Platinum Metals Rev., 1980, 24, (1), 14
The Characterisation of Catalysts
A Survey of Present Industrial Practice
This paper is based upon an introductory address given by Dr. Acres to a residential course on Characterisation of Catalysts organised by Professor J. M. Thomas and Dr. R. M. Lambert for representatives from industry, government research centres and universities and held last year in King’s College, Cambridge.
During the past ten years substantial progress has been made in the development of techniques for the characterisation of heterogeneous catalysts. The advances that have been made in physical techniques that can be applied to the real catalysts used in everyday industrial processes are of particular significance as, if properly applied, they can practically eliminate the worst features of recipe-based industrial catalytic processes.
These physical techniques can be applied to the four stages that make up an industrial catalytic process, namely process design, catalyst production and scale-up, and process trouble shooting.
The Techniques of Physical Characterisation
The principal physical characterisation techniques used by industry are listed in Table I. Generally, they enable fast evaluations to be made of the wide range of catalyst and process variables frequently encountered in industrial situations.
A major factor in the success of the use of physical techniques in industry is the combination of the results obtained from a number of techniques, such as, SEM, N2 Pore Size Distribution and Surface Area, TEM, X-ray, ESCA and Metal Area by Gas Adsorption, in order to build up a full picture of the physical characteristics of a catalyst.
Physical parameters allied with activity-selectivity measurement are the basic tools of catalyst design and process development. Catalyst design is obviously related to process development, and the former usually precedes the latter. The major feature of catalyst design is the utilisation of metal and support chemistry to vary the dispersion, distribution and degree of reduction of the catalytic centres, while process development is concerned with making use of reaction chemistry, reactor operation and catalyst design, to maximise the amount of product formed and, if possible, to suppress side reactions.
The fat hardening reaction chosen to illustrate the interaction of process development and physical characterisation requires the reduction of the trienoate component, linolenate, to below 2 per cent while retaining the total dienoate plus oleate at levels above 82 per cent. At the same time the catalyst must not hydrogenate any of the components to stearate or isomerise the unsaturated bonds to the trans configuration. The normal percentage composition of a liquid vegetable oil such as soya-bean oil is: palmitate 10, stearate 4, oleate 25, linoleate 53, linolenate 7, with an iodine value (I.V.) 135.
This hydrogenation process currently uses a nickel catalyst which will only be displaced if the process becomes more cost effective. This will result from a combination of factors, including greater activity at a lower temperature and more frequent recycling before the catalyst becomes spent.
A full consideration of the process using nickel suggests that for a palladium catalyst to be adopted it would have to operate at 80 to 120°C with a hydrogenation rate of at least 6 ml of hydrogen per minute per mg of palladium, and recycle more than four times. As catalysis is a surface phenomenon, it would be expected that maximum activity will be obtained by choosing the catalyst with the highest metal area. This may be obtained either by increasing the total amount of active metal, that is by increasing metal loading, or by modifying the method of preparation so that the metal is more completely dispersed. Figure 1 shows the effect of increasing metal loading for two methods of preparation, designated Type 1 and Type 3. From Figure 1 it would be expected that Type 3 catalyst will show the highest activity, approximately three times that of the Type 1 catalyst.
However, when the two catalysts are tested against soyabean oil it is found that the contrary activity pattern is obtained, as shown in Figure 2. Although the pattern in Figure 2 would appear to be contrary to the original postulate that the activity of a catalyst should be proportional to its metal area, closer examination shows that the postulate is correct for, with each type of catalyst, activity is proportional to metal area. What has to be explained is how Type 1 catalyst has a higher activity per unit of metal area (24.5 ml H2/min/m2), compared to Type 3 catalyst (2.6 ml H2/min/m2), and how Type 3 catalyst increases in activity per unit area of metal when the metal area is almost constant. At the point of activity increase the overall rate of catalyst activity is 3.8 ml H2/min/m2, but the slope of the tangent to the curve is 25.6 ml/min/m2, very close to that for Type 1 catalyst. This strongly suggests that the mechanism responsible for the high activity of Type 1 catalyst is also responsible for the sharp increase in activity of Type 3 catalyst at high metal loadings.
Transmission electron micrographs of the two catalysts are shown in Figures 3 and 4 where it will be noted that the metal is located differently, giving rise to a large in-pore diffusional resistance to hydrogen and triglyceride molecules with Type 3 catalyst but a negligible resistance to diffusion for Type 1 catalyst. From this an increase in the amount of metal on the exterior of the Type 3 catalyst should be seen where the sudden increase in activity occurs. Using ESCA and measuring the total palladium 3d electron flux for the catalysts it is possible to relate catalyst activity to the exposed palladium metal area and to see how this differs from the total palladium metal area. Figure 5 shows that the general metal distribution shown by TEM is valid; it is apparent from Figure 6 that the curve for the Type 3 catalyst shows a considerable increase in ESCA signal for a small metal area increase, hence it can be concluded that this metal is located on the outside of the carbon particle. Superficial metal area alone is responsible for activity in this reaction and a linear relationship exists for catalyst activity versus ESCA Pd3d electron counts, Figure 7 shows such a relationship.
The activity pattern shown by the two catalysts can now be seen primarily as a function of metal placement and secondarily as a function of metal dispersion. Both catalysts fulfil the requirement of having an activity which is greater than 6 ml H2/min/mg Pd, Type 1 catalyst is 132 ml H2/min/mg Pd and Type 3 catalyst is 18 ml H2/min/mg Pd, both at 5 per cent palladium loading and 180°C, while for 120°C the figures are 108 ml H2/min/mg Pd for Type 1 catalyst and 8 ml H2/min/mg Pd for Type 3 catalyst.
The physical picture that has emerged for the activity of two catalysts is equally important to selectivity. Where metal is held within the pore structure large diffusional resistances to the reactants and products are to be expected. Hydrogen, however, being a smaller molecule and being consumed in the reaction, faces a considerably lower diffusion problem. The residence time of a triglyceride molecule within the pore structure is high compared to that for the outside surface of the particle, and the types of catalyst should show differences in the distribution of fatty acids when used to hydrogenate a soyabean oil. With longer residence times an increased degree of saturation can be expected, together with high trans acid concentrations. Thus Type 3 catalyst would be expected to have a higher stearate/oleate fraction than the Type 1 catalyst and a higher trans acid concentration. The predominence of mass transport limitation should also blur the high selectivity of palladium for the hydrogenation of the trienoate (linolenate).
Table II shows the product distribution for the two catalyst types at temperatures of 80 to 160°C. All three results obtained with Type 1 catalysts have a low linolenate level of about 2 per cent, but Type 3 catalyst is higher at 3 to 4 per cent. For comparable iodine values the Type 1 catalyst also shows a lower stearate level and a higher linoleate level. From the overall physical picture of the Type 3 catalyst, and the product distribution shown in Table II, it can be seen that no amount of further research work will improve it for this reaction; however, Type 1 catalyst is now amenable to fine tuning using the process operating conditions to modify its catalytic behaviour significantly.
|Catalyst||Temperature °C||Product composition* per cent||lodine value (I.V.)|
Three process operating variables can be modified, these are the amount of catalyst used, the temperature of operation and the agitation.
Table III shows the results of such experimentation. At constant temperature, increasing the mass transfer by decreasing agitator speed decreases the selectivity. However, by lowering temperature by 20 per cent, metal loading by 20 per cent and agitator speed by 15 per cent, conditions of near perfect mass transfer are achieved. This result shows what can be obtained in the area of process development by a proper understanding of the physical nature of the catalyst allied with an appreciation of the operating parameters and how they affect mass transfer.
An industrial catalyst has to be optimised to obtain the best activity, selectivity and life for a given process. Suitable physical techniques are chosen specifically to follow the change in the catalyst performance with formulation. Furthermore the industrial catalyst has, of necessity, to be studied in the form in which it is used, under conditions relevant to the process, while measurements have to be obtained in a relatively short period of time. For these reasons, conventional physical techniques for the characterisation of catalysts may have to be substantially altered. Two examples relating to car exhaust emission control catalysts have been chosen to illustrate these points.
Current three-way catalyst systems (T.W.C.) used for controlling exhaust emissions from motor vehicles are capable of removing carbon monoxide, hydrocarbons and nitrogen oxides in a single unit. The reactions involved are highlighted in Table IV.
The Three-Way Catalyst
Physical characterisation has played a major part in the development of this technology, as illustrated by the following two examples.
Metal Area Determination
Efficient dispersion of a supported noble metal catalyst to maximise the available active surface for incoming reactants is a very important requirement. Although gas adsorption techniques now exist for determining the metal area of supported metal catalysts these are not applicable to industrial catalysts. To illustrate this, the measurement of rhodium metal areas is now described.
Rhodium is an essential component of T.W.C. systems and one of its functions is to promote the reduction of nitrogen oxides. The metal is effective at low concentrations, which in turn is a function of its dispersion or metal area. In order to measure the adsorption of nitric oxide at low rhodium dispersions a modified gas adsorption system was developed, in which the detector employed was a quadrupole mass spectrometer. The amount of gas adsorbed against time is determined and a comparative metal area calculated. The sensitivity of the equipment and its ability to handle corrosive gases are important considerations.
The chemisorption of nitric oxide on two rhodium catalysts having significantly different metal dispersions is shown in Figure 8. The effect of metal loading on rhodium dispersion is also shown. The improvement in activity resulting from the better dispersion of rhodium via preparation A is adequately shown by steady state engine test results on the complete catalyst concepts, shown in Table V.
Dynamic Gas Absorption
In some practical catalyst systems the rate at which gas is absorbed/desorbed is an important criterion. T.W.C. systems are an example of this parameter and have been studied using the mass spectrometer system outlined above.
T.W.C. catalysts should have a selectivity on the air:fuel ratio scale as wide as possible, and at the same time have high activity for the removal of the appropriate gases. One method of achieving this is the inclusion in the catalyst design of an oxygen storage component. This stores oxygen under transient oxidising conditions and subsequently releases it under rich conditions, thus enabling the oxidising and reducing capabilities of the catalyst to be extended.
In order to design such T.W.C. systems for various car fuel management systems, three oxygen storage components were developed. The three concepts, A, B and C, were optimised by using the mass spectrometer flow system to measure the uptake of oxygen. The system allowed both the amount and the rate of oxygen uptake to be measured. As shown in Table VI system A has a slow response but large capacity, whereas system C has a low capacity but rapid response. The effect of the different oxygen storage components on the selectivity of a complete catalyst concept is amply illustrated by the results obtained on an actual car, and given in Table VII. Concept B is better than A, and considerably better than C, for this particular vehicle and fuel management system. However, for other cars with different fuel management systems the order may be reversed.
|Total uptake||1 Minute uptake||Total uptake||1 Minute uptake||Total uptake||1 Minute uptake|
|Air Aged at 700°C||3.89||1.52||1.12||0.81||0.26||0.22|
Catalyst Preparation and Scale-Up
The industrial preparation of catalysts can be split conveniently into three related sub-topics, the objective, the problem and the solution.
The objective is to produce catalysts of a consistently high quality on a large scale, in an economically viable manner, and from readily available raw materials.
The problem is that any catalyst preparation involves several successive unit operations such as impregnation, drying and activation, each of which yields a catalyst intermediate or precursor of somewhat undefined nature. We need to be able to characterise each of these if the overall catalyst manufacture and the final product are to be successful. For quality control purposes, ideally such characterisation must be done simply and rapidly, even while the catalyst is being manufactured. The differing nature of these catalyst precursors arises from the chemical and physical changes that occur during preparation and the interactions between the active material and the catalyst support.
In the following examples the use of TPR for characterising catalysts and catalyst precursors is described as it relates to catalyst preparation and scale-up.
The apparatus used is a modified standard temperature programmed gas chromatograph. By following the change in hydrogen concentration in a nitrogen gas carrier as it passes over the catalyst sample, the rate of reduction can be continuously recorded. If the sample temperature is raised at a uniform rate the different components on the catalyst surface reduce at different temperatures and the record of the rate of reduction typically will contain several reduction peaks similar to a conventional chromatographic analysis. From the temperatures at which these peaks occur the composition can be inferred, and from the sizes of the peaks the relative amounts may also be deduced.
Since the apparatus is very sensitive to small uptakes of hydrogen, the hydrogen chemisorbed on a metal surface or absorbed into the bulk of a metal such as palladium can also be measured quantitatively. We thus obtain not only a measure of the changes in surface composition but also a measure of metal dispersion of the final reduced catalyst. Three examples will perhaps suffice to illustrate the type and usefulness of the information obtained.
The first example shows a calcination study of a series of impregnated platinum on silica catalysts. From small scale trial preparations it was found that the quality of the catalyst was very dependent upon calcination time and conditions. Figure 9 shows a series of TPR fingerprints of such a catalyst taken after calcination for different lengths of time. The metal is mobile under such conditions and, as can be seen, the final metal dispersion first decreases and then increases. These changes are paralleled by rather marked changes in the TPR fingerprint.
The catalyst manufacturer would not long remain in business if he used a trial and error approach to making every large batch of catalyst, and large batches often require different treatment times compared to similar smaller batches. By following the systematic variations in the TPR fingerprint which occur with time the manufacturer can now, however, control and optimise the final product and substantially reduce the chances of making a poor batch.
The second example is also concerned with quality control but from a somewhat different aspect. The careful experimentalist can afford the luxury of using nothing but the highest quality reagent grade materials for his preparations, but catalyst manufacture on a large scale may necessitate the use of technical grade raw materials of somewhat uncertain and variable quality, with substantial variations from one batch of raw materials to the next.
Undesirable impurities frequently give bands in the TPR fingerprint and Figure 10 shows the TPR fingerprint of a good and bad catalyst, as indicated by the hydrogen chemisorption values. Quite clearly the bad sample has an extra component; in this case it was known to be sulphate—a common impurity in some alumina supports.
The third TPR example is a case of a catalyst ‘post mortem’. If a catalyst fails prematurely, the user is naturally interested in the reason for this and is uncertain whether to blame the catalyst manufacturer or some defect in his own operation. TPR can be a valuable aid in such a study, not only because possible impurities are often indicated but sometimes because more far-reaching changes may have taken place. An unusual example of such an event is shown in Figure 11, which shows the different TPR fingerprints of a bimetallic platinum-tin on alumina catalyst. The catalyst as-made had the platinum and tin combined, as shown by a single reduction peak. However, after use the alloy combination has been destroyed and two peaks, one platinum and the other tin are recorded. This, as shown, could have happened during air regeneration of the catalyst, particularly if wet air had been used.
These examples illustrate the use of temperature programmed reduction in catalyst preparation and scale-up. In practice all of the physical techniques in Table I will be used.
Process Trouble Shooting
As an example of the combined use of physical techniques in process trouble shooting, a description of the use of XRD, EPMA, SEM and TEM in characterising rhodium-platinum gauzes, used as catalysts in the oxidation of ammonia, will be given.
On occasions rhodium-platinum gauze packs have been reported as giving anomalously low conversion efficiencies after only a short period of plant operation. This effect is normally associated with medium and high pressure plants and until recently could not be accounted for either as a catalyst or as a plant problem.
Figures 12 and 13 show the EPMA distributions of rhodium in a gauze with normal activity and one with reduced activity. These results immediately suggest a rhodium enrichment factor. SEM photographs of three gauzes having normal, intermediate and low activity are shown in Figure 14, and indicate two forms of surface structure, namely platelets and needles. Additional XRD data shows the platelets to be essentially 10 per cent rhodium-platinum, while the needles consist of rhodium oxide (Rh2O3). An explanation of the formation of rhodium oxide has been given by Schmahl and Minzi, who studied the relationship between the decomposition of rhodium oxide and oxygen pressure and temperature, for rhodium-platinum alloys. This work showed that rhodium oxide is stable only below temperatures in the 800 to 900°C range depending upon the partial pressure of oxygen. Thus the problem becomes one of plant operating conditions; both the temperature and the ammonia-to-air ratio must be controlled so that rhodium oxide is not formed.
Finally the use of a special gas reaction cell, on a transmission electron microscope, to study the behaviour of platinum or rhodium-platinum was illustrated. The purpose of the experiments was to examine how surface rearrangements of the gauze take place, and in particular the mechanism of whisker formation.
A film recording these effects was shown at the conference to demonstrate that industry will turn even a transmission electron microscope into a nitric acid plant in order to use all appropriate physical techniques as a means of improving industrial catalytic processes.
The full text of this address, including references to earlier work, and of the other papers given at the Characterisation of Catalysts course are to be published by John Wiley & Sons Limited, Chichester, in 1980.