Characterization and Effect of Ag ( 0 ) vs . Ag ( I ) Species and Their Localized Plasmon Resonance on Photochemically Inactive TiO 2

Commercial TiO2 (anatase) was successfully modified with Ag nanoparticles at different nominal loadings (1%–4%) using a liquid impregnation method. The prepared materials retained the anatase structure and contained a mixture of Ag0 and AgI species. Samples exhibited extended light absorption to the visible region. The effect of Ag loading on TiO2 is studied for the photocatalytic reduction of CO2 to CH4 in a gas–solid process under high-purity conditions. It is remarkable that the reference TiO2 used in this work is entirely inactive in this reaction, but it allows for studying the effect of Ag on the photocatalytic process in more detail. Only in the case of 2% Ag/TiO2 was the formation of CH4 from CO2 observed. Using different light sources, an influence of the localized surface plasmon resonance (LSPR) effect of Ag is verified. A sample in which all Ag has been reduced to the metallic state was less active than the respective sample containing both Ag0 and Ag+, indicating that a mixed oxidation state is beneficial for photocatalytic performance. These results contribute to a better understanding of the effect of metal modification of TiO2 in photocatalytic CO2 reduction.


Introduction
In recent decades, atmospheric CO 2 concentration has been increasing annually mainly due to the extensive use of fossil fuels [1,2].Data from NASA show that in April 2018 the concentration of CO 2 reached 407.45 ppm, an increase of 29.23% compared to the respective concentration for April 1958 [2].As a result, the annual mean global surface temperature increased-as CO 2 is a major greenhouse gas contributing to climate change-by 0.99 • C, the highest rise observed since 1880 [3].Many efforts and experimental research have been focused on the capture of CO 2 from the atmosphere and/or its conversion to other economically valuable compounds.Since CO 2 is a very stable molecule with a C = O direct bond energy of +805 kJ/mol, it requires a high energy input for bond cleavage [4].One of the most promising technologies to convert CO 2 to useful hydrocarbons is the photocatalytic reduction (the chemical transformation of CO 2 through the absorption of photons of appropriate wavelength from a semiconducting material-photocatalyst) which offers a solution for both the energy crisis and global warming [5].The photocatalytic conversion of CO 2 to value-added hydrocarbons and renewable fuels, such as methane or methanol, is part of a greater effort in the literature to utilize small molecules such as H 2 and CO 2 , collectively termed as "artificial photosynthesis" [6][7][8][9][10].
As a semiconductor material, titanium dioxide (TiO 2 ) has attracted tremendous interest due to its good photocatalytic activity, low toxicity, chemical stability, abundance in nature, and low cost [11,12].However, there are several factors which hinder the photocatalytic efficiency of TiO 2 , such as its absorption range which is limited to wavelengths in the UV region (about 4% of total solar spectrum) [13,14].Another significant factor reported in many studies is the recombination of charge carriers (electron-hole recombination) in the bulk or the surface of the TiO 2 material [15].Schneider et al. proposed that nearly 90% of all photogenerated electrons recombine rapidly after the electron excitation process.As a result, this phenomenon triggers low quantum yields (<10%) [16].
Therefore, much effort has been focused on enhancing the photocatalytic activity of TiO 2 , including doping with metal or non-metal species [17,18], surface sensitization [19,20], and noble metal deposition [21][22][23][24][25].In particular, the noble metal deposition has attracted a lot of attention since it can improve e − -h + separation and enhance the semiconductor's absorption in the visible light region [26,27].Noble metals such as Au, Pt, Ag, Pd, Rh, and coinage metals such as Ni etc., have lower Fermi levels than TiO 2 .Therefore, they can act as electron traps, facilitating the separation of charge carriers and preventing their recombination [26].On the other hand, the localized surface electron plasmon resonance effect (LSPR) induced by noble metals contributes to the activation of TiO 2 absorption in the visible region of the light spectrum.
There are several studies in the literature about photocatalytic CO 2 reduction using Ag/TiO 2 composites [28][29][30][31][32].The results vary, depending on the materials' synthesis and reaction conditions.For example, previous studies revealed that CH 3 OH is the major CO 2 reduction product in liquid-phase experiments, while CH 4 is the main product in gas-phase reactions [33].A gas-phase system offers several advantages like the variation of the concentration of reactants in a wider range without dissolution phenomena [34].Furthermore, in gas-solid reactor systems, experiments using high purity conditions can be established as a high vacuum can be applied to eliminate gaseous C-containing impurities from the reaction chamber [35].Performing CO 2 reduction experiments using conditions of high-purity is extremely important as any source of carbon-containing species can lead to an overestimation of the products.Due to the high stability of the CO 2 molecule, these C-impurities are expected to react faster than CO 2 , leading to hydrocarbons that do not originate from the actual photocatalytic process facilitated by the tested material.
Furthermore, a photocatalytic CO 2 reduction experiment can be performed either in batch or in flow-through mode [35].On the one hand, flow conditions lead to faster reaction rates.However, since the concentration of even the major photocatalytic CO 2 reduction products are quite low, it is very difficult to identify and accurately quantify them during flow-through conditions.On the other hand, a batch mode process allows for a more accurate estimation of the products' yields even at lower concentrations, as the amount of molecules accumulate over the total reaction time.Besides the use of H 2 O, both in liquid-and gas-phase, the CO 2 photoconversion into CH 4 can also be carried out by mediating hydrogen at room temperature and atmospheric pressure called photo-methanation [36].Furthermore, literature concerning the effect of Ag oxidation state on CO 2 photoreduction activities is rarely studied.Most of the research on CO 2 photoreduction only focuses on the effect of the Ag metallic co-catalyst which contributes through the electron trapping and LSPR mechanism [33,37].
In this study, Ag nanoparticles were deposited on commercial TiO 2 (anatase) using the liquid impregnation method.This method was employed as it is simple, fast, and of low cost [38,39].The physical and chemical properties of the synthesized Ag/TiO 2 composites were extensively studied and characterized using X-ray powder diffraction (XRD), scanning electron microscopy (SEM), ultra violet-visible (UV-Vis) spectrophotometry, X-ray photoelectron spectroscopy (XPS), and transmission electron microscopy (TEM).Furthermore, the assessment of the photocatalytic CO 2 reduction activity of the aforementioned materials, the study of the influence of Ag loading, and the effect of mixed Ag species and the LSPR on the products' concentration was carried out in a high purity gas-phase photoreactor.

Synthesis of Photocatalysts
All investigated Ag/TiO 2 materials have been successfully prepared by facile impregnation methods.Commercially available TiO 2 (anatase, ≥99.0%,Merck) was used in this study, both as a starting and as a reference material.The Ag decorated TiO 2 photocatalysts were prepared using a modified liquid impregnation method [14].The resulting Ag modified TiO 2 photocatalysts were referred to as Ag/TiO 2 X% (X = 1-4), whereby X is the nominal silver loading.

Powder X-ray Diffraction
X-ray diffraction techniques were used to determine crystallinity and structural changes which may occur during Ag metal deposition.Figure 1 shows the XRD powder patterns of the pristine TiO 2 and the various Ag/TiO 2 samples calcined at 500 • C in air atmosphere.As depicted, all prepared materials exhibited high crystallinity.There were no detectable changes in the crystal structure of the materials during Ag deposition.All materials retained the tetragonal crystal structure of TiO 2 anatase.The main reflexes at 2θ~25 • , 38 • , and 48 • were attributed to the (101), (004), and (200) planes of TiO 2 anatase (JCPDS File No. 83-2243), respectively.For Ag/TiO 2 materials, small reflexes at 2θ~44 • indicated the presence of Ag 0 species with a lattice plane of (200) (JCPDS File No. 04-0783).The emergence of Ag 0 species is due to the partial reduction of Ag + (AgNO 3 ) under high temperature conditions used in the synthetic protocol.The reflex characteristic of the Ag 0 species can be clearly seen for the 2%, 3%, and 4% Ag/TiO 2 composites.Furthermore, there were also small reflexes (2θ~33.2• ) which corresponded to the diffraction pattern of (111) planes of Ag 2 O species (JCPDS File No. .In addition, it can be assumed that Ag was not incorporated into the TiO 2 structure, but deposited on its surface (vide infra) [11].
From the powder XRD data it can be concluded that the as-synthesized materials contained a mixture of Ag 0 and Ag I species, and that the Ag was not completely reduced to Ag 0 during the synthesis.This phenomenon occurs due to the calcination in air atmosphere, where O 2 inhibits the complete reduction of all Ag species.The calcination process at 500 • C decomposes AgNO 3 into different Ag species, NO x gas, and O 2 gas.
The average crystallite size of deposited Ag for each sample was calculated from the broadening of the cubic Ag (200) reflex using the Scherrer equation (Equation (1)): where D is the crystallite size, λ is the X-ray wavelength used in this analysis, β 1/2 is the full width at half maximum, and θ is the main reflex of XRD pattern.Based on this calculation, the size of deposited Ag nanoparticles varies from 28 nm to 37 nm (Ag/TiO 2 4% exhibited the largest crystallite size among the tested samples).Further SEM and TEM analyses were used to support these data (see Figures 5  and 6).From the powder XRD data it can be concluded that the as-synthesized materials contained a mixture of Ag 0 and Ag I species, and that the Ag was not completely reduced to Ag 0 during the synthesis.This phenomenon occurs due to the calcination in air atmosphere, where O2 inhibits the complete reduction of all Ag species.The calcination process at 500 °C decomposes AgNO3 into different Ag species, NOx gas, and O2 gas.
The average crystallite size of deposited Ag for each sample was calculated from the broadening of the cubic Ag (200) reflex using the Scherrer equation (Equation 1): where D is the crystallite size, λ is the X-ray wavelength used in this analysis, β1/2 is the full width at half maximum, and θ is the main reflex of XRD pattern.Based on this calculation, the size of deposited Ag nanoparticles varies from 28 nm to 37 nm (Ag/TiO24% exhibited the largest crystallite size among the tested samples).Further SEM and TEM analyses were used to support these data (see Figures 5 and 6).

Electronic Spectra
To study the optical properties and light absorption effects of Ag species deposited on TiO2, UV-Vis spectra were collected from the tested samples in a pressed disk form.Figure 2a demonstrates the absorption properties of the as-synthesized materials in the wavelength range 250-800 nm.As it is clearly seen, the reference TiO2 sample has only UV light absorption.The deposited Ag species contributed to an increase in the material's absorption towards the visible region as seen in the 420-700 nm wavelength range (see inset graph).This is in accordance with previous work on visible light absorption in Ag nanoparticles species at 410-425 nm due to the surface plasmon resonance effect [40][41][42][43][44][45][46] (Figure 2b).The absorption, with a maximum near 425 nm, can be attributed to the plasmon absorption of spherical Ag nanoparticles [45], whereas the broad absorption reaching down to ~700 nm may be attributed to differently shaped Ag nanoparticles, such as cube-like structures [45], or to the absorption of Ag2O [47].

Electronic Spectra
To study the optical properties and light absorption effects of Ag species deposited on TiO 2 , UV-Vis spectra were collected from the tested samples in a pressed disk form.Figure 2a demonstrates the absorption properties of the as-synthesized materials in the wavelength range 250-800 nm.As it is clearly seen, the reference TiO 2 sample has only UV light absorption.The deposited Ag species contributed to an increase in the material's absorption towards the visible region as seen in the 420-700 nm wavelength range (see inset graph).This is in accordance with previous work on visible light absorption in Ag nanoparticles species at 410-425 nm due to the surface plasmon resonance effect [40][41][42][43][44][45][46] (Figure 2b).The absorption, with a maximum near 425 nm, can be attributed to the plasmon absorption of spherical Ag nanoparticles [45], whereas the broad absorption reaching down to ~700 nm may be attributed to differently shaped Ag nanoparticles, such as cube-like structures [45], or to the absorption of Ag 2 O [47].In order to further investigate the electronic properties of the Ag/TiO2 materials, additional UV-Vis diffuse reflectance spectra were collected for all samples.To this aim, the Kubelka-Munk function, F(R), allows the optical absorbance of the materials to be approximated from their reflectance (R) according to [49] (Figure 3a).The band gap energy of the tested samples can be estimated from the Tauc plot, using equations from the literature [50].In order to further investigate the electronic properties of the Ag/TiO 2 materials, additional UV-Vis diffuse reflectance spectra were collected for all samples.To this aim, the Kubelka-Munk function, F(R), allows the optical absorbance of the materials to be approximated from their reflectance (R) according to [49] (Figure 3a).The band gap energy of the tested samples can be estimated from the Tauc plot, using equations from the literature [50].In order to further investigate the electronic properties of the Ag/TiO2 materials, additional UV-Vis diffuse reflectance spectra were collected for all samples.To this aim, the Kubelka-Munk function, F(R), allows the optical absorbance of the materials to be approximated from their reflectance (R) according to [49] (Figure 3a).The band gap energy of the tested samples can be estimated from the Tauc plot, using equations from the literature [50].The optical band gap energy calculation was determined by extrapolating the linear region of the plot hv vs (F(R)hv) 1/2 as shown in Figure 3b.The calculation shows that the band gap energy of the prepared materials is in the 3.15-3.20eV range.This slight shift indicates that the deposition of Ag (1%-4%) on the surface of TiO 2 does not significantly affect the band gap energy.

Determination and Characterization of Surface Ag on TiO 2
X-ray photoelectron spectroscopy was implemented to investigate the surface chemical state and composition of the as-prepared samples.Figure 4a shows the survey spectrum of the 4% Ag/TiO 2 sample.From the area Ag 3d and Ti 2p photoelectron peaks, and respective sensitivity factors, the surface Ag:Ti atomic percentage was estimated to be 20%.The positive deviation respect to the nominal Ag:Ti ratio suggests that Ag particles are indeed decorating the TiO 2 surface-photoelectron spectroscopy is largely surface sensitive where the Ag is located.Besides the peak from Ti, O, and Ag, traces of C (C 1s) can be detected.This is usually ascribed either to carbon impurities in the titania, or to adventitious carbon during the XPS measurement.
X-ray photoelectron spectroscopy was implemented to investigate the surface chemical state and composition of the as-prepared samples.Figure 4a shows the survey spectrum of the 4% Ag/TiO2 sample.From the area Ag 3d and Ti 2p photoelectron peaks, and respective sensitivity factors, the surface Ag:Ti atomic percentage was estimated to be 20%.The positive deviation respect to the nominal Ag:Ti ratio suggests that Ag particles are indeed decorating the TiO2 surface-photoelectron spectroscopy is largely surface sensitive where the Ag is located.Besides the peak from Ti, O, and Ag, traces of C (C 1s) can be detected.This is usually ascribed either to carbon impurities in the titania, or to adventitious carbon during the XPS measurement.From the high-resolution X-ray photoelectron (XP) spectrum of Ag 3d (Figure 4b), the binding energy (BE) corresponding to the Ag 3d5/2 peak appears at 367.32 eV.In addition, the BE of Ag 3d3/2 is located at 373.32 eV.Those two peaks indicate the formation of Ag I species [51].The peak is asymmetric and broad and therefore it is designated to two Ag species (Ag I and metallic Ag).This is in accordance with results from XRD measurements (see above).The asymmetry and broadness disappears in fully reduced samples where metallic Ag is present only.Furthermore, it is noteworthy, that XPS also detected the presence of potassium most probably originating from impurities From the high-resolution X-ray photoelectron (XP) spectrum of Ag 3d (Figure 4b), the binding energy (BE) corresponding to the Ag 3d 5/2 peak appears at 367.32 eV.In addition, the BE of Ag 3d 3/2 is located at 373.32 eV.Those two peaks indicate the formation of Ag I species [51].The peak is asymmetric and broad and therefore it is designated to two Ag species (Ag I and metallic Ag).This is in accordance with results from XRD measurements (see above).The asymmetry and broadness disappears in fully reduced samples where metallic Ag is present only.Furthermore, it is noteworthy, that XPS also detected the presence of potassium most probably originating from impurities contained in the precursors used in the synthesis of the materials.The shoulder at high BE in the Ag 3d region is assigned to K 2s.This impurity may contribute to the low photochemical activity because this is known for the comparable impact of sodium in CO 2 reduction (see below) [34].
Figure 4c illustrates the Ti 2p spectrum of the same Ag/TiO 2 4% sample.The peak for Ti 2p 3/2 can be fitted with two components, one located at 458.07 eV and the other at 458.62 eV.Both peaks are attributed to Ti IV .The small positive shift (ca.0.6 eV) is due to the strong interaction between TiO 2 and Ag species leading to electron transfer from the former to the latter [52,53].Another peak located at 463.83 eV is assigned to Ti 2p 1/2 .The splitting between Ti 2p 3/2 and Ti 2p 1/2 is 5.75 eV indicates that Ti exists as Ti IV species [54].Meanwhile, no shoulder peaks at BE 462.5 eV and 456.8 eV for Ti 2p 1/2 and Ti 2p 3/2 core level of Ti III appear, indicating that no Ti III species exist on the surface of the material [55].
Figure 4d depicts the fitting of the O 1s peak.An asymmetric peak of O 1s can be deconvoluted into two peaks.These peaks are located at 529.29 eV and 530.28 eV corresponding to the Ti-O [56] and H-O bonds [57], respectively.All those peaks are corresponding to O 2− species.

Electron Microscopy and Energy-Dispersive X-ray Spectroscopy
In order to investigate the surface morphology of the samples, SEM pictures were collected (Figure 5).No significant morphology differences between the pure TiO 2 anatase and the Ag/TiO 2 can be found (not shown).SEM results show agglomerated spherical TiO 2 particles with crystallite diameters in the 100-200 nm range.From Figure 5 it can be clearly seen that the Ag nanoparticles emerge (appearing as white bright spheres) on the surface of TiO 2 .The size of the Ag nanoparticles varies from 2-50 nm.It is in accordance with calculations from the powder XRD data (see above).The energy dispersive X-ray (EDX) analysis confirmed that the actual Ag loading is 1.1%, 1.8%, 2.1%, and 3.6% for Ag/TiO 2 1%, Ag/TiO 2 2%, Ag/TiO 2 3%, and Ag/TiO 2 4%, respectively.Small deviations between nominal and actual loading may be due to the uneven distribution of the Ag nanoparticles on the surface of the Ag-TiO 2 composite.
attributed to Ti .The small positive shift (ca.0.6 eV) is due to the strong interaction between TiO2 and Ag species leading to electron transfer from the former to the latter [52,53].Another peak located at 463.83 eV is assigned to Ti 2p1/2.The splitting between Ti 2p3/2 and Ti 2p1/2 is 5.75 eV indicates that Ti exists as Ti IV species [54].Meanwhile, no shoulder peaks at BE 462.5 eV and 456.8 eV for Ti 2p1/2 and Ti 2p3/2 core level of Ti III appear, indicating that no Ti III species exist on the surface of the material [55].
Figure 4d depicts the fitting of the O 1s peak.An asymmetric peak of O 1s can be deconvoluted into two peaks.These peaks are located at 529.29 eV and 530.28 eV corresponding to the Ti-O [56] and H-O bonds [57], respectively.All those peaks are corresponding to O 2− species.

Electron Microscopy and Energy-Dispersive X-Ray Spectroscopy
In order to investigate the surface morphology of the samples, SEM pictures were collected (Figure 5).No significant morphology differences between the pure TiO2 anatase and the Ag/TiO2 can be found (not shown).SEM results show agglomerated spherical TiO2 particles with crystallite diameters in the 100-200 nm range.From Figure 5 it can be clearly seen that the Ag nanoparticles emerge (appearing as white bright spheres) on the surface of TiO2.The size of the Ag nanoparticles varies from 2-50 nm.It is in accordance with calculations from the powder XRD data (see above).The energy dispersive X-ray (EDX) analysis confirmed that the actual Ag loading is 1.1%, 1.8%, 2.1%, and 3.6% for Ag/TiO2 1%, Ag/TiO2 2%, Ag/TiO2 3%, and Ag/TiO2 4%, respectively.Small deviations between nominal and actual loading may be due to the uneven distribution of the Ag nanoparticles on the surface of the Ag-TiO2 composite.TEM images of the 2% and 4% Ag/TiO 2 are shown in Figure 6.As shown in Figure 6a,b, Ag (appearing as small dark spheres) is distributed unevenly on the TiO 2 covering only selected areas of the TiO 2 surface, which is in accordance with previous research [22].Furthermore, the size distribution of Ag nanoparticles also varies as seen from Figure 6c.From the TEM images (Figure 6c-e) the lattice fringes allow the identification of crystallographic spacing.The d-spacing value of 0.35 nm corresponds to the anatase (101) plane [57,58], whereas the d values of 0.28 nm and 0.16 nm, respectively, represent the plane of ( 111) and (220) for Ag 2 O species [59][60][61].These results strongly support the results from powder X-ray diffraction (see above).
TEM images of the 2% and 4% Ag/TiO2 are shown in Figure 6.As shown in Figure 6a,b, Ag (appearing as small dark spheres) is distributed unevenly on the TiO2 covering only selected areas of the TiO2 surface, which is in accordance with previous research [22].Furthermore, the size distribution of Ag nanoparticles also varies as seen from Figure 6c.From the TEM images (Figure 6ce) the lattice fringes allow the identification of crystallographic spacing.The d-spacing value of 0.35 nm corresponds to the anatase (101) plane [57,58], whereas the d values of 0.28 nm and 0.16 nm, respectively, represent the plane of ( 111) and (220) for Ag2O species [59][60][61].These results strongly support the results from powder X-ray diffraction (see above).

Sample Testing in Photocatalytic CO2 Reduction
As carbon-containing impurities adsorbed onto the surface of the photocatalysts can lead to overestimation of the concentration of the products of CO2 reduction, special attention must be paid to sufficiently remove them.These impurities might originate from the synthesis process (from precursors, solvents, etc.) or from adsorbed species (e.g., atmospheric CO2 and hydrocarbons).As explained in detail in the experimental section, extensive cleaning of the materials in a gas-phase of only water in inert gas was performed.This cleaning approach usually ensures that impurities are sufficiently removed and that the observed products in photocatalytic CO2 reduction originate predominantly from true photocatalytic reaction between both H2O and CO2.However, this is not the case here.All four Ag loaded samples were tested under conditions of high-purity in photocatalytic CO2 reduction and compared to the reference TiO2 material.Only in the case of Ag/TiO2 2% was a CH4 production not originating from residual C-containing impurities identified.In all other samples (Ag/TiO2 1%, 3%, and 4% and reference TiO2), the levels of CH4 produced in the cleaning procedure (i.e., in absence of CO2) and in the actual reaction with CO2 and H2O in the gas-

Sample Testing in Photocatalytic CO 2 Reduction
As carbon-containing impurities adsorbed onto the surface of the photocatalysts can lead to overestimation of the concentration of the products of CO 2 reduction, special attention must be paid to sufficiently remove them.These impurities might originate from the synthesis process (from precursors, solvents, etc.) or from adsorbed species (e.g., atmospheric CO 2 and hydrocarbons).As explained in detail in the experimental section, extensive cleaning of the materials in a gas-phase of only water in inert gas was performed.This cleaning approach usually ensures that impurities are sufficiently removed and that the observed products in photocatalytic CO 2 reduction originate predominantly from true photocatalytic reaction between both H 2 O and CO 2 .However, this is not the case here.All four Ag loaded samples were tested under conditions of high-purity in photocatalytic CO 2 reduction and compared to the reference TiO 2 material.Only in the case of Ag/TiO 2 2% was a CH 4 production not originating from residual C-containing impurities identified.In all other samples (Ag/TiO 2 1%, 3%, and 4% and reference TiO 2 ), the levels of CH 4 produced in the cleaning procedure (i.e., in absence of CO 2 ) and in the actual reaction with CO 2 and H 2 O in the gas-phase were identical (not shown).So, for all samples except Ag/TiO 2 2%, CH 4 formation can be attributed solely to left-over impurities.Consequently, only Ag/TiO 2 2% is further considered in comparison with the reference TiO 2 (Figure 7).
It is remarkable that the reference TiO 2 used in this work was almost inactive in photocatalytic CO 2 reduction (Figure 7), although a broad-band 200 W Hg/Xe light source was used that was able to excite TiO 2 .Not only was the CH 4 yield very low (~1.2 ppm), it was also identical in the cleaning procedure and the CO 2 reduction experiment.The low to non-existent activity may have to do with TiO 2 being present only in anatase structure, because usually P25 with both rutile and anatase components (~80/20%) is used and shows high activity [32,34,35,62].Furthermore, the titania may be highly defective, so that recombination events are predominant.While we cannot clearly give an answer to the inactivity of the reference TiO 2 , it is obvious that the presence of Ag induces photocatalytic activity, the origin of which is further studied.
Catalysts 2019, 9, x FOR PEER REVIEW 9 of 16 phase were identical (not shown).So, for all samples except Ag/TiO2 2%, CH4 formation can be attributed solely to left-over impurities.Consequently, only Ag/TiO2 2% is further considered in comparison with the reference TiO2 (Figure 7).It is remarkable that the reference TiO2 used in this work was almost inactive in photocatalytic CO2 reduction (Figure 7), although a broad-band 200 W Hg/Xe light source was used that was able to excite TiO2.Not only was the CH4 yield very low (~1.2 ppm), it was also identical in the cleaning procedure and the CO2 reduction experiment.The low to non-existent activity may have to do with TiO2 being present only in anatase structure, because usually P25 with both rutile and anatase components (~80/20%) is used and shows high activity [32,34,35,62].Furthermore, the titania may be highly defective, so that recombination events are predominant.While we cannot clearly give an answer to the inactivity of the reference TiO2, it is obvious that the presence of Ag induces photocatalytic activity, the origin of which is further studied.
The 2% Ag loaded sample exhibited considerably higher CH4 production compared to the bare anatase reference TiO2 (Figure 7).Furthermore, almost twice as much CH4 was formed, as in the case of the cleaning procedure, so at least half of all products must originate from CO2 as carbon source.The higher photocatalytic efficiency of the Ag decorated TiO2 sample is often attributed to the fact that the Ag species act as electron scavengers and therefore the recombination of the photo-generated e − -h + pairs is hindered [63].The photo-generated electron transfer from the TiO2 conduction band to the Ag species deposited on the surface is thermodynamically possible due to the higher Fermi energy level of TiO2 compared to Ag [64].As a result, the formation of a Schottky barrier occurs at the interface between Ag and TiO2, which can further improve the charge separation and consequently, the photocatalytic activity [65].The Schottky barrier prevents the back-transfer of electrons from Ag to TiO2, thus they remain trapped in the Ag species.Since the titania itself is inactive in the present work, two explanations may account for the improved activity of the Ag/TiO2 sample: Either recombination in the bare titania is so much favored that charge separation is only possible in presence of Ag, or titania contributes nothing to the observed photoactivity, because it is only due to the localized surface plasmon resonance (LSPR) effect.The LSPR effect induced by the presence of Ag nanoparticles in the visible region can boost the energy of trapped electrons through the strong local electron field of LSPR, making it react with electron acceptors more easily [28].This synergistic effect of the LSPR effect and the e − -h + separation can lead to an enhancement in the photocatalytic activity as shown in Figure 2b.
In order to test how the LSPR effect influences the CH4 production, a CO2 reduction experiment was performed with the Ag/TiO2 2% sample, using a 10 W 365 nm high-power LED while all the other experimental parameters remained the same.As it can be seen from Figure 7, the CH4 The 2% Ag loaded sample exhibited considerably higher CH 4 production compared to the bare anatase reference TiO 2 (Figure 7).Furthermore, almost twice as much CH 4 was formed, as in the case of the cleaning procedure, so at least half of all products must originate from CO 2 as carbon source.The higher photocatalytic efficiency of the Ag decorated TiO 2 sample is often attributed to the fact that the Ag species act as electron scavengers and therefore the recombination of the photo-generated e − -h + pairs is hindered [63].The photo-generated electron transfer from the TiO 2 conduction band to the Ag species deposited on the surface is thermodynamically possible due to the higher Fermi energy level of TiO 2 compared to Ag [64].As a result, the formation of a Schottky barrier occurs at the interface between Ag and TiO 2 , which can further improve the charge separation and consequently, the photocatalytic activity [65].The Schottky barrier prevents the back-transfer of electrons from Ag to TiO 2 , thus they remain trapped in the Ag species.Since the titania itself is inactive in the present work, two explanations may account for the improved activity of the Ag/TiO 2 sample: Either recombination in the bare titania is so much favored that charge separation is only possible in presence of Ag, or titania contributes nothing to the observed photoactivity, because it is only due to the localized surface plasmon resonance (LSPR) effect.The LSPR effect induced by the presence of Ag nanoparticles in the visible region can boost the energy of trapped electrons through the strong local electron field of LSPR, making it react with electron acceptors more easily [28].This synergistic effect of the LSPR effect and the e − -h + separation can lead to an enhancement in the photocatalytic activity as shown in Figure 2b.
In order to test how the LSPR effect influences the CH 4 production, a CO 2 reduction experiment was performed with the Ag/TiO 2 2% sample, using a 10 W 365 nm high-power LED while all the other experimental parameters remained the same.As it can be seen from Figure 7, the CH 4 concentration is much higher when the Hg/Xe lamp (containing wavelengths in both UV and visible range of the light spectrum) was used compared to the CH 4 production under LED illumination.This means that the absorption of the Ag/TiO 2 2% sample in the 420-550 nm range due to the LSPR effect is the main reason for the observed photocatalytic activity of this sample.
The observation of favorable performance of a sample with ~2 wt % Ag is in a good agreement with Zhao et al. [37].They reported that the optimal Ag concentration on TiO 2 was 2 wt % in simultaneous H 2 production and photocatalytic CO 2 reduction.Similar results were also obtained by Liu et al. [28] and Yu et al. [30].Based on Liu et al. the composite of 2.5% Ag/TiO 2 exhibits the best activity in CO 2 reduction compared with other Ag loadings.On the other hand, Yu et al. proposed that 1.5% Ag/TiO 2 is the best material.This is due to the further increase of Ag loading triggering the LSPR peak to become slightly red-shifted and broader, implying that the Ag nanoparticles begin to agglomerate.However, the exact CH 4 production rate for those studies was different due to the difference of both material properties and reactor set up.
Additionally, it must be clarified which oxidation state of Ag is required for high photoactivity.In order to study the effect of the Ag oxidation state, a sample in which all Ag has been reduced to the metallic state (namely Ag 0 /TiO 2 2%) was tested in CO 2 reduction under similar conditions as the Ag/TiO 2 2%, which contained both Ag 0 and Ag I .As it can be seen from Figure 7, a mixed oxidation state is photocatalytically beneficial.One explanation for this higher photocatalytic activity may be the presence of a Ag 2 O-TiO 2 interface that may act as an e -trapping layer compared to the pure Ag 0 loaded sample.As reported in [66] an interfacial charge transfer (IFCT) takes place between Ag 2 O and TiO 2 under band gap irradiation.This is facilitated by the position of the energy bands of Ag 2 O and TiO 2 , as the conduction band of Ag 2 O is more negative than that of TiO 2 .This electron transfer hinders the e − -h + recombination in the Ag 2 O.The photo-produced h + reacts with H 2 O generating OH • and H + .
Furthermore, considering the function of the Ag 0 /Ag I species as a co-catalyst, mixed systems composed of a metallic core and an oxide shell have often been found beneficial for different reactions.For example, a Ni/NiO core shell system, as well as Rh/Cr 2 O 3 or CuO x /Cr 2 O 3 systems (in which the core-shell structure is debated) have been found beneficial for water splitting, particularly for hydrogen evolution [67][68][69][70].While the metal ensures a conducting contact with the semiconductor and may act as the catalytic active site for reduction [70], the oxide (shell) may have other functions, such as prevention of the back reaction or functioning for water oxidation [67,71].Although the structural arrangement of the Ag 0 and Ag I (core-shell, neighboring, far from each other, etc.) is not known in the present case, it is possible that both species adopt important roles in the mechanism.We can only speculate that electron conduction and electron pooling may proceed on Ag 0 , whereas adsorption of CO 2 or stabilization of some relevant intermediates may take place on Ag I .This needs to be clarified in future studies.
Since a general mechanism for photocatalytic CO 2 reduction is not yet clear, an a priori design of a suitable catalyst is not an easy task.However, TiO 2 -based materials with co-catalysts (mixed states, metal loaded, z-schemes, etc.) have the potential to exhibit good efficiencies in producing useful chemicals from CO 2 .The optimum synthesis parameters should be optimized both in terms of efficiency, but also in terms of cost, environmental burden, and sustainability.The necessity for high purity experimental conditions is eminent in combination with (at least) blank experiments in order to reach safer conclusions about novel materials.

General Considerations
Powder XRD data were obtained on a Miniflex 600 device (Rigaku, Tokyo, Japan) in the range of 2θ = 3-80 • (0.2 • step, 30 kV, 10 mA, Cu K α ).UV-Vis spectra were recorded on a UV-2450PC PharmaSpec device (Shimadzu, Kyoto, Japan) for dry-pressed sample disks and on a Lambda 365 device (PerkinElmer, Waltham, MA, USA) in the diffuse reflectance mode for the as-synthesized powders, respectively.XPS data were recorded on a K-Alpha device (ThermoFisher Scientific, Waltham, MA, USA) using a 72 W monochromated Al K α source.All XP spectra were referenced according to the C 1 s peak at 285 eV.Simulations of the experimental photopeaks were carried out using a mixed Gaussian/Lorentzian peak fit procedure according to the CasaXPS software.Semi-quantitative analysis accounted for a nonlinear Shirley background subtraction.SEM measurements were done using a Merlin VP compact device (Zeiss, Oberkochen, Germany) and additional EDX investigations were performed using a Quantax 400 device (Bruker, Billerica, MA, USA).TEM images were recorded either on a JEM-1400 device (Jeol, Akishima, Japan), or on a JEM-ARM200F device (Jeol, Akishima, Japan).Gas chromatograph (GC) measurements during photocatalytic CO 2 reduction were collected using a Tracera GC-2010 Plus device (Shimadzu, Berlin, Germany) equipped with a barrier ionization discharge detector (BID) and a flame ionization detector (FID).

Preparation of Photocatalysts
The Ag decorated TiO 2 photocatalysts were prepared using a liquid impregnation method, which is exemplified in the following for the 1 mol% Ag/TiO 2 sample.TiO 2 powder (7.99 g) and 0.17 g of AgNO 3 precursor were added in 100 mL of distilled water.The theoretical loading amount of Ag was 1 mol% (the molar ratio of Ag to TiO 2 ).The slurry was kept under stirring and heated at 80 • C for 4 h to achieve Ag + equilibrium on the surface of TiO 2 .The stirring process was equipped with reflux apparatus.Then, the slurry was allowed to settle at room temperature overnight.The formed slurry was dried in an oven for 12 h at a temperature of 100 • C. The resulting solid was ground with a mortar and subsequently calcined at 500 • C for 3 h in air.For the preparation of the samples with different Ag loadings, the amount of AgNO 3 was varied accordingly, while all other reagent amounts were kept similar.The resulting Ag modified TiO 2 photocatalysts are referred to as Ag/TiO 2 X% (X = 1-4) [30].The Ag 0 /TiO 2 2% composite was prepared by reducing the Ag/TiO 2 2% sample using 20% H 2 (in Ar) for 1 h at 200 • C.

Photocatalytic Setup
The photocatalytic CO 2 reduction experiments were carried out in a high-purity gas-phase photoreactor setup.A detailed description of the photocatalytic reactor design was given by Mei et al. [6].In brief, the photoreactor is made exclusively of stainless steel components suitable for ultra-high vacuum applications.All of the tube connections were realized using vacuum coupling radiation (VCR) fittings, thus ensuring a grease-free sealing of all interconnecting parts.The sample powder was spread as evenly as possible on a quartz sample holder (diameter ~2.5 cm) placed in the center of the photoreactor.High purity He (6.0) and 1.5% CO 2 /He (5.5/6.0)mixture were used in the experiments and their respective flows towards the reactor were adjusted using a set of two mass flow controllers (MFC).Water enrichment of the gas mixtures was performed in a temperature-controlled stainless-steel saturator.All water-carrying pipes were constantly heated to 120 • C to avoid internal H 2 O condensation.A 200 W Hg/Xe lamp (Newport Oriel) emitting UV and visible light was used as the irradiation source, as well as a 10 W array of 4 high-power LEDs emitting at 365 nm (SeoulVioSys).A water-based filter was used for IR radiation removal from the lamp to avoid sample heating.The product analysis was performed using a Shimadzu Tracera GC 2010 plus gas chromatograph (GC) equipped with a barrier ionization discharge detector (BID-Q Bond 50 m column 5A molecular sieve) and a flame ionization detector (FID-Alumina Bond/KCl 50 m column).Helium (He) at a flow of 25 mL min −1 was used as the carrier gas in the GC.The temperature of the columns was increased from initially 30 • C up to finally 160 • C (heating rate: 20 • C min −1 ).Then the temperature was decreased back to 30 • C (cooling rate: 50 • C min −1 ) and immediately afterwards increased again to 160 • C (heating rate: 20 • C min −1 ).In total, 33 min were needed for a complete measurement with a 5-min cool-down period.This procedure facilitated desorption of molecules from the column.

Sample Pretreatment
Before studying the performance of the samples in photocatalytic CO 2 reduction, extensive cleaning was performed to clean the materials' surface from residual carbon-containing species.These impurities can react with CO 2 and H 2 O molecules to form hydrocarbons, thus leading to overestimation of the efficiency of the materials towards CO 2 reduction.The cleaning process applied involved two steps: In the first step, the samples were re-calcined at 400 • C for 3 h at a rate of 180 K h −1 under a synthetic air environment to remove the majority of the adsorbed C-containing impurities.In the second step, the cleaning process continued with the sample (50 mg) inside the photoreactor in batch mode.The reactor was filled with H 2 O (T H2O = 5 • C)-saturated pure He (at a flow of 10 mL min −1 ) up to a final pressure of 1500 mbar.A gas sample was collected every 45 min from the batch reactor, for a total duration of 6 h, with the first chromatogram taken without light irradiation.As with each collected sample, when the pressure inside the reactor dropped, a pressure-based correction was performed.Subsequently, a flow-through cleaning process took place, where moist pure He flowed continuously (at a rate of 20 mL min −1 ) over the sample under light irradiation for 3 h.Under these circumstances, detection of products from sample cleaning was not accurately possible, so the flow cleaning was followed by a second batch cleaning process.The amount of CH 4 produced from impurities in this second batch cleaning were compared to CH 4 formation in the photocatalytic CO 2 reduction experiment in order to evaluate the relative number of products truly originating from CO 2 .

Photocatalytic CO 2 Reduction
The photocatalytic CO 2 reduction experiment was carried out in batch mode conditions.The experimental conditions and GC analysis were similar to the batch cleaning process, except for the presence of CO 2 as a reactant gas.Thus, the gas-phase feed of the reactor consisted of 1.5% CO 2 diluted in He and saturated with H 2 O at 5 • C. The product analysis was also performed every 45 min with the necessary pressure drop corrections.An extensive set of blank measurements (no photocatalyst, no CO 2 , no light, and combinations of the above) were performed to ensure that no, or sufficiently low, concentration of hydrocarbons were produced at conditions other than a CO 2 reduction experiment.

Conclusions
In this study, Ag nanoparticles deposited on commercial anatase TiO 2 were successfully prepared by liquid impregnation method.As XPS analysis results indicated, the composite contained the mixture of Ag I and Ag 0 on the surface of TiO 2 .It was found that all silver/titania materials had visible light absorption due to surface plasmon absorption of electrons in silver species.The TEM and SEM images showed the uneven dispersion of Ag nanoparticles with a diameter of ~2-50 nm on the TiO 2 surface.Only the sample with 2 wt % Ag was revealed to be an effective material for CO 2 reduction.The LSPR effect was studied using two different irradiation sources containing only UV (365 nm LED) or both UV and visible (Hg/Xe lamp) wavelengths, and it was identified as the main reason for the photocatalytic activity of the Ag/TiO 2 2% sample.In addition, the presence of mixed-valent silver was beneficial for CH 4 formation in photocatalytic CO 2 reduction.A sample in which all silver has been reduced to the metallic state was instead less active.

Figure 1 .
Figure 1.Powder XRD patterns of pristine TiO2 and Ag/TiO2 materials.Small squares (□) at 2θ = 33.2°indicate the presence of Ag I species and the small stars (*) at 2θ = 44.4°indicate the presence of Ag 0 .

Figure 1 .
Figure 1.Powder XRD patterns of pristine TiO 2 and Ag/TiO 2 materials.Small squares ( ) at 2θ = 33.2• indicate the presence of Ag I species and the small stars (*) at 2θ = 44.4• indicate the presence of Ag 0 .

Figure 2 .
Figure 2. (a) UV-Visible spectra of TiO2 and Ag/TiO2 as pressed disks.Inset shows the material absorption magnification at 400-700 n; (b) surface plasmon resonance (SPR) effect of Ag species deposited on TiO2 surface, modified from reference [48].

Figure 2 .
Figure 2. (a) UV-Visible spectra of TiO 2 and Ag/TiO 2 as pressed disks .Inset shows the material absorption magnification at 400-700 n; (b) surface plasmon resonance (SPR) effect of Ag species deposited on TiO 2 surface, modified from reference [48].

Figure 2 .
Figure 2. (a) UV-Visible spectra of TiO2 and Ag/TiO2 as pressed disks.Inset shows the material absorption magnification at 400-700 n; (b) surface plasmon resonance (SPR) effect of Ag species deposited on TiO2 surface, modified from reference [48].

Figure 7 .
Figure 7. CH4 concentration over time for the Ag/TiO2 2% sample in batch cleaning, and CO2 reduction with Hg/Xe lamp and LED, as well as CO2 reduction with metallic state Ag 0 /TiO2 2%.Bare TiO2 is used as a reference.

Figure 7 .
Figure 7. CH 4 concentration over time for the Ag/TiO 2 2% sample in batch cleaning, and CO 2 reduction with Hg/Xe lamp and LED, as well as CO 2 reduction with metallic state Ag 0 /TiO 2 2%.Bare TiO 2 is used as a reference.