Langmuir 2008, 24, 13653-13662
13653
Polysiloxane Nanofibers via Surface Initiated Polymerization of Vapor Phase Reagents: A Mechanism of Formation and Variable Wettability of Fiber-Bearing Substrates De-ann E. Rollings and Jonathan G. C. Veinot* Department of Chemistry, UniVersity of Alberta, Edmonton, Alberta, T6G 2G2 Canada ReceiVed May 23, 2008. ReVised Manuscript ReceiVed September 5, 2008 A detailed study of polysiloxane nanofiber formation by surface initiated polymerization of vapor phase organotrichlorosilane reagents is presented. Substrate composition, substrate activation, reagent quantity, reaction pressure, and reaction time are parameters shown to influence nanofiber synthesis. Stepwise variation of the parameters isolates the role of each on polysiloxane nanofiber growth, and a mechanism for fiber formation is proposed based on these findings. Tunable aqueous wettability of the fibers is also demonstrated in this report, with contact angles varying from 85° to 130° ( 2° depending upon fiber surface density and length. Aqueous contact angles are further increased to >150° by either solution functionalization of calcined fibers or copolymerization with an organofluorosilane reagent.
Introduction Polysiloxanes are inorganic polymers having an alternating silicon-oxygen backbone chain with saturated and/or unsaturated organic side chains. Following their discovery approximately 50 years ago, polysiloxanes, more commonly referred to as silicones, have been studied extensively. This vast interest is due, in part, to their many interesting material characteristics such as high chemical, thermal, and physiological stability, water resistance, low adhesion toward both organic and inorganic materials, and minimal toxicity. The physical characteristics of polysiloxanes are also tunable by straightforward modification of pendant organic side groups. Polysiloxane fibers possess enhanced value due to their large aspect ratio and high surface areas. These qualities, in combination with the more general characteristics noted above, make polysiloxane fibers attractive materials for a variety of applications that may include dry adhesives and selfcleaning surfaces. Common procedures presently employed for preparing nanofibers of various materials include, vapor-liquid-solid growth (VLS),1-3 template-directed synthesis,1,4,5 kinetically controlled solution synthesis,1,6,7 electrospinning,8-10 substrate etching, polymer drawing,11 and vapor deposition. Only a few methods exist for the fabrication of polysiloxane fibers. Klabunde et al. prepared siloxane nanofibers in solution from long chain * To whom correspondence should be addressed. E-mail: jveinot@ ualberta.ca. (1) Xia, Y.; Yang, P.; Sun, Y.; Wu, Y.; Mayers, B.; Gates, B.; Yin, Y.; Kim, F.; Yan, H. AdV. Mater. 2003, 15, 353–389. (2) Verheigen, M. A.; Immink, G.; de Smet, T.; Borgstro¨m, M. T.; Bakkers, E. P. A. M. J. Am. Chem. Soc. 2006, 128, 1353–1359. (3) Wang, X.; Ding, Y.; Summers, C. J.; Wang, Z. L. J. Phys. Chem. B 2004, 108, 8773–8777. (4) Jang, J.; Bae, J.; Park, E. AdV. Funct. Mater. 2006, 16, 1400–1406. (5) Zhang, M.; Bando, Y.; Wada, K.; Kurashima, K. J. Mater. Sci. Lett. 1999, 18, 1911–1913. (6) Jang, J.; Chang, M.; Yoon, H. AdV. Mater. 2005, 17, 1616–1620. (7) Yu, T.; Joo, J.; Park, Y. I.; Hyeon, T. J. Am. Chem. Soc. 2006, 128, 1786– 1787. (8) Li, D.; Xia, Y. Nano Lett. 2003, 3, 555–560. (9) Deitzel, J. M.; Kleinmeyer, J. D.; Hirvonen, J. K.; Beck Tan, N. C. Polymer 2001, 42, 8163–8170. (10) Kameoka, J.; Verbridge, S. S.; Liu, H.; Czaplewski, D. A.; Craighead, H. G. Nano Lett. 2004, 4, 2105–2108. (11) Jeong, H. E.; Lee, S. H.; Kim, P.; Suh, K. Y. Nano Lett. 2006, 6, 1508– 1513.
alkylsilanes with the assistance of pentanone-capped Au nanoparticles as catalysts.12 Poly(methylsiloxane) nanofiber networks on Si wafers have been prepared by McCarthy and Gao by polymerization of CH3SiCl3 with trace H2O in toluene followed by extraction with ethanol.13 McCarthy and Gao have also reported polysiloxane nanofiber formation of Me3SiCl/SiCl4 in the vapor phase under a humidity controlled environment.14 Seeger et al. used vapor phase MeSiCl3 at atmospheric pressure in a humidity controlled environment to produce poly(methylsiloxane) nanofibers.15 Unfortunately, none of these reports provide detailed investigations of parameter influences that would yield formation mechanisms. Only recently has a brief explanation for preferential growth in one dimension been proposed,14 and it is specific to nanofibers formed using R3SiCl reagents. Furthermore, to our knowledge, no report exists outlining the preparation of copolymer polysiloxane nanofibers. We have successfully fabricated homopolymer polysiloxane fibers with vinyltrichlorosilane (VTS) as the precursor as well as block copolymer fibers with VTS and 3,3,3-trifluoropropyltrichlorosilane (FTS). Interest in VTS fibers stems from their functionality and potential derivatization via functionalization of either the vinyl or residual hydroxyl groups, which could afford innovative device applications. Surfaces bearing VTS/ FTS copolymer fibers exhibit both the potential for chemical modification as well as low surface energy and a high surface area structure, attributes that may prove invaluable for mimicking natural phenomena. Our previous contribution outlined the observation of fiber formation on flat surfaces and provided detailed fiber characterization to confirm the fibers were indeed polysiloxane networks.16 The present contribution focuses on parameter optimization, a discussion of and proposed mechanism for the fiber forming process, as well as the influence of fiber topography, morphology, and composition on substrate wettability. (12) Prasad, B. L. V.; Stoeva, S. I.; Sorensen, C. M.; Zaikovski, V.; Klabunde, K. J. J. Am. Chem. Soc. 2003, 2003, 10488–10489. (13) Gao, L.; McCarthy, T. J. J. Am. Chem. Soc. 2006, 128, 9052–9053. (14) Gao, L.; McCarthy, T. J. Langmuir 2008, 24, 362–364. (15) Artus, G. R. J.; Jung, S.; Zimmermann, J.; Gautschi, H.-P.; Marquardt, K.; Seeger, S. AdV. Mater. 2006, 2006, 2758–2762. (16) Rollings, D. E.; Tsoi, S.; Sit, J. C.; Veinot, J. G. C. Langmuir 2007, 23, 5275–5278.
10.1021/la801595m CCC: $40.75 2008 American Chemical Society Published on Web 11/04/2008
13654 Langmuir, Vol. 24, No. 23, 2008
Rollings and Veinot
Scheme 1. Schematic of the Experimental Apparatus Used for Fabrication of Polysiloxane Nanofibersa
a
Photographs show substrates with (a) a fibrous mat extending ∼500 nm off the substrate and (b) a structured film ∼300 nm thick.
Materials and Methods Reagents. Concentrated sulfuric acid (98%) and hydrogen peroxide (30 v/v %) were obtained from Fisher Scientific and used without further purification. Toluene purification was achieved using an Innovative Technologies PureSolv solvent purification system with alumina columns. Vinyltrichlorosilane (VTS), butyltrichlorosilane (BTS), 3,3,3-trifluoropropyltrichlorosilane (FTS), and octyltrichlorosilane (OTS) were used as received from Aldrich Chemical Co. Si (100) test wafers (n-doped (P), 1-100 Ω · cm) were obtained from Evergreen Semiconductor Materials. Optical grade fused quartz wafers were purchased from Esco products, and alumina membrane discs (0.2 µm) were acquired from Whatman. Undoped Ge (100) wafers (10-45 Ω · cm) were purchased from MTI Corp. Bangs Laboratories was the supplier of 0.15 µm silica beads suspended in water, and polyimide-coated fused silica capillary tubes with an inner diameter range of 72.0-74.0 µm and outer diameter of 358.0-359.0 µm were purchased from Polymicro Technologies. The PDMS stamp was fabricated in the University of Alberta Nanofab (Andarr Industries). Substrate Preparation and Activation. Substrates were activated by the removal of organic contaminants with one of three methods: (1) oxygen plasma reactive ion etch (RIE), (2) oxidative piranha solution, or (3) sonication in solvents of increasing polarity. For RIE, the plasma chamber was first purged for 30 min with 80% of 100 sccm of O2 and 75% of 300 W radio frequency at 150 mTorr. The substrate was then treated with O2 plasma under similar conditions for 90 s. “Piranha” cleaned wafers were submerged in the 3:1 H2SO4/ H2O2 solution for 30 min and then rinsed with deionized water and spun-dry at 4000 rpm for 30 s. Care should be taken when handling this solution, as it is highly corrosive and reacts violently with organics. Wafers cleaned by sonication in solvents of increasing polarity (toluene, acetone, isopropyl alcohol, deionized water) were
sonicated for 3 min each. Ge wafers and porous alumina discs were activated using RIE. Following activation, substrates were placed in a vacuum oven (125 Torr, >120 °C) for a minimum of 2 h prior to modification and remained under vacuum until reaction. Commercially available silica beads (Bangs Laboratories) were supported on Si wafers to facilitate their reaction with VTS under reduced pressure. Mono- and multilayers of silica beads were prepared by spin-coating 0.1 mL of 0.15 µm beads suspended in water at 4000 rpm for 30 s onto a Si wafer previously cleaned with piranha. The substrate was placed in the vacuum oven (125 Torr, >120 °C) for >2 h to facilitate cross-linking of the beads to the native oxide surface. Finally, the bead-coated substrates were activated by treatment with RIE, and surface adsorbed water was removed by heating in the vacuum oven for 2 h. The polyimide coating was removed from the commercial silica capillary tubes by submerging capillaries in boiling concentrated sulfuric acid for >10 min. The capillaries were subsequently rinsed 3 times with deionized water and then left to soak in deionized water for 10 min before flame-drying. Since the capillaries were stored in a regular laboratory setting, capillaries were additionally piranha cleaned immediately before being used to remove any contamination adsorbed from the atmosphere. After “piranha” treatment and rinsing with copious amounts of water, the capillaries were placed in the vacuum oven to dry before reaction with VTS. OTS SAMs were fabricated by placing vacuum oven-dried Ge wafers, previously RIE cleaned, into a 1.5 mM solution of OTS in dry toluene under inert atmosphere. Wafers remained in the silane solution for 2 h before being rinsed thoroughly with toluene and dried in the vacuum oven at 120 °C for 1 h. Patterned substrates were fabricated by stamping a “piranha” cleaned wafer with a polydimethylsiloxane (PDMS) mold dip-coated in 3,3,3-trifluoropropyltrichlorosilane (neat). The patterned wafer
Polysiloxane Nanofiber Formation
Langmuir, Vol. 24, No. 23, 2008 13655 Scheme 2. Illustration Depicting the Fiber Forming Processa
a (i) Addition of H2O and VTS. Water adsorbs to the surface of the substrate. (ii) Some h-VTS dissolves in the water adsorbed layer and hydrogen bonds with surface hydroxyl groups. (iii) H-Bonded h-VTS couples with surface hydroxyl groups, eliminates H2O, and then other dissolved h-VTS moieties cross-link with surface bound h-VTS. (iv) Polymeric islands emerge as further cross-linking with incoming VTS occurs. (v) Vapor phase VTS preferentially couples to the islands that have emerged from the adsorbed water layer due to their greater surface area and accessibility; otherwise, VTS dissolves in the water layer and cross-links with other VTS moieties, forming a thin film.
was then placed in the vacuum oven at 120 °C for >1 h to remove any surface adsorbed water. Polysiloxane Nanofiber Synthesis. The reaction apparatus was placed in a 150 °C oven for at least 2 h before being used for a reaction. When equipment was removed from the oven, it was assembled while still hot and then cooled under vacuum to minimize any water adsorption. Having cooled to room temperature, the chamber (∼2.3 L) was backfilled to atmospheric pressure with dry Ar, and activated silicon substrates were placed inside the reactor with an adapted manifold top under flowing argon (Scheme 2-1). The equipment was evacuated again and, while under dynamic vacuum, flame-dried to remove any water adsorbed during the assembly process before finally leaving the chamber at a static base pressure of 125 Torr. Two Schlenk flasks (∼0.07 L) connected externally to the reaction chamber were backfilled with Ar to atmospheric pressure before being charged separately with 2.7 mmol deionized water (50 µL) and 1.5 mmol organotrichlorosilane (200 µL). Next, the deionized water flask was opened to the reaction chamber, and all the water was introduced to the chamber by evaporation using direct flame. After a predetermined amount of time (t1), the Si substrate was covered with a tight sealing, custom designed glass shield and the organotrichlorosilane vapor was introduced to the reaction vessel by opening the stopcock to the silane reagent Schlenk flask. After a second induction time, (t2), which allowed the silane reagent adequate time to partially evaporate, the glass shield was raised, exposing the activated substrate to reagent vapor (t3). Standard reaction times were t1 ) 5 min, t2 ) 10 min, and t3 ) 60 min. Modified substrates were subsequently removed and stored in ambient conditions. VTS/FTS Copolymer Synthesis. Gradient copolymer nanofibers of VTS/FTS were prepared using similar conditions to those described above with the following modifications. After the Si substrate had been exposed to VTS for 10 min, the VTS Schlenk flask stopcock was closed and the flask was backfilled with Ar. Subsequently, 0.92
mmol (150 µL) of FTS was injected into the flask and the protective glass shield in the reaction chamber was lowered to cover the Si substrate. The stopcock was reopened, and 5 min elapsed before the shield was raised to expose the Si wafer to the organotrichlorosilane vapor for 24 h. Scanning Electron Microscopy (SEM). Secondary electron images were obtained by first sputter-coating an ∼1 nm film of Cr onto the substrates using an Edwards Xenosput XE200 instrument. Microscope images were obtained using a JEOL 6301F field emission secondary electron microscope with an accelerating voltage of 5 kV. Contact Angle Measurements. Advancing and receding contact angle measurements were obtained using a First Ten Angstroms FTA100 Series contact angle/surface energy analysis system using deionized water as the probe liquid. The average of three measurements taken at different locations on the substrate ensured representative values were recorded. X-ray Photoelectron Spectroscopy (XPS). A Kratos Axis Ultra instrument operating in energy spectrum mode at 210 W was used for XPS measurements. The base pressure and operating chamber pressure were maintained at e1 × 10-7 Pa. A monochromatic Al KR source was used to irradiate the samples, and the spectra were obtained with an electron takeoff angle of 90°. Wide survey spectra were collected using an elliptical spot with 2 mm and 1 mm major and minor axis lengths, respectively, and 160 eV pass energy with a step of 0.33 eV. Sample compositions were determined from the peaks of the survey spectra with subtracted linear background using the internal instrument values of relative sensitivity factor. Sample charging was minimized using an electron gun. Time of Flight Static Secondary Ion Mass Spectroscopy. Time of flight static secondary ion mass spectrometry was conducted on an Ion ToF IV-100 instrument. Auger Electron Spectroscopy (AES). The Auger measurements and SEM images associated with AES were carried out using a
13656 Langmuir, Vol. 24, No. 23, 2008
Rollings and Veinot
Figure 1. SEM images of fibers grown on substrates of varied composition and morphology: (a) SiOX/Si(100) wafer, (b) GeO2/Ge(100) wafer, (c) porous alumina, (d) high magnification of porous alumina showing fibers inside pores, (e) 0.15 µm silica beads, (f) quartz, (g) capillary tube, and (h) view of fibers located inside the pore of a capillary. All substrates were exposed to VTS/H2O using the standardized reaction conditions shown in Scheme 1. Scale bars are 500 nm with the exception of images (g) and (h) where they represent 50 and 10 µm, respectively.
JAMP-9500F Auger microprobe (JEOL). The instrument is equipped with a Shottky field emitter, which can obtain 3 and 7.5 nm spatial resolutions for SEM and Auger mapping, respectively. The accelerating voltage and emission current for both SEM and Auger imaging were 25 kV and10 nA, respectively. The working distance was 23 mm. The sample was rotated 30° away from the primary electron beam to face the electron energy analyzer. The Auger peaks of Ge LMM (1143 eV), C KLL (263 eV), and O KLL (503 eV) were selected for Auger imaging. The intensity of each pixel in the Auger image was calculated by (P - B)/B, where P and B are the peak and background intensity, respectively. Such intensity definition helps to reduce the edge effect of islands and dots. An auto probe tracking technique was used to compensate for possible drifting of the image during the analysis as a result of power instabilities.
Results and Discussion Polysiloxane fibers were prepared on a variety of substrates. A schematic representation of the general synthetic procedure is shown Scheme 1. Briefly, an O2 plasma reactive ion etched (O2 RIE) Si wafer with native oxide was placed in the reaction apparatus, and the chamber was evacuated to a static base pressure of 125 Torr. Isolated Schlenk flasks connected to the reaction chamber were maintained at atmospheric pressure with Ar and loaded with defined quantities of water and VTS. Water was subsequently introduced into the room temperature reaction chamber upon heating the Schlenk flask and then allowed to equilibrate for a predetermined time, (t1). This procedure facilitated the formation of an adsorbed water layer on the substrate surface. Next, a glass shield was lowered over the substrate, the VTS Schlenk flask stopcock opened, and VTS introduced to the chamber by vapor diffusion. After another set period of time, t2, the shield was raised, exposing the substrate to the water/VTS vapor mixture for time t3. This illustration highlights the important time intervals, t1, t2, and t3, used in a “typical” procedure. The quantities of water and VTS that constitute “typical” reaction conditions are 50 and 200 µL, respectively. Once the reaction is complete, visual inspection of the substrates after removal from the apparatus provides qualitative information about the fibers formed on the substrate. Substrates bearing fibrous mats that extend less than or equal to 1 µm perpendicular from the
substrate are indistinguishable from bare substrates. Thicker fibrous mats (>1 µm) appear uniform across the entire substrate, whereas polymeric films (>200 nm) have a nonuniform appearance. Both thicker fibrous mats and polymeric films readily diffract ambient light and have a cloudy, white appearance. Systematic variation of reagent volumes, reaction time intervals, and chamber base pressure provided valuable insight into the fiber forming process. Additional information was also obtained during our investigation of various substrates, activation methods, and the order of reagent addition. Length, spatial density, and surface topography of the fiber-bearing substrates processed are described as evaluated by SEM. Combined analysis using Fourier transformation infrared spectroscopy (FTIR), X-ray photoelectron spectroscopy (XPS), time of flight mass spectroscopy (ToFSIMS), electron dispersive spectroscopy (EDS), and scanning electron microscopy (SEM) confirm the fibers are networked poly(vinylsiloxane), and the details have been discussed elsewhere.16 Length, spatial density, and surface topography of the fiber-bearing substrates processed are described as evaluated by SEM. These detailed analyses have provided us with the necessary information to propose the most detailed formation mechanism reported to date for fiber structures of this type. The Role of Substrates. After optimizing the fiber growth conditions for hydroxyl terminated native oxide Si wafers, substrates of various shapes, composition, and surface chemistry were investigated using the optimized conditions outlined above. Native surface oxide passivated Ge(100) and fused quartz were chosen as representative “flat” hydroxylated substrates; commercial silica beads (d ) 0.15 µm), fused silica capillary tubes (outer diameter, 358.0-359.0 µm), and porous alumina discs (pore diameter, 0.2 µm) were the structured substrates tested. Fibers were successfully grown on all hydroxyl terminated substrates tested (Figure 1). Fibers grown on quartz, silica, and the capillary tubes typically have larger diameters (∼50-70 nm) than those grown on Si wafers (∼35 nm). The lengths of fibers fabricated on the capillary tube and quartz routinely range from 5 to 10 µm in contrast to 1 µm on Si wafers. An explanation for these dimensional differences is not obvious and is the subject of further study. Nevertheless, these results support our hypothesis
Polysiloxane Nanofiber Formation
Langmuir, Vol. 24, No. 23, 2008 13657
Figure 2. SEM (a) and AES mapping ((b) carbon, (c) oxygen, and (d) germanium) of an OTS functionalized oxide passivated Ge(100) wafer. Black regions in the SEM correspond to areas of high carbon and oxygen concentration in the AES analysis. This is consistent with large agglomerates of OTS. AES mapping color scale is located to the right of the Ge map; colors at the top of the color bar correspond to higher concentrations of the element in question. Inset scale bars represent 500 nm.
that hydroxyl group terminated surfaces are needed for production of polysiloxane nanofibers in this manner. Another intersting feature of this technique is fiber growth within pores. A survey SEM scan of the entire pore length of the porous alumina revealed fiber formation normal to the surface, throughout the pore. To ascertain if surfaces terminated with hydroxyl groups were required for fibers to form, surfaces with minimal hydroxyl group concentration were also evaluated. These surfaces were obtained by passivating Ge(100) and Si(100) wafers with incomplete monolayers of octyltrichlorosilane (OTS). It is well-established that long chain organotrichlorosilanes assemble on hydroxylated surfaces due to the positive van der Waals interactions between the alkyl chains and coupling between the silicon headgroup and the hydroxyl functionality.17-21 It is known that siloxane SAMs initially form as isolated islands before filling in to generate a complete monolayer.22 Therefore, a substrate removed from the deposition solution before full functionalization is achieved would be expected to have islands of cross-linked organosilanes present on the surface. Contact angle measurements of methyl terminated alkyltrichlorosilane SAMs with varied alkyl chain lengths have been reported by Wasserman et al.18 Well-packed SAMs with alkyl chains greater than three carbons exhibit advancing contact angles of ∼110°; the OTS SAM functionalized Ge wafers discussed here exhibit advancing and receding contact angles of 93° and 50°, respectively. The lower contact angles of these SAMs suggest a nonuniform monolayer was formed. Figure 2 shows a representative SEM and corresponding C, O, and Ge Auger electron spectroscopy (AES) mapping images of an OTS SAM functionalized Ge wafer. AES data for the SAM covered substrate confirms a nonuniform coverage of the OTS on the surface as indicated by large regions having high concentrations of Ge concurrent with low concentrations of C. Consistent with this observation, the oxygen concentration is higher in regions where carbon is present because hydrolysis of octyltrichlorosilane replaces the chlorine moieties with oxygen. OTS SAM modified substrates processed using standard polysiloxane fiber growth parameters (Vide supra) produced fiber structures with variable spatial distribution depending on the quality of the original SAM functionalized substrate before exposure to VTS. Substrates coated with more uniform (i.e., higher quality) SAMs (not shown) have sporadic fiber growth (17) Sagiv, J. J. Am. Chem. Soc. 1980, 102, 92–98. (18) Wasserman, S. R.; Tao, Y. T.; Whitesides, G. M. Langmuir 1989, 5, 1074–1087. (19) Carson, G.; Granick, S. J. J. Appl. Polym. Sci. 1989, 37, 2767–2772. (20) Kessel, C. R.; Granick, S. Langmuir 1991, 7, 532–538. (21) Silberzan, P.; Leger, L.; Ausserre, D.; Benattar, J. J. Langmuir 1991, 7, 1647–1651. (22) Onclin, S.; Ravoo, B. J.; Reinhoudt, D. N. Angew. Chem., Int. Ed. 2005, 44, 6282–6304.
Figure 3. SEM images of incompletely functionalized OTS SAM Ge wafers exposed to VTS. It is reasonable that areas where no fiber growth occurs are regions effectively functionalized by OTS. (a) High magnification image showing a single “hole” where no fibers formed and (b) low magnification image displaying the “holes” absent of fibers. Scale bars are 500 nm.
evenly distributed over their surface with typical diameter (∼30 nm) and length (>500 nm). In contrast, wafers incompletely functionalized with OTS have both regions absent of fibers as well as areas of densely packed fibers. Figure 3 presents the SEM images obtained after exposing the Ge wafer, used in Figure 2, to VTS. The substrate possesses “holes” where no fibers are present. We suggest these “holes” are observed in the SEM because surface hydroxyl groups are blocked by the surface bonded OTS islands; no fibers are formed in these regions during exposure to VTS. Dense fibers only grow in regions where native oxide surface hydroxyl groups are accessible for reaction with VTS. As a final confirmation, selected areas of silicon substrates were patterned with FTS to block surface hydroxyl groups. This was accomplished by patterning the surface with FTS using a polydimethylsiloxane (PDMS) stamp. Fiber growth was severely reduced in the regions stamped by PDMS (Figure 4). Influence of Substrate Activation. Oxygen plasma reactive ion etching (O2 RIE) is a procedure commonly used for removing polymer lithography masks23 and organic contamination from substrate surfaces.24 We have previously shown that this technique is also sufficient for activating substrate surfaces in our polysiloxane fiber growth procedure.25 To extend the versatility of our nanofiber synthetic procedure, we tested other methods commonly employed for substrate cleaning. “Piranha” solution is a powerful oxidant comprising a 3:1 mixture of concentrated sulfuric acid and 30 v/v % hydrogen peroxide used extensively in the electronics industry to remove organic contaminates from (23) Williams, K. R.; Gupta, K. J. W. M J. Microelectromech. Syst. 2003, 12, 761–778. (24) Kern, W. J. J. Electrochem. Soc. 1990, 137, 1887–1892. (25) Rollings, D. E.; Tsoi, S.; Sit, J. C.; Veinot, J. G. C. Langmuir 2007, 23, 5275–5278.
13658 Langmuir, Vol. 24, No. 23, 2008
Rollings and Veinot
Figure 4. (a) Schematic depicting the PDMS stamping process used to pattern the Si wafer. Left side of Si substrate stamped with 3,3,3-fluoropropyltrichlorosilane (FTS) before exposure to VTS. (b) SEM image showing a patterned/passivated region of the Si wafer. Nanofibers preferentially grow on areas of the substrate not stamped with FTS. Scale bar is 50 µm.
Figure 5. SEM images of fibers obtained with (a) piranha cleaning and (b) sonication of substrate in increasing polarity of solvents (toluene, acetone, isopropyl alcohol, and water). Scale bars are 500 nm.
various substrates. 26 We observed that wafers cleaned using this solution yield fibers of spatial density, length, and diameter resembling those obtained using substrates cleaned by O2 RIE (Figure 5a). Another cleaning procedure, substrate sonication in solvents of increasing polarity, was also employed. Substrates were freed from organic contaminants by sonicating in toluene, acetone, isopropyl alcohol, and deionized water prior to exposing them to water and VTS vapors. SEM images revealed that these substrates also bore fibers; however, their spatial densities were significantly lower than those observed for substrates activated with RIE or “piranha” solution (see Figure 5b). These observations may be rationalized by considering the efficiency of each cleaning procedure at removing organic contaminants from substrates. “Piranha” and O2 RIE treatments are known to be efficient cleaning methods for removal of organic material from substrate surfaces.24,27,28 Sufficient removal of adventitious materials from the substrate surface increases the accessibility and subsequent reactivity of surface hydroxyl groups. These accessible hydroxyl groups readily form covalent attachments with organotrichlorosilanes due to the formation of the more thermodynamically stable Si-O bond.17-21 Therefore, increased accessibility translates to an overall higher fiber density on the substrate. Influence of Reagent Quantities. After extensive study, it was found that even small variation in the quantity of VTS reagent dramatically affects fiber growth. SEM micrographs (Figure 6) (26) Kern, W. J. Handbook of Semiconductor Wafer Cleaning Technology Science, Technology, and Applications; William Andrew Publishing/Noyes: Park Ridge, NJ, 1993. (27) Mun, S. Y.; Jang, Y. S.; Ko, Y. S.; Huh, S. B.; Lee, J. K.; Jeong, Y. H. J. Electrochem. Soc. 2006, 2006, G866-G869. (28) Sirghi, L.; Kylian, O.; Gilliland, D.; Ceccone, G.; Rossi, F. J. Phys. Chem. B 1996, 110, 25975–25981.
revealed that when 250 nm) with smooth, rounded architectures were obtained. These findings indicate the amount of water initially present on the substrate surface substantially influences the structural formations that are produced. RSiCl3/H2O Ratio. In an effort to determine if fiber formation results from a specific water/silane molar ratio or instead is imparted by the absolute concentration of reagent in the chamber, the ratio of reagents was halved and doubled. When decreased by a factor of 2, a thin, rough, uneven film (∼50 nm) is obtained. When the reagent quantities were doubled, a thick film (200-500 nm) with smooth rounded features was produced. To understand these observations, it is important to consider changes occurring at the wafer surface. When water vapor is introduced, a water layer adsorbs onto surfaces terminated with hydroxyl groups. When altering the water content in the chamber, the thickness of the adsorbed water layer changes because the volume to chamber surface area ratio has been modified. A smaller volume of water decreases the thickness of the adsorbed water layer, whereas a larger volume of water has the potential to increase the thickness of the adsorbed water layer. Given no fibers were observed on substrates when the volumes were doubled and halved, we conclude the thickness of the adsorbed water layer plays a crucial role in the production of fibrous structures. Order of Reagent Addition. To further support our proposal that an adsorbed water layer was required for fiber formation, the order of reagent addition to the reaction chamber was reversed. No fibers formed on the substrate when VTS was admitted to the chamber before the water (not shown). This confirms the necessity of the surface adsorbed water layer for fabrication of organosilicon nanofibers by this method. Influence of t1, t2, and t3. It has also been determined that the time intervals between introduction of reagents, t1 and t2, and substrate exposure time, t3, affect the fiber formation. To reiterate, t1 denotes the time that water was allowed to equilibrate within the reaction chamber prior to introduction of VTS. The second
Polysiloxane Nanofiber Formation
Langmuir, Vol. 24, No. 23, 2008 13659
Figure 6. Oblique SEM images illustrating the influence of vinyltrichlorosilane volume on fiber formation: (a) 0.15 mL, (b) 0.25 mL, and (c) 0.5 mL. Scale bars are 500 nm.
Figure 7. SEM micrographs showing the thin (250 nm) with smooth rounded contours was obtained and no fibers were formed. If t1 > 5 min, dense fiber growth was observed. From this, we conclude that water equilibration time is a major factor in fiber production. Since hot water vapor entered the room temperature apparatus, the majority of the water initially physisorbed to the chamber surfaces. While equilibrating, a portion of the admitted water re-entered the vapor phase; our results imply the adsorbed water layer is thicker at t1 ) 0 then at t1 ) 5 min. Hence, fibers do not form if the water layer is too thick (Vide infra). Likewise, VTS vapor must disperse within the reaction chamber if fiber growth is to be realized. If t2 ) 0 min, a thick (>150 nm) uneven film was obtained and no fiber growth was observed. Extending t2 to 5 min yielded a small number of fibers scattered across the surface; spatially dense fibers were not obtained until t2 ) 8 min. If VTS vapor was allowed to stand for extended periods (i.e., t2 >15 min), uneven films with rough areas and nucleation sites were seen. To rationalize these findings, one must appreciate there is a rapid, large change in reagent concentration when VTS vapor initially enters the reaction vessel. This and our apparatus design causes a high concentration of VTS near the substrate surface before it diffuses throughout the chamber. Because fibers do not evolve until t2 > 8 min, we conclude that fiber growth is optimal when the delivery of VTS to the substrate surface is diffusion limited. While somewhat expected, we also observe that fiber length relies on the time the substrate is exposed to VTS. If t3 ) 5 min, 200 nm fibers were produced, longer fibers (∼500 nm) were obtained after 10 min, and finally micrometer long fibers resulted with t3 > 20 min. There was no obvious difference in the spatial density of the fibers with extended reaction time (t3) upon qualitative evaluation of the SEM micrographs. Figure 8 shows SEM images illustrating the dependence of fiber length on t3.
Proposed Mechanism of Fiber Formation. We have previously established by variable angle FTIR and ToF-SIMS that nanofibers formed using this technique have a polysiloxane framework.25 Using this knowledge, and the information obtained from the systematic investigation outlined above, we have constructed a mechanism for nanofiber formation. It is important to note that reaction conditions very similar to those used herein are often employed for vapor phase formation of siloxane bonded SAMs. In this context, the mechanism presented herein incorporates relevant information learned by researchers studying SAM formations in solution and vapor phase.29-35 When water vapor first enters the chamber, it adsorbs onto the substrate surface, resulting in the formation of an adsorbed water layer. Next, incoming VTS molecules are completely hydrolyzed either in the vapor phase or when adsorbs on the substrate surface. Subsequently, hydrolyzed VTS (h-VTS) molecules are physisorbed to the surface water layer by hydrogen bond interactions. Due to the inherent instability of this state,21 the silanol moiety of the h-VTS reacts with a surface hydroxyl group via condensation reactions, anchoring the h-VTS to the surface. However, all three hydroxyl groups on the h-VTS cannot simultaneously couple to the surface because the density of hydroxyl groups on a native oxide silicon surface is not high enough to accommodate three silanol condensations on one molecule.18 This leaves one or possibly two silanol groups of h-VTS available for coupling with other physisorbed h-VTS moieties. At this point, steric constraints imposed on the alkyl chains of each VTS by the cross-linking reaction between two h-VTSs must be considered. It has been reported that the maximum bond length of a Si-O-Si bond is 3.2 Å when assuming a hypothetical bond angle of 180°, while the van der Waals diameter for C is 3.5 Å.36 Therefore, when two h-VTS molecules cross-link, the alkyl chains must arrange to minimize the interaction between the otherwise overlapping van der Waals radii of the alkyl chains.20,37,38 By adopting this configuration, the remaining hydroxyl group is positioned away from the substrate in a suitable orientation to direct further condensation reactions normal to the substrate surface. Subsequent coupling of these newly surface bonded h-VTS molecules with additional h-VTS leads to the formation of polymeric islands on the surface (29) Hoffmann, P. W.; Stelzle, M.; Rabolt, J. F. Langmuir 1997, 13, 1877– 1880. (30) Dong, J.; Wang, A.; Ng, K. Y. S.; Mao, G. Thin Solid Films 2006, 515, 2116–2122. (31) Tripp, C. P.; Hair, M. L. Langmuir 1992, 8, 1120–1126. (32) Mitchon, L. N.; White, J. M. Langmuir 2006, 22, 6549–6554. (33) Vallant, T.; Brunner, H.; Mayer, U.; Hoffmann, H. J. Phys. Chem. B 1998, 102, 7190. (34) Fadeev, A. Y.; McCarthy, T. J. Langmuir 2000, 16, 7268–7274. (35) Krasnoslobodtsev, A. V.; Smirnov, S. N. Langmuir 2002, 18, 3181–3184. (36) Jorgensen, W. L.; Maxwell, D. S.; Tirado-Rives, J. J. Am. Chem. Soc. 1996, 118, 11225–11236. (37) Stevens, M. J. Langmuir 1999, 15, 2773–2778. (38) Rye, R. R.; Nelson, G. C.; Dugger, M. T. Langmuir 1997, 13, 2965–2972.
13660 Langmuir, Vol. 24, No. 23, 2008
Rollings and Veinot
Figure 8. Cross-sectional SEM images demonstrating the increased length (l) of fibers with longer exposure times to VTS: (a) t3 ) 5 min, l ∼ 200 nm; (b) t3 ) 10 min, l ∼500 nm; and (c) t3 > 20 min, l ∼ 1.1 µm. Scale bars are 500 nm.
of the substrate. Essentially, the initial anchoring of h-VTS creates a nucleation site for island formation.32,39 The newly formed polymeric network protrudes from the substrate surface, so that incoming VTS molecules come in contact with the islands more readily than with the water layer. The vertical polymerization produces complex cross-linked fiber structures, and a “forestgrowth” effect occasionally occurs whereby shorter fibers are “shadowed” by larger ones. If the incoming VTS molecules do not react with the polymeric networks growing away from the surface, they will reach the water layer on the surface, physisorb, and eventually condense with other h-VTS moieties to form a film between the islands. Hence, the critical factor for fiber formation must be the nucleation and growth of the polymeric islands vertical to the surface. An illustration of the fiber forming process is provided in Scheme 2. From this proposed mechanism, it can be seen that the concentration of surface hydroxyl groups dictates the spatial density of the fibers formed because they are nucleation sites for vertical polymerization. The concentration of hydroxyl groups on the native oxide surface of a Si wafer is 5 OH/nm2.18 Based upon our observations of fiber-bearing Si/SiO2 wafers, we conclude that this concentration is adequate to support the formation of dense fibers with relatively uniform diameters. Fiber density is reduced when a surface with fewer accessible hydroxyl groups is employed. The reaction performed on a Ge wafer with an incomplete OTS SAM illustrates this point effectively; surface bound OTS reduces the concentration of surface hydroxyl groups, resulting in a decreased spatial density of fibers (Scheme 2). Following our rational stepwise variation of all the known and controllable reaction parameters, we propose that a major factor governing the nucleation of polymeric islands is the concentration of h-VTS in the surface adsorbed water layer. As stated above, when silanol groups anchor to the surface in the appropriate geometry, nucleation points are created from which nonequilibrium islands form. Incoming VTS and/or h-VTS bond onto the top of these islands and polysiloxane networks are formed that grow away from the substrate surface. Since h-VTS must anchor to the surface if polymeric islands are to form, the concentration of h-VTS molecules in the surface adsorbed water dictates whether rapid nucleation and growth occurs. The effect of water and reagent quantities as well as t1 and t2 may be understood in the context of the concentration of h-VTS in the adsorbed water layer; we have established that fiber growth does not occur if the concentration is outside a narrow range. Recall that if t1 < 5 min, no fiber growth was observed. Consistent with our observations, Cohen et al.40 observed an effect between surface adsorbed water and the solution functionalization of vinyltrimethoxysilane (VTMS). In this report, (39) Rye, R. R. Langmuir 1997, 13, 2588–2590. (40) Yoshida, W.; Castro, R. P.; Jou, J. D.; Cohen, Y. Langmuir 2001, 17, 5882–5888.
the authors determined that organosilane surface coverage decreased when the surface adsorbed water layer had a thickness greater than 2 monolayers. For larger amounts of adsorbed water, it was proposed that penetration of the VTMS through the water layer to the substrate hydroxyl groups became increasingly difficult due to the hydrophobicity of the vinyl group. As a result, VTMS molecules condensed in the thick water layer without attachment to the surface hydroxyl groups, forming a film. It is interesting to note that Cohen and co-workers also observed polymeric island structures forming on the substrates with dimensions of approximately 22 nm in diameter and 10-20 nm in height when under certain conditions. The lack of fibers extending from the surface is likely due to a larger energy penalty associated with the increased polymer-liquid interfacial area with respect to the polymer-vapor interface formed in our system. When the VTS quantity is in large excess (i.e., 0.50 mL) or the VTS is not uniformly dispersed in the reaction chamber (t2 < 8 min), the surface adsorbed water layer becomes saturated in h-VTS and rough, uneven films are produced. In this case, a large number of h-VTS molecules anchor to the surface, producing islands that are so closely spaced that cross-linking between islands occurs concomitantly with vertical polymerization. Polymerization in both directions impedes fiber growth. Our findings suggest that the optimal VTS concentration near the surface is achieved when concentration gradients of VTS are eliminated (t2 > 8 min). VTS is loaded into an ambient pressure flask. When the flask is opened to the reaction chamber, a sudden pressure drop occurs; some VTS rapidly evaporates, entering the chamber to create a large VTS concentration gradient. To minimize the effects of this gradient, a shield covering the substrate prevents the VTS from hydrolyzing in the adsorbed water layer before the system has equilibrated. In summary, the key points governing the production of fibers are as follows: (1) the thickness of the adsorbed water layer, (2) the local concentration of VTS at the substrate surface, and (3) the dissolution and surface attachment of VTS to direct polymerization normal to the surface. Surface Wettability. In our earlier report,25 we demonstrated the significance of surface roughness induced by the fibrous structure on wettability by directly contrasting the advancing contact angles for a substrate bearing fibers and a smooth substrate functionalized with the same material. An important point warranting further discussion in this area is the influence of fiber spatial density and length on roughness and subsequent wettability. Recall that as a droplet of liquid makes contact with a surface, three interfaces with corresponding free energies, γSV, γSL, and γLV, are formed, where SV, SL, and LV, respectively, are the solid-vapor, solid-liquid, and liquid-vapor interfaces. To evaluate the wettability of a substrate, two contact angles are measured: advancing and receding. The advancing contact angle
Polysiloxane Nanofiber Formation
Langmuir, Vol. 24, No. 23, 2008 13661
Figure 9. SEM images and corresponding contact angles demonstrating the effect of fiber length and spatial dispersion on wettability of the substrate: (a) moderately dispersed fibers ∼200 nm in height, (b) sparsely distributed fibers ∼100 nm in height, and (c) densely packed fibers ∼1 µm in height. Scale bars are 500 nm.
is acquired by dispensing a drop of water on the surface and then increasing the volume of the drop by the addition of more water. The receding angle is defined as the angle exhibited by the drop after a portion of the drop volume has been removed and the contact line moved inward. Contact angle hysteresis is the difference between the advancing and receding angles. Young’s equation41 is the simplest expression for rationalizing a liquid droplet’s behavior on a surface. θ is defined as the contact angle formed when a droplet of water comes to rest on a substrate.
cos θ )
γSV - γSL γLV
(1)
For a hydrophilic surface, θ < 90,
γSV - γSL >0 γLV
(2)
and the energy of the solid-liquid interface is less than the energy of the solid-vapor interface. Alternatively, when
γSV - γSL 90, and γSV < γSL, the surface is deemed hydrophobic. However, Young’s model is only applicable for flat, chemically homogeneous substrates. To account for the effects of surface roughness, and/or chemical heterogeneity on wettability, a few different models have been developed.Twowell-knownmodels,theWenzelandCassie-Baxter, are adaptations of Young’s equation that relate interfacial contact area to a substrate’s wetting behavior. Another approach receiving increased attention of late expresses the contact angle in terms of the contact line, that is, the location of the three-phase equilibrium, instead of interfacial contact area. Various reports have been published42-46 rationalizing wettability in this manner in light of the inability of both Wenzel and Cassie models to predict experimental contact angles of certain topographies. The use of the contact line to rationalize a droplet’s response to a surface has many corollaries to the Wenzel and Cassie models; both roughness and surface heterogeneity at the contact line influence the predicted behavior of a water droplet on a substrate. As surface roughness increases, it reaches a critical value, changing the equilibrium at the contact line, thereby increasing (41) Wenzel, R. N. Ind. Eng. Chem. 1936, 28, 988–994. (42) Gao, L.; McCarthy, T. J. Langmuir 2007, 23, 3762–3765. ¨ ner, D.; McCarthy, T. J. Langmuir 2000, 16, 7777–7782. (43) O (44) Extrand, C. W. Langmuir 2003, 19, 3793–3796. (45) Mansky, P.; Lui, Y.; Huang, E.; Russell, T. P.; Hawker, C. Science 1997, 275, 1458–1460. (46) Bartell, F. E.; Shepard, J. W. J. Phys. Chem. 1953, 57, 455.
the advancing contact angle, θa.47 For hydrophobic surfaces, γSL is larger than γSV; accordingly, a smaller linear fraction of asperities on the contact line increases the receding angle, θr. Moreover, a decreased linear fraction of asperities destabilizes the contact line. This instability lowers the energy barrier to displacing the drop; therefore, the drop can travel more freely on the surface.43,13,48,49 For a truly hydrophobic surface, both the advancing and receding angles need to be large. For drops suspended on surface roughness, the θa increases with a larger rise angle of the asperities, and the θr increases as the linear fraction of asperities on the contact line decreases. With collapsed drops, increased roughness results in larger θa than that of smooth surfaces of similar chemical composition and the increased linear fraction leads to lower θr. In both cases, the hysteresis is increased with the asperity contact and the severity of the edge angles (asperity rise). Static contact angle measurements taken for substrates having fibers of various densities and lengths are shown in Figure 9. It is readily seen that fibers of decreased spatial density and length have lower θa and θr than those of long, dense fibers. We deduce that short, sparsely distributed fibers seen in Figure 9b are unable to effectively suspend the drop of water. As a result, the drop conforms to the rough features and the contact line is the interface between the solid and the liquid. The continuity of the contact line increases its stability, and the receding angle decreases. Because fibers are sparse on the surface, the roughness compared to more spatially dense fibers is decreased and hence a smaller advancing contact angle is observed. Conversely, long, densely packed fibers (Figure 9c) exhibit high θa and θr behavior characteristic of drops suspended on top of asperities. Because the water drop is resting on a mat of randomly oriented fibers, an erratic topography exists at the contact line that increases θa. With the drop suspended, the asperities have a smaller linear fraction on the contact line than the collapsed drops. The concomitant decreased contact line stability is marked by the higher θr. Our findings are in accordance with the high advancing and receding contact angles previously obtained for dense, randomly distributed fibrous topographies.13,48,49 The hydrophobicity of fiber-bearing substrates was increased further by the introduction of low surface energy fluorinated species to the fibrous structure via copolymerization. Gradient copolymer fibers were constructed by the sequential addition of 3,3,3-trifluoropropyltrichlorosilane (FTS) to the chamber once a VTS reaction was initiated. With this protocol, extremely long, ¨ ner, D.; Youngblood, J.; (47) Chen, W.; Fadeev, A. Y.; Zhsieh, M. C.; O McCarthy, T. J. Langmuir 1999, 15, 3395–3399. (48) Zhu, L.; Ziu, Y.; Xu, J.; Tamirisa, P. A.; Hess, D. W.; Wong, C. P. Langmuir 2005, 21, 11208–11212. (49) Gao, L.; McCarthy, T. J. Langmuir 2006, 22, 2966–2967.
13662 Langmuir, Vol. 24, No. 23, 2008
Rollings and Veinot
Figure 10. (a) Survey XP spectrum of VTS/FTS copolymer fibers displaying the peaks corresponding to Si, C, O, and F. Inset: high-resolution spectrum of the C 1s region. The higher energy peak observed in the C 1s high-resolution spectrum is due to fluorine atoms bonded to the third carbon of the propyl group on the FTS. (b) SEM of the copolymer fibers. Scale bar is 500 nm.
Figure 11. (a) Stream of water bouncing off the surface of a Si wafer bearing copolymer fibers of VTS/FTS. The reflection of the stream and water droplets is observed on the wafer. (b) Picture of water being sprayed on a bare Si wafer.
very dense fibers were produced (SEM, Figure 10). XPS analysis of the substrate confirms the presence of Si, O, C, and F. In addition, the absence of the Cl peak indicates that all Si-Cl bonds were hydrolyzed. High resolution XPS of the carbon region shows two peaks; the higher energy component corresponds to carbon bonded to fluorine atoms. FTS/VTS copolymer fibers have θa ) 161° and as such are classified as superhydrophobic. Receding angles are not reported because of inherent difficulties associated with obtaining accurate θr for extremely hydrophobic surfaces with the goniometer employed for the present measurements. Qualitatively, copolymer fibers have a larger θr than VTS fibers as water bounces off substrates bearing copolymer fibers (Figure 11a). To better illustrate the transformation, a picture of water streaming onto a bare substrate is also included.
Conclusions Critical parameters that promote siloxane fiber formation have been isolated, and a reasonable mechanism has been presented. It is clear from the present study that water equilibrium and slow addition of VTS are vital if fibers are to be formed. We have also established that substrates must possess hydroxyl terminated surfaces for fibers to grow. Fibers nucleate when hydrolyzed VTS moieties couple with surface hydroxyl groups to form polymeric islands. These islands emerge from the adsorbed water layer as more hydrolyzed VTS react with the polymeric islands. Additional VTS molecules diffusing to the substrate surface preferentially couple to the islands due to the large surface area, resulting in fiber growth normal to the surface. We have also confirmed that hydroxyl groups must be accessible to react with VTS, and hence, substrates must be
cleaned to remove organic contamination. Both oxygen plasma and piranha are suitable for activating oxide terminated substrates; both adequately remove organic contamination and increase the effective concentration of surface hydroxyl groups. Substrates patterned with FTS show fiber growth only in regions not patterned. These results further support our conclusion that surface hydroxyl groups are required for nanofiber growth. The aqueous wetting behavior of substrates bearing fibers was also found to depend on fiber spatial density and length. Advancing contact angles vary from 68° to 131°, and receding angles vary from 44° to 127°. The highest angles were obtained on substrates with spatially dense, long fibers. Substrates bearing VTS/FTS copolymer nanofibers are superhydrophobic, exhibiting advancing contact angles of 161°. Adding to the versatility of the present method, our vapor phase fabrication technique also provides a convenient procedure for preparing fiber structures within pores. Many chromatographic and microfluidic devices require coatings placed on the inside of pores. Polysiloxane nanofibers would be well-suited to these applications in light of their high surface area and tailorable surface chemistry. Optimization of this process is the subject of continuing work in our laboratory. Acknowledgment. The authors thank the Natural Sciences and Engineering Research Council of Canada (NSERC) for funding; G. D. Braybrook for assistance with SEM imaging; the staff of the University of Alberta NanoFab for assistance; and the Alberta Centre for Surface Engineering and Sciences (ACSES) for XPS, AES, and ToF-SIMS analyses. LA801595M
