Nanoscale composition inhomogeneity in silica-aluminas prepared by various methods, страница 2

   The aim of our present work was to investigate more finely the situation by measuring the composition at a smaller scale, i.e. at the nanoscale rather than micrometer level. As ammonia temperature programmed desorption (TPD) did not show clear correlations with the EDS data, the present article had to be restricted to the latter. This aspect will be shortly discussed later.

For the present study, we used two series of sol-gel silica-alumina catalysts, prepared in different ways: (i) laboratory prepared samples and (ii) silica-aluminas prepared in industry according to ``sol-gel techniques'', some of which are probably very different from the original alkoxide-based method. A large range of compositions was covered in both cases. One of our aims was to compare the amplitude of the local Si/Al ratio variation for specimens of similar global chemical compositions, but resulting from two different preparation procedures. To our knowledge,

such an investigation has not yet been performed. In general, there is a strong feeling that, in spite of the wide industrial application of the sol-gel processing method, there is still a lack of data concerning the detailed properties of the final products when they contain two or several metallic elements [9]. This is a general remark, not limited to silica-aluminas. The aim of the work was not to rank samples according to their homogeneity. The commercial interests are too important for a university laboratory to influence them. For this reason, the industrial samples are not identified in this work. Another precaution we took is to report only data on samples

produced before 1991.

2. Experimental

2.1. The samples

Two kinds of sol±gel silica±alumina powders have

been investigated: (i) samples selected from the complete

series of silica±aluminas carefully prepared in

our laboratory starting from alkoxides (see [1,2]) and

(ii) a number of industrial products prepared by

``industrial sol±gel methods'', very often based on

other precursors than alkoxides. Our laboratory prepared

materials [6] were made from distillation-puriÆed

aluminium isopropylate and ethyl silicate, which

were introduced progressively into a 1 N solution of

acetic acid (10 ml/g oxide to produce). The mixture

was reØuxed at 808C for 6 h, then kept overnight at

room temperature. After addition of distilled water

(1 ml/g oxide), this was reØuxed again for 8 h under

vigorous stirring. Stirring was prolonged for 1 h. The

next day, the pH was raised progressively to 7 by an

ammonia solution and kept at that value for 1.5 h. The

material became viscous but stirring was maintained.

This was followed by reØuxing for 2 h, centrifugation

and washing three times (3 ml water/g oxide). The

recovered solid was heated (heating rate: 308C/min) to

5008C and maintained at this temperature for 24 h.

The global chemical composition of the laboratory

prepared powders is accurately known, whereas the

global composition of the industrial specimens, as

indicated by the producer, probably Øuctuates somewhat

during industrial production. Sometimes the

indicated silica/alumina global composition of the

industrial samples corresponds to a large range rather

than to a precise value.

Table 1 shows the global chemical composition of

the investigated samples. Comparable samples are

shown on the same row, but the chemical compositions

may not be exactly the same. Comparison is sometimes

possible with more than one sample for a given

silica/alumina proportion.

2.2. The microanalysis measurements

The specimens for X-ray microanalysis were prepared

by grinding the initial sol±gel silica±alumina

powder, ultrasonically dispersing the resulting powder

in water and depositing the suspension on electron

microscopy grids covered with a continuous carbon

layer, according to the conventional ``droplet'' technique.

The microanalysis was performed in a JEM-

100-C-TEMSCAN analytical electron microscope

(JEOL) equipped with a KEVEX-5100 X-ray EDS

spectrometer. The accelerating voltage was always

100 kV, our constant preoccupation being to accurately

match the thin Ælm conditions required for