Inﬂuence of Photo-Deposited Pt and Pd onto Chromium Doped TiO 2 Nanotubes in Photo-Electrochemical Water Splitting for Hydrogen Generation

: Hydrogen (H 2 ) is considered as an ideal fuel for the future. The photo-electrochemical (PEC) water splitting employing semiconducting materials and induced irradiation, preferably of solar spectrum, presents a viable route for H 2 production. In this work, self-ordered chromium-doped TiO 2 nanotube (CT) was fabricated using in-situ electro-anodization. CT surface modification was then performed by photo-deposition of Pt and Pd particles, producing Pt-CT and Pd-CT catalysts, respectively. Their morphological features, crystallinity, surface composition, and optical absorption have been inspected by scanning electron microscopy (SEM), X-ray diffraction (XRD), energy dispersive X-ray spectroscopy (EDX), Raman, and UV–vis absorption spectroscopy. Linear sweep voltammetry, chronoamperometry, and open circuit potential methods have been applied to study PEC activities of Pt-CT and Pd-CT catalysts in a form of electrodes. It was found that Pt-CT and Pd-CT electrodes possess excellent photo-generated electron/hole ( e − / h + ) separation and transport properties. The enhanced photocurrent responses of 4 and 3 times more than that of CT are revealed for Pt-CT and Pd-CT, respectively. The activity of as-prepared Pt-CT and Pd-CT catalysts was then tested for H 2 generation. The maximum amount of the evolved H 2 followed decreasing order: 1.08 > 0.65 > 0.26 mL cm − 2 h − 1 for Pt-CT, Pd-CT, and CT electrodes, respectively, clearly showing the positive contribution of photodeposited (nano)particles onto CT surface.


Introduction
Hydrogen (H 2 ) is alternative to fossil fuels, superior to renewable energy sources such as wind and solar; H 2 is storable and transportable fuel, and could be easily accessible whenever required [1,2]. There are different methods for H 2 production; such as the petroleum cracking, hydrocarbon reforming, the electrolysis of water, photocatalytic and photoelectrochemical (PEC) water splitting [3,4]. Ever since the discovery of water splitting on TiO 2 by Fujishima and Honda [5] semiconductor-based photocatalysts have triggered broad scientific interest. PEC water splitting is considered as the most environmentally friendly and cost-effective technology to produce H 2 [6]. A wide range of materials, such as TiO 2 , ZnO, WO 3 , ZrO, etc. have been introduced and explored for PEC water splitting process, however, TiO 2 still remains the basic material in the most of applications [7][8][9]. The plausible explanation can be found in its low price, high chemical stability, suitable band-edge positions for required reactions, and relatively long electron lifetime. However, TiO 2 has several limitations, such as fast recombination of photogenerated e − /h + pairs and

Characterization
Our previous results revealed that highly ordered TiO 2 nanotube arrays can be synthesized with the electro-anodization method. However, in the case of CT, amount of K 2 CrO 4 in in-situ electro-anodization can change the morphology of the obtained material. For example, when K 2 CrO 4 concentration > 15 mM was used, a porous film instead of nanotube arrays was observed (SEM images of pristine CT can be found in the ref [12]). Hence, in this work, CT electrode was synthesized according to the optimal conditions established in our previous study [12]. As can be seen from both Figures 1 and 2, highly ordered and uniform crystalline CT nanotubes with the small layer of photodeposited platinum and palladium were successfully fabricated with the average 130 nm in diameter (marked with dashed lines in Figures 1B and 2B), 30 nm in wall thickness, and ∼8-10 µm in length. It should be mentioned that the deposited Pt and Pd did not change the morphology of CT (Figures 1 and 2). A can be seen from Figure 1C,D), showing Pt-CT-2, Pt particles are fairly homogenously distributed over the surface of the CT and are of spherical shape. Accordingly, it can be concluded that shorter period of time used for photo-deposition (e.g., 20 min as in the case of Pt-CT-2) yielded isolated Pt particles onto CTs ( Figure 1D). On the other hand, the increase in photo-deposition-time > 60 min led to a partial agglomeration of Pt particles onto the CT, as can be seen for images corresponding to Pt-CT-4 ( Figure 1E,F), as well as those for Pt-CT-5, and Pt-CT-6 ( Figure S1, Supplementary Materials). In the case of as-prepared Pd-CT materials ( Figure 2 and Figure S2, Supplementary Materials), interesting effects regarding the time employed for photo-deposition can be observed, particularly when comparing to Pt photo-deposition in the same time intervals. Namely, as can be approximated comparing the same photo-deposition time intervals for Pt and Pd, it seems that more Pd particles were present onto CTs. The example of such effect can be observed when comparing Figures 1A and 2A, presenting Pt-CT-1 and Pd-CT-1, respectively, at the same magnification. Accordingly, larger Au particles and sintering were observed on different substrates [30]. However, comparing to Au, Pt binds more strongly to metal oxides and small (nano)particles could be deposited at the surface of substrate. Since the photo-reduction process of Pd (II) in aqueous solution is quite fast [28], more agglomeration was observed comparing to Pt decorated surface in our study. It should be noted that larger or agglomerated particles on the top of CT surface demonstrated that deposited noble metals are not able to penetrate the interlayer and/or concentrate on the outer walls. In such case, agglomerated structures can still act as centers of charge recombination, however, simulta-  The crystal structures of as-prepared electrodes were investigated by XRD. Figure 3A shows the XRD patterns of CT, Pt-CT-6, and Pd-CT-6. As can be seen, in all cases anatase phase is dominant, exhibiting major peaks at 25.3°, 38.0°, 48.1°, 53.9° and 70.3° (2θ) which were well indexed to the lattice planes of (101), (004), (200), (105) and (220), respectively (ICDD-JCPD:01-086-1156) [13], indicating that TiO2 pertains mainly to anatase. The plausible explanation why characteristic peaks corresponding to Cr are not detected in none of tested materials might be due to incorporating into the TiO2 crystal lattice or rather small size of dopant introduced by used electro-anodization method. Our previous work it was demonstrated that higher concentration of Cr did not change the XRD patterns of Ti nanotubes, i.e., TiO2 phase [12]. However, the presence of Cr has recently been confirmed in our previous research by XPS analysis [31]. The representative diffraction peaks In the case of latter material, a distribution of small light dots (Pd particles) can be clearly seen all-over CT surface, which is not so obvious in the case of photo-deposited Pt particles. However, such higher amount deposited onto CT surface presumably resulted with increased agglomeration rate ( Figure 2E, marked with yellow circles). The agglomeration effect is particularly obvious in the case of Pd-CT-6 ( Figure S2, Supplementary Materials; images with the lower magnification, i.e., 5 µm) where presence of larger agglomerates (in size of 3 × 4 µm) are clearly noticeable. Differences in amount and size of deposited Pt and Pd can be plausibly explained by their different interactions due to differences in atomic mass and density of both metals. Similar effect was observed by Pisarek et al. [24] using magnetron sputtering technique to decorate titania nanotubes surface by Pt and Pd nanoparticles. Additional confirmation for effects observed can be also found in other studies on the decoration of TiO 2 -based materials with Pt and Pd [25,26]. It should be emphasized that the agglomeration occurred at both Pt-CT and Pd-CT materials prepared in this study is much lower than that obtained in our previous study at comparable conditions using CT substrate for photo-deposition of Au and Ag as surface-decorative co-catalysts [13]. Such findings are in good agreement with the literature [27][28][29][30]. Hence, depending on the type of metals, metals cluster size as well as the agglomeration rate are different due to parameters such as binding energy between nanoparticles and between nanoparticles and a substrate (CT in our case), photo-reduction speed, and nanoparticle nucleation and growth rate. For example, the interaction of Au-Au is much stronger than that of Au-substrate made of metal oxides [27].
Accordingly, larger Au particles and sintering were observed on different substrates [30]. However, comparing to Au, Pt binds more strongly to metal oxides and small (nano)particles could be deposited at the surface of substrate. Since the photo-reduction process of Pd (II) in aqueous solution is quite fast [28], more agglomeration was observed comparing to Pt decorated surface in our study. It should be noted that larger or agglomerated particles on the top of CT surface demonstrated that deposited noble metals are not able to penetrate the interlayer and/or concentrate on the outer walls. In such case, agglomerated structures can still act as centers of charge recombination, however, simultaneously can cover the top of the nanotubes, consequently decreasing the efficiency of PEC for H 2 production.
The crystal structures of as-prepared electrodes were investigated by XRD. Figure 3A shows the XRD patterns of CT, Pt-CT-6, and Pd-CT-6. As can be seen, in all cases anatase phase is dominant, exhibiting major peaks at 25.3 • , 38.0 • , 48.1 • , 53.9 • and 70.3 • (2θ) which were well indexed to the lattice planes of (101), (004), (200), (105) and (220), respectively (ICDD-JCPD:01-086-1156) [13], indicating that TiO 2 pertains mainly to anatase. The plausible explanation why characteristic peaks corresponding to Cr are not detected in none of tested materials might be due to incorporating into the TiO 2 crystal lattice or rather small size of dopant introduced by used electro-anodization method. Our previous work it was demonstrated that higher concentration of Cr did not change the XRD patterns of Ti nanotubes, i.e., TiO 2 phase [12]. However, the presence of Cr has recently been confirmed in our previous research by XPS analysis [31]. The representative diffraction peaks for the face-centered cubic crystal structure of Pd, namely (111), (220) and (222) [32], were observed at 40 • , 69 • and 87 • (2θ) in the XRD patterns of Pd-CT-6 sample. Similarly, in the case of Pt decorated CTs (Pt-CT-6) diffraction peaks at 40 • and 69 • (2θ), corresponding to the face-centered cubic phase of Pt (220) and Pt (222) [33], respectively, can be observed. Such findings indicate that Pt and Pd at CT surface are present in metallic form.

Photo-Electrochemical Characterization
PEC water splitting aiming at the production of H2 with large positive change in the Gibbs free energy (ΔG° = + 237.2 kJ mol −1 ) is a thermodynamically uphill reaction. In addition, H2 production is a low efficiency process due to the high recombination rates of photogenerated e − /h + pairs in the most cases of applied photocatalytic materials. In order to increase the process efficiency, a sacrificial agent (usually various alcohols) is usually The interactions between co-catalysts (Pt or Pd) and substrate (CT) can have an influence on the size and amount of deposited co-catalysts as well as electron charge transfer between Pt and Pd and CT which may lead to the substantial changes in PEC activity. To understand the interactions between Pt, Pd, and substrate material, the Raman spectra of Pt-CT, Pd-CT, and pristine CT were compared (Figure 3(B1,B2)). The peaks observed for pristine CT at 153, 401, 526 and 643 cm −1 , although latter three are of weaker intensity, are the characteristic Raman bands of anatase which can be assigned to E g , B 1g , doublet of A 1g + B 1g and E g modes of the anatase phase, respectively [34]. The strongest E g mode at 153 cm −1 , arising from the extension vibration of the anatase structure, is well resolved, clearly indicating anatase phase in pristine CT [34,35]. Comparing the Raman spectra of selected Pt-CT and Pd-CT samples at shortest and longest period of time used for the photo-deposition (Pt-CT-1 and Pt-CT-6, and Pd-CT-1 and Pd-CT-6) with the pristine CT showed shifting of peaks in the case of surface decorated materials. Namely, the Raman peaks at 153 and 520 cm −1 of pristine CT shifted towards higher wavenumbers (159 and 526 cm −1 , respectively) in all Pt-CT and Pd-CT recorded samples. In addition, the magnitude of the titanium peaks increased after surface-decoration by Pt and Pd particles. However, it should be emphasized that peaks magnitude decreased with increasing of photo-deposition time in both cases, which is particularly evident in the case of CTs surface decorated with Pd ( Figure 3(B1,B2)). Since Raman spectra demonstrate the existence of chemical and/or physical interactions between Pt and Pd particles with pristine CT, such effect can be assigned to changes of vibrational modes of anatase structure at the CT surface by deposited co-catalysts. The EDX spectra in Figure 3(C1,C2) provide insight in the surface composition of surface-decorated CT with Pt and Pd particles. Hence, the obtained results; high Ti and O content, clearly exhibit the existence of TiO 2 lattice in synthesized nanotubes, while lower Cr content indicates its presence in CTs, although XRD ( Figure 3A) did not reveal its presence in a known crystal polymorph. The detected Pt and Pd clearly show their presence in the surface-decorated electrodes. Furthermore, the presence of carbon, nitrogen, potassium, fluorine, and chlorine in the spectrum of EDX can be assigned to leftovers of precursors used in synthesis and photo-deposition. EDX mapping of the Pt-CT-6 and Pd-CT-6 electrodes, showed in Figures S3 and S4 (Supplementary Materials), demonstrates that noble metals used for CT surface-decoration (Pt and Pd) are uniformly dispersed on the top of CT substrate, in spite of several agglomerates observed through SEM analysis (Figures 1 and 2 and Figures S1 and S2, Supplementary Materials). Generally, particle size and distribution of particles on the surface of substrate (in our case CT) could have a strong influence on the PEC performance [36,37].
The optical properties of all as-prepared Pt-CT and Pd-CT materials were analyzed by recording their UV-Vis spectra in the range of 250-600 nm ( Figure 3D), Figures S5 and S6 (Supplementary Information)). A noticeable shift in the absorption edge of CT with photodeposited Pt and Pd co-catalysts comparing to the pristine CT towards the visible-light region can observed in the normalized UV-vis spectra (Figure 3(D1,D2)). It was found that both Pt-CT and Pd-CT, regardless the time used for the photo-deposition, increased absorb light in the whole UV-vis range, presumably as a consequence of local surface plasmon resonance (LSPR) effect caused by deposited Pd and Pt particles. The optical bandgaps values were estimated by employing the Tauc's model [38]. The same calculation procedure was already used for TiO 2 nanotubes and pristine CT [12], yielding bandgap energy values of~3.2 eV and~2.82 eV, respectively [12]. As can be seen, the surface decoration with Pt and Pd resulted with more red-shifting and consequently further decrease of bandgap energy of as-prepared electrodes; to values of 1.8-2.2 eV for Pt-CT and 2.1-2.8 eV for Pd-CT electrodes (Figure 3(D1,D2)) and Figures S5 and S6 (Supplementary Materials)). Such results indicate on their ability to effectively utilize visible part of solar spectrum.

Photo-Electrochemical Characterization
PEC water splitting aiming at the production of H 2 with large positive change in the Gibbs free energy (∆G • = + 237.2 kJ mol −1 ) is a thermodynamically uphill reaction. In addition, H 2 production is a low efficiency process due to the high recombination rates of photogenerated e − /h + pairs in the most cases of applied photocatalytic materials. In order to increase the process efficiency, a sacrificial agent (usually various alcohols) is usually required, enabling the reactions with holes while the photogenerated electrons would remain yielding H 2 production. Since our previous research revealed that EG (with ∼5 vol%) showed enhancement in H 2 production rates [13], the same sacrificial agent was also used in this work.
The photo-sensitivity of all samples was investigated by measuring LSV (Figure 4 and  Accordingly, it seems that Pd-CT electrodes have more potential for EG oxidation which can be of large interest for studying the EG electro-oxidation in fuel cells application [39]. However, that was not the subject of this study, thus it was not studied in details here. Thus, we investigated their behavior using LSV measurement in the same system, but without EG presence ( Figure S10 (Supplementary Information)), and results revealed the absence of anodic peak in all cases. It should be mentioned that obtained results are in good agreement with literature where Pd-based materials showed superior photo-electrocatalytic activity for EG oxidation at alkaline conditions comparing to their analogues using Pt nanoparticles [9,40]. On the other hand, it is well known that Pt-based electrocatalysts exhibit an excellent performance for EG oxidation in acidic media [41]. Figure 5 shows the CA curves for Pt-CT and Pd-CT electrodes at 1.5 versus RHE in a solution of KOH (1 M) with 5%V EG. During dark period (0-100 s), tested electrodes mostly did not show current density response, indicating that as-prepared electrodes would act as photocatalysts primarily. After switching visible light illumination on, an initial rapid increase in current density can be observed, followed by decreasing within short period of time (50 s) reaching afterwards relatively steady-state conditions, indicat- Accordingly, it seems that Pd-CT electrodes have more potential for EG oxidation which can be of large interest for studying the EG electro-oxidation in fuel cells application [39]. However, that was not the subject of this study, thus it was not studied in details here. Thus, we investigated their behavior using LSV measurement in the same system, but without EG presence ( Figure S10 (Supplementary Information)), and results revealed the absence of anodic peak in all cases. It should be mentioned that obtained results are in good agreement with literature where Pd-based materials showed superior photo-electrocatalytic activity for EG oxidation at alkaline conditions comparing to their analogues using Pt nanoparticles [9,40]. On the other hand, it is well known that Pt-based electrocatalysts exhibit an excellent performance for EG oxidation in acidic media [41]. Figure 5 shows the CA curves for Pt-CT and Pd-CT electrodes at 1.5 versus RHE in a solution of KOH (1 M) with 5%V EG. During dark period (0-100 s), tested electrodes mostly did not show current density response, indicating that as-prepared electrodes would act as photocatalysts primarily. After switching visible light illumination on, an initial rapid increase in current density can be observed, followed by decreasing within short period of time (50 s) reaching afterwards relatively steady-state conditions, indicating that all as-prepared electrodes possess superior long-term stability. Such effect proved the photo-activity of as-prepared electrodes. Figure 5A,C show that the current density of undoped TiO 2 nanotubes increased when Cr was introduced in the TiO 2 nanotube structure, forming CT; from 0.035 mA cm −2 to 0.23 mA cm −2 , respectively. Such results speak in favor to the increased activity under visible light, as well as decreased charge recombination [12]. Generally, the photocurrent density of Pt-CT electrodes (0.49-0.92 mA cm −2 ) is higher than that of Pd-CT electrodes (0.39-0.72 mA cm −2 ). It should be noted that the highest obtained photocurrent density was 0.92 mA cm −2 which was recorded for Pt-CT-3 among all tested electrodes, while Pd-CT-5 showed maximum among Pd-decorated CTs (0.72 mA cm −2 ) ( Figure 5A,C), which is actually 4 and 3 times more than that of pristine CT (0.23 mA cm −2 ) [13]. Moreover, the photocurrent response was also repeatable during on-off cycles upon light illumination ( Figure 5B The OCP transients were also measured for all as-prepared Pt-CT and Pd-CT electrodes in order to reproduce the catalytic conditions without an externally applied bias. As can be seen in Figure 5E,F, during dark conditions in the beginning of OCP experi- The OCP transients were also measured for all as-prepared Pt-CT and Pd-CT electrodes in order to reproduce the catalytic conditions without an externally applied bias. As can be seen in Figure 5E,F, during dark conditions in the beginning of OCP experiments, lasting for 15 min, electrodes reached stable potentials. Upon switching the light on, the potentials rapidly shifted toward more negative values, reaching after certain period stable value. Such behavior, typical for n-type semiconductors, may be assigned to the injection of photogenerated e − by Pt-CT and Pd-CT electrodes into the counter electrode. During the last period within OCP experiments (~9 min) the light was switched off, and the potentials positively shifted and reached near initial values in several minutes (in most cases < 3 min). Such effect indicates on the electron storage properties of as-prepared electrodes. After the light was turned off, OCP decay kinetics reflects the charge recombination kinetics. Therefore, the fitting of a bi-exponential function to OCP decay profiles was employed with two time constants (Equation (1)) to estimate the average recombination lifetime [42]: where V, t, A 1 and A 2 , τ 1 , τ 2 , and τ m are potential decay, time, fitting constants, exponential lifetime component for band-to-band and band-to-surface state band recombination processes, and the harmonic mean of decay lifetime, respectively. The total half-lives are equal to log(2τ m ), which are calculated for all as-prepared Pt-CT and Pd-CT electrodes and summarized in Table S1 (Supplementary Material). Apparently, OCP decay of Pt-CT electrodes is much slower comparing to that of Pd-CTs, indicating that charge recombination at the surface of CT is more efficiently suppressed with the deposition of Pt than with Pd particles. The EIS measurement conducted at the potential of 0.2 V vs. Ag/AgCl has also been employed to study electrochemical reaction process induced by Pt-CT and Pd-CT electrodes. As can be observed (Figure 6), the impedance arc diameter of studied electrodes under light illumination was smaller than that obtained under dark. Furthermore, Pt-CT-3 displayed the smallest impedance arc diameter. Presence of Pt as co-catalyst yielded decrease of electron transfer resistance more than in the case of Pd particles. Accordingly, it can be concluded that Pt-CT electrodes have better electrical conductivity under light illumination, which can be assigned to the surface plasmon resonance effect and the better charge transfer.  Table S1 (Supplementary Material). Apparently, OCP decay of Pt-CT electrodes is much slower comparing to that of Pd-CTs, indicating that charge recombination at the surface of CT is more efficiently suppressed with the deposition of Pt than with Pd particles. The EIS measurement conducted at the potential of 0.2 V vs. Ag/AgCl has also been employed to study electrochemical reaction process induced by Pt-CT and Pd-CT electrodes. As can be observed (Figure 6), the impedance arc diameter of studied electrodes under light illumination was smaller than that obtained under dark. Furthermore, Pt-CT-3 displayed the smallest impedance arc diameter. Presence of Pt as co-catalyst yielded decrease of electron transfer resistance more than in the case of Pd particles. Accordingly, it can be concluded that Pt-CT electrodes have better electrical conductivity under light illumination, which can be assigned to the surface plasmon resonance effect and the better charge transfer.

PEC Water Splitting for H2 Evolution
The utilization of noble metals has been broadly employed for the surface decoration of semiconducting materials to efficiently perform water splitting aiming at H2 generation due to the following reasons: (i) the LSPR effect, which facilitates harvesting of solar irra-

PEC Water Splitting for H 2 Evolution
The utilization of noble metals has been broadly employed for the surface decoration of semiconducting materials to efficiently perform water splitting aiming at H 2 generation due to the following reasons: (i) the LSPR effect, which facilitates harvesting of solar irradiation to generate and transfer plasmatic carriers; and (ii) enabling the trapping of conduction band electrons, thus boosting the charge separation. PEC water splitting with Pt-CT and Pd-CT electrodes in this study was conducted using an inverted burette setup ( Figure 7C) and the evolved H 2 (Figure 7) was quantitatively measured by electrolyte displacement in an inverted burette. As can be seen (Figure 7), Pt-CT electrodes produced much more H 2 than Pd-CT electrodes at comparable conditions; i.e., photo-deposition time used for surface decoration with chosen noble metals. The maximum amount of the evolved H2, 1.08 mL cm −1 h −1 was obtained with Pt-CT-3, the electrode which showed the highest photocurrent density among all studied ( Figure 7A). On the other hand, the maximum evolved H2 amount by Pd-CTs, was 0.65 mL cm −1 h −1 (1.66 times lower than that obtained by Pt-CT-3), achieved by Pd-CT-5. It is worth to note that Pt-CT and Pd-CT electrodes synthetized in this work provide higher amount of the evolved H2 than some of materials used in other studies (Table 1).  The maximum amount of the evolved H 2 , 1.08 mL cm −1 h −1 was obtained with Pt-CT-3, the electrode which showed the highest photocurrent density among all studied ( Figure 7A). On the other hand, the maximum evolved H 2 amount by Pd-CTs, was 0.65 mL cm −1 h −1 (1.66 times lower than that obtained by Pt-CT-3), achieved by Pd-CT-5. It is worth to note that Pt-CT and Pd-CT electrodes synthetized in this work provide higher amount of the evolved H 2 than some of materials used in other studies (Table 1). Generally, when two materials with different work functions (ϕ m ) are in contact, a Schottky barrier is created at the junction, acting as an energy barrier for electron transfer. The electron transfer is facilitated from a material of lower work function, e.g., in our case TiO 2 (4.2 eV), to one that has a higher work function, e.g., both Pt and Pd in our case. Since Pt exhibits higher work function (5.12 eV < ϕ m < 5.93 eV) than Pd (5.12 eV < ϕ m < 5.35 eV) [48,49], it provides low overpotential for water splitting and consequently H 2 evolution, thus Pt has been showed as more effective co-catalyst than Pd when used for surface decoration of CTs. The stability of Pt-CT-3 and Pd-CT-5 electrodes, which showed the highest effectiveness in H 2 generation among Pt-CT and Pd-CTs, respectively, was examined for PEC water splitting during three cycles of 60 min. As can be seen in Figure 7C,D, there are no significant decreases in activity during consecutive cycles, which proves the good stability of the as-prepared electrodes in the PEC water splitting.

Fabrication of Nanocomposites (CT) Using In-Situ Electro Anodization
Titanium plates of 1 mm thickness (1 cm × 3 cm) were used to grow titania nanotubes). The addition of an appropriate amount of K 2 CrO 4 as chromium source in the electrolyte, resulted with the fabrication of CT electrode using in-situ electro anodization, which enable the penetration of Cr into the Ti structures [12]. Prior to anodization, Ti plates were mechanically polished by different grit size of silicon carbide papers (SiC, Tehran, Iran), and thereafter cleaned by rinsing with deionized water. Afterward, the plates were chemically etched with a mixture of HF:HNO 3 :H 2 O in a volume ratio of 1:4:5 for 30 s, and thereafter washed with deionized water. In-situ electro-anodization was performed at room temperature in a solution of EG containing 0.005 M K 2 CrO 4 , 0.1 M NH 4 F, and 1 mL deionized water at 60 V for 6 h. Ti and a graphite electrode were used as the anode and cathode electrodes, respectively. After anodization, Ti electrodes were washed with deionized water and dried at room temperature. Ti electrodes were annealed at 400 • C for 2 h with a ramp of 2 • C min −1 using a F3L-1720 furnace (AZAR Furnaces, Tehran, Iran) to achieve the materials of high crystallinity.

Photo-Deposition Process
Surface decoration of CT with Pt and Pd was accomplished using the photo-deposition method. Solutions were consisted of the appropriate concentrations of metal chloride precursors (K 2 PtCl 6 and PdCl 2 ). Since these components are slightly soluble in water, an appropriate amount of HCl was used to dissolve them. In a typical experiment, 2 mL ethanol as a sacrificial donor was added to 18 mL of the metal chloride solutions. Photo-deposition was performed by illuminating the solution of metal chlorides in which electrodes were immersed in with a high-pressure mercury lamp (400 W, Zhejiang, China). The electrodes were exposed to such process for 10, 20, 40, 60, 100 and 120 min irradiation, and thereafter washed with deionized water. Accordingly, obtained electrodes are denoted onward throughout the manuscript as Pt-CT-1, Pt-CT-2, Pt-CT-3, Pt-CT-4, Pt-CT-5, and Pt-CT-6 corresponding to illumination time of 10,20,30,40,60, and 120 min, respectively. The same analogy is made in the case of Pd-decorated CT electrodes.

Characterization of Materials
The surface morphology and microstructure of as-prepared materials, as well as the surface elemental analysis were studied using a field emission scanning electron microscope equipped with an energy dispersive X-ray (EDX) mapping port Philips XL30 (Eindhoven, The Netherlands). In order to determine phase composition, X-ray diffraction (XRD) patterns were obtained using Philips X'Pert system (Enscheda, The Netherlands). The data were recorded in the 2θ range of 20 • -90 • . UV-Vis spectra were recorded on a JASCO V-570 UV-visible spectrophotometer (Portland, OR, USA) to study of light absorption properties of the samples. In addition, the Raman spectra of the samples were obtained using TakRam N1-541 (Tehran, Iran) at a laser power of 50-200 mW.

Photo-Electrochemical Measurement
All electrochemical and PEC measurements were carried out in a quartz three electrodes cell in 1 M KOH solution containing 5% vol. EG with a potentiostat/galvanostat OGF 500 (Origaflex, Clichy, France), where the saturated Ag/AgCl electrode and the platinum foil electrode worked as the reference electrode and the counter electrode, respectively. The working electrode was made of as-prepared catalysts (Pt-CT and Pd-CT) of which an area of 1 cm 2 was immersed into the electrolyte. A 35-W Xe lamp (200 mW cm −2 , and 420 < λ < 700 nm) (Briteye, Isfahan, Iran), is used for all PEC experiments. Linear sweep voltammetry technique (LSV) with a scan rate of 10 mV s −1 , photocurrent measurement, chronoamperometry (CA) at 1.5 V vs. RHE electrode, and open circuit potential (OCP) have been used to study PEC performance of as-prepared electrodes. All of these techniques were studied in the presence and absence of illumination as well as by light switching on and off. Electrochemical impedance spectroscopy (EIS) measurements for studying the semiconductor/electrolyte interface were performed using an IM6ex system (Zahner-Elektrik, Kronach, Germany). The frequency range was from 100 kHz to 100 mHz, while the amplitude of the imposed sinusoidal voltage was 10 mV. The impedance was recorded at a potential of 0.2 V (vs. Ag/AgCl).

Photo-Electrochemical Water Splitting
A three electrode configuration was used for PEC water splitting and H 2 production. Different Pt-CT and Pd-CT (depending on the time-exposure to noble-metal deposition) electrodes were immersed in a KOH (1 M) solution containing 5 vol % EG and under illumination (200 mW cm −2 , Xe lamp). The cathode (Pt Mesh) was inserted into an inverted burette, where H 2 was collected via electrolyte displacement. The variation of the electrolyte level in the burette, which is proportional to the volume of H 2 production ( Figure 1C), was recorded in predefined time-intervals (t = 0, 7.5, 15, 22.5, 30, 37.5, 45, 52.5 and 60 min).

Conclusions
Self-ordered chromium-doped TiO 2 nanotubes (CT), ∼8-10 µm in length, were fabricated using in-situ electro-anodization, which were afterwards successfully decorated with Pt and Pd metal particles using photo-deposition technique. The thorough inspection of morphological, surface and optical properties, as well as crystallinity of as-prepared materials exhibited following characteristics. CT structures are quite uniform in size and shape, and possess rather uniform distribution of deposited Pt and Pd over CT surface, while the agglomeration of noble metals was found to be photo-deposition time dependent. XRD revealed that TiO 2 pertain to anatase phase, while EDX clearly proved existence of chromium as dopant in CT. In addition, the presence of Pt and Pd as co-catalyst at CT's surface was confirmed by XRD, EDX and Raman. It was found that as-prepared Pt-CT and Pd-CT materials have lower bandgaps (1.8-2.2 eV and 2.1-2.8 for Pt-CT and Pd-CT electrodes, respectively) comparing to the CT (~2.82 eV). The electrochemical characterization indicated good photocatalytic properties of as-prepared materials, as well as their stability during exposure to visible irradiation. Pt-CT-3 (0.92 mA cm −2 ) and Pd-CT-5 (0.72 mA cm −2 ) showed the highest photocurrent density among Pt and Pd-decorated CTs, which is 4 and 3 times more than that of CT. Besides, Pt-CT electrodes showed (i) much lower charge recombination and (ii) better electrical conductivity under visible light illumination, than observed in the case of Pd-CTs. As a consequence of determined electrochemical and semiconducting properties, Pt-CT showed higher H 2 generation in PEC water splitting than Pd-CTs at comparable conditions; even 1.66 times H 2 was produced during 60 min course. It should be noted that both materials showed excellent stability during consecutive cycles; < 2% decrease was observed in the PEC water splitting efficiency during repetitive cycles.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author.