A Dewetted‐Dealloyed Nanoporous Pt Co‐Catalyst Formed on TiO2 Nanotube Arrays Leads to Strongly Enhanced Photocatalytic H2 Production

Abstract Pt nanoparticles are typically decorated as co‐catalyst on semiconductors to enhance the photocatalytic performance. Due to the low abundance and high cost of Pt, reaching a high activity with minimized co‐catalyst loadings is a key challenge in the field. We explore a dewetting‐dealloying strategy to fabricate on TiO2 nanotubes nanoporous Pt nanoparticles, aiming at improving the co‐catalyst mass activity for H2 generation. For this, we sputter first Pt‐Ni bi‐layers of controllable thickness (nm range) on highly ordered TiO2 nanotube arrays, and then induce dewetting‐alloying of the Pt‐Ni bi‐layers by a suitable annealing step in a reducing atmosphere: the thermal treatment causes the Pt and Ni films to agglomerate and at the same time mix with each other, forming on the TiO2 nanotube surface metal islands of a mixed PtNi composition. In a subsequent step we perform chemical dealloying of Ni that is selectively etched out from the bimetallic dewetted islands, leaving behind nanoporous Pt decorations. Under optimized conditions, the nanoporous Pt‐decorated TiO2 structures show a>6 times higher photocatalytic H2 generation activity compared to structures modified with a comparable loading of dewetted, non‐porous Pt. We ascribe this beneficial effect to the nanoporous nature of the dealloyed Pt co‐catalyst, which provides an increased surface‐to‐volume ratio and thus a more efficient electron transfer and a higher density of active sites at the co‐catalyst surface for H2 evolution.


Introduction
Since the pioneering work of Fujishima and Honda in 1972, [1] the splitting of water by using light has received great attention as ap otentialm ean for converting solar energy into clean and renewable vectors or fuels such as hydrogen gas. [2][3][4][5] In the field of photocatalysis and photo-electrochemistry,T iO 2 has attracted enormous research interest [6,7] due to as et of key advantages:T iO 2 is in fact cheap, abundant, corrosion-resistant, environmentally friendly and has as uitable conduction band edge to evolve H 2 by reduction of H 2 Oo ro ther suitable agents. [8,9] However,the high recombination rate of photo-generated electron-holep airs in the semiconductor as well as the low rate of charge transfer to reactants at the solid-liquid junction typically result in al ow photon-to-product conversion efficiency. This, alongw ith the relativelyl arge band gap, impedes the development of cost-effective TiO 2 based photocatalytic and photo-electrochemical units.
Many efforts have been therefore given in the last 40 years to modify TiO 2 in view of enhancing its photo-activity.A ne fficient strategy is to "decorate" the semiconductor surface with co-catalyst nanoparticles of as uitable metal, such as Pt,A uo r Pd, which can act as electron trappings ites to limit the electron-hole pair recombination in TiO 2 .T his can extend the lifetime of chargec arriers and enhancet he photocatalytic activity. [10][11][12][13][14] Conduction band (CB) electron trapping occurs because high work function metals can form ar ectifying Schottky barriera tt he TiO 2 surface. Moreover,m etals such as Pt can also serve as catalytic cathodic sites for hydrogen atom recombination and H 2 evolution. [15] However,b ecause of the high cost and low abundance of Pt, the understanding of the cocatalystm orphology-activity relationship can be key to design more efficient photocatalysts, providing for example ah igher activity per mass of loaded co-catalyst.
To this end, nanoporous metals have attracted considerable interest: their bicontinuous structure, tunable pore sizes, good electricalc onductivity and high structural stability can lead to an improved performance in variousa pplications,f or example, in electrocatalysis, energy storage, or sensing. [16][17][18][19][20][21] In this context, dealloying is am ost straightforwarda pproach to produce nanoporous metals or porousm etal NPs. Dealloying is based on the selective dissolution of al essn oble metal component from a( single phase) metal alloy by chemical or electrochemical methods. [22,23] The dissolution of the less noblec omponent is coupled with diffusion and aggregation of the more noble metal elementa tt he solid/liquid interface: the overall process leaves behind ap orousm etal structure (a metal "sponge") that can be enriched or mainly composed of the nobler element. [24] Nanoporous metals prepared via selective dealloying of solid solutionsp ossess at hree-dimensional (3D) structure of randomly interpenetrating ligaments/pores with sizes between a few nm to severalt ens of mm; these structural features can be preciselyt uned by varying the preparation conditions (such as alloy composition, dealloying time, temperature, and electrochemicalp arameters) or by subsequentathermalc oarsening step. [18,[25][26][27][28][29] In spite of the large application in heterogeneousc atalysis, there is however still ac omparably low number of studies on nanoporous metal co-catalysts for photocatalytic applications. Nguyene tal. have reported on porousA u, [30] AuPt, [31] or PtPd [32] nanoparticlesp roduced on TiO 2 nanotubes (NTs) by chemicald ealloying of dewetted-alloyed nanoparticles. These porousn anoparticles combined to TiO 2 were found to strongly enhancethe photocatalytic activity.
In this work, we fabricate ap hotocatalytic platform based on anodic TiO 2 NTsf unctionalized with ap orousP tc o-catalyst formed via dewetting-alloying-dealloying [33][34][35] of sputtered Pt-Ni bilayers. For this we firstly sequentially coatt he TiO 2 NTs with at hin Pt and Ni metal film (few nm thick) by Ar-plasma sputtering. Then we expose the PtÀNi coated structures to a suitable thermal treatment to induce dewetting-alloying of the Pt-Ni bi-layer. [36] Due to metal atom surface mobility,t he Pt and Ni films break up and mix with each other forming PtÀNi dewetted-alloyed islands.A fterwards these structures are exposed to ac hemical dealloying treatment, under free-corrosion conditions, to selectively remove Ni from the dewetted-alloyed Pt-Ni islands, thereby leaving behind ad ealloyed porous Pt cocatalystatt he TiO 2 NTssurface.
The role of the Pt and Ni film initial thickness, the deposition sequence and dealloying time are herein investigated to achieve control over the final composition and morphologyo ft he dealloyed Pt co-catalyst. By optimizing these parameters, we produced nanoporous Pt/TiO 2 structures that deliver as ignificantly enhanced photocatalytic H 2 evolutionp erformance compared to structures functionalized with comparable loadings of dewetted, non-porous Pt. The improved photocatalytic performance is ascribed to the high specific surfacea rea and nanoporous structure of the Pt co-catalyst.

Results and Discussion
The photocatalystf abrication methoda dopted in this work is exemplified in Scheme1;t he experimental detailsc an be found in the SI. As photocatalytic platform we use arrays of highly-ordered TiO 2 NTs. TiO 2 NTsh ave attracted considerable interesti np hotocatalysis and photo-electrochemistry in the last decades due to their reliable and versatile fabrication method,h igh surfacea rea and one-dimensional (1D) morphology; [37,38] namely the latter can provide attractive charget ransport properties. [39] Moreover, in the context of the present work, these highly-ordered TiO 2 nanotube layers can be used as ad efined, corrugated photo-active surfacet oh ostt he formationo ft he nanoporous Pt co-catalyst by ad ewetting-alloying-dealloying approach. [34,35,40] The TiO 2 NTsw ere grown by self-organizing electro-chemical anodization of aT im etal foil in ah ot HF/o-H 3 PO 4 electrolyte. [41] The SEM image in Figure 1a shows the morphology of as-prepared structures. These TiO 2 NTsh ave an inner diameter of % 70 nm and sidewalls with at hickness of % 10 nm, while the NTsl ayer thickness (NT length) is % 150 nm. The TiO 2 NTsa re then sputter-coated (by Ar plasma sputtering) with thin Pt and Ni metal films. The metal coating is uniformly distributed over the nanotube surface, as one can see from Figure1b; here the structures are sequentially coated with aP tf ilm, nominally 5nmthick,f ollowed by a10nmthickN ifilm.
We screened different Pt:Ni nominal ratios,b ys puttering firstly the Pt film (5 nm) and then Ni (different thicknesses), this to assesst he effect on the morphology of the dewetted state, the feasibility of dealloying the dewetted bimetallic islands, and the resultingp hotocatalytic H 2 evolutiona ctivity.T he samples are labelled as "PtxNiy", where "x" and "y" indicate the nominal thickness of the sputtered Pt and Ni films, respectively. The adoption of this deposition sequence (namely Pt followed by Ni)i sd ictatedb yt he fact that depositing Ni first seems to limit dewetting (metal film agglomeration) especially when using relatively thickN if ilms (showni nF igure S1). This may be ascribed to the susceptibility of Ni to oxidize, for example, by interaction with the underneath oxide surface, resulting in a poor metal atom surface mobility.
To induce dewetting, that is, the rupture of the conformal metal film into metal islands or particles, [33] the PtNi-coated TiO 2 structures were exposed to at hermal treatment in a1 0% H 2 /Ar atmosphere,a t4 50 8Cf or 1h.F or ab i-layer of 5nmP t and 10 nm Ni, dewetting resultsi nt he formation at the TiO 2 NT bottom of % 40 nm sized PtNi NPs, whilet he dewetted metal bilayer forms at the NT mouths (NT top) aP tNi "continuous" decoration network (Figure 1c-d). We chose ar educing atmosphere based on previousw ork where dewetting of metal films such as Ni, Co or Fe, which can be susceptible to oxidation, wasc arriedo ut in H 2 or in vacuum. [42][43][44] The H 2 partial pressure prevents oxidation of the metal film and provides the required atom surface mobility for metal agglomeration and for Pt-Ni intermixing.
After dewetting, the PtNi-decorated TiO 2 NTsw ere immersed in an acidic mixture for 10 mins to induce dealloying, i.e the selectived issolution of Ni (see the Experimental Sectionf or more details). Figure 1e-h show the samples Pt5Ni10, Pt5Ni15 and Pt5Ni20a fter dealloying. For samples Pt5Ni10 and Pt5Ni15 (Figure 1e,f) and particularly for sample Pt5Ni20 (Figure1g,h) one can notice that the dewetted metal particles at the TiO 2 NT bottom are convertedt hrough the dealloying step into nanoporous metal islands. We evaluated also the possibility of Pt losses during the dealloying step and concluded that only negligible amounts of Pt undergo dissolution-see the SI.
To investigate the nanoporous nature of the dewetted-dealloyed Pt co-catalyst, we carriedo ut TEM analysisf or sample Pt5Ni15-dewetted-dealloyed, alongw ith additional SEM investigation at high magnifications. The results are compiled in Figure 2. The high-magnification SEM image in Figure 2a shows that the dewetted-dealloyed co-catalyst is evidently porous, and the pores have as ize is in the 3-20 nm range, with severalv ery smallp ores (e.g. 3-5 nm) and few larger ones (> 10 nm)-the pores are defined by dashed green lines. The TEM image in Figure 2b confirms these results: the homogeneously distributed nm-sized bright areas correspondt ot he pores in the Pt co-catalyst particle. The higherr esolution of TEM allows to detect also pores with as ize of 1-2 nm. Moreover,t he HR-TEMi mage in Figure 2c shows the lattice planes of Pt (111)a nd anatase TiO 2 (101), with as pacingo f2 .2 and 3.5 ,respectively,which fit well to data in the literature.
We characterizedt hese structures by XRD to assess the crystallographic changes occurring during dewetting and dealloying. Figure 3s hows the XRD patterns of structures at different fabrication stages.
The XRD pattern of sample Pt5Ni10 befored ewetting features only the reflections of Ti metal (due to the presence of the Ti metal substrate underneath the TiO 2 NT structure). This confirms the amorphous nature of as-anodized TiO 2 NTs. The absence of Pt andN ir eflections in the as-sputtered samples may on the one hand be ascribed to their low amount at the NT surface (below detection limit), or can on the other hand suggest that as-sputtered Pt and Ni films are amorphous (or features extremely small crystalline domains, whichc an thus not be detected by XRD).
The same sample after dewetting shows ac learly different XRD pattern:firstly,t he thermalt reatment induces the crystallization of the NTsi nto am ixed anatase-rutile TiO 2 phase, as suggested by the appearance of reflectionsa t2 5.28 and 27.48 attributable to the (101) anatase (PDF no:0 0-021-1272)a nd  (110) rutile (PDF no:0 0-021-1276) phases, respectively.M ore importantly,areflection at 42.88 appears (i.e. between the the-oreticalP t(111)a nd Ni(111)p eaks) that can be ascribed to the formationo faPt-Ni alloy (Figure 3b). This confirms that dewetting of Pt-Ni bilayers forms nano-sized Pt-Ni alloy decorations at the TiO 2 NT surface. [36] Beside the Pt-Ni (111)r eflection,a lso the peaks at 47.98 and 49.88 (i.e. between the peaks of Pt( 200) and Ni (200)) can be attributedt oaPtÀNi alloy phase. It is reported that the lattice constant of Pt-Ni alloys is typically lower than that of pure Pt (3.92 )a nd higher than that of pure Ni (3.52 ). [45][46][47][48][49][50] Thus,t he peaks of PtÀNi alloy phase are expected to be slightly shifted towards higher anglesw ith respect to those of pure Pt phase, as in the case of our results.
The XRD datao fs amples Pt5 and Ni10 after dewetting provide afurther confirmation that Pt-Ni alloyingoccurs with dewetting. Here no peak attributable to Pt-Ni alloy phases appears but for sample Ni10-dewetted only an intense diffraction peak at 44.48 can be seen that fits well to the (111)p lane of metallic Ni (JCPDS no.03-065-0380). Thisi ng enerala lso supports that dewetting inducest he agglomeration of the Ni film and crystallization/grain growth in the dewetted Ni islands.F or sample Pt5-dewetted, the main Pt reflection (i.e. (111)) cannot be seen owing to overlap with the strong (101) Ti peak. However,t he formation of crystalline Pt domains upon dewetting is supported by the appearance of the Pt (200) reflection at 46.28.
When comparing the XRD pattern of sample Pt5Ni10-dealloyed with Pt5Ni10-dewetted, it is possible to appreciate the chemicalc hanges inducedb yt he selectiver emoval of Ni on the crystallographic feature of the formed nanoporousm etal: firstly,t he reflection associated to the Pt-Ni alloy decreases in intensity,w hich can be ascribed to ap artial amorphizationo f the dealloyed metal (this is in line with the rearrangement and diffusiono fP ta toms causedb yNi removal);s econdly, such a peak slightly shifts from 42.88 to 42.68,w hich confirms the removal of Ni and the consequente xpansion of the metal lattice. [51] We further characterized the structures by X-ray photoelectron spectroscopy (XPS);a sd iscussed below,t he XPS data (Figure 4) complement the XRD results providing additional evidences of the occurrence of dewetting-alloying and dealloying. Figure 4a shows the XPS spectra in the Pt 4f region for sample Pt5Ni10a td ifferent fabrication stages, that is, as-sputtered,a fter dewetting and after dealloying.W ei ncluded the spectrum of sample Pt5 after dewetting. Compared to sample Pt5-dewetted (reference for pure dewetted Pt), the spectrum of sample Pt5Ni10a s-sputtered is substantially different:t he low XPS signal intensity is caused by the fact that in this sample the Pt film is "buried", that is, coated by the Ni film, while the significantp eak shift towards lower B.E. values can be attributedt oastrong interaction with both the underneath oxide andN ioverlayer.
On the contrary,b oth spectra of samples Pt5Ni10-dewetted and Pt5Ni10-dealloyed only slightly differ from the reference (Pt5-dewetted).T he spectrum of Pt5Ni10-dewetted (thermal treatment in H 2 /Ar) shows the typical signature of Pt metal: [41] that is, two sharp peaks at % 75.1 and 71.7 eV associated to the Pt 4f 5/2 and Pt 4f 7/2 signals, respectively. [52] This confirms the metallics tate of Pt (Pt 0 )i np articles dewetted from both Pt films or Pt-Ni bilayers. Moreover,t he Pt 4f peaks of Pt5Ni10dealloyed show am inor shift towards higherB .E. values (Figure 4a)t hat is causedb yt he selective removal of Ni and is in agreement with resultsreported in previousliterature. [53] The XPS spectra in the Ni 2p region are compiled in Figure 4c-g. Figure 4c showst he spectra of sample Pt5Ni10 assputtered, after dewetting and after dealloying (included is also the spectrum of sample Ni10 after dewetting, as ar eference for dewetting of ap ure Ni film). The fitting of the Ni 2p 3/2 peak for sample Pt5Ni10a s-sputtered (Figure 4e)r eveals a dominant contribution of NiO and Ni(OH) 2 ,w hile the content of Ni metal is relativelyl ow.T he signals of metallicN i, NiO and Ni(OH) 2 peak at 852.5, 854.2 and 853.2 eV,r espectively.T hese resultsc an be ascribed to the high susceptibilityo fN it oa tmospherico xygen, that is, the sputtered Ni metal films undergo immediate oxidation after sputteringw hen exposed to ambient conditions. However,t he fitting of the spectra for sample Ni10 and Pt5Ni10 after dewetting (Figure 4d,f) shows that the main contribution to the Ni 2p 3/2 signal peaks at 853.2 eV,t hat is, the thermal treatment in H 2 /Ar causes the reduction of NiO and Ni(OH) 2 to Ni metal. These results confirm the metallic Moreover, the spectra in Figure 4h show that the longert he duration of the chemical dealloying,t he lower the final Ni content, until reaching ac omplete removal of Ni after a6 0min long dealloying step.
We finally investigated the photocatalytic H 2 generation rate (r H2 )o fT iO 2 NT structures decorated with different co-catalysts,  Figure 5a.T he H 2 evolution was measuredi na gas tight quartz reactor,u nder UV light illumination (LED 365 nm, 100 mW cm À2 )a nd in 20 vol. %e thanol aqueous solution.
The activity of pristine TiO 2 NTsisn egligible, that is, we measured aH 2 evolution rate of 0.06 mLh À1 cm À2 in the absence of co-catalyst. The NT layers used in the present work are rather thin, that is, ca. 200 nm-thick, andt hus althoughi ntense UV light is used forp hoto-activation, only ap oor photo-activity is generally observed. [30,41,[55][56] This can be due to the following factors i) the short thickness of the NT layer is not optimized to fully absorbed the incident photon flux, [56,57] and ii)a high photon flux in the absence of charge transfer co-catalyst (Pt for electron transfer) may generatei nt he NTsahigh density of photo-generated chargec arriers with high recombination rate (short-lived carriers).
The decoration with dewetted Pt NPs leads to higher H 2 evolution rates (12.2 mLh À1 cm À2 ), as also found in previous work. [58,59] This can be ascribed to the formation of aP t/TiO 2 Schottkyj unction that aids TiO 2 CB electron extraction and transfer to the environment, as well as to the intrinsic catalytic activity of Pt for hydrogen recombination. The activity of NTs decorated with dewetted, pure Ni nanoparticles was investigated in ar ecent report of ours. [60] Depending on the Ni film nominal thickness (e.g. 5-10 nm), the resulting r H2 was found to varyi nt he 0.2-0.5 mLh À1 cm À2 range;t his confirms that the dewetted Ni co-catalyst is significantly less active than pure, dewetted Pt. Interestingly,s ample Pt5Ni15-dewetted, that is, NTsf unctionalized with aP t-Ni alloy co-catalyst, shows compared to dewetted Pt as light improvementoft he H 2 evolution rate (17.1 mLh À1 cm À2 ), which can be ascribed to the bimetallic natureo ft he co-catalyst as reported in previous work. [13,[61][62][63] More importantly,t he NTsd ecorated with porousP t (Pt5Ni15-10 minutes long dealloying) show ad ramatic enhancemento ft he H 2 evolution activity,r eachingaproduction rate of 37.2 mLh À1 cm À2 ,w hich is nearly 4times higher than that measured for the dewetted, pure Pt co-catalyst (Pt5-dewetted) in spite of the comparable nominalP tloading.
Please notet hat control experiments proved that if an identical etchingt reatment (dealloying) is applied to sample Pt5dewetted, the resulting activity remains virtually unaltered (see the r H2 of sample Pt5-dealloyed, Figure 5a). Therefore the photocatalytic enhancement hast ob ea scribed to the nanoporous structure of the dealloyed Pt co-catalyst, which providesa ni ncreased surface-to-volume ratio and thus ah igher density of active site at the co-catalyst surface for H 2 evolution (Figure 5b). [64]   We then explored the effect of the dealloying time on the photocatalytic performance. Remarkably,t he data in Figure 5c show that an increase of the dealloying time from 10 to 30 min further enhances the H 2 evolution activityb yafactor 2, hence reaching ap roduction rate of 76.7 mLh À1 cm À2 .A ne xtended etching (60 min) however turnedo ut to be detrimental, leadingt oar H2 comparable to that of structures dealloyed for 10 min.
We have quantified the Pt loading based on ICP-AES data reported in our previous work. [65] An ominally 1nm-thick Pt film corresponds to aP tl oading of 1.41 mgcm À2 .A( nominally) 5nm-thick Pt film (as in the case of the bestp erforming sample Pt5Ni15-dealloyed) corresponds to al oadingo f 7.05 mgcm À2 .T hus, the H 2 evolution activity per unit mass of loaded Pt is 10.9 and 1.7 mL H2 h À1 mg Pt À1 for samples Pt5Ni15dewetted-dealloyed and Pt5-dewetted, respectively.I .e. dealloyed (nanoporous) Pt leads to a6 .4 times highera ctivity compared to the non-porous co-catalyst, in spite of the same precious metal loading.
To confirm the beneficial effect of Pt porousification on the H 2 evolution activity,w et estedaseries of photocatalysts produced by dewetting-dealloying PtÀNi bilayerso fd ifferent nominal thicknesses and Pt:Ni ratios;t he data are summarized in Figure 5d.T he resultsn ot only suggest that aP t:Ni nominal thickness ratio of % 1:3i samost optimized condition to reach ah igh photocatalytic performance but also confirm that for the exploredP t-Ni compositionr atios the activity of the dealloyed, nanoporous co-catalyst is superior than that of nonporousc o-catalysts, that is, dewetted-alloyed Pt-Ni or dewetted, pure Pt decorations.

Conclusions
We reportedo nt he fabricationo fT iO 2 nanotube arrays functionalized with an anoporous Pt co-catalyst produced by a dealloying approachf rom dewetted-alloyed PtÀNi bilayers. For these structures we observed as uperior photocatalytic H 2 generation ability under UV light irradiation comparedt on anotubes loaded with comparable amounts of non-porous co-catalysts such as pure, dewetted Pt or dewetted-alloyed PtÀNi decorations. The photocatalytic enhancemento riginates from the porousm orphologyo ft he Pt co-catalyst, which can enable amore efficient electron transfer kinetics due to alarge surface area and high density of surfacea ctive sites for H 2 evolution. With the present work we provided insights for the synthesis of noble metal-based (co-)catalytic materials with high mass activity for solar energy conversionp rocesses.I nab roader context, the outlined concepts can in principle be extended to the fabrication of al arge variety of nanoporous metals for applicationi nc atalysis, sensing or plasmonics.

Experimental Section
Fabrication of the TiO 2 nanotube layers Self-organized highly ordered TiO 2 nanotube arrays were in the present work formed on Ti substrates by means of electrochemical anodization. Briefly,p rior to anodization, the titanium foils (1.5 2.5 cm 2 ,A dvent Research Materials, 0.125 mm thickness and 99.6 % purity) were cleaned by sonication in acetone, ethanol and deionized water for 15 minutes each step, respectively,f ollowed by drying under aN 2 gas stream. Then, the Ti foils were anodized to fabricate the highly ordered TiO 2 nanotube arrays in ah ot electrolyte based on 3 m HF in o-H 3 PO 4 (Sigma-Aldrich). [41] For the anodization process, at wo-electrode configuration was used where the Ti foil and aP ts heet were the working and counter electrodes, respectively.T he anodization experiments were carried out by applying ap otential of 15 V( for 2h)u sing aD Cp ower supply (VLP 2403 Voltcraft). The electrolyte was kept at 120 8Cd uring anodization. After anodization, the samples were immersed in ethanol for 1h to remove electrolyte remnants and were then dried under a N 2 gas stream.

Formation of the nanoporous Pt co-catalyst
To form the nanoporous Pt co-catalyst on TiO 2 NTs, we used asputter-dewetting-dealloying approach as follows: ·Metal sputtering:aplasma-sputtering machine (EM SCD 500, Leica) was used to sputter Pt and Ni metal thin films using a 99.99 %p ure Pt target (Hauner Metallische Werkstoffe) and a 99.98 %p ure Ni target (Hauner Metallische Werkstoffe). The appliedsputtering current was 16 mA and the pressure inside the sputtering chamber was set to 10 À2 mbar with Ar gas. As discussed below,the optimized sputtering sequence is Pt first followed by Ni. The amountofs puttered metals was in situ controlled by monitoring the metal film nominal thickness using an automated quartz crystal microbalance. ·Thermal dewetting:t he samples were annealed at 450 8Ci na 10 %H 2 /Ar atmosphere (flux = 3Lh À1 )f or 1h,t oi nduce dewetting of the Pt-Ni bi-layers and form alloyedP tNi decorations on the TiO 2 NTs. ·Dealloying:Dealloying was carried out in an acidic medium composed of 37.5 %H NO 3 (70 %), 37.5 %a cetic acid and 25 % H 2 SO 4 (98 %), at room temperature fordifferent dealloying times.

Characterization of the structures
Field-emission scanning electron microscope (FESEM Hitachi S4800) and high resolution transmission electron microscope (HRTEM,P hilips CM30) were employed to characterize the morphology of the samples. X-ray diffraction (XRD) performed with a X'pert Philips MPD (equipped with aP analytical X'celerator detector) using graphite monochromized Cu Ka radiation (l = 1.54056 ) was used to analyze the crystallographic properties of the samples. The chemical composition of the samples was examined by X-ray photoelectron spectroscopy (XPS, PHI 5600 US) and peak positions were calibrated with respect to the C1 speak at 284.8 eV.

Photocatalytic measurements
For the photocatalytic runs we followed ap rotocol described in previous work. [66] The H 2 generation experiments were carried out under illumination with UV light (UV-LED smart Opsytec, 365 nm, 100 mW cm À2 )f or 6h.T he samples were immersed in an aqueous solution of ethanol (20 vol. %) in aq uartz tube cell that was sealed with ar ubber septum. Before running the photocatalytic experiments, the quartz cell was purged with N 2 gas to remove O 2 . 200 mLgas samples were withdrawn at the end of the photocatayt-ic experiments and analyzed by gas chromatography (GCMS-QO2010SE, Shimadzu) to examine the amount of evolved H 2 .T he GC was equipped with at hermal conductivity detector (TCD), a Restek micropacked Shin Carbon ST column (2 m 0.53 mm). GC measurements were conducted at at emperature of 45 8C( isothermal conditions) with the temperature of the injector setup at 280 8Ca nd that of the TCD at 260 8C. The flow rate of the carrier gas (Argon) was 14.3 mL min À1 .