DETERMINATION OF OXIDE SURFACE REACTIVITY VIA SOLID-STATE NUCLEAR MAGNETIC RESONANCE AND AB INITIO METHODS

Open Access
Author:
Washton, Nancy Maurer
Graduate Program:
Chemistry
Degree:
Doctor of Philosophy
Document Type:
Dissertation
Date of Defense:
May 21, 2007
Committee Members:
  • Karl Todd Mueller, Committee Chair
  • Barbara Jane Garrison, Committee Member
  • James David Kubicki, Committee Member
  • John B Asbury, Committee Member
  • Susan Louise Brantley, Committee Member
Keywords:
  • chlorosilane
  • surface modification
  • reaction pathways
  • ab initio
  • aluminosilicates
  • reactive surface area
  • NMR
Abstract:
Although a mature field, the study of oxide surface speciation and reactivity is still open to scientific investigations endeavoring to elucidate the chemical species involved in various forms of reactivity. Silicates, aluminosilicates, and aluminas are unarguably the most important oxide materials for industrial uses, in addition to comprising a large percentage of environmentally important oxides, e.g., quartz, feldspars, clays, gibbsite, etc. We have used solid-state Nuclear Magnetic Resonance (NMR) and ab initio computational methods to probe the surface reactivity of several [alumino]silicate materials. Cohesion for this work is provided by studying the effects of chemically modifying oxide surfaces with a monochlorosilane, either experimentally through the use of NMR, or computationally, through the use of Density Functional Theory (DFT) investigations of the reaction pathways. For two suites of natural aluminosilicate glasses, the number of accessible and reactive sites for covalent attachment of a fluorine containing chlorosilane probe molecule was determined 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. For this suite of samples the rate of acid-mediated dissolution (pH approximately 4) does not correlate to geometric surface area or BET measured surface area, but does depend directly on the number of reactive sites as measured by solid-state NMR. These data imply that the M-OH species bonded to (3,3,3-trifluoropropyl)dimethyl-chlorosilane (TFS) represent loci accessible to proton-mediated dissolution. The chemisorption site for TFS on silicates is known to be non-hydrogen bonded Q3 groups, and therefore these sites are implicated as regions that control or influence the rate of acid-mediated dissolution. The second suite of rhyolites, from Kozushima Japan, constitutes a chronosequence, and is comprised of four rhyolites that have weathered for 1.1, 1.8, 26, and 52 thousand years. For both suites of samples the number of reactive sites on a per gram basis is shown to increase as a function of laboratory and field weathering time. These new data provide a chemically specific and quantifiable proxy for reactive surface area, and our interpretations offer a testable hypothesis for the mechanism of acid-mediated dissolution for aluminosilicate minerals and glasses. The implication that the chlorosilane binding site on aluminosilicate glasses was potentially the rate-determining site for dissolution led to an investigation of the surface species involved in chlorosilane binding. The low surface area of the volcanic glasses precluded their use as samples for detailed surface characterization via NMR, and therefore a set of six [alumino]silicate gels with high surface area were used as models for the natural glasses. This set of six samples included pure silica gel, pure mesoporous alumina, and four aluminosilicate samples with different weight percent Al2O3, ranging from 11 to 58 percent. The gels were modified with 3,3,3-(trifluoropropyl)dimethylchlorosilane (TFS) followed by a quantitative and qualitative assessment of the binding sites via NMR. The number of TFS binding sites as determined by 19F direct polarization NMR decreased as Al2O3 weight percent increased in the samples, which is posited to be a function of the increasing amount of AlVI relative to AlIV. As the proportion of AlVI rises in the sample, phase separated aggregates composed of pure alumina begin to dominate the aluminum speciation, which we hypothesize is less reactive to TFS binding than the AlIV sites. The Cross-Polarization Carr-Purcell-Meiboom-Gill (CP-CPMG) pulse sequence was used to determine the identity of the surface sites involved in TFS binding. The center of mass in the spikelet pattern was assigned to a specific spikelet and taken to represent the two end members, silica gel (12 ppm) and mesoporous alumina (8 ppm). The spikelet pattern of the mixed aluminosilicate gels clearly indicates a center of mass shift to lower frequencies with a concomitant change in spikelet intensity. Ratios of the 12 to 10 ppm spikelets were taken from the measured intensities and compared to ratios predicted by a binomial probability function weighted by the probability of a given Qn silicon site being present on the surface. The model compares very well with experiment, and may be a useful tool for predicting the reactive surface species on aluminosilicates. Our predictive model does not address the etiology of the differences exhibited by these complex materials, and cannot due to its qualitative nature. To-date, experimental methods are incapable of generating a molecular level picture of changes to surface reactivity accompanying isomorphic substitution at the silicon center. However, computational methods are particularly well suited to this endeavor, and have been utilized here in a preliminary investigation of chlorosilane binding to silica surfaces. The gas-phase reaction pathways for condensation between mono-functional organosilanols (TFS and Trimethylsilanol (TMS)) and Q3Si or Q2Si sites have been investigated with DFT. The condensation is known to occur via an SNi mechanism, which is a bimolecular nucleophilic substitution that retains the stereochemistry at the metal center. Three condensation mechanisms were studied for each QnSi species: the neat reaction (i.e., no solvent or additional molecules), an HCl inclusive reaction, and an HCl/ H2O inclusive reaction. The calculated activation energies were higher than predicted for all reactions with the exception of the Q2Si/HCl pathway having a calculated value of 118 kJ/mol. Analysis of the calculated values for Grxn indicate that all of the reaction pathways containing Q3Si species are favorable, but for the Q2Si species only the HCl inclusive pathway is favorable. Addition of solvent effects via the polarizable continuum model (PCM) to the Grxn calculations increases the loss in free energy upon condensation for the Q3Si, while the opposite is true for the Q2Si pathways. Based on these data the Q2Si is the most energetically favorable site for chlorosilane condensation. However, this is inconsistent with the experimental findings that the Q3Si is the dominant species involved in condensation with mono-functional organosilanols. The origin of this discrepancy lies in the reaction mechanism and the chemical species participating in it. Chlorosilane condensation does not occur at temperatures below 300 oC in the absence of molecular water, and therefore the Q2Si/HCl reaction pathway is not physically feasible. Our data suggests that the Q2Si is the favored participant in the absence of water, which is not supported by experimental findings. Furthermore, these data imply that water is a necessary participant in the condensation step, a fact that has been previously suggested but not confidently based on experimental or computational evidence. In addition, once explicit water is included in the reaction pathways the Q3Si becomes the thermodynamically favored participant in the reaction, which is consistent with experimental findings.