Hydrothermal and Mechanosynthesis of Mixed‐Cation Double Perovskite Scintillators for Radiation Detection

This article details work performed on the synthesis and characterization of an inorganic mixed‐cation double halide perovskite, Cs2Ag.6Na.4In.85Bi.15Cl6 (CANIBIC). Single crystals have been created via a hydrothermal reaction, milled into a powder, and pressed into pellets, while nanocrystals have been directly synthesized via mechanosynthesis. A computational model is constructed to predict the X‐ray diffraction pattern of CANIBIC; this model aligns very well with the X‐ray diffraction pattern measured for CANIBIC crystal powder. This model can therefore be developed in the future as a tool to predict lattice parameters and crystal structures of other novel double‐halide perovskites. Photoluminescence spectra obtained from each format show broad emission centered at 630 nm, as is typical for self‐trapped exciton emission; self‐trapped exciton emission is also confirmed by investigating photoluminescence intensity as a function of laser power. Nanocomposites are produced via the loading of nanocrystals of CANIBIC into PMMA. Although nanocomposite disks consisting of a small proportion of CANIBIC nanocrystals in PMMA have a smaller mass attenuation coefficient than a pressed pellet of CANIBIC, these disks have comparatively bright radioluminescence due to their optical transparency. These nanocomposite disks are therefore a particularly useful format for the practical use of the CANIBIC scintillator.


Introduction
Halide perovskites are a class of materials with the composition ABX 3 , where A is a monovalent cation, B is a divalent cation, and X is a halide anion (typically Cl − , Br − , or I − ). [1]Halide perovskite semiconductors have commonly been used as light-emitting diodes (LEDs) [2] and solar cell materials. [1]In recent years, the potential applications of the materials as radiation detectors have been explored. [3]Materials such as CsPbX 3 , [4,5] FAPbX 3 , [6] and MAPbX 3 [7,8] benefit from high mass attenuation coefficients due to the central Pb 2 + cation, and tunable emission by varying the halide composition [9] allowing the optimization of the emitted light to match an associated photosensor.
Lead-based perovskite materials have several significant disadvantages; for example, there are issues with the long-term stability of the perovskite, [10] as well as with the toxicity of the perovskite, precursor chemicals, and synthesis waste materials.These environmental concerns have led to other heavy cations being incorporated into the perovskite structure to replace the Pb 2 + ion, including Sn 2 +, 4 + , [11,12] Bi 3 + , [13] In 3 + , [14] Sb 4 + , [14] Ag + , [15] Na + , [16] and Cu 2 + . [17]The divalent cation B in the perovskite crystal structure may also be replaced by a monovalent or trivalent cation which can be varied within the bulk perovskite structure so as to form a mixed-cation double perovskite.The general structure of the double perovskite crystal lattice, which has also been referred to as elpasolite, [18] is shown in Figure 1a.
Many double perovskites exhibit an extremely broad peak emission due to the formation of lower energy trapping sites, shifting the scintillation emission energy.This occurs in cases of strong coupling of electrons or holes to the crystal lattice, where a charge carrier may be self-trapped as a polaron in its own lattice distortion field. [19]The energy shift depends on the lattice deformation and trapping energy as free excitons move to this trapping site [20] and recombine radiatively. [14]The degree to which this lattice distortion occurs is affected by the heterogeneity of the material. [21]The mechanism of this self-trapped exciton (STE) emission is shown in Figure 1b.The STE mechanism leads to an extremely large Stokes shift in the emission, resulting in low levels of self-absorption and therefore contributing to the high light-yield of these scintillators.However, STE emission has an extended scintillation lifetime compared to single perovskite materials such as FAPbBr 3 [6] due to the coupling of phonons and excitons. [14]STE emission is associated with whitelight scintillators; [14] a broad emission peak over several hundred nanometers is typically seen.The breadth of STE emission can be attributed to the distortion of the self-trapped state with respect to the ground state. [20]This distortion can be described by the Huang-Rhys factor, S, the number of phonons emitted after carrier capture.Halide perovskites are in the strong coupling regime and have a large Huang-Rhys factor of ≈S = 350, compared to the Huang-Rhys factor of S = 75 for GaAs. [22]STE emission has previously been observed in other cubic, perovskite-like halide scintillator materials, such as Cs 2 LiYCl 6 , [23] Cs 2 HfCl 6 , [24] and Cs 2 NaLaBr 3 I 3 . [25]s 2 Ag x Na (1 − x) InCl 6 (CANIC) has been investigated as a white light scintillator, with the B-sites of the double perovskite oc-cupied by Ag + , Na + , and In 3 + .This mixed-cation double perovskite is typically produced with >2% Bi 3 + doping to increase the light yield of the material. [15,16,21,26,27]The material presents a broad photoluminescence (PL) emission at these low Bi 3 + loadings and decays on the order of microseconds owing to the STE recombination. [16]The photoluminescent quantum yield (PLQY) of CANIC can be optimized by varying the Ag + :Na + ratio, as nonradiative processes and thermal quenching are suppressed by the alloying of Na + at 40% loading. [15]urther to this optimization of the monovalent B-site cation, the radioluminescence (RL) and X-ray sensitivity were improved by increasing the doping of Bi 3 + up to 15% by Zhu et al. [28] Cs 2 Ag 0.6 Na 0.4 In 0.85 Bi 0.15 Cl 6 (CANIBIC) is a mixed-cation double perovskite that boasts a broad yellow light emission centered at 625 nm and a large Stokes shift of ≈200 nm originating from STE emission.Figure 1c shows the relative mass attenuation coefficients for a range of common scintillator materials compared to CANIBIC at 15% Bi 3 + loading.CANIBIC also displays excellent radioluminescent light yield and X-ray imaging properties. [28]erovskites are typically synthesized via hydrothermal or solution-growth methods, with nanocrystals of double perovskites grown via room-temperature recrystallization. [29]Nanocrystals of Cs 2 AgIn y Bi 1 − y Cl 6 [10] and Cs 2 Ag x Na 1 − x In y Bi 1 − y Cl 6 [29] have previously been investigated after synthesis via recrystallization and display the typical broad white light STE emission under laser excitation.
To investigate the scintillation properties of nanocrystals of CANIBIC under X-ray illumination, nanometer-scale particles were directly grown via mechanosynthesis.Mechanosynthesis instigates the synthesis process via the milling of precursor materials and milling media.As the energy required for the process is provided by the individual collisions rather than from external heating or solvents required for other synthesis procedures, [30] factors such as the crystal quality and resulting scintillator light yield may be optimized by varying factors such as the milling media size or milling speed.This process has been used previously to directly synthesize perovskite materials such as CsPbX 3 , [31] MAPbI 3 , [32] and BiFeO 3 , [33] with the size of the The inset shows the pellet cross section at higher magnification.f) XRD pattern of a milled CANIBIC crystal (blue), overlaid with a simulated CANIBIC pattern (red).The inset shows the shifting of the 34°peak with respect to Bi 3 + loading. [28]chanosynthesised nanoparticles ranging from 7 nm to 200 μm depending on the milling parameters.Recent work by Xu et al. reports on the mechanosynthesis of Cs 2 AgInCl 6 doped with Na + and Bi 3 + , resulting in a bulk material. [34]e present a study that has improved our understanding of CANIBIC, comparing the responses of hydrothermally grown crystals, sintered pellets, directly mechanosynthesised nanocrystals and plastic nanocomposites in response to both laser excitation and X-ray induced radioluminescence.

Composition
Bulk crystals of CANIBIC were synthesized using the hydrothermal reaction described in the Experimental Section, yielding millimeter-scale single crystals.A photo of CANIBIC crystals scintillating under ultraviolet (UV) illumination is shown in Figure 2a.Scanning electron microscope (SEM) images of the surface of a CANIBIC crystal can be seen in Figure 2b, showing a well-ordered layered structure on the surface of the asgrown crystals.
In this material, the general form of the double perovskite structure Cs 2 B I B III Cl 6 has been altered by having Ag and Na atoms on the B I site and In and Bi on the B III site to form the material Cs 2 Ag 0.6 Na 0.4 In 0.85 Bi 0.15 Cl 6 .The optimal ratios of 40% Ag and 15% Bi have been previously reported by Zhu et al., [28] which achieve the maximum light yield from photoluminescence and scintillation processes, combined with high X-ray mass attenuation due to the addition of Bi.
The relatively complex stoichiometry of the CANIBIC material was investigated using energy dispersive X-ray spectroscopy (EDS) measurements, as shown in Figure 2c.To confirm the required optimal elemental composition of the crystals, multiple EDS measurements were taken with an SEM utilizing a circular backscatter (CBS) detector.EDS measurements were carried out on both single crystal samples and nanocrystals that were fabricated by mechanosynthesis (see Section 3).Table 1 shows the measured composition values x and y corresponding to the relative cation ratios, which are a close match to the required optimal compositions.The composition would be further corroborated with the compositional and scintillation measurements described in the following sections.
To realize millimeter-thick X-ray scintillators, pellets of CANIBIC material were fabricated by conventional sintering.Asgrown crystals were milled into a powder consisting of particles with an average radius of 600 nm and pressed into 10 mm diameter and 1 mm thick wafers.A pellet scintillating under UV illumination is shown in Figure 2d.The pellet was observed to have a smooth surface and to be opaque to visible light.The microstructure of the pressed pellet was investigated by cross-sectional SEM imaging, as shown in Figure 2e.The image confirms a dense and homogeneous material structure with minimal voids, minimizing optical scattering of the scintillation light.The inset image shows that the pellet material is composed of grains with dimensions of the order of 10 μm.

XRD Analysis
The X-ray diffraction (XRD) pattern of a milled bulk powder of CANIBIC crystals was obtained and can be seen in Figure 2f.The inset shows the shift in the 33.7°-34.2°peakwith the change in Bi 3 + due to the shifting geometry of the crystal lattice as shown by Zhu et al. [28] It can be seen that the sample measured yields a peak at ≈34°, indicating a 15% Bi 3 + atomic ratio relative to In 3 + .This was the desired ratio for optimized radioluminescence.CANIBIC is a relatively novel composition with no recognized powder diffraction card standard.As such, a computational crystallographic model was constructed using CrystalMaker X in order to understand and verify the phase formed in this complex, multi-elemental material system.The CANIBIC space group at room temperature is anticipated to be Fm-3m, with a prototypical chemical formula of A 2 B I B III X 6 .The crystallographic model developed describes these occupancy conditions as probabilities rather than specifically discretizing each individual lattice site.This occupancy probability approach is more realistic than what would be expected in the physical CANIBIC powder, as both the orientation and dopant concentration are randomly distributed through the sample.From this model, a diffraction pattern was calculated and overlaid with the experimental XRD spectra collected from ball-milled CANIBIC powder, also shown in Figure 2f.
As observed in Figure 2f, the calculated and experimental XRD spectra align extremely well, with all peaks observed in the spectra being described and predicted by the calculation.There does not appear to be any substantial peak shifting present, suggesting that the site occupancy of the synthesized powders likely matches the intended stoichiometry.Additionally, there are no major deviations in observed peak intensity when comparing the model and experimental data, with subtle differences in peak height likely associated with varying preferred orientations in the as-synthesized powder.This well-fitted calculated spectrum confirms the phase of the synthesised CANIBIC powder to be the cubic Fm-3m space group, with a lattice parameter of a = 10.55 Å.This description of the crystallographic phase is also consistent with previous work reported in literature on CANIC and CANIBIC material systems. [10,15,16,21,28,29,34]As such, the developed crystallographic model appears to accurately describe the CANIBIC powders synthesized in this work and can be utilized as a tool to predict and describe crystal structure and lattice parameters in future work exploring other unique double-halide perovskite compositions.

Nanoparticles
Nanometer-sized particles of CANIBIC were then directly mechanosynthesized; this process did not require corrosive acids and was completed in under 20% of the time required for the bulk hydrothermal synthesis.A vial of the milled material dispersed in toluene is shown in Figure 3a, emitting yellow light under UV illumination.
The synthesized material was separated by size, and imaging was performed using scanning transmission electron microscope (STEM) and SEM techniques to visualize nanoparticle size, shape, and morphology and qualitatively assess composition.Figure 3b shows images from these measurements, clearly showing the wide range of CANIBIC nanocrystal particle sizes produced by mechanosynthesis.The smaller synthesized CANIBIC nanoparticles in Figure 3b(i) exhibit a pill-shaped, acicular particle shape with varying aspect ratios.These nanoparticles contain smaller circular precipitates within the primary acicular particle that exhibit large degrees of visual contrast, suggesting that there are localized regions with either a higher-Z composition or increased thickness.The larger material shown in Figure 3b(ii) appears to have a similar shape as the millimeter-sized single crystals, although now with an average particle size of 500 nm.
To confirm the composition of the various regions within the smaller nanoparticles, EDS spectra were collected on various CANIBIC nanoparticles, shown in Figure S1, Supporting Information.The EDS maps clearly suggest that the small circular precipitates are composed predominately of Ag.Cs appears to be uniformly distributed throughout the acicular particle.Interestingly, the other B I and B III lattice site expected compositions namely Na, In, and Bi, appear to have segregated out of the nanoparticles and exist predominately on the exterior spacing between the nanoparticles.This compositional segregation of Ag precipitates and Na, In, and Bi exteriors is consistent for particles with smaller aspect ratios, which suggests that phase segregation is likely due to competition between strain energy and surface free energy.These results demonstrate that the nanoparticles formed via ball milling are not compositionally homogenous and likely suggest that there is localized phase segregation within the nanoparticles themselves, or a degree of incomplete synthesis.Further to the STEM imaging, the XRD pattern of a drop cast sample of mechanosynthesized nanoparticles was obtained and is shown in Figure S2, Supporting Information.It can be seen that the CANIBIC pattern observed in the bulk crystal XRD measurement is present, alongside CsAgCl 2 and unreacted salts.Nonetheless, these results confirm the presence of sub-50 nm-sized particles consisting of an acicular particle shape with localized regions of compositional segregation, with Ag segregating to circular precipitates within the acicular particle and the remaining B′ and B′′ elements segregating the acicular particle exterior.
The images collected via STEM suggest that there is compositional heterogeneity throughout the pill-shaped nanoparticle structures, which likely corresponds with the formation of several phases in the nanoparticle suspension.This contradicts the experimental XRD spectra collected on the polycrystalline powder which clearly show double-perovskite phase formation without any secondary phases present.This discrepancy in the observed phases present in the powder and the nanoparticles is likely attributed to the particle size effects on surface energy and subsequent phase formation energies.The XRD spectrum shown in Figure 2f was collected on CANIBIC powder, with average particle sizes on the scale of microns.In this case, the volume component of the overall phase formation energy dominates and follows the predicted thermodynamically stable double perovskite phase.As the size of the particles decreases toward the nanometer scale, the surface energy component dominates, which likely alters phase formation energy and results in the observed phase segregation being more energetically favorable than maintaining a single phase.The addition of surfactants used during the milling process to synthesize the nanoparticles may also play a role in altering the phase formation energies, as well as causing a limited amount of re-dissolution and precipitation of some components.Future work will be needed to identify the specific phase formation mechanism in the pill-shaped nanoparticles and the various factors that control this phase segregation including nanoparticle size, composition, and surfactant concentration.However, these results demonstrate the presence of complex phase formation energies present in these multi-component double-perovskite nanoparticles and highlight the dependency of volume/strain and surface energies on energetically favorable phase formation.
As a confirmation of the particle sizes, dynamic light scattering (DLS) measurements of the CANIBIC dispersed in toluene were taken.Figure 3c shows DLS measurements of particle size from size-separated samples of material, corroborating the particle sizes seen in the direct imaging.From this, it was determined that the mechanosynthesis process creates nanoparticles with a range of sizes, from 10-1000 nm, with diminishing mass yields of material as the size decreases observed in the dispersions.
To investigate the response of the nanoparticles in a solid matrix and to obtain reliable radiation measurements that would otherwise be difficult to perform with dispersed particles, nanocomposite plastic disks were produced.The plastic used was the optically transparent and chemically inert polymethyl methacrylate (PMMA), and the nanoparticles were incorporated at 1% mass loading.This amount was chosen to balance the requirement for a high mass attenuation coefficient for photoelectric absorption and the need to minimize incoherent scattering, which would otherwise have a detrimental effect on the optical transmission of the nanocomposite materials.Figure 3d shows a 3.8 cm diameter composite disk loaded with 500 nm radii CANIBIC nanocrystals.The disk is scintillating under UV illumination and can be seen to be transparent.SEM images of these disks were taken to determine the homogeneity of the sample.Figure 3e shows SEM images taken with the CBS detector of the surface of a nanocomposite disk.The white particles visible in the disk are the agglomerates of CANIBIC nanocrystals, and it can be seen that there is an even distribution of the material throughout the PMMA disk.To confirm that the mixed-cation double perovskite cation ratios were preserved in this different method of synthesis, multiple EDS measurements were taken with an SEM utilizing the CBS detector.EDS spectra for both large and small drop casts of nanocrystals are shown in Figure 3f, and the atomic percentages of the elements are given in Table 1.It can be seen that the elemental ratios for the mechanosynthesized material are consistent with atomic percentages measured from the CANIBIC crystals grown via hydrothermal reactions.

Photoluminescence
The optical emission and absorption properties of the CANIBIC materials were characterized using room temperature PL with a 355 nm laser and photoluminescence excitation (PLE) spectra.Figure 4a shows optical data obtained from four different forms of the CANIBIC materials: i) as-grown single crystals; ii) pellets formed from conventionally-sintered bulk powder; iii) large diameter nanocrystals in suspension; and iv) small diameter nanocrystals in suspension.For each sample, a set of three PLE spectra were acquired with emission wavelengths of 550, 625, and 700 nm, where 625 nm represents a wavelength close to the peak PL emission intensity.The peak PL emission wavelengths and PLE absorption edges are listed for each sample in Table 2, together with the corresponding Stokes shifts.The PL emissions across the four sample types were very similar, with a broad "white-light" emission centered at 630 nm with a full width half maximum (FWHM) of 200 nm, which is typical of that observed from STE emission.
The excitation spectra show the same strong absorption edge for each of the four emission wavelengths, confirming that the full width of the emitted peak originates from the same excitation processes.The data show a large separation between the absorption edge and emission peak, again consistent with PL measurements from other STE materials.Across the four different material types, there is a minimal shift in the peak emission wavelength.By contrast, the absorption edge shifts to lower wavelengths as the material type moves from single crystal to nanocrystals, such that the bandgap energy increases from 3.14 eV for bulk material to 3.44 eV for small-dimension NCs.The corresponding Stokes shift increases from 1.18∼ ± ∼0.02 to 1.46 ± ∼0.02 eV as the particle size decreases due to the corresponding reduction in wavelength of the absorption edge.This is evidence that the band gap of CANIBIC is widening with the decrease in particle size, whilst the scintillation spectra remain consistent as the STE emission mechanism is not affected by this change in size.The observed shift in absorption wavelength is consistent with the expected increase in band-gap energy when moving  from bulk to nanocrystal materials, as has been frequently reported in other nanoparticle perovskite systems.For example, the room temperature band-gap energy in single crystal bulk CsPbBr 3 is 2.41 eV, increasing to 2.60 eV for nanoparticles of 5 nm diameter. [35,36].Further confirmation of the STE origin of the emitted luminescence was obtained by investigating the PL intensity variation as a function of the incident laser power intensity.Figure 4b shows the proportional change in PL intensity as a function of laser power, which is expected from STE emission.Optical emission originating from defect-induced radiative recombination would have a nonlinear PL intensity as a function of laser power due to the saturation of the defect sites. [14]he scintillation decay time, combined with the lack of a long-lived afterglow, is an important parameter for an X-ray scintillator.Double perovskite materials that luminesce through STE emission are generally known to exhibit decay times of the order of microseconds, in contrast to the faster nanosecond scintillation of lead halide materials.Time-resolved PL decay spectra were acquired from these CANIBIC materials using a time-correlated single photon counting (TCSPC) system containing a 405 nm pulsed laser with a 40 kHz repetition rate.The TCSPC decay spectra, which are shown in Figure 4c were fitted with a bi-exponential decay equation to give the lifetime and intensity values shown in Table 3.These are consistent with CANIBIC pellets with 15% Bi 3 + loading, as measured by Zhu et al. [28] All of the materials have a dominant decay component at ≈1.5 μs and a shorter, lower-intensity decay component at 400 ns.Previous studies have speculated that the origins of this split decay profile are In 3 + defects. [29]

X-Ray-Induced Radioluminescence
X-ray RL spectra were obtained to assess the intensity and spectral shape of the scintillation light and to characterize the response of the CANIBIC material to ionizing radiation.Figure 5a shows the normalized RL emission spectra acquired for a CANIBIC pellet under illumination by an 80 kV, 100 μA X-ray source.It can be seen that the spectral shape of the RL emission is analogous to the corresponding PL emissions, with the characteristic broadband STE emissions centered at ≈630 nm.
The intensity of the emitted RL is dependent on several interrelated factors, namely: 1) the quantum efficiency of the material for absorption of X-rays, parameterized by the X-ray mean absorption length ; 2) the light yield of the material for ionizing radiation, normally expressed in terms of photons/MeV; and 3) the optical transparency of the material to permit efficient emission of the created scintillation light.As shown in Table 5, the mass attenuation coefficient, mean absorption length, and quantum efficiency for 1 mm thickness of CANIBIC were calculated using the NIST XCOM database.For a photon energy of 60 keV, the mean attenuation length  = 0.48 mm, so that the X-ray interaction is fairly uniform throughout a 1 mm thick scintillator sample.This contrasts with PL measurements, where the excitation photons with an energy above that of the band-gap are absorbed within nanometers of the material surface.Consequently, the overall scintillation brightness from ionizing radiation is strongly dependent on the optical transparency of the scintillator and the ability of the scintillation light to be emitted from the sample.Good optical transparency is primarily achieved by minimizing selfabsorption in the material (through a high Stokes shift) and by reducing optical scattering from crystallite boundaries and defects in the material.In the case of the CANIBIC/PMMA nanocomposite disk, the perovskite mass loading in the PMMA is typically low (e.g., at 1 wt%); however, the optical transparency of the thin PMMA disk is high.Therefore, the effective brightness of the nanocomposite disk is high, despite its low quantum efficiency.
To further understand the X-ray response of the material, the scintillation brightness was measured at various dose rates.CANIBIC crystals, pellets, and disks were illuminated by an  X-ray tube set to a 40 kV voltage, and the tube current was varied between 10 and 200 μA.The scintillation light was collected by a photomultiplier tube (PMT), with the signal representing the total contribution of light from all visible wavelengths.The scintillator response is given in Figure 5b, with all three samples showing a linear response to an increasing X-ray dose.It can be noted that the nanocomposite disk has 50% of the signal amplitude of the pellet in this measurement, displaying the benefit of the higher transparency in the nanocomposite material.These CANIBIC nanocomposite disks are particularly well suited for applications such as X-ray imaging and beam monitoring where thin scintillators are required.
To compare with the decay times from time-resolved PL emission, the time-resolved RL of a CANIBIC pellet was performed using a 40 kV laser-pulsed X-ray source.The RL decay spectra and fit are shown in Figure 5c, with the fit values presented in Table 4 using a triexponential fit.Multiple microsecond components are observed in these spectra, with the most intense component having a decay time of 1.4 μs, decay consistent with the time-resolved PL measurements.Two weaker, longer decay components are also observed, including a very weak 16 μs component with an intensity of 0.4%.This lack of long-lived afterglow observed in this material, for example, at the millisecond time range, is encouraging for use as a scintillator and compares well with other double perovskite scintillators such as Rb 2 CuBr 3 , which shows 2.7% intensity at 20 ms. [17]CANIBIC's afterglow of 0.1% at 0.1 ms is a significant reduction when compared to 1.5% at 3 ms for Tl doped CsI. [28]Reducing afterglow in scintillators helps reduce signal pileup and improves the responses of materials used in real-time X-ray imaging applications.

Gamma Ray Characterization
Gamma ray spectroscopy measurements were performed with a CANIBIC pellet using 241 Am, 57 Co, and 137 Cs gamma sources, with primary emission at energies of 59.6, 122, and 662 keV.
The pulse height spectra are shown in Figure 6a.For each of the gamma ray energies, the mean absorption length and quantum efficiency for 1 mm of CANIBIC material were determined, as shown in Table 5.As discussed previously, the mass attenuation coefficient for each energy is the sum of the attenuation coefficients for all photon interaction permitted processes (i.e., photoelectric absorption, Compton scattering, and Raleigh scattering).The pulse height spectra show a continuum of events for all the sources measured.In the case of 59.6 keV emission from 241 Am and 122 keV emission from 57 Co photoelectric absorption is the dominant interaction in 1 mm of scintillator, although no resolved full energy peak is observed.In the case of 662 keV emission from 137 Cs the dominant interaction is Compton scattering, where the pulse height spectrum will be predominantly a continuum.The data shows a clear increase in channel with increasing gamma ray energy, confirming the expected relationship between photon energy and scintillator brightness.Figure S3, Supporting Information, shows a comparison of 241 Am gamma spectra taken with a CANIBIC pellet and a nanocomposite disk with 1% mass loading of CANIBIC.The materials' mass attenuation coefficients at 59.5 keV and the calculated percentage interaction chance are shown in Table S1, Supporting Information.It can be seen that the two spectra end at approximately the same energy, showing that at reduced loading and for CANIBIC particles of different sizes, the number of photons created per event is consistent.The disk has 46% of the total light yield of the pellet.This is higher than expected, as the quantum efficiency of the nanocomposite disk was calculated to be 3% that of the pellet.This shows the importance of the transmission of the sample, as the low self-absorption of the composite disks results in a relatively higher effective brightness.This is consistent with the result for X-ray sensitivity at 40 kV discussed previously.Figure 6b shows the mass attenuation coefficients for photoelectric absorption and total absorption for a 1% mass-loaded PMMA disk, a 10% loaded disk, and a pellet.Optimizing the stopping power of the material without losing light yield due to internal absorption is a key parameter to investigate.

Conclusion
This paper has described the work performed in the development of the mixed-cation double perovskite CANIBIC.Single crystals of the material were grown via a hydrothermal reaction, and the compositional and scintillation properties were characterized to ensure consistency with previous work.The underlying crystal lattice structure was modeled and measured via XRD analysis for the first time, expanding our understanding of the double perovskite structure.This material was milled and pressed into pellets to create standard samples, with the fabrication process optimized to generate the greatest light yield.These pellets have an improved X-ray response compared to the crystals, showing the benefit of this refining process.The STE emission mechanism of this material was independently confirmed via PL emission and excitation analysis, further corroborating the body of work on this emission regime.
Taking this material a step further, discrete nanocrystals of CANIBIC were successfully mechanosynthesized for the first time.This displays the exciting possibilities mechanosynthesis offers, as no external heating or solvents are required for the synthesis, lowering both material costs and the risk of using hazardous materials.This procedure was iterated to increase the light yield of the dispersions and lower the particle size as we attempted to reach the Bohr radius to investigate the quantum confinement effects that would result.This process resulted in the synthesis of nanocrystals in the size range of 10-1000 nm in size, which were size separated into two distinct samples.The elemental composition of these different-sized materials was confirmed to be consistent via EDS analysis, allowing the comparison of different particle sizes on the scintillation properties of CANIBIC.The absorption edge of the material was observed to shift with particle size with no change to the emitted light spectra, indicative of the STE emission mechanism.For higher energy measurements, 600 nm radii nanocrystals were loaded into PMMA plastic disks at 1% mass loading, and the scintillation properties of the crystals, pellets, nanocrystal dispersions, and nanocomposite disks were compared.
The PL and RL emissions were consistent for all four forms of CANIBIC even with the shifting absorption edge, with the smallest, 10 nm radii nanocrystals outperforming in terms of light yield given the relative amount of material present.Comparing the X-ray sensitivity of the pellets and disks shows that the disks outperform the pellets relative to the mass loading of the samples.This is ascribed to the benefits of the disks' high transmission relative to the pellets due to their lower self-absorption from internal scattering.This shows that nanocomposite disks are the more economical and practical standard sample for X-ray measurements.

Experimental Section
Bulk Synthesis Procedure and Pellet Fabrication: CsCl, AgCl, NaCl, InCl 3 , and BiCl 3 were purchased from Merck.The synthesis was performed following the procedure outlined in Zhu et al. [28] For 2 mmol of the final product, 4 mmol of CsCl, 1.2 mmol of AgCl, 0.8 mmol of NaCl, 1.7 mmol of InCl 3 , and 0.3 mmol of BiCl 3 were mixed in 0.1 mol of a 10 m HCl solution.This mixture was poured into an acid digestion vessel and heated to 180 °C for 12 h, before returning to room temperature at a rate of 3 °C h −1 .The resulting crystals were extracted from the remaining acid, washed with isopropyl alcohol, and dried in a vacuum oven at 60 °C until dry.
For the fabrication of CANIBIC pellets, crystals of CANIBIC were milled in a planetary ball miller at 700 RPM for 5 min.The powder was removed from the milling vessel and pressed into a 10 mm diameter die for 5 min at 100 MPa.
Mechanosynthesis and Nanocomposite Fabrication Procedure: As with the single crystal synthesis process, using precursors purchased from Merck for 2 mmol of material 4 mmol of CsCl, 1.2 mmol of AgCl, 0.8 mmol of NaCl, 1.7 mmol of InCl 3 , and 0.3 mmol of BiCl 3 were placed in a zirconium oxide milling vessel with 2 mm diameter zirconium oxide milling balls, weighing 90% of the milling mass by weight.The precursors were milled for several hours until they had formed a fine powder of CANIBIC, whereupon 1 mL of oleylamine was added to function as a surfactant and encourage smaller particle formation, and milled for several more hours.The resulting nanometer scale CANIBIC crystals were then dispersed in toluene and separated by size for measurements via sedimentation.
The PMMA composites were fabricated by mixing 1.5 g of PMMA (Aldrich) with a dispersion containing 15 mg of the dispersed CANIBIC nanocrystals in a vial with 3 mL of toluene.The mixture was then heated at 56 °C for 1 h on a hot plate and stirred using a magnetic stirrer.The resulting mixture was then sonicated for 30 min to remove bubbles from the mixture and poured into 2 mm deep, 38 mm diameter molds, and left to cast over two days, covered to allow the toluene to recirculate and prevent the surface from distorting.The resulting composite disks were then removed from the molds.
Experimental Procedures: SEM imaging and EDS measurements were taken with a Thermo Scientific Apreo 2 SEM (Thermo Fisher Scientific).STEM images were collected on a ThermoFisher Talos F200X transmission electron microscope operated at an accelerating voltage of 80 kV.HAADF-STEM images were collected on an aberration-corrected ThermoFisher Titan 3 G2 60-300 with a monochromator and an X-field emission gun source used with an accelerating voltage of 80 kV.
XRD patterns were obtained with an X'Pert X-ray diffraction platform (Malvern Panalytical).The simulated XRD spectra were obtained using CrystalMaker X.To ensure simulation accuracy, a model was first constructed on the Cs 2 AgBiBr 6 double perovskite crystal system, which has readily available refined crystallographic parameters. [37][40] This developed model was then altered to describe the CANIBIC occupancy.Several iterations of the model were completed to refine the lattice parameter and accurately describe the experimental XRD spectra.
DLS measurements were taken using a Zetasizer Nano S (Malvern Instruments), with the dispersion contained in square glass cuvettes.Gamma spectroscopy was performed with the samples mounted to a 1.5in.ET 9256kB PMT (ET Enterpises) with optical coupling gel, connected to an Easy-MCA (Ortec).
PL spectra were taken with a 355 nm YAG laser and recorded using a Maya 2000 Pro spectrometer (Ocean Insight) with a 420 nm long-pass filter covering the collecting lens.For low-temperature measurements, drop cast samples of the dispersions were mounted to a cryostat stage connected to a heating plate and liquid nitrogen supply, with the temperature set using a Temperature Measurement Control Unit TIC 304-MA (Cry-oVac).PLE spectra were obtained with a 250-1000 nm xenon lamp (XE1).Double monochromators were in place of both the excitation and emission light paths, with the light collected in an FLS980 spectrometer (Edinburgh Instruments).
Time-resolved PL decay spectra were measured using a PicoQuant fluorescence lifetime spectrometer (PicoQuant), illuminated by a 405 nm laser pulsed with a DDS Function Generator (Aim-TTi).A 420 nm long pass filter covered the collecting lens.RL measurements were obtained with a QE 6500 spectrometer (Ocean Insight), connected to an integrating sphere (Labsphere) to ensure consistency of measurement geometry.The samples were irradiated by an X-ray tube (Hamamatsu) set to 80 kV and 100 μA.The pulsed RL decay measurements were acquired with a laser-excited 40 kV pulsed X-ray source N5084 (Hamamatsu Photonics) coupled to an FLS1000 Photoluminescence Spectrometer (Edinburgh Instruments).
X-ray sensitivity data were taken in a dark box, with the samples mounted to a 90°mirror stand, and light collected with a 1.5-in.PMT (ET Enterprises).The X-ray source was a Mini-X X-ray tube, with the voltage set to 40 kV and the dose varied by adjusting the current between 10 and 200 μA in 10 μA steps.The beam was collimated through a 1 mm diameter mask.
Gamma spectroscopy was performed with samples mounted to a 1.5in.PMT (ET Enterprises) within a light-tight box surrounded by a lead castle.Various sources were positioned adjacent to this setup, with the output of the PMT processed through a data acquisition system containing a preamplifier, amplifier, oscilloscope, and multi-channel analyzer (Ortec).

Figure 1 .
Figure 1.a) Structure of a mixed-cation double perovskite.The red sphere represents the A site cation; the blue, cyan, yellow, and green spheres represent the varying B site cations; and the purple spheres represent the halide anions.b) Nuclear coordinate diagram of the STE emission mechanism (E g refers to the band gap and E PL to the emission energies).c) Mass attenuation coefficient plots for NaI, GADOX, CsPbBr 3 , and CANIBIC.

Figure 2 .
Figure 2. a) Crystals of CANIBIC scintillating under UV illumination at 390 nm.b) SEM image of a CANIBIC crystal surface.c) EDS spectrum of a CANIBIC single crystal.The relevant peaks have been labeled.d) A 10 mm diameter CANIBIC pellet scintillating under UV illumination.e) SEM image of a CANIBIC pellet cross section.The inset shows the pellet cross section at higher magnification.f) XRD pattern of a milled CANIBIC crystal (blue), overlaid with a simulated CANIBIC pattern (red).The inset shows the shifting of the 34°peak with respect to Bi 3 + loading.[28]

Figure 3 .
Figure 3. a) A vial of mechanosynthesized CANIBIC nanoparticles suspended in toluene, scintillating under UV illumination.b) Images of drop casts of size separated nanocrystals: STEM image of small nanocrystals (i).SEM image of a drop cast of large CANIBIC nanoparticles (ii).c) Size distribution of a suspension of CANIBIC NCs measured using DLS.d) A 3.8 cm diameter plastic disk loaded with CANIBIC NCs, scintillating under UV illumination.e) SEM image of a PMMA disk loaded with CANIBIC nanoparticles.The white particles in the image are the perovskite nanocrystals spread across the disk.f) EDS spectra of drop casts of size separated CANIBIC NCs.The relevant peaks have been labeled.

Figure 4 .
Figure 4. a) Normalized PL emission under illumination by a 355 nm laser and excitation spectra measured at different wavelengths for CANIBIC crystals, pellets, and large and small nanocrystal dispersions.b) PL intensity for a CANIBIC pellet at varying laser excitation intensity.c) Normalized time decay profiles for CANIBIC crystals, pellets, and large nanocrystal dispersion under illumination by a pulsed 405 nm laser.

Figure 5 .
Figure 5. a) Normalized RL spectra for a CANIBIC pellet illuminated by an 80 kV, 100 μA X-ray source.b) RL intensity of CANIBIC crystals, pellets, and nanocomposite disks as measured by a PMT at varying dose rates.c) Time decay profiles of a CANIBIC pellet under illumination by a 40 kV pulsed X-ray source, fitted with a triexponential decay curve.

Figure 6 .
Figure 6.a) Gamma spectra taken with a 1 mm thick CANIBIC pellet.b) Relative mass attenuation coefficients of photoelectric absorption and total attenuation for different loadings of CANIBIC in PMMA.

Table 1 .
Elemental composition of the Cs 2 Ag x Na (1 − x) In y Bi (1 − y) Cl 6 materials via EDS analysis.

Table 2 .
CANIBIC materials PL and PLE data.

Table 3 .
PL lifetimes of CANIBIC samples, taken from Figure4cand fit with a biexponential curve.

Table 5 .
Mass attenuation coefficient for photoelectric absorption, mean absorption length, and percentage quantum efficiency for a 1 mm thick CANIBIC pellet.
X/Gamma ray energy[eV]