Surface-luminescence from thermally reduced bismuth-doped sodium aluminosilicate glasses

Journal of Non-Crystalline Solids 358 (2012) 3193–3199

Contents lists available at SciVerse ScienceDirect

Journal of Non-Crystalline Solids journal homepage: www.elsevier.com/ locate/ jnoncrysol

Surface-luminescence from thermally reduced bismuth-doped sodium aluminosilicate glasses Karsten H. Nielsen a, Morten M. Smedskjaer b, Mingyeng Peng c, Yuanzheng Yue b, Lothar Wondraczek a, d,⁎ a

Department of Materials Science, Friedrich-Alexander University of Erlangen Nuremberg, Germany Section of Chemistry, Department of Biotechnology, Chemistry and Environmental Engineering, Aalborg University, Denmark State Key Laboratory of Luminescent Materials and Devices and Institute of Optical Communication Materials, School of Materials Science and Engineering, South China University of Technology, Guangzhou, China d Otto-Schott-Institute, University of Jena, Germany b c

a r t i c l e

i n f o

Article history: Received 9 July 2012 Received in revised form 11 September 2012 Available online xxxx Keywords: Bismuth; Aluminosilicate; Luminescence; Near-infrared; Thermal reduction

a b s t r a c t We report on the effect of hydrogen annealing on the optical properties of bismuth-doped sodium aluminosilicate glasses. The redox state of bismuth in the as-melted glasses is governed by the composition, viz., NIR luminescence is observed only in the glasses with low optical basicity. Upon thermal reduction, visible emission from Bi 3+ and, eventually, minor amounts of Bi 2+ is significantly lowered, depending on heat-treatment time and temperature, and glass composition. Hydrogen treatment was also found to result in a decrease of the NIR emission intensity and, at the same time, formation of metallic bismuth particles in the surface region. Surface-tinting as well as the decrease of visible luminescence follow Arrhenian kinetics, suggesting that hydrogen permeation is the rate-governing process. Upon re-annealing in air, the effects of thermal reduction on the optical properties are reversible only to a limited extent. © 2012 Elsevier B.V. All rights reserved.

1. Introduction The optical properties of bismuth-doped oxide glasses have been arousing significant and renewed interest over the last few years. This has been motivated by various potential applications such as simple coloring [1], third order optical non-linearity (e.g. [2]), and surface conductivity (e.g. [3]) on the one side, and luminescence in the visible (VIS) and near-infrared (NIR) spectral range on the other side [4,5]. In particular, broadband NIR luminescence has been studied extensively for application in novel laser sources and optical amplifiers [5–8]. However, due to the large variety of redox states in which bismuth may be present in oxide glass matrices, the respective origin of luminescence and other optical properties remains debated [4]. It has therefore become important to be able to manipulate the redox state of bismuth in oxide glasses. In silicate glasses, Bi3+ and metallic bismuth are traditionally regarded as the most prevalent species [9]. For instance, voltammetric studies on a soda lime silicate melt at 1250 °C have confirmed the presence of these two species (Bi0 and Bi3+), but also indicated the presence of a third species (presumably Bi5+) [10]. The presence of bismuth in additional oxidation states (e.g., Bi+, Bi2+ and Bi5+) has also been suggested by, e.g., X-ray photoelectron spectroscopy [2,11], indicating that the redox chemistry of bismuth in glasses and glass-forming liquids is more complex than first ⁎ Corresponding author at: Otto-Schott-Institute, University of Jena, Fraunhoferstrasse 6, Jena D-07743, Germany. Tel.: +49 3641 9 48501; fax: +49 3641 9 48502. E-mail address: [email protected] (L. Wondraczek). 0022-3093/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.jnoncrysol.2012.09.021

assumed. Currently, this has prevented knowledge-based tools for exploiting ultrabroad NIR photoluminescence of bismuth-doped glasses, and even the nature of the NIR emitting center remains highly debated [5]. Luminescence from Bi3+ and Bi2+ occurs in the blue and red spectral ranges, respectively [11–13]. The origin of NIR luminescence has been ascribed to Bi+ and subvalent species [4], and Bi ion clusters such as Bi2-, Bi22- [14,15], or Bi53+ [16,17]. In addition, the highly oxidized valence state of bismuth, i.e., Bi5+ [15,18] has also been suggested as a source for NIR photoluminescence. Several different approaches have been attempted to obtain and optimize NIR luminescence from bismuth-doped glasses by manipulating the oxidation state of bismuth. These approaches include controlling the optical basicity of the glass [19–21], controlling the melting atmosphere and temperature [22,23], adding oxidation [24] or reduction agents [23] to the batch, tempering the glass [23], crystallizing the glass [25], and irradiating the glass with femtosecond lasers, γ-rays, or high-energy electron rays [26–28]. Based on these previous efforts it can be stated that the occurrence of NIR-active Bi species is highly sensitive to numerous parameters [29] and that its optimization requires delicate tuning of all parameters which affect the oxidation state of bismuth [23]. Heat-treatment of a glass in a reducing atmosphere offers several levers for such tuning (i.e., gas type, partial pressure, temperature, and duration) [30]. It has been shown that the luminescence properties of rare earth-doped silicate [31], aluminosilicate [32], alkali`borosilicate [33], and alkali aluminosilicate [33] glasses can be tuned through

3194

K.H. Nielsen et al. / Journal of Non-Crystalline Solids 358 (2012) 3193–3199

reduction of the optically active polyvalent ions. However, heattreatment of a bismuth-doped glass fiber in H2 atmosphere with subsequent flame brushing negatively affected the NIR luminescence [29]. On the other hand, thermal reduction of bismuth-doped crystalline phosphors has been applied to increase NIR activity [34]. Heat-treatment of bismuth-doped glasses in a reducing atmosphere is thus a potential approach for manipulating the bismuth oxidation state and consequently the NIR luminescence. It should be noted, though, that the thermal reduction may influence the optical properties differently in different matrix materials and process conditions, especially when metallic particles are created [4,35,36]. In addition to affecting the redox state of the material's surface, such heat-treatments of glasses containing polyvalent ions can also induce ionic diffusion in the glass surface layer [37]. In this work, we investigate the luminescence properties of Bi2O3-doped sodium aluminosilicate glasses which have been heat-treated under reducing conditions. We also study the effect of the [SiO2]/[Al2O3] ratio in the base glass on the optical properties and discuss our findings in terms of the reduction and luminescence mechanisms. 2. Experimental 2.1. Sample preparation As a model system and host for bismuth dopants, sodium aluminosilicate glasses of the type (80 − 2x)SiO2∙(20 + x)Na2O∙xAl2O3 (mol%) were chosen. The optical basicity and the glass transition temperature (Tg) were varied by varying x, but the number of nonbridging oxygen per tetrahedron (NBO/T) was kept constant. Four glasses with x = 0, 5, 10 and 15 (denoted Al0 … Al15) were produced and doped with 0.5 mol% Bi2O5. The optical basicity was evaluated using the model of Duffy and Ingram [38]. For the present case, it increases from 0.55 to 0.64 from Al0 to Al15. Glasses were prepared by melting batches of SiO2, Al2O3, Na2CO3, and Bi2O3 powders in alumina crucibles at temperatures given below with subsequent quenching by casting on a preheated brass plate (or a steel plate). Al0 was melted at 1550 °C for 50 min, whereas Al5, Al10, and Al15 were held in a temperature range between 1450 and 1550 °C for 60–90 min in order to refine and homogenize the melts. After casting, samples were annealed slightly below their respective glass transition temperature (Tg) for 1 h, and naturally cooled by turning off the furnace (Table 1). Samples of ~0.8–1.0 cm in diameter and 2.00± 0.05 mm thickness were obtained by cutting from each slab of glass with a diamond saw. They were then ground with SiC grinding papers under ethanol and finally polished using a cloth with diamond paste.

buffer was placed in the tube to fix the oxygen partial pressure [37]. Treatment temperatures Tr were chosen with respect to Tg of each sample (Table 1). Reduction experiments were conducted in two different gas mixtures, 1/99 H2/N2 and 10/90 H2/N2, for two different durations (2 and 4 h). An additional series of experiments was conducted on Al5 in 10/90 H2/N2 for 0, 1, and 8 h (0 h refers to a dynamic heat-treatment without isothermal hold at the maximum temperature). Furthermore, Al15 was heat-treated in 10/99 H2/N2 at the treatment temperatures of Al0 and Al5, corresponding to 0.94Tg (487 °C) and 0.97Tg (512 °C) of Al15. Re-annealing of thermally reduced Al10 treated for 2 h in 10/90 H2/N2 was conducted in air at Tg for 1 and 44 h, respectively, in order to investigate the reversibility of the thermal reduction process. 2.3. Characterization Calorimetric Tg of the as-prepared glasses was determined by differential scanning calorimetry (DSC, Jupiter STA, Netzsch, Selb, Germany) by applying a two scan procedure as described elsewhere [39]. Optical absorption spectra were recorded on two UV/Vis/NIR spectrometers (Lambda 950, PerkinElmer, Waltham MA, USA, & Cary Bio 50, Varian Inc., Palo Alto CA, USA). Photoluminescence spectroscopy in the Vis spectral range was measured on samples treated in 10/90 H2/N2 atmosphere on a Vis fluorescence spectrometer (Cary Eclipse, Varian Inc., Palo Alto CA, USA). For measurement of photoluminescence in the infrared spectral region, a spectrofluorometer equipped with an IR photomultiplier tube (PMT, H10330A, Hamamatsu, Shizuoka, Japan on Fluorolog-3, Horiba Jobin Yvon, Unterhaching, Germany) was employed on selected samples. Finally, X-ray diffraction patterns were collected on selected samples at a step size of 0.002°/s (Siemens Kristalloflex D500, Bragg-Brentano, 30 kV/30 mA, Cu Kα). 3. Results and discussion 3.1. As-prepared glasses Optical absorption spectra of the as-prepared glasses are shown in Fig. 1. Except for Al0, all glasses appear fully transparent and colorless to the naked eye (inset of Fig. 1). The as-melted Al0 glass exhibits a brown tint. Absorption spectra of Al5, Al10, and Al15 follow a similar trend with a UV cut-off around 330 nm, whereas Al0 exhibits an additional broad absorption band over the spectral range of 330 to ~ 750 nm. No sharp bands can be distinguished in all curves, although for Al0, a shoulder peak is observed around 390 nm. 0.6

2.2. Thermal reduction

0.4

Absorbance

Heat treatment of the glasses was conducted in H2/N2 atmosphere using a custom designed tube furnace (SF17, Entech, Ängelholm, Sweden). Samples were placed in the furnace at room temperature. The furnace was subsequently evacuated and flushed with the treatment gas three times before commencing experiments. Heating and cooling cycles were performed at 10 K/min. An Fe2O3/Fe3O4 oxygen

Al0 Al5 Al10 Al15

0.2

Table 1 Glass transformation temperature (Tg), optical basicity (Λ) [38], annealing temperature after melting [Ta], and treatment temperature during thermal reduction (Tr).

Λ Tg (°C) Tr (°C) Ta (°C)

Al0

Al5

Al10

Al15

0.55 483 487 450

0.58 495 498 450

0.61 507 512 450

0.61 528 534 520

0.0 200

400

600

800

1000

Wavelength (nm) Fig. 1. Optical absorbance as a function of wavelength for the as-prepared glasses. Inset: photograph of the samples with four different Al2O3 contents.

K.H. Nielsen et al. / Journal of Non-Crystalline Solids 358 (2012) 3193–3199

(Fig. 2b). This band is due to the presence of a minor amount of Bi2+ species [19,20,46,47]. As shown in Fig. 2c, weak NIR luminescence occurs from the Al0 sample, but is not observed in any of the other glasses. For silicate glasses, it has been reported that addition of small amounts of Al2O3 may lead to enhanced NIR luminescence [21]. The extent of this enhancement depends on overall optical basicity, local optical basicity, and its additional effect on the redox state of bismuth. The absence of NIR luminescence in Al5, Al10, and Al15 might be a direct result of the higher optical basicity of these glasses [19,23,48].

Glasses containing metallic bismuth colloids have been reported to appear brownish or dark in color [2,40], indicating that metallic bismuth could be present also in the as-prepared Al0 glass. However, the longer melting times for Al5, Al10, and Al15 compared to Al0 should be favorable for the occurrence of more reduced bismuth in those glasses [9]. According to the theory of Duffy and Ingram, decreasing the optical basicity of the glass shifts the redox equilibrium to the more reduced side [38]. Hence, glass composition and not melting conditions appears to determine the bismuth redox state. Moreover, increasing the alumina content in the glass-forming liquid may prevent ionic clustering by the so-called dispersive effect of alumina [41]. This effect has been found in bismuth-bearing glasses [21] and may offer an additional explanation for the above-mentioned observation. Luminescence emission spectra for the as-prepared glasses excited at 310 nm are shown in Fig. 2a, where the broad blue emission band which is typical for Bi3+ is readily identified. The maximum fluorescence intensity increases from Al0 to Al5 and then decreases with further increasing alumina content. At the same time, a red-shift is observed (inset of Fig. 2a). This red-shift of the Bi3+ emission band with increasing [Al2O3] could be due to the decrease in the energy difference between the s and p orbitals with increasing optical basicity [21,42]. A similar red-shift has been found in a bismuth-doped aluminosilicate glass with increasing SrO content [19]. The emission spectra become more asymmetric towards higher wavelength with increasing Al2O3 content (Fig. 2a). A similar asymmetric band shape has been observed for bismuth-doped alkali germanate glasses [43], where it was suggested that red asymmetry results from the presence of bismuth emission centers in lower redox state, i.e., Bi2+. However, this explanation is not in agreement with the fact that the shoulder peak is more pronounced for the glasses with higher optical basicity. Considering the use of a fixed excitation wavelength, the broadening may also be related to emission from higher energy levels, as it has been observed on emission due to a 6p-6 s transition from Tl + in sodium alumino silicate glass [44]. For this system, incorporation of Tl + as a charge compensator for AlO4- was reported to cause a blue shift of the emitted light [45]. This makes the presence of charge-compensating Bi 3+ as an explanation for the observed broadening and red shift of the emission from the glasses of this study unlikely. In the Vis range, an additional weak emission band is observed around 765 nm for the most acidic glasses upon excitation at 310 nm

Following heat-treatment in the hydrogen-containing atmospheres, a brownish tint is developed in all four glasses. This colorization is also reflected by the absorption spectra and photographs of Al5 heat-treated at 498 °C for different durations in 10/90 H2/N2 as shown in Fig. 3. The observed decrease in optical transmission with increasing heattreatment duration is similar to that reported in Ref. [2] for glasses containing bismuth colloids obtained by striking. The evolution of color is attributed to the formation of metallic Bi particles which leads to Vis light scattering and yellow, brown, or even red coloration [4]. Noteworthy, no surface plasmon resonance (SPR) peaks have been observed in this study. In borate and borosilicate glasses, bismuthrelated SPR peaks have been observed at 460 nm [23,24,49] and 560 nm [2] However, in the absence of boron oxide, only weak or no SPR from bismuth colloids could be detected [23,40]. In this work, the apparent absence of well-resolved SPR in the transmission spectra indicates the absence of small metallic bismuth particles (i.e., b 20 nm), whereas the formation of larger particles cannot be excluded [4]. It is found that the coloration can be removed by scratching or polishing the surface layer, indicating that the structural changes responsible for the coloring take place only in the surface layer of the thermally reduced glasses. The tarnishing model has proven successful in describing ion migration depths following thermal reduction [50] as well as for describing absorption data based on light scattering from metallic particles in glass created due to hydrogen permeation [35,36,51]. To describe our data in terms of the tarnishing model, we introduce an effective treatment time tr, which is the time period for which the sample is held at a temperature higher than 0.80Tg (in K) [52] (including, e.g., heating ramps). As shown in the inset of Fig. 3, the

b ex: 310 nm

440 430 420 410 0

Intensity (a.u.)

3.2. Thermal reduction

5

10

15

Al 2O 3 content (mol%)

675

Intensity (a.u.)

Al10 Al15

ex: 310 nm 500

825

c ex: 450 nm

Al5

Wavelength (nm)

750

Wavelength (nm)

Al0

400

Al0 Al5 Al10 Al15

Intensity (a.u.)

Peak position [nm]

a

3195

600

1050

Al0 Al5 1200

1350

1500

Wavelength (nm)

Fig. 2. a) Photoemission in the 350–600 nm wavelength range from as-melted glasses excited at 310 nm. Inset: Position of emission peak maxima as a function of Al2O3 concentration. b) Photoemission in the 660–870 wavelength range. c) Infrared emission spectra for untreated glasses after 450 nm excitation. Infrared luminescence is found only in the most acidic glass (Al0).

3196

K.H. Nielsen et al. / Journal of Non-Crystalline Solids 358 (2012) 3193–3199

a

Bi3+ species is lowered; (2) newly-formed particles on the sample surface cause scattering of emitted and excited light, especially for the low wavelength range (Fig. 3); (3) the amount of optically active Bi3+ species is reduced relative to the amount of optically active Bi2+ species; and (4) significant self-quenching occurs due to increasing overall optical absorption with increasing treatment time. The latter effect results in the asymptotic decrease of VIS emission intensity (Fig. 6), governed by the reactions

8 hours 4 hours 2 hours 1 hour 0 hours untreated

2

1.0

0.5

0

10

ð1Þ

2þ

0

ð2Þ

Bi ➔Bi ➔Bin

1

400

0

20

(tr)0.5 (min0.5)

0

3þ

Bi ➔Bi ➔Bin 0.0

600

800

1000

Wavelength (nm) Fig. 3. a) Photographs of samples Al5 reduced at Tg in 10/90 H2/N2 for various durations. b) Optical absorption spectra for the same samples. Inset: increase in absorbance at 400 nm as a function of the square root of the effective reaction time tr according to the tarnishing model (see text). The line represents a linear fit to the data.

3.3. Effect of treatment temperature The influence of heat-treatment temperature on the optical properties has been investigated for the Al15 glass treated in 10/90 H2/N2 atmosphere. Fig. 7a–b shows the optical transmission and Vis fluorescence spectra. We find that increasing heat-treatment temperature reduces optical transmission as well as fluorescence intensity. Both properties appear to change congruently, indicating that they are governed by the same reduction reaction and directly coupled to each other. For the investigated temperature range, the average rates of absorbance increase at 400 nm and fluorescence intensity appear to follow Arrhenian behavior (Fig. 7c). Hence, we assume that hydrogen permeation is the rate-governing process at this partial pressure [30]. This observation is similar to that of a previous study on the reduction of silver-containing glasses [35]. 3.4. Effect of glass composition The effect of glass composition on optical absorbance is illustrated in Fig. 8, showing the difference in optical absorbance at 400 nm between untreated and treated samples with increasing Al2O3 content in glasses heat-treated at their respective Tg for 2 and 4 h in two different atmospheres. While for treatment in 1/99 H2/N2 atmosphere,

Al5, 8hours in 10/90 H2 /N2

Intensity (a.u.)

(104) (110)

Intensity (a.u.)

(012)

data obtained in this study are well described by the tarnishing model, i.e., there exists a linear relation between the change of absorbance and tr1/2. This implies the occurrence of a diffusion-controlled process. The fact that the linear fit does not go through the origin of the diagram indicates a problem with estimating the effective time, or the presence of a lag phase. For glasses with high bismuth content, a lag phase in the initial phase of thermal reduction has been suggested, during which bismuth atoms are reduced and migrate to form clusters and, at a later stage, colloids [3]. Rhombohedral metallic bismuth crystals [53] can indeed be detected by X-ray diffraction as a result of prolonged thermal reduction (i.e., 8 h) in the Al5 glass (Fig. 4). An estimated particle size of ~ 40 nm is found for these crystallites by application of the Scherrer equation [54] on the diffraction peaks from the (012), (104), and (110) planes. As shown in Fig. 5, thermal reduction of the Al5 glass in 10/90 H2/N2 results in a decrease in the intensity of Vis fluorescence with increasing treatment duration. In addition, the position of the emission band undergoes an apparent red-shift relative to the untreated glass. This could be caused by four effects: (1) the total amount of optically active

where Bin refers to bismuth cluster or ion clusters of n atoms. The weak NIR luminescence intensity diminishes as a result of thermal reduction (Fig. 5). Similarly, the band at 765 nm, which is ascribed to Bi 2+, disappears. Thus, the measured fluorescence spectra do not imply the formation of active, low-redox Bi luminescent centers following thermal reduction.

untreated 0 hour 1 hour 2 hours 4 hours 8 hours

Intensity (a.u.)

Δ Abs @400nm

Absorbance

b3

0

200

400

600

t r (min)

ex: 310 nm Bismuth, JCP2 01-085-1331 350

400

450

500

550

Wavelength (nm) 20

30

40

50

60

70

2θ Fig. 4. X-ray diffraction pattern of sample Al5 treated at Tg for 8 h in 10/90 H2/N2. For reference, the diffraction pattern of rhombohedral bismuth is shown (JCPDS card no. 85-1330).

Fig. 5. Photoemission spectra of Al5 treated in 10/90 H2/N2 for different durations after 310 nm excitation. The arrow illustrates intensity decrease and the red-shift of the emission peak with increasing progress of thermal reduction. Inset: fluorescence intensity normalized to that of the as-prepared sample as a function of the effective reaction time tr, showing an asymptotic reduction process.

K.H. Nielsen et al. / Journal of Non-Crystalline Solids 358 (2012) 3193–3199

Ex: 450 Δ Absorbance @400nm

Intensity (a.u.)

2h 10% H 2 2h 1% H2 1200

1300

1400

1500

10% H 2 4hours

1.2

Untreated

1100

3197

10% H 2 2hours 1% H2 4hours

0.4

0.0

1600

1% H2 2hours

0.8

0

5

Wavelength (nm)

10

15

Al2O3 content (mol%)

Fig. 6. Infrared emission spectra of Al0 treated at 487 °C for 2 h in 10/90 H2/N2 and 1/ 99 H2/N2, respectively.

the impact of composition is below the detection limit, for treatment in 10/90 H2/N2, the absorbance change markedly drops when adding Al2O3 into the Al0 glass. This indicates that formation of coloring Bi-species is suppressed by adding Al2O3, assumedly due to the dispersive effect of alumina. In accordance with the coupling between absorbance and luminescence decrease (Fig. 7), which was found for their temperature dependence, the effect of composition on the change in the luminescence behavior after thermal reduction follows a similar trend (Fig. 9). The decrease in blue luminescence is less pronounced for glasses with higher [Al2O3] following heat-treatment in 10/90 H2/N2 than for glasses with lower [Al2O3], indicating less effective reduction of Bi 3+ species. It should be noted that treatment temperatures were chosen by Tg-scaling for the respective glass (Table 1) in order to keep the same configurational excitation degree for various glasses. By

Fig. 8. Difference in optical absorbance at 400 nm between untreated and treated samples (Δabsorbance@400 nm) as a function of alumina content for different heat-treatment conditions (indicated in the figure).

doing so, a reasonable comparison in optical changes can be made among different glass compositions. Tg increases with increasing Al2O3 content, and hence the heat treatment temperature should be correspondingly increased. As mentioned above, increasing treatment temperature can also lead to further diminishing of blue luminescence. The observed decrease in the effect on blue luminescence with increasing Al2O3 content is therefore driven by neither viscosity (equivalent for all treatment conditions) nor treatment temperature, but could instead be caused by structural variations, free volume, and the presence of hydroxyl groups formed by the permeation of hydrogen into the glass. The formation of colloids by striking has been found to occur by gradual reduction of bismuth in all available oxidation states [2]. The observed tendency in luminescence (Fig. 9) may be explained

Temperature (K) Untreated

Ex. 310 nm

810

c

795

780

765

o

ln(k) (ln(-Δintensity)

487 C o

Intensity (a.u.)

512 C o

534 C

2

R =0.9941

5.7

-0.9

5.4 2

R =1 -1.2

5.1 350

400

450

500

550

-0.6

0.0012375

0.0012650

0.0012925

ln(k) (ln(ΔAbs@400nm))

a

0.0013200

1/Temperature (1/K)

Wavelength (nm)

b Absorbance

1.5

o

534 C 512 o C

1.0

487 o C Untreated

0.5

0.0

400

600

800

1000

Wavelength (nm) Fig. 7. a) Photoemission spectra after excitation at 310 nm for as-melted Al15 samples and after treatment for 4 h in 10/90 H2/N2 at 487 °C, 512 °C and 534 °C respectively. b) Optical absorption spectra of Al15. c) Arrhenius plot of the absorbance change and the change in luminescence intensity. The lines represent linear fits to the averaged data. Coefficients of determination (R2) are given.

3198

K.H. Nielsen et al. / Journal of Non-Crystalline Solids 358 (2012) 3193–3199

Flourescence rest peak intensity (%)

16

4. Conclusions

2hours

12

8

4hours 4

ex: 310nm 0

0

5

10

15

Al2O3 content (mol%) Fig. 9. Maximum visible fluorescence intensity relative to that of the as-prepared sample as a function of alumina content for glasses treated at Tr in 10/90 H2/N2 atmosphere. Labels indicate treatment duration. Fluorescence spectra were recorded after excitation at 310 nm. Lines are guides for the eye.

by the fact that more Bi 5+ species are available for being reduced in the glasses with higher basicity and thus act as a buffer for Bi 3+ reduction.

In summary, we studied the influence of thermal annealing in hydrogen-bearing atmosphere on the oxidation state of bismuth in sodium aluminosilicate glasses with varying optical basicity. Bi3+ and Bi2+ have been detected in the as-melted glasses by fluorescence spectroscopy. Prior to hydrogen treatment, broadband near-infrared (NIR) fluorescence only occurs in the glass system with the lowest optical basicity. Thermal reduction of the glass samples in 10/90 H2/N2 atmosphere at temperatures around Tg leads to a decrease in both the Vis light transmission and the NIR luminescence. Ultimately, thermal reduction results in the formation of metallic bismuth particles near the specimen surface. The decrease of the Vis transparency upon thermal reduction is thus ascribed to light scattering by these particles. The optical absorbance at 400 nm increases with increasing thermal reduction time. This is well-described by the tarnishing model, indicating that particle formation and light scattering are governed by hydrogen permeation. Both the photoemission intensity and the Vis light absorbance increase with increasing thermal reduction temperature and following Arrhenian behavior. The changes which are induced in the glasses by thermal reduction are only partially reversible upon re-annealing of the glass samples in oxidizing atmosphere (e.g., air) on a time-scale of up to 44 h.

3.5. Re-annealing and reversibility

Acknowledgments

The effect of re-annealing in air on the optical properties of a previously reduced sample is shown in Fig 10. It is found that the effect of thermal reduction is only partly reversible and occurs on a different time scale. The majority of the recovery of fluorescence intensity and decrease of coloration is obtained already within about one hour of treatment time. This indicates a relatively fast reaction during this first stage of re-annealing in air. As suggested in earlier studies [53,55], melting and evaporation of bismuth particles could occur in this stage. In this case, the increase in luminescence intensity would be related to the decrease of optical absorption, and hence to more efficient light propagation to Bi 3+ species in deeper regions of the sample. However, the change in absorbance is not in accordance with Ref. [55], in which light transmission at 560 nm was found to change during annealing, but to be reversible upon cooling. As noted before, both luminescence and optical transmission recover only partly in the first stage of re-annealing. Even prolonged annealing time in air up to 44 h did not result in further reoxidation (Fig. 10). This could be a direct result of the expected lower oxygen permeation rates relative to hydrogen permeation rates [56].

Financial support by the Germany Science Foundation (grant no. WO 1220/2) is gratefully acknowledged. KHN further acknowledges the European Region Action Scheme for the Mobility of University Students (ERASMUS) for enabling his stay at the University of Erlangen-Nuremberg. We thank Ralf Keding (Aalborg University), Ning Da, Sebastian Krolikowski and Guojun Gao (all at university of Erlangen-Nuremberg) for their technical assistance in glass preparation and luminescence spectroscopy.

1.4

Absorbance

1.0 0.8 0.6

Ex: 310nm

Intensity (a.u.)

Untreated Reduced Reannealed 2 hours Reannealed 44 hours

1.2

350

400

450

500

550

Wavelength (nm) 0.4 0.2 0.0

400

600

800

1000

Wavelength (nm) Fig. 10. Absorbance spectra of Al15 samples after reduction for 4 h in 10/90 and after subsequent re-annealing in air at Tg for the indicated treatment times. Inset: photoemission spectra of the same samples following excitation at 310 nm.

References [1] W.A. Weyl, Coloured Glasses, fifth ed. Society of Glass Technology, Sheffield, 1999. [2] G. Lin, D. Tan, F. Luo, D. Chen, Q. Zhao, J. Qiu, J. Non-Cryst. Solids 357 (2011) 2312–2315. [3] B. Kusz, K. Trzebiatowski, J. Non-Cryst. Solids 319 (2003) 257–262. [4] M. Peng, C. Zollfrank, L. Wondraczek, J. Phys. Condens. Matter 21 (2009) 285106. [5] M. Peng, G. Dong, L. Wondraczek, L. Zhang, N. Zhang, J. Qiu, J. Non-Cryst. Solids 357 (2011) 2241–2245. [6] K. Murata, Y. Fujimoto, T. Kanabe, H. Fujita, M. Nakatsuka, Fusion Eng. Des. 44 (1999) 437–439. [7] Y. Fujimoto, M. Nakatsuka, Appl. Phys. Lett. 82 (2003) 3325–3326. [8] M. Peng, J. Qiu, D. Chen, Y. Meng, I. Yang, X. Jiang, C. Zhu, Opt. Lett. 29 (2004) 1998–2000. [9] M.B. Volf, Chemical Approach to Glass, Elsevier, Amsterdam, 1984. [10] C. Rüssel, Non-Cryst. Solids 119 (1990) 303–309. [11] M. Peng, N. Da, S. Krolikowski, A. Stiegelschmitt, L. Wondraczek, Opt. Express 17 (2009) 21169–21178. [12] M. Peng, B. Sprenger, M.A. Schmidt, H.G.L. Schwefel, L. Wondraczek, Opt. Express 18 (2010) 12852–12863. [13] M. Peng, L. Wondraczek, Opt. Lett. 34 (2009) 2885–2887. [14] S. Khonthon, S. Morimoto, Y. Arai, Y. Ohishi, J. Ceram. Soc. Jpn. 115 (2007) 259–263. [15] V.O. Sokolov, V.G. Plotnichenko, E.M. Dianov, Opt. Lett. 33 (2008) 1488–1490. [16] R. Cao, M. Peng, L. Wondraczek, J. Qiu, Opt. Express 20 (2012) 2562–2571. [17] E. Ahmed, D. Köhler, M. Ruck, Z. Anorg. Allg. Chem. 635 (2009) 297–300. [18] Y. Fujimoto, J. Am. Ceram. Soc. 93 (2010) 581–589. [19] J. Ren, L. Yang, J. Qiu, D. Chen, X. Jiang, C. Zhu, Solid State Commun. 140 (2006) 38–41. [20] J. Ren, J. Qiu, D. Chen, X. Hu, X. Jiang, C. Zhu, J. Alloy. Compd. 463 (2008) L5–L8. [21] J. Ren, J. Qiu, D. Chen, C. Wang, X. Jiang, C. Zhu, J. Mater. Res. 22 (2007) 1954–1958. [22] X. Jiang, A. Jha, Opt. Mater. 33 (2010) 14–18. [23] S. Khonthon, S. Morimoto, Y. Arai, Y. Ohishi, Opt. Mater. 31 (2009) 1262–1268. [24] S.P. Singh, B. Karmakar, Opt. Mater. 33 (2011) 1760–1765. [25] M. Peng, D. Chen, J. Qiu, X. Jiang, C. Zhu, Opt. Mater. 29 (2007) 556–561. [26] M. Peng, Q. Zha, J. Qiu, L. Wondraczek, J. Am. Ceram. Soc. 92 (2009) 542–544. [27] B.I. Denker, B.I. Galagan, A.M. Musalitin, I.L. Shulman, S.E. Svechkov, E.M. Dianov, Laser Phys. 21 (2011) 746–749.

K.H. Nielsen et al. / Journal of Non-Crystalline Solids 358 (2012) 3193–3199 [28] A.V. Kir'yanov, W. Dvoyrin, V.M. Mashinsky, N.N. Il'ichev, N.S. Kozlova, E.M. Dianov, Opt. Express 19 (2011) 6599–6608. [29] V.G. Truong, L. Bigot, A. Lerouge, M. Douay, I. Razdobreev, Appl. Phys. Lett. 92 (2008) 041908­3. [30] M.M. Smedskjaer, Y. Yue, J. Deubener, H.P. Gunnlaugsson, S. Mørup, J. Non-Cryst. Solids 356 (2010) 290–298. [31] M.M. Smedskjaer, J. Qiu, J. Wang, Y. Yue, Appl. Phys. Lett. 98 (2011) 071911. [32] M. Nogami, K. Watanabe, Y. Ito, H. Ito, H. Nakano, J. Am. Ceram. Soc. 93 (2010) 1663–1667. [33] M.M. Smedskjaer, J. Wang, Y. Yue, J. Mater. Chem. 21 (2011) 6614–6620. [34] J. Xu, H. Zhao, L. Su, J. Yu, P. Zhou, H. Tang, L. Zheng, H. Li, Opt. Express 18 (2011) 3385–3391. [35] J.L. Barton, M. Morain, J. Non-Cryst. Solids 3 (1970) 115–126. [36] C. Estournès, T. Lutz, J.L. Guille, J. Non-Cryst. Solids 197 (1996) 192–196. [37] M.M. Smedskjaer, J. Deubener, Y. Yue, Chem. Mater. 21 (2009) 1242–1247. [38] J.A. Duffy, F.G.K. Baucke, J. Phys. Chem. 99 (1995) 9189–9193. [39] Y. Yue, J. Non-Cryst. Solids 354 (2008) 1112–1118. [40] M.R. Tuzzolo, J.E. Shelby, J. Non-Cryst. Solids 143 (1992) 181–190. [41] K. Arai, H. Namikawa, K. Kumata, T. Honda, Y. Ishii, T. Handa, J. Appl. Phys. 59 (1986) 3430–3436. [42] J.A. Duffy, M.D. Ingram, J. Non-Cryst. Solids 21 (1976) 373–410.

3199

[43] G. Chi, D. Zhou, Z. Song, J. Qiu, Opt. Mater. 31 (2009) 945–948. [44] M.N. Alexander, P.I.K. Onorato, C.W. Struck, G.W. Tasker, D.R. Uhlmann, J. Non-Cryst. Solids 91 (1987) 63–82. [45] M.N. Alexander, P.I.K. Onorato, C.W. Struck, I.R. Rozen, G.W. Tasker, D.R. Uhlmann, J. Non-Cryst. Solids 79 (1986) 137–154. [46] M. Peng, L. Wondraczek, J. Am. Ceram. Soc. 93 (2010) 1437–1442. [47] M. Peng, L. Wondraczek, Opt. Lett. 35 (2010) 2544–2546. [48] T. Murata, T. Mouri, J. Non-Cryst. Solids 353 (2007) 2403–2407. [49] S.P. Singh, B. Karmakar, Mater. Chem. Phys. 119 (2009) 355–358. [50] M.M. Smedskjaer, Y. Yue, J. Non-Cryst. Solids 355 (2009) 908–912. [51] C. Estournès, N. Cornu, J.L. Guille, J. Non-Cryst. Solids 170 (1994) 287–294. [52] L.F. Kirkegaard, M. Korsgaard, Y. Yue, S. Mørup, Glass Sci. Technol. 78 (2005) 1–6. [53] B. Kusz, K. Trzebiatowski, M. Gazda, L. Murawski, J. Non-Cryst. Solids 328 (2003) 137–145. [54] J. Fang, K.L. Stokes, J.A. Wiemann, W.L. Zhou, J. Dai, F. Chen, C.J. O'Connor, Mater. Sci. Eng. B-Adv. 83 (2001) 254–257. [55] E. Haro-Poniatowski, M. Jiménez de Castro, J.M. Fernández Navarro, J.F. Morhange, C. Ricolleau, Nanotechnology 18 (2007) 315703. [56] J.E. Shelby, Introduction to Glass Science and Technology, second ed. The Royal Society of Chemistry, Cambridge, 2005.