ebook img

Sol-Gel Thin Film Formation CJ BRINKER,*1 AJ HURD, GC FRYE, PR SCHUNK and CS ASHLEY ... PDF

16 Pages·2010·5.2 MB·English
by  
Save to my drive
Quick download
Download
Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.

Preview Sol-Gel Thin Film Formation CJ BRINKER,*1 AJ HURD, GC FRYE, PR SCHUNK and CS ASHLEY ...

日本セラミックス協会学術論文誌 創立100周 年記念号 99 [10] 862-877 (1991) The Centennial Memorial Issue of The Ceramic Society of Japan Sol-Gel Thin Film Formation C. J. BRINKER,*1 A. J. HURD, G. C. FRYE, P. R. SCHUNK and C. S. ASHLEY Sandia National Laboratories, Albuquerque, New Mexico 87185- 5 8 0 0 , U S A ゾルゲル法による薄膜の形成 C. J. Brinker,*1 A. J. Hurd, G. C. Frye, P. R. Schunk and C. S. Ashley Sandia National Laboratories, Albuquerque, New Mexico 87185-5800, USA ディップ法 (又はスピニング法) によるゾルゲル薄膜の形成は急激な溶媒の蒸発を伴った重力 (又は遠心力) に よる溶媒の排出を含んでいる. ディップの途中で一定な膜厚プロファイルが形成され, そのプロファイルは構成 溶液の蒸発速度や流れによって生 じる表面張力の勾配によって決まる. ゾルに含まれている無機元素は溶媒の蒸 発とともに濃縮され, 膜形成プロセスの最後の段階で大きくなる毛細管力によって気液界面のメニスカスをゲル 内部へ引き込みち密化される. 生成膜の重要な特性である微構造 (気孔率, 表面積, 気孔径) は無機元素の種類 や大きさ, 膜の生成途中での濃縮や蒸発の相対的な速度, 毛細管力の大きさ, 生成に伴うせん断力に依存 してい る. [Received April 12, 1991] Sol-gel thin film formation by dipping (or spinning) in cal coatings (e. g., 4)-10)), passivation and planariza volves gravitational (or centrifugal) draining accompa tion layers (e. g., 11)-13)), sensors (e. g., 14), 15)), high nied by vigorous solvent evaporation. During dipping a or low dielectric constant films (e. g., 3), 16), 17)), inor steady-state film thickness profile develops, the shape ganic mem-braves (e. g., 18), 19)), electrooptic and of which is established by the evaporation rates of the non-linear optical films (e. g., 20)-22)), electrochrom fluid components and possible surface tension gradient ics (e. g., 23), 24)), semiconducting anti-static coat driven flows. The entrained inorganic species are con ings (e. g., 23)), superconducting films (e. g., 25)), centrated by the accompanying evaporation of solvent strengthening layers (e. g., 26) ,27)), and ferroelec and compacted by capillary pressure developed at the final stage of the deposition process as liquid-vapor trics (e. g., 28), 29). menisci recede into the gel interior. The important Despite significant advances in technologies based microstructural properties of the deposited film (% on sol-gel thin film processing, there has been rela porosity, surface area, and pore size) depend on the tively little effort directed toward understanding the size and structure of the inorganic species, the relative fundamentals of the sol-gel coating process itself. rates of condensation and evaporation during deposi see (e. g., 30)-39)). This paper reports on recent stu tion, the magnitude of the capillary pressure, and the dies that address the underlying physics and chemis accompanying shear stress. try of sol-gel thin film formation by dip- (or spin-) coating. We first briefly review the aspects of sol-gel Key-words: Sol-gel, Dip-coating, Spin-coating, Microstruc chemistry that govern the size and structure of the in ture, Films organic species which, along with the solvent (s), comprise the coating sol. We then discuss the salient features of dip- and spin-coating with consideration 1. Introduction of single component fluids and binary fluid mixtures. HIN films formed using sol-gel techniques We finally address the deposition of inorganic sols represent the oldest commercial application of with regard to time scales and the effects of sol struc sol-gel technology. The first patent based on sol-gel ture, capillary pressure, and the accompanying shear processing was granted to Jenaer Glaswerk Schott & on such properties as refractive index, surface area, Gen. in 1939 for silicate sol-gel films formed by dip and pore size of the deposited films, which we meas coating.1) Rear view mirrors produced by sol-gel thin ure using in situ techniques such as conventional and film techniques have been in production since 1953, imaging ellipsometry39) and gas sorption on surface followed by antireflective (AR) coatings in 1964, acoustic wave substrates.14) The goal of our research and architectural coatings since 1969.2),3) Today sol has been to establish generic schemes that enable gel thin film coatings are being intensively studied tailoring of film microstructures to specific applica for such diverse applications as protective and opti tions. *1 To whom correspondence should be addressed: Inorganic 2. Sol-gel chemistry Materials Chemistry Division 1846, Sandia National Labora tories, Albuquerque, NM 87185-5800, USA The sol-gel process uses inorganic or metal organ 862 C. J. BRINKER et al. Nippon Seramikkusu Kyokai Gakujutsu Ronbunshi The Centennial Memorial Issue 99 [10] 1991 (Journal of The Ceramic Society of Japan) 863 ic precursors.40) In aqueous or organic solvents the ligands such as acetylacetonate( acac)44)o r alcoho l precursors are hydrolyzed and condensed to form in amines,45),46t)o reduce both the effective func organic polymers composed of M-O-M bonds. For tionality44)a nd the rates of hydrolysis and inorganic precursors (salts) hydrolysis proceeds by condensation.44),4S6i)l icon,h owever, is substantiall y the removal of a proton from an aquo ion less electropositiveH.y drolysis and condensation of [MONH2N]z+ to form a hydroxo (M-OH) or oxo silicona lkoxides occur at much lower rates and the (M=O) ligand: condensation pathway can be more easilyi nfluenc e d [MONH2N]z++pH2O by sterica nd chemical factorss uch as the stericb ulk [MONH2N-p](z-p)++pH3O+ ( 1 ) of the alkoxide ligand,t he concentrationo f the acid Condensation reactions involving hydroxo ligands or base catalyst,t he molar ratio (r) of H2O: M in Eq. result in the formation of bridging hydroxyl (M- (2), or the aging times (see,F igs. 1 and 2.47),8)4 ) OH-M) or bridging oxygen (M-O-M) bonds de There is a considerable body of evidence based on pending on the coordination number of M and the acidity of the bridging hydroxyl. Normally monomer ic aquo rations are the only stable species at low pH MOLECULAR VERSUS PARTICLE CHAINS and various monomeric or oligomeric anions the only species observed at high pH.40) At intermediate pH, well-defined polynuclear ions are often the stable so lution species (e. g., [Zr4(OH)8(OH2)16]8+ or [AlO4 Al12(OH)24(OH2)12]7+), but the metal solubility is normally limited there, usually leading to the precipi tation of oxohydroxides, e. g., Al(O)OH41) or Fe(O)OH,42) or oxides, e. g., ZrO243) or TiO2.44) If peptized by the addition of acid or sterically stabi lized by adsorption of organic molecules, these col loidal sols are well-suited for film deposition by dip RLCA ping. D=2.09 The most commonly used organic precursors for ρ/∝r1/r(3-D)〜1 sol-gel film formation are metal alkoxides (M(OR)z), where R is an alkyl group (CxH2x+1).40) Normally the alkoxide is dissolved in alcohol and hydrolyzed by Fig. 1. Possible condensation pathways for silica polymeriza the addition of water under acidic, neutral or basic tion. Aqueous conditions (or for metal alkoxides, high H2O/Si), conditions, although film formation is also possible high pH, and/or high temperature promote the formation of com by deposition of the neat alkoxides followed by ex pact, particulate structures. Low H2O/Si, moderately acidic condi tions, and low temperatures lead to more extended structures posure to moisture. Hydrolysis replaces an alkoxide characterized by a mass fractal dimension D perhaps via reaction- ligand with a hydroxyl ligand: limited cluster aggregation (RLCA). M(OR)z+H2O_??_M(OR)z-1(OH)+ROH ( 2 ) Condensation reactions involving the hydroxyl ligands produce polymers composed of M-O-M or M-OH-M bonds plus, in most cases, the by- products water or alcohol as shown below for silicate condensation: Si(OR)3OH+Si(OR)4 _??_ (RO)3Si-O-Si(OR)3+ROH ( 3 ) 2Si(OR)3OH_??_(RO)3Si-O-Si(OR)3+H2O ( 4 ) The reverse of reactions (3) and (4), viz., siloxane bond alcoholysis and siloxane bond hydrolysis, pro mote bond breaking and reformation processes that, if extensive, permit complete restructuring of the growing polymer. Since metals of interest for sol-gel film formation (Si, Al, Ti, etc.) have coordination numbers (CN) Fig. 2. Average condensation rates (estimated as 1/tgel) for sili 4, complete condensation would lead to compact, cates prepared from tetraethoxysilane (top) or for aqueous particulate oxides. In fact, for electropositive metals silicates47) (bottom). Data in the top figure are from Coltrain et like Ti and Zr, it is difficult to avoid particle forma al.48) where p-TSA is p-toluenesulfonic acid, HOAc, acetic acid; TFA, tri-fluoroacetic acid. In the bottom figure, A and B tion unless the alkoxide precursor is modified, e. g., represent regions of high and low sticking probability, respective chelated with slowly hydrolyzing multidentate ly. 864 Sol-Gel Thin Film Formation 29Si NMR49)-51) and small-angle X-ray scattering (SAXS)52)-54) data indicating that, under many syn thesis conditions, condensation of silicon alkoxides results in the formation of randomly branched polymeric" silicates rather than dense silica parti cles. Silicate polymers are characterized using 29Si NMR by the distribution of Si(OR)x(OH)y(OSi)n species, or Qn distribution where n=4-x-y. The prevalence of Q1-Q3 species in Fig. 340) clearly indi cates that silicates may be very weakly condensed, compared, for example, to a commercial particulate silica sol.47) SAXS probes structure on the (cid:129)`0.5-50nm length scale. The Porod slope, P, of a plot of log scattered in tensity versus log scattering wave vector, K=(4ƒÎ/ ))sin(ƒÆ/2), is related to the mass fractal dimension, D, and the surface fractal dimension, Ds, by the fol Fig. 4. Porod plots (log scattered intensity versus log scattering lowing expression:53) wave vector [K=(4ƒÎ/ƒÉ)sin(ƒÆ/2)]) obtained by SAXS for a vari P=Ds-2D ( 5 ) ety of silicate sols. (A) two-step acid-catalyzed TEOS, (B) two-step acid-base cata For uniform (non-fractal) objects D=3, Ds=2, and lyzed TEOS, (C) one-step base-catalyzed TEOS (H2O/Si=1), P reduces to -4. For mass fractal objects, D=Ds, (D) one-step base-catalyzed TEOS (H2O/Si=2), and (E) com and P=-D. For surface fractal objects, D=3 and mercial particulate silicate sol (LUDOX SM). From Schaefer and P=Ds-6. Porod plots of a variety of silicates pre Keefer.54) pared from metal alkoxides are compared with the Porod plot of a commercial particulate silica sol in lence of fractal silicates over a wide range of process Fig. 4.54) It is generally believed40) that the preva ing conditions is a consequence of kinetically limited growth mechanisms such as reaction-limited cluster- aggregation (RLCA).55) The mass fractal dimension D relates an object's mass M to its radius r:56) (d ) M(cid:129)`rD ( 6 ) where for mass fractal objects, D is less than the dimension of space. The surface fractal dimension Ds relates an object's area A to its size:56) A(cid:129)årDs ( 7 ) where in three dimensions 2(cid:129)…Ds(cid:129)…3. Since in three (c ) dimensions, D<3, the density of a mass fractal ob ject decreases with distance from its center of mass: ƒÏ(cid:129)å 1/r(3-D) ( 8 ) Because density is inversely related to porosity, this relationship requires that, unlike Euclidian objects, fractal objects become more "porous" as their size in (b ) creases. We will show that this property may be ex ploited to tailor the pore structure of films deposited from fractal precursors. 3. Film formation (a ) 3.1 Dip-coating In dip-coating, the substrate is normally withdrawn vertically from the coating bath at a con stant speed Uo (see, Fig. 5a).57) The moving sub Fig. 3. Comparison of 29Si NMR spectra of silicate sols prepared from tetraethoxysilane (TEOS) to the spectrum of a commercial strate entrains the liquid in a fluid mechanical boun aqueous silicate sol (LUDOX HS40). dary layer that splits in two above the liquid bath sur (a) Acid-catalyzed TEOS sol (H2O/Si=2) after three hours at face, returning the outer layer to the bath (see Fig. room temperature; 5b).38) Since the solvent is evaporating and draining, (b) two-step acid-catalyzed TEOS sol (H2O/Si=5); the fluid film has an approximate wedge-like shape (c) multicomponent silicate sol (H2O/Si=5) aged for two weeks at 50(cid:129)Ž and (cid:129)`pH 3; that terminates in a well-defined drying line (x=0, in (d) LUDOX HS40. Fig. 5a). When the receding drying line velocity C. J. BRINKER et al. Nippon Seramikkusu Kyokai Gakujutsu Ronbunshi The Centennial Memorial Issue 99 [10] 1991 (Journal of The Ceramic Society of Japan) 865 SOL-GEL DIP-COATING (b) PORE SIZE CONTROLLED BY: -SIZE, STRUCTURE, COMPOSITION -RATES OF CONDENSATION/ EVAPORATION -CAPILLARY PRESSURE (a) (c) Fig. 5. (a) Schematic of the steady state dip-coating process, showing the sequential stages of structural development that result from draining accompanied by solvent evaporation and continued condensation reactions. (b) Detail of liquid flow patterns (streamlines) during dip coating. U is the withdrawal speed, S is the stagnation point, ƒÂ is the boundary layer, and h is the thickness of the fluid film. (c) Schematic of the imaging ellipsometer apparatus. Shaded area represents the sol reservoir and the entrained film. X is a beam expander; P a polarizer, C a quarter wave compensator; A an analyzer; N a diffusing screen; and V a video camera. From Hurd and Brinker.39) equals Uo, the process is steady state with respect to balance is modulated by the ratio of viscous drag to the liquid bath surface.39) liquid-vapor surface tension (ƒÁLV) according to the The hydrodynamic factors in dip-coating (pure liq relationship derived by Landau and Levich:58) uids, ignoring evaporation) were first calculated cor h=0.94(ƒÅUo)2/3/ƒÁLV1/6(ƒÏg)1/2 (10) rectly by Landau and Levich58) and recently general Figure 6 plots the logarithm of the product of thick ized by Wilson.59) In an excellent review of this top ness and refractive index*2 versus the logarithm of ic, Scriven38) states that the thickness of the deposit Uo, for films prepared from a variety of silicate sols in ed film is related to the position of the streamline which the precursor structures ranged from rather dividing the upward and downward moving layers. A weakly branched polymers characterized by a mass competition between as many as six forces in the fractal dimension to highly condensed particles.57) film deposition region governs the film thickness and The slopes are in the range 0.53 to 0.64, values position of the streamline:5) (1) viscous drag upward bounded by the expectations from Eqs. (1) and on the liquid by the moving substrate; (2) force of (2).*3 This reasonable correspondence between the gravity; (3) resultant force of surface tension in the thicknesses of the deposited films and a theory deve concavely shaped meniscus; (4) inertial force of the loped for gravitational draining of pure fluids sug boundary layer liquid arriving at the deposition gests that the entrainment of the inorganic species region; (5) surface tension gradient; and (6) the dis has little effect on the hydrodynamics of dip-coating, joining (or conjoining) pressure (important for films at least in the early stages of deposition where the en less than 1ƒÊm thick). trained sol is quite dilute. Thus some insight into sol- When the liquid viscosity (ƒÅ) and substrate speed gel film deposition may be gained by closer examina are high enough to lower the curvature of the gravita tion of the details of gravitational draining (and tional meniscus, the deposited film thickness (h) is evaporation) of pure (and binary) fluids. that which balances the viscous drag ((cid:129)åƒÅUo/h) and *2 Since the refractive index (n) is proportional to the volume gravity force (ƒÏgh):38) fraction solids (φ), the product hn is proportional to the h=c1(ƒÅUo/ƒÏg)1/2 (9) mass/unit area and takes into account the film porosity. *3 As discussed in Section 4.3, the particulate sols appear to where the constant c1 is about 0.8 for Newtonian liq order at higher Uo, which may contribute to the lower slopes uids. When the substrate speed and viscosity are low (0.53 and 0.58 compared to 0.62 and 0.64 observed for films (often the case for sol-gel film deposition), this prepared from polymeric sols). 866 Sol-Gel Thin Film Formation FILM THICKNESS DURING DIP COATING h=c1(ƒÅU/pg)1/2 WHEN ƒÅ, U ARE HIGH h=0.94(ƒÅU)2/3/ƒÁLV1/6(pg)1/2 WHEN ƒÅ, U ARE LOW Fig. 7. Thickness profile of dip-coated ethanol film (solid dots). The profile is quite well fit by the form h(cid:129)`xƒË with ƒË=0.50(cid:129)}0.01 (solid line). From Hurd and Brinker.60) E(x)=-Dva1x-1/2 (11) where Dv is the diffusion coefficient of the vapor ( Fig. 6. Product of thickness and refractive index (proportional to film mass/unit area) versus withdrawal rate plotted according 0.1cm2/s) and a1 is a constant. Since thickness va to Eqs. (9) or (10). From Brinker et al.57) ries inversely with evaporation rate, the divergence in the evaporation rate as x(cid:129)¨0 accounts for the blunt profile shown in Fig. 7, where the data are fit to the 3.2 Film thickness profiles during dip-coating: form: pure fluid h(x)(cid:129)`1/E(x)(cid:129)`xƒË, ƒË=0.50(cid:129)}0.01 (12) Previous theories of gravitational draining of pure according to the expectations from Eq. (11). fluids have not taken into account simultaneous The singularity strength (exponent) in Eq. (12) is evaporation. Although the thickness of the fluid en sensitive to the geometry of the film. For coating a trained at the bath surface is apparently not sensitive fiber, we would expect a logarithmic singularity,61) to evaporation, the film is progressively thinned by and as a coated cylinder decreases in radius, the evaporation as it is transported by the substrate profile should pass smoothly from x1/2 toward ln x. away from the coating bath. For depositing sols, thin Although some experimental evidence exists for this ning by evaporation causes a corresponding increase behavior by extrapolating to small cylinder in sol concentration, hence an understanding of diameters,61) it is a difficult proposition to prove ex simultaneous draining and evaporation is essential to perimentally. Surface tension effects make it difficult the underlying physics of sol-gel film deposition. to coat a fiber fast enough for the drying line to be In order to address this problem, Hurd and well-separated from the gravitational meniscus at Brinker39) developed an imaging ellipsometer that al the reservoir surface. lows acquisition of spatially resolved thickness and 3.3 Film fluid profiles during dip-coating: binary refractive index data over the entire area of the fluid depositing film (illustrated schematically in Fig. 5c). The most common coating sols are composed of A thickness profile of an ethanol film obtained by im two or more miscible liquids, e.g., ethanol-water. aging ellipsometry is shown in Fig. 7.60) Instead of Differences in evaporation rates and surface tensions the wedge expected for a constant evaporation rate, alter the shape of the fluid profile in the vicinity of the film profile is distinctly blunt near the drying line the drying line*4 and create connective flows within (x=0 in Figs. 5a and 7), which indicates more rapid the depositing film. Schunk62) has developed a finite thinning, and hence, a greater evaporation rate element model of the complete convective-diffusion there. problem associated with dip-coating an ethanol The most significant factor in the rate of evapora water mixture. His strategy is to combine the theory tion is the rate of diffusion of vapor away from the of convection and diffusion in the liquid with a mass film surface. The blade-like geometry of the deposit transfer model in the gas. Models and constitutive ing film in the vicinity of the drying line led Hurd60) equations are used to account for liquid-vapor to make an analogy between vapor diffusion (away equilibria and surface tension. Thus far thermal from the film) and the electrostatic potential sur effects induced by latent heat of evaporation and rounding a knife blade, viz., the electric field at the diffusion-driven convection are neglected*5 and surface is analogous to the evaporation rate (E) and the charge is analogous to the flux of vapor. Near any *4 In some cases rib-like instabilities are also observed in a sharp boundaries, E diverges, but the vaporized region near the liquid bath surface.61) *5 Bornside et al .63) have shown that for spin-coating with vola mass must remain integrable. For the knife blade ge tile solvents the temperature change of the depositing fluid ometry (infinite sheet), E varies with x as follows:60) is negligible ((cid:129)á1K at 3000rpm). C. J. BRINKER et al. Nippon Seramikkusu Kyokai Gakujutsu Ronbunshi The Centennial Memorial Issue 99 [10] 1991 (Journal of The Ceramic Society of Japan) 867 water is assumed to be relatively non-volatile. Con servation of mass and momentum (of both liquid and volatile species) leads to solutions for the position of the free surface (which locates according to the capil lary hydrodynamic forces and ethanol loss by evapo ration) and the concentration profiles and shown in Fig. 8.62) Preferential evaporation of ethanol enriches the free (liquid-vapor) surface in water. In the vicini ty of the drying line, the more volatile alcohol may be substantially depleted, leaving behind a water-rich film. Corresponding ellipsometric images61) of deposit ing alcohol-water films show two roughly parabolic features (Fig. 9), that correspond to successive drying of the alcohol-and water-rich regions, accord ing to the non-constant evaporation model (Eq. (12)). This suggests that each component has an indepen dent evaporation singularity. If the water-rich phase Fig. 9. Thickness profile of 50:50 propanol:water film is denoted as phase 1 and ethanol is phase 2, then the (volume ratio). The double parabolic profile is due to differential independent profiles are additive: volatilties and surface tension gradient driven flows. x1 is the posi h1=a1x1/2 x>0 (13) tion of the drying line; x2 is the position of the "false" drying line created by the depletion of the ethanol-rich phase. Film thickness h2=a2(x-x2)1/2 x>x2 (14) equals approximately the fringe order times 240nm. From h2=0 x<x2 (15) Hurd.61) where h is the total thickness, and x2 is the position of the "false" drying line created by the substantial depletion of ethanol. dƒÁ/dx=(ƒÁ1-ƒÁ2)dƒÓ1/dx(x>x2) The "foot" feature in Fig. 9 typical of binary sol =0(0<x<x2) (16) vents is not due to differential volatility alone. Since where the surface tension is assumed to follow a sim each component has a different surface tension ƒÁ, sur ple mixing law, ƒÁ=ƒÓ1ƒÁ1+ƒÓ2ƒÁ2 where ƒÓi is the volume face tension gradients are established:61) fraction of component i.*6 Since at the liquid-vapor boundary, the viscous shear force must balance the force imposed by surface tension gradients, ƒÅdu/ dz=dƒÁ/dx(z=h), liquid flows into the water-rich foot with velocity u: u=1/ƒÅ[dƒÁ/dx]z-uo, (17) the so-called "Marangoni effect". The foot slowly grows until this flux is balanced by that of evapora tion from the expanding free surface. The surface tension gradient driven flow of liquid through the thin neck created by the preferential evaporation of alcohol can create quite high shear rates during dip-coating. A striking example is that of toluene and methanol (see, Fig. 10).61) The sur face tension gradient driven flows are strong enough to greatly distort the double parabolic profile, creat ing a "pile-up" of toluene near the drying line. A crude estimate of the surface tension gradient, ƒ¢ƒÁ/ ƒ¢x_??_((10dyne/cm)/10-1cm), leads to a shear rate, du/dz_??_104s-1, in the thin region, from Eq. (17), as suming ƒÅ=0.01P. As we discuss in Section 4.3, these shear fields may be sufficiently strong to align or order the entrained inorganic species. 3.4 Spin-coating Fig. 8. Fluid streamlines and EtOH concentration contours of 30wt% EtOH-70wt% H2O solution during dip-coating predicted Spin-coating differs from dip-coating in that the by a finite element model. The overlying saturation of EtOH was assumed to vary linearly from 10% at the reservoir surface to 0% *6 For ethanol-water mixtures the surface tension does not at the drying line. The change in ethanol concentration along the obey a simple linear mixing law. The surface tension can be free (liquid-vapor) surface results in a steep gradient in surface approximated by 1/(ƒÁ2.7)_??_ƒÓH/(ƒÁH2.7)+ƒÓE/(ƒÁE2.7) where the tension. Substrate is represented in the bottom two plots by the subscripts H and E refer to water and ethanol, respectively. vertical line on the right. From Schunk et al.62) (Terry Garino, private communication). 868 Sol-Gel Thin Film Formation outward, and viscous force (friction), which acts radially inward. The thickness of an initially uniform film during spin-off is described by: h(t)=ho/(1+4ƒÏƒÖ2ho2t/3ƒÅ)1/2 (18) where ho is the initial thickness, t is time, ƒÏ is the den sity, and ƒÖ is the angular velocity. Even films that are not initially uniform tend monotonically toward uniformity, sooner or later following Eq. (18). Equation (18) pertains to Newtonian liquids that do not exhibit a shear rate dependence of the viscosi ty during the spin-off stage. If the liquid is shear-thin ning (often the case for aggregating sols), the lower Fig. 10. Thickness profile of 50:50 methanol: toluene film shear rate experienced near the center of the sub (volume ratio). A steady state "bubble" of toluene forms due to strate causes the viscosity to be higher there and the strong surface-tension-gradient-driven flows. The thinning of the film to be thicker. This problem might be avoided by film creates a rather high shear rate. From Hurd.61) metering the liquid from a radially moving arm dur ing the deposition/spin-up stage. 3.5 Effects of condensed phases The previous sub-sections have largely ignored STAGES OF THE BATCH SPINNING PROCESS the effects of the entrained inorganic species, viz., polymers or particles. These species are initially con centrated by evaporation of solvent as they are trans ported from the coating bath toward the drying line within the thinning fluid film during dipping or dur DEPOSITION SPIN-UP ing the four stages of spin-coating. They are further concentrated (compacted) at the final stage of the deposition process by the capillary pressure Cp creat ed by liquid-vapor menisci as they recede into the drying film (see, Fig. 5a): Cp=2ƒÁcos(ƒÆ)/rp (19) EVAPORATION where ƒÆ is the contact angle of the receding meniscus SPIN-OFF within the emptying pore and rp is the pore size. During dip-coating, the shape of the entrained Fig. 11. The four stages of the batch spin-coating process. From Bornside et al.63) fluid profile is altered very little by the presence of the condensed phase (see, Fig. 12a).60) Thus by con sideration of the conditions that establish the steady depositing film thins by centrifugal draining and state fluid profiles, we are able to quantify the chang evaporation (see, Fig. 11). Bornside et al.,63) divide ing environment of the inorganic species within the spin coating into four stages: deposition, spin-up, thinning film. spin-off and evaporation, although for sol-gel coat Above the stagnation point, all fluid elements are ing, evaporation normally overlaps the other stages. moving upward (see, Fig. 5b), so all the entrained in An excess of liquid is dispensed on the surface dur organic species that survive past the stagnation point ing the deposition stage. In the spin-up stage, the liq are incorporated in the final deposited film. Steady uid flows radially outward, driven by centrifugal state conditions in this region require conservation force. In the spin-off stage excess liquid flows to the of non-volatile mass, thus the solids mass in any perimeter and leaves as droplets. As the film thins, horizontal slice of the thinning film must be the rate of removal of excess liquid by spin-off slows constant:60) down, because the thinner the film, the greater h(x)ƒÓs(x)=constant (20) resistance to flow, and because the concentration of where ƒÓs is the volume fraction solids (see, Fig. 13a). the non-volatile components increases, raising the From Eq. (20), we see that ƒÓ varies inversely with h. viscosity. In the final stage, evaporation takes over Since for a planar substrate there is a parabolic thick as the primary mechanism of thinning. ness profile, ƒÓ should vary as 1/h_??_x-1/2 in the thin According to Scriven,38) an advantage of spin-coat ning film (see, Fig. 12b). When coating a fiber, we ing is that a film of liquid tends to become uniform in expect ƒÓ(x)(cid:129)`(lnx)-1. thickness during spin-off and, once uniform, tends to The rapid concentration of the entrained inorganic remain so provided that the viscosity is not shear-de species by evaporation is more evident from consider pendent and does not vary over the substrate. This ation of the mean particle (polymer) separation dis tendency is due to the balance between the two main tance <r>, which varies as the inverse cube root of ƒÓ, forces: centrifugal force, which drives flow radially <r>(cid:129)`x1/6. As shown in Fig. 13b,61) this is a very C. J. BRINKER et al. Nippon Seramikkusu Kyokai Gakujutsu Ronbunshi The Centennial Memorial Issue 99 [10] 1991 (Journal of The Ceramic Society of Japan) 869 PROFILE-CONCENTRATION RELATIONSHIP Fig. 12a. Thickness profile of a titanate sol during dip-coating The conservation of nonvolatile mass provides a relation as determined by imaging ellipsometry. Position x is defined in between the thickness profile h(x) and the concentration ƒÓ(x). Fig. 5a. From Brinker et al.37) Fig. 13a. Fluid profile-concentration relationship for sol-gel dip coating according to Eq. (20). From Hurd.61) EVAPORATION-INDUCED CROWDING Fig. 12b. Volume fraction solids (ƒÓ) of the titanate sol versus position x as determined from the refractive index according Half the distance between particle neighbors is to the Maxwell Garnett relationship (open circles). Solid line is de covered in the last 2% of the entrained film (100msec). rived from data in Fig. 12a. according to Eq. (20). From Brinker et al.37) Fig. 13b. Fluid profile-mean separation relationship during sol gel dip-coating. From Hurd.61) precipitous function: half the distance between parti cle (polymer) neighbors is traveled in the last 2% of side et al.,63) in their studies of spin-coating, as the deposition process ((cid:129)`0.1s). This evaporation sumed the following relationship: driven transport may be viewed like that induced by ƒÅ=ƒÅo(1-ƒÔA)4+ƒÅs (21) centrifugation or electrophoresis: the centrifugal ac where ƒÅ is the viscosity of the sol, ƒÅs is the viscosity celeration needed to cause an equivalent rate of of the solvent, ƒÅo is the viscosity of the polymer, and crowding is as much as 106g!*7 Since rheological ƒÔ A is the mass fraction of solvent. From this relation properties of suspensions are often concentration ship, we see that the viscosity will change abruptly in dependent,64) the increasing concentration of the in the vicinity of the drying line during dip-coating or organic species is expected to alter the rheology during the evaporation stage of spin-coating. from Newtonian (dilute conditions) to shear-thin The short time of the deposition process ((cid:129)`10s) ning (aggregated systems) or thixotropic (ordered and, more importantly, the extremely short time in systems). The viscosity also exhibits concentration which the inorganic species are in close proximity dependence, at least for high concentration. Born ((cid:129)á1s) establish important time scales for the dip *7 This assumes that there exists no steric barriers to concen and spin-coating processes and distinguish sol-gel tration; often aggregation/network formation will interupt thin film formation from bulk (monolithic) gel forma this dramatic compaction process. tion. For bulk systems, the gelling sol is normally 870 Sol-Gel Thin Film Formation maintained at a constant concentration; the gelation, ACOUSTIC WAVE SENSOR aging, and drying stages occur sequentially. Often the time required to prepare a monolithic bulk gel is on the order of days to weeks. During dip-coating (and spin-coating) the drying stage completely over laps the gelation and aging stages. The depositing sol (which initially is often less concentrated than sols intended for bulk gels) remains rather dilute un til the final stage of the deposition process at which point it is rapidly concentrated. There are only frac tions of seconds available for condensation reactions (Eqs. (3) and (4)) to occur. We anticipate several consequences of the short time scale of the film deposition process. (1) There is little time available for reacting species to "find" Fig. 14. Schematic illustration of surface acoustic wave (SAW) apparatus used to acquire gas adsorption-desorption data from low energy configurations. Thus (for reactive sys sol-gel thin films. tems) the dominant aggregative process responsible for network formation may change from reaction limited (near the reservoir surface) to transport quired using a surface acoustic wave (SAW) tech limited near the drying line (see, Fig. 13b). The nique developed by Frye and co-workers.14) The film reduced fractal dimensionality of transport-limited is deposited on a piezoelectric quartz substrate (see, aggregates should result in more porous networks Fig. 14) which acts as a feedback element of an oscil (see, Eq. (8)). (2) For sols composed of repulsive lator circuit. The substrate contains two sets of inter particles, there is little time available for the parti digital transducers; the application of an alternating cles to order as they are concentrated in the thinning field produces alternating strain, launching a surface film. (3) There is little time available for condensa acoustic wave. Changes in the mass of the film due to tion reactions to occur. Thus gelation may actually N2 adsorption or desorption cause a change in the occur by a physical process, through the concentra SAW velocity which is sensed as a change in the tion dependence of the viscosity (e.g., Eq. (21)), resonant frequency of the oscillator circuit. The ex rather than a chemical process. (In some systems treme sensitivity of the SAW device ((cid:129)`80pg/(cm2 this is evident by the fact that the deposited film is film)) permits accurate adsorption data to be ob quickly re-solubilized when immersed in solvent.) tained easily on a 1cm2 area of film less than 100nm (4) Since the gels are most likely weakly condensed thick. compared to bulk gels, they are more easily compact 4.1 Influence of precursor structure and reactiv ed; first by evaporation and then by the capillary ity pressure exerted at the final stage of the deposition The possible structures of the inorganic precur (drying) process (see, Fig. 5a). Furthermore the sors range from weakly branched polymers to highly maximum capillary pressure is likely to be higher condensed particles (see, e.g., Fig. 1).65) For fractal due to the collapse of the network (and hence reduc objects the packing efficiency (i.e., the volume frac tion in pore size) that precedes the final stage of tion solids, ƒÓ) depends on the fractal dimension, drying. Of course, greater capillary pressure also pro size, and condensation rate. The fractal dimension motes greater compaction of films compared to bulk and size dictate steric constraints. Mandelbrot56) has gels. shown that if two structures of radius R are placed in dependently in the same region of space, the mean number of intersections M1 ,2 is expressed as: 4. Control of film microstructure M1 2∝RD1+D2-d (22) From the preceding discussion we expect that the where D1 and D2 are the respective fractal dimen final film structure should depend on the competition sions and d is the dimension of space (d=3). Thus if between such phenomena as evaporation and capilla each object has a fractal dimension less than 1.5, the ry pressure, which tend to compact the structure, probability of intersection decreases indefinitely as R and condensation reactions and aggregation process increases. These structures are mutually transpar es which tend to stiffen the structure, resisting its ent: during film formation, they should freely inter tendency to collapse. In this section we document penetrate as they are forced into close proximity by how these factors, along with the structure of the en the increasing concentration. Alternatively, if the trained inorganic species, influence film microstruc fractal dimension of each object exceeds 1.5, the ture. probability of intersection increases algebraically We characterize the film microstructure (surface with R. The structures, though porous, are mutually area, % porosity, and pore size distribution) in situ opaque; when concentrated they are unable to inter by analyzing N2 adsorption-desorption isotherms ac penetrate, much like an assemblage of "tumble C. J. BRINKER et al. Nippon Seramikkusu Kyokai Gakujutsu Ronbunshi The Centennial Memorial Issue 99 [10] 1991 (Journal of The Ceramic Society of Japan) 871 weeds". molecules larger than 0.4nm, the kinetic diameter of These concepts of mutual opacity or transparency the N2 molecule. Because the condensation rate is assume that every intersection results in irreversible low under these conditions, we expect that, although sticking"; conditions chemically equivalent to an D>1.5, the precursors can interpenetrate as they infinite condensation rate. In fact, the condensation are concentrated on the substrate surface by evapora rates of silicates are quite low (e.g., 10-4 tion, leading to densely packed, but compliant, 1/mole-sec)66) and very dependent on [H+] (see, configurations with molecular-scale pores. During Fig. 2).47),48). Thus the probability of sticking at any the final stage of drying the capillary pressure is enor point of intersection is(cid:129)á1, causing film structures to mous due to the small pores (Eq. (19)), causing fur be generally more compact than expected from Eq. ther compartion of the compliant network. (22). For example, deposition of rather weakly bran The differences in time scales of bulk and thin film ched silicate sols with D=1.9 (see, sample (b) in sol-gel processing are evident from the comparison Fig. 3 and sample (A) in Fig. 4) at(cid:129)`pH 2, where of adsorption-desorption isotherms of the film and the condensation rate is minimized, produces dense bulk xerogel (dried gel) prepared from identical sili films characterized by a Type ‡U N2 adsorption cate precursors (Fig. 15). Whereas the film is charac desorption isotherm (see, Fig. 15b)67) and surface terized by a Type ‡U isotherm and low surface area, area equivalent to the geometric area of the film.14) indicating no accessible porosity, the bulk xerogel This indicates that there is no porosity accessible to isotherm (Fig. 15a) is of Type ‡T, indicative of a microporous material, and the N2 BET surface area BULK AND THIN FILM SAMPLES PREPARED exceeds 800m2/g. The long duration of the gelation, FROM IDENTICAL PRECURSORS HAVE DIFFERENT MICROSTRUCTURES aging, and drying stages inherent in the bulk gel process allows the structure to stiffen at an early (a) point of the drying stage through continued conden sation reaction: the result is the gel is compacted less by evaporation during the initial stage of drying, so the pores are larger and the capillary pressure smaller (Eq. (19)) during the final stage of drying. Both increased stiffness and reduced capillary pres sure cause the bulk sample to be much more porous than the film. The interplay between precursor structure and condensation rate is well illustrated by a series of films prepared from more highly branched fractal clusters characterized by D=2.4 and a higher Q4/ (Q2+Q3) ratio (see, sample (c) in Fig. 3).17) Figure 16 shows that dip-coating at pH_??_3.5, where the con MICROPOROUS BULK SILICA XEROGEL (b) NON-POROUS SILICA FILM Fig. 15a. Type ‡T N2 adsorption-desorption isotherm for a bulk Fig. 16. Reciprocal relationship between the hydrodynamic silica xerogel prepared from TEOS using a two-step acid-cata radius of multicomponent silicate clusters (D=2.4) deposited lyzed process (H2O/Si=5). From Brinker et al.37) near pH 3 and the corresponding refractive indices of the deposit Fig. 15b. Type ‡U N2 adsorption-desorption isotherm (acquired ed films. Time refers to the aging times at 50(cid:129)Ž and pH 3 em by SAW technique) for a silicate film prepared from the identical ployed to grow the inorganic clusters prior to film deposition. sol as in Fig. 15a. From Frye et al.14) From Brinker et al.17)

Description:
ture, capillary pressure, and the accompanying shear on such properties as refractive index, surface area, and pore size of the deposited films, which we meas ure using in situ techniques such as conventional and imaging ellipsometry39) and gas sorption on surface acoustic wave substrates.14) The
See more

The list of books you might like

Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.