Natural raw clay, involved herein, was sampled from Marrakesh-High Atlas region. Detailed characterizations performed by Rhouta et al. [28] on this clay revealed that it is made up of fine clay fraction (<2 μm) (>65%) along with accessory minerals, namely quartz (< 5%) and carbonates (≈30 wt%) in the forms of calcite (Mg_0.03Ca_0.97CO₃) and ankerite (Ca_1.01Mg_0.45Fe_0.54(CO₃)₂). The extracted fine fraction (<2 μm) was found to be exclusively composed of fibrous clay minerals, namely palygorskite (≈95%) exhibiting predominant dioctahedral character and deficiency into zeolitic water along with sepiolite (≈5%). The composition of this palygorskite was found on the basis of 26 oxygens to be (Si_7.97Al_0.03)(Mg_2.17Al_1.46Fe_0.40Ti_0.05)(Ca_0.03Na_0.07K_0,03)O_20.18(OH)_1.94(H₂O)_3.88, 2.43 H₂O. Its Cation Exchange Capacity (CEC), Brunaeur-Emmet-Teller method (BET) specific surface area and total porous volume were assessed to be 21.2 meq∙100 g⁻¹, 116 m²∙g⁻¹ and 0.458 cm³∙g⁻¹ respectively.

The Ti precursor considered herein is the titanyl sulfate (TiOSO₄∙2H₂O) purchased from Aldrich and used without further treatment.

In contrast to our previous works [24] [25] [26] [27] [30] in which the synthesis of clay minerals supported photocatalysts implied several laborious and time-consuming steps including purifying, homoionisation and fine fraction recovering before its functionalization, the elaboration herein was accomplished according to a one-pot route starting directly from raw clay. Indeed, given amounts of raw palygorskite clay and TiOSO₄∙2H₂O powders were introduced in an alumina jar of planetary ball mill PM100 from Retsch for being grinded using six alumina balls of 20 mm in diameter. The synthesis of TiO₂ supported on palygorskite fibers was achieved by in-situ reacting carbonates accessorily present (≈30 wt%) in raw clay with TiOSO₄∙2H₂O precursor according to different conditions. Indeed, the amount of each reactant was chosen in such a way to have a total mass of reactants of 50 g per run by considering two raw clay/TiOSO₄ mass ratios of 0.6/1 and 3.3/1, which corresponded to mixtures in which both compounds were in stoichiometric or in excess proportions, respectively. The grinding was performed at a rate of 500 rpm for different times (10, 15, 20 and 30 min). As-grinded samples were thereafter air annealed for 1 h at different temperatures (400˚C, 500˚C, 600˚C and 700˚C). Samples were designated Pal_TiO₂(x)_G_y_T_z by referring to the composite nature of the sample (Pal_TiO₂), the reactant mass ratio (x = TiO₂/Pal), the grinding time (G_y) and the post-heat-treatment temperature (T_z). The obtained powdered materials were finally stored for further uses.

The structural changes of TiO₂ upon heating of as-grinded samples were analyzed in situ versus the temperature by XRD over the two-theta range 2_60 deg using a Bruker D8 Advance diffractometer equipped with a Vantec Super Speed detector and an Mid-Infrared (MIR) radiation heating chamber (Bragg-Brentano configuration; Ni filtered Cu K_α radiation). The diffractograms were recorded every 50˚C from the room temperature to 950˚C. An isotherm was maintained at each level for 40 min to record the pattern then the temperature was increased using a ramp of 1 deg.s⁻¹ up to the next level. Also XRD at room temperature was recorded on TiO₂_Pal samples prior to air annealing at different temperatures in the same angular range using a Seifert XRD 3000TT diffractometer equipped with a graphite monochromator (Bragg-Brentano configuration; Cu K_α radiation).

The microstructure, morphology and uniformity of the samples were analyzed at different steps of the synthesis by a Scanning Electron Microscope (SEM; Leo-435VP) equipped with an Oxford energy dispersive spectrometer (EDS, (EDS, KEVEX Si (Li) assisted software Brüker)). Also, a JEOL JEM 2010 TEM equipped with a Tracor EDS analyzer was used for characterizing clay particles and performing local elemental composition.

The photocatalytic activity of different samples was evaluated using three kinds of set-ups: a laboratory-scale set-up, an indoor solar simulator and an outdoor solar pilot. As far as the first set-up is concerned, the degradation reaction was performed in a batch quartz reactor (40 × 20 × 36 mm³) placed in a thermostated chamber (25˚C) under the UV light of a lamp (HPLN Philips 125 W) emitting at 365 nm. The reactor was irradiated with a photon flux of about 100 mW∙cm⁻² by adjusting the distance to the lamp so that it simulates the UV intensity of solar spectrum on the earth [31]. This lamp was chosen because the OG absorption is negligible at this wavelength and, as a result, the direct photolysis of the solution (i.e. without photocatalyst) was found negligible for more than 24 h. The photoactivity of different samples was assessed at pH around 6 by measuring the decomposition rate of OG in aqueous solutions containing the supported photocatalyst according to a procedure previously reported [32] and adapted to the dispersion of supported powder photocatalysts in previous papers [26] [33]. The photocatalyst powder of TiO₂_Pal nanocomposites was added to 25 cm³ of OG solution (10⁻⁵ M) in an amount equal to 0.8 g∙dm⁻³. This catalyst mass was found as optimum to avoid excess of catalyst and to ensure an efficient absorption of photons [34]. The dispersion was agitated with an inert Teflon magnetic stirrer. To determine the dye concentration, aliquots were taken from the mixture at regular time intervals and centrifuged at 12,500 rpm for 5 min. The OG concentration in the supernatant was determined by measuring the absorbance at 480 nm using a UV-VIS-NIR spectrophotometer (Perking Elmer lambda 19).

Sunlight is particularly weak in the UV spectral region and, in addition, it changes along the day due to day-night and seasonal cycles [35] [36]. Thus, prior optical and kinetic studies have to be performed to determine on one hand optimal catalyst mass allowing the maximum (≈90%) of the absorption of photons (considered as reactant) and, on the other hand, kinetic constants to achieve process efficiency in an outdoor solar pilot [35] [37] [38]. Therefore, optical properties of photocatalysts suspensions differing in mass concentrations were studied by an optical experimental setup described in details elsewhere [37] [39]. Practically, the light supplied by a source simulating solar radiation (maximum intensity equal to 1000 W∙m⁻² in the full spectral range, i.e. 50 W_uv∙m⁻² in the UV range) came across a Polymethyl methacrylate (PMMA treated to be UVA transparent, Sunactive^® GS2458) cell 2 cm thick (i.e. optical length) containing under stirring the TiO₂_Pal dispersions whose concentrations ranged from 0.1 to 4 g∙L⁻¹. This cell was placed at the inlet of an integrating sphere. The direct radiation as well as scattered radiation transmitted hence collected from the integration sphere were directed via an optical fiber towards a spectrophotometer permitting the detection of wavelengths ranging from 250 to 1100 nm (ultraviolet, visible and near-infrared). The transmission of light was hence assessed versus catalyst powder concentration to determine the mass concentration at which the 90% absorbency was reached. Afterwards, kinetic study of OG dye degradation was carried out versus mass concentration of catalysts powders dispersion by using an indoor solar simulator set-up. For a given mass concentration, the catalyst dispersion was kept under stirring in a 150 mL quartz bucker (suspension height of 5 cm), then it was irradiated by sunlight simulator (1000 W∙m⁻²) under constant photons flow. At different time intervals, aliquots were picked up from media, centrifuged and resulting supernatants analyzed by spectrophotometry to determine the dye concentration according to the way described above.

Afterwards, the efficiency of TiO₂_Pal catalysts towards the photodegradation under sunlight of OG dye was tested by using an outdoor solar pilot described in details by Janin et al., [40] (picture given in the supplementary Figure 1S.). Briefly, a mass concentration of TiO₂_Pal that permitting 90% of photons absorption was dispersed under stirring in an OG dye solution and stored in a stainless-steel tank (with a maximum volume of 50 L). Then 6 L of effluent were fed into a flat solar panel (30 × 100 × 2 cm³) from the bottom to the top via a liquid pump and recirculated into the thank. This panel was covered with a special PMMA plate to prevent evaporation of compounds and to ensure watertight. This flat reactor was oriented southwards and inclined at an angle around 42˚ in order to maximize the average solar irradiation throughout the year. This was controlled by two equipments: a pyranometer whose wavelength domain varies between 380 to 2800 nm and an UV radiometer (UVA-Sensor CT-UVA 3) measuring the corresponding spectrum in the range of 310 - 400 nm.

Figure 1 shows an overview of the room temperature XRD diagrams of the two

starting materials after a grinding for 30 min at 500 rpm: titanyl oxysulfate precursor (TiOSO₄) and palygorskite-rich raw clay (Pal). Also, the XRD diagrams of a mixture of TiOSO₄ and Pal clay in a mass ratio of 0.6/1 after grinding for 30 min at 500 rpm is reported (Pal_TiOSO₄).

The diffractogram pertaining to the milled titanium precursor revealed only the reflections at the 2θ angular positions of 10.67˚ (8.35 Å), 10.95˚ (8.20 Å), 18.61˚ (4.79 Å), 19.07˚ (4.69 Å), 21.43˚ (4.16 Å), 25.44˚ (3.50 Å), 25.70˚ (3.47 Å), 28.47˚ (3.13 Å), 29.25˚ (3.05 Å), 37.70 (2.38 Å), 38.61 (2.33 Å) corresponding to the compound TiOSO₄ (ICDD file N˚14-503) without evidence for the formation of any other crystalline phase, especially TiO₂. In the same sense, the XRD pattern of the grinded Pal raw clay showed several reflections corresponding to the fibrous mineral, i.e. palygorskite, at the angular positions described by Rhouta et al., [30]. The characteristic basal reflection is at 2θ ≈ 8.38˚ (10.56 Å) in addition to the peaks ascribed to the accessory minerals, namely the carbonates in the form of calcite (C) Mg_0.03Ca_0.97CO₃ (ICDD file: 01-089-1304) and ankerite (A) Ca_1.01Mg_0.45Fe_0.54(CO₃)₂ (ICDD: 01-084-20 2066) as well as quartz (Q) (ICDD file: 03-065-0466) whose strongest reflections were observed at 29.46˚ (3.03 Å), 30.82˚ (2.90 Å) and 26.64˚ (3.34 Å), respectively.

This result showed that both the palygorskite and the accessory phases, constituting the raw Pal clay, were apparently not affected by the grinding operation in the mechanosynthesis jar since their crystallographic structure remained stable. On the other hand, these grinding conditions of the Pal_TiOSO₄ mixture caused a drastic decrease in the XRD intensities of TiOSO₄ reflexions at the point to disappear, as well as those of the carbonates (calcite and ankerite), while preserving the clay structure confirmed by the persistence of the corresponding peaks. The disappearance of TiOSO₄ reflexions can be due to its amorphisation but the major event during grinding is the trigger of reaction (1), according to which TiOSO₄ reacts with the carbonates present among the accessory minerals in the raw clay Pal:

TiOSO₄∙2H₂O + Pal/(Na₂, Ca, Mg)CO₃ → TiO₂/Pal + (Na₂,Ca,Mg)SO₄∙2H₂O + CO₂↑ (1)

The occurrence of this reaction was further confirmed by the production of the hydrated sulfate compound CaSO₄∙0.5H₂O evidenced by the presence in the composite sample Pal_TiOSO₄ of the characteristic peaks noted S on the diffractogram (Figure 2) at angular positions 2θ of 14.71˚ (6.00 nm), 29.71˚ (3.00 nm) and 31.86˚ (2.80 nm) in agreement with the ICDD file 01-081-1848. The near-consumption of the TiOSO₄ precursor revealed by XRD may be due to the fact that the low value of the mass ratio TiOSO₄:clay (0.6:1) would be the driving force of reaction (1).

Furthermore, the reaction (1) also showed that titanium dioxide has to be formed but no diffraction peaks of this oxide was observed in Figure 3 suggesting that the as-prepared TiO₂ phase must be amorphous after the mechanosynthesis step (TiOSO₄: clay = 0.6:1; 500 rpm; 30 min). Subsequently, a mixture of TiOSO₄ and Pal raw clay with the mass ratio 3.3/1, i.e. with Ti precursor in excess was grinded for 30 min at 500 rpm. Then, XRD patterns of this composite sample were recorded in-situ in air as a function of the temperature (Figure 2).

For as-milled sample, the diagram recorded at 50˚C revealed that, as mentioned above, amorphous TiO₂ and the hydrated CaSO₄ product were formed during the grinding stage as a result of the above reaction (1). In this respect, a significant proportion of the carbonates were consumed by the reaction with TiOSO₄ (that was in excess) has shown by the decrease of their XRD intensity. This was further supported by a huge decrease in carbonate content determined

by coulometric analysis. Indeed, the carbonate content has been reduced approximately by a factor of 2, from 28 wt% ± 2 wt% in the raw clay Pal to 13 wt% ± 2 wt% in the grinded TiOSO₄/Pal (3.3/1) mixture.

By increasing the temperature, it can be seen that the crystallization of anatase TiO₂ started from 450˚C as shown by the emergence of the corresponding peaks at angular positions 2θ of about 25.44˚ (3.51 Å) and 38.53˚ (2.33Å) (ICDD Data Sheet 01-021-1272). As the temperature increased further, the anatase peaks became more intense and well resolved, indicating the improvement of its crystallization. The anatase peaks remained observable up to 700˚C, without the appearance of any XRD peaks of the less photoactive rutile variety. This strongly denoted the remarkable stability of the anatase formed under the conditions of this study. Palygorskite peaks were observable up to 400˚C, beyond which their intensities severely reduced. This can be ascribed to the occurrence of structural collapse of the clay mineral. However, this can be also due to an absorption filter induced by the conformal coverage of anatase particles on palygorskite fibers as suggested by Rhouta et al., [28] and Bouna et al., [29] and confirmed by the TEM analyses shown in the next section.

In order to examine the effect of grinding time on the reactivity of TiOSO₄ with the carbonates present in the Pal raw clay, a mixture of the two compounds in a ratio of 0.6/1 was ground at 500 rpm for 10, 15, 20, 25 and 30 min and then air annealed at 600˚C for 1 h. The superimposition of the corresponding XRD diagrams (Figure 3) clearly showed the formation of anatase TiO₂ for all the samples investigated regardless grinding time. Nevertheless, the anatase crystallinity appeared to improve with the grinding time as evidenced by the gradual narrowing of it characteristic peak at about 2θ around 25˚. In this respect, by referring to quartz, which is an insensible and stable accessory mineral under the applied grinding conditions, the increase of the intensity ratio of the main anatase/quartz peaks is noteworthy. This probably reflected an increase in the amount of anatase phase formed with the increase of grinding time.

XRD results were further supported by TEM characterizations. Indeed, depending on the grinding conditions, TEM analysis after the first step showed palygorskite fibers of 30 - 70 nm in diameter and 650 nm in length mixed with TiOSO₄ platelets of 400 - 600 nm with an average thickness of 3 - 5 µm. At this stage, no evidence for crystallized anatase nanoparticles (NPs) was found confirming this oxide phase was amorphous. After annealing at 600˚C in air for 1 h, TEM revealed that palygorskite fibers were entirely wrapped with homogeneous monodisperse and spherical TiO₂ anatase NPs 10 - 15 nm in diameter (Figure 4(a)). Only a few palygorskite fibers were not covered as revealed by EDS analysis. For most of them, a granular and conformal TiO₂ coverage formed a uniform cladding (Figure 4(b)).

The photocatalytic activity of different TiO₂–Pal composite samples elaborated by mechanosynthesis was tested towards the degradation under UV light of OG dye by using the laboratory set-up described in the experimental section (2.3). The results in Figure 5 showed that photocatalytic efficiency significantly increased with the amount of TiOSO₄ used in photocatalyst synthesis and thus the amount of TiO₂ present in TiO₂_Pal nanocomposite. This was evidenced by on one hand the strong decrease of OG concentration for instance in the first 10 min, revealing a high initial degradation rate of the pollutant around 2.6 × 10⁻² s⁻¹ for TiO₂/Pal mass ratio of 3.3/1 with respect to that determined for the mass ratio of 0.6/1 (≈1.5 × 10⁻² s⁻¹). On the other hand, the almost total elimination of the OG dye from the solution was achieved within about 2 h when using TiO₂/Pal mass ratio of 3.3/1 whilst almost 10% of the dye remained after the same time of UV irradiation in presence of TiO₂/Pal mass ratio of 0.6/1 (but a plateau has not yet been reached).

The influence of annealing temperature on the photocatalytic activity towards the removal of OG dye of TiO₂_Pal was afterwards studied on the most active one corresponding to TiO₂ /Pal mass ratio = 3.3/1. Figure 6 interestingly showed that the unheated as-milled Pal_TiOSO₄ mixture (sample Pal_TiO₂_(3.3)_30) already exhibited photoactivity under UV irradiation. Indeed, 50% of the initial quantity of OG decreased after 2 hours of irradiation. This is due to the formation

of photoactive TiO₂ resulting, as beforehand mentioned, from the heterogeneous solid-state reaction between TiOSO₄ and the carbonate phases (calcite + ankerite) present as accessory minerals in the Pal raw clay (reaction 1). Although the TiO₂ formed in as-grinded material was amorphous, as ascertained above by XRD, this is not exceptional since it was already shown that amorphous TiO₂ can be active in photocatalysis under UV irradiation [41]. Indeed, from a general point of view, even if amorphous semiconductors have less efficient electronic properties than their crystallized counterparts, they nevertheless find specific applications as in photovoltaic.

Samples annealed at 400˚C and 500˚C showed quite similar photo-degradation curves of the OG and paradoxically showed a lower photoactivity compared to as-grinded sample, allowing the removal of only 15% after 2 hours of illumination. This annealing temperature range is therefore insufficient to ensure a good crystallization of TiO₂ in the form of anatase but, on the other hand, it seems to be sufficient for inducing a detrimental diffusion of species that trap excitons (Pal clay cations for example), thus affecting the photocatalytic efficiency of the composite. Calcination at 600˚C promoted the most efficient active palygorskite-supported TiO₂ photocatalyst since it allowed almost complete degradation of the OG after 2 hours of UV irradiation. This improvement was attributed to the good crystallinity of TiO₂ anatase upon the sample calcination at 600˚C as confirmed above by XRD. On the other hand, when the sample was annealed at 700˚C, photocatalytic activity again decreased. This reduction in photocatalytic activity may be due to the growth of TiO₂ particles beyond the optimal nanocrystalline size, which promotes the recombination of charge carriers. Also, a partial phase transformation of the more active anatase TiO₂ phase into the less photoactive rutile variety as reported by Bouna et al., [29] and Rhouta et al., [28] is not excluded although none of the corresponding peaks have yet been detected at 700˚C, the presence of other crystalline phases can hide them.

Owing to its high photoactivity under UV evidenced above by the tests on the laboratory set-up, the TiO₂_Pal nanocomposite with mass ratio TiO₂/Pal of 3.3/1 was thereafter retained for performing the study of photocatalytic activity under sunlight towards the removal of OG dye.