Grätzel‐Type TiO2 Anatase Layers as Host for Pt Single Atoms: Highly Efficient and Stable Photocatalytic Hydrogen Production

Single atoms (SAs) represent not only a new frontier in classic heterogeneous catalysis but are also increasingly investigated as co‐catalysts in photocatalytic reactions. In contrast to classic catalysis, many photocatalytic platforms require only a very low SA loading density to reach a saturated photocatalytic activity. As a result, an optimized light harvesting/carrier transport combination in the supporting semiconductor becomes the key aspect for the overall photocatalytic efficiency. In this work, it is demonstrated that Grätzel type mesoporous TiO2 layers represent an ideal host for Pt single‐atoms (SAs) that allow for a highly effective photocatalytic H2 generation. Using a layer with an optimized geometry, structure, as well as Pt SA loading, a photocatalytic H2 production is achieved of up to ≈2900 µL h−1 (under irradiation at λ = 365 nm and I = 65 mW cm−2) – a performance that is far superior to previous Pt SA/TiO2 structures based on TiO2 nanotubes, nanosheets, or metal organic frameworks. Moreover, such SA/substrate combination provides a highly stable H2 production over time. The present work thus introduces the use of this classic TiO2 nanostructure as the most effective host for Pt SAs and its use for highly efficient photocatalytic H2 production from aqueous solutions.


Introduction
Single atom catalysts (SACs), over the last decade, have emerged as a new frontier in virtually any field of catalysis.SACs have found applications in classic thermal catalysis, electrocatalysis DOI: 10.1002/aenm.202302998and, most recently, in photocatalysis.The most researched direction of photocatalysis is water splitting to produce green H 2 as an eco-friendly fuel. [1]In photocatalysis, electron-hole pairs are generated in semiconductors using super-band-gap illumination (hv>E g ).The photogenerated charge carriers can then migrate to the semiconductor surface, where they may trigger redox reactions, such as H 2 evolution (2H + + 2e − → H 2 ).However, charge transfer at the interface is kinetically slow for many semiconductors, [2] i.e., often a co-catalyst is needed to promote the charge transfer and thus the hydrogen evolution reaction (HER).5] However, the use of large amounts of precious metals is economically unfeasible.[8] The overwhelming majority of such SA studies focuses on SA reactivity and underlying principles, as well as on anchoring techniques on different substrates.Moreover, a considerable amount of efforts are dedicated towards achieving a high SA loading, [9][10][11] although recent work has shown that under typical solar illumination conditions (AM 1.5) of anatase TiO 2 , already a very low loading of SAs (≈10 5[ SAs μm −2 ) is sufficient for maximized performance.At this (sufficient loading) point, the H 2 production is limited by the light harvesting and utilization features of the semiconductor substrate. [12]This means that for the overall activity of a photocatalyst, photoelectron generation and harvesting become more important than the SA density.With regard to efficient substrates, some recent work investigated SAs on various light-harvesting structures, such as titania nanopowders, [13] nanotubes, [14,24] nanosheets, [15,49] metal organic framework structures, [16] or hybrid structures. [17]Some of the nanostructures, such as anodic nanotubes [18] and nanosheets [19,20] were produced on a conductive substrate, e.g. on fluorine doped tin oxide (FTO), which provides the advantage of multipurpose applicability in photocatalytic and photoelectrochemical reactions.
In the present work, we demonstrate that classic Grätzel-type mesoporous anatase layers that have been extensively used as electron collecting scaffolds in non-aqueous solutions in dyesensitized solar cells [21,22] can also be utilized as a most effective host for Pt SAs in photocatalytic H 2 generation from aqueous solutions.The layers produced on FTO and then thermally annealed (sintered) can achieve, when optimally decorated with Pt single atoms, up to 2900 μL h −1 H 2 (when tested in 50 vol.% methanol solution under irradiation at  = 365 nm and I = 65 mW cm −2 ) -this is remarkably superior to any other TiO 2 nanostructure reported in previous works. [23,24,25,17]hese SA-based photocatalytic structures combine a large surface area, short hole diffusion path, and excellent electrolyte penetration, [26,27] while their geometry allows for a H 2 production rate that is also remarkably stable over time.
The samples were then loaded with Pt SAs by a reactive deposition approach [28] using H 2 PtCl 6 solutions with different Pt concentrations.Figure 1d,h show a high-resolution top view and cross-sectional scanning electron microsope (SEM) image of a 14 μm thick titania sample (annealed at 550 °C) after Pt was deposited from a 0.05 mM H 2 PtCl 6 solution.Clearly, no Pt deposition in form of nanoparticles is evident from the SEM, as expected for homogeneous SA deposition, while energy-dispersive x-ray diffraction (EDX) analysis of the layer found a Pt content of 0.16 at.%.When the composition is analyzed at different depths from the layer surface, i.e., near the top (i), in the middle (ii), and near the bottom (iii), as indicated in Figure 1f, a very uniform Pt content across the layer is found (Figure 1g).High-resolution SEM images (Figure 1h) taken and at different depths (i-iii) further reveal no Pt agglomerates or nanoparticles formation within the TiO 2 layer.The absence of any SEM visible Pt nanoparticles was further confirmed for all samples produced in this work and at different Pt loadings.These observations are in line with a uniform Pt decoration throughout the entire layer with highly dispersed Pt species.
The presence of Pt SAs in the above sample can be identified by HAADF-STEM.As shown in Figure 2a and Figure S2 (Supporting Information), Pt species deposited on TiO 2 in the form of SAs and few-atom clusters can be identified, and corresponding elemental mapping in the same area shows a uniform Pt distribution.An average SA density of 5 × 10 5 μm −2 was derived from HAADF-STEM images -this finding is well in line with previous reports on reactive deposition of Pt SAs on titania . [12]arying the concentration of H 2 PtCl 6 (0.005 mM, 0.05 mM and 2 mM) during Pt SA deposition results in a Pt content of 0.06, 0.16 and 0.25 at.%, respectively (as determined by EDX in Figure 2b).The fact that strong changes in the precursor concentration only lead to a relatively small increase in the Pt SA loading is in accord with earlier results, that illustrated the availability of only a limited number of attachment sites for Pt SAs on many titania substrates., I.e., at a high precursor concentration of 2 mM a saturation of the Pt loading occurs. [29]Correspondingly, HAADF-STEM (Figure S3, Supporting Information) shows that at the highest Pt loading of 0.25 at.% (produced using 2 mM H 2 PtCl 6 ) more Pt clusters are present along with a higher Pt SA density of ≈ 8 × 10 5 μm −2 .
X-ray photoelectron spectroscopy (XPS) was employed to evaluate the chemical nature of the deposited Pt surface species (Figure 2c).In the Pt4f region, all samples show mainly a doublet peak Pt 4f 7/2 / 4f 5/2 at 72.4 eV / 75.9 eV.This Pt4f peak position is typical for SA Pt anchored on TiO 2 surface [30,31] and corresponds to a formal charge (+) on Pt of approximately  ≈ 2, i.e., to Pt SAs that are two-fold oxygen-coordinated with the TiO 2 surface.Moreover, the coordination of Pt SAs with a TiO 2 surface has been previously investigated by x-ray absorption near edge structure (XANES) and extended X-ray absorption fine structure (EXAFS) measurements [13] (Figure S4, Supporting Information).XANES spectra for Pt SA/TiO 2 samples at Pt L3edge show a white line peak corresponding to positively charged Pt centers (Figure S4a, Supporting Information).Here, a peak shift to higher energies and an unexpectedly high intensity (as compared to PtO 2 ) can be associated with Pt size effects on the electronic structure, as previously reported in. [53]EXAFS peaks at ≈1.6 Å (Figure S4b, Supporting Information) further confirm that Pt SAs are coordinated with oxygen atoms.XPS analysis of the layer composition yields a Pt surface content of 0.15 at.%, 0.42 at.%, and 0.52 at.% after Pt SA deposition using 0.005 mM, 0.05 mM, and 2 mM H 2 PtCl 6 , respectively (Figure 2d).A slight discrepancy between Pt contents evaluated by XPS and EDX can be ascribed to the high surface sensitivity of XPS (≈10 nm analysis depth), as compared to EDX (few μm analysis depth).
The samples prepared under different annealing and Pt SA loading conditions were then evaluated for their photocatalytic H 2 production in 50 vol.%methanol (MeOH) solutions (the effect of MeOH concentration is additionally compared in Figure S5, Supporting Information).Figure 3a shows results for 14 μm thick layers annealed at 350, 450, 550, 650, and 750 °C, and decorated with Pt SAs using 0.05 mM H 2 PtCl 6 .For all layers, the amount of H 2 produced linearly increases with time, and thus, from the slope of the curves the production rates can be determined (Figure 3b).The rates of H 2 generation show an increase with annealing temperature, reaching the highest activity at 550 °C, and subsequently decrease for higher annealing temperatures.This variation in performance can be ascribed to crystallization and sintering of the layers at different temperatures, as will be discussed further below.Based on the annealing results, we then investigated all further samples using an annealing temperature of 550 °C, and varied the TiO 2 layer thickness and the Pt loading (using precursor concentrations of 0.005 mM, 0.05 mM, and 2 mM H 2 PtCl 6 ).The photocatalytic activity for all these layers is shown in Figure 3c.From the data it is evident that for the layers with increasing thickness, the rate of H 2 production increases significantly, and apparently saturates at a thickness of 8-14 μm (Figure 3d).Similarly, increasing Pt SA loading from 0.06 up to 0.16 at.% (as measured by EDX after reactive deposition using 0.005 mM and 0.05 mM H 2 PtCl 6 ) provides a drastic boost to performance, while using higher Pt loading (0.25 at.% after reactive deposition using 2 mM) does not yield a clear benefit.To characterize the specific surface area, we carried out dye adsorption/desorption tests. [32,33]As outlined in detail in the Supporting information (Figure S6, Supporting Information) and in literature, [52] these data can be correlated with Brunauer-Emmett-Teller (BET), and the obtained estimates of surface area were summarized in Table 1-2.Based on these experiments, TiO 2 layers annealed at 450 °C and 550 °C have a similar specific surface area of 84 m 2 g −1 (as confirmed by BET in Figure S7a,b, Supporting Information).The layer annealed at 650 °C has a slightly lower (≈77 m 2 g −1 ), and the sample annealed at 750 °C -a significantly lower specific surface area (≈43 m 2 g −1 ).This decrease of the surface area of the layers at higher annealing temperatures can be attributed to sintering of nanoparticles, as also evident in SEM (Figure S8, Supporting Information), and this also directly explains the poor H 2 production from this sample.On the other hand, low performance after 350 °C annealing is likely due to incomplete removal of the organic binder at low temperature. [34] Light absorption properties of the mesoporous TiO 2 layers of different thicknesses (4-14 μm) were compared in the wavelength range of 300-700 nm (Figure S9, Supporting Information).All studied samples exhibit a nearly identical absorbance at  < 380 nm (Figure S9a, Supporting Information), which corresponds to band-to-band electron excitation in TiO 2 .The opti- cal bandgap of TiO 2 in mesoporous layers estimated from the Tauc plots has a similar value of ≈3.2 eV for all thicknesses (Figure S9b, Supporting Information), in line with bandgap values reported for anatase TiO 2 . [35]Additionally, incident photon to current conversion effeiciency (IPCE) spectra recorded for these samples yield a similar bandgap of ≈3.2 eV (Figure S10, Supporting Information).Based on previous studies, [36,37] a complete absorption in the UV (super band-gap region) is expected for > 5 μm thick porous TiO 2 -however, this thickness may be different in mesoporous structures due to extended light path within the layer as a result of light scattering and refraction.For our structures, only the increase from 4 to 8 μm in TiO 2 layer thickness leads to changes in the absorption; for 14 μm no additional light absorption can be observed over the entire spectrum.Overall, the differences in the optical properties between layers are most prominent in the visible range (at  > 380 nm), which does not contribute to the H 2 evolution reaction.
Based on nearly identical UV light absorption in 4-14 μm thick TiO 2 layers (Figure S9a, Supporting Information), it becomes clear that the superior H 2 evolution performance by thicker TiO 2 layers is not related to the quantity of photons absorbed, but instead to the internal charge carrier distribution.The fact that for the thicker samples (8-14 μm) still an increase in H 2 production occurs, in spite of the limited light penetration depth (i.e., absorption depth), shows that reactive electron transfer sites (Pt SAs) are accessible via electron diffusion (see schematic in Figure 4a).Owing to a high electron mobility and lifetime in anatase, [38,39,40] electron diffusion lengths up to several tens of micrometers can be reached. [41,42]Ultimately, favorable (i.e., more sparse) redistribution of photoexcited carriers among the charge transfer sites within thick TiO 2 layers allows to reduce carrier recombination losses.In support of this theory, intensitymodulated photocurrent spectroscopy (IMPS) measurements (Figure S11, Supporting Information) indicate a relatively fast electron transport within the layers, which is comparable to that in oriented anodic TiO 2 nanotubes and monocrystalline TiO 2 nanoflakes. [43,44,45]he fact that the H 2 production rate increases with Pt SA loading from 0.06 to ≈ 0.16 at.% Pt, but does not increase any further for 0.25 at.%, indicates that for this loading (5 × 10 5 SA μm −2 ), a sufficiently effective surface density of electron transfer sites is established.Based on the H 2 evolution (≈2900 μL h −1 under irradiation at  = 365 nm and I = 65 mW cm −2 ), and assuming a uniform distribution of both Pt SA sites (≈5 × 10 5 μm −2 ) and photogenerated electrons within the 14 μm thickness of the irradiated layer area (≈0.44 cm 2 ), a TOF of ≈0.79 #H 2 site −1 s −1 can be determined.
It is noteworthy that for TiO 2 layers with the lowest Pt SA loading of 0.06 at.% (e.g., after reactive deposition from 0.005 mM H 2 PtCl 6 ), not only an inferior H 2 evolution performance is observed, but also an apparent change in color can be found in the illuminated area of the sample (Figure S12, Supporting Information) which can be ascribed to self-reduction of TiO 2 by accumu- lated photoelectrons via the reaction: Ti 4+ + e − → Ti 3+ (leading to blue-black titania). [46]This does not occur at sufficient Pt SA decoration, that is 0.16 and 0.25 at.%.
A key factor in photocatalytic reactions is the suitable coupling of Pt SAs with the semiconductor -, i.e., an electrical contact needs to be established for successful electron transfer from the substrate to the Pt SA. [47,48] Therefore, the electronic coupling of SAs was evaluated under cathodic polarization in dark (see Linear sweep voltammetry (LSV) measurements in Figure S13, Supporting Information).Evidently, the cathodic current densities increase with Pt loading.Notably, as compared to the pristine TiO 2 layers, even the slightest Pt loading provides a significantly superior electron transfer ability.This is consistent with good electrical connection between SAs and TiO 2 layers for all Pt loadings. [12]oreover, we evaluated the H 2 evolution rates of the Pt SA loaded TiO 2 layers in long-term photocatalytic tests, as shown in Figure 4b.The layers with both moderate and high Pt SA loading (produced using 0.05 mM and 2 mM H 2 PtCl 6 ) maintain the initial performance after the entire experiment in 4 consecutive runs, i.e., they show a remarkable stability of active sites and the TiO 2 layer structure.This may also be attributed to the carrier distribution illustrated in Figure 4a, i.e., the straightforward difference in light penetration depth and electron diffusion length that may partially prevent light induced agglomeration. [25]An excellent long-term stability of HER was also confirmed in 24 h measurements (Figure S14, Supporting Information).SEM characterization after 24 h of photocatalytic tests reveals that some SAs agglomerated into Pt clusters (Figure S15, Supporting Information).Corresponding XRD pattens confirm that the crystallinity of TiO 2 support remains intact (Figure S16, Supporting Information).
The extraordinary performance of Grätzel-type mesoporous structure is particularly apparent, when this structure is compared to other nanostructures that were previously used and reported as hosts for SA Pt in photocatalytic H 2 generation.Figure 5 shows a comparison to nanoflakes, [23] nanoparticles, [24] nanosheets, [25] metal-organic framework-derived TiO 2 , [17] compact layers [25] and nanotubes [24] under similar irradiation conditions and with optimized Pt SA loading in all cases, indicating that mesoporous TiO 2 layers significantly outperform all previously reported TiO 2 structures.This can be ascribed to specific features of the Grätzel-type TiO 2 structures, such as excellent light absorption, charge transport/redistribution (i.e., swift hole transport to the surface and long range electron diffusion), and large surface area available for Pt SA anchoring.

Conclusion
The present work shows that Grätzel-type TiO 2 layers widely used in solar cells represent an outstanding platform for Pt SA cocatalyzed photocatalytic H 2 generation.The mesoporous layer structure provides an excellent light absorption and a very high surface area for Pt SA anchoring.Annealing of such layers at 550 °C shows the highest benefit for performance via preservation of surface area and establishing interparticle connections.The layers can be easily decorated with Pt SA using a simple reactive deposition approach.Surface decoration with Pt SA cocatalytic sites at an optimal density of 5 × 10 5 μm −2 leads to the best performance for layers with thickness of >10 μm.The obtained H 2 production of ≈2900 μL h −1 significantly outperforms previously reported structures.We assign the key beneficial factor in terms of performance and stability to charge carrier redistribution, i.e., photoelectron reaching SA centers even in light shaded areas of the photocatalyst.

Experimental Section
Preparation of Mesoporous TiO 2 Layers: Titania nanoparticle paste (Ti-Nanoxide T/SP, particle size 15-20 nm) was purchased from Solaronix.The paste was coated via doctor blade technique on the transparent conducting oxide glass (FTO, fluorine doped tin oxide layer over glass, 8 Ω sq −1 , Solaronix).The films were then annealed at different temperatures for 1 hour.After sintering, the TiO 2 mesoporous films were obtained.
Pt Single Atom (SA) Deposition: For Pt SA deposition, 10 mL of 0.005, 0.05 and 2 mM H 2 PtCl 6 (Metakem) solution in aqueous MeOH (50 vol %) solution was prepared in a quartz cell.The solution was purged with argon for 15 min and the prepared TiO 2 mesoporous film was immersed in the solution and sealed.Then, the solution was kept under stirring in dark for 1 h.After 1 h the TiO 2 film was rinsed with DI water and dried in a N 2 stream.
Structural Characterization: The crystallographic phase of mesoporous TiO 2 film was identified by X-ray diffraction (XRD, X'pert Philips MPD with a Panalytical X'celerator detector) using graphite monochromatized Cu K radiation (wavelength 1.5406 Å).The morphology and elemental composition of the samples were studied by scanning electron microscopy (SEM, Hitachi S-4800) and energy dispersive X-ray spectroscopy (EDX).High-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) and EDX mapping were carried out by a high-resolution transmission electron microscope (Spectra 200, Thermo Fisher Scientific).X-ray photoelectron spectroscopy (XPS, PHI 5600, US) was used to analyze the chemical composition of the samples.Ti2p peak (located at 458.5 eV) was used to calibrate the binding energy.
Brunauer-Emmett-Teller (BET) Measurement: The specific surface area (S BET ) of the TiO 2 layers annealed at 550 °С and peeled off the substrates was evaluated based on Brunauer-Emmett-Teller theory for N 2 isotherm (with respect to Rouquerol criteria). [50]N 2 adsorption/desorption isotherm was measured at 77 K on a volumetric gas adsorption analyzer (Autosorb iQ XR, Anton-Paar Quanta Tec, USA) up to 0.965.Prior to analysis, the sample was degassed under high vacuum (10 −7 Pa) at 200 °C for 12 hours, while high purity (99.999%)N 2 and He gases were used for the measurements.

Dye Adsorption/Desorption Tests:
The absolute surface area of differently prepared TiO 2 layers was compared based on dye adsorption/desorption tests. [32,33]The samples were immersed in 0.3 × 10 −3 M Ruthenium based dye: cis-bis(isothiocyanato) bis(2,2-bipyridyl 4,4-dicarboxylato) ruthenium(II) bistetrabutylammonium (D-719, Eversolar, Taiwan, the same as the commonly called N719 dye) solution in acetonitrile and tert-butanol (1:1) and kept in dark for 24 hours at room temperature.The dye-sensitized samples were then rinsed with acetonitrile to remove the non-chemisorbed dye molecules, and dried under nitrogen stream.Afterward, the samples were immersed in 5 mL of 10 mM KOH solution for 24 hours.The concentration of desorbed dye was measured using UV-vis spectrophotometer (Lambda XLS+, PerkinElmer) and calculated using the Beer-Lambert Law equation: where A was the absorbance at 515 nm,  was the molar extinction coefficient ( was 14100 L mol −1 cm −1 at 515 nm), and l was the path length in the solution.Then the molar amount of dye absorbed per unit area (cm 2 ) of the samples (M) was calculated as follow: where V was the volume of the KOH solution.
Intensity-Modulated Photocurrent Spectroscopy (IMPS): Intensitymodulated photocurrent spectroscopy (IMPS) was performed using Zahner IM6 electrochemical workstation.IMPS studies were performed in a three-electrode setup with TiO 2 layers as a working electrode, Pt as a counter electrode, and Ag/AgCl as a reference electrode.As an electrolyte, air-saturated 0.1 M Na 2 SO 4 aqueous solution was used.During the measurements, a constant potential of 0.5 V (versus Ag/AgCl) was applied to the working electrode.Simultaneously, it was subjected to UV LED illumination ( = 358 nm) with periodically oscillating light intensity (with 10% amplitude from the average light intensity).By changing the oscillations frequency in the range 10 −1 -10 3 Hz, Nyquist plots of real versus imaginary photocurrent conversion efficiency (H) were recorded.The charge transport time constant ( c ) was further evaluated as follows. [51] c = (2f min ) −1 (3) where f min was the modulation frequency at the lowest point of the IMPS plots.
Linear Sweep Voltammetry (LSV): The electrochemical properties were investigated using an AutoLab PGSTAT 302N potentionstat-galvanostat (Metrohm) with a three-electrode system.The TiO 2 samples were used as working electrodes, Ag/AgCl and Pt as the reference and counter electrode, respectively.The surface area of the working electrode (immersed in an aqueous 0.1 M Na 2 SO 4 solution) was 0.5 cm 2 Prior to the measurement, the electrolyte was purged with argon for 30 minutes.
Photocatalytic Hydrogen Evolution: Photocatalytic H 2 evolution was studied under UV irradiation in the presence of MeOH as a hole scavenger.TiO 2 mesoporous film was immersed in 10 mL 50 vol.% MeOH solution in a quartz cell, and purged with Ar gas for 15 min and sealed.Then the TiO 2 mesoporous film was irradiated with UV radiation ( = 365 nm, I = 65 mW cm −2 ) and H 2 producedwas determined in different intervals of time using a gas chromatograph (GCMS-QO2010SE, SHIMADZU) with a thermal conductivity detector.

Figure 1 .
Figure 1.Characterization of the Pt SA decorated (using 0.05 mM H 2 PtCl 6 ) mesoporous TiO 2 layers: a-c) Cross-sectional SEM images of the layers with different thicknesses, d) SEM image of the surface, e) XRD patterns after annealing at different temperatures, g) EDX analysis of Pt content and h) corresponding SEM characterization of the surface and at different layer depths i-iii (as indicated in (f)).

Figure 2 .
Figure 2. a) HAADF-STEM image of the Pt SA decorated (using 0.05 mM H 2 PtCl 6 ) TiO 2 layer (14 μm thick, annealed at 550 °C) with highlighted SAs and clusters, and corresponding EDX elemental mapping, b) Pt content measured by SEM-EDX, c) XPS spectra in the Pt 4f region for the Pt SA decorated (using 0.005 mM, 0.05 mM and 2 mM H 2 PtCl 6 solutions) TiO 2 layers, and d) Pt content measured by XPS.

Figure 3 .
Figure 3. Photocatalytic H 2 evolution performance by Pt SA-decorated mesoporous TiO 2 layers as a function of a,b) the annealing temperature (using 14 μm layer thickness and 0.05 mM H 2 PtCl 6 concentration in each case), and c,d) TiO 2 layer thickness and H 2 PtCl 6 concentration (using annealing temperature of 550 °C in each case).

Figure 4 .
Figure 4. a) Schematics of light harvesting and charge transport within the TiO 2 layer.Here, light penetration is limited to a certain depth.However, photogenerated electrons can diffuse within the layer and reach the available Pt SA electron transfer sites.b) Cycling stability of the photocatalytic H 2 evolution performance by Pt SA-decorated (using 0.05 and 2 mM H 2 PtCl 6 concentrations) 14 μm thick TiO 2 layers.

Figure 5 .
Figure 5.Comparison of photocatalytic H 2 production by various TiO 2 nanostructures with optimized Pt SA loading irradiated for 3 h ( = 365 nm, power density 65 mW cm −2 ) in 50 vol% MeOH aqueous solutions.The red bar corresponds to Pt SA-loaded (0.05 mM) 14 μm TiO 2 (annealed at 550 °C) from this work.

Table 1 .
Dye adsorption (per 1 cm 2 sample area) for 14 μm thick TiO 2 layers annealed at different temperatures and corresponding absolute surface area estimation.