The low salinity effect observed on sandstone model surfaces

Enhanced oil recovery has been documented in sandstone reservoirs where solutions of decreased salinity have been used for water flooding. Injected water with salinities (total dissolved solids) of 8000 ppm or less has increased oil production by incremental amounts ranging from 5 to 38% over conventional water flooding in some reservoirs. Explanations of the mechanism for the low salinity effect have proposed the presence of clays, polar compounds in the crude oil and the activity of divalent ions such as Ca2+ but there is no clear and consistent molecular level explanation for the effect. We used atomic force microscopy to investigate the fundamental properties of surfaces during the low salinity response and especially the role of illite clay. We designed model systems to reflect the minerals present in sandstone, namely illite attached to a flat SiO2 surface, in the form of an oxidized silicon wafer, freshly cleaved mica and an ozone cleaned SiO2 surface. We used tips that had been functionalized with hydrocarbons terminated either with carboxylate (-COOH) or methyl (-CH3), to simulate the polar and nonpolar components of crude oil. We used artificial sea water (ASW), 36,500 ppm total dissolved solids (TDS) for the high salinity solutions and diluted it to 1400 ppm TDS for the low salinity solutions. The adhesion maps derived from the interaction of the functionalized tips with the model surfaces show that reduced salinities have little or no effect on illite adhesion: there is only a 0-2% change. Adhesion on the fresh, clean muscovite surface varies considerably and the CH3 tip records a 10-20% decrease in response to low salinity solutions. In contrast, the carboxylate tip records a distinct response to salinity decrease, with a 20-30% drop. On a clean SiO2 surface, the low salinity response is evident, reproducible and repeatable through two cycles of high to low salinity changes. The range in decreased adhesion for these model mineral surfaces corresponds well with what is observed for core plug and field tests. (C) 2012 Elsevier B.V. All rights reserved.