Brine films remaining on mineral surfaces in deep reservoirs during CO2 sequestration are expected to influence multiphase flow, diffusion, and reactions, but little is known about their behavior. Using synchrotron X-ray fluorescence (XRF), we measured thicknesses of KCsI2 brine films on two difference roughness mica surfaces under conditions representative of geological CO2 sequestration (7.8 MPa and 40°C) to understand the influences of mineral surface roughness and capillary potential. Brine thicknesses measured on the Mica 1 (smooth) and Mica 2 (rough) mica surfaces ranged from 23 to 8 nm and 491 to 412 nm, respectively, over the small range of tested capillary potentials (0.18–3.7 kPa). Within these potentials, brine film thicknesses on mica were governed by surface roughness and only weakly influenced by capillary potentials. In comparing drainage and rewetting isotherms, some film thickness hysteresis was observed, possibly indicative of changes in mica wettability.
 Geologic carbon sequestration is a potentially important method for mitigating global warming from anthropogenic CO2 emissions, achieved by injecting CO2 (captured from large stationary sources) into deep geologic reservoirs [Raupach et al., 2007; Intergovernmental Panel on Climate Change, 2005]. Upon injection, reservoir pores become variably saturated with supercritical (sc) CO2 and the native fluid phase, typically brine, with their distribution dependent on capillary pressure and wetting history [Plug and Bruining, 2007; Pentland et al., 2011]. Among CO2 storage mechanisms, residual trapping immobilizes the nonwetting fluid (supercritical (sc) CO2) through capillary forces in reservoir pores [Hesse et al., 2008; Perrin and Benson, 2010; Szulczewski et al., 2012]. In multiphase flow, the coexistence of immiscible fluid phases in pore networks results in complex dependence of relative permeabilities and saturations on capillary pressure history. Some studies have concluded that thin (1–100 nm) adsorbed water films can exist on rock surfaces, depending on bulk properties, interfacial properties, and the disjoining pressure [Basu and Sharma, 1996; Buckley et al., 1989; Hirasaki, 1991; Tokunaga, 2012], and that rough surface microtopography supports thicker films [Tokunaga et al., 2000]. Thus, the determination of wetting film thicknesses on capillary (disjoining) pressure is important for understanding scCO2 behavior in reservoirs [Kim et al., 2012].
 Very few measurements of brine film thicknesses under geologic CO2 sequestration conditions have been reported, but studies on other immiscible fluid combinations are abundant. Ellipsometry and atomic force microscopy (AFM) have been applied for measuring water films on mineral surfaces in equilibrium with nonwetting fluids (air or oil) [Basu and Sharma, 1996; Gaebel et al., 2009; Gupta and Sharma, 1992]. To our knowledge, neither ellipsometry nor AFM techniques have been developed for use on films in scCO2. Moreover, ellipsometry cannot be applied on rough surfaces. Recently, Loring et al.  used in situ infrared spectroscopy to observe water films on forsterite (Mg2SiO4) in equilibrium with scCO2. They measured water film thicknesses (0.1–2 nm) with scCO2 at several water saturations, based on integration of the absorbance peak corresponding to HOH bending vibration in the scCO2 phase at 50°C and 18 MPa. However, their approach does not permit accurate control of capillary (disjoining) pressures at lower magnitude values associated with scCO2 entry into reservoir pores. Tokunaga et al.  developed an approach to measure brine film thicknesses on solid surfaces that relied on synchrotron X-ray fluorescence (XRF) of selenate ion dissolved in the film, and conducted experiments with air as the nonwetting phase. That approach was applied for measuring film thicknesses (0.5–10 µm) and hydraulic diffusivities on silica (root-mean-square roughness (Rrms), 9 µm) over a range of controlled potentials.
 Recently, we extended the X-ray-based approach to determine thicknesses of brine films on silica surfaces under confinement with scCO2 (7.8 MPa and 40°C), measuring synchrotron XRF of two tracers, I− and Cs+. The obtained area-averaged film thicknesses on the rough quartz surface ranged from 265 to 249 nm over controlled potentials (0.18–3.7 kPa). Film thicknesses on a smooth (1.6 nm Rrms) silica surface were about 2 nm over the same range of potentials. During the rewetting, the hysteresis of film thickness on silica surfaces was insignificant [Kim et al., 2012]. In the present study, scCO2-confined brine films were measured on another common mineral, mica, to further explore influences of capillary potentials and roughness.
2. Experimental Methods
2.1. Materials and Experimental Procedure
 Details of the experimental system and procedures were presented in the previous study [Kim et al., 2012]. Here, operational aspects specific to the experiments on mica are noted. Smooth muscovite mica (V1) discs (0.2 mm thickness) were purchased from Axim Mica. The smooth specimen used for film measurements was designated Mica 1. The rough mica window, Mica 2 was prepared from smooth mica discs by abrasion with microfinish diamond sandpaper (grain size 30 μm). The mica roughness was measured with an atomic force microscope (AFM, Veeco Multimode™) in contact mode (1 Hz scan rate, Np-10 tip) [Kim et al., 2012]. The roughnesses of samples were obtained after leveling with NanoScope IV (Digital Instruments) software, version 5.12. In addition, a larger scanning area of 720 µm × 540 µm was obtained by laser profilometry (Zygo New View6k). We used the average value from four regions on individual mineral surfaces for reporting roughness. The roughness of each window is presented in Figure 1 with respect to scanning lengths. The area of AFM scanning ranged from 10 µm × 10 µm to 150 µm × 150 µm (specific AFM images are shown in supporting information). Over the 100 µm × 100 µm scan area (similar in size to the synchrotron X-ray beam used in the film measurements), roughnesses of the smooth Mica 1 and roughened Mica 2 were 28 and 146 nm, respectively. Surface roughness was observed to increase with the measurement length scale, although larger-scale roughness on natural rock fracture surfaces increases to greater magnitudes [Brown and Scholz, 1985].
 Sapphire discs (1.8 mm thickness, purchased from Sapphire Jewel Company) were selected to support the mica samples because of their low attenuation of hard X-rays and their relatively high strength [Bai et al., 1996]. The mica samples were mounted onto supporting sapphire window, and sealed into the film chamber (mica surface on the interior side).
 Potassium iodide (99.999% purity) and Cesium iodide (99.999% purity) were purchased from Alfa Aesar. We prepared the brine solution (KCsI2) to contain Cs+ (1 M), K+ (1 M), and I− (2 M) for measuring Cs+ and I− in brine films through synchrotron XRF. The 99.99% pure CO2 gas was used to prepare scCO2.
 Experiments were conducted at beamline 11-2 of the Stanford Synchrotron Radiation Lightsource (SSRL), SLAC National Accelerator Laboratory. The main components of this system consist of an Isco syringe pump (Teledyne) for total pressure control, a preequilibration (CO2-brine) chamber (300 mL Parr reactor), a sight glass reservoir (capillary pressure control chamber), and a brine film chamber [Kim et al., 2012]. The schematic design and photograph of the film chamber are shown in supporting information. The system was kept at 40°C and 7.8 MPa (Isco pump) during experiment. Prior to pressurize, the entire system was flushed with CO2 at atmospheric pressure. Then, Isco pump delivered the brine to fill the entire system. The system was then filled with scCO2-saturated brine and pressurized to 7.8 MPa. Finally, brine-equilibrated scCO2 was infused from the preequilibration chamber to the brine chamber and capillary pressure control chamber to replace about half of the scCO2-saturated brine from the reservoir. The capillary pressure was controlled by adjusting the height of the brine-scCO2 interface in the capillary potential control chamber with a resolution of 10 Pa (∼1 mm), to as much as 3.7 kPa (350 mm) below the center of the film chamber [Kim et al., 2012]. A porous ceramic tube pressed against the inside window surface (mica) provided capillary continuity between the bulk brine (capillary pressure control chamber) and the brine layer on window, and allows film flow to occur symmetrically over a 4.6 mm radius.
2.2. Measurement of Brine Film With Synchrotron XRF
 Brine films were measured using synchrotron XRF of dissolved iodide in brine, measured with a Canberra 32-element Ge detector. The operating conditions used for this study at beamline 11-2 were a 37 keV monochromatic energy, 45° beam angle on sample, 100 (horizontal) µm × 100 (vertical) µm beam size, scan time 1 s on individual points, 0.25 mm step size, and 6 mm × 5 mm scanning area. The present brine film chamber (Figure 1) has a modified front flange and recessed bolts in order to minimize blocking of fluorescent X-rays; a limitation in our previous chamber design [Kim et al., 2012].
 The tracer fluorescence intensity is proportional to the thickness of brine on solid when solutes are nonreactive [Tokunaga et al., 2000; Kim et al., 2012]. Cs+ and I− spiked calibration filter paper standards for film thicknesses were described in the previous study [Kim et al., 2012]. The areal I− and Cs+ concentrations in calibration standards chosen ranged from 0 to 45 µmol m−2 and corresponded to equivalent brine thicknesses of 0, 4.5, 15, and 45 nm for the 1 M Cs+ and 2 M I− solution used in this study. As explained later, only XRF data from iodide measurements were used for final determinations of films on mica. We calibrated for individual sample windows to determine best correspondence with film measurements.
3. Results and Discussions
 The normalized calibration curve parameters are shown in Table 1. The slopes for I− and Cs+ fluorescence intensity on both Mica surfaces were measured with three different standard papers. The different calibration parameters obtained from Mica 1 (28 nm, Rrms) and Mica 2 (146 nm, Rrms) may reflect differences in geometric configuration and window thicknesses. We calculated the greater brine film thicknesses on Mica 2 through extrapolation of its calibrated range. During experiments with brine films, XRF maps were obtained for four elements (Ag, Mo, Cs, and I) for purposes of delineating suitable measurement regions and for determining film thicknesses. Figure 2 shows four elemental fluorescence maps of normalized Kα count rate obtained for the experiment on Mica 1 at 0.21 kPa (after 62 min of equilibration). The Ag and Mo maps indicate the locations of the front silver gasket/shield and stainless steel window frame, respectively. The Cs and I maps show the brine film on the mica surface in the plateau regions. The selected area for film measurements was consistently 2.25 mm2 area (1.5 mm × 1.5 mm) with the plateau for Cs and I (and minimum region of Mo and Ag) XRF. From our previous study, we found that equilibration of films between capillary pressure steps usually took around 1 h during drainage [Kim et al., 2012].
Table 1. Summary of Calibration Curves Obtained With Each Window (Y = aX + b)
The normalized count rate was obtained from the average value of ∑Ii/I0 within the 2.25 mm2 measurement area, where I0 is the ion chamber count within the data acquisition time and Ii is the count from the specific X-ray fluorescence peak within the same time interval in the ith detector channel. The Canberra 32-element Ge detector was operated with 32 channels.
Smooth Mica 1
0, 4.5, 15 nm
Rough Mica 2
0, 15, 45 nm
 Table 2 shows the obtained brine film on mica surfaces (smooth and rough) during drainage and rewetting processes with XRF intensity from the two tracer ions (Cs+ and I−). Calculated brine film thicknesses obtained from Cs+ and I− on smooth Mica 1 were significantly different (ranging from 0.1 to 7.5 nm), and largely attributable to Cs+ adsorption. A portion of the Cs+ associated with the films is involved in balancing mica's negative surface charge of about 3.5 µmol charge/m2 [Gaines et al., 1957; Osman and Suter, 2000], and amounts to the Cs+ contained in a f × (3.5 nm) thick film. Here, f is the fraction of cation exchange sites occupied by Cs+ (instead of K+). The selectivity for Cs+ is greater than for K+, but the extent of this selectivity for our synthetic mica is unknown. Nevertheless, the excess in equivalent film thickness attributable to Cs+ sorption onto cation exchange sites alone would be constrained between about 2 and 3.5 nm. However, Cs+ entry into mica edge interlayers also contributes to its accumulation on this substrate [Liu et al., 2003; McKinley et al., 2004]. Thus, the adsorption behavior of Cs+ on mica (more evident for Mica 1) was responsible for the greater apparent film thicknesses obtained with Cs+ relative to I−. The exception to this relation observed at 0.21 kPa capillary pressure is not understood. It should be mentioned that the electric double layer thickness for the 2 M ionic strength solution is much less than 1 nm, such that anion exclusion effects are insignificant. Overall, these considerations lead to the conclusion that the I− fluorescence provided more reliable measurements of film thicknesses.
Table 2. Film Thickness Measurements Obtained With Two Different Ions (Cs+ and I−), for Mica Surfaces During Drainage
Sample (Cal Number)
Capillary Pressure (kPa)
Brine Film Thickness/Standard Deviation (nm)
Difference (Cs+ − I−) (nm)
Mica 1 (Cal 1)
16.5 ± 0.3
23.3 ± 0.8
16.7 ± 0.4
11.4 ± 0.5
14.2 ± 0.3
10.3 ± 0.4
14.6 ± 0.4
8.7 ± 0.4
15.5 ± 0.6
8.0 ± 0.2
11.4 ± 0.3
4.1 ± 0.2
12.8 ± 0.3
9.5 ± 0.4
14.1 ± 0.4
9.1 ± 0.3
12.8 ± 0.3
12.9 ± 0.6
Mica 2 (Cal 2)
519 ± 13
491 ± 36
488 ± 16
478 ± 46
486 ± 12
472 ± 34
361 ± 9
442 ± 26
387 ± 12
412 ± 28
387 ± 14
448 ± 24
439 ± 10
425 ± 25
444 ± 8
457 ± 36
 Relative differences in film thicknesses measured from the tracer ions (Cs+ and I−) on the rough Mica 2 were small, ranging from 2 to 18% (see Table 2), because the portion of Cs+ required for charge balance in the much thicker films was small. The fairly consistent results obtained with the two tracers on Mica 2 reflect the dominating influence of capillarity over adsorption contributions on rough surfaces at low capillary pressures. In view of the uncertain excess Cs+ sorption on mica surface, we reported the final measured values of film thicknesses determined from only I−-based calculations on each surface.
 The measured brine film thicknesses for both the smooth Mica 1 and rough Mica 2 at different capillary potentials during the drainage and rewetting process are shown in Figure 3. The error bars describe twice of the standard deviations among the individual points within the film measurement region. As capillary potential increases, the scCO2 pressure increases relative to brine's pressure and represents scCO2 invasion into the originally brine-filled reservoir pore.
 The Mica 2 (rough) maintained thicker films, decreasing from 491 to 412 nm with the increase of capillary potential. The brine thickness on the Mica 1 (smooth) was also decreased from 23 to 4 nm over the same range of capillary potentials. These results of brine film thicknesses are identical with previous findings of roughness effects, with rougher surfaces providing greater capillary contributions to the overall film thicknesses on silica samples [Kim et al., 2012; Tokunaga et al., 2000]. In topographic depressions on mineral surfaces, the equilibrium menisci are governed by equilibration of adsorptive and capillary energies [Philip, 1978; Shahraeeni and Or, 2010]. Therefore, the capillary contribution to film thickness is the dominant factor on rougher surfaces near zero capillary potentials associated with large radii of curvature because of greater retained water mass in deeper channels. In addition, the effect of changing capillary pressure over a small range (∼3.7 kPa) on film thickness was small, and the variances of measurements were similar to changes in brine film thickness. Nevertheless, brine film thicknesses at the largest controlled capillary pressure (3.7 kPa) were clearly less than the thicknesses at the smallest controlled capillary pressures.
 The drainage and rewetting isotherms show some hysteresis on both mica surfaces, in contrast to indistinct hysteresis on silica [Kim et al., 2012]. The observed film thickness hysteresis on mica may be partly related to changing wettability. Observations of contact angle hysteresis on mica surfaces under CO2 sequestration conditions have been reported by others [Broseta et al., 2012; Chiquet et al., 2007], and reaction between scCO2 and silanol groups in silica have been proposed as explanations for reduced wettability [Dickson et al., 2006; Kim et al., 2012]. Mica surface reactions probably influenced our measurements, but are beyond the scope of this study.
 The possibility that true hydrostatic equilibrium, however, was not achieved at each step may have influenced results in this study. Comparing film thicknesses on both rough mica (Rrms: 146 nm) and silica (Rrms: 330 nm, previous study [Kim et al., 2012]), we observed thicker brine films on mica. This is consistent with mica having a higher surface charge density and being more hydrophilic than the nearly uncharged silica under geologic CO2 sequestration conditions. Thicker aqueous phase films may enhance reactions between brine and mica surfaces.
 Figure 3 includes a comparison between the experimental results of brine thicknesses on the smoother Mica 1 surface with a Derjaguin-Landau-Verwey-Overbeek (DLVO) model for adsorbed layers. Tokunaga  calculated the brine films on silica with scCO2 as the confining fluid with DLVO model. The brine film thickness f dependence on disjoining pressure Π can be expressed as the sum of contributions from van der Waals dispersion (VDW), ΠvdW(f), and electrostatic interactions (EL) Πel(f) [Tokunaga, 2012]. The brine thickness from VDW interactions is
where A132 represents the Hamaker constant for the interactions between (1) mineral surface, (3) aqueous film, and (2) CO2. The A132 in this study can be calculated to be −1.8 × 10−20 J for mica-H2O-CO2 using equation (2). Hamaker constants for CO2 (A22, 4.63 × 10−21 J) and water (A33, 3.74 × 10−20 J) at 313 K and 7.8 MPa were estimated with the Lifshitz equation [Israelachvilli, 1991] based on their relative permittivities ( : 1.153 [Moriyoshi et al., 1993] and : 73.20 [Bradley and Pitzer, 1979]) and their refractive indices ( : 1.062 [Sun et al., 2003] and : 1.332 [Harvey et al., 1998]) under the experimental conditions. The Hamaker constant for mica (A11) was estimated to be equal to 1.355 ×10−19 J, the value reported for room temperature and pressure by Hough and White . The Πel in a symmetric electrolyte solution between parallel charged plates was related to f with the Gregory compression approximation [Gregory, 1975] and was recently applied to predict scCO2-confined film thicknesses on silica in Tokunaga  and Kim et al. . The electrostatic contribution to film thickness is very small (<1 nm) because of the brine's high ionic strength (2.0 M), regardless of the assumed electrostatic potential at the mica surface. Therefore, the DLVO-predicted film thicknesses are largely associated with dispersion (van der Waals) interactions in the mica-brine-scCO2 system. Note that the measured water film thicknesses on the smooth mica surface (Rrms: 28 nm) are slightly greater than the DLVO-based predictions in Figure 3. It should be noted that the calculated film thicknesses from the DLVO model are less accurate at high ionic strength and that surface roughness probably enhanced the measured “smooth” surface film thickness values.
 Measurements of brine film thicknesses on mica were conducted to investigate the influences of surface roughness and capillary pressure under geologic CO2 sequestration conditions (7.8 MPa and 40°C). Thicker films were retained on the rough surface relative to the smooth surface; as expected based on the importance of capillary film retention on rough surfaces at low capillary potentials. The observed hysteresis in drainage and rewetting isotherm may reflect wettability changes on mica. The measurements of film thicknesses on the smoother mica surface were found to be in fair agreement with predictions based on a DLVO-based model. Further studies on brine films wetting different mineral surfaces are needed over wider ranges of capillary potentials to understand film behavior under geologic CO2 sequestration.
 This research is supported as part of the Center for Nanoscale Control of Geologic CO2, an Energy Frontier Research Center funded by the U.S. Department of Energy (DOE), Office of Science. This research was carried out under U.S. DOE contract DE-AC02–05CH11231. Portions of this study were carried out at the Stanford Synchrotron Radiation Lightsource (SSRL), a Directorate of SLAC National Accelerator Laboratory and an Office of Science User Facility operated for the U.S. DOE, Office of Science by Stanford University. The SSRL Structural Molecular Biology Program is supported by the DOE Office of Biological and Environmental Research, and by the National Institutes of Health, National Institute of General Medical Sciences (including P41GM103393) and the National Center for Research Resources (P41RR001209). In addition, we thank Kevin Knauss for use of the AFM, and the anonymous reviewers for their valuable comments.