Monitoring the Structure Evolution of Titanium Oxide Photocatalysts: From the Molecular Form via the Amorphous State to the Crystalline Phase

Abstract Amorphous TixOy with high surface area has attracted significant interest as photocatalyst with higher activity in ultraviolet (UV) light‐induced water splitting applications compared to commercial nanocrystalline TiO2. Under photocatalytic operation conditions, the structure of the molecular titanium alkoxide precursor rearranges upon hydrolysis and leads to higher connectivity of the structure‐building units. Structurally ordered domains with sizes smaller than 7 Å form larger aggregates. The experimental scattering data can be explained best with a structure model consisting of an anatase‐like core and a distorted shell. Upon exposure to UV light, the white TixOy suspension turns dark corresponding to the reduction of Ti4+ to Ti3+ as confirmed by electron energy loss spectroscopy (EELS). Heat‐induced crystallisation was followed by in situ temperature‐dependent total scattering experiments. First, ordering in the Ti−O environment takes place upon to 350 °C. Above this temperature, the distorted anatase core starts to grow but the structure obtained at 400 °C is still not fully ordered.


Experimental section
Synthesis procedure The adopted synthesis method, 1 so called 'direct injection', is a simple synthesis procedure to obtain high surface area titanium oxide (TixOy) photocatalysts involving simultaneous test for photocatalytic water splitting. The method involves direct injection of titanium(IV) ethoxide [Ti(OEt)4] precursor into the water-methanol reaction mixture in a photocatalytic reaction cell equipped with a lamp to provide radiation in ultraviolet (UV) range. With this method, synthesised catalyst is exposed to UV light for 2 h mimicking direct injection synthesis method. The synthesis product in suspension is referred to as h-UV-TixOy. To have a non-UV counterpart, samples were also synthesised via 'ex situ' synthesis method, this time in the absence of UV light. The non-UV synthesis product in suspension is referred to as h-TixOy. The two synthesis routes are explained below in detail. The samples resulting from both of following synthesis routes were collected for stationary suspension measurements in sealed capillaries. Protective atmosphere was maintained in the case of air sensitive (suspension subjected to UV irradiation) samples. In order to have dry powder samples, the sediments were separated from the suspensions by decanting off the liquid part. The remaining parts were dried overnight (for 3 h in case of powders obtained as counterparts to suspensions measured at the synchrotron source due to limited time) at 60 °C in an oven to obtain d-UV-TixOy and d-TixOy samples.
Synthesis procedure under UV light exposure (h-UV-TixOy): Synthesis of h-UV-TixOy samples involves injection of 525 μL Ti(OEt)4 (99.98%, Sigma Aldrich, CAS: 3087-36-3) as metal oxide precursor into a mixture of 180 mL deionized water and 20 mL MeOH (Sigma Aldrich, CAS: 67-56-1) in a glass photocatalytic reaction vessel. This concentration was calculated aiming 1 g⋅L -1 TiO2 catalyst in the end. The reaction vessel is equipped with a Peschl Ultraviolet TQ150 150 W middle pressure Hglamp having peak emission at 366 nm for UV light exposure. The UV lamp is inserted to the reaction vessel with a water cooled jacket. Following the injection of the precursor into the reaction solution, the UV lamp is turned on. During illumination, the reaction suspension was stirred using a stirring bar and kept under continuous Ar purging (50 mL⋅min -1 ). To be consistent with the previous photocatalytic study, the irradiation period was kept at 2 h.

Synthesis procedure without UV light exposure (h-TixOy):
To observe the effects of UV light exposure, the synthesis was also carried out in the absence of UV light. The same amount of the precursor was injected into the reaction mixture prepared in the same way. In this case, samples were synthesised in a beaker rather than the reaction vessel. The reaction mixture was stirred for 2 h under continuous Ar purging. In this way h-TixOy 'non-UV' samples were obtained which are interesting for us to understand the nature of the amorphous TixOy material independent of the effects induced by UV light exposure. These samples together with the dried version, d-TixOy, are studied extensively including the in situ temperature-dependent experiments in the present work.
Experimental parameters for the total scattering experiments Room temperature PDFs: Data collection for the powder samples, suspensions and the liquid precursor were performed at I15-1 beamline at Diamond Light Source (Diamond). The energy was 77 keV (λ= 0.16167 Å), the sample to detector distance was 200 mm (Qmin = 0.2 Å -1 and Qmax, instrumental = 38 Å -1 ), and the beam size on the sample was 700 x 150 m. Data acquisition was performed with a Perkin Elmer XRD 4343 CT detector. Collected scattering data were integrated using the DAWN software package. 2 Scattering data obtained from an empty capillary and a capillary filled with the water-methanol mixture were used as backgrounds for powder and suspension measurements. The program PDFgetX3 3 implemented in the xPDFsuite 4 was used for processing PDFs from the integrated scattering data (qmax= 22 Å -1 ). DebyePDFCalculator within Diffpy-CMI 5 (qmax= 22 Å -1 , ADPs= 0.003, delta2= 1.0, Qdamp= 0.0258 Å -1 , Qbroad= 0.0118 Å -1 ) was used to simulate PDFs for non-periodic models.
Characterisation TEM imaging was performed using a Titan Themis instrument (Thermo Fisher Scientific) operated at 300 kV with a Cs corrector for the image forming lens. EELS data were acquired in a probe corrected Titan Themis instrument (Thermo Fisher Scientific) operated at 300 kV using a Gatan Quantum ERS energy filter with the entrance aperture collecting electrons up to 35 mrad. Multivariate statistical analysis was performed on the datasets to highlight the different spatial contributions of the Ti L2,3 edges. 7 High resolution TEM images shown in the Supporting Information were taken with a Hitachi HF-2000 instrument. Specific surface area was determined by nitrogen sorption experiments with a Quantachrome NOVA 3200e instrument after degassing approximately 150 mg powder samples at 150 °C overnight. Data were evaluated by the BET (Brunauer-Emmett-Teller) method using the adsorption data in the relative pressure range of 0.05 to 0.2. From the specific surface area, approximate particle sizes can be calculated using the relation Dp= 6000·σ -1 ·As -1 (assuming spherical particles, Dp = particle size in nm, σ = specific density in g·cm -3 , As = specific surface area in m 2 ·g -1 ). Dynamic light scattering (DLS) data were recorded on a Malvern Zetasizer Nano-ZS using laser radiation with a wavelength of 633 nm. The scattered light was measured at a backscattering angle of 173°. The suspension collected after synthesis was ultrasonicated for 30 min. The results of this collection was compared to the ones collected from the middle portion of the suspension let settle for 3 h. Thermogravimetry (TG) and differential scanning calorimetry (DSC) measurements were carried out using a NETZSCH STA 449 F3 Jupiter thermal analyser for the qualitative analysis of crystallisation temperatures. The measurements were carried out under an air flow of 40 mL⋅min −1 using approximately 10 mg powder heated in an aluminium oxide crucible with heating rates of 2, 5 and 10 °C⋅min −1 . Fourier transform infrared (FTIR) spectroscopy measurements were performed using a Perkin Elmer Spectrum Two spectrometer with an attenuated total reflectance (ATR) unit. For construction of the clusters Avogadro molecular editor 8 and for visualisation of the structures Diamond software 9 is used.

Figure S9
Nitrogen adsorption/desorption isotherms of d-TixOy sample after degassing approximately 160 mg powder at 150 °C overnight (15.9 wt% loss). The isotherm is of type Ib according to the IUPAC definition. Separation of the processes of supermicropore filling and monolayer formation on the outer surface is difficult for such isotherms. Therefore, the common B.E.T. method was not used for determination of the specific surface area of the solid. The specific surface area was determined using non-local density functional theory (NLDFT) method with the model for nitrogen adsorption at 77 K on silica (assuming cylindrical pore) implemented in Quantachrome NovaWin software package. Figure S10 STEM images and EELS spectrum images acquired at d-UV-TixOy and d-TixOy in the areas marked with yellow and purple boxes. The pixel size was 2.7 nm to prevent beam damage. Left shows the colour code for the Ti-L3 position. The spatial variation within each sample is smaller than between them. The spectra given in Figure 6b are averaged over the area shown in yellow and purple.

Figure S15
Refinement of the PDF obtained from d-TixOy at 400 °C using anatase structure. 12 The calculated PDF is plotted in red colour on top of the experimental PDF given in black. Note that the difference (calculated PDF intensity subtracted from that of the experimental PDF) curve is given in green colour with an offset below.

Table S1
Refined parameters for the PDF obtained from d-TixOy at 400 °C using anatase-TiO2 structure (qmax= 22 Å -1 , qdamp= 0.0375 Å -1 , qbroad= 4.067*10 -6 Å -1 , sratio= 1.0, Rw= 0.277266) Figure S16 High-resolution transmission electron micrographs obtained after in situ temperature dependent total scattering experiments at different magnifications (a,b) and additional electron diffraction patterns (c). Sample was cooled from 400 °C to room temperature. Figure S18 DSC and TG data collected under air flow at heating rates of 2, 5 and 10 °C⋅min -1 (represented by blue, red and black curves respectively) for d-TixOy. The temperature values corresponding to crystallisation peaks are used to calculate the activation energy for the crystallisation process. 15 Respective Kissinger plot is given as an inset.