Photoreduction of a Pd-Doped Mesoporous TiO 2 Photocatalyst for Hydrogen Production under Visible Light

: Photoreduction with visible light can enhance the photocatalytic activity of TiO 2 for the production of hydrogen. In this article, we present a strategy to photoreduce a palladium-doped TiO 2 photocatalyst by using near-UV light prior to its utilization. A sol-gel methodology was employed to prepare the photocatalysts with different metal loadings (0.25–5.00 wt% Pd). The structural and morphological characteristics of the synthesized Pd-TiO 2 were analyzed by using X-ray Diffraction (XRD), BET Surface Area (S BET ), TemperatureProgrammed Reduction (TPR), Chemisorption and X-ray Photoelectron Spectroscopy (XPS). Hydrogen was produced by water splitting under visible light irradiation using ethanol as an organic scavenger. Experiments were developed in the Photo-CREC Water-II (PCW-II) Reactor designed at the CREC-UWO (Chemical Reactor Engineering Centre). It was shown that the mesoporous 0.25 wt% Pd-TiO 2 with 2.5 1eV band gap exhibits, under visible light, the best hydrogen production performance, with a 1.58% Quantum Yield being achieved.


Introduction
In the last few years, researchers have been investigating clean and emission-free renewable energy resources [1]. In this respect, hydrogen is an energy vector that has attracted the attention of many scientists in both academia and industry [2][3][4][5][6][7]. Hydrogen is one of the most abundant elements on earth that can be obtained from a variety of organic sources [8]. Besides its abundance, hydrogen has a high calorific value (143 MJ/kg) and clean emissions, with the combustion of hydrogen producing water as the only by-product [9,10].
Photocatalysis provides a promising method for hydrogen production [11]. This method involves photons and semiconductors that generate electron-hole pairs, facilitating water splitting [12]. Photoexcited electron-hole pairs can be separated efficiently using sacrificial agents, which allow the formation of hydrogen with reduced electron-hole pair recombination [13][14][15][16][17]. Nowadays, however, this process faces challenges in being implemented using visible light, given its low photon conversion efficiency [18].
In order to achieve photocatalytic water splitting, feasible photocatalysts must meet the following criteria: (a) They must display suitable band gaps to absorb visible light, (b) They must be chemically stable under redox conditions, (c) They must have a low cost, (d) They must be recyclable, (e) They must be chemically resistant and (f) They must be adaptable for large-scale hydrogen production [19,20].
Until today, synthesized titanium dioxide (TiO2) has been the most frequently used semiconductor due to its oxidation and chemical resistance properties, accessibility, and affordability [21]. TiO2 contains three allotropic phases: anatase, rutile, and brookite [22]. However, anatase appears to be the most desirable phase for photocatalytic hydrogen production [23]. Synthesized mesoporous semiconductors may also add significant beneficial features to photocatalysis given that they may contribute with: (a) favorable electronic structures, (b) adequate properties for light absorption, (c) sufficient electron transport properties, (d) good photo-semiconductor excitation and (e) suitable chemical species transport properties [24].
Photocatalytic reactions can be realized via sunlight irradiation [25]. TiO2 is however, limited by its 3.2 eV band gap, being able the use about 4% of the solar spectrum only. Thus, modified TiO2based photocatalysts with smaller band gaps are required for more efficient sunlight utilization [26]. One possible approach to modify photocatalysts is by loading a noble metal as a co-catalyst. For instance, it has been proven that loading a noble metal on TiO2 is an effective method to improve its photoactivity [27]. In this respect, in prior studies, our group has investigated the effect of platinum on TiO2 under near-UV light [28,29].
Nevertheless, one can consider the use of less expensive noble metal co-catalysts with lower Fermi levels than TiO2, such as palladium. In this respect, palladium is anticipated to decrease the electron-hole recombination, leading to an energy band gap reduction [30]. Abdelaal et al. [31], for instance, prepared a Pd-TiO2 photocatalyst for methylene blue degradation under a 300 W Xenon lamp irradiation. Once the Pd metal was added, authors observed a diminished band gap of 3.04 eV. Likewise, Espino et al. [32], doped TiO2 with palladium for sodium diclofenac, isoproturon, and phenol degradation in water using a 400 W mercury lamp. They noticed a decreased band gap of 2.78 eV.
In the present study, a mesoporous Pd-TiO2 photocatalyst active under visible light is considered. A key preparation step, photoreducing the Pd-TiO2 using near-UV light is implemented to enhance its photocatalyst activity [33]. It is shown, that the near-UV light photoreduced mesoporous Pd-TiO2 displays a 2.5 1eV energy band gap. The near-UV light photoreduced Pd-TiO2 photocatalyst is evaluated in a Photo-CREC Water-II Reactor (PCW-II Reactor) under visible light [34]. Ethanol is employed as an organic scavenger to help with electron-hole separation [33]. The promoted mesoporous Pd-TiO2 energy band gap reduction enhances light absorption in the visible region [35]. This leads as is reported in the present study, to enhanced hydrogen production and increased Quantum Yields (QY).

Adsorption-Desorption Isotherms
The BET-surface area (SBET), the pore diameter (Dp), and the pore volume (Vp) of the prepared photocatalysts are reported in Table 1. These mesoporous photocatalysts display an increase in the specific surface area versus the one obtained for Degussa P-25. For the TiO2 doped with palladium in amounts greater than 0.25 wt%, a decrease in the specific surface area and an increase in the average pore diameter were noticed, with this being attributed to a moderate blocking of the TiO2 pores with Pd [36]. On the other hand, and when utilizing the Barrett-Joyner-Halenda (BJH) methodology, the mesoporous pore size distribution was found to be unimodal for the 0.25 and 0.50 wt% Pd-500 °C thermally treated TiO2 photocatalysts, with pore sizes in the 18-22 nm range. However, and for the photocatalysts with Pd loadings equal or larger than 1.0 wt%, the unimodal pore size distribution evolved towards a bimodal pore size distribution, with a second peak at 16-35 nm.

Hydrogen Chemisorption
By employing hydrogen pulse chemisorption, it was possible to determine the fraction of dispersed Pd on the photocatalyst [37]. Specifically, by assuming that only one hydrogen molecule chemisorbs on a single Pd site, the metal percent dispersion was calculated as reported in Table 2. This shows that increasing the metal loading decreases the metal dispersion, with the metal loading remaining at a high 75% for the 0.25 wt% Pd-TiO2. The mass fractions of anatase and rutile were determined from the relative XRD diffraction band intensities, using an anatase sample with a 100% TiO2 crystalline phase as a reference [25]. XRD patterns obtained for the different photocatalysts, are shown in Figure 1. XRD peaks for DP25, anatase and rutile are also given in Figure 1 in order to compare them with those of the Pd-TiO2 photocatalysts. XRD peaks at 25°, 38°, 48°, 54°, 63°, 69°, 70.5° and 75° 2θ diffraction angles were assigned to anatase (101) Figure 2 reports a comparative analysis of the XRD diffractograms for the photocatalysts before and after reduction. One can see a peak at 34° of the 2θ angle scale, which corresponds to (002) reflections of a tetragonal palladium oxide phase [JCPDS . Furthermore, the peaks at 40° and 46° of the 2θ angle scale relate to the Pd° [JCPDS No. 87-0638]. Thus, there is a structural difference in the semiconductor material after the reduction process, with the absence of XRD detectable palladium oxides and the formation of metallic palladium h k l (1 1 1) and (2 0 0).   Furthermore, the average size of the crystallites was calculated based on XRD peak broadening using the Scherrer Equation. The calculated crystallite sizes were between 9 and 14 nm and they are reported in Table 3. Additionally, the lattice constants of the tetragonal anatase unit cells were calculated for the anatase phase (h k l) = (1 0 1). The resulting a = b = 3.7779 and c = 3.4888 parameters at 25° in the 2θ angle scale, showed that pure anatase was present in the photocatalysts [38].

Temperature Programmed Reduction (TPR)
Temperature Programmed Reduction (TPR) is one key parameter that could influence photocatalyst performance. TPR showed that once the photocatalyst was under hydrogen flows, the palladium oxide species were reduced [39], with four TPR peaks being observed. The first negative peak at 68 °C was attributed to the decomposition of palladium β-hydride, which occurred as soon as the hydrogen and palladium came into contact, at the beginning of the analysis. This negative peak at 68 °C may be larger at higher Pd loadings in excess of 0.25 wt% [40], with this being attributed to the higher Pd dispersions. The second broad peak was assigned to palladium oxide reduction. This peak started at 200 °C and was completed at 300 °C. This broad peak was attributed to the broad palladium particle size distribution (18-22 nm) [41], with larger particle sizes moderately increasing the palladium oxide reduction temperature.
Furthermore, as reported in Figure 3, the 0.25 wt%Pd-TiO2 semiconductor showed double peaks in the 400-600 °C range, which were attributed to the Ti +4 ions surface reduction [42]. Similar trends to the ones reported in Figure 3 were found for all doped photocatalysts, with these consisting of 0.25 wt% to 5.00 wt% Pd-TiO2. It should be noted, however, that only the second peak in the 200-300 °C range was considered in all the calculations, in order to establish the amount of reducible palladium. In summary, one can see that for the Pd-TiO2 of the present study, the Pd reduction temperature was above 350 °C, which suggests strong metal-support interactions, potentially leading to high photocatalytic hydrogen production activity [43]. Figure 4 illustrates the UV-visible absorption spectra of the Pd-TiO2 at different metal loadings while applying the Kubelka-Munk (K-M) model, following the Tauc plot methodology. When using the Kubelka-Munk (K-M) method, the bad gap is determined by drawing a straight line in the (α hv) 0.5 versus (hv) plot with α representing the absorption coefficient, h being the Planck constant (6.34 × 10 34 J s/photon) and v denoting the radiation frequency [44]. Following this, the location of the band gap is determined at the straight line intersection with the x-axis. Figure 4 reports the Tauc plots of the doped TiO2 with 0.25 wt%Pd, 0.5 wt%Pd,1 wt%Pd, 2.5 wt%Pd, and 5 wt%Pd loadings. For instance, one can observe in Figure 4, that for the 0.25 wt% Pd-TiO2, a linear extrapolation yields a 2.51 eV band gap. This 2.51 eV band gap corresponds to a 494 nm photon wavelength (λ) given λ = hc/Ebg, with h being the Planck constant (6.34 × 10 34 J s/photon), c being the speed of light under vacuum (3.00 × 10 8 m/s 2 ), and Ebg being the electron band gap. One should note, that 2.51 eV represents a significantly reduced Ebg. with respect to the 2.99 eV band gap for mesoporous TiO2 without Pd, as reported in Table 4. It is also observed in Table 4 that using Pd loadings above 0.25 wt%, yields to a reversed trend in the Ebg, with band gaps increasing steadily instead. In this respect, one could assign the band gap decrease of the 0.25 wt% Pd-TiO2 to the Fermi level changes. These changes can be assigned to the sp-d orbital exchange interactions between the band electrons and the localized d electrons of the Pd 3d ions substituting the Ti 4+ cations. The s-d and p-d exchange interactions give rise to a downward shift of the conduction band edge and an upward shift of the valence band edge, leading to a band gap narrowing [45][46][47][48]. However, at higher than 0.25%wt Pd loadings, it is speculated that the band gap increase is due to the dominant d-d transitions over the sp-d transitions.

Band Gap
Thus, one can consider, that low noble metal loadings (e.g., 0.25%wt Pd-TiO2) facilitate both charge collection and light absorption [49]. Low Pd loadings give rise to localized energy levels in the band gap of the TiO2. In this case, the valence band electrons of the TiO2 are excited at wavelengths longer than 400 nm [50]. Alternatively, excessive noble metal loading may lead to smaller photocatalyst specific surface areas, with larger metal crystallites formed with PdO inclusions [51]. In this respect, XPS confirmed that the PdO presence shields incident photons, blocking light absorption and preventing the generation of semiconductor electron-hole pairs [52].

X-ray Photoelectron Spectroscopy (XPS)
For chemical state identification, quantitative XPS analyses were performed on: a) 0.25 wt%-5 wt% Pd-TiO2 before photoreduction and b) 0.25%Pd-TiO2 after photoreduction. The Pd 3d3/2 and Pd 3d5/2 spin orbital splitting photoelectrons were observed in both the photoreduced and nonphotoreduced 0.25 wt% Pd-TiO2 photocatalysts. Figure 5 reports the photoreduced and non-photoreduced 0.25 wt% Pd-TiO2 photocatalyst XPS peaks. XPS peaks were analyzed via band deconvolution, at the 334.54 eV, 336.38 eV, 339.69 eV, and 341.54 eV characteristic binding energies.   Table 5 reports the observed binding energies for the 0.25 wt% Pd-TiO2 with the full-widths-athalf-maximum (FWHM) and percentual areas. Thus, one can see that according to Table 5 and Figure 5, the 0.25 wt% Pd-TiO2 photocatalyst displays in the XPS, the two most intense peaks, at 334.43 and 339.69 eV. These peaks are assigned to the metallic Pd. Furthermore, there are two other weaker recorded peaks at 336.28 and 341.54 eV, which are attributed to 3d5/2 and 3d3/2 binding energies of the PdO species. Table 5 reports a comparison between the XPS binding energies of the 0.25 wt% Pd-TiO2 photocatalyst before and after 60 min. of near-UV photoreduction. One can see that there is a significantly increased in Pd° content ranging from 49.8% up to 81.7% after UV irradiation, with the reaming PdO amounting to 18.3% only. This increased Pd° is the result of near-UV electrons diminishing the oxidized Pd species.
Furthermore, the XPS analysis of the 0.25%wt Pd-TiO2 photocatalysts, also showed Ti 2p and O 1 s bands at peaks of 454 and 526 eV, respectively (not displayed). These bands were assigned to the titanium oxide.

Macroscopic Irradiation Energy Balance (MIEB)
The MIEB were developed in a carefully selected 6000 cm 3 Photo-CREC Water-II Reactor control volume. This control volume as described in Figure 6, contained uniformly suspended Pd-TiO2 photocatalysts. On this basis, the rate of photon absorption was calculated as [27]: where, Pa = the rate of absorbed photons, Pi = the rate of photons reaching the reactor inner surface, Pbs = the rate of backscattered photons exiting the system, and Pt = the rate of transmitted photons in Einstein/s. It is desirable to obtain a high rate of absorbed photons for photocatalytic processes. For detailed calculations of the MIEB refer to Appendix A.
The Pa (rate of absorbed photons) was calculated as shown in Table 6. One can thus see, that when a 0.25 wt% Pd-TiO2 is used, there is a significant Pa increase versus the Pa value when undoped TiO2 is utilized. However, one can also observe that larger loadings than 0.25 wt% Pd on TiO2 yield a mildly increased Pa with 1.00 wt% Pd on TiO2 giving a Pa maximum. Larger than 1.00 wt% Pd on TiO2 yield a modest Pa decrease, with a consequently diminishing absorption efficiency. In summary, TiO2 photocatalysts doped with Pd considerably augment the absorbed visible irradiation photons, and this is the case when compared to the undoped TiO2.

Hydrogen Production
The Pd-TiO2 photocatalysts of the present study were evaluated in the PCW-II with respect to their ability to enhance hydrogen production, under the following conditions: (a) by utilizing a photocatalyst concentration of 0.15 g/L, (b) by using a 2.0 v/v% of ethanol as organic scavenger and (c) by utilizing a pH = 4 ± 0.05. Figure 7 reports the cumulative hydrogen volume produced, using TiO2 doped with different Pd loadings under visible light. It was shown that Pd-TiO2 semiconductors consistently enhance hydrogen production, with the best performance obtained with the 0.25 wt% Pd. This mesoporous semiconductor may have an increased scavenging effect of photogenerated electrons and therefore, prevent electron-hole pairs recombination [53].  Furthermore, it was also observed that when the metal loading of the Pd-TiO2, was augmented to 0.5 wt% Pd and above, a decreased rate of hydrogen production was obtained. One should notice as well, that the MIEB as reported in Table 6, showed relatively stabilized visible light absorption for various Pd doped TiO2 photocatalysts. Thus, one can conclude that hydrogen production differences cannot be assigned to changes in electron and hole pairs generation [54], but rather to a most effective trapping of electrons in 0.25 wt%Pd-TiO2 than in photocatalysts with larger Pd loadings. Figure 6 also displays that after only 6 h of visible light irradiation, a maximum volume of 4.7 cm 3 STP (standard temperature and pressure) of hydrogen is obtained when using the 0.25 wt% Pd on TiO2. This volume is approximately 4 times higher than the volume produced with undoped mesoporous TiO2.

Precursor Near UV-Light Photoreduction
Palladium is present in a metallic state during the sol-gel photocatalyst preparation. However, palladium can be oxidized during the photocatalyst precursor calcination preparation step. Confirmation of this is given by the X-ray Diffraction Analysis where at 34° (111) of the 2θ angle scale, there is indication of the presence of PdO. As well, the XPS also shows that 50.2% of palladium is present as PdO after photocatalyst precursor calcination as reported in Table 5. Thus and on this basis, one can conclude that palladium species on the semiconductor requires further reduction, to ensure that a substantial amount of palladium species is present as Pd°.
Therefore, a special and additional photocatalyst pretreatment was implemented in the present study, to ensure that most palladium was appropriately reduced to Pd°. Pd° promotes the high photocatalytic activity of TiO2, by generating a Schottky junction between the metal and the photocatalyst. The metal particles trap and store the photogenerated electrons, reducing the rate of the electron hole recombination [55].
With this end and as described in Figure 8, a 15W BLB UV-Lamp was employed to irradiate the prepared semiconductor during 1 h. This PdO photoreduction using near UV light can be described as per Equation (2), with the resulting palladium being present as Pd° on the TiO2 structure.
PdO +2 e − → Pd° E° = 0.915 V (2) It is speculated that photoreduction as per Equation (2), is a very efficient process with photogenerated electrons migrating from the outer TiO2 particle surface to the TiO2 mesoporous inner surface. Formed electrons can reduce the PdO into Pd° [56].
Regarding the present studies, following photoreduction, the near-UV lamp was replaced by a visible light lamp. It was then observed that when the photocatalyst was photoreduced prior to its utilization, this led to an important increase in hydrogen production, under visible light irradiation. Figure 9 displays an enhanced cumulative hydrogen production under visible light in the Photo-CREC Water Reactor II at different loadings (0.25, 0.50, 1.00, 2.50 and 5.00 wt%). It is interesting to see that the same consistent trends were observed in previous studies of our research team, using near-UV light [57]. It has to be noted as well, that the lower the Pd loadings, the higher the hydrogen production. Furthermore, when comparing Figures 7 and 9, it can be observed that the photoreduced Pd-TiO2 phtocatalyst displayed significantly increased hydrogen production rates. Particularly for the 0.25 wt% Pd-TiO2 after photoreduction, the maximum hydrogen volume produced was 8.0 cm 3 STP. This is equivalent to a 1.7 times increased hydrogen production rate. At this low palladium loading, a good metal dispersion of 75% was also observed, according to the chemisorption studies, with a slightly decreased surface area and average pore size [58].
In contrast, when using the 0.50 and 5.00 wt% Pd-TiO2, lower metal dispersions were observed, with larger metal crystallite sizes being detected. This was in line with their diminished photocatalytic hydrogen production activity [59]. To explain these results, one can consider that under visible light and using a Pd-TiO2 photocatalyst, photons are both absorbed and scattered. The MIEB as reported in Section 2.2, showed that higher Pd loadings (2.5 and 5.0 wt%-Pd) do not enhance the absorption of visible light significantly. This phenomenon can be assigned to the presence of larger metal crystallites and TiO2 particle agglomerates. This limits photons from reaching the Pd° active metallic sites and from being absorbed [60]. The opposite of this was observed at lower than 1.00 wt% Pd loadings, where the photon absorption increases, positively impacting the semiconductor photoactivity.
On the other hand, it can be hypothesized as well, that the less effective photoreduction of palladium may occur for 2.5 and 5.0wt% Pd on TiO2, due to the oversupply of noble metal. In this case, layers of PdO could still be present on the TiO2, shielding the TiO2 from light absorption. The formation of such sites could increase the photocatalyst reflectivity leading to visible light scattering [61]. As well, this phenomenon could also be attributed to the partial blocking of semiconductor pores, which may decrease the TiO2 specific surface area, as reported in Table 1.
Given that hydrogen is produced under visible light, the photogenerated holes created by the noble metal react with the organic scavenger ethanol forming byproducts such as acetaldehyde, ethane, CO2 and methane, in a progressive increment as is shown in Figure 10.

Quantum Yield (QY) Evaluation
For the photoconversion of organic species, one can consider a Quantum Yield (QY) parameter. The QY calculation methodology is reported in Appendix D.

Effect of Pd Addition on Quantum Yields
The QY calculation for the TiO2 photocatalysts calculation of the Pt transmitted photons, the Pi incident photons, and the Pbs backscattered photons, as described in Section 2.2. Table 7 and Figure 11 report the QY% for the mesoporous photocatalysts doped with palladium at different metal loadings (0. 25   One can observe in Table 6 that for all the prepared Pd-TiO2 photocatalysts, the QYs% obtained while being irradiated with visible light were in the 0.10-1.13% low range. These low QYs% were assigned to the lack of ability of the Pd-TiO2 photocatalysts to produce hydrogen under visible light with only 49.8 wt% of the loaded palladium as Pd°. However, Table 6 also shows that when the photocatalysts were photoreduced with near-UV irradiation, the QYs% increased, reaching QY% values as high as 1.58%. These results demonstrate the importance of the noble metal photoreduction using near-UV irradiation (photoreduction). One should note that photoreduction was critical to making Pd-TiO2 photocatalysts active under visible light for hydrogen production.
Few papers in the technical literature report QYs of a comparable magnitude. Some authors have used Pd-TiO2 (0.3 wt%) and 3NbTi/Pt-Pd as photocatalysts, achieving maximum QYs between 0.43 and 1.2% [62][63][64]. As well, the absorbed photon rates were measured and calculated using a numerical solution of a radiation equation. These modeled absorption rates may involve significant errors. In this case, as stated before in the present study, the QYs% were determined by using an experimentally evaluated Pa, determined from MIEB macroscopic balances, as reported in Section 2.2.
Furthermore, Figure 12 reports the consistent QY% trends observed for Pd-doped TiO2 photocatalysts as follows: (a) During the first hour of irradiation, the QY% increased progressively until they reached a stable value; and (b) During the six hours of irradiation that followed, the QY% remained unchanged, with the photocatalysts under study exhibiting a stable performance. It can also be observed that there is a significant increase of QY% when using 0.25 and 0.50 wt% Pd-TiO2, whereas higher Pd loadings led to a decrease in the QY%. It is also interesting to see in Figure 12, that the photoreduced semiconductors of the present study display good and stable QYs%, showing their significant ability to produce hydrogen. This photocatalyst stability was also established with 4 consecutive hydrogen production photocatalytic runs, each lasting 6 h or the equivalent of 24 h under visible light irradiation.
Based on the reported results, it is thus anticipated, that further research with mesoporous TiO2 photocatalysts doped with noble metals will be valuable, particularly in the case of Pd-doped TiO2 using ethanol as a scavenger. This will likely lead to stable and efficient photocatalytic processes for hydrogen production via water splitting. These Pd-TiO2 photocatalysts may also present significant cost advantages versus other noble metal dopants such as Ta, Nb or Pt.

Experimental Methods
The photocatalyst of the present study was synthesized using a sol-gel methodology [53]. Different techniques were utilized for its characterization as follows: (a) Specific Aurface Area (BET), (b) Chemisorption, (c) X-ray Diffraction, (d) TPR Analysis, (e) XPS and (f) UV-Vis Absorption. These techniques allowed the determination of the surface area, the pore size distribution and the pore size, the phase composition, the band gap, the metal dispersion, the temperature-programmed reduction and the Pd° crystallite size of the photocatalyst.

Photocatalyst Synthesis
The synthesis of the photocatalyst of the present study via the sol-gel method requires the following: (a) ethanol USP (C2H5OH) obtained from commercial alcohols, (b) hydrochloric acid (HCl, 37% purity), (c) pluronic F-127, (d) anhydrous citric acid, (e) titanium (IV) isopropoxide, and (f) palladium (II) chloride (PdCl2, 99.9% purity). All the reagents were purchased from Sigma Aldrich (Oakville, Ontario, Canada). Detailed information about the photocatalyst synthesis methodology is reported in Appendix B.
According to the methodology proposed by Das et al. [65], 20 mL ethanol was acidified with 1.65 g of hydrochloric acid. This was followed by the addition of 1 g of copolymer pluronic F-127. The mixture was stirred for 30 min until complete dissolution was reached. 0.315 g of anhydrous citric acid was mixed with 1 mL of water and added to the initial solution. The resultant mixture was stirred for 2 h. Following this, 1.42 g of titanium IV isopropoxide in ethanol added dropwise [29].
Furthermore, and after this, PdCl2 was added to the resulting sol-gel suspension, given its solubility under the conditions of the sol-gel suspension. PdCl2 was incorporated at different  concentrations to achieve 0.25-5.00 wt% Pd loadings. The resulting sol-gel suspension was stirred for 24-48 h and then calcined under an air atmosphere at 500 °C for 6 h [66]. This allowed the copolymer to be removed, with an ordered mesoporous titanium framework being formed [67,68]. However, and considering that palladium oxidizes during the calcination step, the resulting photocatalyst had to be reduced in a subsequent step. To accomplish this, the synthesized semiconductor was placed in a flow reactor unit under an atmosphere of 1 cm 3 /s of Ar/H2 (g) (90/10%, Praxair) at 500 °C for 3 h [69]. Given that this reduction with hydrogen was incomplete, a further and critical Pd-TiO2 photoreduction step was implemented in the Photo-CREC Water-II (PCW-II), by exposing the photocatalyst to near-UV light at room temperature for 60 min.

Equipment
The Photo-CREC Water-II (PCW-II) Reactor is an innovative unit for hydrogen production. It is a slurry batch reactor with a total volume of 6 L. As seen in Figure 13, it is composed of: (1) A 15-W fluorescent visible light lamp, (2) A Pyrex glass inner tube where the lamp is placed, (3) A black polyethylene outer tube, (4) A centrifugal pump, (5) Two sampling ports where the photocatalyst suspension is always kept sealed under agitation, one for the liquid phase and the other one for the gas phase, (6) A hydrogen storage tank, and (7) Silica windows for irradiation measurements [70]. Refer to Appendix C for a detailed lamp characterization.

Photocatalyst Characterization
Nitrogen adsorption and desorption isotherms at −195 °C were measured using a Micrometrics ASAP 2010 Surface Area and Porosity analyzer (Norcross, GA, U.S.A). The samples were initially degassed in a vacuum at 300 °C for 3 h. The BET-surface area was determined from the BET plot. The pore volume was measured at a relative pressure of P/Po = 0.99. To estimate the pore size distribution, the N2 physisorption BJH (Barrett-Joyner-Halenda) method was used [71].
Pulse chemisorption with the Micromeritics AutoChem II Analyzer (Norcross, GA U.S.A) was utilized to determine the fraction of dispersed metal. On the other hand, the XRD spectrum for each material was measured in a Rigaku Rotating Anode X-ray Diffractometer (Rigaku, Auburn Hills, MI, United States) rated at 45 kV and 160 mA, to identify the crystalline phases of the materials. The diffractograms were taken in the 2θ angle scale (20°-80° range), with a step size of 0.02° and a dwell time of 2 s/step.
The H2 Temperature Programmed Reduction (TPR) analyses of the Pd-TiO2 photocatalysts were evaluated in a Micromeritics AutoChemII Analyzer. 250 mg of the photocatalyst were placed in the U-tube with a gas reduction mixture of Ar/H2 (g) (90/10%). The reaction temperature was kept within a 0° to 600 °C range with a flow rate of 50 mL min −1 . The amount of H2 consumed during the reduction was measured by a thermal conductivity detector (TCD) [42].
To determine the band gap, a UV-VIS-NIR Spectrophotometer (Shimadzu UV-3600, Nakagyoku, Kyoto, Japan) was used, with the BaSO4 as a reference [48]. By using the Kubelka-Munk (K-M) methodology, Tauc plots were developed, in order to establish the corresponding band gaps for each photocatalyst [72]. Furthermore, to ascertain the composition and the oxidation/reduction state of palladium, an X-ray Photoelectron Spectroscopy (XPS) analysis was utilized [73].
Furthermore, to measure the irradiation of the visible light lamp inside the Photo-CREC Water-II, the Stellar Net EPP2000-25 Spectrometer (StellarNet Inc., StellarNet, Inc.,Tampa, Florida, U.S.A.) was used. The light source used was a fluorescent mercury Philips lamp (15 W) with an output power of 1.48 W and an average emitted photon energy of 274.5 kJ/photon mole.

Hydrogen Production
The Photo-CREC Water-II Reactor was used to evaluate the Pd-TiO2 photocatalyst utilized during the water splitting runs. 6000 mL runs of water were used as the main reagent. As well, ethanol served as an organic scavenger. The pH was set to 4 ± 0.05 with H2SO4 [2M]. A BLB UV lamp was turned on for 60 min to photoreduce the PdO present in the semiconductor. After the first hour, the UV lamp was replaced by the Phillips Visible light lamp to initiate the reaction. Then, the water splitting run continued during (6) hours of irradiation [69].
The photocatalyst was loaded at a concentration of 0.15 g/L and sonicated for 10 min to avoid particle agglomeration. This ensured a homogeneous distribution of the photocatalyst throughout the reactor. To avoid any undesirable oxidation reaction, an argon flow was circulated for 10 min to keep an inert atmosphere.

Conclusions
Palladium-doped TiO2 photocatalysts, photoreduced with near-UV irradiation prior to their utilization, are shown to favor hydrogen production under visible light. The structural properties of the best performing photocatalysts (0.25 wt%Pd-TiO2) were established using X-ray Diffraction (XRD), BET Surface Area (SBET), Temperature Programmed Reduction (TPR), Chemisorption and Xray Photoelectron Spectroscopy (XPS), and UV-Vis Reflectance Spectroscopy. It was shown that the near-UV photoreduced 0.25 wt%Pd-TiO2 enhances hydrogen formation, reaching in 6 h, 8.0 cm 3 STP of produced hydrogen. The 0.25 wt% Pd on TiO2 also displays in all cases, a zero-order kinetics, with a highest QY% of 1.58%. These positive findings set favorable prospects for further research, including the development of a kinetic model which's availability is considered critical for the scaling up of a Photo-CREC-Water II unit for hydrogen production. The various terms in Equation (1) can be calculated as follows: (a) Pi is the rate of photons reaching the slurry suspension:  And Pa-wall is the rate of photons absorbed by the inner Pyrex glass surface.
(b) Pbs represents the difference between Pi and Pt/c→0+. It is the rate of photons transmitted at a photocatalyst concentration approaching zero [28] as follows: The described macroscopic balances were originally developed for near-UV irradiation in photocatalytic reactors [70]. However, given that these macroscopic balances are not photon wavelength dependant, they were extended to visible light irradiation.

Appendix B. Photocatalyst Synthesis-EISA Method
In this research, the photocatalyst synthesis uses the Evaporation-Induced-Self-Assembly (EISA) Method. This approach has been utilized in the preparation of semiconductors when adding a template to control the material pore size [74]. This methodology was applied in the present study, given that: (a) It is easy to use, (b) it allows close control of the pore size distribution with the selected copolymer template, (c) it enhances the specific surface area, (d) It is relatively inexpensive, and e) it does not require special equipment [75]. Figure A1 describes the photocatalyst synthesis based on a sol-gel method with the use of a polymeric template [pluronic F-127 (PEO106PPO70PEO106)], a metallic precursor [titanium IV isopropoxide] and an active metal [PdCl2] precursor. The mixing of these components leads to the formation of micelles and subsequently to a mesoporous structure. Subsequent calcination and reduction yield: (a) An adequate template removal, and (b) The formation of mesopores with highly dispersed palladium nanoparticles [76]. Figure A1. Description of the Four Consecutive Steps during Pd-Doped TiO2 Photocatalyst Preparation: (a) Ethanol enters the interface between the hydrophilic (PEO) and hydrophobic (PPO) chains and attaches to the hydrophobic core, (b) Ahydrophobic block (PPO) is placed in a central location surrounded by the PEO tails forming micelles, (c) The PEO tails become attached to the TiO2, leaving a hybrid outer layer with a dominant TiO2 composition, and (d) Calcination of the resulting photocatalyst precursor yields TiO2 with a 3D mesoporous structure [57]. Figure A2 reports the spectrum of the polychromatic BLB Ushio near-UV lamp, with an observed output power of 1.61 W and a 325.1 kJ/mole of photons average.   The axial distribution of the radiative flux was determined using the Stellar Ney EPP2000-25 Spectrometer. Figure A4 displays the observed axial visible lamp radiation distribution with the radiation significantly decreasing, in the lower and upper axial lamp positions [77].

Appendix D. Quantum Yield Calculation
The QY is defined as the formed H· molar rate over the absorbed photons molar rate. Thus, it describes the extent of the photochemical utilization for hydrogen production of absorbed photons [34]. The QY is calculated according to the following equation: where, • is the rate of moles of hydrogen radicals formed at any time during the photocatalyst irradiation.
To calculate the QY, the Pa or the moles of absorbed photons is needed. The Pa was determined by using the MIEB in the Photo-CREC Water-II Reactor [69], as described in Section 2.2.