Introduction

Fuel cell as a serious potential alternative to the conventional engine for powering transportation vehicles suffers from lack of fuel flexibility as hydrogen is merely the only favorable fuel. However, hydrogen infrastructure, distribution and storage are very complicated and prohibitively expensive. As such, developing an onboard reformer of existing commercial fossil fuel to produce hydrogen is a prominent intermediate solution.

The development of an onboard reformer necessitates the search for an effective reforming catalyst. The traditional Ni–alumina-based catalyst is a well-established catalyst for methane steam reforming. It has been further developed on different fronts. At the one hand, loading promoters and/or active metals such as Cu, La, Cr, K, Ca, Mo, Pd and Mn on the catalyst support lead to enhanced catalytic reactivity as compared with traditional Ni/Al2O3 catalyst (Lee et al. 2004; Martinez et al. 2004; Wang et al. 2003; Juan-Juan et al. 2004; Ayabe et al. 2003; Dias and Assaf 2003; Chen et al. 2004; Quincoces et al. 2002; Zhang et al. 2003; Ho et al. 2002; Wang et al. 2004). On the other hand, replacing alumina support with other supports of higher oxygen ionic conductivity improves resistance to carbon formation (Wang et al. 2004; Nimwattanakul et al. 2006; Matsumura and Nakamori 2004; Guo et al. 2004; Zhu and Flytzani-Stephanopoulos 2001). For example, Descorme (Descorme et al. 2000) have reported that lanthanum gallate doped with non-reducible elements such as gadolinium and samarium has positive impact on the catalyst activity.

Incorporating catalytically active metals in some stable ceramic structures such as perovskite and pyrochlore has been the subject of many research studies. For example, Liu (Liu and Krumpelt 2005) have used different lanthanum chromites as reforming catalysts for diesel. Ruthenium at low concentration levels (5 %) has been incorporated on the B-site of such perovskites. During the reforming process, the near-surface ruthenium pops up to the surface and undergoes reduction. When testing for reforming activity, such perovskites have shown unusual reactivity and excellent stability. The catalysts performed well or sometimes even better than the high-cost and highly active rhodium-based catalysts. Many other studies reported similar results using different perovskite-based catalyst doped with active metals on the B-site (Mota et al. 2012; Villoria et al. 2012; Provendier et al. 2001; Erria et al. 2006; Dinka and Mukasyan 2007; Mawdsley and Krause 2006; Dinka and Mukasyan 2007). The common factor in all of these catalysts is the well dispersion of the active metals and their reduction at the surface during the reforming process.

Yttrium strontium titanate-based perovskite has shown excellent performance when used as anode for solid oxide fuel cell. This is mainly attributed to the mixed ionic conductivity that such material has at high temperature. Fergus (Fergus 2006) has measured the ionic conductivity of strontium titanate perovskite after doping the A-site at different doping elements namely Y, Nb, Sm and Eu at different dopant concentrations. In all cases, A-site deficiency is created. Among the various oxides analyzed, the highest conductivity was for Y0.08Sr0.88TiO3. This material is chosen for our analysis. Further doping at the B-site can be conducted with different active metals (Kondakindi et al. 2010; Dinka and Mukasyan 2007; Navarro et al. 2007; Mawdsley and Krause 2008). Such further doping is meant to improve the activity of the material. The prepared catalysts were benchmarked against Ni–alumina catalyst. As the latter catalyst is developed for steam reforming, testing was firstly conducted under steam reforming. Thereafter, the potential application of YST-based catalyst for autothermal reforming of heavy hydrocarbons at high temperature is explored.

Experimental

The catalysts were prepared by the modified sol–gel method, as that reported elsewhere (Lu et al. 2007). Titanium isopropoxide was dissolved in ethylene glycol. The solution was diluted by ethanol and heated at 70–80 °C while constantly stirring using magnetic stirrer. Solution became clear after some time. Citric acid was slowly added as complexing agent. Strontium nitrate, yttrium nitrate and catalytically active metal nitrates were separately dissolved in water. Aqueous metal nitrate solution was then added to the titanium isopropoxide solution. Ammonia was used to adjust pH to 6–7. Solution was left overnight until a gel is formed. The gel was dried in the oven for several hours. The resulting dried gel was crushed using mortar and pestle, transferred to a crucible and heated in a furnace to 600 °C for 1 h to remove organic substances. The resulting powder was mixed again in a mortar and calcined at 800 °C for 5 h and later at 600 °C for 10 h before cooling to room temperature. The resulting powder was characterized using X-ray diffraction (XRD) and surface area (BET) analysis methods. Samples phase analysis was carried out using powder X-ray diffraction with D8 advance diffractometer equipped with Ni-filtered Cu Kα radiation supplied from Bruker. The specific surface area was measured using surface area and porosimetry analyser supplied from Micromeritics, model ASAP2020. The samples were degassed at 120 °C overnight before the measurements. The total surface areas were estimated using the BET equation over a range of 0.05–0.3. Particle size was measured using transmission electron microscopy (TEM) supplied by JOEL, model JEM 1400.

The schematic diagram of the activity testing reactor is given in Fig. 1. Three gas lines are available for feeding nitrogen, hydrogen and air/oxygen. Water and liquid fuel are fed through two separate lines through two HPLC pumps. All lines lead to an oven and finally to the reactor furnace. The oven allows uniform adiabatic heating up to 200 °C. Coil mixers are installed in every line for mixing. Additional coil mixers are also integrated after water and fuel are met together and thereafter such that all reactants are thoroughly mixed and heated before entering the reactor. Once introduced in the three-zone furnace the reactants mix in the SiC-filled top zone. The SiC particles helped ensuring uniform mixing and heating. After being mixed, the reactants pass through a vacant zone up to the catalyst to ensure immediate exposure to catalyst, thus avoiding pre-ignition of fuel at intermediate temperatures. The catalyst is loaded in the middle zone of the tube. The sample is pelletized, crushed and sieved through to get granules having size in the range of 0.2–0.5 mm. The appropriate amount (in this case 0.5 g) is diluted with SiC particles such that the total volume of the mixture is 3 ml. Once the reformates exit the furnace, they are directed to an online gas chromatographer (GC) for gas analysis.

Fig. 1
figure 1

Schematic diagram of a benchtop microreactor. Three gas lines with integrated coil mixers met together. Gas mixture is further mixed in an additional coil mixer. Two HPLC pumps feed water and fuel through two separate lines. These liquid lines also include individual coil mixers. After liquids are met, they are also mixed through coil mixer, and all reactants are further mixed before entering reactor tube. After reactants are introduced to the reactor furnace, they are exposed to the catalyst in the middle of the reactor tube. After reaction products exit the furnace, the heavy liquids are trapped in a chiller and gas products are directed to a gas chromatographer for gas analysis

Full size image

The gas analysis is carried out on a gas chromatograph (GC) supplied by Agilent, model 7890A. The GC is equipped with three columns, one of them is molsieve (HP-Plot 5A) and the two others are general-purpose columns (PlotQ, HP-1). Two detectors are integrated: thermal conductivity (TCD) and flame ionization (FID) detectors. The carrier gas used was argon. The method used is a modified version of that reported elsewhere (Thormann et al. 2008). The first column (PlotQ) is opened 2 min after sample injection and left open for 10 min. The oven temperature was hold at 60° for the first 12 min and subsequently increased to 180° at a rate of 10 K/min. The latter temperature profile is necessary for the least volatile compounds (high-C compounds) to evaporate and for having sufficient resolution of the different reaction products elution times. A fast cooling rate of 50 K/min back to 60° is applied immediately after the high temperature (180°) is reached. The molsieve column is then opened so that the permanent gases are trapped. Nitrogen is fed in the reactor as a carrier gas and as balancing flow to determine the volume of the different reformates at GC.

Results and discussion

The materials used for the reforming catalytic reaction are: Y0.08Sr0.88Ti0.9M0.1O3, where M = Ni, Ru and Co. These materials will be subsequently designated as YSTM, where M = N, R and C stands for nickel, ruthenium and cobalt, respectively. The A-site deficiency is intended to enhance the ionic conductivity and yttrium at the concentration reported in this study provides the highest ionic conductivity (Fergus 2006). The doping elements on the B-site are meant to boost the catalytic activity of the catalysts. Nickel is used as it has proven to be an excellent transition metal for reforming at high temperature. Ruthenium has been chosen to represent the family of precious metals. The latter choice is the cheapest among many other precious metals such as platinum and rhodium. Cobalt was picked as it has been reported that it has superior activity than nickel yet not as expensive as ruthenium (Llorca et al. 2004; Song and Ozkan 2009; Urasaki et al. 2008; Virginie et al. 2008; Casanovas et al. 2008; de la Pen˜a O’Shea et al. 2008). The reaction has been conducted at high temperatures within the recommended range of 700–900 °C. Upper range of 800–850 °C was chosen as the conductivity increases with increasing temperature. High-temperature (900 °C) evaluation was avoided as it is not practical commercially due to the limitations and constraints in cost, design and energy supply. The traditional Ni–alumina catalyst has been chosen as the benchmarking catalyst. As it is mainly developed for steam reforming, steam reforming was conducted on all catalysts at T = 800 °C to have a meaningful comparison. Autothermal reforming has also been conducted as it has the advantage of design simplicity, compactness and least energy requirement thus mostly celebrated commercially.

Upon preparation of the catalysts, characterizations have been conducted on the fresh catalyst including X-ray diffraction, specific surface area and transmission electron microscopy. The catalyst precursor has been calcined for 10 h at different temperatures of 600, 650, 700 and 800 °C to study the effect of calcination temperature on the catalyst structure. Figure 2 shows a typical XRD pattern of YSTN catalyst. It is shown that low crystallinity is formed at 600 °C. At 650 °C well-defined peaks are formed along with the presence of an unknown second phase. The second phase formation persisted at higher calcination temperature of 700 °C and disappeared at 800 °C. As such we choose the starting calcination temperature of 800 °C, but for only 5 h, a second calcination at 600 °C for 10 h is conducted before furnace cooling to room temperature. It was preferred not to hold the temperature for longer time at 800 °C to avoid coarsening and reduction in specific surface area. The same calcination profile was adopted for all catalysts. The XRD patterns are shown in Fig. 3 for the different catalysts. As shown all catalysts showed single-phase crystallinity. The specific surface area was measured to be between 10 and 20 m2/g. The particles are very fine and nanoscale in size as shown in Fig. 4a for a TEM micrograph. A small highlighted portion is further shown in Fig. 4b at higher magnification. Even though particles are nanoscale in size, the catalyst showed only modest specific surface area.

Fig. 2
figure 2

XRD pattern for YSTN after calcination for 10 h at 600, 650, 700, 800 °C. Amorphous phase is clearly observed for calcination temperature of 600 °C. Tiny second phase is present for calcination temperature of 650 and 700 °C and disappear at 800 °C

Full size image
Fig. 3
figure 3

XRD pattern for the different catalysts (YST, YSTC, YSTN and YSTR) after calcination at 800 °C for 5 h followed by dwelling at 600 °C for 10 h. The XRD patterns showed single phase for all catalysts

Full size image
Fig. 4
figure 4

a TEM micrograph of a typical fresh catalyst (YSTC). b Particle morphology at higher magnification

Full size image

Activity of the catalysts was evaluated for steam and autothermal reforming of heavy hydrocarbon oil namely dodecane, a diesel surrogate free of aromatics and sulfur. Very high LHSV = 5 was chosen in the hope of distinguishing the best-performing catalysts from the less performing ones. Only 0.5 g of catalyst diluted by SiC particles is loaded in the middle zone of the reactor tube. SiC particles are also used to fill the first zone of the reactor tube to help further mixing fuel with air and water and provide uniform heating of the mixture to intermediate temperature. The reaction temperature was 800 °C, while the GHSV was maintained at 22,000 h−1. Steam reforming was conducted on YST material systems and Ni–alumina catalysts with S/C = 3. Typical gas analysis data are given in Table 1. The product gas concentrations on dry basis and free of nitrogen are shown in Fig. 5a–d and average values are listed in Table 2. The results are surprisingly very similar despite the high LHSV with the perovskite catalysts performing as well as the Ni–alumina commercial catalyst. The hydrogen concentrations are 70 % or higher for the different catalysts which is within expectation. What is, however, noticeable is the high methane concentration for YSTC catalyst. This indicates that the catalyst is not as effective reforming catalyst as more methane is slipping from catalytic reforming reaction. It is not surprising that CO2 concentration is as low as 10–11 %. This is in agreement with thermodynamic calculations that suggest reduction in CO2 concentrations with increasing temperature and S/C ratio and reducing reactor pressure (Twigg 1989). In our case, reactor pressure is 2.5 bar and reaction temperature and S/C ratio are high (T = 800 °C and S/C = 3). Carbon monoxide concentration is consistently higher than that of CO2 and is around 15 % for Ni–alumina and YSTR and higher than 18 % for YSTN and YSTC. High concentration of CO is favorable as CO can be a further source of hydrogen through water gas shift reaction that can be conducted downstream at low temperature.

Table 1 Typical gas analysis data including compound type, retention time, area and normalized concentrations
Full size table
Fig. 5
figure 5

Gas product (CH4, CO, CO2 and H2) concentrations for steam reforming at 800 °C and S/C = 3 for a YSTN, b YSTN, c YSTR and d Ni–alumina

Full size image
Table 2 Conversion and gas product distribution for the different in-house (YSTR, YSTN, YSTC) and commercial (Ni–alumina) catalysts under steam reforming at 800 °C and S/C = 3
Full size table

The gas product concentrations listed in Table 2 are for dry basis and free of nitrogen. A better way of understanding catalyst performance could be obtained from the hydrogen yield calculated as the ratio of H2 mol to moles of input fuel which is given in Fig. 6. Carbon monoxide yield is also important in view of the fact that it can be converted back to H2 through water gas shift reaction. As such, Fig. 6 also shows CO yield appended to H2 yield. Table 3 lists the hydrogen, CO and syngas (H2 + CO) yields. Figures 5, 6 and Tables 2, 3 suggest that reforming dodecane with YSTR resulted in very low methane slip, highest hydrogen yield and concentration and best syngas yield that has even outperformed the commercial Ni–alumina catalyst. Activity of the catalysts follows the order YSTR > YSTN > YSTC.

Fig. 6
figure 6

H2 and syngas (H2 and CO) yield for the different catalysts under steam reforming of dodecane at T = 800 °C, S/C = 3. Upper theoretical limit for H2 and syngas yield is also illustrated as a horizontal line

Full size image
Table 3 H2, CO and syngas yields for the different in-house (YSTR, YSTN, YSTC) and commercial (Ni–alumina) catalysts under steam reforming. (S/C = 3, T = 800 °C)
Full size table

Conversion or carbon balance is calculated as:

$${\text{Conversion}} = \frac{{\varSigma {\text{moles}}\,{\text{of}}\,{\text{C}}\,{\text{in}}\,{\text{reformate}}}}{{{\text{Total}}\,{\text{mole}}\;{\text{of}}\,{\text{C}}\,{\text{in}}\,{\text{feed}}}}\times100$$

Conversions are also listed in Table 3. The latter quantity is 90 % for Ni–alumina and YSTR and around 85 % for YSTN and YSTC. Failing to have full conversion may be due to the non-uniformity in mixing, the pre-combustion of fuel and the hot-spot formation due to the non-uniform temperature distribution and inhomogeneous fuel mixing.

The steam reforming reaction is given by:

$${\text{C}}_{n} {\text{H}}_{m} + n{\text{H}}_{2} {\text{O }} \to n{\text{CO}}_{x} + \, \left( {n + m/2} \right){\text{H}}_{2} \quad \Delta {\text{Hr}} > \, 0$$
(1)

Assuming that dodecane is 100 % converted and no CO is produced, the reaction would become:

$${\text{C}}_{12} {\text{H}}_{26} + \, 36{\text{H}}_{2} {\text{O }} \to \, 12{\text{CO}}_{2} + \, 37{\text{H}}_{2} + 12{\text{H}}_{2} {\text{O}}$$
(2)

That is, theoretically the H2 yield is 37. A line at H2/fuel mole ratio = 37 is also constructed in Fig. 6 to define the theoretical limit for H2 yield. This theoretical limit is also valid for syngas production as CO is absent in the products of reaction (2). As such, both Ni–alumina and YSTR catalysts are of more than 90 % of the theoretical limit.

Current focus is more on autothermal reforming not only because of cost reasons but also because ATR favors design compactness and simplicity and also better thermal efficiency. Autothermal reforming activity at S/C = 3 and O/C = 0.32 and at two temperatures 800 and 850 °C are given in Figs. 7, 8 and in Table 4. Again the results are similar and quite comparable with the commercial catalyst. The latter is mainly developed for methane steam reforming; however, it performed quite well for ATR and under heavy hydrocarbon, namely dodecane. It is interesting to note that the perovskite catalyst has nearly the same performance as Ni–alumina catalyst irrespective of the doping element. However, YSTN and YSTR are consistently slightly better than YSTC and do sometimes over perform Ni–alumina catalyst although by slight amount.

Fig. 7
figure 7

H2 and syngas (H2 and CO) yield for the different catalysts under autothermal reforming of dodecane at T = 800 °C, S/C = 3 and O/C = 0.32. Upper theoretical limit for H2 and syngas yield is also illustrated as a horizontal line

Full size image
Fig. 8
figure 8

H2 and syngas (H2 and CO) yield for the different catalysts under autothermal reforming of dodecane at T = 850 °C, S/C = 3 and O/C = 0.32. Upper theoretical limit for H2 and syngas yield is also illustrated as a horizontal line

Full size image
Table 4 Conversion and H2, CO and syngas yields for the different in-house (YSTR, YSTN, YSTC) and commercial (Ni–alumina) catalysts under autothermal reforming. (S/C = 3, O/C = 0.32, T = 800, 850 °C)
Full size table

The autothermal reforming reaction is given by:

$${\text{C}}_{n} {\text{H}}_{m} + {\text{O}}_{2} + n{\text{H}}_{2} {\text{O }} \to n{\text{CO}}_{x} + \, \left( {n + m/2} \right){\text{H}}_{2} \quad \Delta {\text{Hr}} \cong 0$$
(3)

The reaction assuming that the entire carbon monoxide is converted into CO2 through the water gas shift reaction, would be

$${\text{C}}_{12} {\text{H}}_{26} + \, 36{\text{H}}_{2} {\text{O}} + 3.78{\text{O}}_{2} \to \, 12{\text{CO}}_{2} + \, 29.5{\text{H}}_{2} + 19.5{\text{H}}_{2} {\text{O}}$$
(4)

The highest theoretical hydrogen yield is 29.5 and is given in both Figs. 7 and 8 as a horizontal upper limit for syngas yield. The syngas yield for all in-house prepared catalysts is nearly 90 % of the theoretical limit with YSTR having highest value.

Overall the performance of the catalysts is nearly the same for both temperatures (800 and 850 °C). Improvement in activity is not noticed with increasing temperature as predicted by thermodynamics. This was mainly due to the problem with the equipment design, lacking a separate preheater due to space limitation. As reaction temperature is increased, the top (mixing/preheating) zone temperature is also increased above the set value, and thus pre-combustion and inhomogeneous mixing are more pronounced at 850 °C. This is also reflected by the slight lower conversion at 850 °C as compared with that at 800 °C. Overall the YST-based catalysts are promising for autothermal reforming of dodecane with Ru the best active dopant, followed by Ni and lastly by Co. As dodecane is one of the main diesel surrogates, the testing of the catalyst for diesel reforming may also show promising performance. Such analysis is left for future activity.

Conclusion

YST-based perovskite catalysts were prepared and characterized for structural properties. Single-phase crystallinity and specific surface area between 10 and 20 m2/g were measured, and the particles were nanoscale in size. The activity was tested at two temperatures 800, 850 °C and under both steam and autothermal reforming. Three types of doping on B-site gave very similar activity, conversion and hydrogen and syngas yields that were quite comparable with those of commercial catalyst. Over the period of 12–24 h, the catalysts were stable and no degradation was noticed. Such degradation may potentially emerge upon adding sulfur to fuel. Such analysis is undergoing and will be reported in the future. The high activity and stability of these catalysts doped with precious or transitional cost-effective metals provide a potential alternative family of reforming catalysts for commercial transportation fuel in an onboard reformer.