Local structure change of luminescent Ag zeolite-A and -X studied by in situ XAFS and IR spectroscopy

The in situ X-ray absorption fine structure (XAFS) for the structural changes of Ag clusters produced in the cavity of luminescent zeolites by thermal treatment of Ag zeolite-A and Ag zeolite-X has been studied. The following procedures are compared: (i) samples are heated and cooled to room temperature under atmosphere (under air); (ii) samples are heated and cooled to room temperature in a vacuum and then exposed to air. It was confirmed that the Ag clusters were broken when the Ag zeolite was exposed to air for Ag zeolite-X, which complements our previous results for Ag₁₂-A. It is suggested that the deformation of the Ag clusters plays an important role in the generation of a strong photoluminescence band, and Ag clusters may not be direct species producing the strong photoluminescence. The local structure of the Ag ions was found to be slightly different from that of the unheated species. The difference may originate from the formation and breakdown of Ag clusters in the zeolite cavity.

1. Introdution

Zeolites show unique properties owing to the cavities and cages existing in the crystalline aluminosilicates. Several types of cations can be easily exchanged with Na⁺ ions in the Na-type zeolites and their unique physical and chemical properties will depend on the substituted cations (Kim et al., 2003).

Ag zeolites are typical functional materials, and can be used in information storage, catalysis, pressure and chemical sensors, and anti-bacterial materials. In the field of nanoscale science, the importance of metal nanoparticles and nanoclusters increases. Nanoparticles and clusters are produced simply in the zeolite cavity, and their particle sizes can be well controlled (Kim & Seff, 1987; Sun & Seff, 1994; Seifert et al., 2000; Calzaferri et al., 2003).

It has been reported that Ag-type zeolite-A, when cooled to room temperature and mixed with air after being heated at 500°C for 24 h under vacuum, shows a strong photoluminescence (PL) band around 2.18 eV (Hoshino et al., 2008). The same behaviour can be observed in the atmosphere. The intensity of the PL band is known to be affected by the conditions of the heating steps and the types of atmospheric gases exposed to after cooling (Nakamura et al., 2014). However, the detailed mechanism of the PL has been unclear. On the other hand, when the Ag-type zeolite is heated, the lattice water molecules in the cavity are removed and the Ag⁺ ion is reduced to Ag⁰; `Ag clusters' are produced inside the zeolite (Hoshino et al., 2008).

Recently the relation between Ag oligomers and PL was studied (De Cremer et al., 2009, 2010; Coutino-Gonzalez et al., 2013). The degree of Ag loading in the framework and the type of framework (A, X and Y) can tune the ionization potential, which correlates with the optical properties.

X-ray absorption fine structure (XAFS) is a powerful tool for studying the local structure in nanoscale materials and is suitable for in situ measurements (Hoshino et al., 2008; Nakamura et al., 2014; Altantzis et al., 2016). So far it has been reported that Ag₄ clusters are observed as the origin of the green/yellow emission in Ag zeolite-A and of the yellow emission in Ag zeolite-X (Altantzis et al., 2016; Fenwick et al., 2016); and Ag₃ clusters are observed as the origin of the green emission in Ag zeolite-Y (Coutino-Gonzalez et al., 2015).

In a previous paper (Nakamura et al., 2014) we investigated by XAFS the relation between the existence of Ag clusters and the enhancement of the PL intensity in order to clarify the PL mechanisms for Ag zeolite-A. We carried out in situ XAFS measurements under the same conditions as the PL measurements, and attempted to expose the zeolites to various gases – oxygen, nitro­gen, water vapour and their mixtures – after heating, in order to study the origin of the structural change of the Ag clusters. The result indicates that the Ag clusters produced by heating are destroyed by the introduction of air or a gas mixture of H₂O and N₂, but the Ag clusters remain intact when O₂ or N₂ are introduced. Coutino-Gonzalez et al. reported that the X-ray beam has a great influence on the dynamics of formation and destruction of luminescent Ag clusters in zeolites (Coutino-Gonzalez et al., 2014), but our result for the introduction of O₂ and N₂ gas is contradicted. We suppose that the existence of H₂O and N₂ gas is important for the destruction of Ag clusters and the effect of X-rays is not essential for cluster destruction.

In this paper, we further investigate how Ag clusters are destroyed after exposure to air in Ag zeolite-A and -X from XAFS and discuss the mechanism of PL in Ag-zeolites. Two procedures were applied: (i) Ag-zeolites are heated at a certain temperature in the atmosphere and cooled in air (including H₂O and N₂); (ii) Ag-zeolites are heated under vacuum and cooled to room temperature (RT) and then exposed to air.

2. Experimental

2.1. Sample preparation

Fully Ag⁺-exchanged zeolite A (hydrated 12Ag-A) powder samples were prepared by immersing the hydrated 12Na-A in 0.1 M AgNO₃ solution for 24 h at 25°C. After careful filtration the 12Ag-A powder was dried under air at RT in a darkroom. The sample is called `Ag<sub>12</sub>-A' in this paper. Analogously, the fully Ag<sup>+</sup>-exchanged zeolite X (hydrated 86Ag-X) and zeolite Y (51.2Ag-Y) were prepared by a similar procedure and named `Ag₈₆-X' and `Ag_51.2-Y', respectively. Detailed sample preparation procedures are reported in previous papers (Hoshino et al., 2008; Miyanaga et al., 2013).

Ag₁₂-A, Ag₈₆-X and Ag_51.2-Y were heated at 400 or 500°C in atmosphere or in vacuum to produce the Ag clusters. After keeping at 400 or 500°C for 24 h, they were cooled to RT in the atmosphere. In the vacuum process the sample was exposed to air after cooling to RT. In situ XAFS measurements were performed for each process.

2.2. PL measurements

PL spectra were obtained using a UV-VIS spectrometer (SPM-002; KLV) where a 405 nm violet laser (SU-61-405; audio-technica) and 365 nm LED (Ocean Optics LLS-LED365) were used for excitation. A high-vacuum chamber (RVX-3) was used to measure the PL spectra under various atmospheric conditions. PL was obtained after 24 h exposure to air.

2.3. XAFS measurements and data analysis

X-ray absorption spectra of the Ag K-edge (25.5 keV) were measured in transmission mode at beamline NW10A at Photon Factory, KEK, Japan. A Si(311) monochromator was used. For in situ XAFS measurements the RVX-3 vacuum chamber was used (Miyanaga et al., 2016). The powder sample was mounted on the sample cell in the RVX-3 chamber, in which temperature can be controlled from RT to 500°C and pressure can be controlled by vacuum pump (∼10⁻⁵ Pa). In order to discuss quantitatively the change of local structure around Ag, we performed non-linear least-squares fitting (curve-fitting) to EXAFS spectra using the equation

where r_j is the interatomic distance between X-ray absorbing Ag and photoelectron scattering atoms in the jth shell, and N_j is the coordination number of the jth shell. f_j(k,r_j) and φ_j(k) are the backscattering amplitude and phase shift functions, respectively, which were calculated using the FEFF8.10 code (Ankudinov et al., 1998). The k-range in the curve-fitting method is from ∼3.0 to 9.0 Å⁻¹ for the Ag K-edge. The fits were performed in k-space to the k²χ(k) function after back-transforming the R-space Fourier transform with the Hamming window. λ(k) is the photoelectron mean free path obtained by FEFF 8.10 calculation (Ankudinov et al., 1998). Ag foil and AgNO₃ were used as standard samples and S₀ ² (reduction factor by many-body effect) was determined as 0.83. As the origin of the photoelectron kinetic energy, E₀, in the fits we used the single value of 4.3 eV as the energy shift from the absorption edge.

As will be discussed later, we applied three-shell fitting for the EXAFS data. The number of unknown parameters was nine (r, N and σ for each shell). On the other hand, the number of possible fitting parameters was ten, obtained from the refinement of XAFS standards and criteria.

The EXAFS oscillation function was extracted from the X-ray absorption spectra and Fourier transformed using the XANADU code (Sakane et al., 1993). In order to obtain the structural parameters, the EXAFS function was fitted in k-space by a non-linear least-squares method using the theoretical parameters calculated by FEFF 8.10 (Ankudinov et al., 1998).

3. Results

3.1. Photoluminescence of Ag₁₂-A, Ag₈₆-X and Ag_51.2-Y

First, we present the results of the PL measurements. Fig. 1 shows the PL curves for luminescent species of Ag₁₂-A, Ag₈₆-X and Ag_51.2-Y, which were cooled to RT after heating to 400 or 500°C. Ag₁₂-A and Ag₈₆-X show a strong PL band at 2.18 eV and an intermediate band at 2.24 eV, respectively, after excitation by the 405 nm laser (a). On the other hand, a strong PL band was observed at 2.40 eV for Ag_51.2-Y after 365 nm excitation in addition to the 2.18 eV and 2.24 eV bands for Ag₁₂-A and Ag₈₆-X (b).

Figure 1 Photoluminescence (PL) spectra for Ag12-A, Ag86-X and Ag51.2-Y after heating under atmosphere after excitation by (a) 405 nm laser and (b) 365 nm LED. The relative intensities between (a) and (b) are not comparable.

3.2. XAFS for Ag₁₂-A under atmosphere

In this subsection, we present the EXAFS results for Ag₁₂-A heated to 500°C, kept for 24 h and then cooled to RT under atmospheric conditions. Fig. 2 shows the Ag-K X-ray absorption near-edge structure (XANES) (a), k²χ(k) spectra (b) and its Fourier transform (FT) (c) for Ag₁₂-A unheated at RT in the atmosphere (black line), heated at 500°C (red line), and cooled to RT after heating to 500°C (blue line). After the heating, the intensity of the white peak in the XANES becomes smaller and the energy shifts to slightly higher. After exposure to air at RT the XANES returns to that of the unheated species. It is found that the structure around 3.5 Å⁻¹ in (b) shows a clear change at 500°C, which is the characteristic feature of Ag clusters formation (Hoshino et al., 2008). In Fig. 2(c), the first FT peak around 1.8 Å is assigned to Ag—O₁ (nearest neighbour oxygen from the zeolite framework and lattice water molecule in the cage). The second FT peak around 2.8 Å includes the contribution from Ag—Ag and Ag—O₂ (further next nearest oxygen atom in the zeolite framework). The intensity of the FT peaks measured at 500°C is smaller than that measured at RT because of the large Debye–Waller factor. The fact that the k²χ(k) spectra for the sample cooled to RT after heating to 500°C are almost the same as the unheated one indicates that the Ag clusters produced in the process of heating to 500°C were broken by mixing with air, as discussed previously (Nakamura et al., 2014). On the other hand, their FT spectra are slightly different and a detailed discussion will be given later.

Figure 2 Ag-K (a) XANES spectra, (b) k^2χ(k) spectra and (c) Fourier transform for Ag12-A unheated at RT in atmosphere (black line), heated for 24 h and measured at 500°C (red line), and cooled to RT after heating to 500°C (blue line).

Table 1 shows the structural parameters r and N for O₁, O₂ and Ag for Ag₁₂-A for the process of heating under atmospheric conditions. During the heating process, r_O1 decreases from 2.38 Å to 2.28 Å and N_O1 is also reduced from 3.7 to 3.3, indicating the removal of lattice water molecules in the cavity. Even in the unheated species, a small portion of the Ag—Ag pair exists in the cavity naturally (r_Ag = 2.87 ± 0.01 Å and N_Ag = 1.3 ± 0.3). In the heating process, N_Ag increases from 1.3 ± 0.3 to 3.1 ± 0.3 which means that the formation of Ag clusters and the size of the clusters is almost the same as for the Ag₄ cluster, as previously reported (Fenwick et al., 2016). At this stage, Ag₁₂-A does not show a strong PL yet.

Table 1
Structural parameters r, N and σ for Ag₁₂-A for the process that heated under atmosphere

	rO1 (Å)	NO1	σO1 (Å)	rO2 (Å)	NO2	σO2 (Å)	rAg (Å)	NAg	σAg (Å)
Unheated sample measured at RT	2.38	3.7	0.13	2.87	0.9	0.10	2.87	1.3	0.11
Heated at 500°C for 24 h	2.28	3.3	0.15	2.70	1.7	0.16	2.81	3.1	0.17
Cooled to RT after heating in air	2.37	3.4	0.13	2.82	1.1	0.10	2.87	1.2	0.11

In the final process of cooling to RT after heating to 500°C, water molecules are incorporated into the cavity again. The parameters are almost the same as for the unheated case: for example, N_O1 (3.4 ± 0.3) and N_Ag (1.2 ± 0.3) are almost the same as for the unheated N_O1 (3.7 ± 0.3) and N_Ag (1.3 ± 0.3).

3.3. XAFS for Ag₁₂-A under vacuum

Next, we discuss the case where Ag₁₂-A was heated under vacuum and air was introduced after cooling to RT. Fig. 3 shows the Ag-K XANES (a), k²χ(k) spectra (b) and its FT (c) for Ag₁₂-A unheated species measured at RT in atmosphere (black line), unheated under vacuum (blue line), heated at 500°C (red line) for 24 h, and cooled to RT in vacuum (green line). After the heating, the intensity of the white peak in the XANES becomes smaller and the energy shifts to slightly higher. The behaviour is almost the same as that under atmosphere. As discussed in the previous subsection, the characteristic structure around 3.5 Å⁻¹ shows the existence of Ag clusters under vacuum state, see Fig. 3(b). In Fig. 3(c), a clear contribution from Ag—Ag contact can be seen in the vacuum condition and the Ag—Ag contribution increases with heating.

Figure 3 Ag-K (a) XANES spectra, (b) k^2χ(k) spectra and (c) Fourier transform for Ag12-A unheated species measured at RT in atmosphere (black line), unheated under vacuum (blue line), heated at 500°C for 24 h (red line), and cooled to RT in vacuum (green line).

Table 2 shows the structural parameters for Ag₁₂-A after heating in vacuum and 12 h after air introduction. During the first evacuation process, a part of a water molecule is removed from the cavity and the Ag cluster is grown. After continuous heating to 500°C, the Ag clustering process proceeds. However, N_Ag (= 2.8 ± 0.3) is slightly smaller than for the process in air discussed in the previous subsection (N_Ag = 3.1 ± 0.3) but the reason for this is still unclear. The structural parameters after 12 h of air exposure are almost the same as the result for cooling to RT in air as in the previous subsection (Table 1).

Table 2
Structural parameters r, N and σ for Ag₁₂-A for the process that heated in vacuum and 12 h after air introduction

	rO1 (Å)	NO1	σO1 (Å)	rO2 (Å)	NO2	σO2 (Å)	rAg (Å)	NAg	σAg (Å)
Vacuum at RT	2.29	3.2	0.13	2.82	1.7	0.10	2.83	2.0	0.11
After 24 h in vacuum at 500°C	2.28	2.9	0.14	2.71	2.1	0.16	2.80	2.8	0.17
RT in vacuum after heating	2.27	3.4	0.13	2.82	1.4	0.12	2.81	2.6	0.11
Exposure to air for 12 h	2.37	3.2	0.13	2.88	1.0	0.10	2.86	1.3	0.11

3.4. XAFS for Ag₈₆-X under atmosphere

In this subsection we discuss the result for Ag₈₆-X under atmosphere. Fig. 4 shows the Ag-K XANES (a), k²χ(k) spectra (b) and its FT (c) for Ag₈₆-X unheated at RT in atmosphere (black line), heated at 400°C (red line), and cooled to RT after heating to 500°C (blue line). After the heating, the intensity of the white peak in the XANES becomes smaller and the energy shifts to slightly higher value. After air exposure at RT the XANES returns to that of the unheated species. In the case of zeolite-X, there is a characteristic feature around 3.5 Å⁻¹ even in the unheated state in (b) and it becomes prominent after heating to 400°C. In (c) peaks for the heated to 400°C sample shift to shorter than unheated and after heated sample. Both spectra for (a) and (b) for the cooled to RT sample are almost the same as for the unheated one but slightly different as discussed for Ag zeolite-A.

Figure 4 Ag-K (a) XANES spectra, (b) k^2χ(k) spectra and (c) Fourier transform for Ag86-X unheated at RT in atmosphere (black line), heated for 24 h and measured at 400°C (red line), and cooled to RT after heating to 400°C (blue line).

Table 3 shows structural parameters for Ag₈₆-X for the process that heated under atmosphere. The structural change of r_O1 is almost the same as that for Ag₁₂-A. On the other hand, r_O2 and r_Ag are different; both are longer than for Ag₁₂-A. N_Ag for Ag₈₆-X is smaller than that for Ag₁₂-A.

Table 3
Structural parameters r, N and σ for Ag₈₆-X for the process that heated under atmosphere

	rO1 (Å)	NO1	σO1 (Å)	rO2 (Å)	NO2	σO2 (Å)	rAg (Å)	NAg	σAg (Å)
Unheated sample measured at RT	2.38	3.8	0.14	3.06	0.7	0.08	2.98	1.7	0.15
Heated at 400°C for 24 h	2.28	3.1	0.16	2.71	1.0	0.17	2.84	2.3	0.20
Cooled to RT after heating in air	2.37	3.5	0.14	3.05	0.6	0.08	2.98	1.5	0.15

Values of r_Ag for the unheated sample and clusters are 2.98 ± 0.01 Å and 2.84 ± 0.01 Å, longer than for Ag₁₂-A (2.87 ± 0.01 Å and 2.81 ± 0.01 Å). Ag atoms are strongly connected to O atoms in the framework; however, r_O1 is similar to that for Ag₁₂-A. The cavity size for zeolite-X is larger than that for zeolite-A; then r_Ag of the Ag clusters is longer than that for Ag₁₂-A.

3.5. XAFS for Ag₈₆-X under vacuum

Fig. 5 shows the Ag-K XANES (a), k²χ(k) spectra (b) and its FT (c) for Ag₈₆-X unheated at RT in atmosphere (black line), RT in vacuum (red line), heated for 24 h and measured at 400°C (yellow line), cooled to RT after heating to 400°C (blue line), and after 8 h exposure to air (green line). The qualitative change of structure is the same as for that under atmospheric conditions. After the heating, the intensity of the white peak in the XANES becomes smaller and the energy shifts to slightly higher. After air exposure at RT the XANES returns to that of the unheated species. It is noted that the change in XANES for Ag₈₆-X under vacuum is smaller than for other systems.

Figure 5 Ag-K (a) XANES spectra, (b) k^2χ(k) spectra and (c) Fourier transform for Ag86-X unheated at RT in atmosphere (black line), RT in vacuum (red line), heated for 24 h and measured at 400°C (yellow line), cooled to RT after heating to 400°C (blue line), and after 8 h exposure to air (green line).

Table 4 shows the structural parameters for Ag₈₆-X for the process that heated under vacuum and 8 h after air introduction. N_Ag (= 2.8 ± 0.3) is larger than that for atmospheric conditions in the previous subsection (N_Ag = 2.3 ± 0.3). This cluster size (N_Ag = 2.8 ± 0.3) corresponds to `Cluster Type I' discussed in our previous paper (Miyanaga et al., 2013) for Ag₈₆-X, where the cluster size depends on the filling rate of Ag ions in the zeolite-X cavity. On the other hand, the result for this cluster size (N_Ag = 2.8 ± 0.3) corresponds to that for the Ag₄ cluster as proposed by Altantzis et al. (2016) and Fenwick et al. (2016). From the coordination number, it is not so easy to determine which cluster model, Ag₄ or Ag₈, is adequate.

Table 4
Structural parameters r, N and σ for Ag₈₆-X for the process that heated in vacuum and 12 h after air introduction

	rO1 (Å)	NO1	σO1 (Å)	rO2 (Å)	NO2	σO2 (Å)	rAg (Å)	NAg	σAg (Å)
Vacuum at RT	2.29	3.4	0.14	2.94	0.4	0.10	2.84	2.1	0.15
After 24 h in vacuum at 500°C	2.29	3.2	0.15	2.73	3.0	0.18	2.82	2.8	0.18
RT in vacuum after heating	2.29	2.5	0.12	2.84	1.4	0.15	2.82	1.9	0.12
Exposure to air for 8 h	2.37	3.3	0.14	3.02	0.5	0.08	2.96	1.5	0.15

3.6. IR spectra of Ag₁₂-A

Fig. 6 shows the infrared (IR) transmission spectra of Ag₁₂-A for the sample exposed to air (a) and oxygen (b). The black line is for the unheated Ag zeolite. Due to the complicated structure of zeolite, the bands are superimposed on the various Si—O and Al—O contributions, so it is extremely difficult to distinguish between them. Only the assignment of the absorption band around the 1000 cm⁻¹ region as ν(T—O) (T = Si or Al) (Rodrigues, 1995) is possible. IR bands at relatively higher wavenumber are assigned to the ν_as (asymmetric) mode, and the lower region to the ν_s (symmetric) mode. A bending mode should appear in the lower energy region but is not shown here. The green line shows the spectrum of Ag₁₂-A cooled to RT after heating for 24 h at 500°C in vacuum (measured in a vacuum). Compared with the unheated case, the absorption band has changed drastically. While the individual band assignment is not possible as mentioned above, by considering the results of XAFS it is reasonable to assume that the change is due to the formation of Ag clusters. For cluster formation, it is necessary that Ag⁺ ions (at least partially) are reduced to Ag⁰; also Ag which is to be aggregated away from the site of the Ag⁺ ion was located. This reduction of the ions and change of position also affects the vibrations of the framework in each case. The changes in the distances between the framework and the Ag atoms and the changes in the electrostatic attraction of Ag⁺ can perturb the vibration of the T—O bond. In other words, it can be said that the IR is sensitive to such a change in the zeolite framework. The green lines in Fig. 6 are the spectra for the case of introducing air (a) and oxygen (b) into the zeolite, which are cooled to RT after the vacuum heating. After introduction of the air, the spectrum greatly changed again, and is much closer to that of the unheated spectrum (black line). Whereas for XAFS the air intake after heating substantially coincides with that for the unheated one, IR spectra for both states are not completely the same as each other. A distinct band appeared at 1140 cm⁻¹ and the strength of the main band at 1000–900 cm⁻¹ decreased slightly. After the introduction of oxygen, the situation is different. The structures at 1100 cm⁻¹ and 1180 cm⁻¹ still remain after the introduction of O₂ gas, indicating that Ag clusters are not destroyed by O₂ gas. This supports our previous XAFS results (Nakamura et al., 2014).

Figure 6 IR spectra of Ag12-A for (a) air exposure and (b) oxygen exposure.

4. Discussion

In this section we discuss the above results for Ag₁₂-A and Ag₈₆-X. As already discussed in the previous paper for Ag₁₂-A (Hoshino et al., 2008; Nakamura et al., 2014), Ag clusters are produced in the zeolite cavity by heating in air or in vacuum, and the qualitative results are the same for Ag₈₆-X. For Ag₁₂-A, the Ag cluster is grown only by evacuating the sample at RT. On the other hand, for Ag₈₆-X, the formation of Ag clusters is incomplete by only evacuating the sample at RT, but is completed by heating to ∼400°C.

The formation of Ag clusters in the zeolite cavity is confirmed by the increase of the coordination number of Ag—Ag (N_Ag) from the present XAFS analysis. On the other hand, XANES also shows that the valence changes to Ag⁰; the intensity of the white line peak becomes smaller and the energy shifts to slightly higher as the samples are heat-treated both in atmosphere and in vacuum. These XANES behaviours indicate the existence of Ag clusters or a component of Ag⁰ as discussed previously (Suzuki et al., 2005). Although unfortunately we did not study X-ray diffraction (XRD), transmission electron microscopy and electron diffraction for the present Ag-zeolite samples, Altantzis et al. presented direct observations of Ag clusters using XRD and HAADF-STEM for the Ag zeolite-X as similarly heat-treated as in our present study (Altantzis et al., 2016).

In Table 5, the local structure parameters for Ag—Ag interactions are summarized along with the PL emission energy. The interatomic distances for Ag clusters produced by heating are similar among the three zeolites, i.e. they are 2.82 ± 0.01 Å. On the other hand, the coordination number is different: N_Ag for Ag₁₂-A is 3.0 ± 0.3, N_Ag for Ag₈₆-X is 2.3 ± 0.3. The cluster models estimated from N_Ag are Ag₄ for Ag₁₂-A and Ag₃ for Ag₈₆-X.

Table 5
Interatomic distances (r_Ag ± 0.01 Å) and coordination numbers (N_Ag ± 0.3) for the Ag—Ag interaction for the unheated species, and for Ag clusters and cluster deformed species; the PL energy at 365 nm excitation is also listed

	Interatomic distance / coordination number
	Unheated	Ag clusters	Deformed	PL energy
Ag12-A	2.87 Å / 1.3	2.81 Å / 3.0	2.87 Å / 1.2	2.18 eV
Ag86-X	2.98 Å / 1.7	2.83 Å / 2.8	2.97 Å / 1.5	2.24 eV

As for Ag₁₂-A (Nakamura et al., 2014), the Ag clusters are broken and the local structure around the Ag atoms returns to almost a similar state as for the unheated species by cooling to RT under air, existing also for Ag₈₆-X, but the local structure around the Ag atoms does not completely coincide with that for the unheated species. This result is supported by IR spectra where clear differences appear between the unheated species and after exposure to air (Fig. 6). The IR structure assigned to the framework of zeolite after heating and air exposure is different from that for the unheated species.

Ag₁₂-A and Ag₈₆-X show a strong enhancement of PL intensity by cooling with air after heating to ∼400°C, but no PL enhancement appears in the unheated species and heated species under the existence of Ag clusters (Nakamura et al., 2014).

The Ag—Ag interatomic distances r_Ag after cooling with air (luminescent species, after full exposure to air) are 2.87 ± 0.01 Å for Ag₁₂-A and 2.98 ± 0.01 Å for Ag₈₆-X; it is interesting that these orders are the same as those of the packing ratio of the Ag atoms in the cavity (i.e. the number of Ag atoms to volume of unit cell, A > X). That is, highly packed species have contracted structures and show shorter Ag—Ag distances.

On the other hand, the energies of the PL are 2.18 eV for Ag₁₂-A and 2.24 eV for Ag₈₆-X excited by 365 nm laser [as shown in Fig. 1(b) and Table 5, which may correspond to the result that the lower the packing ratio, the higher the PL energy, as reported by De Cremer et al. (2009). The longer the Ag—Ag distance in the Ag zeolite species, the higher the PL emission energy.

We have discussed the local structure change around Ag atoms for Ag₁₂-A and Ag₈₆-X during heating (cluster formation process) and following cooling with air (cluster deformation process) by XAFS analyses. The structure model of Ag clusters is almost the same as that proposed previously. However, in the luminescent Ag zeolite species, Ag clusters no longer exist, and they are deformed to similar structures as the unheated species; whereas the local structure of deformed species is rather similar to that of the unheated species but confirmed to be different. At the present stage, we cannot estimate the difference between them, but it should be a key point to reveal the luminescent species and the mechanism of this system.

5. Conclusion

We studied the in situ XAFS for the structural changes of Ag clusters in luminescent Ag₁₂-A and Ag₈₆-X produced by thermal treatment in atmosphere and in vacuum. It was confirmed that the Ag clusters were broken when the zeolite is exposed to air for Ag₈₆-X, which complement our previous results for Ag₁₂-A. After the air exposure, the strong PL still appears; it is suggested that the deformation of the Ag clusters plays an important role in the generation of the strong PL band and Ag clusters may not be direct species producing the strong PL. It is found that the local structure of Ag ions is slightly different (bond distances are slightly shorter) than for the unheated species which are induced by the formation and breakdown of Ag clusters; this is key point of the PL enhancement mechanism.