『Abstract
　For two suites of volcanic aluminosilicate glasses, the accessible and reactive sites for covalent attachment of the fluorine-containing (3,3,3-trifluoropropyl) dimethylchlorosilane (TFS) probe molecule were measured by quantitative ¹⁹F nuclear magnetic resonance (NMR) spectroscopy. The first set of samples consists of six rhyolitic and dacitic glasses originating from volcanic activity in Iceland and one rhyolitic glass from the Bishop Tuff, CA. Due to differences in the reactive species present on the surfaces of these glasses, variations in the rate of acid-mediated dissolution (pH 4) for samples in this suite cannot be explained by variations in geometric or BET-measured surface area. In contrast, the rates scale directly with the surface density of TFS-reactive sites as measured by solid-state NMR. These data are consistent with the inference that the TFS-reactive M-OH species on the glass surface, which are known to be non-hydrogen-bonded Q³ groups, represent loci accessible to and affected by proton-mediated dissolution. The second suite of samples, originating from a chronosequence in Kozushima, Japan, is comprised of four rhyolites that have been weathered for 1.1, 1.8, 26, and 52 ka. The number of TFS-reactive sites per gram increases with duration of weathering in the laboratory for the “Icelandic” samples and with duration of field weathering for both “Icelandic” and Japanese samples. One hypothesis is consistent with these and published modeling, laboratory, and field observations: over short timescales, dissolution is controlled by fast-dissolving sites, but over long timescales, dissolution is controlled by slow-dissolving sites, the surface density of which is proportional to the number of TFS-reactive Q³ sites. These latter sites are not part of a hydrogen-bonded network on the surface of the glasses, and measurement of their surface site density allows predictions of trends in reactive surface area. The TFS treatment method, which is easily monitored by quantitative ¹⁹F solid-state NMR, therefore provides a chemically specific and quantifiable proxy to understand the nature of how sites on dissolving silicates control dissolution. Furthermore, ²⁷Al NMR techniques are shown here to be useful in identifying clays on the glass surfaces, and these methods are therefore effective for quantifying concentrations of weathering impurities. Our interpretations offer a testable hypothesis for the mechanism of proton-promoted dissolution for low-iron aluminosilicate minerals and glasses and suggest that future investigations of reactive surfaces with high-sensitivity NMR techniques are warranted.』

1. Introduction
2. Methods
　2.1. Samples
　2.2. Sample preparation
　2.3. Surface modification
　2.4. Nuclear magnetic resonance
3. Results
　3.1. Observations from the “Icelandic” suite (including Bishhop Tuff)
　3.2. Observations from the Japanese chronosequence suite
4. Discussion
5. Conclusions
Acknowledgments
Appendix I. Clay reactivity to silanization
References