Magnetically modified nanogold-biosilica composite as an effective catalyst for CO oxidation

2.1 The temperature-dependent biosynthesis of gold nanoparticles

Cultivation of Diadesmis gallica was performed according to (Schröfel et al., 2011; Guillard and Lorenzen, 1972). The gold precursor (1 mM HAuCl₄, Carl Roth GmbH + Co. KG, ≥ 99,5%) was mixed with concentrated DG suspension in the ratio 1:1 (Konvičková et al., 2016). Samples were then incubated at 25 °C (DG/Au(25)) and 5 °C (DG/Au(5)) until the colour of the mixture turned ruby red due to the presence of AuNP. After the biosynthesis of AuNP, both samples were separated from the culture media via sedimentation and dried at 37 °C for 24 h.

2.2 Magnetic modification of DG/Au(5) sample

DG/Au(5) sample was magnetically modified by perchloric acid stabilized ferrofluid containing NPs of magnetic iron oxides, using two different methods, i.e., with and without the use of methanol. The first procedure using methanol is the traditional approach employing a transfer of the sample to methanol followed by the addition of ferrofluid (Safarik et al., 2007). The second later discovered method (without methanol) is based on the simple mixing of the sample with ferrofluid (Safarik et al., 2012). In more detail, in the first procedure, 10 mL of methanol (p.a.) and 1 mL of ferrofluid (concentration 46.2 mg mL⁻¹, pH 1.5) was added to 0.5 g of DG/Au(5). Next, the sample was stirred for two hours in the dark. Magnetically modified sample DG/Au(5)MF (M = methanol, F = ferrofluid) was separated from methanol and suspension residues by a magnetic separator. The sample was washed twice in methanol and dried. In the second procedure, 0.6 g of DG/Au(5) and 2 mL of ferrofluid at the concentration of 46.2 mg mL⁻¹ and pH 1.5 were mixed thoroughly and subsequently dried to form DG/Au(5)F composite. Both magnetic composites were stabilized at 60 °C for 24 h.

Magnetic parameters of magnetically modified samples DG/Au(5)MF and DG/Au(5)F were measured by vibrating sample magnetometer (VSM) EV9 at room temperature. Displacement hysteresis loops represent the dependence of the magnetization (M) on the applied magnetic field (H).

2.3 Calcination of magnetically unmodified and modified bionanocomposites

The oxidation of CO was selected as a model reaction for the catalytic activity testing of prepared samples. Given the used methodology and the catalytic process itself, tested samples required several pre-treatments before their using in the catalysis, namely calcination, desired size fraction preparing and activation in the stream of helium. Since the catalytic reaction itself was carried out up to 330 °C, the temperature of calcination was set to 450 °C in air for 4 h. Sufficient difference between the maximum reaction temperature and the temperature of calcination was set with respect to the stability of samples during the whole catalysis procedure.

2.4 Morphological characterization of nanogold-biosilica composites and size distribution of nanoparticles

The diatom cells before and after the biosynthesis of AuNP were observed using an optical microscope (Olympus, CX 31). Size and shape of AuNP in samples were studied by transmission electron microscopes (FEI TecnaiSphera 2) and (Jeol 1200 EX) operating at 120 keV. Small amounts of powder samples were suspended in ethanol and fixed on the grids. The AuNP size distribution was calculated using JMicrovision: Image Analysis Toolbox (www.jmicrovision.com), when at least 250 NP were evaluated per sample.

2.5 Spectroscopic characterization of nanogold-biosilica composites

Contents of gold, silica and iron in composites were determined using atomic emission spectrometry with inductively coupled plasma (ICP-AES, ICP CirosVision, Spectro). The crystalline structure of samples was characterized using X-ray diffraction analysis using Brucker D8 Advance diffractometer (Brucker AXS) working with the Bragg-Brentano geometry and position-sensitive Vantec detector and Krystalloflex K780generator. Cobalt lamp CoKα (λ = 1.7889 Å) was used as a source of radiation. Crystalline phase was identified by PDF-2Release 2004 databases (International Centre for Diffraction Data).

2.6 CO catalytic oxidation

The bionanocomposites after calcination were pressed to tablets under the compressive force of 40 kN, followed by crushing and sieving to get a required fraction in the range from 0.16 to 0.315 mm. Prior to oxidation measurement, the sample was activated in the stream of He during 1 h at 300 °C. The activation is performed in order to expose the active sites and remove the impurities adsorbed from the air.

CO catalytic oxidation was carried out in a tubular reactor (i.d. of 5 mm) with an integral catalyst bed in the 60–330 °C temperature range under the atmospheric pressure. The feed flow rate was 100 mL min⁻¹ and contained 2500 ppm CO and 2 mol.% O₂ in He. The catalyst sample of 0.1 g was layered between the inert glass fractions to ensure uniform flow of the gas mixture so that the plug flow was ensured. The input and output concentrations of CO were analysed using gas chromatograph AGILENT 7890A with thermal conductivity detector.

CO conversion was chosen as a measure for the catalytic activity and was calculated using the Eq. (1), where constant total volume flow due to low CO concentration is assumed; $c_{\mathit{COin}}$ is the inlet CO concentration and $c_{\mathit{COout}}$ is the outlet CO concentration.

3.1 Characterization of nanogold-biosilica composites after the temperature-dependent biosynthesis

The probable mechanism of AuNP fabrication using brown algae has been described in previous publications (Schröfel et al., 2011; Konvičková et al., 2016). Gold ions present in the suspension are reduced using biomolecules (for example extracellularly polysaccharides) contained in the silica algae. Biosynthesized AuNPs are subsequently associated and stabilized on the surface of diatomaceous frustules. During biosynthesis, the influence of two different temperatures (25 °C and 5 °C) on the size and homogeneity of nanogold particles in the composite was tested. The samples of DG with AuNP for both temperatures changed its colour from brown through purple to dee purple (Fig. A.1 Supplementary Data) due to the presence of bionanogold particles.

Whole bioreduction process of gold ions in DG/Au(25) sample at 25 °C took app. 48 h. Biosynthesis of AuNP at 5 °C took 96 h. That was the expected result because, as it is generally known, the temperature influences activity and mobility of ions and thus subsequent formation of NPs (Kumari et al., 2016).

TEM (transmission electron microscopy) micrographs revealed that spherical AuNP biosynthesised at 5 °C have narrow size and more homogeneous size distribution than AuNP biosynthesised at 25 °C. The average particle size is around 15 nm for DG/Au(5) sample and 17 nm for DG/Au(25) (Figs. A.2 and A.3 Supplementary Data). The smaller size of AuNP prepared at 5 °C may be caused by the increase in the activation energy of Au cluster nucleation which leads to a slower growth of Au nuclei in the solution (Polte, 2015). Because AuNP in DG/Au(5) sample were smaller (mostly 14 nm) and their size distribution was homogenous, higher efficacy of prepared bionanocomposite for catalysis in lower temperature can be expected (Dobrosz et al., 2005).

ICP-AES analysis revealed the total content of gold in DG/Au(25) and DG/Au(5) samples as 7.61 ± 0.76 wt.% and 6.46 ± 0.65 wt.%, respectively. Regarding standard deviation, it can be considered that the content of gold is almost identical in both samples. Catalytic activity will be influenced mainly by NPs size and dispersity on frustules surface.

3.2 Calcination and its effect on the nanogold-biosilica composites

Prepared bionanocomposite is a complex system full of organic parts, which are unstable at high temperatures and can decrease the catalytic activity of samples (Yang et al., 2015). For this reason, the calcination is a necessary step. The calcination of the sample guarantees its stability during the CO oxidation process itself, which also works at elevated temperatures (up to 300 °C), therefore the calcination temperature should be at least 450 °C. During the calcination process, the size of NPs could increase (Kratošová et al., 2016). Therefore, it is useful to prepare initial NPs with the smallest possible size distribution before calcination and catalysis.

3.2.1

3.2.1 DG/Au(25) and DG/Au(5) after calcination

TEM analysis confirmed the growth of NPs on the biosilica surface of DG in DG/Au(25) and DG/Au(5) after calcination. For both samples, NPs size distribution was fitted by the lognormal curve, which pointed on the polydispersity of AuNP anchored on frustules (Fig. 1). The size of NPs in DG/Au(25) increased to an average size of 21 nm after calcination, with most occurred NPs of 19–21 nm (see Fig. 1b). High temperature during the calcination process caused an increase of AuNP size by approximately 4 nm.

Close

Fig. 1

TEM micrograph of AuNP associated on diatom frustules of DG/Au(25) (a) and DG/Au(5) (c) after the calcination and size distribution histogram of DG/Au(25) (b) and DG/Au(5) (d).

Export to PPT

The average size of AuNP in DG/Au(5) was still 15 nm after the calcination, but now in the polydisperse system of nanogold on silica frustules. The nanogold size was around 16–18 nm, which is evident from the histogram in Fig. 1d.

With regard to the HAuCl₄ precursor in biosynthesis experiments, we may suppose some poisoning effect of the residual chlorides. Residual chloride anions may cause the agglomeration of AuNP by forming Au-Cl-Au bridges and change the particle size distribution during the calcination process (Dobrosz et al., 2005). Anyhow, it seems to be useful to wash the samples to remove chloride ions after the biosynthesis of AuNP.

During the calcination, organic parts are burned at high temperature leading to the loss of weight which was in our samples 48% and 57% for DG/Au(25) and DG/Au(5), respectively. Due to the calcination, the content of gold in samples was higher than in non-calcined, i.e.16.6 ± 1.7 wt.% and 15.5 ± 1.5 wt.% in DG/Au(25) and DG/Au(5), respectively.

The crystallographic analysis confirmed that only the gold remained after the calcination in the samples, the silica was not detected due to the amorphous nature (Wang et al., 2013). The crystalline structure of AuNP in samples DG/Au(25) and DG/Au(5) before and after calcination was studied using X-ray diffraction (XRD). The XRD pattern (Fig. 2) showed 3 characteristic diffraction peaks at 2 theta values 45°, 52° and 77° corresponding to (1 1 1), (2 0 0) and (2 2 0) planes of cubic crystal structure of gold (PDF number 04-0784) for DG/Au samples. Intensities of Au peaks are higher after calcination which is related to growing particles size and crystalline character.

Close

Fig. 2

XRD patterns of DG/Au(25) (a) and DG/Au(5) (b) before – black line and after calcination – red line.

Export to PPT

3.2.2

3.2.2 Magnetically modified sample of DG/Au(5) after calcination

To improve properties regarding stability and catalytic efficiency, the treatment of composite with ferrofluid was assessed as an optimisation step. It was previously reported (Vigneron and Caps, 2016) that promoter reducible oxides (FeO_x, TiO₂, NiO) can improve the efficacy and thereby they turn on the catalytic cycle because they are also the source of oxygen needed for the CO oxidation.

For this experiment, DG/Au(5) sample was only used because of better physical properties (smaller size of NPs, narrower distribution, higher homogeneity) as described above. Finally, two types of samples were prepared: DG/Au(5)MF and DG/Au(5)F by a different method of magnetic modification (with and without the use of methanol).

The TEM revealed a heterogeneous size distribution of spherical AuNP for both magnetically modified samples after calcination. The average size of nanogold in DG/Au(5)MF was 23 nm. The most frequently represented size of AuNP was 23 nm (Fig. 3b). The average size of spherical nanogold in DG/Au(5)F was 20 nm. The most frequently poses size of AuNP was approximately 16–20 nm (Fig. 3d).

Close

Fig. 3

TEM micrograph of AuNP associated on diatom frustules of DG/Au(5)MF (a) with size distribution histogram (b) and DG/Au(5)F (c) with size distribution histogram (d) after calcination.

Export to PPT

Burning of organic parts during calcination process leads to the loss of weight in samples which was 52% and 43% for DG/Au(5)F and DG/Au(5)MF, respectively. The total amount of Au and Fe in magnetically modified bionanocomposite after calcination was measured by ICP-AES. Preparation of nanogold-biosilica composites proceeded at laboratory conditions and therefore we talk about low-volume preparation of sample (app. 1 g). Two samples with AuNP were prepared using the same method of biosynthesis (at 5 °C) for each magnetic modification (with and without methanol). DG/Au(5)F contained 3.70 ± 0.37 wt.% Au and 13.80 ± 1.00 wt.% Fe after the calcination. In the case of DG/Au(5)MF sample, 6.52 ± 0.66 wt.% Au and 10.40 ± 0.80 wt.% Fe was measured. Differences in content of Au in the magnetically modified samples may be caused by the irregular distribution of Au due to the static method of preparation. This deficiency can be avoided by continuously mixing the solution at the time during the bioreduction of Au ions. The lower content of Fe in the case of DG/Au(5)MF was predicted because 1 mL of ferrofluid was added to 0.5 g of pure nanogold-biosilica composite, while in another modification protocol 0.6 g of composite was mixed with 2 mL of ferrofluid obtaining DG/Au(5)F sample. Theoretically, the amount of iron is 1.7× higher than in the sample modified with methanol. This result has not been achieved due to the different method of preparation. Method of magnetic modification of sample with methanol has a longer tradition and was earlier verified on various samples as well as on diatoms (DG) (Safarik et al., 2007; Kratošová et al., 2013). Surprisingly, the direct mixing of ferrofluid with bionanocomposite was less effective than in the case of methanol method.

Ferrofluid used for composites modification contains magnetic iron oxide nanoparticles (Safarik et al., 2012). Using XRD analysis for DGAu(5)MF and DGAu(5)F magnetite and maghemite diffraction lines were not observed before their calcination (Fig. 4), but only in the case of DGAu(5)F sample the presence of ferrosilite (FeSiO₃) was revealed. XRD study proved these expectations and showed presence of magnetite in magnetically modified samples after their calcination (Fig. 4). Higher intensities of iron oxides peaks can be caused by growing particles size and crystalline character.

Close

Fig. 4

X-ray diffraction of DGAu(5)F (a) and DGAu(5)MF (b)- red line and after calcination - black line.

Export to PPT

However, due to similar positions of maghemite and magnetite diffraction lines in the XRD pattern, it is difficult to differentiate between them.

The response of magnetic NPs in ferrofluid to external magnetic fields reflects characteristic different spin dynamical processes compared to those of their bulk counterparts. In order to test if the obtained magnetically modified DG/Au(5)MF and DG/Au(5)F could be used in the magnetic separation procedures, samples were tested using VSM. The test with magnetization measurements as a function of the applied magnetic field was performed. At room temperature, the field-dependent magnetization curves of all prepared magnetic composites show their ferromagnetic properties (Fig. 5) and thus their ability to be influenced and manipulated in the external magnetic field, which could be beneficial in future application of catalysts.

Close

Fig. 5

Magnetic curves of magnetically modified samples in the external magnetic field for DGAu(5)MF (blue line) and for DGAu(5)F (red line) after calcination.

Export to PPT

Regarding magnetization curves, samples have been already saturated in the applied external magnetic field (15.000 Oe). Both the curves have similar shapes, assuming similar magnetic behaviour of samples. The values of magnetization saturation (Ms) and coercive field (Hc) are shown in the Table 1.

Table 1 Basic magnetic parameters of DG/Au(5)MF and DG/Au(5)F composites.

Sample	Ms [emu/g]	Hc [Oe]
DG/Au(5)MF	4.95	1.45
DG/Au(5)F	4.09	3.86

3.3 CO catalytic oxidation over prepared bionanocomposites

The catalytic activity of non-magnetic (DG/Au(25), DG/Au(5)) and magnetically modified (DG/Au(5)F, DG/Au(5)MF) nanogold-biosilica samples was tested for catalytic oxidation of CO at temperatures 60–330 °C (Fig. 6).

Close

Fig. 6

Temperature dependences of CO conversion in the presence of bionanocomposites as the catalyst.

Export to PPT

The conversion of CO at 300 °C was 58% in the presence of DG/Au(25) and 60% when using DG/Au(5). The sample DG/Au(5) with smaller and uniform AuNP were presented (Table 2), did not show a significant difference in catalytic activity compared to the sample DG/Au(25) with bigger NPs.

The lower catalytic efficiency of both samples in CO oxidation could be caused by residual Cl⁻ on the nanogold catalyst lead to deactivation of active sites in nanogold-biosilica composite. This phenomenon was confirmed in the work by (Chen et al., 2017). Moreover, the TEM analysis of samples after catalytic tests confirmed the unification of size and distribution of NPs for both samples as it is seen in Fig. 7. In the DG/Au(5) sample, the average size of AuNP was 20 nm and in the sample of DG/Au(25) was 23 nm after the catalytic test. The most prevalent size of NPs was 22 nm for both samples.

Close

Fig. 7

TEM micrograph of AuNP associated on diatom frustules of DG/Au(25) (a) with size distribution histogram (b) and DG/Au(5) (c) with size distribution histogram (d) after catalysis.

Export to PPT

Results show that the size of AuNP increased for both samples (DG/Au(25) and DG/Au(5)) after catalytic tests. The reaction between HAuCl₄ and some acidic oxides, e.g. with negatively charged silica surfaces (isoelectric point 1, 2) is usually inactive (Dobrosz et al., 2005) and leads to relatively weak interaction between AuNP and inactive supports. This can cause the agglomeration of NPs, which may lead to decreasing catalytic activity in CO oxidation. To avoid agglomeration, several precipitation agents such as ammonia or urea have been investigated in the deposition-precipitation method. The use of ammonia causes the formation of gold ammine cationic complex, which reacts with the negatively charged silica surface (Dobrosz et al., 2005). Another approach is the limitation of particles size by capturing in the pore with controlled uniform pore system as present in zeolites or mesoporous silica (Chen et al., 2017). Stabilization of NPs via cellular debris seems to be sufficient at normal temperatures but to avoid unwanted growth at high temperatures it is probably better to anchor the particles more efficiently. On the other hand, it was reported that trioctahedral AuNP the size of 45 nm and 100 nm could be catalytically active as well, although the optimum of AuNP for CO oxidation is considered up to 5 nm (Sreedhala et al., 2015).

It was expected that magnetic modification would increase the catalytic activity of the bionanocomposites. Generally, the molecules of CO are adsorbed on the AuNP surface in Au/oxide-matrix based nanocomposite and the dissociation of O₂ occurs at the metal/support interface (between AuNP and SiO₂). Molecules of O₂ are stable at room temperature and, therefore, higher temperature is needed for their dissociation. In the case of AuNP on inactive support, the active sites on the AuNP surface are inhibited by the CO molecules due to the missing dissociated oxygen. By adding the reducible oxide to metal/inactive-supports, the new interface between AuNP and reducible oxides is formed and regarded as the location of the most active sites in CO oxidation. The release of oxygen from the matrix causes the presence of oxygen vacancy at the interface of AuNP/reducible oxide, where others O₂ can be dissociated. This process is called oxygen activation via a redox-type mechanism. Dissociated O₂ molecules then react with the adsorbed CO molecules on the AuNP surface via Langmuir-Hinshelwood mechanism and the catalytic process continues in the cycle. In this way, the reaction runs at lower temperatures, since oxygen is provided from promoter reducible oxide, in our case probably from FeO_x. Moreover, modification with promoter reducible oxides could inhibit the unwanted growth of NPs at high temperature (Beck et al., 2009).

DG/Au(5)F showed excellent catalytic effect in comparison with the nonmagnetic sample when 98% conversion of CO at 300 °C was observed. Moreover, CO molecules were completely converted at 330 °C. However, DG/Au(5)MF prepared using the method with methanol, did not confirm the increase of catalytic activity in CO oxidation. CO conversion at 300 °C was only 15% (18% CO conversion at 330 °C). The conversion was even lower than for unmodified samples. With regard to these results, it was surprising that the TEM analysis revealed the similar size distribution after catalytic experiments (Fig. 8). The average and the most frequently represented size was 23 nm in both magnetically modified samples. AuNP size distribution was fitted with the lognormal curve for both samples, which pointed to the polydisperse system.

Close

Fig. 8

TEM micrograph of AuNP associated on diatom frustules of DG/Au(5)MF (a) with size distribution histogram (b) and DG/Au(5)F (c) with size distribution histogram (d) after catalysis.

Export to PPT

We can assume that the method of magnetic modification influences the catalytic activity of the sample. Methanol used in the preparation of DG/Au(5)MF sample caused a decrease in catalytic activity. CO oxidation was found to be directly related to the number of perimeter sites, i.e., the number of bifunctional gold-support sites (Vigneron and Caps, 2016). The ratio of Fe to Au (Table 2) was higher in the case of DG/Au(5)F, which may indicate a larger amount of active sites at the Au/promoter reducible oxide interface.

There are several studies, which are focused on preparation of metallic NPs and their using for CO oxidation. For example work of (Qian et al., 2007), deposition-precipitation method for preparing of AuNP (7–15 nm) from HAuCl₄ precursor supported on SiO₂ were used. 100% CO conversion was achieved at 526 °C. 100% CO conversion at 60 °C were reported by (Ren et al., 2012). As nanocatalyst, they used Au/CeO₂/SiO₂ nanocomposite prepared by the deposition-precipitation method. AuNP size was 2–5 nm. In the work of Xu et al., the catalytic activity of Au/SiO₂ and Au/FeO_x/SiO₂ nanocomposites prepared by deposition-precipitation method was studied as well. Sizes of AuNP were 3–13 nm for Au/SiO₂ sample and 1–3 nm for Au/FeO_x/SiO₂ sample. Total conversion of CO over Au/SiO₂ and Au/FeO_x/SiO₂ nanomaterials was observed at 475 °C and 227–262 °C, respectively (Xu et al., 2010). Reducible oxides enhanced electron interactions, lead to the enlargement of the reactive species and finally to the increase of nanocomposite catalytic activity in CO oxidation (Comotti et al., 2006; Chen and Goodman, 2006; Vigneron and Caps, 2016). Only in one study, biosynthesized metallic nanoparticles were tested for CO oxidation. Ag nanoparticles prepared using plant extract were anchored on chemically prepared CeO₂ support. Moreover, the size of Ag nanoparticles (18 nm) was optimized by microwave irradiation. Completely conversion of CO over Ag/CeO₂ nanomaterial was achieved at 200 °C (Yang et al., 2015).

Bionanocomposites prepared by bioreduction and magnetically modified had comparable and satisfactory catalytic activity with regard to the above-mentioned articles despite to higher particle size. With respect to the physical and chemical methods of NPs preparation described in other works, the preparation of the bionanocatalyst in this study seems to be simple, environmentally friendly and economical undemanding. The main advantageous of AuNP biosynthesis using DG is forming and direct embedding of the nanogold on the 3D structure of the biosilica frustules in one step. It is not necessary chemically prepared matrix for NPs anchoring or stabilizer agents. The final nanomaterial is functional bionanocomposite (magnetically responsive), which shows catalytic activity in CO oxidation and can be further utilized as nanocatalyst for other degradation reactions (organic or inorganic compounds).