Special data handling techniques for Fourier-transform infrared photo-acoustic spectra. Estimation and characterization of the N contents on ammoniated trichlorosilylated silica gel, using partial least-squares regression and curve fitting on the Si—H stretching vibration

Abstract

The reaction of silica gel with trichlorosilane (HSiCl₃) and subsequently with gaseous NH₃, resulting in a surface covered with Si—O—Si(H)(NH₂)₂ and NH₄Cl species, has been studied. The Si—H stretching vibration (2270 cm^–1) of the trichlorosilylated silica perturbs, upon ammonia adsorption, into three overlapping bands.

It is shown that partial least-squares regression (PLSR) is able to predict the concentration of the different surface species, using the Si—H stretching vibration region as the only spectral dataset. It is noticeable that this vibration has a constant intensity, since the Si—H species are generated during the trichlorosilylation step, and not during the ammoniation.

The mechanisms, causing this perturbation are investigated in more detail, using curve fitting and ²⁹Si cross-polarisation magic-angle-spinning (CP MAS) NMR.

The occurrence of an ammoniated secondary species [(Si—O)₂—Si(H)(NH₂)] on the silica surface is proposed and its infrared (IR) absorbing frequency is determined. An introduction to PLS for the spectroscopist is included.