Preparation of TiO2–based hollow microspheres by spray drying and their use as novel active pigments for photocatalytic coatings

The use of TiO2 nanoparticles in organic-based photocatalytic coatings imposes several challenges: poor activity under visible light, binder photo-degradation, need for UV activation and toxicity concerns. Here we present a scalable two-step synthesis route to prepare TiO2-based hollow microspheres (HoS) as alternative photocatalyst to commercial TiO2 nanoparticles. The hollow microspheres of TiO2 or WO3-doped TiO2 (3 mol% WO3) are synthesized via template-assisted spray-drying followed by calcination. The structure and composition of the powders are characterized and their photocatalytic performance is assessed using methylene blue photodegradation under UV irradiation as model reaction. XRD analysis reveals the presence of anatase and TiO2(B) phases, indicating the heterostructured nature of the samples. The results of the dye photo-degradation tests confirm the photocatalytic functionality of the TiO2-based HoS. Moreover, the introduction of WO3 (TiO2/ WO3 HoS) leads to an enhancement of the performance, approaching that of commercial (Aeroxide P25, ~21 nm) nanoparticles. The most active TiO2/WO3 HoS are incorporated into an acrylic formulation and the resulting coatings tested towards pollutant abatement under UV light. A coating containing P25 nanoparticles undergoes an activation process where binder degradation leads to increased TiO2 exposure and a rise in catalytic activity but possibly at the expense of coating stability. By contrast TiO2/WO3 HoS-acrylic coatings exhibit catalytic activity similar to the initial activity of P25 containing coatings, but does not cause the same partial binder


Introduction
The rising concern about environmental pollution and its adverse effects on human health has encouraged the development of sustainable technologies for the removal of pollutants in e.g. urban environments [1]. Photocatalytic oxidation processes make use of inorganic semiconductors to create highly oxidizing species that promote the degradation of pollutants, and photocatalytic processes have been put into practice for the efficient abatement of contaminants in air-and aqueousphase systems [2][3][4][5][6][7][8]. Titanium dioxide, in anatase or rutile form [9], is extensively used as photocatalyst due to its high oxidation capability and chemical stability [10]. The irradiation of TiO 2 with UV light (wavelength, λ < 390 nm) causes the promotion of electrons (e − ) from the valence band (VB) of the semiconductor to the conduction band (CB), resulting in the creation of a positive hole (h + ) in the valence band [11]. The electrons and holes stimulate the generation of highly oxidizing species such as hydroxyl radicals and superoxide radicals (*OH and *O 2 − , respectively), from reaction with oxygen or water.
These species then promote the subsequent degradation of compounds in contact with the photo-catalyst [12]. TiO 2 nanoparticles (<100 nm) are the preferred form of the photocatalysts because they provide large active surface area and exhibit higher efficiency due to a lower electron/ hole pair recombination as the result of a shorter diffusion pathway for the charge carriers [13]. Most practical applications of photocatalytic materials demand the immobilization of the active phase. An example is the formulation of the TiO 2 nanoparticles into an organic-based coating that can be applied to surfaces and render these active towards photo-degradation of pollutants such as volatile organic compounds (VOC), nitrogen oxide gases (NO x ) and bacteria [14][15][16][17][18][19][20][21][22][23]. However, there are challenges associated with such incorporation of the photocatalytic material in an organic matrix. The film-forming agent (binder) and other components of the coating formulation hinder the access of light and reactants to the TiO 2 surface resulting in an inefficient utilization of the photons and lead to lower performance compared to the photocatalyst powder on its own [24]. Additionally, the photocatalyst will also cause photodegradation of the organic components of the film [25][26][27][28]. In fact, such initial photo-degradation has been shown to be the mechanism of activation of photocatalytic coatings as the binder degradation allows light and reactants to access the photocatalyst and thereby renders the coating surface photocatalytically active [17,29].
The high surface area of nanosized TiO 2 results in a large photocatalyst-binder interfacial area, and the photo-degradation of the large amount of binder in contact with the catalytic pigments may cause gradual breakdown of the coating structure and compromise its stability. Also, TiO 2 is a suspected human carcinogen, particularly in the case of nanoparticles [30]. Hence, it is relevant to evaluate alternatives to nanosized TiO 2 for use in photocatalytic coatings.
An alternative to TiO 2 nanoparticles is engineering of TiO 2 -structures that combine high surface area, short diffusion pathways for charge carriers and micrometric size to avoid the adverse health effects of nanoparticles. An example could be the synthesis of hollow TiO 2 microspheres using spray-drying technology. Spray-drying allows the synthesis of hollow microspheres via template-assisted procedure by combining a metal oxide precursor and an organic compound as sacrificial template [31,32]. The spray-dried structures are mesoporous after subsequent calcination and have a surface area adequate for catalytic applications, and the hollow morphology ensures improved light harvesting as the nm-scale thickness of the sphere walls limit the internal scattering resulting in superior performance compared to massive particles [33].
Previous studies [34][35][36] have explored the production of TiO 2based hollow microspheres by spray-drying and their use for photocatalytic applications in powder form. However, to the best of our knowledge, a comprehensive study of their use as alternative material to traditional TiO 2 nanoparticles upon integration into coating structures has not been reported. In this work, we present a simple, cost-effective, and scalable methodology for the production of TiO 2 -based hollow microspheres (HoS) using spray-drying technology, namely the calcination of spray-dried titanyl sulfate/citric acid composite spheres. This allows production of hollow particles with high surface area and a mesoporous structure. The structure and composition of the synthesized materials are elucidated using different characterization techniques, and the photocatalytic performance is evaluated by performing photodegradation of a probe pollutant (methylene blue) under UV light as a model reaction. Finally, the optimized photocatalysts are incorporated into acrylic-based coatings and their photocatalytic functionality within a coating is assessed.
In a typical synthesis of TiO 2 HoS, 20 g of titanyl sulfate (19 mmol) and 5 g of citric acid (24 mmol) were dissolved in 200 mL of deionized water (citric acid/metal molar ratio: 1.25). For the preparation of TiO 2 / WO 3 HoS an additional 0.15 g of ammonium metatungstate hydrate (0.6 mmol) was incorporated into the solution (WO 3 content 3 mol%). The solution was stirred overnight to ensure the formation of the Ti(IV)citrate complexes. The presence of sulfuric acid in the titanium precursor causes a substantial drop of the pH to ~1, which makes the solution unsuitable for spray-drying. Therefore, ammonia solution was added drop-wise until achieving neutralization. In further experiments the citric acid/metal salt molar ratio was varied from 0.30 to 2.50 by increasing the concentration of citric acid in the solution, whereas the concentration of the other reagents was maintained.
Spray-drying of freshly prepared solutions was carried out using a Buchi B-290 mini spray dryer. The temperature inlet was 200 • C and the pump feed 10% (~3 mL/min). Air was used as the drying gas with an aspirator rate of 100%. A two-fluid nozzle (0.5 mm core diameter) was used at a pressure drop of 1.35 bar (rotameter height 60 mm).
After collection of the powder product (metal salt/template solid particles), an amount (0.5 to 2 g) of the resulting spray-dried powder was placed in a ceramic crucible and calcined in a muffle furnace (Nabertherm LT 9/11) with air flow (600 L/h) at 500 • C (ramp 1 • C/ min) for 5 h. The effect of the calcination settings on the structure and performance of the resulting photocatalysts was evaluated by varying the calcination temperature from 500 • C to 800 • C and the ramping rate from 1 • C/min to 10 • C/min. As standard procedure, the obtained powders were used as prepared for powder characterization and coating formulation. Only in the case of calcination of large batches of powder (>10 g), an acid wash with 2 M sulfuric acid was performed to remove traces of unreacted precursor. For this purpose, the samples were first sonicated in 85 mL of acid, following by washing with deionized water (twice) and finally with ultrapure water (Milli-Q) using centrifugation (11,000 rpm for 5 min). The washed powders were dried overnight at 100 • C. This treatment was introduced as calcination of large batches involved thicker powder layers, whereby oxygen supply to the entire sample is hindered and carbon deposition from citric acid decomposition becomes a risk. As result, the acid-washed powders featured comparable degree of purity to small batches after calcination (see Fig. S7), and the two types of samples are therefore used indistinctly in the present work.
Aeroxide® P25 TiO 2 nanoparticles (~21 nm and mixture of ~80% anatase and ~20% rutile, according to the supplier) were obtained from Sigma-Aldrich and used without further purification as reference photocatalytic material.

Transmission electron microscopy (TEM)
TEM analyses of the powder samples were carried out using a Jeol JEM-2100 electron microscope with an LaB6 electron gun operating at an accelerating voltage of 200 kV and filament current of ~100 μA. The samples were supported on formvar/carbon-coated copper grids by application of an aliquot of a dilute slurry in hexane, following by evaporation of the solvent in the air.

Scanning electron microscopy (SEM) and energy dispersive X-ray spectroscopy (EDS)
SEM analyses were performed on a Thermo Scientific Prima E microscope operated at an accelerated voltage of 30 kV. Representative SEM images were acquired with the XT Microscopy Server software using secondary-and backscattered-electron imaging modes. The powder samples were mounted on aluminum stubs (Agar Scientific) using adhesive carbon tape. EDS microanalysis was carried out using a Thermo Scientific UltraDry detector and processed with Pathfinder software. Novatouch LX 2 apparatus at 77 K. Prior to the measurement, samples were dried/degassed under vacuum at 350 • C for 4 h. The specific surface area (S BET ) was calculated by the Brunauer-Emmett-Teller (BET) method in the relative pressure range (P/P 0 ) between 0.05 and 0.3. The total pore volume (V p ) was calculated from the amount of adsorbed nitrogen at the relative pressure (P/P 0 ) = 0.99. The Barrett-Joyner-Halenda (BJH) pore size distribution was obtained from the desorption branch of the N 2 adsorption isotherm for calculation of the volume and surface area of mesopores (2-50 Å pore diameter).

X-ray diffraction (XRD)
The powder diffraction patterns of the prepared samples and P25 nanoparticles were acquired with a Huber G670 powder diffractometer in the 2θ range 3 to 100 • in steps of 0.005 • using Cu-Kα1 radiation (λ = 1.54056 Å) and a scanning time of 10 min. The data was collected in transmission mode from a rotating flat plate sample inclined 45 • relative to the primary beam. The powders were supported on gold ring holders and covered with cellulose-based Scotch tape, which does not cause sharp peaks in the diffraction pattern. The XRD profiles were analyzed using the Match! software for powder diffraction, allowing background subtraction, phase identification and semi-quantification through Rietveld refinement.

Particle size analysis
The evaluation of the particle size of the synthesized hollow microspheres was achieved by means of laser diffraction measurements using a Malvern Mastersizer 3000 equipped with a Hydro SV wet dispersion unit.

Thermogravimetric analysis (TGA)
TGA measurements of precursor powder (metal salt/template) and calcined hollow microspheres were obtained with a Netzsch STA 449 F1 Jupiter apparatus. The samples were heated from room temperature to 900 • C with a ramp of 10 • C/min under N 2 flow. Once the temperature reached 900 • C, the gas flow was changed to 30 vol% O 2 for 10 min at the same temperature.

UV-visible (UV-vis) spectroscopy
The UV-vis spectra of the synthesized samples and P25 nanoparticles were acquired with a Cary 60 UV-Vis spectrophotometer within the range of 200-800 nm. For this purpose, the powders were pelletized by first grinding a mixture of sample and KCl (volume ratio 1:3) in a mortar. Subsequently, a small portion of the powder mixture was loaded in a pellet ring holder and subjected to 10 tons pressure using a manual hydraulic press. The resulting 1 cm diameter pellets were inserted between two portions of UV-transparent tape for optical analysis. The band gap, E g , was estimated using the Tauc relationship (Eq. (1)) [38]: Here α is the absorption coefficient, hν is the energy of photon, A is a characteristic constant and n is a factor related to the nature of the transition, being 2 and 0.5 for indirect and direct allowed transitions, respectively.

Formulation of photocatalytic coatings
Photocatalytic acrylic coatings containing P25 or TiO 2 /WO 3 hollow microspheres (HoS) were prepared by dispersing ~0.4 g of powder photocatalyst in ~8 g of a 1:1 (w/w) mixture of a base acrylic coating and o-xylene. The acrylic coating was first produced using a high-speed disperser (Dispermat) by mixing proprietary raw materials: acrylic resin solution, plasticizer, rheological agents, dispersive agents, color pigments, extender and solvents. Surface-treated rutile (coated in the manufacturing process with silica and alumina to prevent chalking), and calcium carbonate were chosen as white pigment and extender, respectively. The base coating formulation was adjusted to reproduce a flat finish coating for indoor application and it was further diluted with o-xylene in order to favor the ultrasonic dispersion of the hollow microspheres into the coating by lowering the viscosity of the system. The TiO 2 -based HoS/base coating/o-xylene mixture was homogenized using a Hielscher ultrasonic disperser (200 W, 26 kHz), which generated ultrasonic pulses of 100% amplitude and 50% duration (0.5 s sonication, 0.5 s rest) using a titanium sonotrode (7 mm diameter) immersed ~1 cm in the liquid. The sonication treatment was continued for 5 min while the mixture was kept in a water bath to minimize heating and solvent evaporation. Following this, the fineness of the grind of the prepared coatings was measured with a Hegman gauge (0-100 μm range), revealing the absence of agglomerates larger than ~20 μm. Moreover, the digital microscopy analysis of model coatings containing only TiO 2based HoS and acrylic binder allows confirming that the hollow microspheres are not broken up as result of the ultrasonic dispersion process (Fig. S14).
The prepared coatings, containing 5 wt% photocatalyst and 20 vol% of solid content, were applied on glass microslides (Elka, 76 × 26 × 1 mm) using a Baker applicator (300 μm wet-film thickness, WFT). Prior to use, the glass microslides were sequentially cleaned with soapy water, ethanol and deionized water, flushed with compressed air and finally dried at room temperature. After application, the coatings were allowed to dry with ventilation in the darkness for 24 h until further use. Following this procedure, and according to the solid content, coatings with ~40 μm dry film thickness (DFT) were obtained.

Lab-scale photo-reactor
The photocatalytic activity of the photocatalytic powders and coatings were carried in a custom-made lab-scale photo-reactor described elsewhere [29]. The main body of the setup consists on an aluminum cabinet (400 × 400 × 510 mm). An Osram Ultra-Vitalux 300 W UV mercury vapor lamp served as UV irradiation source, emitting mainly UVA (315-400 nm) and UVB (280-315 nm) in a lesser extent (13.6 W and 3.0 W, respectively), while the remaining power produces visible light and infrared radiation. The UV irradiance 15 cm from the light source, corresponding to the surface of the irradiated solution, was measured with an RS Pro UV AB tester, resulting in 11 mW/cm 2 (about twice the magnitude of the UV component of sunlight, ~5 mW/cm 2 ), being a fraction of it attenuated when passing through the aqueous media. Alternatively, a Philips EcoClassic 140 W halogen lamp with negligible UV contribution was used as visible light source. The visible light irradiance (>390 nm) was 12.5 mW/cm 2 , measured with an RS Pro Solar Power Meter, while the UV irradiance was 13 μW/cm 2 (~0.1% of the total irradiance). A double-walled glass cell connected to a thermostatic water circulation bath (Brinkmann Lauda Ecoline RE104) set at 20 • C is used as reaction vessel. All the photodegradation experiments were conducted at 20 ± 1 • C.

Methylene blue photo-degradation experiments
The evaluation of the photocatalytic activity was asserted by performing photo-degradation of methylene blue as model pollutant, as done in our previous research work [29]. Briefly, the photo-degradation experiments were carried out using 5 mg/L methylene blue aqueous solutions. In a typical experiment with powder photocatalysts, 100 mg of powder were dispersed in 150 mL of 5 mg/L methylene blue solution (667 mg TiO 2 /L). After stirring in the dark for 30 min at 500 rpm, the lamp was turned on and ~ 1 mL aliquots of the powder-dye solution slurry were taken from the reaction vessel every 2-15 min (UV light tests), or every 30 min (halogen light tests). The aliquots were filtered through nylon syringe membranes (Ahlstrom-Munksjö, 0.2 μm pore size) and the concentration of remaining methylene blue in solution was estimated by measuring 500 μL in an Agilent Cary 60 UV-Vis spectrophotometer calibrated at 665 nm. The evaluation of the photocatalytic activity of the coatings was performed under UV light similarly to the powder experiments. For this purpose, the coated glass slides were immersed in 150 mL of methylene blue solution inside the reaction vessel and placed on a 25 mm high glass plate used as sample holder. The samples were located ~1 cm below the liquid surface. The dye solution was stirred in the dark for 30 min and then UV-irradiated for 180 min. 500 μL aliquots were taken every 5-30 min to monitor the photo-degradation process spectrophotometrically.

Water contact angle (WCA) -Digital microscopy analysis
A Keyence VHX-6000 digital microscope was used for the visual inspection of the dry films, as well as for performing sessile WCA measurements on coatings subjected to UV exposure, as done in our previous publication [29]. For the latter purpose, the coatings, applied on glass slides, were exposed to UV light in the same glass cell used for the photodegradation tests, but immersing them in Milli-Q water instead of dye solution. After a given exposure time, the sample was taken from the photo-reactor and gently flushed with compressed air until dryness. The WCA was measured by placing a 10 μL water droplet on the coating surface, followed by acquisition of the corresponding side-view picture with the microscope rotated 90 o . The droplet profile was fitted to an ellipse using ImageJ software. The results are shown as the average of three individual measurements carried out in different, evenly distributed spots on the coatings. After completion of the WCA measurement, the sample was returned to the photo-reactor for continuation of the UV exposure.

Scanning electron microscopy (SEM) analysis of acrylic coatings
As-prepared and UV-exposed photocatalytic coatings were characterized using SEM. To this end, the coatings were applied on flexible plastic film to obtain free films after drying and detachment from the plastic substrate. The free films were exposed to UV light by fastening to glass microslides with standard adhesive tape. The as-prepared and UVirradiated photocatalytic coatings were mounted on aluminum stubs (Agar Scientific) with adhesive carbon tape and covered with an aluminum grid (Agar Scientific) to minimize charging effects, resulting in a 2 × 2 mm area of sample available for analysis. SEM images were acquired mainly in backscattered-electron imaging mode at low accelerating voltage (1-1.5 kV) in order to reduce possible beam damage and charging. ImageJ software was employed for digital treatment of the collected SEM images with the purpose of quantifying the porosity of the coatings. Quantification was carried out by converting the SEM images into binary pictures using a threshold filter for selection of the pore area. The apparent porosity was expressed as percentage of 2D pore area (%) by dividing the measured pore area by the total area of the analyzed SEM image.

Fourier-transform infrared (FTIR) spectroscopy
FTIR analysis of the as prepared and UV-irradiated photocatalytic coatings was carried out using a Thermo Scientific Nicolet iS5 FTIR spectrophotometer. Samples were scanned in the frequency range of 400-4000 cm − 1 after collection of the air spectrum as background. Data acquisition was carried out using the Thermo Scientific OMNIC software. Characteristic spectral absorption bands were normalized using the alkyl C -H stretching band appearing at ~2950 cm − 1 for comparative studies. The FTIR spectra of selected raw materials used in the coatings formulation and non-photocatalytic acrylic coating were acquired in order to facilitate the interpretation of the absorption bands.

CIELAB color analysis
The color of the photocatalytic coatings before and after UV exposure for 180 min was evaluated using an X-Rite Ci6x spectrophotometer with a measuring circular area of 50 mm 2 (8 mm diameter). To this end, the coatings were exposed to UV light using the same UV lamp and photoreactor described in Section 2.4 immersed in Milli-Q water, as done for the WCA analysis (Section 2.6). Results were expressed in terms of the CIELAB color coordinates (L*, a*, b*) and their corresponding variations (ΔL*, Δa*, Δb*) as well as the total color change (ΔE*). Measurements were carried out under the standard illuminant D65 and the 10 o standard observer. Fig. 1a and b display TEM images of TiO 2 and TiO 2 /WO 3 HoS prepared by spray drying and subsequent calcination at 500 • C. The figure illustrates the formation of microsized spherical particles and the hollow structure is evident for some of the particles. Laser diffraction analysis (Fig. S1) shows a relatively narrow particle size distribution centered at 1-2 μm, indicating that the spray-drying process is capable of producing hollow microspheres of a reasonably uniform size. Fig. S2 shows the effect of calcination temperature on the morphology of the TiO 2 HoS and illustrates that higher calcination temperature (700-800 • C) leads to a rougher surface and possibly a more compact structure. Fig. 1c and d show selected backscattered electron BE-SEM images of TiO 2 and TiO 2 /WO 3 HoS for 500 • C calcination. The results illustrate the hollow spherical nature of the synthesized microsized particles, in agreement with TEM data. Moreover, the BE-SEM analysis of multiple particles reveals the presence of dark areas within most of the microspheres (Fig. S3a), corroborating that the hollow morphology is representative of the spray-dried and calcined powder. This finding is further supported by the detection of broken microspheres that reveal their internal cavity (Fig. S4). The SEM examination at higher magnification (Fig. S3b) also reveals that the hollow microspheres display a corrugated external surface. Overall, the microscopy analysis indicates that the main synthesis product are single-shell, hollow microspheres, although some massive particles are also present in the powder. No evidences of hollow multishelled (HoMS) structures are found, as typically obtained with sequential templating synthesis using carbonaceous spheres as sacrificial template [39].

Structural characterization of TiO 2 and TiO 2 /WO 3 hollow microspheres (HoS)
EDS analysis was carried out for the TiO 2 /WO 3 HoS sample. The EDS spectrum and the elemental mapping (Fig. S5) corroborates the widespread and homogeneous distribution of tungsten into the TiO 2 -based microspheres. Moreover, the semi-quantitative EDS analysis is in good agreement with the intended composition (97 mol% Ti, 3 mol% W). The EDS profile analysis of a single microsphere (Fig. S6) confirms the presence of Ti and W in intimate contact and also provides qualitative proof of the hollow structure due to the profile depletion at the center of the particle.
Thermogravimmetric analysis (TGA) of the pre-calcined precursor powder (TiOSO 4 /citric acid (CA) composite particles) and TiO 2 -based HoS was performed in order to monitor the calcination process and the fraction of unreacted precursor in the samples after calcination at 500 • C (Fig. S7). The onset of the thermal decomposition of the composite precursor is observed at ~200 • C, most likely due to the degradation of ammonium citrate and ammonium sulfate [40]. A second mass loss process begins at ~470 • C, which could be attributed to the decomposition of TiOSO 4 to anatase [41]. Fig. S7 also shows the TGA curves for TiO 2 -based HoS and the effect on the acid-washing step on the content of unreacted precursor. Overall, the powders display a minimal fraction of unreacted precursor (~3 wt%) after calcination at 500 • C, which persists even after acid-wash. Nonetheless, the interference of these unreacted species on the photocatalytic activity is expected to be negligible, given their high chemical resistance. Unreacted precursor species are therefore expected to act as purely as spectators under reaction conditions. Fig. 2a shows the N 2 -physisorption isotherms for TiO 2 and TiO 2 /WO 3 HoS, compared to P25. The broad hysteresis loops displayed by the synthesized powders (type IV isotherm [42]) confirm the mesoporous character of the hollow microspheres, whereas the curve shape for P25 (type II isotherm [42]) reveals the low fraction of mesopores in the nanoparticles. The BET areas of the TiO 2 and TiO 2 /WO 3 HoS (141 and 147 m 2 /g, respectively) are also substantially higher than for P25 (56 m 2 /g), which potentially makes the HoS appropriate for surface mediated reactions such as catalytic applications. Fig. 2b displays the pore size distribution curves for TiO 2 , TiO 2 /WO 3 HoS and illustrates that the synthesized HoS powders have their pores in the size range of 4-8 nm in the mesoporous range. The influence of the calcination settings and the template/metal ratio on the textural properties of the hollow microspheres was investigated for the TiO 2 HoS. The outcome of this study is presented in Figs. S8-S10. The calcination temperature has the most noticeable effect (Figs. S8-S9), leading to a decrease of the BET area and the pore volume by more than one order of magnitude when increasing from 500 • C to 700 • C. This result is in agreement with the TEM illustrations in Fig. S1, as the substantial decline in surface area at 700 • C could be attributed to excessive roughening of the microspheres surface due to crystal growth. Moreover, a fast temperature ramp rate also leads to a 50% decrease of the surface area (Fig. S10a), which is consistent with the precursor decomposition being shifted towards higher temperature. This may cause a more rapid thermal degradation of the TiO 2 -precursor/template, resulting in collapse of the hollow particle structure [31]. A 50% drop in surface area is also seen when the citric acid/TiOSO 4 molar ratio was increased to 2.5 (Fig. S10b). This is also attributed to a greater tendency towards collapse of the hollow structures, possibly because the greater template concentration causes greater disordering of the metal species. This illustrates the role of template/metal ratio for the synthesis protocol.
Consequently, the screening of the spray-drying/calcination conditions identified the following optimal parameters for the synthesis of TiO 2 -based HoS with high specific surface area: i) CA/Ti ratio = 1.25, ii) calcination at 500 • C and iii) temperature ramp rate of 1 • C/min. These settings were subsequently adopted for the synthesis of TiO 2 /WO 3 HoS. Fig. 3 shows the XRD patterns of TiO 2 and TiO 2 /WO 3 HoS as well as P25 nanoparticles. The commercial P25 nanopowder displays characteristic diffraction peaks attributable to anatase and rutile, with a ratio 87:13, according to the semi-quantitative Rietveld refinement, while the crystal size of anatase, based on the (101) diffraction peak (2θ = 25.4 degree), is ca. 25 nm (Table S1). These findings are in agreement with the information provided by the supplier and with previous structural analysis reported in the literature, which suggests the presence of anatase and rutile as individual nanoparticles and a small fraction forming an anatase/rutile heterojunction structure [43,44].
The TiO 2 HoS and TiO 2 /WO 3 HoS with 500 • C calcination exhibit near-identical XRD profiles having contributions of anatase and a TiO 2 (B) phase in an approximate 90:10 ratio (Table S1). The presence of TiO 2 (B) is revealed after close examination of the XRD patterns at 2θ = 26-36 degree (Fig. S11). TiO 2 (B) has a monoclinic structure [45] and has been previously identified as a constituent phase of TiO 2 HoS prepared using template-assisted hydrothermal synthesis [46]. By contrast, the XRD profile of the TiO 2 HoS sample calcined at 600 • C is predominantly attributable to anatase as a single phase, suggesting that transformation from TiO 2 (B) to anatase occurs in the 500-600 • C range (note that a small peak at 2θ = 27.5 degree is observed in Fig. S11, which could be tentatively assigned to rutile (110) signal). In addition to the phase changes, an increasing calcination temperature also leads to an increased crystallite size (Table S1) and a loss of surface area (Fig. S9) most likely associated with partial collapse of the hollow sphere structure. The presence of WO 3 was not detected in the XRD pattern of the TiO 2 /WO 3 sample, indicating that W is either in the form of an amorphous WO 3 phase or that W is integrated into the anatase or TiO 2 (B) structures in the form of a solid solution.
A comparative analysis of the opto-electronic behavior of the synthesized samples was carried out in order to evaluate the effect of WO 3 in the TiO 2 /WO 3 HoS specimen. For TiO 2 /WO 3 HoS the corresponding Tauc plot (Fig. S12) is shifted ca. 0.1 eV downwards compared to the pure TiO 2 HoS, which is consistent with a lowering of the bandgap caused by incorporation of WO 3 [47]. The lower band gap of TiO 2 /WO 3 HoS should enable a larger utilization of the solar spectrum and enhanced photocatalytic performance under visible light.

Photocatalytic tests with powder samples
The photocatalytic activity of the synthesized TiO 2 -based HoS was evaluated by methylene blue photo-degradation experiments with aqueous slurries of catalyst powder under UV irradiation. Fig. 4a displays the normalized concentration of methylene blue versus irradiation time for P25, as well as for TiO 2 and TiO 2 /WO 3 HoS (500 • C calcination). Compared with P25 and TiO 2 HoS, TiO 2 /WO 3 HoS exhibits lower C/C 0 value after establishing adsorption equilibrium in the dark for 30 min, which suggests that the introduction of WO 3 sites improves the dye adsorption capacity. Upon illumination the time to complete discoloration decreases in the order TiO 2 HoS ≫ TiO 2 /WO 3 HoS > P25. The reaction kinetics were fitted to a pseudo-first order reaction using the relation between time and reactant concentration given by Eq. (2) [4,48,49]: Here k represents the rate constant (in min − 1 ), C/C 0 is the relative reactant concentration and t is the irradiation time (in min). The values of k are 0.49 ± 0.07 min − 1 , 0.19 ± 0.03 min − 1 and 0.06 ± 0.01 min − 1 for P25, TiO 2 /WO 3 HoS and TiO 2 HoS, respectively. Interestingly, the rate constant for the reaction on TiO 2 HoS is almost an order of magnitude lower compared to P25. Nevertheless, the incorporation of WO 3 has a substantial positive effect on the photocatalytic behavior and enables a three-fold increase of the rate constant compared to the unmodified TiO 2 HoS. There may be several factors contributing to the lower activity of  TiO 2 HoS compared to P25. One factor is that although the TiO 2 HoS expose a larger surface area than P25 much of this surface is in the form of mesoporosity and not all of this surface will be illuminated. A smaller photo-activated external surface may well be part of the reason for the lower activity of TiO 2 HoS. Another factor may be the dissimilar phase compositions. Anatase, rutile and TiO 2 (B) display comparable band gaps (3.2 eV for anatase and TiO 2 (B), 3.0 eV for rutile) [50]. However, theoretical calculations show that the energy levels of the valence band and conduction bands of TiO 2 (B) are higher than for anatase, resulting in hole migration from the valence band of anatase to TiO 2 (B), as well as electron transfer from the conduction band of TiO 2 (B) to anatase [50]. These factors could also impact the photocatalytic properties of the TiO 2 HoS negatively given the TiO 2 (B) content. The promotional effect achieved with TiO 2 /WO 3 HoS compared to TiO 2 HoS is noteworthy. Here, an important factor is most likely the smaller band gap of TiO 2 /WO 3 HoS evident from the Tauc analysis (Fig. S12b, Table S2). A smaller band gap will lead to greater utilization of the visible part of the incoming light. The effect is further supported by experiments carried out under visible light. Fig. S13 shows the analogue dye photo-degradation tests under visible light (halogen lamp), which reveal the superior performance of TiO 2 /WO 3 HoS, surpassing both TiO 2 HoS and P25 nanoparticles. The faster photodegradation kinetics of TiO 2 /WO 3 HoS compared to the two pure TiO 2 photocatalysts may be explained considering i) the reduction of the bandgap, ii) the introduction of trapping centers, which minimize the recombination of charge carriers; and iii) the improved light absorption ability of HoS-based structures (due to enhanced light scattering) [33]. The combination of these features results in an increase of the photocatalytic efficiency under visible light. The improved photocatalytic capability of TiO 2 /WO 3 HoS encourages their use as functional pigment for the formulation of photocatalytic coatings in the following stage of the research. Fig. 5a shows the results of the methylene blue photo-degradation tests with acrylic coatings containing 5 wt% P25 or TiO 2 /WO 3 HoS (P25-acrylic and TiO 2 /WO 3 -acrylic, respectively). In agreement with the similar results for slurries of these two materials (Fig. 4), a similar methylene blue bleaching is observed with both coatings, achieving 70-80% photo-degradation after 180 min of irradiation. This indicates that the hollow microspheres represent a suitable alternative to commercial nanoparticles in formulated coatings. Previous work [29] has shown that only a limited part of the contained TiO 2 surface is accessible in a coating system, and this may partly negate the differences in external surface area between P25 nanoparticles and the HoS. Fig. 5b presents semi-logarithmic representation of normalized dye concentration vs. time for both photocatalytic coatings, where the slope according to Eq. (2) will correspond to the rate constant, k. The P25-acrylic coating shows a transition between two linear regimes occurring after ca. 90 min. Here the rate constant increases from k 1,P25 = (5.2±0.1)×10 − 3 min − 1 to k 2,P25 = (9.3±0.1)×10 − 3 min − 1 . This behavior has been explained [29] in terms of an activation of the coating when binder photo-degradation leads to pore development and resulting breakthrough in exposure of additional TiO 2 surface, which raises the catalytic activity. By contrast, the kinetics of the dye photo-degradation on TiO 2 /WO 3 HoS-acrylic is described by a single linear behavior, with a kinetic rate constant of k HoS = (5.7 ± 0.2) × 10 − 3 min − 1 .

Photocatalytic tests with acrylic coatings
The absence of a break in the reaction rate for TiO 2 /WO 3 HoS-acrylic indicates that the coating containing hollow microspheres does not undergo the same activation process where binder degradation exposes additional catalytic surface. This behavior could be advantageous, as the binder degradation that creates porosity for the nanoparticle containing coating may eventually compromise the structural integrity of the coating. The exposure of additional TiO 2 surface is evident from watercontact angle (WCA) measurements because the breakthrough of additional TiO 2 surface creates a gradual drop in WCA [29]. Fig. 6 shows the WCA vs. time curves for both the P25-acrylic and the TiO 2 /WO 3 HoSacrylic coatings. Fig. 6 reveals that the WCA for P25-acrylic remains invariable during the first ca. 60 min of UV exposure, followed by the onset of a decrease in WCA. The drop in WCA is attributable to the exposure of a TiO 2 -rich surface after partial binder photo-erosion and arises as the high energy OH groups created by illumination of the TiO 2 surface interact strongly with water and cause the decline in WCA [29]. By comparison, the WCA for TiO 2 /WO 3 HoS-acrylic shows no significant change in the course of the UV exposure test, which is consistent with the constant dye photo-degradation kinetics and the absence of a breakthrough of a TiO 2 rich surface. These results illustrate that the micrometer-sized TiO 2 /WO 3 HoS cause far less binder photodegradation than TiO 2 nanoparticles, which is attributable to the far smaller binder/HoS interface of micrometer-sized particles. The main cause for the dissimilar evolution of the WCA for coatings containing nanoparticles or TiO 2 /WO 3 HoS must be the lower exposure of TiO 2 surface for the HoS containing coating due to comparatively less binder photo-degradation. However, the surface properties of the HoS, such as the higher surface acidity of the WO 3 -doped photocatalyst [51,52], could also play a role.
The kinetic studies suggest that the initial exposure of accessible photo-catalytic surface within the coating is similar for both HoS and P25 containing coatings, which yield similar catalytic properties (k HoS / k 1,P25 = 62%). In fact the initial performance of HoS and P25 are more similar in coating systems than in slurry experiments. This may arise from a higher utilization of specific surface area with hollow microspheres compared to nanoparticles, most likely due to the less favored interaction of the film-forming agent with the mesoporous surface of TiO 2 /WO 3 HoS. This assumption is tentatively supported by the comparable oil absorption values of P25 and TiO 2 /WO 3 HoS (33 and 39 g oil/100 g, Table S3).
The most striking difference is that the nanoparticle containing coating undergoes an activation phenomenon that raises photocatalyst accessibility and rate constant significantly (78%), expectedly at the expense of a weakening of the coating structure due to binder degradation. While a fully activated coating containing TiO 2 nanoparticles exhibits a higher catalytic activity, the micrometer-sized HoS may yield less coating damage due to reduced binder degradation.

Effect of UV exposure on physical/chemical properties of TiO 2 /WO 3 HoS-acrylic coating
The different tendencies towards photo-erosion of the binder indicated by the kinetics and WCA results may thus suggest lower UV impact on the structure of the TiO 2 /WO 3 HoS-acrylic coating compared to a nanoparticle-loaded analogue. The surface features of the coatings subjected to UV exposure were also examined with SEM. Fig. 7 presents the backscattered-electron (BE) SEM images of the TiO 2 /WO 3 HoSacrylic coatings before and after UV treatment for 180 min, while the analogue comparison for the P25-acrylic coatings has been described in our preceding publication [29]. The BE-SEM imaging analysis reveals substantial differences between the two systems. As reported previously [29], the UV irradiation of P25-acrylic induces pore formation, evidenced by the appearance of dark spots in BE mode that represents voids near the surface, as result of the degradation of the film [53]. It is inferred that such alteration of the coating morphology leads to major exposure of photocatalytically active TiO 2 sites and is responsible of the promotional effect observed in dye photo-degradation, as well as the improved hydrophilicity upon UV exposure assessed in WCA analysis. In contrast, the UV exposure of the TiO 2 /WO 3 HoS-acrylic coating causes less pore development. The analysis focused on different areas of the film (Fig. S15) confirms that the trend derived from Fig. 7 is representative of the coating. The fraction of pore 2D area, obtained from digital treatment of the BE-SEM images, shows minimum variation after UV exposure, remaining below ~2% (Fig. S16a). For comparison, the P25acrylic coating (Fig. S16a) exhibits a larger increase of pore 2D area (~15%) as previously reported, underlining the major effect of the photocatalytic process on the binder stability when TiO 2 nanoparticles are incorporated as pigments. On the other hand, the BE-SEM analysis of the TiO 2 /WO 3 HoS-acrylic coating also reveals a significant increase of white spots after UV irradiation, which can mainly be attributed to TiO 2 /WO 3 HoS due to their microsize and spherical shape (Figs. 7 and S15). The magnitude of the variation after UV exposure is shown in Fig. S16b, resulting in an increase of ~50%. In light of this finding, it is reasonable to assume that the photocatalytic degradation of the binder in the TiO 2 /WO 3 HoS-acrylic coating is primarily limited to the immediate environment of the hollow microspheres, enabling exposure of active sites to some extent without causing pore formation, in contrast to the nanoparticle-containing coating. Therefore, this behavior leads to less deterioration by UV irradiation of HoMS-based photocatalytic coatings.
The structural integrity of the photocatalytic coatings was also visually assessed by digital microscopy. Fig. S17 presents representative images of P25-acrylic and TiO 2 /WO 3 HoS-acrylic coatings as obtained and after UV exposure for 180 min. P25-acrylic exhibits more prominent changes as a result of the UV irradiation, evidenced by the major exposure of TiO 2 pigment (white areas in the image) due to extensive binder photo-degradation. In contrast, minor changes are perceived for the TiO 2 /WO 3 HoS-acrylic coating. On the other hand, the color evaluation of the coatings after UV exposure (Table S4) reveals that both coatings undergo comparable color change after irradiation, mainly due to some yellowing (Δb* ≈ +0.5). It should be noted that the as-prepared TiO 2 /WO 3 HoS-acrylic coating already exhibits a slight yellow coloration compared to the as-prepared P25-acrylic (b* values of 1.57 and 0.14, respectively), which can be caused by the different optical properties of the TiO 2 /WO 3 HoS pigments compared to P25 as result of the introduction of WO 3 .
Furthermore, the chemical change to the coatings resulting from UV exposure was evaluated by means of FTIR spectroscopy. The FTIR results of the as-prepared and UV-treated P25-acrylic and TiO 2 /WO 3 HoSacrylic coatings, as well as selected raw materials and nonphotocatalytic acrylic coatings for comparison, are compiled in Figs. S18-S20. The characteristic vibration modes exclusively attributed to the acrylic binder (peaks between 1300 and 900 cm − 1 , Fig. S18) were used as descriptor of the photo-degradation of the film-forming agent by evaluation of the IR absorbance decrease after UV treatment [27]. The decline in the bands arising from the acrylic binder is significantly higher for the P25-acrylic coating (~25% based on the absorption band at ~1146 cm − 1 , Fig. S20) while TiO 2 /WO 3 HoS-acrylic shows negligible variation, similarly to the behavior exhibited by the acrylic coating without photocatalytic pigment. Consequently, the FTIR evaluation verifies the lower impact of the UV irradiation after 180 min on the physical/chemical properties of TiO 2 -based HoS-acrylic coating compared to the nanoparticle-containing sample, in line with the WCA, digital microscopy and SEM findings.

Conclusions
In this work, the synthesis of TiO 2 -based hollow microspheres using template-assisted spray-drying has been successfully accomplished as novel approach to prepare functional pigments for photocatalytic coatings formulation aimed to indoor architectural use. This methodology combines easily scaled-up production and precise control of the composition and the textural properties of the resulting nanostructures by direct refinement of the synthesis parameters. The hollow microsized pigments offer a lower inherent toxicity compared to nano-sized pigments. The optimal photocatalyst, TiO 2 /WO 3 HoS, displays high surface area and adequate mesoporous structure for the adsorption and conversion of reactants, leading to photocatalytic activity comparable to commercial P25 nanoparticles, as asserted by the dye photo-degradation experiments performed under UV irradiation. In the early stage of the photo-degradation tests (<90 min), the TiO 2 /WO 3 HoS-acrylic coating shows comparable kinetics to that of the P25-based coating. Thereafter (>90 min), the latter exhibits a UV activation step that leads to faster reaction rate, but at expenses of partial binder photo-degradation. In contrast, this process does not take place in the case of the HoS-based coating in the time frame of the UV exposure experiments. Thus, the TiO 2 /WO 3 HoS-acrylic coating undergoes less binder damage induced by UV irradiation and can be regarded as a more relevant photocatalytic system for applications in real scenarios, as the activation step will be hindered in the absence of intense UV light sources. The morphology of the hollow microspheres minimizes the interaction with the binder, which lessens binder degradation.
Overall, the excellent photocatalytic properties of the prepared hollow microspheres via a simple, cost-effective and continuousproduction process; and the sustainability implications of using microsized pigments as alternative to more harmful nanoparticles pave the way for the development of a new generation of nanostructured materials that can provide enhanced photocatalytic functionality to organicbased coatings. Future work within this topic should address stability analysis, by performing accelerated UV aging tests, and explore strategies to enhance the performance under visible light (i.a. doping and synthesis of hollow multishelled (HoMS) particles). Moreover, it should be verified that the catalytic performance established here for a model reaction (dye bleaching) also translates into activity for air purification reactions (i.e. VOC and NO x abatement) and for coating self-cleaning ability under realistic illumination (e.g. sunlight).