Wettability is a key parameter influencing capillary pressures, permeabilities, fingering mechanisms, and saturations in multiphase flow processes within porous media. Glass-covered silicon micromodels provide precise structures in which pore-scale displacement processes can be visualized. The wettability of silicon and glass surfaces can be modified by silanization. However, similar treatments of glass and silica surfaces using the same silane do not necessarily yield the same wettability as determined by the oil-water contact angle. In this study, surface cleaning pretreatments were investigated to determine conditions that yield oil-wet surfaces on glass with similar wettability to silica surfaces treated with the same silane, and both air-water and oil-water contact angles were determined. Borosilicate glass surfaces cleaned with standard cleaning solution 1 (SC1) yield intermediate-wet surfaces when silanized with hexamethyldisilazane (HMDS), while the same cleaning and silanization yields oil-wet surfaces on silica. However, cleaning glass in boiling concentrated nitric acid creates a surface that can be silanized to obtain oil-wet surfaces using HMDS. Moreover, this method is effective on glass with prior thermal treatment at an elevated temperature of 400°C. In this way, silica and glass can be silanized to obtain equally oil-wet surfaces using HMDS. It is demonstrated that pretreatment and silanization is feasible in silicon-silica/glass micromodels previously assembled by anodic bonding, and that the change in wettability has a significant observable effect on immiscible fluid displacements in the pore network.
 Micromodels are planar microfluidic devices with at least one transparent face that enables visualization of fluids within spatially structured pore networks created by etching or other microfabrication techniques [Berejnov et al., 2008; Buckley, 1991; Gunda et al., 2011; Javadpour and Fisher, 2008; Karadimitriou and Hassanizadeh, 2012; Lenormand, 1999]. Processes that are studied include, but are not limited to, displacement of a wetting fluid by an immiscible nonwetting fluid (drainage), or displacement of a nonwetting fluid by an immiscible wetting fluid (imbibition). For visualization in micromodels, the two fluids are distinguished from one another by their native differences in color (e.g., when one is a dark-colored oil), by differences in refractive index (when both are clear and colorless), or using colored or fluorescent dyes to facilitate distinction of two otherwise clear colorless fluids [Chomsurin and Werth, 2003; Grate et al., 2011; Lenormand, 1999]. Displacements in porous media are relevant to petroleum recovery [Abdallah et al., 2007; Buckley, 1991, 2001; Dandekar, 2006; Zhao et al., 2010], nonaqueous phase liquid contaminant transport [Bradford et al., 1999; Demond and Roberts, 1991; Dwarakanath et al., 2002; Hofstee et al., 1998; O'Carroll et al., 2004; Powers et al., 1996], and geological carbon sequestration [Benson and Cole, 2008; Chalbaud et al., 2010; Chiquet et al., 2007; Espinoza and Santamarina, 2010; Krevor et al., 2011; Zhang et al., 2011b].
 Wettability is a key parameter influencing capillary pressures, permeabilities, fingering mechanisms, and final saturations in displacement processes within porous media [Abdallah et al., 2007; Anderson, 1986; Lenormand, 1999]. The wettabilities of surfaces in micromodels for studying multiphase flow phenomena are typically dependent on the materials of micromodel fabrication. Micromodels have been developed in materials such as glass, epoxy resins, silicon, and most recently, polydimethylsiloxane (PDMS). Epoxy resin [Bergslien et al., 2004; Buckley, 1991; Lenormand, 1999] and PDMS models are oil-wet, although PDMS can be treated with plasma to create a water-wet surface temporarily [Berejnov et al., 2008; Bhattacharya et al., 2005; Javadpour and Fisher, 2008]. Glass micromodels, typically prepared by wet etching, are water-wet [Buckley, 1991; Conrad et al., 1992; Javadpour and Fisher, 2008; Lenormand, 1999]. The etched pore network in the surface is sealed with either a flat glass plate or a symmetrically etched pore network in glass, creating micromodels where all surfaces are made of the same material, and hence have the same wettability. Silicon micromodels are prepared by dry etching techniques in silicon wafers [Chomsurin and Werth, 2003; Gunda et al., 2011; Lenormand, 1999; Willingham et al., 2008; Zhang et al., 2011a, 2011b], yielding structures with vertical walls and much higher precision than wet-etched glass. The silicon surfaces are oxidized to create a water-wet silica interface, and the models are sealed with a borosilicate glass plate by anodic bonding at (typically) 400°C. We will call these silicon-silica/glass micromodels. While this technique provides strongly bonded devices with near-perfect structures, it does create models with different materials for the pore network and the cover plate.
 The combination of silicon-silica and glass is normally of little consequence for obtaining micromodels of uniform wettability so long as both surfaces are thoroughly cleaned to provide water-wet surfaces. In our treatment, we use the definition by Anderson , where water-wet corresponds to oil-water contact angles from 0 to 75, intermediate-wet from 75 to 115, and oil-wet from 115 to 180°. Glass and silica surfaces can be modified with silanes to alter the wettability, a process that is called “silanization” [Arkles et al., 2009; Fadeev, 2006; Grate and McGill, 1995; Jain et al., 2002; McGill et al., 1994; Menawat et al., 1984; Shahidzadeh-Bonn et al., 2004; Wei et al., 1993]. Alkylsilane or arylsilane reagents in the liquid form, or adsorbed from the vapor phase, react with surface silanols to create new siloxane bonds that covalently bond the alkyl or aryl groups to the surface, altering surface wettability. We have previously reported the air-water and oil-water contact angles obtained from diverse silanes on silica surfaces and found a linear correlation relating these contact angles (and their cosines) [Grate et al., 2012]. This study identified silanes that produce intermediate-wet and oil-wet surfaces on silica, and demonstrated the surface modification of an assembled silicon-silica/glass micromodel to obtain an intermediate-wet micromodel.
 Here, we turn our attention especially to the preparation of oil-wet surfaces on glass in comparison with silica. When using silanes intended to yield hydrophobic oil-wet surfaces, we have found that the resulting silica and glass surfaces do not necessarily have the same wettabilities even when treated under the same conditions. These differences between silica and glass represent a potential issue for multiphase flow studies in surface-modified silicon-silica/glass micromodels, where all surfaces should ideally have the same wettability. Glass and silica are different materials and their surfaces change with thermal treatments that affect surface silanol coverage. The availability of surface silanols then influences the silanization process, the surface density of attached organic groups, and the wettability outcome. For example, in the field of capillary gas chromatography, capillaries of glass, fused silica, and quartz have all been examined and noted to have different characteristics depending on the material, its thermal history, and the chemical pretreatments such as basic etching or acid leaching, and these treatments influence subsequent silanization processes [Bartle et al., 1981; Grob, 1986]. The advantageous features of silicon-silica/glass micromodels, and the importance of wettabilities in displacement processes, have prompted us to study the silanization of glass surfaces in comparison with silica surfaces. Our goal is to develop surface modification procedures that can be applied in an anodically bonded silicon-silica/glass micromodel, where both the glass and silica surfaces of a micromodel will be similarly oil-wet.
2. Experimental Methods
 In our experiments, the borosilicate glass was Pyrex 7740 obtained from Sensor Prep Services (Elburn, Ill.) and the silica surfaces were thermal oxide grown on silicon wafer material (Virginia Semiconductor Inc., Fredericksburg, Va.). Clean surfaces for silanization reactions on 3 cm × 1 cm rectangles of silicon were obtained with a series of water and organic solvent rinses prior to final cleaning by standard cleaning solution 1 for 30 min (SC1, a 70–80°C solution of five parts deionized (DI) water, 1 part 27% ammonium hydroxide, and 1 part 30% hydrogen peroxide) [Kern, 2008], UV-ozone [Kern, 2008] for 30 min using a Jelight model 342 UV-ozone cleaner, concentrated nitric acid (68%, boiling point 120°C, cleaning conditions specified in Tables 1 and 2), or aqua regia for 10 min. Surfaces cleaned with liquid reagents were thoroughly rinsed with deionized water. For silanization reactions, cleaned dry samples were put in an aluminum foil lined glass jar and the silane was drop cast over the surfaces. The jar was closed and placed in a 90°C oven for 15 h. Samples were solvent-rinsed with water, methanol, and hexanes after silanization and allowed to dry thoroughly before contact angle measurements. All contact angles were determined on a Ramé-Hart Model 500 Advanced Goniometer (Netcong, N. J.) as described in detail previously [Grate et al., 2012]. Static contact angles were stable within minutes for all surfaces except glass surfaces modified with silanes yielding hydrophobic interfaces. The latter sometimes required several minutes to reach stable contact angles. Measurements were typically made on three locations on each of three samples for nine total measurements per silane/material combination.
Table 1. Contact Angle Measurements on Silanized Silica and Glass Surfaces Using Silanes Intended to Yield Oil-Wet Surfacesa
3.1. Contact Angles of Silanized Glass and Silica Surfaces
 Using silanes previously identified [Grate et al., 2012] to yield oil-wet surfaces on silica, we found that silanization reactions producing hydrophobic oil-wet surfaces on silica do not necessarily yield oil-wet surfaces on glass. Moreover, this may not be apparent from air-water contact angles, which may be similar on the two materials; instead these differences are revealed by the oil-water contact angles. Examples collected from several experiments are shown in Table 1. On silica, silanization with hexamethyldisilazane (HMDS) gave air-water and oil-water contact angles of 97 and 140°, respectively, the latter of which is clearly in the oil-wet range. However, on glass cleaned with aqua regia prior to silanization we observed air-water and oil-water contact angles of 97 and 91°, respectively. The glass surface is not oil-wet by the oil-water contact angle measurement. In a repeat experiment on glass, cleaned with SC1, contact angles were 96 and 114°, respectively. Again, the silanized glass is not oil-wet, and again, the oil-water contact angle on the glass is quite different from that on silica (140°). On the other hand, glass cleaned in boiling concentrated nitric acid prior to silanization with HMDS had air-water and oil-water contact angles of 100 and 150°, respectively, which is clearly in the oil-wet range.
 Our rationale for using boiling concentrated nitric acid (boiling point 120°C) was that it would act as an oxidizer to thoroughly clean the surfaces, while simultaneously hydrating and cleaving surface siloxane linkages to create a surface with abundant surface silanols for reaction with silanes. In addition, alkali metal ions migrate to the surface of heated glass, and these ions can be exchanged out of the glass by acid treatment, leading to a surface enriched in silica [Elmer, 1980; Jang et al., 2001; Wright et al., 1980]. Wright et al.  noted that leaching glass with boiling 20% hydrochloric acid (∼110°C) resulted in surfaces that were nearly pure silica, while Jang et al.  noted that boiling glass surfaces in concentrated nitric acid reduced surface concentrations of sodium, calcium, and aluminum atoms.
 With dodecyltriethoxysilane (see Table 1), another silane that is expected to produce hydrophobic surfaces, results similar to those with HMDS were observed. Air-water contact angles on silica and glass were both 102°, while the oil-water contact angles were 148 and 110° on silica and glass, respectively (glass not cleaned in boiling nitric acid). Again, the glass surface was not oil-wet; it was quite different from the silica surface treated with the same silane, and the difference was only apparent in the oil-water contact angle. The oil-water contact angle appears to probe the surface differently than the air-water contact angle, at least for the glass surfaces.
 The results in Table 1 were collected from experiments on glass that has been prepared for silanization by cleaning and treatment methods, as we sought conditions to achieve oil-wet glass surfaces. Hot concentrated nitric acid was the most promising pretreatment method for glass to enable silanizations leading to oil-wet surfaces. Therefore, we set up a series of more directly comparable experiments using the same pretreatment methods on silica and glass surfaces followed by identical simultaneous silanization reactions with HMDS. In these experiments, we also investigated the effect of prior thermal treatments of 400°C. Micromodels are assembled by anodic bonding at 400°C, and therefore we needed to investigate silanizations on surfaces with similar prior thermal histories. Results are shown in Table 2. HMDS is a monofunctional silane with regard to reactions with surface silanols. It can only bond to the surface and not to other adjacent silanes, and thus creates a monolayer on the surface. Trichloro and trialkoxysilanes, by contrast, can bond to the surface and polymerize with one another, potentially leading to disordered silane multilayers. HMDS therefore represents a controlled test for silane without major risk of siloxane polymer formation on the surface.
 In the first experiments listed in Table 2, we examined whether boiling the concentrated nitric acid was necessary. Glass and silica surfaces, with no prior 400°C treatment, were soaked in concentrated nitric acid at room temperature for 1 week. After silanization with HMDS, air-water and oil-water contact angles on silica were 101 and 154°, respectively, while those on glass were 102 and 148°, respectively. This treatment gave oil-wet surfaces with similar contact angles on silica and glass, and the results were consistent with the prior boiling nitric acid treatment on glass with no prior 400°C treatment as seen in Table 1 (air-water and oil-water contact angles of 100 and 150° on glass). However, in a similar experiment where the glass was first baked at 400°C for 25 min before room temperature nitric acid treatment for 1 week, silanization with HMDS produced glass surfaces with air-water and oil-water contact angles of 95 and 102°, respectively. This glass surface is not oil-wet, and baking the glass at high temperature clearly did have an effect on the subsequent silanization and surface wettability results. Heating glass at high temperature dehydrates the glass with adjacent surface silanols condensing to form siloxane linkages. Hence, there are fewer surface silanols for subsequent modification by the hydrophobic silane.
 However, when glass with a history of baking at 400°C is subsequently treated with boiling concentrated nitric acid, and then silanized, HMDS yields oil-wet surfaces on both silica and glass. The air-water and oil-water contact angles on silica were 95 and 140°, respectively, while those on glass were 98 and 141°, respectively. These results with HMDS provide a method for obtaining similar oil-wet surfaces on silica and glass that have prior exposure to high temperatures such as those associated with anodic bonding of silicon-silica/glass micromodels.
3.2. Oil-Wet Micromodel
 Based on our studies with HMDS, we modified the interior surface of a silicon-silica/glass micromodel that had been assembled by anodic bonding at 400°C in our laboratories, using methods described in the supporting information. These micromodels have been described previously [Grate et al., 2012; Zhang et al., 2011a, 2011b]. Figure 1 compares the displacement of water by hexadecane in micromodels with water-wet and oil-wet surfaces. Displacements in water-wet micromodels were described in detail previously [Zhang et al., 2011a]. At low flow rates, displacement in the water-wet model is unstable and is characterized by capillary fingering with multipore blobs left undisplaced. By contrast, hexadecane displaced water from the oil-wet micromodel with ease, displacing the vast majority of the water with occasional (mostly) single-pore residual water left behind. The entry pressure for the hexadecane into the oil-wet pore network was only approximately 50 Pa, compared to ∼3500 Pa in water-wet pore networks with the same pore geometry. Figure 2 shows hexadecane and water in contact with each other and a fluid channel boundary in the entry section of the oil-wet micromodel. The wetting of the surface by hexadecane in preference to water is apparent, and stands in contrast with similar images previously published by Grate et al.  for water-wet micromodel surfaces. The oil wet surface was stable for at least several days as repeated displacement experiment were performed, with alcohol rinses to wash the model in between experiments.
 Our observations can be interpreted in terms of surface coverage by the silane, which varies depending on the initial surface chemical composition of the glass. In this interpretation, thermal histories and pretreatments leading to lower densities of surface silanols, and hence lower densities of silane residues after reaction with HMDS, would lead to intermediate-wet surfaces, while surfaces that yield higher densities of silane residues after silanization would lead to oil-wet surfaces. The oil-water contact angle is more sensitive to these surface coverage differences than the air-water contact angle. Some precedence for this interpretation can be found in the work by Menawat et al. , who treated soda-lime glass samples, previously cleaned in warm nitric acid for 20 min, with various monofunctional organosilanes. Silane concentrations in liquid-phase silanization reactions were varied, leading to differences in contact angles that were interpreted as differences in surface silane coverage. The oil-water advancing contact angles (oil used was xylene) showed large differences in contact angle with concentration, from 70 to 100° at 0.05 mg/L silane concentration in n-hexane, up to values approaching 160° at silane concentrations from 5 to 40 mg/mL. Variations in air-water contact angles were not so extreme. In one case, on increasing the t-butyldiphenylchlorosilane concentration from 0.05 to 40 mg/mL, the air-water contact angle varied from about 68 to 72°, while the oil-water contact angle varied from 100 to 155°.
 In summary, in this technical note, we have demonstrated that borosilicate glass and silica surfaces do not necessarily have the same wettabilities when silanized under the same conditions with the same silane. By finding pretreatment procedures that yield similar oil-wet surfaces on glass and silica after silanization, even after prior exposure to elevated temperatures (400°C), we have addressed a practical problem that is relevant to the surface modification of anodically bonded silicon-silica/glass micromodels. It is, in general, more difficult to obtain oil-wet surfaces on glass after silanization, especially glass with a history of prior high temperature treatment. In particular, we find that SC1 does not provide glass surfaces that yield oil-wet surfaces after silanization. However, treatment with boiling concentrated nitric acid does restore 400°C treated glass to a state where it can be silanized with HMDS to yield oil-wet surfaces similar to that of HMDS-silanized silica surfaces. Therefore, we generally recommend boiling with concentrated nitric acid as a treatment to prepare the interior surfaces of silicon-silica/glass micromodels prior to silanization reactions. To prepare oil-wet micromodels with HMDS, we specifically recommend this pretreatment, and we have demonstrated that it is feasible to do so.
 The Carbon Sequestration Initiative of the Laboratory Directed Research and Development Program at the Pacific Northwest National Laboratory (PNNL) supported this research. A portion of this research was carried out in the William R. Wiley Environmental Molecular Sciences Laboratory (EMSL), a national scientific user facility sponsored by the Department of Energy's Office of Biological and Environmental Research and located at PNNL. PNNL is a multiprogram national laboratory operated for the DOE by Battelle Memorial Institute.