Local structure and structural rigidity of the green phosphor β-SiAlON:Eu²⁺

The continued replacement of traditional illumination sources with solid-state lighting (SSL) is driven by their exceptional efficiency compared to incandescent and fluorescent lamps and the absence of toxic-heavy metals, such as mercury.^1–3 One of many strategies for assembling a SSL device is to integrate ultraviolet (UV) or near-UV light emitting diodes (LED) with phosphors emitting in the blue, green, and red regions of the visible spectrum.^2,3 Eu²⁺ inserted in β-Si_3−xAl_xO_xN_4−x is a material that shows exceptional promise as a green-emitting phosphor for such applications.^4,5 The compound crystallizes in the β-Si₃N₄ crystal structure (space group 163, $P63$⁠), shown in Fig. 1(a), with three-dimensional corner-sharing of (Si,Al)(O,N)₄ tetrahedra. This high symmetry structure contains one crystallographically independent cation position (Wyckoff 6c) and two independent anion positions (Wyckoff 6c and 2b).

Incorporating Eu²⁺ in the β-SiAlON host produces an ideal excitation and emission spectrum for use in SSLs, as highlighted by the contour plot in Fig. 1(b). The excitation band is strong and broad, spanning from the UV (<300 nm) to the blue-green region of the visible spectrum (∼475 nm), allowing excitation by a variety of commercial LEDs. Intense emission, with quantum yields that are 80% or greater, occurs in the green region of the visible spectrum (λ_em = 540 nm) and corresponds to the allowed Eu²⁺ 5d to 4f (⁠$8Hj$ → $8S7/2$⁠) transition.

The substitution of the luminescent center in most rare-earth inserted phosphors occurs on an obvious crystallographic site based on size and valence. In β-SiAlON, a substitution site is not obvious due to the size mismatch between Eu²⁺ and the cations in the structure. As a result, the inclusion of the rare-earth must occur at an interstitial, and the Si:Al or N:O ratios must change to maintain electroneutrality. Rare-earth solubility is limited (<2 mol. %),⁶ making direct identification of the interstitial site extremely challenging by traditional characterization techniques like X-ray diffraction. Scanning transmission electron microscopy (STEM) has tentatively suggested that Eu²⁺ occupies the large hexagonal channels along the [001] direction coordinated by SiN₄ tetrahedra (Wyckoff position 2a).⁷ The structural studies carried out here are intended to enhance understanding of such Eu²⁺ insertion in this unique phosphor.

Recent research has suggested that highly efficient phosphors, such as β-SiAlON:Eu²⁺, contain densely packed polyhedral units and high symmetry, which result in structures^8,9 that limit photoluminescence quenching and other non-radiative relaxation pathways. Because structural rigidity in a solid is rather subjective, a proxy such the Debye temperature (Θ_D) proves very useful.¹⁰ In the case of stoichiometric β-Si₃N₄, Θ_D is experimentally measured to be very high, close to 900 K, which indicates great promise for luminescence applications.¹¹ However, the inclusion of the luminescent center (Eu²⁺) to produce the phosphor requires co-substitution of Al³⁺ and O²⁻, potentially impacting Θ_D. A second goal reported here is to demonstrate that the changes in Θ_D can be monitored by first-principles density functional theory (DFT).

β-SiAlON:Eu²⁺ can be readily prepared using many synthetic methods.^12–14 However, the concentration of Al³⁺ and O²⁻ can be strictly controlled using a high-pressure and high-temperature route. For the studies carried out here, a mixture of α-Si₃N₄, Al₂O₃, and Eu₂O₃ powders was ground with an alumina mortar and pestle following the stoichiometric ratio β-Si_3−xAl_xO_xN_4−x:Eu²⁺. Powders were reacted in a BN crucible at 1950 °C for 12 h under 0.92 MPa of N₂ (purity >99.9995%). The products were then annealed for 8 h in an Ar atmosphere, ground to a fine powder, and washed with a mixture of HNO₃ and HF.

This route produces a pure-phase product for samples containing Eu²⁺ as well as Eu²⁺-free compounds, as identified by a Rietveld refinement of high-resolution synchrotron X-ray powder diffraction data [Fig. 2(a)] measured at beamline 11-BM, Advanced Photon Source.^15–17 Due to the similar X-ray scattering power of the substituted cations and anions (e.g., Al³⁺ for Si⁴⁺ and O²⁻ for N³⁻), the total composition was not refined based on the X-ray diffraction data. Fortunately, the neutron scattering path lengths of N (9.36 fm) and O (5.80 fm) are sufficiently different to unequivocally distinguish composition and site occupancy of the anions.¹⁸ Because Eu has a large neutron absorption cross-section (4530 b for 0.0253 eV), a sample was prepared following the same synthetic protocol, maintaining the total composition while omitting Eu²⁺ to prevent absorption. Neutron powder diffraction was performed at NPDF¹⁹ (Los Alamos Neutron Science Center, Los Alamos National Laboratory); time-of-flight data were collected at 295 K using four sets of detector banks, located at $±40°,±90°,±119°$⁠, and $±140°$⁠.

Rietveld refinement of neutron scattering data confirms the incorporation of a small oxygen concentration (∼5%) on both crystallographic anion sites [Fig. 2(b)]. At such small O²⁻ content, there is not an obvious site preference. Although the anion concentration was refined, the scattering contrast between Al³⁺ (3.45 fm) and Si⁴⁺ (4.15 fm) is not sufficient to allow refinement of the cations, so the ratio of cations was fixed as the nominally loaded composition. The final refined composition was Si_2.875Al_0.125O_0.16(1)N_3.84(1). Comparing the refined anionic O²⁻:N³⁻ ratio with the loaded cation Si⁴⁺:Al³⁺ ratio indicates that the as prepared compound is nearly charge neutral within standard deviation.

To determine the substitution site of Eu²⁺ in the bulk, a combined approach of first-principles computational modeling and X-ray absorption spectroscopy was employed. DFT calculations based on the Vienna ab initio Simulation Package (VASP)^20,21 were used to generate the potential energy surface for β-Si₃N₄ (space group P6₃), as illustrated in Fig. 3. Visualization of the (001) plane reveals two favorable interstitial sites for Eu substitution: a 12-coordinate site (Wyckoff 2a) and a second 12-coordinate site located in the center of the unit cell (Wyckoff 6c). The polyhedral volume of the 6c site (∼20 ų) is half the size of the 2a site (∼40 ų), making it more likely for Eu²⁺ to incorporate in the large void.

The location of Eu²⁺ in the crystal structure was experimentally confirmed by measuring the extended X-ray absorption fine structure (EXAFS) at the Eu L₃ edge. Fluorescence spectra at 295 K and 20 K were collected at the Advanced Photon Source beamline 20-BM, and all data were processed using the Demeter software suite,²² which employs IFEFFIT and FEFF6. Additional experimental details are presented in the supplementary material.²³ The absorption energy is consistent with Eu²⁺, and the X-ray absorption near-edge structure (XANES) shows that a majority (>90%) of Eu is in the divalent oxidation state.

The starting model for the EXAFS refinement was a DFT-optimized 2 × 2 × 4 supercell of β-Si₃N₄ with an interstitial Sr atom. Sr (⁠$rSr2+=1.31 Å$⁠) was used instead of Eu (⁠$rEu2+=1.30Å$⁠) due to the difficulty of treating f electrons within DFT. The supercell was optimized until the residual forces were less than 10^–2eV/Å. Using this model, $Δr$ of the outer-shell N was initially >0.1 Å, so FEFF calculations were performed on an expanded unit cell (with the unit cell parameters increased by 2%) to ensure proper representation of atomic potentials by muffin-tin radii implemented in FEFF.

Fitting the Eu L₃ EXAFS demonstrates that the spectrum is consistent with Eu²⁺ in the distorted 2a interstitial site of β-Si₃N₄ (Fig. 4). The coordination environment and structural parameters around Eu²⁺ are presented in Table I. Notably, the Debye-Waller factors (⁠$σN2, σSi2$⁠) are quite large and do not change markedly between 295 K and 20 K. These large mean-squared displacements are dominated by static disorder in the structure owing to the distortion of the β-Si₃N₄ structure required to incorporate Eu²⁺. This is apparent in the intensity of the oscillations of the EXAFS, as the intensity is not enhanced at low temperature. Additionally, the first coordination shell (N_a + N_b) determined by EXAFS is larger than the DFT-optimized structure. This finding is supported by analysis of the Eu²⁺ bond valence sum (BVS),²⁴ which is 2.0 at 20 K (BVS_DFT = 2.3), indicating optimal bonding.

DFT calculations of elastic constants²⁵ allowed Θ_D to be approximated.¹⁰ The substitution of Al³⁺ and O²⁻ in β-SiAlON was modeled following a previously determined site preference^4,26–28 for β-Si_3−xAl_xO_xN_4−x (x = 0, 0.5, 1, 1.5, 2, and 2.5). Incorporating Al³⁺ and O²⁻ decreases the structural rigidity, decreasing Θ_D from 956 K for x = 0 to 710 K for x = 2.5 (Fig. 5). This is a result of phonon mode softening due to chemical disorder and the lower valences of Al³⁺ and O²⁻ in the crystal structure. At the low level of substitution experimentally studied here (x = 0.125), the reduction in Θ_D is not significant, with interpolation suggesting a value near Θ_D = 920 K. Rietveld refinement of neutron diffraction data provides reliable atomic anisotropic displacement parameters (ADPs) necessary for an experimental estimate of Θ_D at the high-temperature limit to compare with calculation.^8,23,29 For this sample, it is estimated that Θ_D ≈ 725 K. Although this is lower than calculated by DFT, significant static disorder and site mixing between Si⁴⁺ and Al³⁺ are missed by the Rietveld analysis and likely account for such a sharp drop in the experimental Θ_D. Nevertheless, staying at low x values is clearly advantageous to maximize the quantum yield, and it has also been shown to narrow emission and improve color in β-SiAlON:Eu²⁺.³⁰ 

In summary, the position of Eu²⁺ in the crystal structure was confirmed by fitting the Eu L₃ EXAFS using a DFT-optimized starting structure; the Eu²⁺ luminescent center occupies a distorted 12-coordinate interstitial site in the parent β-Si₃N₄ crystal structure (Wyckoff 2a). There are minor differences in bond lengths determined by EXAFS and DFT, but the presence of local distortions around Eu²⁺ in the interstitial is captured well by both approaches. DFT calculations of Θ_D have been employed as a proxy for structural rigidity and suggest that at the low level of (Al/O) substitution required to incorporate small amounts of Eu²⁺, the structure remains rigid and should retain high performance.

M.W.G. thanks the Natural Sciences and Engineering Council of Canada for support through a NSERC Postgraduate Scholarship, and the U.S. Department of State for an International Fulbright Science & Technology Award. The research made use of the shared facilities at the Materials Research Laboratory (MRL) (DMR-1121053). Calculations were conducted at the UCSB Center for Scientific Computing, supported by the California Nanosystems Institute (NSF CNS-0960316), Hewlett-Packard, and the MRL. This work benefited from the use of the Advanced Photon Source at Argonne National Laboratory was supported by the U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences (DE-AC02-06CH11357), and neutron scattering at the Lujan Center, funded by DOE Office of Basic Energy Sciences at the time of measurement; LANL is operated by Los Alamos National Security LLC (DE-AC52-06NA25396).