1 Introduction

Ethanol steam reforming (ESR) reaction (C2H5OH + 3H2O → 6H2 + 2CO2) using an appropriate catalyst is an efficient route for renewable hydrogen production and has been highlighted in the literature [1,2,3,4,5,6]. In addition, ethanol is atmospheric carbon neutral since the amount of CO2 produced by steam reforming is consumed by the biomass growth, and this offers a nearly closed carbon loop not contributing to the greenhouse gas emissions [7, 8]. The selection and development of a suitable catalyst for ESR is a key aspect, once the catalyst must be stable, active, selective, and maximize hydrogen production while simultaneously avoiding the formation of byproducts (CO and CH4). Coke formation is a major issue in the ESR, which may lead to a decrease in catalytic activity and selectivity towards hydrogen, as well as to catalyst deactivation, limiting the industrial application [9, 10]. Carbon formation on the catalyst surface may take place via several reactions, such as ethanol dehydration to ethylene, followed by polymerization to coke; the “Boudouard” reaction; and decomposition of hydrocarbons (methane and ethylene) [11, 12]. The extent of each reaction depends on both chosen catalyst and reaction conditions.

Noble metal catalysts, such as Rh, Ru, Pd and Pt, can effectively break C–C and C–H bonds with a relatively good stability, but the fluctuating price and high cost limit their large-scale application [13, 14]. On the other hand, catalysts containing transition metals, mainly Ni [15,16,17] and Co [18,19,20], have been largely investigated as active catalysts for ESR reaction and exhibited catalytic activities comparable to those based on noble metals. Ni-based catalysts have been widely employed on commercial scale in reforming processes for more than 40 years [1], especially due to the excellent capability for C–C and C–H bond cleavage and low cost compared to expensive noble metals [21, 22]. However, Ni-based catalysts deactivate by carbon deposition and aggregation of active Ni particles more severely than noble metal catalysts, shortening their lifetime in the ESR reactions [1, 23]. Preventing the deactivation of Ni catalysts still remains a major challenge.

To minimize coke deposition and metal sintering, several alternatives have been suggested such as the addition of small amount of noble metals [7, 24]. Generally, it is possible to enhance the resistance to coke deposition and to prevent nickel sintering by adding small amount of noble metal promoters [25, 26]. Palma et al. [27] concluded that addition of small amount of Pt and Rh improves the catalytic performance and coke resistance in the ethanol reforming.

Pereira et al. [28] studied the effect of introducing small loading of Rh and Ru into Co/SiO2 catalysts for the auto-thermal reforming of ethanol, aiming a synergistic effect between Co and Rh or Ru. Authors concluded that the noble metal facilitates the reduction of cobalt under experimental conditions of oxidative steam reforming of ethanol due to the intimate contact between Co and the noble metal (Ru or Rh) phases in the silica-supported bimetallic systems. Doping supported transition metal catalysts with palladium has been presented as promising to enhance the reduction of Ni under mild conditions and to improve the air-resistibility of Ni originated from the stabilization effect of Pd [29, 30].

Here we studied the influence of palladium on the catalytic performance of Ni/SiO2 in the ethanol steam reforming for hydrogen production. Ni/SiO2 and Pd–Ni/SiO2 catalysts were prepared by incipient wetness impregnation method and characterized systematically by different techniques (XRD, XRF, N2 adsorption–desorption, H2-TPR and XPS) in order to understand the structure–activity relationship.

2 Experimental

2.1 Catalysts Preparation

Ni/SiO2 and Pd–Ni/SiO2 catalysts were synthesized by incipient wetness impregnation method [31]. The nominal loading amount of Ni and Pd were 10 and 1 wt.%, respectively. Commercial SiO2 gel powder (Sigma-Aldrich, 60–200 mesh) used as support was previously calcined at 650 °C for 6 h under heating rate of 5 °C/min in a muffle furnace. Nickel was impregnated on the silica support (1.001 cm3/g pore volume, obtained by N2 adsorption/desorption experiments) with an aqueous solution of nickel precursor salt Ni(NO3)2.6H2O Sigma-Aldrich. After impregnation, the sample was dried in muffle furnace at 110 °C overnight followed by calcinations in 2 steps: firstly at 350 °C for 3 h (heating rate of 2 °C/min) and then at 650 °C for 5 h (5 °C/min) under static atmosphere. 1%Pd-10%Ni/SiO2 catalyst was prepared by sequential impregnation of palladium over Ni/SiO2 catalyst using a palladium nitrate aqueous solution (Pd(NO3)2, 10 wt.% in 10 wt.% nitric acid, Sigma-Aldrich). Subsequently, the sample was dried and calcined similarly. This first calcination step was carried out in order to prevent nickel to redissolve in the palladium solution during the second impregnation step. All catalysts were crushed and sieved to obtain the fraction between 0.18 and 0.12 mm, and from now on 10%Ni/SiO2 and 1%Pd-10%Ni/SiO2 catalysts are denominated as NiSiO and PdNiSiO, respectively.

2.2 Catalysts Characterization

Chemical composition analysis was performed by X-ray fluorescence spectroscopy (XRF) using a Rigaku spectrometer RIX 3100 model apparatus equipped with a rhodium standard tube as source of radiation. Around 300 mg of each sample was pelletized and analyzed quantitatively.

Textural properties were determined by nitrogen adsorption/desorption experiments at liquid nitrogen temperature using a Micromeritics ASAP2010 gas adsorption instrument. All samples were degassed under vacuum at 300 °C for 24 h prior to the measurements. The specific surface area was calculated using the Brunauer–Emmett–Teller (BET) method in a relative pressure range of 0.05–0.3. The pore size distribution was determined from desorption branches by the Barrett–Joyner–Halenda (BJH) method.

X-ray powder diffraction (XRD) measurements were performed in a Miniflex Rigaku diffractometer. The XRD patterns were collected using CuKα radiation (λ = 1.5406 Å) operated at 30 kV and 15 mA in the range from 10 to 80° with a step size of 0.05° and counting time of 1 s per step. Crystalline phases were identified by using JCPDS (Joint Committee on Powder Diffraction Standards) database. The mean crystallite size was calculated using Scherrer equation. XRD patterns of the catalysts reduced ex-situ at the same conditions employed prior to the catalytic tests (i.e., 600 °C for 2 h under 50 mL/min of H2) were also obtained in order to analyze crystallographic features of the materials resulting from the activation pretreatment.

Temperature-programmed reduction with hydrogen (H2-TPR) experiments were performed on a system equipped with a TCD detector employing a mixture of 1.53 vol.% H2/Ar (30 mL/min) flowing through the sample. Prior to H2-TPR analysis, the samples were pretreated at 150 °C for 30 min under argon flow of 30 mL/min and then cooled down to room temperature in argon gas flow before reduction. The temperature was linearly raised up to 1000 °C at 10 °C/min, and hydrogen consumption was continuously monitored by thermal conductivity detector (TCD). A trap of molecular sieve ensured the water retention prior to the detector, so that only hydrogen consumption was measured.

X-ray photoelectron spectroscopy (XPS) was performed to determine the chemical state and the surface composition of the calcined catalysts. XPS analysis was carried out using an ESCALAB 250 spectrometer (Thermo Scientific), employing monochromatic Al Kα (1486.6 eV) as X-ray source. The C 1 s signal at 284.6 eV was binding energy reference. Spectra were analyzed using a Gaussian-Lorenzian peak shape obtained from CasaXPS® software version 2.3.15.

Coke deposition on spent catalysts was determined by thermogravimetric analysis (TGA). Each sample was heated from room temperature (10 °C/min) up to 1000 °C under flow of air (60 mL/min). Quantification of coke was calculated according to the equation:

$$C=\frac{{m}_{coke}}{{m}_{usedcatalyst}\, x \, t}$$

In which mcoke is the mass of coke on the catalyst calculated from TGA profile; musedcatalyst is the mass of catalyst remaining after TGA analysis, and t is the time-on-stream.

2.3 Catalytic Performance Test

The reforming reaction at stoichiometric conditions was conducted in a continuous fixed-bed quartz reactor at atmospheric pressure, and temperature from 200 up to 600 °C in steps of 100 °C. Initially, 200 mg of catalyst diluted with 400 mg of carborundum (SiC) was reduced in situ at 600 °C for 2 h flowing pure H2 (50 mL/min) and then cooled down to room temperature under He flow (50 mL/min). This reduction conditions was established based on H2-TPR analysis. Reactant mixture with H2O/ethanol molar ratio = 3/1 was injected with a syringe pump (0.05 mL/min) into a vaporizer at 200 °C and mixed with a nitrogen flow (50 mL/min) before entering the reactor. The gas hourly space velocity (GHSV) of ethanol and water was 48,500 h−1. Gaseous products were analyzed on line by gas chromatography (Shimadzu, model GC-2014) containing two columns (RT-QPLOT and Carboxen 1010) and equipped with a TCD using He as a carrier gas. The liquid phase was collected in the condensers and subsequently analyzed by a Shimadzu Prominence high-performance liquid chromatography (HPLC) equipped with a Bio-Rad Aminex HPX-87H column, using 0.01 M H2SO4 as eluent at 0.6 mL/min, and with UV and refractive index detectors. Time-on-stream stability tests were carried out at 500 °C in order to estimate deactivation of the catalysts. Ethanol conversion and selectivity to different gaseous products were calculated according to the equation used by Mondal et al. [7]. All the catalytic tests were carried out in duplicates and the values obtained for ethanol conversion showed a standard deviation below 3.0%.

Ethanol equilibrium conversion as a function of temperature was calculated taking into account the law of mass action described by Soave–Redlich–Kwong equation of state, and the equilibrium constant derived from the standard Gibbs free energy for the ESR reaction. According to thermodynamics analysis, ethanol equilibrium conversion is approximately total over the entire temperature range tested in this work (see Figure S1 in supplementary information, SI).

3 Results and Discussion

Table 1 shows results of textural and physicochemical properties of the catalysts and SiO2 support. The chemical compositions measured by XRF revealed that Ni and Pd contents are in good agreement with nominal composition, indicating that the synthesis method was appropriate to incorporate the metal loading.

Table 1 Physicochemical and textural properties of the catalysts and SiO2 support
Full size table

The nitrogen adsorption/desorption measurements and the Barrett Joyner Halenda (BJH) pore size distributions are shown in Fig. 1. All isotherms are similar to type IV according to the IUPAC classification, indicating the mesoporous nature of the materials [32]. In all the cases it was found a narrow pore size range with mean diameter about 100 Å, as evidenced in the BJH pore size distribution profiles (Fig. 1). Textural properties (surface area, pore volume, mean pore diameter) of the SiO2 support and catalysts obtained from the adsorption–desorption isotherms of nitrogen are also summarized in Table 1. BET surface area and pore volume of SiO2 support decreased after incorporation of the metals (Ni and Pd), which can be attributed to the partial plugging of SiO2 mesopores by NiO and PdO crystals limiting the accessibility of N2 molecules.

Fig. 1
figure 1

N2 adsorption–desorption isotherms and pore size distribution of the catalysts and SiO2 support

Full size image

The successful syntheses of the catalysts were also confirmed by X-ray diffraction (XRD) patterns in Fig. 2. All XRD patterns presented a diffuse peak around 2θ = 23°, corresponding to (101) diffraction plane, which agrees with that expected for α-cristobalite silica (JCPDS no. 39–1425). The XRD patterns of NiSiO and PdNiSiO catalysts exhibited the diffraction peak relative to silica, and additional peaks at 2θ = 37, 43 and 62°. These peaks correspond to (111), (200) and (220) diffraction planes related to NiO phase (JCPDS no. 41-1107). This result also suggests that the Pd addition did not modify the nature of the nickel phase. Importantly, a small peak located at 2θ = 34.2° in the XRD pattern of PdNiSiO catalyst corresponding to (101) diffraction plane can be attributed to PdO phase (JCPDS no. 41-1107). The mean crystallite sizes for PdO and NiO were estimated by using the Scherrer equation and the values are given in Table 1. Therefore, the second calcination after addition of palladium did not affect the dimension of NiO crystallites. The catalysts were reduced ex-situ at the same conditions employed prior to the catalytic tests and were immediately analyzed by XRD. Diffractograms are displayed in Fig. S2 (SI) and we verified absence of diffraction peaks for metallic palladium phases for both NiSiO and PdNiSiO. This finding suggests that Pd0 is highly dispersed on SiO2 surface. On the other hand, metallic Ni phase (JCPDS no. 65-2865) was identified with mean crystallite size of 195 and 167 Å for NiSiO and PdNiSiO catalysts, respectively (i.e., smaller for the bimetallic catalyst). Therefore, these mean coherent domain sizes suggest an estimate of Ni dispersion higher on the bimetallic catalyst after reduction.

Fig. 2
figure 2

XRD patterns of the calcined catalysts and SiO2 support

Full size image

H2-TPR experiments were performed to determine the appropriate reduction temperature to activate the catalysts prior the tests, as well as to infer about the chemical interaction between metals (Ni and Pd) and SiO2 support. The reduction profiles of the catalysts and SiO2 support are depicted in Fig. 4. All of the catalysts exhibited a H2-TPR profile similar and relatively complex formed by two consecutive temperature regions. The reduction peaks in the low temperature range (around 300–600 °C) can be attributed to the reduction of Ni species (NiO → Ni0), which are free nickel oxide species and have a weak interaction with the silica support; whereas the reduction peaks in the high temperature range (620–900 °C) corresponds to a stronger interaction of NiO species with support [33, 34]. This result is in good agreement with literature [35,36,37,38,39,40] and clearly demonstrate different degrees of interaction between NiO species and SiO2 support. For PdNiSiO catalyst, a small reduction peak at lower temperature (about 95 °C) can be assigned to the reduction of PdO (detected previously by XRD measurement) to metallic Pd [41, 42]. In addition, we detected a significant consumption of hydrogen at room temperature for PdNiSiO catalyst. It is well known that the reduction of PdO normally takes place at room temperature [43,44,45], therefore the peak at 95 °C suggests some PdO having some stronger interaction with nickel and/or SiO2. Our results indicate that PdO species reduces predominantly at room temperature, but some fraction of PdO reduces at 95 °C.

In the case of PdNiSiO catalyst, the main peak for NiO reduction (350–600 °C) is larger than the corresponding one for NiSiO, as depicted in Fig. 3. That happens because part of NiO species that reduces at high temperature on NiSiO (600–850 °C) shifts to lower temperature on PdNiSiO (at the range of 500–600 °C roughly). Therefore, by adding Pd on NiSiO a fraction of NiO is more easily reduced, concentrating this reduction at the range of 350–600 °C (and not at 350–800 °C as observed for NiSiO). According to several studies reported in literature [8, 42, 46,47,48] this occurs because palladium catalyzes the reduction of Ni species (with strong and weak interaction with support) by shifting the Ni2+ peak towards lower temperatures as a result of H2 spillover process.

Fig. 3
figure 3

H2-TPR profiles of the calcined catalysts and SiO2 support

Full size image

Quantitative analysis of H2-TPR profiles revealed a high reduction degree for nickel and palladium species on all the catalysts, as shown in Table 2 by comparing the theoretical and experimental H2 uptake to reduce NiO and PdO species into Ni and Pd, respectively. The total amount of consumed hydrogen was similar for both catalysts, indicating that the reduction degree was independent from the presence of palladium. The noble metal increased the ability of NiO to reduce at lower temperature but did not affect the reduction degree. The total H2 consumption by PdO species (19 μmol) was calculated by the combined reduction of Pd2+ species at room temperature and species having a stronger interaction with the support at 95 °C, and it was equivalent to the theoretical requirement for complete reduction of PdO species taking into account the real loading of Pd as provided by XRF analysis (Table 1).

Table 2 Quantitative results for H2-TPR analysis
Full size table

The surface species were investigated by XPS analysis and the XP spectra of Ni 2p, Si 2p and O 1 s regions for NiSiO are depicted in Fig. 4. Peaks related to impurities were not detected. The spectrum relative to nickel exhibited two peaks around 855.7 and 861 eV corresponding to Ni 2p3/2 line, followed by two relatively intense peaks at 873.7 and 880.3 eV which are associated to Ni 2p1/2 line. These binding energies are indicative of the existence of Ni species in the form of bivalent state as NiO [48]. The Si 2p and O 1 s binding energies were found to be 103.4 and 532.7 eV, respectively, characteristic of silicon dioxide [49]. These findings indicate a similar oxidation state of the elements on surface and in the bulk, as evidenced previously by XRD analysis (Fig. 1).

Fig. 4
figure 4

XP spectra of the calcined NiSiO catalyst

Full size image

The XP spectra of the bimetallic PdNiSiO catalyst are presented in Fig. 5. The spectrum of palladium can be deconvoluted into two pairs of doublets. The doublet centered at 338.1 and 344.0 eV correspond to Pd2+ species [50], confirming that palladium exists on the surface as PdO. Conversely, the doublet located at 336.6 and 342.1 eV agrees with those reported for metallic Pd (Pd0). This finding can be attributed to the reduction of PdO species during XPS measurements probably resulting from exposing the sample to the X-ray beam.

Fig. 5
figure 5

XP spectra of the calcined PdNiSiO catalys

Full size image

The Ni 2p region was decomposed into four contributions typical of Ni2+ species in the NiO form, similar to NiSiO catalyst (Fig. 4). The spectrum relative to the Si 2p region can be fitted to two peaks at 102.8 and 104.6 eV. The first peak is associated to the silicon atom bonded to hydroxyl group (Si–OH), while the second one is related to -O-Si–O- on the surface silicon dioxide [51, 52]. The spectrum of O 1 s showed two components centered at 532.1 and 533.8 eV which are associated to Si–O-Si and Si–O-H bonds, respectively [53, 54].

3.1 Evaluation of Catalysts Performance

The catalytic activity in the ESR reaction was evaluated at the temperature range of 300–600 °C. Ethanol conversion and product distribution for both NiSiO and PdNiSiO catalysts are presented in Figs. 6 and 7, respectively. On all catalysts, the conversion of ethanol increased significantly as a function of the reaction temperature, reflecting the endothermic feature of the reforming reaction. For NiSiO catalyst, the total conversion of ethanol was obtained from 400 °C. On the other hand, PdNiSiO catalyst led to total ethanol conversion from 500 °C. At the temperature range of 300–600 °C the equilibrium conversion of ethanol is ≥ 99.6% (Figure S1 in Supplementary Information), therefore a more suitable comparison of NiSiO and PdNiSiO can be done at 300 and 400 ºC. Both NiSiO and PdNiSiO presented similar ethanol conversion at 300ºC (≈42%), but the bimetallic catalyst showed higher H2 selectivity (≈50%) than the monometallic one (≈40%). At 400ºC, NiSiO shows higher activity than PdNiSiO (XEtOH = 100% vs 90%, respectively), and only at 500ºC the bimetallic material reaches the same activity of NiSiO for ethanol conversion (≈100%). The bimetallic PdNiSiO seems to be slightly less active than NiSiO at low temperatures, but more selective to H2. Moreover, the rates of coke formation calculated after 30 h on stream indicate that the bimetallic PdNiSiO is less prone to deactivation, as will be discussed afterwards.

Fig. 6
figure 6

Effect of temperature on ethanol conversion (filled symbols) and product selectivity (empty symbols) over NiSiO catalyst

Full size image
Fig. 7
figure 7

Effect of temperature on ethanol conversion (filled symbols) and product selectivity (empty symbols) over PdNiSiO catalyst

Full size image

Regarding product distribution, the main products formed over the catalysts were H2, CO, CO2 and CH4. With respect to H2 production, PdNiSiO catalyst exhibited an initial H2 selectivity of 50% at 300 °C, while the initial H2 selectivity obtained over the NiSiO catalyst was 40% under the same ethanol isoconversion condition (approx. 42%). This result suggests that the addition of palladium have positive effects on the initial H2 selectivity at lower temperature (300 °C) and can be attributed to the ability of Pd to promote steam reforming and water gas shift reaction, and to suppress coke formation reactions [7]. However, the maximum H2 selectivity was equal for both catalysts (SH2 = 60%) and was obtained at 600 °C. For NiSiO catalyst, CO and CH4 selectivities slightly decreased by increasing temperature from 400 to 500 °C, while selectivity to H2 and CO2 enhanced, suggesting that water gas shift reaction (CO + H2O → CO2 + H2) and methane steam reforming (CH4 + H2O → CO + 3H2) were favored. Methanation and WGS reaction are the main parallel side reactions in the ESR, and significantly affect the product distribution [55]. Similar behavior was observed over PdNiSiO catalyst.

Above 500 °C, both catalysts exhibited an increase in CO selectivity, while the CO2 concentration diminished, implying in the reverse water gas shift reaction (CO2 + H2 → CO + H2O) [56]. Interestingly, for PdNiSiO catalyst one can observe a decrease in the selectivity to H2 and CO when the temperature increases from 300 to 400 °C. That can be attributed to the methanation reaction (CO + 3H2 → CH4 + H2O) [57], for which both H2 and CO are converted into CH4, which is a coproduct experimentally quantified at that temperature range. Methanation is an endothermic reaction and thus is favored by the high temperature. The presence of intermediate products (acetaldehyde or ethylene) was not detected in the product stream at the range of temperature studied. This result is in agreement with previous studies which evidenced that temperatures in the range of 450–600 °C led to higher ethanol conversion and H2 selectivity [58,59,60,61].

It was expected that loading palladium in the NiSiO catalyst could increase both ethanol conversion and selectivity to H2 in the product stream. However, this promotion effect was not observed over all temperature range studied. On the other hand, the incorporation of Pd into NiSiO catalyst caused a significant increase in the initial H2 selectivity and decrease in the concentrations of CO in the gaseous products at 300 °C. This result can be attributed to the presence of well-dispersed palladium species which are responsible for facilitating the reduction of nickel oxide species and shifting its reduction to lower temperatures, and thus enhancing the chemical activity of Pd-promoted NiSiO catalyst. This hypothesis is supported by XRD and H2-TPR analysis, as discussed previously.

In order to verify the relevance of NiSiO and PdNiSiO catalysts, their performance was compared to previous studies over similar nickel-based catalysts in the ESR reaction (see Table 3). The differences in the results can be explained by the nature of the catalysts and also the different operating conditions. Comparing our results to those reported in the recent literature (Table 3), most of findings indicate selectivity to H2 in between 48 and 70%, and our results fit into this range. However, concerning the selectivity to CO, NiSiO and PdNiSiO lead to lower values (5.5–6.0%) than most of reported studies (9–30%), which is interesting once low concentration of CO is desired. We can conclude that supported-Ni promoted catalyst are suitable materials for hydrogen production by ethanol steam reforming, but that CO selectivity should be reduced for fuel cell applications.

Table 3 Results reported in literature for ethanol conversion and hydrogen selectivity over various catalysts, and NiSiO and PdNiSiO of this present study
Full size table

3.2 Stability Test

The ESR is considered as a harsh reaction, generally causing serious catalyst deactivation after a long-term operation. Therefore, stability is a key issue in the development of catalysts for H2 production by ESR. This work evaluated the stability of the catalysts during 30 h on stream at 500 °C. Figure 8 shows ethanol conversion and selectivity to compounds (H2, CO, CO2 and CH4) as a function of time for NiSiO and PdNiSiO catalysts. Both materials presented ethanol conversion of 100% during 30 h on stream. H2 selectivity decreased significantly after 8 h for NiSiO and PdNiSiO; on the other hand, CO and CO2 selectivities remained approximately constant and CH4 selectivity slightly increased as a function of time. These results are attributed to the occurrence of methanation reaction which is favored at high temperatures. After 8 h on stream, the product distribution remained quite constant until 30 h. The results clearly demonstrate a similar stability for both catalysts in the ESR reaction.

Fig. 8
figure 8

Time-on-stream stability results for NiSiO catalyst (upper) and PdNiSiO (down) at 500 °C for 30 h in the ethanol steam reforming

Full size image

The catalyst deactivation during the ESR process is mainly caused by carbon deposition. To check the deactivation, the spent catalysts (after 30 h on stream) were characterized by TGA analysis and the profiles are illustrated in Fig S3 (SI). For both NiSiO and PdNiSiO catalysts the weight loss occurred at the range of 550–650 °C, which can be attributed to the combustion of graphitic/filamentous coke on the catalyst surface [68]. Saturated C–C bond such as paraffinic or amorphous carbon tends to oxidize at lower temperatures (Toxidation = 400–550 °C) than unsaturated hydrocarbon such as olefins, graphitic and filamentous carbon (Toxidation = 550–650 °C). As reported in literature, the carbon deposited on the catalyst surface during ESR reaction can have amorphous or filamentous nature [69]. Amorphous carbon leads to a more severe deactivation if compared to filamentous carbon [66]. In this investigation, both NiSiO and PdNiSiO catalysts were stable after 30 h on stream and presented some coke formation which was quantified in terms of mgcoke gcat−1 h−1 (as detailed in the sequence). Also, TGA results of the spent catalysts suggested a filamentous nature of the coke as previously discussed. Such carbon does not block Ni0 active sites, and for this reason the catalytic activity remains constant (Fig. 8). However, excessive carbon accumulation may result in plugging of the reactor and pressure drop, which was not observed after 30 h on stream at the experimental conditions. Therefore, we suggest that the possible route for coke formation on NiSiO and PdNiSiO has been derived from Boudouard reaction (2CO → CO2 + C) and methane decomposition (CH4 → C + 2H2).

The coke formation rates were calculated from TGA profiles and results are summarized in Table 4. We verified that coke deposition on NiSiO catalyst (1.41 mgcoke gcat−1 h−1) was higher than on PdNiSiO (1.13 mgcoke gcat−1h−1), suggesting that addition of Pd suppresses coke formation. This result is in good accordance with previous works [8, 69, 70] showing that noble metals are more resistant to carbon formation than non noble metals. Table 4 allows a comparison between our results and other Ni-based catalysts reported in the recent literature [9, 71,72,73,74,75] under similar reaction conditions. PdNiSiO and NiSiO exhibited improved resistance to coke deposition, once catalysts reported in literature showed coke formation rates of one or even two orders of magnitude higher than reported in the present study. These results suggest that the catalysts developed in this work are promising and competitive candidate for ESR reaction.

Table 4 Carbon formation rates during time-on-stream stability tests in the ethanol steam reforming. Data from literature are shown for comparison
Full size table

4 Conclusions

In this study, NiO/SiO2 catalyst promoted with small amount of palladium was successfully synthesized by incipient wetness impregnation method and tested in the steam reforming of ethanol for hydrogen production. Palladium catalyzed the reduction of Ni species interacting strongly with the silica support. All catalysts displayed total ethanol conversion and high H2 selectivity (~ 60%) above 500 °C. The characterization results of the spent catalysts revealed that the loss of activity after 8 h on stream for both catalysts is due to the formation of filamentous/graphitic carbon. The Pd-promoted catalyst exhibited higher resistance to coke deposition and higher stability in comparison with the unpromoted material. Comparing our results with those reported in the recent literature, we can conclude that Ni/SiO2 and Pd–Ni/SiO2 catalysts are suitable materials for hydrogen production by ethanol steam reforming reaction, but CO selectivity should be decreased for fuel cell applications.