Single crystal growth, optical absorption and luminescence properties under VUV-UV synchrotron excitation of type III Pr3+:KGd(PO3)4

Scintillator materials are widely used for a variety of applications such as high energy physics, astrophysics and medical imaging. Since the ideal scintillator does not exist, the search for scintillators with suitable properties for each application is of great interest. Here, Pr3+-doped KGd(PO3)4 bulk single crystals with monoclinic structure (space group: P21) are grown from high temperature solutions and their structural, thermal and optical properties are studied as possible candidates for scintillation material. The change in the unit cell parameters as a function of the Pr3+ level of doping and temperature is studied. Differential thermal analysis reveals that KGd0.942Pr0.058(PO3)4 is stable until 1140 K. The 5d3, 5d2 and 5d1 levels of Pr3+ with respect to the 3H4 ground state are centred at 166, 196 and 218 nm, respectively, in this host. The luminescence of KGd0.990Pr0.010(PO3)4, by exciting these 5d levels, shows intense emissions centred at 256 and 265 nm from the 5d1 to 3F3,4 and 1G4 levels of Pr3+ with a short decay time of 6 ns. The 6P3/2,5/2,7/2 → 8S7/2 transitions of Gd3+ appear after exciting the 5d levels of Pr3+ and the 4 f levels of Gd3+, showing an energy transfer between Pr3+ and Gd3+.


Results and discussion
Bulk single crystal growth. Table 1 shows the crystal growth experiments carried out for different levels of Pr 3+ doping and the crystals obtained.
The saturation temperature of all solutions was around 993 K and no significant changes were observed with Pr 3+ doping at the levels studied in this work. The saturation temperature is slightly higher than that previously reported for the crystal growth of Ce:KGd(PO 3 ) 4 crystals 18 grown in similar experimental conditions. The crystals obtained were generally transparent, free of inclusions and cracks, and slightly greenish due to Pr 3+ doping. The sizes obtained were a* × b × c* = 7.6-13.6 mm × 17.1-24.1 mm × 10.7-13.6 mm and the weights ranged from 2.7 to 6.9 g. As can be seen in Table 1, the crystal dimension in b direction was always higher than in a* and c* directions. This faster growth in the b crystallographic direction has already been reported by us for different doping KGd(PO 3 ) 4 crystals [16][17][18] and is in agreement with the non-presence of the (010) crystalline face. The growth rate of the {hkl} form is inversely proportional to interplanar spacing d hkl and the sequence in this crystal is d 001 > d 100 > d 020 . The chosen orientation of the seed also reinforces this behaviour, together with the thermal gradients in the solution. The crystal growth rate varies from 6.1 × 10 −3 to 17.4 × 10 −3 g·h −1 .
As an example, Fig. 1 shows an as-grown Pr:KGd(PO 3 ) 4  Using the atomic percentage of each element present in the chemical compound as obtained from the EPMA (electron probe microanalysis) results, the chemical formula of each crystal was determined and the distribution coefficient of Pr 3+ in KGd(PO 3 ) 4 Table 2 shows, for the five doping levels studied, the Pr 3+ atomic ratio with respect to Gd 3+ in the crystal, the chemical formula, the number of Pr 3+ ions per unit cell volume in the crystal and the distribution coefficient of Pr 3+ (K Pr ). For the crystals obtained from solutions with 2.00 and 5.00 atomic % of Pr 3+ in the solution, the atomic percentage of each element was measured at several points along the a* and c crystallographic directions in a plate perpendicular to b crystallographic axis, obtaining up to about 85 results per sample. These results (see Fig. S1 in Supporting Information) showed that the variation of the measured values of Pr 3+ atomic concentration along the crystallographic directions is of the same order as the error in the measurements. Therefore, the results indicate the uniformity of Pr 3+ atomic concentration of the growing crystals along the a* and c crystallographic directions, up to 5.8 atomic % of Pr 3+ in the crystal. In the samples obtained from solutions with 0.25, 0.50 and 1.00 atomic % of Pr 3+ in the solution, the EPMA measurements were carried out far from the undoped KGd(PO 3 ) 4 seed, i.e. in the last stages of the crystal growth. Taking into account the values of the Pr 3+ distribution coefficient in KGd(PO 3 ) 4 and their error, it can be observed that the Pr 3+ distribution coefficient is not far from the unit in any of the concentrations studied. Besides, the results do not show any significant tendency for its value (K Pr ) to decrease or increase as the level of Pr 3+ doping in the solution increases.
Structural characterization. X-ray powder diffraction analysis was carried out to study the evolution of the unit cell parameters of KGd 1-x Pr x (PO 3 ) 4 depending on the Pr 3+ doping concentration. The refinement of the unit cell parameters was carried out using the TOPAS program 20 . Table 3 shows the unit cell parameters of the crystals studied, while Fig. 2 shows the evolution of these parameters as a function of the praseodymium content in KGd(PO 3 ) 4 . It can be seen that there is an ascending linear behaviour of the unit cell parameters on increasing the Pr 3+ content in the crystals. As Table 3 and Fig. 2 show, the a and β parameters remain practically the same, the b and c parameters increase slightly and the unit cell volume clearly increases when the Pr 3+ concentration in the crystal increases. This behaviour is expected because the ionic radius of Pr 3+ with coordination VIII is higher than the ionic radius of Gd 3+ with the same coordination (1.126 Å and 1.053 Å, respectively 21 ).
thermal stability. Figure 3 shows the thermogram obtained for KGd 0.942 Pr 0.058 (PO 3 ) 4 in both the heating and cooling processes in the range 500-1273 K. The weight change during the experiment was not significant.
An endothermic peak beginning at 1140 K can be observed in the heating process, which is attributed to the incongruent melting process of KGd 0.942 Pr 0.058 (PO 3 ) 4 . This temperature is so similar to the 1142 K obtained for undoped KGd(PO 3 ) 4 22 that it can be said that there is no appreciable difference in the incongruent melting temperature of KGd(PO 3 ) 4 with the Pr 3+ doping, at least up to 5.8 atomic % of Pr 3+ in the crystal. Meanwhile, no heat   exchange in the cooling process of the sample was observed, which means that no crystalline phase transitions were produced during the process. Figure 4 shows the X-ray powder diffractogram of KGd 0.942 Pr 0.058 (PO 3 ) 4 at room temperature, its evolution with the temperature up to 1273 K and the cooling process up to room temperature. The temperatures written to the right of the graph are used as labels, since it is expected that the temperature distribution in the sample support during these measurements was not homogeneous. This could lead to a partial incongruent melting when the thermocouple of the diffractometer chamber indicated 1093 K.    www.nature.com/scientificreports www.nature.com/scientificreports/ The diffraction standard patterns of KTb(PO 3 ) 4 (89-1424 ICDD database 23 and GdPO 4 (83-0657 ICDD database 24 , together with a Pt diffraction peak (the sample holder was of Pt), are also shown in Fig. 4. The diffraction standard pattern of type III KTb(PO 3 ) 4 (space group: P2 1 ) was used due to the non-existence of the type III KGd(PO 3 ) 4 powder diffraction standard pattern in the version of the ICDD database used. Hence at room temperature all diffraction peaks correspond to the monoclinic crystalline phase of type III KGd(PO 3 ) 4 together with a diffraction peak belonging to the Pt crystalline phase of the sample holder. Till 1078 K, there are no extra peaks of any other crystalline phase, and the diffraction peaks belonging to KGd(PO 3 ) 4 present a decrease in their sharpness and intensity, related to the loss of crystallinity. At 1093 K a small peak appeared at 29.2° and was identified as GdPO 4 . At 1108 K, the two crystalline phases (KGd(PO 3 ) 4 and GdPO 4 ) are coexistent, and at 1123 K the KGd(PO 3 ) 4 crystalline phase has totally decomposed. In Ponceblanc et al. 25 , differences in the phase transition temperature of the same compound were also observed depending on both the heating rates and technique used. From 1093 to 1108 K, as the intensity of the type III KGd(PO 3 ) 4 peaks decreased, the intensity of the GdPO 4 peaks increased. From 1123 K to 1273 K only the diffraction peaks of GdPO 4 can be observed, meaning that this crystalline compound is stable at this range of temperatures. Throughout the cooling process from 1273 K until room temperature (see the last four diffractograms) there are no significant changes, so the GdPO 4 remains stable. This means that the phase transition is not reversible, as expected for an incongruent melting, and in our case the solidification of the liquid phase leads to an amorphous phase.
Therefore, according to the differential thermal analysis and X-ray powder diffraction results, KGd 0.942 Pr 0.058 (PO 3 ) 4 decomposes at 1140 K into GdPO 4 and liquid phase, which probably consisted of a mixture of phosphorus and potassium oxides, since the sample weight remained practically constant.
The studies on KGd(PO 3 ) 4 22 and KYb 0.029 Gd 0.971 (PO 3 ) 4 17 are comparable to that presented in our work. The results for the first compound show that KGd(PO 3 ) 4 decomposes irreversibly at 1142 K into Gd(PO 3 ) 3 , GdPO 4 , Gd 2 P 4 O 13 and an amorphous phase, and that at room temperature after the cooling process only GdPO 4 remains. Regarding the second case, KYb 0.029 Gd 0.971 (PO 3 ) 4 decomposes irreversibly at 1130 K into Gd(PO 3 ) 3 , at 1223 K Gd(PO 3 ) 3 , GdPO 4 , Gd 2 P 4 O 13 , GdP 5 O 14 and an amorphous phase coexist, and at room temperature after the cooling process the GdPO 4 and Gd 2 P 4 O 13 crystalline phases remain. Thus the difference in the thermal evolution observed for the KGd 0.942 Pr 0.058 (PO 3 ) 4 is that this crystal is decomposed into a unique crystalline compound, GdPO 4 , and a liquid phase. The intermediate Gd(PO 3 ) 3 crystalline compound observed in the previous studies is not observed in our case, and neither are the crystalline phases Gd 2 P 4 O 13 and GdP 5 O 14 from the previous works present in our case. Only the GdPO 4 crystalline phase is observed to be stable till room temperature in all three studies 17,22 .
Another compound whose thermal decomposition has been studied in the literature is KLa(PO 3 ) 4 26 , which decomposes into La(PO 3 ) 3 , LaPO 4 and an amorphous phase containing phosphorus and potassium oxides. Thus, Linear thermal expansion tensor. Bearing in mind the X-ray powder diffractograms measured in the range from room temperature up to 773 K and using the Le Bail method 27 , the unit cell parameters at different temperatures in the P2 1 space group were refined. The parameters relating to goodness of fit are R wp and R exp , whose values must fulfil the expression R wp ≤ 2·R exp for it to be considered that a good fit is obtained. In all cases these parameters are around R wp = 21.06 and R exp = 18.08. Table 4 shows the unit cell parameters of KGd 0.942 Pr 0.058 (PO 3 ) 4 at different temperatures, while Fig. 5 shows the relative thermal evolution of these parameters with respect to those at room temperature as a function of temperature. It can be seen that the unit cell parameters follow a linear trend. The a and b parameters clearly increase with temperature, as does the c parameter but in minor proportion, while the β parameter decreases slightly. From these results, the linear thermal expansion coefficients in each crystallographic direction can be calculated using the expression α = (ΔL/ΔT)/L RT , where ΔL/ΔT is the slope of the linear fit of the change of each unit cell parameter with the temperature, and L RT is the unit cell parameter at room temperature, 303 K. The linear thermal expansion coefficients in the crystallophysical system X 1 ||a, X 2 ||b and X 3 ||c* are α 11 = 12.00 × 10 -6 K −1 , α 22 By diagonalizing the tensor, the thermal expansion values in the principal axes of the tensor X 1 ' , X 2 '||b and X 3 ' are obtained, these being α' 11 = 12.43 × 10 −6 K −1 , α' 22 = 12.40 × 10 −6 K −1 and α' 33 = 7.03 × 10 −6 K −1 . The X 1 ' axis is at 16.31° clockwise from the a axis, while X 3 ' is at 14.44° from the c axis with the positive b axis pointing toward the observer (see Fig. 6). In Fig. 7a, a broad band centred at 218 nm (45872 cm −1 , 5.69 eV) is observed, which corresponds to the electronic transition from the 3 H 4 ground state of Pr 3+ to its lowest 5d level (5d 1 ) in KGd(PO 3 ) 4 . From the crystallographic point of view, there is only one site expected, with C 1 point symmetry, for the Pr 3+ and Gd 3+ ions in   28 . Therefore all Pr 3+ ions have the same crystal field, and consequently only one band for the 3 H 4 → 5d 1 transition is expected. This transition of Pr 3+ has been systematically studied in many hosts and it can be predicted by considering the study carried out by Dorenbos 29 . In this work, the value of the 3 H 4 → 5d 1 transition of Pr 3+ in phosphate hosts varies from 212 nm for LaP 3 O 9 30 , through 222 nm for YPO 4 30,31 , to 224 nm for YP 3 O 9 30 . In addition, since the average energy difference of the first spin-allowed 4 f → 5d transition of Pr 3+ ( 3 H 4 → 5d 1 ) with respect to the transition of Ce 3+ ( 2 F 5/2 → 5d 1 ) in the same host 29 is 12240 ± 750 cm −1 and because the 2 F 5/2 → 5d 1 transition of Ce 3+ in KGd(PO 3 ) 4 is centred at 302.5 nm (33058 cm −1 , 4.10 eV) 18  It should be noted that, of the optical absorption measurements, only the 5d 1 absorption band was identified out of all the 5d x levels of Pr 3+ in KGd(PO 3 ) 4 . The reason for this is that, although the UV limit of the equipment used is 175 nm, the measurements were carried out in air atmosphere, and below 190 nm the air absorption hid the other 5d x levels. The 5d 2 and 5d 3 energy levels, together with the 5d 1 energy level already determined, of Pr 3+ in KGd(PO 3 ) 4 were quantified in the UV-VUV synchrotron measurements (see next section) by studying the excitation spectra for several emission wavelengths. optical emission. Figure 8 shows the emission spectra of Pr:KGd(PO 3 ) 4 under λ exc = 218 nm (45872 cm −1 , 5.69 eV), 196 nm (51020 cm −1 , 6.33 eV) and 166 nm (60241 cm −1 , 7.47 eV) at room temperature.
Also in Fig. 8a, parity-allowed transitions from the d to f levels of Pr 3+ , which are of great interest for scintillation applications, are also observed. These broad, intense bands correspond to the 5d 1 → 3 H 5 , 5d 1 → 3 H 6 , 5d 1 → 3 F 3,4 and 5d 1 → 1 G 4 electronic transitions of Pr 3+ , centred at 229, 239, 256 and 265 nm, respectively. The emission band corresponding to the electronic transition from the 5d 1 level to the 3 H 4 ground state of Pr 3+ does not appear, probably due to the self-absorption effect, as occurs in AREP 2 O 7 hosts (A = Na, K, Rb, Cs; RE = Y, Lu) 32 . We should also note that the significantly weaker broad bands centred at 358 and 435 nm correspond to the 5d 1 → 1 D 2 and 5d 1 → 3 P 2 transitions. The most intense band originating in a 5d level is the broad band located around 256-265 nm, which corresponds to the overlapping of three electronic transitions, 5d 1 → 3 F 3,4 and 5d 1 → 1 G 4 . By exciting the 5d 1 level of Pr 3+ at 210 nm in LiYF 4 at 10 K, it can be observed how 5d 1 → 3 H 4 , 5d 1 → 3 H 5 , 5d 1 → 3 H 6 , 5d 1 → 3 F 3,4 and 5d 1 → 1 G 4 transitions appear at 220, 230, 245, 255 and 272 nm, respectively, while 5d 1 → 1 D 2 and 5d 1 → 3 P 2 transitions do not appear. It should also be noted that the band that corresponds to the 5d 1 → 3 H 5 transition is the most intense, while that corresponding to the 5d 1 → 1 G 4 transition is the least 33 . Under direct 4 f → 5d 1 excitation (280 nm) of Pr 3+ in Lu 3 Al 5 O 12 , Lu 3 Al 4 GaO 12 and Lu 3 Al 3 Ga 2 O 12 hosts, www.nature.com/scientificreports www.nature.com/scientificreports/ the emission bands corresponding to the 5d 1 → 3 H J and 5d 1 → 3 F J electronic transitions appear centred at 310 and 360 nm, respectively, with the first band being the most intense. The bands corresponding to the 5d 1 → 1 G 4 , 5d 1 → 1 D 2 and 5d 1 → 3 P J transitions do not appear. The emission peaks corresponding to some of the 4 f → 4 f electronic transitions of Pr 3+ are insinuated in the visible range from 480 to 760 nm 34 . The emission spectrum of La 0.999 Pr 0.001 PO 4 at 300 K under the direct 4 f → 5d 1 excitation (193 nm) of Pr 3+ shows intense emission bands centred at about 230, 240 and 255 nm corresponding to the 5d 1 → 3 H 4 , 5d 1 → 3 H 5 and 5d 1 → 3 H 6 , 3 F 2 electronic transitions, respectively. Two very weak, broad emission bands appear centred around 375 and 440 nm corresponding to the 5d 1 → 1 D 2 and 5d 1 → 1 I 6 , 3 P J transitions. It should be noted that an emission peak centred at approximately 610 nm, with an intensity similar to the two previous bands, corresponds to the 1 D 2 → 3 H 4 electronic transition. The most intense emission bands correspond to the 5d 1 → 3 H 4 and 5d 1 → 3 H 5 transitions, and the least intense to the 5d 1 → 1 D 2 transition 35 .
Finally, it is important to note that in Pr:KGd(PO 3 ) 4 , no 4 f → 4 f transitions of Pr 3+ were observed under excitations in the 120-248 nm range. The assignation of the Gd 3+ and Pr 3+ transitions was checked by comparing the Pr:KGd(PO 3 ) 4 emission spectra with those of the undoped KGd(PO 3 ) 4 excited at λ exc = 218, 196 and 166 nm (see Fig. S3 in Supporting Information) and by consulting the Dieke's diagram 38 , the extended Dieke's diagram 39 , and the work carried out by Wegh et al. 40 and by Yang et al. 36 .
In the excitation spectrum of Pr:KGd(PO 3 ) 4 crystal under VUV-UV radiation from 120 to 248 nm for λ emi = 265 nm corresponding to the 5d 1 → 1 G 4 electronic transition (see Fig. 9a), the excitation of the 5d 3, 5d 2 and 5d 1 levels of Pr 3+ in KGd(PO 3 ) 4 is produced at wavelengths around 166, 196 and 218 nm, respectively. Some 4 f levels of Gd 3+ were also excited giving rise to this Pr 3+ emission, which could be explained by an energy transfer from Gd 3+ to Pr 3+ .
As for the excitation spectrum for the Gd 3+ emission at λ emi = 592 nm (see Fig. 9b), although this emission could also be observed by excitation of the 5d energy levels of Pr 3+ and the consequent energy transfer to Gd 3+ , it seems that it is more favoured when the 6 G J levels of Gd 3+ are excited.
Given the calculations explained in our previous work on type III Ce:KGd(PO 3 ) 4 single crystals 18 , the energy of the exciton creation (E ex ) and the approximate energy difference from the bottom of the conduction band to the top of the valence band (E VC ) of the type III KGd(PO 3 ) 4 host were predicted at 7.57 eV (164 nm) and at 8.17 eV (152 nm), respectively. Therefore, as already mentioned in the previous work, the E ex band could appear in the excitation spectra (Fig. 9), but it would not be appreciated due to an overlapping with the 8 S 7/2 → 4 F 9/2 transition of Gd 3+ . Figure 10 shows the energy levels scheme for Pr 3+  Decay time measurements. Time profiles were recorded in two time regimes to measure the fast and slow components of the decay curves of the emission at 256 nm and 312 nm, respectively. These curves are shown in Fig. 11. Pr:KGd(PO 3 ) 4 crystals were excited at 218 nm ( 3 H 4 → 5d 1 of Pr 3+ ) and 166 nm ( 3 H 4 → 5d 3 of Pr 3+ and 8 S 7/2 → 6 H 13/2 of Gd 3+ ) under pulsed synchrotron radiation. In order to improve the photon flux reaching the sample, the excitation radiation was not exactly monochromatic but had a bandwidth of around 7%.
As can be seen in Fig. 11a, for the fast component all the decay curves can be fitted to single exponential decays with a time constant of around 6 ns. This value can be attributed to the lifetime of the 5d 1 level of Pr 3+ in KGd(PO 3 ) 4 , which is significantly shorter than the lifetimes obtained in Ce:KGd(PO 3 ) 4 18 . The lifetimes obtained for the 5d 1 emitting electronic state of Pr 3+ in different hosts are usually longer than 6 ns, as observed in Table 5. The shortening of this lifetime in the KGd(PO 3 ) 4 host could be related to the energy transfer from Pr 3+ to Gd 3+ , this being an additional depopulation channel through a non-radiative decay of the 5d 1 emitting state of praseodymium.
It can also be seen that no significant changes in lifetime with the Pr 3+ doping content of KGd(PO 3 ) 4 came about, at least up to 5.8 atomic % Pr:KGd(PO 3 ) 4 . And no important quenching of the emission is expected due to the concentration effect, so no shortening of the lifetime. This can be seen in the long lifetime of the LiPrP 4 O 12 phosphate (Table 5), where Pr 3+ is not a doping element but a host element.
As previously stated, the emission at around 312 nm observed in Fig. 8 corresponds to some of the 4 f → 4 f electronic transitions of Gd 3+ . The decay time of these electronic transitions (lifetime could be in μs or ms) is significantly slower than those from the 5d energy levels and could not be measured at the DESIRS www.nature.com/scientificreports www.nature.com/scientificreports/ beamline (DESIRS-6.65 m Monochromator) due to the interpulse duration of 1118 ns of the synchrotron radiation. Figure 11b shows its slow decay time. By fitting to a single exponential decay, the lifetime obtained is 9 ms. As reported previously in the literature, the origin of this slow component could be attributed to the Gd 3+ emission corresponding to the electronic transition 6 P J → 8 S 7/2 along with some contribution from the trapping effect, since the lifetime of the emitting 6 P 7/2 level of Gd 3+ is of the same order (4.9 ms in NaY 0.80 Gd 0.20 PO 4 and 6.36 ms in NaGd(PO 3 ) 4 ) 41,42 as the 9 ms.  It has been demonstrated that KGd 0.942 Pr 0.058 (PO 3 ) 4 is thermally stable up to 1140 K, where it suffers an irreversible decomposition into a unique crystalline compound, GdPO 4 , and a liquid phase. The X 1 ' principal axis of the thermal tensor of this crystal is at 16.31° clockwise from the a crystallographic direction when the positive b axis (parallel to the X 2 ' principal axis) is pointing toward the observer and the X 3 ' principal axis is at 14.44° clockwise from the c axis. The absorption bands corresponding to the 3 H 4 → 5d 1 , 3 H 4 → 5d 2 and 3 H 4 → 5d 3 electronic transitions of Pr 3+ in KGd(PO 3 ) 4 are centred at 218, 196 and 166 nm, respectively. The 1 S 0 energy level of Pr 3+ overlaps with the 5d 1 level of the same ion in this host, preventing the non-radiative relaxation from the 5d 1 level to the 1 S 0 energy level and the radiative relaxation between 4 f levels that would deteriorate the scintillation efficiency of these crystals. Under ultraviolet excitation, an intense, broad emission band located around 256-265 nm was observed in all grown crystals, corresponding to the 5d 1 → 3 F 3,4 and 5d 1 → 1 G 4 electronic transitions of Pr 3+ . The lifetime of the 5d 1 level of the Pr 3+ in type III KGd(PO 3 ) 4 was measured for the emission band centred at 256 nm and a lifetime of around 6 ns was obtained, which is of great interest for scintillator applications. In most cases this lifetime is shorter than the lifetime obtained for the same level of Pr 3+ in other hosts. Moreover, the emission spectra show a broad band in the visible range corresponding to the 5d 1 → 3 P 2 transition of Pr 3+ with enough intensity under 218 nm excitation, which could be an appropriate transition for use in scintillator applications. It should also be noted that the 6 P 3/2,5/2,7/2 → 8 S 7/2 electronic transitions of Gd 3+ were observed centred at 305, 312 and 323 nm by exciting the 5d levels of Pr 3+ , although it would be interesting to study whether the same behaviour occurs under X-ray excitation. A platinum stirrer with a diameter of 18 mm, located at 12-14 mm below the solution surface and rotating at 55 rpm with a change of rotation direction every 50 s, was used to mix the solution. This stirring was needed because of the high level of dynamic viscosity of the growth solution, around 19 Pa·s 43 . An a* oriented KGd(PO 3 ) 4 seed was placed in contact with the surface of the solution at 12 mm from the solution centre, rotating with the same angular velocity as the stirrer. The use of this crystallographic direction in the KGd(PO 3 ) 4 seeds leads to the growth of high crystalline quality crystals 16,22 . The crystallographic c direction of the seed was oriented in radial direction, while its b direction was tangential to the rotation movement in order to achieve a good aerodynamic orientation of the crystal during its movement. Note that the morphology of this crystal usually presents an edge perpendicular to the b crystallographic direction, while it has natural {001} faces. With the aim of further improving the mixing of the solution, high thermal gradients (around 11 K·cm −1 in depth, with the hottest point at the bottom) were applied.
Once the solution was homogeneous, its saturation temperature was determined by accurately measuring the growth/dissolution rate of the KGd(PO 3 ) 4 seed depending on the temperature, which will then be used to start the growth of the single crystal. Beginning at the saturation temperature of the solution, cooling rates of 0.1 K·h −1 for the first 15 K and 0.05 K·h −1 for the next 10-15 K were applied to create supersaturation and grow the single crystal. At the beginning of the crystal growth experiments, the cooling rate was higher in order to initiate growth and not lose contact between the crystal seed and the solution. During this initial cooling rate regime, the supersaturation of the solution increases gradually because of the difficulty in mixing the solution due to its high viscosity. After decreasing the temperature of the furnace 15 K, a second cooling rate was applied that was slower than the first to avoid an additional increase in the supersaturation of the solution, since this could induce nucleation in different points of the solution and also inclusions of solution inside the crystals. The growth rate can be maintained even with a slower cooling rate due to the accumulated supersaturation of the solution and the larger crystal surface.
After finishing the thermal cooling ramps, the crystal was removed from the solution and maintained at a few mm above the surface of the solution while the furnace was cooled to room temperature at a rate of 20-25 K·h −1 .  www.nature.com/scientificreports www.nature.com/scientificreports/ Electron probe microanalysis (EPMA) with wavelength dispersive spectrometry (WDS) was used to determine the Pr content of the crystals. In this non-destructive technique, an electron beam is focused on the sample and the characteristic X-rays emitted (specific to each element of the sample) are dispersed by crystals (WDS) before being recorded and compared with the emission of standard compounds containing the elements to be analysed. The X-rays of the sample and the standards are obtained under the same measurement conditions. The equipment used was a JEOL JXA-8230. The standards used were an undoped KGd(PO 3 ) 4 single crystal for K, Gd, P and O measurements and an REE-1 for determining the Pr concentration. An accelerating voltage of 20 kV and a current of 20 nA were applied, with measuring time of 10 s for K, P, Gd and O and 100 s for Pr peaks and 5 s and 50 s for background measuring, respectively. Kα lines of K, P and O and Lα lines of Gd and Pr were used. The dispersive crystals were PETJ for K, PETH for P, LDE1 for O, and LIFL for Gd and Pr measurements. The detection limit of Pr 3+ was around 105 ppm.
Structural characterization and thermal stability. The evolution of the unit cell parameters of Pr:KGd(PO 3 ) 4 with the Pr 3+ content was studied by X-ray powder diffraction measurements, using a D5000 Siemens X-ray powder diffractometer in vertical θ-θ configuration with the Bragg-Brentano geometry. The X-ray diffraction patterns of undoped KGd(PO 3 ) 4 and 1, 2 and 5 atomic % Pr:KGd(PO 3 ) 4 in solution were obtained using Cu Kα radiation and recorded in the 2θ range from 10 to 70°. The measurements were made with a step size of 0.03° and a step time of 7 s. The unit cell parameters were refined using the TOPAS program 20 , the Le Bail method 27 and the crystal data for undoped type III KGd(PO 3 ) 4 studied by Parreu et al. 28 (171710 ICSD database).
The thermal stability of the KGd(PO 3 ) 4 doped with praseodymium was studied by X-ray powder diffraction. The equipment used was a Bruker-AXS D8-Discover diffractometer equipped with a Cu source, a parallel incident beam (Göbel mirror), a HI-STAR GADDS (general area detector diffraction system) and a MRI BTS-Solid temperature chamber with a platinum ribbon heating stage. The powder samples were placed in the centre (occupying an area of ~1 × 1 mm 2 ) on the platinum ribbon. This stage was covered with a beryllium dome to maintain temperature. The sample was heated and cooled at a rate of 10 K·min −1 . Diffraction patterns in the heating and cooling cycles were recorded every 15 K between 1048 and 1273 K and twice at room temperature, one diffraction pattern before the heating process and the other after the cooling process. The measurements were made in the 2θ range from 18 to 52° (one frame) with a detector-sample distance of 15 cm, an exposition time of 300 s per frame and a delay time of 60 s before each frame.
To complement the study of the thermal stability of the KGd(PO 3 ) 4 doped with praseodymium, differential thermal and thermogravimetric analyses (DTA-TGA) were performed using a TA Instruments SDT 2960 Simultaneous DSC-TGA. Al 2 O 3 was used as the reference material, and the heating and cooling rates were at 10 K·min −1 with an air flux of 90 cm 3 ·min −1 .
The evolution of the unit cell parameters of the crystals grown from a 5 atomic % Pr:KGd(PO 3 ) 4 in solution with temperature was also studied by X-ray powder diffraction. The equipment was the same D5000 Siemens X-ray powder diffractometer previously used to study the Pr:KGd(PO 3 ) 4 unit cell parameters, but with an Anton-Paar HTK10 temperature chamber with a platinum ribbon heating stage. The sample was placed in the centre (occupying an area of ~9 × 5 mm 2 ) on the platinum ribbon. The diffraction patterns were recorded at temperatures of 298, 323, 373, 473, 573, 673 and 773 K (in which the monoclinic P2 1 crystalline phase of KGd(PO 3 ) 4 is stable), in the 2θ range from 10 to 70° with a step size of 0.03°, a step time of 5 s and a delay time of 300 s before each measurement. As before, the unit cell parameters were refined using the TOPAS program 20 , the Le Bail method 27 and the crystal data for undoped type III KGd(PO 3 ) 4 studied by Parreu et al. 28 (171710 ICSD database). optical characterization. The bulk single crystals obtained were cut in plates perpendicular to the crystallographic a*, b and c* directions with a diamond saw. The plates were initially lapped and then polished with Al 2 O 3 particle solutions to a size of 0.1 μm using a Logitech polishing machine. These plates were used for the optical absorption and emission studies. The unpolarized optical absorption of Pr 3+ in KGd(PO 3 ) 4 was studied using a CARY 5000 UV-Vis-NIR spectrophotometer at room temperature in the wavelength range from 205 to 2475 nm, while the unpolarized optical absorption of undoped KGd(PO 3 ) 4 was studied in the wavelength range from 190 to 315 nm.
The emission spectroscopy was studied under vacuum ultraviolet-ultraviolet (VUV-UV) excitation in the wavelength range from 120 to 248 nm (10-5 eV). Experiments were performed in the DESIRS beamline at SOLEIL Synchrotron, France (proposal number 20151215, standard). The samples were placed in a vacuum chamber which can be evacuated to a pressure below 2 × 10 −5 bars. A lithium fluoride window at the entrance of the vacuum chamber separates it from the synchrotron line. The monochromatized synchrotron light reached the sample at an angle of 90°. The emitted light from the sample was collected at 45°, focused with a silica lens and analysed with an Ocean Optics Jaz spectrometer (minimum spectral resolution 0.3 nm). The emission spectra were recorded in the range from 192 to 886 nm. To obtain the excitation spectra, the intensity obtained for a particular emission wavelength was plotted in front of the excitation wavelength in the excitation wavelength range from 120 to 248 nm.
Lifetime measurements were also carried out in the DESIRS beamline of SOLEIL Synchrotron, France (proposal number 20161324, standard) in a single bunch mode operation for pulsed radiation. The same configuration in the vacuum chamber as in previous measurements was used. The light was guided with an optical fibre to an ANDOR spectrograph (Shamrock 193i) with a grating of 150 lines·mm −1 , coupled to an iStar Intensified Charge Coupled Device (ICCD) (model DH734-18F-03). When the level of vacuum in the chamber was lower than 2 × 10 −5 bars, the window between this chamber and the synchrotron was removed in order to increase the photon flux reaching the sample.