Quenching of the red Mn4+ luminescence in Mn4+-doped fluoride LED phosphors

Red-emitting Mn4+-doped fluorides are a promising class of materials to improve the color rendering and luminous efficacy of white light-emitting diodes (w-LEDs). For w-LEDs, the luminescence quenching temperature is very important, but surprisingly no systematic research has been conducted to understand the mechanism for thermal quenching in Mn4+-doped fluorides. Furthermore, concentration quenching of the Mn4+ luminescence can be an issue but detailed investigations are lacking. In this work, we study thermal quenching and concentration quenching in Mn4+-doped fluorides by measuring luminescence spectra and decay curves of K2TiF6:Mn4+ between 4 and 600 K and for Mn4+ concentrations from 0.01% to 15.7%. Temperature-dependent measurements on K2TiF6:Mn4+ and other Mn4+-doped phosphors show that quenching occurs through thermally activated crossover between the 4T2 excited state and 4A2 ground state. The quenching temperature can be optimized by designing host lattices in which Mn4+ has a high 4T2 state energy. Concentration-dependent studies reveal that concentration quenching effects are limited in K2TiF6:Mn4+ up to 5% Mn4+. This is important, as high Mn4+ concentrations are required for sufficient absorption of blue LED light in the parity-forbidden Mn4+ d–d transitions. At even higher Mn4+ concentrations (>10%), the quantum efficiency decreases, mostly due to direct energy transfer to quenching sites (defects and impurity ions). Optimization of the synthesis to reduce quenchers is crucial for developing more efficient highly absorbing Mn4+ phosphors. The present systematic study provides detailed insights into temperature and concentration quenching of Mn4+ emission and can be used to realize superior narrow-band red Mn4+ phosphors for w-LEDs.


Powder X-ray diffraction
Di raction angle 2θ (°) 25 45 55 65 75 35 15 Intensity (arb. u.) Figure S1 Powder X-ray diffraction (XRD) patterns of K 2 MnF 6 and the K 2 TiF 6 :Mn 4+ (x%) phosphors. The diffraction patterns are in agreement with the literature references for hexagonal phase K 2 MnF 6 (PDF 04-015-4092) and K 2 TiF 6 (PDF 00-008-0488). and carbon (C) peaks originate from the aluminium holder and carbon tape beneath the sample. The platinum (Pt) peak is due to the platinum layer sputtered onto the phosphor particles for the SEM-EDX measurements.

Influence of excitation wavelength on the observed thermal quenching behavior
The excitation wavelength used in photoluminescence (PL) measurements can have a large influence on the temperature dependence observed for the PL intensity I PL . Because the Mn 4+ excitation bands broaden and redshift with temperature ( Figure S3a), exciting at the 4 A 2 → 4 T 2 band maximum (450 nm) or 4 A 2 → 4 T 2 band onset (405 nm) results in very different temperature dependences ( Figure S3b). Consequently, different quenching temperatures T ½ and activation energies ∆E are obtained. As the true T ½ and ∆E are obtained when effects due to band broadening and shifting are minimized, the preferred excitation wavelength is at or close to the band maximum when measuring the temperature dependence for I PL . However, even then a change in the emission intensity by variations in the absorption strength at the excitation wavelength can introduce an error in T ½ . In general, temperature-dependent lifetime measurements provide a more reliable value for T ½

Quenching by multi-phonon relaxation
In the configurational coordinate diagram, the parabolas of the Mn 4+ 2 E and 4 A 2 states are at the same equilibrium position and luminescence quenching due to direct crossover from the 2 E excited state to the 4 A 2 ground state is not possible (see configuration coordinate diagram in Figure 4a). The 4 A 2 ground state may however be reached by multiphonon relaxation. In some works on Mn 4+ -doped oxides, thermal quenching of the Mn 4+ luminescence was attributed to multi-phonon relaxation [2][3][4][5] . The temperature dependence of the luminescence intensity and emission lifetime was explained with a multi-phonon relaxation process involving more than 40 phonons. Also in Mn 4+ -doped fluorides, a high number of phonons is necessary to reach the 4 A 2 ground state via multi-phonon relaxation. In a fluoride, the maximum phonon energy ν max is ~500 cm −1 , so around 32 high energy vibrations are needed to bridge the ~16000 cm −1 energy gap between the 2 E and 4 A 2 states 6 . For such high numbers of phonons (p > 30), it is unrealistic that non-radiative multiphonon relaxation is responsible for thermal quenching, as typically multi-phonon relaxation can only compete with radiative decay if the energy difference between the ground and excited state is equal to or less than 5 times the ν max of the surrounding lattice. Moreover, if quenching would occur through multi-phonon relaxation, it is expected that the T ½ is relatively similar for the different Mn 4+ -doped fluoride phosphors since all hexafluorometallates will have around the same maximum phonon energy ν max (Ref. 7). There is however a large spread in the T ½ of Mn 4+ -doped fluorides, varying from e.g., T ½ = 403 K in Cs 2 HfF 6 :Mn 4+ to T ½ = 518 K in K 2 SiF 6 :Mn 4+ (see also Table 1) 8 . Finally, the non-radiative decay rate due to multi-phonon relaxation increases with temperature as the non-radiative decay rate at low temperatures multiplied by (n + 1) p (Ref. 9). This implies that also at low temperatures multi-phonon relaxation should be effective as phonon emission can always occur. The observation of quantum efficiencies close to 100% at ambient temperature is not consistent with multi-phonon relaxation. We conclude that thermal quenching of the Mn 4+ luminescence cannot be due to multi-phonon relaxation from the 2 E state. Figure S5 shows the luminescence quenching temperature of Mn 4+ -doped fluorides and Mn 4+ -doped oxides as function of the ligand-to-Mn 4+ charge-transfer (CT) transition energy. The quenching temperatures and CT energies were collected from luminescence measurements presented in this work and the literature. The results in Figure S5 show that the Mn 4+ luminescence quenching temperature and ligand-to-Mn 4+ CT transition energy are not correlated. The data displayed in Figure S5 is also listed in Table S1. We want to note that for most Mn 4+ -doped oxides the CT excitation band overlaps with the 4 A 2 → 4 T 1 excitation band, which introduces an uncertainty in the CT energies reported in Table S1.

Mn 4+ luminescence spectra measured at elevated temperatures
Additional proof for thermal quenching by crossover via the 4 T 2 state is obtained from Mn 4+ luminescence spectra measured at elevated temperatures. Figure S7 shows emission spectra of K 2 SiF 6 :Mn 4+ (commercial phosphor) measured at T = 573 and 673 K. Besides the characteristic 2 E → 4 A 2 emission lines, the luminescence spectra in Figure S7 exhibit some additional weak emission bands/lines at wavelengths shorter than 600 nm. These emissions are assigned to the Mn 4+ 4 T 2 → 4 A 2 and 2 T 1 → 4 A 2 transitions. An excitation spectrum recorded for λ em = 530 nm at T = 473 K (see Figure  S8) confirms that the emission band centered at 530 nm is related to Mn 4+ . The observation of Mn 4+ 4 T 2 → 4 A 2 emission in K 2 SiF 6 :Mn 4+ at 573 K shows that the 4 T 2 excited state is indeed thermally populated at elevated temperatures, and consequently can play a role in the thermal quenching process. Upon further heating to 673 K, the intensities of the 4 T 2 → 4 A 2 and 2 E → 4 A 2 emissions decrease (green spectrum in Figure S7), and after cooling to 573 K, most of the 4 T 2 → 4 A 2 and 2 E → 4 A 2 emission intensity is regained (blue spectrum in Figure S7). These measurements indicate that the emission intensity decrease between 573 and 673 K is due to thermal quenching of both the 4 T 2 → 4 A 2 and 2 E → 4 A 2 emission, and not due to chemical degradation of the phosphor (the small difference in intensity at 573 K before and after heating to 673 K is however attributed to phosphor degradation). The fact that both the 4 T 2 → 4 A 2 and 2 E → 4 A 2 emission are quenched upon raising the temperature from 573 to 673 K shows that the loss in 2 E → 4 A 2 emission intensity is not accompanied an increase in the 4 T 2 → 4 A 2 emission intensity, as is sometimes observed for Cr 3+ (isoelectronic with Mn 4+ ) [27][28][29] . Instead, in K 2 SiF 6 :Mn 4+ the 4 T 2 → 4 A 2 emission is quenched by non-radiative relaxation via the crossing of the 4 T 2 state and 4 A 2 ground state.  Figure S8 PL excitation (blue, λ em = 530 nm) and emission (red, λ exc = 360 nm) spectra of K 2 SiF 6 :Mn 4+ at T = 473 K. The red and blue arrows indicate the excitation and emission wavelengths used for recording the spectra, respectively. The excitation spectrum (blue) of the weak emission band centered at 530 nm consists of two excitation bands that are assigned to the 4 A 2 → 4 T 1 and 4 A 2 → 4 T 2 transitions of Mn 4+ .

Bandwidth of 4 A 2 → 4 T 2 excitation band in Mn 4+ -doped fluorides
To investigate the variation in the offset ∆R for Mn 4+ -doped fluorides, we compare the bandwidth of the 4 A 2 → 4 T 2 excitation band in K 2 TiF 6 :Mn 4+ , K 2 SiF 6 :Mn 4+ and Cs 2 HfF 6 :Mn 4+ . The width of the 4 A 2 → 4 T 2 excitation band is controlled by the displacement of the 4 T 2 state and therefore gives a good indication of ∆R. In Figure S9 it is observed that the full width at half maximum intensity (fwhm) of the 4 A 2 → 4 T 2 band increases from 3104 cm −1 in K 2 TiF 6 :Mn 4+ to 3251 cm −1 in Cs 2 HfF 6 :Mn 4+ . This shows that the width of the 4 A 2 → 4 T 2 excitation band, and thereby ∆R, varies per fluoride host lattice. The energy difference between the fwhm values is however small compared to the difference in 4 A 2 → 4 T 2 energy, which indicates that the 4 T 2 level energy is more important for the quenching temperature.  Table S2.  Figure S11 PL excitation (blue, λ em = 636 nm) and emission (red, λ exc = 450 nm) spectra of K 2 TiF 6 :Mn 4+ (15.7%) at T = 4 K. For the excitation spectrum only the spectral region of the 4 A 2 → 2 T 1 and 4 A 2 → 2 E transitions is shown.

Energy transfer between Mn 4+ ions
The results presented in Figure 5 show that concentration quenching via energy migration is weak in K 2 TiF 6 :Mn 4+ . We attribute this to the fact that energy transfer between Mn 4+ neighbors probably has to occur via exchange interaction 14,37 .
The very small oscillator strength of the zero-phonon line prevents efficient resonant energy transfer via dipole-dipole interaction. Energy transfer via exchange interaction (wavefunction overlap) is possible but only active for very small (<5 Å) distances between the Mn 4+ ions and is therefore limited to transfer between nearest neighbors (nearest neighbor distance is 4.7 Å in K 2 TiF 6 ). With a Mn 4+ doping concentration of e.g., 5%, most Mn 4+ ions will not have Mn 4+ neighbors within a distance of 5 Å 38 . Energy transfer between Mn 4+ ions will therefore be very inefficient and as a consequence energy migration to quenching sites is limited in K 2 TiF 6 :Mn 4+ . This situation will also apply to other Mn 4+ -doped fluoride phosphors. We therefore expect concentration quenching by energy migration in general not to be an issue for the use of Mn 4+ -doped fluorides in w-LEDs. The quenching that is observed for higher Mn 4+ concentrations is explained by an increase in the amount of quenching centers with increasing manganese concentration.