Double-Nozzle Flame Spray Pyrolysis as a Potent Technology to Engineer Noble Metal-TiO 2 Nanophotocatalysts for Efﬁcient H 2 Production

: Noble metal-TiO 2 nanohybrids, NM 0 -TiO 2 , (NM 0 = Pt 0 , Pd 0 , Au 0 , Ag 0 ) have been engineered by One-Nozzle Flame Spray Pyrolysis (ON-FSP) and Double-Nozzle Flame Spray Pyrolysis (DN-FSP), by controlling the method of noble metal deposition to the TiO 2 matrix. A comparative screening of the two FSP methods was realized, using the NM 0 -TiO 2 photocatalysts for H 2 production from H 2 O/methanol. The results show that the DN-FSP process allows engineering of more efﬁcient NM 0 -TiO 2 nanophotocatalysts. This is attributed to the better surface-dispersion and narrower size-distribution of the noble metal onto the TiO 2 matrix. In addition, DN-FSP process promoted the formation of intraband states in NM 0 -TiO 2 , lowering the band-gap of the nanophotocatalysts. Thus, the present study demonstrates that DN-FSP process is a highly efﬁcient technology for ﬁne engineering of photocatalysts, which adds up to the inherent scalability of Flame Spray Pyrolysis towards industrial-scale production of nanophotocatalysts.


Introduction
Photocatalytic water splitting using solar light as a primary energy source is among the most sustainable energy technologies to produce H 2 [1]. To optimize photocatalytic H 2 production by photoactive nanomaterials, a successful engineering strategy is the construction of heterojunctions, comprising of at least two photocatalysts which would offer key advantages, i.e., such as maximum charge separation and optimal energy bands' positioning [2]. Within this context, TiO 2 -based photocatalysts have been widely studied as potent catalytic materials due to their low cost, non-toxicity and photochemical stability [3]. However, due to the fast recombination of photo carriers and backward reactions, the photocatalytic efficiency of bare TiO 2 is considered to be low for H 2 evolution [4]. Literature data show that noble metal nanoparticles (NM 0 ), e.g., such as Au 0 , Pt 0 , Pd 0 , Ag 0 , when appropriately anchored onto TiO 2 nanoparticles, can contribute key advantages: (a) Act as electron acceptors to suppress electron-hole (e − /h + ) recombination, through the formation of a Schottky junction [5]. Taking into account the work function (ϕ) of these metals (ϕ Pt = 5.65 eV, ϕ Pd = 5.30 eV, ϕ Au = 5.21 eV, ϕ Ag = 4.26 eV) and TiO 2 (ϕ TiO2 = 4.20 eV) [6] a Schottky-type upward band bending occurs at the NM 0 -TiO 2 . Thus, in TiO 2 -NM 0 the photoexcited electrons will be preferentially transferred from the TiO 2 semiconductor to the metal, until the Fermi levels of the metal and the semiconductor are aligned [6]. The formed upward band-bending promotes accumulation of electrons from the conduction band of TiO 2 to the noble metal states [5].
(b) Localized Surface Plasmon Resonance (LSPR), which can occur in the case of large enough Au 0 and Ag 0 particles, could further enhance the photocatalytic activity [7]. Noble metal particle size, morphology and surrounding environment, play a crucial role in the effectiveness of LSPR [8]. In this case, the so-called hot-carriers can be injected from the noble metal to the semiconductor [9].
(c) Noble metals, particularly Pt 0 , can serve as reaction sites, i.e. due to low activation energy for surface reduction of H + towards H 2 [10,11].
In last decades, numerous studies have been devoted to optimization of the TiO 2 photoactivity, exploiting the advantages of Pt 0 , Pd 0 , Au 0 , Ag 0 nanoparticles [12][13][14][15][16]. In brief, Keller et al. had examined the H 2 evolution efficiency of Pt 0 and Au 0 /TiO 2 (anatase/rutile) photocatalysts using as substrate water/methanol mixture [13]. Nunez et al. have tested the activity of Pt 0 /TiO 2 photocatalysts to produce H 2 from H 2 O/methanol mixture, and, using a focused irradiation set-up, they reported a H 2 photoproduction of 16.0 mmol g −1 h −1 by a 2%Pt 0 /TiO 2 [14]. Waterhouse et al. have examined the catalytic efficiency of NM 0 /TiO 2 (NM 0 = Au 0 , Pt 0 , Pd 0 ) heterojunction in H 2 generation using different alcohol/H 2 O mixtures, where the co-catalyst activity followed the order Pd 0 >Pt 0~A u 0 [15]. Lai et al. [16] have used an ultrasonication-assisted in situ deposition strategy to decorate plasmonic Ag 0 on TiO 2 nanotube arrays and tested for H 2 production from water [16]. In all the aforementioned works, crucial parameters for effective H 2 evolution were found to be: (1) the nature and content of the NM 0 co-catalyst, (2) surface, crystallographic, and porosity properties of the TiO 2 support, (3) the anatase/rutile ratio, (4) the strong metal-support interactions, (5) the NM 0 dispersion over the TiO 2 matrix.
It is now well documented that NM 0 contribution to H 2 generation by TiO 2 depends critically on the synthetic procedure adopted to produce an appropriately interfaced heterostructure. Thus, a successful noble metal deposition strategy onto TiO 2 should meet, at least, the following prerequisites: (1) proper metal NM 0 -TiO 2 interfacial association, i.e., simple physical adsorption of metal onto the TiO 2 matrix is not enough, (2) high dispersion of noble metal onto the TiO 2 is a key parameter to achieve optimal photocatalytic performance, (3) optimal interfacing of rutile/anatase nano phases, (4) high crystallinity and surface area of TiO 2 semiconductors. The fulfillment of the four requirements has been achieved by some synthesis methods which-though successful-have the drawbacks of being multistep, time-consuming, or achieving only part of the four requirements.
Herein, we have used Flame Spray Pyrolysis technology (FSP) to synthesize NM 0 -TiO 2 nanoparticles in one step. FSP is an established gas-phase combustion process that has been utilized for the production of NM 0 -TiO 2 NPs under controlled conditions [17,18]. More specifically, Huang et al. have synthesized a 10 wt% Pd 0 /TiO 2 catalyst for methane conversion [19], while Li et al., have prepared TiO 2 -supported Pt sub-nanoclusters for CO oxidation [20]. Zhao et al. have produced Pt 0 /TiO 2 catalysts with surface-supported isolated Pt atoms controlling the loading at a very low level with precursor/solvent formulation and flame temperature [21]. To the best of our knowledge, there are quite a few works where FSP technology was used to produce nanophotocatalysts dedicated for H 2 production. Within this context, Selli et al. have investigated the photocatalytic H 2 production from water, using Au 0 /TiO 2, nanoparticles [22]. They attribute the observed higher activity of FSP-made TiO 2 nanoparticles to the higher surface area and improved crystallinity, i.e., compared to commercial titania [22]. Later, the same group synthesized fluorinated-Pt/TiO 2 materials for steam-reforming of methanol [23]. The introduction of monovalent fluoride during the synthesis, as well as the short residence time of the growing oxide particles in the hottest zone of the flame, led to the formation of nanocrystals with smaller anatase and rutile unit cell volume, favoring the photocatalytic activity [23]. In a similar work, the catalytic efficiency of FSP-made Cu-Pt 0 /TiO 2 composites, in H 2 production from methanol photo-steam-reforming [24] was examined. Recently, our group engineered PdO/Pd 0 /TiO 2 nano-heterostructures, for formic acid dehydrogenation, using a novel Sequential-Deposition FSP method [25]. This work demonstrated that the sequence of metal-deposition on the TiO 2 , i.e., concurrent formation/deposition of {Pd 0 NP/TiO 2 } vs. sequential depiction of Pd on pre-formed TiO 2 particles made a significant difference in their catalytic H 2 production form HCOOH [25].
Herein, we have further exploited the potential of FSP technology, producing TiO 2 -NPs decorated with noble metal particles (Pt 0 , Pd 0 , Au 0 , Ag 0 ) with two different FSP-methods, see Figure 1; One Nozzle (ON-FSP) and Double-Nozzle Flame Spray Pyrolysis (DN-FSP). A systematic screening of the two FSP methods was accomplished, using different NM 0loadings onto the TiO 2 matrix. So far, DN-FSP methodology has been used originally by Pratsinis et al. [26] to produce Pt/Ba/Al 2 O 3 particles as catalysts for NO x storage reduction. Later, Baümer and Mädler synthesized Co/Al 2 O 3 Fischer-Tropsch catalysts [27], while Grunwaldt and Mädler produced CoMo/Al 2 O 3 hydrotreating catalysts [28]. Grunwaldt et al. examined the catalytic behavior of MnO x /γ-Al 2 O 3 and FeO x /γ-Al 2 O 3 in CO removal [29]. In that work, it was proven that the nanocatalysts produced by DN-FSP were more efficient [29]. So far, however, the potential of DN-FSP to produce TiO 2 -based photocatalysts has not been studied. Despite this, from all the cited FSP works [24][25][26][27], it becomes obvious that the use of DN-FSP methodology allows production of particles with better controlled properties. This is due to the versatility of the method, i.e., DN-FSP offers a configuration where the parameters of the two Nozzles can be optimized independently [30]. In the present work, one FSP Nozzle was used for the optimization of TiO 2 particle size and phase, while the second Nozzle was operating under conditions optimized for the production of the controllable size of the noble metal particles. Adjustment of the geometry of the two Nozzles, including distance and relative angle of the two flames, allows controllable deposition mass of the noble metal up to TiO 2 particles [31].
Overall, in the present work, our specific aims were: (1) to develop a Double-Nozzle FSP methodology, to produce TiO 2 nanocatalysts, decorated with controllable amount of noble metal particle [NM 0 = Pt 0 , Pd 0 , Au 0 , Ag 0 ]; (2) evaluation of NM 0 -TiO 2 composites for the photocatalytic H 2 production from H 2 O/methanol mixture. More specifically, we focus on the comparison of photocatalytic performance of nanocatalysts synthesized by the two FSP methods. Our data provide a comprehensive insight into the beneficial role of DN-FSP method for synthesis of NM 0 -TiO 2 , with high H 2 production yields.

Materials
The solvents used, xylene and acetonitrile, were of commercial grade obtained from Merck. Ti (titanium(IV) isopropoxide-TTIP, 97% purity) and noble metal organic precursors for the FSP particles synthesis, had purities of 98% (Platinum(II) acetylacetonate, Palladium(II) acetylacetonate, dimethyl acetylacetonate Gold (III), Silver(I) acetylacetonate), by Alfa Aesar. Gas flows (purity > 99.999%) were supplied with mass flow controllers (Bronkhorst El-Flow) calibrated at 273.5 K and 1.013·105 Pa. Catalytic experiments were carried out using Milli Q water and methanol of HPLC-grade and they were purchased from Merck.

Preparation of the NM 0 -TiO 2 Nanohybrids
The Flame Spray Pyrolysis set-up used, has been described in detail in our previous works [25,32,33]. Herein, two different FSP protocols (One-Nozzle, Double-Nozzle) were optimized and implemented, (Figure 1) to obtain the desired NM 0 /TiO 2 nano-hybrids: One-Nozzle FSP Process (ON-FSP): In ON-FSP process, TTIP and noble metal precursors' solutions (NM 0 ) were mixed in xylene/acetonitrile solvent mixture (2.2/1.0 ratio) at a constant Ti-concentration of 0.64 M (Table S1 in the Supplementary Materials). By varying the NM 0 concentration in the solvent mixture, the NM 0 content (0-5 wt% vs. Ti) was adjusted. A liquid precursor solution, containing both Ti and NM 0 atoms, was supplied to the capillary tube at 5 mL min −1 using a syringe pump. A premixed supporting flame mixture of CH 4 and O 2 (2.5/5 L/min) was ignited and stabilized during ON-FSP process ( Figure 1a). The collection of the particles was realized with the aid of a vacuum pump (Busch, Mink MM 1202 AV), using a glass-fiber filter (ALBET-Lab Science, GF6, ∅ 25.7 cm) which was put in the distance of 69 cm. Double-Nozzle FSP Process (DN-FSP): TiO 2 nanoparticles were produced by the DN-FSP reactor. One of the Nozzles, herein codenamed as Burner-1, (see Figure 1b) was dedicated to the formation of the TiO 2 nanoparticles. A 0.64 mol/L Ti precursor solution, prepared with titanium-V isopropoxide diluted in a mixture of 110/50 xylene/acetonitrile, was fed at 5 mL/min with a syringe pump (Lambda Vit-Fit) to the FSP atomizer and dispersed with 5 L/min of oxygen at 1.8 bar pressure drop. The spray was ignited and stabilized by a premixed pilot flame of 5 L/min oxygen and 2.5 L/min methane. The second Nozzle, herein codenamed as Burner-2, (see Figure 1b) was dedicated to the noble metal production. The appropriate amount of noble metal precursor (Table S1) diluted in a mixture of 110/50 xylene to acetonitrile was fed at 3 to 7 mL/min with a syringe pump to the FSP atomizer and dispersed with 3 L/min of oxygen at 1.8 bar pressure drop. The spray was ignited and stabilized by a premixed pilot flame of 5 L/min oxygen and 2.5 L/min methane. The collection of the particles was realized with the aid of a vacuum pump (Busch, Mink MM 1202 AV), using a glass-fiber filter (ALBET-Lab Science, GF6, ∅ 25.7 cm) which was put at a distance of 69 cm from the burner. Herein, the as-produced materials are codenamed according to their w/w noble metal content, i.e., TiO 2 , or 0.1% NM 0 /TiO 2 , 0.25% NM 0 /TiO 2 , 0.5% NM 0 /TiO 2, 5% NM 0 /TiO 2 , where NM 0 = Pt 0 , Pd 0 , Au 0 , Ag 0 respectively (Table 1).

Characterization of the Nanomaterials
X-Ray Diffraction (XRD): The NM 0 -TiO 2 nanomaterials were characterized by powder X-ray diffraction (XRD) on a PANalytical X'PertPRO diffractometer using CuKα radiation, equipped with an X'Celerator detector. The patterns were recorded in the 2-theta (2θ) range from 20 • to 80 • , in steps of 0.02 • and a counting time of 2 s per step. The average crystal size of anatase was calculated using the Scherrer Equation (1) [34]: where d = crystallite size (nm), Ka shape constant (~0.9), λ = the wavelength of Cu Kα radiation (1.5406 Å), β = full width at half maximum, and θ = diffraction angle. Energy Band-Gap Estimation: Diffuse-Reflectance UV-Vis (DR-UV/Vis) spectra were recorded by a PerkinElmer (Lamda 35) spectrophotometer in the wavelength range of 240-800 nm, using as background standard powder BaSO 4 . The allowed indirect bandgap of ON and DN noble metal-TiO 2 nanoparticles was calculated, using the Kübelka-Münk Equation (2) [35]: where hν is the phonon energy, C 1 is a proportionality constant. The commercial AEROXIDE ® TiO 2 P25 was used as reference to compare the band-gap values of TiO 2 FSP samples.
Brunauer-Emmett-Teller (BET) Analysis: The determination of specific surface area (SSA) of the synthesized nanoparticles was realized using a Quantachrome NOVAtouch LX2 porosimeter. The N 2 adsorption-desorption isotherms were recorded at 77 K. The outgassed process was performed at 150 • C for 16 h under vacuum, before the measurements. The data points of the absorption, in the relative pressure P/Po range of 0.1−0.3, were used to calculate the specific surface area (SSA), while the absorption data points in the P/Po range 0.35−0.99 were used for the estimation of pore radius by the BJH method. The total pore volume was obtained at the P/Po = 0.99 data point. The complete isotherm adsorption-desorption BET data sets are presented in Figures S5 and S6 of Supplementary Materials. For the calculation of the average primary particle size, d BET , it was assumed that the NM 0 -TiO 2 nanoparticles were monodisperse spheres, according to Equation (3): where ρ = the weighted density of the particles with TiO 2 (rutile) = 4.23 g/cm 3 , and TiO 2 (anatase) = 3.84 g/cm 3 . Scanning Transmission Electron Microscopy (STEM): The morphological characterization of the samples was performed by Scanning Transmission Electron Microscopy (STEM) using a high-angle annular dark-field detector accompanied by a FEI CM20 microscope operated at 200 kV and provided 0.2 nm resolution. Gatan GIF 200 Energy Filter software was used to analyze obtained results. For every measurement, the sample was ground in a mortar and dry-loaded onto a support film (Lacey Carbon, 300 mesh, (Cu)).
X-ray Photoelectron Spectroscopy (XPS): XPS spectra were collected using a SPECS GmbH spectrometer, under ultrahigh vacuum conditions with a base pressure of 2-5 × 10 -10 mbar in a SPECS instrument equipped with a monochromatic twin Al-Mg anode X-ray source and a multichannel hemispherical sector electron analyzer (HSA-Phoibos 100). In all the experiments, the monochromatized Mg Kα line was set at 1253.6 eV and the analyzer pass energy at 20 eV. The binding energies were calculated concerning the energy of the C 1 s core level at 284.5 eV. Samples were placed on silicon substrates and left in a high vacuum before being placed in the main chamber for XPS measurement. The energy resolution was set to 1.18 eV, and the photoelectron take-off angle was 45 • with respect to the surface normal. Recorded spectra were the average of 4 scans with the energy step set to 0.1-0.2 eV and dwell time of 1 s. The spectral analysis included a Shirley background subtraction and peak deconvolution employing mixed Gaussian-Lorentzian functions in a least-squares curve-fitting program, WinSpec, developed at the Laboratoire Interdisciplinaire de Spectroscopie Electronique, University of Namur, Belgium XPS [36,37].
Photocatalytic H 2 -Evolution Experiments: The photocatalytic reactions were performed into a 250 mL pyrex immersion-well reactor (Photochemical Reactors Ltd., UK, Model 3210), bearing two angle sockets and one vertical socket, cooled by tap water circulation at T = 25 • C, during the whole photocatalytic procedure. The light source was an inlet xenon lamp, equipped with a power supply of 300 Watt (TOPTION Ltd., Xi'an, China, model: TOPL-X300). In each experiment, 50 mg of the catalyst was dispersed in 150 mL water/methanol mixture 20% v/v (final concentration of catalyst 0.33 g/L). Before the reaction start, the suspension was bubbled with Ar gas (99.9997%) for at least 1 h, to remove O 2 . During the photocatalytic reaction, methanol oxidation results in CO 2 and CH 4 generation [10]. At regular time intervals, standard gas volumes were withdrawn from the headspace of the reactor with the aid of a gas-tight syringe (Hamilton Company © , Reno, NV, USA, model 1750) and analyzed with a Gas Chromatography System combined with a Thermo-conductive Detector (GC-TCD-Shimadzu GC-2014, Carboxen-1000 column, Ar carrier gas) [38,39].
Recycling tests of NM 0 -TiO 2 catalysts were realized by washing the used catalyst at least 3 times with milli-Q water and ethanol. Then the powder was collected with centrifugation at 9000 rpm, dried at 70 • C overnight and reused.

Characterization of the FSP-Made TiO 2 -Based Photocatalysts
X-ray Diffraction and STEM: X-ray diffraction profiles are shown in Figure 2a [25]. Rietveld analysis, shows that, in all cases, the anatase/rutile phase ratio did not change, i.e., it was A: R~90:10. In general, no differences were observed in the XRD patterns of TiO 2 , among the different NM 0 -TiO 2 . No XRD diffraction peak of any noble metal was resolved. Taking into account the TEM images where the metal particles with sizes in the range 2-4 nm are resolved, such small sizes give poor diffraction peaks in XRD.
The structural information obtained from the XRD data, summarized in Table 1, show that the particle diameter was 15 nm for ON-FSP and 10 nm for DN-FSP NM 0 /TiO 2 materials, i.e., consistently smaller TiO 2 particle sizes were obtained by DN-FSP methodology. This is in accordance with the d TEM values, which show an average size distribution of 16 STEM images for TiO 2 nanoparticles show spherical TiO 2 structure with an average size close to 15 nm and 10 nm, consistent with the analysis of XRD and BET data. Figure 2d shows that DN-FSP allows formation of finely-dispersed Pt particles on the TiO 2 matrix, while ON-FSP produces more coarse Pt particles (Figure 2c). Taking into consideration the size distribution of Pt nanoparticles onto TiO 2 support, it seems that DN-FSP produces smaller particle size of noble metal (~2 nm) as compared to ON-FSP process (nanoparticles ranged from 4 to 5 nm). This is a clear evidence that DN-FSP technology prevents the formation of agglomerations and favors the formation of finely-dispersed particles.
X-ray Photoelectron Spectroscopy; Representative broad-scan XPS data for FSP-made Au 0 /TiO 2 Pd 0 /TiO 2 materials, are presented in Figure 3. In all cases, characteristic Ti 2p3/2 and Ti 2p1/2 energies were observed at 458 and 463.4 eV, assigned to Ti 4+ in octahedral coordination [40][41][42]. In Figure 3d, the Au 4f spectrum is characterized by one pair of peaks, each pair attributed to the spin-orbit coupling (Au 4f 7/2 and Au 4f ), while the XPS analysis did not show any oxidation of the gold. The pair (BEs of 84 eV and 87.3 eV) is assigned to Au (Au 5/2 ) [41,42]. Similarly, the Pd 3d XPS data for Pd 0 /TiO 2 (Figure 3e  The O 1s peaks at~531 eV and 532 eV, correspond to lattice oxygen and OH hydroxyl groups, respectively [43,44]. Additionally, the binding energy at 528.5 eV corresponds to adsorbed oxygen [45]. The peak~534 eV corresponds to the metal-O bond, which is resolved only in Pd/TiO 2. As analyzed in detail previously [25], this XPS feature is indicative of the strong metal-support interactions between Pd and TiO 2 [25]. It is worth mentioning that in the case of Pt/TiO 2 and Ag/TiO 2 , no XPS signal of Pt or Ag was detected due to low concentration and the fine dispersion of noble metal. Diffuse-Reflectance UV-Vis Spectroscopy (DRS-UV-Vis); The DRS-UV-Vis spectra for the ON-FSP and DN-FSP 0.5%NM 0 /TiO 2 nanoparticles are shown in Figure 4a. For loadings of 0.1, 0.25 and 5%, see Figures S5 and S6 and Table S3 of the Supplementary  Materials. Table 2 summarizes the estimated band-gap (E g ) values. Pristine TiO 2 nanoparticles exhibited the typical photo-response in the UV part with a wavelength lower than 380 nm [45][46][47]. The typical absorption-edge spectrum of TiO 2 at~400 nm is due to rutile fraction [47][48][49]. The spectra for both ON-FSP and DN-FSP particles show a trend towards lower band-gaps, which increases proportionally with the percentage of noble metal loading (Figure 3 and Figures S5 and S6 in Supplementary Materials). The energy-shifts manifested as long tails in the absorbance edge of DRS-UV-Vis profile, can be attributed to (a) the formation of Ti 3+ states [12,48], or/and (b) the Schottky barrier, which contributes to the interface charge transfer interaction between Pt 0 or Pd 0 and TiO 2 [47]. The Au 0 /TiO 2 materials show a characteristic band at 550 nm due to the localized surface plasmon resonance (LSPR) by Au 0 [8]. In the case of Ag 0 /TiO 2 , the LSPR is manifested only at high Ag-loading, i.e., 5% w/w, for the DN-FSP particles ( Figure S6 in Supplementary Materials).
Using the Kübelka-Münk transformation (Equation (2)), the band-gap (Eg) values have been estimated and listed in Table 2. Interestingly, the data reveal that the Eg values follow the trend: (a) E g (Prisitne_TiO 2 ) > Eg(TiO 2 ) ON > Eg(TiO 2 ) DN (b) E g (Au/TiO 2 ) ON >E g (Ag/TiO 2 ) ON > E g (Pt/TiO 2 ) ON > E g (Pd/TiO 2 ) ON (c) E g (Au/TiO 2 ) DN > E g (Ag/TiO 2 ) DN > E g (Pt/TiO 2 ) DN > E g (Pd/TiO 2 ) DN The data in Table 2 show that in the case of DN-FSP materials, the Eg values have a tendency to be smaller than the Eg of the ON-FSP materials. Our Electron Paramagnetic Resonance data (not shown) reveal that in the case of DN-FSP, the produced TiO 2 particles contain a higher amount of reduced Ti 3+ centers. It is well known that surface-reduced Ti 3+ centers create additional intraband states inside the TiO 2 lattice [1,12,46] that favors the decrease of E g . Interestingly, enhancement in absorption intensity of DN-FSP nanoparticles is visually evidenced as the more intense of color (see inset photos of Figure 4a,c) [48,49]). This is in accordance with Pratsinis and Mädler, who have shown that DN-FSP produces highly-dispersed metal particles with better adhesion on the oxide support [26,27]. Table 2. Band-gap values of (Eg) of nanocatalysts 0.5% NM 0 /TiO 2 (One and Double-Nozzle FSP).

Material
One-Nozzle FSP Double-Nozzle FSP

Photocatalytic Hydrogen Production of NM 0 -TiO 2 NPs
The time kinetics for photocatalytic H 2 production are presented in Figure 5a,b, while the normalized rate of produced H 2 (mmol/g·h) is illustrated as column bars. The GC-TCD analysis of the gas products showed that H 2 was the main gas product, while lowconcentration CO 2 and CH 4 were detected as a result of methanol and water reaction path [50]. The concentration of H 2 , CO 2 and CH 4 was 30,000 ppm, 2000 ppm and 12 ppm respectively, calibrated by a standard gas mixture ( Figure S1a,b in Supplementary Materials) [51,52]. Blank tests (no catalyst, dark), did not produce any gas, confirming that the H 2 yield is clearly a photocatalytic process.
The % w/w loading effect of different metals is evidenced by the H 2 production rates listed in Table S4 in Supplementary Materials, with the best one being the 0.5% w/w. In all cases, higher noble metal loading had a negative effect in H 2 production. This can be attributed to shielding of active catalyst sites, inhibition of light penetration by excessive metal deposition and lower cocatalyst dispersion on the surface of the TiO 2 [10].
Comparison between the different metals indicate that the photocatalytic performance follows the trend Pt 0 > Pd 0 > Au 0 >> Ag 0 . This can be explained by taking into consideration the work functions of, Pt 0 , Pd 0 , Au 0 , Ag 0 and TiO 2 (ϕ Pt = 5.65 eV, ϕ Pd = 5.30 eV, ϕ Au = 5.21 eV, ϕ Ag = 4.26 eV, ϕ TiO2 = 4.20 eV) [6]. It is well known that higher energy differences between [metal work function] and [E CB of TiO 2 ] favor stronger Schottky barriers [1,6]. Thus, the upward Schottky band-bending is larger in the case of Pt 0 , thus, the electrons are trapped more efficiently in the conduction band of TiO 2 . Literature data confirm that Pt 0 is one of the best co-catalysts for H 2 production [53][54][55][56]. Fu et al. studied the H 2 production activity from the photocatalytic reforming of glucose over different noble metal-loaded TiO 2 photocatalysts [53]. Their data show that activity followed the order: Pt/TiO 2 > Au/TiO 2 > Pd/TiO 2 > Rh/TiO 2 > Ag/TiO 2 > Ru/TiO 2 which is in accordance with our data for FSP-made TiO 2 catalysts. Along the same lines, Selli et al. examined the photocatalytic activity of pristine and noble metal (Ag, Au, Au-Ag alloy and Pt) and modified TiO 2 catalysts [54], and found that the rate of H 2 production follows the trend Pt/TiO 2 > Au/TiO 2 > Au-Ag/TiO 2 > Ag/TiO 2 . They attributed the low performance of Ag/TiO 2 to the Φ Ag value, which is very close to TiO 2 , resulting in less efficient charge separation [54].  The difference in H 2 production rates of other metals is smaller, with Ag to present the lowest relative improvement (r Ag (ON-FSP) = 0.62 mmol/g·h vs. r Ag (DN-FSP) = 0.95 mmol/g·h. At this point, we should mention that in a previous work [14], using Pt/TiO 2 , a high H 2 production rate of 16 mmol/g·h has been achieved using an optimized irradiation set-up to focus 100% of the light in the sample. Thus it is clearly proven that in addition to the particle optimization, the proper adjustment of the photoreactor setup is detrimental.
Herein, the higher photoactivity of nanomaterials engineered with DN-FSP vs. ON-FSP process, can be due to a combination of reasons, such as; (1) better {metal particle-oxide particle} adhesion (2) lower noble metal particle size, (3) better dispersion of noble metal onto the TiO 2 surface, (4) higher SSA of catalyst, i.e., as confirmed by our XRD, TEM and BET data. Literature confirms that the synthesis method could affect photocatalytic activity, i.e., as detailed in the recent comprehensive review article of Domen [1]. According to Jiang et al. [57], the different deposition processes for the development of Pt nanoparticles on TiO 2 could lead to the selective deposition of noble metal on suitable trapping sites that shorten the transferring pathway of photoexcited electrons to the surface [57]. In this context, FSP-made TiO 2 particles achieve superior photocatalytic activity as a consequence of higher noble metal dispersion, inherently, achieved by the FSP technology [54]. In pertinence to the present study, Grunwaldt et al. [29], when comparing the efficiency of MnOx/Al 2 O 3 and FeOx/Al 2 O 3 heterojunctions engineered by ON-FSP and DN-FSP process, found a higher rate of catalytic CO conversion by the DN-FSP materials [29]. This was attributed to the minimization of mixed phases and higher dispersion MnO x and FeO x onto the matrix material [29].
On the Photocatalytic Mechanism; Taking under consideration, the data on the characterization of the present NM 0 -TiO 2 materials and the photocatalytic activity results, a possible mechanism can be proposed, for the observed trends. Under UV light irradiation, electrons (e − ) are excited from the Valence Band (VB) of TiO 2 to the Conduction Band (CB). In the presence of the noble-metal particle, the electron are transferred from the TiO 2 to the metal. As a consequence, positive holes (h + ) are accumulated in the VB, which, when methanol is present as hole scavenger, they can oxidize CH 3 OH to CO 2 and possibly HCOOH [50]. The noble metal particle acts as an effective electron-acceptor channel, with high e-storage capacity, converting the surface absorbed H + to H 2 [1,10], as observed in Figure 6. Thus, in the presence of the NM 0 the separation of h + −e − pairs is favored, increasing the lifetime of charged carriers, and preventing recombination. In every case, the photoexcitation of TiO 2 electrons occurs as a result of hv > Eg TiO2 (3.2 eV), from the VB to CB. The Fermi energy positioning vs. the metal-work function and the resulting Schottky barrier, play a key role because the photoexcited electrons should travel from the TiO 2 to the metal particle, then from the metal surface to reduce H + to H 2 . The energy barriers are inversely proportional to the metal-work function, and they follow the trend [6]: E Fermi (Ag) > E Fermi (Au) > E Fermi (Pd) > E Fermi (Pt).
This trend is in accordance with the H 2 production efficiency, as depicted in Figure 6.

Conclusions
In the present research, two FSP-process configurations, i.e., One-Nozzle FSP and Double-Nozzle FSP were used for controlled synthesis of the NM 0 -TiO 2 nanocatalysts which demonstrate efficient H 2 generation by H 2 O/methanol. In all NM 0 -TiO 2 materials, the photocatalytic performance followed the trend NM 0 = Pt 0 > Pd 0 > Au 0 > Ag 0 , which is in accordance with the formation of higher Schottky barriers upon contact of TiO 2 with Pt 0 , Pd 0 , Au 0 , Ag 0 , respectively. The present data show that Double-Nozzle FSP is superior vs. One-Nozzle FSP for engineering of finely-dispersed noble metal on TiO 2 support. Very low noble metal loadings, i.e., 0.5% are optimal for the photocatalytic performance. Overall, the present data provide strong evidence that DN-FSP is a low-cost, scalable, one-step technology for production of efficient photocatalysts.