1. Introduction

Pandora’s box of bismuth has been opened slowly since first finding of strange red luminescence from BaSO₄:Bi in the end of nineteenth century, about a century later confirmed due to a less known state Bi^{2+ [}1]. The opening has brought us surprises, hopes and sometimes confusions. This mainly comes from the diversity of the element in valence states and easy formation into clustering entities [2–6]. It, on the other hand, endows the element of bismuth with unpredicted optical properties. For instance, the peculiar divalent bismuth when stabilized in crystals can have ultraviolet or blue absorptions and intense orange or even red luminescence [1,3,7–11]. This enables a potency to improve color rendering index of commercial white light LEDs when combined. Oscillation strength of ²P_1/2→²P_3/2 of Bi²⁺ absorptions was reported even larger than 0.03 [9], comparable to Eu²⁺, though the transition is theoretically forbidden in parity. It’s possibly because the transition probability has been strongly promoted by intense coupling to surrounding crystal field as the ion of Bi²⁺ is built into the condensed solids. For another instance, very recently, polycation Bi₅³⁺ was found unexpectedly luminescing in mid infrared [6]. It is also weird the luminescence quenches in chloride glass but not crystalline sample at room temperature [6,12]. For one more instance, in last decade a new family of bismuth doped materials has been found and soon become a hot research area because of attractive optical properties [5,13–17]. It features in extraordinary absorptions spectrally ranging from UV, visible to near infrared (NIR), and superbroad emissions in 1000 to 1600nm [13–21]. And it can provide net optical gain in the regime where traditional rare earth cannot work, therefore it has been considered for developing new type of laser devices and next generation superbroad fiber amplifiers [22–30]. So far, inspiring reports have been published subsequently on the devices [26–30]. Nevertheless, there is still no direct and convincing evidence to prove which exact valence state of bismuth is responsible for the NIR emission, though a consensus has been made that the valence should be lower than + 2 [5,9,31–34]. Only recently, bismuth doped glasses have been demonstrated to convert lights in 200-900nm uniquely into NIR lights, implying a possible application as solar spectral converters or concentrator [35]. The NIR bismuth emission centers can be even in situ created by femtosecond laser illumination besides treating in reducing conditions, such as CO [9,36,37]. In some cases different bismuth species coexist in the sample and they could be easily converted into each other. Coexistence of non- or multiple- NIR emission centers can result in additional optical loss of bismuth based fiber devices, and deteriorate the beam quality of output laser [23–25]. In bismuth doped pure silica or aluminosilicate fibers, Bufetov et al have clearly identified that there exist different but single types of bismuth active centers (BAC) [26–29]. BAC in doped pure silica fiber, when uncooled, demonstrates high laser efficiency of 58%; another type of BAC in aluminosilicate fiber exhibits lower efficiency of ~20% [27,28]. Though in the presence of single type of NIR emission centers the efficiency of the lasers can be different depending on their properties, it is desirable to precipitate a single type of NIR emission centers in the materials to improve the quality of laser beam. For that, more knowledge on how to selectively stabilize a specific Bi emission species for activity at a specific spectral range in different glasses will be preferred. In other content, this can help understand the puzzling nature of NIR emission in a different way.

Normally, valence states of species could be identified easily with traditional instrumentations such as XPS, EXAFS or EELS. However, these become less sensitive or reliable for heavy metal especially for bismuth. Possibly due to the easy conversion between bismuth species and lower melting point of bismuth, sometimes artifact can be caused unintentionally during measurements. For instance, we found after XPS the measurement bismuth doped germanate glass has changed into black from original transparent brown. When measuring TEM and EELS of bismuth particles, the particles are readily melted as exposed directly to electron beams. In view of these, alternative techniques should be consulted, of which optical spectroscopy is mainly concerned in this work. This is motivated by the facts that no visually artificial information could be produced during our optical measurements and the technique is hypersensitive to the valence state of activators and the changes of local environment around them. So it will be meaningful to accumulate spectroscopic data of different kinds of bismuth species. As one of systematic studies, here in this work we will focus mainly on Bi³⁺ doped glasses.

Since last century Bi³⁺ doped crystals have been noticed as one of series works on fluorescence properties of ions with ns² configuration [2,4,38–43]. Unlike the Bi³⁺ doped crystals, Bi³⁺ doped glasses have been paid less attention partially due to low luminescence efficiency. To today, Bi³⁺ doped borate, silicate, phosphate and germanate glasses have been reported [39–43]. However, these reports mainly emerged between 1970s and 1990s and concentrated on absorption and emission spectra [39–43]. Except these, no systematic studies have been performed. Electronic configuration of Bi³⁺ is [Xe]4f¹⁴5d¹⁰6s². The ground state is ¹S₀ with 6s² configuration, and the excited states from 6s6p configuration are ³P₀, ³P₁, ³P₂ and ¹P₁ in a sequence of energy increasing. The transitions ¹S₀→³P₀ or ³P₂ are spin forbidden and ¹S₀→³P₁ or ¹P₁ are lifted by spin-orbit coupling. So, the latter two have relatively higher absorption strength than the former two. Backward radiative transition ³P₁→¹S₀ is Laporte allowed and the decay time usually is counted between 10⁻⁶ and 10⁻⁸s. For an instance, Blasse et al found LaBO₃: Bi³⁺ and CaSO₄:Bi³⁺with lifetimes of 0.8μs and 60( ± 10) ns, respectively [2,4]. Reisfeld and Boehm reported Bi³⁺ doped germanate glass with a decay time of 350ns at room temperature [41]. However, Parke and Webb found the lifetime of Bi³⁺ emission varying between 2.7 and 3.9μs in borate, silicate and phosphate glasses [43]. The temporal character of the glasses if it was true would be comparable to Bi²⁺ doped crystals, e.g. Sr₂P₂O₇:Bi²⁺ (less than 10μs) [7–11]. This increases the uncertainty to accurately assign a transition of bismuth in specific valence. To avoid this, plus in view of the striking deviation from the reports on Bi³⁺ doped materials and the lack of complete spectroscopic data, it seems necessary to initialize the reinvestigation partly to redefine the lifetime of Bi³⁺ in these glasses.

Here in this work, we have found the emission lifetime of Bi³⁺ doped borate, silicate, phosphate glasses actually does not show obvious divergence from doped phosphors and germanate glass, and it is much shorter than Bi²⁺ lifetime, and in the meantime, we have unveiled the origin of the weak red emission from bismuth doped calcium phosphate glass, which could not be identified by Parke and Webb at that moment [43]. Moreover, the red emission can be observed in bismuth doped magnesium phosphate glass. In the case of accommodation with either alkali or alkali earth oxide none of NIR emission centers can survive in borate and silicate glasses when prepared in air.

2. Experimental procedure

All glass samples were melted in air on a scale of 30g by a conventional melting and quenching method. As a reference, blank specimen was prepared in parallel for each series of glasses. Analytic reagents H₃BO₃, Li₂CO₃, Na₂CO₃, K₂CO₃, MgO, CaCO₃, SrCO₃, BaCO₃, SiO₂, P₂O₅, Al(OH)₃ and Bi₂O₃ were selected as raw materials. Samples were weighed according to molar compositions xM_zO·(100-x-y)B₂O₃·yBi₂O₃ (M = Li, Na, K, Ba; x = 5, 10, 15, 20, 25, 30, 33, 35; y = 0, 0.25, 0.5, 1.0, 1.5, 2.0, 2.5), 30Na₂O·(70-x)SiO₂·xBi₂O₃ (x = 0, 0.5) and 50M_zO·(50-x)P₂O₅·xBi₂O₃ (M = Na, Mg, Ca, Sr, Ba; x = 0, 0.5). Each borate sample was melted in platinum crucibles at 1200°C for 20 min. The samples of phosphate and silicate were melted at 1100°C for 30 min and at 1400 °C for 6h, respectively. Samples 30Li₂O·69.5B₂O₃·0.5Bi₂O₃ and 35Li₂O·64.5B₂O₃·0.5Bi₂O₃ are easily devitrified and 35K₂O·64.5B₂O₃·0.5Bi₂O₃ is quite hygroscopic in ambient. So, these are neglected in this study. Samples were annealed before cut and polished for optical measurements.

Static excitation and emission spectra and emission decay spectra were measured with a high resolution spectrofluorometer Edinburgh Instruments FLS 920 equipped with a red sensitive single photon counting photomultiplier (Hamamatsu R928P) in Peltier air-cooled housing. A microsecond pulsed xenon flashlamp μF900H with an average power of 60W, which can measure decays from 1μs to 10s, was used to measure the decay curves of red emission from bismuth doped phosphate glasses. For measuring lifetime of Bi³⁺ emission in borate and silicate glasses, a hydrogen filled nanosecond flashlamp nF900 applicable to a lifetime of 100ps to 10μs was employed. Fourier transform infrared (FTIR) spectra were recorded on a Bruker Vector 33 spectrometer on samples embedded in KBr pellets. All the measurements were performed at room temperature.

3. Results and discussion

3.1 Bi³⁺ doped borate and silicate glasses

Parke and Webb examined [43] the absorption, emission and decay times of bismuth doped x Na₂O·(100-x)B₂O₃ (x = 10, 15, 20, 30, 35), 20Li₂O·80B₂O₃, 20K₂O·80B₂O₃, 3Na₂O·7SiO₂ and M_xO·P₂O₅ (M = Li, Na, Mg, Ca, Sr, Zn) glasses. For bismuth doped borate glasses 10Na₂O·90B₂O₃ and 15Na₂O·85B₂O₃, they found brown coloration though not uniformly distributed throughout the samples. We extended the study to a wider range xM_zO·(100-x-y)B₂O₃·yBi₂O₃ (M = Li, Na, K, Ba; x = 5, 10, 15, 20, 25, 30, 33, 35; y = 0, 0.25, 0.5, 1.0, 1.5, 2.0, 2.5), however, we could only get clear colorless samples in the series. For bismuth doped Na₂O-B₂O₃ binary glasses, only weak luminescence exists when soda content x increases from 5 to 20, and it is intensified markedly as x keeps increasing to 35, as can be seen in Fig. 1 . This basically agrees with what Parke and Webb found in the glasses. They noticed that, only when x is larger than 15 can relatively stronger fluorescence be recorded. They also found absorption peak depends more on the concentration than the nature of alkali ion. For instance, the absorption peaks at 230nm (43500cm⁻¹ due to the transition of ¹S₀→¹P₁) same in 20Li₂O·80B₂O₃, 20Na₂O·80B₂O₃ and 20K₂O·80B₂O₃ glasses. However, it moves towards lower energy side (225nm (44400 cm⁻¹) → 240nm (41700cm⁻¹)) as the concentration of alkali increases from 15 to 35 in sodium borate glasses [43]. Nevertheless, this comment does not hold all true for excitation and emission peaks, as we will discuss below (see Figs. 1-3 ). Because of instrumentation limit, excitation in the spectral range of 200 to 250nm cannot be recorded meaningfully.

 

Fig. 1 Excitation (1’, 2’, 3′, λ_em = 429nm) and emission (1, 2, 3, λ_ex = 305nm) spectra of xNa₂O·(99.5-x)B₂O₃·0.5Bi₂O₃ (x = 25, 30, 35) glasses.

Download Full Size | PDF

 

Fig. 3 (A) Excitation (a, λ_em = 396nm) and emission (b, λ_ex = 299nm) spectra and (B) fluorescence decay curve (λ_ex = 299nm, λ_em = 395nm) of 30Na₂O·69.5SiO₂·0.5Bi₂O₃ glass. Inset depicts the sample image after annealed but before polished.

Download Full Size | PDF

Figure 1 denotes the excitation and emission spectra of xNa₂O·(99.5-x)B₂O₃·0.5Bi₂O₃ (x = 25, 30, 35) glasses. When x increases, emission and excitation peak exhibits different behaviors. The emission shifts blue though slightly (430nm → 428nm → 426nm), while the excitation obviously shifts red (309nm → 315nm → 319nm, see Fig. 1). At the same time stokes shift decreases (9106cm⁻¹ → 8382 cm⁻¹ → 7874 cm⁻¹). Same scenario does appear in bismuth doped Li₂O-B₂O₃, K₂O-B₂O₃ and BaO-B₂O₃ glasses. Results are collected in Table 1 . From Figs. 1-2 and Table 1 we can notice that changes of alkali or alkali earth content doesn’t cause significant shift of emission peaks for the glass series of BaO-B₂O₃ and Na₂O-B₂O₃ particularly when comparing with the cases of Li₂O-B₂O₃ and K₂O-B₂O₃. For glasses with equal content of alkali oxide, vis. 25M₂O·74.5B₂O₃·0.5Bi₂O₃ where M stands for Li, Na and K, the emission of Bi³⁺ shifts towards low energy in sequence of Li → Na → K. Similar case happens in NIR emissive bismuth doped glasses [5,19–21,44].

Table 1. Excitation (λ_ex in nm), emission (λ_em in nm), Stokes shift (SS in cm⁻¹), Huang-Rhys parameter S and fluorescence lifetime (τ₁ in ns by this work; τ₂ in μs by Parke and Webb) of Bi³⁺ doped glasses. – means unobserved or unreported.

View Table | View all tables in this article

 

Fig. 2 Excitation (curves 1’, 2’, 3′, λ_em = 438nm; 4’, 5′, λ_em = 416nm) and emission (curves 1, 2, 3, 4, 5, λ_ex = 302nm) spectra of xK₂O·(99.5-x)B₂O₃·0.5Bi₂O₃ (x = 20, 25, 30) and xLi₂O·(99.5-x)B₂O₃·0.5Bi₂O₃ (x = 20, 25) glasses.

Download Full Size | PDF

Figure 3(a) denotes excitation and emission spectrum of 30Na₂O·69.5SiO₂·0.5Bi₂O₃ glass. It reveals one excitation peak at 299nm and one emission peak at 395nm for the transparent colorless sample (see inserted image). The emission peak differs from 385nm by Parke and Webb [43]. The Stokes shift is 8129cm⁻¹, comparable to 8382cm⁻¹ of 30Na₂O·69.5B₂O₃·0.5Bi₂O₃ but smaller than 9478cm⁻¹ of 30K₂O·69.5B₂O₃·0.5Bi₂O₃ glass.

When measuring lifetime of Bi³⁺ in borate and silicate glasses, Parke and Webb selected pulsed exciting light with pulse width, that is, FWHM of 200ns and they obtained decay time ranging from 2.7 to 3.9μs [43]. In this study based on the technology of time correlated single photon counting, we chose hydrogen filled gas nanosecond flashlamp with typical pulse width of 1ns. Typical decay curve of Bi³⁺ emission of doped silicate glass is illustrated as Fig. 3 (b). And it complies well with single exponential decay equation, fit to which, respectively, yields 427ns and 475ns for the two glasses. The rest data are summarized as column “τ₁” in Table 1. The typical lifetime of Bi³⁺ doped borate and silicate glasses is around 500ns.

Besides types and concentrations of alkali or alkali earth, dopant concentration also affects excitation and emission of Bi³⁺ as shown in Fig. 4 . Increase in concentration leads to weakening intensities of emission and excitation peaks and simultaneously the redshift of these peaks. The excitation peak varies from ~309nm to ~330nm, and the emission moves from ~425nm to 440nm, perhaps due to self absorption of Bi³⁺. Lifetimes lie between 470ns and 516ns. This is very similar to NIR emission Bi doped germanate glasses, but the dependence of optical properties on bismuth content is much stronger. For example, increase of bismuth content from 0.01% to 2% even leads to the shift of emission from 1100nm to 1310nm [45].

 

Fig. 4 Excitation (curves 1’, 2’, 3′, 4’, 5′, 6’, λ_em = 430nm) and emission (curves 1, 2, 3, 4, 5, 6, λ_ex = 305nm) spectra of 35Na₂O·(65-x)B₂O₃·xBi₂O₃ (x = 0.25, 0.5, 1.0, 1.5, 2.0, 2.5) glasses.

Download Full Size | PDF

3.2 Mechanism for the compositional dependence of Bi³⁺ emission in the borate and silicate glasses

On the basis of spectroscopic data reported on Bi³⁺ doped crystals and glasses [2,4, 38–43], it can be noticed that, the first excitation band clearly corresponds to the dominate absorption transition from ¹S₀ to ³P₁ at either room temperature or even lower temperature for instance 4.2K [40], and the nature of emission transition strongly depends on temperature. After electrons have been raised to some vibrational level of the excited state of ³P₁, part of them will, at lower temperatures, relax to the lower lying ³P₀ state via a radiationless transition and therefore the forbidden ³P₀→¹S₀ transition can be observed. However, at higher temperatures, electrons on the state of ³P₀, if there has ever been any electron populated on the state, will tend thermally depleted to ³P₁. As a consequence the emission due to the transition of ³P₁→¹S₀ preponderates [39–41]. So, at room temperature the excitation peak in borate and silicate glasses (see column “λ_ex” of Table 1) is attributed to the transition of ¹S₀→³P₁, and the emission is to ³P₁→¹S₀ (see column “λ_em” of Table 1)_.

Inspecting Figs. 1-2 and Table 1 shows that, as stated earlier for Bi³⁺ doped alkali or alkali earth borate binary glasses, excitation peak regularly shifts red and emission peak shifts blue with concentration of alkali or alkali earth increasing. As a result, Stokes shift decreases. How does all this happen? To elucidate it, information on glass microstructural changes is helpful.

Figure 5 shows FTIR spectra of xBaO·(99.5-x)B₂O₃·0.5Bi₂O₃ (x = 20, 25, 33) and xNa₂O·(99.5-x)B₂O₃·0.5Bi₂O₃ (x = 25, 30, 35) glasses. The spectra reveal the almost same trend, that is, as the content of modifier ions increases, amount of BO₄ anion groups grows at the expense of BO₃, possibly due to the introduction of additional oxygen along with the modifiers. This should be the reason why the Stokes shift decreases. It’s known that the transitions of Bi³⁺ occur between 6s² and 6s6p configurations [2, 3, 8–43]. Unlike rare earth, they are easily influenced by ligand fields around Bi³⁺ ion. Coupling strength of electron and phonon can be qualitatively measured by Huang-Rhys factor S and it can be estimated from S = 1/2(SS/ħω + 1), where ħω is the maximum phonon energy, and SS means Stokes shift [8, 11]. Here, if the maximum phonon is taken as ~1350cm⁻¹ corresponding to typical B-O stretching vibration of trigonal BO₃ unit in borate glasses, S can then be calculated and listed in Table 1. Based on the data above, schematic configurational coordinate diagrams are constructed for two extreme cases: (a) lower content of alkali oxide with higher S and (b) higher content of alkali oxide with lower S, as denoted by Fig. 6 (a) and 6(b), respectively. For each series of doped samples M₂O-B₂O₃ (M = Li, Na, K), for case (a), Bi³⁺ ion will experience stronger interaction with environment, and it leads to a bigger offset ΔQ between ¹S₀ and ³P₁, and shorter wavelength absorption, longer wavelength emission, and hence larger Stokes shift. Case (b) is opposite to case (a) and it results in smaller Stokes shift. Figure 6 can also qualitatively apply to Bi³⁺ doped glass of BaO-B₂O₃.

 

Fig. 5 FTIR spectra of xBaO·(99.5-x)B₂O₃·0.5Bi₂O₃ (x = 20, 25, 33) and xNa₂O·(99.5-x)B₂O₃·0.5Bi₂O₃ (x = 25, 30, 35) glasses

Download Full Size | PDF

 

Fig. 6 Configurational coordinate diagrams of bismuth doped alkali borate glasses: (A) for lower and (B) for higher content of alkali oxide. ³P₀ and ³P₂ states are marked as broken lines since the transitions from ¹S₀ to the states are forbidden. Horizontal lines denote the vibrational levels. Blue solid lines refer to the nonradiation relaxations. Upward red solid lines refer to the excitation processes and downward black solid lines to the emission process. ΔQ means the offset of parabolas.

Download Full Size | PDF

3.3 Red emission from bismuth doped phosphate glasses

Parke and Webb noticed red emission from bismuth doped calcium phosphate and did not perform more optical measurements [43]. But they did suggest the red emission seems unlikely to be the simple transition of ³P₁→¹S₀ of Bi³⁺. The origin of the emission cannot be identified at that time. Here, beside calcium phosphate glass, the red emission can be observed in magnesium phosphate glass, though relatively weaker. It cannot be recorded in 50Na₂O · 49.5P₂O₅ · 0.5Bi₂O₃, 50Sr₂O · 49.5P₂O₅ · 0.5Bi₂O₃, and 50BaO · 49.5P₂O₅ · 0.5Bi₂O₃ glasses when prepared in air.

Figure 7 depicts the excitation, emission and decay curves of 49.5CaO · 50P₂O₅ · 0.5Bi₂O₃ (CAB, hereafter) glass. Red emissions indeed were observed in the glasses and they slightly depend on excitation wavelengths, as curves c-d of Fig. 7(a) show. Excitations into 285nm, 345nm and 535nm, respectively, lead to an emission peaking at 712nm, 709nm and 709nm. Accordingly, excitation spectra are different though all composed of three peaks when the monitored emissions change (see curves a and b). For the emission at 712nm, excitation peaks lie at 285, 340 and 520nm; and for the emission at 640nm, the peaks at 285nm, 331 and 495nm. And the peak at 285nm always is the strongest. Decay curves were measured for different situations (see exemplarily curves a-c of Fig. 7(b)). For the emission at 712nm and the excitation at 285nm, decay curve follows well simple exponential equation and it produces a lifetime of 4.417μs. Similarly, lifetimes are 6.041μs and 7.082μs for the cases of Em709nm@Ex345nm and Em704nm@Ex500nm. Figure 7 implies the existence of more than a single type of bismuth emission centers in the glass.

 

Fig. 7 (A) Excitation (a: λ_em = 712nm; b: λ_em = 640nm) and emission (c: λ_ex = 285nm; d: λ_ex = 337nm; e: λ_ex = 516nm) spectra and (B) Decay curves (a: λ_em = 712nm, λ_ex = 285nm, τ = 4.417μs; b: λ_em = 709nm, λ_ex = 345nm, τ = 6.041μs; c: λ_em = 704nm, λ_ex = 500nm, τ = 7.082μs) of 49.5CaO·50P₂O₅·0.5Bi₂O₃ glass.

Download Full Size | PDF

The excitation, emission and decay curves of 49.5MgO · 50P₂O₅ · 0.5Bi₂O₃ (MPB) glass are quite analogous to the glass CPB. The emission peaks of the MPB glass appear at shorter wavelength when compared to CPB glass, and they strongly depend on excitation wavelengths. Upon excitations at 281, 335 and 507nm the emission shifts from 694nm, 681nm to 676nm. Excitation spectrum of the emission at 694nm comprises three peaks at 281nm, 335nm and 507nm, similar to the glass CPB. Lifetimes are measured to be 4.799μs, 6.351μs and 7.182μs for the cases of Em694nm@Ex281nm, Em681nm@Ex335nm and Em676nm@Ex507nm, respectively.

3.4 The origin of red emission from bismuth doped phosphate glasses

The features of excitation and emission spectra and fluorescence lifetime of CPB and MPB glasses are distinctly different from Bi³⁺ doped either glasses or crystals. The excitation spectrum spans from ultraviolet to visible, and the emission appears at longer wavelength of ~700nm, and the lifetime is about 10 times longer than Bi³⁺, and about 100 times shorter than bismuth NIR emission centers (typically several hundred microseconds) [2,4,13–21,31–37]. This means that red luminescence center in bismuth doped CPB and MPB glasses intrinsically differs from Bi³⁺ or bismuth NIR emission centers. It, however, highly resembles Bi²⁺. Table 2 summarizes the spectroscopic data of Bi²⁺ doped crystals and glasses reported so far in literatures [1,3,7–11].

Table 2. Excitation (λ_ex in nm), emission (λ_em in nm) and fluorescence lifetime (τ_Bi2+ in μs) of Bi²⁺ doped glasses and crystals. “–” means unobserved or unreported.

View Table | View all tables in this article

As shown in the table, three absorption peaks corresponding to ²P_1/2 → ²P_3/2 (1), ²P_1/2 → ²P_3/2 (2) and ²P_1/2 → ²S_1/2, can be usually detected in Bi²⁺ doped compounds. Accordingly, emission occurs via ²P_3/2 (1) →²P_1/2 [1,3,7–11]. The absorption of ²P_1/2 → ²S_1/2 lies between 231 and 286nm, and the two absorptions of ²P_1/2 → ²P_3/2 (1) and ²P_1/2 → ²P_3/2 (2) can be separated from each other in the crystals and they appear in the spectral range of 380 to 648nm. The emission of ²P_3/2 (1) →²P_1/2 appear in 586 to 716nm with a lifetime varying between 8 and 25μs. In the compound of Ca_1.994P₂O₇:0.3% Bi, since there is more than one type of Bi²⁺, two absorption peaks of the ²P_1/2 → ²S_1/2 transition were observed at 237 and 281nm [9]. This similarity encourages us to assign the absorptions at 285nm and 340nm in CPB glass and 281nm and 335nm in MPB glass to the transition from ²P_1/2 to ²S_1/2, since as mentioned above more than one type of Bi²⁺ exists in these glasses. The lifetime of CPB and MPB glasses is in the same order as Bi²⁺ in crystals. We attribute the broad absorption covering the range of 400 - 650nm to ²P_1/2 → ²P_3/2 (1) and ²P_1/2 → ²P_3/2 (2) and the red emission of CPB and MPB glasses to ²P_3/2 (1) →²P_1/2. Unlike Bi²⁺ doped crystals, the transitions ²P_1/2 → ²P_3/2 (1) and ²P_1/2 → ²P_3/2 (2) cannot be well resolved, and it is possibly because Bi²⁺ in the glasses experiences stronger coupling strength from lattice especially as comparing CPB glass to the crystal sample with almost same composition Ca_1.994P₂O₇:0.3% Bi [9]. This may also explain why the emission from CPB glass locates at a longer wavelength than the crystal.

When checking over Bi doped alkali or alkali earth borate glasses and phosphate glasses no NIR luminescence can be detected in the case of 808nm excitation at room temperature. This agrees with foregoing investigations and it confirms again that higher content of alkali or alkali earth oxide is harmful to the stabilization of NIR emission bismuth centers though it is favorable for the luminescence of Bi³⁺ or in some cases Bi²⁺.

4. Conclusions

In all, the reinvestigation on bismuth doped alkali or alkali earth borate, silicate and phosphate glasses has defined that the decay time of Bi³⁺ is typical ~500ns in borate and silicate instead of several microsecond reported by Parke and Webb, and it has revealed that Bi²⁺ ion is responsible for the red emission of bismuth doped calcium phosphate glass, the nature of which Parke and Webb confused about. The red emission corresponds to ²P_3/2 (1) →²P_1/2 of Bi²⁺ and it also occurs in magnesium phosphate glass. None of bismuth doped alkali or alkali earth borate, silicate and phosphate glasses can accommodate bismuth NIR emission centers, perhaps because of high content of alkali or alkali earth oxide in the glasses. It can, however, stabilize Bi³⁺ ion in borate and silicate glasses. The composition dependence of Bi³⁺ optical properties can be explained with the configurational coordinate models. Stronger coupling between Bi³⁺ and lattice can lead to redshift of emission and blueshift of excitation and eventually larger Stokes shift. In either magnesium or calcium phosphate glass only Bi²⁺ ion survives. Beside accumulation of spectroscopic data on different valent bismuth ions, this work can guide us on how to manipulate precipitation of some specific bismuth ions by modulating glass components.