1
Liquid-Phase Synthesis of Nanocatalysts

Jean-Francois Hochepied

ENSTA ParisTech UCP, MINES ParisTech MAT/SCPI, 828 Bd des Maréchaux, 91762 Palaiseau Cedex, France

1.1 Introduction

The control of catalysts structure at the nanoscale is the key to increase performances and improve the fundamental knowledge about reaction mechanisms. Thanks to powerful nanocharacterization tools, especially high-resolution transmission electron microscopy [1], chemists can check their ability to control critical parameters as particle size, composition, shape, exposed crystalline faces, and particle-support interfaces. This has boosted studies linking processes, nanostructures, catalytic properties measurements, and modelization, paving the way for the rational design of nanocatalysts.

Liquid-phase processes offer a compromise between industrial constraints and fine control of nanostructures. It is impossible to encompass in a few pages all processes and materials relevant to catalysts and many are considered with more details elsewhere in this book, so the point of view is to focus on two families of materials - metallic oxides and metals - and show by selected examples how they can be shaped and interfaced at the nanoscale using simple and industrially relevant processes to create nanocatalysts. In the case of oxides - either catalysts or catalysts supports - we will consider both particles and porous structures, whereas we will rather focus on particles in the case of metals, considering they can be either colloidal catalysts or supported by oxides.

Liquid-phase syntheses are bottom-up approaches consisting in condensing soluble species. The formation of solid can be described in terms of nucleation and growth (eventually followed by agglomeration) and basic theories and concepts help to understand the strategies relevant to catalysts design.

Roughly, two key parameters, supersaturation S and solid surface tension ? (or free energy), are sufficient to provide expressions for homogeneous nucleation and growth rates in solution. Surface tension is related to the energy needed to create interfaces. In the case of a liquid (drop model), the surface chemical potential of a droplet with radius r is given by the following equation:

where Vm is the molar volume of the condensing molecule. Similar expressions can be obtained with solids, keeping in mind that surface tension is no more isotropic if they are crystallized but depends on the exposed faces. In any case, the important point is the ?/r dependence law. Supersaturation may be defined as the ratio between the actual quotient of the reaction in solution and the equilibrium constant, and gives the driving force of precipitation. In addition, in nucleation an additional free energy term comes from the creation of a solid surface, so the expression of the free enthalpy ?Gi for the condensation of i soluble units A into a cluster Ai can be written as follows:

where s is the surface covered by one unit. Here a simplified approach considers a critical germ i* defined by the maximum of ?Gi and derive a kinetics expression for the nucleation rate J. In fact, in a more rigorous approach, the thermodynamical system with all population from i?=?1 to i?=?Nmax should be considered [2], but the derived expression for nucleation kinetics is the same as in the simplified approach. The critical germ can be considered as the smallest possible solid particle; in theory, it practically does not exist in solution, but its formation is the kinetics bottleneck. So in this approach, the nucleation rate J, that is, the number of germs produced per second and per volume unit, is as follows:

The expression is simple, but unfortunately supersaturation and surface tension are in general not easy to determine. If mixing of solutions is used, even with rapid mixers characteristic times for mixing are in general longer than characteristic times for nucleation. Surface tension depends on the germ facies, which is by no means similar to the equilibrium (Wulff) facies. Nevertheless, the expression gives some clues about sensitiveness of nucleation kinetics to supersaturation and surface tension. In the case of heterogeneous nucleation, a prefactor (f<1) is used in the expression of ?Gi to quantify the fact that it is generally easier than homogeneous nucleation (the interfacial energy between substrate and germ is lower than between solvent and germ). It is important to keep in mind that if heterogeneous nucleation is wanted, the supersaturation must be moderate to avoid homogeneous nucleation, which explains the strategies used for controlling subtrate-nanoparticles interfaces.

If we consider the evolution of a crystalline germ, crystal growth depends on the nature of the faces. If we consider perpendicular growth (addition of new layers on faces), high-energy faces grow faster than low-energy faces, because they can grow by continuous incorporation of matter from the solution, whereas low-energy faces grow by 2D nucleation. So the surface proportion of high-energy faces tends to lower during crystal growth. Trying to modify the relative growth rates of different faces is the basis of kinetic facies tuning. In order to tune the facies of nanoparticles, several strategies are possible. Some recipes produce germs with well-defined facets, additives that will selectively "poison" some surfaces and prevent their growth are also known more or less empirically. If we consider general expressions, nucleation, especially when initial supersaturation is very high, can consume the main part of reactants and only a small fraction can remain available for growth. This seems favorable if one just wants to synthesize nanoparticles, but a problem is the first precipitate may frequently be metastable (according to the Ostwald rule of stages [3], less stable products are kinetically favored), transforming into the stable product via redissolution under conditions of much lower supersaturation, hence more favorable to growth. It is also important to keep in mind that even when there is no more average supersaturation after precipitation, nanoparticles may still evolve due to size polydispersity: Small particles are more soluble than bigger ones due to the 1/r dependence law of their surface chemical potential; hence, in a medium supposed to be at the solubility "equilibrium," the average solubility level is in fact undersaturated for small particles that tend to dissolve, whereas supersaturated for big particles that tend to grow: This phenomenon is known as Ostwald ripening [4]. The same applies for facies: The chemical potential of high-energy faces is higher than that of low-energy faces, and the facies tends to change in favor of low-energy faces. Uncontrolled ripening is therefore in general detrimental to catalysts activity by lowering the surface area and exposing less active faces (Figure 1.1).

Figure 1.1 Ripening of nanoparticles in solution without apparent supersaturation. An important cause for catalysts performance loss.

1.2 Metallic Oxides

1.2.1 (Co)Precipitation by Mixing (Aqueous Solutions) and Ripening

The mixing of a concentrated aqueous solution of metallic salts and a basic solution is the most direct way to (co)precipitate metallic (hydr)oxides. The strategy is simple: generate a very high supersaturation favoring nucleation over growth, which is relatively easy when resulting (hydr)oxides are poorly soluble. In general, as precipitated products (often metastable, sometimes amorphous) need post-treatment (ripening and calcination) to be chemically and crystallographically stable, rebuild their surfaces and meet the specifications needed for the application in catalysis (Figure 1.2).

Figure 1.2 From soluble species to solid (hydr)oxides: the frequent occurrence of intermediate metastable solid (in green) must be considered in the size and shape control of final particles (in red).

The way mixing is done is critical and can change drastically the products starting from the same solutions and considering the same final bulk conditions, as evidenced, for instance, with boehmite (mesoporous or fibrillar) precipitated by mixing aluminum nitrate with soda [5]. The most direct way of mixing two solutions consists in the injection of the basic solution into a batch containing the metallic acidic solution, but even with performant mixers the physicochemical conditions (pH, concentrations, and consequently supersaturation) of the bulk and locally at the injection point vary from the beginning to the end of the injection (even if rapid), with the risk that corresponding precipitated particles may be different. To circumvent this, in order to stabilize the physicochemical conditions in the mixing zones, a separated double-jet system is appropriate, as evidenced, for instance, for nickel hydroxide precipitation where the bulk pH was shown to control particle size and crystallinity [6]. The extreme case of double-jet consists in rapid (static) (micro)mixers, where both fluids are injected and mixed in a confined volume. The design of microreactors at various scales (from laboratory-scale microfluidics to industrial mixers) has been recently boosted by the increasing computing power for hydrodynamic modelization and...