Photoluminescence of CsMI3:Eu2+ (M = Mg, Ca, and Sr) a spectroscopic probe on structural distortions

The photoluminescence of Eu ions doped in iodidoperovskites, CsMI3 (M = Mg, Ca, Sr), was studied. Eu occupies the alkaline earth site in each of these compounds. The emission spectra of the doped perovskites are characterized by 4f5d 4f transitions of Eu located in the yellow range for CsMgI3 and the blue range for CsCaI3 and CsSrI3, respectively. All excitation spectra provide a well-resolved fine structure arising from different FJ (J = 0,. . .,6) levels of the 4f 6 core of the excited state configuration, indicating a weak exchange interaction with the 5d electron. The decay times and temperature dependence of the luminescence were also analyzed. Both CsMgI3:Eu 2+ and CsSrI3:Eu 2+ show a redshifted emission with respect to CsCaI3:Eu . This is explained by trigonal distortions of the [EuI6] 4 octahedra in the former two compounds. The energetic order of the resulting crystal field states is justified in terms of the angular overlap model of ligand field theory. This series of Eu doped earth alkaline iodidoperovskites may serve as a textbook example for the detailed understanding of the structure–luminescence relationships of Eu ions, which is very important for tailoring functional materials for respective applications.


Introduction
The luminescence of Eu 2+ has been a topic in many papers and reviews due to its broad-banded 4f-5d transitions that are parity allowed and thus interesting for many applications like, e.g. LEDs or detectors such as scintillators and X-ray storage phosphors. The nature of the 5d orbitals and their interaction with the local crystal field afford a systematic impact upon the emission color of the respective Eu 2+ -activated material. For example, in fluorides, such as CaF 2 and SrF 2 , Eu 2+ shows violet emission, 1 whereas in KMgF 3 , the 5d states are located at higher energies such that 4f-4f line emission from the excited 6 P J and 6 I J states is also observed in the UV range. 2,3 On the other hand, in nitridosilicates such as M 3 Si 6 O 12 N 2 :Eu 2+ (M = Sr, Ba), 4 M 2 Si 5 N 8 :Eu 2+ (M = Ca, Sr, Ba) 5 or MSi 2 O 2 N 2 :Eu 2+ (M = Ca, Sr), 6,7 Eu 2+ shows yellow to orange luminescence. Further examples include Eu 2+ -activated CaS or SrS, showing a red emission. 8 The examples mentioned above show how the luminescence of Eu 2+ can be affected only by varying the composition of the host material. Interesting reviews that summarize and discuss the data of many of the Eu 2+ -activated compounds presented in the literature were published by Rubio 9 and Dorenbos 10 during the last two decades.
Both for basic research and applications in high-energy radiation detection, halides are interesting host materials for Eu 2+ . As already described above, many of the studies presented in the literature have been attempted for fluorides, and chlorides have already been subject to many investigations. Unlike Eu 2+ -activated fluorides, however, the heavier halides only exhibit broad-banded 5d-4f emission in the violet to blue range due to the nephelauxetic effect shifting the position of the barycenter of the 4f 6 5d 1 states below the lowest excited 4f 7 state. Moreover, crystal field splitting shifts the lowest 4f 6 5d 1 state yet to lower energies, as was indicated by Blasse. 11 Typical examples from the literature are MZnCl 4 :Eu 2+ (M = Sr, Ba), which emit at 402 nm and 396 nm, respectively. 12 Similar to the results presented in this paper, Gahane et al. presented the optical properties of Eu 2+ ions doped in the ternary chlorides CsMCl 3 (M = Mg, Ca, Sr). 13 Many other chlorides are presented in ref. 10 and references given therein.
The number of reported Eu 2+ -activated bromides, however, is already very small and mainly includes detailed investigations of the binary alkali and alkaline earth halides. 10 More recently, however, several ternary bromides and iodides have attracted a lot of attention due to their potential application as scintillating materials. Typical examples for bromides as scintillators include CsCaBr 3 :Eu 2+ 14 and CsSrBr 3 :Eu 2+ . 15,16 A detailed analysis of the photoluminescence properties of Eu 2+ in the series CsMBr 3 (M = Mg, Ca, Sr) has been recently published by us. 17 Their range of applicability of these materials is, however, limited due to their high sensitivity against moisture.
In contrast to the rest of the halides, the luminescence of Eu 2+ doped in ternary iodides is currently only described in a superficial manner in the literature. 10 Recently, the applicability of CsMI 3 (M = Ca, Sr) as scintillators has been investigated and presented by the group of Melcher. [18][19][20][21] Due to the higher covalence of Eu-I bonds compared to other halides, the 5d-4f emission is described to shift to lower energies and thus located in the blue or even bluish-green region. 10 Since the iodides described in this paper mainly contain heavy ions, their applicability as scintillating materials seems absolutely plausible, as the previously mentioned publications also nicely illustrate. The higher doping concentrations of Eu 2+ required for applications result in a loss of resolution of the spectra and thus do not allow a detailed analysis of the fine structure. These features are, however, important for a basic understanding of the structure-luminescence relationship. Another aim of this paper is to demonstrate the photoluminescence of Eu 2+ doped in CsMgI 3 , which has not been presented so far in the literature.
To the best of our knowledge, no detailed analysis of the Eu 2+ -activated ternary iodides CsMI 3 (M = Mg, Ca, Sr) with emphasis on the systematics of the photoluminescence properties in correlation with the structures and chemical compositions of these compounds has been attempted before. Moreover, it seems interesting to us to compare the properties of Eu 2+ in iodides with those of the bromides of similar composition. Such an analysis is very important also from an application perspective as it allows the predictability and tunability of the photoluminescence properties of Eu 2+ by distinct changes in the size of the site the Eu 2+ ions occupy as well as in the structural features of a compound. This is not only interesting for scintillating materials but also very important for other applications such as in LED technology.

Preparation
CsI (Merck, 99.5%) was dried at 200 1C in a dynamic vacuum in order to remove any traces of moisture. The anhydrous iodides MI 2 (M = Ca, Sr) could be obtained by first dissolving the respective hydrates CaI 2 Á4H 2 O (Acros Organics, 99%) and SrI 2 Á xH 2 O (Heraeus, 99.5%) in concentrated HI (Acros Organics, 57 wt%) to remove traces of MCO 3 (M = Ca, Sr) that may form upon standing of the hydrates in air. The thus obtained pure hydrates of the iodides were then dried in a dynamic vacuum at 200 1C to obtain pure anhydrous iodides. EuI 2 was prepared from Eu 2 O 3 (smart elements, 99.995%) using the ammonium iodide route in order to avoid any traces of moisture and oxygen. 22 After slow decomposition at a heating rate of 4 1C h À1 of the thus prepared ternary compound (NH 4 ) 3 EuI 6 ÁxH 2 O to 300 1C in a dynamic vacuum, EuI 2 was directly obtained as a slightly yellowish powder. The phase purity of all starting materials was checked by X-ray powder diffraction (Siemens D5000, Cu K a radiation).
The doped ternary iodides CsMI 3 :Eu 2+ (M = Ca, Sr) were prepared by fusing the respective starting materials CsI and MI 2 (M = Ca, Sr) at their corresponding molar ratios together with 0.1 mol% EuI 2 in sealed silica ampoules under vacuum using the Bridgman technique. For CsMgI 3 :Eu 2+ , CsI 3 (Alfa Aesar, 98%), Mg metals (smart elements, 99.8%) and 0.1 mol% EuI 2 were used as starting materials instead. All mixtures were heated at a rate of 20 1C h À1 to their respective melting points and kept in the melt for 36 h. The mixtures were then slowly recrystallized (2 1C h À1 ) to 100 1C below their melting points Fig. 1 Measured X-ray powder diffraction patterns of (a) CsMgI 3 :0.1% Eu 2+ , (b) CsCaI 3 :0.1% Eu 2+ and (c) CsSrI 3 :0.1% Eu 2+ in comparison to the respective patterns according to the literature. 23,24 The marked references denote the in situ formation of hydrates during the XRD measurements. and finally cooled to room temperature at 10 1C h À1 . The phase purity and crystallinity of the products were also verified by X-ray powder diffraction (see Fig. 1). Due to the hygroscopic nature and sensitivity to oxygen of all the previously described compounds, every step had to be performed under a dry Ar atmosphere using a glove box (Braun).

Optical measurements
Photoluminescence emission and excitation spectra were obtained using a Fluorolog3 spectrofluorometer (Jobin Yvon) equipped with a 450 W xenon lamp and double Czerny-Turner monochromators allowing a resolution down to 0.05 nm and a photomultiplier detection system R928P (Hamamatsu) coupled with a photon counting system. A liquid helium closed-cycle cryostat was used to perform temperature-dependent measurements at 10 K. The emission spectra were corrected for photomultiplier sensitivity and the excitation spectra were corrected for lamp intensity. For decay time measurements, a frequency-tripled Nd:YAG laser (Spectra Physics Quanta Ray Indi 40) with a repetition rate of 10 Hz and a pulse width of 8 ns was used (l ex = 355 nm). For the detection of the desired emission wavelength, a single monochromator (Jobin Yvon) was employed that allows a resolution down to 5 nm. The decay signal was recorded using a photomultiplier tube that was attached to an oscilloscope (Tektronix, 500 kHz). The crystal structures of all iodides presented in this paper can be derived from the perovskite structure and have already been described in the literature. 23,24 The measured powder diffraction patterns are depicted in Fig. 1 and indicate the phase purity of the synthesized compounds. Due to the fact that the XRD measurements could not be performed without the total exclusion of air, some other reflections reveal an increase in intensity after every subsequent measurement on the same samples. These reflections can be assigned to the in situ formation of hydrates of the respective iodides due to their hygroscopic nature in air. Moreover, all XRD patterns reveal the presence of broad reflections in the range between 2y = 151 and 221, which is an artifact due to the fixation of the powders with silicon grease. No effects of doping with EuI 2 can be detected as it is expected due to the very low concentration of only 0.1 mol% that is below the sensitivity limit of the XRD device. Hence, no change in the crystal structures of the three iodides is induced by doping of EuI 2 .

Crystal structures and coordination spheres of
The symmetry data and average M-I distances (M = Mg, Ca, Sr) important for the discussion of the luminescence spectra are summarized in Table 1.
CsMgI 3 crystallizes in the CsNiCl 3 structure type, which is a hexagonally distorted variation of the perovskite type structure with the space group P6 3 /mmc (no. 194). 23 The average Mg-I distance is 2.90 Å and the structure is depicted in Fig. 2. The Cs + ions are 12-fold coordinated by I À ions in the form of anticuboctahedra whereas the Mg 2+ ions are six-fold coordinated in the form of octahedra. The local site symmetry of D 3d (see Table 1), however, already indicates a severe trigonal distortion. This is induced by the face-sharing pattern of the [MgI 6 ] 4À octahedra that form linear chains along the c axis and lead to a high anisotropy of the crystal structure. The effect of anisotropy on the photoluminescence properties of Eu 2+ has already been discussed in the isostructural compound CsCdBr 3 . 25,26 CsCaI 3 crystallizes in the orthorhombic GdFeO 3 structure type with the space group Pnma (no. 62) and is isostructural to CsSrBr 3 24 (see Fig. 3). The average Ca-I distance is 3.10 Å. The Cs + ions are (8+2)-fold coordinated by I À ions and the Ca 2+ ions are six-fold coordinated by I À ions in the form of tilted octahedra along the c axis with local C i symmetry (see Table 1). However, the deviation from octahedral symmetry can be assumed to be small, which is a reasonable basis for our interpretation of the luminescence spectra (see Section 4.2). CsSrI 3 crystallizes in a filled PuBr 3 structure type, which is characterized by an orthorhombic crystal structure with the  24 and is depicted in Fig. 4. In this structure, the Cs + ions are only 8-fold coordinated by I À ions in the form of two-fold capped trigonal prisms. The Sr 2+ ions are six-fold coordinated in the form of tilted octahedra along the c axis with local C 2h symmetry (see Table 1). The average Sr-I distance is 3.37 Å in this compound. With regard to different structures, a remarkable development can be noted as the distortion of the [MI 6 ] 4À octahedra (M = Mg, Ca, Sr) increases in the series CsCaI 3 o CsSrI 3 o CsMgI 3 and ends in a face-sharing pattern of the octahedra in CsMgI 3 . Along with this development, the deviation from a perfect octahedral environment for Eu 2+ in these three compounds also increases and trigonal distortion has to be taken more into account. We will review this aspect in Section 4. The Eu 2+ -activated iodides presented in this paper show a bright luminescence upon UV excitation at room temperature, which is the blue region for CsCaI 3 :Eu 2+ and CsSrI 3 :Eu 2+ and greenishyellow in the case of CsMgI 3 :Eu 2+ . For all three materials, the transitions can be interpreted with the fact that the Eu 2+ ions occupy exclusively the respective alkaline earth sites. For the respective Eu 2+ -activated bromides CsMBr 3 (M = Mg, Ca, and Sr), 17 we have introduced several energetic values based on the notation and definitions of Dorenbos that afford a more detailed picture of the electronic states of Eu 2+ in the respective compounds. 10,27,28 We will also use this notation in this paper.

CsMgI 3
CsMgI 3 :0.1% Eu 2+ shows a bright yellow luminescence upon UV excitation at room temperature. Low temperature luminescence spectra as well as a photograph of the luminescent sample at room temperature are shown in Fig. 5. To the best of our knowledge, no luminescence properties of Eu 2+ -activated CsMgI 3 have been reported so far. The emission spectrum at 10 K is characterized by a broad asymmetric Gaussian band with a maximum localized at 17 330 cm À1 (577 nm). The full width at half maximum of G 10K = 2430 cm À1 is relatively large, indicating a strong electron-phonon coupling. The emission band can be assigned to 4f 6 5d 1 (t 2g ) -4f 7 ( 8 S 7/2 ) transitions of Eu 2+ substituting for Mg 2+ ions assuming O h site symmetry for a first approximation. To a certain extent, the asymmetry of the band can be understood from the large difference in the ionic radii between Eu 2+ and Mg 2+ (r(Eu 2+ ) = 1.17 Å and r(Mg 2+ ) = 0.72 Å). 29 Although the occupation of the Cs + site may also be plausible (r(Cs + ) = 1.88 Å), 29 it can be excluded due to the fact that the coordination number is much higher. This would lead to a highly blue-shifted emission and is well established in the scintillator CsI:Eu 2+ , where the coordination number of the Cs + sites is 8 and the respective Eu 2+ emission is already located at 22 470 cm À1 (445 nm). 30 Moreover, we performed EXAFS measurements on the Eu 2+ -doped isostructural compounds CsMgCl 3 and CsMgBr 3 , which we will publish in another paper. The measurements definitely prove the sole occupation of the Mg 2+ sites at concentrations below 3%.
The assumption of octahedral site symmetry for the Eu 2+ ions serves well for a crude interpretation of the luminescence spectra. However, as we will show below, for a more detailed interpretation of the electronic properties of Eu 2+ in this compound, this approximation is indeed too simplified.
The excitation spectrum of CsMgI 3 :0.1% Eu 2+ at 10 K, as shown in Fig. 5, is characterized by two broad bands that contain a detailed fine structure. Their presence is easily understood from the splitting of the 5d orbitals in an octahedral crystal field that gives rise to a ''t 2g '' and an e g state, the latter being located at higher energy. We marked the t 2g state with quotation marks since the assumption of O h symmetry for Eu 2+ is too simplified in this case, as we will explain below. The fine structure arises due to the remaining 4f 6 core in the excited state configuration of Eu 2+ and can be identified with the different 7 F J states (J = 0,. . .,6). Interestingly, the fine structure is also observable in the e g excitation band. This observation has so far been only reported for the binary alkali halides 31 and Eu 2+ -activated CsMBr 3 , 17 but not in the case of other host lattices. It indicates a weak exchange  interaction between the 4f electrons and the 5d electron in the excited state 32 and becomes much less resolved at room temperature. The positions of the different 7 F J states are compiled in Table 2 and their position relative to the 7 F 0 state fits well with the values anticipated from Eu 3+ . 33 The good resolution of the different 7 F J states allows a relatively precise determination of the Stokes shift DS(2+,CsMgI 3 ) = 3870 cm À1 .
A detailed observation on the excitation spectrum of the Eu 2+related emission depicted in Fig. 5 reveals the presence of another band located at 27 770 cm À1 (360 nm). This remarkable feature has already been observed in Eu 2+ -activated isostructural compounds such as CsCdBr 3 25,26 and CsMgBr 3 . 17 As we have already indicated in Section 3.1, this can be understood if trigonal distortion of O h to D 3d symmetry (see Table 1) is taken into account leading to a splitting of the anticipated ''t 2g '' state into an a 1g and an e g state. Due to that reason, we will mark the anticipated octahedral field state ''t 2g '' with quotation marks throughout this paper. More conclusions from this effect, e.g. the given energetic order of the a 1g and the e g state, will be discussed in Section 4.2. The energetic difference between the a 1g and the e g state is the trigonal crystal field splitting, e trig (7,2+,CsMgI 3 ) = 4310 cm À1 .

CsCaI 3
CsCaI 3 :0.1% Eu 2+ is characterized by strong blue luminescence upon UV irradiation at room temperature. Fig. 6 depicts the low temperature luminescence spectra as well as a photograph of the luminescent sample at room temperature. It has been recently reported that Eu 2+ -activated CsCaI 3 is a promising scintillator with a light output comparable to NaI:Tl + . 18,19 The emission shows a slightly asymmetric Gaussian band peaking at 22 370 cm À1 (447 nm), which is in very good agreement with the value reported in the literature (450 nm). 18 The full width at half maximum amounts to G 10K = 640 cm À1 . The emission can be clearly identified by the 4f 6 5d 1 (t 2g ) -4f 7 ( 8 S 7/2 ) transition of Eu 2+ substituting for Ca 2+ ions under the assumption of an approximate octahedral site symmetry. The difference in ionic radii (r (Eu 2+ ) = 1.17 Å and r (Ca 2+ ) = 1.00 Å) 29 explains the observed asymmetry of the Gaussian band. A distinct fine structure is observed in the excitation spectra arising due to the different 7 F J states from the 4f 6 core. As there are only two broad bands observed in the excitation spectrum, the assumption of an octahedral environment for Eu 2+ seems legitimate, giving rise to a t 2g and an e g state. Similar to CsMgI 3 :Eu 2+ (see Fig. 5), the fine structure is also observable in the e g state. This indicates, again, a weak exchange interaction between the 4f 6 electrons and the 5d electron. 32 The positions of the different 7 F J levels are compiled in Table 3. Within the experimental error, they are in good agreement with the values expected from the respective energies known from Eu 3+ . 33 Due to the well-resolved fine structure, it is also possible to determine the Stokes shift accurately, which is DS(2+,CsCaI 3 ) = 1310 cm À1 . The octahedral crystal field splitting, 10 Dq, can be also determined and is given by e cfs (7,2+,CsCaI 3 ) = 11 050 cm À1 using the nomenclature used by Dorenbos. 10

CsSrI 3
Similar to Eu 2+ -activated CsCaI 3 , CsSrI 3 :0.1% Eu 2+ shows blue luminescence upon excitation with UV light at room temperature. The luminescence spectra at 10 K and a photograph of the luminescent sample are shown in Fig. 7. Similar to CsCaI 3 , Eu 2+activated CsSrI 3 also shows efficient radioluminescence and is thus a promising candidate for scintillation materials. 20,21 Table 2 Positions of the 4f 6 ( 7 F J )5d 1 states in CsMgI 3 :0.1% Eu 2+ and energy differences DE J-0 of the 4f 6 ( 7 F J )5d 1 (J = 1-6) states relative to the 4f 6 ( 7 F 0 )5d 1 state 4f 6 ( 7 F J )5d 1 (t 2g ) 4 f 6 ( 7 F J )5d 1 (e g ) showing blue luminescence upon UV irradiation at room temperature. Table 3 Positions of the 4f 6 ( 7 F J )5d 1 states in CsCaI 3 :0.1% Eu 2+ and energy differences DE J-0 of the 4f 6 ( 7 F J )5d 1 (J = 1-6) states relative to the 4f 6 ( 7 F 0 )5d 1 state 4f 6 ( 7 F J )5d 1 (t 2g ) 4 f 6 ( 7 F J )5d 1 (e g ) The emission band arising from the 4f 6 5d 1 -4f 7 ( 8 S 7/2 ) transition is an almost perfect Gaussian band with a maximum at 22 010 cm À1 (454 nm). The value is in very good agreement with the value found in the literature (452 nm). 20 The full width at half maximum is slightly larger than that in the case of CsCaI 3 and given by G 10K = 710 cm À1 . However, the slight asymmetry cannot be explained by a mismatch between the ionic radii of Eu 2+ and Sr 2+ in this case, as they are almost equal (r(Eu 2+ ) = 1.17 Å, r(Sr 2+ ) = 1.18 Å). 29 Moreover, it turns out that the emission is slightly redshifted to the 5d-4f emission of Eu 2+ in CsCaI 3 , which does not correspond to the expectation when considering the decrease of the ionic radii from Ca 2+ to Sr 2+ and in turn the weaker crystal field splitting. As for the other iodides, the fine structure due to the different 7 F J levels is also observed in CsSrI 3 in all excitation bands (see Table 4 for their positions). The agreement with the values reported for Eu 3+ is also satisfactory. 33 The Stokes shift determined thereof reads DS(2+,CsSrI 3 ) = 1090 cm À1 . However, similar to the case of CsMgI 3 :0.1% Eu 2+ (cf. Fig. 5), another isolated band can be observed at 29 320 cm À1 (341 nm). This indicates a similarity of the site symmetry for the Eu 2+ ions in CsSrI 3 compared to CsMgI 3 . The larger disposal of the [SrI 6 ] À octahedra with respect to CsCaI 3 (see Fig. 3 and 4) thus seems to induce a trigonal distortion resulting in the sub site symmetry D 3d to first order. The subsequent symmetry distortion to the real site symmetry C 2h is assumed to be very weak and thus will be neglected for our considerations. With the assumption of D 3d as the site symmetry for the Eu 2+ ions in CsSrI 3 , the appearance of another isolated band in the excitation spectrum (see Fig. 7) seems plausible, leading to the splitting of the anticipated ''t 2g '' state into an a 1g and an e g state. The excited e g state retains its degeneracy. We will justify the order of the states a 1g and e g shown in Fig. 7 in Section 4.2. From this interpretation, it is possible to deduce the trigonal crystal field splitting in CsSrI 3 given by e trig (7,2+,CsSrI 3 ) = 3960 cm À1 .

Temperature-dependent emission
The temperature dependence of the 5d-4f emission of Eu 2+ in the series CsMI 3 (M = Mg, Ca, Sr) was also investigated. The respective emission spectra are depicted in Fig. 8-10. In all three compounds, Eu 2+ shows an intense luminescence even at room temperature and a slight blueshift with increasing temperature is observed. In the case of CsMgI 3 :0.1% Eu 2+ (see Fig. 8), an asymmetric wing of the emission on the red side of the spectrum evolves with increasing temperature. All these observations agree with the behavior we have already reported for the respective bromides. 17 In Fig. 11, the evolution of the integrated intensities with increasing temperature is depicted. For all three compounds the photoluminescence signal is slightly quenched with higher temperatures. In fact, the integrated intensities of the emission of Eu 2+ in CsCaI 3 and CsSrI 3 decrease to only 88% and 85% of the initial value when heated to 300 K, whereas in the case of CsMgI 3 :0.1% Eu 2+ , it decreases to 71%. This indicates that the quenching temperature T 1/2 , at which the integrated intensity Fig. 7 Emission spectrum (cyan line, E ex = 25 000 cm À1 ) and excitation spectrum (black line, E em = 21 980 cm À1 ) of CsSrI 3 :0.1% Eu 2+ at 10 K. Note the splitting of the anticipated ''t 2g '' state into the trigonal field states a 1g and e g . Inset: CsSrI 3 :0.1% Eu 2+ showing blue luminescence upon UV irradiation at room temperature. Table 4 Positions of the 4f 6 ( 7 F J )5d 1 states in CsSrI 3 :Eu 2+ (0.1%) and energy differences DE J-0 of the 4f 6 ( 7 F J )5d 1 (J = 1-6) states relative to the 4f 6 ( 7 F 0 )5d 1 state 4f 6 ( 7 F J )5d 1 (''t 2g '') 4f 6 ( 7 F J )5d 1 (e g ) has reached 50% of the maximum signal, is well above 300 K in all three doped compounds. Unfortunately, we could not perform comparable measurements above 300 K so we are only able to state this about the quenching temperature. In fact, the low Stokes shifts for Eu 2+ in these compounds and very low phonon energies in iodides (below 200 cm À1 ) 34 make non-radiative transitions very inefficient and a weak electronphonon coupling should be expected. This behavior is clearly observed in Fig. 11.

Decay times
The decay curves of the luminescence signal of Eu 2+ in CsMI 3 (M = Mg, Ca, Sr) are depicted in Fig. 12. They were recorded upon excitation into the ''t 2g '' states (l ex = 355 nm) at room temperature. All decay curves show a mono-exponential behavior in the recorded time range. Since the luminescence is also not highly quenched at room temperature (see Section 3.3), it may also be assumed that the shortening of the decay time of Eu 2+ by phonons is relatively small and, thus, a good estimate of the radiative lifetime is obtained by the room temperature measurements. Their values are compiled in Table 5.
From the decay times, it is possible to obtain direct information about the electric dipole matrix element m 5d-4f , since where e is the elementary charge. The matrix element h5d|r|4fi represents the radial integral between these two states and is related to the decay time by 35 where A denotes the total transition probability per unit time, E em (7,2+,A) is the emission energy of Eu 2+ in compound A in    cm À1 and w is a correction factor that includes the refraction index n of the host medium, For CsCaI 3 , a refraction index of 2.0 has been deduced from band structure calculations, 19 whereas for CsSrI 3 , a refraction index of 1.79 was calculated. 36 Due to the similarity in the structures of CsSrI 3 and CsMgI 3 , a refraction index of 1.79 was also assumed for CsMgI 3 . With these values, it is possible to estimate the radial integrals for the 5d-4f transition for Eu 2+ in the iodides CsMI 3 (M = Mg, Ca, Sr), as summarized in Table 5.

Photoluminescence of Eu 2+ doped in CsMI 3 (M = Mg, Ca, Sr)
Eu 2+ shows very intense luminescence in the series CsMI 3 (M = Mg, Ca, Sr), which goes from blue in CsCaI 3 and CsSrI 3 to yellow in CsMgI 3 . The important luminescence data obtained from the spectra presented in Section 3 are compiled in Table 6. As previously mentioned, all important equations to derive the respective values have already been described by us in another publication and can be found in the literature. 17 However, similar to the case of the respective bromides, 17 a fine structure due to the 7 F J states ( J = 0,. . .,6) arising from the 4f 6 core in the excited 4f 6 5d 1 state can be observed in the crystal field states ''t 2g '' and e g . This clearly indicates that the exchange interaction between the 5d electron and the other six 4f electrons is very small. 32 Other conditions of such observations are high site symmetries, good crystal quality, low dopant concentrations and low temperature measurements. 27 Examples, in which Eu 2+ is also known to exhibit well-resolved excitation spectra with such a fine structure, include Sr 3 (PO 4 ) 2 , 37 Ba 2 B 5 O 9 Br 38 and binary alkali halides. 31 As the fine structure seems to be very well resolved even in the e g state, which was so far only known for the alkali bromides and iodides, 31 we also presume that covalence of the bond between Eu 2+ and the respective (halide) anions has an important effect upon the strength of the exchange interaction. This makes sense since a higher covalence exhibits a larger nephelauxetic effect and thus leads to a larger energy gap between the excited 4f states and 5d states inhibiting an admixture. Therefore, the exchange interaction will be small as well as the radiative lifetimes will be lowered. Both effects are indeed observed for the system CsMI 3 (M = Mg, Ca, Sr). Since the covalence of the Eu-I bonds in these compounds should not vary very much, it may be assumed that the bond length and, thus, the size of the alkaline earth site are the most important factors in this case which influence the luminescence properties. As we will show below, this approximation is, however, too crude for CsMgI 3 and CsSrI 3 .
We also attempted to analyze the emission energies E em (7,2+,A) and the positions of the 7 F 0 levels within the ''t 2g '' and e g states, denoted as E t 2g ex 7; 2þ; 7 F 0 ; A À Á and E eg ex 7; 2þ; 7 F 0 ; A À Á , respectively, with respect to the difference in the ionic radii between Eu 2+ and the respective alkaline earth ion M 2+ (M = Mg, Ca, Sr). The plots are illustrated in Fig. 13.
The curves shown in Fig. 13 nicely illustrate the nature of the different states and indicate the impact of the crystal structure upon the photoluminescence properties of Eu 2+ in iodidoperovskites. Since the behavior of the emission energies and the excitation energies of the ''t 2g '' states is strikingly similar, it can be directly concluded that emission occurs from the lowest excited hypothetical ''t 2g '' state. It turns out for Eu 2+ -activated CsMgI 3 and CsSrI 3 that both the energy of the ''t 2g '' states and the emission energy are lower than those in the case of CsCaI 3 . The energies of the e g state, however, remain relatively constant with respect to changes in the size of the alkaline earth site. This observation does not agree with the naive expectation that it should increase with decreasing Eu-I bond length due to the anti-bonding nature of the e g state as it was observed in the respective bromidoperovskites. 17 This behavior can, however, be understood taking into account the large bond length between Eu 2+ and I À , which would be 3.37 Å according to the sum of the Shannon radii at 6-fold coordination. 29 With the anti-bonding nature of the e g this value should yet be larger. Therefore, the repulsive interaction between the 5d electron of Eu 2+ and the 5p electrons of I À ions is negligibly small giving rise to the relatively constant behavior.
A more detailed view in Fig. 13 already reveals the special role that CsCaI 3 plays within this series. In fact, the emission energy and the energy of the 7 F 0 levels within the t 2g state go through a maximum, whereas the respective energy of the 7 F 0 levels within the e g state goes through a minimum in that case. We emphasize again to keep in mind that in the case of CsCaI 3 , it is valid to speak of a t 2g state. The special dependence of the energies upon the size of the alkaline earth site can be directly correlated with the structures of the iodides. As indicated in Section 3.1, the tendency of tilting of the [EuI 6 ] 4À octahedra increases from CsCaI 3 over CsSrI 3 to CsMgI 3 . In the last compound, the octahedra are face-sharing (see Fig. 2). This goes along with an increasing trigonal distortion of the [EuI 6 ] 4À octahedra to D 3d symmetry and the fact that the Eu 2+ ions experience stronger repulsive forces due to the alkaline earth ions in the second coordination sphere. This is not the case for CsCaI 3 . The subsequent symmetry reduction to the real site symmetry C 2h in CsSrI 3 is assumed to be very weak and will be neglected for the following considerations. The t 2g state splits into an a 1g and an e g state in D 3d symmetry. Therefore, the lowest excited state and the emission energies of Eu 2+ in CsMgI 3 and CsSrI 3 should be lower than those in the case of CsCaI 3 . These aspects are well reproduced by the spectra (see Fig. 13) and are, in fact, the reason for the strange behavior of CsSrI 3 :Eu 2+ showing a slightly redshifted emission with respect to CsCaI 3 :Eu 2+ . In the latter Eu 2+ -activated compound, the effect of trigonal distortion is negligibly small, as we will also show in the following section.
The discussion above indicates that the size of the alkaline earth site is not the only important parameter of the luminescence properties of Eu 2+ in iodidoperovskites. Especially in the case of CsMgI 3 :Eu 2+ , the emission is extraordinarily redshifted. We have reported a similar behavior for CsMgBr 3 :Eu 2+ . 17 Both compounds exhibit a high anisotropy and a pseudo-onedimensional characteristic along the hexagonal c axis. Together with the small size of the Mg 2+ site inducing a large crystal field splitting, the anisotropy synergetically leads to a large redshift. It has been discussed in the literature that the structural feature of linear chains induces a preferential orientation of one of the 5d orbitals into the direction of the chain. 37,39 Due to the similar behavior of CsSrI 3 :Eu 2+ with respect to CsMgI 3 :Eu 2+ and the observed splitting in the excitation spectra, we assign the extraordinary redshift for CsMgI 3 :Eu 2+ to the effect of the trigonal distortion induced by the tilts of the [EuI 6 ] 4À octahedra rather than to a preferential orientation of the 5d orbitals. In general, this is an important issue for applications especially in the LED technology, where it is desired to have green-to redemitting phosphors excitable in the blue range. Therefore, not only the size of the alkaline earth site can help here but also a larger anisotropy and with that, a larger distortion.
Based on this interpretation, it is possible to deduce the position of the undistorted ''t 2g '' state and, thus, the octahedral crystal field splitting, e undist. cfs (7,2+,A) = 10 Dq, without distortion. Since the energy of the barycenter of the undistorted ''t 2g '' state is conserved, the resulting a 1g state is lowered by 2/3 e trig (7,2+,A), whereas the resulting e g state is increased by 1/3 e trig (7,2+,A). Assuming that the distortion of the initial e g state in O h symmetry is negligible, the octahedral crystal field splitting for the undistorted case can be derived as follows: where E eg ex ð7; 2þ; AÞ is the average energy of the excitation band assigned to the e g state in O h symmetry and E trig a 1g ð7; 2þ; AÞ and E trig eg ð7; 2þ; AÞ are the average energies of the respective trigonal states as found in the excitation spectra of the compounds. The values for the octahedral crystal field splitting without distortion are compiled in Table 7.
As expected, the values in Table 7 increase with decreasing size of the alkaline earth site, thus inducing a smaller Eu-I bond length. Upon comparison with the 10 Dq values known from the respective bromidoperovskites, 17 the derived values for the iodides are all smaller. This is in agreement with theoretical considerations since I À is a larger ion and more polarizable and, thus, a softer ligand. It should be noted, however, that for CsMgBr 3 :Eu 2+ , the crystal field splitting was directly derived without considering trigonal distortion. 17 Therefore, the corrected value for 10 Dq in the case of CsMgBr 3 :Eu 2+ will be smaller (13 230 cm À1 ). It is, however still larger than the value of the respective iodide. The crystal field splitting derived using eqn (4) and (5) is in the range known from alkali halides 31 making our interpretation plausible.

Energetic order of trigonal crystal field states in CsMgI 3 :Eu 2+ and CsSrI 3 :Eu 2+
Throughout this paper, we argued that the trigonal distortion leads to a splitting of the t 2g state into an a 1g and an e g state with the latter being higher in energy. In this section, we want to justify this statement using the angular overlap model (AOM) known from ligand field theory. A well-written review about this approach has been published by Schäffer 40 and the reader can refer to it for the details of this approach. Since the 5d orbitals are much more diffuse and more involved in metal-ligand bonding than the 4f orbitals of Eu 2+ , the latter were neglected in the AOM approximation. In the case of CsMgI 3 :Eu 2+ and CsSrI 3 :Eu 2+ , the trigonal distortion of the [EuI 6 ] 4À octahedra evolves along the C 3 axis, thus leading to the disposal of the ligands with respect to the normal z axis by a small polar angle Y (see Fig. 14). The strong tilting pattern of the respective octahedra in the structures of CsMgI 3 and CsSrI 3 (see Fig. 2 and 4) induces a stronger repulsion between the ligands, which makes the trigonal distortion also plausible from an electrostatic perspective. For the AOM approximation, it is assumed that the Eu-I bond lengths do not change during the trigonal distortion, i.e. only the angular part of the overlap integral is important in this consideration. If the Eu 2+ ion is placed into the origin of the coordinate system, the positions of the ligands can be parametrized by the disposal Y of the Eu-I bond with respect to the z axis and their azimuthal angle F relative to the x axis. The numbering of the ligands as denoted in Fig. 14 leads to the parametrization shown in Table 8.
The distortion angle Y is not known quantitatively, thus, it will be denoted by a small angle y. With the parametrization in Table 8 the factors by which the angular overlap integrals are diminished from the maximum value can be deduced in the coordination frame shown in Fig. 14. They arise due to the fact that for every spherically symmetric ligand, the coordination frame has to be rotated such that the 5d orbitals have maximum overlap with the ligand orbitals. This procedure leads to a 5 Â 5 transformation matrix that has to be evaluated for every ligand separately. Its general form has been deduced by Schäffer. 41 For the calculation in the case of CsMgI 3 and CsSrI 3 , some other assumptions have still to be taken into account. Within the AOM approximation, only sand p-like interactions are considered and parametrized by the interaction terms e s and e p . Moreover, for small angles y, the functions sin y and cos y can be approximated very well by their Taylor expansions up to second order. With these assumptions, the energies of the 5d xz , 5d xy and 5d yz orbitals transforming as t 2g states in a perfect octahedral environment can be evaluated during the trigonal distortion: e yz = 4e p + y 2 (9e s À 11e p ) e xy = 4e p À 2e p y 2 In accordance with the group theoretical expectation, 5d xz and 5d yz degenerate forming the e g state, whereas the 5d xy orbital transforms as an a 1g state in D 3d symmetry and affords a different energy in the AOM. It should be noted that no assumptions about the symmetry and degeneracy of states have been included into this calculation, i.e. the results of the AOM are totally independent of the group theoretical considerations. The energy splitting between the a 1g and the e g state is given by e AOM trig (7,2+) = 9y 2 (e s À e p ) Since I À ions are weak p donors, it may be assumed that e s c e p . Thus, eqn (9) affords a positive result indicating that the a 1g state is lower in energy than the e g state. However, eqn (9) allows another conclusion as well. As the degree of tilt of the [MI 6 ] 4À octahedra (M = Mg, Ca, and Sr) increases in the series CsCaI 3 o CsSrI 3 o CsMgI 3 , the repulsion between the I À ligands also increases, and so does y. In the case of CsCaI 3 :Eu 2+ , y is negligibly small and therefore, no splitting due to trigonal distortion is observed in the excitation spectra (see Fig. 9). The 5d xz , 5d xy and 5d yz orbitals degenerate in that case forming the t 2g manifold with an energy of 4e p . This agrees with the well-known result from ligand field theory. 40 In the cases of CsMgI 3 :Eu 2+ and CsSrI 3 :Eu 2+ , however, the angle y has to be taken into account. The [MgI 6 ] 4À octahedra in CsMgI 3 are tilted to a much larger degree than the respective octahedra in CsSrI 3 . Therefore, the splitting due to the trigonal distortion should be lower in the latter compound, which is indeed observed (see Table 9). The behavior of the curves shown in Fig. 13 is also well understood in terms of the AOM. When regarding eqn (6)- (8), it turns out that the energies of the hypothetical ''t 2g '' state, E t 2g ex 7; 2þ; 7 F 0 ; A À Á , should decrease with increasing y. This is exactly the observed behavior. CsCaI 3 :Eu 2+ provides the limiting case y -0, thus leading to a maximum for E t 2g ex 7; 2þ; 7 F 0 ; A À Á in this series.    Fig. 15 finally represents a summarizing schematic overview of the evolution of the 5d states of Eu 2+ in CsMI 3 (M = Mg, Ca, and Sr). It shows the irregular behavior of the emission energies illustratively. The centroid shift, e c (7,2+,A), should only be regarded qualitatively and was intentionally not determined for iodidoperovskites. Since a strong degree of tilt for the [MI 6 ] 4À octahedra is not reasonable in terms of Pauling's rules 42 due to the larger repulsion, this effect indicates a larger covalence of the M-I bond and thus, the Eu-I bond. Moreover, the effect of trigonal distortion has to be included into the formula such that a simple approach as attempted for the bromidoperovskites 17 does not seem very reasonable. We will discuss the effect of higher covalence of the Eu-I bond in CsMgI 3 :Eu 2+ and CsSrI 3 :Eu 2+ in terms of the decay times in Section 4.4.

Temperature-dependent emission
All Eu 2+ -activated iodidoperovskites CsMI 3 (M = Mg, Ca, and Sr) show efficient luminescence even at room temperature. As we have already indicated in Section 3.3, the quenching temperature T 1/2 , at which the integrated intensity has decreased to 50% of its initial value at 10 K, is well above 300 K. In fact, the intensities in CsCaI 3 :Eu 2+ and CsSrI 3 :Eu 2+ at room temperature only decrease to 88% and 85%, respectively, compared to their 10 K values. Together with the low Stokes shifts, this indicates a weak electron-phonon coupling. This is predominantly due to the fact that the highest phonon energy in iodides is relatively low (below 200 cm À1 ). 34 The 5d-4f emission of Eu 2+ in CsCaI 3 and CsSrI 3 exhibits a blueshift with higher temperatures, which arises due to the thermal population of the higher 7 F J levels within the lowest excited 5d state.
Interestingly, the Stokes shift DS(2+,CsMgI 3 ) and the FWHM G at 10 K are both very large in the case of CsMgI 3 :Eu 2+ (see Table 6). Moreover, the emission is highly asymmetric. This behavior is already familiar from CsMgBr 3 :Eu 2+ . 17 As a possible explanation anomalous luminescence could be assumed, which arises if the 5d states are closely located at the bottom or inside of the conduction band of the host lattice. Anomalous luminescence of Eu 2+ is typically characterized by Stokes shifts in the range of 4000-10 000 cm À1 , large FWHMs (4000-6000 cm À1 ), shortened decay times and low quenching temperatures T 1/2 . 43 However, the large covalence of the Eu-I bond in CsMgI 3 will shift the 5d states to low energies. Due to that fact and the large crystal field splitting due to the small Mg 2+ site and the trigonal distortion, it does not seem very probable that anomalous luminescence occurs in this compound. The experimental results also contradict the presence of anomalous luminescence since the luminescence is still very efficient even at room temperature and a fine structure due to the 7 F J levels can even be observed in the high-energy excitation band assigned to the e g crystal field state (see Fig. 5 and 11). Thus, we exclude anomalous luminescence of Eu 2+ in CsMgI 3 . Nevertheless, the large Stokes shift and FWHM at 10 K indicate a large reorganization of the electron density in the excited state. Obviously, this is related to the huge difference between the ionic radii of Mg 2+ and Eu 2+ (DR = 0.45 Å 29 ) and the relatively short Mg-I bond length (see Table 1). Upon excitation into the diffuse 5d orbitals of Eu 2+ , the Eu-I bond length is expected to change drastically inducing the large Stokes shift and, subsequently, the large FWHM. A quantitative correlation between the Stokes shift and the size of the alkaline earth site is, however, not very reasonable since it is not known whether the Eu-I distance increases or decreases in the excited state. In general, an increase of the bond length is assumed due to the more antibonding nature of the excited state. However, as it has been shown for the Ce 3+ -activated elpasolite CsNaLuCl 6 by pressure-dependent luminescence measurements, the potential surface of the excited 5d state can also be located at lower metal-ligand distances than for the ground state. 44 For CsSrI 3 :Eu 2+ , however, the Stokes shift DS(2+,CsSrI 3 ) and the FWHM G at 10 K are very low (see Table 6), although the structure of the host compound is very similar to that of CsMgI 3 . Moreover, a lower Stokes shift of 1090 cm À1 and 1420 cm À1 , 17 respectively, is observed compared to CsSrBr 3 :Eu 2+ . In general, an increase of the Stokes shift is expected upon increase of the polarizability of the ligands. This trend is clearly observed when comparing CsCaBr 3 :Eu 2+ (DS(2+,CsCaBr 3 ) = 1080 cm À1 (ref. 17)) with CsCaI 3 :Eu 2+ (DS(2+,CsCaI 3 ) = 1310 cm À1 , see Table 6). Our explanation for the reverse effect in the Sr-based compounds relies on the fact that the average Sr-I bond length is relatively large (3.37 Å, see Table 1 24 ). Since Eu 2+ ions have approximately the same ionic radius as Sr 2+ ions, they will be readily incorporated into the structure without severe distortion effects. If the Eu-I bond length is therefore relatively large in the ground state, an excitation into the 5d states will not change the bond length drastically and vibrational relaxation upon reorganization of the electron density is highly reduced giving rise to a small Stokes shift. Calculations of the electron density distribution in the ground and the first excited state and the resulting bond lengths could give some further insight here.

Decay times and radial integrals
The measured decay times and deduced values of the radial integrals (see Table 5) of Eu 2+ in CsMI 3 (M = Mg, Ca, Sr) are in good agreement with the values reported in the literature. 35 Although they have been measured at room temperature, the low vibrational energy in iodides makes vibrational coupling rather inefficient such that the decay time is not significantly shortened with respect to temperatures as low as 10 K. Their dependence upon the size of the alkaline earth site is plotted in Fig. 16.
The behavior of the decay times of Eu 2+ with varying differences in ionic radii of Eu 2+ and the alkaline earth ion is very similar to the variation of the emission energies and thus, the energies E t 2g ex 7; 2þ; 7 F 0 ; A À Á indicate that the ''t 2g '' state is the emissive state. As has already been discussed above, Eu 2+activated CsSrI 3 and CsMgI 3 show an exceptional behavior also in the decay times and conclusively, also in the radial integrals. In fact, the decay times of Eu 2+ in these two compounds are very short compared to the values known from oxides. 35 Obviously, this is related to the larger disposal of the octahedra in both structures. Such disposal is related to a larger covalence of the Eu-I bond. Due to the nephelauxetic effect, this lowers the barycenter of the 5d states inhibiting an admixture with the excited 4f 7 states of Eu 2+ . Thus, the decay times of CsMI 3 :Eu 2+ (M = Mg, Sr) are lowered compared to the decay time of Eu 2+ in CsCaI 3 .
The decay times detected for CsCaI 3 :Eu 2+ and CsSrI 3 :Eu 2+ are shorter compared to the respective bromides, 17 in accordance with the expectations regarding the higher covalence and the stronger nephelauxetic effect in iodides. In the case of CsMgI 3 :Eu 2+ , the decay time is strikingly similar to the value found for CsMgBr 3 :Eu 2+ . 17 Taking the fact into account that the ''t 2g '' excitation band is at almost the same energy in both compounds, one may conclude that the synergic effect of covalence and trigonal distortion leads to roughly the same position of the lowest excited state of Eu 2+ in CsMgBr 3 and CsMgI 3 and thus, to a similar decay time.
The radial integrals are also in a reasonable range. They are significantly larger for CsMgI 3 :Eu 2+ and CsSrI 3 :Eu 2+ than for CsCaI 3 :Eu 2+ . This indicates a larger overlap between the 4f and the 5d wavefunctions in the former two compounds, which is induced by the increasing anisotropy in both structures. In the case of CsMgI 3 :Eu 2+ , the small size of the Mg 2+ site still increases the value of the radial integral and thus, lowers the decay time.
In the cases of CsCaI 3 :Eu 2+ and CsSrI 3 :Eu 2+ , the low Stokes shifts allow the possibility of self-absorption that might lead to lengthening of the decay time. This effect can be excluded here, however, due to the low concentration of only 0.1 mol% that leads to a high probability of isolation of the Eu 2+ ions.

Conclusions
The photoluminescence properties of Eu 2+ -activated CsMI 3 (M = Mg, Ca, Sr) were presented and analyzed in detail. The crystal structures of the compounds can be derived from a perovskite structure type with six-fold coordinated alkaline earth sites substituted by Eu 2+ . All emission spectra are characterized by an asymmetric Gaussian-shaped band that can be assigned to 4f 6 ( 7 F J )5d 1 -4f 7 ( 8 S 7/2 ) transitions. All excitation spectra provide a distinct fine structure in both the ''t 2g '' and the e g state during approximation of an octahedral symmetry for the Eu 2+ ions. It can be assigned to the different 7 F J levels arising from 4f 6 core in the excited state using a decoupled scheme and thus assuming a weak exchange interaction between the 4f electrons and the 5d electron. This behavior is in agreement with the behavior observed for the respective bromides 17 and is otherwise only known for the alkali bromides and iodides. 31 In the cases of Eu 2+ -activated CsMgI 3 and CsSrI 3 , the excitation spectra reveal a non-negligible trigonal distortion of the [EuI 6 ] 4À octahedra leading to a splitting of the supposed ''t 2g '' state into an a 1g and an e g state. Their energetic order has been justified in terms of the angular overlap model of ligand field theory. The trigonal distortion leads to several irregular effects like a redshifted emission and excitation of Eu 2+ in CsSrI 3 with respect to Eu 2+ -activated CsCaI 3 that would not be anticipated assuming a simple octahedral symmetry. The asymmetry of the emission of Eu 2+ in CsMgI 3 and CsSrI 3 can be understood taking the splitting of the ''t 2g '' state into account.
An analysis of the behavior of the emission and excitation energies of Eu 2+ ions dependent upon the size of the alkaline earth site substituted indicates that emission occurs from the lowest excited supposed ''t 2g '' state. It turns out that the lowest excited states are slightly bonding whereas the high-energy e g state is almost non-bonding due to the large Eu-I bond length in the excited state.
All Eu 2+ -activated iodides presented in this paper show very efficient photoluminescence even at room temperature. The quenching temperatures T 1/2 are all well above room temperature, which is easily understood taking the low phonon energies in iodides into account that makes vibrational coupling rather inefficient. The Stokes shifts are also rather small except in the case of CsMgI 3 :Eu 2+ . Here, the large Stokes shift is most probably induced by the large amount of reorganization of the electron density in the excited state due to the low Eu-I bond length in the ground state. The values of the decay times are all shorter, or, in the case of CsMgI 3 :Eu 2+ , similar to the decay times detected for the respective bromides. 17 This is in accordance with the expectations. The decay times of CsMgI 3 :Eu 2+ and CsSrI 3 :Eu 2+ are extraordinarily short, which can also be correlated with their special structure. In fact, the larger disposal of the octahedra in these compounds indicates a larger covalence of the M-I bond and thus, the Eu-I bond, which is directly reflected in the lower decay times relative to CsCaI 3 :Eu 2+ . In general, the materials investigated in this work allow an extremely detailed insight into the Eu 2+ structureluminescence relationship due to the high resolution of the spectra and may, thus, serve as a textbook example in this context.