1. Introduction

Luminescent Ag nanoclusters consisting of only a few Ag atoms and dispersed in different hosts, such as liquids, zeolites, sol gels and glasses [1–6], attract increasing attention due to very high luminescence quantum yield (QY) of up to 60% [1,7,8], which is comparable to that of semiconductor quantum dots. The color of luminescence of Ag nanoclusters can be tuned from UV, throughout VIS, to near-IR by selection of the proper host and the number of Ag atoms comprising the Ag nanoclusters [1,2,9]. The luminescence of these Ag nanoclusters may be used in highly demanded devices, such as luminescent lamps, bio-compatible nano-labels, flexible displays, color phosphors for light emitting diodes (LEDs), and also for down shifting of solar spectrum in photovoltaics, and some other applications [1,2,10]. Despite a large amount of publications on synthesis of luminescent Ag nanoclusters there is no satisfactory understanding of the mechanism of luminescence of Ag nanoclusters, and only recently an energy level diagram for the Ag nanoclusters has been proposed [11].

Recently we have reported on glasses doped with luminescent Ag nanoclusters across the whole glass bulk [10,12], which emit a very broad white color emission spectrum covering all visible range. A broad range of the emission lifetimes of Ag nanoclusters dispersed in this oxyfluoride glass host, varying from ns to tens of µs, has been found [11]. The width and lifetimes range suggest a broad distribution of Ag nanoclusters size/shape/surroundings. Transmission electron microscopy (TEM) studies visualized the Ag nanoclusters and confirmed their size distribution [10]. Based on quantum chemistry calculations and electron spin resonance (ESR) studies, the Ag nanoclusters in this oxyfluoride glass have been argued to be mostly diamagnetic tetramers Ag₄²⁺ with different elongation along the diagonal of the tetramer [9,12].

To the best of our knowledge, no study of polarization memory (PM) of luminescence of Ag nanoclsuters, dispersed in any host, has been carried out up to date. It is shown further that PM of luminescence and its temperature dependence can prove and provide information about an energy transfer between the Ag nanoclusters, which determines the origin of broad white luminescence band of the Ag nanoclusters. Based on the energy transfer mechanism between the Ag nanoclusters, we suggest the new routes for increasing QY of luminescence of Ag nanoclusters.

We have observed that the PM of the Ag nanoclusters emission in the blue range of the spectrum is stronger compared to the green and red ranges. Lowering the temperature of the sample resulted in increase of the PM value in the blue and simultaneous decrease in the green and red. An enhancement by a factor of ten of integral intensity of Ag nanoclusters luminescence with lowering the temperature down to 25 K has been detected. The enhancement is several times larger in the blue part of the spectrum compared to the green and red parts. These findings have suggested a temperature activated energy transfer from Ag nanoclusters emitting in the blue to adjacent Ag nanoclusters emitting in the green and red.

2. Experimental techniques

Ag nanoclusters doped glasses have been prepared by a melt-quenching technique, as described elsewhere [10,13]. In brief, the glass components powders were batched in a Pt crucible, melted for some hours at about 1000 to 1500°C and subsequently cast into a preheated mold, and allowed to cool down to the room temperature. Glass samples with several chemical compositions have been prepared. In particular, in this paper we study a glass which has the following composition: 33(SiO₂) 9.5(AlO_1.5) 32.5(CdF₂) 19.5(PbF₂) 5.5(ZnF₂), mol%, doped with 5 wt% of AgNO₃. This glass will be called further on a basic glass.

Emission spectra have been measured using an Andor Technologies Newton DU970 EMCCD camera attached to a Shamrock SR303i spectrometer. Excitation of the emission was done using either a 355 nm (1 mW) monochromatic line of an Ar-ion laser or a tunable Xe (300 W) source, which was linearly polarized by a Glan-Thompson prism. The spectral response of the setup to different polarizations of luminescence was taken into account. Low temperature measurements of luminescence have been carried out in a helium flow optical cryostat from the room temperature down to 3.7 K.

3. Results and Discussion

Figure 1 shows a CIE chromaticity diagram for the luminescence of the basic glass, which indicates the attainable colors of the luminescence, when excited at used UV wavelengths and temperatures. The white area in Fig. 1 corresponds to white light with different degrees of warm/cool tint. A broad gamut of white light luminescence with different tint can be generated by this Ag nanoclusters doped glass with varying UV excitation wavelength and ambient temperature. Excitation at longer/shorter wavelengths and higher/lower samples temperatures results in a warmer/cooler white luminescence, respectively. An insert in Fig. 1 shows an example of the white luminescence, having a green-blue tint, excited in the basic glass at 350 nm at 5K temperature.

 

Fig. 1 CIE chromaticity diagram, based on 1931 (2°) color matching functions [16], for emission of Ag nanoclusters dispersed in a basic oxyfluoride glass at the room (open symbols) and 14 K (filled symbols) at indicated excitation wavelengths. The white area in the CIE diagram corresponds to a white light gamut. Insert shows a photo of white luminescence at 5 K and excitation at 350 nm by a Xe lamp.

Download Full Size | PDF

Some typical spectra of this white luminescence are presented in Fig. 2(a) at particular excitation wavelengths and temperatures. It is seen in Fig. 2(a) that blue shifting of the excitation wavelength results in a blue shift of Ag nanoclusters emission spectrum. The shift indicates a distribution of Ag nanoclusters parameters, such as number of Ag atoms comprising the nanocluster, shape, size and different surroundings. The room temperature QY in these glasses reaches a value above 20% [10]. Notable, the luminescence intensity undergoes an order of magnitude increase when samples are cooled down to liquid He temperatures. The temperature dependent increase of the intensity of Ag nanoclusters luminescence, or luminescence QY, is found to depend on the excitation wavelength. The increase is higher for shorter excitation wavelengths as it is seen from Fig. 2(b), where the ratio of luminescence intensity at 14 K, I_14K, to intensity at the room temperature, I_RT, is plotted against the wavelength. The spectral dependence of this ratio has a prominent maximum, which red shifts upon red shifting of the excitation wavelength. Note in the case of 355 nm excitation, the growth of luminescence intensity with lowering the temperature is substantial for the whole visible spectral range.

 

Fig. 2 (a) Emission spectra excited at 355 (blue curves) and 440 nm (green curves) at the room (solid lines) and 14 K (dashed lines) temperatures. (b) Spectral changes of luminescence with lowering temperature from the room temperature down to 14 K, represented by a respective ratio on Y-axis, when excited at the indicated wavelengths. The experiments were carried out with a basic glass.

Download Full Size | PDF

The temperature dependence of luminescence, shown in Fig. 2, points at a temperature activated excitation energy transfer between the adjacent Ag nanoclusters within this glass host. Assuming such an energy transfer, we studied the spectral dependence of PM of luminescence of Ag nanoclusters at different temperatures; the results are shown in Fig. 3 . The PM is defined according to the following equation:

 

Fig. 3 (a,b,c) Spectra of polarization memory (PM) of luminescence at the room (red curves) and 14 K (blue curves) temperatures excited at 355 (a), 400 (b) and 435 (c) nm. (d) Spectra of PM of luminescence at the low temperature of 14 K excited at the indicated wavelengths. The experiments were carried out with a basic glass.

Download Full Size | PDF

Where $I_{\left|\right|}$ and $I_{\perp }$ are the luminescence intensity components with polarizations parallel and perpendicular to the linear polarization of the exciting light, respectively. Linear polarization memory of luminescence describes a property of emitted light to preserve the direction of polarization matching that of the exciting light. In particular, the polarization direction of the emitted light may be randomized due to sensitizer-activator energy transfer. In this case the light absorber (sensitizer) and the emitter (activator) are physically different entities, and the excitation energy is transferred from the sensitizer to the activator resulting in change of polarization of the emitted light. By observing the dependence of the polarization memory of luminescence on excitation wavelength, temperature and emission wavelength one can detect energy transfer, identify the sensitizer and the activator entities and establish transfer mechanisms.

Figure 3(a-c) shows that PM is largest in the blue (labeled as “Blue”), and then it decreases in the green (labeled as “Green”). Further in the red (labeled as “Red”), the PM shows some increase at the room temperature or stays invariant at 14 K. Also, decreasing sample temperature down to 14 K results in an increase of PM values in the blue and green, while the PM value decreases in the red. A prominent shoulder at about 550 nm is seen in PM spectral dependence at low temperature and at relatively long excitation wavelength of 435 nm, Fig. 3(c); this shoulder appears and shifts to the red with red shifting of excitation wavelength, Fig. 3(d).

Based on the above observations we propose a mechanism for Ag nanoclusters luminescence. According to notations of Fig. 3, we consider three types of Ag nanoclusters depending on whether their emission occurs in the blue, green or red parts of the spectrum, which will be called further, for simplicity, as the Blue, Green and Red Ag nanoclusters, respectively. These luminescent Ag nanoclusters have been argued to be mostly the Ag₄²⁺ tetramers with different degree of elongation along a tetramer’ diagonal.

Figure 4 shows a three-dimensional configuration coordinate diagram (CCD) of photoexcitation and luminescence of adjacent Blue, Green and Red Ag nanoclusters, depicted by the corresponding colors as described in the caption to the Fig. 4. This CCD explains the experimental findings of Figs. 1-3. The Q1 and Q2 are configuration coordinates of the nanoclusters. The blue, green and red hemispheres in Fig. 4 represent the potential wells of the excited states of the adjacent Ag nanoclusters, which emit the light of the blue, green and red colors, as indicated by down-headed arrows of respective colors.

 

Fig. 4 Three-dimensional configuration coordinate diagram (CCD) of proposed spatial arrangement of adjacent excited states of Ag nanoclusters emitting in the blue (blue hemisphere), green (green hemisphere) and red (red hemisphere). Two temperature activated energy transfer routes from the upper lying to the lower lying hemispheres/nanoclusters are indicated by dashed violet curved arrows. Absorption transitions are shown by solid straight up-headed arrows of the respective colors, and the emission transitions from the blue, green, red emitting Ag nanoclusters are shown by the blue, green and red color down-headed arrows, respectively. The projections of the hemispheres to the zero energy plane are indicated by the dashed ellipses, respectively. Orange wavy arrows show phonon assisted relaxation of Ag nanoclusters within the corresponding potential well. Shaded rectangles indicate barriers for the hops from the Blue to the Green and to the Red Ag nanoclusters according to the Route 1.

Download Full Size | PDF

Maxima or nearly flat parts in PM curves in Fig. 3 correspond to an emission from a Blue (centered about 450 nm), Green (centered about 525 to 600 nm) and Red (675 to 750 nm) Ag nanoclusters, whereas descending parts of PM curves are due to energy transfer by emission of phonons. Therefore, pump at 355 to 435 nm results in excitation of mostly Blue nanoclusters, while a pump at 435 to 480 nm results in excitation of mostly Green and partly Red nanoclusters; the corresponding excitation transitions are shown in Fig. 4 by up-headed arrow of the respective color.

When Blue or Green nanoclusters are excited, a photoexcited electron then relaxes by emission of phonons to the bottom of the blue/green/red potential well, respectively, as indicated by wavy orange arrows in Fig. 4. When the Blue nanoclusters are excited, excited electrons have either a possibility to hop to an adjacent Green and further to the Red Ag nanoclusters, producing green/red luminescence, respectively, as marked by Route 1. Or, the excited electrons inside the blue well have a possibility indicated as Route 2, which involves a single hop to another adjacent Red Ag nanocluster which emits in the red.

Since the Route 1 involves two hops, it will result in a larger loss of PM of luminescence compared to a single hop described by Route 2. With lowering the temperature, the probability of multiple thermal activated hops decreases, therefore the radiative recombination takes place mainly within Blue Ag nanoclusters. Such lowering of the temperature, in particular to 14 K in Fig. 3(a), results in increase of PM for the Blue region of the emission spectrum and flattening of PM spectrum at longer wavelengths.

As the excitation wavelength shifts to the red, the Green and Red Ag nanoclusters can be excited directly, resulting in larger values of PM, as seen in Fig. 3(c,d) compared to Figs. 3(a,b).

An appearance and red shifting of the long wavelength shoulder between 525 to 650 nm on the PM curves in Fig. 3(c,d), when exciting at progressively longer wavelength, add important information about the origin of emitting Ag nanoclusters. Similar step-like descending behavior of the PM has been known for example, for hot luminescence of semiconductors, such as GaAs [14]. Such step-like loss of PM indicates energy transfer between the adjacent Ag nanoclusters with higher energy to Ag nanoclusters with lower energy, when the energy transfer is mediated by a transversal optical phonon, ω_TO, of the host [14]. The energy shift between the onsets of decrease of PM and the onsets of the flat PM regions in Fig. 3(d), which are approximately around 475-525 nm, 575-625 nm and 630-700 nm, correspond to 2000 cm⁻¹ and 1400 cm⁻¹ and 1600 cm⁻¹, respectively. It is known from the Raman scattering data, that these oxyfluoride glasses have largest TO phonon energies of 800-1200 cm⁻¹, corresponding to the breathing mode of SiO₂ tetrahedra with bridging and non-bridging oxygen atoms [15]. Therefore the Blue to Green nanoclusters transition/energy transfer may require simultaneous absorption of two such ω_TO phonons, while transition/energy transfer from the Green to the Red clusters is at about 1500 cm⁻¹ and therefore requires an absorption of one or two such phonons. This indicates a larger barrier for the transition from the Blue to the Green nanoclusters compared to the barrier for the transition from the Green to the Red nanoclusters, as schematically shown in CCD in Fig. 4 by shaded rectangles.

This also explains the temperature dependence given in Fig. 2(b), where intensity of Blue emitting nanoclusters increases with lowering temperature much stronger than for the Green and Red clusters. With such lowering, a phonon population number decreases according to the Bose-Einstein formula and this results in a decrease of the probability of simultaneous absorption of two TO phonons, which is required for the electron hop from the Blue to Green well. The broadening of the PM features in Fig. 3 is due to a disorder naturally present in a disordered matrix glass, resulting in different surroundings for a single Ag nanocluster or a nanoclusters group.

At present we investigate means to improve the quantum yield of the white luminescence of Ag nanoclusters doped glasses. In general, four routes to higher room temperature QY are the optimization of the glass melting and casting conditions, glass chemical composition, silver doping concentration and wavelength of the exciting light.

4. Conclusion

In conclusion, we report on investigation of polarization memory of luminescence and its temperature dependence for luminescent Ag nanoclusters dispersed within oxyfluoride glass host. We propose a model explaining a broad white color photoluminescence of the Ag nanoclusters by temperature activated energy transfer/hops from the adjacent Ag nanoclusters emitting in the blue to either Ag nanoclusters emitting in the green and further in the red, or directly to Ag nanoclusters emitting in the red. The means for further enhancement of quantum yield of this white luminescence are also suggested.