Main group ions with an ns 2 np 0 configuration, like Sb 3+ and Bi 3+ , are classic activator ions used in a variety of phosphors and scintillators. 1 The magnitude of the luminescent Stokes shift for these ions can vary greatly depending on the degree of reorganization that occurs in the excited state, which in turn depends upon the structure of the host. When an activator ion is placed on a site that is compressed with respect to its preferred environment, reorganization of the excited state is suppressed, leading to a small Stokes shift. Conversely, if it is placed on a large site, relaxation of the coordination sphere of the activator ion in the excited state can be extensive, resulting in a large Stokes shift. Chloride double perovskites, with the general formula Cs 2 MM'Cl 6 are favorable phosphor host structures for a variety of luminescent centers, including Mn 2+ , Yb 3+ , Eu 3+ , and Cr 3+ . [2][3][4][5][6][7][8] For example, in a pair of recent reports, Tang et al. substituted Sb 3+ and Bi 3+ into the vacancy-ordered double perovskite Cs 2 SnCl 6 , and observed photoluminescence (PL). 9,10 Incorporating up to B1 mol% of Sb 3+ produced orange-red light (l max = 602 nm) with a maximum photoluminescent quantum yield (PLQY) of 37% whereas doping with Bi 3+ at levels that approach 7 mol% resulted in a blue emission (l max = 455 nm) with a PLQY as high as 80%. Nanocrystals of Cs 2 SnCl 6 :Sb 3+ have also been prepared that show dual emission: a low temperature blue emission (l max = 438 nm) that disappears upon warming to room temperature and an orange-red emission at 615 nm that is only present when Sb 3+ is introduced. 11 The quantum efficiency of the nanocrystals is relatively low (PLQY = 8.3%).
The PL characteristics of Sb 3+ and Bi 3+ are known to be highly sensitive to their crystallographic environment, but when these ions are doped into Cs 2 SnCl 6 , the coordination environment is unclear. The aliovalent doping of the trivalent ions for Sn 4+ requires a compensating charged defect. Recent DFT calculations have suggested the likeliest defect is a chloride vacancy. 10 If the Sb 3+ /Bi 3+ ion and the chloride vacancy are located in the same octahedron, it will reduce the coordination environment around the dopant ion from a 6-coordinate octahedron to a 5-coordinate square pyramid. To better understand the structureproperty relationships in this class of materials, it would be highly desirable to study a compound where the trivalent ion coordination environment is unambiguous. Cs 2 M + M 3+ Cl 6 double perovskites are an obvious choice because an isovalent substitution for M 3+ does not require a compensating defect.
Studies of double perovskite hosts have been difficult because many of these compounds have intrinsic moisture instabilities. 12,13 For example, the double perovskites Cs 2 NaMCl 6 :Sb 3+ and Cs 2 NaMBr 6 :Sb 3+ (M = Sc, Y, La) are incredibly hygroscopic, complicating characterization and limiting potential applications. 12,13 Attempts by Blasse et al. to synthesize Cs 2 NaSbX 6 (X = Cl, Br) were performed under ultra-dry conditions, but measurements indicated the presence of the thermodynamically stable Cs 3 Sb 2 X 9 phases. While this phase is non-luminescent it could provide a non-radiative decay pathway. It seems that the incorporation of significant amounts of Sb 3+ rapidly destabilizes most Cs 2 NaMX 6 (X = Cl, Br) systems, which necessitates the investigation of alternative host structures. 14,15 Developing a chemically stable double perovskite with an isovalent doping site for the Sb 3+ cation would alleviate the need for charge compensating chloride vacancy defects, allowing comparative analysis of the effect of coordination number on the photoluminescent properties. This goal led us to study the optical properties of Sb 3+ ions doped into the lead-free halide double perovskite host Cs 2 NaInCl 6 . Herein, we show that Cs 2 NaInCl 6 :Sb 3+ is not only useful as a model compound, it is a promising rareearth free blue phosphor.
Cs 2 NaInCl 6 :Sb 3+ was synthesized by precipitation from an HCl(aq) solution. For a typical synthesis, 20.0 mL of concentrated HCl(aq) (Fisher Scientific, 37%) and 2.0 mL of phosphinic acid (H 3 PO 2 , Sigma-Aldrich, 50 wt% in H 2 O) were heated to 80 1C. Caution should be taken when heating, as phosphinic acid can undergo autoignition upon decomposition into phosphine gas, which occurs near 110 1C. To this solution, 1.00 mmol of In 2 O 3 (Alfa Aesar, 99.994%), 2.00 mmol of NaCl (GFS Chemicals, 99%), and varying amounts (0.001-1.00 mmol) of Sb 2 O 3 (Acros Organics, 99+%) were added. The solution was heated and stirred to allow the reagents to dissolve. Next, 4.00 mmol of CsCl (Acros Organics, 99+%) was added, immediately triggering the precipitation of Cs 2 NaInCl 6 :Sb 3+ . This reaction can be described by eqn (1):
The precipitate was then filtered using a porous fritted funnel, washed several times with neat ethanol (Decon Labs Inc., 200 proof), and dried overnight via vacuum filtration.
Powder X-ray diffraction (PXRD) data were collected on a Bruker D8 Advance powder diffractometer (40 kV, 40 mA, sealed Cu X-ray tube) equipped with a Lynxeye XE-T position-sensitive detector. The data were collected with an incident beam monochromator (Johansson type SiO 2 -crystal) that selects only Cu Ka 1 radiation (l = 1.5406 Å). Rietveld refinements of laboratory PXRD data were carried out using the TOPAS-Academic (Version 6) software package to determine the crystal structure. Thermogravimetric analysis (TGA) was performed on a Thermogravimetric Analyzer TGA Q50. Samples were heated under a nitrogen stream of 50 mL per minute with a heating rate of 25 1C per minute between 25 1C and 900 1C.
UV-visible diffuse reflectance spectroscopy (DRS) data were collected from 178-890 nm with an Ocean Optics USB4000 spectrometer equipped with a Toshiba TCD1304AP (3648-element linear silicon CCD array). The spectrometer was used with an Ocean Optics DH-2000-BAL deuterium and halogen UV-vis-NIR light source and a 400 mm R400-7-ANGLE-VIS reflectance probe. The detector was calibrated using a Spectralon Diffuse Reflectance Standard.
Steady-state photoluminescence (PL) data were obtained using a Jovin Horiba FluoroMax4 (xenon source, 0.5 nm excitation and emission slit widths, 1 nm step size) equipped with a solid-state sample holder. Luminescent data was analyzed using the FluorEssence (v3.5) software powered by Origin. Temperature-dependent emission spectra were collected by mixing the samples with an optically transparent silicone resin (GE Silicones, RTV615) and depositing the combination onto a quartz slide (Chemglass). Then, using a Janis cryostat (VPF-100) for a temperature-controlled environment from 300-600 K, the emission was measured in 20 K increments (l ex = 340 nm). Photoluminescent lifetime data was collected using a Horiba DeltaFlex System with a NanoLED N-330 nm (l ex = 330 nm). Internal photoluminescent quantum yield (PLQY) measurements were performed with a Jovin Horiba FluoroMax4 equipped with a Quanta-j integrating sphere (15 cm) and a PTFE sample cup. BaSO 4 powder dispersed in a silicone resin was used as the blank reference sample. All samples and the blank were excited at 335 nm (l max ) and the absorbance and luminescence signals integrated from 325-345 nm and 370-570 nm, respectively. Radiometric, sphere, and dark count corrections were applied during data acquisition, while corrections for filters and integration time differences were applied in the FluorEssencet analysis package for Quantum Yield (FluorEssence v3.8.0.60, Origin v8.6001). Additional details are available in the ESI. † A white light-emitting phosphor-converted light emitting diode (pc-LED) device incorporating the Cs 2 NaInCl 6 :Sb 3+ phosphor was mixed with lab-made red-emitting Sr 2 Si 5 N 8 :Eu 2+ and commercially available green-emitting b-SiAlON:Eu 2+ in the same silicon resin mentioned above and cured in a custom brass mold to form a phosphor cap. This cap was placed on a 370 nm LED driven by a 20 mA current, and an AvaSphere-50-IRRAD spectrophotometer was used to obtain the pc-LED luminescence spectrum and performance characteristics.
Solution precipitation yields single-phase samples characteristic of the double perovskite crystal structure with sharp diffraction peaks (Fig. 1). Analysis indicates that all compounds have Fm% 3m crystal symmetry (Fig. S1, S2 and Table S1, ESI †). The In 3+ and Na + sites are fully ordered in Cs 2 NaInCl 6 , evidenced by the presence of a strong (111) peak at B14.51 2y. The rock salt ordering of octahedral-site cations can be imagined as a three dimensional checkerboard ordering of Na + and In 3+ cations separated by bridging chloride ions. Rietveld refinement yielded bond distances of: d(Cs-Cl) = 3.7271(1) Å, d(Na-Cl) = 2.782(4) Å, and d(In-Cl) = 2.485(4) Å. Sb 3+ incorporation does not change the average crystal structure, aside from a slight increase in the lattice parameter. Indeed, the lattice parameters increase to 10.54211(8) Å for the highest concentration of Sb 3+ substitution achieved in a phase-pure sample compared to 10.53344(4) Å in the undoped Cs 2 NaInCl 6 sample (Table 1). Although the 6-coordinate Shannon radius of Sb 3+ (0.90 Å) is slightly smaller than In 3+ (0.94 Å), the increase in lattice parameter with increased Sb 3+ content is consistent with the observation that chloride double perovskites with Sb 3+ have larger lattice parameters than their In 3+ analogs. 15,16 As shown in previous reports, the amount of Sb 3+ incorporated into the structure is much smaller than the Sb/In ratio in solution prior to precipitation. This may be related to the hygroscopicity of the Sb 3+ end-member, Cs 2 NaSbCl 6 . [12][13][14] Since the full Cs 2 NaIn 1Àx Sb x Cl 6 solid solution cannot be prepared, the amount of Sb 3+ incorporated into the Cs 2 NaInCl 6 :Sb 3+ phosphor was estimated using the lattice parameter changes found in the Cs 2 AgIn 1Àx Sb x Cl 6 system. 15 In that report, the lattice parameter increased by 0.195 Å as x increased from 0 to 1. If we assume a similar linear increase in the lattice parameter of the Cs 2 NaIn 1Àx Sb x Cl 6 solid solution, we can estimate the Sb 3+ content in these samples (Table 1).
The product remains phase pure until reaching sample (9), where peaks arising from Cs 3 Sb 2 Cl 9 appear in the powder diffraction patterns shown in Fig. 1. Analyzing the thermal stability of the samples by TGA indicates the samples are thermally stable up to 550 1C, as shown in Fig. S3 (ESI †). Increasing the incorporation of Sb 3+ does not significantly impact the thermal stability; if anything, the thermal stability is slightly enhanced by antimony doping. The samples also did not show any signs of degradation when stored under ambient conditions (room temperature, in air), as evidenced by the lack of changes in the PXRD patterns and optical measurements taken over multiple weeks, (see Fig. S4 and Table S2, ESI †).
The optical properties of the phase-pure samples (1-8) were studied by first analyzing the UV-visible diffuse reflectance spectra (Fig. S5, ESI †). After converting into pseudo-absorbance via the Kubelka-Munk function, 17 provided in eqn (2),
where a is the optical absorption coefficient, and R is the reflectance, the data were replotted in Fig. 2. The undoped Cs 2 NaInCl 6 host is a wide bandgap insulator. The band gap is difficult to estimate with a high degree of precision as it occurs near the short wavelength limit of our spectrometer. Extrapolating the upturn in absorbance that occurs below 250 nm, an estimate of E g E 5.1 eV (Fig. 2) can be made. Previously reported transmission measurements on single crystals found an upturn of absorbance at a slightly lower energy of 4.8 eV. 18 As soon as Sb 3+ is incorporated, characteristic 5s 2 -5s 1 p 1 absorption transitions of the [SbCl 6 ] 3À octahedra appear (Fig. 2 and Fig. S6, ESI †). The broad absorption feature centered B260 nm is assigned as the spin-allowed 1 S 0 -1 P 1 transition, and the peak at 280 nm is assigned to the spin and parity forbidden 1 S 0 -3 P 2 transition. 12 The transitions at 317 nm and 333 nm can be assigned to the 1 S 0 -3 P 1 transition (spin forbidden, parity allowed). This transition is split into two peaks by coupling between the lattice vibrations and empty T 1u excited state orbitals; a dynamic Jahn-Teller distortion of the excited state. 5 The intensity of these transitions increases with increasing Sb 3+ content, corroborating their assignment to absorption from the dopant ion.
Analyzing the photoluminescent excitation spectra provides more information on the electronic structure across this series of compounds (see Fig. 3). The two excitation maxima at 317 nm and 333 nm, which correspond to the Jahn-Teller split 5s 2 -5s 1 p 1 transitions of the [SbCl 6 ] 3À octahedra, lead to the same broad emission. These correspond nicely with the two main absorption features observed in the Kubelka-Munk pseudo-absorption spectra (Fig. 2). The weak excitation observed at 280 nm corresponds to the 1 S 0 -3 P 2 transition. Interestingly the relative intensities of the three peaks are quite different in the absorbance and excitation spectra. Not surprisingly, the efficiency of the emission is much higher when both the excitation and emission are associated with the 1 S 0 -3 P 1 transitions. The 1 S 0 -1 P 1 transition, which produces a strong absorbance at B260 nm, does not lead to emission. This may be related to its proximity in energy to the band edge, which could allow for thermal excitation into the conduction band and subsequent energy migration. It is also possible that a rapid, radiative return to the ground state (fluorescence) with minimal Stokes shift occurs before crossing over to the triplet excited state.
Cs 2 NaInCl 6 doped with Sb 3+ exhibits bright blue photoluminescence (PL), centered at 445 nm with a full-width-athalf-maximum (FWHM) of B80 nm (0.51 eV) (Fig. 4). As observed in other double perovskites doped with Sb 3+ , this blue emission can be attributed to local Sb 3+ excited state relaxation via a 3 P 1 -1 S 0 pathway. The magnitude of the Stokes shift is indicative of the extent of an excited state reorganization of the [SbCl 6 ] 3À octahedra. 12 The relatively small Stokes shift (0.94 eV) observed in this compound follows a trend observed in other Sb 3+ -doped double perovskites, which show an increasing Stokes shift with increasing ionic radii of the 6-coordinate 3+ cation (Table 2). 19 The exception is Cs 2 SnCl 6 :Sb 3+ , where the Stokes shift does not follow the trend extrapolated from the double perovskites. We hypothesize that this is due to the Sb 3+ adopting a 5-coordinate environment ([SbCl 5 ] 2À ), which is not surprising given the attraction between the negatively charged Sb 0 Sn and positively charged V Cl defects. The stereoactive lone pair of Sb 3+ makes the square pyramidal geometry more likely than trigonal bipyramidal. The asymmetric coordination environment allows for larger reorganization of the excited state, and thus, a larger Stokes shift.
Analyzing the optical properties of Cs 2 NaInCl 6 :Sb 3+ reveals an increase in the Stokes shift as the radius of the trivalent ion that Sb 3+ replaces increases. While the extent of structural relaxation around the Sb 3+ dopant is not known, it is reasonable to assume that the antimony ion has more freedom to relax in the excited state as the lattice parameter of the host double perovskite increases, thereby lowering its energy and red shifting the ensuing emission. 1 The emission position in Cs 2 NaInCl 6 :Sb 3+ is also independent of the excitation wavelength, indicating that the emission arises from a consistent radiative decay process (Fig. S7,ESI †). The emission characteristics of Cs 2 NaInCl 6 :Sb 3+ are similar to the industry standard blue-emitting phosphor, BaMgAl 10 O 17 :Eu 2+ (BAM:Eu 2+ ), which has a 0.90 eV Stokes shift (340 nm excitation, 452 nm emission, FWHM = 55 nm). 20 However, the BAM:Eu 2+ phosphor deteriorates over extended usage due to oxidation of the luminescent center from Eu 2+ to Eu 3+ . 21 Not only is this degradation mechanism not operative in the Cs 2 NaInCl 6 :Sb 3+ system, the absence of rare-earth ions is an attractive feature. The maximum emission intensity was observed for the sample (7), corresponding to an approximately 3% substitution of Sb 3+ for In 3+ . The slight decrease in luminescence intensity for sample (8) (estimated [Sb 3+ ] E 4%), suggests concentration quenching begins to play a role for the higher Sb 3+ contents. The lower emission intensity of sample ( 9) can also be attributed in part to the presence of Cs 3 Sb 2 Cl 9 . This may give rise to a non-radiative deactivation pathway, 22 not to mention the expected decrease in emission because a non-negligible part of the sample has formed a non-emissive phase. The CIE coordinate diagram (Fig. 5a) for Cs 2 NaInCl 6 :Sb 3+ corroborates the expected blue emission, with CIE coordinates of (0.148, 0.067). The PLQY of this compound was determined to be 79(5)% for sample (6), an impressive efficiency for a phosphor prepared from solution (Fig. S8 and Table S3, ESI †).
The thermal stability of the photoluminescence was explored by collecting temperature-dependent PL measurements in 20 K intervals between 300 K and 600 K. These measurements indicate the temperature at which the photoluminescence intensity drops to 50% of its room-temperature value (T 50 ) is 405 K, as shown in Fig. 5b. Minimal shifts were also seen in the emission maximum with temperature, indicating excellent color stability (Fig. S9, ESI †). This is consistent with locally excited systems since the shape of the [SbCl 6 ] 3À octahedra will not significantly change with temperature.
Photoluminescent lifetime measurements were conducted to further understand the decay mechanism. The data plotted in Fig. 5c for sample (6) were fit to a biexponential, following eqn (3), resulting in a lifetime of t 1 = 0.090 ms accounting for 46% of the decay, and a longer lifetime t 2 = 1.016 ms accounting for the rest.
(3)
These lifetime values are reasonably close to those reported for Cs 2 SnCl 6 :Sb 3+ system, which had t 1 = 0.154 ms and t 2 = 0.821 ms, where the two lifetimes were attributed to 3 P 1 -3 P 0 and 3 P 1 -1 S 0 transitions, respectively. 9 To explore the potential of this phosphor for use in lighting applications, a prototype pc-LED was fabricated by combining a UV-LED chip (l ex = 370 nm) with a mixture of the Cs 2 NaInCl 6 :Sb 3+ , a lab prepared red-emitting Sr 2 Si 5 N 8 :Eu 2+ and a commercially available green-emitting b-SiAlON:Eu 2+ . The device was driven by a 20 mA current to yield the corresponding emission spectrum plotted in Fig. 5d. The fullspectrum warm white light produced using Cs 2 NaInCl 6 :Sb 3+ possess excellent color quality with a color rendering index (R a ) was also fabricated. The luminescence spectrum and CIE coordinates of the pc-LED can be seen in Fig. S10 (ESI †). 23 The CIE coordinates of the BAM:Eu 2+ containing pc-LED closely resembles that of the pc-LED composed of Cs 2 NaInCl 6 :Sb 3+ , indicating a similar broad band spectrum can be produced from both blueemitters. The resulting R a and CCT of the pc-LED using the BaMgAl 10 O 17 :Eu 2+ synthesized here is 95.3 and 4479.2 K. The R a of the BAM:Eu 2+ containing pc-LED is slightly better; however, the CCT of this device is significantly higher than the pc-LED using Cs 2 NaInCl 6 :Sb 3+ . The nearly 500 K difference in CCT highlights the ability of Cs 2 NaInCl 6 :Sb 3+ to be used in conjunction with UV-LEDs to produce a broad spectrum, warm white light.
Based on the results reported here, it could be imagined that a variety of moisture stable double perovskites could successfully host the Sb 3+ activator in a 6-coordinate environment. To this end, Sb 3+ was doped into Cs 2 NaBiCl 6 and Cs 2 AgInCl 6 hosts using modifications of previously reported synthesis methods. 5,24,25 The samples were each irradiated with light from a broadband Ultra-Violet Products UVSL-25 Mineralight Lamp (365 longwave, 254 shortwave) excitation source to test for luminescence. However, no luminescence is observed at room temperature in Cs 2 NaBiCl 6 samples doped with Sb 3+ . The lack of observed luminescence in the Bi 3+ containing system may arise from the close alignment of the energy levels of Sb 3+ and Bi 3+ . These orbitals tend to hybridize efficiently in double perovskites, as demonstrated by the pronounced band gap bowing effect seen in Cs 2 AgBi 1Àx Sb x Br 6 and Cs 2 AgBi 1Àx Sb x Cl 6 solid solutions. 26,27 This energetic alignment allows for facile energy transfer between the Sb 3+ activator and the host, which can enhance concentration quenching. In Cs 2 AgInCl 6 :Sb 3+ , trace (o1%) amounts of Sb 3+ lead to an intense yellow-white luminescence, reminiscent of Cs 2 Ag 0.60 Na 0.40 InCl 6 :Bi 3+ or Cs 2 AgIn 1Àx Bi x Cl 6 phosphors. 24,28 It appears as though selftrapped excitonic emission leads to the broad photoluminescence observed in both systems. These alternative hosts, while non-toxic and stable, do not have the correct electronic structure to localize the excited state on Sb 3+ -dopants, illustrating the importance of the host structure in the design of new phosphors.
During the review process, another paper describing photoluminescence in Cs 2 NaInCl 6 :Sb 3+ appeared in the literature. 29 The photophysical properties reported in both studies are generally in good agreement with each other.
Bright blue luminescence (l max = 445 nm, FWHM E 0.51 eV) with an internal PLQY of 79(5)% is observed when Sb 3+ ions are doped into the ordered double perovskite Cs 2 NaInCl 6 host. In samples precipitated from HCl(aq) solution, approximately 5% of the In 3+ ions can be replaced with Sb 3+ , but the PL starts to decrease when the Sb-content gets larger than B3%. When compared to Cs 2 SnCl 6 :Sb 3+ phosphors where the Sb 3+ ions are likely 5-coordinate, the Stokes shift in Cs 2 NaInCl 6 :Sb 3+ is smaller (0.94 eV vs. 1.38 eV) leading to blue emission rather than orange-red, and the quantum efficiency is more than doubled (79% vs. 37%). The emission characteristics of Cs 2 NaInCl 6 :Sb 3+ phosphors are comparable to the commercial blue phosphor, BaMgAl 10 O 17 :Eu 2+ (BAM:Eu 2+ ), which makes it an attractive, rare-earth free alternative to commercial blue phosphors.
This journal is © The Royal Society of Chemistry 2020