Photoreduction of Ag+ by diethylaniline in colloidal

Microporous and Mesoporous Materials 194 (2014) 183–189

Contents lists available at ScienceDirect

Microporous and Mesoporous Materials journal homepage: www.elsevier.com/locate/micromeso

Photoreduction of Ag+ by diethylaniline in colloidal zeolite nanocrystals Florence Luchez a, Zakaria Tahri b, Vincent De Waele a,b, Ivan Yordanov c, Svetlana Mintova c, Alain Moissette a, Mehran Mostafavi b, Olivier Poizat a,⇑ a

Laboratoire de Spectrochimie Infrarouge et Raman (UMR 8516 de l’Université et du CNRS), Université Lille 1 Sciences et Technologies, Bât. C5, 59655 Villeneuve d’Ascq cedex, France Laboratoire de Chimie Physique (UMR 8000) CNRS-Université Paris Sud 11, 91405 Orsay, France c Laboratoire Catalyse & Spectrochimie, ENSICAEN – Université de Caen – CNRS, 6 bd Maréchal Juin, 14050 Caen cedex, France b

a r t i c l e

i n f o

Article history: Received 19 December 2013 Received in revised form 1 April 2014 Accepted 2 April 2014 Available online 13 April 2014 Keywords: Colloidal zeolite Zeolite nanocrystal Silver cation photoreduction Silver nucleation Transient absorption spectroscopy

a b s t r a c t The photoinduced formation and the subsequent early nucleation steps of silver atoms (Ag0) in nanosized zeolite Beta crystals stabilized in aqueous colloidal suspensions are studied by steady-state UV–Vis and Raman spectroscopy, and by transient absorption spectroscopy. The reduction of extra-framework silver cations is initiated by photoinduced electron transfer using organic electron donors such as N,N-diethylaniline or triphenylamine. The Ag0 species are formed in less than 100 ns and are found to be stable for more than one microsecond before beginning to aggregate, leading first to Ag+2 species in 1.1 ls. The data suggest that the reduction of extra-framework Ag+ arises only if the electron donor species is adsorbed on the zeolite particle surface or within the channels. Ó 2014 Elsevier Inc. All rights reserved.

1. Introduction Nanoclusters of silver atoms and ions are the object of considerable interest with regard to their exceptional redox and optical properties that give rise to various applications in catalysis, optoelectronics, trace analysis, chemical and biological sensoring [1–8]. These diverse properties are mainly due to the large surface area of the material, which dominates the contributions arising from the reduced-size volume of the material. Thus, for all applications, the sensitivity and efficiency of silver clusters increase with their surface/volume ratio. To limit the coalescence of silver clusters into larger particles, a variety of methods have been reported in the literature. Among them, stabilization of small clusters by protecting agents such as micelles [9], surfactants [10,11], vesicles [12], ligands [13–16], and polymers [17], or by deposition on silica or latex nanospheres to form composite silver-coated nanoparticles [5,18,19], has been demonstrated. Alternatively, the stabilization of small ionic silver clusters confined within zeolites has been proposed [1,2,20,21]. Indeed, the growth of Ag particles inside the cages and channels of zeolites is sterically limited by the channel size, thus preventing aggregation into large nanoparticles. The production of silver nanoclusters in zeolites is usually carried out on bulk materials by reduction of charge-compensating silver cations introduced into the zeolite by ion-exchange, also called ⇑ Corresponding author. Tel.: +33 3 20 43 40 85; fax: +33 3 20 43 67 55. E-mail address: [email protected] (O. Poizat). http://dx.doi.org/10.1016/j.micromeso.2014.04.008 1387-1811/Ó 2014 Elsevier Inc. All rights reserved.

extra-framework cations because they do not belong to the structural lattice of the zeolite but are located in the pores and channels. Chemical reduction of metal cations by reducing agents such as H2 or NaBH4 [22–24], thermal treatment under high temperature [2,25], or reduction by c-radiolysis [2,26–28] have been reported. Photoreduction of the extra-framework silver cations upon UV-irradiation has also been recently achieved [29]. Silver clusters were characterized mainly by EPR [26,27], UV–Vis absorption [22,24,27,28], emission [29], and time-resolved QXAFS [24]. A strong dependence of the final cluster size and shape on the zeolite framework and also on the water and metal content in the framework structure was found. A critical parameter controlling the cluster formation is found to be the diffusion of metal cations and small oligomers through the pores and channels. However, very few studies have addressed the question of the kinetics of cluster formation in zeolite matrices. In solution, numerous studies by time-resolved absorption spectroscopy have been devoted to the silver nucleation mechanisms induced by pulse and gamma radiolysis. Kinetic data and reference transient UV–Vis absorption spectra of small silver clusters and nanoparticles in aqueous solutions [30–32], alcohols [33–35], micromulsion [36], and supercritical ethane [37] are available. In zeolites, by analyzing the growth of the plasmon resonance absorption band at 410 nm, Cvjeticanin and Petranovic showed that the diffusion of Ag+ cations is the rate-determining step of the aggregation process, which takes place with a time constant of several seconds that depends on the zeolite content [28].

184

F. Luchez et al. / Microporous and Mesoporous Materials 194 (2014) 183–189

Recently, Shimizu and coworkers studied the formation of silver clusters in Ag-MFI materials upon H2 reduction and their dispersion upon reoxidation by following the structural change of Ag [24]. In these two studies, the kinetics was investigated on the second to hour time scale, and only the last stages of the aggregation process were observed. To our knowledge, no time-resolved study of the nucleation process of metal clusters in zeolite matrix with a time-resolution capable of measuring the earliest stages of the nucleation process has been reported up to now. All the work consecrated so far to the elaboration and characterization of silver clusters confined within zeolites was performed using bulk micron-sized crystals of zeolite in powder forms or in water suspensions. A new approach for the stabilization of metal nanoparticles in zeolite has been recently proposed based on the use of nanosized crystals prepared from colloidal precursor solutions under mild hydrothermal synthesis conditions [38–39]. The colloidal suspensions were shown to contain zeolite nanoparticles stable over months and monodisperse with sizes typically in the range 50–100 nm. The large surface/volume ratio of such particles ensures high sensitivity for applications based on chemical interactions such as chemical sensing. On the other hand, from the practical point of view, a manifest advantage of these colloidal suspensions is their optical transparency, which offers a unique possibility to perform transient absorption spectroscopy by directly measuring the transmission of the sample [40–42]. This experimental configuration is more convenient than the transient diffuse reflectance configuration generally used for studying the photochemistry in zeolites. In this paper we report on the early-step nucleation mechanisms of silver confined in zeolite Beta nanocrystals (BEA type structure) stabilized in aqueous colloidal suspensions. The nucleation process follows the photoinduced reduction of silver cations previously introduced into the zeolite framework by ion-exchange. Photoreduction of Ag+ is performed via the pulse laser excitation of an electron donor species introduced as solute into the colloidal suspension of zeolite nanocrystals. The photoinduced processes are followed in real-time by transient UV–Vis absorption spectroscopy using a laser flash photolysis experimental setup. Two different electron donor species, N,N-diethylaniline (DEA) and triphenylamine (TPA), of similar redox properties but different morphology, were considered. We observe that the initial reduction of Ag+ into Ag0 occurs only if the electron donor shows specific interaction with the zeolite nanocrystals. The silver atoms are found to be stable for a few microseconds in the BEA type framework structure before they start aggregating.

2. Experimental 2.1. Preparation of the zeolite suspensions doped with silver cations Colloidal zeolite Beta was synthesized according to a procedure previously reported [43] from a clear precursor suspension of freshly freeze-dried colloidal silica (Aldrich, SM-30, 30 wt.% in water), aluminum isobutoxide (Aldrich) and tetraethyl ammonium hydroxide (TEAOH, Aldrich), of overall composition: 64.5TEAOH: 1.4Al2O3: 100SiO2: 1500H2O. Initially, the aluminum isobutoxide was dissolved in a solution of TEAOH in double-distilled water, and then the silica source was added to the clear suspension and stirred for additional 60 min. The precursor suspension was hydrothermally treated at 100 °C for 5 days, leading to a BEA sample denoted as BEA-1. The crystalline materials were purified in two steps consisting of high-speed centrifugation at 24,500 rpm for 120 min and removal of the mother liquor, and then redispersion in double-distilled water using an ultrasonic bath. The purified BEA-1 zeolite suspension was subjected to ion exchange with

AgClO4 overnight, and purified twice by centrifugation (10,000 rpm for 10 min). The corresponding Ag+ doped zeolite sample, denoted as Ag-BEA-1, has the following general chemical composition: [(Ag+)x(TEA+)y](Alx+ySi64xyO128), where TEA+ is the tetraethylammonium cation. The chemical composition of AgBEA-1 is determined by inductively coupled plasma atomic emission spectroscopy (ICP-AES, VARIAN-VISTA); Si/Al = 15 and Si/ Ag = 76. The crystallinity and purity of this compound were determined by X-ray powder diffraction using a STOE Study diffractometer in Debye–Scherrer geometry with CuKa radiation. The X-ray diffraction pattern contains all Bragg reflections expected for the BEA type structure and confirms that the sample is crystalline [44]. The particle size distribution curve of sample Ag-BEA-1 contains one sharp peak centered at 70 nm [44]. Additional information for the crystal size and morphology of the samples was obtained by scanning electron microscopy (SEM) using a Phillips XL 30 microscope. 2.2. Amine photoreductors N,N-diethylaniline (DEA, P99%, Aldrich) and triphenylamine (TPA, 99%, Acros Organics) were used as received. Being not soluble in pure water, these amines could not be directly added to the AgBEA-1 colloidal solution. Instead, solutions of DEA or TPA in ethanol were added to the colloidal solution under argon. The overall concentrations of Ag+ and amine were 1.2 103 and 1.0 103 mol L1, respectively, in the final solutions. The S0 ? S1 extinction coefficient (e) of DEA and TPA being about 103 M1 cm1 at 266 nm, an optical density (OD) value of 1.0 is obtained for both solutions at this wavelength, which is adequate for laser flash photolysis measurements using a pump excitation at 266 nm. 2.3. Spectroscopic experiments Steady-state UV–Visible-NIR absorption spectra were recorded at room temperature between 200 and 1800 nm using a Cary 6000i spectrometer. Colloidal solutions were studied in quartzsuprasil cells. Raman spectra were obtained with 1064 nm excitation on a Bruker RFS 100/S Fourier Transform Raman spectrometer (4 cm1 spectral resolution). Nano-microsecond transient absorption experiments were performed using a conventional laser flash photolysis setup. Excitation pulses at 266 nm (7–8 ns, 1 mJ) were provided by a 20-Hz Nd:YAG laser (DIVA II, Thales laser). The probe light was provided by a Xe flash lamp (XBO 150 W/CR OFR, OSRAM). Samples (OD 1.0 at 266 nm) were placed in a quartz cell (10 10 mm2 section) and deaerated by bubbling N2 apart from those used for measuring the effect of quenching by O2. The transmitted light was analyzed with a photomultiplier (R1477-06; Hamamatsu) coupled to a digitalized oscilloscope (TDS 540, Tektronix). Kinetic measurements at each wavelength were accumulated over 9 laser shots. 3. Results The BEA type framework structure displays two types of intersecting channels with aperture sizes of 0.55 0.55 and 0.76 0.64 nm2. To examine the possibility of photoreducing the extra-framework silver cations in the Ag-exchanged Beta nanocrystals and to investigate the processes of intrazeolite silver cluster formation, two electron donor amines, DEA and TPA, were used. Whereas these amines have similar low gas phase ionization potential (6.98 and 6.75 eV, respectively), which denotes comparable electron donor properties, they differ notably in shape and size. In principle, the rod-shape DEA amine (0.47 nm large, 0.86 nm long) is small enough to enter and diffuse within the channels of


zeolite Beta whereas the spherical and bulky TPA molecule (9.9 10.2 nm2) cannot penetrate into the intrazeolite volume. Another difference between the two amines concerns the molecular shape. Due to the propeller-like trigonal structure of TPA, the N atom is notably more sterically hindered and the lone electron pair less accessible than in DEA. According to the redox properties of the isolated Ag+ cation (Ered = 1.8 V [15,16,45]) and DEA/TPA amines (Eox = +0.76/0.86 V), spontaneous reduction of Ag+ by either DEA or TPA is thermodynamically unlikely (DG +2.6 eV). However, if the amine is photoexcited in the lowest excited singlet (S1) state (DE00 4 eV) or triplet state (DE00 2.7 eV), the electron transfer becomes favorable (DG < 0 eV) and thus Ag+ is expected to be photoreduced. The results of laser flash photolysis experiments in the presence of DEA as electron donor will be considered first. For a rigorous interpretation of the data and particularly a clear identification of the photoinduced processes involving the extra-framework Ag+ cations, preliminary investigations of the photoreactivity of the precursor reactants alone were carried out in similar experimental conditions (concentration, solvent mixture composition, pump excitation wavelength and intensity). On one hand, pulse excitation at 266 nm of a pure colloidal solution of Ag-BEA-1 in N2-purged water/ethanol 99:1, in the absence of electron donor amine, did not lead to any transient absorption signal, at any time delay, in the 280–700 nm probed spectral window. This is consistent with the fact that the zeolite material does not absorb at 266 nm. On the other hand, transient absorption spectra of the amine DEA (103 M) were recorded at different time delay after excitation at 266 nm in ethanol (Fig. S1) and in mixtures of water/ethanol 99:1 (Fig. 1) and water/ethanol 75:25 (Fig. S2) in the absence of zeolite. In all cases, the spectra are characterized

Fig. 1. Laser flash photolysis spectra obtained after 266 nm excitation of DEA (103 M) in N2-purged (top) and aerated (bottom) water/ethanol 99:1 solution; (A) spectral evolution from 0.30 to 35 ls; (B) pure triplet state spectrum extracted after subtracting 2 ls trace from 0.30 ls trace.

185

by the superimposition of three spectral features. The former comprises two broad absorption bands at 340 and 460 nm that decay in a few hundred nanoseconds in N2-purged solutions but much faster in aerated solutions. These bands, very weak in water but much stronger in ethanol, correspond likely to the triplet state T1 spectrum of DEA [46]. The second transient feature is an absorption band also appearing at around 460 nm but displaying a much sharper and narrower shape than the triplet component present at the same position. Its lifetime is a few tens of microseconds and is not sensitive to the presence of air in the solution. This band is unambiguously assigned to the radical cation DEA+, which is known to occur from efficient monophotonic ionization in the lowest excited singlet state S1 (p-p⁄) in polar solvents [46–48]. The third transient feature is a broad and structureless absorption covering the whole visible domain and peaking in the red above 700 nm. Such absorption signal has the characteristic shape of the solvated electron, e solv [49] produced in the photoionization process. As for the T1 state spectrum, the decay of the e solv band is much faster in the presence of air. In N2-purged water/ethanol 99:1 solution, the decay time ses amounts to 280 ± 20 ns. The intensity of the DEA+ and e solv spectral contributions relative to that of the T1 state spectrum increases with the amount of water in the solution, which is consistent with the increase in polarity of the solution (Figs. S1 and S2). In pure ethanol (Fig. S1A), the early time spectra correspond essentially to the T1 state species, which reveals the predominance of the intersystem crossing process over photoionization in this moderately polar solvent. At time delay longer than 500 ns, the T1 state has disappeared and the residual spectrum characterizes the presence of the DEA+ species formed to a minor extent. In contrast, in water/ethanol 99:1 (Fig. 1), photoionization is the dominant process and the T1 state spectrum is perceptible only after subtraction of the radical cation spectrum (Fig. 1B). Note that, in this solvent mixture where the DEA+ species is produced at high concentration, a clear evolution of the DEA+ band shape is observed up to 35 ls, characterized by the decrease of the 460 nm band and the appearance of a more structured shape. The final spectrum at 35 ls is similar to that reported for the radical cation of N,N,N0 N0 -tetraethylbenzidine, TEB+ [50], and keeps a constant intensity for hundreds of ls. It is, indeed, known that the DEA radical cation is highly reactive and decays via a second-order process leading to the dimeric radical cation TEB+ [51,52]. Finally, the excitation at 266 nm of DEA (103 M) in the colloidal solution of pure zeolite Beta (BEA-1) led to strictly similar results as those found in the absence of zeolites, i.e., the photophysics of DEA is unchanged. Similar laser flash photolysis measurements with pulse excitation at 266 nm were done for Ag-BEA-1 water/ethanol 99:1 colloidal solution containing the amine DEA (103 M). Before analyzing the transient absorption spectra, it is worth noting that, during these measurements, a yellow coloration of the sample was rapidly appearing upon irradiation. Indeed, steady-state absorption spectra recorded at the end of the measurements revealed the existence of a new absorption band in the range 400–500 nm, maximizing at 410 nm and increasing with the irradiation time. It corresponds to the typical surface plasmon resonance band of silver and reveals the formation of silver nanoparticles by reduction of extra-framework Ag+ cations [24,28]. The plasmon band keeps a well-defined shape over a period of several weeks, indicating that the silver nanoparticles are stabilized in the colloidal BEA zeolite. Figs. 2 and 3 show typical series of transient absorption spectra recorded in the 0–35 ls time delay range following 266 nm excitation for aerated and N2-purged solutions, respectively. The kinetics followed at key wavelengths and their best fit using exponential functions are displayed in Fig. 4. In the presence of air (Fig. 2), the spectra and their time evolution are fairly similar to those observed for DEA in 99:1 water/ethanol mixture in the absence of zeolite (Fig. 1), showing mainly the DEA+ radical cation

186


Fig. 2. Laser flash photolysis spectra of DEA (103 M) in aerated water/ethanol 99:1 Ag-BEA-1 colloidal solution obtained from 0.3 to 35 ls after 266 nm excitation.

Fig. 3. (A) Laser flash photolysis spectra of DEA (103 M) in N2-purged water/ ethanol 99:1 colloidal solution of Ag-BEA-1 obtained from 0.16 to 3.5 ls (top) and 3.5 to 35 ls (bottom) after 266 nm excitation; (B) triplet state spectrum extracted after subtracting 400 ns trace from 160 ns trace (the intensity is multiplied by 3).

band at 460 nm and the broad absorption of the solvated electron extending in the red region. When the solution is deaerated (Fig. 3), we still observe the DEA+ 460 nm band and e solv red absorption. However the relative intensity of the latter is significantly weaker than in the absence of zeolite (Fig. 1, top). The 460 nm absorption is quite broad at very short delay time and narrows somewhat from 160 to 550 ns, thus suggesting the presence in the 160 ns spectrum of the triplet state T1 of DEA in addition to the DEA+ contribution. As a confirmation, the T1 state spectrum can be identified after subtraction of the DEA+ and e solv spectral contributions from the 160 ns spectrum (Fig. 3B). However, the most interesting spectral feature in the 160 ns spectrum is a new intense transient absorption band at around 355 nm, which is not observed in the aerated solution. This band is located at roughly the same position as the UV band of the T1 state spectrum of DEA (kmax 340 nm) but is clearly much more intense than the T1 state contribution displayed in Fig. 3B. In addition, as noted above, the T1 contribution to the 460 nm has disappeared at 550 ns while the 355 nm band has dropped only slightly. The intense absorption peaking at 355 nm in the 160 ns

spectrum is thus characterizing a distinct transient species. This conclusion is further confirmed by comparing the kinetics at different probe wavelengths. The 355 nm band rises slightly up to 500 ns, and then disappears within 3 ls (Fig. 4C). It leads to a new transient band peaking at 300–310 nm, close to the high energy edge of the investigated spectral window, well distinguishable in the 3.5 ls spectrum (Fig. 3). The growth of this new band is hardly perceptible, as it is almost superimposed on the former band at 350 nm, which is notably more intense. Nevertheless, a slight increase in absorption is discernible at 310 nm in a short period of time, i.e., t 6 1 ls (Fig. 4D). Then the 310 nm band decreases up to 35 ls and also shifts to higher energy, which indicates that it is probably replaced by another transient absorption lying more in the UV region, outside of the investigated spectral window. No further spectral evolution is perceptible after 35 ls up to 500 ls, which corresponds to the full time domain probed in this study. The time dependence of the OD has been fitted at all wavelengths using sums of exponential functions. Three spectral regions can be distinguished. At wavelengths longer than 550 nm, where only the e solv species contributes to the transient absorption, all decays can be fitted with a single exponential kinetics of time constant ses = 260 ± 20 ns (150 ± 20 ns for the aerated sample), as illustrated for example by the 700 nm kinetic trace (Fig. 4A). This value is comparable to the 280 ± 20 ns found above in the same solvent in the absence of Ag-BEA-1 nanocrystals. In the region 400–500 nm, the observed time dependence of the transient signal is more complex (Fig. 4B). The band at 460 nm, due to both the radical cation DEA+ and triplet state T1 of DEA, is superimposed on the blue part of the broad e solv absorption. A first decay component of time constant 270 ± 20 ns, close to time ses , corresponds likely to the disappearance of the broad background signal due to the e solv species. During this decay, as seen in Fig. 3A, the band at 460 nm is essentially translated downwards but does not significantly change in intensity. In the same time, as discussed above, the minor contribution of the T1 state species to the band at 460 nm is also fading, with a decay kinetics that cannot be differentiated from that of the e solv decay. At longer times (Fig. 3B), a slight decay of the residual DEA+ absorption can be fitted with a second order kinetics, which might point to the existence of a process of dimerization of the radical cation, as observed in solution. Such a dimerization process is indeed definitely evidenced from the resonance Raman spectrum recorded after irradiation (Fig. 5A). This spectrum shows bands at 1599, 1438, 1342, 1228, 1161, 412 and 217 cm1, which are characteristic of the TEB+ cation radical [53]. As a confirmation, a similar Raman spectrum is recorded upon irradiation of a mixture of TEB and AgBEA (Fig. 5B). At longer time delays, the TEB+ spectrum appears stable for milliseconds. Finally, below 400 nm, the spectral evolution concerns the still unassigned 355 and 310 nm absorption bands and shows the presence of at least four kinetic contributions of time constants s1–s4. At 355 nm (Fig. 4C), the kinetics is characterized by a fast, hardly measurable rise of time constant s1 6 80 ns followed by a much slower two-exponential decay of time constants s2 = 1.1 ± 0.3 ls (75%) and s3 = 11.4 ± 0.5 ls (25%), respectively. At 310 nm (Fig. 4D), a rise time of 0.9 ± 0.3 ls is clearly identified, which correlates within the experimental accuracy to the main decay component s2 found for the band at 355 nm. This rise is followed by a two-exponential decay with time constants of 11 ± 0.5 ls (65%), similar to the decay time s3 found at 355 nm, and s4 = 1.5 ± 0.3 ls (35%), respectively. The 11.4 ls contribution to the 355 nm decay is likely due to the overlapped band-foot of the 310 nm absorption. The fact that two time constants s3 and s4 are involved in the decay kinetics at 310 nm indicates that at least two different transient species with somewhat similar spectra are contributing to the absorption in this region. The simplest reaction scheme that can account for all these kinetic features characterizing the


187

Fig. 4. Kinetic traces of the transient absorption spectra presented in Fig. 3 and the best fits (red line) for four characteristic wavelengths: (A) 700 nm: decay time 260 ± 20 ns; (B) 460 nm: decay time 270 ± 20 ns (80%) + second-order component (20%); (C) 355 nm: rise time 680 ns, decay times 1.1 ± 0.3 ls (75%) and 11.4 ± 0.5 ls (23%); (D) 310 nm: rise time of 0.9 ± 0.3 ls, decay times 1.5 ± 0.3 ls (35%) and 11 ± 0.5 ls (65%). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Fig. 5. FT Raman spectra of irradiated N2-purged Ag-BEA-1 sample containing DEA (A) and TEB (B); probe wavelength 1064 nm.

spectral region of 300–400 nm requires the contribution of five chemical intermediates (A–E) involved in four consecutive steps according to Scheme 1. Note that none of the time constants s1–s4 corresponds to the decay times found for the esolv (time ses ) or DEA+ species, indicating that the chemical intermediates absorbing in the spectral region 300–400 nm evolve independently. Before discussing the assignment of species (A–E) and the nature of the four chemical processes by which they are related, let us examine the case of the TPA photoreductor. The photophysics and photochemistry of TPA in solution has been well studied by flash photolysis experiments [54,55]. In the absence of any quencher, the T1 state of TPA undergoes intramolecular electrocyclization, yielding the N-phenyldihydrocarbazole (DHC) T1 state that deactivates to its metastable ground state with a time constant of 0.4 ls. The later species is characterized by a strong absorption band at

610 nm. The transient absorption spectra recorded following 266 nm excitation of TPA (103 M) in N2-purged ethanol in the absence of zeolite particles (Fig. S3) are consistent with this reaction scheme. In fact, it shows the appearance of the typical DHC absorption at 600–610 nm from the initial triplet species (430 nm) with a risetime of 0.5 ± 0.04 ls. Unlike the case with DEA, the time evolution of transient absorption spectra following 266 nm excitation of TPA (103 M) in the Ag-BEA-1 water/ethanol suspension (Fig. S4) appears strictly similar to that observed in pure solvent, with a DHC risetime of 0.44 ± 0.04 ls. This analogy reveals that, in contrast to what was found with DEA, no new photoinduced process is arising upon excitation of TPA in the Ag-BEA-1 suspension. This observation is corroborated by the fact that no yellow coloration of the colloidal solution was perceived during irradiation. Additionally, the steady-state absorption measurements did not reveal the presence of any absorption band growing with the irradiation time. 4. Discussion The results presented above reveal that light irradiation of the amine DEA dissolved in a colloidal solution of silver-exchanged zeolite nanoparticles Ag-BEA-1 leads to the formation of Ag clusters by photoreduction of the extra-framework Ag+ ions. The clusters are characterized from steady-state absorption measurements by the typical surface plasmon band of silver at 410 nm. The large bandwidth of this band and the absence of scattering tails in the red part of the spectrum are indicative of cluster size of few nanometers whose optical absorption corresponds purely to the dipolar response. On the other hand, laser flash photolysis measurements of Ag-BEA-1 colloidal solutions containing DEA show the existence

Scheme 1. Four consecutive step reaction scheme characterizing the spectral evolution observed following pulsed laser irradiation of Ag-BEA-1 colloids.

188


in the nano-microsecond time domain of transient absorption bands in the blue region of the spectra (k < 400 nm), which are not observed in the case of Ag+ -free zeolite BEA-1 colloidal solutions. It is thus logical to link these transient blue bands to the presence of silver in the zeolite framework and to assign them to the early reduced silver precursors of the long-lived clusters absorbing at 410 nm. The time dependence of these transient absorption features in the blue domain reveals the appearance in less than 100 ns (s1) of a first strong band at 355 nm that leads in 1.1 ls (s2) to a new band at about 310 nm, itself evolving in 1.5 ls (s4) to a close-lying but slightly less intense band, which finally disappears in 11 ls (s3) and leads to transient bands below 300 nm (Scheme 1). By analogy with previous investigations of the formation and nucleation mechanisms of silver clusters induced in aqueous or alcoholic solutions by pulse and gamma radiolysis [28–34], the absorption at 355 nm (Scheme 1, species B) can be ascribed with confidence to the monoatomic-reduced silver, Ag0, and the subsequent absorption at 310 nm (Scheme 1, species C) to the charged dimer Ag+2 formed by further reaction of Ag0 with a neighboring Ag+ cation. Still by analogy with the reactivity in solution, the second species absorbing around 310 nm (Scheme 1, species D) is possibly the doubly charged Ag2+ 3 cluster produced by reaction between Ag+2 and Ag+ [31–33]. The absorption of Ag2+ 3 is indeed reported to be hardly distinguishable from that of Ag+2 in the region above 300 nm, and also to differ from it only by a sharper and narrower band shape on the red edge [32], as observed here. The next step in the spectral evolution is found to lead to new species absorbing below 300 nm (Scheme 1, species E), which is consistent with the absorption spectra reported for silver aggregates larger than Ag2+ 3 in solution [30–36]. For example, 0 + Ag2+ 4 , which can arise from dimerization of Ag2, capture of Ag by Ag+3, or disproportionation of Ag2+ 3 [31–33], is characterized by an absorption band with a maximum at around 270 nm [32]. Time s3, which characterizes the disappearance of all transient signals in the region 310 nm, is thus probably a mean value for various processes. In summary, the first steps of the intrazeolite silver nucleation process are quite comparable to those in solution and can be approximated by the series of chemical processes listed in Scheme 2. Note that the absence of any noticeable spectral evolution in the 35–500 ls time domain, in particular the non-observation of the emergence of the silver plasmon band, is consistent with the results of previous kinetic studies in aqueous solution. The decay of the Ag2+ 4 species is indeed reported to occur only in the millisecond time domain and the early buildup of the silver plasmon absorption takes place in tens to hundreds of milliseconds [31]. The oxidation power of the ionic clusters produced in the first stages of the aggregation depends strongly on the cluster size. Whereas silver atoms are strongly reducing in aqueous solution, with an Ag+/Ag0 redox potential value of 1.8 V [45], the oxidation potential of Ag+ increases rapidly with increasing cluster size [31,45], becoming positive from Ag2+ 4 [31] and reaching the bulk metal value of +0.8 V in larger clusters. Therefore the final silver clusters are likely present in the form of neutral, metallic Agn species.

Scheme 2. First chemical steps of the silver nucleation process following Ag+ photoreduction in Ag-BEA-1 colloids.

It is interesting to note that the values of s2, s3, and s4 are at least 10 times larger than the values reported for similar reactions in bulk water [32]. This means that these diffusion-controlled reactions have slowed down in the zeolite. However, evaluating from a quantitative point of view the rates of transport diffusion through the zeolite channels is a complex question. The mechanisms by which molecules flow through nanoporous materials are currently not well understood. Several factors can influence the intrazeolite diffusion dynamics with relative extents hardly assessable. In principle, the confinement and one-dimensional aspect of the channels are supposed to favor high diffusion rates through the so-called ‘‘single-file diffusion mechanism’’ [56,57], which has been predicted theoretically and observed experimentally for small molecules in zeolites [58–60] but is still far from being well understood. Most simulations involve hard sphere fluids in a model cylindrical pore with hard walls. In realistic microporous materials with atomically detailed walls, the rate of diffusion will be limited by frictional effects depending on variables such as the host lattice internal topology (pore diameter, tortuosity, flexibility, presence of structural defects), the guest entities structure, the existence of host–guest electrostatic interactions. . . Other factors such as the possible inhomogeneous distribution of the photogenerated Ag0 population or the obstruction of the channels by the bulky DEA molecules may also affect the intrazeolite diffusivity. A key question to be addressed relates to the nature of the reduction process that leads to the formation of early Ag0 species. An important point to note is that the kinetics of appearance of Ag0 (rise component at 355 nm, time s1 6 80 ns) is notably faster than the decay of the solvated electron band (700 nm absorption, time ses = 260 ns). This disparity indicates that the esolv population associated with the transient absorption in the red region cannot be considered as responsible for the reduction of the intrazeolite Ag+ cations. The ses time being similar to that measured for the e solv produced upon photoionization of DEA in the absence of Ag-BEA-1, it probably also characterizes electrons produced in the solvent bulk and decaying before having time to diffuse to the Ag-BEA-1 nanocrystals. It suggests that two types of DEA molecules must be distinguished. A first group consists of those free DEA molecules located in the solvent bulk, producing e solv upon photoionization but unable to photoreduce the extra-framework Ag+. Besides, a second population of DEA molecules must be lying in direct interaction with the zeolite particles to account for the observed fast photoreduction of the extra-framework Ag+ ions. On the basis of the intensity of the Ag0, + e solv, and DEA absorption bands observed in the shortest time spectrum (160 ns) and given the molar extinction coefficient of these 5 species, concentration values of [Ag0] = 2 105 M, [e M, solv] = 10 and [DEA+] = 3 105 M can be estimated. The total amount of DEA+ corresponds nicely to the sum of the amounts of Ag0 and e solv, as expected in the assumption that DEA is involved in two independent photoinduced processes, DEA + hm ? DEA+ + e and solv DEA + Ag+ + hm ? DEA+ + Ag0. The fact that, in contrast to the DEA case, no silver photoreduction is observed when using the amine TPA as electron donor could be interpreted as due to a size restriction effect on the amine reactivity, suggesting that the silver reduction reaction is an intrazeolite process. Indeed, DEA alone can in principle penetrate the BEA channels and could, upon excitation to the S1 or T1 state, directly reduce the Ag+ species lying in its vicinity within the channels. The possibility that charge-compensating Ag+ ions could migrate towards the zeolite nanoparticle surface and diffuse outside to meet the amine species is not credible because it would result in local rupture of the charge neutrality in the zeolite framework. However, though the dynamics of intrazeolite molecular diffusion remains a complex and debated issue, it is generally recognized that aromatic molecules have difficulty to enter zeolite channels when they are filled with water. An alternative hypothesis would be that the amine molecules remain adsorbed at the surface


of the zeolite nanocrystals and that only DEA, less bulky than TPA, can have enough electronic interaction with the zeolite lattice to allow fast reduction of the intrazeolitic Ag+ cations. In any case, the important point is that, whatever its nature, a specific interaction exists between part of the DEA population and the Ag-BEA-1 particles so that photoreduction of intrazeolitic Ag+ by DEA occurs. This reaction may be either the direct photoreduction of Ag+ by the excited S1 and/or T1 states of DEA (Ag+ + S1/T1 DEA ? Ag0 + DEA+), in which case the electron acceptor and donor species must be in contact within the zeolite porous volume, or a two-step process involving first the photoionization of adsorbed DEA and production of electrons within the zeolite channels + (S1/T1 DEA ? e intrazeolite + DEA ), and the subsequent intrazeolite + + reduction of Ag (Ag + eintrazeolite ? Ag0). This two-step mechanism is the only possible in the hypothesis that DEA is adsorbed at the surface of the zeolite nanocrystals. In this case, the short diffusion distance to be covered by the intrazeolite electrons to reach the extra-framework Ag+ could account for the fast reduction rate of this process (s1 6 80 ns). 5. Conclusions The early-step nucleation mechanisms of silver confined in zeolite Beta nanocrystals stabilized in colloidal suspensions, followed in real-time by transient UV–Vis absorption spectroscopy, is reported. The photoreduction of silver cations in zeolite Beta via the light excitation of an electron donor species is clearly demonstrated with N,N-diethylaniline as donor but not with the bulky triphenylamine. The formation of Ag0 occurs in less than 100 ns and is found to be stable for more than one microsecond before beginning to aggregate, leading first to Ag+2 species in 1.1 ls. The data suggest that the reduction of extra-framework Ag+ arises only if the electron donor species is adsorbed on the zeolite particle surface or within the channels. Acknowledgment The authors gratefully acknowledge funding from the ANR through the Blanc SIMI 8 program (Project ‘‘TAR-G-ED’’, Contract ANR-13-BS08-0002-01). Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.micromeso. 2014.04.008. References [1] [2] [3] [4] [5] [6] [7] [8]

G.A. Ozin, A. Kuperman, A. Stein, Angew. Chem. Int. Ed. 28 (1989) 359–376. T. Sun, K. Seff, Chem. Rev. 94 (1994) 857–870. G.M. Koretsky, M.B. Knickkelbein, J. Chem. Phys. 107 (1997) 10555–10566. N. Aoyama, K. Yoshida, A. Abe, T. Miyadera, Catal. Lett. 43 (1997) 249–253. T.-H. Lee, R.M. Dickson, Proc. Natl. Acad. Sci. USA (2003) 3043–3046. M. Muniz-Miranda, J. Raman Spectrosc. 35 (2004) 839–842. K. Himizu, J. Shibata, A. Satsuma, J. Catal. 239 (2006) 402–409. T. Baba, Y. Iwase, K. Inazu, D. Masih, A. Matsumoto, Microporous Mesoporous Mater. 101 (2007) 142–147. [9] A. Henglein, J. Phys. Chem. 83 (1979) 2209–2216. [10] R. Humphry-Baker, M. Grätzel, P. Tundo, E. Pelizetti, Angew. Chem. Int. Ed. 18 (1979) 630–631. [11] J.H. Fendler, Chem. Rev. 87 (1987) 877–899.

189

[12] K. Monserrat, M. Grätzel, P. Tundo, J. Am. Chem. Soc. 102 (1980) 5527–5529. [13] G. Schmid, Chem. Rev. 92 (1992) 1709–1727. [14] X. Li, J. Zhang, W. Xu, H. Jia, X. Wang, B. Yang, B. Zhao, B. Li, Y. Ozaki, Langmuir 19 (2003) 4285–4290. [15] S. Remita, M. Mostafavi, M.O. Delcourt, J. Phys. Chem. 100 (1996) 10187– 10193. [16] I. Texier, S. Remita, P. Archirel, M. Mostafavi, J. Phys. Chem. 100 (1996) 12472– 12476. [17] L. Longenberger, G. Mills, J. Phys. Chem. 99 (1995) 475–478. [18] M. Litorja, C.L. Haynes, A.J. Haes, T.R. Jensen, R.P. Van Duyne, J. Phys. Chem. B 105 (2001) 6907–6915. [19] E.I. Isaeva, V.V. Gorbunova, T.B. Boitsova, M.P. Sukontseva, A.Y. Men’shikova, Y.O. Skurkis, Russ. J. Gen. Chem. 75 (2005) 1340–1345. [20] P.A. Jacobs, J.B. Uytterhoeven, J. Chem. Soc. Faraday Trans. 1 (75) (1979) 56–64. [21] G.A. Ozin, F.J. Hugues, Phys. Chem. 87 (1983) 94–97. [22] R. Kellerman, J. Texter, J. Chem. Phys. 70 (1979) 1562–1563. [23] V.S. Gurin, N.E. Bogdanchikova, V.P. Petranovskii, Mater. Sci. Eng. C 18 (2001) 37–44. [24] K.I. Shimizu, K. Sugino, K. Kato, S. Yokota, K. Okumura, A. Satsuma, J. Phys. Chem. C 111 (2007) 1683–1688. [25] Y. Suzucki, N. Matsumoto, T. Ainai, T. Miyanaga, H. Hoshino, Polyhedron 24 (2005) 685–691. [26] J. Michalik, T. Wasowicz, J. Sadlo, E.J. Reijerse, L. Kevan, Radiat. Phys. Chem. 47 (1996) 75–81. [27] E. Gashard, J. Belloni, M.A. Subramaniau, J. Mater. Chem. 6 (1996) 867–870. [28] N.D. Cvjeticanin, N.A. Petranovic, Zeolites 14 (1994) 35–41. [29] G. De Cremer, Y. Antoku, M.B.J. Roeffaers, M. Sliwa, J. Van Noyen, S. Smout, J. Hofkens, D.E. De Vos, B.F. Sels, T. Vosch, Angew. Chem. Int. Ed. 47 (2008) 2813– 2816. [30] M. Mostafavi, N. Keghouche, M.O. Delcourt, J. Belloni, Chem. Phys. Lett. 167 (1990) 193–197. [31] A. Henglein, R. Tausch-Treml, J. Colloid Interface Sci. 80 (1981) 84–93. [32] E. Janata, A. Henglein, B.G. Ershov, J. Phys. Chem. 98 (1994) 10888–10890. [33] M. Mostafavi, G.R. Dey, L. François, J. Belloni, J. Phys. Chem. A 106 (2002) 10184–10194. [34] G.R. Dey, K. Kishore, Radiat. Phys. Chem. 72 (2005) 565–573. [35] J.A. Jacob, S. Kapoor, N. Biswas, T. Mukherjee, Colloids Surf. A 301 (2007) 329– 334. [36] S. Kapoor, R. Joshi, T. Mulherjee, Chem. Phys. Lett. 396 (2004) 415–419. [37] N.M. Dimitrijevic, D.M. Bartels, C.D. Jonah, K. Takahashi, T. Rajh, J. Phys. Chem. B 105 (2001) 954–959. [38] S. Mintova, Coll. Czech. Chem. Commun. 68 (2003) 2032–2054. [39] J. Kecht, B. Mihailova, K. Karaghiosoff, S. Mintova, T. Bein, Langmuir 20 (2004) 5271–5276. [40] S. Mintova, V. De Waele, U. Schmidhammer, E. Riedle, T. Bein, Angew. Chem. Int. Ed. 42 (2003) 1611–1614. [41] S. Mintova, V. De Waele, M. Hölzl, U. Schmidhammer, B. Mihailova, E. Riedle, T. Bein, J. Phys. Chem. A 108 (2004) 10640–10648. [42] U. Schmidhammer, V. De Waele, S. Mintova, E. Riedle, T. Bein, Adv. Funct. Mater. 15 (2005) 1973–1978. [43] G. Majano, S. Mintova, O. Ovsitser, B. Mihailova, T. Bein, Microporous Mesoporous Mater. 80 (2005) 227–235. [44] Z. Tahri, F. Luchez, I. Yordanov, O. Poizat, A. Moissette, V. Valtchev, S. Mintova, M. Mostafavi, V. De Waele, Res. Chem. Intermed. 35 (2009) 379–388. [45] A. Henglein, Ber. Bunsen Ges. Phys. Chem. 81 (1977) 556–561. [46] F. Saito, S. Tobita, H. Shizuka, J. Photochem. Photobiol. A 106 (1997) 119–126. [47] K.D. Cadogan, A.C. Albrecht, J. Phys. Chem. 73 (1969) 1868–1875. [48] H. Masuhara, N. Mataga, Acc. Chem. Res. 14 (1981) 312–318. [49] B.D. Michael, E.J. Hart, K.H. Schmidt, J. Phys. Chem. 75 (1971) 2798–2805. [50] S.A. Alkaitis, M. Grätzel, J. Am. Chem. Soc. 98 (1976) 3549–3554. [51] T. Mizoguchi, R.N. Adams, J. Am. Chem. Soc. 84 (1962) 2058–2061. [52] A.N. Frolov, A.V. El’tsov, U.V. Kul’bitskaya, A.L. Ponyaev, Zh. Org. Khim. 16 (1980) 2467–2473. [53] V. Guichard, A. Bourkba, O. Poizat, G. Buntinx, J. Phys. Chem. 93 (1989) 4429– 4435. [54] R. Rahn, J. Schroeder, J. Troe, K.H. Grellmann, J. Am. Chem. Soc. 85 (1963) 1881–1882. [55] E.W. Förtser, K.H. Grellmann, H. Linschitz, J. Am. Chem. Soc. 95 (1973) 3108– 3115. [56] J.K. Percus, Phys. Rev. A 9 (1974) 557–559. [57] D.G. Levitt, Phys. Rev. A 8 (1973) 3050–3054. [58] V. Kukla, J. Kornatowski, D. Demuth, I. Girnus, H. Pfeifer, L.V.C. Rees, S. Schunk, K.K. Unger, J. Kärger, Science 272 (1996) 702–704. [59] K. Hahn, J. Kärger, V. Kukla, Phys. Rev. Lett. 76 (1996). 2762(1)–2762(4). [60] H. Jobic, K. Hahn, J. Kärger, M. Bee, A. Tuel, M. Noack, I. Girnus, G.J.J. Kearley, Phys. Chem. B 101 (1997) 5834–5841.