Mn⁵⁺-activated Ca₆Ba(PO₄)₄O near-infrared phosphor and its application in luminescence thermometry

Abstract

The near-infrared luminescence of Ca₆Ba(PO₄)₄O:Mn⁵⁺ is demonstrated and explained. When excited into the broad and strong absorption band that spans the 500–1000 nm spectral range, this phosphor provides an ultranarrow (FWHM = 5 nm) emission centered at 1140 nm that originates from a spin-forbidden ¹E → ³A₂ transition with a 37.5% internal quantum efficiency and an excited-state lifetime of about 350 μs. We derived the crystal field and Racah parameters and calculated the appropriate Tanabe–Sugano diagram for this phosphor. We found that ¹E emission quenches due to the thermally-assisted cross-over with the ³T₂ state and that the relatively high Debye temperature of 783 K of Ca₆Ba(PO₄)₄O facilitates efficient emission. Since Ca₆Ba(PO₄)₄O also provides efficient yellow emission of the Eu²⁺ dopant, we calculated and explained its electronic band structure, the partial and total density of states, effective Mulliken charges of all ions, elastic constants, Debye temperature, and vibrational spectra. Finally, we demonstrated the application of phosphor in a luminescence intensity ratio thermometry and obtained a relative sensitivity of 1.92%K⁻¹ and a temperature resolution of 0.2 K in the range of physiological temperatures.

Introduction

Mn⁵⁺ optical centers have the [Ar]3d² electron configuration and always encounter a strong crystal field when tetrahedrally coordinated in crystals. Their lower electronic states have the ³A₂ < ¹E < ¹A₁ < ³T₂ < ³T₁ progression in energy. The ground state (³A₂) is orbitally non-degenerate and the first excited state ¹E has almost no nuclear displacement with respect to the ³A₂ state and can be split by the low-symmetry ligand field¹. The ¹E energy of approximately 8000–9000 cm⁻¹ is strongly affected by a nephelauxetic effect. At low temperatures, emission occurs solely from the spin-forbidden ¹E → ³A₂ electronic transition of a genuine electric dipole origin. At these temperatures, the emission from the spin-allowed ³T₂ → ³A₂ transition is almost negligible since the low energy orbital of the ³T₂ state is localized more than 1000 cm⁻¹ above the ¹E state, which results in its low population. Therefore, Mn⁵⁺ emissions appearing in the near infrared (NIR) spectral range at wavelengths longer than 1100 nm and have a narrow spectral band (FWHM < 10 nm) that can be split into two bands with an energy difference of up to 300 cm⁻¹. These emission bands are usually accompanied by vibronic sidebands and have decay times of a few tens to a few hundred microseconds.

The ultranarrow-band NIR luminescence of Mn⁵⁺ is especially favorable for NIR lasers^2,3,4 and the development of narrow-band NIR light sources for the selective identification of chemical analytes⁵. Recent research suggests that Mn⁵⁺ activated nanoparticles are excellent probes for deep-tissue luminescence imaging and luminescence thermometry in the second biological transparency window (1000–1350 nm) and that they show high photo- and chemical stabilities^6,7. One of their features, especially favorable from the biomedical application perspective, is that they exhibit broad and strong absorption bands from the spin-allowed electronic transitions that span the 500–1000 nm range^{8,9,10,11,12,13}, which facilitates their excitation by the wavelength of the first optical biological window. They have a higher quantum efficiency (QE) than Ag₂S or Ag₂Se quantum dots, and do not contain toxic heavy metals like InAs or PbS quantum dots, nor suffer from photobleaching and photoblinking. In comparison to the lanthanide activated NIR bioimaging nanoprobes, Mn⁵⁺ nanoparticles are brighter due to higher values of the d-d absorption cross-section compared to the spin-forbidden f-f absorptions in the NIR-excited lanthanides. All these advantages of Mn⁵⁺ indicate that it is an ion with a very important and, so far, unexplored application potential worth large-scale intensive research.

Emissions from Mn⁵⁺ centers have been demonstrated in a considerably smaller number of host materials compared to Mn²⁺ and Mn⁴⁺ optical centers¹⁴. To facilitate Mn⁵⁺ emission, the material needs to provide both the tetrahedral environment for Mn⁵⁺ and a sufficiently large energy bandgap compared to the Mn⁵⁺ transitions’ energies. More importantly, the host material must provide the stabilization of the Mn 5+ valence state, which imposes more constraints on the materials’ structure and composition than the 2+ and 4+ valance states. For these reasons, the majority of Mn⁵⁺ emitting host materials contain electropositive ions such as alkaline earth metals and PO₄^3- and VO₄^3- groups (or SiO₄⁴⁻ group with a charge compensation). The typical examples of such materials are Mn⁵⁺ activated Li₃PO₄¹⁵, Sr₅(VO₄)₃F^4,16, Ba₅(PO₄)₃Cl¹⁷, Sr₅(PO₄)₃Cl^1,17,18, Ca₂PO₄Cl^1,17, Ca₂VO₄Cl^1,17, Sr₂VO₄Cl^1,17, Y₂SiO₅¹⁹, and M₂SiO₄ (M = Ba, Sr, Ca)²⁰.

Recently, Kim et al.²¹ have introduced the Ca₆Ba(PO₄)₄O:Mn⁵⁺ as a new blue pigment that shows coloration due to intense Mn⁵⁺ absorption, but its luminescent properties have not been analyzed. Also recently, the efficient yellow emission from Eu²⁺ activated Ca₆Ba(PO₄)₄O has been demonstrated^22,23,24, implying that Ca₆Ba(PO₄)₄O is an interesting host material for luminescent ions. Considering Ca₆Ba(PO₄)₄O structure and composition, one can observe seven electropositive ions (six Ca and one Ba) surrounding the rigidly connected PO₄ tetrahedra. Thus, one may assume that the Mn⁵⁺ emission, in this new host-activator combination, would be efficient and of high energy due to the nephelauxetic effect. For this reason, we prepared the Ca₆Ba(PO₄)₄O:Mn⁵⁺ powder for this research and observed the intense NIR emission under the 650 nm excitation. The emission energy of 8772 cm⁻¹ is among the highest energies detected for Mn⁵⁺ activated phosphors. Considering the potential use of Ca₆Ba(PO₄)₄O for both yellow and NIR phosphors and the absence of data on its electronic and vibrational properties, we calculated and explained its electronic band structure, the partial and total density of states (DOS), effective Mulliken charges of all constituent ions, elastic constants, Debye temperature, and vibrational spectra of undoped and Mn⁵⁺ doped material. Then, we proceeded with the detailed characterization and analysis of Ca₆Ba(PO₄)₄O:Mn⁵⁺ luminescence that includes the measurements of material absorption, excitation and emission spectra, emission decays, concentration quenching, quantum yield, the temperature dependence of emission band shift, bandwidth and decay, and the calculation of the crystal field parameters. This versatile characterization and analysis of Ca₆Ba(PO₄)₄O:Mn⁵⁺ undoubtedly confirm the high potential of this material in NIR applications.

Experimental

The conventional solid-state reaction was employed for the preparation of Ca₆BaP_4-4xMn_4xO₁₇ (x = 0.005, 0.0075, 0.0125, 0.01, 0.015, 0.02) powder samples. Stoichiometric amounts of CaCO₃ (Alfa Aesar, 98%), BaCO₃ (Alfa Aesar, 99.8%), (NH₄)H₂PO₄ (Alfa Aesar, 98%), and MnO (Aldrich, 99.99%) were thoroughly mixed in an agate mortar for 1 h with an appropriate amount of ethanol. Mixtures of the raw materials were placed in alumina crucibles and heated in an air atmosphere at 600 °C for 6 h, ground in an agate mortar, and further calcinated at 1280 °C for 10 h.

The crystal structure of powders was examined by powder X-ray diffraction (PXRD) using the Rigaku SmartLab instrument (Cu-Kα_1,2 radiation; λ = 0.1540 nm) at room temperature. Data were recorded over the 6°−130° 2θ range, with a 0.01° step size and 1 min/° counting time. All PXRD data were analyzed by the Rietveld method implemented in TOPAS Academic software^25,26. Raman scattering measurements were performed using micro – Raman system TriVista 557 equipped with a triple monochromator and CCD detector (monochromator configuration 900/900/1800 points per millimeter) with 1.5 cm⁻¹ resolution. For excitation, Ar laser line at 514.5 nm has an incident power of less than 60 mW to minimize the heating effect. Laser beam was focused on the samples by means of microscopic lenses with 100× magnification. Spectra were recorded in the range of 100–1200 cm⁻¹. Measurements of diffuse reflectance were performed on the Thermo Evolution 600 spectrometer equipped with an integrated sphere, using BaSO₄ as a reference over the 220–1350 nm wavelength range. The photoluminescence emission and emission decays were measured using the FLS1000 Fluorescence spectrometer (Edinburgh Instruments; 0.1 nm spectral resolution) supplied with R5509-72 photomultiplier tube from Hamamatsu in nitrogen-flow cooled housing for near-infrared range detection. For measurements of the emission spectra and decays, the 668 nm laser diode excitation is used in the continuous and pulsed mode, respectively. The temperature of the sample was controlled using a THMS 600 heating-cooling stage from Linkam (0.1 K temperature stability and 0.1 K set point resolution). Emission decays at low temperatures were recorded at 1136 nm for all samples. The luminescence quantum efficiencies were measured using FLS980 Fluorescence Spectrometer from Edinburgh Instruments equipped with 450 W Xenon lamp, R5509-72 photomultiplier tube from Hamamatsu in nitrogen-flow cooled housing for near-infrared range detection, and calibrated integrating sphere for the direct absolute efficiency reading. The Al₂O₃ powder was used as a scattering reference. Thermometry was performed using a custom-made Peltier-based heating stage in the 20–100 °C range (0.02 °C precision). An Ocean Insight LSM-635A LED was used as the excitation source and is controlled by the Ocean Insight LDC-1 single channel driver and controller. The bifurcation optical Y cable was used for measuring PL emission spectra by Ocean Insight NIRQuest+ Spectrometers.

Results and discussion

The structure of Ca₆Ba(PO₄)₄O and Ca₆Ba(PO₄)₄O:Mn⁵⁺

The starting model used for Rietveld refinement and detailed structural analysis of the two key materials – the Ca₆Ba(PO₄)₄O host and the sample containing 0.5%Mn – was the previously published crystal structure of Ca₆Ba(PO₄)₄O determined from synchrotron powder diffraction data²². Refined parameters included the zero-point error, background polynomial terms, peak shape function terms, unit cell parameters, an isotropic atomic displacement parameter per atom type and atomic fractional coordinates, using bond valence sum restraints on the two P atoms. The key crystallographic parameters are summarized in Table 1. Ca₆Ba(PO₄)₄O and Ca₆Ba(PO₄)₄O:0.5%Mn adopt monoclinic space group C2/m, with one Ba, two Ca, two P and seven O atoms in the asymmetric unit. Ba atoms are 12-coordinate, the two crystallographically unique Ca atoms are 7- and 8-coordinate, while both unique P atoms adopt tetrahedral coordination environments. Dopant Mn⁵⁺ ions replace P⁵⁺ on these two sites, which lie on a mirror plane (Wyckoff site 4 m in space group C2/m). In Ca₆Ba(PO₄)₄O, the average cation-oxygen bond lengths in the two tetrahedra are 1.534(13) and 1.529(15) Å, while bond angles range from 105.1(8) to 112.9(7)° and 105.3(7) to 112.8(9)°, with bond valence sums of 5.0(1) for both P atoms. In Ca₆Ba(PO₄)₄O:0.5%Mn, the coordination environments remain similar, as expected given a low doping level. Average bond lengths are 1.534(22) and 1.504(27) Å, while bond angles range from 103.4(16) to 112.8(12)° and from 105.5(13) to 112.5(11)°, with bond valence sums of 5.0(2) and 5.4(2) for P1 and P2 sites, (see Tables 2 and 3), respectively. The final Rietveld fits obtained are shown in Fig. 1a, b, while the unit cell and the P atom environments are given in Fig. 1c, d, respectively.

Table 1 Summary of structural data for Ca₆Ba(PO₄)₄O and Ca₆Ba(PO₄)₄O:Mn⁵⁺ (space group C2/m)

Table 2 P–O bond lengths (Å)

Table 3 O–P–O angles and their difference, δ, to regular tetrahedron angle of 109.5°

Fig. 1: The structure of Ca₆Ba(PO₄)₄O:Mn⁵⁺.

a Rietveld fits for Ca₆Ba(PO₄)₄O, R_wp = 5.31%. b Rietveld fit for Ca₆Ba(PO₄)₄O:Mn⁵⁺, R_wp = 8.59%. In each case the blue curve represents the observed pattern, the red curve is the calculated pattern and the difference curves are shown in grey, while blue tick marks represent the positions of the Bragg peaks. c The crystal structure of Ca₆Ba(PO₄)₄O, with the unit cell viewed along the b-axis; d Geometries of two PO₄ tetrahedra with average P–O bond lengths in Ca₆Ba(PO₄)₄O:Mn⁵⁺

Electronic properties

A detailed calculation of the electronic properties of the Ca₆Ba(PO₄)₄O was performed to verify if its electronic structure is suitable to facilitate the Mn⁵⁺ emission. The electronic configurations for all chemical elements in Ca₆Ba(PO₄)₄O were as follows: O – 2s²2p⁴, P - 3s²3p³, Ca - 3s²3p⁶4s², Ba - 5s²5p⁶6s². The following parameters were used for the calculations: energy 10⁻⁵ eV per atom; maximal force 0.03 eVÅ⁻¹; maximal stress 0.05 GPa; maximal displacement 0.001 Å. K-points set was 2 × 2 × 2 for geometry optimization and 3 × 3 × 3 for the DOS calculations; the energy cut-off was 340 eV.

The structures obtained by geometry optimizations using the generalized gradient approximation (GGA) and the local density approximation (LDA) calculations are shown in Table 1 and are in excellent agreement, LDA especially, with the data obtained by Rietveld analysis of PXRD.

The band gaps in both GGA and LDA calculations, Fig. 2a, are direct, equal to 4.365 eV (GGA) and 4.475 eV (LDA) and have similar values to the reported bandgaps of 4.75 eV for the Ca₄(PO₄)₂O and 5.0 eV for the Ba₂Ca(PO₄)₂^27,28. Since the density-functional theory-based calculations (DFT) always tend to underestimate the true band gaps, the calculated values should be considered as lower band gap estimates.

Fig. 2: Electronic properties of Ca₆Ba(PO₄)₄O:Mn⁵⁺.

a The electronic band structure of Ca₆Ba(PO₄)₄O. b Partial and total density of states of Ca₆Ba(PO₄)₄O

The conduction band is due to the Ca 4s, 3d and Ba 6s states, Fig. 2b. The top of the valence band is remarkably flat, thus indicating extremely high effective masses of holes. The valence band consists of two sub-bands, which are separated by a narrow gap of about 0.5 eV. The upper one – from about −4.2 eV to 0 eV is made predominantly by the oxygen 2p states. The lower one (from −4.7 eV to about −7 eV) consists of the O 2p states and P 3p states, highly hybridized with each other. The P 3s states make another narrow band between −7.5 eV and −8.5 eV. The O 2s and P 3s states are spread between −22 eV and −17.5 eV, making several clearly seen maxima. The Ba 5s and 5p states are peaked at about −26 eV and −11 eV, respectively, whereas the Ca 3s and 3p states are localized deep in energy at about −38 eV and −20 eV, respectively.

The considered crystal has a high degree of covalency of chemical bonds, which can be assessed by calculating effective Mulliken charges. Since there are two crystallographically inequivalent P⁵⁺ ions, two Ca²⁺ ions and seven O²⁻ ions and since two types of calculations (GGA/LDA) were run, we give the ranges, in which these charges fall for all ions. They are −1.07-1.09 (in units of proton charge) for the oxygen ions, +2.25 + 2.34 for the phosphorus ions, +1.25 + 1.31 for the calcium ions, and +1.39 + 1.47 for the barium ions. The deviation from the formal charges is especially large for the P and O ions, whereas the Ca and Ba effective charges are closer to their formal ones. This is consistent with the P–O bonds being more covalent than the Ca–O and Ba–O ones.

With the calculated elastic constants, it is possible to estimate the Debye temperature using the following equations:

$$\theta _D = \frac{h}{k}\left( {\frac{{3n}}{{4\pi }}\frac{{N_A\rho }}{M}} \right)^{1/3}v_m$$

(1)

$$v_m = \left[ {\frac{1}{3}\left( {\frac{2}{{v_t^3}} + \frac{1}{{v_l^3}}} \right)} \right]^{^{ - 1/3}}$$

(2)

$$v_l = \sqrt {\frac{{3B + 4G}}{{3\rho }}} ,\;v_t = \sqrt {\frac{G}{\rho }}$$

(3)

where h = 6.626 × 10⁻³⁴ J·s is the Planck’s constant; k_B = 1.381 × 10⁻²³ JK⁻¹ is the Boltzmann constant; N_A = 6.022 × 10²³ mol⁻¹ is the Avogadro’s number; ρ is the crystal’s density; n is the number of atoms per one formula unit (twenty eight in the case of Ca₆Ba(PO₄)₄O), and M is the formula weight. The average, transverse and longitudinal sound velocities are denoted by v_m, v_t, v_l, correspondingly. The B (bulk modulus) and G (shear modulus) values are calculated as the average values of the corresponding Voigt and Reuss (denoted with the V and R subscripts, respectively) values from Table 4. With these equations and calculated elastic parameters, the Debye temperature for Ca₆Ba(PO₄)₄O was estimated to be 496 K (GGA) and 551 K (LDA).

Table 4 Calculated elastic constants (in GPa) for Ca₆Ba(PO₄)₄O

Vibrational spectra

For an undistorted tetrahedron (T_d symmetry), the four fundamental vibrational modes (all Raman active) are a₁ + e + 2t₂^1,29,30. The ν₁(a₁) is the totally symmetric stretching mode, ν₂(e) is bending deformation, and ν₃(t₂) and ν₄(t₂) vibrations are the two stretching t₂ modes. With symmetry lowered from T_d to C_s, several bands appear for each vibrational mode. Raman scattering spectra of the Ca₆Ba(PO₄)₄O and Ca₆Ba(PO₄)₄O:Mn⁵⁺ powders are shown in Fig. 3a, b, respectively. The following assignment is made for the Ca₆Ba(PO₄)₄O: ν_L (cm⁻¹) = 123, 153, 183, 215.4, 262.5, and 294.7 are the lattice modes; ν₁(a₁) (cm⁻¹) = 944 and 985.3; ν₂(e) = 434.5 (cm⁻¹); ν₃(t₂) (cm⁻¹) = 1040.1, and 1074.2; ν₄(t₂) (cm⁻¹) = 584.3 and 608.3. In Raman scattering spectra of the Ca₆Ba(PO₄)₄O:Mn⁵⁺ powder, Fig. 3b, additional vibrations from the MnO₄³⁻ ion are clearly visible: ν₁(a₁) (cm⁻¹) = 807; ν₂(e) (cm⁻¹) = 314.4; ν₃(t₂) (cm⁻¹) = 851.4; ν₄(t₂) (cm⁻¹) = 343.9, 356.2, and 368.5; ν₁₊ ν_L/ ν₃₊ ν_L (cm⁻¹) = 1128.5 and >1128.5. They agree with the data reported by Gonzalez-Vilchez and Griffith²⁹ for vibrational modes of the MnO₄³⁻ molecular ion.

Fig. 3: Vibrational properties of Ca₆Ba(PO₄)₄O:Mn⁵⁺.

a Raman scattering spectra of Ca₆Ba(PO₄)₄O powder. b Raman scattering spectra of Ca₆Ba(PO₄)₄O:Mn⁵⁺ powder

Photoluminescence properties

Figure 4a, b displays the absorption, color (inset in Fig. 4a), and emission spectra of Ca₆Ba(PO₄)₄O:0.5%Mn⁵⁺ powder. The observed absorption and emission, the blue coloration of the Mn⁵⁺ doped powder, and the characteristic vibrational modes of MnO₄³⁻ in the Raman scattering spectrum of Ca₆Ba(PO₄)₄O:Mn⁵⁺ (Fig. 3b) all unambiguously demonstrate the presence of Mn⁵⁺ in the sample.

Fig. 4: Optical properties of the Ca₆Ba(PO₄)₄O:Mn⁵⁺ powder.

a The Kubelka−Munk transformation of the Ca₆Ba(PO₄)₄O:Mn⁵⁺ diffuse reflectance The inset shows photographs of the Ca₆Ba(PO₄)₄O (white) and Ca₆Ba(PO₄)₄O:Mn⁵⁺ (blue) powders. b Emission spectra measured at −190 °C and 10 °C. c Tanabe–Sugano diagram for 3d² electron configuration in tetrahedral coordination. d Emission spectrum of the Ca₆Ba(PO₄)₄O:Mn⁵⁺ measured at room temperature (black line) and the fit to the Gaussian of the ³T₂ →³A₂ emission peak (red line) showing its maximum 1062 nm/9416 cm⁻¹. Spectra shown in logarithmic scale. e Temperature dependence of the excited state lifetime (symbols – experimental data, solid line – the fit to Eq. (4)). The inset shows emission decay measured at 208 K. f Temperature dependence of the ¹E emission peak spectral position (symbols – experimental data, solid line – the fit to Eq. (5)). g The estimate of configurational diagram based on the spectroscopic data with calculated Stokes shift (E_Stokes) and Huang–Rhys parameter (S). h Photoluminescence internal quantum efficiency (QE) of Ca₆Ba(PO₄)₄O:Mn⁵⁺ powders for different concentrations of Mn. The inset shows linear dependence of the log₁₀(QE/concentration) vs log₁₀(concentration) for data equal and above critical concentration (0.75%) with a slope of −1.97 indicating that a multipolar dipole-dipole mechanism is responsible for the concentration quenching of emission. i The linear dependence of Ca₆Ba(PO₄)₄O:Mn⁵⁺ emission intensity on excitation power

Figure 4a depicts the Kubelka–Munk transformation of the Ca₆Ba(PO₄)₄O:Mn⁵⁺ powder diffuse reflection measured between 220 and 1350 nm. The O²⁻ → Mn⁵⁺ charge-transfer band appears at around 301 nm (33,222 cm⁻¹) as expected for the tetraoxo-coordinated Mn^{5+ 31} and the peak at the lower wavelength (225 nm) is associated with the intrinsic host absorption. The strong absorption around 639 nm (15,649.5 cm⁻¹) is associated with the ³A₂ → ³T₁(³F) electronic transition, which is electric dipole-allowed in an undistorted tetrahedral symmetry and is composed of three overlapping components due to the removal of the orbital degeneracy of the ³T₁(³F) state with the site symmetry lowering from T_d to C_s. The weak shoulder at about 943 nm (10,604.5 cm⁻¹) corresponds to the symmetry forbidden ³A₂ → ³T₂(³F) transition (in T_d site symmetry) and becomes partially allowed with a symmetry lowering. The electric dipole-allowed ³A₂ → ³T₁(³P) transition that corresponds to a two-electron jump is located at approximately 369 nm (27,100 cm⁻¹) and is barely visible due to the much more intense charge transfer band. The spin-forbidden transitions to the singlet states ¹A₁(¹G) at 740 nm (13,513.5 cm⁻¹) and ¹E(¹D) at 1140 nm (8772 cm⁻¹) are weak, sharp, and only weakly depend on the host materials properties. The transitions to ¹T_1,2 singlet states are difficult to observe in the spectrum since they are very weak and superimposed on the main and stronger bands.

Emission spectra of Ca₆Ba(PO₄)₄O:Mn⁵⁺ powder measured at −190 °C and 10 °C are shown in Fig. 4b with blue and red lines, respectively, and are typical for emissions from transitions of 3d² electronic configuration in a tetrahedral environment as described by the Tanabe–Sugano diagram, Fig. 4c. The spectra show ultranarrow emission bands (FWHM = 3 nm (20 cm⁻¹) at −190 °C; FWHM = 5 nm (35 cm⁻¹) at 10 °C) from the ¹E → ³A₂ intraconfigurational transition (1140 nm), followed by vibrational sidebands with progressions of ≈320 cm⁻¹ ((ν₂(e)) and ≈800 cm⁻¹ ((ν₁(a₁)). This indicates the coupling of the ¹E excited state and the non-totally symmetric ν₂(e) mode of MnO₄³⁻, i.e., a dynamic Jahn–Teller effect.

The very small splitting of ¹E emission band is due to only weakly distorted MnO₄ tetrahedra (see Table 3) and it is barely visible with our instrument resolution at the emission spectrum measured at low temperatures (−190 °C). The low-intensity broad emission band from the ³T₂(³F) → ³A₂ transition is centered at 1062 nm (9416 cm⁻¹) and can be resolved only spectral deconvolution, Fig. 4d.

Temperature dependence of the ¹E lifetime and emission peak spectral position are shown in Fig. 4e, f, respectively. The ¹E level emission decays show lifetime values of about 350 μs at room temperature and 560 μs at low temperatures. The temperature dependence of lifetime, Fig. 4e, shows that a low-temperature lifetime value is approximately the value of a radiative lifetime. Considering that the excited state is separated in energy from the ground state by 8772 cm⁻¹, almost eight quanta of the highest vibrational frequencies of the phosphor (≈1100 cm⁻¹) are needed to bridge the gap. Thus, a multiphonon non-radiative relaxation is not probable as the emission quenching mechanism. The ¹E emission deactivation through the crossing with a charge transfer band is also not probable due to very high energy difference. Therefore, we assume that the thermal quenching of the ¹E state population takes place by a thermally activated cross-over via ³T₂ state, see Fig. 4g, similarly to Mn⁴⁺ activated phosphors. The temperature dependence of the emission lifetime, shown in Fig. 4e, can be described by the following equation^32,33,34,35:

$$\tau \left( T \right) = \frac{{\tau _{R0} \cdot {{{\mathrm{tanh}}}}(h\nu /2k_BT)}}{{1 + \left( {\tau _{R0} \cdot {{{\mathrm{tanh}}}}(h\nu /2k_BT)/\tau _{NR}} \right) \cdot \exp \left( { - \Delta E/k_BT} \right)}}$$

(4)

where $\tau _{R0}$ = 560 ± 19 μs is the radiative lifetime at T = 0 K, k_B = 0.69503476 cm⁻¹K⁻¹ is the Boltzmann constant, $h\nu$ = 448 ± 90 cm⁻¹ is the average energy of phonon coupled to the ¹E → ³A₂ transition, 1/$\tau _{NR}$ = 1527 ± 120 ms⁻¹ is the nonradiative decay rate, $\Delta E$ = 1631 ± 200 cm⁻¹ is the activation energy of the process (the cross-over via the ³T₂ state), and T represents the temperature. The smaller the configuration coordinate parabola offset between the ground state (³A₂) and the ³T₂ state, the larger the cross-over energy $\Delta E$ (activation energy of the process) needed to activate the non-radiative de-excitation process. Thus, the ¹E → ³A₂ emission of Mn⁵⁺ activated phosphors, which have large ³T₂ energies and smaller Stokes shifts, will start to quench at higher temperatures.

The shift of ¹E emission band with energy is shown in Fig. 4f. It can be described by the following equation:^36,37

$$\delta E\left[ {{{{\mathrm{cm}}}}^{ - 1}} \right] = \alpha \cdot \left( {\frac{T}{{\theta _D}}} \right)^4 \cdot {\int}_0^{T/\theta _D} {\frac{{x^3}}{{e^x - 1}}dx}$$

(5)

where $\theta _D$ = 783 ± 12 K is the Debye temperature of the host material, $x = \hbar \omega _D/k_BT = \theta _D/T$, $\omega _D$ is Debye cut-off frequency, and α = −650 ± 17 cm⁻¹ represents the electron-phonon coupling coefficient. The relatively high Debye temperature indicates a rigid structure which favors efficient emissions from optical centers³⁸.

The concentration dependence of an internal quantum efficiency (QE) is given in Fig. 4h. The largest value of 37.5 ± 2.0% is recorded for the 0.5% Mn⁵⁺ doped sample, after which the concentration quenching of emission occurs. This is a relatively high value for an NIR-emitting phosphor, and comparable to one obtained in Ba₃(PO₄)₂⁷. The log₁₀(QE/concentration) vs log₁₀(concentration) plot has a −1.97 slope, which is close to −2, which undoubtedly indicates that a multipolar electric dipole-dipole mechanism is responsible for the concentration quenching of emission. The linear dependence of Ca₆Ba(PO₄)₄O:Mn⁵⁺ emission intensity on excitation power, Fig. 4i, is expected for the typical downshifting photoluminescence emission process.

The 3d² electronic configuration of Mn⁵⁺ in a tetrahedral environment is described by the Tanabe–Sugano model for 3d⁸ electronic configuration in octahedral symmetry³⁹, see Fig. 4c. The crystal field and Racah parameters are calculated from the following equations using data from diffuse reflection and emission spectra^40,41:

$$Dq = \frac{{E\left( {{}^3A_2 \to {}^3T_2} \right)}}{{10}} = \frac{{10604.5}}{{10}}\;{{{\mathrm{cm}}}}^{ - 1} = 1060\;{{{\mathrm{cm}}}}^{ - 1}$$

(6)

$$\begin{array}{l}x = \displaystyle\frac{{E\left( {{\,}^3A_2 \to {\,}^3T_1} \right) - E\left( {{\,}^3A_2 \to {\,}^3T_2} \right)}}{{Dq}}\\\quad =\, \displaystyle\frac{{15649.5 - 10604.5}}{{1060.45}} = 4.757\end{array}$$

(7)

$$B = \frac{{x^2 - 10x}}{{15 \cdot (x - 8)}} \cdot Dq = 544\;{{{\mathrm{cm}}}}^{ - 1} \to \frac{{10Dq}}{B} = 19.5$$

(8)

$$\begin{array}{l}C = \displaystyle\frac{1}{2} \cdot \left( {E\left( {{\,}^3A_2 \to {\,}^1E} \right) - 10Dq - 8.5B + \frac{1}{2}\sqrt {400Dq^2 + 40DqB + 49B^2} } \right)\\\quad =\, 2292\;{{{\mathrm{cm}}}}^{ - 1}\end{array}$$

(9)

$$C/B = 4.21$$

(10)

By comparing the obtained Dq, B and C parameters with literature data, Table 5, one can observe that Ca₆Ba(PO₄)₄O provides the smallest Dq and the largest B parameters amongst all phosphate hosts, and that Li₃VO₄ is the only host with a smaller Dq (considering available data).

Table 5 Comparison of the Dq, B and C parameters (all in cm⁻¹) for the tetrahedrally coordinated Mn⁵⁺ ions in different crystalline solids

By considering the obtained parameters and the configuration coordinate diagram, Fig. 4g, the relatively small value of Huang–Rhys parameter S = 0.53 is found for the assumed coupling to the ν₁₊ ν_L/ ν₃₊ ν_L vibrational mode with energy $\hbar \omega = 1128.5\;{{{\mathrm{cm}}}}^{ - 1}$.

The Slater parameters are calculated from Racah parameters by the simple relations^42,43:

$$F^{\left( 2 \right)} = 49F_2 = 7\left( {7B + C} \right) = 42271{{{\mathrm{cm}}}}^{ - 1}$$

(11)

$$F^{(4)} = 441F_4 = 441\frac{C}{{35}} = 28877\;{{{\mathrm{cm}}}}^{ - 1}$$

(12)

Both values are considerably reduced from the free-ion values of $F^{\left( 2 \right)} = 91427{{{\mathrm{cm}}}}^{ - 1}$ and $F^{(4)} = 56625\;{{{\mathrm{cm}}}}^{ - 1}$¹⁸.

As it follows from the Tanabe–Sugano diagram for the 3d² configuration in the tetrahedral crystal field (Fig. 4c), the energy separation between the ground state ³A₂ and the first excited state ¹E (in the strong crystal field) is practically independent on the crystal field strength (both states are parallel to each other). At the same time, this energy interval is very close to the energy interval between the ³F and ¹D states of the free ion, which is determined by the Racah parameters B and C, which vary from host to host because of the covalent effects. As a result, the nephelauxetic effect is dominating in this case.

Application in luminescence thermometry

We have tested the performance of Mn⁵⁺ activated Ca₆Ba(PO₄)₄O (the sample containing 0. 5% Mn since it showed the best quantum efficiency) as a NIR luminescent thermometer operating in the second biological window and in the physiological temperature range. As can be seen from Fig. 5a, when temperature increases, the broad emission peak from the ³T₂ level in the 950 nm to 1030 nm range also increases in intensity, while the intensity of the narrow emission peak from the ¹E level around 1140 nm decreases with temperature. This occurs due to thermalization between ¹E and ³T₂ levels where the energy difference between these two levels ($\Delta E_T$) is bridged by thermally exited electrons. Consequently, a simple Boltzmann-type relation for the luminescence intensity ratio (LIR) between the two abovementioned emission intensities applies^44,45:

$$LIR\left( T \right) = \frac{{I({\,}^3T_2)}}{{I({\,}^1E)}} = B \cdot \exp \left( { - \frac{{\Delta E_T}}{{k_BT}}} \right)$$

(13)

where B is a temperature-invariant constant and T represents temperature. The fit of Eq. (13) (full line, Fig. 5b) to experimental LIR data (diamond markers, Fig. 5b) is almost perfect (R² = 0.997) and provides an energy difference $\Delta E_T\;{{{\mathrm{of}}}}\;1216\;{{{\mathrm{cm}}}}^{ - 1}$ that agrees with the energy difference obtained from spectroscopy (Fig. 4g). To experimentally determine the uncertainty in the LIR (error budget), 50 emission spectra were acquired at each temperature. Then, the measurement distribution mean was used as the LIR value while a standard deviation ($\sigma _{LIR}$) was used as an uncertainty in LIR as shown in the insert of Fig. 5b on the LIR value distribution measured at 30 °C).

Fig. 5: Luminescence thermometry with Ca₆Ba(PO₄)₄O:Mn⁵⁺ powder.

a Photoluminescence emission spectra of Ca₆Ba(PO₄)₄O:Mn⁵⁺ powder measured at different temperatures. b Luminescence intensity ratio (LIR) as a function of temperature (experimental data - diamond markers; fit to Eq. (13) - full line). The insert shows the LIR distribution histogram measured at 303.15 K (30 °C) – filled diamond marker. c Calculated absolute and relative sensitivities (marked values at 303.15 K (30 °C)). d Experimentally obtained values for temperature resolution – filled diamond marker represents the value at 303.15 K (30 °C)

The absolute (S_a) and relative (S_r) sensitivities of the thermometer were then calculated from the following equations:

$$S_a\left[ {{{{\mathrm{K}}}}^{ - 1}} \right] = \left| {\frac{{\partial LIR}}{{\partial T}}} \right|,\;S_r\left[ {{{{\mathrm{\% K}}}}^{ - 1}} \right] = 100{{{\mathrm{\% }}}} \cdot \left| {\frac{{\partial LIR}}{{\partial T}}\frac{1}{{LIR}}} \right|$$

(14)

and presented in Fig. 5c (blue dots represent values obtained at a temperature of 30 °C). The relative sensitivity value varies from 2.35%K⁻¹ to 1.26%K⁻¹ over the measurement range, being 1.92%K⁻¹ at 30 °C. These are relatively high values⁴⁶, especially for luminescence thermometers operating in the second biological transparency window (>1000 nm). For example, Gschwend et al.⁴⁷ achieved a relative sensitivity of 0.43%K⁻¹ for an LIR thermometer based on Mn⁵⁺-activated Ba₃(PO₄)₂, while Shen et al.⁴⁸ achieved a relative sensitivity of 1.3%K⁻¹ for an LIR thermometer based on Ag₂S quantum dots.

The temperature resolution (uncertainty in measured temperature, $\delta T$) is determined as a ratio between experimentally obtained LIR uncertainty ($\sigma _{LIR}$) and absolute sensitivity ($S_a$) for a given temperature, Fig. 5d:

$$\delta T = \frac{{\sigma _{LIR}(T)}}{{S_a(T)}}$$

(15)

and it has an average value of 0.21 K. Finally, repeatability of measurement (R_M) is quantified as⁴⁶:

$$R_M = 1 - \frac{{max\left| {\overline {LIR} - LIR_i} \right|}}{{\overline {LIR} }}$$

(16)

where $\overline {LIR}$ is the average LIR measured at a certain temperature over all $LIR_i$ acquired. Based on experimental data, an R_M value of 0.97 (97%) is obtained.

Conclusion

Because of its rigid structure, appropriate crystal sites for doping, and sufficiently large energy band gap to accommodate the energy levels of dopant ions, Ca₆Ba(PO₄)₄O is an excellent host for Eu²⁺, Sm²⁺ and Mn⁵⁺ luminescence centers. In this host, Mn⁵⁺ provides ultranarrow emission in the near-infrared spectral range at 1140 nm that can be easily excited over the broad spectral range that spans 500–1000 nm covering the entirety of the first biological transparency window and making this material an excellent near-infrared phosphor and non-toxic blue/turquoise pigment. The phosphor has an internal quantum efficiency of 37.5%, which is a high value considering the quantum efficiencies of inorganic NIR phosphors. The thermal quenching of the ¹E emission takes place by a thermally activated cross-over via the ³T₂ state with an activation energy of 1631 cm⁻¹. The optimal Mn⁵⁺ doping concentration is 0.5%. For higher doping concentrations, quantum efficiency decreases due to non-radiative deexcitation caused by a dipole-dipole electric interaction. Based on the available literature data, the Ca₆Ba(PO₄)₄O:Mn⁵⁺ phosphor provides the smallest Dq and the largest B parameters amongst all phosphate hosts. This material is one of the best single-doped ratiometric luminescence thermometry sensors for use in a second biological transparency window due to the opposite temperature dependence of the ¹E and ³T₂ emission intensities. It provides a relative sensitivity of 1.92%K⁻¹ and a temperature resolution of 0.2 K in the range of physiological temperatures, with a measurement repeatability of 97%. These findings, particularly the high value of quantum efficiency and strong absorption, tuneability of emission wavelength by changing the Mn⁵⁺ environment, long emission decay times that allow for time-gated measurements, and strong temperature susceptibility of emission, demonstrate the great potential of Mn⁵⁺-activated phosphors for NIR applications. When reduced to nano dimension, Ca₆Ba(PO₄)₄O:Mn⁵⁺ has great potential for bioimaging and biothermal imaging applications in the second biological transparency window. Future research will concentrate on the synthesis and applications of Ca₆Ba(PO₄)₄O:Mn⁵⁺ nanoparticles.