Enhanced photocatalytic, electrochemical and photoelectrochemical properties of TiO2 nanotubes arrays modified with Cu, AgCu and Bi nanoparticles obtained via radiolytic reduction

Graphical abstract


Introduction
The photocatalysts in the form of ordered arrays of TiO 2 nanotubes (NTs), electrochemically synthesized directly on titanium surface, attract much attention due to their high specific surface area, enhanced charge carriers transfer, large number of active sites [1], high chemical and mechanical stability [2]. TiO 2 NTs were utilized for photocatalytic purification of water [3] and gas [4,5] phases, hydrogen generation [6,7] and in non-photocatalytic applications such as gas sensors [8,9], biomedical materials [10], etc.
Various methods such as non-metal (e.g. N [11], B/N [12], S [13]) and transition metal (e.g. Zr [14], Fe [15,16], Cr [17,18], Zn [19]) doping were used to enhance the photocatalytic and photoelectrochemical activity of TiO 2 nanotubes. Another approach to this issue is the modification of nanotubes' surface with metal (noble and non-noble) nanoparticles [20]. Pd-modified TiO 2 NTs exhibited enhanced photoactivity in methyl orange degradation process [21], Pt nanoparticles were used to improve photocatalytic (degradation of methyl orange) [22], photoelectrochemical (degradation of galactose) [23] and electrochemical [24] properties of titania nanotubes. Ag-, Au-and Cu-decorated NTs possessed higher photocatalytic activity, comparing to bare NTs, in a process of degradation of "Congo red" [25]. The photocatalytic activity (degradation of methylene blue) of TiO 2 and WO 3 -TiO 2 NTs was improved by the deposition of Co [26] or Pt [27] nanoparticles. It has been established that metal nanoparticles present on the surface of TiO 2 (in the case of difference in their Fermi levels) can act as traps for photoinduced electrons, preventing from fast charge carriers recombination [28][29][30]. Moreover, the nanopar- ticles can induce additional electronic states in the bandgap of TiO 2 , and, as a consequence, the photons with lower energy than inherent band gap of titanium dioxide can cause the generation of electron-hole pairs. Additionally, metal nanoparticles, which often possess catalytic properties, can act as active sites for photocatalytic processes [31][32][33]. Also the modification of TiO 2 with bimetallic nanoparticles attracts much attention, due to large variety of possible combinations (different types and amounts of metals) which allows facilitating the design of materials with desired properties. It was often observed, that bimetallic nanoparticles exhibit synergistic properties compared to their monometallic counterparts [34]. Titania modified with bimetallic nanoparticles such as PdPt [35], AgPt, AuAg [36], AuCu [29,36], AuPd [32], AgCu [30] NiAu [37] displayed higher catalytic and photocatalytic activity in the comparison with the samples decorated with only one metal.
A large number of examples of the use of TiO 2 nanotubes decorated with monometallic Cu or Ag nanoparticles can be found in the literature [2,31,[38][39][40]42,[44][45][46][47][48][49][50]60]. According to previously discussed examples, it is supposed that in the case of modification of TiO 2 with bimetallic AgCu nanoparticles, a synergistic effect could be observed. However, there are very few studies on modification of titania with bimetallic AgCu nanoparticles. TiO 2 P25 was co-impregnated with Cu and Ag and subsequently tested in the process of photocatalytic Acid Orange 7 removal [61]. AgCu nanoalloys were also used as sensitizers for metal-cluster-sensitized solar cells [62]. In both cases, the synergistic effect was observed. Recently, Ag@CuO nanoparticles were generated on TiO 2 by radiolysis [30]. It has been found that the photocatalytic activity of Ag@CuO/P25 is higher under UV light, but lower under visible light compared to the activity of CuO/P25 and Ag/P25 [30]. The application of AgCu nanoparticles for modification of ordered TiO 2 nanotubes arrays has not yet been reported.
Bismuth is one of the most promising non-noble metals used to enhance photocatalytic activity of TiO 2 . However, in most research, the properties of Bi-doped TiO 2 photocatalysts (mixed Bi-Ti-oxide phases, Bi 2 O 3 -TiO 2 composites) obtained via sol-gel [63][64][65][66] and hydrothermal [67] methods were examined. There is also an example of TiO 2 nanotubes ordered arrays modified with Bi 2 O 3 using dip coating method, exhibiting high photoelectrocatalytic properties [68] and an example of Bi-doped TiO 2 NTs (used for energy storage) obtained via direct anodization of Bi-Ti alloy. Only a few reports concern the fabrication of TiO 2 modified with zero-valent bismuth nanoparticles. However, TiO 2 P25 modified with zero-valent Bi clusters obtained via radiolytic reduction exhibited enhanced photoactivity (compared with pure TiO 2 ), combined with high stability [69]. Although the TiO 2 nanotubes arrays modified with zerovalent Bi nanoparticles (used as photoelectrochemical sensors for organic compounds detection) was described [70], for the best of our knowledge no reports concerning photocatalytic applications of this kind of arrangement were published.
In this regard, the aim of this work was to investigate the properties of highly ordered TiO 2 nanotubes arrays decorated with monometallic (Bi, Cu, Ag) and bimetallic (AgCu) nanoparticles obtained using gamma radiolysis technique. The ratio of Cu to Ag in bimetallic nanoparticles was adjusted to the value 3:1 based on preliminary experiments performed with the use of P25 TiO 2 powder , which is in accordance with the most suitable ratio value mentioned in the literature [62]. The samples were characterized by FE-SEM, STEM, EDS, XRD, XPS techniques. The influence of the amount of metals' modifiers on electrochemical, photoelectrochemical and photocatalytic (degradation of phenol) performance of modified nanotubes was investigated.

Materials and chemicals
Titanium foil (thickness 0.127 mm, 99.7% purity) purchased from Sigma-Aldrich was used as substrate material. Isopropanol, acetone and methanol (p.a., POCh) were used for cleaning Ti foil surface. NH 4 F (p.a.) and ethylene glycol (99.0%, p.a.) purchased from POCh were the components of electrolyte used for TiO 2 nanotubes preparation. CuSO 4 ·5H 2 O, Ag 2 SO 4 , Bi 2 O 3 (p.a., FLUKA) were used as precursors of metals used for surface modification of the synthesized titania nanotubes. Isopropanol and ethanol (p.a. Sigma-Aldrich) were used as metal precursors' solvents. Phenol (≥99%, Sigma-Aldrich) was used as model organic water contaminant. Na 2 SO 4 (99.5%) used for electrochemical and photoelectrochemical measurements was purchased from POCh Avantor. All the reagents were used without further purification. Deionized water was used as solvent.

Synthesis of ordered TiO 2 nanotubes layers
Ordered TiO 2 nanotubes arrays (NT) were synthesized via simple electrochemical anodization of titanium foil, according to the procedure described in details in our previous paper [5], which is schematically illustrated in Fig. 1a. Ti foil samples (dimensions 2 × 3 cm) were successively ultrasonically cleaned in acetone, isopropanol, methanol, rinsed with water and subsequently dried in air and anodized for 1 h at the applied potential of 30 V in the electrolyte containing ethylene glycol (98%, v/v), water (2%, v/v) and NH 4 F (0.09 M) using 2-electrode system with a cylindrical Pt mesh cathode. Ti foils were immersed in the electrolyte to the level of 2/3 of their length and placed inside the Pt mesh cylinder. Afterwards, the as-synthesized nanotubes were rinsed with water, ultrasonically treated (5 min in water) in order to remove surface precipitates, dried in air (24 h at 80 • C) and calcined for 1 h at 450 • C (heating rate 2 • C min −1 ).

Modification of TiO 2 NTs with metals via radiolysis technique
Pure TiO 2 nanotubes were modified with mono-and bimetallic nanoparticles (Cu, Ag, AgCu, Bi) by direct surface adsorption of metals ions or metal precursor particles (Bi) from alcoholic solutions or suspensions followed by radiolytic reduction (Fig. 1b).  (Table 1). In order to prepare metal-modified NTs, a specified amount of the precursor solution was dropped onto the pure nanotubes' surface using  It should be noted that only one side of each sample was modified. After precursors' deposition, the samples were subsequently placed in the glove-box and additional amount of alcohol (which scavenges the oxidizing • OH radicals induced by water radiolysis to lead to alcoholic reducing radicals [69]) was added under N 2 atmosphere. The samples were then placed in plastic boxes (sealed with parafilm in order to avoid the access of oxygen from air) and exposed to irradiation for 5 h. Metal nanoparticles were synthesized by radiolytic reduction of the metal precursors using a 60 Co panoramic ␥-source located at the Laboratoire de Chimie Physique, in Orsay (dose rate = 3 kGy h −1 , dose = 15 kGy). After drying in air, the samples were ready for further experiments.

Characterization techniques
The non-modified and meal-modified TiO 2 nanotubes arrays were characterized by field-emission scanning electron microscopy combined with energy dispersive X-ray spectroscopy (FE-SEM/EDS), scanning transmission electron microscopy combined with energy dispersive X-ray spectroscopy (STEM/EDS), X-ray diffractometry (XRD) and X-ray photoelectron spectroscopy (XPS).
FE-SEM images and EDS spectra were recorded using JEOL JSM-6360LA microscope equipped with JED-2300 energy dispersive Xray analyzer. The accelerating voltage was 5 kV at working distance 6 mm. Surface elemental analysis was performed in five different areas of the sample in order to determine the average composition.
STEM images and EDS surface mapping were registered using HITACHI HD2000 microscope (accelerating voltage 200 kV, emission current 20 A). To prepare samples for analysis, the nanotubes were scratched from the substrate (Ti foil) and dispersed in ethanol. The suspension was subsequently dropped on a carbon-covered copper microgrid and dried in vacuum.
The XRD patterns were recorded using Rigaku SmartLab X-ray diffractometer with copper K␣ target (40 kV, 30 mA, = 1.5404 Å). The scanning range was 2 = 10-90 • at a scan step of 0.01 • . Estimated crystal size was calculated according to Sherrer's formula.
XPS analysis was conducted using Jeol JPC-9010 MC X-ray photoelectron spectrometer with Mg K␣ X-ray source. 20 scans were performed for analyzing Ti, O and C content, whereas 50 scans were performed to analyze the presence of metals (Cu, Ag, Bi) used for surface modification of TiO 2 NTs.

Photocatalytic phenol decomposition measurements
The photocatalytic activity under UV-vis irradiation of nonmodified and metal-modified TiO 2 nanotubes was studied in the process of water purification using phenol (concentration 60 mg L −1 ) as a model pollutant. Photocatalytic activity tests were performed using a simple experimental set-up containing 300 W xenon lamp (Oriel LSH302) equipped with water IR cut-off filter (8 cm length). A quartz cuvette (dimensions 1 × 1 cm) was placed in front of the water filter and acted as a photoreactor. To perform photocatalytic experiment, the photocatalyst sample with the dimensions of 1 × 3 cm (cut from larger sample) and with active part with the dimensions of 1 × 2 cm was placed at the back wall of the cuvette using a paperclip as a holder. Subsequently, the cuvette was filled with 3.5 mL of phenol solution. After 5 min in dark (to ensure the adsorption equilibrium of phenol) the reaction system was irradiated for 60 min. Magnetic stirring and bubbling with oxygen was ensured during the whole period of photocatalytic experiment.
Small aliquots (0.5 mL) of the solution were collected from the reaction system after 20, 40 and 60 min of irradiation in order to analize the phenol degradation by high-performance liquid chromatography (HPLC) and to determine the concentration of organic carbon by total organic carbon (TOC).
The chromatographic system used for HPLC was Agilent 1260 infinity quaternary LC equipped with UV-detector (set at 254.4 nm), the column Adsorbosphere C18 reverse phase (5 m, l = 150 mm, ID: 4.6 mm; Alltech) combined with an All-Guard cartridge system TM (7.5 × 4.6 mm; Alltech) and the autosampler. The mobile phase (flow rate 1 mL min −1 ) consisted of 75% water and 25% acetonitrile. The peaks areas corresponding to phenol and its main degradation by-products (dihydroxybenzenes) were calculated in direct concentration (mg L −1 ) and in concentration of organic carbon (mol L −1 ) based on calibration curves.
The TOC measurements were performed using a Shimadzu TOC-LCSH analyzer. Prior to analysis, the samples were diluted in water in order to reach the required sample's volume. The inorganic carbon present in the samples was removed by acidification and air purging and subsequently TOC was measured by IR after complete catalytic oxidation on Pt catalyst at 680 • C.

Determination of electrochemical and photoelectrochemical properties
The electrochemical and photoelectrochemical activity of pure and modified titania nanotubes was studied using an AutoLab PGStat 302 N potentionstat-galvanostat system (Methrom Autolab) in the standard three-electrode assembly, with titanium foil covered by nanotubes as a working electrode, while Ag/AgCl/0.1 M KCl and Pt mesh as reference and counter electrodes, respectively. The photoelectrochemical cell was equipped with a quartz window and a cooling jacket that keeps the temperature constant at 23 • C (±1 • C) using thermostat (Julabo F-12). Prior to analysis, 0.5 M Na 2 SO 4 solution of electrolyte was purged with argon gas for about 1 h. Ar-cushion above the electrolyte was applied during the measurements. The transient photocurrent measurements were carried out at + 0.5 V vs. Ag/AgCl/0.1 M KCl bias voltage. A high-pressure 150 W xenon lamp (Osram XBO 150) equipped with AM1.5 filter and the automated light chopper with a period of 20 s was used as a light source. The light intensity was adjusted to 100 mW cm −2 (Ophir).

Results and discussion
3.1. Morphology and structure

FE-SEM and STEM analysis combined with EDS
FE-SEM images of the non-modified sample (NT) are presented in Fig. 2a-c. The observation confirmed that the surface of Ti substrate was covered with the array of ordered, smooth and elongated nanotubes. However, some scratches and residual precipitates were present on the face of nanotubes, which is in accordance with our previous observations [5]. The FE-SEM images were used to estimate the dimensions of the TiO 2 nanotubes and to evaluate the total amount of electrochemically synthesized titanium dioxide. The average tube's length was approximately 2.3 m and inner diameter and wall thickness were 90 and 16 nm, respectively. The estimated amount of TiO 2 (one side of the active part of the sample, 2 × 2 cm) was 3.28 mg. This value was used to calculate the content of metals used for surface deposition (Table 1).
In the case of metal-modified samples, FE-SEM and STEM observations (Fig. 2d, e) did not indicate the presence of metal nanoparticles, which is not surprising, taking into account several factors. Radiolysis technique leads to the production of homoge- nously distributed metal nanoparticles with the size up to several nanometers [52,55], in the case of Bi not exceeding 1.2 nm [69]. These nanoparticles are probably deposited mainly on the walls' surface of the nanotubes [52]. Additionally, in the case of copper nanoparticles the small contrast between Cu and TiO 2 can be a factor that hinders observation [29,55]. In contrast, another deposition techniques, such as chemical reduction [47,48], photodeposition [38,40,71], electrodeposition [42] led to the formation of larger nanoparticles which were present also on the top surface of nanotubes and could be easily observed using SEM technique.
The presence of the metals used for deposition was confirmed by FE-SEM/EDS analysis (performed for samples with the highest amount of modifiers: Bi-NT IV, Cu-NT IV, AgCu-NT IV). The results of this analysis are presented in Table 1. However, the amount of the deposited metals was much lower than expected, i.e., 0.003, 0.19 and 0.25/0.054 mol% of Bi, Cu and Cu/Ag respectively, due to low precision of EDS analysis for component of low content. The atomic ratios of O to Ti were also much lower than stoichiometric value of two, reaching 0.71, 0.67 and 0.69 for Bi-NT IV, Cu-NT IV and AgCu-NT IV respectively. This is not surprising and has been already observed for other titania samples [72], due to limitation of the EDS method to elements of large atomic number. Low yield of X-ray absorption is noticed for light elements, e.g., oxygen. STEM-EDS mapping analysis performed for sample AgCu-NT III revealed the homogeneous distribution of metal modifiers on the surface of TiO 2 nanotubes (Fig. 3).

XPS analysis
The presence of metal nanoparticles as well as chemical composition of surface layer of bare-and metal-modified samples was investigated by X-ray photoelectron spectroscopy (XPS). Six main elements were analyzed in details using narrow scanning, i.e., titanium, oxygen, carbon, bismuth, copper and silver and XPS data is summarized in Table 1 (Bi, Cu and Ag content) and in supplementary materials (Tables S1 and S2). It was found that samples did not differ strongly both in their compositions and chemical state of elements. All the samples possessed carbon on the surface in the range of 18-27 mol%, which could result either from the used electrolyte, or/and adsorbed CO 2 from the air. It should be mentioned that this amount of carbon is quite low in comparison with titania samples prepared by sol-gel and hydrothermal methods, in which organic precursors of titania (e.g., titanium isopropooxide) are usually used [73]. The atomic ratio of oxygen to titania for pure NT achieved almost the stoichiometric value of 2.0 reaching 1.97. While, for all the modified samples atomic ratios were slightly lower than two (1.61-1.94), probably due to reductive conditions during metal radiolytic deposition on TiO 2 nanotubes.
The molar content of the deposited metals on the surface layer of NT depended on the kind of the modifiers. It was found that amount of deposited bismuth was lower than expected values (Table 1). It is suggested that either Bi deposition was incomplete, because of too short time used for radiolytic reduction (the same time was applied for all the samples independently on precursor amount) or Bi clusters penetrated and deposited mainly inside the nanotubes. In the case of copper deposition, larger amount of the metal precursor used for deposition resulted in its larger content on the surface of titania nanotubes, with only one exception for Cu-NT III sample. In the case of deposition of two metals (Ag and Cu), the increase in amount of the metal precursors resulted in proportional increase of their content on the surface layer of NTs. It should be noted that in the case of Cu and AgCu modified samples, the amount of the deposited copper (measured by XPS) was much higher (e.g. 5.6%) than the calculated value (e.g. 0.94%). This phenomenon is not surprising because in the case of surface modification, the modifier is not incorporated into bulk TiO 2 , but accumulates on its surface and, as a consequence, the modifier to titanium ratio increased. This was observed also for titania samples modified with other metals, e.g. Pt [33]. All AgCu-NT samples were prepared using metals' precursors solution with Cu:Ag ratio equal to 3:1. However, the Cu to Ag ratio determined by XPS was much higher, varying from about 16:1 to 23:1. These results can suggest that bimetallic nanostructures with a core rich in silver and a shell rich in copper are formed. Similar nanoparticles were recently obtained on TiO 2 -P25 by radiolysis [30].
According to some theoretical and experimental studies, AgCu clusters tend to form core-shell structures in which copper is located in the core of nanoparticles while silver atoms segregate on the surface and create shell [74][75][76]. This is related to such properties of Cu and Ag atoms as relative strength of bonds, surface energies, size and electronegativity [34,77]. However, here we are   [52,77] and [78]). out of equilibrium conditions and the radiolytic reduction of metal ions with relatively low gamma dose rate (a few kGy h −1 ) favors the formation of core-shell structures in which the more noble metal is reduced first and form the core of the bimetallic nanoparticle [78]. This is the result of fast electrons' transfer from less noble metals to more noble ones, privileging their reduction. Hence, the more noble silver atoms are first reduced and create the core of nanoparticle. The core is then covered with the less noble metal (copper) forming the shell (Fig. 4). Copper is sentitive to air. Therefore, the core-shell nanoparticles propably turn into Ag@CuO nanoparticles [30].
XPS data after deconvolution of titanium and oxygen peaks are summarized in Table S2 (supplementary materials), and exemplary spectra are shown in Fig. 5. It was found that titanium existed mainly in Ti +4 form (>95%). The non-modified TiO 2 nanotubes contained only ca. 2% of reduced titanium (Ti +3 ). In general, Ti 3+ formation can be induced by reducing TiO 2 with a suitable reductant in gas or liquid phase [79]. In the case of TiO 2 nanotubes obtained with organic electrolytes, oxygen vacancies can be created during calcination step due to high carbon content in as-prepared nano-tubes [80]. Modification with metals only slightly increased the content of Ti +3 , due to reductive conditions during metal depositions, indicating that dose of gamma radiation was well assigned. For two samples with the largest amount of copper and silver (Cu-NT IV and AgCu-NT IV) titanium existed only in the oxidized form of +4. Oxygen states varied for all the tested samples. O 1s region could be deconvoluted into two peaks at binding energy of 529.7-529.9 eV and 530.8-531.7 corresponding to lattice oxygen in TiO 2 and surface oxygen in Ti-OH, respectively. It was found that pure TiO 2 nanotubes and copper modified Cu-NT II sample possessed the largest amount of oxygen in the form of surface hydroxyl groups (35.2 and 36.9%, respectively). Interestingly, the samples (Cu-NT IV and AgCu-NT IV) possessing titanium in only oxidized form (Ti +4 ) exhibited also the lowest amount of oxygen in the form of hydroxyl group, i.e., the largest amount of oxygen in TiO 2 form (lattice oxygen) proving the lack of oxygen vacancies (Ti +3 ).
It should be underlined that the positions of titanium peaks did not shift after modification with metals indicating surface modification of titania, but not metal substitution.

XRD analysis
The crystalline composition of non-modified and metalmodified TiO 2 NTs films was examined by XRD analysis. The exemplary XRD patterns are shown in Fig. 6. Titanium foil consisted mainly of pure titanium with a small amount of crystalline titania, i.e., anatase and rutile in the amount of 5.7 and 7.8 wt.%, respectively. Pure titania nanotubes consisted mainly of amorphous and anatase titania (crystallite size of ca. 35 nm) with a small amount of rutile phase (<5 wt%). It was impossible to determine the precise crystalline composition of the nanotubes, due to very intensive titanium peaks from the support. According to [81], the crystallites of anatase tend to grow along the length and the curvature of the nanotubes rather than across the thickness of the tubes' walls. As mentioned before, the estimated crystallites' size has been calculated according to Scherrer's formula from the broadening of anatase (1 0 1) reflection [82,79]. This method has been already used to determine the crystallite size of anatase in TiO 2 nanotubes [83][84][85]. The accuracy of the calculations has been estimated to be about 20%. The crystallite size of anatase was in the range of 36-40.4 nm (summarized in Table 1) for all the metalmodified samples, indicating that the kind of modifiers and gamma radiation used for metal reduction/deposition did not influence significantly crystalline properties of titania support. Unfortunately, it was impossible to detect crystallites of deposited metals, possibly due to their very small amount, high level of dispersion and small cluster size, which was consistent with previous studies on titania nanotubes modification with small metal clusters [45,48,49,60,86].

Photocatalytic activity and mechanism discussion
Phenol is a representative of phenolic compounds, which when presents in water, causes severe environmental problems [87] and it is often selected as a model water pollutant in the study of photocatalytic processes [88][89][90][91][92]. In this regard, it was used to determine the photocatalytic activity of Cu-, AgCu-, and Bi-modified TiO 2 nanotubes. The results of photocatalytic activity measurements of all the examined samples are presented in Fig. 7. The activity is expressed as the efficiency of phenol degradation (1-c/c 0 ) after 60 min of irradiation and as apparent first order kinetic rate constant, calculated from ln(c 0 /c) versus time plot [90]. High efficiency of phenol degradation (90% and more) was observed for all the examined samples. It should be noted that the photolysis had a big influence on the degradation process. About 50% of phenol loss was observed with the absence of photocatalyst. However, it is still much less than in the case of the presence of the less active sample. In this respect, the comparison between the activities of different photocatalysts can be performed.
Among Cu-modified nanotubes, only the samples decorated with larger amount of Cu nanoparticles (Cu-NT III and IV) exhibited enhanced photocatalytic activity in phenol degradation process compared with bare nanotubes. The efficiency of phenol degradation in the presence of NTs modified with bimetallic (AgCu) nanoparticles was higher in comparison with monometallic samples modified with the same amount of Cu and all the samples were more active than non-modified nanotubes. The sample AgCu-NT III was the most active. Among Bi-modified NTs, only Bi-NT II and III samples were more photoactive than bare nanotubes. For comparison, Ag-modified sample was prepared (Ag-NT). The amount of deposited silver was the same as the amount of copper in the most active Cu-modified sample (Cu-NT III). However, this amount of Ag-modifier caused the decrease of the photoactivity of TiO 2 nanotubes. This is probably due to exceeding of the optimal Ag content. Wongwisate et al. [93] observed that TiO 2 modified with relatively small amount of silver (about 0.075 mol%, calculated value) exhibited the highest photoactivity in the initial stage of 4-chlorophenol degradation process. However, silver nanoparticles, when they are present in excessive amount, they can become electron-hole recombination centers and, simultaneously, decrease the photoactivity due to screening effect [61,93]. In contrast, modification of TiO 2 with relatively high amount of Cu (0.6 mol%, calculated value) caused the increase in its photocatalytic activity in methyl orange degradation [55]. Moreover, TiO 2 nanotubes modified with 5.6 mol% Cu (value estimated based on EDX analysis) exhibited higher photoactivity in "Congo red" degradation than pure NTs [25]. For more detailed comparison of the prepared samples, another parameters (initial phenol degradation rate and TOC removal efficiency) were considered as more suitable for revealing the differences between the photocatalytic activities of non-modified and metal-modified TiO 2 nanotubes. Fig. 8a and b show the initial phenol degradation rate in the presence of Cu-, AgCu-and Bi-modified samples, calculated for the first 20 min of irradiation. The initial degradation rate for Ag-NT sample (value not shown in charts) was 1.27 mg dm −3 min −1 , which was lower than for non-modified NTs. The value of this parameter was enhanced for all Cu-modified samples, wherein Cu-NT III sample was the most active (initial degradation rate about 8% higher than for bare NTs). The presence of silver in Cu nanoparticles caused further increase of the photoactivity, however, the synergistic effect was not observed for the sample decorated with the highest amount of modifier (AgCu-NT IV). Among all Cu-and AgCu-modified samples, the highest value of initial phenol degradation rate (about 12% higher than for non-modified nanotubes) have been reported for AgCu-NT I sample and was only slightly lower for AgCu-NT II and III samples (increase of about 9-9.7% compared to bare nanotubes). Among Bi-modified nanotubes, the highest increase of initial phenol degradation rate was observed for Bi-NT III sample (about 10%) and only slightly lower for Bi-NT II sample (about 8.7%). The behavior of the samples modified with the lowest and the highest amount of Bi was similar to non-modified nanotubes.
As it can be seen from Fig. 8c, the TOC removal efficiency after 60 min of irradiation was relatively low (about 68%) in the presence of non-modified TiO 2 nanotubes. The modification of nanotubes with Cu nanoparticles caused a significant increase in the value of this parameter, reaching over 90% for the samples with higher amount of modifier (Cu-NT II, III and IV). It should be noted that nanotubes modified with bimetallic, AgCu nanoparticles did not show a meaningful increase of TOC removal efficiency (except from AgCu-NT I sample) with respect to the samples modified only with copper. However, AgCu-NT IV sample exhibited the highest TOC removal performance (about 95%) among all Cu-and AgCumodified nanotubes. Lower effectivity of TOC removal (but still higher than for bare nanotubes) was observed in the case of samples modified with the average amount of Bi nanoparticles (about 78% for samples Bi-NT II and III), Fig. 8d. The decrease in TOC removal below the level of pure nanotubes was observed for samples with the lowest and the highest Bi-loading (Bi-NT I and IV respectively).
As can be concluded, for each TiO 2 modification, an optimal amount of the modifier causes an increase of the photocatalytic activity. In this regard, non-modified nanotubes and metalmodified samples, which exhibited the highest efficiency for phenol degradation, and relatively high values of initial phenol degradation rate and TOC removal efficiency (AgCu-NT III and Bi-NT II), were chosen for further mechanism discussion.
Phenol decomposition in the presence of pure and metalmodified nanotubes was in accordance with a typical scheme of photocatalytic degradation in UV/TiO 2 system in which the fast initial substrate decay was followed by creation of primary intermediates (hydroquinone, catechol and benzoquinone). After a certain time period, the concentration of primary intermediates reached an optimum and then decreased in parallel with the decrease of the phenol content [54,82,[94][95][96]. As it can be seen from Fig. 9 (and Fig. 8a and b), the initial phenol degradation rate was similar for AgCu-NT III and Bi-NT II samples and was higher than for bare nanotubes. The difference in degradation rate increased in the next time period and finally slightly decreased. About 2.5% difference in phenol degradation efficiency between modified-and non-modified nanotubes was observed after 60 min of UV irradiation. The plots of primary intermediates' concentration vs. time were similar for all the discussed samples, however, after 20 min of irradiation, slightly lower intermediates' concentration was observed for Bi-modified sample and much lower concentration for AgCu-modified nanotubes. These results suggest that the subsequent photodegradation steps were accelerated by modification of NTs with metal nanoparticles (especially AgCu). This is in agreement with TOC measurements (Figs. 8c and d,9). It has been observed that the total organic carbon concentration decreased monotonously during the photocatalytic process, however, the degradation rate for pure nanotubes was lower than for Biand AgCu-modified samples. After 60 min of irradiation the differences in TOC loss reached 10.1 and 25% (for Bi-and AgCu-modified samples respectively) in comparison with pure nanotubes. Due to the fact that, for all examined samples, the concentrations of phenol and primary intermediates were relatively similar at the end of the process, the higher TOC level can be attributed to the presence of compounds which were formed in subsequent steps of the photocatalytic process. As it was mentioned before, the modification of TiO 2 with metal nanoparticles can lead to more efficient electron-hole separation and to enhance • OH and O 2 • − radicals formation [30,32,33,[53][54][55]93]. O 2 • − radicals can be subsequently transformed, via H 2 O 2 , to • OH radicals which are thought to be the most responsible for photocatalytic degradation of phenol. • OH radicals can be simultaneously generated via direct oxidation of water molecules by photogenerated holes [95]. Thus, higher efficiency of • OH radicals creation can lead to enhanced photoactivity. Guo et al. [92] suggested that the attack of • OH on phenyl ring is the first stage of photocatalytic process which leads to the formation di-and trihydroxybenzenes and, subsequently, to opening of the phenyl ring and forming maleic acid among other intermediate products (Fig. 10).

Linear voltammetry and chronoamperometry
The photoelectrochemical behavior of pure and modified titania electrodes exposed to illumination from solar simulator is presented in Fig. 11. The photocurrent densities in all the examined samples increased with the increasing potential applied to the electrode. As shown, TiO 2 samples with loaded Bi, Cu, and AgCu nanoparticles exhibit improved photoactivity compared to unmodified titania. However, in the case of the two most active TiO 2 NTs samples: Bi-NT II and AgCu-NT II, the saturated photocurrent was over two times higher compared to pristine TiO 2 NTs. As it was discussed earlier, for Bi-NT II the highest TOC removal efficiency was observed whereas AgCu-Nt II sample is characterized with one of the highest initial phenol degradation rate. Thus, enhanced photoelectrochemical activity was reported for the same materials that exhibited improved photocatalytic properties than unmodified titania. In order to investigate the photoelectrode stability, the current was measured at + 0.5 V vs. Ag/AgCl/0.1 M KCl bias voltage for 15 min. Fig. 11b shows the transient photocurrent response for pure and metal modified TiO 2 nanotube arrays by on-off cycles registered under UV-vis radiation. The run of chronoamperometry curve exhibits the rapid increase and decrease of current when the irradiation was switched on and off, respectively. The dark current densities were negligible for the all tested samples. As it could be seen all the electrode materials were characterized with the resistance towards photocorossion processes that enables their application for long term processes induced by light.

Cyclic voltammetry
Cyclic voltammetry (CV) was performed to characterize materials in contact with a deareated electrolyte. Curves registered for pure and Bi, Cu and AgCu modified titania are presented in Fig. 12. The working electrode was polarized from the rest potential in the anodic direction up to 1 V and back up to −1.0 V vs. Ag/AgCl/0.1 M KCl. In general, the CV shape for all the samples is typical for anatase structure of titania NT materials that is characterized with a very low capacitive current in the anodic potential range but in the cathodic range much rich electrochemical activity is observed [97]. In the negative range, two reduction peaks were recorded. According to Bertoluzzi et al. [98] the first one, located at about −0.25 V could be assigned to the filling of narrow deep trap states. On the reverse scan this signal is accompanied with a broader anodic peak that represents relatively slow depopulation of these states [99]. The next signal, recorded at about −0.7 V could be related with alteration in electronic structure of the oxide, e.g. incorporation of additional states within the bandgap that change conductivity as well as optical properties [97]. According to Pelouchova et al. [100], this cathodic peak could be ascribed to Ti 4+ reduction combined with cation intercalation and is related with significant alterations in the material conductivity. On the other hand, for silver modified TiO 2 NT reduction peak located below −0.5 V was identified as hydrogen evolution [101] despite its generally regarded that this reaction takes place at more negative potentials.
As it could be clearly noticed, CV curves registered for titania samples differ from each other taking into account both positions of cathodic peaks and values of current density. The observed differences result from the presence of metal nanoparticles and increase of Ti 3+ amount comparing to pristine titania NTs as described above. According to Jiang et al. [1], the observed change in double redox activity in the case of modified TiO 2 comes from two possible localization of metal nanoparticles: in the internal walls of the nanotubes and at the top surface of the nanoporous structure. Thus, similarly to non-metal doped titania [102], the presence of metal nanoparticles strongly affects the electrochemical activity (e.g. some peaks are shifted or new peaks arises), whereas increased   amount of Ti 3+ can be responsible for the increase of charging current [103].

Conclusions
In summary, TiO 2 nanotubes were modified with Cu, AgCu and Bi nanoparticles by using radiolysis technique. EDS and XPS techniques confirmed the presence of metal modifiers, however, the STEM observation did not provide any information about the size and distribution of nanoparticles, probably due to their very small dimensions. Comparing the initial Cu to Ag ratio (3:1) in precursors' solution with the Cu to Ag ratio (16:1 to 23:1) in AgCu modified samples Ag core -Cu shell nanoparticles were obtained. It was observed that surface modification with metal nanoparticles leads to enhanced photocatalytic activity under UV-vis irradiation, and this effect depends on the amount of deposited metal. The efficiency for phenol degradation in the presence of NTs modified with bimetallic (AgCu) nanoparticles was higher in comparison with monometallic samples modified with the same amount of Cu. Among all Cu-and AgCu-modified samples, the highest value of initial phenol degradation rate (about 12% higher than for non-modified nanotubes) is obtained with the bimetallic sample modified with the smallest amount of metals (Cu-0.13 mol%, Ag-0.05 mol%). For Bi-modified samples, the most suitable Bi content was 0.31 and 0.63 mol% in terms of photocatalytic activity. Metal nanoparticles act as traps for excited electrons decreasing the charge carrier recombination rate, and increasing the photocatalytic activity. The photoelectrochemical experiments performed under the influence of simulated solar light irradiation confirmed the enhanced photoactivity of metal-modified nanotubes. The saturated photocurrent for the most active Bi-and AgCu-modified samples, was over two times higher than for pure nanotubes. All the examined materials were resistant towards photocorrosion which enables their application for long term photoinduced processes.