An orthophosphate semiconductor with photooxidation properties under visible-light irradiation

An orthophosphate semiconductor with photooxidation properties under visible-light...

(Parte 1 de 3)

An orthophosphate semiconductor with photooxidation properties under visible-light irradiation

The search for active semiconductor photocatalysts that directly split water under visible-light irradiation remains one of the most challenging tasks for solar-energy utilization1–6. Over the past 30 years, the search for such materials has focused mainly on metal-ion substitution as in In1−xNixTaO4 and (V-,Fe- or Mn-)TiO2 (refs 7,8), non-metal-ion substitution as in TiO2−xNx and Sm2Ti2O5S2 (refs 9,10) or solid-solution fabrication as in (Ga1−xZnx)(N1−xOx) and ZnS–CuInS2–AgInS2

(refs 1,12). Here we report a new use of Ag3PO4 semiconductor, which can harness visible light to oxidize water as well as decompose organic contaminants in aqueous solution. This suggests its potential as a photofunctional material for both water splitting and waste-water cleaning. More generally, it suggests the incorporation of p block elements and alkali or alkaline earth ions into a simple oxide of narrow bandgap as a strategytodesignnewphotoelectrodesorphotocatalysts.

It is known that the photolysis of water using semiconductors involves photogenerated electrons and holes migrating to the surface of the semiconductor and serving as redox sources that then react with adsorbed water, leading to the splitting of water. Both light absorption and suitable redox potentials are prerequisites for the direct splitting of water. As semiconductors with suitable redox potentials are rare and redox power is weakened by bandgap narrowing, a widely used approach is to treat the water splitting reaction in terms of two coupled half-reactions5,6,10,12–14. This emphasizes the significance of developing visible-light-sensitive photofunctional materials with the alternative ability to evolve H2 or O2 from aqueous solutions containing sacrificial reagents. They can later be combined to form a complete water splitting system.

Different compounds have been explored for this purpose. Water oxidation is found to be more challenging because the formation of one molecular oxygen involves the transportation and reaction of four electrons or holes5,15.

The silver orthophosphate used in this study was prepared by a simple ion-exchange method in the case of the powder samples and by an in situ electrochemical deposition method in the case of the thin-film samples. The crystal structure of the material was first investigated in 1925 when Wyckoff established it as having a cubic structure type16. Later, as a result of progress in the X-ray diffraction (XRD) technique as well as the interest in the ionic conductivity of Ag3PO4 (refs 17,18), more accurate

1Photocatalytic Materials Center, and International Center for Materials Nanoarchitectonics (MANA), and Innovative Center of Nanomaterials Science for Environment and Energy (ICNSEE), National Institute for Materials Science (NIMS), 1-2-1 Sengen, Tsukuba, Ibaraki 305-0047, Japan, 2Research School of Chemistry, Australian National University (ANU), Canberra ACT 0200, Australia, 3Research School of Biology, Australian National University (ANU), Canberra ACT 2601, Australia, 4Ecomaterials and Renewable Energy Research Center (ERERC), Department of Physics, Nanjing University, Nanjing, 210093, China. *

crystal structures involving the determination of the silver and phosphorus fractional coordinates were established. Ag3PO4 is of body-centred cubic structure type with a lattice parameter of

∼6.004 Å (ref. 17). The structure consists of isolated, regular PO4 tetrahedra (P–O distance of ∼1.539Å) forming a body-centred cubic lattice. Six Ag+ ions are distributed among twelve sites of two-fold symmetry. XRD patterns of our samples (see Fig. 1a) confirmed this crystal structure. The particle size of the crystals was estimated to be ∼0.5–2µm from scanning electron microscopy (SEM) images such as that shown in Fig. 1b. The ultraviolet– visible diffuse reflectance spectrum (Fig. 1c) indicated that the golden coloured Ag3PO4 can absorb solar energy with a wavelength shorter than ∼530nm. Further analysis of the absorption spectrum revealed an indirect bandgap of 2.36eV as well as a direct transition of 2.43eV (see Supplementary Fig. S1). The bandgap satisfies the energy criterion thermodynamically for the uphill reactions involved in water splitting.

The electrode potential of Ag/Ag3PO4 is between the reduction potential of H+ and Ag/AgNO3 (Fig. 2a; ref. 19), which means

Ag3PO4 cannot split water to release H2. It is, however, a potential photofunctional material possessing strong photooxidative capabilities in the presence of the sacrificial reagent silver nitrate.

O2 evolution over Ag3PO4 in an aqueous silver nitrate solution under visible-light irradiation was therefore tested. The typical amount of evolved O2 as a function of time is shown in Fig. 2b. For comparison purposes, the performance of two recognized light-sensitive semiconductors for O2 evolution, BiVO4 (prepared according to ref. 20) and WO3 (Commercial, Wako; ref. 21), under the same experimental conditions, are also shown. It was found that Ag3PO4 is the highest performing, with vigorous bubbles of O2 being observed as soon as light irradiation commenced (Supplementary Movie S1).

The wavelength dependence of the O2 evolution was then further investigated to prove whether or not the reaction really was driven by light11,12. Figure 2c shows the ultraviolet–visible diffuse reflectance spectrum of the Ag3PO4 along with the apparent quantum yield of O2 evolution as a function of the incident light wavelength. The apparent quantum yield decreased with increasing wavelength and the longest wavelength suitable for

O2 evolution was found to coincide with the absorption edge of the semiconductor Ag3PO4. This indicates that the oxygen



Absorbance (a.u.) Wavelength (nm) a c b

Ag P O

NONESEI15.0 kVX12,0001 µmWD 10.0 m θ

Figure 1 | Crystal structure, particle morphology and optical property of Ag3PO4. a, XRD patterns of the Ag3PO4 powders. Inset: Schematic drawing of the crystal structure. b, SEM image of the prepared Ag3PO4 powders. c, Ultraviolet–visible diffusive reflectance spectrum of the Ag3PO4 samples.

evolution reaction is indeed driven by light and that the light absorption property of the semiconductor governs the reaction rate. The extremely high quantum yield at wavelengths less than ∼480nm, significantly higher than previously reported values7,10–12,20, shows that the recombination of photoexcited electrons and holes within the material is very weak.

Inconsiderationoftheknowledgethatsilversaltsarewellknown todecomposeonlightirradiation,supplementaryexperimentssuch aspurewatersplittingand18O-isotope-labelledwateroxidationand so on (see Supplementary Experimental S1, Fig. S2 and Table S1), were also carried out and the results confirm that the oxygen observediswithoutdoubtderivedfromtheoxidationofwater.

Organic dye decomposition9,2,23 experiments using silver nitrate as an electron acceptor were also carried out to further certify the strong photooxidative ability of the semiconductor. Figure 2d(I) shows the variation of methylene blue concentration with il- lumination time over Ag3PO4(Brunauer–Emmett–Teller (BET), 0.98m2 g−1). For comparison purposes, the results of methylene commercial TiO2−xNx (BET,43m2 g−1) and methylene blue photolysis under the same conditions (see the Methods section and

SupplementaryFig.S3)arealsoshown.Itwasfoundthattheprocess of methylene blue decomposition over Ag3PO4 was dozens of times quicker than that over the reference materials. It is noted here that only ∼14% of the methylene blue was adsorbed on the Ag3PO4 powders when equilibrium adsorption states were reached during thedark reactionand thepHvalue ofthesolution waskeptconstant (the pH of deionized water) during the reaction. This shows once again that Ag3PO4 has very high photooxidation activity indeed. The inset in Fig. 2d(I) shows the colour changes of the methylene blue solutions during the photooxidation process.

In view of the fact that organic dyes can also harvest visible light, experiments with methylene blue photodegradation over

Ag3PO4 powders under different monochromatic visible-light irradiation conditions were designed to test the possibly evolved effects of photosensitization (Fig. 2d(I)). From ultraviolet–visible absorption spectra (see Supplementary Fig. S4) it is clear that the methylene blue dyes are most excited by a wavelength of approximately 600–700nm and least excited by a wavelength of

350–500nm, whereas the Ag3PO4 semiconductors can be excited only by a wavelength shorter than ∼530nm, as described above.

Therefore, we first used monochromatic visible light centred at a wavelength of 420.4nm (1λ = ±14.9nm) to excite the Ag3PO4 semiconductor but minimize the excitation of the methylene blue.

The methylene blue was still exhausted very quickly. In contrast, when monochromatic visible light centred at a wavelength of 639.3nm (1λ = ±16.2nm) was used to excite the methylene blue molecules but not the Ag3PO4 semiconductor, it is difficult to observe any methylene blue degradation at all. The effect of photosensitization during methylene blue decomposition over

Ag3PO4 is thus negligible. In other words, methylene blue decomposition can definitely be attributed to the intrinsic strong photooxidative activity of the Ag3PO4. It is worth mentioning that even without sacrificial reagent the Ag3PO4 still shows strong photooxidative ability (see Supplementary Figs S2a and S5). The results of total organic car- bon (TOC) measurements as well as decomposition experiments using other organic compounds (see Supplementary Fig. S6) in- dicate that the strong photooxidative ability of Ag3PO4 is also effective in these cases.

Unfortunately, as the electrode potential of Ag/Ag3PO4 is lower than that of the hydrogen electrode the Ag3PO4 semiconductor



Light on

C BTime (min)


Absorbance (a.u.) Wavelength (nm)

Apparent quantum yield (%)

Time (h) IAmount of evolved O µ mol)

Time (min)

V (versus NHE) (pH = 0)

Ag/Ag3PO4 Ag/AgNO3


Light c d

(I) WO3:72µmolh−1. c, Ultraviolet–visible diffuse reflectance spectrum and apparent quantum yields of the Ag3PO4 semiconductor plotted as a function of wavelength of the incident light. Apparent quantum yields were plotted at the centre wavelengths of the band-pass filters, with error bars showing the deviation of the wavelengths (1λ=±15nm). d, Variation of methylene blue concentration as a function of illumination time (with the time of light on set as 0) under visible light (λ>400nm) (I) with powder samples Ag3PO4 (A), TiO2−xNx (B), BiVO4 (C) and methylene blue photolysis (D), and under various monochromatic visible lights with Ag3PO4 (I) at wavelengths of λ=420.4nm (1λ=±14.9nm) (E) and λ=639.3nm(1λ=±16.2nm)(F). The inset shows the colour changes of the methylene blue solutions corresponding to the four filled square points in A.

would decompose during the water oxidization if no sacrificial reagent was involved (see Supplementary Fig. S7). Silver phosphate decomposition can, however, be halted or delayed. Inspired by the facts that: (1) the aforementioned redox scheme can be viewed as a short-circuited photoelectrochemical cell24; and (2) water oxidization on the Ag3PO4 semiconductor occurs automatically once it is illuminated (λ < 530nm), we designed a macroscopic photoelectrochemical cell (see Supplementary Fig. S8) in which

Ag3PO4 thin films deposited onto an inert conducting substrate are used both as a water oxidization reactor and as the macroscopic electrode when halting/delaying the process of photocorrosion or restoring the material is needed. The latter function is warranted by adding a solar cell as an alternative to the sacrificial reagent. The difference in the electrode potential between Ag/Ag+ and O2/H2O justifies the cell configuration.

Figure 3a shows an SEM image of the in-situ-deposited Ag3PO4 thin film. Cyclic voltammetry was used to investigate the effect of light on the current density (Fig. 3b), in terms of the photocurrent and dark-current response from the sample. The positive photocurrent at anodic potentials suggests that the Ag3PO4 electrode is an n-type semiconductor. The onset potential of this semiconductor electrode is 0.9–1.0VNHE. The dark current is very weak up to about 1.3VNHE, where electrocatalytic oxygen evolution starts. The proven contribution of light irradiation to the current densitybetween1.0and1.3VNHE indicatesthattheAg3PO4 thinfilm can be used positively as a photoanode. As Ag3PO4 oxidizes water automatically while simultaneously itself being reduced under light

3O2), a change in irradiation conditions also changes the state of the photoelectrode. That is why the onset potential shows slight variation in Fig. 3b. Moreover, the bias voltage here can be set either to transfer the photoexcited electrons to the external circuit or to re-oxidize the reduced Ag. The incident photo-to-current efficiency here (see Supplementary Fig. S9) is therefore an indirect measure of the ability to delay the photocorrosion, instead of a measure of oxygen evolution activity.

As well-crystallized Ag3PO4 semiconductor can form at room temperature, in an aqueous solution containing silver cations and phosphate anions, the sample preparation method contains an inbuilt rejuvenation mechanism; that is, when the fabricated

(Parte 1 de 3)