Thin Coatings of α‐ and β‐Bi₂O₃ by Ultrasonic Spray Coating of a Molecular Bismuth Oxido Cluster and their Application for Photocatalytic Water Purification Under Visible Light

Abstract

Thin coatings of Bi₂O₃ were deposited on glass substrates by ultrasonic spray coating of THF solutions of the molecular precursor [Bi₃₈O₄₅(OMc)₂₄(DMSO)₉] ⋅ 2DMSO ⋅ 7H₂O (OMc=O₂CC₃H₅) followed by hydrolysis and subsequent annealing. Depending on the synthetic protocol, the bismuth oxido cluster was transformed into either α‐ or β‐Bi₂O₃. The as‐synthesized Bi₂O₃ coatings were characterized by powder X‐ray diffraction (PXRD), thickness measurements, diffuse reflectance UV‐Vis spectroscopy (DRS), photoluminescence (PL) spectroscopy, Raman spectroscopy and scanning electron microscopy (SEM). The thin coatings (thickness: 5–16 μm) were compared with regard to their performance in photocatalytic rhodamine B (RhB) decomposition under visible light irradiation. The β‐Bi₂O₃ coatings, that showed the highest photocatalytic activity, were used for the photocatalytic decomposition of other pollutants such as triclosan and ethinyl estradiol. In addition, the interplay between the photooxidation that is induced by the excitation of the catalyst using visible light and the photosensitized decomposition pathway was studied by degradation experiments of aqueous rhodamine B solutions using β‐Bi₂O₃ coatings.

In the light: A preparation route for thin films of α‐ and β‐Bi₂O₃ by ultrasonic spray coating of a bismuth oxido cluster easy to apply for various substrates is presented. The as‐prepared Bi₂O₃ coatings were evaluated with regard to their photocatalytic activity in the decomposition of organic model dyes as well as pollutants, which are relevant to wastewater treatment using visible light.

1. Introduction

The preparation, modification and photocatalytic evaluation of semiconductors are an ongoing focus of research since the report of the UV light‐driven photocatalytic activity of TiO₂ in the early 1970s.1 Several bismuth‐based chalcogenides (e. g. Bi₂O₃,2, 3 Bi₂S₃,4 BiOCl)5 and heterometallic oxides (e. g. BiWO_6, 6 BiVO_4, 7, 8, 9, 10 Ba(BiO₃)₂,11 NaBiO₃)12 are suitable as advanced photoredox catalysts targeting visible‐light response.13 Bismuth oxide, which is classified as a non‐toxic compound with a high LD₅₀ value of 5000 mg/kg (rat, oral),14 is an emerging candidate for the photocatalytic degradation of organic compounds in wastewater such as dye pollutants, biocides or drug residues. The decomposition of the biocide triclosan and the medicinal drug ethinyl estradiol in wastewater is an important research target. Although their concentrations are still below the legal limit, their inlet as well as their outlet concentrations to sewage treatment plants are already above the detection limits of the routinely used analytical instruments, which was exemplarily monitored for two German cities.15 Residuals like the antimicrobial agent triclosan can be a problem for the so‐called secondary treatment, that is based on the use of microorganisms in sewage plants. Ethinyl estradiol is known for its high estrogenic activity even at concentrations already present in the environment, which can disrupt amphibian mating behavior for example. Therefore, residuals of ethinyl estradiol might contribute to the worldwide problem of amphibian decline.16 Previous investigations on dispersions of tetragonal β‐Bi₂O₃ nanoparticles have proven their excellent efficiencies by decomposing the initial amount of ethinyl estradiol and triclosan.3

In addition to the tetragonal β‐Bi₂O₃, the monoclinic α‐Bi₂O₃ and the cubic δ‐Bi₂O₃ polymorphs exhibit favorable visible‐light response.17 However, β‐Bi₂O₃ revealed the best photocatalytic performance under visible light irradiation in comparative studies on single‐phase oxides with regard to the decomposition of organic molecules like methyl orange and 4‐chlorophenol.18 Recently, we have shown that nanoparticles of the metastable polymorph β‐Bi₂O₃ are easily accessible by a controlled hydrolysis process under mild conditions of polynuclear bismuth oxido clusters due to their structural relationship between the well‐defined precursor and the tetragonal phase β‐Bi₂O₃.2

However, an implementation of photocatalysts based on bismuth oxide nanoparticles into sewage treatment plants or their use in decentral solutions for wastewater treatment is hampered by cost‐intensive separation of the catalyst nanoparticles from the purified water and their potential nanotoxicity, which is getting important in case of particle leaching.19 Therefore, recent studies focused on immobilization concepts with main emphasis given to well‐established photocatalysts like titanium dioxide.20, 21, 22 So far, a limited number of studies has been reported for the production of single‐phase β‐Bi₂O₃ coatings including low‐pressure chemical vapor deposition (CVD) starting from bismuth(III) tert‐butoxide. The resulting semiconductor films were shown to be suitable for UV‐light induced oxygen evolution from water in the presence of a sacrificial agent.23 In addition, a high visible‐light‐harvesting potential and incident photon‐to‐electron conversion efficiency were shown for β‐Bi₂O₃ nanoporous films obtained by radio frequency magnetron sputtering using Bi targets24 and sol‐gel spin coating of a solution of bismuth(III) nitrate pentahydrate in the presence of additives such as an ethanolic solution of triton X‐100 and citric acid.25 The water purification efficiency of single‐phase tetragonal β‐Bi₂O₃ coatings synthesized by spray pyrolysis using bismuth(III) acetate as the precursor was evaluated for the degradation of organic dye solutions like methyl orange, acid blue 113, rhodamine B and indigo carmin under UV and visible light.26, 27 Photochemical solution deposition using the photosensitive bismuth(III)‐N‐methyldiethanolamine complex is another suitable preparation method in order to immobilize tetragonal β‐Bi₂O₃. Efficient photodegradation of an aqueous methylene blue solution with a UV light source was demonstrated in the presence of the as‐prepared films.28, 29 In addition, in situ growth of Bi₂O₃ on a Cu surface starting with bismuth(III) nitrate pentahydrate provides β‐Bi₂O₃ films as well.30 To prove the photoelectric conversion performance, the immobilized catalysts were placed into a cell under chopped visible light irradiation.

Herein, we report an alternative preparation method for immobilized Bi₂O₃ polymorphs, based on ultrasonic spray coating.31 This cold spray technique enables an easy scale‐up from the laboratory scale to dimensions much larger in size, which would be needed for wastewater treatment plants. In addition, ultrasonic spray coating is quite flexible with regard to the choice of material characteristics of the substrates. Although dispersions might be used in ultrasonic spray coating, the use of molecular precursors, which are soluble in organic solvents, is favorable as this produces more uniform films. The most common starting material bismuth(III) nitrate pentahydrate is not suitable because it is only soluble with appropriate amounts under highly acidic aqueous conditions. Therefore, the previously reported bismuth oxido cluster [Bi₃₈O₄₅(OMc)₂₄(DMSO)₉] ⋅ 2DMSO ⋅ 7H₂O, which is easily accessible and soluble in a broad range of solvents due to its organic ligand shell working as solubility enhancer, was used in this study.32 We present the ultrasonic spray coating of this cluster followed by a hydrolysis and annealing protocol. Thus, thin coatings of α‐ and β‐Bi₂O₃ were prepared and their photocatalytic activity for the degradation of organic molecules such as rhodamine B under visible light irradiation was studied. Even more relevant to wastewater treatment, degradation experiments on the biocide triclosan and the medicinal drug ethinyl estradiol were performed using the as‐prepared photocatalytic coatings.

2. Results and Discussion

2.1. Thin Coating Formation

Thin coating formation of photoactive semiconductors was realized using a cold spray technique. The general concept of the preparation of Bi₂O₃ thin films by this ultrasonic spray coating starting with the well‐defined bismuth oxido cluster [Bi₃₈O₄₅(OMc)₂₄(DMSO)₉] ⋅ 2DMSO ⋅ 7H₂O is summarized in Scheme 1. First, the molecular precursor [Bi₃₈O₄₅(OMc)₂₄(DMSO)₉] ⋅ 2DMSO ⋅ 7H₂O was dissolved in THF to give a colorless solution that was spray coated on a glass substrate. Constant heating of the substrate to 80 °C throughout the spray process ensured the removal of the solvent and film formation of the cluster on the substrate. The as‐deposited bismuth oxido cluster was then hydrolyzed by dipping the substrate into an aqueous sodium hydroxide solution for 60 min. The samples were annealed at 370 °C in a tube furnace for 15 min under argon flow resulting in single‐phase β‐Bi₂O₃ coatings (Figure 1a, β‐Bi₂O₃‐1). Coatings of the monoclinic phase (Figure 1a, α‐Bi₂O₃‐1) are obtained by an annealing protocol using higher temperatures (550 °C for 15 min). Comparison of the powder X‐ray patterns of α‐Bi₂O₃ films resulted in strong deviations of the observed and calculated intensities of the reflections. Further analysis revealed a preferred orientation of the crystallites and applying the model for oriented platelets of Toraya and Marumo resulted in a (010)‐oriented platelets with the values P1=0.98 and P2=0 after fitting the data with PowderCell (Figure 1b).33, 34

Scheme 1.

General concept for the preparation of i) α‐Bi₂O₃ and ii) β‐Bi₂O₃ coatings by ultrasonic spray coating starting with a well‐defined bismuth oxido cluster.

Figure 1.

a) Powder XRD patterns of as prepared thin coatings of α‐Bi₂O₃‐1 and β‐Bi₂O₃‐1 (references: β‐Bi₂O₃, ICDD 00‐027‐0050, red line; α‐Bi₂O₃, ICDD 01‐070‐8243, green line).35, 36 b) i) Measured pattern of α‐Bi₂O₃‐1 (y_obs, black line), the calculated intensities of the reflections for ii) 94.7 wt% α‐Bi₂O₃ with a preferred (010)‐orientation (y_calc1, red line) and for iii) 5.3 wt% Bi₂O₂CO₃ (y_calc2, green line) as well as iv) the sum pattern (y_calc1+y_calc1, blue line).

The Bi₂O₃ coatings show absorption in the visible‐light region (Figure 2a), whereby β‐Bi₂O₃ shows the higher absorption edge at 552 nm. The band gaps determined based on Tauc plots using (αhν)^2/n (n=1) versus hν (Figure 2b) are (2.78±0.03) eV for α‐Bi₂O₃‐1 and (2.34±0.03) eV for β‐Bi₂O₃‐1. The visible light response of the monoclinic coating is in the range of values as reported for α‐Bi₂O₃ in literature (2.8 eV).37 The direct band gap value for the as‐prepared β‐Bi₂O₃ coating is lower than the value for β‐Bi₂O₃ nanoparticles [(2.44±0.03) eV], that were prepared starting from the analogous bismuth oxido cluster,3 and slightly higher than reported for porous β‐Bi₂O₃ (2.27 eV) by Luo et al.38

Figure 2.

a) UV/Vis diffuse reflectance spectra of the as‐prepared samples α‐Bi₂O₃‐1 and β‐Bi₂O₃‐1. b) Tauc plots (for direct band gap transition) calculated from the UV/Vis diffuse reflectance spectra of the pure polymorphs.

As exemplarily shown in Figure 3 for β‐Bi₂O₃‐1, the coatings have a block‐like structure. The breaking‐up of the layers is due to the removal of solvent residuals during the annealing process. The associated volume reduction led to tensions and microcracking in the end, similar to nature, when soils dry out, and well known for TiO₂ films after thermal treatment.39 This effect can also be observed for α‐Bi₂O₃‐1 (Figure S1). Therefore, the glass substrates are not completely covered with Bi₂O₃ as evidenced by EDX mapping. However, microstructuring and its positive impact on the specific surface area can be beneficial for the catalytic properties. The thickness of the coating is (8±4) μm according to the eddy current method using electrically conductive substrates like a silver sheet as well as a molybdenum sheet coated with β‐Bi₂O₃. In comparison, analyzing SEM images (Figure S2) gives (5±1) μm coating thickness on a glass substrate. Exemplarily, the scotch tape adhesion test was performed on sample β‐Bi₂O₃‐1 to get a qualitative idea of the adhesion of the Bi₂O₃ coatings on the substrates. The as prepared coatings are not fully resistant to external forces, since the scotch tape came off with some Bi₂O₃ stuck to it (Figure S3). Nevertheless, the coatings remain adhered to the substrate in stirred solutions without mass loss during photocatalytic testing as shown by differential weighing. In addition, AAS measurements of the remaining test solutions yielded a negligibly small mass of (1.9±0.3) μg Bi in total (Bi₂O₃ in solution after 10.5 h of catalysis: 0.004 % coating mass from β‐Bi₂O₃‐3).

Figure 3.

SEM images of the sample β‐Bi₂O₃‐1 magnified by factors of a) 100×, b) 500× and c) 1000×. Corresponding EDX mapping results are shown for d) Bi (magnification: 589×).

2.2. Photocatalytic Testing and Optimization of the Bi₂O₃ Coatings

For comparison, the photocatalytic activity of the different polymorphs of the Bi₂O₃ coatings was evaluated in a reactor using 35 mL rhodamine B solution (1×10⁻⁵ M) under visible light irradiation (Figure S4). Before illumination, the time for establishing the adsorption‐desorption equilibrium was determined experimentally for the test solution (Figure S5). In addition, a photolysis experiment with a blank glass substrate was performed. The sample α‐Bi₂O₃‐1 shows only a low photodegradation performance. After 600 min, the RhB conversion was 24 % for α‐Bi₂O₃‐1, which is only a slight improvement over photolysis (17 %). The thin coating of β‐Bi₂O₃ (β‐Bi₂O₃‐1) was more active with a conversion of 33 % after 600 min, which corresponds to a reaction rate constant of 0.60 ⋅ 10⁻³ min⁻¹ (Figure S6, Table S1). The higher activity of β‐Bi₂O₃ as photocatalyst in comparison to α‐Bi₂O₃ (Table S1) is in accordance with previous studies investigating decomposition of several pollutants under visible light irradiation using Bi₂O₃ dispersions.18, 40, 41, 42, 43, 44

For the Bi₂O₃ coatings we observed an incomplete coverage of the substrate, which was believed to limit the photocatalytic activity. Therefore, yet another preparation pass was performed using a β‐Bi₂O₃‐1 sample as starting substrate. The deposited [Bi₃₈O₄₅(OMc)₂₄(DMSO)₉] ⋅ 2DMSO ⋅ 7H₂O layers on top of the Bi₂O₃ coating were treated in analogy to the preparation method used before. The annealing process was performed at 370 °C under argon atmosphere to give single‐phase β‐Bi₂O₃ (β‐Bi₂O₃‐2, Figure 4). Moreover, the as‐obtained β‐Bi₂O₃‐2 samples were also used as starting material for a third preparation cycle. The powder XRD patterns of thus prepared β‐Bi₂O₃‐3 show the tetragonal phase, whereby the small shoulder at 2Θ=27.3° (*) might be indicative of the monoclinic phase as minor impurity (Figure 4). Thus it is demonstrated that a step by step deposition of β‐Bi₂O₃ layers is possible. The mass of Bi₂O₃ per unit area on the substrate was determined to be 2.8 mg ⋅ cm⁻² (β‐Bi₂O₃‐1), 5.6 mg ⋅ cm⁻² (β‐Bi₂O₃‐2) and 8.4 mg ⋅ cm⁻² (β‐Bi₂O₃‐3). Simultaneously with the mass of the coatings the thickness increased (β‐Bi₂O₃‐1: (5±1) μm, β‐Bi₂O₃‐2: (11±6) μm, β‐Bi₂O₃‐3: (16±4) μm) due to the additional deposits, which was monitored by SEM (Figure S2). EDX analysis revealed that the films are composed of bismuth and oxygen only and no residuals of sodium from the hydrolysis step were detected (Figure S7).

Figure 4.

Powder XRD patterns of the as‐obtained coatings after performing the preparation cycle once (β‐Bi₂O₃‐1), twice (β‐Bi₂O₃‐2) and thrice (β‐Bi₂O₃‐3) (references: β‐Bi₂O₃, ICDD 00‐027‐0050, red line).35

The morphology of the superimposed coatings (Figure 5) is similar to the observed block‐like structure of β‐Bi₂O₃‐1 (Figure 3). Nevertheless, the positions of the stacked blocks are shifted resulting in a complete coverage of the substrates with Bi₂O₃. An optical distinction between the two‐stacked coatings (β‐Bi₂O₃‐2) and the three‐stacked coatings (β‐Bi₂O₃‐3) does not become obvious. The small channels at the surface are probably a result of the evaporation of residual solvent during the annealing process. UV/Vis diffuse reflectance spectra of the as‐obtained samples do not differ significantly from β‐Bi₂O₃‐1 (Figure S8a). Calculation of the optical band gaps based on Tauc plots give (2.32±0.02) eV for both β‐Bi₂O₃‐2 and β‐Bi₂O₃‐3 (Figure S8b). With regard to the separation efficiency of photo‐generated charges, no significant PL of the coatings is detected (Figure S9), which might be attributed to a low recombination rate of photo‐induced electrons and holes.45, 46

Figure 5.

SEM images of two‐stacked coating β‐Bi₂O₃‐2 [magnification: a) 500×, b) 1000×] and three‐stacked coating β‐Bi₂O₃‐3 [magnification: a) 500×, b) 1000×].

The photocatalytic efficiency of the as‐prepared samples β‐Bi₂O₃‐2 and β‐Bi₂O₃‐3 was evaluated by experiments on the degradation of rhodamine B under visible light irradiation. The visible light spectra of the decolorization of a dye solution is exemplarily given for β‐Bi₂O₃‐3 (Figure 6). Notably, a slight hypsochromic shift becomes obvious, which might be indicative of a stepwise decomposition of N,N,N′,N′‐tetraethylrhodamine (rhodamine B) to give de‐ethylated rhodamine intermediates.47 This process is described as a photosensitized mechanism, whereby excited electrons of the organic molecule are transferred to the conduction band of the semiconductor. It was already shown for β‐Bi₂O₃ nanoparticles that the photodegradation is a result of the interplay of the photosensitized degradation pathway in combination with the direct oxidation of the chromophoric system by photogenerated active species.3, 48, 49 The experiment in aqueous solution was repeated using a cut‐off filter at 550 nm (OG550) to clarify the influence of the photosensitized mechanism. This cut‐off filter nearly completely prevents the direct excitation of the semiconductor, but an excitation of rhodamine B is still possible (λ=554 nm). The photocatalytic performance of β‐Bi₂O₃‐3 is significantly lower (24 % rhodamine B conversion after 600 min) in comparison to the experiment using a cut‐off filter at 420 nm (Figure S10). Nevertheless, the degradation efficiency of the photocatalyst is still higher as compared to the photolysis of rhodamine B. Thus, it is concluded that the photosensitized degradation pathway does apply here indeed but exhibits only a minor contribution to the decolorization of rhodamine B using the as‐prepared β‐Bi₂O₃ coatings.

Figure 6.

Visible light spectra monitoring the photodegradation of rhodamine B (1 ⋅ 10⁻⁵ M) using β‐Bi₂O₃‐3 as photocatalyst (t=−30 min: start of stirring in the dark, t=0 min: start of irradiation with visible light).

Upon additional coating of the film β‐Bi₂O₃‐1, which was prepared by a single preparation pass, the photocatalytic efficiency was significantly increased (Figure 7, Figure S11), however the photocatalytic efficiencies of the films β‐Bi₂O₃‐2 and β‐Bi₂O₃‐3 are similar. Using a cut‐off filter at 420 nm, the rhodamine B conversion after 600 min was approximately 75 % for β‐Bi₂O₃‐2 as well as for β‐Bi₂O₃‐3, which is a significant improvement as compared to β‐Bi₂O₃‐1 (33 %). The enhanced decolorization determined by UV/Vis spectroscopy is in good accordance with the lowered non‐purgeable organic carbon (NPOC). Starting from 3.36 mg ⋅ mL⁻¹ (t=−30 min) a value of 2.42 mg ⋅ mL⁻¹ was achieved in the presence of β‐Bi₂O₃‐1 after 10 h of illumination, which was further lowered with the photocatalyst β‐Bi₂O₃‐3 (NPOC=1.54 mg ⋅ mL⁻¹) proving a efficient oxidation of the dye to CO₂. It is concluded that a complete coverage of the substrate, which is associated with an higher amount of active centers at the surface, seems to be the key factor in order to maximize the photocatalytic efficiency of the Bi₂O₃ coatings, whereas the thickness of the coating seems to be negligible. The as‐prepared sample β‐Bi₂O₃‐2 is favorable with regard to the amount of decolorized dye per irradiation time and per catalyst mass employed (Figure S12, Table S1). The catalytic activity of Bi₂O₃ itself in the sample β‐Bi₂O₃‐3 is reduced to 63 % in comparison to the semiconductor material in the sample β‐Bi₂O₃‐2 due to the higher mass percentage embedded in the bulk material of the coating.

Figure 7.

Time‐dependent photodegradation of an aqueous rhodamine B solution (1 ⋅ 10⁻⁵ M) under visible light irradiation (t>0 min) without catalyst and in the presence of β‐Bi₂O₃‐1, β‐Bi₂O₃‐2 and β‐Bi₂O₃‐3 as photocatalysts.

The XRD pattern of the sample β‐Bi₂O₃‐3 shows the formation of a minor amount of Bi₂O₂CO₃ in addition to β‐Bi₂O₃ after 10 h of visible light irradiation in a RhB solution (Figure 8), which is confirmed by Raman spectroscopy (Figure S13). The reaction of β‐Bi₂O₃ with CO₂, either formed in situ upon release during the photodegradation of organic compounds or as a result of contact with CO_2(aq) under light irradiation, was discussed in previous studies on β‐Bi₂O₃ particles.3, 50, 51, 52, 53 In contrast, XRD studies of the samples α‐Bi₂O₃‐1 and β‐Bi₂O₃‐1 after the degradation experiments do not show any indication of photocorrosion (Figure S14, Figure S15), which is most likely the result of the lower photocatalytic activity (Table S1). Noteworthy, Bi₂O₂CO₃ itself is an efficient photocatalyst,54 which might explain the negligible loss of activity. However, the β‐Bi₂O₃/Bi₂O₂CO₃ mixture can be easily regenerated by annealing at 370 °C for 1 h in a furnace under an argon atmosphere to give single‐phase β‐Bi₂O₃. The recycled sample shows nearly the same photodegradation efficiency as the pristine sample (Figure S16).

Figure 8.

Powder XRD patterns of the sample β‐Bi₂O₃‐3 i) before photocatalysis, ii) after 10 h of rhodamine B degradation under visible light and iii) after regeneration at 370 °C for 1 h under an argon atmosphere (references: β‐Bi₂O₃, ICDD 00‐027‐0050, red line; Bi₂O₂CO₃, ICDD 00‐041‐1488, blue line).35, 55

2.3. Photocatalytic Degradation of Further Pollutants

To probe the coatings for the degradation of other pollutants than rhodamine B, β‐Bi₂O₃‐3 samples were illuminated with visible light in 35 mL of 4 ⋅ 10⁻⁵ M aqueous solutions of ethinyl estradiol and triclosan (Figure 9). The adsorption‐desorption equilibriums are reached throughout a dark phase of 30 min (Figure S17, Figure S18). In the presence of β‐Bi₂O₃‐3 as immobilized photocatalyst, the initial concentrations are reduced to 53 % for ethinyl estradiol and 17 % for triclosan (Figure S19, Figure S20). In case of triclosan over 80 % of the initial pollutant is removed from solution in the presence of β‐Bi₂O₃‐3 including adsorption on surfaces and photolysis, whereas 15 % are decomposed solely during the irradiation time without using a catalyst. It can be summarized, that the photocatalytic conversion of triclosan after 600 min exceeds its photolysis more than three times. Therefore, the immobilized photocatalysts can be a beneficial contribution for supplementary cleaning stages decreasing the concentration of the antimicrobial pollutant in wastewater. This can protect the so called secondary treatment, which makes use of aquatic microorganisms, from damage. However, the NPOC value (5.77 mg ⋅ mL⁻¹) was not decreased, which shows that the biocide is not fully oxidized to CO₂. In case of ethinyl estradiol the NPOC was reduced from 9.61 mg ⋅ mL⁻¹ to 5.00 mg ⋅ mL⁻¹, which indicates partial removal of photooxidation products in the form of CO₂ from water samples.

Figure 9.

Time‐dependent photodegradation of aqueous ethinyl estradiol and triclosan solutions (4 ⋅ 10⁻⁵ M) under visible‐light irradiation (t>0 min) without catalyst and in the presence of β‐Bi₂O₃‐3 as photocatalyst.

3. Conclusions

Thin coatings of α‐Bi₂O₃ and β‐Bi₂O₃ are accessible starting from the well‐defined bismuth oxido cluster [Bi₃₈O₄₅(OMc)₂₄(DMSO)₉] ⋅ 2DMSO ⋅ 7H₂O under mild conditions using ultrasonic spray coating. A modified hydrolysis and annealing protocol adapted from a nanoparticle synthesis was successfully transferred to the preparation route for immobilized Bi₂O₃ on different substrates composed of float glass, silver or molybdenum. Therefore, an implementation of the concept for various other substrate materials seems straightforward. As might be expected a significantly higher photocatalytic activity is observed for coatings composed of tetragonal Bi₂O₃ (β‐Bi₂O₃‐1) as compared to coatings composed of monoclinic Bi₂O₃ (α‐Bi₂O₃‐1). As‐prepared samples by a repeated pass of the coating process (β‐Bi₂O₃‐2, β‐Bi₂O₃‐3) show significantly enhanced activity, whereby the crucial factor is a full coverage of the substrate for generating an higher amount of active sites rather than coating thickness. Thus a degradation efficiency of 74 % of rhodamine B decomposition after 600 min of visible light irradiation was obtained. As expected the value is lower than observed for nanoparticle dispersions used as photocatalyst, but from a technological point of view the approach is promising for applications in sewage treatment plants. Even more important than studies on model compounds are photodegradation experiments using wastewater pollutants like the antimicrobial agent triclosan or ethinyl estradiol with its estrogenic activity. We have demonstrated partial removal of photooxidized ethinyl estradiol as well as rhodamine B in the form of CO₂. The as‐prepared coatings β‐Bi₂O₃‐3 are able to decompose nearly half of the initial concentration of ethinyl estradiol and more than 80 % in case of triclosan under visible light irradiation, which offers potential for further implementation in treatment stages.

Experimental Section

Chemicals

Bismuth(III) nitrate pentahydrate (Co. AlfaAesar) and sodium methacrylate (Co. AlfaAesar) were used as received. [Bi₃₈O₄₅(OMc)₂₄(DMSO)₉] ⋅ 2DMSO ⋅ 7H₂O was prepared according to a literature procedure.32

Spray Coating

Ultrasonic spray coating56 as a cold spray technique57 was carried out with a spray coating setup ExactaCoat Inert (Co. Sono‐Tek) under argon atmosphere. The ultrasonic nozzle can move over a maximum area of 40×40 cm² and was adjusted within a sample‐to‐nozzle distance of 36 mm. The substrate is heated (maximum: 250 °C) during the deposition process (heating plate: 20×20 cm²). A piezoelectric quartz crystal is used to control the nozzle, and vibration with a fixed ultrasonic frequency (120 kHz) is provided. The solution passes (maximum: 19 mL min⁻¹) through an opening diameter of 1 mm and forms a thin liquid film on the nozzle tip. The amplitude of the wave generated by the vibration depends on the electrical power of 1.6 W being adjusted at the piezocrystal. The nozzle head translates with a velocity of 25 mm ⋅ s⁻¹ in 3 mm displaced traces.

Materials Characterization

Powder X‐ray diffraction (XRD) was carried out with a STOE‐STADI‐P diffractometer equipped with a Ge(111)‐monochromator. The X‐ray source was CuK_α1‐radiation (40 kV, 40 mA). Energy dispersive X‐ray spectroscopy (EDX) and scanning electron microscopy (SEM) were performed using a NovaNano SEM (Co. FEI) with the following parameters: pressure (∼10⁻⁵ mbar), work distance (4 to 6 mm) and acceleration voltage (5 to 30 kV) using a Si Drift Detector XFlash 3001 (Co. Bruker AXS). Diffuse reflectance UV‐Vis spectroscopy was performed using a Carry 60 UV‐Vis (Co. Agilent Technologies) equipped with a Barrelino™ (Co. Harrick Scientific Products) remote diffuse reflection probe. The band gap E_g of the semiconductor was estimated according to the equation (αhν)=A ⋅ (hν−E_g)^n/2, where α is the absorption coefficient of the material, hν is the photon energy and A represents a proportionality constant. The parameter n indicates the nature of the fundamental optical transition. A direct transition is associated with a value of $\mathrm{n}=1$ , whereas a semiconductor with an indirect transition is described with $\mathrm{n}=4$ .58 PL spectra of the samples were recorded using a Cary Eclipse (Fa. Varian) with an excitation wavelength of 380 nm. In addition, the fluorescence spectrophotometer was equipped with a 380 nm band‐pass filter between radiation source and sample along with a 400 nm long‐pass filter in front of the detector. The coating thickness measurement was carried out with a FISCHERSCOPE MMS PC2 (Co. Helmut Fischer GmbH) equipped with a plug‐in module PERMASCOPE and a probe FTA3.3‐5.6HF using an eddy current measuring method (DIN EN ISO 2360). The carbon content was determined using a FlashAE 1112 (Co. Thermo Scientific). The photocatalytic dye decomposition was analyzed by in situ UV‐Vis spectroscopy using a Carry 60 UV‐Vis (Co. Agilent Technologies) equipped with fiber optics. A Thermo Scientific iCE 3000 System was used for atomic absorption spectroscopy (AAS). The non‐purgeable organic carbon (NPOC) was quantified in acidified solutions after purging the samples with oxygen by using a multi N/C 3100 TOC analyzer (Co. Analytik Jena). Raman spectra were recorded using a inVia Raman microscope (Co. Renishaw), and the laser excitation of a Nd : YAG (Nd : Y₃Al₅O₁₂) solid state laser (λ=532 nm, P _max=50 mW) was focused on the sample.

Preparation of β‐Bi₂O₃ Coatings Starting from Ultrasonic Spray‐Coated [Bi₃₈O₄₅(OMc)₂₄(DMSO)₉] ⋅ 2DMSO ⋅ 7H₂O

[Bi₃₈O₄₅(OMc)₂₄(DMSO)₉] ⋅ 2DMSO ⋅ 7H₂O is dissolved in tetrahydrofuran (25 g ⋅ L⁻¹). The colorless solution is used to produce thin films of the bismuth oxido cluster on float glass (23×25 mm, thickness: 1000 μm) by ultrasonic spray coating. The film is deposited repeating the spray process thirty times (30 spray cycles) with a flow rate of 0.5 mL ⋅ min⁻¹. The substrate is being completely scanned by the spray head above throughout each spray cycle and is preheated to 80 °C to enable the evaporation of the solvent. In a typical procedure, each bismuth oxido cluster coating is hydrolyzed with 0.025 M aqueous NaOH (50  mL) for 60 min and meanwhile the layer becomes turbid. After drying for 12 h at 40 °C, the sample is annealed in a furnace at 370 °C for 15 min (argon flow rate: 100 L ⋅ h⁻¹) and cooled down under argon atmosphere to give a yellow/orange coating with a mass of (16±4) mg (2.8 mg ⋅ cm⁻²) deposited on the substrate (β‐Bi₂O₃‐1). The deposition of additional films is realized by repeating the spray coating process followed by hydrolysis and the annealing process. The samples β‐Bi₂O₃‐(1+x) (x=0, 1) are used as substrates. Yellow/orange coatings with a mass of (32±5) mg (5.6 mg ⋅ cm⁻²) for the sample β‐Bi₂O₃‐2 and (48±2) mg (8.4 mg ⋅ cm⁻²) for the sample β‐Bi₂O₃‐3 are obtained after performing the preparation pass once more and twice more respectively. The residual carbon content is 0.12 % for β‐Bi₂O₃‐1.

Coatings on a silver sheet (23×25 mm, thickness: 250 μm, purity: 99.9 %) and a molybdenum sheet (23×25 mm, thickness: 100 μm, purity: 99.95 %) were prepared according to the as described procedure in order to perform the coating thickness measurements.

Preparation of α‐Bi₂O₃ Coatings Starting from Ultrasonic Spray‐Coated [Bi₃₈O₄₅(OMc)₂₄(DMSO)₉] ⋅ 2DMSO ⋅ 7H₂O

Starting from [Bi₃₈O₄₅(OMc)₂₄(DMSO)₉] ⋅ 2DMSO ⋅ 7H₂O the spray, hydrolysis and drying protocols are analogous to those for the β‐Bi₂O₃ films. After drying, the sample is annealed in a furnace at 550 °C for 15 min (argon flow rate: 100 L h⁻¹) and cooled down under argon atmosphere to give a yellow coating with a mass of (16±3) mg (2.8 mg ⋅ cm⁻²) deposited on the glass substrates (α‐Bi₂O₃‐1). The residual carbon content is 0.18 %.

Evaluation of Photocatalytic Activity

The photocatalytic activity of the as‐prepared β‐Bi₂O₃ coatings was evaluated in a water cooled glass reactor (T=15 °C, Figure S21) using the coated carrier material and 35 mL of an aqueous solution of 1 ⋅ 10⁻⁵ M rhodamine B (pH=4.7), 4 ⋅ 10⁻⁵ M triclosan (pH=5.2), and 4 ⋅ 10⁻⁵ M ethinyl estradiol (pH=4.8). The reactor is equipped with a 300 W xenon lamp (type Cermax® VQTM ME300BF, Co. Perkin Elmer). A hot mirror filter (λ≤700 nm) is located within a distance of 12.3 cm at one side of the reactor and directly illuminates an area of 4.5 cm². A cut‐off filter (λc (τi=0.50)=(420±6) nm, GG420, Co. Schott) is used to remove the UV light. Before illumination, the solutions are stirred for 30 min in the dark to establish the adsorption‐desorption equilibrium of the dye at the catalyst surface. The progress of photodegredation was studied by UV‐Vis spectroscopy. The illumination process is interrupted by stopping the stirring and darkening the light beam by a cover prior to the UV‐Vis measurement. The interval of measurement is timed every 10 min for the first 30 min, every 15 min between 30 to 120 min, every 30 min between 120 and 180 min and every 60 min to the last measurement after 600 min. The concentration of the pollutant is determined by calculating the area under the UV‐Vis curve from 450–600 nm for RhB, 200–256 nm for triclosan and 235–315 nm for ethinyl estradiol. The photodegradation is plotted as a function of the irradiation time. Following the way of pseudo‐first order reaction kinetics the logarithmic concentration of rhodamine B is plotted versus time to calculate the reaction rate constant k.

Supporting information

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re‐organized for online delivery, but are not copy‐edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

Supplementary

M. Hofmann, L. Rößner, M. Armbrüster, M. Mehring, ChemistryOpen 2020, 9, 277.