1. Introduction

Over the last few years photo-excited phenomena in dielectric or semiconductor materials have attracted interest due to their importance in the understanding of fundamental processes as well as to their potential applications in the development of integrated optical and photonic devices. Recent experiments have shown that intense laser pulses produce formation and decoloration of color centers in metal-glass composites, which consequently result promising tools for ultrafast optical switching [1–3]. Processes induced by weak light also find application as optical control techniques in many practical situations. For instance, light-induced phase transformations can be used both to modify light propagation in photonic-band gap materials [4] and to develop rewritable optical memories [5, 6]. Non-thermal light-induced atomic desorption [7, 8], shortly LIAD, allows the loading with suitable vapor densities of hollow-core fibers [9] and atom chips [10, 11].

Light interaction plays also a key role in tuning the optical response of metal nanoparticles, which are characterized by surface plasmon resonances dependent on their size, shape, and environment [12]. Earlier studies have demonstrated that photothermal effects ending in atomic desorption are a powerful tool to control number, size and form of metallic clusters deposited on surfaces [13–15]. In these experiments cluster formation and evaporation are not entangled processes because the atoms once desorbed cannot be recycled.

In this work we investigate light-induced effects on Rb atoms, layers and clusters simultaneously embedded in nanoporous glass. Under this confined condition we find that light-desorbed atoms remain trapped and contribute to change the Rb phase equilibrium inside the pores. This generates a loop where nanoparticles and atomic layers form a nearly closed system whose evolution is controlled by optical radiation. Frequency tuning of light induces atomic desorption either from the pore surface or from nanoparticles allowing atoms to move from clusters to layers and vice versa. Indeed we have recently observed in Rb loaded-porous silica that green-blue light desorbs adatoms, i.e. individual atoms lying on a surface, from the pore walls whereas red-near infrared (NIR) light desorbs atoms from clusters [16]. In the first case extremely weak light fluxes not only drive some atoms out of the sample but also increase the atomic mobility inside the silica nanocavities, and a large part of the detached atoms condense on the pore walls forming new clusters. Whereas LIAD effect produces cluster formation, resonant absorption of red-NIR light induces cluster evaporation. This makes the nanocomposite either blue or transparent respectively. We demonstrate that these photoinduced transformations between a high and a low transmission state can be used both for light writing or bleaching and for color sequence recording in Rb-loaded porous glass.

2. Experimental

The porous glass used in our experiments is 96% silica with a random interconnected network of elongated pores about 17 nm in diameter; the porosity is approximately 50% of the total glass volume [17]. The sample (30×15×1mm³) (Fig. 1(a)) is inserted in a Pyrex cell formed by a cylinder (length 7.9 cm; diameter 3.4 cm) close at the ends by glass windows. It is mechanically fixed close to one of the cell windows by a Pyrex ring sealed to the glass tube. The cell preparation is as follows: it is first connected to a vacuum system, and kept for a few days at a temperature slowly increasing up to 420 K in order to speed up desorption of impurities and water; then rubidium metal is distilled in a side arm welded to the cell body through a short capillary, finally the cell is sealed off. The cell is kept all the time at room temperature and the sample is continuously exposed to the alkali vapor supplied by the reservoir. During the porous glass loading, atoms from the cell volume slowly diffuse inside the pores where they stick to the surface forming atomic layers and clusters.

 

Fig. 1. (a) Picture of nanoporous glass sample fixed close to one of the cell windows. (b) Experimental set-up. The “writing” or “bleaching” laser beams illuminate the porous glass sample (PG). The reading laser beam measures the PG transmission, while a 780 nm laser beam monitors the Rb vapor density (PD1, PD2: photodiodes). (c) Light induced cluster formation/evaporation processes. Green-blue light desorbs atoms from the silica pore walls whereas NIR light detaches atoms from clusters. In the first case the desorbed atoms build up clusters, in the second case surface layers.

Download Full Size | PDF

Figure 1(b) shows a sketch of the experimental apparatus. The Rb vapor density is monitored by means of a probe laser beam, generated by a free-running diode laser, resonant with the Rb D2 line (780 nm). The probe laser is scanned across the D2 line. The spectrum is continuously acquired and then elaborated to extrapolate the vapor density. A second probe beam from an external cavity diode laser, tuned at 785 nm, i.e. the reading beam, measures the porous glass transmission. Other laser sources, emitting in the blue-green and the NIR spectral regions, are used in turn to illuminate the porous glass sample. Their frequency and intensity are selected in such a way either to write or to bleach the porous glass sample.

A schematic picture of the induced processes is shown in Fig. 1(c). When light desorbs atoms from the pore surface their mobility increases and, due to the restricted geometry of the host matrix, they have a high probability to form Rb nanoclusters on the glass walls. Since these particles show surface plasmon resonance absorption bands in the red-NIR region [16], the porous glass turns blue. On the contrary, when light induces cluster evaporation the desorbed atoms land on the glass surface and the sample becomes transparent again. During the illumination time, a fraction of the desorbed atoms flows out of the sample and the Rb vapor density inside the cell rises. In the dark the porous glass traps atoms back again and the equilibrium between phases is slowly re-established. Light makes porous glass “breathing” and the loading-empting processes may be repeated at will with constant features.

3. Results and discussion

Indirect information about the photo-induced phase transformations of Rb inside the nanopores can be extrapolated by looking at the atomic density changes in the cell, during illumination. Indeed it has been observed that light ejects alkali metal atoms from porous glass and consequently increases the external vapor pressure [18, 19]. One of the most important parameter for the evaluation of the involved processes is the desorbing rate. It is defined as the number of desorbed atoms per second, immediately after the light is switched on, and it is extracted from the absorption signal [20]. In Fig. 2 the desorbing rate as a function of light wavelength is shown. It is characterized by an essentially non resonant monotonic increase with the light frequency, plus a broad resonance in the red-NIR region. This behavior clearly demonstrates the presence of two distinct desorption mechanisms, even if their relative weights are affected by the atomic distribution close to the outer surface of the sample.

 

Fig. 2. Rb Desorbing Rate as a function of desorbing light wavelength. Black curve is the sum of two functions: an exponential one (blue line) and a Gaussian one (red line), which describe photodesorption from the pore surface and clusters respectively. Laser intensity is 5 mW/cm² uniform all over the surface of the porous glass sample.

Download Full Size | PDF

The non-resonant contribution is due to non-thermal atomic desorption from the glass pores [18] while the resonant part is due to Surface-Plasmon Induced Desorption [21]. The first process is based on local electronic excitations at the surface defects [22] and can be explained in the framework of the Menzel-Gomer-Redhead theory [23, 24]. This is consistent with the experimental results, which show that the yield of alkali metals desorbed from borosilicates increases monotonically with the photon energy [16, 25]. The second process is instead produced by the excitation of surface plasmon oscillations, which cause atomic desorption from metallic nanoparticles. The desorption dependence on photon energy is mainly dominated by the dipolar surface plasmon excitation which contributes to the electric field enhancement at the cluster surface [26]. As the desorbing light intensity grows the particle temperature rises and thermal evaporation takes place [14]. As follows from the data, the efficiency of LIAD from glass increases greatly towards the UV nevertheless it occurs also for red-NIR light. Therefore whereas green-blue light excites only surface layers, red-NIR light interacts both with clusters and layers.

Whereas the desorbing rate concerns with adatoms and clusters close to the outer surface of the sample, the change of the porous glass transparency under illumination directly describes atomic redistribution within the bulk matrix, as illustrated in Fig. 3. Here the acronym “LIAD” recorded on the sample after exposure to green light at 532 nm is shown (Fig. 3(a)). The light intensity is 20 mW/cm² and the exposure time is about two minutes. Printing is done by uniformly illuminating the sample through a black mask with carved letters placed in front of the cell window. The strong coloration is due to the light-induced increase of Rb cluster density inside the pores with respect to the equilibrium condition (see Fig. 1(a)). When light is switched off, atoms slowly move from clusters to the pore surface. The color slowly disappears with a characteristic time of the order of 10 hours. Nevertheless the bleaching process can be made faster by using NIR light. Figure 3(b) shows a detail of the two letters “IA” after illumination with a laser beam at 808 nm. The light intensity is 2.7 W/cm² and the illumination time is about 40 seconds. The beam has been focused on the letter “I” which results almost erased. The evolution of the sample in the dark is reported in Fig. 3(c).

 

Fig. 3. (a) The acronym LIAD is recorded on the porous glass after green illumination (wavelength: 532 nm; intensity: 20 mW/cm²). (b) Detail of letters IA after illumination of letter I with NIR light (wavelength: 808 nm; intensity: 2.7 W/cm²). (c) Time evolution of the sample after the illuminations; each photo is taken one hour after the other.

Download Full Size | PDF

We find that the rate of cluster formation at the beginning of illumination is proportional to the number of incident photons. This is evident by looking at Fig. 4 where the relative transmission decrease rate at 785 nm as a function of light intensity is shown for green illumination at 532 nm. This parameter is defined as the derivative of the sample transmission, normalized to the equilibrium value in the dark, calculated at the time when the light is switched on. As in the case of the desorbing rate, the relative transmission decrease rate is directly determined from the experimental data. The linear dependence, which we observe for the light induced cluster formation, indicates a non-thermal process and is common to photodesorption from porous glass [18]. This result therefore provides direct experimental evidence that these two phenomena are strongly correlated.

 

Fig. 4. Relative Transmission Decrease Rate as a function of light intensity. The change of porous glass transmission is recorded at 785 nm when the sample is uniformly illuminated by green light at 532 nm.

Download Full Size | PDF

We observe that porous glass loaded with rubidium shows another peculiar property: it “remembers” the color sequence of light pulses. This means that its answer to the same perturbation depends on the wavelength of a previous illumination. We monitor this effect by “reading” both the sample transmission at 785 nm and the Rb vapor density in the cell, when the porous glass is alternately exposed to a writing beam at 488 nm and a bleaching beam at 808 nm. In Fig. 5(a) a sequence of three light pulses, namely NIR-blue-NIR, is applied to the glass. A small portion of the sample is illuminated in order to investigate the change of local equilibrium phase in the matrix. The blue and NIR spots are 0.3 cm² and 0.1 cm² respectively and are overlapped on the porous glass. The illumination time is 40 s and the dark time between two consecutive illuminations is 280 s. The light intensity is 5.6 mW/cm² and 2.2 W/cm² for blue and NIR light respectively. The first NIR light pulse increases the Rb vapor density in the cell, and slightly decreases the glass transparency. This means that at thermal equilibrium the bleaching beam works as a writing beam because the largest part of the adsorbed atoms forms layers within the bulk matrix. The blue pulse causes a large decrease of the glass transparency and a small vapor density increase. The second NIR pulse, that has the same intensity of the first one, partially bleaches the porous glass and increases the vapor density more than the first NIR pulse. Moreover the desorption dynamics is slower. The changes between the first and the second NIR pulses are due to the structural phase change induced by the blue pulse on the composite, and reflect the fact that the cluster number is increased with respect to the equilibrium condition in the dark. This is evident when looking at the inset of Fig. 5(a), where two NIR pulses are sent to the cell. In this case the signals are identical, the first pulse does not affect the second one.

 

Fig. 5. Porous glass exposed to a sequence of NIR-blue-NIR (a) and blue-NIR-blue light pulses (b). The light intensities are 5.6 mW/cm² and 2.2 W/cm² and the illuminated areas are 0.3 cm² and 0.1 cm² for blue (488 nm) and NIR (808 nm) light respectively. The insets show sequences of two red (a) and blue pulses (b). Black curves give the glass transmission at 785 nm and red curves the vapor density inside the cell.

Download Full Size | PDF

In Fig. 5(b), the sequence blue-NIR-blue is reported. The first blue pulse writes on the sample and drives some atoms in the cell volume, as in Fig. 5(a). The NIR pulse induces a large change in the vapor density and bleaches the sample. The second blue pulse again gives a small peak in the vapor density and writes via cluster formation on the sample. In this case the NIR pulse has only essentially modified the glass transparency, as shown in the inset of Fig. 5(b), where the effect of two blue pulses is reported. Therefore the effect produced by NIR light on the nanocomposite is determined by the equilibrium between Rb layers and clusters. Changing the ratio between these atomic distributions modifies the optical response of the glass sample to a NIR light pulse.

4. Conclusion

In conclusion we extend the investigation of the properties of metals embedded in porous matrices [27–30] by taking into account their optical response. We find that light, even below the mW/cm² range, builds up clusters with atoms detached from the pore surface. This is due to the combined effect of nanoconfinement and atomic photodesorption. The increase of cluster density brings the sample transmission to a lower level. Then if cluster evaporation is induced by resonant absorption of light, atoms move from clusters to the pore surface and the transmission increases again. Both these processes are reversible and depend on light frequency and intensity. Moreover we demonstrate that they supply a method for storing and erasing images as well as for preserving memory of a previous illumination. This result increases the number of applications of light-induced phase transformations for optical control in nanostructured materials and opens perspectives in the field of optical information.