Photocatalyzed preferential oxidation of CO under simulated sunlight using Au–transition metal oxide–sepiolite catalysts

Elena Rodríguez Aguado*a, Juan Antonio Ceciliaa, Antonia Infantes-Molinaa, Aldo Talonb, Loretta Storarob, Elisa Morettib and Enrique Rodríguez-Castellón*a
aDepartamento de Química Inorgánica, Cristalografía y Mineralogía, Facultad de Ciencias, Universidad de Málaga, 29071 Málaga, Spain. E-mail:;
bDipartimento di Scienze Molecolari e Nanosistemi, Università Ca’ Foscari Venezia, INSTM Venice Research Unit, Via Torino 155/B, 30172 Mestre Venezia, Italy

Received 2nd November 2019 , Accepted 27th December 2019

First published on 30th December 2019

In the present study a series of Au–transition metal oxides supported on a clay mineral such as sepiolite were tested in the preferential oxidation of CO in an excess of H2 under simulated solar light irradiation and in the absence of light, at 30 °C and atmospheric pressure. Transition metal oxides (ZnO, Fe2O3, NiO, MnO2, and Co3O4) were dispersed over the sepiolite surface where, subsequently, Au nanoparticles with an average particle size between 2 and 3 nm were successfully deposited–precipitated. The obtained photocatalysts were characterized by XRD, XRF, DRUV-Vis, N2 adsorption–desorption and HRTEM in order to evaluate the optical, structural and chemical properties of the prepared samples. Despite the low amount of gold (nominal 1.0 wt%), the catalysts exhibited an outstanding behavior under light irradiation, with reaction rates between 4.5 and 5.2 mmol COox gcat−1 h−1 for the Au–NiSep, Au–CoSep and Au–ZnSep samples. These photocatalysts exhibited a high dispersion of the respective transition metal oxides over the sepiolite support and the presence of low-coordinated hemispherical gold nanoparticles. The superior photocatalytic efficiency of these samples was ascribed to the reduction of the electron–hole pair recombination of photogenerated charge carriers by the excitation of the localized surface plasmon resonance of the Au nanoparticles. The BET surface area and the gold particle size seemed to be relevant factors affecting the catalytic performance.

1. Introduction

Hydrogen is not only an essential and extensively used compound in the chemical industry but is also considered an attractive clean energy vector, especially when combined with efficient fuel cells for both stationary power generation and transportation.1,2 Specifically, polymer electrolyte membrane fuel cells (PEMFCs) have received much attention from the transportation industry to drive the first generation of hydrogen fuelled vehicles.3,4 The hydrogen generated on board by catalytic reforming of methanol,5 ethanol or biofuels must be extremely pure, containing less than 10 ppm CO in order to protect the platinum anode from poisoning by CO adsorption and subsequent deactivation.6 Among the well-known techniques for hydrogen purification, such as pressure swing adsorption (PSA),7 separation by using metal membranes8 or CO methanation,9,10 the preferential oxidation of CO in a H2 stream (CO-PROX) is considered one of the most effective and economical methods for reducing the concentration of CO before entering the PEMFCs.11 Catalysts used in this process must meet a number of requirements in order to avoid competitive side reactions such as H2 oxidation.12 Ideally, they should work at low and medium temperatures (25–90 °C) and be highly active and selective toward CO2 formation. A wide variety of promising catalysts based on supported noble metals (e.g., Au and Pt)13,14 and transition metal oxides (e.g., CuO–CeO2 and Co3O4)15,16 have been investigated for the thermal oxidation of CO. Among all, well dispersed gold nanoparticles on reducible metal oxides exhibit better performance at low temperatures than the rest of the noble metals.17,18 However, the high activity of these catalysts still remains a topic of discussion. It has been related to the presence of low coordinated corner sites,19 nanometric gold (2–4 nm) and electronic effects of the support on the Au nanoparticles.20 However, it is generally admitted that when the Au nanoparticles are dispersed on a reducible metal oxide, the electron density on the surface of the Au nanoparticles is increased due to the electron transfer from the metal oxide. This promotes the adsorption of CO at the metal/support interface on Au–O vacancy sites where the oxidation reaction takes place with the aid of reactive oxygen provided by the support.17

Overall, previous research studies show three very important considerations when designing a gold based CO-PROX catalyst: the surface area of the catalyst to increase catalytic site availability,21 the substrate material for oxygen transport20 and the presence of highly dispersed ultra-fine particles with diameters lower than 5 nm.22,23

More recently, the use of photocatalysts has also been proposed in order to perform the reaction under standard conditions by using UV-Vis irradiation, avoiding the use of thermal sources to input energy into the system.24 Many of the studies have been focused on the use of supported gold nanoparticles on semiconductors such as TiO2.25–27 In the UV region, the gold nanoparticles serve as charge traps for conduction band electrons generated in the semiconductor, thereby reducing the electron–hole pair recombination and lifetimes which results in more catalytically active sites.25 In the visible range, the gold nanoparticles exhibit an absorption band at around 560 nm which is caused by the localized surface plasmon resonance (LSPR) effect.28 This phenomenon is a resonant photon-induced coherent oscillation of charges at the metal–dielectric interface, established when the photon frequency induced by the incident electromagnetic radiation matches the frequency of metal surface electrons.23,29 This effect essentially enhances the localized electric field in the proximity of gold nanoparticles, which in turn interact with the near surface of the semiconductor allowing the formation of electron–hole pairs.

Several studies have been devoted to the so-called photo-CO-PROX reaction, most of them being focused on the use of gold nanoparticles supported on semiconductor TiO2, due to its well-known photocatalytic properties.30 Dai et al.31 studied the preferential oxidation of CO in a H2-rich stream over Au/TiO2 systems under UV irradiation. The results showed that UV radiation not only promoted the oxidation of CO, but also increased the selectivity of CO oxidation from 35.6% to 37.5%. They also studied the effect of visible radiation in the mentioned reaction at room temperature for the same system prepared again by the deposition–precipitation method. It was found that the visible light irradiation promoted the adsorption of CO and its activation at the surface of Au species, increasing CO conversion from 29.5 to 38.5%.32 As it is known that metal oxide supports play an important role in the dispersion of Au nanoparticles and in the adsorption behavior of reactants, Au/TiO2 and Au/Al2O3 systems were compared.33 The results reflected that the visible light irradiation promoted the oxidation of CO and its selectivity to CO2 in the presence of H2 over Au/TiO2, whereas it also promoted the oxidation of CO but suppressed the selectivity of CO oxidation over Au/Al2O3. This fact was ascribed to the different electron transfer processes in both systems. For Au/TiO2, the excited electron in the Au 6sp band is transferred to the conduction band of TiO2 with the formation of a transient electron-rich state at the Au surface. While for Au/Al2O3, the excited electrons are transferred to the HOMO of Al2O3 resulting in the formation of a transient electron-deficient state at the Au surface, exerting therefore different effects on CO adsorption and oxidation.

Yang et al.34 also studied a polyaniline (PANI) assembled Au/TiO2 catalyst aimed at building up a good interface to facilitate electron transfer as an effective way of promoting CO oxidation over Au/TiO2. The results showed that coupling the LSPR effect of Au nanoparticles and the electron donor effect of PANI led to a better CO-PROX activity under visible light than Au/TiO2. More recently, Zhang et al.35 have studied the system Au/ZIF-8-TiO2, a gold nanoparticle catalyst supported on metal–organic framework modified TiO2. The results reflected its better performance in CO oxidation than Au/TiO2 at room temperature in the dark and under visible light irradiation, which was ascribed to the higher surface electron density of both Au and TiO2, attributed to the electron transfer between ZIF-8 and TiO2.

Within this context, the present work proposes the dispersion of a series of reducible metal oxides (ZnO, Fe2O3, NiO, MnO2, and Co3O4) on a low cost fibrous clay mineral (sepiolite), with the subsequent incorporation of gold by the deposition–precipitation method with NaOH. This commonly used method allows the formation of well dispersed nanometric gold particles, which is key to ensuring high CO oxidation efficiency.36,37 Sepiolite is a clay mineral with Si12O30Mg8(OH)4(H2O)4·8H2O as the unit cell formula and is considered a promising supporting material because of its strong mechanical stability, high abundance, cost-effectiveness and adsorption capacity. It is a 2[thin space (1/6-em)]:[thin space (1/6-em)]1 type of layered clay mineral having fibrous morphology with molecular sized channels and a high specific surface area. Therefore, the great dispersion and stability of metal oxides and gold nanoparticles are expected which may provide a greater proportion of catalytically active sites.38

In summary, this work aims to evaluate the oxidation capacity of Au-decorated transition metal oxides dispersed on a fibrous clay mineral, such as sepiolite, in the preferential CO oxidation in an excess of H2 at room temperature and atmospheric pressure, in the dark and under simulated solar light irradiation.

2. Experimental

2.1. Synthesis of transition metal oxides supported on sepiolites

The starting commercial sepiolite named Pangel S9 was supplied by Tolsa S.A. (Spain). The sepiolite was used as a support to disperse several transition metal oxides (ZnO, Fe2O3, NiO, MnO2, and Co3O4). The supported oxides were prepared by the following procedure: first, an aqueous suspension of sepiolite was prepared by adding 3.0 g of the clay to 150 mL of distilled water under stirring. Then, an adequate amount of the corresponding metal (Fe3+, Zn2+, Co2+, Ni 2+, and Mn2+) nitrate was added to the sepiolite suspension. The amount of precursor salts was adjusted to obtain a metal loading of 10 wt%. After that, the suspension was heated to 80 °C, while a solution of diluted NaOH (0.1 M) was slowly added until pH = 10 (ref. 39) was reached to precipitate the cations as their respective hydroxides. Once pH value was stable, the solution was maintained at that temperature for 24 h under stirring. The obtained solid was then filtered, washed with deionized water to remove the Na+ cations and dried at 80 °C for 24 h. Finally, the solids were calcined at 400 °C for 2 h to obtain the metal oxides supported on sepiolites.

2.2. Synthesis of gold catalysts Au–MSepiolite (M = Mn, Fe, Ni, Co, Zn)

The gold containing sepiolites were prepared by the deposition–precipitation method. Prior to the gold deposition, the surface acidity of the support was adjusted to pH = 6 in order to enhance the metal deposition and to obtain small and stable particles. The modified sepiolite was dispersed in a 1 M NaNO3 solution, the pH adjusted to 6 with 0.1 M HNO3 and kept overnight under stirring. After that, the solid was filtered and dried at 70 °C. An adequate amount of HAuCl4·3H2O to obtain 1.0 wt% gold in the solid was dissolved in 100 mL of Milli-Q water. The pH of the solution was adjusted to 6 by slowly adding an aqueous solution of NaOH (0.1 M) using a syringe pump (1 mL h−1). The solution was then heated to 80 °C and the modified clay added, keeping the mixture under continuous stirring for 1 h.40 After cooling to room temperature, the solid obtained was separated by filtration, washed several times with deionized water to remove chlorides (AgNO3 test) and sodium ions, and finally dried at 100 °C. Thus, the synthesized catalysts will be designated hereafter as Au-Sep, Au-ZnSep, Au-CoSep, Au-NiSep, Au-FeSep and Au-MnSep, where the chemical symbol of the transition metal element stands for the corresponding metal oxide and Sep is the abbreviation of sepiolite.

2.3. Catalyst characterization

Elemental bulk composition was determined by X-ray fluorescence with a wavelength dispersive X-ray fluorescence spectrometer (ARL ADVANTXP) and the UNIQUANT software. The X-ray tube was set at 60 kV.

X-ray powder diffraction (XRD) patterns of fresh catalysts were collected on a PAN analytical X'Pert Pro automated diffractometer. Powder patterns were recorded between 10° and 80° 2θ, with a step size of 0.0167° (2θ) and an equivalent counting time of ∼60 s per step, in the Bragg–Brentano reflection configuration by using a Ge (111) primary monochromator (Cu Kα1) and an X'Celerator detector.

The size and morphology of the nanoparticles were determined by high resolution transmission electron microscopy (HRTEM) using a TALOS F200× instrument. TEM analysis was performed at 200 kV and 5.5 μA and scanning transmission electron microscopy (STEM) using a HAADF detector, at 200 kV and 200 nA. Image J software was used to calculate the average particle size distribution.

Diffuse reflectance UV-Vis (DRUV-Vis) spectra were collected with a Perkin Lambda 35 UV-Vis spectrophotometer. The absorption coefficient (α) was calculated as follows: α = ln(1/T)/d, where T is the measured transmittance and d is the optical path length. Band gap energy, Eg, was determined thoroughly using the α value (m−1) from the plot of (αhν)1/2 versus photon energy (), where h is Planck's constant and ν is the frequency (s−1). The intercept of the tangent to the absorption curves was used to estimate the band gap value.

N2 adsorption–desorption isotherms at −196 °C were measured to determine the textural properties of the materials by using ASAP 2010 apparatus of Micromeritics. Prior the measurements, the samples were outgassed overnight at 110 °C and 10−4 mbar. The N2 isotherms were used to determine the specific surface areas through the BET equation and the total pore volume was calculated from adsorbed N2 at a relative pressure of P/P0 = 0.95. The microporosity of the samples was determined using Boer's t-plot method. The pore size distribution was estimated from the desorption branch of the isotherm using Non-local Density Functional Theory (NLDFT).

2.4. Photocatalytic test

Photocatalytic CO-PROX tests were carried out using laboratory flow apparatus with a fixed bed reactor operating at atmospheric P. The catalyst (0.15 g) was placed in a quartz cell with a cooling water system. Before the catalytic experiments, the sample was pre-treated in situ at 200 °C under flowing filtered atmospheric air for 4 h. The gas hourly space velocity, GHSV, was 22[thin space (1/6-em)]000 h−1. The feed consisted of 1.25% CO, 1.25% O2 and 50% H2 (vol%) balanced with He. The temperature of the quartz cell was controlled at about 30 °C (measured using a thermocouple placed inside the catalyst bed). During the testing process, visible light (Sunlight Solar Simulator, AM1.5G filter, 100 W Xenon arc lamp, Abet Technologies) was introduced into the surface of the quartz cell. For testing the thermocatalytic activity of the catalyst under dark conditions, the quartz cell was wrapped with Al foil to shut down light irradiation.

The carbon monoxide and oxygen conversions were calculated based on the CO (eqn (1)) and O2 (eqn (2)) consumption, respectively:

image file: c9dt04243a-t1.tif(1)
image file: c9dt04243a-t2.tif(2)
The selectivity towards CO2 was estimated from the oxygen mass balance as follows (eqn (3)):
image file: c9dt04243a-t3.tif(3)
The excess of oxygen factor (λ) (eqn (4) used was 2 because this value was previously found to be optimal for CO-PROX.41
image file: c9dt04243a-t4.tif(4)

3. Results and discussion

3.1. Catalyst characterization

The porous nature and textural parameters of the starting and modified sepiolites were evaluated by N2 adsorption–desorption at −196 °C. Fig. 1 includes the corresponding isotherms which, according to the IUPAC classification, can be regarded as type II, as observed by the increase in adsorbed N2 at high relative pressures, being characteristic of macroporous solids.42 However, a pronounced increase of adsorbed N2 is also observed at low relative pressures, and so they present microporosity as well. The type H3 hysteresis loop is quite narrow in all cases which is typical of non-rigid aggregate particles, such as clay minerals, or macroporous structures that are not completely filled with pore condensates.42 The specific surface area of the starting sepiolite estimated from the BET equation, Table 1, is 182 m2 g−1. After metal oxide and gold incorporation SBET values slightly decrease in all cases with the exception of Au–NiSep and Au–CoSep. The thermal treatment required for the decomposition of the metal precursor with the formation of an oxide on the sepiolite support may cause structural changes associated with dehydroxylation and a partial collapse of the layered structure, which could explain this loss of surface area.38 In addition, since macroporosity may be assigned to voids between fibers, this distinct decrease can also be ascribed to the partial filling of these interfiber voids by the metal oxides, whereas the observed decrease in the volume of micropores could be attributed to a slight partial collapse of the sepiolite structure.43 In the case of Ni and Co based sepiolites the volume of micropores increases, which is associated with the formation of voids originated in the space between metal oxide particles.
image file: c9dt04243a-f1.tif
Fig. 1 N2 adsorption–desorption isotherms of the studied materials.
Table 1 Textural properties of the studied materials Au–Sep and Au–MSep
Sample SBET (m2 g−1) Sext (m2 g−1) t-plot (m2 g−1) Vp (cm3 g−1) Vmicropores (cm3 g−1)
Sepiolite 182 134 48 0.607 0.021
AuSepiolite 164 135 29 0.338 0.012
AuNiSep 201 132 69 0.349 0.031
AuZnSep 173 135 39 0.333 0.016
AuCoSep 214 126 88 0.337 0.040
AuMnSep 129 105 24 0.253 0.010
AuFeSep 182 159 22 0.379 0.009

Fig. 2 shows the pore size distribution of all studied materials determined by the NLDFT method,44 where, as observed from the isotherm profiles, it presents a certain contribution of microporosity in all cases, with pore sizes being lower than 2 nm and the pore size distribution being sharper in the case of Ni and Co based sepiolites, as mentioned above. The presence of microporosity may be attributed to the channels formed by the inversion of the tetrahedral layer in the clay.45 Furthermore, an important contribution of pores between 2 and 100 nm confirms the presence of meso- and macroporosity in all cases. However, the total pore volume included in Table 1 reflects a distinct decrease from 0.60 cm3 g−1, for the starting sepiolite, to values lower than 0.35 cm3 g−1 for the rest of the samples. This fact suggests that part of the space among the sepilolite fibers is occupied by the metal oxides, as also observed in TEM micrographs.

image file: c9dt04243a-f2.tif
Fig. 2 Pore size distribution of bare sepiolite and supported sepiolite catalysts.

Fig. 3 compiles the X-ray patterns of the catalysts supported on sepiolite. The diffractogram of Au–sepiolite shows sepiolite as the only crystalline phase in agreement with the values in the standard card (JCPDS, PDF card no. 98-015-6199). In the rest of the diffractograms, the main diffraction peaks of sepiolite are maintained but suffer from a significant loss of intensity due to the incorporation of transition metal oxides in the catalytic system. The deposition of these transition metal oxides by precipitation hardly causes modifications in the sepiolite structure since the 110 reflection at 7.48°, which corresponds to the interlayer distance in the clay structure,38 is maintained in all cases. In situ variable-temperature synchrotron studies46 demonstrated the folding of the Sep structure above 320 °C, whereas TGA-DSC studies confirmed a mass loss of ∼15% in the form of water at 400 °C.38 This fact could suppose a slight shift of the 110 reflection as a consequence of the folding of sepiolite after thermal treatment. The widening of the diffraction peaks between 10° and 20° 2θ can also confirm the sepiolite folding. Other studies showed that heating Sep to 500 °C did not affect the clay structure.47 Furthermore, gold is not detected in any of the cases due to its low content, high dispersion and the presence of small particles of Au crystals. The presence of the corresponding transition metal oxides is only detected in the case of Au–ZnSep where the main reflections corresponding to the standard card no. 98-007-6641 are assigned to ZnO. The rest of the diffractograms do not show the presence of these supported oxides, pointing out that they could be highly dispersed over the clay support or that transition metal oxide amorphous phases are formed.

image file: c9dt04243a-f3.tif
Fig. 3 XRD patterns of gold supported sepiolites.

Gold content was determined by X-ray fluorescence. In all cases, Au loading is close to the theoretical value (1.0 wt%) with the exception of the bare sepiolite, with a content of 0.06 wt%. This almost negligible content can be ascribed to the weak interactions between the clay mineral and the gold precursor during the synthesis, since the isoelectronic point (iep) of the sepiolite suspension is below the pH at which the deposition–precipitation process takes place. Sepiolite point of zero charge (pzc) has been reported to be between 2 and 8,48–51 this variation being ascribed to the surface heterogeneity of this mineral and the presence of impurities along with the different pretreatments which it has undergone. The isomorphic substitution of Al3+ instead of Si4+ leads to negatively charged adsorption sites which repel the anionic gold species involved in the deposition–precipitation process.48 So, at pH = 6 the surface of sepiolite is negatively charged, hindering the interaction with the anionic gold species. If the synthesis were performed at lower pH, the surface would be positively charged but the adsorbed anionic gold species would present a greater proportion of poisonous and difficult to remove Cl ions.37 However, the incorporation of a relatively high amount of transition metal oxides into sepiolite seems to generate positively charged sites facilitating the interaction with the anionic gold species.

The photo-responsive behaviour of ZnSep and Au–ZnSep evaluated by diffuse reflectance spectroscopy (DRUV-Vis) is shown in Fig. 4. Both samples show an absorption peak in the UV region (below 400 nm) which corresponds to the intrinsic absorption of semiconductor ZnO. The band gap values go from 3.17 eV for the support, ZnSep, to 3.09 eV for the sample Au–ZnSep. This slight decrease after gold incorporation indicates the presence of a strong interaction between the gold NPs and ZnO, expanding its photo-response to longer wavelengths. After gold incorporation, the spectrum of the sample Au–ZnSep also exhibits a broad band in the visible region with the maximum positioned between 520 and 570 nm and attributable to the LSPR effect of the Au nanoparticles loaded on ZnO.28

image file: c9dt04243a-f4.tif
Fig. 4 DRUV-Vis spectra of the bare ZnSep support and the Au–ZnSep catalyst.

The exact position and shape of the surface plasmon band depends on several factors such as the dielectric constant of the medium, the surrounding environment, and the particle size and the shape of the Au NPs which determine the light absorption range.52 As the shape and/or size of a NP changes, the density of the electromagnetic field at the NP surface also changes. These combine to induce a shift in the oscillation frequency of the conduction electrons and a strong field enhancement of the electromagnetic fields near the rough surfaces of Au NPs.53,54 There are a number of reports on the association of the LSPR effect and the NP size; overall, as the particles grow bigger, the absorption band broadens and a red shift to a longer wavelength occurs.55 Therefore, the position and intensity of this peak suggest the presence of very small nanoparticles as also observed by TEM, which are suitable for the studied reaction.56

The rest of the samples were studied by this technique as well, but any absorption peak due to plasmonic gold nanoparticles was distinguishable, which is ascribed to the high dilution of the coloured samples. The band gap values of Au–NiSep and Au–CoSep were not possible to determine for the same reason.

HRTEM micrographs of Au–Sep and Au–MSep (M = Zn, Fe and Ni) catalysts are included in Fig. 5. These micrographs show the fibrous structure of sepiolite decorated with the corresponding transition metal oxides, which are markedly dispersed over the surface. The Au–Sep sample only shows the fibrous structure of the clay, with any supported gold particle being difficult to identify over its surface since the gold content is almost negligible (XRF results) as a consequence of the abovementioned synthesis drawbacks. The micrographs of Au–NiSep, Au–ZnSep, Au–CoSep and Au–MnSep (not shown) show an uneven distribution of the transition metal oxides over the sepiolite fibers with areas showing a great proportion of particle agglomeration. It should be noted, however, that in the Au–FeSep micrograph iron(III) oxide is uniformly distributed with less agglomerated areas. In all cases the gold nanoparticles are mainly deposited on the oxide surface and to a lesser extent on the fibrous structure of the clay, as seen in Fig. 6. These micrographs with higher magnification have been recorded in STEM mode in order to achieve a better contrast among the transition metal oxides and gold nanoparticles. The Energy Dispersive X-Ray (EDX) image of the Au–ZnSep sample is also included in Fig. 6 as the representative image of the Au–MSep catalysts. Gold particle size distribution was measured from the selected regions of STEM micrographs, where the contrast was better, and the corresponding histograms are included in Fig. 6. On the whole, the histograms show a relatively narrow particle size distribution, mainly indicating the coexistence of highly dispersed gold particles from 1 to 6 nm. The average particle size is centred at around 1.9, 2.2, 2.9 and 3.3 nm for Au–NiSep, Au–CoSep, Au–ZnSep and Au–FeSep, respectively, which has been widely reported to be suitable for CO preferential oxidation.23,56 These small particles involve low-coordinated Au atoms, which are known to be the most reactive surface structures since the lower the metal coordination is, the higher the d states are in energy, and the stronger they interact with adsorbates.19 This is because the high lying metal d-states in low coordinated Au atoms are better positioned for interactions with the valence state of the adsorbate than the low-lying states of highly coordinated atoms.57,58 However, in the case of Au–MnSep, the gold nanoparticles are agglomerated exhibiting irregular shapes and sizes. The shape of the rest of the gold nanoparticles is spherical, with some elongated rods found in the case of Au–ZnSep. The change in the shape of the gold nanoparticles leads to significant changes in their optical characteristics.54 The accumulation of charges at the NP surface during electron oscillations is different along the rod axis than along the perpendicular direction, being maximum for the transverse plasmons and minimum for the longitudinal plasmons. The restoring forces are proportional to the charge accumulation leading to lower resonance frequencies for axial oscillations of electrons. This provokes a red shift in the resonance band with deviations from spherical shapes,59 which results in impoverished photocatalytic efficiency.

image file: c9dt04243a-f5.tif
Fig. 5 HRTEM micrographs of Au–Sep and Au–MSep (M = Zn, Fe and Ni).

image file: c9dt04243a-f6.tif
Fig. 6 STEM, EDX images and particle size distribution of Au–MSep (M = Ni, Fe and Zn) sepiolites.

3.2. Catalytic test

Au transition metal oxides supported on sepiolite have been tested in the preferential CO oxidation in a H2-rich stream at 30 °C and atmospheric pressure. Fig. 7 and Table 2 show CO conversion and CO2 selectivity for the studied materials under simulated solar light irradiation and in the absence of light. In general, the results clearly show a higher CO conversion in light mode in all cases, mainly when it comes to Au–NiSep, 90% CO conversion is achieved. In addition, Au–Co, Au–Zn and Au–Fe based sepiolites also display a good performance, achieving 88, 78 and 77% CO conversion, respectively, with CO oxidation rates of 5.2, 5.1 and 4.5 mmol COox gcat−1 h−1, respectively. These rates are greater than those obtained in several studies based on noble metals supported on reducible oxides (1.0–4.9 mmol COox gcat−1 h−1) even though a comparison is not straightforward since, in spite of working in an excess of H2, operating conditions are different.60 However, these values are very close to those obtained in previous studies (2.9–6.0 mmol COox gcat−1 h−1) under the same reaction conditions by using the system Au/TiO2, also prepared by the deposition–precipitation method.26,27 The most striking result is that obtained for Au–MnSep, achieving only 20% CO conversion. It should be noted that Au–sepiolite has also been tested and is found to be inactive in both reaction modes (Table 2), suggesting that metal transition oxides play an important role in the gold deposition process and its subsequent dispersion on the catalyst. Indeed, CO oxidation negligibly proceeded on impregnated sepiolite samples with transition metal oxides before gold deposition, indicating the critical role of Au sites.
image file: c9dt04243a-f7.tif
Fig. 7 (A) CO conversion and (B) CO2 selectivity in the preferential oxidation of CO in an excess of hydrogen under dark and simulated solar light irradiation conditions for Au–MSep (M = Mn, Fe, Co, Ni and Zn) catalysts. Operating conditions: 30 °C; GHSV = 22[thin space (1/6-em)]000 h−1; feed gas: 1.2% CO, 1.2% O2 and 50% H2 (vol%, He balance).
Table 2 Catalytic performance of Au–MSep catalysts
Sample CO Conversion (%) CO2 Selectivity (%) Reaction rate (mmol COox gcat−1 h−1)
Dark Light Dark Light Dark Light
Au–MnSep 2 20 100 65 0.1 1.2
Au–FeSep 65 77 97 45 3.8 4.5
Au–CoSep 42 88 100 65 2.4 5.1
Au–NiSep 60 90 100 60 3.5 5.2
Au–ZnSep 46.2 78.1 100 80 2.7 4.5
Au–Sep 0.2 0.2 100 100 0.01 0.01

CO2 selectivity is the result of the simultaneous oxidation of CO and H2, both occurring in the presence of molecular oxygen catalysed by metal oxides. The desired reaction for hydrogen purification is clearly the CO oxidation, while H2 oxidation is highly undesirable due to the consumption of hydrogen (needed to feed a PEMFC). Nevertheless, as 100% selectivity is not attainable, H2 oxidation usually takes place and H2O forms, reducing the activity and selectivity of the catalyst. CO2 selectivity values reflect that all the catalysts display a higher activity for CO oxidation than for H2 oxidation in both modes, typical of Au based systems.61,62 However the observed CO2 selectivity values are lower under simulated light irradiation. This is probably due to the simultaneous increase of hydrogen oxidation due to light irradiation, even though the factors governing the photocatalytic activity in the CO-PROX reaction are not yet fully understood. Under simulated solar light irradiation selectivity values range from 45% for Au–FeSep to 80% for Au–ZnSep, and so the latter is regarded as the most active photocatalyst, if taking into account that all catalysts have the same theoretical metal loading (both gold and transition metals, wt%). However, CO2 selectivity is rather higher under dark conditions (traditional CO-PROX mode) with values very close to 100% in all cases.

If the reaction rates are considered (see Table 2 and Fig. 8), in all cases the catalytic performance is improved under simulated solar light irradiation, Au–CoSep and Au–NiSep being the most active catalysts. They exhibit the highest BET surface area (214 and 201 m2 g−1, respectively), while the catalyst with lowest observed catalytic activity, excluding the sample Au–Sep, is Au–MnSep with the lowest specific surface area (129 m2 g−1). The role of the BET surface area in the catalytic performance under simulated solar light irradiation is analyzed in Fig. 9, evidencing a direct relationship between the reaction rate and the BET surface area of the final photocatalysts.

image file: c9dt04243a-f8.tif
Fig. 8 CO reaction rates of the studied Au–MSep catalysts.

image file: c9dt04243a-f9.tif
Fig. 9 Evolution of the reaction rate in the Photo-PROX of CO as a function of the BET surface area for the Au–MSep (M = Mn, Fe, Co, Ni and Zn) catalysts.

In summary, the surface area seems to be a key factor for these kinds of samples, affecting the catalytic efficiency. In view of the results obtained, it is suggested that in the case of Au–NiSep, Au–CoSep and Au–ZnSep, Au clusters are strongly bonded to the oxide surface at the interfacial structure. The DRUV-Vis results of the sample Au–ZnSe confirm the decrease of the band gap after gold incorporation from 3.17 to 3.09 eV. In TEM images the presence of nanometric gold atoms is clearly observed in low coordinated positions at the surfaces of transition metal oxides which, as pointed out earlier, are more reactive than coordinated metal atoms because of their higher lying d-states.19,63

In agreement with numerous studies, the normal CO-PROX reaction over reducible oxides is likely to occur at the interface between the Au particles and the metal oxide. At the interface, the lattice oxygen is thought to be involved in the CO oxidation by migration from the bulk to the Au-oxide perimeter where oxygen reacts with CO weakly adsorbed on low-coordinated gold atoms. Then, the oxygen transferred to the Au perimeter reacts with adsorbed CO to form CO2, generating oxygen defect sites on the support surface, which can be restored by the gas-phase oxygen. The Au/support interfacial perimeter sites, where a cationic site of the support and one Au atom (or more) are localized at the Au perimeter, act as dual adsorption centers to bind molecular oxygen in a peroxo-type configuration.14,64 Thus, small Au particles with a high number of interfacial perimeter sites per Au atom will exhibit the highest catalytic activity in the CO oxidation as in our case.

Therefore, in the dark mode (normal CO-PROX mode under standard conditions), the differences in the activity for CO oxidation not only can be attributed to the particle size and surface area but also to the ability of the different supports to supply reactive oxygen species. The reaction rates included in Table 2 show a substantial catalyst support effect. The sample Au–FeSep is the most active catalyst in dark mode followed by Au–NiSep, Au–ZnSep and Au–CoSep, and so Fe2O3 presumably has the greatest ability to store and supply reactive oxygen.

However, in light mode the tendency is quite different, as discussed earlier, Au–NiSep and Au–CoSep being the most active catalysts closely followed by Au–ZnSep and Au–FeSep. In that sense, the photochemical response in the preferential oxidation of CO under UV-Vis irradiation has been studied for Au/TiO2 systems.52,65 The gold nanoparticles serve as charge traps by the excitation of the localized surface plasmon resonance (LSPR), thereby reducing the electron–hole pair recombination in the semiconductor and lifetimes, which results in more catalytically active sites. However, the mechanism by which energy is transferred still remains a topic of discussion.66,67 Two main mechanisms have been cited in the literature: direct charge transfer and local electric field enhancement. The former proposes a charge transfer mechanism in which an excited plasmon in Au generates conduction band electrons with sufficient energy to traverse the Schottky barrier and enter the conduction band of the adjacent semiconductor.68 The latter alternative mechanism considers that elevated electric fields (enhanced by plasmonic absorption and concentration) near the interface excite the electron–hole pairs in the semiconductor, leading to enhanced photocatalytic activity.69

These proposed mechanisms for light irradiation could explain the improved behavior of the Au–CoSep catalyst, with reaction rates going from 2.4 in dark mode to 5.1 mmol COox gcat−1 h−1 under simulated sunlight irradiation. The promoting effect of light irradiation on the CO oxidation efficiency of the samples Au–NiSep and Au–ZnSep is also very pronounced, with reaction rates from 3.5 to 5.2 mmol COox gcat−1 h−1 and from 2.7 to 4.5 mmol COox gcat−1 h−1, respectively. These results point out that the reaction mechanism in light mode for these samples is mainly governed by the reduction of the electron–hole pair recombination of photogenerated charge carriers in the semiconductor by the excitation of the localized surface plasmon resonance of the Au nanoparticles.

4. Conclusions

A series of catalysts comprised of gold supported on transition metal oxides supported in turn on a low-cost natural clay material such as sepiolite were successfully synthesized. The high specific surface area of the sepiolite support promoted a high dispersion of the transition metal oxides and gold nanoparticles, which is related to the number of available catalytic sites and the catalytic activity. The highly dispersed gold nanoparticles presented suitable average particle sizes between 1.9 and 3.3 nm. Catalytic results reflected the promoting effect of light irradiation on the preferential oxidation of CO at atmospheric pressure and 30 °C, showing the outstanding behavior of the Au–MSep (M = Ni, Co and Zn) catalysts in spite of the low amount of noble metal (1 wt%). The enhanced photocatalytic activity of these samples under simulated solar light irradiation was ascribed to the suitable growth of gold nanoparticles over the dispersed transition metal oxides, which act as concentrators of the energy of incoming photons due to the localized surface plasmon resonance effect.

Conflicts of interest

There are no conflicts to declare.


Thanks to the project RTI2018-099668-BC22 of Ministerio de Ciencia, Innovación y Universidades of Spain, the project UMA18-FEDERJA-126 of Junta de Andalucía and FEDER funds. A. I. M. thanks the Ramon y Cajal Program RyC2015-17870 (Spanish Ministry of Economy and Competitiveness, Spain).


  1. S. Chen, A. Kumar, W. C. Wong, M.-S. Chiu and X. Wang, Appl. Energy, 2019, 233–234, 321–337 CrossRef.
  2. B. Chutichai, S. Authayanun, S. Assabumrungrat and A. Arpornwichanop, Energy, 2013, 55, 98–106 CrossRef CAS.
  3. Ó. González-Espasandín, T. J. Leo, M. A. Raso and E. Navarro, Renewable Energy, 2019, 130, 762–773 CrossRef.
  4. F.-C. Wang and W.-H. Fang, Int. J. Hydrogen Energy, 2017, 42, 10376–10389 CrossRef CAS.
  5. O. Özcan and A. N. Akın, Int. J. Hydrogen Energy, 2019, 44, 14117–14126 CrossRef.
  6. B. Qiao, J. Liu, Y.-G. Wang, Q. Lin, X. Liu, A. Wang, J. Li, T. Zhang and J. Liu, ACS Catal., 2015, 5, 6249–6254 CrossRef CAS.
  7. E. H. Majlan, W. R. Wan Daud, S. E. Iyuke, A. B. Mohamad, A. A. H. Kadhum, A. W. Mohammad, M. S. Takriff and N. Bahaman, Int. J. Hydrogen Energy, 2009, 34, 2771–2777 CrossRef CAS.
  8. S. Kozhakhmetov, N. Sidorov, V. Piven, I. Sipatov, I. Gabis and B. Arinov, J. Alloys Compd., 2015, 645, S36–S40 CrossRef CAS.
  9. M. Krämer, M. Duisberg, K. Stöwe and W. F. Maier, J. Catal., 2007, 251, 410–422 CrossRef.
  10. M. V. Konishcheva, D. I. Potemkin, P. V. Snytnikov and V. A. Sobyanin, Int. J. Hydrogen Energy, 2019, 44, 9978–9986 CrossRef CAS.
  11. A. P. Mishra and R. Prasad, Bull. Chem. React. Eng. Catal., 2011, 6, 1–14 CAS.
  12. T. V. Choudhary, C. Sivadinarayana, C. C. Chusuei, A. K. Datye, J. P. Fackler and D. W. Goodman, J. Catal., 2002, 207, 247–255 CrossRef CAS.
  13. K. Liu, A. Wang and T. Zhang, ACS Catal., 2012, 2, 1165–1178 CrossRef CAS.
  14. G. C. Bond and D. T. Thompson, Gold Bull., 2000, 33, 41–50 CrossRef CAS.
  15. G. Marbán and A. B. Fuertes, Appl. Catal., B, 2005, 57, 43–53 CrossRef.
  16. L. Zhong, T. Kropp, W. Baaziz, O. Ersen, D. Teschner, R. Schlögl, M. Mavrikakis and S. Zafeiratos, ACS Catal., 2019, 9, 8325–8336 CrossRef CAS.
  17. N. Bion, F. Epron, M. Moreno, F. Mariño and D. Duprez, Top. Catal., 2008, 51, 76–88 CrossRef CAS.
  18. M. Haruta, Faraday Discuss., 2011, 152, 11 RSC.
  19. T. Jiang, D. J. Mowbray, S. Dobrin, H. Falsig, B. Hvolbæk, T. Bligaard and J. K. Nørskov, J. Phys. Chem. C, 2009, 113, 10548–10553 CrossRef CAS.
  20. M. M. Schubert, S. Hackenberg, A. C. van Veen, M. Muhler, V. Plzak and R. J. Behm, J. Catal., 2001, 197, 113–122 CrossRef CAS.
  21. F. Moreau and G. C. Bond, Catal. Today, 2007, 122, 215–221 CrossRef CAS.
  22. G. V. Hartland, Chem. Rev., 2011, 111, 3858–3887 CrossRef CAS PubMed.
  23. S. Link and M. A. El-Sayed, Int. Rev. Phys. Chem., 2000, 19, 409–453 Search PubMed.
  24. R. T. T. Kamegawa, M. Matsuoka and M. Anpo, 11th International Symposium on Hybrid Nano Materials toward Future Industries, Nagaoka, Japan, 3–5 February 2006.
  25. D. A. Panayotov and J. R. Morris, Surf. Sci. Rep., 2016, 71, 77–271 CrossRef CAS.
  26. E. Moretti, E. Rodríguez-Aguado, A. I. Molina, E. Rodríguez-Castellón, A. Talon and L. Storaro, Catal. Today, 2018, 304, 135–142 CrossRef CAS.
  27. E. Rodríguez-Aguado, A. Infantes-Molina, A. Talon, L. Storaro, L. León-Reina, E. Rodríguez-Castellón and E. Moretti, Int. J. Hydrogen Energy, 2019, 44, 923–936 CrossRef.
  28. C. J. Orendorff, T. K. Sau and C. J. Murphy, Small, 2006, 2, 636–639 CrossRef CAS PubMed.
  29. S. Sarina, E. R. Waclawik and H. Zhu, Green Chem., 2013, 15, 1814 RSC.
  30. K. Nakata and A. Fujishima, J. Photochem. Photobiol., C, 2012, 13, 169–189 CrossRef CAS.
  31. W. Dai, X. Zheng, H. Yang, X. Chen, X. Wang, P. Liu and X. Fu, J. Power Sources, 2009, 188, 507–514 CrossRef CAS.
  32. J. Liu, R. Si, H. Zheng, Q. Geng, W. Dai, X. Chen and X. Fu, Catal. Commun., 2012, 26, 136–139 CrossRef CAS.
  33. K. Yang, J. Liu, R. Si, X. Chen, W. Dai and X. Fu, J. Catal., 2014, 317, 229–239 CrossRef CAS.
  34. K. Yang, Y. Li, K. Huang, X. Chen, X. Fu and W. Dai, Int. J. Hydrogen Energy, 2014, 39, 18312–18325 CrossRef CAS.
  35. Y. Zhang, Q. Li, C. Liu, X. Shan, X. Chen, W. Dai and X. Fu, Appl. Catal., B, 2018, 224, 283–294 CrossRef CAS.
  36. X. Bokhimi and R. Zanella, J. Phys. Chem. C, 2007, 111, 2525–2532 CrossRef CAS.
  37. F. Moreau and G. C. Bond, Catal. Today, 2007, 122, 260–265 CrossRef CAS.
  38. A. Al-Ani, R. Gertisser and V. Zholobenko, Appl. Clay Sci., 2018, 162, 297–304 CrossRef CAS.
  39. J. G. Carriazo, L. M. Martínez, J. A. Odriozola, S. Moreno, R. Molina and M. A. Centeno, Appl. Catal., B, 2007, 72, 157–165 CrossRef CAS.
  40. A. Álvarez, S. Moreno, R. Molina, S. Ivanova, M. A. Centeno and J. A. Odriozola, Appl. Clay Sci., 2012, 69, 22–29 CrossRef.
  41. A. Arango-Díaz, E. Moretti, A. Talon, L. Storaro, M. Lenarda, P. Núñez, J. Marrero-Jerez, J. Jiménez-Jiménez, A. Jiménez-López and E. Rodríguez-Castellón, Appl. Catal., A, 2014, 477, 54–63 CrossRef.
  42. M. Thommes, K. Kaneko, A. V. Neimark, J. P. Olivier, F. Rodríguez-Reinoso, J. Rouquerol and K. S. W. Sing, Pure Appl. Chem., 2015, 87, 1051–1069 CAS.
  43. L. M. Martínez, M. I. Domínguez, N. Sanabria, W. Y. Hernández, S. Moreno, R. Molina, J. A. Odriozola and M. A. Centeno, Appl. Catal., A, 2009, 364, 166–173 CrossRef.
  44. J. Landers, G. Y. Gor and A. V. Neimark, Colloids Surf., A, 2013, 437, 3–32 CrossRef CAS.
  45. J. A. Cecilia, E. Vilarrasa-García, C. L. Cavalcante, D. C. S. Azevedo, F. Franco and E. Rodríguez-Castellón, J. Environ. Chem. Eng., 2018, 6, 4573–4587 CrossRef CAS.
  46. J. E. Post, D. L. Bish and P. J. Heaney, Am. Mineral., 2007, 92, 91–97 CrossRef CAS.
  47. N. Degirmenbasi, N. Boz and D. M. Kalyon, Appl. Catal., B, 2014, 150–151, 147–156 CrossRef CAS.
  48. M. Alkan, Ö. Demirbaş and M. Doğan, J. Colloid Interface Sci., 2005, 281, 240–248 CrossRef CAS PubMed.
  49. M. Alkan, G. Tekin and H. Namli, Microporous Mesoporous Mater., 2005, 84, 75–83 CrossRef CAS.
  50. M. Kosmulski, J. Colloid Interface Sci., 2011, 353, 1–15 CrossRef CAS PubMed.
  51. M. A. Dinamarca, C. Ibacache-Quiroga, P. Baeza, S. Galvez, M. Villarroel, P. Olivero and J. Ojeda, Bioresour. Technol., 2010, 101, 2375–2378 CrossRef CAS PubMed.
  52. L. Du, A. Furube, K. Yamamoto, K. Hara, R. Katoh and M. Tachiya, J. Phys. Chem., 2009, 113, 6454–6462 CAS.
  53. K. L. Kelly, E. Coronado, L. L. Zhao and G. C. Schatz, J. Phys. Chem. B, 2003, 107, 668–677 CrossRef CAS.
  54. S. Eustis and M. A. El-Sayed, Chem. Soc. Rev., 2006, 35, 209–217 RSC.
  55. M. M. Alvarez, J. T. Khoury, T. G. Schaaff, M. N. Shafigullin, I. Vezmar and R. L. Whetten, J. Phys. Chem., 1997, 101, 3706–3712 CrossRef CAS.
  56. A. Primo, A. Corma and H. Garcia, Phys. Chem. Chem. Phys., 2011, 13, 886–910 RSC.
  57. B. Hammer and J. K. Norskov, Nature, 1995, 376, 238–240 CrossRef CAS.
  58. T. V. W. Janssens, B. S. Clausen, B. Hvolbæk, H. Falsig, C. H. Christensen, T. Bligaard and J. K. Nørskov, Top. Catal., 2007, 44, 15 CrossRef CAS.
  59. M. A. Garcia, J. Phys. D: Appl. Phys., 2012, 45, 389501 CrossRef.
  60. Y. Yoshida and Y. Izumi, Catal. Surv. Asia, 2016, 20, 141–166 CrossRef CAS.
  61. M. Haruta, N. Yamada, T. Kobayashi and S. Iijima, J. Catal., 1989, 115, 301–309 CrossRef CAS.
  62. R. M. Torres Sanchez, A. Ueda, K. Tanaka and M. Haruta, J. Catal., 1997, 168, 125–127 CrossRef.
  63. C. Lemire, R. Meyer, S. Shaikhutdinov and H. J. Freund, Angew. Chem., Int. Ed., 2004, 43, 118–121 CrossRef PubMed.
  64. C. D. Powell, A. W. Daigh, M. N. Pollock, B. D. Chandler and C. J. Pursell, J. Phys. Chem. C, 2017, 121, 24541–24547 CrossRef CAS.
  65. T. Hasobe, H. Imahori, P. V. Kamat and S. Fukuzumi, J. Am. Chem. Soc., 2003, 125, 14962–14963 CrossRef CAS PubMed.
  66. S. Linic, P. Christopher and D. B. Ingram, Nat. Mater., 2011, 10, 911–921 CrossRef CAS PubMed.
  67. W. Hou and S. B. Cronin, Adv. Funct. Mater., 2013, 23, 1612–1619 CrossRef CAS.
  68. Y. Tian and T. Tatsuma, J. Am. Chem. Soc., 2005, 127, 7632–7637 CrossRef CAS.
  69. D. B. Ingram, P. Christopher, J. L. Bauer and S. Linic, ACS Catal., 2011, 1, 1441–1447 CrossRef CAS.

This journal is © The Royal Society of Chemistry 2020