*3.3. Oxygen Species: Bonding Strength and Mobility*

The bonding strength of surface oxygen species is determined by such methods as pulsed microcalorimetry and temperature-programmed desorption (TPD) of O2. Experimental data are compared with calculations for model structures of surface sites using the semiempirical interacting bonds method [35,99–104]. The oxygen mobility was estimated using such methods as oxygen isotope heteroexchange with 18O2 or C18O2 [24,68,69,105–107], steady state isotopic transient kinetic analysis (SSITKA) [108,109] and modeling of reforming processes kinetic relaxations [19,104,108].

For catalysts based on fluorite, perovskite and spinel oxides containing transition and rare-earth cations with variable charges, the bonding strength of surface oxygen species depends on their stoichiometry. In the initial oxidized state, it is ~150–200 kJ/mol O2, corresponding to M–O forms of oxygen [35,99,100]. After removing more than one monolayer of oxygen its binding strength increases to ~400 kJ/mol for spinels (MnCrOx), ~500 kJ/mol for perovskites (PrFeOx, LaPrMnCrOx) and ~650 kJ/mol for LnCeZrO fluorites, respectively, corresponding to M2O bridging forms of the surface oxygen [99–103]. In the stationary state of these catalysts in the reactions of fuels reforming, only bridging forms of oxygen are present on the surface being regenerated by CO2 or H2O pulses forming also CO and H2 as products [101–103]. For PrCeZrO fluorite and MnCrOx spinel layers supported on mesoporous MgAl2O4 the coverage by reactive oxygen species decreases while their bonding strength increases due to interaction with support [104].

Diffusion coefficients of oxygen for these catalysts are high enough (up to 10–12 cm2/s at 700 ◦C, Table 1, Figure 1) to provide fast oxygen migration to the metal–support interface, which is required to transform activated fuel fragments into syngas.

Even though for stoichiometric spinel MnCr2O4 the oxygen diffusion in the bulk is not too fast, it is at least close to that of La-doped ceria-zirconia, which allows to suggest usage of much less expensive spinel for the catalysts design. Oxygen diffusion coefficients along domain boundaries in Pt-supported fluorites exceed by 1–3 order of magnitude those for the bulk diffusion (Table 1). The amount of oxygen involved into this fast diffusion channel decreases when using oxygen exchange with C18O2 or SSITKA instead of heteroexchange with 18O2, apparently reflecting also its lower amount (oxygen storage capacity) in real reaction conditions of biofuels reforming.


**Table 1.** Oxygen self-diffusion coefficients in the bulk (*Dbulk*) and along grain boundaries (*Dinterface*) at 700 ◦C [19,24,68,69,104–109].

**Figure 1.** Temperature dependencies of oxygen self-diffusion coefficients *DO* for ceria and ceriazirconia based samples prepared via modified Pechini route.

For PrNi0.5Co0.5O3–Ce0.9Y0.1O2−<sup>δ</sup> nanocomposite the method of isotopic heteroexchange of oxygen with C18O2 also demonstrated coexistence of fast and slow channels of oxygen diffusion [106]. Fast migrations go through perovskite–fluorite interfaces as well as via Ce0.65Pr0.25Y0.1O2−<sup>δ</sup> nanodomains (Table 1), while diffusion through perovskite domains is slow.

### **4. Catalytic Properties**

#### *4.1. Catalysts Based on Cerium–Zirconium Mixed Oxides*

In methane dry reforming (MDR) Ni-supported biphasic ceria-zirconia sample prepared in supercritical ethanol without adding AA complexion was very fast completely deactivated in MDR due to coking [20]. In contrary, a high and stable performance of catalysts with Ni supported by impregnation on single-phase Ce0.5Zr0.5O2 oxide prepared either by modified Pechini route or in supercritical alcohols with addition of AA (*vide supra*)

was maintained even at 600 ◦C [20–24,89]. At 700 ◦C, the effective first-order rate constants for these catalysts were in the range of 1–4 s–1 cm–2, being close to values for Pt or Ru- supported Pr (Pr+Sm)–doped ceria-zirconia (Figure 2) and exceeding by an order of magnitude k values for Ni supported on these fluorites (~0.2 s–1 cm–2 at 700 ◦C) [87]. This stresses importance of the spatial homogeneity of ceria-zirconia mixed oxide for ensuring a high oxygen mobility and, hence, stability to coking, which is usually not taken into account.

**Figure 2.** Specific rate constants temperature dependence for methane dry reforming on metal -supported SmPrCeZrO (filled symbols) and PrCeZrO (empty symbols) catalysts. Feed 7% CH4 +7% CO2 in He.

Mixed Ce0.75Zr0.25O2 oxides doped with Ti, Nb and Ti+Nb (Table 2) were prepared in supercritical isopropanol with addition of AA. 5 wt.% Ni were supported either by impregnation (I) or via one-pot synthesis from mixed solutions in supercritical conditions (O) [24]. For samples with Ni added in one-pot route its surface content estimated as Ni/Ce+Zr atomic ratio from X-ray photoelectron spectroscopy data was twice as low in comparison with impregnated samples. This indicates Ni incorporation into the bulk of fluorite particles for one-pot samples reflected in the increase of oxygen diffusion coefficients (Table 2) due to generation of additional oxygen vacancies. Such disordering also leads to stronger sintering of samples during calcinations step, so specific surface area for one-pot samples was twice as low [24]. However, both reagents' conversions and reaction rate related to the surface concentration of nickel atoms estimated by hydrogen chemisorption (TOF, turn-over frequency) of one-pot samples were quite close to those of impregnated samples of the same chemical composition (Table 2), which indicates on a higher dispersion of Ni on the surface of one-pot samples. TOF values strongly depend on the support composition, and the catalyst with titanium and niobium co-doped ceriazirconia support prepared by impregnation is three times more active than that with the unmodified ceria-zirconia support, apparently due to optimized interaction of Ni with more disordered doped ceria-zirconia.

**Table 2.** Kinetic parameters of MDR and oxygen mobility for catalysts prepared in supercritical conditions [24].


<sup>1</sup> I—impregnated sample, O—one-pot sample.

Oxygen diffusion coefficients (*Do*) estimated by the isotope exchange method, of the order of 10<sup>−</sup>15–10–14 cm2/s (Table 2), are quite close to those for Ni/Ce0.5Zr0.5O2 with single-phase oxide support prepared via modified Pechini route (Figure 1). They are high enough for efficient oxygen transport to the metal-support interface and contribute to a high value of catalytic activity and stability against catalyst coking.

For catalysts based on SmPrCeZrO2 oxide support, in MDR the rate constant was higher for supported Ru (~7 s–1 m–2 at 850 ◦C) than for Pt (Figure 2) [87]. Ni-supported catalyst is much less active (*k*~0.2 s–1 cm–2 at 700 ◦C), which is explained by coking of the catalyst. For supported LaNiO3 activity is higher by an order of magnitude (*k*~1.8 s–1 cm–2 at 700 ◦C). This is explained by decoration of Ni nanoparticles by LaOxCO3 species preventing coking. Ru+Ni-supported catalyst demonstrates the highest activity (Figure 2) since NiRu clusters are not coked [87]. For Ru co-supported with LaNiO3, specific activity is significantly lower being identical with that for Ru alone supported on this fluorite. According to CO chemisorption data the metal surface area for this catalyst (~0.05 m2/g) is much lower than that for Ru + Ni–loaded sample (0.5 m2/g). Hence, the rate constant related to the metal surface is even higher for Ru+LaNiO3–loaded catalyst due to stronger metal–support interaction/interface.

For Pt/Lnx(Ce0.5Zr0.5)1−xO2−<sup>y</sup> catalysts specific rate of hydrogen production in methane partial oxidation into syngas (POM) in diluted feed (Figure 3) is the highest for La-doped catalysts due to a higher content of Ptn+ cations transformed into clustered metal species in reaction media [13,18,72,94,95].

**Figure 3.** Specific rate of H2 production in partial oxidation of methane for Pt/Lax(Ce0.5Zr0.5)1−xO2−<sup>y</sup> O2−<sup>y</sup> (1), Pt/Prx(Ce0.5Zr0.5)1−xO2−<sup>y</sup> (2) and Pt/Gdx(Ce0.5Zr0.5)1−xO2−<sup>y</sup> (3) catalysts. Feed 1% CH4 + 0.5% O2 in He, 770 ◦C, contact time 5 ms.

In the autothermal reforming of acetone (Figure 4) [95], specific catalytic activity of Pt-supported doped ceria-zirconia catalysts correlates with the oxygen mobility required to prevent coking, being the highest for Pt/Pr0.3Ce0.35Zr0.35O2 catalyst [18].

While testing catalysts in real concentrated feeds in POM, methane steam reforming and MDR, the temperature gradient along the length of reactor equipped with the catalyst fraction or granules caused by exothermicity or endothermicity of these reactions emerges, which complicates data analysis. To avoid this problem, catalytic layers were supported onto the inner walls of single channels cut from corundum honeycomb monoliths, which allowed to avoid any temperature gradients and estimate rate constants (Table 3) [109–112]. Among Pt/Lnx(Ce0.5Zr0.5)1−xO2−<sup>y</sup> catalysts the highest activity was revealed for Pr-doped catalyst possessing the highest oxygen mobility, which stresses importance of this characteristic.

**Figure 4.** Specific rate of H2 production in autothermal reforming of acetone for Pt/Prx(Ce0.5Zr0.5)1−xO2−<sup>y</sup> and Pt/Gdx (Ce0.5Zr0.5)1−xO2−<sup>y</sup> catalysts.

**Table 3.** Effective first-order rate constants (s<sup>−</sup>1) of CH4 reforming at 700 ◦C on separate corundum channels with supported active components [111,112].


<sup>1</sup> Feed 7% CH4 + 3.5% O2, N2 balance; <sup>2</sup> Feed 7% CH4 + 21% H2O, N2 balance; <sup>3</sup> Feed 7% CH4 + 7% CO2, N2 balance.

To accelerate preparation of ceria-zirconia based catalysts, an automated workstation was used allowing wet impregnation of La-doped γ-Al2O3 by mixed solutions, so series of samples with supported CeZrO layers doped by Pr or Sm and promoted by Pt, Ru, Cu, Cu + Ni were made [113]. The ruthenium-promoted Ru/Ce0.4Zr0.4Sm0.2 O2−y/La-γ-Al2O3 catalyst demonstrated the best activity and coking stability in ethanol steam reforming, which was explained by a high mobility and reactivity of oxygen in this sample.

#### *4.2. Catalysts Based on Perovskite Oxides*

#### 4.2.1. Reactions of Methane Reforming

A lot of ABO3 perovskites with partial substitution of La in A sublattice for alkalineearth (Ca, Sr) or rare-earth (Ce, Pr) cations as well as containing in B sublattice, along with Ni, such metals as Fe, Co, Rh and Ru, were studied in these reactions [30–48,50,51,53–60] as well as in diesel fuel reforming [49,61]. Main attention was paid to the effect of perovskite initial composition on segregated in reaction conditions Ni (or Ni alloys) dispersion, interaction of a metal component with remaining oxide matrix and ability of the latter to activate oxidants and provide oxygen species diffusion to the metal-support interface. Clearly, formation of Ni nanoalloys with Co, Fe, Ru, Rh for reduced perovskites doped in B sublattice prevents coking. For Fe-containing perovskites [28,32,34,39,57] incomplete reduction of Fe cations in reaction conditions helps to stabilize a part of perovskite Ln-Fe-O matrix strongly interacting with segregated nanoalloys. In a similar way, substitution

of La in A sublattice for Ce and Pr cations able to change their charge state increases oxygen mobility and reactivity in remaining perovskite matrix, thus improving resistance to coking [32,34,58].

A high and stable performance in MDR without any coking was demonstrated also for LaMnO3-based perovskite with *in-situ* exsolved Ni nanoparticles strongly interacting with perovskite surface layers [114].

Detailed review on methane dry reforming over perovskite derived catalysts is presented in [39], where basically the same trends in ensuring high and stable performance related to oxygen mobility and strong interaction of metal nanoparticles with remaining oxide matrix are analyzed.
