Nanoscale composition inhomogeneity in silica-aluminas prepared by various methods, страница 6

range, as indicated for samples of Fig. 4(d) and

Fig. 5(b)) known for the industrial ones. The meaning

of A is that of a Si/Al weight concentration ratio that

should ideally be measured at the nanoscale level on

all the explored points of an ideally homogeneous

specimen having this composition. B is simply the

arithmetic average of the experimentally measured

values of the Si/Al weight concentration ratio on all

the analysed points for each specimen. It is evident

that extreme (inÆnite) values of the Si/Al ratio sometimes

recorded (as those shown in Fig. 3(d) and

Fig. 4(a)) could not be mathematically taken into

consideration for the calculation of B.

Impurities were detected in many of the investigated

specimens, both in laboratory and in industrial

ones, as shown in Table 2. In some samples, these

were present in relatively high quantities. The impurities

were very inhomogeneously distributed among

the analysed points. Nano-size domains contained

elements other than Si and Al, whereas other ones

were free from impurities. This left little hope to

quantify impurity concentration, either chemically

or through EDS, because this made it impossible to

relate the corresponding values to the non-uniformity

of the Si±Al ratio.

It is striking that particles composed of pure silica

were found in the industrial samples and that parts of

the laboratory samples were very rich in silica. Most of

the silica±aluminas are extremely heterogeneous at

the scale of our measurements. Exceptions are the

laboratory samples of Fig. 3(c) and Fig. 4(c). Nevertheless,

relative Øuctuations of10% to20% around

the average values were observed for these samples

(5<Si/Al<7 for the former case and 2.5<Si/Al<2.7 for

the latter, when neglecting a few more diverging

points).

4. Discussion

The heterogeneity of composition of all silica±

aluminas at the nanometer level is disappointingly

high. This is true even for samples prepared in small

batches under intense agitation and in well-controlled

conditions. These samples are possibly slightly more

homogeneous than the industrial ones, but it is very

doubtful that this would be sufficient for ensuring a

satisfactory control of the uniformity of acidity. The

conclusion is that all the investigated samples contain

practically all the catalytic sites detected in the measurements

reported in the introduction [1±6]. Only the

proportion of these sites is different. The presence of

inhomogeneously distributed impurities could possibly

enhance this effect. The industrial catalysts generally contain less impurities than the laboratory prepared ones. As indicated in the introduction, the ammonia TPD

measurements made on some of these samples did not

allow any clear correlation with the degree of inhomogeneity.

They just show the general trends widely

described in literature, or which can be deduced from

the results of other more elaborate techniques used for

the identification and quantification of acid sites. This

is understandable, because four main parameters are

varying in our set of just 20 samples: overall composition,

heterogeneity, and nature and amount of impurities.

A statistical study should involve sets of about

10 samples with the same overall Al, Si and impurity

content, to be significant.

There is obviously a need for more elaborate

approaches to preparation of silica±aluminas with

uniform acidity. In this we agree with an excellent

article published recently, which addressed the problem

in a very broad way [17]. Very few investigators

searched for rationally designed methods to control

the hydrolysis of the starting alkoxides of silicon and

aluminium so that both react simultaneously in the

right proportions. In principle, this could be achieved

by selecting carefully the nature of the alkoxy groups

of silicon and aluminiumto ``tune'' their reactivity and

by controlling all the reaction parameters (solvent

used, temperature, rate of water addition, agitation

etc.). On the other hand, the extremely large number of

studies devoted to the sol±gel preparation of silica±

aluminas may have nearly exhausted the possibilities