Carbon supported catalysts in low temperature steam reforming of ethanol: study of catalyst performance

Abstract

Three different metals (Pt, Pd and Ni) on multi-walled carbon nanotube and graphite supports were tested as catalysts in ethanol reforming experiments at low temperatures. The carbon nanotube based catalysts outperformed their counterparts supported on graphite in both ethanol conversion and hydrogen evolution. The palladium catalyst in both supports was rather inactive, whereas the nickel/nickel oxide catalyst showed excellent performance. The most active catalyst (Ni/CNT) showed ethanol conversion and hydrogen evolution already at 200 °C with complete conversion at 450 °C. The measured selectivity for H₂ production was ∼42% with almost 100% conversion. Analysis of the spent catalyst samples revealed a considerable increase in carbon concentration caused by the formation of soot and coke on each sample but also carbon nanofibers and nanotubes were found to grow on the catalyst nanoparticles in the course of the steam reforming reactions.

---

Introduction

As the world's crude oil storages will run out in the near future, the demand to find new, and eventually sustainable energy sources is growing. One possible “green” source of energy is hydrogen with its versatile application for fuel cells as well as for “green” feedstocks in the chemical (ammonia synthesis, desulfurization of crude oil) and food (hydrogenation of unsaturated fatty acids) industries. Although hydrogen is considered as an environmentally friendly fuel or reactant, it is not necessarily green. Whether it is or not, actually depends on the method of synthesis and the feedstock it is obtained from.¹ Currently, steam reforming (SR) is the most efficient method (efficiency up to 85%)² to produce hydrogen, however the most common feedstock's are fossil based hydrocarbons, i.e. the process is not renewable. Using ethanol as a raw material instead of fossil hydrocarbons has several advantages; it is non-toxic, free from sulfur, can be bio-waste based leading to truly CO₂-free H₂ and it can be reformed at significantly lower temperatures (<450 °C) than hydrocarbons (>500 °C).^3–7

Depending on the support, different catalysts (Pt, Pd, Rh, Ir, Ru, Ni, Co and Cu) have been demonstrated to be active in steam reforming of ethanol (SRE), from which Ni seems to be the best with nearly perfect ethanol conversion and excellent hydrogen selectivity (up to 96%) at 800 °C.⁸ The most common catalysts are supported on metal oxides (CeO₂ and Al₂O₃)^5,6,9,10 however, some recent studies also report on the use of carbon based supports for steam reforming of ethanol^7,11,12 and hydrocarbons.^13–15 Carbon nanotubes (CNTs) have been found as promising catalyst support materials during the past decade. CNTs have relatively high specific surface area and large porosity as those of activated carbons. At the same time, they also show excellent thermal and electrical conductivity similar to that of graphite (GC). Furthermore, carbon nanotubes are rather inert in chemical reactions and can be easily grown on templates to form scaffolds and hierarchical structures with engineered pore systems.^11,16,17 Carbonaceous supports, however, are more sensitive to heat than metal oxides and may face deactivation due to catalytic oxidation and gasification at relatively low temperatures.¹⁸

The aim of this work is to study and explain the differences in the activity and stability of commonly used Pt, Pd and Ni catalyst metals supported on different carbon based materials in the steam reforming of ethanol at low to moderate temperatures.

Experimental

Catalyst preparation, characterization and activity test

The catalyst materials were prepared as described earlier.¹⁹ In brief, multi-walled carbon nanotubes (CNTs, purchased from Sigma-Aldrich, >90% carbon basis, outer diameter of 10–15 nm, inner diameter of 2–6 nm and length of 0.1–10 μm) were pretreated by refluxing in cc. nitric acid for 8 h, followed by washing with deionized water and drying. The graphite (GC, Aldrich, <20 μm) was used as received. To anchor Pt, Pd or Ni nanoparticles, the powders were suspended in toluene by ultrasonic agitation (3 h) then the metal precursor (Pt(acac)₂, Pd(acac)₂ or Ni(acac)₂) was added to the dispersion. The amount of the precursor was adjusted to reach a corresponding nominal metal loading of 2 wt% in the products. The mixture was stirred at room temperature overnight then dried at 80 °C. The samples were calcined in air at 300 °C for 1 h then reduced in 15% H₂/Ar atmosphere at 500 °C for 5 min. The Ni(acac)₂ decorated samples were decomposed in two steps; first at 185 °C for 2 h and then at 380 °C for 1 h, which is followed by reduction at 250 °C for 15 min.

The average particle size was determined with transmission electron microscopy (TEM, Leo 912 Omega, acceleration voltage of 120 kV, LaB6 filament) by analyzing more than 135 particles of each specimen. The elemental concentrations were measured from 6 different locations in each sample with an EDX analyzer (Inca, Oxford Instruments) installed on a Zeiss Ultra plus field emission scanning electron microscope. The crystal structure of active metals was determined with powder X-ray diffraction (XRD, Siemens D5000, CuKα). Chemisorption of H₂ was measured with a Micromeritics AutoChem 2910 apparatus. The catalyst (100 mg) was heated (10 °C min⁻¹) under an Ar flow to 340 °C where kept for 30 min under 5% H₂/Ar flow, flushed with Ar for 30 min and cooled to room temperature. Hydrogen temperature-programmed reduction (TPR-H₂) measurements were performed using an adsorption analyzer AutoChem 2920 (Micromeritics, USA). The catalyst (100 mg) was pretreated under a He flow at 120 °C for 30 min (heating rate 10 °C min⁻¹). The TPR profile was measured using 5% H₂/Ar (flow rate 25 mL min⁻¹) and a heating ramp of 10 °C min⁻¹ from the ambient temperature up to 500 °C. The change in H₂ concentration was monitored with a TCD detector.

The activity tests of the samples were performed as follows. First, the samples (100 mg) were pretreated and reduced by heating to 350 °C with a heating rate of 10 °C min⁻¹ under 20% H₂/N₂ atmosphere, followed by reduction at 350 °C for 30 min and cooling down in reductive atmosphere. The catalytic activity of the samples was tested from 150 °C to 450 °C (heating rate of 10 °C min⁻¹) and the formed products were analyzed by the means of FTIR spectrometry and thermal conductivity analysis (XMTC H₂ analyzer). A water–ethanol mixture with a molar ratio of 3 : 1 at a feed rate of 0.09 mL min⁻¹ was vaporized at 180 °C and introduced to the reaction zone by N₂ carrier. The total gas flow rate of the reactants and the carrier gas was 600 mL min⁻¹ corresponding to ∼30 mL min⁻¹ ethanol and ∼90 mL min⁻¹ of water as reactants, and 480 mL min⁻¹ N₂ as balancing gas (at 180 °C).^11,20 The selectivity of hydrogen (S_H2) was calculated with the following equation:¹¹

Results and discussion

The activity of each catalyst was tested in low temperature SRE. From the ethanol conversion (Fig. 1) it is clear that only CNT supported catalysts had considerable activity with each metal, whereas in the case of the graphite supported catalysts only nickel had any activity (∼50% at 450 °C). Of all the studied catalysts Ni/CNT showed the highest ethanol conversion ∼100% at 450 °C, whereas Pd had very limited activity in SRE. The main product for each catalyst was H₂ but also considerable amounts of other products such as CH₄, CO₂, CO and CH₃CHO (up to 2.5 vol%, Fig. S1–S3†) were formed. The yield and selectivity of the catalysts towards H₂ was moderate (∼13% and ∼33% on Ni/GC and ∼19% and ∼42% on Ni/CNT, respectively). The ethanol conversion over the CNT supported catalyst was ∼100% in comparison with ∼50% on the graphite based catalysts indicating that CNTs are more feasible than GC in the studied reaction temperature range. The relatively high amounts of methane (∼2 vol%), CO (∼0.8 vol%) and CO₂ (∼1 vol%) are understandable considering the high activity of nickel in the methanation reaction.²¹ In addition, the Pt/CNT catalysts with a relatively high conversion (∼70% at 410 °C) showed poor selectivity towards H₂ as also methane (∼2.3 vol%) and CO (∼1.4 vol%) formed in considerable quantities. It is worth mentioning that the Pt/CNT catalyst was the only catalyst that showed decrease in ethanol conversion at high temperatures (above 410 °C), while for all the other catalysts the maximum conversion was at the end of studied temperature range.

Fig. 1 Comparison of activities of different carbon-supported metal catalysts i.e. carbon nanotube (CNT) (a and b) and graphitic carbon (GC) (c and d) in steam reforming of ethanol (results of nickel decorated samples adapted from ref. 20).

The results obtained are in line with the reference data (under identical operating conditions and reactor geometry) for Ni, Co and Pt supported on CNTs¹¹ and also for oxide supported catalysts such as Ru/CeO₂, Pt/CeO₂, Ni/Y₂O₃ and Ni/La₂O₃ (Table 1).^22–24 As reported earlier for CNT supported metals, the most active and selective catalysts were Ni and Co, having ∼100% conversion already at 375 °C and ∼40% H₂ selectivity at 450 °C, although at significantly higher catalyst loadings (i.e. ∼10%) compared to the samples (2%) studied in this work. The selectivity towards H₂ and CH₄ over 2% Pt/CNT catalyst was practically the same as for the previously reported ones with a 10% metal loading. It is also worth mentioning that at 300 °C the 10% Ni/CNT had a clear decrease in ethanol conversion caused by deactivation of the catalyst by coke formation, however this phenomenon was not detected (Fig. 1a and b) in this work. The particle size reported for 10% Ni/CNT was slightly higher than 2% Ni/CNT catalyst, (∼5 nm) and the coke formation on the spent catalyst was confirmed by TEM.¹¹

Table 1 Comparison of low temperature SRE catalyst activities to the values reported in the literature with the steam to ethanol molar ratio of 3 : 1. For comparison, the total gas flow rates are normalized to STP condition, i.e. to standard temperature (0.0 °C) and pressure (100 kPa)

Catalyst	gcat (g)	Reactant flow rate (mmol min^−1)		Total gas flow (mL min^−1)	T (°C)	X (EtOH) (%)	S (H2) (%)
		EtOH	H2O
2% Ni/CNT	0.1	0.827	2.484	362	450	100	42
2% Ni/GC	0.1	0.827	2.484	362	450	50	33
2% Pt/CNT	0.1	0.827	2.484	362	410	70	16
10% Ni/CNT (ref. 11)	0.1	0.827	2.484	362	450	100	40
10% Pt/CNT (ref. 11)	0.1	0.827	2.484	362	450	90	20
10% Rh/CNT (ref. 11)	0.1	0.827	2.484	362	450	20	13
10% Co/CNT (ref. 11)	0.1	0.827	2.484	362	450	100	40
1% Ru/CeO2 (ref. 22)	0.05	0.046	0.138	64	450	>90	57
3% Pt/CeO2 (ref. 23)	0.05	0.220	0.661	1000	300	100	39
20% Ni/Y2O3 (ref. 24)	4	0.459	1.378	N.A.	320	93	53
15% Ni/La2O3 (ref. 24)	4	0.459	1.378	N.A.	320	>99	49

---

The side product distribution curves for the Ni catalysts (Fig. 2a and b and S1†) show that dehydrogenation of ethanol takes place first leading to the formation of acetaldehyde (150–325 °C). Above 350 °C, the concentration of acetaldehyde decreases and simultaneously methane, CO and CO₂ start to form indicating the decomposition of acetaldehyde and the onset of the water-gas shift reaction. Over Pt/CNT, decomposition of ethanol to CH₄ and CO is more pronounced than dehydrogenation and ethanol reforming (Fig. 2c and S2†).

Fig. 2 Comparison of product distributions calculated on particular conversions of most active catalysts (a) Ni/CNT, (b) Ni/graphite and (c) Pt/CNT.

The reasons for the superior activity of CNT supported catalysts compared to graphitic carbon catalysts are associated with differences in the texture and microstructure of the two support materials. Although CNTs and graphite have basically the same crystal structure and chemical composition (sp² hybridized carbon with hexagonal layered lattice) the surface for CNTs is curved and has a significantly larger specific surface area than that of the graphite microparticles. Usually, a high specific surface area correlates with high activity because the catalyst particles tend to have a smaller size (and consequently a larger active area) compared to those synthesized on less nanostructured surfaces. Accordingly, the porous CNT powder-like support is more suitable for creating better dispersed catalysts having high activity in SRE (Table 2).

Table 2 Surface properties (adapted from ref. 20), metal dispersion, and temperature programmed reduction of nickel decorated catalysts

Catalyst	Specific surface area (m^2 g^−1)	Pore volume (cm^3 g^−1)	Pore size (nm)	Ni dispersion (%)	TPR-H2
					Tmax	n (H2) (mmol g^−1)
Ni/CNT	284	0.8	10.2	9.6	346 °C	0.89
					391 °C
Ni/GC	12	0.02	7.9	6.7	279 °C	0.54

---

The dispersion of nickel catalyst measured by H₂ chemisorption correlates with the average particle size determined by TEM (Tables 2–4). Temperature programmed reduction (Table 2, Fig. S4†) of nickel catalysts showed one reduction peak for graphite and two peaks for the CNT supported catalyst similar to that reported earlier for the Ni/CNT catalyst.¹⁸ The first H₂ consumption peak (∼280 to ∼350 °C) in both catalysts is related to the reduction of nickel oxide,^25,26 while the second peak is due to the reduction of iron residuals in the CNTs (catalyst for CNT growth). Reduction of nickel occurs at lower temperatures when supported on graphite than on CNTs. The consumption of hydrogen was lower for the low surface area GC than for the higher specific surface area CNT indicating that not only the metal but also the support is being somewhat reduced during the TPR measurements in the case of CNTs. The presence of acetaldehyde and the lack of ethylene (Fig. 2 and S1–S3†) side-products seem to support this, since acetaldehyde is a typical side-product of SRE for a reducible support as found earlier for CeO₂.²²

Although, the existence of iron residuals in the samples suggests a possibility that Fe could act as a co-catalyst,²⁷ it is very unlikely at least for two reasons. Metal impurities on the surface of CNTs have been removed by the acidic pre-treatment, thus any iron left in the CNTs after acid washing is typically in the inner hollow cavity of CNTs, which is hard to be accessed by the reactants. In addition, the activity of both pristine and acid pre-treated CNTs in SRE has been found negligible,²⁰ thus the promoting effect of Fe residuals may be ruled out for our samples.

Transmission electron micrographs of the catalysts (Fig. 3) show the structural differences between different supports. Similar to that reported earlier,¹⁷ the support had a great effect on the particle size. The catalysts on graphite were found to be larger, and had a broader size distribution than on CNTs. Apparently, the acidic pre-treatment of CNTs form suitable locations on the substrate for the nucleation and growth of small metal particles. Similar phenomenon has been reported earlier with palladium decorated carbon nanofibers where the particle size of palladium increased with decreasing acidity.²⁸ Therefore, on CNTs we found systematically smaller metal particles than on graphite. On both studied supports, the palladium particles had the highest average size, whereas the nickel and platinum particles had similar size distributions on similar supports.

Fig. 3 TEM micrographs of carbon nanotubes (a–c) and graphite (d–f) decorated with Pt (first row), Pd (middle row) and Ni (last row). Image on the left hand side shows the material before and on the right hand side after the use in reaction.

Although, the original particle size of different metals/metal oxides supported on CNTs do not differ significantly, their activities are merely different showing maximum ethanol conversion of 100% for Ni, 75% for Pt and only 10% for Pd. These results are in reasonable agreement with the recent study by Lin and co-workers claiming the high d-character of the metal–metal bonding and the promoting effect on the C–C and C–H scission due to the high Ni–C bonding stability make Ni a highly active metal for the steam reforming of ethanol.²⁹

X-Ray diffraction analysis (Fig. S5†) shows highly broadened reflections for the catalyst nanoparticles as expected for the very small diameters also seen by TEM. Pt and Pd are found in its metallic form, while Ni showed no reflections mainly because of the small size and concentration in the sample. The elemental concentrations (Tables 3 and 4) of the samples were measured before and after the reaction and the actual metal concentrations in the fresh and spent samples were close to the nominal value (2 wt%). The amount of oxygen according to EDX was found to decrease significantly in the spent catalysts, indicating that the support must have undergone reduction either during the pretreatment (reduction right before the reaction) and/or during the reaction thus loosing oxygen from the surface. In addition, an increase in the carbon content in the used catalyst was also observed (Tables 3 and 4) particularly for those being active in SRE. Carbonaceous products depositing within the catalyst nanocomposites as a consequence of side reactions e.g. decomposition of methane, polymerization of ethylene and Boudouard reaction³⁰ can explain well the increased carbon content. The formation of soot/coke on the CNT supported catalyst during the SRE reaction has also been reported earlier.^11,13 A closer examination of the formed carbon content with TEM revealed that not only amorphous carbon but also carbon nanotubes on both Ni/CNT and Ni/GC (Fig. 3c and f) and on Pd/GC (Fig. 3e) were formed.

Table 3 Particle sizes and the elemental concentrations of CNT supported catalysts before and after the use in steam reforming of ethanol

	Particle size (nm)		Elemental concentrations (wt%)
	Fresh catalyst	Spent catalyst	Fresh catalyst	Spent catalyst	ΔC^a
a (mused − mfresh)/mfresh × 100%.
Pt/CNT	1.9 ± 0.7	1.9 ± 0.6	Pt 2.3 ± 0.7	Pt 2.8 ± 1.0	1.8
			C 86.9 ± 0.9	C 88.5 ± 1.9
			O 8.8 ± 1.8	O 4.6 ± 1.9
			Al 1.3 ± 0.3	Al 2.3 ± 0.9
			Fe 0.8 ± 0.3	Fe 1.8 ± 0.7
Pd/CNT	2.7 ± 2.1	2.4 ± 2.4	Pd 3.6 ± 2.3	Pd 2.5 ± 0.5	2.2
			C 83.0 ± 1.9	C 84.8 ± 3.4
			O 11.3 ± 3.5	O 8.7 ± 1.2
			Al 1.1 ± 0.4	Al 3.1 ± 2.0
			Fe 1.0 ± 0.9	Fe 1.0 ± 0.4
Ni/CNT	2.0 ± 0.8 (ref. 20)	2.0 ± 0.7	Ni 1.4 ± 0.4	Ni 1.7 ± 0.7	10.1
			C 84.1 ± 1.2	C 92.6 ± 0.4
			O 12.9 ± 1.9	O 3.6 ± 0.8
			Al 0.9 ± 0.2	Al 1.3 ± 0.2
			Fe 0.8 ± 0.4	Fe 0.9 ± 0.3

---

Table 4 Particle sizes and the elemental concentrations of graphite supported catalysts before and after the use in steam reforming of ethanol

	Particle size		Elemental concentrations (wt%)
	Fresh catalyst	Spent catalyst	Fresh catalyst	Spent catalyst	ΔC^a
a (mused − mfresh)/mfresh × 100%.
Pt/GC	3.9 ± 2.4	5.1 ± 1.6	Pt 1.7 ± 0.9	Pt 3.0 ± 1.7	11.2
			C 86.5 ± 2.1	C 96.2 ± 1.8
			O 11.9 ± 2.5	O 0.8 ± 0.3
Pd/GC	9.8 ± 4.7	9.4 ± 3.9	Pd 2.6 ± 1.8	Pd 3.1 ± 1.5	4.6
			C 86.7 ± 0.9	C 90.7 ± 2.3
			O 10.6 ± 1.8	O 6.2 ± 1.6
Ni/GC	3.3 ± 1.4 (ref. 20)	4.5 ± 1.8	Ni 2.0 ± 0.6	Ni 2.5 ± 2.1	9.5
			C 88.8 ± 1.6	C 97.2 ± 2.5
			O 9.2 ± 1.3	O —

---

Growth of carbon nanofibers/tubes in the course of the steam reforming reaction is not surprising considering that alcohols are excellent precursors for catalytic chemical vapor deposition of filamentous carbons on catalyst nanoparticles of the iron group elements even at moderate temperatures.^31–34 Typically, the carbon content is mainly graphite below 400 °C and carbon nanotubes above that with steam/ethanol ratios under 4 : 1.³⁵

According to the vapor–liquid–solid model,^32,36,37 it is the decomposition of the organic feedstock limiting the growth at the low temperature regime, thus the relatively reactive oxygenates compared to hydrocarbons are easier to dehydrogenate and also more suitable for growing carbon from those. In agreement with our results, noble metals tend to form less carbon nanofibers/tubes than nickel catalysts because of the poor solubility of carbon in those.^22,38,39

As expected, the catalyst size on the CNT supported catalysts did not change during the steam reforming reaction. As reported¹⁹ for thermally aged Pt and Pd supported CNTs and GC catalysts, the CNTs provide stable support for both metals whereas in case of Pt/GC the particle size increases due to surface diffusion of the metal.

Whether the formation of carbon structures on the catalysts caused any deactivation or not should be studied in the future. However, one could suppose that the catalysts that were active in growing larger amounts of well-structured carbon nanofibers/tubes (as the side-products of the reforming reaction) are less deactivated than the catalysts that produce soot/coke, since these latter ones passivate the surface of the metal, i.e. poison the catalyst. This seems to be supported by the fact, that above ∼425 °C the ethanol conversion and hydrogen production are getting decreased over Pt/CNT, which can be a consequence of the amorphous carbon found by TEM on the spent catalyst.

Conclusions

In this paper, the influence of different carbon supports and catalyst nanoparticles (Pt, Pd, Ni/NiO) on the steam reforming of ethanol were studied, and the differences found in their activity and stability was elucidated. The CNT supported catalysts showed generally much better activity and stability than their counterparts supported on GC. The higher specific surface area and porous structure can explain the differences between the two support materials. Furthermore, the acidic pre-treatment of the nanotubes, which is known to promote the formation of smaller metal nanoparticles as compared to pristine carbon surfaces, can also contribute to the better catalytic activity of catalyst on such supports. The differences in the activity of the applied metal (oxide) catalyst nanoparticles are explained by several factors. On one hand, the d-character of the metal–metal bonding and the promoting effect on the C–C and C–H scission may give a reasonable account on the high activity of Ni in the reaction. On the other hand, the solubility of carbon (which is a side-product of the reaction) in the catalyst seems to have an important effect on the overall performance of the catalyst. According to the observations made, metals known to have typically good carbon solubility at elevated temperatures (such as Ni and Pd) can tolerate more the deposition of carbon on their surface. This is because the carbon can diffuse in the catalyst particle and then precipitate in the form of nanofibers/nanotubes thus avoiding the build-up of continuous coke and poisoning of the surface.

Acknowledgements

A.-R. Rautio is grateful for the post-graduate position received from the Graduate School in Electronics, Telecommunications and Automation (GETA).

References

1. J. A. Turner, Science, 2004, 305, 972.
2. J. D. Holladay, J. Hu, D. L. King and Y. Wang, Catal. Today, 2009, 139, 244.
3. A. Heinzel, B. Vogel and P. Hübner, J. Power Sources, 2002, 105, 207.
4. Q. Ming, T. Healey, L. Allen and P. Irving, Catal. Today, 2002, 77, 51.
5. D. K. Liguras, D. I. Kondarides and X. E. Verykios, Appl. Catal., B, 2003, 43, 345.
6. S. Duan and S. Senkan, Ind. Eng. Chem. Res., 2005, 44, 6381.
7. D. Janas, S. K. Kreft and K. K. K. Koziol, J. Ind. Eng. Chem., 2015, 25, 222.
8. M. Ni, D. Y. C. Leung and M. K. H. Leung, Int. J. Hydrogen Energy, 2007, 32, 3238.
9. F. Aupretre, C. Descorme and D. Duprez, Catal. Commun., 2002, 3, 263.
10. A. C. Basagiannis, P. Panagiotopoulou and X. E. Verykios, Top. Catal., 2008, 51, 2.
11. P. K. Seelam, M. Huuhtanen, A. Sapi, M. Szabo, K. Kordas, E. Turpeinen, G. Toth and R. L. Keiski, Int. J. Hydrogen Energy, 2010, 35, 12588.
12. G. Özkan, S. Gök and G. Özkan, Chem. Eng. J., 2011, 171, 1270.
13. T. Hou, L. Yuan, T. Ye, L. Gong, J. Tu, M. Yamamoto, Y. Torimoto and Q. Li, Int. J. Hydrogen Energy, 2009, 34, 9095.
14. E. Lopez, J. Kim, A. M. Shanmugharaj and S. H. Ryu, J. Mater. Sci., 2012, 47, 2985.
15. P. Tolmacsov, A. Gazsi and F. Solymosi, Appl. Catal., A, 2009, 362, 58.
16. N. Halonen, A. Rautio, A.-R. Leino, T. Kyllönen, G. Tóth, J. Lappalainen, K. Kordás, M. Huuhtanen, R. L. Keiski, A. Sápi, M. Szabó, Á. Kukovecz, Z. Kónya, I. Kiricsi, R. Vajtai and P. M. Ajayan, ACS Nano, 2010, 4, 2003.
17. A. Sapi, R. Remias, Z. Konya, A. Kukovecz, K. Kordas and I. Kiricsi, React. Kinet. Catal. Lett., 2009, 96, 379.
18. A.-R. Leino, M. Mohl, J. Kukkola, P. Mäki-Arvela, T. Kokkonen, A. Shchukarev and K. Kordas, Carbon, 2013, 57, 99.
19. K. Kordas, A.-R. Rautio, M. Mohl, P. Mäki-Arvela, J.-P. Mikkola, D. Murzin, L. Ge, P. M. Ajayan and R. Vajtai, On the interaction of metal nanoparticles with supports, Unpublished data. Submitted to Topics in Catalysis at the proceedings of the Nordic Symposium on Catalysis.
20. P. K. Seelam, PhD thesis, University of Oulu, Finland, 2013.
21. R. Razzaq, H. Zhu, L. Jiang, U. Muhammad, C. Li and S. Zhang, Ind. Eng. Chem. Res., 2013, 52, 2247.
22. A. Erdöhelyi, J. Raskó, T. Kecskés, M. Tóth, M. Dömök and K. Baán, Catal. Today, 2006, 116, 367.
23. P. Ciambelli, V. Palma and A. Ruggiero, Appl. Catal., B, 2010, 96, 190.
24. J. Sun, X.-P. Qiu, F. Wu and W.-T. Zhu, Int. J. Hydrogen Energy, 2005, 30, 437.
25. A. Irionda, V. L. Barrio, J. F. Cambra, P. L. Arias, M. B. Guemez, M. C. Sanchez-Sanchez, R. M. Navarro and J. L. G. Fierro, Int. J. Hydrogen Energy, 2010, 35, 11622.
26. T. de Freitas Silva, C. G. M. Reis, A. F. Lucredio, E. M. Assaf and J. M. Assaf, Fuel Process. Technol., 2014, 127, 97.
27. L. Chen, C. K. S. Choong, Z. Zhong, L. Huang, T. P. Ang, L. Hong and J. Lin, J. Catal., 2010, 273, 197.
28. A. V. Tokarev, E. V. Murzina, K. Eränen, H. Markus, A. J. Plomp, J. H. Bitter, P. Mäki-Arvela and D. Yu Murzin, Res. Chem. Intermed., 2009, 35, 155.
29. J. Lin, L. Chen, C. K. S. Choong, Z. Zhong and L. Huang, Sci. China: Chem., 2015, 58, 60.
30. A. Haryanto, S. Fernando, N. Murali and S. Adhikari, Energy Fuels, 2005, 19, 2098.
31. H. Wang, Y. Liu, L. Wang and Y. N. Qin, Chem. Eng. J., 2008, 145, 25.
32. H. Sugime and S. Noda, Carbon, 2012, 50, 2953.
33. S. Murayama, R. Kojima, Y. Miyauchi, S. Chiashi and M. Kohno, Chem. Phys. Lett., 2002, 360, 229.
34. N. Halonen, A. Sápi, L. Nagy, R. Puskás, A.-R. Leino, J. Mäklin, J. Kukkola, G. Tóth, M.-C. Wu, H.-C. Liao, W.-F. Su, A. Shchukarev, J.-P. Mikkola, Á. Kukovecz, Z. Kónya and K. Kordás, Phys. Status Solidi B, 2011, 248, 2500.
35. F. Diaz Alvarado and F. Gracia, Chem. Eng. J., 2010, 165, 649.
36. N. Halonen, K. Kordás, G. Tóth, T. Mustonen, J. Mäklin, J. Vähäkangas, P. M. Ajayan and R. Vajtai, J. Phys. Chem. C, 2008, 112, 6723.
37. M. S. Dresselhaus, Y.-M. Lin, O. Rabin, M. R. Black and G. Dresselhaus, Nanowires, in Springer Handbook of Nanotechnology, ed. B. Bhushan, Springer-Verlag, Berlin, 2004, p. 105.
38. G. L. Selman, P. J. Ellison and A. S. Darling, Platinum Met. Rev., 1970, 14, 14.
39. J. R. Rostrup-Nielsen and J.-H. Bak Hansen, J. Catal., 1993, 144, 38.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c5ra07282d

---