On the Optical Response of Tellurium Activated Zinc Selenide ZnSe:Te Single Crystal

: In this study, the light output of a zinc selenide activated with tellurium (ZnSe: Te) single crystal was measured for X-ray radiography applications. A cubic crystal (10 × 10 × 10 mm) was irradiated using X-rays with tube voltages from 50 to 130 kV. The resulting energy absorption efficiency, detective quantum efficiency, and absolute luminescence efficiency were compared to published data for equally sized GSO: Ce (gadolinium orthosilicate) and BGO (bismuth germanium oxide) crystals. The emitted light was examined to estimate the spectral compatibility with widely used optical sensors. Energy absorption efficiency and detective quantum efficiency of ZnSe: Te and BGO were found to be similar, within the X-ray energies in question. Light output of all three crystals showed a tendency to increase with increasing X-ray tube voltage, but ZnSe: Te stood at least 2 EU higher than the others. ZnSe: Te can be coupled effectively with certain complementary metal – oxide – semiconductors (CMOS), photocathodes, and charge-coupled-devices (CCD), as the effective luminescence efficiency results assert. These properties render the material suitable for various imaging applications, dual-energy arrays included.


Introduction
From its inception, medical imaging has undergone consistent advancements, with the X-ray imaging acquisition mechanism playing a fundamental part. In this respect, scintillating materials that convert X-rays to light photons, combined with a suitable means of detecting the latter ones, not only "survived" the rampant technological progress, but rather remain central for the scientific interest [1].
Most scintillating coatings are inorganic compounds of crystalline structure, solutions of organic compounds, or polymeric ones [2].

Materials and Methods
In an attempt to determine ZnSe: Te crystal's applicability as a scintillator under X-ray radiography conditions, we concentrated our efforts on extracting and comparing a series of luminescence and scintillation properties.
A single crystal, cubic-shaped, edge 10 mm, with polished surfaces was obtained from Advatech [27]. Energy absorption efficiency (EAE), detective quantum efficiency (DQE), absolute luminescence efficiency (AE or LE), spectral matching factors (SMF), as well as effective efficiencies (EE) with various detectors were determined and compared with previously published data for GSO:Ce and BGO crystals of the same form [9]. The measurements were performed under polyenergetic X-ray excitation by a BMI Merate X-ray tube, in the range of 50-130 kVp, while keeping the current-time product at 63 mAs. Except from the tube's inherent filtration, equivalent to 2 mm Al, another 20 mm of Al were added in order to account for the attenuation by a typical human chest [32].
The energy absorption efficiency was determined as a function of different kVp of the X-ray tube. It is a measure of the energy absorbed locally, used for projection imaging detectors. This parameter is the fraction of incident X-ray energy absorbed locally at the points of X-rays interaction within the scintillator used in projection imaging detectors. This is an indication of what fraction of the incident X-ray photons contributes to the output signal formation and is expressed by [33]: where Φ 0 (E) is the X-ray photon fluence (photons per unit of area) incident on the scintillator, multiplied by the corresponding photon energy (E), giving the incident X-ray energy fluence. µ att (E)/ρ is the X-ray total mass attenuation coefficient and µ en (E)/ρ is the total energy absorption mass attenuation coefficient. W denotes crystal thickness (equal to 10 mm) and d is the density (in g/cm 3 ) [34,35]. Signal-to-noise ratio (SNR) and detective quantum efficiency (DQE), related to absorbed X-rays and optical photons have been previously investigated by Swank [36,37], under monoenergetic X-ray radiation, in terms of the noise-equivalent absorption and the statistical factor. Input SNR for poly-energetic X-ray radiation was first investigated by Nishikawa and Yaffe [38]. Dick and Motz [39] and Ginzburg and Dick [40] have also experimentally measured zero-frequency DQE under monoenergetic radiation for commercial rare earth screens. In this study, the input SNR 2 , the absorbed SNR 2 , and the absorbed DQE (at zero-frequency) for the scintillators under investigation were modelled for use in the energy range of 50-130 kVp.
For energy integrated detectors, Equation (2) was used to model the input SNR 2 [36][37][38][39][40][41][42][43]: where Ψ 0 (E) = Φ 0 (E)E for each energy value. The numerator is equal to the square of the first statistical moment of the distribution of X-ray energy. This distribution is expressed by the X-ray spectral distribution of Ψ 0 (E). This is also equal to the square of the total incident X-ray energy fluence (input signal) times energy. The denominator is equal to the second moment of the aforementioned distribution, which has been considered to express the noise of the input energy fluence. The SNR 2 of absorbed X-ray energy fluence indicates the SNR 2 of the X-ray energy fluence which is absorbed (i.e., detected) by the scintillator. It can be calculated from Equation (3) [41][42][43]: The detective quantum efficiency of absorbed X-rays (DQE abs ) describes the degradation of the SNR from the input to absorption within the scintillator mass. It shows the fraction of input X-ray energy fluence effectively absorbed by the scintillator. DQE abs may be expressed as follows [39][40][41][42][43]: The absolute luminescence efficiency is a measure of how much light the crystal emits (expressed as light energy flux . Ψ λ , units µW m −2 ) for a given X-ray exposure rate incident on it ( . X, units mR s −1 ) and is expressed in efficiency units, EU = µW m −2 /(mR s −1 ) [33]. Obviously, preference is given to scintillators that achieve high AE values, thus tending to decrease the patient radiation dose.
In our experimental setup, the crystal was placed at a surface-tube focal spot distance of 72.5 cm. A dosimeter (RTI Piranha P100B) located at the same distance was used to provide the exposure rate values. For the emitted light flux, a photomultiplier tube (PMT) was used (EMI 9798 with extended sensitivity S20 photocathode), coupled to the crystal's "back" surface via a light integrating sphere (Oriel 70451), in order to minimize errors arising from light non-uniformities. No high voltage and voltage divider were used to bias the dynodes, since the PMT was connected as a biased photodiode (22 V between photocathode and the first dynode), whose current was measured by a femtoamp meter (Keithley 6430) [31]. AE is given by Equation (5) [44]: where i elec is the electrometer's output current in pA, S is the irradiated crystal's surface in mm 2 and η p is the photocathode's peak photosensitivity expressed in pA/W. α s is the spectral compatibility factor expressing the compatibility of the scintillator's emission spectrum to the spectral sensitivity of the photocathode. c g is the geometric light collection efficiency of the experimental setup, expressing the fraction of the crystal's emitted light that impinges eventually on the PMT entrance window, after passing through the integrating sphere. The value of 15.6 was estimated for our specific setup. Concluding, AE is expressed in units of µW m −2 /mR s −1 , where µW m −2 corresponds to the light energy flux ( . Ψ λ ) and mR s −1 to the exposure rate ( . X). For simplicity, the notation efficiency unit (EU) was used: 1 EU = 1 µW m −2 /(mR s −1 ).
AE is an efficiency measure of the scintillator itself, but when it comes to incorporating it in a complete system, it is the effective efficiency that better describes the system's performance, taking into consideration the light detector's spectral response. It is the product of absolute efficiency with the spectral matching factor [34,45]: where α s is the spectral matching factor of the optical detector, which is given by: where S p is the spectrum of the light emitted by the scintillator, S D is the spectral sensitivity of the optical photon detector, and λ denotes photon wavelength [34]. The crystal's emission spectrum was obtained from Ryzhikov et.al [16]. For the sake of completeness, a variety of popular light detectors were considered; spectra extracted from their datasheets [46][47][48][49].

Results and Discussion
The EAE for all three crystals is plotted versus the selected kVp in Figure 1. The attenuation coefficients used in Equation (1) were calculated from the respective elementary ones (derived from the NIST XmuDat database [50,51]) and are drawn in Figure 2 for all three materials. Shown in Figure 3 is the ratio of the attenuation coefficients µ en /µ att that appears in Equation (1), as a function of photon energy.   Decreasing of EAE while the tube voltage increases is generally obvious in Figure 1. The ability of a crystal to absorb the incident X-ray energy is expressed by µ att (the probability that a photon is absorbed by the crystal, via any interaction mechanism), but more importantly by µ en (the probability that a photon's energy is imparted to electrons within the material). With increasing of the tube high-voltage, the emitted X-ray spectrum shifts to higher energy regions. In these regions the attenuation coefficients generally diminish, with the exception of some areas, characteristic for every material, where the probability of photoelectric interaction maximizes locally (Figure 2), namely the K-edge, L-edge, etc. The K-edges, characteristic for GSO and BGO are found at 50.2 keV and 90.5 keV respectively, whereas ZnSe has K-edges outside the region of interest (below 13 keV, where photons have been attenuated by the Al filter). However, both GSO and BGO, at their respective K-edge energies, exhibit a ratio µ en /µ att sink of more than 60%. This K-edge sink contributes to the abrupt EAE loss of GSO which is observed in Figure 1 from 60 kVp onwards. Conversely, at 40 and 50 kVp the EAE of GSO is slightly higher than that of the other two crystals, in accordance with the ratio µ en /µ att before the K-edge sink. At tube voltages higher than 90 kVp, there is an increasing tendency for EAE of GSO and this can be attributed to the gradual recovery of the ratio µ en /µ att after the K-edge energy.
On the other hand, from Figure 1 it is clear that ZnSe and BGO have a commensurate efficiency in absorbing energy, both retaining values above 0.8 up to 100 kVp and consecutively showing a gradual decrease. ZnSe exhibits a slightly lower efficiency than BGO in tube voltages up to 100 kVp, but this behavior is reversed thereafter; yet, their difference is always below 0.1. This reverse beyond 100 kVp can be attributed to the fact that µ en /µ att of BGO has a K-edge sink at energies pertaining to X-ray spectra obtained with tube voltages from 100 kVp to 140 kVp (i.e., 90.5 keV), whereas ZnSe does not. Figure 4 shows the variation of input SNR 2 with X-ray tube voltage in the range 50-130 kVp. Data shown in this figure were calculated for Ψ 0 (E) and constant Air Kerma. The SNR 2 increases as a function of tube voltage because it depends on the product Ψ 0 (E) times energy.  The calculated results depend on the EAE of the scintillators (see Figure 1). ZnSe:Te demonstrates the maximum SNR abs 2 values, especially at high tube voltages (more than 110 kVp). The DQE absorbed values as a function of set tube voltage for the three crystals are shown in Figure 6, which also demonstrates that for all scintillators the DQE absorbed values are equal to EAE (see Figure 1). Again, ZnSe:Te demonstrates the highest DQE absorbed values at high tube voltages (more than 110 kVp).  All three materials exhibit a constantly increasing AE with increasing of the kVp, but ZnSe: Te clearly outperforms having at least 2 EU higher than GSO: Ce and 4-7 EU higher than BGO, in every point of the graph. Obvious is the proportionality between AE and kVp for both ZnSe:Te and GSO:Ce in the tube high-voltage region from 50-80 kVp. At higher voltages, the relation is non-proportional, something that Balcerzyk et al. [11] attribute to the crystal's structure, as far as GSO: Ce is concerned.
Revisiting the remark expressed in the introductory section, that the material's light yield varies within a broad interval [18], one can note that it still remains higher than the values reported for BGO and GSO:Ce, even in its "least efficient" version (see Table 1, footer note). This is verified by the AE-kVp graph, even though ZnSe: Te has lower density and effective atomic number than the other ones.
Specifically, at the end of its proportionality region, i.e., 80 kVp, ZnSe: Te attains a value of 8 EU, as opposed to the 5.8 EU of GSO: Ce and the 2.6 EU of BGO. At the highest tube voltage applied in our experiments, 130 kVp, the absolute efficiency differences were even greater, measuring 10.4 EU, 8 EU and 3.3 EU respectively. Also prominent in the same graph is the non-proportional behavior of BGO in the whole energy region examined herein, as well as the plateau that it exhibits above 100 kVp, with the AE value stabilizing below 3.3 EU.
The linear relationship between light output and exposure rate is depicted in Figure 8 for all three materials. Linear interpolation lines have been drawn for the data points, with the R 2 value (i.e., the coefficient of determination) being at least 0.996 in any case. The lines' slopes come in agreement with the relationship between the three crystals' AE values (Figure 7). The higher slope of the ZnSe: Te line corresponds to higher AE values, whereas the other two materials stand lower, with GSO: Ce slope being closer to that of ZnSe: Te.
In Figures 9 and 10, the emitted spectrum of ZnSe: Te [16], is juxtaposed with the spectral responses of several commercially available detectors. For ease of comparison, everything is normalized. Luminescence of ZnSe: Te peaks at 640 nm [16].   From Table 2, that summarizes the SMF with every such detector, comes out the perfect match that takes place between this crystal and several CCD and CMOS detectors, a fact that verdicts towards the usability of ZnSe: Te in digital X-ray systems. A nearly total overlap, above 0.96, is found with the monolithic 0.25µm CMOS and the CMOS RadEye high resolution that are currently used in a number of commercial radiography systems, both industrial and medical. Mediocre compatibility instead was registered with the PMT EMI 9798 extended photocathode (E-S20) used in our experiments, since the SMF was only 0.60. In any case, this value is taken into consideration for the latter calculations. More tangible data regarding this "usability", provide the EE values. These are plotted versus tube's kVp in Figure 11 for every detector mentioned before and verify the perfect compatibility with several of them. EE values nearly equal to the AE ones are obtained (~5 EU at 50 kVp and~10 EU at 130 kVp) when ZnSe: Te is combined, e.g., with the CMOS hybrid blue sensor, the CCD with ITO gates and microlenses, the GaAs photocathode, in addition to the two above-mentioned commercial CMOS.

Conclusions
The output luminescence signal of a ZnSe: Te single crystal was examined within the radiographic energy region for possible applications in novel radiographic applications (such as dual energy). To this aim, the spectral compatibility with various commercial sensors was also considered. BGO and GSO: Ce crystals of equal dimensions were used, for the purpose of comparison. In terms of EAE and DQEabs, both ZnSe and BGO demonstrated commensurate performance, especially in tube voltages up to 100 kVp. The maximum luminescence efficiency was obtained at the maximum examined X-ray energy (130 kVp) since the light output of ZnSe: Te constantly increased within the energy range in question. The emitted optical photons of ZnSe: Te were found to be optimally detected by a wide range of complementary metal-oxide-semiconductors (CMOS), photocathodes, and charge-coupled-devices (CCD). These properties, together with those mentioned in the introductory section (brief decay times and convenient filtering capabilities), qualify the use of ZnSe: Te as a high spectral energy photon absorber in dual-energy arrays, improving the overall detection capability.

Conflicts of Interest:
The authors declare no conflict of interest.