Photochromism in the periodic nanospace of zeolite LTA by the transfer of photoexcited electrons of adsorbed Na atoms

Abstract

Cationic sodium (Na) clusters incorporated into a dehydrated Na-form LTA (Na-LTA) zeolite by the adsorption of Na atoms exhibit diamagnetism regardless of their adsorption amount. These clusters preferentially form in β-cages under the condition of dilute adsorption. In this study, photochromism was observed on the dilute Na-adsorbed Na-LTA, which showed a color change from yellow-green to dark blue-green at room temperature. This phenomenon resulted in dissociation of the diamagnetic Na clusters in β-cages upon light irradiation and thereby in the formation of two metastable Na₄³⁺ clusters with C_3v symmetry in α-cages from the photoabsorption and electron spin resonance spectra. We also observed photochromism in a sample of insufficiently dehydrated Na-LTA with dilute Na adsorption—i.e., the initial white color changed to blue, then blue-green and finally dark moss-green upon irradiation with UV light. This phenomenon consists of two steps. First, products in the β-cages, formed by the reaction of residual water molecules with adsorbed Na atoms, release electrons upon irradiation. The electrons transfer to the α-cages, forming the metastable Na₄³⁺ clusters. Next, the diamagnetic Na clusters are formed in vacant β-cages by the thermal transfer of electrons from the Na₄³⁺ clusters in the α-cages. Na clusters including multiple electrons were also formed in the α-cages by long exposure to UV light irradiation, which had a similar effect to increasing the adsorption amount of Na atoms onto the dehydrated Na-LTA. After termination of irradiation, the sample required a week to change its color to light yellow.

---

Introduction

Zeolites, which are inorganic microporous crystals with regular nanopores or nanochannels smaller than 2 nm, have been applied to adsorbents, ion exchangers, catalysts, membranes and so on.^1–4 While researchers in various fields have sought to expand the applications of zeolites, the astounding phenomenon of ferromagnetism on arrayed potassium (K) clusters incorporated into aluminosilicate zeolite LTA crystals in the absence of magnetic elements has attracted researchers in the fields of nanomaterials and nanotechnologies.⁵ (Three-character codes in bold font indicate the framework topology authorized by the International Zeolite Association.) Many valuable results have been achieved with new zeolite-based composite materials.⁶ High-density arrayed nanoparticles of semiconductors and metals in cage-type zeolite crystals are typical examples.^7–11 By using channel-type aluminophosphate AFI crystals, carbon nanotubes of 0.4 nm in diameter and aromatic molecules arranged one-dimensionally in a preferred orientation have been achieved that give rise to superconductivity and non-linear optical properties, respectively.^12–14

The origin of ferromagnetism on the K clusters in K-form LTA with Si/Al = 1 (K-LTA) and the detailed electronic structure of these clusters have been studied across decades, because the many-body problem of electrons (electron correlation) has been a quite difficult issue to clarify.^15–21 In contrast, there is a paucity of research on sodium (Na) clusters in Na-form LTA with Si/Al = 1 (Na-LTA), because they merely exhibit diamagnetism.^22,23 The locations of the incorporated Na and K clusters in the crystal structure of LTA also differ. Independent of the amount of K atoms adsorbed onto K-LTA, K clusters are formed only in α-cages,²⁴ as shown in Fig. 1, whereas Na clusters are formed in β-cages first, and then in α-cages.²⁵ The numbers of Na clusters in the β-cages take a downward turn when the amount of Na atoms adsorbed onto Na-LTA is increased further.^23,25 This means that the ground state energies of the Na clusters in the two types of cages are competing.

Fig. 1 Average crystal structure of dehydrated Na-LTA. Left: models of the α- and β-cages with Na⁺ ions at 6R, 8R, and 4R. Right: lattice points of the framework Si and Al atoms and the connections between them omitting O atoms. The lattice constant is a ≈ 2.46 nm, when the alternative arrangement of Si and Al atoms is considered.

Before describing our research strategy, we will explain the characteristics of the alkali–metal clusters in zeolites and the related features of Na-LTA. Alkali–metal clusters in aluminosilicate zeolites are cationic, because the valence s-electron of the adsorbed alkali–metal atom, which is simply called “electron” hereinafter, spreads over the charge-compensating alkali–metal cations for the negatively charged zeolite framework. For example, Na₄³⁺ clusters were detected using electron spin resonance (ESR) spectroscopy in Na atom-adsorbed Na-form FAU with Si/Al = 2.0–2.4 (Na-FAU).^26,27 Thus, the distribution of Na⁺ in the initial Na-LTA strongly affects the structures and electronic states of the Na clusters within Na-LTA. In the case of dehydrated Na-LTA which has a chemical composition of (Na₁₂Al₁₂Si₁₂O₄₈)₈ per unit cell (UC), its structure (the space group Fm3̄c with a lattice constant of a ≈ 2.46 nm) is illustrated in Fig. 1.^28–30 (Detailed structural features are given in Section S1†.) The center of the 6-ring (6R) (Wykoff position: 64g) is occupied by a Na⁺ ion. There are four sites occupied by Na⁺ at 8R (Wykoff position: 96i), whose occupancies are 1/4, i.e., one Na⁺ ion is positioned per 8R. Residual Na⁺ ions are located at the inner wall of the α-cage, i.e. near the center of the 4-ring (4R) (Wykoff position: 96h), with an occupancy of 1/12. Regarding the distribution of the cations, the sites of the K⁺ ions and their occupancies affected the structural and electronic properties of the incorporated K clusters in K-form LTAs with Si/Al = 1–3.^31–35 Hereinafter, we use the term pseudo unit cell (pUC), Na₁₂Al₁₂Si₁₂O₄₈ (a ≈ 1.23 nm), for convenience, because one α-cage and one β-cage are in the pUC.

Returning to our strategy, some organic dye molecules alter their molecular and/or electronic structures into their isomers upon external stimulations, e.g. light, heat, electric field and so on. This phenomenon with a change of color (optical properties) is called chromism.^36–39 Since chromism is usually a reversible phenomenon, the ground state energies of the two isomers before and after the stimulations are close. The Na clusters in the two types of cages of Na-LTA are competing in terms of their ground state energies as mentioned. Creation and annihilation of Na clusters in the periodic nanospace of zeolites by external perturbations might be a new type of chromism. However, despite its potential use in new applications, this fascinating phenomenon has been largely overlooked.

Here, we focused on the dehydrating conditions and the amount of adsorbed Na atoms for sample preparation to induce effective photochromism in Na-LTA. To evaluate this phenomenon, photoabsorption and electron spin resonance (ESR) spectroscopy techniques were adopted as primary tools. We determined the time dependence of the photoabsorption spectra at a second-scale time resolution on light irradiation to obtain information on the dynamic variation of the electronic states of the Na clusters.

Experimental section

Sample preparation

The Na-LTA used in this study was obtained from Tosoh, Tokyo, or synthesized in-house according to the previous procedure.⁴⁰ The latter Na-LTA was characterized by PXRD, scanning electron microscopy (SEM), thermogravimetry and differential thermal analysis (TG-DTA). Their results are described in detail in Section S2.1.† H₂O in the cages of Na-LTA is resistant against dehydration. Thus Na-LTAs dehydrated under two types of dehydrating conditions, which are hereafter referred to as Na-LTA(m) and Na-LTA(s), were used as hosts for Na adsorption. Na-LTA(m) was dehydrated under a mild condition of heating at 673 K for 2 h under evacuation using a rotary pump, with a final vapor pressure of ca. 1 × 10⁻¹ Pa. Na-LTA(s) was dehydrated under a severe condition of heating at 773 K for more than 15 h under evacuation using diffusion and rotary pumps, and the final pressure was lower than 1 × 10⁻³ Pa. Based on the FT-IR spectra as shown in Fig. 2, Na-LTA(s) had less residual H₂O. (See Section S2.2 for details.†)

Fig. 2 FT-IR spectra of hydrated and dehydrated Na-LTAs.

Na metal distilled at least three times in vacuo was sealed with dehydrated Na-LTA in an evacuated quartz or borosilicate glass tube. Na metal was adsorbed onto dehydrated Na-LTA through its vapor phase by heating the sample glass tube at 323–373 K in an electric furnace. Optical spectra of the samples with stepwise increased amounts of Na atoms could be obtained by measuring the spectra and adsorbing Na alternately using one sample.

Characterization

Because the powdered samples are sealed in glass tubes to prevent the rehydration and oxidation of Na atoms, the samples in the glass tubes were measured as they were. The obtained diffuse reflection (DR) spectra of the powdery samples were converted into absorption spectra using the Kubelka–Munk (K–M) function.⁴¹ The relationship among the K–M value, absorption coefficient, and diffuse reflectivity, which were denoted as A_K–M, K, and r, respectively, was defined as A_K–M = K/s = (1 − r)²/2r, where s is the scattering coefficient that can be interpreted as a reciprocal of the average size of powder particles. The K–M function cannot be adopted to very weak diffuse reflectivity.²⁴ Therefore, the Na-LTA powder was ground using a mortar and pestle to reduce the particle size. The ground Na-LTA powder was used to obtain the photoabsorption spectra in Fig. 4. In our experience, A_K–M higher than 40 in a UV-vis spectrum is in danger of underestimating the absorption intensity or spectral deformation.

To realize the measurement of DR spectra before and under additional light irradiation to induce photochromism, we used the system described in Section S3.1.† In brief, even if the sample is simultaneously irradiated with light for inducing photochromism and light for the spectral measurement, only the signal from the measurement light will be extractable by using a lock-in amplifier. Before measuring the spectra including photochromic information, we light-irradiated the sample for more than 30 min until the sample color (photoabsorption) became constant.

To detect dynamic phenomena in the optical absorption spectra at a time scale of seconds, we constructed an optical system equipped with a diode array (Section S3.1†). Spectra were measured at both RT and 77 K. We also used a Cary5000 spectrophotometer (Agilent Technologies) to measure the DR spectra in cases where additional light irradiation to induce photochromism was unnecessary.

X-band ESR spectra were measured in a first derivative mode using an ESP300E spectrometer (Bruker) because ESR spectroscopy is known to be a powerful tool for detecting metastable and/or transient unpaired electrons under light irradiation.^42,43 Photochromism was induced by using blue LEDs or monochromatic Xe lamp light through the ESR cavity windows (Section S3.2†).

FT-IR spectra were measured to detect residual H₂O in the cages of Na-LTA through their –OH stretching bands. The spectra were obtained by using a Nicolet Magna750 spectrometer in diffuse reflectance geometry while keeping the powder samples in the quartz glass tubes. Powder X-ray diffraction (PXRD) patterns were obtained using Rigaku SmartLab and PANalytical Empyrean X-ray diffractometers; scanning electron microscopy (SEM) images were obtained using a Hitachi S4800 scanning electron microscope; and thermogravimetry and differential thermal analysis (TG-DTA) curves were obtained using a Rigaku Thermo plus EVO2 TG-DTA8120. For the direct observation of light irradiation-induced color variation, optical microscopy time-lapse images were recorded using an Olympus SZX10. All the above methods are described in detail in Section S3.3.†

Results and discussion

Photochromic properties of Na-adsorbed Na-LTA(s)

As a typical example, the color of the prepared sample with Na atoms adsorbed on Na-LTA(s), which is labeled Na/Na-LTA(s), used for the later ESR measurement varied from yellow-green to dark bluish-green as shown in Fig. 3. Photoabsorption spectra of Na/Na-LTA(s) with various amounts of adsorbed Na atoms before and under light irradiation at room temperature (RT) are shown in Fig. 4. With increased amounts of adsorbed Na atoms from samples (a) to (e), photoabsorption increased. The photoabsorption intensity higher than 40 plotted by the broken curves in (e) may be underestimated. (This is common to all the following spectra.) The derivation of the adsorbed amounts of Na atoms for samples (a)–(e), along with the confirmation of the crystallinity of these samples after grinding, dehydration, and Na adsorption, is described in Section S4.1.†

Fig. 3 Photochromism of Na/Na-LTA(s) used for the ESR measurement. Color change was induced by white LED light used to take these photos of the sample. The total time of light irradiation is given under each image.

Fig. 4 Photoabsorption spectra of Na/Na-LTA(s). Blue and red spectra were obtained before and under light irradiation of 3.2 eV, respectively. Amounts of adsorbed Na atoms per pUC of samples (a)–(e) were ca. 0.2, 0.3, 0.4, 0.6, and 1.0, respectively. The spectral baselines of (b)–(e) are shifted.

At the initial stage of Na atom adsorption ((a) and (b)), absorption bands at 3.1 and 2.8 eV labeled D₁ and D₂, respectively, appeared in this order. These bands originate from the Na clusters in β-cages.²⁵ Under light irradiation whose photon energy resonated to these bands, new bands labeled C₁ and C₂ appeared at 2.0 and 1.65 eV, respectively, instead of the decrease of the D₁ and D₂ bands. The origin of these newly appeared bands will be discussed later. With increasing Na adsorption (samples (c)–(e) in Fig. 4), the bands labeled P, S₁, and S₂ appeared at 2.0, 1.4 and 1.2 eV, respectively, and grew irrespective of light irradiation. These three bands were from the Na clusters in α-cages, whose origins were assigned to the surface plasmon-like collective excitation of multiple electrons (P band) and individual one electron-like excitation between quantized electronic levels (S₁ and S₂ bands).²⁵ It was interpreted that four or six electrons were included in these clusters configuring a spin singlet state providing a diamagnetic property.²³ With the growth of these bands, the spectra became insensitive to light irradiation. These results indicate that photochromism occurs efficiently under the condition of dilute Na adsorption. The Na clusters in β-cages are the key, and we can easily imagine that the generated species are formed in vacant α-cages. The possibility of photochromism at 77 K was also examined, but the C₁ and C₂ bands did not appear under irradiation. The spectrum at 77 K, however, was narrowed compared with that at RT (Fig. S8†). This indicates that an electron–phonon (vibration/displacement of Na⁺ ions) interaction was present and stabilized these Na clusters against light irradiation at 77 K.

ESR study of Na-adsorbed Na-LTA(s)

In consideration of the above results, we focused on Na/Na-LTA(s) with dilute Na adsorption for the ESR measurements in which the Na clusters in β-cages are the main species (Fig. S9†). The amount of adsorbed Na was estimated to be ca. 0.1 atom per pUC from the photoabsorption spectrum of this Na/Na-LTA(s) using Smakula's equation, as described in Section S4.3.†^34,44

ESR spectra of this sample before and under light irradiation are shown in Fig. 5a. Measurement conditions are given in Table S1.† Meaningful fine structures already exist in the spectrum before light irradiation. We first estimated the density of the unpaired electrons contributing to this spectrum by using the ESR spectrum of paramagnetic copper(II) sulfate pentahydrate (CuSO₄·5H₂O) crystals as a reference. The estimated density was ca. 4 × 10⁻⁴ per pUC (β-cage); the derivation is described in Section S4.4.† As the adsorbed Na atoms were estimated to be ca. 0.1 per pUC, most of the electrons in the Na clusters in β-cages were ESR-silent. This finding is consistent with the previous report that the Na clusters in β-cages are diamagnetic, meaning that they include two electrons in a spin singlet state.²³ A few thermally generated ESR active species must be the origin of the fine structures already existing before irradiation. As can be seen in the expanded region of Fig. S9,† very weak photoabsorption corresponding to C₁ and C₂ bands could be observed.

Fig. 5 ESR spectra of Na/Na-LTA(s) measured at RT: (a) before (blue) and under (red) light irradiation at 3.1 eV; (b) photoinduced component of (a); and (c) differential (second derivative) of the spectrum of (b).

Upon light irradiation, the fine structures that appear in the ESR spectrum increased their intensities. To elucidate the origin of the photoinduced ESR active species, its component was extracted by subtracting the ESR spectrum before irradiation from that under irradiation, as shown in Fig. 5b. The photoinduced ESR spectrum is centered at the g-value of 2.002 ± 0.002, which is close to the free electron value of g = 2.0023, with ten hyperfine structures (HFSs) in a regular separation of A_Na1 = 36.8 ± 0.2 G. Because meaningful traces could be observed on these HFSs, we differentiated the spectrum to obtain the second derivative spectrum, as plotted in Fig. 5c. A set of four weak components with a small splitting of A_Na2 = 5.7 ± 0.1 G can be seen for each intense HFS located near the spectral center. From the analogy of the ESR spectrum of naphthalene anions, whose hydrogen atoms (nucleus spin I_H = 1/2) can be classified into two symmetric positions, in an organic solution, the present photoinduced ESR spectrum was interpreted.^45–47 Two types of Na⁺ ions with different hyperfine interactions of A_Na1 and A_Na2 exist. Because the nucleus spin of Na is I = 3/2 (100% abundance), the ten HFSs and the four differential peaks on each HFS correspond to three equivalent Na⁺ ions and one Na⁺ ion, respectively. Details are described in Section S4.5.† By simulation, we could completely reproduce the experimental spectrum as shown in Fig. S11.† Combining the results of photoabsorption and the ESR spectra, the photogenerated species is the Na₄³⁺ cluster in the α-cage consisting of two types of Na⁺ ions.

Compared to the Na₄³⁺ clusters in the β-cages of Na-FAU observed as 13 lines of HFSs,²⁷ which have the symmetry of T_d, the present photogenerated Na₄³⁺ clusters have a different point symmetry of D_3h or C_3v. The latter point symmetry is more likely, because the three equivalent Na⁺ ions are the main constituent of the cluster and the Na⁺ ion with the small constant, A_Na2, must locate away from the main Na⁺ ions. Considering this point symmetry and the initial Na⁺ sites of the dehydrated Na-LTA, we propose the local structure of the present Na₄³⁺ clusters with the C_3v symmetry illustrated in Fig. 6. (For a clearer image, see the ESI animation.†) The [111] axis of the Na-LTA has a three-fold symmetry. Therefore, the Na⁺ ions from the three 8Rs labeled Na1 make the main and equivalent contribution to the cluster, and the contribution of the Na⁺ ion attributed to 6R, which is labeled Na2, to the [111] axis is minor.

Fig. 6 Schematic illustration of the local structure of the α-cage where the metastable Na₄³⁺ cluster is generated. (a) and (b) are viewed from the [010] and [111] directions, respectively. Na1 and Na2 are the Na⁺ ions of the cluster, and the hazy translucent green region represents the electron cloud of the unpaired electron.

Photochromic properties of Na-adsorbed Na-LTA(m)

As described in the Experimental section, a small amount of water molecules (H₂O) (≈0.1 pUC) remains in mildly dehydrated Na-LTA, Na-LTA(m). By the stepwise increment of the Na adsorption amount onto this Na-LTA(m), the photoabsorption assigned to the stretching mode of the hydroxyl groups (–OH) decreased gradually, as shown in Fig. 7a. In contrast, in the UV region, the photoabsorption band at 5.5 eV labeled W in Fig. 7b gradually increases. (The adsorption amount of Na seems to be ≈0.1 atom per pUC for the sample, giving the curves numbered 4 in Fig. 7.) Although the reacted species between the residual H₂O and the adsorbed Na atoms are the origin of the W band, their concrete form is uncertain. Therefore, we call these species W for convenience. When band D₁ at 3.1 eV appeared after the growth of the W band, no OH stretching vibration could be observed in the FT-IR spectrum. Species W appeared to be located in β-cages, because a small amount of H₂O (low site occupancy of H₂O) was detected in β-cages by the analysis of the PXRD (powder X-ray diffraction) pattern of dehydrated Na-LTA.³⁰ Na clusters were formed in vacant β-cages (without species W) after all residual H₂O molecules were consumed for the reaction with the adsorbed Na atoms. The following investigations of Na-adsorbed Na-LTA(m) (Na/Na-LTA(m)) were conducted for the samples, providing only the W band before light irradiation.

Fig. 7 (a) FT-IR and (b) UV-vis-near IR spectra of Na-LTA(m) (spectrum 1) and those with a stepwise increment of adsorbed Na atoms (spectra 2–4). Spectra labeled with the same numbers in the two figures are from the same sample.

The inset of Fig. 8a shows the spectra of Na/Na-LTA(m) with dilute Na adsorption before and just after prolonged UV irradiation. The adsorbed amount of Na is estimated to be ≈0.1 atom per pUC. The actual color of the sample varied from white to light blue, to bluish green, and finally to dark moss-green upon irradiation, as shown in Fig. 9. In the spectrum just after UV irradiation, photoabsorption was detected at around 2 and 3 eV, and the W band decreased. The spectral shape at around 2 eV differs from that of Na/Na-LTA(s) under light irradiation (sample (a) in Fig. 4). These features were investigated as follows.

Fig. 8 (a) Dependence of the absorption spectra of Na/Na-LTA(m) on light irradiation duration, t, at RT. Monochromatic Xe lamp light of 5.5 eV was irradiated. On the left side, the base line was stepwise shifted for the spectra for t ≥ 10 min irradiation. The inset shows the spectra before and just after the sufficient UV-light irradiation using a D₂ lamp. (b) Dependence of the photoabsorption intensities at 1.65, 2.00, and 2.9 eV extracted from (a).

Fig. 9 Photochromism of Na/Na-LTA(m). Color change was induced by a D₂ lamp. The total time of light irradiation is given under each image. The last image for the irradiation time of 5 min (right side) was obtained by increasing the sensitivity of the CMOS camera to intensify the sample color.

The dependence of the absorption spectra of this sample on the duration of UV light irradiation (5.5 eV) was determined at RT as shown in Fig. 8a. To make clear the dependence of spectral variation on the irradiation duration, the absorption intensities at 1.65, 2.0, and 2.9 eV are plotted as shown in Fig. 8b. The absorptions at 1.65 and 2.0 eV appeared within 1 min and grew proportionally to the irradiation duration until 3–4 min. After 4 min, the growth became slower. The species giving these photoabsorptions were directly generated by irradiation. In contrast, absorption at 2.9 eV was negligible at the initial 1–2 min of irradiation, but once the absorption appeared, its growth was accelerated after 3 min. The growth features of the 2.9 eV photoabsorption imply that the species of its origin were not directly generated by irradiation. More specifically, the formation of intermediates upon irradiation results in the characteristic absorption growth.

Looking more closely at Fig. 8b, we can see that the ratio between the intensities of 1.65 and 2.0 eV is initially invariant, and the spectral profile at 1.3–2.4 eV within 10 min of irradiation (Fig. 8a) is identical to that of Na/Na-LTA(s) which indicates the C₁ and C₂ bands upon irradiation (samples (a) and (b) in Fig. 4). Thus, the absorption bands appearing at 2.0 and 1.65 eV at the early stage of irradiation can be assigned to C₁ and C₂, respectively. The spectrum at 4 min of irradiation was deconvoluted as shown in Fig. S12.† The ratio between these bands was 1 : 1.80 (Table S3†), and their origin will be discussed later. At the increased irradiation duration, the absorption at 1.65 eV peaked after ca. 10 min and then decreased slightly at 120 min. In contrast, the absorption at 2.0 eV continued to increase until the end of the experiment and became stronger than that at 1.65 eV. As a result, the spectral shape in this region was close to that of Na/Na-LTA(s) with increased Na adsorption under irradiation (sample (c) in Fig. 4). These behaviors mean that UV light irradiation first has the effect of generating metastable Na clusters in α-cages, and then has an additional effect similar to an increase in the Na adsorption amount, which leads to the partial formation of Na clusters that include multiple electrons in α-cages. The delayed growth of photoabsorption at 2.9 eV can be easily interpreted using a model in which the photogenerated Na clusters in α-cages (C₁ and C₂ bands) act as an intermediate for the generation of Na clusters in vacant β-cages that give D₁ and D₂ bands.

As described in the Introduction section, chromism is generally a reversible reaction. We next examined the reverse reaction after the termination of UV light for the colored Na/Na-LTA(m). The actual color changed from dark moss-green to green, and then yellow-green, and finally yellow, as shown in Fig. 10a. Fig. 10b shows the spectra of this reaction. The absorption at 1.4–2.3 eV decreased, maintaining its spectral shape and diminishing at ca. 1 day. In contrast, the D₁ and D₂ bands were quite stable. The D₂ band had a lifetime of ca. 10 days, but the lifetime of the D₁ band was impossible to determine. With the decrease of all these bands at 1–4 eV, a 2–3-fold increase in band W was observed. After these photoabsorption measurements, we could return the yellow sample to white within 1 h by heating at 400 K. We also confirmed that the reverse reaction was not promoted by light irradiation at a wavelength of 600 nm, which resonates with the C₁, C₂ and P bands. These results indicate that the reverse reaction occurs only thermally.

Fig. 10 (a) Photos and (b) absorption spectra of Na/Na-LTA(m) after terminating UV-light irradiation at RT. The time elapsed after cessation of light irradiation is shown below each image in (a). The spectrum of 0 h in (b) and that in the inset of Fig. 6 are identical. The upper and lower spectra in (b) correspond to 0–24 h and 1–10 d after termination, respectively.

The photochromic properties of this sample at 77 K were also examined. As shown in Fig. 11, the C₁ and C₂ bands appeared upon UV light irradiation as well as at RT with their spectral width reduced. The reduced width indicates the presence of electron–phonon interaction on these photogenerated Na clusters in α-cages as well as those in β-cages (Fig. S8†). These bands grew intensely, exceeding a value of 40 in the vertical axis without changing the spectral shape, indicating that the P band made no contribution. Obviously, the D₁ and D₂ bands could not be detected. These results indicate that the Na clusters with multiple electrons in α-cages and the Na clusters in β-cages do not form at 77 K, but they form thermally at RT. The heights of the thermal potential barriers for the formation of these two types of clusters are estimated to be about 20 meV, because the thermal energies at 77 and 300 K are k_BT = 6.6 and 26 meV, respectively.

Fig. 11 Dependence of the absorption spectra of Na/Na-LTA(m) at 77 K on the duration of monochromatic UV light (5.5 eV) irradiation. The vertical axis value of the right-side spectra is one-tenth that of the left-side spectra.

ESR study of Na-adsorbed Na-LTA(m)

As the photogenerated Na clusters in α-cages giving the C₁ and C₂ bands are the single species at 77 K, the photoinduced ESR signal at the low temperature can be attributed directly to these species. ESR spectra of Na/Na-LTA(m) (W band only) were measured at 80 K before and under UV light irradiation, as shown in Fig. 12. Before irradiation, only a single peak related to species W was detected. During irradiation, a photoinduced ESR signal was detected. At first glance, the spectrum of this irradiated sample resembles that in Fig. 5b, specifically, in that the major ten HFSs are roughly separated from each other by ca. 36 G. In addition, four very minor lines of the HFS might exist as depicted in Fig. S13.† Therefore, Na₄³⁺ clusters with the C_3v symmetry are formed at 80 K as well as at RT. The origin of the C₁ and C₂ bands appearing in the 77 K spectra is the Na₄³⁺ cluster in the α-cage.

Fig. 12 ESR spectra of Na/Na-LTA(m) measured at 80 K. Blue and red spectra were obtained before and under light irradiation, respectively.

The spectrum at 80 K, however, differs from that at RT in several respects. The ten HFSs do not indicate ideal relative intensities (Scheme S1†) with regular separation as observed at RT. Despite the use of powdery crystals of Na-LTA, the ESR spectrum at RT could be analyzed successfully by adopting the model of paramagnetic species in a solution. This means that the hyperfine splitting constants and g-values are isotropic at RT. In general, the isotropic properties on these constants can be realized by the thermal rotational motion of the paramagnetic species.⁴⁷ The Na₄³⁺ cluster with the C_3v symmetry in the α-cage might exist in a similar environment at RT. In the α-cage, there are eight equivalent locations, where the Na₄³⁺ cluster can exist. At 80 K, the Na₄³⁺ cluster is frozen to one location, resulting in the appearance of anisotropic features in the ESR spectrum. The suppression of the growth of the D₁, D₂ and P bands at 77 K is consistent with this model. Although the HFS splitting constant at 80 K is anisotropic, the g-value was isotropic, because the spectral shape has inversion symmetry at the spectral center of g ≈ 2.002.

Electronic structures and the mobilities of Na clusters

The aforementioned experimental results indicate that the electrons in the diamagnetic Na clusters or species W in β-cages transfer to the adjacent α-cages through their photoexcited state. Thermal transfers of the electrons among α-cages and from α-cages to β-cages are also apparent. The detailed electronic structures of the Na clusters in α- and β-cages, their stabilities, and the transfer mechanism among the cages are the points to be discussed.

In Na/Na-LTA(s) with dilute Na adsorption, Na clusters including two electrons in a spin-singlet state (diamagnetism) are formed in β-cages rather than those with one electron. If all the Na⁺ ions at the eight 6Rs of one β-cage participate in cluster formation (Fig. 1), the diamagnetic Na cluster in the β-cage can be expressed as Na₈⁶⁺. Because the electron–phonon interaction is observed as the spectral width dependence of the D₁ and D₂ bands on temperature (Fig. S8†), the polaron effect would be an appropriate concept to explain the stability of the diamagnetic Na clusters in β-cages. In general, the polaron effect is the local lattice deformation (displacement) screening the charge carrier (electrons or holes).⁴⁸ Lattice deformation deepens potential energy (ΔE) for the carrier. This effect has been introduced for the collective interpretation of the electronic states of the alkali–metal clusters in zeolites.²¹ The Na cluster in the β-cage with an electron (paramagnetic) (Scheme 1B1) and that with two electrons (diamagnetic) (Scheme 1B2) can be defined as a polaron and a bipolaron, respectively. The relative stability of the two types of polarons is determined by the two competing factors of the potential energy lowered by the polaron effect (ΔE_gain = ΔE_bipolaron − 2ΔE_polaron) and the increased energy (U) by the Coulomb repulsive force between the two electrons in the bipolaron.²¹ The actual generation of diamagnetic Na clusters (bipolaron) in β-cages indicates that ΔE_gain > U is satisfied.

Scheme 1 Illustrated models of the formation of the diamagnetic Na cluster in the β-cage (B1 and B2), the proposed transfer mechanism of a photo-excited electron in the β-cage (C1–C3, D1, and D2), and the aggregated electrons in the α-cage (E). Black closed circles indicate the initial Na⁺ positions of dehydrated Na-LTA in A.

The photoabsorption energies of the D₁ and D₂ bands were explained by the model of optical transition of the electron between the quantized levels of 1s and 1p in a spherical quantum well with infinite depth.^21,25 A detailed explanation of this model is provided in Section S6.† The 1s level is non-degenerate, while the 1p level has three-fold orbital degeneracy. These levels can be expressed as A_1g and T_1u, respectively, which are the Mulliken symbols in group theory. As the electrons are in the 1s level at the ground state, the optically allowed 1s–1p transition energy can be derived as 3.2–3.7 eV from Fig. 13a,²¹ assuming the diameter of the well to be equal to the inner diameter of the β-cage, 0.65–0.70 nm, as shown in Fig. 13b. Because the relative intensities of the D₁ and D₂ bands depend on the Na adsorption amount (Fig. 4), not only the above-mentioned Na₈⁶⁺ clusters but also diamagnetic Na clusters with different structures might be formed in the β-cages, although the details of their structure remain to be clarified. We will next consider the light irradiation-induced dissociation process of these diamagnetic Na clusters in β-cages. By photoabsorption, only one electron can be transferred from the diamagnetic Na cluster in the β-cage to an α-cage through the excited state. If the single residual electron in the β-cage—e.g., Na₈⁷⁺ (see Scheme 1C1)—has a long lifetime, it should be detected in the ESR spectrum, but the photoinduced ESR component could be resolved perfectly by the single model of the Na₄³⁺ clusters with C_3v symmetry (Fig. S11†). Thus, the ESR active one-electron state (polaron state) does not exist in the β-cage. The remaining electron should transfer spontaneously from the β-cage to the α-cage by thermal excitation (or through a photoexcited state by further irradiation), as illustrated in Scheme 1C2.

Fig. 13 (a) Dependence of the 1s–1p electronic transition energy on the diameter of the spherical quantum well with infinite depth.²¹ The energies of C₁, C₂, D₁, and D₂ bands, and the inner sizes of the α- and β-cages are shown for reference. Approximated energy potentials for (b) the diamagnetic Na cluster in the β-cage and (c) the photogenerated Na₄³⁺ cluster with C_3v symmetry in the α-cage. For (b) and (c), the diameter, d, indicates the inner size of the cages. Small black arrows indicate electrons and colored arrows indicate optical transitions between the 1s and 1p levels.

The electron transferred to the α-cage was distributed as the metastable Na₄³⁺ cluster with C_3v symmetry observed in the ESR spectra. The primary HFS coupling constant, A_Na1, for the three equivalent Na⁺ ions is close to the constant of Na₄³⁺ in the β-cages of Na-FAU but is small compared to the constant of an Na atom, 316 G.^27,49,50 Most of the electron wave function is distributed in the inner part of the main three Na⁺ ions (Na1) of the cluster. Regarding the HFS, a single peak was observed in the ESR spectrum of the K clusters in the α-cages of K-LTA due to the participation of many K⁺ ions with small HFS coupling constants.³⁴ Consequently, a single peak, whose profile was an envelope function of many HFSs, appeared in the ESR spectrum. A similar behavior can be seen in the ESR spectra of color centers in some alkali halides, e.g. NaCl, KI, RbBr, etc.⁵¹

We will next discuss the optical transitions of the metastable Na₄³⁺ clusters in α-cages. The energies of the C₁ and C₂ bands are the intermediate of the 1s–1p transition energies for an electron in the spherical quantum wells with the inner diameters of α- and β-cages (Fig. 13a). Since the size of the Na₄³⁺ cluster in the α-cage is obviously smaller than the effective size of the cage, we need to revise the spherical well model to apply. The electronic excited state of this cluster, 1p, must have a broadened confinement size compared to the 1s ground state, as illustrated in Fig. 13c, in order to decrease the 1s–1p transition energy. Namely, the electron wave function at the excited state spreads over some left-behind Na⁺ ions in the same α-cage which did not participate in the cluster at the electronic ground state. The appearance of two absorption bands of C₁ and C₂ can be explained by considering the symmetry of the Na cluster. For convenience, let us start from the spherical well model. When the symmetry of the potential well is lowered to C_3v, the 1s level of this cluster remains as A_1g, while the 1p levels in T_1u split into A_2u and E_u which are non-degenerate and doubly degenerate, respectively, as shown in Fig. 13c. Thus, the transitions from A_1g to A_2u and from A_1g to E_u have relative transition probabilities (oscillator strengths) of 1 : 2. This is consistent with the actual relative intensity of the two bands, C₁ : C₂ = 1 : 1.80 (Table S3†).

Considering the finding that the Na₄³⁺ clusters in the α-cages are mobile at RT, they can transfer thermally to equivalent locations of both intra and inter α-cages via the Na⁺ ions at 8Rs, as depicted in Scheme 1C3. The mobility of the Na₄³⁺ clusters is the origin of the unique secondary dynamical process observed in the Na/Na-LTA(m) sample. In this sample, species W in the β-cages supply electrons to the α-cages upon UV light irradiation, resulting in the formation of Na₄³⁺ clusters (Scheme 1D1 and D2). The secondary process of the formation of Na clusters in β-cages corresponds to a reverse reaction of Na/Na-LTA(s) photochromism. In the photochromism of Na/Na-LTA(m), the metastable Na₄³⁺ clusters in the α-cages serve as intermediates for the thermal formation of the Na clusters in the empty β-cages. Since the Na clusters in the β-cages are diamagnetic, two paramagnetic Na₄³⁺ clusters in the α-cages contribute to the formation of one diamagnetic Na cluster in the secondary process.

Finally, we consider the formation of Na clusters with multiple electrons in α-cages. Under UV light irradiation of Na/Na-LTA(m), the number of metastable Na₄³⁺ clusters increases with time. Several mobile Na₄³⁺ clusters can encounter each other and fuse together at the same α-cage as shown in Scheme 1E. The fused Na cluster including multiple electrons gives a surface plasmon-like transition observed as band P in Fig. 8a. The present adsorbed Na amount is lower than that required for forming Na clusters with multiple electrons without light irradiation as in, for example, sample (d) or (e) of Fig. 4. Thus, the Na clusters with multiple electrons generated forcedly by light irradiation are unstable and diminish with a lifetime of about one day after termination of UV light.

Conclusions

The unique photochromic phenomena observed in dehydrated Na-LTA are induced in the case of dilute adsorption of Na atoms. Depending on the dehydrating conditions, two types of species are formed in β-cages as initial nanostructures. Diamagnetic Na clusters with photoabsorption at ca. 3 eV (400 nm) are formed in β-cages without H₂O (Scheme 1B2). They dissociate into two paramagnetic Na₄³⁺ clusters with C_3v symmetry in α-cages upon light irradiation (Scheme 1C2). These clusters have two photoabsorption peaks at 1.6–2.1 eV (600–800 nm), reflecting their symmetry. In contrast, the reacted species between residual H₂O and adsorbed Na atoms, named W, has photoabsorption at 5.5 eV (230 nm) (Scheme 1D1). Under irradiation with UV light, the Na₄³⁺ clusters in α-cages are also generated (Scheme 1D2), and these are mobile at RT (Scheme 1C3). The mobility is the driving force for the formation of diamagnetic Na clusters in β-cages where H₂O is absent and Na clusters with multiple electrons in α-cages (Scheme 1E). The photochromic properties observed in the Na clusters of dehydrated Na-LTA are expected to be applied to optically responsive materials and solid-basic photocatalysts.

Author contributions

Tetsuya Kodaira: conceptualization, formal analysis, investigation, methodology, resources, writing – original draft, and writing – review and editing. Takuji Ikeda: formal analysis, investigation, and writing – review and editing.

Data availability

All the data that support the findings of this study are available within the main article and its ESI.†

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

T. K. gratefully thanks Mr T. Oh for his experimental assistance, Prof. Y. Nozue for his useful comments and for providing his handmade optical choppers, and Prof. T. Nakano for his useful information on alkali–metal clusters in zeolites.

References

1. J. Čejka, H. van Bekkum, A. Corma and F. Schüth, Introduction to Zeolite Science and Practice, Elsevier, Amsterdam, 3rd edn, 2007.
2. Y. Li and J. Yu, Nat. Rev. Mater., 2021, 6, 1156–1174.
3. E. Perez-Botella, S. Valencia and F. Rey, Chem. Rev., 2022, 122, 17647–17695.
4. N. Rangnekar, N. Mittal, B. Elyassi, J. Caro and M. Tsapatsis, Chem. Soc. Rev., 2015, 44, 7128–7154.
5. Y. Nozue, T. Kodaira and T. Goto, Phys. Rev. Lett., 1992, 68, 3789–3792.
6. F. Laeri, F. Schüth, U. Simon and M. Wark, Host-Guest-Systems Based on Nanoporous Crystals, Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim, 2003.
7. N. Herron, Y. Wang, M. M. Eddy, G. D. Stucky, D. E. Cox, K. Moller and T. Bein, J. Am. Chem. Soc., 1989, 111, 530–540.
8. G. D. Stucky and J. E. Mac Dougall, Science, 1990, 247, 669–678.
9. Z. K. Tang, Y. Nozue and T. Goto, J. Phys. Soc. Jpn., 1991, 60, 2090–2094.
10. T. Kodaira and T. Ikeda, Eur. Phys. J. D, 2003, 24, 299–302.
11. V. V. Poborchii, V. P. Petranovskii, I. A. Glukhov and A. A. Fotiadi, Microporous Mesoporous Mater., 2023, 348, 11295.
12. Z. K. Tang, L. Zhang, N. Wang, X. X. Zhang, G. H. Wen, G. D. Li, J. N. Wang, C. T. Chan and P. Sheng, Science, 2001, 292, 2462–2465.
13. S. Cox, T. Gier and G. Stucky, Chem. Mater., 1990, 2, 609–619.
14. L. Werner, J. Caro, G. Finger and J. Kornatowski, Zeolites, 1992, 12, 658–663.
15. Y. Nozue, T. Kodaira, S. Ohwashi, T. Goto and O. Terasaki, Phys. Rev. B: Condens. Matter Mater. Phys., 1993, 48, 12253–12261.
16. T. Nakano, Y. Ikemoto and Y. Nozue, Eur. Phys. J. D, 1999, 9, 505–508.
17. T. Nakano, Y. Ikemoto and Y. Nozue, Mol. Cryst. Liq. Cryst., 2000, 341, 461–466.
18. H. Kira, H. Tou, Y. Maniwa and Y. Murakami, AIP Conf. Proc., 2001, 590, 525–528.
19. T. Nakano, D. Kiniwa, F. L. Pratt, I. Wantanabe, Y. Ikemoto and Y. Nozue, Phys. Rev. B: Condens. Matter Mater. Phys., 2003, 326, 550–555.
20. T. Nakano and Y. Nozue, J. Comput. Methods Sci. Eng., 2007, 7, 443–462.
21. T. Nakano and Y. Nozue, Adv. Phys.: X, 2017, 2, 254–280.
22. Y. Nozue, T. Kodaira, S. Ohwashi, N. Togashi, T. Monji and O. Terasaki, Stud. Surf. Sci. Catal., 1994, 84, 837–842.
23. T. Kodaira, Y. Nozue, S. Ohwashi, N. Togashi and O. Terasaki, Surf. Rev. Lett., 1996, 3, 717–720.
24. T. Kodaira, Y. Nozue, S. Ohwashi, T. Goto and O. Terasaki, Phys. Rev. B: Condens. Matter Mater. Phys., 1993, 48, 12245–12252.
25. T. Kodaira, Y. Nozue and T. Goto, Mol. Cryst. Liq. Cryst., 1992, 218, 55–60.
26. J. A. Rabo and P. H. Kasai, Prog. Solid State Chem., 1975, 9, 1–19.
27. M. R. Harrison, P. P. Edwards, J. Klinowski, J. M. Thomas, D. C. Johnson and C. J. Page, J. Solid State Chem., 1984, 54, 330–341.
28. J. J. Pluth and J. V. Smith, J. Am. Chem. Soc., 1980, 102, 4704–4708.
29. J. M. Bennett, C. S. Blackwell and D. E. Cox, J. Phys. Chem., 1983, 87, 3783–3790.
30. T. Ikeda, F. Izumi, T. Kodaira and T. Kamiyama, Chem. Mater., 1998, 10, 3996–4004.
31. T. Oh, J.-S. Yu, T. Ikeda and T. Kodaira, Solid State Commun., 2002, 123, 387–390.
32. T. Ikeda, T. Kodaira, T. Oh and A. Nisawa, Microporous Mesoporous Mater., 2003, 57, 249–261.
33. T. Kodaira, Y. Murakami and S. Inoue, Phys. Rev. B: Condens. Matter Mater. Phys., 2005, 359–361, 1445–1447.
34. T. Kodaira and Y. Murakami, J. Chem. Phys., 2007, 127, 094704.
35. T. Kodaira, S. Inoue and Y. Murakami, Eur. Phys. J. D, 2007, 43, 267–270.
36. M. Irie, Chem. Rev., 2000, 100, 1683–1684.
37. A. K. Sasmal and T. Pal, J. Indian Chem. Soc., 2021, 98, 100073.
38. A. V. Shchegolkov, S.-H. Jang, A. V. Shchegolkov, Y. V. Rodionov, A. O. Sukhova and M. S. Lipkin, Nanomaterials, 2021, 11, 2376.
39. T. Yamamoto, A. Takahashi and H. Otsuka, Bull. Chem. Soc. Jpn., 2024, 97, uoad004.
40. L. D. Rollmann, E. W. Valyocsik and R. D. Shannon, in Inorganic Syntheses, ed. S. L. Holt Jr, Wiley, New York, 1984, vol. 22, pp. 61–68.
41. P. Kubelka and F. Munk, Z. Tech. Phys., 1931, 12, 593–609.
42. T. Sekiya, Y. Takeda, H. Takeda, S. Ohya and T. Kodaira, J. Phys. Soc. Jpn., 2012, 81, 124701.
43. A. Tashiro, S. Kanegawa, O. Sato and Y. Teki, Polyhedron, 2013, 66, 167–170.
44. D. L. Dexter, Phys. Rev., 1956, 101, 48–55.
45. N. M. Atherton and S. I. Weissman, J. Am. Chem. Soc., 1961, 83, 1330–1334.
46. M. Iwaizumi, M. Suzuki, T. Isobe and H. Azumi, Bull. Chem. Soc. Jpn., 1968, 41, 732–734.
47. C. Karunakaran and M. Balamurugan, in Spin Resonance Spectroscopy: Principles and Applications, ed. C. Karunakaran, Elsevier, Amsterdam, 2018, vol. 4, pp. 169–228.
48. C. Franchini, M. Reticcioli, M. Setvin and U. Diebold, Nat. Rev. Mater., 2021, 6, 560–586.
49. R. Catterall and P. P. Edwards, Adv. Mol. Relax. Interact. Processes, 1978, 13, 123–169.
50. M. Allegrini, E. Arimondo and L. A. Orozco, J. Phys. Chem. Ref. Data, 2022, 51, 043102.
51. H. Pick, in Optical properties of solids, ed. F. Abelès, North-Holland Publishing, Amsterdam, 1972, vol. 9, pp. 653–754.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d4nr04272g

---