Comparing Pt/SrTiO₃ to Rh/SrTiO₃ for hydrogen photocatalytic production from ethanol

Abstract

Photocatalytic hydrogen production from ethanol as an example of biofuel is studied over 0.5 wt% Rh/SrTiO₃ and 0.5 wt% Pt/SrTiO₃ perovskite materials. The rate of hydrogen production, r_H2, over Pt/SrTiO₃ is found to be far higher than that observed over Rh/SrTiO₃ (4 × 10⁻⁶ mol of H₂ g ⁻¹_catal. min⁻¹ (1.1 × 10⁻⁶ mol of H₂ m ⁻²_catal. min⁻¹) compared to 0.7 × 10⁻⁶ mol of H₂ g ⁻¹_catal. min⁻¹ (5.5 × 10⁻⁸ mol of H₂ m ⁻²_catal. min⁻¹), respectively, under UV excitation with a flux equivalent to that from the sun light (ca. 1 mW cm⁻²). Analyses of the XPS Rh3d and XPS Pt4f indicate that Rh is mainly present in its ionic form (Rh³⁺) while Pt is mainly present in its metallic form (Pt⁰). A fraction of the non-metallic state of Rh in the catalyst persisted even after argon ion sputtering. The tendency of Rh to be oxidized compared to Pt might be the reason behind the lower activity of the former compared to the later. On the contrary, a larger amount of methane are formed on the Rh containing catalyst compared to that observed on the Pt containing catalyst due to the capacity of Rh to break the carbon–carbon bond of the organic compound.

Introduction

Photo-catalytic production of hydrogen from renewables is poised to be one of its main sources in the future once successful catalytic materials are made possible. The reaction requires the presence of a semiconductor with bang gap energy within the energy of the solar radiation, a conductor such as a noble metal to accept electrons from the conduction band in addition to hydrogen-containing compounds [1]. Ultimately, the desired compound for hydrogen production is water. Next to water are alcohols and, in particular, ethanol because it is provided from biomass and is therefore renewable [2].

SrTiO₃ is stable in water as well as in presence of organic compounds; more importantly it endures corrosion under UV excitation and unlike other non oxygen containing compounds is already oxygen terminated and therefore cannot be over oxidized. It has two band gaps: one indirect at 3.25 eV, similar to anatase the perovskite structure with TiO₂, and one direct at 3.75 eV [3]. The indirect band gap is between the upper valence band, mainly composed of O2p, and the empty states Ti3d (t_2g). The direct band gap is between O2p and Ti3d (e_g) levels [4] (Fig. 1).

Fig. 1

Top view (a), side view (b) and perspective view (c) of SrTiO₃. Small black spheres (Sr²⁺), large gray spheres (O²⁻), smaller yellow spheres in c are those of Ti⁴⁺ cations. Also indicated in d are the electronic transitions between O2p and Ti3d levels (redrawn from Ref. [4])

Numerous works have addressed the photo-catalytic activity of SrTiO₃ alone, with other transition metals as well as doped with other ions [5–11]. Results differ strongly from one study to the other due to difference in crystallinity, metal dispersion, effect of dopant and reaction conditions. In a recent work [6], the reactivity of SrTiO₃ with different particle sizes was tested for hydrogen production from water. It was found that bulk material (particles dimension >100 nm) was more active than 30 nm size particles which in turn were more active than 6.5 nm size particles (producing 28, 19.4 and 3 μmol of H₂ g ⁻¹_catal. h⁻¹, respectively [at 26.3 mW cm⁻² with λ in the (250–380 nm)]. Reasons for this decrease are attributed to an increase in the water oxidation overpotential for the smaller particles and reduced light absorption due to quantum size effect. In another work [7], probing the anisotropy of the reactivity of SrTiO₃ microcrystals indicated that both reduced and oxidized products are formed preferentially on {100} surfaces. This anisotropy was explained as being due to differences in the electronic band structure. Because direct optical transitions for charge carriers having momentum vectors in the <100> direction overlap well with the spectral distribution of the absorbed photons, more photogenerated carriers are moving toward {100} surfaces than other surfaces and, as a result, the {100} surfaces are more active. Other work [8] has addressed the photocatalytic water splitting activity for hydrogen production over the mesoporous-assembled SrTiO₃ nanocrystal-based photocatalysts with various hole scavengers: methanol, ethanol, 2-propanol, d-glucose, and Na₂SO₃. Pristine mesoporous-assembled SrTiO₃ photocatalysts exhibited higher photocatalytic activity in hydrogen production than the non-mesoporous-assembled commercial photocatalysts or commercial SrTiO₃. These results indicate that the mesoporous assembly of nanocrystals with high pore uniformity plays a significant role, affecting the photocatalytic hydrogen production activity.

Moreover, it was seen that the Pt co-catalyst enhances the visible light harvesting ability of the mesoporous assembly with an optimum Pt loading of 0.5 wt% on the mesoporous-assembled SrTiO₃ photocatalyst providing the highest photocatalytic activity, with hydrogen production rate of 276 μmol h⁻¹ g_catal. and a quantum efficiency of 1.9 % under UV light irradiation. Other works indicated that Rh (1 %)-doped SrTiO₃ photocatalyst loaded with a Pt co-catalyst (0.1 wt%) gave 5.2 % of the quantum yield at 420 nm for the H₂ evolution reaction from a methanol solution (10 vol%) [9]. Another work [10] focused on doping SrTiO₃/TiO₂ with N ions (using hexamethylenetetramine) and optimized photocatalytic activity of hydrogen production (average hydrogen production rate = 5.1 mmol g ⁻¹_catal.  h⁻¹ with 2 wt% loaded Pt) under visible light was seen although little is known about the catalyst stability and comparison with the non-doped semiconductor material. A recent computational study (using the hybrid DFT method) of doping SrTiO₃ with metal cations coupled with experimental study showed that co-doping Cr and La ions had considerable enhancement effect on hydrogen production of water/methanol solution [11]. The reason invoked is that doping with La ions raised the Fermi level of Cr ions and stabilizes its oxidation state of Cr³⁺ which is needed for the red shift of the band gap to extend light absorption into the visible region.

In this work, we have conducted a study of Rh/SrTiO₃ and Pt/SrTiO₃ materials in addition to monitoring their photo-catalytic reaction to further understand the extent of their activity. In particular, we attempt to answer which of the two metals is more active when added to the semiconductor support. We find that Rh has a weaker effect than Pt on hydrogen production (which is similar to the case where these metals are deposited on TiO₂ anatase [12]). Although we have not conducted detailed study of the particle size effect of the semiconductor, we do not find considerable difference in the reaction rates upon changing its morphology.

Experimental

SrTiO₃ was prepared by the sol–gel method where TiCl₄ was added to a strontium-nitrate solution in stoichiometric amounts. After the addition of TiCl₄ to the strontium-nitrate solution, the pH was raised with sodium hydroxide to a value between 8 and 9 where strontium hydroxide and titanium hydroxide precipitated. The precipitate was left to stand for about 12 h at room temperature to ensure completion of the reaction after which it was filtered and washed with de-ionized water until neutral pH (~7). The resulting material was then dried in an oven at 100 °C for a period of at least 12 h. Next the material was calcined at 800 °C. X-ray diffraction techniques were used to indicate formation of SrTiO₃ (Fig. S1). Several SrTiO₃ materials were tested in addition to a commercial SrTiO₃ (Sigma-Aldrich). Rh metals were impregnated from a solution containing RhCl₃ in 1 N HCl. The resulting catalyst differed from the initial SrTiO₃ as its particle size was far smaller and its BET surface area higher. Pt metals were impregnated the same manner (from a PtCl₄ precursor). Other techniques used to study the material included XPS, XRD, TEM, and UV–vis. Photoreaction was conducted in a batch reactor (100–250 mL) containing 10–25 mg of materials under stirring conditions with a UV lamp flux of about 1 mW cm⁻². Analysis was conducted using GC-TCD equipped with a Porapack Q with a N₂ carrier gas isothermally at 50 °C (N₂ flow rate = 20 mL min⁻¹). At these conditions, the products were eluted from the column with the following order: hydrogen, oxygen, methane, carbon dioxide, ethylene, ethane, propylene, propane, acetaldehyde, followed by ethanol. The powder XRD patterns of the samples were recorded on a Philips X’pert-MPD X-ray powder diffractometer. A 2θ interval between 10° and 90° was used with a step size of 0.010° and a step time of 0.5 s. The diffractometer was equipped with a Ni-filtered Cu K_α radiation source (λ = 1.5418 Å). The X-ray source was operated at 45 mA and 40 kV. Sample preparation for the X-ray analysis involved gentle grinding of the solid into a fine powder and packing of approximately 0.1–0.3 g of the sample into an aluminum sample holder with light compression to make it flat and tight. XRD patterns of the samples were recorded with the X’pert HighScore Plus software and saved in XRDML text format for further manipulation and processing. X-Ray Photoelectron spectroscopy was conducted using a Thermo Scientific ESCALAB 250 Xi, equipped with a mono-chromated AlK_α X-ray source, Ultra Violet He lamp for UPS, ion scattering spectroscopy (ISS), and reflected electron energy loss spectroscopy (REELS). The base pressure of the chamber was typically in the low 10⁻¹⁰ to high 10⁻¹¹ mbar range. Charge neutralization was used for all samples (compensating shifts of ~1 eV). Spectra were calibrated with respect to C1s at 284.7 eV. The Rh3d, Pt4f, O1s, Sr3d, Ti2p, C1s and valence band binding energy regions were scanned for all materials. Typical acquisition conditions were as follows: pass energy = 20 eV and scan rate = 0.1 eV per 200 ms. Ar ion bombardment was performed with an EX06 ion gun at 2 kV beam energy and 10 mA emission current; sample current was typically 0.9–1.0 μA. The sputtered area of 900 × 900 μm² was larger than the analyzed area: 600 × 600 μm². Self-supported oxide disks of approximately 0.5 cm diameter were loaded into the chamber for analysis. Data acquisition and treatment were done using the Avantage software.

Results

Figure 2 presents a TEM image of the Rh/SrTiO₃ catalyst. Rh particles with mean size of 3–4 nm are seen. SrTiO₃ is composed of small crystallites with sizes between 10 and 20 nm. The BET surface area of the material is found to be ca. 13 m² g ⁻¹_catal. Figure S2 presents high resolution image of an Rh particle; the spots in the FT image unambiguously identify it by its crystalline structure. Figure 3 presents TEM of two types of Pt/SrTiO₃: one where Pt was deposited on commercial SrTiO₃ and the other on SrTiO₃ that was prepared by the sol–gel method followed by annealing at 800 °C. The commercial SrTiO₃ is composed of particles with dimension larger than 1 μm while the one prepared by the sol–gel method is made of typical perovskite (cubic) particles with a mean size of about 50 nm; Pt particles cannot be seen at this resolution.

Fig. 2

TEM of 0.5 wt% Rh/SrTiO₃. Rh can be seen as dark particles in the main figure as well as the inset

Fig. 3

TEM of 0.5 wt% Pt/SrTiO₃. a Commercial SrTiO₃ and b SrTiO₃ prepared by the sol–gel method; the inset in b presents the [100] cubic structure of SrTiO₃ evidence of a high degree of crystallinity

XP spectra Ti2p, Sr3d, O1s, and Rh3d (or Pt4f) were collected among other lines; here we focus on the main lines relevant to this study. Figure 4 presents XPS Rh3d and Ti2p of the as-prepared catalyst (a, b) and Ar ions sputtered (c, d). The presence of Rh³⁺ (mainly as Rh₂O₃) and Rh metal can be seen with binding energy = 307.0 and 308.8 eV, respectively, in line with reference elements [13]. To further confirm the presence of Rh metal as well to probe the extent of its reduction, the same area was Ar ions sputtered for 2 min (2 kV, 10 mA; sample current 10 mA). Clearly, the signal from Rh metal increased compared to that of Rh³⁺ due to reduction of Rh ions under bombardment. However, only about 50 % of Rh ions are reduced to metallic Rh. Our objective was not to reduce all Rh ions but to observe for the effects of mild ion sputtering on the reduction of the noble metal. Associated with Rh ions reduction is a reduction of Ti⁴⁺ ions to Ti³⁺ and Ti²⁺ ions (b, d) as has been seen in numerous work for Ar ions reduction of TiO₂ [14, 15].

Fig. 4

XPS Rh 3d and Ti2p of 0.5wt% Rh/SrTiO₃. a, b As-prepared and c, d after Ar ions sputtering

Figure 5 presents the Sr3d, Ti2p and Pt4f of the as-prepared 0.5 wt% Pt/SrTiO₃ catalyst. Sr3d is typical of the doublet (3d_5/2, 3d_3/2) of Sr²⁺ ions (spin orbit splitting close to 2 eV), the narrow lines of the Ti2p_3/2,1/2 and their binding energy positions indicate the absence of Ti³⁺ while the Pt4f_7/2,5/2 is that of metallic Pt (binding energy of Pt_7/2 at 72.5 eV with a spin orbit splitting of 3.4 eV). The presence of Pt metal in the as-prepared catalyst is in sharp contrast to that of the Rh/SrTiO₃ where a large fraction of Rh is found to be in the form or Rh³⁺ ions. This is typical of Pt deposited on reducible metal oxides where it is often largely present in its metallic form [16, 17].

Fig. 5

XPS Sr3d, Ti2p and Pt 4f of the as-prepared 0.5 wt% Pt/SrTiO₃. The Sr to Ti atomic ratio is about 1

Figure 6 presents results of the hydrogen production using photons of 3.3 eV in a batch reactor containing 0.5 wt% Pt/SrTiO₃ and 0.5 wt% Rh/SrTiO₃. Hydrogen production is seen together with methane. The production rate is comparable to that observed on Au/TiO₂ anatase previously and far higher than that of Au/TiO₂ rutile [18]. The ratio Rh/Ti is found equal to 0.07 while that of Pt/Ti is equal to 0.16. Two main observations are clear from Fig. 6. First Pt/SrTiO₃ is about six times higher than Rh/SrTiO₃ based on weight and about 20 times higher based on area. Even if we account for difference in the M to Ti⁴⁺ ratios, the rate of hydrogen production is still far higher for the Pt-containing catalysts compared to that of the Rh one. Second, Rh/SrTiO₃ produces large amounts of methane when compared to Pt/SrTiO₃. The larger amounts of methane can be rationalized by the fact that Rh is more active in breaking the C–C bond compared to Pt [19–21]. In this case, the ratio H₂ to CH₄ is equal to 2 for Rh/SrTiO₃ while it is equal to 8 in the case of Pt/SrTiO₃.

Fig. 6

Hydrogen and methane photo-production from ethanol over 0.5 wt% Rh/SrTiO₃ and 0.5 wt% Pt/SrTiO₃. BET surface area of Rh/SrTiO₃ = 13 m² g⁻¹ while that of Pt/SrTiO₃ is 3.5 m² g⁻¹. The numbers inside the figure indicate the slope (the rate in moles of hydrogen per g_catal. per minute)

The reaction occurs due to electron excitation from the valence band (VB) O2p to the conduction band (CB) Ti3d upon UV illumination, as presented in Fig. 1. Alcohols are known as hole scavenges [22–24]. Ethanol is oxidized to acetaldehyde via two electron injection into the VB [25] with a time scale in the nanosecond range [26]. Mechanistically ethanol is first dissociatively adsorbed on the surface forming ethoxides (according to Eq. 2) and then upon two electron injections (through α-oxy radical [18, 25]) acetaldehyde is formed. The hydrogen ions released through this process are reduced to one hydrogen molecule [27, 28]. Reactions 1–4 represent these steps.

$${\text{SrTiO}}_3+\text{UV}\to {\text{e}}^{-}+{\text{h}}^{+}$$

(1)

$${\text{CH}}_3{\text{CH}}_2\text{OH}\to {\text{CH}}_3{\text{CH}}_2\text{O}\left(a\right)+\text{H}\left(\text{a}\right)$$

(2)

$${\text{CH}}_3{\text{CH}}_2\text{O}\left(a\right)+2{\text{h}}^{+}\to {\text{CH}}_3\text{CHO}+\text{H}\left(\text{a}\right)$$

(3)

$$2\text{H}\left(a\right)+2{\text{e}}^{-}\to {\text{H}}_2$$

(4)

(a) adsorbed

The reaction over Pt/SrTiO₃ is similar to that over Au/TiO₂ where the main products are H₂ and acetaldehyde [1, 18], with traces of methane. However, over Rh/SrTiO₃ the reaction proceeds to CH₄. Therefore, a large fraction of CH₃CHO farther reacts with CH₄ and CO.

$${\text{CH}}_3\text{CHO}\to {\text{CH}}_4+\text{CO}$$

(5)

As presented in Fig. 6, CH₄ is formed. Therefore, one should except if the reaction of Eq. 5 is complete, equal amounts of H₂ and CH₄ would be formed. We have conducted numerous runs and have found that in all cases CH₄ concentration was lower than that of H₂. Recent DFT studies on SrTiO₃ indicated the possibility of a pathway for ethanol decomposition involving CH₃CO radical (with the release of three hydrogen ions) [29]. In this case, CH₃CO may react with surface oxygen making acetate species or may split to CH₃ radical and CO; CH₃ radicals may react with OH radicals giving methanol, the latter would easily decompose to CO and hydrogen [30]. This route may explain the higher ratio H₂ to CH₄ observed. To extract all hydrogen atoms from ethanol, the formation of CH₄ is not desired and at present work is in progress to find a method to more efficiently break the carbon–carbon bond before obtaining acetaldehyde at 300 K under photon irradiation in the presence of water providing additional hydrogen and oxygen ions to complete the reforming reaction, as in high temperature ethanol steam reforming [31].

Conclusions

The rate of photo-catalytic reaction of ethanol to hydrogen is found to be far higher on Pt/SrTiO₃ compared to Rh/SrTiO₃. The most likely reason for the higher activity of the Pt containing catalysts is the ease by which metallic Pt is formed compared to Rh (where a large fraction of the as-prepared material is in Rh³⁺). Ar ions sputtering Rh/SrTiO₃ considerably reduced Ti in SrTiO₃, but only reduced about half of Rh ions to metallic Rh. Another important difference is noticed between the two catalytic systems and it is related to reaction selectivity. Rh/SrTiO₃ produces large amounts of CH₄ compared to Pt/SrTiO₃; this is most likely related to the capacity of Rh to break the carbon–carbon bond compared to that of Pt.