Catalysis at the nanoscale level feature Catalysis at the nanoscale level Nicola Pernicone Catalyst Consultant, Via Pansa 7, 28100 Novara, Italy, e-mail: [email protected] As the techniques presently used for routine measurements of nanoparticle size are largely unsatisfactory, especially for supported metal catalysts, new guidelines and procedures are described that allow us to obtain more accurate and reliable values of nanodimensions by chemisorption. Improved values of turnover frequencies will result, with benefi t for studies on the mechanism of catalytic reactions. Examples taken from both old and recent studies show that nanoscience has been applied in catalysis for a long time. It is quite probable that an extensive use of nanotechnologies in catalyst R-D will bring unexpected industrial achievements in the future. Meanwhile, nanoscience principles can be profi tably applied to the improvement of many current petrochemical catalysts. It has been recently remarked[1] that the addition of the thus making it diffi cult to change them one at a time. In “nano” prefi x to many technical terms is a quite recent practice, a skillful manufacturer of such catalysts must take practice that began with the introduction of the nanometer into account a complex network of experimental variables in as a microscopic length unit in place of the Angstrom. his work, aimed at optimizing catalyst performances. In this Though nanoscience was formally born with that event, it is work he clearly needs effective and not too expensive means nevertheless clear that nanosize materials were being studied for a reliable measurement of the size of metal nanoparticles. much earlier, especially in the area of catalysis. A similar As the present situation of such measurements is not picture has also been recently outlined by other authors[2,3]. satisfactory, a brief discussion of this subject now follows. In fact industrial catalysis has involved nanoparticles since its beginning at the dawn of the 20th century. For example, The measurement of metal particle nanosize nanoparticles are typically present in supported metal and related properties catalysts. Here, metal particles of a size usually in the range Physical techniques 1–10 nm are deposited on the external surface and/or in TEM (Transmission Electron Microscopy) seems to be the the porous texture of inert (with some notable exceptions) optimal technique for measuring the size of nanoparticles, materials. There is still much to improve in this area, with as it allows us to look at them directly. However its use is short-term benefi ts for chemical, petrochemical and refi ning problematic. For example, the analysis is carried out on a industries. However, most of the research funds are presently microscopic amount of catalyst, hardly representative of the set aside for very advanced, if not sometimes extravagant whole sample. Many observations must be made on each of manufacturing techniques, which often only fi nd industrial several independently loaded samples in order to get reliable application after decades, and even then only in the most favorable cases. I will try to show how nanoscience concepts Main variables for the control of metal particle can be applied to the improvement of several very important nanosize in supported metal catalysts catalysts presently used in the petrochemical industry, with carrier surface area examples mostly taken from my personal experience. Some nature of the precursor old data will be discussed too, to demonstrate that such chemical nature of the carrier surface concepts have always been applied in industrial catalysis. metal concentration The control of metal particle nanosize is of the utmost metal distribution importance for the performance of any industrial catalyst dispersion promoters based on supported metals. Such control is currently exerted impregnation procedure through a keen manipulation of several different variables, thermal treatments which are summarized in Table 1. It should be stressed that some of these variables are often interconnected, Table 1 196 197 Volume 7, no. 6, 2003 Catalysis at the nanoscale level feature data, with the considerable expenditure of time that entails. correct size measurement. On the other hand, when the Moreover, small or fl at nanoparticles may be lost, due to metal particles are smaller than about 2 nm, they do not insuffi cient contrast. However, a careful use of TEM is very contribute to the diffraction “visible” lines, thus giving an valuable in some cases. overestimated particle nanosize. However, in most cases it XRD (X-Ray Diffraction) techniques are more practical, is possible to determine, using a suitably tailored Rietveld but they too have some specifi c limitations[4]. Briefl y, when analysis[4], the content of undetectable metal nanoparticles, the metal nanoparticles are relatively large, approaching to which a reasonable average size can be assigned, for or exceeding 10 nm, each of them may consist of several example on the basis of ASAXS data[5]. By this way a surface nanodomains, so that the size measured by X-ray line average size of the whole population of metal particles broadening will be smaller than the real one. In such cases, can be calculated. Incidentally, it can be remarked that which can be recognized by HRTEM (High Resolution the widely used Scherrer equation, apart from the above- Transmission Electron Microscopy) analysis, SAXS mentioned problems, gives a volume average particle size, (Small Angle X-ray Scattering) can be used to ensure the completely useless in catalysis (the difference from the correct surface average can be dramatic in the frequent intermezzo 1 cases of wide particle size distribution). In conclusion, XRD How can the carrier infl uence the properties of techniques, when used by qualifi ed specialists, allow us to metal nanoparticles? get reliable values of metal particle nanosize in the large majority of metal-carrier couples, the use of TEM being Metal particle size is a very important property of supported metal restricted to few cases. catalysts, but several other properties are no less important, and For catalytic purposes, once the correct surface average some are even more so, such as lattice disorder or work function. metal particle nanosize Φav is known, it is convenient to The regulation of such properties can be obtained by tuning calculate the number of total surface metal atoms per a range of variables, some of which are interconnected, thus catalyst gram (NMe,S), given by the equation making it very diffi cult to change one at a time. NMe,S = k Cm fm / dm Φav (1) The choice of the carrier plays a crucial role, of course. I do where k is a constant depending on metal particle shape and not refer to the fi rst-step choice (among carbon, alumina, silica on the extent of contact with the carrier surface (usually k and so on), as it is usually straightforward for a catalyst specialist, [6] = 5), Cm is the surface density of metal atoms , fm is the but to the very numerous commercial or home-made samples of metal weight fraction in the catalyst and dm is the metal true each type of carrier. density. In this way an easier connection with chemisorption For example, it is a general rule that, when the surface data can be reached. area of the carrier is high, small metal nanoparticles are usually obtained. However it is not easy to change the carrier surface Chemisorption area without affecting other properties, such as the chemical The methods traditionally used by the catalytic community nature of the carrier surface. Something can be done for oxidic for measuring NMe,S are based on chemisorption. In practice carriers, in that their surface area can be decreased by suitable, what is experimentally measured is the number of probe not too drastic, thermal treatments, provided that surface molecules that disappear from the gaseous phase in contact hydroxylation does not change too much. For active carbons with the catalyst under the chosen experimental conditions this approach is precluded, as the decrease of surface area is (temperature and pressure). Unfortunately nobody knows connected with graphitization, which defi nitely alters the physico- what the real fate is of the disappearing probe molecules chemical properties of the material. because although UHV adsorption studies on monocrystals The chemical species present on the carrier surface interact, can give some information on the geometrical aspects of more or less, with the metal nanoparticles deposited thereon. This the interaction with the metal surface, this can hardly be phenomenon is currently called “metal-support interaction” (MSI) extrapolated to industrial catalysts. In practice, to calculate and is usually stronger when the metal particles are smaller. It NMe,S from the experimental data (the chemisorbed STP gas regulates the mobility of metal atoms and clusters on the carrier volume per catalyst gram Vg) the following assumptions are surface. In practice, when a strong metal-support interaction currently made: (SMSI) occurs, smaller (and more resistant to sintering) metal (a) The degree of coverage is 1 (very hard assumption, it nanoparticles are usually obtained. could be much lower). For the carriers usually employed in industrial catalysis (b) The chemisorption stoichiometry (ratio between metal the intensity of MSI decreases from alumina to silica to active atom and chemisorbed molecule, or atom in the case carbon. However the surface properties of active carbons can of dissociation) is arbitrarily taken as 1 in most cases, be widely changed by proper oxidizing and reducing treatments, probably for the sake of simplicity. thus allowing us to tune the metal particle nanosize and the (c) All the disappeared probe molecules interact only resistance to sintering. with surface metal atoms (unfortunately spillover and subsurface chemisorption cannot always be neglected). 197 Volume 7, no. 6, 2003 Catalysis at the nanoscale level Catalysis at the nanoscale level Errors as high as 100% may result, which of course are technique, no chemisorption method is better than another.