A multinuclear solid state NMR study of the sol–gel formation of amorphous Nb₂O₅–SiO₂ materials

Abstract

Multinuclear ¹H, ¹³C, ¹⁷O, ²⁹Si MAS and ⁹³Nb static NMR is reported from a series of sol–gel prepared (Nb₂O₅)_x(SiO₂)_1−x materials with $x=0.03$, 0.075 or 0.30. ¹³C NMR shows that by 500 °C the organic precursor fragments have been removed although some residual carbon remains as a separate phase. The ²⁹Si NMR typically shows three Q-species (Q^2,3,4) in the initial gels, and that with increasing heat treatment the average n of the Qⁿ-species increases as the organic fragments and hydroxyl groups are removed. ¹⁷O shows unequivocally that the $x=0.03$ and 0.075 samples are not phase separated, while at the much higher niobia-content of $x=0.30$ Nb–O–Nb signals are readily detected, a definite indication of the atomic scale phase separation of Nb₂O₅. The $x=0.03$ and 0.075 samples heated to 750 °C are thus representative of amorphous niobium silicates. Comparison is made to other sol–gel prepared metal silicates especially with another Group Va metal tantalum. The effects of tantalum and niobium on the silica network are very different and it is suggested here that most of the niobium is present as NbO₄, forming part of the silicate network.

Introduction

The sol–gel preparation method [1] is being increasingly used to produce novel materials with interesting physical and chemical properties. Some of these materials are of great technological interest because they combine novel chemical properties with controllable porosity and large surface areas. Silica-based materials are some of the most important sol–gel formed materials, especially with the addition of secondary oxides (e.g. TiO₂, ZrO₂) which introduce different surface acid sites, making them ideal catalysts. Properties of MO_x–SiO₂ binary materials are highly dependent on the structure and dispersion of the two oxides. Recently the relatively little studied Nb₂O₅–SiO₂ binary system has been generating significant interest for a range of chemical processes including potential for heterogenous catalysis [2], [3]. The uses of this system include methanol oxidation, ethanol dehydrogenation, partial methane oxidation, selective adsorption and use as a biosensor after immobilising mediator species (see [4], [5] and references therein). Despite the chemical interest in this system there is still relatively little known about the structural basis of these properties. For the Nb₂O₅–SiO₂ system it is still unclear as to whether catalytic properties are related to surface niobium atoms that are actually in the silica framework, or a separate surface dispersed phase, or a mixture of both of these effects.

Solid state NMR is an obvious probe of the local structure of binary (MO_x–SiO₂) sol–gel produced oxides. For example the loss of organic fragments with heat treatment can be monitored by ¹³C, as well as ¹H, which can also determine the hydroxyl-content [6]. ²⁹Si is a popular probe of the connectivity of silicate networks for both crystalline and amorphous materials through the shift difference between the Qⁿ species (where n is the number of bridging oxygens on a given SiO₄ unit) [7]. ⁹³Nb is still a relatively little studied nucleus (nuclear spin $(I)=9/2$) even though it is 100% naturally abundant. The high spin and relatively high Larmor frequency means that the second-order quadrupolar broadening of the central transition should, for a site of the same distortion, be only ~28% broader than the commonly studied quadrupolar nucleus ²⁷Al. However ⁹³Nb resonances tend to be much broader than those typically observed for ²⁷Al. For example, the values reported for the ⁹³Nb quadrupole interactions from a range of niobates (Table 10.6 in Ref. [7]) are all ⩾13.6 MHz. This is an indication that polarisation effects of the inner shells of electrons are magnifying the electric field gradient at the nucleus in this heavier atom (the Sternheimer antishielding effect).

Oxygen is uniformly distributed throughout the structure in (MO_x–SiO₂) materials. Over the last decade there has been a large increase in the number of reports of ¹⁷O NMR from inorganic solids as it has been realised that oxygen can be an extremely informative nucleus [7]. It is an $I=\frac{5}{2}$ quadrupole nucleus, with a relatively small quadrupole moment and large chemical shift range ~1000 ppm in diamagnetic inorganic materials. Its major handicap is its natural abundance of only 0.037% which normally necessitates isotopic enrichment, but for many oxides this is relatively straightforward to achieve by using labelled water in a sol–gel process. For simple oxides ¹⁷O has been used to follow their sol–gel formation in a number of cases such as TiO₂ [8] and ZrO₂ [9]. The initial changes in structure in the sol–gel preparation of MO_x–SiO₂ materials have been followed by solution-state ¹⁷O NMR for SiO₂–TiO₂ and SiO₂–ZrO₂ [10]. Once formed and dried, the amorphous gel is a porous solid which continues to change its structure on heating. In (MO_x–SiO₂) gels a key question is the dispersion of the metal oxide within the silica framework, which can be monitored by ¹⁷O NMR measurements of the relative number of M–O–M, M–O–Si and Si–O–Si links. The large shift range of the ¹⁷O resonances allows all these different fragments in the TiO₂–SiO₂ system to be readily resolved in the 1D MAS NMR spectra [11]. ¹⁷O MAS NMR has been used to demonstrate the complete mixing of titanium with the silica [12]. The reactivity of the titanium alkoxide can be modified by complexing with acetylacetone to encourage a greater degree of Ti–O–Si crosslinking [13]. In addition to using acetylacetone atomic dispersion can be further improved by functionalising the precursor (e.g. R_xSi(OR)_4−x), with ¹⁷O NMR spectra showing up to five resonances [14]. For the ZrO₂–SiO₂ system zirconia acts as a network modifier, and it is retained in much higher concentrations in the silica phase than TiO₂. Both Zr–O–Zr and Zr–O–Si bonds are well separated in ¹⁷O NMR spectra and can clearly be observed [15]. A ¹⁷O NMR spectrum of a sample containing 10 mol% ZrO₂ showed, in addition to the main Si–O–Si peak, a Zr–O–Si resonance at −150 ppm. At higher zirconia concentrations phase separation occurs, but the silicate phase retains a significant level of zirconium-content as deduced through the intensity of the Zr–O–Si resonance [16].

Here the behaviour of Nb₂O₅ added to SiO₂ will be compared to that of another Group Va ion Ta⁵⁺. For Ta₂O₅–SiO₂ gels even at high Ta₂O₅ concentrations (~25 mol%) where significant phase separation occurs, appreciable Ta–O–Si bonding is still detected by ¹⁷O NMR [17]. NMR is used in this paper to examine where the niobium is located in such silica gel-produced samples (i.e. within the silica structure or as a separate phase) and how this varies with niobium concentration and processing temperature. The emphasis here differs from most previous structural studies of Nb₂O₅–SiO₂ materials as the ability of niobium to enter the silicate network is directly examined, rather than the formation of surface layers which has been the focus of most of the previous work e.g. [18].