Gold nanoparticles supported on magnesium oxide for CO oxidation
© Carabineiro et al; licensee Springer. 2011
Received: 31 August 2010
Accepted: 22 June 2011
Published: 22 June 2011
Skip to main content
We’re sorry, something doesn't seem to be working properly. Please try refreshing the page. If that doesn't work, please contact us so we can address the problem.
© Carabineiro et al; licensee Springer. 2011
Received: 31 August 2010
Accepted: 22 June 2011
Published: 22 June 2011
Au was loaded (1 wt%) on a commercial MgO support by three different methods: double impregnation, liquid-phase reductive deposition and ultrasonication. Samples were characterised by adsorption of N2 at -96°C, temperature-programmed reduction, high-resolution transmission electron microscopy, energy-dispersive X-ray spectroscopy and X-ray diffraction. Upon loading with Au, MgO changed into Mg(OH)2 (the hydroxide was most likely formed by reaction with water, in which the gold precursor was dissolved). The size range for gold nanoparticles was 2-12 nm for the DIM method and 3-15 nm for LPRD and US. The average size of gold particles was 5.4 nm for DIM and larger than 6.5 for the other methods. CO oxidation was used as a test reaction to compare the catalytic activity. The best results were obtained with the DIM method, followed by LPRD and US. This can be explained in terms of the nanoparticle size, well known to determine the catalytic activity of gold catalysts.
It is well known from the literature that for gold to be active as a catalyst, a careful preparation is needed to obtain nanoparticles well dispersed on the support [1–4]. Compared with other supports, MgO is considered as "inactive" [5–8] since it is basically an irreducible oxide, such as Al2O3. These materials have low ability to adsorb or store oxygen at low temperatures .
However, Margitfalvi et al.  prepared Au/MgO catalysts with high activity for low temperature CO oxidation. The activity of these catalysts was further increased by modification with ascorbic acid in a relatively narrow concentration range. These authors suggested that the addition of ascorbic acid slightly changes the ionic/metallic gold ratio and suppresses formation of carbonate, which is responsible for deactivation . Gates and co-workers [10, 11] also managed to produce a Au/MgO catalyst that was active for CO oxidation at 30°C by bringing Au(CH3)2(acac) (acac is acetylacetonate) in contact with partially dehydroxylated MgO and by treatment in flowing helium at 473 K, during which the original mononuclear Au(III) species decomposed, gold being reduced and aggregated. The catalyst underwent rapid deactivation due to the formation of carbonate-like species on the support and on gold, but could be reactivated by treatment in flowing helium, which led to the removal of the carbonate-like species .
Heinz et al.  showed that small clusters of gold (Au20 and Au8) are active towards CO oxidation. In fact, for Au8 clusters, it was found that the oxidation of CO at -33°C is activated after deposition on defect sites of the MgO support [13, 14]. Guzman and Gates [15–17] showed, by X-ray absorption spectroscopy, the presence of both cationic and reduced gold in MgO-supported gold clusters during CO oxidation. Molina and Hammer  showed by DFT calculations that O2 can bind simultaneously to both metal centres (Au and Mg) with CO bonded to another nearby Au centre. Broqvist et al.  proved also by DFT calculations that Cl was a poison for Au/MgO catalysts in CO oxidation, while Na was a promotor. Goodman and co-workers  showed a direct correlation between the concentration of F-centre surface defects in the MgO support and the catalytic activity for CO oxidation of the subsequently deposited Au, implying a critical role of surface F-centres in the activation of Au in Au/MgO catalysts.
Grisel and Nieuwenhuys  found that Au/MgO catalysts supported on alumina were extremely active, achieving 50% CO conversion at room temperature and full conversion at approximately 250°C. It is, however, worth to note that those materials had 5% Au loading, while 1% Au was used in this study. Moreover, these authors used 2% CO in the gas feed for the CO oxidation experiments, while we used 5% CO. Szabó et al. [22–24] also reported that Au/Al2O3 catalysts modified by MgO exhibited high activity in the sub-ambient and ambient temperature ranges for CO oxidation.
Co-precipitation (CP) [1–5, 25–31] and deposition-precipitation (DP) [1–4, 6, 21, 22, 29, 31] are the most common methods to prepare oxide-supported gold catalysts. In this study, less usual Au loading methods were used, such as double impregnation (DIM)  and liquid phase reductive deposition (LPRD) , to prepare Au nanoparticles. To the best of our knowledge, the only reports on the use of DIM is the work of Bowker et al.  dealing with TiO2 samples and our previous work on CeO2[34, 35] and ZnO  catalysts. This method represents an environmentally and economically more favourable route to the preparation of high activity gold catalysts, in comparison to the traditional deposition-precipitation (DP) method . As far as we know, LPRD has only been used by Sunagawa et al.  to prepare Pt and Au catalysts on Fe2O3, FeOOH, ZrO2 and TiO2 supports, and also by us for CeO2 and TiO2. US was only used by our group to prepare very active Au/ZnO catalysts .
The aim of this study is to compare the activity for CO oxidation of Au/MgO catalysts prepared by these unusual methods. This is a simple model reaction to evaluate gold catalysts that has many potential applications, namely in CO removal from H2 streams for fuel cells and gas sensing [1–4, 34, 36, 37].
Commercial MgO (p.a., Merck) was used as received and after a treatment at 400°C, in N2, for 2 h.
Au was loaded on the MgO support by the double impregnation method (DIM) , liquid phase reductive deposition (LPRD)  and ultrasonication (US) . Briefly, the first method (DIM) consists in impregnating the support with an aqueous solution of the gold precursor (HAuCl4) and then with a solution of Na2CO3 that precipitates gold hydroxide within the pores of the catalyst [32, 34–36]. The second procedure (LPRD) consists of mixing a solution of HAuCl4 with a solution of NaOH (with a ratio of 1:4 in weight) that hydroxylates the Au3+ ions, before the support is added to the solution [33, 37, 38]. Au3+ ions are reduced to metallic Au0 by electron transfer from coordinated OH- ions on the surfaces of support particles through their catalytic action . US consists in dissolving the Au precursor in water and methanol, and sonicating for 8 h, reducing gold . In all these methods, a washing procedure is carried out to eliminate residual chloride, which is well known to cause sinterization of Au nanoparticles, turning them inactive [1–4, 37]. Further details can be found elsewhere [34–38].
The materials were analysed by adsorption of N2 at -196°C in a Quantachrom NOVA 4200e apparatus.
Temperature-programmed reduction (TPR) experiments were performed in a fully automated AMI-200 Catalyst Characterization Instrument (Altamira Instruments, Pittsburgh, PA, USA), equipped with a quadrupole mass spectrometer (Dymaxion 200 amu, Ametek). Further details can be found elsewhere [34–38].
High-resolution transmission electron microscopy (HRTEM) measurements were performed with a JEOL 2010 microscope with a point-to-point resolution better than 0.19 nm. The sample was mounted on a carbon polymer-supported copper micro-grid. A few droplets of a suspension of the ground catalyst in isopropyl alcohol were placed on the grid, followed by drying at ambient conditions. The average gold particles and the particle size distribution were determined from a count of at least 250-300 particles. Semi-quantitative estimation of gold loading was performed by energy-dispersive X-ray spectroscopy (EDXS).
X-ray diffraction (XRD) analysis was carried out in a PAN'alytical X'Pert MPD equipped with a X'Celerator detector and secondary monochromator. Rietveld refinement with PowderCell software  was used to identify the crystallographic phases present and to calculate the crystallite size from the XRD diffraction patterns. Further details can be found elsewhere [34–38].
Catalytic activity measurements for CO oxidation were performed using a continuous-flow reactor. The catalyst sample (0.2 g) was placed on a quartz wool plug in a 45-cm long silica tube with 2.7 cm i.d., inserted into a vertical furnace equipped with a temperature controller. Feed gas (5% CO, 10% O2 in He) was passed through the catalytic bed at a total flow rate of 50 ml · min-1 (in contrast with most literature studies that use 1% CO or less [1–4, 31]). The composition of the outgoing gas stream was determined using a gas chromatograph equipped with a capillary column (Carboxen 1010 Plot, Supelco) and a thermal conductivity detector. Further details can be found elsewhere [34–38].
The BET surface area obtained for the MgO sample by N2 adsorption at -196°C was 32 m2 · g-1. This value is smaller than those reported in the literature [9, 23]. Both the thermal treatment of the support at 400°C and/or addition of gold by any of the methods described did not produce significant changes in the BET surface area.
The Au particle size could not be determined for any of the gold-loaded samples through XRD analysis, since the characteristic XRD reflection was absent in these materials. This can be due to the low loading (1 wt%) and small size of Au particles present in these catalysts, as it will be seen by HRTEM.
Gold nanoparticles of 6 nm were reported in literature for Au/MgO catalysts prepared by CP . Smaller values of approximately 4 nm were however obtained by CP and DP on Mg(OH)2[5, 41, 42]. Sizes of approximately 4 nm were also obtained for Au on MgO prepared from a gold complex . Au nanoparticles smaller than 5 nm were obtained on MgO modified with ascorbic acid [9, 23]. Other techniques like impregnation produced gold particles of 8 nm on MgO . Values of approximately 9 nm were obtained for gold on MgO with cube morphology . Gold deposited on MgO/alumina yielded particles ranging from 2.7 to 4.6 nm [21, 22, 24, 44].
Semi-quantitative estimation of gold loading was performed by EDXS, approximately 0.9% being found for all samples.
It can be argued that there are gold catalysts that achieve full CO conversions already at room temperature, but it has to be taken into account that most studies in literature use 1% CO or less [1–4] (while we used 5% of this gas). Also, the majority of authors use higher loadings of Au [1–4] (while we used 1 wt%). Nevertheless, it is possible to see, in our case, that CO conversion increases up to four times by addition of gold (for MgO with Au loaded by DIM), when compared to the unloaded samples.
Schubert et al.  reported activities of 13 × 10-4 and 3.8 × 10-4 molCOgAu -1 · s-1 at 80°C for Au/Mg(OH)2 and Au/MgO catalysts, respectively, both prepared by CP, while Haruta's group obtained 1.2 × 10-4 molCOgAu -1 · s-1 at -70°C for a Au/Mg(OH)2 prepared by DP . Our values for the DIM catalyst, ranging from 1.7 × 10-4 to 3.8 × 10-4 molCOgAu -1 · s-1 at 25 and 100°C (Figure 5b), respectively, are similar to the literature value obtained with Au/MgO catalyst, but below the value obtained for the Au/Mg(OH)2 material . Nevertheless, it was shown that the heat-treated samples (that have MgO instead of Mg(OH)2) have similar activity, meaning that the here reported DIM materials have similar catalytic activity to those reported in the literature, although with double Au content (1% Au, instead of 0.5% Au reported in ). LPRD and US showed smaller values.
Au was loaded (1 wt%) on a commercial MgO support by three different methods: double impregnation (DIM), liquid-phase reductive deposition (LPRD) and ultrasonication (US). CO oxidation was used as a test reaction to compare the catalytic activity. The best results were obtained with the DIM method, which showed activities of 1.7 × 10-4 to 3.8 × 10-4 molCOgAu -1 · s-1 at 25 and 100°C. This can be explained in terms of the nanoparticle size, well known to be related with the catalytic activity of gold catalysts. This sample had the narrowest size range (2-12 nm) and the lowest average size (5.4 nm). Samples prepared by other methods (LPRD and US) showed broader size ranges (3-12 nm) and larger average gold sizes (> 6.6 nm).
energy-dispersive X-ray spectroscopy
high-resolution transmission electron microscopy
liquid-phase reductive deposition
Authors thank Fundação para a Ciência e a Tecnologia (FCT), Portugal, for financial support (CIENCIA 2007 program for SAC), and project PTDC/EQU-ERQ/101456/2008, financed by FCT and FEDER in the context of Programme COMPETE. We also acknowledge CONACYT Grant No 79062, PAPIT-UNAM IN100908 (Mexico) and by RFBR grant 09-03-00347-a (Russia).
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.