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.
Similar content being viewed by others
Explore related subjects
Discover the latest articles, news and stories from top researchers in related subjects.Avoid common mistakes on your manuscript.
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
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:
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
for 373 ≤ T ≤ 1073 K where Rg 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:
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].
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 Kα 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 = 31o and 36.6o) and Co3O4 at 2θ = 31.2o and 36.7o while NiO is located at 2θ = 43.2 o. The metal aluminates, CoAl2O4 (2θ = 36.7o) as well as NiAl2O4 (2θ = 37o; 44.8o), exhibited strong intensities but the CeO2 was identified as a separate oxide phase on the ceria-containing catalyst at 2θ = 28.6o. 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.
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, Ed, 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.
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 CHy 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 CxH1-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:
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 Koxi 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, Koxi 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.
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:
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.
Spectroscopic studies ([25]; [24]) have shown that propane dissociatively chemisorbs on metal sites yielding unsaturated carbonaceous species, CxH1-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
where Ni–Co is the reduced metal catalyst and NiO–CoOz 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 CxH1-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:
where PRij 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 αij and βij 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:
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 CxH1-x consumption during dry reforming has been proposed by [8], namely
Reducing step:
Re-oxidizing step:
or
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, CxH1-x which undergoes hydrogenation to CH4 followed by simultaneous oxidation of H2, CH4 and the residual surface CxH1-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:
It has been assumed that CH4 also goes through successive dehydrogenation similar to steps (9) to (14) and may be summarized as:
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].
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 CxH1-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
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.
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
where rmax 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 rmax = 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 CxH1-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
where C is not formed. Equations (9) and (20) also suggest that the empirical composition of the CxH1-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:
where i:j = H2:CO, αij, βij and λij are non-negative model parameters. We note that αij 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.
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:
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 αij and βij 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].
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 Koxi has been substituted for KDR 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.
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 CxH1-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.
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 CxH1-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 CxH1-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 CxH1-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:
where ATOC 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 ATOC (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 ATOC, 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.
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).
References
Adesina AA (2012) The role of CO2 in hydrocarbon reforming catalysis: friend or foe? Curr Opin Chem Eng 1:272–280
Alenazey F, Cooper CG, Dave CB, Elnashaie SSEH, Susu AA, Adesina AA (2009) Coke removal from deactivated Co–Ni catalyst using different gasifying agents: an analysis of the gas–solid reaction kinetics. Catal Commun 10:406–411
Althenayan FM, Foo SY, Kennedy EM, Dlugogorski BZ, Adesina AA (2010) Bimetallic Co–Ni/Al2O3 catalyst for propane dry reforming: estimation of reaction metrics from longevity runs. Chem Eng Sci 65:66–73
Bartholomew CH (2001) Mechansims of catalyst deactivation. Appl Catal A 212:17–60
Buyeskaya OV, Wolf D, Baerns M (2000) Ethylene and propene by oxidative dehydrogenation of ethane and propane: `performance of rare-earth oxide-based catalysts and development of redox-type catalytic materials by combinatorial methods’. Catal Today 62:91–99
Emeis CA (1993) Determination of integrated molar extinction coefficients for infra red absorption bands of pyridine adsorbed on solid acid catalysts. J Catal 141:347–354
Fogler HS (1999) Elements of chemical reaction engineering. Prentice-Hall, Englewood
Foo SY, Cheng CK, Nguyen TH, Adesina AA (2011) Evaluation of lanthanide-group promoters on Co–Ni/Al2O3 catalysts for CH4 dry reforming. J Mol Catal A 344:28–36
Foo SY, Chen CK, Nguyen TH, Kennedy EM, Dlugogorski BZ, Adesina AA (2012) Carbon deposition and gasification kinetics of used lanthanide-promoted Co–Ni/Al2O3 catalysts from CH4 dry reforming. Catal Commun 26:183–188
Gao J, Guo J, Liang D, Hou Z, Fei J, Zheng X (2008) Production of syngas via autothermal reforming of methane in a fluidized bed reactor over CeO2–ZrO2/SiO2 supported Ni catalysts. Int J Hydrogen E 33:5493–5500
Glockler B, Gritsch A, Morillo A, Kolios G, Eigenberger G (2004) Autothermal reactor concepts for endothermic fixed-bed reactions. Chem Eng Res Des 82:148–159
Guo J, Hou Z, Gao J, Zheng X (2008) Syngas production via combined oxy-CO2 reforming of methane over Gd2O3-modified Ni/SiO2 catalysts in a fluidized bed reactor. Fuel 87:1348–1354
Guo J, Hou Z, Gao J, Zheng X (2008) Production of syngas via partial oxidation and CO2 reforming of coke oven gas over a Ni catalyst. Energy Fuels 22:1444–14448
Hardiman K, Ying T, Adesina AA, Kennedy EM, Dlugogorski BZ (2004) Performance of a Co–Ni catalyst for propane reforming under low steam-to-carbon ratios. Chem Eng J 102:119–130
Hardiman K, Cooper CG, Adesina AA, Lange R (2006) Post-mortem characterization of coke-induced deactivated alumina-supported Co–Ni catalysts. Chem Eng Sci 61:2565–2573
Holladay JD, Hu J, King DL, Wang Y (2009) An overview of hydrogen production technologies. Catal Today 139:244–260
Kolios G, Frauhammer J, Eigenberger G (2000) Autothermal fixed-bed reactor concepts. Chem Eng Sci 55:5945–5967
Ma L, Jiang C, Adesina AA, Wainwright MS, Trimm DL (1996) Simulation studies of autothermal reactor system for H2 production from methanol steam reforming. Chem Eng J 62:103–111
Mohamedali M, Henni A, Ibrahim H (2018) Recent advances in supported metal catalysts for syngas production from methane. Chem Eng 2:9–32
Natesakhawat S, Oktar O, Ozkan US (2005) Effect of lanthanide promotion on catalytic performance of sol–gel Ni/Al2O3 catalysts in steam reforming of propane. J Mol Catal A 241:133–146
Pena MA, Gomez JP, Fierro JLG (1996) New catalytic routes for syngas and hydrogen production. Appl Catal A 144:7–57
Sachtler WMH, Reijen L (1962) Chemisorption bonds on metals. J Res Inst Catalysis Hokkaido Univ 10:87–104
Sengodan S, Lan R, Humphreys J, Du D, Xu W, Wang H, Tao S (2018) Advances in reforming and partial oxidation of hydrocarbons for hydrogen production and fuel cell applications. Renew Sustain Energy Rev 82:761–780
Shahid G, Sheppard N (1990) Infrared spectra and the structures of the chemisorbed species resulting from the adsorption of propene and propane on a Pt/SiO2 catalyst. Spectrochim Acta A 46:999–1010
Sheppard N (1988) Vibrational spectroscopic studies of the structure of species derived from chemisorption of hydrocarbons on metal single-crystal surfaces. Ann Rev Phys Chem 39:589–644
Suthinmporn K, Kawi S (2011) Promotional effect of alkaline earth over Ni–La2O3 catalyst for CO2 reforming of CH4: role of surface oxygen species on H2 production and carbon suppression. Int J Hydrogen Energy 36:14435–14446
Wang Y, Yao L, Wang S, Mao D, Hu C (2018) Low-temperature catalytic CO2 dry reforming of methane on Ni-based catalysts: a review. Fuel Proc Tech 169:199–206
Wilhelm DJ, Simbeck DR, Karp AD, Dickenson RL (2001) Syngas production for gas-to-liquids applications: technologies, issues and outlook. Fuel Proc Tech 71:139–148
Zhang J, Wang H, Dalai AK (2007) Development of stable bimetallic catalysts for carbon dioxide reforming of methane. J Catal 249:300–310
Acknowledgements
Financial support of the Australian Research Council is gratefully acknowledged. FMA also appreciates scholarship and study leave from the Saudi Aramco Company.
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Work done while authors were with the Reactor Engineering & Technology Group, School of Chemical Engineering, University of New South Wales, Sydney, NSW Australia 2052.
Rights and permissions
Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.
About this article
Cite this article
Althenayan, F.M., Adesina, A.A. Oxy-dry reforming of propane over Ce-promoted Co–Ni/Al2O3 catalyst. Appl Petrochem Res 8, 239–251 (2018). https://doi.org/10.1007/s13203-018-0214-9
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s13203-018-0214-9