Reactive adsorption of toxic organophosphates parathion methyl and DMMP on nanostructured Ti/Ce oxides and their composites

2.1 Sample synthesis

The pure Ti and Ce oxides and their composites were prepared by the modified homogeneous hydrolysis of the metal salts with urea as the precipitating agent. Homogeneous hydrolysis was proven to be feasible for the production of various metal oxides and hydroxides with the well-developed nanostructure and considerably high surface area, and was described in detail elsewhere (Stengl et al., 2010; Subrt et al., 2006). In the original procedure, titanium oxysulphate (TiOSO₄) was dissolved in 2 L of hot distilled water acidified with H₂SO₄, and then diluted to 4 L and 300 g of urea was added. The heating of the solution leads to thermal decomposition of urea and a gradual pH increase in the solution. After 6 h of heating, a white precipitate of TiO₂ was formed. However, in our case, employing this synthesis led to undesired precipitation of Ce(SO₄)₂ instead of oxides; thus, this procedure cannot be used.

Therefore, in the case of TiO₂/CeO₂ composites the original method was modified to remove sulphate ions in synthesis as follows: Appropriate amount of TiOSO₄ (30 g for pure TiO₂) was dissolved in 2 L of hot water acidified with H₂SO₄. Then 100 mL of NH₄OH solution was continuously added to the solution while vigorously stirred, and the white precipitate was formed. This precipitate was washed with water several times to remove sulphate ions and subsequently dissolved by addition of 30 mL of nitric acid. Then a corresponding amount of Ce(NO₃)₃·6H₂O was dissolved in the solution. This mixture was diluted to a final volume (4 L), and 300 g of urea was added. Heating of the solution at 85 °C for 6 h leads to a formation of the precipitate, which was washed with water by decantation, filtered, dried, and annealed at 500 °C. The amount of starting TiOSO₄ and Ce(NO₃)₃·6H₂O was in Ti:Ce molar ratios 1:0 (1Ti0Ce_U), 8:2 (8Ti2Ce_U), 1:1 (1Ti1Ce_U), 2:8 (2Ti8Ce_U), and 0:1 (0Ti1Ce_U).

2.2 Characterization methods

Diffraction patterns were collected using diffractometer Bruker D2 equipped with a conventional X-ray tube (Cu Kα radiation, 30 kV, 10 mA) and the LYNXEYE 1-dimensional detector. The XRD data were processed using DiffracPlus Topas and Eva software (Bruker AXS).

FEI Nova NanoSEM 450 scanning electron microscope (SEM) was used to examine the morphology of the samples. Typically, the powdered sample was dispersed in ethanol and treated in an ultrasonic bath for 10 min before deposition on the silicon wafer chip.

FEI Talos F200X High-resolution transmission electron microscope (HRTEM) was used for structure and morphology analyses. The microscope was equipped with Super-X EDS system with four silicon drift detectors (SDDs). The samples were deposited on the 300 mesh regular copper grid.

The surface area (BET) and porosity (BJH) of samples were determined using nitrogen physisorption method with a Coulter SA3100 (Beckman) instrument.

The Raman spectra were acquired with DXR Raman microscope (Thermo Scientific). 32 two-second scans were accumulated with laser 532 nm (6 mW) under the 10× objective of Olympus microscope.

Infrared spectra were recorded on Thermo Nicolet NEXUS 670 FTIR spectrometer in region 4000–500 cm⁻¹ with the Praying Mantis (Harrick) diffuse reflection (DR) accessory.

X-ray photoelectron spectroscopy (XPS) measurements were performed using a SPECS PHIBOS 100 hemispherical analyzer with a 5-channel detector and a SPECS XR50 X-ray source equipped with an Al and Mg dual anode. The calibration of the binding energy (BE) was performed using the C 1s main component set to 285 eV. Survey spectra were recorded with pass energy of 40 eV, and the high-resolution spectra were recorded as the sum of 10 acquisitions with a pass energy of 10 eV. A Shirley background profile was used for data processing in CasaXPS software.

Concentrations of parathion methyl and 4-NP were determined by the LaChrom HPLC system (Merck/Hitachi). The system consisted of an L-7100 pump, L-7400 variable wavelength UV/Vis detector operating at 230 nm, Rheodyne 7725i injection valve with 20 μL sampling loop, and the Luna column (Phenomenex, Torrance, CA, USA) 150 mm × 4.6 mm, packed with PFP stationary phase, 5 μm. The mixture methanol (HPLC-grade, Labscan, Dublin, Ireland)/water 80/20 (v/v), at a flow rate 1 mL min⁻¹ was used as mobile phase.

2.3 Degradation test for parathion methyl

The testing procedure was described in detail in several previous studies (Janos et al., 2014; Stengl et al., 2011). The constant amounts of the sorbent (50 mg) were dispersed in 400 μL of heptane in a series of glass vials. Then a pesticide solution (100 μL, 10.000 mg L⁻¹) in heptane was added to vials (=1 mg of the pesticide/50 mg of sorbent). The vials were sealed and covered to protect the reaction mixture from sunlight. The reaction was terminated by an addition of propan-2-ol (2 mL) at pre-determined time intervals (0.5, 8, 16, 32, 64, 96, and 128 min). The sorbent was separated by centrifugation (4000 rpm for 7 min) and the supernatant was decanted and transferred to a 50 mL volumetric flask. The sorbent was re-dispersed in 4 mL of methanol and centrifuged again. The extraction with methanol was repeated three times. All the supernatants were combined and analysed immediately by liquid chromatography (HPLC).

2.4 Reactive adsorption of DMMP

The behaviour of the liquid DMMP on the surfaces of prepared samples at ambient temperature was observed by FTIR diffuse reflectance spectroscopy (DRIFTS) using Thermo Nicolet NEXUS 670 FTIR spectrometer. In a typical experiment, the powder sample (50 mg) was placed in a DR cuvette (Praying Mantis, Harrick) and a droplet (7 μl) of a liquid DMMP (97%, Sigma–Aldrich) was dosed by an automatic micropipette. The acquisition of spectra was started immediately and repeated in selected time intervals (0, 2, 5, … n + 5 until 30, … n + 10 until 60, … n + 20 until 120, … n + 40 until 240 min, and after 24 h). The one spectrum consists of 128 scans with acquisition time 1.48 min and resolution of 4 cm⁻¹. The spectra were baseline corrected without smoothing. All the presented spectra and kinetic curves, representing integrated peak area as a function of time, were processed using OMNIC Spectra (version 8.3) software.

3.1 Material characterization

The XRD patterns (Fig. 1) of the as-prepared titania (1T0Ce_U) and ceria (0Ti1Ce_U) contain only diffraction lines of anatase TiO₂ and cubic CeO₂, respectively. The modification of the synthesis (removing sulphates from the reaction mixture) has been proven as feasible for a preparation of pure oxides without undesired contamination. The relatively low intensity and broadening of the lines indicated small primary crystallites of the samples, which is typical for homogeneous hydrolysis using urea as the precipitating agent (Subrt et al., 2006). In the case of TiO₂/CeO₂ composites, the sample with Ti:Ce molar ratio 2:8 (2Ti8Ce_U) had all typical diffraction lines of cubic CeO₂ with one additional line assigned to anatase TiO₂. Similarly, the equivalent molar ratio of starting Ti and Ce precursors (1Ti1Ce_U) led to a formation of two-phase system TiO₂—CeO₂. The sample with the small amount of Ce (8Ti2Ce_U) exhibited substantially broadened low-intensity lines of anatase without distinct lines of CeO₂. The semi-quantitative phase analysis, crystallite size estimation and lattice parameters obtained from Rietveld refinement are listed in Table 1. Evidently, the mixing of Ti/Ce oxides led to a reduction in mean crystallite size of both titania and ceria. The strong interaction of Ti with Ce ions in composites is further manifested in changes of lattice parameters; note a gradual increase in cell parameter c of anatase (see Table 1).

Close

Figure 1

XRD patterns of the prepared Ti/Ce oxides.

Export to PPT

Raman spectroscopy is a sensitive probe for the phase analysis, evaluation of unit cell distortion, and the variations in the elemental environment of the nanostructured inorganic compounds (Siu et al., 1999). The results of the analysis (Fig. 2) correspond well with XRD. The typical Raman bands in positions 144 cm⁻¹ (E_g), 197 cm⁻¹ (E_g), 397 cm⁻¹ (B_1g), 516 cm⁻¹ (B_1g + A_1g), and 640 cm⁻¹ (E_g) confirmed pure anatase in sample 1Ti0Ce_U. A strong band at position 463 cm⁻¹ (sample 0Ti1Ce_U) belongs to F_2g mode of cubic CeO₂ structure (space group Fm-3m); the appearance of the less prominent broad bands at ca 256 cm⁻¹ and 600 cm⁻¹ is related to displacement of the oxygen atoms from their ideal fluorite lattice positions and to defect-induced longitudinal optical mode (LO) of ceria caused by formation of the oxygen vacancies, respectively (Sellick et al., 2013; Tiseanu et al., 2014; Zhang et al., 2006). The sample 2Ti8Ce_U with the small amount of Ce exhibited only bands assigned to anatase TiO₂, which may indicate Ce incorporation in TiO₂ lattice. The samples with Ti/Ce molar ratios 1:1 and 2:8 contained vibrational bands of both anatase and ceria proving the formation of the two-phase composite system. Nevertheless, the strong interaction between Ti and Ce ions and distortion of the structure resulted in a significant gradual shift of the main anatase band from 144 cm⁻¹ (sample 1Ti0Ce_U) to 151 cm⁻¹ (sample 2Ti8Ce_U).

Close

Figure 2

Raman spectra of the prepared Ti/Ce oxides.

Export to PPT

Diffuse infrared Fourier transform (DRIFT) spectra of the pure titania (see Fig. S1 in the supplementary material) contain only broad band centred at about 3440 cm⁻¹ and band at 1631 cm⁻¹, which were assigned to ν(OH) stretching and δ(OH) bending of adsorbed water and surface hydroxyl groups. With increasing amount of Ce the new bands appeared in the samples at positions ∼1510 cm⁻¹, 1317 cm⁻¹, and 1054 cm⁻¹ assigned to surface bounded carbonates, which are likely to be formed on CeO₂ surfaces upon CO₂ adsorption from ambient air (Li et al., 1989; Vayssilov et al., 2011). No other remnants from the synthesis i.e. nitrates or sulphates were registered.

The surface area (A_BET) and total pore volume (V_tot) of the samples obtained from nitrogen physisorption (see Table 1) differ substantially between pure oxides and TiO₂/CeO₂ composites. Pure ceria had the lowest both surface area (76 m² g⁻¹) and pore volume (0.07 cm³ g⁻¹). Pure titania possessed slightly higher surface area 97 m² g⁻¹ and almost three times as large pore volume 0.2 cm³ g⁻¹. Surprisingly, in the TiO₂/CeO₂ composites increasing amount of Ti leads to a gradual increase in the surface area and pore volume. The composite with the highest amount of Ti (8Ti2Ce_U) possessed the highest A_BET = 146 m² g⁻¹ and V_tot = 0.36 cm³ g⁻¹. The co-precipitation of CeO₂ and TiO₂ in close intimacy led to the inhibition of their crystallization, which consequently resulted in decrease in the crystallite sizes (see XRD results above). This also reflected in increased surface area and pore volume and also in significantly changed pore size distribution (Fig. 3). Pure ceria had relatively low porosity with the narrow size distribution of pores <4 nm with a contribution of micropores (see Table 1). Pure titania had larger pores in range ca 2–7 nm with the substantially higher volume of adsorbed gas. Interestingly, the mixing of TiO₂ with CeO₂ led to a significant change in the pore distribution and amount of adsorbed gas. The largest pore distribution (mesopores of ca 6–12 nm) had samples with the equivalent and 2:8 Ti:Ce molar ratio, respectively. Significantly higher pore volume than all other samples had the sample with a moderate amount of Ce (8Ti2Ce_U). The gradual increase of Ce in sample changed the hysteresis loop from type H4 of ceria (often associated with slit-like pores) to type H2 of titania, which may result from ink-bottle-like pores of varying radius, generated by agglomerates or compacts of spherical particles of non-uniform size and arrangements (Adelkhani et al., 2011).

Close

Figure 3

The pore size distribution and isotherms (inset) of the prepared Ti/Ce samples obtained from the nitrogen physisorption.

Export to PPT

The XPS data provided valuable information about the surface chemical states of Ti, Ce, and O and their interactions in TiO₂/CeO₂ composites. The electronic structure of CeO₂ is characterized by unoccupied 4f states of Ce⁴⁺ (4f⁰) while Ce³⁺ has an occupied 4f¹ configuration. The different 4f configurations for Ce⁴⁺ and Ce³⁺ result in different core levels and also valence bands. The XPS spectrum of partially reduced cerium in Ce 3d consists of three 3d3/2–3d5/2 spin-orbit-split doublets representing different 4f⁰ (Ce⁴⁺) configurations in the final photoemission state and two other doubles of 4f¹ (Ce³⁺) states. In more detail (Fig. 4a), the highest binding energy peaks $u^{‴}$ and $v^{‴}$ (green), located at about 916.5 eV and 898.0 eV, respectively, result from the Ce 3d⁹ 4f⁰ O 2p⁶ final state. The satellite peak $u^{‴}$ associated with the Ce 3d3/2 is characteristic of the presence of Ce⁴⁺. The lowest binding energy states (blue), u (901.0 eV), v (882.2 eV), u″ (907.0 eV), and v″ (888.5 eV), respectively, are the results of the Ce 3d⁹ 4f² O 2p⁴ and Ce 3d⁹ 4f¹ O 2p⁵ final states (Masek et al., 2009). The Ce³⁺ oxides have Ce 3d3/2 and Ce 3d5/2 spectra composed of two multiplets. The highest binding energy peaks (dark yellow), u′ (902.9 eV) and v′ (884.9 eV), belong to the Ce 3d⁹ 4f¹ O 2p⁶ final state. The lower binding energy states (red), u⁰ and v⁰ respectively located at 899.0 eV and 880.6 eV, correspond to Ce 3d⁹ 4f² O 2p⁵ final state. Practically identical components were observed in the TiO₂/CeO₂ mixed sample (see Fig. S2a in the supplementary material), but the calculated Ce⁴⁺/Ce³⁺ ratio decreased by the addition of Ti from 4.8 to 4.0. Therefore, the increase in a relative number of Ce³⁺ is observed (see Table 2). The pure TiO₂ sample had single Ti 2p3/2 peak located at BE 458.9 eV which corresponds to TiO₂ state (Fig. 4b). However, the Ti 2p doublet in sample 1Ti1Ce_U (Fig. 4c) was significantly shifted to a lower BE 459.7 eV, which suggests the strong interaction of Ti with Ce and subsequent reduction in the Ti states. The peak was tentatively deconvoluted to single peaks, where ca. 37% of Ti remained in TiO₂ state (BE 458.9 eV), and about 63% of Ti has been transformed to sub-stoichiometric Ti^x<4+ oxides with BE ca 457.7 eV and 456.2 eV, respectively. The dominant single O 1s peak at BE 529.1 eV of pure ceria (Fig. 4d) represents CeO₂ (Ce⁴⁺), while the second peak with a lower intensity centred at ∼531.1 eV was associated with Ce³⁺ ions (Beche et al., 2008). A shoulder at higher BE (532.1 eV) was related to hydroxyl (—OH) groups on the surface of the sorbent. The O 1s peak of the pure titania (Fig. S2b) consists of dominant peak related to Ti⁴⁺ (BE 529.8 eV) and a shoulder at higher BE (531.2 eV) related to the surface (—OH) groups. In TiO₂/CeO₂ composite, a single broad peak was tentatively deconvoluted according to several expected components (Fig. S2c). The sample composition is presented in Table 2. Only Ti, Ce, O, and C are present on the surface of samples; note that the lowest carbon pollution was on TiO₂/CeO₂ composite. The overall survey spectra of the TiO₂/CeO₂ sample measured using Al and Mg anodes are presented in Fig. S2d.

Close

Figure 4

XPS spectra of the (a) Ce 3d region of ceria (0Ti1Ce_U), (b) Ti 2p region of pure titania (1Ti0Ce_U), (c) Ti 2p region of 1Ti1Ce_U composite, (d) O 1s region of pure ceria.

Export to PPT

SEM investigation showed that ceria (sample 0Ti1Ce_U, Fig. 5a) formed micron-sized platelets clumped to flower-like agglomerates. Contrarily, titania (sample 1Ti0Ce_U, Fig. 5b) consist of small primary nanoparticles which constitute to irregular “foam” clusters. That is attributed to modification of the synthesis route since standard urea precipitation of TiOSO₄ provides regular mono-dispersed spherical agglomerates of primary nanoparticles (Bakardjieva et al., 2005). This confirmed that the anion of the starting metal precursor significantly affects the crystallisation process, and removing of the sulphate ions, in our case, led to the irregular morphology of precipitated titania particles. The mixing of a moderate amount of Ti with Ce precursor (sample 2Ti8Ce_U, Fig. 5c) led to a formation of two phases. However, the morphology of ceria flower-like particles is seriously disrupted, suggesting again, strong interaction of Ti with Ce. In the case of the equivalent amount of Ti and Ce (1Ti1Ce_U, Fig. 5d) and higher Ti concentration (not shown) there are hardly recognizable two distinct phases, and both ceria and titania formed irregular structure resembling foam.

Close

Figure 5

SEM micrographs of (a) ceria (0Ti1Ce_U), (b) titania (1Ti0Ce_U), (c) composite 2Ti8Ce_U, (d) composite 1Ti1Ce_U.

Export to PPT

TEM analysis of both titania (Fig. S3a and 3b) and ceria (Fig. S3c and 3d) showed that the agglomerates observed on SEM consist of randomly oriented nano-sized primary particles, which are smaller in the case of ceria. Also, the TiO₂/CeO₂ composite (1Ti1Ce_U, Fig. 6) had a relatively uniform structure of agglomerated nanoparticles with mean sizes ∼6 to 8 nm which is coherent with calculated sizes from XRD (Table 1). There are no distinct two phases, but the EDS mapping (Fig. 6b) revealed very homogeneous dispersion of Ti and Ce in the nanostructure of the sample. This confirmed easy and efficient mixing of the elements using our modified procedure. The interplanar distances 0.32 nm (Fig. 6b) and 0.28 nm (Fig. 6c) are typical for (1 1 1) and (2 0 0) planes, respectively, of cubic ceria.

Close

Figure 6

TEM micrographs of (a) 1Ti1Ce_U, (b) EDS map of the same place as in (a), (c and d) high-resolution TEM of the same sample.

Export to PPT

3.2 Degradation of parathion methyl

Experimental dependencies for the degradation of parathion methyl were evaluated using the following equation, that was successfully applied for the description of the CWA degradation on reactive sorbents (Stengl et al., 2012a):

It is assumed in this model that two processes (faster and slower) are effective in the pesticide degradation (Stengl et al., 2012a). q_t is the residual amount of pesticide in time t, k₁ and k₂ are the pseudo-first order rate constants, q₁ and q₂ are model parameters expressing the ratio, in which the faster and slower processes are involved, and q_∞ is the residual (undestroyed) amount of pesticide in infinite time. Under analogous assumptions, an equation describing the creation of 4-nitrophenol may be written as follows:

The normalised degradation curves for parathion methyl (red squares) and curves for a production of the main degradation product 4-NP (blue squares) on Ti/Ce pure oxides and their composites are presented in Fig. 7. Although the degradation of organophosphorus compounds, such as DMMP, on the surface of TiO₂ without illumination was described (Rusu and Yates, 2000), titania prepared by modified homogeneous hydrolysis was almost inactive in degradation of parathion methyl as seen in Fig. 7a. In opposite, nano-ceria prepared by our procedure proved ability to degrade parathion methyl (ca. 50% of parathion methyl was degraded in 80 min) and produce an equivalent amount of 4-NP as the primary degradation product (Fig. 7b).

Close

Figure 7

Degradation of parathion methyl on the sorbents (a) titania (1Ti0Ce_U), (b) ceria (0Ti1Ce_U), (c) 8Ti2Ce_U, (d) 1Ti1Ce_U, (e) 2Ti8Ce_U.

Export to PPT

It is widely accepted that the surface sites, such as corners, edges, impurities or point defects, and surface hydroxyl (—OH) groups, are responsible for reactive adsorption of the organophosphorus compounds on the nano-sized metal oxides (Chen et al., 2010; Klabunde et al., 1996). The XPS results (see above) showed that both titania and ceria contain a considerable amount of the surface hydroxyl groups on their surfaces. Nevertheless, the IR results (Fig. S1) suggested that the relative amount of adsorbed water and surface hydroxyl groups is higher on ceria. The (—OH) group is a strong nucleophile that attacks the phosphorus atom (Chen et al., 2010), which consequently leads to degradation of the parathion methyl to intermediates (i.e. 4-NP). Thus, the higher abundance of these groups on ceria may contribute to its better performance. However, the good degradation efficacy of ceria compared to titania stems from several aspects. The abundance of Ce³⁺ cations in the unique structure of ceria is compensated by the presence of oxygen vacancies. Ab initio simulation techniques suggest that the oxygen vacancies act as active sites for water dissociation (Molinari et al., 2012). The degree of reduction in the surface is, therefore, an important parameter for the reactive adsorption of pollutants.

Interestingly, the degradation efficiency of the TiO₂/CeO₂ composites is moderately higher than that of pure CeO₂. As can be seen from Table 3, the highest degree of conversion was achieved with TiO₂/CeO₂ samples with the prevalent amount of Ce (2Ti8Ce_U, Fig. 7e) and the equivalent amount of Ce and Ti (1Ti1Ce_U, Fig. 7d) in the structure. Moreover, the initial rate of conversion is substantially higher on these two samples pointing out to their better overall reactivity towards PM. As shown in XPS results, the strong interaction of Ce with Ti leads to a partial reduction of Ti⁴⁺ ions and a slight increase of Ce³⁺/Ce⁴⁺ ratio. The reduced sites then can act as the active sites for pesticide degradation. It implies that by the composition of the Ti/Ce mixed sample can, to some extent, control the degree of reduction in the surface and consequently the efficiency of the reactive sorbent.

Note that the TiO₂/CeO₂ composites possessed higher surface area (Table 1) and considerable wider pore size distribution (Fig. 3) compared to pure ceria and titania, which may contribute to their better performance. Nevertheless, the sample with the highest surface area and pore volume (8Ti2Ce_U) exhibited lower efficacy compared to pure ceria. Therefore, the surface area is not likely the sole parameter controlling the sorbent performance.

The most active prepared TiO₂/CeO₂ composite (rate constant k = 0.669 min⁻¹) was substantially more efficient in PM degradation than TiO₂/FeO₂ composites (rate constant k = 0.026 min⁻¹) (Henych et al., 2015), and almost as active as Manganese (IV) oxide (rate constant k = 0.888 min⁻¹) (Stastny et al., 2016) using similar synthesis procedure and identical testing procedure.

3.3 Reactive adsorption of DMMP

The behaviour of the nerve agent simulant DMMP vapour on the metal oxide surfaces at various temperatures was extensively studied during the last two decades (Chen et al., 2010; Mitchell et al., 2004, 1997; Panayotov and Morris, 2009; Rusu and Yates, 2000; Trubitsyn and Vorontsov, 2005). In this work, a liquid DMMP, without subsequent UV or Vis illumination at ambient temperature, was used as a model compound to study the reactivity of the metal oxide surfaces. The obtained DRIFT fingerprint spectra of the DMMP reactive adsorption on selected Ti/Ce oxides in various time intervals are presented in Fig. 8. Tentative assignment of the bands was made according to the literature (Moss et al., 2005; Rusu and Yates, 2000). The adsorption of DMMP is realized through binding of the electron-rich phosphoryl oxygen (P⚌O) to the surface at either Lewis acid sites or with hydroxyl groups by hydrogen-bond formation. For temperatures higher than 200 K is the adsorption dissociative, leading to the formation of an adsorbed non-volatile methyl methylphosphonate passivating the adsorption sites, while the second methyl group presumably forms methanol (Mitchell et al., 2004; Rusu and Yates, 2000).

Close

Figure 8

Fingerprint DRIFT spectra of DMMP reactive adsorption on selected Ti/Ce samples. (a) titania (1Ti0Ce_U), (b) ceria (0Ti1Ce_U), (c) 2Ti8Ce_U, (d) 1Ti1Ce_U.

Export to PPT

Thus, the fate of P⚌O vibration may indicate the reactivity of the sorbent surface. The P⚌O stretching vibrations of DMMP gas occur at 1276 cm⁻¹ (Rusu and Yates, 2000), whereas in a liquid phase (measured by ATR technique) it were found at ∼1260 cm⁻¹. The surface adsorption should lead to further downshift of the wavenumber. The P⚌O vibration of DMMP on the surface TiO₂ (Fig. 8a) measured immediately upon application (time 0) was found at ∼1265 cm⁻¹. After 15 min, the only change in the spectra is downshift of the bands, as DMMP is being bounded the surface; however, no significant change in intensity was observed. The P⚌O band is further downshifted in the course of reaction suggesting the strong interaction of DMMP with the surface. After 24 h, the P⚌O band was found at 1233 cm⁻¹, but only slightly changed in intensity, that is consistent with the degradation of PM (see above), which suggests the very low activity of as prepared titania for organophosphates decomposition. Contrarily, when DMMP was applied onto ceria surface (Fig. 8b), the P⚌O band was registered at 1243 cm⁻¹, which corresponds to surface adsorbed DMMP and which is 20 cm⁻¹ downshifted compared to titania. This suggests that DMMP is much faster adsorbed on ceria surfaces. Furthermore, the P⚌O band progressively lose its intensity, which is accompanied with emerging of a new O—P—O band at 1082 cm⁻¹. As the reactive adsorption took place, the P⚌O double bond is being transformed to O—P—O species, which has an intermediate character between a single and double bond (Rusu and Yates, 2000). As a consequence is expected that all other bands (P—O, C—O, C—P) will be influenced. The shoulder of P⚌O band is still clearly visible even after 24 h, which indicates that all available active sites for DMMP were used, and no further DMMP can be transformed. In the case of TiO₂/CeO₂ composites with the prevalent amount of Ce (2Ti8Ce_U, Fig. 8c) and the equivalent amount of Ti and Ce (1Ti1Ce_U, Fig. 8d) the P⚌O vibration was found at 1252 cm⁻¹ and 1253 cm⁻¹, respectively. These values lie somewhat between those found for pure ceria and titania, which correspond to the combination of TiO₂ and CeO₂ in the sample. After 15 min the P⚌O band downshifted to 1245 cm⁻¹ and decreased in intensity as the P⚌O band undergoes transformation. This trend continues with prolonged time and a new band assigned to O—P—O species appeared at 1096 cm⁻¹. Note the decrease in intensity of the other bands (C—O and CH₃O), while the bands related to CH₃P remain very stable. This may indicate dissociation of DMMP, including liberation of the methoxy groups, to the final surface product methyl methylphosphonate. The kinetics of disappearance of the P⚌O band (calculated from the integrated peak area), as a main indication of the reactive adsorption, is presented in Fig. S3. As can be seen, the TiO₂/CeO₂ composites (2Ti8Ce_U, 1Ti1Ce_U) were more efficient in the elimination of the P⚌O band compared to pure ceria. These results are consistent with the results of PM degradation (see above).

The DRIFT spectra of the CH stretching region of the selected samples are presented in supplementary Fig. S4. As can be seen, the intensities of the CH₃ bands on the surface of titania (Fig. S4a) are substantially lower than those on pure ceria (Fig. S4b) and TiO₂/CeO₂ composites (Fig. S4c and S4d). Therefore, much less of DMMP is being adsorbed on titania. With prolonged time, the intensities of CH₃ bands on all samples decreased as methoxy species are liberated from DMMP molecule. The normalised curves describing the change in the sum of all CH groups as a function of time can be seen in Fig. S5. Their decrease in time is correlated with degradation of the P⚌O band and suggests the overall reactivity of the sorbent surfaces.