Controlled Growth of Silicon Dioxide from “Nanoholes” in Silicon

Xinqiao Jia and Thomas J. McCarthy*. Polymer Science and Engineering Department, University of Massachusetts,. Amherst, Massachusetts 01003. Received ...
0 downloads 0 Views 409KB Size
Langmuir 2003, 19, 2449-2457

2449

Controlled Growth of Silicon Dioxide from “Nanoholes” in Silicon-Supported Tris(trimethylsiloxy)silyl Monolayers: Rational Control of Surface Roughness at the Nanometer Length Scale Xinqiao Jia and Thomas J. McCarthy* Polymer Science and Engineering Department, University of Massachusetts, Amherst, Massachusetts 01003 Received June 3, 2002. In Final Form: November 11, 2002 Controlled growth of silicon dioxide (SiO2) using tetrachlorosilane (SiCl4) and water as precursors on tris(trimethylsiloxy)chlorosilane (trisTMSCl)-modified silicon wafer templates was studied. By manipulating the kinetics of the vapor-phase reaction of trisTMSCl with silicon wafers, surfaces with varying densities and distributions of unreacted silanols (in nanoholes) were obtained. Subsequent treatment with SiCl4/ H2O led to site-specific growth of silica from the nanoholes that was monitored by atomic force microscopy (AFM). Different nanoscale structures with varying surface roughness and wettability were fabricated by controlling the growth kinetics. Modification of the newly grown silica with tridecafluoro-1,1,2,2tetrahydrooctyldimethylchlorosilane (FDCS) allowed the growth kinetics to be followed by X-ray photoelectron spectroscopy. Chemical etching effectively removed the organic residues, resulting in hydrophilic silica surfaces with nanoscale roughness. Further modification with FDCS rendered the surfaces hydrophobic. Water contact angle analysis and AFM clearly indicate that nanometer scale topography has a profound effect on surface wettability.

Introduction The deposition of silicon dioxide (SiO2) thin films is important in a variety of applications. In the semiconductor industry, it is widely used as interinsulators for largescale integrations,1 liquid-crystal displays,2 and gate insulators for thin-film transistors.3 Deposition of thin SiO2 layers on the surface of porous Vycor glass gives highly perm-selective membranes for H2 separation.4 In the coatings industry, SiO2 is a useful material because of its hardness, transparency in the visible spectrum, chemical stability, high electrical resistivity, and gas shielding ability.5 Growth of SiO2 from tin oxide disks templated with organic molecules leads to sensors with clear molecular recognition functions.6 In the past few years, tremendous effort has been put into the fabrication of nanostructures, and silica nanostructures have attracted considerable attention because of their potential application in mesoscopic research, the development of nanodevices, and the potential use of large surface area structures for catalysis.7 Silica nanowires and nanotubes,8 polymer/silica nanocomposites,9 and silica core-shell particles10 are several examples. Recently we prepared nanoscopic silicon dioxide posts on silicon wafers.11 Arrays of nanopores oriented normal to the surfaces were used * Corresponding author. E-mail: [email protected]. (1) Homma, T.; Katoh, T.; Yamada, Y.; Murao, Y. J. Electrochem. Soc. 1993, 140, 2410. (2) Chou, J. S.; Lee, S. IEEE Trans. Elec. Dev. 1996, 43, 599. (3) Kitaoka, M.; Honda, H.; Yoshida, Y.; Takagawa, A.; Kawahara, H. Int. Conf. Thin Film Phys. Appl. 1991, 109. (4) Kim, S.; Gavalas, G. R. Ind. Eng. Chem. Res. 1995, 34, 168. (5) Inoue, Y.; Takai, O. Thin Solid Films 1998, 316, 79. (6) Tanimura, T.; Katada, N.; Niwa, M. Langmuir 2000, 16, 3858. (7) Legrand, A. P., Ed. The Surface Properties of Silica; Wiley: New York, 1998. (8) Wang, Z. L.; Gao, R. P.; Gole, G. L.; Stout, J. D. Adv. Mater. 2000, 12, 1938. (9) Hsiue, G. H.; Kuo, W. J.; Huang, Y. P.; Jeng, R. J. Polymer 2000, 41, 2813. (10) Scha¨rtl, W. Adv. Mater 2000, 12, 1899.

as nanoscopic reaction vessels and silicon dioxide was formed within the pores defined by a cross-linked polystyrene scaffold. Reactive ion etching was used to remove the organic matrix, leaving free-standing silicon dioxide posts on the silicon substrate. We are interested in using silicon dioxide growth to modify the surface characteristics of flat silicon surfaces, including roughness, wettability, and nanoscale surface topography, and this was the main focus of this study. Chemical modification of metal oxide surfaces12 with organosilanes has been utilized to control adhesion, wettability, adsorption, catalytic, and other surface properties. Monofunctional silanes form covalently attached monolayers on these surfaces, while certain trifunctional silanes form self-assembled monolayers.13 We recently prepared closely packed tris(trimethylsiloxy)silyl (trisTMS) monolayers with nanoholes of cross section of ∼0.5 nm2 and used these substrates for the synthesis of uniformly mixed binary monolayers on oxidized silicon wafers.14 The adsorption of carboxylic acid end-functionalized polystyrene (PS-COOH) onto (sub)monolayers of trisTMS with different surface coverages was examined and the effectiveness of these adsorbed layers in suppressing dewetting of a thin polystyrene film was examined.15 In this report, we describe the controlled growth of silicon dioxide using tetrachlorosilane (SiCl4) and water as precursors and trisTMSCl-modified silicon wafers as templates. Surfaces with varying nanoscale topography, (11) Kim, H. C.; Jia, X.; Stafford, C. M.; Kim, D. H.; McCarthy, T. J.; Tuominen, M.; Hawker, C. J.; Russell, T. P. Adv. Mater. 2001, 13, 795. (12) Leyden, D. E., Ed. Silanes, Surfaces, and Interfaces; Gordon and Breach: New York, 1986. (13) Hoffmann, P. W.; Stelzle, M.; Rabolt, J. F. Langmuir 1997, 13, 1877. (14) Fadeev, A. Y.; McCarthy, T. J. Langmuir 1999, 15, 7238. (15) Stafford, C. M.; Fadeev, A. Y.; Russell, T. P.; McCarthy, T. J. Langmuir 2001, 17, 6547.

10.1021/la020515z CCC: $25.00 © 2003 American Chemical Society Published on Web 02/01/2003

2450

Langmuir, Vol. 19, No. 6, 2003

roughness, and wettability were fabricated. The resulting surface structures were thoroughly characterized by contact angle, ellipsometry, X-ray photoelectron spectroscopy, and atomic force microscopy. Experimental Section General Information. All reagents were used as received. Ethanol, 2-propanol, toluene, chloroform (all HPLC grade), concentrated sulfuric acid, hydrogen peroxide (30%), and sodium dichromate were obtained from VWR Scientific. Carbon tetrachloride (anhydrous), hexadecane (anhydrous), and o-xylene (HPLC) were purchased from Aldrich. Organosilanes were obtained from Gelest. House-purified water (reverse osmosis) was further purified by using a Millipore Milli-Q system that involves reverse osmosis, ion exchange, and filtration steps (18 × 106 Ωcm). Silicon wafers were obtained from International Wafer Service (100 orientation, P/B doped, resistivity from 20 to 40 Ωcm). Contact angle measurements were made with a Rame´-Hart telescopic goniometer equipped with a Gilmont syringe and a 24-gauge flat-tipped needle. Water, purified as described above, was used as the probe fluid. Advancing (θA) and receding (θR) contact angles were recorded while the probe fluid was added to and withdrawn from the drop, respectively. Contact angles reported are an average of at least three measurements taken at different locations on the sample; all values for each sample were in a range of (3°. X-ray photoelectron spectra (XPS) were obtained on a Perkin-Elmer Physical Electronics 5100 using Mg KR excitation (15 kV, 400 W). The pressure in the analysis chamber was less than 10-8 Torr during analysis. Spectra were taken at a takeoff angle of 75° (between the plane of the surface and the entrance lens of the detector optics) at pass energies of 89.45 and 35.75 eV for survey and C1s region, respectively. Atomic force microscopy (AFM) measurements were performed on a Dimension 3100 microscope system (Digital Instruments, Inc.). Silicon cantilevers with a typical resonant frequency of 300 kHz and spring constant of 42 N/m were used to acquire images in TappingMode at room temperature under ambient conditions. The scanning rate was around 1 Hz. “Flattening” (software that flattens long-scale topography) was applied to the raw images before performing roughness analysis. Images were captured on 1 × 1 µm areas. Roughness analysis was an average from three random areas per sample. Film thickness was measured with a Rudolph Research AutoEL-II automatic ellipsometer equipped with a helium-neon laser (λ ) 6328 Å) at an incidence angle of 70° (from the normal to the plane). Chemical Modification of Silicon Dioxide. Silicon wafers were cut into 1.2 × 1.2 cm2 pieces and submerged in a freshly prepared mixture of 7 parts concentrated sulfuric acid containing dissolved sodium dichromate (∼3-5 wt %) and 3 parts of 30% hydrogen peroxide. Plates were exposed to the solution overnight, rinsed with copious amounts of water, and placed in a clean oven at 120 °C for 1 h. Samples cleaned and dried in this fashion were used immediately. A vapor phase reaction was applied to modify the surfaces with monochlorosilanes (trisTMSCl and FDCS) following a described procedure.14 In a typical experiment, the substrates were placed in a custom-made wafer holder (slotted glass cylinder) and suspended in a reaction tube containing 0.5 mL of silane. There was no direct contact between the silane liquid and the substrates. The reaction was carried out at 68 ( 1 °C for varying times. After the silanization reaction, wafers were rinsed with 1× 10 mL of toluene, 2× 10 mL of 2-propanol, 2× 10 mL of ethanol, 1× 10 mL of ethanol/water mixture (1:1), 2× 10 mL of water, 2× 10 mL of ethanol, and 2× 10 mL of water (in this order) and then dried in an oven at 120 °C for 30 min. Growth of Silicon Dioxide from TrisTMS Monolayers. Silicon dioxide was grown from trisTMSCl-modified silicon wafers by a chemical vapor deposition (CVD) method in a vacuum system controlled by a Mano-Watch (Model MW-1000, Instruments for Research and Industry, I2R, Inc.) device. Samples were placed in a sample holder in a reaction vessel which was subsequently flushed with nitrogen, evacuated to 20 mTorr, and equilibrated to 50 ( 1 mTorr. SiCl4 vapor was introduced from a reservoir cooled with a mixture of o-xylene and liquid nitrogen (-23 °C). The deposition took place at 25 ( 2 °C and 50 ( 1 mTorr. After

Jia and McCarthy 1 min of SiCl4 exposure, the SiCl4 reservoir was closed and the reaction vessel was purged with nitrogen and evacuated to 20 mTorr (step A). The reaction vessel was then opened to the air for 5 min, during which period an equilibrium amount of water re-adsorbed onto the sample surface. The reaction vessel was again evacuated to 20 mTorr and reequilibrated to 50 mTorr (step B). The growth of SiO2 was controlled by repeating the A-B sequences. After the desired number of cycles, the substrates were removed from the reaction system, rinsed with ethanol and water, and then dried in an oven at 120 °C for 30 min. AFM images were acquired to monitor the evolution of surface topography. Samples that were subsequently treated with FDCS using the vapor phase reaction described above were characterized by contact angle and XPS. To determine the thickness of the chemisorbed silica, the CVD of silica was carried out on silicon wafers without the preadsorbed trisTMS.

Results and Discussion Growth of Silicon Dioxide from Unmodified (Smooth) Silicon Wafers. The growth of SiO2 from unmodified silicon wafers was studied in order to gain insight into the growth kinetics and film quality. To grow SiO2 from trisTMS-templated silicon surfaces, two conditions need to be satisfied: the deposition method must not destroy or contaminate the trisTMS monolayer; the film thickness must be controllable at a thickness scale comparable to the height of the trisTMS groups. Deposition of thin film SiO2 has been widely studied. Silica films can be deposited by either liquid-phase deposition (LPD)16,17 or chemical vapor-phase deposition (CVD).18,19 A selective SiO2 film can be formed by adding boric acid (H3BO3) to a supersaturated hydrofluosilicic acid (H2SiF6) solution at room temperature and this is a promising candidate for many advanced applications. The film, however, is contaminated with fluorine.16 Using tetraethyl orthosilicate (TEOS) as a silicon source eliminates fluorine contamination, but the growth kinetics is complicated by many parameters and the film is not uniform at the nanometer scale.18 Conventional CVD using organosilicon compounds has been extensively studied,18 but it usually involves high temperature, which is not suitable for our organic (trisTMS) system. Plasma-enhanced CVD offers an alternative to using high temperatures,19 but it is not selective to surface chemistry and it is likely that the trisTMS monolayer will be plasma-degraded. Atomic layer control of thin film growth can be effected by using selflimiting surface reactions and a binary reaction sequence. Brunner et al.20 reported a procedure that involves the formation of an alkylsiloxane monolayer through selfassembly of a trifunctional silane from solution (step A) followed by UV/ozone oxidation of the hydrocarbon groups (step B). Repeated application of this A-B cycle resulted in a linear thickness increase of 2.7 Å per cycle. Klaus et al.21 have used a binary reaction sequence that involves the exposure of a silica surface in ultrahigh vacuum to SiCl4 and H2O vapor in an alternating sequence. By using pyridine as a catalyst for both cycles, pinhole-free and homogeneous SiO2 thin films were produced at room temperature with thickness control at the molecular level. We modified this atomic layer deposition method so that it was unnecessary to use catalyst and a complicated high vacuum system and could be carried out in a conventional (16) Chang, P. H.; Huang, C. T.; Shie, J. S. J. Electrochem. Soc. 1997, 144, 1144. (17) Usami, K.; Hayashi, S.; Uchida, Y.; Matsumura, M. Jpn. J. Appl. Phys. 2 1998, 37, L97. (18) Katada, N.; Tokyya, T.; Niwa, M. J. Phys. Chem. 1994, 98, 7647. (19) Inoue, Y.; Takai, O. Thin Solid Films 1999, 341, 47. (20) Brunner, H.; Vallant, T.; Mayer, U.; Hoffmann, H. Langmuir 1996, 12, 4614. (21) Klaus, J. W.; Sneh, O.; George, S. M. Science 1997, 278, 1934.

Controlled Growth of Silicon Dioxide

Figure 1. Total SiO2 film thickness for deposition on smooth silicon wafers versus number of SiCl4/water cycles at 50 mTorr and 25 °C.

organic laboratory. Our cleaning procedure for silicon wafers not only removes organic impurities but also results in a homogeneous and fully hydrated surface, with 4-5 SiOH groups/nm2 remaining after the sample is dried at 120-130 °C.22 This surface spontaneously adsorbs an equilibrium amount of water when exposed to atmosphere. Previous research has shown that hydrolysis of chlorosilanes (with the exception of certain (fluoroalkyl)chlorosilanes23) to silanols at the solid-gas interface by surface water on hydrated silica is required before condensation with the surface silanols.24 Based on these observations, a reaction procedure consisting of repeated application of an A-B sequence (as described in the Experimental Section) was devised to grow SiO2 in a controlled manner. During step A, in which SiCl4 is introduced at controlled vapor pressure, SiCl4 molecules in the vapor phase hydrolyze to form silanols in the presence of surfaceabsorbed water and subsequently condense with surface silanols on the substrate to form siloxane bonds. If this deposition is allowed to proceed for longer than 5 min, the growing process stops because the surface-adsorbed water is consumed. During step B, in which air is introduced (the system had been evacuated and the valve connected to the SiCl4 reservoir closed between A and B cycles), readsorption of water on the surface occurs, hydrolyzing residual surface chlorosilanes. The reaction chamber is then flushed with nitrogen and evacuated (∼20 mTorr) before the next A-B sequence. This allows for continuous SiO2 growth. A 1-min reaction time was chosen for step A and was used for all experiments described here. Film thickness was measured by ellipsometry and at least three different locations were examined on each sample. A refractive index of 1.462 was used for both the native oxide layer and the newly grown SiO2. Figure 1 shows the total SiO2 film thickness deposited on the Si (100) wafer versus the number of A-B cycles. The SiO2 film thickness increases linearly with the increase of the number of reaction cycles and the measured growth rate is 0.8 nm per A-B cycle. Tapping mode AFM studies of the deposited SiO2 films show that surface topography is comparable with the roughness of the initial SiO2 layer on the silicon substrate (Rqsroot-mean-square roughnesss (22) Zhuravlev, L. T. Langmuir 1987, 3, 316. (23) Tripp, C. P.; Veregin, R. P. N.; Hair, M. L. Langmuir 1993, 9, 3518. (24) Tripp, C. P.; Hair, M. L. Langmuir 1995, 11, 1215.

Langmuir, Vol. 19, No. 6, 2003 2451

of 0.20 nm) during the early deposition (first several cycles). There is a slight roughening effect with more numerous deposition cycles (Rq ) 0.50 nm is observed for a film prepared with 10 A-B cycles). This roughening effect is due to the existence of multiple reaction possibilities and the lack of a self-limiting mechanism involved in this system. Partially hydrolyzed species have different reactivities than completely hydrolyzed ones and the native surface silanols have different reactivity than the nascent ones. These differences suggest that control at the atomic level will be difficult; however, the constant growth rate observed in Figure 1 implies that the kinetics are well controlled and reproducible and that the roughening effect is present to only a limited extent. There is no detectable surface area increase for a film deposited with 30 repeat cycles. The thickness of the trisTMS monolayer is ∼0.7 nm, which is comparable to the SiO2 growth rate. This suggests that selective topography changes can be induced by using trisTMS monolayers and this simplified low-tech CVD process. Increasing the temperature of the SiCl4 liquid and decreasing the pressure of the reaction system (increasing the SiCl4 partial pressure) results in faster growth rates and rougher films. Decreasing the reaction time of step A or decreasing the SiCl4 vapor pressure may lead to slower growth rate and smoother films. In the following section, the number of cycles is the only parameter that is varied to control the growth of SiO2 from trisTMSCl-modified silicon wafers. Growth of Silicon Dioxide from trisTMSCl-Modified Silicon WaferssAFM Analysis. The reaction of monochlorosilanes with silicon dioxide (silicon wafers) is a useful route to covalently attached monolayers with a variety of controllable structures.25 The strong bonding in this system sets it apart from self-assembled monolayers prepared with alkyltrichlorosilanes, which are strongly bonded with neighbors (lateral siloxane bonds and van der Waals interactions between alkyl chains) and to a small extent with the surface. As a result, the distance between molecules is significantly greater than in selfassembled monolayers.12 Using a very bulky organosilane, trisTMSCl, we have prepared14 complete monolayers of trisTMS on Si wafers (eq 1). These monolayers are

“complete” in the sense that they are incapable of incorporating any additional trisTMS groups but can react with smaller silanes14 or end-functional polymers.15 Thus the monolayer contains what we call “nanoholes” with a maximum cross-sectional area of ∼0.5 nm2. The kinetics of the trisTMSCl reaction is interesting and useful to the studies reported here. Figure 2 shows water contact angle data for the reaction as a function of time. Significant hydrophobization occurs within 1 h, and the reaction continues over days and is not complete until 5 days. The size of the nanoholes on the trisTMS-templated surface can be tuned by varying the reaction time. Three different reaction times representing different reaction stages were chosen to study the growth of silica: 3, 60, and 120 h with surface coverages of 60%, 81%, and 96% (calculated from the Israelachvili equation14), respectively. These surfaces (25) Fadeev, A. Y.; McCarthy, T. J. Langmuir 1999, 15, 3759.

2452

Langmuir, Vol. 19, No. 6, 2003

Jia and McCarthy Table 2. AFM Data for no. of reaction cycles 0 1 2 4 6 8 10 12

81TRIS

Rq,a nm

Ra,b nm

Rmax,c nm

∆S,d %

0.1(0)e

0.1(0)e

1.2(0.1)e

1.0(0.3) 2.3(0.5) 5.7(0.5) 6.0(0.4) 7.5(0.6) 8.6(0.2) 4.3(0.3)

0.6(0.2) 1.0(0.3) 4.3(0.3) 4.9(0.2) 6.4(0.6) 7.6(0.5) 3.5(0.2)

0(0)e 0(0) 1(0) 6(2) 10(1) 13(3) 15(3) 7(2)

15.3(1.6) 25.0(3.5) 29.6(1.9) 33.3(3.0) 36.2(3.2) 41.2(3.9) 29.4(2.4)

a Root-mean-square roughness. b Mean roughness. c Peak-tovalley distance. d Percent surface area increase. e Values in parentheses are standard deviation of three measurements.

Figure 2. Kinetics of the reaction of trisTMSCl with silicon wafers and graphic descriptions of the areas covered with TRIS groups.15 Water contact angle data (θA, b;θR, O). Table 1. AFM Analysis for

60TRIS

no. of reaction cycles

Rq,a nm

Ra,b nm

Rmax,c nm

∆S,d %

0 1 2 3 4 5

0.2(0)e 2.5(0.2) 4.4(0.4) 5.1(0.5) 4.8(0.4) 2.6(0.2)

0.1(0)e 1.8(0.1) 3.3(0.2) 4.4(0.3) 3.7(0.4) 2.1(0.3)

1.6(0.1)e 15.0(1.6) 26.6(1.3) 27.2(2.9) 29.5(3.4) 17.0(2.6)

0(0)e 5(0.0) 9(1) 15(3) 7(2) 7(1)

a Root-mean-square roughness. b Mean roughness. c Peak-tovalley distance. d Percent surface area increase. e Values in parentheses are standard deviation of three measurements.

are abbreviated 60TRIS, 81TRIS, and 96TRIS and collectively “TRIS surfaces”; pictorial representations of them are shown in Figure 2. SiO2 was grown from trisTMSCl-modified silicon wafers using the method described above for smooth silicon wafers. This requires that water from the vapor phase adsorb in the nanoholes and be available for reaction with SiCl4. Several data suggest that this is the case: (1) the calculations using the Israelachvili equation14 to determine surface coverage assumed a mixture of trisTMS groups and silanols and the probe fluid was water. Clearly the silanols affect the contact angle and are thus assessed by liquid water. (2) Contact angle hysteresis studies14 suggest that water penetrates the nanoholes, giving rise to contact line pinning and hysteresis. Only molecules with molecular cross sections greater than ∼0.5 nm2, e.g. dimethyladamantane, do not sense silanols. AFM was utilized to characterize the evolution of surface topography; contact angle and XPS were employed to follow the reaction kinetics. Figure 3 and Table 1 summarize the AFM data for the sequential reaction of SiCl4/ water with 60TRIS, the surface 60%-covered with trisTMS groups. As compared to the originally smooth trisTMS functionalized surface, nanoscale SiO2 clusters are observed on the surfaces treated with SiCl4/H2O. After the first reaction cycle, SiO2 appears as individual spherical caps uniformly distributed across the surface with diameters and heights in the range 50-66 and 7-15 nm, respectively (Figure 3a). As the number of repeating cycles increases, the surface density and the size of the spherical caps increase and the space between them is gradually filled up. After three repeating cycles, the average diameter and height of the SiO2 spherical caps has increased to 60-80 nm and 20-27 nm, respectively, and some of them

have impinged to form aggregates (Figure 3c). We note that because of the convolution effect of the AFM tip (1525 nm in diameter) with the surface nanoscale structures, SiO2 clusters may be smaller than they appear in AFM images during the early deposition stage. The relative height difference, however, is not affected. Closely packed 2D aggregation patterns are developed after five repeating cycles (Figure 3d). Water completely spreads on this surface, indicating that it is fully covered by SiO2 clusters and that no trisTMS groups are accessible to the probe fluid. The roughness analysis gives more insight into the growth process. AFM roughness analysis was performed on 1 × 1 µm scan areas and the results reported in Table 1 are averages from three different scan areas on the same sample. Four parameters were used to measure the surface roughness. Rq (root-mean-square roughness) is the standard deviation of the Z values within a given area. Ra (mean roughness) represents the arithmetic mean of the deviations from the center plane. Rmax describes the difference in height between the highest and lowest points on the surface relative to the mean plane, and ∆S is the percentage increase of the surface area over the projected surface area. It is clear that surface roughness increases with the growth of SiO2, reaching a maximum after three reaction cycles, with 15% of surface area increase. Further growth leads to decreases in roughness and ∆S due to the impingement of SiO2 nanoclusters. We note that peaks as high as 29.5 nm (after four reaction cycles) are observed and maximum peak height decreases to 17.0 nm after the fifth cycle. Figure 4 is a graphic description of this process. Silica nucleates in nanoholes by reaction of SiCl4 with residual silanols and grows up and out from the SiO2 surface forming mushroom-shaped caps that eventually impinge. Nucleation and growth are competitive and occur at similar rates; there are sections with no nucleation when significant size nanoclusters are formed and no abnormally large clusters are observed until impingement. The growth of silica from the more densely covered 81TRIS and 96TRIS surfaces was also studied and the results are consistent with the observations and analyses made above for the 60TRIS surface, but with predictable and explainable differences. AFM micrographs and roughness data for 81TRIS are shown in Figure 5 and Table 2 and for 96TRIS are shown in Figure 6 and Table 3. The smaller nanohole sizes give rise to a lower density of nucleation sites, therefore slower initial SiO2 growth rates. After four SiCl4/water cycles for 81TRIS, and 12 SiCl4/ water cycles for 96TRIS, fewer SiO2 nanoclusters have been nucleated than after one SiCl4/water cycle for 60TRIS (compare Figures 3, 5, and 6). It requires 12 and 28 reaction cycles for SiO2 to completely cover 81TRIS and 96TRIS surfaces, respectively, compared with only five cycles in the case of 60TRIS. Complete impingement of the nano-

Controlled Growth of Silicon Dioxide

Figure 3. AFM height images of controlled growth of SiO2 from (b) 2; (c) 3; (d) 5. Images are 1 × 1 µm.

Langmuir, Vol. 19, No. 6, 2003 2453

60TRIS

surfaces. Number of SiCl4/water reaction cycles: (a) 1; Table 3. AFM Data for

96TRIS

no. of reaction cycles

Rq,a nm

Ra,b nm

Rmax,c nm

∆S,d %

0 6 12 16 20 25 28

0.1(0)e 4.9(0.3) 9.5(1.0) 13.3(1.3) 13.4(2.4) 11.4(0.3) 10.8(0.5)

0.1(0)e 2.1(0.2) 7.5(1.0) 10.7(1.0) 10.9(2.0) 9.2(0.3) 8.7(0.7)

1.1(0)e 39.0(4.0) 59.1(5.6) 82.0(3.9) 80.9(7.5) 69.8(2.9) 61.9(4.5)

0(0)e 3(1) 19(2) 22(5) 24(4) 18(2) 11(2)

a Root-mean-square roughness. b Mean roughness. c Peak-tovalley distance. d Percent surface area increase. e Values in parentheses are standard deviation of three measurements.

Figure 4. Graphic description of build-up of SiO2 using SiCl4/ water reaction cycles on a TRIS monolayer.

clusters is difficult to assess by AFM but is trivial to assess by contact angle, which is discussed below. The size of the nanoclusters logically depends on the number of nucleation sites. From fewer nucleation sites, the individual SiO2

clusters grow much larger and taller before they eventually impinge. The diameter for SiO2 nanoclusters varies in the range of 80-100 nm for 81TRIS surfaces, and 90-160 nm for 96TRIS surfaces. Roughness also depends on initial TRIS surface coverage (compare Tables 1, 2, and 3), with rougher surfaces forming from surfaces with fewer and

2454

Langmuir, Vol. 19, No. 6, 2003

Figure 5. AFM height images of controlled growth of SiO2 from (b) 4; (c) 8; (d) 12. Images are 1 × 1 µm.

smaller nanoholes. As was the case with 60TRIS, the roughness and surface area increase to a maximum with repeated reaction cycles and then decrease upon impingement of the nanoclusters. Note the difference in height range between the three TRIS surfaces. On 60TRIS, the highest value obtained for Rmax is ∼30 nm compared to ∼40 nm on 81TRIS and 82 nm on 96TRIS. With fewer nucleation sites, the nanoclusters can grow much taller before they impinge. Growth of Silicon Dioxide from TrisTMSCl-Modified Silicon WaferssContact Angle and XPS Analysis. The TRIS surfaces vary in wettability with 60TRIS, 81 TRIS, and 96TRIS, exhibiting water contact angles of θA/θR ) 68°/57°, 86°/71°, and 102°/90°, respectively. All are hydrophobic and water drops slide relatively easily on them due to their low hysteresis.26 The differences in (26) Oner, D.; McCarthy, T. J. Langmuir 2000, 16, 7777.

Jia and McCarthy

81

TRIS surfaces. Number of SiCl4/water reaction cycles: (a) 2;

their contact angles are due to the relative amounts of trisTMS and silanol groups that the probe fluid assesses (they have been named “60, 81, and 96” due to the percentage of trisTMS assessed as calculated by the Israelachvili equation). As silica is grown on these surfaces using sequential SiCl4/water reactions, both the chemical composition and the topography changes. Hydrophilic (clean silica exhibits water contact angles of θA/θR ) 0°/0°) silica fills the hydrophilic nanoholes and covers the hydrophobic trisTMS groups, giving rise to hydrophilic islands of varying height in a sea of hydrophobic trisTMS groups. The surfaces are binary composites with elevated hydrophilic and depressed hydrophobic regions. This will confound any analysis using classical wettability equations for composition or roughness, so no analysis of this sort was carried out. Figure 7 shows the contact angle changes (advancing and receding) for all three TRIS surfaces. These data reflect what is seen in the AFM images, and all

Controlled Growth of Silicon Dioxide

Figure 6. AFM height images of controlled growth of SiO2 from (b) 12; (c) 25; (d) 28. Images are 1 × 1 µm.

surfaces become more hydrophilic with increasing SiO2 coverage. Water spreads (completely wets - θA ) 0°) 60TRIS, 81TRIS, and 96TRIS surfaces after 5, 12, and 28 reaction cycles, respectively. After these numbers of sequential reactions, no trisTMS groups are accessible to the probe fluid, indicating that the nanoclusters are completely impinged on each other, giving pure silica surfaces. We note the precipitous decreases in θA on all surfaces in the late stages of the deposition. θA decreases from 44°, 51°, and 49° to 0° on going from 4 to 5, 10 to 12, and 25 to 28 reaction cycles for 60TMS, 81TMS, and 96TMS, respectively. Small amounts of residual trisTMS groups can dramatically decrease wettabilty as predicted by the Israelachvili equation. Only small changes occur in XPS spectra upon SiCl4/ water reactions because there is little compositional change: native silicon dioxide is indistinguishable from nascent SiO2 and silica is always contaminated with some carbon impurities so the carbon in the trisTMS groups is

Langmuir, Vol. 19, No. 6, 2003 2455

96

TRIS surfaces. Number of SiCl4/water reaction cycles: (a) 6;

not easily characterized. To follow the deposition, the surface silanols were labeled with a fluorinated monochlorosilane, tridecafluoro-1,1,2,2-tetrahydrooctyldimethylchlorosilane (FDCS). This allows the monitoring of silanol concentration, thus silica surface area, by following the fluorine atomic concentration by XPS. Equation 2 describes the surface labeling reaction. The reaction was carried

out in the vapor phase at 68 °C for 24 h, conditions that give complete coverage of the SiO2 surface by perfluoroalkyl groups. The resulting surface is interesting in that it is a quasi two-dimensional composite of trisTMS and perfluoroalkyl groups, with the FDCS domain varying in height at the nanoscale. Figure 8 shows surface fluorine concentration for samples of all three TRIS surfaces

2456

Langmuir, Vol. 19, No. 6, 2003

Figure 7. Water contact angle data for 60TRIS (θA, 2; θR, 4), 81 TRIS (θA, b; θR, O) and 96TRIS (θA, 9; θR, 0) as a function of the number of SiCl4/water reaction cycles.

Jia and McCarthy

Figure 9. Water contact angle data for TRIS/SiO2/FDCS surfaces as a function of the number of SiCl4/water reaction cycles: 60TRIS/SiO2/FDCS (θA, 2; θR, 4), 81TRIS/SiO2/FDCS (θA, b; θR, O), 96TRIS/SiO2/FDCS (θA, 9; θR, 0). Table 4. Roughness Effects on Wettability (Perfluoroalkyl Surfaces)

c

Figure 8. XPS fluorine concentration on TRIS/SiO2FDCS composite surfaces versus the number of SiCl4/water reaction cycles: 60TRIS, 2; 81TRIS, b; 96TRIS, 9.

modified with SiCl4/water and subsequently labeled with FDCS versus the number of reaction cycles. Spectra were recorded at a 75° takeoff angle (rather than a more shallow, more surface-selective angle) to minimize shadowing by the elevated silica islands. Reacted 60TRIS, 81TRIS, and 96 TRIS exhibit fluorine atomic concentrations of 9.2%, 4.2%, and 1.3%; this trend is expected as the nanoholes decrease in size and number in this progression. Fluorine atomic concentration increases for all three surfaces with number of reaction cycles as the percentage of silica on the surface increases and more fluorinated label is incorporated (Figure 8). The increase in fluorine content levels as the silica nanoclusters impinge and only silanols are present on the surface. Wenzel’s equation27 predicts that surface roughness not only enhances the hydrophilicity of hydrophilic surfaces but also enhances the hydrophobicity of hydrophobic ones. Again this is a complex mixed surface (silicone/fluoroalkyl) with topography-dependent composition, so using equations to quantify roughness or surface composition based on contact angle data is not a sensible endeavor. The water (27) Wenzel, R. N. J. Phys. Colloid Chem. 1949, 53, 1466.

Rq,a nm

θA

θR

∆S,b %

RWc

0.1 4.9 9.5 13.3

108° 118° 143° 150°

95° 92° 93° 90°

0 3 19 22

1 1.52 2.59 2.80

a Root-mean-square roughness. b Percent surface area increase. Wenzel’s roughness.

contact angle data are presented in Figure 9 for all of the surfaces described in Figure 8. The general trend is that with increasing surface roughness, the advancing contact angle and the hysteresis increase. The receding angles change very little; the low receding contact angles (and high hysteresis) for the samples of 96TRIS/SiO2/FDCS reflect their roughness (Table 3). The trisTMS groups used to template the growth of SiO2 can be easily removed by oxidation using a mixture of H2SO4/H2O2/Na2Cr2O7. This gives hydrophilic silica surfaces with varying nanoscale roughness. All of these surfaces are completely wet by water and other probe fluids, so roughness differences cannot be assessed by contact angle. To carry out contact angle analysis we have modified several of these surfaces with FDCS to prepare one-component rough hydrophobic surfaces. Table 4 shows contact angle data for four FDCS-modified surfaces with various degrees of roughness: 96TRIS modified by treatment with 0, 6, 12, and 16 cycles of SiCl4/water. These surfaces exhibit peak-to-valley distance values ranging from 1.1 to 82 nm and ∆S values of 0-22% (the complete roughness data are summarized in Table 3). The receding contact angles do not change much with roughness, but the advancing contact angles increase markedly to as high as 150°. Wenzel’s roughness (Rw) values can be calculated by using the advancing contact angle data and the equation Rw ) cos θrough/cos θsmooth, and these are shown in Table 4. Wenzel’s roughness is the actual surface area (contour) divided by the projected (2D) surface area, which should be the same measure as ∆S determined from AFM. The ∆S value for the roughest surface is only 22% (WR ) 1.22), yet the RW value determined from the contact angle is 2.8. This indicates that ∆S values greatly underestimate roughness that affects wettability. We have reported that molecular scale topography influences wettability.25

Controlled Growth of Silicon Dioxide

Summary The controlled growth of silicon dioxide nanoclusters from trisTMSCl-modified silicon wafers was carried out by sequential exposure of the substrates to SiCl4 vapor and atmosphere (water). Surfaces with different topographies and wettabilities were prepared by controlling the surface density of trisTMS groups (by reaction kinetics) and the number of SiCl4/water sequential reactions. Chemisorption of FDCS on the newly grown silica clusters resulted in binary mixed surfaces with one component

Langmuir, Vol. 19, No. 6, 2003 2457

varying in height from 10 to 80 nm. Chemical etching effectively removes the organic residues, resulting in hydrophilic rough silica surfaces. Further modification with FDCS indicates that wettability can be controlled by nanoscale roughness. Acknowledgment. We thank the Office of Naval Research and the NSF-sponsored Materials Research Science and Engineering Center for financial support. LA020515Z