*2.1. Catalyst Characterization*

Catalyst physical characterization details are presented in Table 1. These MgAl2O4-supported metal catalysts have been characterized in our previous publications, and used as catalysts for steam reforming of methane [20], gasifier-derived hydrocarbons including benzene [19] and complex mixtures including tar [21], and biomass-derived ethylene glycol [37] and aqueous products produced via fast pyrolysis [38]. It was reported that the metals form small and stable clusters when supported on MgAl2O4 (approximately 1, 2, and 7 nm Ir, Rh, and Ni median cluster sizes, respectively) even under steam reforming conditions at 850 ◦C [20,21]. In this study, we report additional catalyst characterization include TPR as well as the hydrogen adsorption over the temperature range of interest for natural gas steam reforming (i.e., 600–850 ◦C).

**Table 1.** Catalyst physiochemical characterization. Scanning transmission electron microscopy (STEM) images for the Ir and Rh catalysts are included in the supplementary information (Figure S1).


aBrunauer–Emmett–Teller method.

Figure 1 shows TPR profiles for the Rh-, Ir-, and Ni-supported catalysts used in this study. Results indicate an easier reduction for Rh with a single reduction peak at 112 ◦C. Ni requires the highest temperature to reach a similar reduction level (peak centered at 725 ◦C) and shows evidence of at least two reduction stages (shown by overlapping peaks). The Ir catalyst shows several reduction stages at 168, 220, 253, and 613 ◦C. For all the three catalysts (Rh, Ir, and Ni) after 800 ◦C was reached, no further hydrogen was chemisorbed, indicating a total (or near total) reduction of the metal clusters. However, the temperature needed to produce reduced metal clusters (measured by hydrogen consumption) varied significantly. Results in the literature sugges<sup>t</sup> Rh disperses very well on a wide variety of supports, with low temperature single peaks being characteristic for Rh [39,40], thus corroborating our results. The several reduction stages required for Ir can be ascribed to a broader particle size distribution or a broad distribution of oxidation states. Reports in the literature indicate that smaller Ir particles require a higher reduction temperature compared to those for larger particles [41,42]. Nevertheless, the reduction of Ir-based catalysts is a relatively complex phenomenon that largely depends on the processing conditions and thermal history of each sample (Ir reduction is an autocatalytic process and IrOx are volatile species detrimental to metal dispersion) [43]. The Ni catalyst TPR profile also is consistent with literature results for catalysts with a similar particle size (6 nm) [44,45]. Before reaction catalysts are reduced at 850 ◦C for 16 h in order to ensure a degree of reduction for the metal clusters. We note that in our prior reports we have shown how the MgAl2O4-supported Ir, Rh, and Ni clusters are fully reduced when the catalyst was reduced under hydrogen at 850 ◦C. However, we note that the total reduction cannot be determined, especially for Ni (because of its lower dispersion), when under operando conditions.

H2 activation, adsorption strength relative to carbon, and coverage on metal surfaces are important metrics for steam reforming relevant reactions [33,41,46]. However, obtaining accurate characterization under operando conditions presents technical and experimental challenges. In an attempt to study H2 adsorption capacity relative to metal active area, we combined volumetric H2 adsorption at 600 ◦C (as representative of reforming conditions) with metal dispersion, as determined from metal particle sizes revealed by the STEM imaging (2 nm, 1 nm, and 7 nm for the Rh, Ir, and Ni, respectively) [20,21]. The H2 adsorption data presented in Figure 2 was calculated as H-coverage evolution (mol H/ molMsurf, M= Rh, Ir, Ni) for accumulated H2 added (by pulses) into the adsorption cell (μmol). These adsorption results indicate that bare Rh and Ir surfaces adsorb H2 completely until saturation, which is indicated

by the inflexion point and the asymptotical part of the plots (after adding ~40 and ~20 mmol of H2, respectively). Saturation is reached at coverages of H-Msurf of 1 and 0.7 for Rh, and Ir, respectively, which indicates a surface stoichiometry of 1:1 (H:Rhsurf) and 0.7:1 (H:IrSurf). In the case of Ni, the adsorption pattern is dissimilar in three main aspects: (1) H-uptake is low (0.2 molH/molNisurf) compared to Rh or Ir, (2) there is no clear inflexion point for the adsorption curve, and (3) the calculated stoichiometry H-Nisurf is low (only 20% of surface nickel atoms adsorb hydrogen). H2-uptake at 800 ◦C is lower compared to the uptake at 600 ◦C (data shown Figure S2). At 800 ◦C, hydrogen saturates the metal surfaces at 1:1 (Rh), 0.5:1 (Ir), and 0.08:1 (Ni).

**Figure 1.** Temperature programmed reduction (TPR) profiles for MgAl2O4-supported Ir, Rh, and Ni catalysts. Prior to TPR the catalysts were calcined at 500 ◦C for 4 h. TPR conditions were 50 mg of catalyst, temperature ramping rate 5 ◦C/min, feed rate of 50 mL/min of 5% H2/Ar gas.

**Figure 2.** Hydrogen surface saturation curves over MgAl2O4-supported Rh (blue), Ir (green), and Ni (red) catalysts at 600 ◦C. Coverage was calculated using accumulated H2-uptake and metal dispersion data obtained from STEM imaging (Table 1). 50 mg of catalyst was calcined before analysis and reduced at the conditions used in the catalytic experiments (16 h at 850 ◦C, 100 mL/min, 10% H2 in N2). H2 was pulsed using a 100 μL gas loop.

The physicochemical characterizations of the catalysts shown in this section are significant for the following reasons:


We found that while full surface coverage with a stoichiometry 1:1 H-Rhsurf can be achieved at reforming temperatures in the 600–800 ◦C range for the Rh catalyst, this coverage is significantly lower for Ir (0.7:1 H-Irsurf) and even lower for Ni (0.2:1 H-Nisurf).

#### *2.2. Steam Reforming Activity Comparison Using Natural Gas Simulant*

Steam reforming using a simulant gas mixture representative of natural gas (94.5% methane, 4% ethane, 1% propane, 0.5% butane) was performed over MgAl2O4-supported 5% Ir and 5% Rh catalysts. Conversion of methane and ethane is plotted separately and shown in Figure 3 (propane conversion in Figure S3). Short contact times (τ = 4.5 ms) favor low conversion levels. In the 600–875 ◦C temperature range, methane conversion follows a linear trend for both catalysts (Figure 3a). In spite of similar methane conversion, conversion of ethane (Figure 3b) and propane (Figure S3) is significantly lower for the Ir catalyst in the lower temperature range (<750 ◦C). These differences in activity are important for two main reasons: (1) gas feeds with higher concentration of C2+ hydrocarbons lower the overall performance of Ir-based catalysts (more hydrocarbons remain unreacted) and (2) hampered hydrocarbon hydrogenolysis/reforming activity is an indication that the underlying mechanism is different for Ir compared to Rh in spite of similar methane reforming activity. We note that the methane conversions presented in Figure 3a are below those of equilibrium (with the equilibrium conversions indicated with the dotted line). We also note that the equilibrium ethane conversion shown in Figure 3b is near completion for the entire range of conditions investigated.

**Figure 3.** Comparison of (**a**) methane conversion and (**b**) ethane conversion for the Rh and Ir catalysts when steam reforming using natural gas simulant feedstock (S/C = 1.5 mol, τ = 4.5 ms).

We note that H2/CO ratio is an important metric for reforming catalysts when the steam reforming reaction is integrated to downstream processes. In practice, several strategies can be used to tune the H2/CO ratio of the reforming product (e.g., decreased S/C ratio during SMR, CO2 recirculation, incorporating an additional WGS step) towards reactions of interest (e.g., Fischer–Trospch). The H2/CO ratio for RhMgAl2O4 was evaluated over a wide range of conversions (20%–100%) at the range of operative temperatures of SMR in solar thermochemical applications (600–800 ◦C) by changing the contact time of the methane stream (for τ values from 1.2 to 10 ms) at a constant S/C ratio (S/C = 3). Results for conversion vs. temperature, and H2/CO ratio vs. conversion are presented in the supplementary information as Figure S8. Interestingly, Figure S8 shows that independently on the temperature, H2/CO ratio decreases exponentially with methane conversion reaching a minimal value of 5 at the highest conversion (ca. 100%). This will lead to the conclusion that WGS equilibrium determines the resulting H2/CO ratio. As the H2 partial pressure increases as an e ffect of higher reforming conversion, more CO is formed through the reverse water gas shift reaction (RWGS, CO2 + H2→CO + H2O).

#### *2.3. Methane and Ethane Steam Reforming Activity Comparison*

Based on the results described above, the Ir catalyst shows lower activity than Rh towards the steam reforming of higher hydrocarbons. With the aim of gaining knowledge on fundamental di fferences between the di fferent hydrocarbon constituents, we performed a set of separate experiments using either methane or ethane feed. Figure 4 shows the metal surface normalized catalytic rate (molesCH4/molmetal\*s) for the steam reforming of methane and ethane for the Rh, Ir, and Ni supported catalysts as a function of temperature. These results show that for the Ir catalyst, the ethane steam reforming rates (Figure 4b) are significantly lower compared to methane steam reforming rates (Figure 4a). For the Rh and Ni catalysts, ethane conversion is more facile than methane conversion.

**Figure 4.** Comparison of turnover steam reforming rates for the Rh, Ir, and Ni catalysts using (**a**) methane, and (**b**) ethane feeds (S/C = 3, τ = 30 ms; methane feed = 15.9 vol%, ethane feed = 14.8 vol%).

Over typical metal supported catalysts activation of the initial C–H bond of methane is more di fficult than for ethane [47]. Dehydrogenation of the initial C–H and the creation of C radical + H for ethane is slightly more favorable than for methane (C–H bond strength 101 vs. 105 kcal/mol, respectively). Further, the C–C bond of ethane is much weaker (90.1 kcal/mol) [48]. These di fferences between methane and ethane typically make ethane more reactive than methane. Few literature studies can be used to make a direct comparison or to determine trends with respect to reactivity of methane

vs. ethane reforming because most studies use methane as the only model reactant. Specific studies on ethane reforming have yielded contradictory results depending on the metal used as the catalyst. On one hand, Schädel et al. [35] showed that over Rh, ethane reacts faster than methane, which is consistent with our findings for Rh and Ni catalysts. On the other hand, Graf et al. [34] reported that over Pt, methane reforming occurs the fastest, which is similar to our findings for Ir. Similar trends, where ethane reactivity is higher compared to methane, have been found in other metal-catalyzed reactions such as ethane combustion [49] and ethane hydrogenolysis over Pt and Pd [50]. Interestingly, Figure 3b also shows that the Ir catalyst is relatively inactive at low temperatures but then the catalytic rates quickly accelerate at higher temperatures (>750 ◦C). This rapid transition from 'inactive' to 'active' prompted our study to pursue a more detailed understanding for ethane reactivity over Ir catalyst. The following section is dedicated to hypothesizing the possible reason(s) behind the low reactivity of ethane over Ir catalysts at temperatures lower than 750 ◦C.

#### *2.4. Ethane Steam Reforming over Rh vs. Ir: Mechanistic Insights*

Results shown in Figure 4 indicate that over Ir catalyst at 600 ◦C, surface normalized ethane reforming rates (0.6 s<sup>−</sup>1) are much slower compared to methane reforming rates (3 s<sup>−</sup>1). This di fference in catalytic rate could be attributed to several factors: (1) potential formation of carbon deposits over Ir surface when ethane is fed, (2) decreased number of active sites due to competitive adsorption (e.g., ethyl, O, OH, H, C1, and C2 species), and the (3) chemical identity of the metal and its bonding characteristics. To compare hydrocarbon reactivity over metal surfaces, catalytic measurements should be performed under the same coverage [51]; however, this comparison is not always easy to achieve. As a simplified approach to compare ethane reforming activity over Ir vs. Rh, we performed catalytic measurements at 600 ◦C using the same concentration of ethane in the gas over similar amounts of metal exposed (Table 2). The average Rh cluster size (2 nm) was twice that of the Ir cluster (1 nm); however, the Ir molecular weight (192 g/mol) is twice that of Rh (103 g/mol). Thus, the amount of exposed metal is similar for both catalysts on a molar basis. With similar number of exposed metal sites, the observed product selectivity and associated reaction rates should be determined by the underlying chemical reaction mechanism.


**Table 2.** Catalytic results for ethane steam reforming over MgAl2O4-supported Rh and Ir catalysts at 600 ◦C (S/C = 2.75 mol, τ = 28.3 ms).

a Based on TEM particle size measurements. b Below limit of detection of CNHS elemental analysis instrument (0.3%). c 0.28% measured for the dilution of 1:10 catalyst:Al2O3.

Shown in Table 2 are ethane steam reforming results using the Rh- and Ir-supported catalysts. For both catalysts steady state is reached after 15 min, and catalytic activity is reported after 1 h and 2 hours' time-on-stream (see Figure S4). The main reaction products detected by gas chromatography analysis include H2, CO2, CO, methane, and ethylene. Reactivity of ethane over Rh is di fferent than Ir in four main aspects: (1) conversion is significantly higher (60% vs. 8%), over a similar exposed metal surface area, (2) ethylene formation is markedly lower (50 ppm vs. 2000 ppm), (3) post mortem analysis showed significant less accumulated carbon, and (4) methane is the main reaction product (methane selectivity of 32% vs. 9.6%).

Methane is the primary hydrocarbon product over Rh when under ethane steam reforming conditions. Methane can be formed as a hydrogenation product of CO or CO2 [33] or as a product of ethane hydrogenolysis (according to the Sinfelt–Taylor mechanism) [52,53]. These two reactions are related and depend on the metal's carbon/hydrogen adsorption characteristics and its ability to breakdown adsorbed C2 radicals [54]. In Section 3.1 we reported marked differences in H2 adsorption and coverage for the Rh catalyst vs. Ir or Ni. Although the H2 adsorption experiment does not allow comparison for H/C adsorption, they demonstrate improved H retention for the Rh catalyst. This is a particularly useful property that favors the aforementioned methane formation mechanisms (hydrogenation of CO/CO2 species and hydrogenolysis of ethane) under reforming conditions. Parallel to methane formation, the presence of ethylene product is an indicator of formation (and slow decomposition) of ethyl radicals on the catalyst surface. Higher formation of ethylene over the Ir catalyst is a strong indicator that the overall catalytic ethane reforming rate is hampered by a slow C–C bond scission capacity compared to Rh.

The marked difference in conversion for the Ir and Rh catalysts shown in Table 2 makes it difficult to fairly compare product selectivity. Conversion for the Ir catalyst was increased, by increasing the contact time (from 28 to 166 ms), to compare product selectivity at similar conversion levels. As shown in Figure 5, under these conditions methane is the main reaction product for both catalysts (60% vs. 30% methane selectivity for Ir and Rh, respectively), with CO and CO2 also being produced as main products, and ethylene being produced via incomplete decomposition of ethane. Figure 5c shows changes in selectivity as a function of change in contact time over the Ir catalyst. Methane selectivity increases as ethane conversion increases (selectivity towards methane increases from 8 to 31% for ethane conversion increasing from 8 to 50%). Interestingly, selectivity towards ethylene formation decreases as the conversion increases (2 to 0.6%), which is significantly higher than selectivity towards ethylene over the Rh catalyst at a similar conversion (0.02%, Insert in Figure 5c). These changes in product selectivity—increased methane and decreased ethylene—at higher ethane conversion levels can be explained in a scenario where C–C bond scission forming methane is the main and primary route for ethane decomposition before reforming reactions (CO and CO2 formation) take place, and is similar for both metals (Figure S5).

**Figure 5.** Product selectivity for ethane steam reforming at 600 ◦C (S/C = 2.75 mol, τ = 28 ms [Rh]; τ = 167 ms [Ir]; ethane feed = 16 vol %) over Ir/MgAl2O4 catalyst at 50% ethane conversion (**A**), over Rh/MgAl2O4 catalyst at 59% ethane conversion (**B**). Variation of carbon selectivity with ethane contact time over Ir/MgAl2O4 catalyst (**C**).

These observations for ethane reforming over Ir compared to Rh (higher ethylene formation, lower selectivity towards methane) can be combined to sugges<sup>t</sup> that hydrogenolysis of ethane is relatively slow over Ir and facile over Rh. This hypothesis is additionally supported with several control experiments performed over Ir catalyst and summarized into the following four main points:


**Figure 6.** Logarithmic plot for catalytic rate vs. partial pressure of (**a**) ethane and (**b**) water.

#### *2.5. Ethane Reforming Kinetic Measurements*

Methane reforming kinetics have been extensively reported in the literature. Results indicate that methane decomposition and reforming rates are first order in methane; C–H activation is the rate limiting step and occurs on metal surfaces similar to conventional hydrocarbon chain reaction mechanism [32,55]. Ethane activation over metal surfaces is more complex, involving a series of partially dehydrogenated intermediates that lead to C–C bond breaking [41,56]. Strongly adsorbed H is considered to be competitive with hydrocarbon adsorption and limits its reactivity (in general, and over metal surfaces, the ethane hydrogenolysis rate is inhibited by excess H2) [50,57,58]. One might think that interaction of highly dehydrogenated species in the absence of active hydrogen could lead to a higher coke formation; however, there is limited experimental evidence that shows that co-feeding hydrogen prevents coke formation better than water does [59]. Regarding the activation of the oxidant (e.g., steam), it is been largely reported that methane has zero order dependence on water (or CO2). This indicates that after C–H activation, all the catalytic steps, including dehydrogenation of CHx species, CO/CO2 formation, and WGS reaction, should have no kinetic relevance.

To study the influence that the concentrations of reactants have in catalytic rate, we varied ethane and water concentrations in the gas feed at 600 ◦C. Kinetic results for catalytic rate vs. partial pressure are shown in Figure 6. Calculated reaction orders for both ethane and water and activation energies are shown in Table 3. The reation order to ethane is less than one and zero or near zero to water. For methane steam reforming, the reaction is first order to methane, and the rate limiting steps have been found to be adsorption and activation of the C–H bond [36]. Thus, there is less dependance of hydrocarbon partial pressure on reaction rate for ethane steam reforming than for methane steam reforming.


**Table 3.** Kinetic parameters calculated for ethane steam reforming reaction at 600 ◦C.

> a Measured at 550, 600, and 650 ◦C.

The results presented in Table 2 indicate that the Ir catalyst accumulated more carbon on the surface than Rh after the ethane reforming reaction at 600 ◦C; this presumably is due to a slower C–C scission rate. Analysis of the hydrogenolysis mechanism suggests that if a surface step is hampering the catalytic cycle, the initiation step (adsorption/activation of C–H bond) loses kinetic importance (resulting in a lower reaction order). Our low reaction orders for the Ir catalyst (nethane = 0.29 and nwater = −0.3) are consistant with high carbon coverage of the catalyst surface; in this catalyst, the reaction of adsorbed H with C2 species occur before the C–C bond breaking step, causing formation of ethylene gas. In the ethane hydrogenolysis reaction, the hydrogen adsorption capacity of the metal (M–H coordination), its adsorption strength, and the coverage relative to other species (C2, C1) are important factors in overall catalytic rate and coke formation. The initiation and termination reactions in hydrocarbon chain mechanisms on metals require a high affinity of the metals for H. Our previous analysis in this section indicate that the slowest step is related to the creation (and accumulation) of adsorbed carbon species (C1 and C2); these results might be related to a poor capacity to activate and retain H. The low reaction order found for ethane reforming (0.29), a higher ethylene formation level, and higher carbon accumulation (Table 2) support the conclusion that surface ethyl decomposition (part of ethane hydrogenolysis reaction) is superseding the global reaction order.

Ethane steam reforming studies at 600 ◦C on Rh performed by Graf et al [59] indicated that in the presence of H2, ethane reacts to form methane at a much higher rate than towards the formation of reforming products; additionally, their analysis of reaction products suggests that ethane reforming and hydrogenolysis proceed in a parallel fashion. Our analysis of reforming products differs in two important key aspects from the conclusions by Graf et al. First, by changing the contact time, we found that a higher ethane conversion is accompanied by a higher selectivity towards methane and a lower formation of ethylene (Figure 5c, Figure S5). This indicates that hydrogenolysis reaction proceeds reforming reactions; in other words, reforming steps are sequential to ethane hydrogenolysis. Second, the amount of H available and its adsorption strength to hydrogenolysis reaction becomes critical, more than what it is the presence of oxidant species [41,58]. In this scenario, surface active hydrogen is necessary to hydrogenate C1 species on catalyst surface that will cause excessive carbon buildup if they don't undergo reforming reactions. Several key factors for improved catalysis are (1) an optimal amount of steam, (2) a fast C–C bond scission rate, and (3) a high H retention capacity by the catalyst. These key factors will avoid the accumulation of highly dehydrogenated carbon species (especially C and CH [80 kcal/mol]) that promote coke formation.
