1. Introduction

There has been much interest in samarium (Sm) and europium (Eu) doped glasses due to their efficient luminescence and their persistent spectral hole burning characteristics [1–6]. These ions are most stable in their trivalent state in glasses that have been prepared by conventional glass melting techniques. However, it is well known that, in many host glasses, the trivalent Sm³⁺ and Eu³⁺ ions can be converted to their divalent form (Sm²⁺ and Eu²⁺) upon exposure to high energy radiation. This valence change can be optically detected because the dominant emission bands of trivalent and divalent forms of Sm and Eu can be readily distinguished [7,8]. In the case of Sm, all dominant bands of Sm³⁺ and Sm²⁺ ions are situated in the red region of the spectrum, which means that there is a good match to silicon based detectors used in optical measurements. The valency conversion of these ions has been reported in phosphate, borate and aluminate containing glasses under X-ray, fs-laser, γ and β-irradiation [5,6,8–20]. This conversion is usually accompanied by the formation of defects in the glass which include electron centers and hole centers. Electronic transitions of these defects cause high absorbances in the UV and the visible regions, which results in photodarkening of the glass. As defect centres are paramagnetic, electron spin resonance (ESR) can be used to investigate the nature of these defects. It should be stressed here that the valency conversion has been usually observed in glasses which are host to “oxygen-associated trapped hole centers [21]” such as POHC [22–25], BOHC [8,17,26] or (Al-OHC) [6,20,27,28]. Valency conversion is usually reversible. It has been reported that optical-illumination [26,29,30] as well as annealing the glass at high temperatures [6,29,31] may result in the reverse conversion of divalent to trivalent ions. The photodarkening is usually reversible as well. Annealing or illuminating the sample may reduce the X-ray induced absorption, probably by removing color centres [10,11,28,31–34].

Recently, we demonstrated that the X-ray irradiation induced valency conversion of Sm³⁺-ions in glasses can be a promising dosimetric technique for the measurement of spatially resolved high doses in Microbeam Radiation Therapy (MRT) [30,31,35]. MRT is an experimental form of radiation treatment which guarantees less damage to normal tissue in comparison with other kinds of radiotherapy. This is based on the markedly different responses of tumor and normal cells to this form of treatment. The synchrotron generated X-ray beam is collimated and applied in the form of an array of planar microbeams (typically~20–50 μm width) usually spaced 100–400 μm apart. As a result, the spatial dose distribution has high dose and low dose areas that alternate. While the ‘peak dose’ (~150–600 Gy) provides lethal radiation for damaging tumors, the ‘valley dose’ (~3–30 Gy) spares sufficient minimally irradiated normal tissue, including the central nervous system which has extraordinary resistance to damage. This tissue is capable of repairing the irradiation damaged zones. The exact mechanisms underlying this effect are not well understood. It is suggested that the surviving blood vessels in the valley zones repair the tissue microvasculature through an angiogenesis process; spared tumor tissue on the other hand would be ablated (as suggested) by the migration of lethally irradiated tumor cells to ‘valley zones’. Accordingly, the accurate measurement of peak-to-valley dose ratio (PVDR) is of crucial importance to assure that inadequate normal tissue is maintained [36–41]. However, the accurate, simultaneous recording of peak and valley doses that differ by hundreds of Grays, and the large dose gradients (hundreds of Grays over several microns) in the whole X-ray energy range of interest for MRT (50-250 keV) is an extremely challenging task. No current detector can satisfactorily meet all these requirements [42], and intensive research towards the development of detectors suitable for MRT is currently underway [43–49].

In our earlier work, we examined various Sm³⁺ doped glasses for the presence of Sm³⁺ → Sm²⁺ conversion under the influence of X-ray irradiation for the purposes of developing high-dose high-resolution detector plates suitable for MRT. Among a large variety of glasses we had examined, we found useful Sm³⁺ to Sm²⁺ conversion only in fluoroaluminate (FA) and fluorophosphate (FP) glasses [30,31,35,50]. We showed that both types may be used in the measurement of high-dose to several thousand Grays and provide high spatial resolution required for MRT. The detection is based on the X-ray induced conversion of trivalent Sm³⁺ to the divalent form Sm²⁺. Photoluminescence (PL) spectra of Sm²⁺ ions can be easily distinguished from those of Sm³⁺ ions and hence we can measure the dose which is proportional to response ratio R(t) = PL(Sm²⁺)/PL(Sm³⁺). A side effect of X-ray irradiation is the formation of defect centers and photodarkening of these glasses. In our case, photodarkening is an undesirable effect and makes the calculation of the response ratio complicated as discussed in our previous work [31]. We also showed that the effects of previous X-ray exposure, including the valency conversion of Sm ions, along with photodarkening may be erased by intense optical illumination [30] or annealing at temperatures sufficiently exceeding the glass transition temperature T_g. Annealing at temperatures around or just below T_g results in the stabilization of the Sm²⁺ ionic environment and therefore is not effective for erasure [31].

The defect centers in FP glasses include phosphorus-oxygen hole centers (POHCs) and defects such as PO₂, PO₃ and PO₄ complexes which consist of electrons trapped on phosphate group precursors [22–25,51]. These defects are generally called phosphorus-oxygen electron centers (POECs) [30]. On the other hand, the precise nature of the defects in FA glasses is uncertain as FA glasses are not as well studied in the literature.

In this paper, we investigate different processes occurring in a Sm-doped fluorophosphate glass under the influence of X-ray irradiation, including the formation of defect centers and their correlation with samarium valency conversion. The investigation is based on ESR and optical absorption spectroscopy. We examine the X-ray irradiated FP glasses doped with different Sm-ion concentrations and also study the effect of thermal annealing on defects. We have deliberately chosen to study FP rather than FA glasses inasmuch as FP glasses are among the more thoroughly investigated glasses, and their properties are much better understood. Indeed, FP glasses without RE doping have also been used in dosimetry [52]. FP glasses are therefore a better candidate for optical absorption and ESR studies than FA glasses. Prior knowledge on defects in FP glasses has allowed us to associate different features and bands of the spectra to well known POHC and POEC defects and hence provide a better understanding of the physical processes that take place in these glasses during the Sm³⁺ to Sm²⁺ ion conversion under X-ray irradiation.

2. Experimental

Fluorophosphate (FP) glasses can be thought of as a combination of fluoride and phosphate glasses with a variety of possible cationic species. Samples used in the present study were synthesized and prepared based on the FP10 composition published by Ebendorff-Heidepriem [25]. The FP10 batch composition is given in mol% as 10.0Sr (PO₃)₂-34.4AlF₃-10MgF₂-30.4CaF₂-15.2SrF₂. The FP10 glasses were doped with Sm³⁺ by adding SmF₃ with concentrations varying from 0.001 to 0.5 mol%. Assuming full Sm ionization, it gives us a variation of Sm³⁺ concentration from 0.001 to 0.5 at. %. The quenched glass samples were cut in smaller pieces suitable for ESR and optical absorption spectroscopy experiments. The glass transition temperature T_g of the FP glasses used in this work was measured by using a temperature modulated differential scanning calorimeter (TMDSC), and was found to be approximately 460°C. Annealing experiments were carried out at temperatures 100°C to 550°C using a temperature controlled furnace.

The X-ray irradiation was performed using the emission produced by a commercial FAXITRON X-ray set with a tungsten anode and 0.76 mm Beryllium filtration placed approximately 5 cm from the anode. The X-ray tube operates at 110 kVp (mean energy ~45 keV, calculated using reference [53]) with an approximate dose rate of 50 Gy/min. The quoted dose rate represents dose in air at the surface of the sample, which is the usual manner in which dose is reported for MRT.

The steady-state photoluminescence (PL) spectra were measured from 200 nm to 1200 nm, using an ASEQ fiber input mini-spectrometer with spectral resolution better than 1 nm. The excitation source for all the photoluminescence spectra was a laser diode with an emission wavelength at 405 nm, which can be used to excite both the Sm³⁺ and Sm²⁺ ions [54]. The intensity of excitation was kept as low as possible to minimize the effect of Sm²⁺→ Sm³⁺ reconversion during the measurements. The transmittance spectra were recorded using a Perkin-Elmer Lambda 900 spectrophotometer. The samples were polished flat for these measurements. Electron Spin Resonance (ESR) spectroscopy measurements were carried out using a standard Bruker EMX 10/2.7 instrument working at X-band frequency (~9.8 GHz) so as to obtain the first derivative ESR. All samples were prepared to have the same geometry 1.5 mm × 1.5 mm × 6 mm (to avoid sample shape dependence in ESR measurements) and carefully placed in the same position inside the cavity for each measurement. Further, ESR measurements were also checked for reproducibility. The background signal was recorded and subtracted after each single spectral recording. The ESR signal intensities were normalized to the mass of the samples. ESR measurements were conducted at room temperature, following X-ray irradiation and after each step of thermal annealing.

3. Results and discussion

3.1. ESR spectra

Figure 1 presents a typical ESR signal of an X-ray irradiated Sm-doped FP glass sample. Prior to irradiation, we could not detect any significant ESR signal. It is worth noting the change of scale in Figs. 1(a) and 1(c) (the “wings”) in comparison with Fig. 1(b). In other words, the central part of the ESR signal consists of very strong and narrow lines which are usually associated with POHC. Meanwhile, the weaker wings are commonly related to PO₂, PO₃ and PO₄ complexes which readily capture electrons [22,24,25,51]; we refer to these defects as POEC [30].

 

Fig. 1 The electron spin resonance (ESR) signal of FP glass doped with 0.2% of Sm³⁺ and X-ray irradiated for 2 hours (total dose of ~6 kGy). The spectra were measured after annealing the irradiated sample at 100 °C and cooling back to room temperature. The experimental data (thick solid lines) are approximated by a sum of five doublets and one singlet (symbols). Two doublets (L₁ and L₂) and the singlet L₃ have Lorentzian lineshapes while the other three doublets (Г₁–Г₃) are Gaussians. The singlet and the individual components of each doublet are shown by thin solid lines and are marked by superscript (1) or (2). Note the change of scale (compression over the x-axis and stretching over y-axis by a factor of 50) in the wings, (a) and (c), of the graph. The lower scale is shown for a nominal frequency of 9.85 GHz.

Download Full Size | PDF

For future interpretation and numerical comparisons, we have presented the ESR spectra as combinations of elementary lines such as Lorentzians and Gaussians. In order to have a self-consistent interpretation, we used a unique set of Lorentzians and Gaussians (characterized by positions and widths) for the whole set of ESR spectra obtained in all our experiments. Figures 1(a)-1(c) illustrate typical examples of these efforts. The intense central part of the ESR spectrum is presented as a sum of the first derivatives of five Lorentzians while the weaker signals in the wings use the first derivatives of six Gaussians. The positions and width of these Lorentzian and Gaussian elementary lines are summarized in Table 1.The weighting factors for the latter lines are used as adjustable parameters and their values are discussed below. We wish to stress that we have used the minimum number of lines required to fit all our spectra, and the quality of fitting is illustrated by Figs. 1(a)-1(c).

Table 1. The unique set of Lorentzians and Gaussians (characterized by positions and widths) used for approximating the whole set of ESR spectra obtained in our experiments.

View Table

We have already mentioned that the weaker signals in the wings of ESR spectra are usually associated with PO₂, PO₃ and PO₄ complexes. In our particular case, this implies that the six Gaussians may be interpreted as belonging to three doublets (Г₁, Г₂ and Г₃) related to these complexes. The spin-Hamiltonian parameters of Г₁, Г₂ and Г₃ and their assignments to PO complexes are shown in Table 1.

There is also a good consensus that the strong and narrow central lines are associated with POHC [22,24,25,30,51]. Therefore, in the following we will analyze only the summed and integrated strength of the POHC related ESR signals, which is simply proportional to the total concentration of POHC. However, we made an attempt to deconvolute the ESR signal and obtained five Lorentzians, tentatively belonging to two doublets (L₁ and L₂) and one singlet (L₃) whose spin-Hamiltonian parameters are summarized in Table 1. Tentatively, we assume that the doublets L₁ and L₂ may be attributed to two different types of POHC, i.e. r-POHC and l-POHC [24,51]. In the r-POHC (which was labeled as the stable form of POHC by Griscom [51]), the unpaired spin is shared between the two non-bridging oxygens [23] in the structure. On the other hand, l-POHC was initially reported to be stable only at low temperatures. However, Origlio et al. showed that this structure can be observed at room temperature as well. The weaker singlet with g = 2.030 cannot be found in non-annealed samples, and appears only after thermal treatment. Unfortunately, there is no reported reliable assignment to any particular structural unit. However, it might be tentatively ascribed to so-called OHC (oxygen related centers of unknown structure) [25]. As a conclusion of this discussion, it is worth noting that all Lorentzians strongly overlap. Consequently, despite the above description being the most probable, it may not be unique and needs to be studied further.

3.2 Effect of Sm doping concentration on defects

Figure 2 presents a series of ESR spectra of FP glasses that had been X-ray irradiated for 2 hours. The samples have different concentrations of Sm³⁺. It is clearly seen that the increase in the Sm³⁺ doping concentration (C₀) reduces the ESR signal intensity. Figure 3 shows that the POHC related ESR signal, as well as POEC related Gaussian doublets Г₁, Г₂ and Г₃ decrease exponentially with increasing C₀. However, POEC related components decrease at a faster rate with C₀ in comparison with those of POHC.

 

Fig. 2 Variation of ESR spectra of FP glass samples as a result of changing the concentration of Sm³⁺ (C₀) in the range of 0–0.5 at.%. All the samples were X-ray irradiated for 2 hours prior to the ESR measurement. Symbols are approximation of experimental data based on the approach presented in Fig. 1 and Table1. All the signal intensities are normalized to the mass of the samples.

Download Full Size | PDF

 

Fig. 3 Variation of ESR signal components ascribed to POHC and POEC according to Table1 versus Sm doping concentration (C₀). All the samples were X-ray irradiated for 2 hours prior to the ESR measurement. ESR signal intensities were normalized to the mass of the samples. I is the intensity of POHC related Lorentzian and POEC related Gaussian lines presented in Table 1. In case of Lorentzians, I is the summed intensity of L₁−L₃. Note that the first derivative of these lines sum up to simulate the ESR signal (symbols in Fig. 2). I₀ is the corresponding intensity in the undoped glass irradiated for the same time (same dose). Lines are the fits using the formulas and the fitting parameters as shown in the figure. (The maximum C₀ value along the x-axis is 1 × 10²⁰ cm⁻³.)

Download Full Size | PDF

The above data may be explained by using a model that accounts for the chemical reactions occurring in the glass under X-ray irradiation. High energy X-rays create a large number of electron-hole pairs, which can be captured at pre-existing traps (precursors). Due to the photogeneration process (a primary projectile photoelectron ionizing the glass medium) a hole capturing reaction typically takes place in the vicinity of an electron capturing reaction. The reduction of RE ions usually takes place together with the formation of “oxygen-associated trapped hole centers [21]” nearby [6,20,26]. There are at least three “primary” reactions which consume X-ray generated electrons and holes

It should be emphasized that Sm²⁺ ions created as a result of X-ray irradiation have a metastable ionic environment since the relaxation of glass structure surrounding the Sm²⁺-ions cannot take place at room temperature (far below the glass transition temperature) as we discussed in our previous work [31]. These ions are usually referred to as (Sm³⁺)⁻ or (Sm²⁺)* to distinguish them from Sm²⁺ in thermally reduced (or structurally relaxed) glass [13,19]. Consequently, we should also include the inverse “secondary” reaction of Sm²⁺ to Sm³⁺ reconversion due to the capture of holes.

In all these reactions, the term PO has the general meaning of being a precursor for POHC or POEC creation by capturing the appropriate charge carrier on the PO bond. Equations (1) and (2) clearly show that Sm³⁺ ions and some of PO precursors compete for electrons in the vicinity of reaction defined by Eq. (3), which captures holes. Therefore, it is reasonable to assume that POEC formation in the glass would be suppressed by increasing the Sm³⁺ doping concentration (C₀). Mathematically, this dependence may be expressed as

Similarly, another pair of reactions defined by Eqs. (3) and (4) obviously compete for holes while, the electrons would be captured by the nearest neighbouring POEC precursors according to Eq. (2). The competition for holes between Eqs. (3) and (4) can be expressed as

Assuming that time dependent Sm²⁺ and Sm³⁺ concentrations C₂(t) and C₃(t) should be proportional to initial concentration of Sm³⁺, C₀, one can present

3.3 Effect of thermal annealing on defects

Annealing is known to be an effective method of re-converting Sm²⁺ to Sm³⁺ and for the elimination of X-ray induced defect centers [31]. Figure 4 represents the evolution of ESR spectra of the same sample (FP doped with 0.2% of Sm³⁺) that has been subject to a step-by-step annealing. This treatment involves a series of 30 minute sequential annealing processes at increasing temperatures (100 °C − 300 °C) interrupted by cooling down to room temperature after each step to perform the necessary measurements. It is apparent from this figure that the central “POHC related” part of the ESR spectra decreases rapidly while the “POEC related” wings remain almost constant as a result of thermal annealing.

 

Fig. 4 The evolution of EPR spectra of the same sample (FP doped with 0.2% of Sm³⁺) experiencing a step-by-step annealing treatment carried out at increasing temperatures (100°C−300°C) and cooled back to room temperature after each step. The time duration for every annealing step is 30 min. The sample was X-ray irradiated for 2 hours prior to annealing. The experimental ESR data (thick solid lines) are approximated by a sum (symbols) of functions presented in Table 1 and Fig. 1.

Download Full Size | PDF

Optical transmittance spectra were also recorded after each step of annealing and induced absorbance calculated (typical induced absorbance spectrum is not shown but can be found in [30]). To interpret the data, we performed a so-called “band separation” analysis [57] on induced absorbance spectra and approximated it as a sum of six Gaussians (G1−G6) based on the approach described in our previous work [30] where we associated these bands with POHC and POEC defects according to comparison with the results reported in reference [22].

More detailed analysis maybe done by studying the behaviour of Gaussian and Lorentzian components of the ESR signal which is presented in Figs. 5(a) and 5(c) in comparison with induced absorbance bands shown in Figs. 5(b) and 5(d). The parameters of the ESR components used here are the same as before and are listed in Table 1. The parameters of induced absorbance bands can be found in [30] Table 1. Notice that there is a good correlation of ESR data in Figs. 5(a) and 5(c) with the corresponding data on induced absorbance in Figs. 5(b) and 5(d).

 

Fig. 5 The variation of ESR signal components (a) and (c) and induced absorbance bands (b) and (d) (symbols) versus annealing temperatures (100°C−300°C) related to the same sample of Fig. 4 (doped with 0.2% of Sm³⁺ and X-ray irradiated for 2 hours prior to annealing). Symbols in (a) and (c) correspond to the intensity of lines presented in Table1 used for approximation of experimental data of Fig. 4. Symbols in (b) and (d) correspond to the intensity of bands G1−G6 introduced in [30]. (a) and (b) correspond to POHC related bands while (c) and (d) to POEC related bands. All the intensities are normalized to their value at room temperature (20°C) just after irradiation for 2 hours. Lines are guides to eye.

Download Full Size | PDF

Surprisingly, some of the POEC related signals seem to increase with the annealing temperature T_A while others remain almost constant or decrease. This may be interpreted as an indicator that some electrons released from one POEC band may be recaptured by another POEC band, inasmuch as there is no other process in the glass that would sink electrons. Meanwhile, Figs. 5(a) and 5(b) clearly show that the concentration of POHC monotonicallydecreases with increasing T_A, which is reflected in the decrease of corresponding signals. It seems that the thermal annealing at moderate temperatures (T_A ~350 °C) is sufficiently efficient for destroying POHCs. TL glow curve previously obtained in FP glass shows a peak related to phosphorus-oxygen based defects in the same temperature range [35]. Similar results have been observed for another kind of oxygen-associated trapped hole center (NBOHC) in silica glass [33]. One possible mechanism for the destruction of POHCs could be through the following chemical reactions (however, further investigations may be required):

Furthermore, the ESR spectra recorded in a wider range (g = 1−3.5) illustrated in Fig. 6 reveal some useful and interesting information. The trace of an extremely broad ESR signal (wider than 5000 Gauss) is noticeable in Fig. 6. This very broad signal shows a correlation with the presence of Sm²⁺ ions (inset of Fig. 6) and disappears in undoped samples or samples annealed at very high temperatures (550 °C) where only Sm³⁺ ions are present as shown in Fig. 6. The signal was reproducible. We suggest that the observed signal is probably related to Sm²⁺ ions which are non-Kramer ions and expected to produce very broad ESR lines [58].

 

Fig. 6 (a) ESR signal of undoped and doped (0.2% of Sm³⁺) FP samples recorded in a very wide range. All the samples were X-ray irradiated for 2 h prior to annealing and ESR measurements. The annealing duration was 30 min. Inset shows the corresponding photoluminescence spectra (shifted vertically to facilitate the comparison). Narrow ESR lines observed in the range g = 1.7−2.6 are the same kind of lines shown in Fig. 4 related to X-ray induced defects. Note the wide range deviation of the ESR signal in samples which show Sm²⁺ photoluminescence.

Download Full Size | PDF

It must also be noted that at 550 °C, which is above the glass transition temperature, the ESR signals (related to defects or Sm²⁺ ions) almost totally disappear as shown in Fig. 6. This implies the ionic reconstruction and erasure of almost all consequences of X-ray irradiation. On the other hand, the annealing at temperatures less than T_g (and above 350 °C) is known to increase the brightness of Sm²⁺ PL (for example, the annealed sample at 400 °C shown in the inset in Fig. 6). This phenomenon was discussed in our previous work [31] and was attributed to the formation of stable Sm²⁺ ions due to ionic structural rearrangement at elevated temperatures. It is interesting to note that we see the very broad ESR signal ascribed to Sm²⁺ ions in the samples annealed at 400 °C as well (Fig. 6).

4. Conclusion

X-ray irradiation of Sm-doped fluorophosphate glasses results in Sm³⁺ to Sm²⁺ ion conversion along with the formation of a number of defects in the glass structure. The difference in the photoluminescence signatures of Sm³⁺ and Sm²⁺ ions can be used for high resolution dosimetry required for microbeam radiation therapy (MRT). Towards developing such a detector, we studied the nature of X-ray induced defects and their dependence on Sm doping concentration and thermal annealing by using ESR and optical absorbance spectroscopy. We showed that the intense central part of the ESR spectrum, usually associated with so-called POHC (phosphorus-oxygen hole center), may be presented as a sum of Lorentzians Meanwhile, much weaker wings of the spectrum, usually associated with PO electron centers (which we refer to as POEC), may be approximated by a sum of Gaussians. We observed that both POHC and POEC related signals decrease exponentially with increasing Sm doping concentration, while the POEC related signals show a faster decay with the Sm-concentration. We were able to interpret these experimental results by a model that is based on competition between various defects and Sm-ions for the electrons and holes generated by the absorption of X-rays. The model suggests that the valency reduction of Sm³⁺ ions (electron centers) occurs together with the formation of POHCs (hole centers) within their vicinity. Sm³⁺ ions prevent the competing POEC precursors (POEC precursors within the Sm³⁺ capture volume) from capturing electrons, which results in the reduction of these defect centers with increasing Sm doping concentration. POHC formation would also be somewhat suppressed, which is due to the less dominant reaction of Sm²⁺ to Sm³⁺ reconversion that prevents the nearby POHCs from capturing holes. Annealing the irradiated glass at increasing moderate temperatures (up to 300 °C) results in the reduction of the central “POHC related” part of the ESR spectra while the “POEC related” wings almost survive. POHC and POEC related induced absorbance bands exhibit almost the same behavior. These results can be explained by considering that X-ray induced Sm²⁺ ions have a metastable structure which can be easily destroyed along with POHCs near them under moderate temperature annealing. On the other hand, since there is not such a reaction affecting POECs, the overall POEC concentration remains almost constant. Annealing at temperatures between 350 °C and the glass transition temperature (~460 °C) leads to the formation of stable Sm²⁺ ions (which produce a broad ESR signal) whereas above the glass transition temperature, almost all Sm²⁺ ions become reconverted back to Sm³⁺ and almost all defects become annealed out; there is no marked ESR signal. Our results suggest that the samarium valency conversion is correlated with the formation and destruction of defect centers which should be considered in designing Sm-doped FP glass plate detectors for MRT. It is also clear that annealing above the glass transition temperature can return the irradiated sample back to its original unirradiated state for reuse; a distinct advantage in MRT since once calibrated, the same Sm-doped FP glass can be reused many times.