1. Introduction

Long persistent luminescence (LPL) is a phenomenon that the photoluminescence (PL) can continue for an obvious length of time when the excitation source is removed. This phenomenon has found a wide range of applications in emergency signing, indoor and outdoor decoration or lighting, optical data storage, temperature sensors, and medical functions [1–4]. During the past few decades, the research has focused mainly on UV or visible-light-induced LPL [5–7]. However, with the development of laser technique, high energy density and ultrafast femtosecond lasers has been frequently used in microfabrication of structures, high storage density recording and preparation of novel functional photonic devices, etc [8,9]. Therefore, femtosecond laser pulses induced LPL attract more attentions. In 1998, Qiu et al. first demonstrated that femtosecond laser pulses could induce LPL in activating ions (Eu²⁺, Ce³⁺, Tb³⁺, Pr³⁺ and Mn²⁺) doped- and oxygen-deficient Ge-doped glasses [10–14]. Their work has demonstrated the possibility of writing a three-dimensional image pattern emitting LPL in glasses, which enables applications in the field of display devices.

In contrast to UV or visible-light-induced LPL, LPL induced by femtosecond laser pulses is a more complicated phenomenon. It has been reported that nanosecond laser pulse gives a local temperature rise to 12400 K for LiF, 11800 K for DKDP, and 11500 K for SiO₂ using black-body radiation [15]. Also using black-body radiation, our team measured the internal temperature of fused silica induced by one femtosecond laser pulse, and it was found that even 20 ns later after the irradiation of a single laser pulse, the internal temperature still remained 4500 K, which can easily lead to the transformation of the host structure. Therefore, extrinsic defects may be introduced during laser irradiation. However, little discussion has been made about extrinsic defects in previous papers. In this work, we paid close attention to the PL features, through which we hoped to find the types of extrinsic defects, and to know if they have impacts on LPL process. And it will be very interesting if defects induced by an ultrashort pulse are stable at room temperature and can be released by another laser, which can find applications in the fabrication of rewritable three-dimensional optical memory devices. Therefore, research on LPL phenomenon induced by femtosecond pulses is not only to elucidate the LPL mechanism but also to give a deeper insight into the dynamic processes of the light-matter interaction.

As an alternative of rare-earth ions, transition metal Mn²⁺ ions have over decades been of large interest as active centers in inorganic phosphors, mainly for lighting applications. They exhibit a 3d⁵ electronic configuration and possess various PL features based on d-d transition, depending on coordination environment [2, 16,17]. Rare earth ions (Eu²⁺, Ce³⁺) were usually co-doped so as to improve the luminescent duration of Mn²⁺ on a basis of energy transfer [5]. However, for economic benefit, it is better to improve Mn²⁺ ions LPL duration without co-doping. It has been reported that Mn²⁺ could exhibit LPL in sodium borate glasses and alumino-phosphofluoride glasses induced by femtosecond laser pulses [12,13]. Unfortunately, their persistent time is typically less than one hour and peak wavelength shorter than 600 nm, which has limited their applications especially in medical functions, for example, as tracer particles for in vivo medical imaging, which demands longer persistent time and emission wavelength [4, 12,13]. In this paper, we select sodium germanate glass as the target host for its low phonon energies, and chemical durability [18]. As Cr³⁺ and Mn²⁺ exhibits the outstanding persistent luminescence in gallate, we also introduced gallium on purpose into sodium germanate glass, hoping to improve Mn²⁺ persistent luminescence properties [19–21]. In particular, we also investigated the LPL properties of Mn²⁺ in the same host glass-ceramics, which has never been reported before.

2. Experimental

8Na₂O–12Ga₂O₃–80GeO₂: xMn²⁺ (x = 0-1.5) samples were produced through conventional melting at high temperature and subsequent quenching. For this, analytical grade reagents Na₂CO₃ and MnCO₃, and high pure GeO₂ and Ga₂O₃ were used as the starting materials. The batches were thoroughly mixed and filled into crucibles. Each batch was then melted in a resistively heated furnace at 1550 °C for 30 min under ambient atmosphere. The melts were poured onto a preheated stainless steel plate and splash-cooled by pressing with another stainless steel stamp. The solidified samples were subsequently annealed at 350 °C for 10 h. Thus we obtained visually transparent and colorless glass samples. For referencing, labels G-x are used, where x stands for the nominal amount of added Mn²⁺ in mol%. Differential scanning calorimetry was used to determine the peak temperature of crystallization, Tc = 695 °C for the precursors. Crystallization treatments of this latter sample were performed at this temperature for holding times 5 h. In the following, these samples are labelled GC-x.

The laser used for this investigation was a 150-fs regeneratively amplified Ti: sapphire laser operating at a wavelength of 800 nm and a frequency of 1 kHz. The output laser beam was focused into the interior of the sample by a 20 × objective lens with a numerical aperture of 0.45. By using an XYZ stage, we can precisely control the irradiated position. All the PL and LPL spectra in this paper were induced at a laser power of 40 mW if not mentioned. The PL spectra were recorded upon excitation in the sample. The LPL spectra were recorded 30 seconds later after a 1 mm × 1 mm area was scanned. The sample was scanned by femtosecond laser pulses at a speed of 1mm/s with an interval of 10 μm.

PL and LPL spectra were recorded by a Horiba Jobin Yvon iHR 550 spectrometer with a Symphony CCD detector. Absorption spectrum was measured on a UV–2600 (Shimadzu) spectrophotometer. All of the measurements were carried out at room temperature.

3. Results and discussion

Figures 1(a) and 1(b) depict the PL features of the Mn²⁺-activated and non-activated samples (x = 0-1.5) under UV-light (254 nm) and femtosecond laser excitation, respectively. The Mn²⁺-activated and non-activated samples exhibited an intrinsic PL band peaking at 3.2 eV with a bandwidth of 0.25 eV under UV excitation, which was assigned to the triplet to singlet (T₁→S₀) transition of intrinsic Ge-related oxygen deficient centers [≡ Ge…Ge ≡, GODC (II)] [22]. When subjected to femtosecond laser pulses, the Mn²⁺-activated and non-activated samples exhibited two extrinsic PL bands around 1.85 and 2.2 eV while the PL band at 3.2 eV could not be observed. In addition to the intrinsic and extrinsic PL bands, the Mn²⁺-activated samples showed similar spectra features under both UV and femtosecond laser excitation, which were characteristic emission of Mn²⁺ ions. For clearer demonstration (taking G-0.2 as an example), the PL band of G-0.2 under femtosecond laser irradiation, was divided into six parts by the deconvolution using Gaussian functions. The bands ([II] and [VI]) are due to extrinsic defects as mentioned above. The band II has the same position (1.85 eV) and bandwidth (0.13 eV) in both G-0 and G-0.2,and so does band VI with peak at 2.2 eV and bandwidth of 0.24 eV. It is obvious that the ratio between Peak II and Peak VI is higher in G-0.2 than in G-0. It is due to the emission band VI overlapping with the shoulder of the excitation band of Mn²⁺ ions which extended to 550 nm, therefore, the emission band VI was partly absorbed by Mn²⁺ ions [2]. The other four bands ([I], [III], [IV] and [V]), which evidenced for the presence of at least four emission centers, could be ascribed to the ⁴T₁→⁶A₁ transition of octahedrally coordinated Mn²⁺ (while tetrahedrally coordinated Mn²⁺ usually results in green emission) [12,16,17]. The positions of the sharp emission peaks ([I], [III] and [IV]), centered at 1.75, 1.88 and 1.98 eV, remained almost constant with different concentration of Mn²⁺, whereas the position of the broad band [V] red-shifted from 2.09 to 1.98 eV with increasing Mn²⁺ content, which was attributed to the increasing interaction of Mn²⁺−Mn²⁺ pairs [2].

 

Fig. 1 PL features of Mn²⁺-activated (G-x, x = 0.05-1.5) and non-activated (G-0) samples under UV-light (254 nm) excitation (a) and femtosecond laser irradiation (b).The Gaussian profiles of the PL spectrum of G-0 and G-0.2 under femtosecond laser irradiation are also shown in the insets of Fig. 1(b).

Download Full Size | PDF

The extrinsic band at 1.85 eV was attributed to non-bridging oxygen hole centers (≡ Ge−O∙, NBOHC) while the other extrinsic band at 2.2 eV was probably due to Ge-related E_δ' center, which was a paramagnetic state with an unpaired electron delocalized over five germaniums [23,24]. Using black-body radiation as above mentioned, the internal temperature of G-0.2 was measured to be 4850 K, 20 ns later after the irradiation of one femtosecond laser pulse. While GODC (II) were reported to be stable below 200 °C and start to decrease from 300 °C, and were almost absent at 500 °C [25]. Therefore, it is safe to say that the structure of the intrinsic GODC (II) was destroyed in the region under femtosecond laser irradiation as the localized temperature (4850 K) induced by femtosecond laser pulse is much higher than 500 °C. As a result, the intrinsic PL band around 3.2 eV which could be seen under UV light was absent in the PL spectra induced by femtosecond laser pulses.

In addition, absorption spectra were recorded in Fig. 2(a) (taking G-0.2 as an example). A small absorption peak at 412 nm could be attributed to the ⁶A₁→⁴A₁/⁴E transition of Mn²⁺ [17, 26]. After femtosecond laser irradiation, absorption in the whole energy range increased and the absorption edge shifted to longer wavelength. The overall enhancement of the absorption band implied that femtosecond laser pulses probably induced more defects than what we have detected, for example oxygen vacancies and gallium and germanium vacancies which act as effective electron and hole trapped centers respectively. The absorption band around 500 nm was extensively enhanced and was probably ascribed to the ⁵E→⁵T₂ transition of Mn³⁺ [26]. Consequently, the irradiated region became purple in color while the un-irradiated region still exhibited colorless with high transparency (Fig. 2(b)) [27]. The red-shift of absorption edge after irradiation was probably due to trapped electron centers, located behind the conduction band and non-bridging oxygen, which binds electron less tightly than bridging oxygen [22]. No measurable loss of absorption on time scales of many minutes to hours after irradiation at room temperature. However, when the irradiated sample was annealed, the absorption band decreased whereas still higher than that of un-irradiated one. The inset of Fig. 2(a) shows the difference absorption spectrum of G-0.2 after femtosecond laser irradiation and after annealing at temperature from 200 °C to 400 °C. It is obvious that the difference absorption spectrum increases with increasing annealing temperature. After annealing at 400 °C, absorption band of Mn³⁺ is scarcely detectable. From the above results, the oxidization of Mn²⁺ to Mn³⁺ after the femtosecond laser irradiation should be a nonlinear optical process as no absorption band in the wavelength region near 800 nm.

 

Fig. 2 (a): Absorption spectra of G-0.2: a: before laser irradiation; b: after laser irradiation; c: annealing at 400°C for 1 hour after laser irradiation. Inset: difference absorption spectrum of G-0.2 after femtosecond laser irradiation and after annealing at temperature from 100 °C to 400 °C. Figure 2(b): photograph of G-0.2 after irradiation.

Download Full Size | PDF

Upon removal of the femtosecond laser pulses, all Mn²⁺-activated glasses exhibit visible LPL as shown in Fig. 3(a). Obviously, LPL duration strongly depends on dopant concentration. Red shift was observed with increasing concentration of Mn²⁺. The sample G-0.2 exhibited the longest LPL lifetime among all samples. And the LPL decay spectra of G-0.2 are depicted in Fig. 3(b). The LPL spectra exhibited the similar shape with its PL band under UV light excitation. Apparently, the two extrinsic defects did not show any LPL feature. In addition, no any visible LPL was seen in non-activated samples. Therefore, Mn²⁺ ions are the only origin of LPL properties. The peak at 1.88 eV was labled as P₁ and 1.98 eV as P_2. The ratio of the peak intensity of P₁ to that of P₂, shown in the inset, was increasing from 0.53 to 0.59 with decay time, which indicated a convolution of slow and rapid decay processes originating from different emission centers [1,2]. The LPL can be seen by naked eye for more than one hour in the dark. It has to be noted that, the LPL in this paper emits from a small area of only 1 mm × 1 mm. The LPL process was thermally assisted. We did not detect any LPL at 0 °C, so we scanned the sample in ice water mixture to make sure that the pre-scanned area remained un-thermally activated and the whole scanned area emit LPL at the same time upon proper temperature.

 

Fig. 3 (a) LPL feature of Mn²⁺-activated glass (2 min later after the removal of exciting laser). (b) LPL spectra of G-0.2 at different delay time from 120 to 525 seconds.

Download Full Size | PDF

We also investigated the LPL property of glass-ceramics resulting from the precursor. We have found that the GC-0.5 exhibit the longest LPL lifetime among all the glass-ceramic samples. Figure 4(a) shows the XRD pattern of the samples G-0.5 and GC-0.5. As expected, there are no obvious diffraction peaks in G-0.5 while sharp diffraction peaks in GC-0.5. This indicates that the glass samples are amorphous and thermal treatment of the glass specimen resulted in the precipitation of a nanocrystalline secondary phase, which corresponds to albite NaAlGe₃O₈ (ICSD #61170). The simulated pattern of NaAlGe₃O₈ (ICSD#61170) is shown at the bottom of Fig. 4(a) for reference. We use Scherrer’s equation to obtain a rough estimate of the corresponding crystallite size. The strongest diffraction peak with 2θ = 27.021 was selected for the calculation, yielding an average size of 60 nm for GC-0.5. For GC-0.5, the morphology and size distribution of those nanocrystals was also analyzed by HR-TEM in our previous work, where irregular crystallites with size of 20-50 nm were observed [2].

 

Fig. 4 (a) XRD pattern of the samples G −0.5 and GC-0.5. The simulated pattern of NaAlGe₃O₈ (ICSD#61170) is shown at the bottom for reference. (b) PL spectra under femtosecond laser irradiation and UV excitation as well as the LPL feature (30 seconds after removal of exciting laser) of GC-0.5. (c) Samples G-0.2 and GC-0.5 and their emission states of LPL (30 seconds after the removal of exciting laser).

Download Full Size | PDF

The PL spectra of GC-0.5 under femtosecond laser irradiation and UV-light (254 nm) excitation as well as the LPL feature are plotted in Fig. 4(b). A green emission band appeared at 2.38 eV in both the PL spectra, indicating re-precipitation of Mn²⁺ in a less-strong ligand field (tetrahedral coordination) [28]. GC-0.5 also exhibited intrinsic PL band at 3.2 eV under UV light and could not be detected under femtosecond laser irradiation. In addition, the PL bands of Mn²⁺ at 1.88 eV and 1.75 eV under femtosecond laser irradiation are much stronger than that under UV light excitation. However, such phenomenon did not appear in the glass sample. Furthermore, the extrinsic band at 1.85 eV was strongly enhanced comparing to that of glass samples, implying that femtosecond laser pulses gave rise a much higher number density of such defects in glass-ceramics than in glass samples. Noteworthy, the green PL band was absent in LPL spectra. Figure 4(c) displayed the photographs of G-0.2 and GC-0.5 and their emission states of LPL 30 seconds after the removal of the exciting laser.

In Qiu’s work, the intensity of the LPL decreased in inverse proportion to the time. However, in this experiment, we found that the decay curve is more fitted by the equation of $I\text{(}t\text{)=}{I}_0/{\text{(1+γ}t\text{)}}^{\text{n}}$, where I(t) is the afterglow intensity, I₀ and t are initial afterglow intensity and time, and γ and n are fitting parameters, respectively [29]. The decay curve of G-0.2 and GC-0.5 were shown in Fig. 5(a), and their fitted n equaled to 0.55, and 0.59 respectively. The starting point of the decay curve was 30 seconds after the removal of the exciting laser. The power dependence of the integrated LPL intensity (30 sec after removal of exciting laser) was also measured. The LPL intensity increased fast with increasing power less than 10 mW and then remained almost unchanged to 160 mW for G-0.2. When the power was higher than 160 mW, lasers could not be focused within the sample, instead the surface was damaged. As for GC-0.5, its LPL intensity increased slower with increasing power and reached its maximum at 40 mW. And then the intensity started to decrease from 60 mW. It was obvious that the maximum intensity of GC-0.5 was much less than that of G-0.2. However, the UV-induced LPL was more efficient in crystalline structure than in glassy network [2]. We thought that femtosecond laser pulses gave rise to the localized crystallization of glassy structure while the already crystalline structure of glass ceramics was destroyed upon laser irradiation. Therefore, femtosecond laser-scanned glass and glass-ceramics were subjected to XRD measurement. Unfortunately, no crystallization was observed in glass, while the crystalline structure in glass-ceramics was really destroyed since the scanned area shows no diffraction peak. Nevertheless, LPL properties depend on a variety of factors, e.g. density of defects, the depth distribution of defects, the structure of the sample, etc. The reasons should be investigated further.

 

Fig. 5 (a) Decay curves and (b) power dependence of the integrated LPL intensity of G −0.5 and GC-0.5. (30 seconds after the removal of exciting laser).

Download Full Size | PDF

In this paper, both glass and glass-ceramics are dielectric materials, the conduction band is initially empty and the matter is transparent to the laser light. Electrons from the valence band must be first transferred in the conduction band by nonlinear ionization processes, such as multi-photon ionization or tunneling ionization [30]. As the green PL band was absent in LPL spectra, it indicated that the LPL mechanism was not due to a tunneling process, in which electrons were thermally activated from traps and tunneled through the conduction band and then captured by the activator ions. If the tunneling process was applied, the green PL band of tetrahedrally coordinated Mn²⁺ should be in the LPL spectra since the higher-lying excited level, from which this emission occurs, is much closer to the conduction band as compared to the gap between the red-emitting level and the conduction band. On the basis above mentioned, we propose a mechanism responsible for the generation of LPL, as schematically shown in Fig. 6.

 

Fig. 6 Schematic of the mechanism of persistent luminescence processes in the Mn²⁺ doped glasses and glass-ceramics.

Download Full Size | PDF

Upon irradiation with femtosecond laser pulses, Mn²⁺ ions were excited into the excited states (path 1), lying in the conduction band. At the same time, electrons in the valence band were also excited into the conduction band (path 2). Both path 1 and path 2 are via multi-photon ionization or tunneling ionization. Path 1 and Path 2 creates the electrons with low energy. These electrons were subsequently excited in the conduction band by the light field (inverse bremsstrahlung) until a critical energy for collisional excitation is reached [30]. In collisional excitation, one electron from the highest conduction-band state collided with an electron from the valence band and produced two electrons in the lowest-lying conduction-band state, which was the basis of avalanche process [30]. Some electrons, in the lower end of the conduction band, transited to shallow defects D_s (path 3) while other portion of electrons transited to deep traps (path 4). When excitation source was removed, electrons started to escape from the traps. The activated electrons were directly trapped on the excited states of octahedrally coordinated Mn²⁺ (path 5). Once trapped, they recombined with the parent ions (path 6), emitting their potential energy in the form of red light. Path 5 enabled that activated electrons was not trapped by higher-lying excited level of tetrahedrally coordinated Mn²⁺. As a result, there was no green LPL. However, electrons in deep traps could not be stimulated at room temperature. Therefore, the content of Mn³⁺ increased after laser irradiation. Once annealed, electrons in D_d were thermally heated and they were captured by Mn³⁺, leading to the decrease of the absorption band of Mn³⁺. The defects might prefer gathering in the vicinity of manganese sites. It, therefore, enabled energy transfer possible from the defects to the activators.

4. Summary

Femtosecond laser-induced PL and LPL phenomenon in Mn²⁺-activated sodium gallium germanate glass and glass-ceramics were carefully investigated. The LPL, with emission wavelength of deep red, can be seen by naked eye from an area of only 1mm² for more than one hour in the dark at room temperature, which enabled applications especially in medical functions, for example, as tracer particles for in vivo medical imaging. By scanning the sample in proper temperature, we can make the scanned area emit LPL at the same time. If we scanned the sample to an area of more than 1cm², the LPL can be observed for several hundred hours. Femtosecond laser pulses also induced two extrinsic PL bands at 1.85 and 2.2 eV, which were associated with non-bridging oxygen hole centers (NBOHC) and probably Ge-related E_δ' center respectively. The induced extrinsic defects did not exhibit LPL, and the LPL were attributed to the ⁴T₁→⁶A₁ transition of Mn²⁺ ions. In glass ceramics, a green emission band appeared in PL spectra while absent in LPL spectra. Accordingly, we proposed a LPL mechanism, in which the de-trapped electrons were directly trapped on the excited states of Mn²⁺ instead of tunneling through the conduction band first. The LPL was power dependent, and glass exhibited more efficient LPL than glass-ceramics. However, we could not assess the depth of the electronic traps using thermo-luminescence spectra as the defects induced by femtosecond laser were temperature dependent. It was uncertain whether the intrinsic defects or extrinsic defects played a more important role in LPL process. Therefore, detailed mechanisms should be investigated further.