Effect of Microgravity on the Growth of SilicaNanostructures

David D. Smith,* Laurent Sibille,† Raymond J. Cronise, Arlon J. Hunt,‡Steven J. Oldenburg,§ Daniel Wolfe,§ and Naomi J. Halas§

Biotechnology Science Group, Microgravity Sciences and Applications Department,NASA Marshall Space Flight Center, SD-48, Huntsville, Alabama 35812

Received May 3, 2000. In Final Form: August 30, 2000

The effect of microgravity on the growth of silica nanoparticles via the sol-gel route is profound. In fourdifferent recipes that typically produce silica nanoparticles in unit gravity, low-density gel structures wereinstead formed in microgravity. These observations suggest that microgravity reduces the particle growthrate, allowing unincorporated species to form aggregates and ultimately gel. Hence microgravity favorsthe formation of more rarefied structures, providing a bias toward diffusion-limited aggregation. Moreover,these results add to evidence that the growth of silica nanoparticles occurs not simply through monomeraddition but by the attachment of smaller primary particles and aggregates.

Introduction

The process of the formation of structures from coagu-lating ensembles is fundamentally important since thecollective behavior of the constituents often results indramatically improved or unusual mechanical, thermal,chemical, and optical properties. Examples include col-loidal dispersions, sol-gels, thixotropic clays, liquidcrystals, ferrofluids, and colloidal crystals, which spanthe order parameter from random to quasi-fractalline tohighly ordered (yet tenuous) crystals. These structuresare typically characterized by weak collective interactionsthat result in mechanical softness and sensitivity tothermal fluctuations.1 Hence their formation is highlysensitive to gravity-induced perturbations such as flow-induced shear, sedimentation, convection, and hydrody-namic pressure. In this study we examine the effect ofmicrogravity on the formation of silica structures, specif-ically particles and gels.

There is previous evidence that the formation of this“soft” matter is altered in microgravity. The first com-mercially available products from space (still availablefrom NIST) were the monodisperse latex sphere standardsof Vanderhoff et al., who demonstrated that emulsionpolymerization of latexes in space resulted in bettermonodispersity, increased uniformity, and reduced co-agulation.2 In addition it has been hypothesized that inunit gravity, buoyancy-driven fluid flows and sedimenta-tion deleteriously perturb sol-gel substructures prior togelation, and these perturbations are “frozen” into theresulting microstructure.3,4 Hence, sol-gel pores may beexpected to be smaller, more uniform, and less rough when

formed in microgravity. Wessling et al.5 have reportedthat the formation of polyurethane foams in low gravityreduced the average void size, increased the pore round-ness, and narrowed the standard deviation in pore size.Using a shadowgraphic technique, Leontjev et al.6 ob-served fluid flows due to convection and sedimentationduring the formation of polyacrylamide gels and deducedfrom electrophoretic separations that the resulting poresize distributions were narrower for gels formed inmicrogravity. More recently Zhu et al.7,8 have shown thatcolloidal crystals of poly(methyl methacrylate) (PMMA)formed in microgravity are an order of magnitude largerand that completely different polymorphs can result.Okubo et al. have studied the kinetics of the formation ofcolloidal silica particles (both from aqueous silicates andfrom alkoxides) during parabolic aircraft flights (∼23 s ofmicrogravity per parabola) using dynamic light scatteringand transmission measurements and have found that theformation rate of silica particles is considerably reducedin microgravity.9

Stable silica nanoparticle dispersions may be formedeither by polymerization of silicic acids in an aqueoussystem or through hydrolysis and condensation of siliconalkoxides (the sol-gel or Stober route). These two routesare distinguished from one another by the mechanism ofparticle formation. Comparison of nuclear magneticresonance (NMR) spectra obtained from Ludox, a com-mercial aqueous silicate, with acid-catalyzed siliconalkoxides has demonstrated that solutions of the formerare dominated by monomers and tetrafunctionalized

* To whom correspondence should be addressed: (256) 544-8762(fax); (256) 544-7778 (phone); david.d.smith@msfc.nasa.gov.

† Universities Space Research Association.‡ Lawrence Berkeley National Laboratory, Energy and Environ-

ment Division.§ Department of Electrical and Computer Engineering and

Department of Chemistry, Rice University.(1) Rajagopalan R. In Ordering and Phase Transitions in Charged

Colloids; Arora, A. K., Tata, B. V. R., Eds.; VCH Publishers: New York,1996; p v.

(2) Vanderhoff J. W.; El-Aasser, M. S.; Micale, F. J.; Sudol, E. D.;Tseng, C. M.; Sheu, H. R. Polym. Prepr. 1987, 28, 455; Mater. Res. Soc.Symp. Proc. 1987, 87, 213.

(3) Noever, D. A. Microgravity Sci. Technol. 1994, 3, 14.

(4) Sibille, L.; Smith, D. D.; Cronise, R. J.; Noever, D. A.; Hunt, A.J. Proceedings of the Space Technology and Applications InternationalForum, 1st Conference on Commercial Development of Space, Albu-querque, NM, January 7-11, 1996; American Institute of Physics:Woodbury, N.Y., p 451.

(5) Wessling, F. C.; McManus, S. P.; Mathews, J.; Patel, D. J.Spacecraft Rockets 1990, 27 (8), 324.

(6) Leontjev, V. B.; Abdurakhmanov, Sh.D.; Levkovich, M. G.Proceedings of the AIAA Microgravity Science Symposium, Moscow,May 13-17, 1991; AIAA: Washington, DC, p 274.

(7) Zhu, J. X.; Li, M.; Phan, S. E.; Russel, W. B.; Chaikin, P. M.;Rogers, R.; Meyer, M. 3rd Microgravity Fluid Physics Conference, 1996;American Institute of Physics: Woodbury, N.Y., p 397.

(8) Zhu, J. X.; Li, M.; Rogers, R.; Meyer, W.; Ottewill, R. H.; STS-73Space Shuttle Crew; Russel, W. B.; Chaikin P. M. Nature 1997, 387,883.

(9) Okubo, T.; Tsuchida, A.; Kobayashi, K.; Kuno, A.; Morita, T.;Fujishima, M.; Kohno, Y. Colloid Polym. Sci. 1999, 277, 474.

10055Langmuir 2000, 16, 10055-10060

10.1021/la000643s CCC: $19.00 © 2000 American Chemical SocietyPublished on Web 12/02/2000

species,whereasdi-andtrifunctionalizedspeciesdominatefor alkoxides.10 Moreover, comparison of small-angle X-rayscattering (SAXS) measurements of Ludox with acid- andbase-catalyzed alkoxides shows that only aqueous silicatesols are uniform, whereas alkoxides generate fractalparticles.11 As Brinker points out,10 these results illustratethat sols derived from aqueous silicates are fully hydro-lyzed and grow by classical monomer addition resultingin uniform polymeric particles, whereas sols derived fromsilicon alkoxides grow through cluster aggregation andretain a fractal inner morphology even while the particlescoarsen through surface tension reorganization.

Two distinct regimes characterize particle growth:diffusion-limited, in which the transport of mass to thegrowing structure is the dominant limitation to growth,and reaction-limited, in which the efficiency of attachmentlimits thegrowthprocess.These tworegimesareuniversal;the structures formed in one regime are strikingly similareven from vastly different material systems.12 In general,diffusion-limited conditions result in a reduction in thegrowth rate because there is a decrease in the frequencyof collisions. Moreover, those species that do collide do nothave the chance to attach in a manner that minimizessurface energy; i.e., exterior sites are favored. As a result,aggregates formed in diffusion-limited conditions aredistinguished by lower fractal dimensions. Reaction-limited growth, on the other hand, is characterized bymore compact structures. The sticking coefficient is smallenough that species are able to sample attachment sitesfor energetically favorable configurations.

In this study the formation of silica Stober particles inmicrogravity is examined using transmission electronmicroscopy (TEM). Microgravity allows diffusion-limitedconditions to persist in recipes that typically are reaction-limited, essentially expanding the parameter space underwhich diffusion-limited conditions prevail and providingus with a snapshot of the aggregation process that wouldnot normally be accessible. In the case of silica nano-structures, microgravity provides a bias toward diffusion-limited cluster-cluster growth, altering structure for-mation and generally resulting in lower fractal dimensions.

MethodologyFour different recipes were developed in laboratory prepara-

tions using the Stober method.13,14 Silica Stober particles grownon the ground are of good quality in the range 100-700 nm.However, at certain precursor concentration ratios the particlesare either polydisperse, bimodal, rough, or partially aggregated.Hence the recipes were carefully chosen to examine these “failureconditions”, essentially spanning a large portion of the parameterspace over which Stober particles may be produced. In theselaboratory preparations, the alkoxide was freshly distilled andthe samples were sealed in quartz glass under N2 to preventwater in the atmosphere from influencing the reaction. The firstrecipe (R1) was a control sample chosen to produce the bestpossible particles in terms of monodispersity and sphericity. Thesecond recipe (R2) was chosen to produce the smallest Stoberparticles, which tend to be rough, irregular, and less monodis-perse. The third recipe (R3) was chosen to produce a bimodal sizedistribution, while the fourth recipe (R4) was chosen to producelarge irregular (nonspherical) particles. TEM images of the

laboratory-grown particles are shown in Figure 1. The stoichi-ometry of each recipe is shown in Table 1. Note that only recipesR3 and R4 contain additional water (the ammonium hydroxidereagent also contains water).

For the space-flight experiment, each 5 mL recipe was dividedinto two parts and loaded into coupled polyurethane (Hydex)syringes separated by a breakable Parafilm seal to enable mixingof the reactants. The first part consisted of tetraethyl orthosilicate(TEOS) and half the ethanol, while the second part consisted ofwater, ammonium hydroxide (30% NH3), and the remainingethanol. Each batch was also divided into ground and spacesamples, which were stoichiometrically identical. The onlydifference in the growth conditions between the ground and spacesamples was the presence or absence of gravity. The designatedspace samples (12 total, 3 per recipe) were then placed in theGelation of Sols, Applied Microgravity Research (GOSAMR)hardware, and activated aboard the space shuttle orbiter (MissionSTS-95) after microgravity conditions had been established. TheGOSAMR hardware, built by 3M Corporation and refurbishedfor this experiment, essentially consists of a set of modules, eachof which contains eight coupled syringe cartridges. Upon activa-tion a battery-powered motor-driven lead screw with a reversingactuator drives the syringe cartridges back and forth, which mixesthe solutions after breaking the barrier seals between them. Uponreturn of the flight samples, an ultraprobe sonicator was usedto obtain diluted suspensions of the samples in ethanol, andthese were allowed to evaporate onto carbon-coated copper TEMgrids. TEM images were recorded with a Philips JEOL 2010.

Results and DiscussionVisual inspection revealed that each of the space-grown

samples had formed marginally coherent low-density gels,and that these gels coexisted in the syringe with regionsof solvent. Ground-grown samples, on the other hand,remained in suspension. The resulting ground controlparticles were slightly different in size and polydispersityfrom the laboratory preparations (shown in Figure 1);presumably because for the ground-grown (and space-grown) samples the TEOS was not freshly distilled, thesamples could not be sealed under N2 and/or the reactionvessel was Hydex rather than quartz. TEM images forrecipes R1 and R2 are shown in Figure 2. Note that thereis a dramatic difference between the ground-grown silicastructures and those grown in microgravity. Whereasgrowth in unit gravity produces Stober particles, growthin microgravity favors loose gel structures. In fact forrecipes R1 and R2 it was difficult to find any Stoberparticles at all in the space-grown samples. However, allspace-grown samples did form gels, and these gels had acommon form and scale that was nearly recipe-indepen-dent. The particles making up the backbone of the gelwere elongated with diameters of approximately 10 nmand lengths of about 50 nm. These gels are similar to thestructures observed by Yoshida, who added a calcium saltand a sodium hydroxide solution to a polymerizing silicicacid sol and heated the mixture in an autoclave to effectnonuniform particle growth.15

For the recipes containing added water, R3 and R4, theparticles making up the gel backbone were slightly widerand less elongated. In addition, a few large spherescoexisted with the more prevalent gel structure as shownin Figure 3. These spheres tended to be smaller (in some

(10) Brinker, C. J. In The Colloid Chemistry of Silica; Bergna, H. E.,Ed.; American Chemical Society: Washington, D.C, 1994; p 361.

(11) Schaefer, D. W.; Martin, J. E.; Keefer; K. D. In Physics of FinelyDivided Matter; Bocarra, N., Daoud, M., Eds.; Springer-Verlag: Berlin,1985; p 31.

(12) Lin, M. Y.; Linday, H. M.; Weitz, D. A.; Ball, R. C.; Klein, R.;Meakin, P. Nature 1989, 339, 360.

(13) Stober, W.; Fink, A.; Bohn, E. J. Colloid Interface Sci. 1968, 26,62.

(14) Averitt, R. D.; Sarkar, D.; Halas, N. J. Phys. Rev. Lett. 1997, 78,4217.

(15) Yoshida, A. In The Colloid Chemistry of Silica; Bergna, H. E.,Ed.; American Chemical Society: Washington, D.C., 1994; p 51.

Table 1. Stoichiometry of Silica Sol-Gel Recipes

recipe TEOS (mL) ethanol (mL) water (mL) NH4OH (mL)

R1 0.140 4.21 n.a. 0.654R2 0.153 4.59 n.a. 0.245R3 0.446 3.81 0.576 0.170R4 0.335 3.58 0.420 0.665

10056 Langmuir, Vol. 16, No. 26, 2000 Smith et al.

cases only about half as large) and have much larger sizedistributions than Stober particles formed on the ground.As shown by Bogush and Zukowski,16,17 the coexistenceof large monodisperse spheres (50-250 nm) with smaller(∼10 nm) aggregating primary particles implies that thegrowth of silica Stober particles does not occur by theclassical nucleation and growth model, where nucleationoccurs over a span of time limited by the decreasingavailability of monomer. Rather Bogush and Zukowskideduced that nucleation of particles proceeds continuouslythroughout the reaction. The smaller primary particlesform by the classical monomer addition growth mechanismand then aggregate because of their small size, until theybecome colloidally stable. Bogush and Zukoski proposethat the resulting stable aggregates are the building blocksfor the formation of Stober particles, collecting smalleraggregates and newly formed particles as they aretransported through the solution. Therefore, in this view,reaction-limited conditions must persist to maintainsmooth spherical particles. The final structure then

coarsens through surface tension reorganization to formthe resulting Stober particle.

In contradiction to Bogush and Zukoski, Harris et al.18,19

and also van Blaaderen and Vrij20 have argued that ifgrowth continued to occur through aggregation of sub-particles, smooth spherical particles cannot result. In theirview, Stober particles initially grow by aggregation ofsubparticles but monomer addition later fills in thenonuniformities, resulting in a smooth particle. Theirregular shape of smaller Stober particles is a remnantof the aggregation mechanism not yet enveloped by thesubsequent monomer growth. The size difference betweenthe subparticles and the resulting Stober particles cer-tainly supports this view, since this difference is likelytoo small to yield a smooth surface even when reaction-limited conditions prevail. Furthermore, although only afew Stober particles formed in microgravity, those thatdid form were smooth. Hence, the fact that smooth

(16) Bogush, G. H.; Zukoski, C. F. In Proceedings of the 44th AnnualMeeting of the Electron Microscopy Society of America; Bailey, G. W.,Ed.; San Francisco Press: San Francisco, CA, 1986; p 846.

(17) Bogush, G. H.; Zukoski, C. F. In Ultrastructure Processing ofAdvanced Ceramics; Mackenzie, J. D., Ulrich, D. R., Eds.; Wiley: NewYork, 1988; p 477.

(18) Harris, M. T.; Basaran, O. A.; Byers, C. H. In UltrastructureProcessing of Advanced Ceramics; Mackenzie, J. D., Ulrich, D. R., Eds.;Wiley: New York, 1988; p 843.

(19) Harris, M. T.; Brunson, R. R.; Byers, C. H. J. Non-Cryst. Solids1990, 121, 397.

(20) Van Blaaderen, A.; Vrij, A. In The Colloid Chemistry of Silica;Bergna, H. E., Ed.; American Chemical Society: Washington, D.C.,1994; p 83.

Figure 1. Four recipes R1-R4, prepared in laboratory glassware.

Microgravity and Silica Nanostructure Growth Langmuir, Vol. 16, No. 26, 2000 10057

particles are obtained even in the absence of reaction-limited conditions further supports this view.

Thus, it is not valid to consider silica sol-gels as eitherparticulate or polymeric; they are both. The small ∼10nm subparticles are primarily polymeric, representing thesolubility limit of the molecule as a result of its increasingsize and degree of cross-linking.21 As pointed out by Baileyand Mecartney,22 upon falling out of solution thesepolymeric precursors collapse, ultimately resulting in acompact particle due to continued hydrolysis and con-densation. On the other hand, Stober particles are at leastto some extent particulate, initiated from stable “seeds”formedbyaggregationof subparticles (or solublemicrogels,in Bailey and Mecartney’s view), and later smoothed outdue to continued addition of monomers (and other solublespecies).

Colloid stability therefore plays an important role insilica particle formation and morphology to the extentthat it determines the size of the aggregates (or microgels)that constitute and augment the particle early in the

growth process. According to the Derjaguin, Landau,Verwey, and Overbeek (DLVO) theory,23 colloid stabilityis greatly affected by ionic strength, and the presence ofwater stabilizes these constitutive aggregates at smallerradii. Hence particle nucleation and growth is more readilyestablished in the presence of water, which in part explainsthe greater population of Stober particles in the water-containing recipes R3 and R4. In addition, the kinetics ofthe formation of these aggregates depends on the degreeto which the system is diffusion-limited or reaction-limited.Hence microgravity results in a decrease in the rate offormation of these constitutive aggregates due to a biastoward diffusion-limited conditions. Significant monomerdepletion (into subparticles and smaller soluble species)then occurs faster than the time it takes for stableaggregates to form and initiate Stober particle growth,leading to a preponderance of unstable subparticles andaggregates that eventually compose the loosely formedgel that we observe.

Although colloid stability is not affected by microgravitydirectly, one may be tempted to think that microgravity

(21) Flory, P. J. Principles of Polymer Chemistry; Cornell UniversityPress: Ithaca, NY, 1953.

(22) Bailey, J. K.; Mecartney, M. L. Colloids Surf. 1992, 63, 151.(23) Verwey, E. J. W.; Overbeek, J. Th. G. Theory of the Stability of

Lyophobic Colloids; Elsevier: New York, 1948.

Figure 2. TEM photos of silica Stober particles formed from recipes R1 (A and B) and R2 (C and D) on the ground (A and C) andin microgravity (B and D). Low-density gels were always observed in microgravity, but only in recipes R3 and R4 could Stoberparticles frequently be found.

10058 Langmuir, Vol. 16, No. 26, 2000 Smith et al.

affects stability indirectly, in that reduced mass transportresults in structures that are more extended and henceless stable, resulting in an increase in the critical radiusfor a stable seed. However, the fact that a few Stoberparticles did form in microgravity (Figure 2) and that theywere not larger in size (they were generally smaller) thanthose formed on the ground counters such a conclusion.Thus, microgravity probably does not increase appreciablythe critical seed radius, but rather results in an increasein the average time it takes to form such a nucleus owingto decreased mass transport. Conceptually the resultinggel is not unlike an infinite Stober particle wheremonomers and other soluble species are depleted beforethey can fill in the voids, resulting in a nonuniformitysomewhat reminiscent of the fractal remnants formedearly in the growth of (noninfinite) Stober particlesthemselves.

Since diffusion is present at any level of gravity, whereasbuoyancy-driven convection is not, the effect of micro-gravity on Stober particle growth can be understood mostreadily through its effect on diffusion. According to theStokes-Einstein relation, the diffusion coefficient Dm fora particle of mass m undergoing Brownian diffusion is

where reff is the effective radius of the particle and d(g)is the mass fractal dimension (which depends in somemanner on gravitational acceleration g). Note that thediffusion coefficient only depends on gravity indirectly,through the dependence of the fractal dimension ongravitationally dependent transport mechanisms (convec-tion and sedimentation). Various experiments and com-puter simulations have demonstrated that in general bothdiffusion-limited conditions and cluster-cluster aggrega-tion produce more extended structures than reaction-limited monomer growth conditions, resulting in smallerfractal dimensions.24,25 Because in the Stober route thedominant growth mechanism changes with substructuresize, the bias toward diffusion-limited conditions obtainedin microgravity leads to a larger decrease in d for largersubstructures; i.e., the fractal dimension of the smallsubparticles (and smaller species) is not decreased ap-

(24) Meakin, P. In On Growth and Form; Stanley, H. E., Ostrowsky,N., Eds.; Martinus-Nijhoff: Boston, MA, 1986; p 111.

(25) Meakin, P. Annu. Rev. Phys. Chem. 1988, 39, 237.

Figure 3. TEM photos of silica particles formed from recipes R3 (A and B) and R4 (C and D) on the ground (A and C) and inmicrogravity (B and D). The photos suggest that cluster aggregation is an important mechanism for the formation of Stoberparticles; i.e., the basic building block for the formation of a particle is a smaller particle or aggregate.

Dm ∝ 1reff

) 1m1/d(g)

(1)

Microgravity and Silica Nanostructure Growth Langmuir, Vol. 16, No. 26, 2000 10059

preciably compared to that of the large aggregates.Accordingly, from eq 1 it can be seen that the diffusioncoefficient is reduced in microgravity, and to a greaterdegree for larger substructures (aggregates). Hence thecollision frequency of larger substructures, and conse-quently the Stober particle induction rate, is reduced inmicrogravity. For Stober particles that do manage to form,however, monomer addition again becomes important andthe diffusion coefficient of incorporating species increases.The effect of microgravity is therefore to further increasethe difference in the growth rates for different growthmechanisms. Equivalently, microgravity suppresses thecoagulation of subparticles and aggregates more dramati-cally than it does their formation through addition ofmonomers and other small soluble species.

ConclusionThe importance of the aggregation of unstable clusters

and subparticles to the formation and growth of silicaStober particles makes the effect of microgravity on Stoberparticle growth profound. Rather than simply retardingstructure growth (in this case a silica sol) as would beexpected for a singular growth mechanism, a pathway toan entirely different structure becomes available. Micro-gravity favors diffusion-limited conditions, which slowsthe formation of stable particle-forming aggregates.Monomers are consumed more by unstable subparticlesand aggregates than by Stober particles. Eventuallycluster-cluster aggregation is the only remaining growthmechanism which yields more extended structures, lead-ing to a decrease in the fractal dimension and ultimatelyto gel formation. We observed gels in microgravity at TEOS

concentrations as low as 2.8%. These results suggest thatmicrogravity favors the formation of more rarefied sol-gel structures, providing a bias toward diffusion-limitedaggregation. Indeed the softest of matter may only beachievable in microgravity, where entropic reductions andperturbations to structure formation are minimized.

Notably, our results are strikingly different than thoseof Vanderhoff et al., who achieved improved monodis-persity in latex particles.2 The difference arises becausetheir experiments involved emulsion polymerizations,which are stabilized in microgravity. Our future experi-ments in microgravity will investigate particle formationthrough emulsions, as well as seeded growth experimentsto promote monodisperse silica particle growth. In addi-tion, this study demonstrates that experiments in micro-gravity may help to clarify the mechanisms involved whereroutes to multiple structures are possible, i.e., to betterresolve competing mechanisms. For example, a compari-son of the growth of particles formed from aqueous silicatesin microgravity with those formed from silicon alcoxidesmay provide insight into the relative importance of clusteraddition mechanisms.

Acknowledgment. The authors gratefully acknowl-edge discussions with W. Ford, M. Ayers, M. S. Paley, andD. B. Wurm, as well as the contribution of J. Glenn, whoperformed the microgravity experiments. Funding supportwas provided by the Space Product Development office atMarshall Space Flight Center, NASA, the Office of NavalResearch, the National Science Foundation (ECS-9801707),and the Robert A. Welch Foundation.

LA000643S

10060 Langmuir, Vol. 16, No. 26, 2000 Smith et al.