1. Introduction

The use of composite scintillating materials is a versatile approach for the detection of ionizing radiation. Comprised of micro- or nano-sized polycrystalline scintillator particles dispersed within an organic or inorganic matrix, composite scintillators can be more easily manufactured and are a cost effective solution for complex detection systems [1–3]. In general, single-crystals are the ideal material standard for radiation detection for their high density, brightness, and optical transparency [4]. Though high performing, the cost to manufacture and integrate single crystals can be quite high, making this approach unrealistic for high volume applications. The relative ease and cost efficiency of composite manufacturing gives these alternative scintillating materials an edge over their idealistic counterparts.

Several variants of composite scintillators have been developed over the years, and have attracted great interest within the radiation detection community. Glass ceramics have gained considerable attention due to the combination of the desirable characteristics of sintered ceramics (high light yield, transparency) and glass (low-cost, geometric flexibility) [5]. Glass-ceramic materials are formed through the nucleation of nanocrystals within a glass matrix; additionally, these materials can be doped with rare earth ions to promote more efficient energy transfer mechanisms versus glass scintillators [5–7]. In particular, oxy-fluoride based glass ceramics have emerged as strong potential candidates for scintillation detectors [6,7].

Nanocomposites are a similarly interesting scintillator material due to the tunable characteristics associated with particle size. With a high surface-to-volume ratio, substantial surface effects and quantum behavior can manipulate the scintillation properties of the crystals [3]. Studies have reported shifts in the emission spectra, light yield, and decay times with the nano-crystalline form of bulk crystals [8–11]. Optical transparency has also been realized with sufficiently small particle sizes [11].

Currently, similar types of materials are used in both digital and computed radiography (DR, CR) in the form of flat panel imaging plates. With imaging plates, the phosphor or scintillator powder is mixed with a polymer binder, which is dissolved in a solvent, and is then cast onto a plate. Upon evaporation of the solvent, a homogenous layer of unstructured photo-stimulable material remains [12,13]. BaFBr:Eu²⁺ and Gd₂O₂S:Tb³⁺ are examples of photo-stimulable materials used in CR and DR, respectively [13]. Beyond the relative simplicity of imaging plates, more complex geometries such as pixelated structures are also possible with composites [14].

Consequently, however, due to differences in the refractive indices between the scintillator and binder, internal optical scattering can severely affect the light transport properties of the composite. For flat-panel imaging plates, a degree of scattering is necessary to localize the scintillation event by preventing lateral photon drift. In contrast, in pixelated systems, boundaries between each pixel confine the scintillation photons to within the active scintillator area, and therefore remove the need for scattering. Excess scattering within the pixel will then only serve to diminish the light output and degrade the overall quality of the detector. This is undesired especially for medical imaging applications, where a reduced light output requires longer exposure times with correspondingly higher doses to the patient. It is therefore of interest to develop composites with reduced optical scattering.

By closely matching the refractive indices of the scintillator and binder material, scattering at material interfaces can be substantially reduced. Given by Snell’s Law and the Fresnel equations, a reduction in the quantity Δn (refractive index difference between two materials) reduces both the angle of refraction and total reflection; these effects reduce the perturbation of the optical path caused by the interface and can lead to an improvement in the overall mean free path. For many scintillators, a perfect index match is not attainable due to both dispersion and birefringence. Therefore, in the case of crystals with non-cubic symmetry, matching the refractive index to an average of the optical axes at the emission wavelength can provide a significant improvement to the transmissivity of the composite for the wavelength of interest.

Previous studies attempting to achieve this condition have considered fluoride crystals such as BaF₂ and BaFCl:Eu²⁺ due to their relatively low refractive index [15,16]. The index of refraction of BaFCl and BaF₂ have been reported between 1.647 to 1.67, and 1.47 to 1.51, respectively over the visible range [15–18]. Polymer binders generally have refractive index values below 1.70. Unlike BaF₂, a cubic crystal, BaFCl is a biaxial crystal which forms in the tetragonal matlockite (PbFCl) structure, and thus a best-fit to the average index must be employed [19].

In this paper, we explore composite fabrication with micron-sized polycrystalline Sm²⁺-doped BaFCl dispersed in a polymer matrix with a comparable index. Using the index values reported in literature [16–18] for BaFCl (n_BaFCl = 1.647-1.67) and the index of the chosen matrix (n_matrix = 1.665), Δn is predicted to be less-than or equal-to 0.018. While nanocomposites are a current topic of many scientific pursuits, we investigate larger sized scintillator particles as they provide the advantages of retaining bulk properties of the crystal as well as being more sensitive to ionizing radiation. Additionally, the use of Sm²⁺-doping is advantageous for its emission in the red end of the spectrum; this characteristic enables light transmission outside of the highly absorbing region of the polymer binder, often located in the UV-blue range. BaFCl:Sm²⁺ has also been identified as a potential system for optical data storage due to its spectral hole burning properties, making it a versatile material for a variety of applications [20,21].

Herein, we report on the fabrication and performance of a translucent composite scintillator utilizing BaFCl:Sm²⁺. In addition, ray-tracing simulations are performed to understand the interplay between composite thickness, irradiation energy deposition, and visible light transmission.

2. Experimental

Under inert atmosphere, BaF₂ (Strem Chemicals, 99.99%), BaCl₂ (Alfa Aesar, 99.998%), and SmF₃ (Alfa Aesar, 99.9%) were added and mixed in a glassy carbon crucible. SmF₃ was introduced at a 1 mol% doping concentration with respect to BaFCl. The mixed constituents were transferred to a Carbolite CWF1200 furnace and heated to a dwell temperature of 1100°C, and remained at the process temperature for 1 hour. Cooling occurred at a natural rate. A typical synthesis produced ∼22 g of material. Synthesized polycrystalline BaFCl:Sm²⁺ was removed from the inert environment for post processing. The polycrystalline material was crushed and sieved to isolate particle size subsets of >75 µm, 53-75 µm, 38-53 µm, and <38 µm. Photoluminescence (PL) and photoluminescence excitation (PLE) spectra were measured using a Horiba-PTI QuantaMaster spectrometer containing a 75 W xenon arc lamp source with excitation and emission monochromators. The spectral bandwidth was set to 0.5 nm for excitation and emission, and a step size of 0.1 nm was used. Both the PL and PLE spectra were corrected for the wavelength-dependent response of the spectrometer. X-ray diffraction (XRD) measurements were made on a Rigaku Ultima III diffractometer that uses a fine line sealed Cu Kα (λ = 1.5406Ǻ) x-ray tube. The diffraction measurements were taken with the <38 µm subset BaFCl:Sm²⁺ powder attached to a background-free silicon slide.

BaFCl:Sm²⁺ powder obtained from the synthesis was centrifugally mixed under vacuum with isopropyl alcohol (IPA) and Norland Optical Adhesive 1665 (n = 1.665) in a Thinky Planetary Vacuum Mixer. The IPA was included to facilitate homogenous distribution of the scintillator particles within the optical adhesive, and to prevent excessive heat generation due to friction. The remaining solvent was then allowed to evaporate from the composite prior to curing in order to reduce the production of spatial voids within the material. The solvent-evaporated composite mixtures contained fifty percent BaFCl:Sm²⁺ by volume. Using all size subsets, composites were subsequently cast into cylindrical molds Ø1.0 cm × 1.3 mm thick for spectroscopic analysis, and molds Ø6.35 cm × 600 µm thick for radiography using only the <38 µm subset. For the large-scale composites, one composite was cast onto a polished copper back-plate, and one was cast onto a polished copper back-plate electro-plated with silver. Each composite was cured with a 365 nm UV lamp.

Radio luminescence (RL) spectra were measured with a custom-built system using a PANalytical x-ray generator with a molybdenum target operating at 50 kV and 40 mA, producing x-rays with an effective energy of ∼25 keV. The luminescence output was guided via a fiber optic bundle to a Princeton SpectraPhysics spectrometer. Transmission spectra were acquired on a Cary5000 UV-VIS spectrophotometer. The reflectance spectra of electroplated silver and polished copper were measured using the Diffuse Reflectance Accessory (DRA) for the Cary5000. All measurements were carried out at room temperature.

Radiographs were acquired using the Los Alamos National Laboratory (LANL) Microtron x-ray generator operating at 120 Hz with an incident electron energy of 10 MeV. The source was a tungsten target producing a bremsstrahlung distribution. No filter was used to remove lower energy x-rays. Samples were placed 307.8 cm from the source, and the integration time was 66 ms.

3. Results and discussion

Alternate phases of BaFCl can be formed during a solid-state melt-based synthesis and can lead to shifts in the emission spectrum [22–24]. This has been attributed to a fluorine rich melt created by volatile loss of BaCl₂. XRD analysis confirmed the resulting BaFCl scintillator powder synthesized in this work was phase pure. Shown in Fig. 1(a), the XRD pattern of the BaFCl:Sm²⁺ powder matched with the known BaFCl phase (PbFCl) [19]. The crystal structure diagram in Fig. 1(b) illustrates the tetragonal shape of the BaFCl unit cell, which is the origin of the birefringent nature of the crystal. In Fig. 1(c), the powder under examination was illuminated with a long-wave 365 nm UV lamp; here, the characteristic red emission of the Sm²⁺ activator ion can be seen.

 

Fig. 1. (a) XRD analysis of the BaFCl:Sm²⁺ powder with known BaFCl phase peaks using the PbFCl (tetragonal matlockite) crystal structure as Ref. [19]. XRD analysis indicated phase pure BaFCl was formed. (b) Crystal structure of the BaFCl system generated in VESTA 3 [25]. (c) BaFCl:Sm²⁺ powder under 365 nm UV illumination.

Download Full Size | PDF

To elucidate the valence state of the Sm ion within BaFCl, size-dependent PL/PLE spectra were acquired at room temperature and are shown in Fig. 2. The absence of Sm³⁺ transitions indicated that Sm³⁺ was quantitatively reduced to Sm²⁺ during the synthesis in argon atmosphere. The peaks located in Fig. 2(a) at 687 nm, 702 nm, 728 nm, 766 nm, and the doublet near 814 nm are attributed to ⁵D₀→⁷F_0,1,2,3,4 transitions of the Sm²⁺ ion, respectively [26]. The highest energy phonon in BaFCl has been reported at 294 cm⁻¹ [27], and the ⁵D₁→⁵D₀ multi-phonon relaxation therefore involves at least 5 phonons (see inset of Fig. 2(a)). This makes radiative relaxation of the ⁵D₁ excited state competitive and leads to pronounced ⁵D₁ luminescence in this low phonon energy material, as seen by the ⁵D₁→⁷F_0,1,2,3 transitions located at 629 nm, 642 nm, 663 nm, and 694 nm, respectively. Shown in the inset of Fig. 2(a), the transition energies of BaFCl:Sm²⁺ match reasonably well with the transition energies in LaCl₃:Sm²⁺ reported by Dieke et al [26], further supporting the assignment to Sm²⁺ transitions.

 

Fig. 2. (a) Room temperature PL of BaFCl:Sm²⁺. The ⁵D₀→⁷F_0,1,2,3,4 and ⁵D₁→⁷F_0,1,2,3 transitions were present in the spectral range analyzed and match well to the transition energies reported for LaCl₃:Sm²⁺ reported by Dieke et al [26]. (b) PL intensity (integrated over the full spectral range in Fig. 2(a)) as a function of particle size, normalized to the largest particle subset (75 µm). (c) Room temperature PLE of BaFCl:Sm²⁺.

Download Full Size | PDF

Incorporating samarium into the BaFCl lattice in its divalent state is an unfavorable process. Sm ions generally assume a trivalent state naturally, and thus require a reducing agent to provide the additional electron. Reducing Sm³⁺ to Sm²⁺ can be achieved through the use of a reducing atmosphere during the synthesis procedure or by low power (<1 mW/cm²) UV and X-ray stimulation of the BaFCl:Sm³⁺ compound [28]. The latter process has been shown to be a reversible effect through higher power (∼100 mW/cm²) excitation, indicating potential for optical data storage [29–31]. In the present work, the inert atmosphere and glassy carbon crucible provided a reducing environment during the melt, facilitating the Sm³⁺ reduction.

In Fig. 2(b), the integrated luminescence intensity is displayed, normalized to the largest particle subset (>75 µm). The figure illustrates the size dependency of the scintillator powder luminescence; specifically, the luminescence is quenched at smaller particle sizes. This could be due to quenching through surface defects or possibly a result of defects induced by physical crushing of the material to reduce the particle sizes. Finally, size dependent PLE is shown in Fig. 2(c).

Small scale composites fabricated with the <38 µm BaFCl:Sm²⁺ particle subset are shown in Fig. 3(a). Both samples in the figure are from the same composite mixture, and have nominally identical compositions. The left sample was cast onto a standard glass microscope slide and the right was cast onto aluminum foil and subsequently peeled off. Characterization measurements were performed with the non-glass-backed sample (right). Under 365 nm UV back illumination (Fig. 3(a) bottom), the characteristic Sm²⁺ emissions are visible throughout the entirety of the material; this suggests light generated at the front end (facing the light source) of the composite is able to propagate to the back face (facing the camera). The ability for light to traverse the thickness of the material is indicative of a relatively low index mismatch, coupled with a lower absorption at the emission wavelengths. The smallest size particle subset is chosen for the advantages of better homogenous mixing during the composite fabrication process and for small pixelated feature considerations. It was also determined the <38 µm subset was optimal within the four size-subsets as it provided the highest degree of transmissivity. Due to the large bin of <38 µm, it is possible there are particles <<38 µm which improve the overall transmission, similar to the effect of nanoparticles in transparent nanocomposites. Understanding the full distribution of particle sizes within the subset could provide a pathway for enabling composites with greater transmissivity.

 

Fig. 3. (a) Images of the BaFCl:Sm²⁺ composite under ambient light exposure (top) and UV (365 nm) back illumination (bottom). Both samples shown have the same composition and dimensions. (b) Radio-luminescence (red trace) and transmission (black trace) of the 1.3 mm thick BaFCl:Sm²⁺ composite. The emission lines of the scintillator are outside the highly absorbing region of the composite, improving the overall light transport.

Download Full Size | PDF

In Fig. 3(b), the transmission is displayed from 300-800 nm with the radio-luminescence spectrum overlaid. To calculate the absorption coefficient, α, the transmission was corrected for reflections at the air-material interfaces and subsequently converted to absorptivity. From the comparison of the emission and transmission spectra, the emission wavelengths of the composite are beyond the highly absorbing region, providing longer propagation lengths for the scintillation photons. This property is a key factor which influences the maximum thickness achievable for the composite.

The absorptivity is inversely proportional to the mean free path (µ_m), or the average distance the photon travels between interaction events. Applying the Beer-Lambert Law, the maximum thickness imposed on a material such that at least 5% of the photons generated at one end reach the opposite side is approximately equal to 3µ_m. At the location of emission maximum, α is equal to 40.16 cm⁻¹, corresponding to a mean free path of 249 µm. Since photons are generated throughout the material during an irradiation event, it is expected that these composites can be scaled to thicknesses greater than 747 µm. From the RL spectra, it is also evident that the Sm²⁺ ion retains its divalent state when exposed to the X-irradiation, as seen by the similar ⁵D_J→⁷F_J transitions present in the PL spectra.

Large-scale composites, shown in Fig. 4, were fabricated and cast onto a silver coated copper back-plate (Composite A-left) and a polished copper back-plate (Composite B-right). Due to the large area of the forms and casting method used, an uneven surface was created during the casting process; this led to variations in the thickness for each sample. Average thicknesses were measured to be 836 µm and 488 µm, with ranges of 650 µm to 1.1 mm and 420 µm to 660 µm, for Composite A and B, respectively. Thicknesses were measured at 20 random locations on each composite using a micrometer, excluding the edges. An effect of the reduced thickness compared to Composite A can be seen in Fig. 4(a), where under ambient light exposure the copper back-plate of Composite B is visible through the bulk of the material. The similarity in the color of the composite and silver back-plate for Composite A made a visible inspection of the translucency inconclusive.

 

Fig. 4. BaFCl:Sm²⁺ composites A (on silver-coated copper) and B (on polished copper) under (a) ambient light exposure, and (b) unfiltered bremsstrahlung x-ray exposure from a tungsten target. (c) Reflectance of polished copper and electro-plated silver on copper. (d) Digital conversion to a color map normalized to the maximum emission in 16-bit representation. The brighter response of Composite A is likely due to a greater average thickness, measured to be 836 µm versus 488 µm for Composite B. Both composites have a diameter of 6.35 cm.

Download Full Size | PDF

In Fig. 4(b), when exposed to unfiltered bremsstrahlung x-ray irradiation from a tungsten target, Composite A exhibits a sizeable brightness improvement over Composite B. Over the relevant 620-750 nm emission wavelength range, the reflectance of the polished copper and electro-plated silver, shown in Fig. 4(c), are comparable and were determined to have minimal effect on the brightness. To further address the comparison between Composite A and B, the radiograph from Fig. 4(b) was converted to a digital color map of the intensity distribution, normalized to the highest intensity value (Fig. 4(d)). Using the color map, average intensity values were calculated to be 0.8076 and 0.5130 for A and B, respectively. Comparing the ratio of intensities (I_Avg-A /I_Avg-B) to the ratio of the thicknesses (t_Avg-A/t_Avg-B) indicated a gain of 1.57 in brightness from the 1.7 gain in thickness; this observation demonstrates that the increased brightness can be correlated to the increase in average thickness. Figure 4(d) also illustrates the degree of thickness uniformity across the large-scale composites; controlling this parameter at these scales is currently under investigation.

To understand the interplay between deposited x-ray energy and transmitted scintillation light as a function of composite thickness, ray tracing simulations using the FRED Optical Engineering software were performed. In the FRED simulations, where replicate models of the large scale composites were created, the composite thickness was varied from 0.1 mm to 2.0 mm. A total of 1 million rays at 687 nm were used in each simulation; here, 1000 initial ray positions were randomly generated within the composite volume, and 1000 random ray directions were sampled at each position. Internal transmittance values for the material were gathered from the transmission data, taking the reference thickness to be 1.3 mm. The simulation integrated the fraction of power passing through the top face of the composite at each thickness, providing an upper bound for the composite transmittance at that thickness value.

Shown in Fig. 5, the fraction of power integrated at one end of the composite is displayed as a function of thickness (red curve). As would be expected, the curve follows an exponential decay function, following the Beer-Lambert Law. Also shown in the plot is the fraction of deposited energy into the composite for select irradiation energies (black curves) calculated using the NIST-XCOM database [32]. By comparison, we can see there is a distinct tradeoff between the energy deposited into the material and the transmission of 687 nm photons as the thickness is increased. At lower energies, the relative gain in energy deposition is substantially greater than for higher energies, and could explain the observation in Fig. 4, where a thicker composite produced a brighter image. For example, the gain in energy deposition as the thickness is increased from 488 µm to 836 µm at 59 keV is approximately equal to 1.419, and matches fairly well with the relative brightness gain of 1.57. Due to the unfiltered bremsstrahlung source used to produce the radiographs, it is likely there were lower energy contributions to the energy deposition that are not displayed in Fig. 5; these other contributions would provide additional gain to the deposited energy, and thus improve the agreement between experiment and theory. Further improvements to the total transmissivity could be accomplished though closer index-matching of the scintillator and matrix, surface morphology control or precise size distribution control. The scalable nature of these composites indicate a potential for both low and high energy radiation detection applications.

 

Fig. 5. Fractional deposited energy into the composite as a function of thickness for select irradiation energies (black lines) calculated using the NIST-XCOM database [32]; and fractional transmitted power of scintillation light as a function of thickness (red). At lower energies, the gain in energy deposition is greater than the loss in transmission; this supports the observation from Fig. 4, a greater brightness with increased thickness.

Download Full Size | PDF

4. Conclusions

We have demonstrated the ability to synthesize polycrystalline BaFCl:Sm²⁺ in the melt, and have fabricated translucent scintillating composites utilizing a sub-38 µm particle size subset of the compound. The translucent appearance of the composites is indicative of a relatively low index mismatch (Δn ≤ 0.018) between the BaFCl:Sm²⁺ and the polymer matrix. This enhanced transmissivity allows the composite thickness to be increased, and thus improves energy deposition at higher irradiation energies. Due to the flexibility in thickness constraints, these composites show promise for both low energy and high energy radiation detection applications.