for the sec-ond.In both cases the pore size distribution is essentially identical to that provided by CPD of the wet gel(Fig.1),but the total pore volume is actually greater(25%for hexane-washed,7%for heptane-washed).The e ect of the solvent washing step of the DSB process was examined.Surprisingly,APD after solvent washing alone provided a totalporeFig.2.Pore volume distribution data for gels subjected to APD:(a)washed with hexane,reacted with TMCS;(b)washed with heptane,reacted with TMCS;(c)washed with hexane only;(d)washed with heptane only.SA±surface area;TPV±total porevolume.Fig.1.Pore volume distribution data for silica gel(ethanol±water solvent)subjected to APD or CPD.SA±surface area; TPV±total pore volume.nd et al./Journal of Non-Crystalline Solids283(2001)11±1713volume that is signi®cantly higher than that re-sulting from the DSB process (Fig.2),as well as that resulting from CPD of the gel with its mother liquor (Fig.1).This increase corresponds primarily to a higher population of intermediate mesopores.Hexane provides more pore volume than heptane.The feasibility of substituting azeotropic distil-lation with toluene for solvent washing was con-sidered.It can be seen in Fig.3that when the gel is subjected to distillation in toluene,reaction with TMCS and then APD,a pore volume distribution nearly equal to that provided by the DSB process (Fig.2)is observed.Also presented in Fig.3are porosity data for a gel subjected only to toluene distillation and then APD.This treatment pro-vided signi®cantly more large and intermediate mesopores in comparison to the sample reacted with TMCS.However,toluene distillation alone did not provide as much total pore volume as hexane washing alone (Fig.2).Also considered in this study was the substitu-tion of a di erent silylating reagent for the TMCS employed in the DSB process.TMMS provides a potential advantage over TMCS in that it is more stable in ethanol or water;this allows the surface modi®cation reaction to be conducted in the mo-ther liquor,potentially eliminating the solvent ex-change step.Porosity data for a gel subjected to reaction with 0.55mol of TMMS (per mole of TEOS)and then APD are presented in Fig.4.Reaction conditions were identical to those em-ployed with TMCS (20h at 50°C).This material exhibits a higher total pore volume than that provided by the DSB process (1.05vs 0.96cm 3/g),due primarily to a larger volume of intermediate mesopores.It is important to emphasize that sol-vent exchange is not necessary,and was not uti-lized,to produce this material.In a related series of experiments gels were treated with increasing amounts of TMMS.It was observed that the vol-ume of large mesopores increased further as the amount of TMMS was increased to 0.82mol (per mole of TEOS).To examine the issue of wettability with respect to the TMMS-reacted surface,a gel treated in this manner was washed with heptane prior to APD.The porosity data for this material are also pre-sented in Fig.4.It can be seen that the additional solvent washing reduced the total pore volume only slightly;however,there was a substantial shift in the pore volume from large to intermediate mesopores.TMMS-treated gels were also subjected to CPD and a `hybrid'drying treatment.The porosity data for the materials resulting from these treatments are also presented in Fig.4.The gel subjected to CPD exhibited a lower pore volume than the gel subjected to APD.In comparison to the TMMS-treated gel that was subjected directly to APD,there was a substantial loss of pore volumecor-Fig.3.Pore volume distribution data for gels subjected to APD:distilled with toluene,reacted with TMCS;distilled with toluene only.SA ±surface area;TPV ±total porevolume.Fig.4.Pore volume distribution data for silica gels:(a)reacted with TMMS,subjected to APD;(b)reacted with TMMS,wa-shed with heptane,subjected to APD;(c)reacted with TMMS,subjected to CPD;(d)reacted with TMMS,exposed to liquid CO 2,subjected to APD.SA ±surface area;TPV ±total pore volume.nd et al./Journal of Non-Crystalline Solids 283(2001)11±17responding to intermediate mesopores.In the hy-brid treatment the gel was subjected to initial ¯ooding with CO2in the CPD cell,but was then removed from the cell and subjected to APD.The total pore volume and pore size distribution pro-vided by this processing were nearly identical to those provided by CPD alone.4.DiscussionAPD of the`two-step'silica gel produced in this study provides a material with a total pore volume nearly equal to that provided by CPD of this gel (Fig.1).However,with CPD much more of this pore volume corresponds to intermediate mesop-ores(50±100 A)and large mesopores(>100 A).It may be concluded that the larger mesopores shrink irreversibly to smaller mesopores during APD. The observation that the pores do not shrink all the way into the micropore range is probably due to this gel being relatively sti ,as a result of using base in the second step of the processing,and of aging at50°Cinstead of room temperature. When applied to the two-step gel employed throughout this study,the DSB process provides a pore size distribution that is essentially identical to that provided by CPD.However,the total pore volume provided by the DSB process was actually greater(by25%).Thus,the DSB process is not just a viable alternative to CPD;it can provide more mesoporosity than CPD with some precursor silica gels.This unexpected result is at least partly due to the relatively sti gel used here.The importance of the solvent washing step in the DSB solvent was investigated.Washing with an aprotic solvent followed by APD(Fig.2)pro-vided a total pore volume signi®cantly higher than that for the gel subjected to APD directly(Fig.1). Deshpande et al.[11]report that utilizing a solvent with a lower surface tension reduces shrinkage of the gel during APD.The results of the present study are consistent with this observation.The washing procedure used in the present study re-sulted in a mixture of hexane and ethanol in which the hexane content did not exceed75vol.%.The surface tensions of the pure liquids are0.0184and 0.0227N/m at20°C,respectively[12],so the sur-face tension of the mixture is%0.02N/m throughout the drying process.The mother liquor of the gel contains%15%water(added to promote hydrolysis during the sol±gel processing).Since the gel is washed with ethanol before being washed with hexane,the concentration of water should be quite low.However,this residual water will be the last solvent to evaporate during APD.Since the surface tension for pure water is0.073N/m,the surface tension of the solvent in the gel during the last stages of APD could be signi®cantly higher than0.020N/m.It was noted above that washing with an aprotic solvent followed by APD provided a total pore volume signi®cantly higher than that for the complete DSB process(Fig.2).From this it may be concluded that reaction with TMCS actually causes a loss of porosity.Hydrochloric acid is a by-product of the reaction between TMCS and the silica gel,and acid catalyzes the hydrolysis of sili-con alkoxide groups(including the trimethylsiloxy group).Thus,it is conceivable that the elimination of silanol groups is not complete under these conditions,and that further condensation poly-merization limits the`spring-back'that would otherwise occur in the last stages of APD. Another possible explanation for the lower porosity of the gels treated with TMCS relates to the contact angle between the solvent and the surface of the silica network.Greater shrinkage stresses are expected to develop when the solvent wets the silica network more e ectively,as mani-fested by a lower contact angle[6].Stein et al.[13] have demonstrated that reaction of silica with TMCS provides a hydrophobic surface that is ef-fectively wetted by non-polar organic solvents.In contrast,hexane(or heptane)should not wet the hydrophilic surface of the original silica gel(i.e., before reaction with TMCS);thus,less shrinkage would be expected when the gel is subjected only to solvent exchange and APD.The feasibility of substituting azeotropic distil-lation with toluene for solvent washing in the DSB process was also considered in this study.When this substitution was made(Fig.3),a pore volume distribution nearly equal to that provided by the DSB process(Fig.2)was observed.However, when the gel was subjected only to toluenend et al./Journal of Non-Crystalline Solids283(2001)11±1715distillation and APD,signi®cantly more large mesopores were preserved.This may result from a strengthening of the silica network during the distillation,which involves a much higher tem-perature(110°Cvs50°C)than that employed in the gel processing or the DSB process.Alterna-tively,this e ect may be due to more complete removal of the ethanol and water from the gel.In either case,azeotropic distillation is a convenient substitute for the time-consuming and wasteful solvent washing procedure.Also considered in this study was the substitu-tion of a di erent silylating reagent for the TMCS employed in the DSB process.The motivation for this search was the desire to conduct the reaction in the mother liquor of the gel,thus allowing elimination of the solvent exchange step alto-gether.Material produced with TMMS(Fig.4) exhibited a higher total pore volume than that from the DSB process(Fig.2),due primarily to a higher population of intermediate mesopores.In light of the discussion above the greater porosity provided by the TMMS-based processing may be attributed to two factors.First,since the reaction of TMMS with silica gel produces methanol in-stead of hydrochloric acid as a by-product,more complete elimination of silanol groups on the silica network would be expected.Secondly,the reaction with TMMS renders the surface of the silica net-work hydrophobic,which in this case is poorly wetted by the polar mother liquor.To test this last issue a gel reacted with TMMS was washed with heptane before being subjected to APD.The sub-stantially lower population of large mesopores in this material(Fig.4)belies greater shrinkage under these conditions,which is consistent with better wetting of the hydrophobic surface by the non-polar solvent.At the outset of this project it was assumed that CPD would provide the highest pore volumes. This assumption proved to be incorrect;almost all of the APD-based processing schemes provided higher total pore volumes,and greater populations of intermediate and large mesopores,than CPD-based processing.As noted above,very high pore volumes will not be obtained with a relatively sti gel,such as the one employed throughout this study.On the other hand,the head-to-head com-parison between APD-based and CPD-based processing in this study does point out that partial, irreversible shrinkage of the silica network can occur during CPD.Upon re¯ection it seemed likely that the gel was being damaged during the initial¯ooding of the CPD cell with the liquid carbon dioxide,which raises the pressure abruptly to a value in excess of850psi.To test this hy-pothesis,TMMS-treated gels were subjected to CPD or a hybrid drying treatment.The gel sub-jected to CPD exhibited less porosity than that subjected to APD(Fig.4).The porosity provided by the hybrid processing was nearly identical to that provided by CPD alone.Taken together, these results con®rm that the gel structure can be damaged by the pulse of high-pressure CO2at the beginning of the CPD process.5.Summary and conclusionsThe conventional means of producing silica in the aerogel form is to subject silica gel to CPD. Unfortunately,CPD is much less practical than APD.In this study,the DSB process(®rst reported by Deshpande,Smith and Brinker),which involves exchange of the mother liquor with an aprotic solvent,reaction with TMCS and then APD, provided a greater total pore volume,and more mesopores with diameters>50 A,than CPD. Surprisingly,the high porosity was found to be due primarily to the solvent exchange step;sub-sequent reaction of the gel with TMCS,which creates a hydrophobic surface on the silica net-work,actually reduces the porosity.This e ect is due to either the catalysis of additional hydrolysis by hydrochloric acid,a by-product of the TMCS reaction,or by enhanced wetting of the non-polar solvent on the hydrophobic silica surface. Reaction of the gel with another silylating re-agent,TMMS was also considered in this study. This reagent is stable in the mother liquor;thus, reaction of the gel with TMMS need not be pre-ceded by solvent exchange.APD of TMMS-trea-ted gels provided signi®cantly higher porosity than the DSB process.This discovery has important commercial consequences;silica aerogel can now be produced without the need for CPD(whichnd et al./Journal of Non-Crystalline Solids283(2001)11±17requires a high-pressure cell)or solvent exchange (which is time consuming and generates much solvent waste).The extra porosity provided by TMMS-treated gels is consistent with the expla-nations provided above for the decrease in pore volume associated with the use of TMCS.In theory no process that concludes with APD should provide porosity in excess of that provided by CPD.However,this was commonly observed in the present study.Subsequently it was discovered that some loss of porosity does occur when CPD is conducted with carbon dioxide as the supercritical ¯uid;damage to the silica network occurs during the initial¯ooding of the CPD cell,which abruptly raises the pressure to850psi.AcknowledgementsThis work was funded by the National Science Foundation,under grant number DMR-9971534. One of us(T.M.H.)would also like to thank C.J. Brinker of Sandia National Laboratories for sponsorship of a sabbatical featuring work with the DSB process.References[1]M.Schmidt,F.Schwertfeger,J.Non-Cryst.Solids225(1998)364.[2]D.M.Smith,A.Maskara,U.Boes,J.Non-Cryst.Solids225(1998)254.[3]D.Buttner,E.Hummer,J.Fricke,in:J.Fricke(Ed.),Aerogels,Springer,Berlin,1985,p.116.[4]S.Kistler,Nature127(1931)741.[5]R.Deshpande,D.Smith,C.J.Brinker,US Patent No.5,565,142,issued1996.[6]D.M.Smith,G.W.Scherer,J.M.Anderson,J.Non-Cryst.Solids188(1995)191.[7]D.M.Smith,D.Stein,J.M.Anderson,W.Ackerman,J.Non-Cryst.Solids186(1995)104.[8]C.J.Brinker,K.Keefer,D.Schaefer,C.Ashley,J.Non-Cryst.Solids48(1982)47.[9]C.J.Brinker,K.Keefer,D.Schaefer,R.Assink,B.Kay,C.Ashley,J.Non-Cryst.Solids63(1984)45.[10]E.P.Barrett,L.G.Joyner,P.P.Halenda,J.Am.Chem.Soc.73(1951)373.[11]R.Deshpande,D.Hua,D.Smith,C.J.Brinker,J.Non-Cryst.Solids144(1992)32.[12]R.C.Weast(Ed.),CRC Handbook of Chemistry andPhysics,63rd Ed.,CRC,Boca Raton,FL,1982,p.F-36.[13]D.J.Stein,A.Maskara,S.Haereid,J.M.Anderson,D.M.Smith,in:A.K.Cheetham,C.J.Brinker,M.L.Mecartney,C.Sanchez(Eds.),Better Ceramics Through ChemistryVI,Materials Research Society,Warrendale,PA,1994, p.643.nd et al./Journal of Non-Crystalline Solids283(2001)11±1717。