Supporting Information for Facet Effect of Single-Crystalline Ag 3 PO 4 Sub-microcrystals on Photocatalytic Properties Yingpu Bi, Shuxin Ouyang, Naoto Umezawa, Junyu Cao, and Jinhua Ye* International Center for Materials Nanoarchitectonics (MANA), and Photocatalytic Materials Center, National Institute for Materials Science (NIMS), 1-2-1 Sengen, Tsukuba, Ibaraki 305-0047, Japan *To whom correspondence should be addressed. Email: jinhua.ye@nims.go.jp Experimental Section 1. Synthesis of rhombic dodecahedral and cubic Ag 3 PO 4 sub-microcrystals Scheme S1. Schematic illustration of the growth process of the Ag 3 PO 4 crystals with different morphologies: (A) rhombic dodecahedral Ag 3 PO 4 crystals, (B) cubic Ag 3 PO 4 crystals. The rhombic dodecahedral Ag 3 PO 4 crystals were prepared by a simple precipitation process (shown in Scheme S1A). In a typical synthesis, CH 3 COOAg (0.2 g) was solved in aqueous solution. Na 2 HPO 4 aqueous solution (0.15 M) was added with drop by drop to the above solution, and golden yellow precipitation would be formed. The obtained samples for morphology and structure analysis were washed with water to remove the CH 3 COO - and dried under atmosphere. The Ag 3 PO 4 cubes were prepared by using silver-ammino complex as the silver ions source (shown in Scheme S1B). In a typical synthesis, AgNO 3 (0.2 g) was solved in aqueous solution. Ammonia aqueous solution (0.1 M) was added with drop by drop to the above solution to form a transparent solution. Then, Na 2 HPO 4 aqueous solution (0.15 M) was added, and olivine Ag 3 PO 4 crystals with cubic structure have been synthesized. 2. Photocatalytic Reactions In all catalytic activity of experiments, the samples (0.2 g) were put into a solution of MO or RhB dyes (100 ml, 8 mg/l), which was then irradiated with a 300W Xe arc lamp equipped with an ultraviolet cutoff filter to provide visible light with λ 420 nm. The degradation of organic dyes was monitored by UV/Vis spectroscopy (UV-2500PC, Shimadzu). Before the spectroscopy measurement, these photocatalysts were removed from the photocatalytic reaction systems by a dialyzer.
3. Characterizations SEM images were taken using a field-emission scanning electron microscope (JSM-6701F, JEOL) operated at an accelerating voltage of 5 kv. An energy-dispersive (ED) detector was equipped with this field-emission scanning electron microscope and operated at an accelerating voltage of 15 kv. The X-ray diffraction spectra (XRD) measurements were performed on a Rigaku RINT-2000 instrument using Cu Kα radiation (40 kv). The XRD patterns were recorded from 10 to 90 with a scanning rate of 0.067 / s. UV/Vis absorption spectra were taken at room temperature on a UV-2550 (Shimadzu) spectrometer. The infrared spectra were obtained on a Fourier Transform Infrared (FTIR) Spectrometer (Shimadzu IRprestige-21). The surface area of the sample was measured by the BET method (Shimadzu Gemini 2360, Micromeritics). 4. Calculations Computational details. Our electronic structure calculations were based on the density-functional theory (DFT) +U approach [1]. The exchange-correlation energy functional was represented by the local-density approximation (LDA) [2]. Projector-augmented wave pseudopotentials were employed as implemented in the VASP code [3,4]. The valence configurations of the pseudo potentials are 4d 10 5s 1 for Ag, 2s 2 2p 4 for O, and 3s 2 3p 3 for P. We used an energy cutoff of 500 ev in the plane-wave basis set expansion. Monkhorst-Pack k-point sets of 4 4 4 was used for a 16-atom unit cell of cubic Ag 3 PO 4 (space group P43n). The on-site Coulomb repulsion (Hubbard U) was applied both for Ag d and O p states. First, we estimated U for a single Ag + ion by subtracting atomic total energies E tot of three different occupations of d states [5]: U ion (Ag + d) = (E tot (d 10 ) - E tot (d 9 )) (E tot (d 9 ) - E tot (d 8 )) = 13.92 ev where the self-interaction correction was taken into account with the Perdew-Zunger type LDA correlation energy functional [2]. Similarly, U ion for a single O 2- ion was given by U ion (O 2- p) = (E tot (p 6 ) - E tot (p 5 )) (E tot (p 5 ) - E tot (p 4 )) = 9.5 ev. We then multiply a common parameter α with U ion both for Ag and O to describe the screening effect in bulk Ag 3 PO 4. The parameter α was determined so as to reproduce an experimental band gap of Ag 3 PO 4 (2.45 ev [6]), which results in U bulk (Ag d)= 14.62 ev and U bulk (O p)= 9.98 ev. The band gap correction is important in our calculations because the computational band from LDA (0.1 ev) significantly underestimes the experimental value. The cell volume and atomic positions of the unit cell were relaxed until the residual forces were below 0.02 ev/å. The obtained lattice constant from LDA+U was a= 5.74 Å in comparison with an experimental value (6.005 Å) [7]. Model construction. The relaxed unit cell was extended to construct the surface models shown in Fig.3A and 3B. In the case of (100) surface, a 48-atom slab model was created by extending the 16-atom unit cell alogn z-direction by three times. Two oxygen atoms on one side are moved to the other side of the slab to satisfy the stoichiometry on each side of the slab. It was then expanded by 2 2 in the xy plane to construct the 192-atom supercell including a vacuum region with the thickness (L vac ) same as the Ag 3 PO 4 layer shown in Fig. 3B. Thus, the lattice vectors of the supercell can be described as A 1 = 2a 1, A 2 = 2a 2, and A 3 = (3 + L vac /a)a 3 where a 1 = ax, a 2 = ay, and a 3 = az are the lattice vectors of the 16-atom unit cell. For modeling (110) surface, we have created a slab with the lattice vectors described as A 1 = a 1 +a 2, A 2 = 2a 3, and A 3 = (3+ L vac / a 1 +a 2 )(a 1 +a 2 ). Four atoms each of Ag and O on one side of the slab were moved to the other side for the stoichiometry reason. These surface structures were fully relaxed until the total energy difference was converged within 0.001 ev where the mid layer (the second layer) was fixed to represent a bulk region and k-point was sampled at Γ. References [1]. Anisimov, V. I., Zaanen, J. and Andersen, O. K. Phys. Rev. B 1991, 44, 943.
[2]. Perdew, J. P. and Zunger, A. Phys. Rev. B 1981, 23, 5048. [3]. Kresse, G. and Hafner, J. Phys. Rev. B 1993, 47, 558. [4]. Kresse, G. and Furthmüller, J. Phys. Rev. B 1996, 54, 11169. [5]. Janotti, A.; Segev, D.; and Van de Walle, C. G. Phys. Rev. B 2006, 74, 045202. [6]. Yi, Z.; Ye, J.; Kikugawa, N.; Kako, T.; Ouyang, S.; Williams, H. S.; Yang, H.; Cao, J.; Luo, W.; Li, Z.; Liu, Y. and Withers, R. L. Nature Mater. 2010, 9, 559. [7]. Wyckoff, R.W.G., Crystal structure of silver phosphate and silver arsenate (Ag3XO4) American Journal of Science, Serie 5(1,1921-1938) 1925, 10, 107.
Additional Figures and Discussions Figure S1. (A,B) SEM images of rhombic dodecahedral Ag 3 PO 4 sub-microcrystals with different magnifications. Figure S2. (A,B) SEM images of cubic Ag 3 PO 4 sub-microcrystals with different magnifications.
Figure S3. (A) Geometrical models of ideal rhombic dodecahedron, (B) geometrical models of cube, (C) crystal model of Ag 3 PO 4 rhombic dodecahedron (which is cleaved from a 2 2 2 supercell), (D) crystal model of Ag 3 PO 4 cubes. Figure S4. EDS patterns of Ag 3 PO 4 rhombic dodecahedrons (A) and cubes (B).
Figure S5. FTIR spectra of Ag 3 PO 4 products with different morphologies. (a) rhombic dodecahedrons, (b) cubes, (c) particles prepared by directly reacting AgNO 3 with Na 2 HPO 4. FTIR analysis of the as-prepared Ag 3 PO 4 crystals The infrared spectra of Ag 3 PO 4 rhombic dodecahedron and cubes has been investigated and shown in Figure S5. For comparison, the infrared spectra of spherical Ag 3 PO 4 particles prepared by directly reacting AgNO 3 with Na 2 HPO 4 has also been listed in this figure. It can be clearly seen that the infrared spectrums of rhombic dodecahedral and cubic Ag 3 PO 4 crystals are highly identical to that of spherical Ag 3 PO 4 particles although the absorption peaks intensities and positions have been slightly shifted. These observations clearly confirm that there are no CH 3 COO - or NH 4 - ions adsorbed on the surfaces of both Ag 3 PO 4 rhombic dodecahedron and cubes, and their catalytic active sites can be accessible for the photocatalytic reactions.
Figure S6. (A, B) SEM images of Ag 3 PO 4 products prepared by reacting CH 3 COOAg with NaH 2 PO 4, (C, D) SEM images of Ag 3 PO 4 products prepared by reacting CH 3 COOAg with H 3 PO 4. Results and discussion As shown in Figure 1A and Figure S1, Ag 3 PO 4 rhombic dodecahedrons with perfect and regular structures have been fabricated by directly mixing CH 3 COOAg with Na 2 HPO 4. Furthermore, the morphology changes of Ag 3 PO 4 products prepared with NaH 2 PO 4 and H 3 PO 4 have also investigated (note that when Na 3 PO 4 was used in this system, only a mixture of Ag 2 O and Ag 3 PO 4 has been obtained). Figure S6A and 6B show the typical SEM images of the as-prepared Ag 3 PO 4 products by replacing Na 2 HPO 4 with NaH 2 PO 4, indicating that this sample exhibit irregular rhombic dodecahedral morphology and non-uniform diameters. It is also clear from Figure S6B that all corners and edges of these products were slightly truncated. Furthermore, as shown in S6C and 6D, when H 3 PO 4 solution was used in this synthesis system, the obtained products are dominated by particles with irregular morphologies and no any rhombic dodecahedron has been observed. Based on the above result, it can be concluded that ph values of reaction system play an important role in determining the morphology and structure of Ag 3 PO 4 products.
Figure S7. (A, B) SEM images of Ag 3 PO 4 products prepared by reacting [Ag(NH 3 ) 2 ] + with NaH 2 PO 4, (C, D) SEM images of Ag 3 PO 4 products prepared by reacting [Ag(NH 3 ) 2 ] + with H 3 PO 4. Results and discussion During the preparation process of Ag 3 PO 4 cubes, the [Ag(NH 3 ) 2 ] + complex has been gradually decomposed by H + ions from Na 2 HPO 4 and release Ag + ions to form Ag 3 PO 4 crystals. As a result, it is important to know if the release rates of Ag + ions will have any influence on the crystal growth and formation process of Ag 3 PO 4 products. Thereby, the morphology changes of Ag 3 PO 4 products prepared with NaH 2 PO 4 and H 3 PO 4 have been studied. As shown in Figure S7A,B, when NaH 2 PO 4 was added into silver-ammino complex solution, cubic Ag 3 PO 4 nanostructure with large diameter distributions between 200 to 700 nm has been prepared. However, it should be noted that all corners and edges of these cubes were slightly truncated. Furthermore, when H 3 PO 4 solution was used, only Ag 3 PO 4 particles with spherical morphology have been prepared and no any perfect cubes has been observed in this sample (shown in Figure S7C,D). Based on the above results, we speculate that the morphology changes of Ag 3 PO 4 products should be due to the increased release rates of Ag + ions from [Ag(NH 3 ) 2 ] + complex as a result of the more H + ions contained in NaH 2 PO 4 and H 3 PO 4, which may be disadvantage for the growth of Ag 3 PO 4 cubes.
Figure S8. SEM images of Ag 3 PO 4 products prepared with different amounts of Na 2 HPO 4 solutions: (A, B) 0.01 M, (C, D) 0.03 M. Results and discussion In order to completely decompose [Ag(NH 3 ) 2 ] + complex to release Ag + ions, an excess amount of Na 2 HPO 4 solution must be introduced into the synthesis system. Thereby, it is important to know the relationship between amounts of Na 2 HPO 4 and morphology changes of Ag 3 PO 4 products, which may help us understand the growth process of Ag 3 PO 4 cubes. As shown in Figure S8A and 8B, when 0.01 M Na 2 HPO 4 was added into the reaction system, the obtained Ag 3 PO 4 products with irregular morphologies and a small quantity of cubes have been synthesized. However, after further addition of Na 2 HPO 4 (0.03 M), the percentages of Ag 3 PO 4 cubes with sharp corners, edges, and smooth surfaces in the samples have been greatly increased. These demonstrations clearly indicates that the Na 2 HPO 4 has reacted with the [Ag(NH 3 ) 2 ] + complex to release Ag + ions, which react with PO 3-4 ions to gradually form Ag 3 PO 4 cubes.
Scheme S2. Schematic illustration of the synthesis process of the Ag 3 PO 4 crystals with different morphologies: (a) [Ag(NH 3 ) 2 ] + complex for synthesizing Ag 3 PO 4 cubes, (b) CH 3 COOAg for synthesizing Ag 3 PO 4 rhombic dodecahedrons. Results and discussion At present, we speculate that the difference in the morphology for Ag 3 PO 4 crystals prepared from CH 3 COOAg and [Ag(NH 3 ) 2 ] + complex can be explained in terms of the various growth process. Furthermore, we identified a schematic illustration to explain some possible reasons for their differences. 1. As shown in Scheme S2a, when Na 2 HPO 4 was added into [Ag(NH 3 ) 2 ] + solution, the free Ag + ions have been gradually released from [Ag(NH 3 ) 2 ] + complex through the neutralization reaction between H + from Na 2 HPO 4 and NH 3 from [Ag(NH 3 ) 2 ] +, which subsequently reacted with PO 4 3- anions to form Ag 3 PO 4 crystals. More specifically, Na 2 HPO 4 could rationally control the release rate of Ag + ions and the growth rate of Ag 3 PO 4 crystals, which may promote the formation of Ag 3 PO 4 {100} planes and cause the resultant products with uniformly cubic structures. In contrast, in the presence of AgNO 3 precursor, the free Ag + ions present in this reaction system in considerable concentrations, and the growth process of Ag 3 PO 4 crystals is spontaneous and uncontrollable, which leads to the rapid nucleation and same growth rates of various planes of Ag 3 PO 4 crystals, resulting in an irregularly spherical morphology. 2. On the other hand, when CH 3 COOAg was used as the precursor (shown in Scheme S2b), the CH 3 COO - anions might adsorb preferentially with the Ag + ions of Ag 3 PO 4 {110} planes, which can effectively enhance their stability and reduce their growth rates during the synthesis process. Thereby, the growth rate in the [110] direction is much lower than that in other facet directions, finally, Ag 3 PO 4 rhombic dodecahedrons exposed with {110} facets have been fabricated. Moreover, as shown in Figure S6A,B, when Na 2 HPO 4 was replaced by NaH 2 PO 4, the as-prepared Ag 3 PO 4 products transformed into irregular rhombic dodecahedral crystals. Furthermore, when H 3 PO 4 solution was used, the obtained products are dominated by spherical particles with irregular morphologies (shown in Figure S6C,D). Thereby, we consider that in these cases, the decreasing of ph values (two and three protons contained in NaH 2 PO 4 and H 3 PO 4 ) may facilitate the generation of CH 3 COOH molecule, which may lose the ability of selective adsorption on the {110} planes and results in the Ag 3 PO 4 products with irregular shapes.
Figure S9. The intensity and wavelength distribution of the irradiation light employed in the organic decomposition experiments. Integral intensities were measured under the actual experimental conditions. Figure S10. (A,B) SEM images of irregular Ag 3 PO 4 particles prepared by directly reacting AgNO 3 with Na 2 HPO 4, (C) XRD pattern of irregular Ag 3 PO 4 particles, (D) Ultraviolet visible diffusive reflectance spectrum of irregular Ag 3 PO 4 particles.
Figure S11. The ultraviolet visible diffusive reflectance spectrums of N-doped TiO 2. Figure S12. (A) the ultraviolet visible diffusive reflectance spectrums of Ag 3 PO 4 rhombic dodecahedrons and cubes, and the intensity and wavelength distribution by using monochromatic visible light centred at a wavelength of 420.4 nm ( λ=±14.9 nm). The photocatalytic activities of Ag 3 PO 4 rhombic dodecahedrons and cubes for (B) MO and (C) RhB degradation under visible light λ=420.4 nm ( λ=±14.9 nm).
Figure S13. The adsorption properties of MO (A) and RhB (B) over Ag 3 PO 4 rhombic dodecahedra and cubes under dark conditions. Results and discussion At present, the surface defect of Ag 3 PO 4 rhombic dodecahedrons and cubes cannot be accurately detected as a result of the intrinsic limitations of Ag 3 PO 4 semiconductor in Electron microscope characterizations. However, it has been generally considered that the surface defects lead to the formation of surface active sites, and the surface defects and active sites are associated with each other. Thereby, we performed another experiment to further investigate their surface active sites by using adsorption measurements of MO and RhB under dark conditions. As shown in Figure S13A,B, it can be clearly seen that Ag 3 PO 4 rhombic dodecahedrons offers more active sites to adsorb dye molecule than cubes, indicating that the rhombic dodecahedrons possess more surface defect, which is highly consistent with the calculation results about the higher surface energy of {110} facets than that of {100} facets.
Figure S14. Surface energies of reduced and ideal surfaces of {110} and {100} facets. Surface energy for oxygen reduced surface. To investigate the stability of each surface under an oxygen reduced condition, we have studied the surface energy of reduced surfaces. Herein, we have employed a model structure, in which one oxygen atom was removed from each side of {100} and {110} slab shown in Figure 3. The geometries of the reduced surfaces were fully relaxed with the mid layer fixed in the same way as the ideal surfaces. The surface energy γ of the reduced surface was computed by γ = (E surf + μ O )/A, E surf = (E slab ne bulk )/2, where E slab is the total energy of the reduced slab, and μ O is the chemical potential of oxygen which is directly related to the O 2 partial pressure in ambient. Thus, the surface energy is a function of μ O. The result is shown in Figure S14. Here, we measure μ O with respect to the total energy of a single oxygen molecule, and the surface energy of the ideal surfaces are also shown in this figure. A system which gives lower surface energy is more likely to be realized, and thus the ideal surface is more realistic in a higher μ O while the reduced surface predominates in a lower μ O. The intersection of the two lines for the ideal and reduced surfaces gives the transition value of μ O, which is higher in {110} plane than in {100}, indicating the {110} surface is more easily reduced compared to the {100} surface. In other word, oxygen vacancies start to form on {110} surface even under a relatively high O 2 pressure in ambient. Therefore, the abundance of oxygen vacancies as well as the higher surface energy of a {110} surface possibly contribute to the higher reactivity of the Ag 3 PO 4 rhombic dodecahedral crystals.