3.1.

Preparation of Mono- and Bimetal-Modified DAP

DAP was modified with NPs of Ag and Cu by the photodeposition method, and the data obtained for hydrogen evolution during photodeposition are shown in Fig. 1. Photoreduction of cations by photogenerated electrons on the surface of an irradiated semiconductor in the presence of a hole scavenger is one of the oldest and most often used methods for preparation of metal-modified semiconductors.^19,60–62 The fastest hydrogen evolution (the shortest induction period of ca. 7 min) was observed for Cu deposition on DAP. The induction period (intersection with the $x$-axis), during which NPs of noble metal are formed, differs depending on the noble metal and properties of titania. For example, photodeposition of Pt and Au is very fast with often undetectable induction periods (almost linear evolution of hydrogen from the beginning of irradiation).^{50,57,58,63,64} However, for less noble metals, a longer irradiation time is necessary to reduce respective cations and to form NPs, e.g., 8 min for Cu and 45 min for Ag (during deposition on OAPs).⁶⁵ Cu is known for its stronger than Ag activity for methanol dehydrogenation due to its larger work function [4.61 to 4.67 eV (Ref. 66) and 4.14 to 4.46 eV (Ref. 67), respectively].^50,65 The difference between the metal work function and the electron affinity of titania (3.41 to 4.5 eV)^68–71 is decisive for generation of a Schottky barrier (the electronic potential barrier formed at the metal–semiconductor heterojunction), i.e., the greater the difference between the electron affinity of titania and the work function of metal is, the higher is the Schottky barrier. Additionally, the higher the Schottky barrier is, the higher is the transfer and trapping of photogenerated electrons by metal,⁷² and thus the higher is the hydrogen generation rate. DAP/Cu/Ag and DAP/Ag-Cu bimetallic photocatalysts showed much lower activity than that of Cu-modified DAP. The poor activity of a codeposited sample (worse than expected: resultant reaction rate between that of two single-modified samples) is probably caused by possible formation of an AgCu alloy [Positive shift of XPS peak for Ag $3_{\mathrm{d}5/2}$ from 368.1 to 368.2 eV for DAP/Ag and DAP/Ag-Cu, respectively (as was proposed for Ag–Cu bimetallic nanoparticles.)⁷³]. An increase in the overvoltage toward the hydrogen evolution reaction was reported for an AgCu alloy electrode in comparison to electrodes composed of single metals (a silver electrode and a copper electrode), which was surprising since silver and copper have similar properties concerning the overvoltage.⁷⁴ Therefore, a probably larger overvoltage for hydrogen evolution could be the reason for the low level of photocatalytic activity of the DAP/Ag-Cu sample. The DAP/Cu/Ag sample exhibited the lowest activity among bimetallic samples. It is thought that sequential deposition should result in partial coverage of one metal by another since during this deposition, NPs of predeposited metal (here Cu) are negatively charged (as they are an electron reservoir). Similarly, in the case of silver photodeposition on gold-modified titania, the formation of core(Au)-shell(Ag) was reported.⁵⁸ Thus, partial deposition of Ag on the surface of Cu (DAP/Cu/Ag) inhibited proton adsorption on the surface of Cu NPs and, consequently, in hydrogen evolution. Interestingly, sequential deposition of Cu on DAP/Ag (DAP/Ag/Cu) resulted in the highest rate of hydrogen evolution among all samples ($0.53\mu \mathrm{mol}{\min }^{-1}$ versus 0.04, 0.08, 0.14, and $0.44\mu \mathrm{mol}{\min }^{-1}$ for DAP/Ag, DAP/Cu/Ag, DAP/Ag-Cu, and DAP/Cu, respectively). Although the induction period was slightly longer than that for single Cu deposition (DAP/Cu), since Ag (present on titania surface) disturbed in reduction of Cu(II), the linear evolution of hydrogen after 30 min of irradiation indicated an electronic interaction between two modifiers. To clarify the possible interactions detailed characterization of samples were performed.

Fig. 1

Hydrogen evolution during photodeposition of metals on DAP sample.

3.2.

Characterization of Ag- and Cu-Modified DAP

The color of samples was brown and violet during photodeposition of Ag and Cu, respectively, which confirmed that zero-valent metallic deposits were formed since LSPR of small spherical NPs of Ag and Cu appears at 410 to 430 (Refs. 75 and 76) and 560 nm,⁷⁷ respectively. In the case of sequential metal deposition, the samples took the color of the second metal, i.e., deposition of Cu on DAP/Ag resulted in DAP/Ag/Cu with violet color and deposition of Ag on DAP/Cu caused brown coloration of DAP/Cu/Ag. The color of sample during codeposition was dark gray. The color of dried samples (insets of Fig. 2) was different than that during photodeposition, which could be caused by surface oxidation of these less noble metals (In contrast, the color of Au- and Pt-modified titania did not change after contact with air.).^57,65 Similar observations on color change after sample drying were obtained for deposition of Ag and Cu on OAPs⁶⁵ and commercial titania samples.^42,48

Fig. 2

DRS spectra of modified DAP samples taken with (a) ${\mathrm{BaSO}}_4$ and (b) DAP as a reference. Inset: photograph of dried samples.

Photoabsorption properties (DRS spectra) are shown in Fig. 2. The intrinsic interband absorption of titania is observed at wavelengths shorter than 400 nm ($E_g\cong 3\mathrm{eV}$), as shown in Fig. 2(a). Slight photoabsorption observed for DAP at wavelengths longer than 400 nm could be caused either by impurity or a reduced form of titanium (Ti(III)). DAP possesses very high purity since only three compounds are used for its synthesis: ${\mathrm{TiCl}}_4$, ${\mathrm{O}}_2$, and Ar. Therefore, only chloride could be considered as its impurity. After synthesis, DAP was washed thoroughly with Milli-Q water and thus only a small content of chlorine ($<0.5\mathrm{wt}$. % of titania) was detected on the surface of DAP by XPS analysis. The possibility of formation of Ti(III) is also low since synthesis of DAP is performed in excess of oxygen in the system. On the other hand, the conditions of DAP synthesis, i.e., rapid heating and quenching of the reaction mixture (ca. 1-s residence time in the oven), which are necessary to avoid the formation of octahedron (thermodynamically favorable, OAPs) instead of decahedron, could result in formation of some lattice defects, e.g., reduced form of titanium [Ti(III)].

NPs of noble metals can absorb visible light as is clearly shown in Fig. 2(b). Two maxima in the LSPR peak could be observed for DAP/Ag, which indicates the high heterogeneity of silver deposits: fine nanosized NPs (LSPR at ca. 410 nm) and large particles (LSPR at ca. 590 nm).^49,78–80 The photoabsorption properties of Cu-modified samples are quite different than reported ones, where generally, due to rapid copper oxidation, no LSPR peak of Cu could be detected (no absorption at 400 to 600 nm), and a single broad peak at 600 to 1000 nm indicated the presence of CuO.^81,82 For a single DAP/Cu sample, the broad peak from 400 to 800 nm could be caused by: (1) the interfacial charge transfer (IFCT) from the valence band of titania to the ${\mathrm{Cu}}_x\mathrm{O}$ ($X:1,2$) clusters at 400 to 500 nm, (2) the interband absorption of ${\mathrm{Cu}}_2\mathrm{O}$ at 500 to 600 nm, (3) LSPR of zero-valent Cu at ca. 550 to 570 nm, and (4) $2E_g\to 2T_{2g}$ transitions of ${\mathrm{Cu}}^{2+}$ located in the distorted or perfect octahedral symmetry at 600 to 800 and 740 to 800 nm, respectively.^51,83,84 Considering the shape and width of the DRS peak of DAP/Cu with the maxima at ca. 700 nm, it is thought that copper could exist in all reported forms: Cu(0), Cu(I-II), Cu(I), and Cu(II). Interestingly, photoabsorption properties of bimetallic photocatalysts differed, and maxima at shorter wavelengths (ca. 510 nm) could indicate that Cu(I) is the predominant form in those samples.

Morphology of samples was investigated by microscopic observation, and exemplary STEM images for single-modified samples are shown in Fig. 3. It is clear that Cu forms small nanoclusters of ca. 2 nm, which are uniformly distributed on DAP. On the other hand, aggregation of silver results in the formation of polydispersed NPs (3 to 40 nm), which confirms the broad LSPR peak of silver (Fig. 2). Aggregation of Ag NPs results in the copresence of unmodified titania particles in the final product, as clearly shown in the right part of Fig. 3(a) (marked by a white, dashed rectangle), which should be crucial for the resultant photocatalytic activity.

Fig. 3

STEM images of (a) DAP/Ag (marked region shows unmodified titania particle) and (b) DAP/Cu.

To investigate morphology of bimetallic samples, STEM-EDS was applied, and exemplary images for DAP/Ag/Cu and DAP/Ag-Cu samples are shown in Figs. 4 and 5. In the DAP/Ag/Cu sample, Ag formed larger NPs than Cu, which was uniformly distributed on the surfaces of both titania and silver NPs. However, a few places with a high density of Cu were also observed, suggesting the formation of large Cu NPs, as shown in Fig. 4(b). It should be pointed out that larger NPs of Cu were only formed in the vicinity of Ag NPs. It is thought that the interaction between both metals during Cu deposition could result in the formation of aggregates of Cu since during photodeposition silver is negatively charged (working as an electron sink).

Fig. 4

(a) and (b) STEM images with respective EDS images of DAP/Ag/Cu taken at different view. Mapping colors: Ti, green; Ag, red; Cu, yellow; SE-mode, scanning mode of STEM.

Fig. 5

(a)–(d) STEM images with respective EDS images of DAP/Ag-Cu taken at different view. Mapping colors: Ti, green; Ag, red; Cu, yellow; SE-mode, scanning mode of STEM.

In contrast to DAP/Ag/Cu, no large Cu NPs were formed in the codeposited sample (DAP/Ag-Cu). This sample had uniformly distributed very fine (nanosized) Cu and Ag nanoclusters, as shown in Fig. 5. Interestingly, several large Ag deposits ($>100\mathrm{nm}$) were also observed in this sample, but they did not initiate the formation of large Cu NPs in their vicinity [Fig. 5(d)]. It is proposed that due to competition between the two metals during photodeposition, copper occupied the surface of titania first, forming either single nanoclusters or an alloy with silver. Similarly, copper hindered direct deposition of silver on titania. Therefore, large deposits of silver were also observed in this sample.

Surface composition of samples and oxidation states of elements were determined by XPS, and the data obtained are summarized in Table 1. Titanium, oxygen, carbon, silver, and copper were analyzed in detail, and a fraction of the oxidation states of elements from the deconvolution of XPS peaks is shown in Table 2. Exemplary XPS peaks for Ti $2{\mathrm{p}}_{3/2}$, O 1s, C 1s, Cu $2{\mathrm{p}}_{3/2}$, and Ag $3{\mathrm{d}}_{5/2}$ of DAP/Cu/Ag sample are shown in Fig. 6. The ratio of oxygen to titanium exceeded two, reaching ca. 7.6 and 3.4 to 7.2 for bare and modified DAP samples, respectively. Enrichment of the titania surface with oxygen has been often reported, e.g., a ratio of 4.6 for titania samples prepared by the microemulsion method,⁴¹ a ratio of 2.5 for titania prepared by laser ablation,⁸⁵ and a ratio of 2.2 to 5 for OAP prepared by HT (depending on duration of HT).⁶⁵ Significant excess of oxygen on the surface of DAP (7.6 ratio) is reasonable since anatase particles are formed under the continuous flow of oxygen. The deconvoluted oxygen peak indicates the presence of three forms of oxygen at ca. 529.4, 531.6, and 533.1 eV. The first peak is related to oxygen in the crystal lattice of ${\mathrm{TiO}}_2$, the second peak to $\mathrm{C}═\mathrm{O}$, ${\mathrm{Ti}}_2{\mathrm{O}}_3$ and OH groups bound with two titanium atoms, and the third one is related mainly to hydroxyl groups bound to titanium and carbon ($\mathrm{Ti}─\mathrm{OH}$, $\mathrm{C}─\mathrm{OH}$).^86,87 The content of lattice oxygen on the surface of DAP amounted to only ca. 20%, thus the surface was mainly composed of hydroxyl groups (or other compounds containing oxygen, e.g., carbon dioxide from air). Modification of the DAP surface with NPs of noble metals resulted in a decrease in the content of hydroxyl groups (except for codeposited sample: DAP/Ag-Cu), which is reasonable since surface modifiers displaced other adsorbed species. This competition of adsorption on the titania surface between hydroxyl groups and metal deposits can be crucial for photocatalytic performance, especially in the case of reactions performed in organic media. For example, for cyclohexane oxidation, deposition of gold NPs on the titania surface resulted in a decrease in the amount of available hydroxyl groups, which in consequence decreased the amount of generated hydroxyl radicals during irradiation and thus the efficiency of oxidation.⁸⁸ Slightly smaller content of lattice oxygen (${\mathrm{TiO}}_2$) in codeposited sample (DAP/Ag-Cu) indicates that some particular bimetallic nanostructure with surface oxygen enrichment was formed. Deconvolution of Cu and Ag peaks confirms that this sample differed significantly from others, having the largest Cu content, the smallest Ag content, the smallest content of ${\mathrm{Cu}}^{2+}$, the largest content of Cu(0), and the largest content of ${\mathrm{Ag}}^{2+}$. The most surprising aspect of the codeposited material is that it contains the largest fraction of Cu and the smallest fraction of Ag of any of the bimetallic composites: it is expected that the materials prepared by sequential deposition, DAP/Cu/Ag, and DAP/Ag/Cu, should have larger Ag and Cu content, respectively. Interestingly, the DAP/Ag-Cu sample has an even larger content of Cu on the surface (14 wt. %) than single-modified DAP with Cu (DAP/Cu, 12.9 wt. %). It is thought that during codeposition, segregation of two metals could result in the formation of a silver core and Cu discontinuous shell (fine Cu clusters deposited on the surface of silver). Similar metal segregation for Ag-Cu composites has already been reported, e.g., during radiolytic reduction–deposition of Ag and Cu on titania P25.⁸¹ However, based on obtained STEM-EDS images, it is difficult to confirm/reject the possibility of formation of an Ag(core)-Cu(shell) due to the small sizes of metallic clusters (nanosized) uniformly dispersed on the support (Fig. 5). It is highly possible that the mixture of Ag-Cu alloyed nanoclusters, core–shell nanoclusters, and monometallic nanoclusters/NPs (nanoclusters of Cu and large deposits of Ag) could coexist in the codeposited sample.

Table 1

XPS analysis of oxygen, titanium, carbon, silver, and copper for various DAP samples, and fraction of oxidation states of C from deconvolution of XPS peak of C 1s

Table 2

Fraction of oxidation states of Ti, O, Cu, and Ag from deconvolution of XPS peaks of Ti 2p3/2, O 1s, Cu 2p3/2, and Ag 3d5/2

Fig. 6

XPS results for O 1s, C 1s, Ti 2p3/2, Cu 2p3/2, and Ag 3d5/2 for DAP/Cu/Ag sample.

NPs of silver and copper have positively charged surfaces with $1+$ being the predominant form (${\mathrm{Ag}}^{+}$ and ${\mathrm{Cu}}^{+}$). Sequential deposition significantly influenced the form (charge) of predeposited metals, i.e., DAP/Ag/Cu has the highest content of zero-charged silver (27.3%) since at the beginning of Cu deposition, electrons from titania are sinking in predeposited silver, and subsequently deposited Cu stabilized Ag NPs. Similarly, a decrease in the content of $2+$ charged Cu (from 12.8% to 5.6%) and increase in the content of $1+$ charged Cu (from 82% to 93.7%) was observed in the case of sequential deposition of Ag on DAP/Cu.

Titanium in all samples existed mainly in ${\mathrm{Ti}}^{4+}$ form, e.g., 98% in DAP (details in Table 2). Deposition of metals slightly changed the surface composition of the sample, and the largest content of oxygen vacancies (${\mathrm{Ti}}^{3+}$) was observed for samples in which Cu was deposited first, e.g., ca. 3.9% in the DAP/Cu sample and ca. 3.3% in the DAP/Cu/Ag sample. On the other hand, deposition of silver decreased the content of ${\mathrm{Ti}}^{3+}$, suggesting preferential deposition of silver on lattice defects. Similar behavior was proposed for gold NPs, which were preferentially formed on lattice defects of titania.^28,89

3.3.

Photocatalytic Activity of Ag- and Cu-Modified DAP

DAP modification with silver and copper resulted in significant enhancement of UV/vis photocatalytic oxidation of acetic acid, as shown in Fig. 7. DAP/Ag exhibited the highest photocatalytic activity, indicating that an enhancement in activity did not correlate with the work function of metals. Therefore, not only formation of the Schottky barrier (inhibition of ${\mathrm{e}}^{-}/{\mathrm{h}}^{+}$ recombination), but also other factors were crucial for overall photocatalytic performance. Previously, it was shown that the content of surface hydroxyl groups influenced the formation of reactive oxygen species (ROS), and thus photocatalytic activity.^90,91 For example, Cu/OAPs possessing the highest content of hydroxyl groups on its surface also exhibited the highest photocatalytic activity among other OAPs samples modified with Ag, Pt, and Au.⁶⁵ Interestingly, in this study, neither the work function of metal nor the content of hydroxyl groups correlated with photocatalytic activity, i.e., DAP/Ag has the highest activity despite the lowest content of hydroxyl groups on the surface. In contrast to methanol dehydrogenation (anaerobic conditions), decomposition of acetic acid was carried out in the presence of oxygen. In this regard, the possible heterojunctions between oxides of noble metals (formed on the surface of metallic deposits) and titania could result in enhanced photocatalytic activity. Similar heterojunctions between titania and CuO, ${\mathrm{Cu}}_2\mathrm{O}$, ${\mathrm{Ag}}_2\mathrm{O}$, and AgO have already been proposed for inhibition of the recombination of charge carriers.^53,81,84,92 It is also possible that heterojunction $\mathrm{CuO}/{\mathrm{Cu}}_2\mathrm{O}\text{-}{\mathrm{TiO}}_2$ is less active than metal-modified titania $\mathrm{Ag}/{\mathrm{TiO}}_2$ or $\mathrm{Ag}\text{-}\mathrm{AgO}/{\mathrm{TiO}}_2$. (Copper exists mainly in oxidized form, but silver coexists in metallic and oxidized form.) To clarify this matter, a series of photocatalysts possessing the same composition and being different only by the oxidation state of metals will be investigated in future research.

Fig. 7

Oxidative decomposition of acetic acid on bare and modified DAP with noble metals under UV/vis irradiation.

The two most active samples (DAP/Ag and DAP/Ag/Cu) differed significantly in their properties, i.e., DAP/Ag had the highest content of ${\mathrm{Ag}}^{+}$ (89.5%) among all samples, whereas DAP/Ag/Cu had the lowest content of ${\mathrm{Ag}}^{+}$ (67.7) and the highest content of zero-valent Ag (27.3%). Therefore, different mechanisms for both samples should be proposed, i.e., interparticle charge transfer between: (1) ${\mathrm{TiO}}_2$ and ${\mathrm{Ag}}_2\mathrm{O}$ for DAP/Ag, and (2) ${\mathrm{TiO}}_2\text{-}\mathrm{AgO}/\mathrm{Ag}\text{-}\mathrm{CuO}/{\mathrm{Cu}}_2\mathrm{O}$ for DAP/Ag/Cu. The correlation between photocatalytic activities of bimetallic samples under anaerobic conditions (methanol dehydrogenation, Fig. 1) and aerobic conditions (Fig. 7) suggests that the morphology of bimetallic deposits is crucial for photocatalytic performance. Therefore, it is proposed that the most active sample (DAP/Ag/Cu) has Cu nanoclusters uniformly distributed on the surface of both titania and Ag NPs (Fig. 4). However, participation of larger NPs of Cu, deposited in the vicinity of Ag NPs, in the overall photocatalytic performance could not be neglected and will be investigated in detail in our future study.

Photocatalytic activity under visible light irradiation was first tested for irradiation at wavelengths longer than 420 nm to ensure inactivity of bare titania and to keep overall plasmonic properties of silver. Photocatalytic activity of modified samples [Fig. 8(a)] generally correlated with their photoabsorption properties (Fig. 2). DAP/Ag sample had the strongest absorption of visible light and its photocatalytic activity was the highest. However, observed photocatalytic activity of bare DAP under irradiation with a visible light was surprising and made the discussion on the photocatalytic performance of other modified samples, which showed lower photocatalytic activity than that of DAP, difficult suggesting that those modifiers hindered the intrinsic activity of bare DAP (This vis-response could be caused by the presence of adsorbed ${\mathrm{Cl}}^{-}$ or ${\mathrm{Ti}}^{3+}$, since small conductivity for bare DAP samples was observed even up to 545 nm, and this phenomenon will be investigated in detail in our future study.). Therefore, activity tests were repeated for a narrower range of irradiation by using another cut-off filter (Y48) to ensure the transmission of light with wavelengths longer than 450 nm [Fig. 8(b)]. Indeed, application of longer wavelengths of irradiation eliminated the visible response of bare DAP. The photocatalytic activity of bimetallic samples (DAP/Ag/Cu, DAP/Cu/Ag, DAP/Ag-Cu) corresponds with the photoabsorption properties presented in Fig. 2. DAP/Ag showed the highest visible activity among tested samples. Interestingly, DAP/Cu showed slightly higher activity than bimetallic samples. Similar behavior was observed for other plasmonic photocatalysts (Au-Ag, Au-Cu, Ag-CuO) under visible light irradiation, i.e., decrease in activity after addition of the second modifier.^58,81,93,94 For example, in the case of bimetallic photocatalysts containing silver and gold, single-modified samples showed higher photocatalytic activity (either ${\mathrm{TiO}}_2/\mathrm{Au}$ or ${\mathrm{TiO}}_2/\mathrm{Ag}$ depending on titania support properties) than bimetallic photocatalysts in which two metals formed a bimetallic nanostructure, e.g., core–shell.⁵⁸ Only photocatalysts with NPs of both metals separately deposited on titania (Ag NPs and Au NPs) showed higher photocatalytic activity due to the ability of absorption of more photons by LSPR of two metals (different positions of LSPR). It was proposed that the deposition of two metals close to each other (bimetallic nanostructure) resulted in electron transfer from one metallic deposit to another one (recombination of charge carriers), instead of being transferred to the conduction band of titania. Similarly, modified titania P25 with Ag and CuO showed the best photocatalytic performance for single-modified titania with CuO, then with Ag, and finally with bimetallic NPs.⁸¹ Although, in this study, opposite performance of single-modified samples was noticed, i.e., higher activity of ${\mathrm{TiO}}_2/\mathrm{Ag}$ than that of ${\mathrm{TiO}}_2/\mathrm{Cu}(\mathrm{Cu}/{\mathrm{Cu}}_2\mathrm{O}/\mathrm{CuO})$, all bimetallic photocatalysts exhibited worse photocatalytic activities than monometallic ones, which could indirectly support the mechanism of action of plasmonic photocatalysts through the charge transfer mechanism rather than energy transfer one. The opposite performance of monometallic photocatalysts (activity: ${\mathrm{TiO}}_2/\mathrm{Ag}>{\mathrm{TiO}}_2/\mathrm{Cu}$) than the reported one (activity: ${\mathrm{TiO}}_2/\mathrm{Cu}>{\mathrm{TiO}}_2/\mathrm{Ag}$), could result from the properties of copper where Cu was mainly in the form of CuO. Therefore, the heterojunction between two semiconductors, CuO and ${\mathrm{TiO}}_2$, could result in enhanced photocatalytic performance. (It should be pointed that P25 consists of anatase and rutile, which influences the properties and photocatalytic activity of resultant photocatalysts by either preferential deposition of metal NPs at the anatase/rutile junction or by charge transfer between two crystalline components and/or modifiers.) In this study, Cu is mainly in the form of ${\mathrm{Cu}}_2\mathrm{O}$. Therefore, it is proposed that for activation of titania under visible light Cu in the form of CuO is more recommended.

Fig. 8

Photocatalytic activity of bare and modified DAP under visible light irradiation: (a) $\lambda >420\mathrm{nm}$ and (b) $\lambda >455\mathrm{nm}$.