3.1 Characterisation of the fresh solids

Chemical analysis of the prepared solids shows that the expected Ln/Cu atomic ratios were obtained within a tolerance below 10% in all cases (Table 1). However, a slight excess of copper with respect to the lanthanide stoichiometry can be noticed. Probably the rare earth salts used suffered from hydration, giving rise to a lack of the Ln-component. The specific surface areas of the Ln₂CuO₄ mixed solids were found to be very close to that of La₂CuO₄, about 0.5 m² g^–1. The observed phases are the so-called K₂NiF₄-type structures, La₂CuO₄, Pr₂CuO₄ and Nd₂CuO₄. Besides very weak diffraction lines characteristic of compounds originating from the corresponding lanthanide precursors, La₂O₃, PrO₂ and Nd(OH)₃, were identified.

The TPR profiles of the three Ln₂CuO₄ are given in Fig. 1 . The mixed oxides are reduced within the temperature range 350–750 °C, but with marked differences between each other. La₂CuO₄ exhibits two main peaks at 420 and 495 °C, which may be attributed to reduction of Cu²⁺ to Cu⁰ via the intermediate formation of Cu⁺ [14]. When Pr or Nd substitutes for La, the reduction occurs at higher temperature, with maxima at 540–625 °C for Pr₂CuO₄ and 600–650 °C for Nd₂CuO₄. The calculation of the H₂ consumption up to 750 °C shows an almost complete reduction of Ln₂CuO₄ in Cu⁰ and Ln₂O₃, the reduction percentage in metallic copper being 104, 95 and 97%, respectively, for La₂CuO₄, Pr₂CuO₄ and Nd₂CuO₄.

3.2 Catalytic activity of Ln₂CuO₄

The T80 and T50 temperatures for the NO conversion are plotted in Fig. 2 as a function of the ionic radius of the rare earth ion. The dashed lines represent the activation observed between the first temperature rise of the test and after catalytic stabilisation, i.e. when two complete cycles of heating and cooling under reaction conditions had been run. This phenomenon of catalytic activation under reaction conditions had already been described in previous work [5]. The values plotted here are those after stabilisation of the catalysts. The T80 and T50 values for Nd₂CuO₄ and Pr₂CuO₄ are by around 100–140 K higher than for La₂CuO₄, which therefore is the most active of the three catalysts.

In addition to N₂, some N₂O is formed. The selectivity to N₂O is close to 100% at low temperature (200 °C) and then decreases with temperature. It is about 10% at 450 °C. However, the maximum N₂O production (≈100 ppm) was observed at around 400 °C for Pr₂CuO₄ and Nd₂CuO₄, as it is observed for this type of solids [6].

After stabilisation of the activity, the apparent activation energies given in Table 2 are equal with values around 45 kJ mol^–1 for the three Ln₂CuO₄ oxides, whatever the lanthanide ion is. Before activation under reaction conditions the values differ between the catalysts, 103 kJ mol^–1 for Pr₂CuO₄, 88 kJ mol^–1 for Nd₂CuO₄ and 31 kJ mol^–1 for La₂CuO₄. However, in the latter case, the Arrhenius plot varies strongly during the temperature rise and is not really reliable. In the low temperature domain, it is higher than 150 kJ mol^–1.

The activity per gram of La₂CuO₄ shown in Fig. 3 was found to be 4 times higher than for the oxide containing Nd and still 2–3 times higher than for the Pr-based oxide, depending on the temperature at which the activity was measured. Thus, as seen in Fig. 3, the catalytic activity seems to increase almost linearly with the ionic radius of the rare-earth element.

3.3 Modifications of Ln₂CuO₄ under reaction conditions

We had previously found [5] that La₂CuO₄ is subject to modifications under CO + NO reaction conditions, leading to the formation of metallic copper and lanthanum hydroxide resulting from hydration of the oxide after exposure to air. In order to check whether this reduction is restricted to lanthanates or is also possible for other rare earth oxides, we recorded X-ray diffraction patterns of the Pr₂CuO₄ and Nd₂CuO₄ samples issued from the catalytic test. In Table 3, their phase composition is compared to La₂CuO₄ in the same state.

We can state that the initial Ln₂CuO₄ binary oxide has been completely transformed during the activation under the CO + NO mixture. As previously observed for La₂CuO₄ [5], the principal phases after reaction and exposition to air are the corresponding rare earth hydroxides Ln(OH)₃ and metallic copper. The observed activation under reaction conditions thus corresponds to a reduction of the initial cuprate.

3.4 Discussion

The active sites for the CO + NO reaction should be the same for La₂CuO₄, Pr₂CuO₄ and Nd₂CuO₄. In fact, we found the same apparent activation energy for the three solids after the catalysts had reached their activated state under reaction conditions. We attribute this coincidence to the presence of metallic copper, the common component of the three samples. Indeed for La₂CuO₄ [7], the surface concentration of metallic copper was found to be related to the catalytic activity, and it was concluded that copper could be considered as the main active species in the reaction. It can be remarked that the presence of metallic copper does not exclude redox mechanisms involving copper in different valence states and occurring at the surface or at the interface with the lanthanide oxide. In the present work, the catalytic activities of the three solids after stabilisation are different and related to the ionic radius of the lanthanide. It suggests that the number of active copper sites should be different for the three solids, which implies some differences in the reduction steps of the original cuprates and in the formation of the copper particles. If we consider the ionic radii of the rare earth element with a coordination number of 12, they are of 0.1286 nm for Pr³⁺ and 0.1276 nm for Nd³⁺ compared to 0.1320 nm for La³⁺ [15]. These differences in ionic radii are small, but they correspond to different energetics of the cuprates crystal lattice. The smaller ionic radii for the Pr- and Nd-based catalysts result in a more compact structure, which makes the oxygen less labile and consequently the reduction of the oxide more difficult. This is evidenced by the TPR study under hydrogen, which clearly shows that the smaller the ionic radius, the higher the reduction temperature. Cu⁰ is formed less easily when La is substituted by Pr or Nd. However, the link between this difference in reducibility observed by TPR and the changes in catalytic activity requires further discussion. In TPR processes, both thermodynamic and kinetic factors are important to control the initial formation of copper nuclei and then the interface propagation inside the bulk [16]. However, in the present catalytic study, the differences of activities in the CO + NO reaction cannot be related directly to a variation in the reduction rate constants of the cuprates, because the activities were measured on solids after stabilisation, i.e. after formation of the copper particles. In these conditions, the decrease in activity from La to Nd should be related to an increase in the copper particle sizes, which means a lower number of active sites. Indeed, the easier reducibility of La₂CuO₄ at low temperature favours the formation of a high number of copper nuclei thus inducing the formation of a dispersed copper phase. On the other hand, for the less reducible oxide (Nd₂CuO₄), the reduction proceeds at higher temperature. There are a smaller number of nuclei and the reduction therefore results in the growth of large particles, all the bigger as the high temperature favours their sintering. Consequently, the number of active sites is decreased and thus the catalytic activity.