Oxy-dry reforming of propane over Ce-promoted Co–Ni/Al₂O₃ catalyst

Abstract

This paper reports the production of syngas from two types of O₂-assisted dry reforming of propane, namely oxidative (O₂-dosed) dry reforming (ODR) and dry (CO₂-dosed) partial oxidation (DPOX). Reaction runs were conducted over alumina-supported bimetallic Co–Ni promoted with CeO₂ at 120 kPa and 793–893 K. Ceria promotion improved the carbon deposition resilience of the Co–Ni catalyst. Physicochemical attributes were obtained from liquid N₂ adsorption, H₂ chemisorption and temperature-programmed desorption runs for NH₃, CO₂, CH₄ and C₃H₈. Rate behavior under ODR, DPOX and pure dry reforming could be described consistently with empirical models that are structurally similar to Langmuir–Hinshelwood type relations. Inferences from these models allowed the postulation of the same overall reaction network for the three types of reactions albeit with variation in rate-controlling steps depending on the different product species. On the whole, DPOX seemed to be a superior option for the manufacturing of syngas for downstream olefin FT production due to reduced variability in the H₂:CO ratio and the closeness to unity (0.72–0.95) of the exiting syngas over the range of O₂ partial pressure used.

Introduction

Synthesis gas (H₂/CO) is the primary feedstock for clean fuels manufacturing in a variety of gas-to-liquid (GTL) conversion processes [16, 21, 28]. Although steam reforming (SR) of hydrocarbons is most commonly employed, the partial oxidation (POX), and dry reforming (DR) routes have also been discussed [23, 27]. In particular, autothermal reforming [10, 11, 17, 18] which combines the exothermic POX with the endothermic SR reactions in a single reactor module in order to reduce overall energy demand and minimize carbon deposition has also attracted considerable attention within the past decade [12, 13, 19]. However, with the recent interest in the role of greenhouse gases on climate change and possible carbon tax on manufacturing plants, development of efficient technologies for CO₂ utilization in the petrochemicals and gas processing industries has taken on added urgency. Thus, the CO₂ (dry) reforming option for syngas production has assumed a central stage in new green technology initiatives [1]. Similar to steam reforming, it is also an endothermic reaction. Nevertheless, the product H₂:CO ratio is lower than that from SR and more suitable for downstream olefin or GTL synthesis.

In this study, the co-feeding of oxygen with the CO₂ and hydrocarbon as reactants to the reformer has been proposed in order to take advantage of the exothermicity of the accompanying hydrocarbon oxidation to reduce energy requirement for the dry reforming. In particular, O₂ addition will also improve material economy through reactive excoriation of the carbon residue that inevitably accompanies dry reforming. This is especially relevant with C₂₊ alkanes as the hydrocarbon source due to a greater propensity for carbon laydown than with CH₄. A further benefit of this approach in comparison with the autothermal reforming is that the additional energy for steam generation is avoided. The combined pure DR and POX of propane are given by

$${\text{C}}_{3} {\text{H}}_{8} + \frac{3}{2}{\text{CO}}_{2} + \frac{3}{4}{\text{O}}_{2} = \frac{9}{2}{\text{CO}} + 4{\text{H}}_{2}$$

(1a)

Thus, for the oxy-dry reforming operation (where the relative partial pressure of O₂ and CO₂ may be varied to achieve different levels of oxygen-assisted reforming), the reaction rewrites in general form as:

$${\text{C}}_{3} {\text{H}}_{8} + \left( {3 - 2\gamma } \right){\text{CO}}_{2} + \gamma {\text{O}}_{2} = \left( {6 - 2\gamma } \right){\text{CO}} + 4{\text{H}}_{2}$$

(1b)

where γ is the amount of O₂ supplied per mole of C₃H₈ fed with γ = 0 for pure dry reforming and γ = 1.5 for the strict partial oxidation option. Given that pure propane DR is an endothermic reaction (ΔH₂₉₈ = 622 kJ mol⁻¹) while propane POX is exothermic (ΔH₂₉₈ = − 227 kJ mol⁻¹), the heat of reaction, ΔH, for Eq. (1b) may vary from positive to negative values for 0 ≤ γ ≤ 1.5 in addition to being a function of temperature, T. Indeed, ΔH(T), may be given by

$$\Delta {\text{H}}\left( T \right) = \mu_{0} + R_{\text{g}} \left. {\left( {\mu_{1} T + \frac{{\mu_{2} }}{2}T^{2} + \frac{{\mu_{3} }}{3}T^{3} - \frac{{\mu_{4} }}{T}} \right)} \right|_{298}^{T} \left( {{\text{kJ mol}}^{ - 1} } \right)$$

(2)

for 373 ≤ T ≤ 1073 K where R_g is the universal gas constant (8.413 J mol⁻¹K⁻¹) and μ₀ to μ₄ are dependent on γ as displayed in Table 1. It is immediately evident that Eq. (1b) yields theoretical H₂:CO ratio of 2/(3-γ) which is more suitable for downstream Fischer–Tropsch synthesis or olefin production plant (i.e., 0.67–1.33 for realistic values of γ) than the higher values (> 3) obtained from typical steam reforming operation. Furthermore, using the parameter expressions provided on Table 1, Eq. (2) may be readily simplified to the linear form:

$$\Delta {\text{H}}\left( T \right) = \left( {0.1951 - 6.2 \times 10^{ - 3} \gamma } \right)T + \left( {542.72 - 558.1\gamma } \right)$$

(3)

with an error bound of ± 5% (compared to its full nonlinear version). However, the possible side reactions such as reverse water–gas shift reaction, carbon deposition, H₂ and CO oxidation may impact upon the true product distribution. Therefore, a key objective of this investigation is to obtain empirical O₂ partial pressure in the feed that will optimize the H₂:CO product ratio while maintaining maximal reaction rate and minimal carbon laydown. Promotion of the Co–Ni catalyst with ceria has been adopted in the present investigation because rare-earth oxides have been found to reduce carbon deposition during hydrocarbon dehydrogenation [5], steam reforming [20, 29] and dry reforming [8, 9].

Table 1 Values of thermodynamic parameters in Eq. (1b)

Experimental details

The ceria-promoted Co–Ni catalyst was prepared by co-impregnation of the aqueous solution containing calculated amounts of Co(NO₃)₂, Ni(NO₃)₂ and Ce(NO₃)₃ onto γ-alumina (support) which had been previously treated at 1073 K for 6 h. The nitrate addition was followed by 3 h of stirring at 303 K (at constant pH of two) using a computer-controlled Mettler–Toledo Excellence titrator (model T90) and subsequent 24 h drying of the slurry in an oven was performed at 403 K. The resulting solid containing 2.5Ce:5Co:10Ni (wt % basis) on alumina was further calcined at 1073 K for 5 h at 5 K min⁻¹. The calcined solid was then crushed and sieved to 180–250 μm particles for characterization and reactor studies.

The BET surface area and pore volume for the catalyst were obtained from liquid N₂ adsorption–desorption measurements at 77 K on Quantachrome Autosorb-1 unit. H₂ chemisorption was carried out on Micromeritics AutoChem 2910 (equipped with a thermal conductivity detector, TCD, and gas ports with mass flow controllers for preparation, carrier and analysis gases) to determine metal dispersion and particle size while acid and basic site concentration and strength were obtained on the same instrument using NH₃ and CO₂ temperature-programmed desorption (TPD) at 973 K for 4 different heating rates (15, 20, 25, 30 K min⁻¹). TPD runs were also performed for C₃H₈ and CH₄ under identical conditions to determine the adsorption characteristics (heat of desorption and site coverage) of these species, especially since CH₄ may be produced via direct C₃H₈ decomposition during reforming. Metal oxide phase identification was performed on ThermoCahn TG-2121 TGA unit under H₂-temperature-programmed reduction (TPR). TPR runs employed 50% H₂/Ar mixture flowing at 55 ml min⁻¹ at 973 K under a temperature ramp of 10 K min⁻¹. XRD analysis was also conducted on a Philips X’pert system using an Ni-filtered Cu K_α radiation (λ = 1.542 Å) at 40 kV and 30 mA to complement H₂-TPR data. Actual elemental composition of the fresh catalyst was determined via XRF analysis. The residual total organic carbon (TOC) content on used catalyst specimens was obtained from a Shimadzu Total Organic Carbon Analyzer (solid module SM-5000A).

Catalyst evaluation runs for the Ce-promoted catalyst were carried out in a 15 mm ID quartz fixed-bed reactor at temperatures between 793 and 893 K and constant pressure of 120 kPa. Prior to each run, the catalyst (ca. 0.5 g) was reduced in H₂ for 2.5 h. Feed gas to the reactor contained CO₂, C₃H₈ and O₂ at a total flow rate of 100 ml min⁻¹(at 298 K). In one set of experiments, feed with different CO₂:C₃H₈ ratio, $R_{{{\text{CO}}_{2} }}$ (from 0 to 9) but regulated to keep a constant feed O₂:C₃H₈ ratio, $R_{{{\text{O}}_{2} }}$, of 0.75 was employed. This was termed, the oxidative dry reforming (ODR) or O₂-dosed dry reforming set since CO₂ partial pressure was varied. The fixed $R_{{{\text{O}}_{2} }}$ value of 0.75 corresponds to the stoichiometric requirement in Eq. (1a). The influence of oxygen partial pressure was investigated in the second set of runs by varying the feed O₂:C₃H₈ ratio, $R_{{{\text{O}}_{2} }}$, from 0 to 1.5 with the feed CO₂ content, $R_{{{\text{CO}}_{2} }}$ kept constant at 1.5 for each run [the stoichiometric coefficient of CO₂ in Eq. (1a)]. This second set of runs was referred to as the dry partial oxidation (DPOX) or CO₂-dosed POX. Feed mixture composition in all runs was carefully chosen to be well outside the hydrocarbon explosion limits. Additional experimental details are provided in a previous paper [3].

Results and discussion

Properties of fresh and used catalyst samples

XRF analysis revealed an elemental composition of the promoted catalyst as 2.6 wt % Ce, 5.5 wt % Co and 10.5 wt % Ni and balance (others) in reasonable agreement with the intended recipe of 5Co:10Ni and 2.5Ce:5Co:10Ni. However, traces of I, Yb, Si, Th, Te, Fe in the range 0.0043–1.09 wt % were probably due to original impurities in the AR grade Sigma-Aldrich chemicals.

BET and pore volume data displayed in Table 2 suggest that the addition of ceria to the Co–Ni/Al₂O₃ catalyst did not have any marked effect on its porosity or surface area—an indication of negligible increase in crystallite size. XRD runs were performed on a Phillips X’Pert X-ray diffractometer using Ni-filtered CuKα radiation (λ = 1.542 Å) at 40 kV and 30 mA. The X’Pert Pro software was used for the qualitative determination of the metallic oxide phases. The X-ray diffraction patterns of calcined catalysts (cf. Fig. 1) showed that both ceria-promoted and undoped Co–Ni catalysts contained similar cobalt and nickel oxides, with NiCo₂O₄ (2θ = 31^o and 36.6^o) and Co₃O₄ at 2θ = 31.2^o and 36.7^o while NiO is located at 2θ = 43.2 ^o. The metal aluminates, CoAl₂O₄ (2θ = 36.7^o) as well as NiAl₂O₄ (2θ = 37^o; 44.8^o), exhibited strong intensities but the CeO₂ was identified as a separate oxide phase on the ceria-containing catalyst at 2θ = 28.6^o. The absence of Ce aluminate and Ce–Co–Ni oxide phases indicates negligible perovskite formation. H₂ chemisorption data collected at 383 K and detailed in Table 2 revealed that the metal particle size and dispersion were barely affected by the addition of ceria suggesting that the latter was unreduced under the conditions employed.

Table 2 Physical properties of the undoped and Ce-promoted catalyst

Fig. 1

X-ray diffraction patterns for the Co–Ni catalyst and the Ce-promoted counterpart

Average crystallite size for all the metal oxides computed from Scherrer equation for the XRD data was 122.2 nm (undoped) and 147.1 nm (promoted). These values compare well with those from H₂ uptake data. H₂-TPR profile shown in Fig. 2 suggests the presence of nearly identical metal phases in both types of catalyst. The peak at 401 K may be attributed to the removal of interstitial water. The principal broad peak at about 725 K is due to the reduction of a key metal oxide phase containing Co₃O₄, NiCo₂O₄ and NiO. However, the low temperature shoulder peaks (473–513 K) may be ascribed to the presence of XRD-amorphous Ni₂O₃ and Co₂O₃ (arising from the thermodynamically feasible 2NiO + Co₃O₄ + CoO + O₂ ↔ Ni₂O₃ + 2Co₂O₃ at T ≥ 673 K) while the higher temperature peaks suggest the reduction of the Ni-Co aluminates (at 973 K). Since CeO₂ is irreducible below 1000 K, a peak corresponding to ceria reduction was absent from the TPR profile. The results of NH₃-TPD summarized in Table 3 indicate that both catalysts have relatively weak acid sites characterized by NH₃ heat of desorption, E_d, of 104.5 kJ mol⁻¹ (for unpromoted) and 73.5 kJ mol⁻¹ (Ce-doped) characteristic of Lewis acid sites with peak temperature below 673 K [6]. Interestingly, the acid site concentrations on both catalysts are close (6–7 μmol m⁻²) implicating the location of acid sites at the Ni (and Co)-support boundary since they have nearly identical composition and phases on both catalysts. However, the CO₂-TPD data suggest that the basic site concentration on the promoted catalyst (0.35 μmol m⁻²) is nearly twice that of the undoped specimen (0.19 μmol m⁻²). Basic sites may be the contiguous conjugate centers of the metal-support boundary acid sites in the Ni-Co/Al₂O₃ catalyst. The ceria phase in the promoted sample also offers additional basic sites due to free oxygen at the CeO₂ surface. Moreover, while the acid site strength on the ceria-containing catalyst is somewhat weaker than the undoped counterpart, the latter appears to have weaker basic site strength (66.5 kJ mol⁻¹) than the promoted sample (78.5 kJ mol⁻¹). Thus, the ratio of the acid:basic site density in the undoped catalyst is about twice that of the ceria-promoted sample confirming the latter’s stronger basic character.

Fig. 2

TPR patterns of fresh 0 and 2.5 wt % Ce–5Co:10Ni catalyst samples

Table 3 Catalyst properties arising from temperature-programmed desorption of NH₃ and CO₂ (acid-base attributes) as well as C₃H₈ and CH₄ for the catalysts

Table 3 also shows the results of the C₃H₈ and CH₄-TPD runs at 1023 K using four heating rates from 15 to 30 K min⁻¹. It is apparent that the surface capacity for C₃H₈ adsorption is similar for both catalysts (3.4 and 3.8 μmol m⁻² undoped and promoted samples, respectively) suggesting that the reduced metal crystallites are the C₃H₈ adsorption sites. Nonetheless, the heat of desorption for the reactant is significantly different with 102 kJ mol⁻¹ for the Co–Ni catalyst and 74 kJ mol⁻¹ for the Ce–Co–Ni system. The smaller C₃H₈ desorption strength on the promoted catalyst implicates a reduced carbonaceous residue on the ceria-containing sample. This trend is also reinforced by the CH₄-TPD runs which gave heat of desorption values of 82 and 46 kJ mol⁻¹ for the unpromoted and promoted catalysts as may be seen on Table 3. The somewhat higher surface capacity for CH₄ adsorption on the promoted catalyst (19.5 μmol m⁻²) than the Co–Ni catalyst (11.4 μmol m⁻²) suggests the possibility of additional CH₄ chemisorption on ceria sites. Furthermore, the ratio of surface site concentration between CH₄ and C₃H₈ on the unpromoted catalyst is 3.35 (i.e., 11.4/3.4) indicating that C₃H₈ may reasonably fragment into three CH_y adspecies (not necessarily in a single step) on the catalyst surface. However, the enhanced adsorption capacity due to ceria addition is also consistent with a surface site ratio (between CH₄ and C₃H₈) of about 5 (19.5/3.8 = 5.13) for the promoted catalyst. Perhaps the ceria phase offered adsorption centers that are more suitably sized for CH₄ than C₃H₈ (since CH₄ is smaller at 40 nm than the C₃H₈ molecule at 49 nm). Interestingly, an increase in the surface adsorption capacity for CH₄ as compared to C₃H₈ (ca. 70%) in the promoted catalyst corresponds reasonably well with the area coverage based on the species size. These attributes foreshadow superior carbon resistance character due to ceria promotion.

Oxidative reforming runs: effect of feed CO₂:C₃H₈ ratio ($R_{{{\text{CO}}_{2} }}$)

Although several studies have focused on the effect of catalyst on dry reforming [12], there is a paucity of information on the kinetic implications of the rate data and plausible empirical models. The latter may be culled from the commercially available modeling software following optimized data regression. Figure 3a shows the rate behavior (for H₂, CH₄ and C₃H₈) as a function of the principal reactant ratio, $R_{{{\text{CO}}_{2} }}$ in the feed while the oxygen content was kept constant at $R_{{{\text{O}}_{2} }}$ = 0.75 for each run. Since the stoichiometric feed CO₂:C₃H₈ ratio required by Eq. (1b) is 1.5 (i.e., 3-2γ), it is evident that the range of $R_{{{\text{CO}}_{2} }}$ employed (0–9) included sub- and supra-stoichiometric conditions. All three species displayed a hyperbolic rate decrease with $R_{{{\text{CO}}_{2} }}$ suggesting a pseudo-first-order dependency on feed composition. The initial high H₂ production may be attributed strong dissociative C₃H₈ chemisorption on metal sites at the high C₃H₈ partial pressure (corresponding to low $R_{{{\text{CO}}_{2} }}$ values). The Ni-support boundary contains acid sites where hydrogen abstraction from the alkane molecule can take place producing C_xH_1-x type unsaturated carbonaceous species. The similarity in the $R_{{{\text{CO}}_{2} }}$ dependency of H₂, CH₄ rate curves implicates the direct formation of these species from C₃H₈ adsorption. Thus, the appropriate empirical rate equation may be given as:

$$r_{i} = \frac{{r_{i}^{0} }}{{\left[ {1 + K_{\text{oxi}} R_{{{\text{CO}}_{2} }} } \right]}}$$

(4)

where the numerator, $r_{i}^{0}$, corresponds to the consumption or net production rate of these species (i = H₂, CH₄ and C₃H₈) for the direct partial oxidation of propane at $R_{{{\text{CO}}_{2} }}$= 0 (with $R_{{{\text{O}}_{2} }}$ = 0.75) while K_oxi is an empirical constant. Although Eq. (4) was not derived from formal mechanistic considerations, it is structurally similar to a conventional Langmuir–Hinshelwood rate model where the denominator is the sum of adsorption terms which constitutes the resistance to reaction and the numerator, $r_{i}^{0}$, is the driving force and depends on $R_{{{\text{O}}_{2} }}$ (i.e., the O₂ partial pressure) in this case. Indeed, K_oxi may be regarded as the ratio of adsorption equilibrium constants for CO₂ and C₃H₈ since both are implicated in the feed variable, $R_{{{\text{CO}}_{2} }}$. Estimates of the parameters in Eq. (4) based on the nonlinear regression of the rate data (using SigmaPlot 10 modeling software) are summarized in Table 4.

Fig. 3

a Rate profiles for H₂ and CH₄ production and C₃H₈ consumption during oxidative reforming using $R_{{{\text{O}}_{2} }}$ of 0.75 at 873 K. b Rate profiles for CO production and CO₂ consumption during oxidative reforming using $R_{{{\text{O}}_{2} }}$ of 0.75 at 873 K

Table 4 Estimates of parameters for Eq. (4) for describing the rate profiles for H₂, CH₄ and C₃H₈ during oxidative dry reforming

In Fig. 3b, both CO production and CO₂ consumption curves revealed a turning point at $R_{{{\text{CO}}_{2} }}$ between 2 and 3, which is above the stoichiometric value of 1.5 indicating that CO₂ is not only less strongly adsorbed than C₃H₈ on the catalyst, but was also involved in CO formation. The adsorption of CO₂ on weak basic sites is evidenced from Table 3. These sites are present on both unpromoted and ceria-containing (0.35 μmol m⁻² _BET) catalysts in a ratio of about 1:2. The higher heat of desorption on the promoted catalyst is consistent with the formation of a bidentate carbonate species on the ceria catalyst where surface oxygen species most likely enhanced CO₂ adsorption [26]. The bidentate species would react with surface carbon during dry reforming yielding lower carbon formation compared to the unpromoted Co–Ni catalyst. The separate ceria phase is evident from X-ray diffractogram. The weak adsorption of CO₂ would require much higher gas phase CO₂ partial pressure to ensure favorable coverage in order to compensate for the stronger chemisorption of C₃H₈ for the surface reaction to proceed at reasonable rate. This would account for the appearance of the CO and CO₂ rate maxima beyond the stoichiometric value ($R_{{{\text{CO}}_{2} }}$ = 1.5) during reforming. As a result, the rate behavior for both CO production and CO₂ consumption may be captured by the unimodal relation:

$$r_{j} \left[ {{\text{or}} \left( { - r_{j} } \right)} \right] = \frac{{[r_{j}^{0} + k_{1} R_{{{\text{CO}}_{2} }} ]}}{{[1 + k_{2} R_{{{\text{CO}}_{2} }} ]^{2} }} {\text{j }} = {\text{CO or CO}}_{ 2}$$

(5)

where k₁ is a pseudo-kinetic constant and k₂ is combination of adsorption equilibrium constants in the steps for CO formation and CO₂ utilization rates, while $r_{j}^{0}$ is the rate for species j when $R_{{{\text{CO}}_{2} }}$  = 0 (i.e., at $R_{{{\text{O}}_{2} }}$ = 0.75). Nonlinear parameter estimation provided the values displayed in Table 5.

Table 5 Estimates of parameters for Eq. (5) for the CO and CO₂ rate profiles during oxidative dry reforming

Spectroscopic studies ([25]; [24]) have shown that propane dissociatively chemisorbs on metal sites yielding unsaturated carbonaceous species, C_xH_1-x, (or atomic C) and H adatoms. Moreover, both H₂ and CH₄ may further react with CO₂ and O₂ contributing to the overall exponential decrease in their respective production rates with increasing $R_{{{\text{CO}}_{2} }}$ value. Even so, the declining CO production rate with increasing CO₂ partial pressure ($R_{{{\text{CO}}_{2} }}$> 3) indicates a reduction in the surface concentration of the precursor carbonaceous species during reforming. The reactivity of the carbonaceous species is also dependent on its composition (value of x) and age on the catalyst surface as observed by Hardiman et al. [14]. The CO₂ consumption rate profile (cf. Fig. 3b) has been corrected for the CO₂ supplied to the reactor. The slow drop in its utilization rate after $R_{{{\text{CO}}_{2} }}$> 3 may be associated with the low concentration of surface carbonaceous species occasioned by the reduced C₃H₈ partial pressure as well as catalyst activity loss due to metal site re-oxidation at high $R_{{{\text{CO}}_{2} }}$ and in the presence of O₂ via

$${\text{C}}_{3} {\text{H}}_{8} + 5{\text{CO}}_{2} + \frac{{\left( {z - 1} \right)}}{2}{\text{O}}_{2} + {\text{Ni}} - {\text{Co}} = 8{\text{CO}} + 4{\text{H}}_{2} + {\text{NiO}} - {\text{CoO}}_{z}$$

(6)

where Ni–Co is the reduced metal catalyst and NiO–CoO_z is the oxidized catalyst (0 ≤ z ≤ 1.5) to accommodate the multiple oxidation states of Co. Indeed, the formation of NiCo₂O₄ and Co₃O₄ (NiO–CoOz for z = 1.5 and 1.33, respectively) is consistent with thermodynamics and the XRD phases observed in Fig. 2. Interestingly, CH₄ which may also be formed from the hydrogenation of the C_xH_1-x surface precursor reached nearly negligible rates at $R_{{{\text{CO}}_{2} }}$ ≥ 5 suggesting minimal surface carbonaceous coverage at high CO₂ partial pressure. Figure 4 illustrates the product ratio behavior with respect to the feed composition. It is apparent that the H₂:CO curve approached an asymptotic limit of 0.48 at $R_{{{\text{CO}}_{2} }}$ ≥ 5. This limiting value is in agreement with an estimate of 0.5 implicated in Eq. (6). Consistent with the preceding discussion, the CH₄:CO and CH₄:H₂ ratio are also decay curves. Hence, the product ratio profile may be captured by the 3-parameter hyperbolic decay model:

$${\text{PR}}_{ij} = {\text{PR}}_{ij}^{\infty } + \frac{{\alpha_{ij} }}{{\left[ {1 + \beta_{ij} R_{{{\text{CO}}_{2} }} } \right]}}$$

(7)

where PR_ij is the product ratio of species, i to j, $PR_{ij}^{\infty }$ is the product ratio of i to j at infinite (high) CO₂ partial pressure (i.e., $R_{{{\text{CO}}_{2} }}$≥ 5), with α_ij and β_ij as empirical rate constants associated with the formation of species i and j. These estimates are supplied in Table 6. Since $PR_{ij}^{\infty }$ is zero for both CH₄:CO and CH₄:H₂, it may be considered as the contribution from the non-catalytic gas–solid reaction described in Eq. (6). Nonetheless, at $R_{{{\text{CO}}_{2} }}$= 0, the partial oxidation of C₃H₈ would have proceeded with $R_{{{\text{O}}_{2} }}$= 0.75 which is smaller than the stoichiometric requirement of 1.5, obtained from the reaction:

Fig. 4

Product ratio as a function of the feed composition under oxidative dry reforming of propane at 873 K with $R_{{{\text{O}}_{2} }}$ of 0.75

Table 6 Parameter estimates for Eq. (7) for the product selectivity ratio variation with respect to $R_{{{\text{CO}}_{2} }}$ during oxidative dry reforming

$${\text{C}}_{3} {\text{H}}_{8} + \frac{3}{2}{\text{O}}_{2} = 3{\text{CO}} + 4{\text{H}}_{2}$$

(8)

Consequently, an H₂:CO ratio of 1.25 obtained from Fig. 4 suggests that the ceria promoter may have served as a source of supplementary oxygen and, hence, its closeness to the theoretical value of 1.33 based on Eq. (8). The participation of lanthanide (L) promoters in redox reactions, L⁴⁺/L³⁺, for C_xH_1-x consumption during dry reforming has been proposed by [8], namely

Reducing step:

$${\text{L}}_{a} {\text{O}}_{b} + {\text{C}}_{x} {\text{H}}_{1 - x} \leftrightarrow {\text{L}}_{a} {\text{O}}_{b - x} + x{\text{CO}} + \frac{{\left( {1 - x} \right)}}{2}{\text{H}}_{2}$$

(R1)

Re-oxidizing step:

$${\text{L}}_{a} {\text{O}}_{b - x} + x{\text{CO}}_{2} \leftrightarrow {\text{L}}_{a} {\text{O}}_{b} + x{\text{CO}}$$

(R2a)

or

$${\text{L}}_{a} {\text{O}}_{b - x} + \frac{x}{2}{\text{O}}_{2} \leftrightarrow {\text{L}}_{a} {\text{O}}_{b}$$

(R2b)

where a = 1 or 2 and b = 2 or 3.

In view of these observations, the key steps during oxygen-assisted dry reforming may be proposed as the initial dehydrogenation of propane to H₂, atomic surface carbon, C, surface carbonaceous deposit, C_xH_1-x which undergoes hydrogenation to CH₄ followed by simultaneous oxidation of H₂, CH₄ and the residual surface C_xH_1-x species to water, CO and H₂. However, as the CO₂ partial pressure increased, reaction between deposited carbon and CO₂ would yield improved CO production. The additional oxidation of CO to CO₂ by oxygen is possible but has not been explicitly listed in view of the high partial pressure of CO₂ as $R_{{{\text{CO}}_{2} }}$ increased. Consequently, the following reactions may be posited:

$${\text{C}}_{3} {\text{H}}_{8} \to \frac{2}{x}{\text{C}}_{x} {\text{H}}_{1 - x} + {\text{C}} + \left( {\frac{5x - 1}{x}} \right){\text{H}}_{2} 0\, < \,x\, < \, 1$$

(9)

$${\text{C}}_{x} {\text{H}}_{1 - x} + \frac{(5x - 1)}{2}{\text{H}}_{2} \to x{\text{CH}}_{4}$$

(10)

$${\text{C}} + \frac{1}{2}{\text{O}}_{2} \to {\text{CO}}$$

(11)

$${\text{C}} + {\text{CO}}_{2} \to 2{\text{CO}}$$

(12)

$${\text{C}}_{x} {\text{H}}_{1 - x} + \frac{x}{2}{\text{O}}_{2} \to x{\text{CO}} + \frac{{\left( {1 - x} \right)}}{2}{\text{H}}_{2}$$

(13)

$${\text{C}}_{x} {\text{H}}_{1 - x} + x{\text{CO}}_{2} \to 2x{\text{CO}} + \frac{{\left( {1 - x} \right)}}{2}{\text{H}}_{2} .$$

(14)

$${\text{H}}_{2} + \frac{1}{2}{\text{O}}_{2} \to {\text{H}}_{2} {\text{O}}$$

(15)

$${\text{H}}_{2} + {\text{CO}}_{ 2} \to {\text{CO}} + {\text{H}}_{2} {\text{O}} .$$

(16)

It has been assumed that CH₄ also goes through successive dehydrogenation similar to steps (9) to (14) and may be summarized as:

$${\text{CH}}_{4 } + {\text{CO}}_{2} \to 2{\text{CO}} + 2{\text{H}}_{2}$$

(17)

Dry partial oxidation runs: effect of oxygen co-feeding ($R_{{{\text{O}}_{2} }}$)

The influence of O₂ addition during dry reforming was studied using a feed containing CO₂:C₃H₈ ratio, $R_{{{\text{CO}}_{2} }}$ = 1.5. The experimental range for the O₂:C₃H₈ ratio, $R_{{{\text{O}}_{2} }}$, was 0–1.5 to avoid conditions for total hydrocarbon oxidation [cf. Eqn (8)]. Figure 5a shows the reaction rate envelopes for CO₂ and CO while the profiles for H₂, CH₄ and C₃H₈ are displayed in Fig. 5b. As may be seen in Fig. 5a, CO production rate peaked at $R_{{{\text{O}}_{2} }}$ of about 0.75 while CO₂ consumption rate was initially flat but dropped at $R_{{{\text{O}}_{2} }}$ > 0.75. However, in Fig. 5b, all three products appear to crest around the same feed O₂ partial pressure, $R_{{{\text{O}}_{2} }}$= 0.75 implicating similar rate-controlling steps. It would, therefore, seem that O₂ and C₃H₈ most likely chemisorb on similar sites, the electron-deficient metal–support interface centers. Indeed, the left skew of the rate-composition profiles indicates that O₂ was more strongly adsorbed than the hydrocarbon. The experimental heat of desorption for C₃H₈ was 74 kJ mol⁻¹ in this study compared with 200 kJ mol⁻¹ on Ni for O₂ [22].

Fig. 5

a Reaction rate profiles for CO and CO₂ under dry partial oxidation using $R_{{{\text{CO}}_{2} }}$ of 1.5 at 873 K. b Reaction rate profiles for H₂, C₃H₈ and CH₄ under dry partial oxidation using $R_{{{\text{CO}}_{2} }}$ of 1.5 at 873 K

The intercept on the rate-axis corresponding to the reaction rate value for pure dry reforming using $R_{{{\text{CO}}_{2} }}$ = 1.5 suggests that the introduction of O₂ initially enhanced product formation rate as well as C₃H₈ consumption (with concurrent dry reforming) most likely through additional oxygen gasification of the surface carbon moieties [cf. Eqn (13)]. Alenazey et al. (2011) have demonstrated that O₂ is a more aggressive carbon gasifier than CO₂ and, hence, the steady rise in H₂, CH₄, C₃H₈ and CO rate as carbon-ladden metal sites was re-oxidized to the metal atoms, thereby improving catalyst activity in the region, 0 ≤ $R_{{{\text{O}}_{2} }}$≤ 0.75. Nevertheless, at $R_{{{\text{O}}_{2} }}$ > 0.75, continuing oxidation probably caused metal site oxide formation resulting in a loss of catalyst activity which combined with a concomitant lowering of C₃H₈ partial pressure to reduce surface concentration of C_xH_1-x and the associated decline in species formation rate in this region, 0.75 ≤ $R_{{{\text{O}}_{2} }}$≤ 1.5. The earlier plateau in the CO₂ consumption rate curve indicates that carbon gasification via Eq. (14) was taking place at the constant $R_{{{\text{CO}}_{2} }}$ value employed albeit the increased O₂ partial pressure was becoming a more significant contributor to species formation until at $R_{{{\text{O}}_{2} }}$ > 0.75; CO₂ became rather uncompetitive with O₂ as the key oxidant for surface carbonaceous species. Hence, the drop in CO₂ consumption rate in the upper range was 0.75 ≤ $R_{{{\text{O}}_{2} }}$≤ 1.5.

As previously alluded to, the appearance of a maximum in the rate envelopes for H₂, CH₄, C₃H₈ and CO suggests the applicability of an empirical kinetic model similar to Eq. (5), namely

$$r_{j} \left[ {{\text{or}} \left( { - r_{j} } \right)} \right] = \frac{{[r_{j}^{0} + k_{3} R_{{{\text{O}}_{2} }} ]}}{{[1 + k_{4} R_{{{\text{O}}_{2} }} ]^{2} }} \quad j \, = {\text{ H}}_{2} ,{\text{ CH}}_{4} ,{\text{ C}}_{3} {\text{H}}_{8} \,{\text{and}}\,{\text{CO}}$$

(18)

where, as before, k₃ is a pseudo-kinetic constant and k₄ is a combination of adsorption equilibrium constants in the steps for each of H₂, CH₄, C₃H₈ and CO production, while $R_{{{\text{CO}}_{2} }}$  = 1.5. Table 7 contains the associated parameter estimates from the nonlinear regression fit of the data. The relatively low correlation coefficient values for H₂ and CH₄ (0.87 and 0.92, respectively) may be due to the inability of the model to comprehensively capture the secondary interaction between these two species, especially at near zero $R_{{{\text{O}}_{2} }}$ values where secondary H₂ production from further CH₄ dehydrogenation was probably somewhat more dominant than reaction of either with O₂. Even so, this is not a significant impairment to the model fidelity.

Table 7 Parameter estimates for Eq. (18) for species reaction rate profiles during dry partial oxidation

The CO₂ rate profile with an initial plateau followed by a decline, however, is a characteristic inverted logistic behavior (the S-curve occurs frequently in natural processes where performance seemed stalled or latent at the beginning before an exponential acceleration/deceleration until maturity or termination eventuates (such as in complex learning/adaptation systems). These features capture the catalyst performance progression from a substantially dry reforming condition to the predominantly oxidation stage. Thus, the CO₂ rate data were fitted to the 3-parameter logistic model given by

$$( - R_{{{\text{CO}}_{2} }} ) = \frac{{r^{\rm{max} } }}{{[1 + \eta (R_{{{\text{o}}_{2} }} )^{v} ]}}$$

(19)

where r^max is the maximum value for CO₂ consumption rate (during pure dry reforming, i.e., $R_{{{\text{O}}_{2} }}$= 0 and $R_{{{\text{CO}}_{2} }}$  = 1.5) while η and υ are model constants. Nonlinear regression of the rate data provides r^max = 5.05 × 10⁻⁷ mol m⁻² s⁻¹, η = 0.58 and υ = 5.14 with a correlation coefficient of 0.991 implicating a good fit of the rate model.

Figure 6 illustrates the plots for product ratio trend with respect to the feed oxygen variable, $R_{{{\text{O}}_{2} }}$. Unlike the data in Fig. 4, where the H₂:CO ratio decreased almost exponentially with CO₂ partial pressure, Fig. 6 shows that this reaction index experienced an initial decrease going through a minimum at $R_{{{\text{O}}_{2} }}$ = 0.5 before an upturn. Although both H₂ and CO rates exhibited an upward trend in the region, 0 ≤ $R_{{{\text{O}}_{2} }}$ ≤ 0.5, it would seem that CO production contributed via Eqs. (11) & (12) which were unaccompanied by stoichiometric H₂ production steps such as Eqs. (13) & (14) played a significant role leading to a declining trend in the H₂:CO ratio. However, a change in the rate-controlling step(s) for CO production is signaled at the minimum point where the reduction in coverage or absence of surface carbon atom, C, probably shut down CO contribution from Eqs. (11) & (12). Even so, H₂ and CO formation still continued via other steps such that the H₂:CO ratio now increased almost linearly over the range, 0.5 ≤ $R_{{{\text{O}}_{2} }}$≤ 1.5. Clearly, within the window, 0 ≤ $R_{{{\text{O}}_{2} }}$≤ 0.5, C₃H₈ partial pressure would be higher than in the upper O₂ range, 0.5 ≤ $R_{{{\text{O}}_{2} }}$≤ 1.5, making it more realistic that propane decomposition via Eq. (9) which produces atomic C as accompaniment to C_xH_1-x and H₂ would be more likely than the alternative decomposition step operative in the region, 0.5 ≤ $R_{{{\text{O}}_{2} }}$≤ 1.5, namely

$${\text{C}}_{3} {\text{H}}_{8} \to \frac{2}{x}{\text{C}}_{x} {\text{H}}_{1 - x} + {\text{CH}}_{4} + (1 + x){\text{H}}_{2} .$$

(20)

where C is not formed. Equations (9) and (20) also suggest that the empirical composition of the C_xH_1-x adspecies depends, in part, on the hydrocarbon feed composition. This has also been experimentally corroborated by Hardiman et al. [14]. Thus, the H₂:CO product ratio over the entire $R_{{{\text{O}}_{2} }}$ range may be captured by the linear combination of the two regimes as:

$${\text{PR}}_{i;j} = \frac{{\alpha_{ij} }}{{(1 + \beta_{ij} R_{{{\text{o}}_{2} }} )}} + \lambda_{ij} R_{{{\text{o}}_{2} }}$$

(21)

where i:j = H₂:CO, α_ij, β_ij and λ_ij are non-negative model parameters. We note that α_ij is the H₂:CO ratio for pure (O₂-free feed) dry reforming using $R_{{{\text{CO}}_{2} }}$= 1.5. Estimates of these model constants based on data regression to Eq. (21) are provided in Table 8.

Fig. 6

Product ratio profiles for dry partial oxidation using $R_{{{\text{CO}}_{2} }}$ of 1.5 at 873 K

Table 8 Parameter estimates for Eq. (22) for the product selectivity ratio variation with respect to $R_{{{\text{O}}_{2} }}$ during dry partial oxidation

The curves for CH₄:CO ratio and CH₄:H₂ ratio shown in Fig. 6 are nearly parallel to peaks co-located at $R_{{{\text{O}}_{2} }}$= 0.5. Although differing in magnitude, the observed trends for the two profiles invite the application of a common nonlinear model:

$${\text{PR}}_{i:j} = \frac{{[{\text{PR}}_{ij}^{0} + \alpha_{ij} R_{{{\text{O}}_{2} }} ]}}{{[1 + \beta_{ij} R_{{{\text{O}}_{2} }} ]^{2} }} \quad i, \, j \, = {\text{ H}}_{ 2} ,{\text{ CH}}_{ 4} \,{\text{and }}\,{\text{CO}}$$

(22)

with ${\text{PR}}_{ij}^{0}$ as the product ratio of i to j at $R_{{{\text{O}}_{2} }}$= 0 (i.e., pure dry reforming using $R_{{{\text{CO}}_{2} }}$= 1.5), while α_ij and β_ij are empirical constants associated with the formation rate of species i and j. Regression analysis of the product selectivity ratio data yielded the parameter estimates in Table 8.

Figure 7 illustrates the Arrhenius plots for the reaction specs. Activation energy values for H₂, CO, CO₂ and CH₄ were obtained as 30.2, 30.5, 45.3 and − 127 kJ mol^−1, respectively. The similarities in magnitude of the activation energy estimates for H₂, CO and CO₂ suggest that their production may not be the rate-determining step in the complex reaction network. However, the negative activation energy for CH₄ confirms its linkage with the carbon deposition step [cf. Eqn (9)]. Bartholomew [4] found that carbon deposition rate (from alkanes) decreased with increasing temperature in the range 773–873 K. As a result, CH₄ formation from Eqs. (9) to (10) would be characterized by negative activation energy as obtained in this case. Indeed, Hardiman et al. [14, 15] have reported a similar observation in their steam reforming investigation at low feed steam:propane ratio. Eqn (9) would, therefore, appear to be one of the kinetically dominant reaction steps in this study. Even so, it is instructive that all rate data used have been checked for the absence of negligible transport resistances [7].

Fig. 7

Arrhenius plots for the reaction species using $R_{{{\text{CO}}_{2} }}$ of 4.0 and $R_{{{\text{O}}_{2} }}$ of 1.5

Comparison between the oxygen-containing runs with pure dry reforming

Figure 8a, b shows the reaction rate profiles under conventional dry reforming conditions. While H₂ and CH₄ production as well as C₃H₈ utilization rates decreased exponentially with increased CO₂ partial pressure, the rate profiles for CO, CO₂ are characterized by an optimum at $R_{{{\text{CO}}_{2} }}$= 3 suggesting conformity with stoichiometric requirements. However, the rate values at $R_{{{\text{CO}}_{2} }}$= 0 suggest that the direct decomposition of C₃H₈ produced only H₂ and CH₄ with practically no CO formation. This indicates that ceria participation as an oxygen source was not significant suggesting that the redox reaction involving the promoter is triggered only in the presence of a gas phase oxygen carrier (such as CO₂ or O₂) as described by redox equations (R1) to (R2). Furthermore, a CH₄:C₃H₈ ratio of 0.58 during decomposition implicates substantial carbon deposit on the catalyst surface. As may be expected, this ratio decreased with increased CO₂ partial pressure as later substantiated by carbon content analysis of the used catalyst. Not surprisingly, the C₃H₈, H₂ and CH₄ rate dependency on $R_{{{\text{CO}}_{2} }}$ was captured by Eq. (4) where K_oxi has been substituted for K_DR in the denominator. However, the unimodal behavior describing the CO and CO₂ rate was adequately depicted by Eq. (5). The relevant parameter estimates from regression analysis are provided in Tables 9 and 10. Figure 9 also displays the product selectivity ratio under pure dry reforming conditions. The hyperbolic decay pattern is similar to what was observed during oxidative dry reforming in Fig. 4. As a result, the 3-parameter hyperbolic decay model given by Eq. (7) was also used to fit the data. The resulting parameter estimates are listed in Table 11.

Fig. 8

a Reaction rate profiles during pure dry reforming at 873 K. b Reaction rate profiles during pure dry reforming at 873 K

Table 9 Estimates of parameters for Eq. (4) based on the rate profiles for H₂, CH₄ and C₃H₈ during pure dry reforming

Table 10 Estimates of parameters for Eq. (5) using the CO and CO₂ rate data for pure dry reforming

Fig. 9

Product ratio profiles during pure dry reforming at 873 K

Table 11 Parameter estimates for Eq. (22) for the product selectivity ratio variation with respect to $R_{{{\text{CO}}_{2} }}$ during pure dry reforming

Based on the foregoing discussion, the introduction of a specified O₂ dosage into the C₃H₈/CO₂ feed mixture enhanced catalyst performance through increased carbon gasification (with O₂), presumably via the same rate-controlling step(s) applicable during pure dry reforming. However, the reaction behavior during dry partial oxidation was influenced by the stronger O₂ chemisorption and, hence, the reduced surface coverage of C_xH_1-x species needed for product formation resulted in unimodal rate profiles for all reaction products.

Although the three sets of experiments reported in this study have employed different feed conditions (O₂-dosed dry reforming, CO₂-dosed partial oxidation and pure dry reforming), the participation of common reacting species, namely CO₂, C₃H₈, O₂, H₂ and CH₄ over the same catalyst allows some important comparisons to be made with respect to role of feed CO₂:O₂ ratio in determining the relative significance of steps involved in oxygen-assisted hydrocarbon reforming, trend and magnitude of product selectivity ratio as well as possible switching of rate-controlling steps. It is apparent from Figs. 10a, b which pool together the rate data for oxidative dry reforming and dry partial oxidation that the individual species rate has a discernible dependency on the CO₂:O₂ feed ratio implying that the reaction pathway illustrated by Eqs. (9)–(17) provides a plausible account of product formation even if different rate-limiting steps may be necessary for individual components.

Fig. 10

a Reaction rate envelope as a function of the feed CO₂:O₂ ratio across the oxidative dry reforming and dry partial oxidation conditions at 873 K. b Reaction rate envelope as a function of the feed CO₂:O₂ ratio across the oxidative dry reforming and dry partial oxidation conditions at 873 K

The rapid decay of the H₂ formation rate with increasing CO₂:O₂ feed ratio is consistent with the consumption of initial H₂ production (via C₃H₈ decomposition on the reduced metal sites) with both CO₂ and O₂ as required by Eqs. (15)–(16). On the other hand, the profiles for both C₃H₈ and CH₄ in Fig. 10a exhibit peaks at about CO₂:O₂ feed ratio of about three implicating similarity in rate-controlling steps for these two species. In particular, the existence of a peak in the rate profile means that two surface entities were involved in rate-determining step. In the case of C₃H₈, this may be conjectured as the two vacant metal sites required for dissociative chemisorption. For CH₄ formation, the rate-limiting step would be the surface hydrogenation of C_xH_1-x species. By same token, the CO and CO₂ rate envelopes in Fig. 10b suggest a commonality between the controlling steps for CO production and CO₂ utilization with both curves having their individual maximum at CO₂:O₂ feed ratio = 3.

Product selectivity ratio considerations also reveal that the H₂:CO ratio has a relatively small variability (0.72–0.96 with an average value of 0.81) for the dry partial oxidation runs than the wide variation observed (0.48–1.25 with an average of 0.68) for the oxidative dry reforming. This would suggest that the dry partial oxidation mode of operation should be more suited for syngas produced as feed to olefin manufacturing Fischer–Trospch plants. In general, however, H₂ and CO production rates were about the same magnitude for both types of operation with the dry partial oxidation exhibiting milder variation than the oxidative dry reforming. Interestingly, the pure dry reforming H₂ and CO production rates too are in the same size range as the two forms of O₂-assisted dry reforming modes. Nonetheless, the DPOX runs have superior heat utilization as thermodynamics calculations indicate.

Post-reaction carbon analysis

The residual carbon content on the used catalysts from the DPOX runs is plotted in Fig. 11. The high total organic carbon (TOC) content registered in the low $R_{{{\text{O}}_{2} }}$ region is consistent with the previous discussion regarding high propane dehydrogenation to surface C_xH_1-x at low $R_{{{\text{O}}_{2} }}$. The sigmoid exhibited a rapid drop in region 0.5 ≤ $R_{{{\text{O}}_{2} }}$ ≤ 1.25 which incidentally corresponds to the zone where maximum reaction rate occurred. It would seem that a more resistant carbonaceous deposit was formed at low $R_{{{\text{O}}_{2} }}$ while under O₂-rich conditions a more reactive carbon was present. Hardiman et al. [15] conducted temperature-programmed oxidation–reduction experiments to characterize coke on used catalysts from propane steam reforming and confirmed the existence of both naphthalenic and aliphatic C_xH_1-x species on the used samples. A similar observation was also reported during propane dry reforming over unpromoted Co–Ni catalyst [3]. The left-sided S-curve seen in Fig. 11 is a reflection of the different reactivities with oxygen for these two forms of carbonaceous pool. Even so, the S-shaped TOC(%) profile across the range of $R_{{{\text{O}}_{2} }}$ values used in this study may be captured by the 3-parameter sigmoid expression:

$${\text{TOC}} = \frac{{A_{\text{TOC}} }}{{\left\{ {1 + e^{{\rho \left( {R_{{{\text{O}}_{2} }} - R_{{{\text{O}}_{2} }}^{\text{crit}} } \right)}} } \right\}}} .$$

(23)

where A_TOC is the catalyst capacitance for carbon deposition, while ρ is the oxygen-induced anti-coking factor for the catalyst and $R_{{{\text{O}}_{2} }}^{\text{crit}}$ is the critical O₂:C₃H₈ ratio when TOC is at 50% of A_TOC (and it is the point at which the transition occurs from one type of coke to another in this case). Nonlinear regression of the data gave parameter estimates for A_TOC, ρ and $R_{{{\text{O}}_{2} }}^{\text{crit}}$ as: 85.61%, 4.17, and 0.93, respectively. Alenazey et al. [2] have shown that the value of $R_{{{\text{O}}_{2} }}^{\text{crit}}$. depends on the type of carbon gasification agent (where O₂ may be replaced by the required gasifying agent) used for the coke removal.

Fig. 11

TOC results for spent ceria-promoted catalysts from dry partial oxidation runs carried with $R_{{{\text{CO}}_{2} }}$ of 1.5 at 873 K

Conclusions

For the first time, a study of the oxygen-assisted dry reforming of propane has been carried out using Ce-promoted Co–Ni/Al₂O₃. Catalyst characterization revealed that the common physicochemical properties (BET area, pore volume, H₂ chemisorption) of the Ce-promoted and undoped catalysts were essentially the same with the exception of the increase in surface basicity (as measured by NH₃ and CO₂ temperature-programmed desorption) of the ceria-containing sample. XRD pattern confirmed the presence of a CeO₂ phase in the latter catalyst. Dry partial oxidation (CO2-dosed partial oxidation) appeared to be superior to both oxidative dry reforming and pure dry reforming in terms of the value and uniformity in H₂:CO product ratio if the syngas is to be employed for olefin or GTL manufacture. Reaction runs with varying oxygen partial pressure showed that even with a low CO₂:C₃H₈ ratio ($R_{{{\text{CO}}_{2} }}$ = 1.5) where carbon deposition was significant, the addition of O₂ improved the H₂:CO ratio in the product stream above that obtained for pure dry reforming. Furthermore, analysis of the TOC content in the used catalysts points to the beneficial effects of autodry reforming operation in terms of significant reduction in carbon residue and, thus, the increased longevity of the catalyst. Kinetics of the product formation may be classified into two generic types, namely a hyperbolic decay dependency on $R_{{{\text{CO}}_{2} }}$ (for H₂, CH₄ and C₃H₈) and a Langmuir–Hinshelwood type relation (for CO and CO₂) during oxidative and pure dry reforming reactions. The rate behavior of participating species during dry partial oxidation was determined by different rate-limiting steps since the form of the Langmuir–Hinshelwood rate model involves the interaction of two surface entities in the rate-controlling step (an exponent of two for the denominator).