Host selection

In search of the ideal host for Pr³⁺ ions

Pr³⁺-based phosphors are a large group of materials, where the host lattice plays an important role in how the material behaves. These host materials can include compounds like fluorides, chlorides, phosphates, silicates, borates, and others. However, not all of them allow visible-to-UVC upconversion to occur. Because of this, choosing the right host for Pr³⁺ ions is a key step in making efficient upconverting materials.

First of all, the host determines the energy of emitted radiation in the UV range. Because the 5d electrons are not shielded from the environment like the 4f electrons, the configuration of the 4f5d levels in the crystal field of ligands is significantly impacted by the crystal structure and host material composition. The energy of the lowest, emitting 5d level of Ln³⁺ ions is determined by the parameter called redshift (D(A)), which is the difference between the energy of the first 4f → 4f5d transition for free Pr³⁺ ions and those incorporated into host A. The magnitude of D(A) depends on various parameters, including the ligand type, coordination number, polyhedron type, and the average distance between Ln³⁺ and ligand atoms [1]. It could be calculated according to the formula:

D(A) = E_5d(free)  - E_5d(A)                                                                                                      (1)

where, E_5d(free) is the position of the lowest 5 level of RE³⁺ as a free ion, and E_5d(A) likewise for RE³⁺ ion doped into compound A (see Figure 1). To obtain emission within the range of 200 – 280 nm, the value of E_5d(A) should be between 36,960 – 51,250 cm^-1 (taking into account a Stokes shift of approximately 2500 cm^-1).

Figure 1. Schematic illustration of crystal field splitting of 5d levels in octahedral field.

However, in the case of Pr³⁺ions, E_5d(A) cannot be higher than the energy of the ¹S₀ level, which is usually around 47,000 cm^-1. Otherwise, instead of broadband 4f5d → 4f emission, strong narrow bands corresponding to ¹S₀ → ³H_J, ³F_J, ¹D₂, ¹G₄, ¹I₆, and ³P₀ → to ³H_J, ³F_J transitions will be observed. This phenomenon is called ‘photon cascade emission’ (PCE) or ‘quantum cutting’ (QC) (see Figure 2). To avoid quenching of 4f5d → 4f emission by the PCE process, E_5d(A) of Pr³⁺ should not be higher than 46,000 cm^-1. Consequently, hosts that provide appropriate splitting of 5d orbital must be characterized by D(A) between 15,580 cm^-1 and 24,680 cm^-1.

Figure 2. Energy level diagrams presenting luminescence processes under VUV excitation for Pr³⁺ ions with different energies of the lowest 5d level.

Peter Dorenbos collected and summarized information about crystal field splitting of Ln³⁺ 5d levels for over 300 compounds [2]. According to his study, fluorides, which stand out from low phonon energies, have small values of D(A), meaning they are more suitable for the PCE process rather than broadband UVC emitters. Although chlorides or bromides exhibit a larger redshift of 5d levels, they are very often hygroscopic, and by absorbing water, their luminescence can be quenched. On the other side, oxide matrices, due to the huge variety of materials including simple and complex oxides, are characterized by a wide range of D(A) values (10 000 – 30 000 cm^-1). The above properties combined with high physical and chemical stability make Pr³⁺-doped oxides, such as phosphates, silicates, and borates, the most convenient materials for UVC light sources.

In addition to affecting the energy of the 5d configuration, the host also significantly influences the efficiency of visible-to-UVC upconversion. The first key factor is the bandgap energy (E_g) between the valence and conduction bands in the matrix. For effective UVC emission, the 5d levels of Pr³⁺ ions must be energetically below the conduction band edge to prevent energy loss via ionization. The simplified scheme of the ionization process is presented in Figure 3A. Another influential factor is the Stokes shift (ΔS), the energy difference between absorption/excitation and emission bands. ΔS affects the probability of non-radiative relaxation of 5d states via thermally induced crossover quenching: when ΔS is large (Figure 3B, left), quenching is efficient, favoring visible emission. When ΔS is small (Figure 3B, right), quenching occurs only at high temperatures, allowing most energy to be emitted as UVC radiation. This suggests that host compounds inducing a small ΔS are optimal for blue-to-UVC upconversion with Pr³⁺.

Figure 2. Schematic presentations of factors influencing the efficiency of the Vis-to-UVC upconversion process. (A) The bandgap energy between the valence (VB) and conduction (CB) bands. (B) The decay time of the intermediate level (τ_int). (C) Stokes shift (ΔS) of UVC emission.

The decay time of the intermediate level (τ_int) is another crucial parameter influencing UC luminescence. Unlike other factors, τ_int is directly linked to the upconversion process itself. Generally, a longer τ_int increases the probability of second-photon absorption, thus enhancing UC process efficiency (see Figure 3C). Most studies identify the ³P₀ level as an intermediate state in upconversion. Consequently, the conventional approach to enhancing Vis-to-UVC UC has been to increase the ³P₀ excited state decay times by selecting host crystals with lower phonon energies. In such hosts, multiphonon relaxation (MPR) from the ³P₀ to the lower ¹D₂ level and phonon-assisted cross-relaxation (CR) [³P₀, ³H₄ → ¹D₂, ³H₆] processes are reduced, leading to an extended ³P₀ lifetime (see the ESA 1 or ETU 1 mechanism in Figure 4). This approach was applied by Cates et al. [3]  in Lu₇O₆F₉:Pr³⁺, which showed a higher UVC UC luminescence compared to Y₂SiO₅:Pr³⁺. Despite some improvements, gains in UC efficiency have been lower than expected, leading to further investigation of the ¹D₂ level as an intermediate. This level can be effectively fed in a high-phonon energy host by MPR and CR processes (see the ESA 2 or ETU 2 mechanisms in Figure 4). Because the ¹D₂ level has a significantly longer decay time than the ³P₀ level, the enhancement of UC luminescence can be observed, which was subsequently demonstrated in β-Y₂Si₂O₇:Pr³⁺ [4]. Our recent studies also further supported this strategy. We developed a high-phonon Sr₃(BO₃)₂:Pr³⁺ phosphor with efficient ³P₀ → ¹D₂ MPR, achieving a significant increase in UC luminescence compared to Y₂SiO₅:Pr³⁺ [5]. These findings open up promising potential for high-phonon energy compounds in efficient visible-to-UVC upconversion, which have so far been underestimated.

Figure 4. Schemes of possible upconversion mechanisms: (A) excited state absorption (ESA) and (B) energy transfer upconversion (ETU) observed for low- (1) and high-phonon energy (2) materials. For simplicity, the CR processes were omitted in the ETU mechanism.

Bibliography: 

[1] Dorenbos, P. 5d-Level Energies of Ce³⁺ and the Crystalline Environment. I. Fluoride Compounds. Phys Rev B 2000, 62, 15640–15649, doi:https://doi.org/10.1103/PhysRevB.62.15640.

[2] Dorenbos, P. The 5d Level Positions of the Trivalent Lanthanides in Inorganic Compounds. J Lumin 2000, 91, 155–176, doi:https://doi.org/10.1016/S0022-2313(00)00229-5.

[3] E. L. Cates, A. P. Wilkinson, J.-H. Kim, Visible-to-UVC upconversion efficiency and mechanisms of Lu₇O₆F₉:Pr³⁺ and Y₂SiO₅:Pr³⁺ ceramics, J Lumin 2015, 160, 202, doi: https://doi.org/10.1016/j.jlumin.2014.11.049.

[4] E. L. Cates, F. Li, Balancing intermediate state decay rates for efficient Pr³⁺ visible-to-UVC upconversion: the case of β-Y₂Si₂O₇:Pr³⁺, RSC Adv 2016, 6, 22791, doi: https://doi.org/10.1039/C6RA01121G.

[5] P. Zdeb, N. Rebrova, P. J. Dereń, Discovering the Potential of High Phonon Energy Hosts in the Field of Visible-to-Ultraviolet C Upconversion, J Phys Chem Lett 2024, 15, 9356. doi:https://doi.org/10.1021/acs.jpclett.4c02053.