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

Abstract

The photoluminescence of Eu²⁺ ions doped in iodidoperovskites, CsMI₃ (M = Mg, Ca, Sr), was studied. Eu²⁺ occupies the alkaline earth site in each of these compounds. The emission spectra of the doped perovskites are characterized by 4f⁶5d¹ → 4f⁷ transitions of Eu²⁺ 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 ⁷F_J (J = 0,…,6) levels of the 4f⁶ 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:Eu²⁺ and CsSrI₃:Eu²⁺ show a redshifted emission with respect to CsCaI₃:Eu²⁺. 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 Eu²⁺ doped earth alkaline iodidoperovskites may serve as a textbook example for the detailed understanding of the structure–luminescence relationships of Eu²⁺ ions, which is very important for tailoring functional materials for respective applications.

---

1. Introduction

The luminescence of Eu²⁺ 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 Eu²⁺-activated material. For example, in fluorides, such as CaF₂ and SrF₂, Eu²⁺ shows violet emission,¹ whereas in KMgF₃, the 5d states are located at higher energies such that 4f–4f line emission from the excited ⁶P_J and ⁶I_J states is also observed in the UV range.^2,3 On the other hand, in nitridosilicates such as M₃Si₆O₁₂N₂:Eu²⁺ (M = Sr, Ba),⁴ M₂Si₅N₈:Eu²⁺ (M = Ca, Sr, Ba)⁵ or MSi₂O₂N₂:Eu²⁺ (M = Ca, Sr),^6,7 Eu²⁺ shows yellow to orange luminescence. Further examples include Eu²⁺-activated CaS or SrS, showing a red emission.⁸ The examples mentioned above show how the luminescence of Eu²⁺ can be affected only by varying the composition of the host material. Interesting reviews that summarize and discuss the data of many of the Eu²⁺-activated compounds presented in the literature were published by Rubio⁹ and Dorenbos¹⁰ during the last two decades.

Both for basic research and applications in high-energy radiation detection, halides are interesting host materials for Eu²⁺. 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 Eu²⁺-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 4f⁶5d¹ states below the lowest excited 4f⁷ state. Moreover, crystal field splitting shifts the lowest 4f⁶5d¹ state yet to lower energies, as was indicated by Blasse.¹¹ Typical examples from the literature are MZnCl₄:Eu²⁺ (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 Eu²⁺ ions doped in the ternary chlorides CsMCl₃ (M = Mg, Ca, Sr).¹³ Many other chlorides are presented in ref. 10 and references given therein.

The number of reported Eu²⁺-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 CsCaBr₃:Eu^2+ 14 and CsSrBr₃:Eu²⁺.^15,16 A detailed analysis of the photoluminescence properties of Eu²⁺ in the series CsMBr₃ (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 Eu²⁺ doped in ternary iodides is currently only described in a superficial manner in the literature.¹⁰ Recently, the applicability of CsMI₃ (M = Ca, Sr) as scintillators has been investigated and presented by the group of Melcher.^18–21 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 Eu²⁺ 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 Eu²⁺ doped in CsMgI₃, which has not been presented so far in the literature.

To the best of our knowledge, no detailed analysis of the Eu²⁺-activated ternary iodides CsMI₃ (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 Eu²⁺ 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 Eu²⁺ by distinct changes in the size of the site the Eu²⁺ 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.

2. Experimental

2.1. Preparation

CsI (Merck, 99.5%) was dried at 200 °C in a dynamic vacuum in order to remove any traces of moisture. The anhydrous iodides MI₂ (M = Ca, Sr) could be obtained by first dissolving the respective hydrates CaI₂·4H₂O (Acros Organics, 99%) and SrI₂·xH₂O (Heraeus, 99.5%) in concentrated HI (Acros Organics, 57 wt%) to remove traces of MCO₃ (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 °C to obtain pure anhydrous iodides. EuI₂ was prepared from Eu₂O₃ (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 °C h⁻¹ of the thus prepared ternary compound (NH₄)₃EuI₆·xH₂O to 300 °C in a dynamic vacuum, EuI₂ 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_α radiation).

The doped ternary iodides CsMI₃:Eu²⁺ (M = Ca, Sr) were prepared by fusing the respective starting materials CsI and MI₂ (M = Ca, Sr) at their corresponding molar ratios together with 0.1 mol% EuI₂ in sealed silica ampoules under vacuum using the Bridgman technique. For CsMgI₃:Eu²⁺, CsI₃ (Alfa Aesar, 98%), Mg metals (smart elements, 99.8%) and 0.1 mol% EuI₂ were used as starting materials instead. All mixtures were heated at a rate of 20 °C h⁻¹ to their respective melting points and kept in the melt for 36 h. The mixtures were then slowly recrystallized (2 °C h⁻¹) to 100 °C below their melting points and finally cooled to room temperature at 10 °C h⁻¹. 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).

Fig. 1 Measured X-ray powder diffraction patterns of (a) CsMgI₃:0.1% Eu²⁺, (b) CsCaI₃:0.1% Eu²⁺ and (c) CsSrI₃:0.1% Eu²⁺ 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.

2.2. 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 (λ_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).

3. Results

3.1. Crystal structures and coordination spheres of CsMI₃ host lattices (M = Mg, Ca, Sr)

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 2θ = 15° and 22°, which is an artifact due to the fixation of the powders with silicon grease. No effects of doping with EuI₂ 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 EuI₂.

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.

Table 1 Overview of the local site symmetries of the M²⁺ sites and average M–I distances in CsMI₃ (M = Mg, Ca, and Sr)^23,24

Compound	Local site symmetry of M^2+ site	Average M–I distance/Å
CsMgI3	D 3d	2.90
CsCaI3	C i	3.10
CsSrI3	C 2h	3.37

---

CsMgI₃ crystallizes in the CsNiCl₃ structure type, which is a hexagonally distorted variation of the perovskite type structure with the space group P6₃/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⁻ ions in the form of anti-cuboctahedra whereas the Mg²⁺ 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 [MgI₆]⁴⁻ 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 Eu²⁺ has already been discussed in the isostructural compound CsCdBr₃.^25,26

Fig. 2 Representation of the crystal structure of CsMgI₃ indicating the anisotropy due to the face-sharing [MgI₆]⁴⁻ octahedra along the hexagonal c axis.²³

CsCaI₃ crystallizes in the orthorhombic GdFeO₃ structure type with the space group Pnma (no. 62) and is isostructural to CsSrBr₃ ²⁴ (see Fig. 3). The average Ca–I distance is 3.10 Å. The Cs⁺ ions are (8+2)-fold coordinated by I⁻ ions and the Ca²⁺ ions are six-fold coordinated by I⁻ 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).

Fig. 3 Representation of the crystal structure of CsCaI₃ with emphasis on the slightly tilted [CaI₆]⁴⁻ octahedra.²⁴

CsSrI₃ crystallizes in a filled PuBr₃ structure type, which is characterized by an orthorhombic crystal structure with the space group Cmcm (no. 63)²⁴ and is depicted in Fig. 4. In this structure, the Cs⁺ ions are only 8-fold coordinated by I⁻ ions in the form of two-fold capped trigonal prisms. The Sr²⁺ 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.

Fig. 4 Representation of the crystal structure of CsSrI₃ with indication of the large disposal of the [SrI₆]⁴⁻ octahedra.²⁴

With regard to different structures, a remarkable development can be noted as the distortion of the [MI₆]⁴⁻ octahedra (M = Mg, Ca, Sr) increases in the series CsCaI₃ < CsSrI₃ < CsMgI₃ and ends in a face-sharing pattern of the octahedra in CsMgI₃. Along with this development, the deviation from a perfect octahedral environment for Eu²⁺ in these three compounds also increases and trigonal distortion has to be taken more into account. We will review this aspect in Section 4.

3.2. Photoluminescence of CsMI₃:0.1%Eu²⁺ (M = Mg, Ca, and Sr)

The Eu²⁺-activated iodides presented in this paper show a bright luminescence upon UV excitation at room temperature, which is the blue region for CsCaI₃:Eu²⁺ and CsSrI₃:Eu²⁺ and greenish-yellow in the case of CsMgI₃:Eu²⁺. For all three materials, the transitions can be interpreted with the fact that the Eu²⁺ ions occupy exclusively the respective alkaline earth sites. For the respective Eu²⁺-activated bromides CsMBr₃ (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 Eu²⁺ in the respective compounds.^10,27,28 We will also use this notation in this paper.

3.3. CsMgI₃

CsMgI₃:0.1% Eu²⁺ 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 Eu²⁺-activated CsMgI₃ 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⁻¹ (577 nm). The full width at half maximum of Γ_10K = 2430 cm⁻¹ is relatively large, indicating a strong electron–phonon coupling. The emission band can be assigned to 4f⁶5d¹(t_2g) → 4f⁷(⁸S_7/2) transitions of Eu²⁺ substituting for Mg²⁺ 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 Eu²⁺ and Mg²⁺(r(Eu²⁺) = 1.17 Å and r(Mg²⁺) = 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:Eu²⁺, where the coordination number of the Cs⁺ sites is 8 and the respective Eu²⁺ emission is already located at 22 470 cm⁻¹ (445 nm).³⁰ Moreover, we performed EXAFS measurements on the Eu²⁺-doped isostructural compounds CsMgCl₃ and CsMgBr₃, which we will publish in another paper. The measurements definitely prove the sole occupation of the Mg²⁺ sites at concentrations below 3%.

Fig. 5 Emission (orange line, E_ex = 24 390 cm⁻¹) and excitation spectrum (black line, E_em = 18 020 cm⁻¹) of CsMgI₃:0.1% Eu²⁺ at 10 K. Note the splitting of the anticipated “t_2g” state into the trigonal field states a_1g and e_g. Inset: CsMgI₃:0.1% Eu²⁺ showing yellow luminescence upon UV irradiation at room temperature.

The assumption of octahedral site symmetry for the Eu²⁺ 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 Eu²⁺ in this compound, this approximation is indeed too simplified.

The excitation spectrum of CsMgI₃:0.1% Eu²⁺ 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 Eu²⁺ is too simplified in this case, as we will explain below. The fine structure arises due to the remaining 4f⁶ core in the excited state configuration of Eu²⁺ and can be identified with the different ⁷F_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 halides³¹ and Eu²⁺-activated CsMBr₃,¹⁷ 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 state³² and becomes much less resolved at room temperature. The positions of the different ⁷F_J states are compiled in Table 2 and their position relative to the ⁷F₀ state fits well with the values anticipated from Eu³⁺.³³ The good resolution of the different ⁷F_J states allows a relatively precise determination of the Stokes shift ΔS(2+,CsMgI₃) = 3870 cm⁻¹.

Table 2 Positions of the 4f⁶(⁷F_J)5d¹ states in CsMgI₃:0.1% Eu²⁺ and energy differences ΔE_J–0 of the 4f⁶(⁷F_J)5d¹ (J = 1–6) states relative to the 4f⁶(⁷F₀)5d¹ state

4f^6(^7FJ)5d^1(t2g)			4f^6(^7FJ)5d^1(eg)
State ^7FJ	Position/cm^−1	ΔEJ–0/cm^−1	State ^7FJ	Position/cm^−1	ΔEJ–0/cm^−1
^7F0	21 200	0	^7F0	35 240	0
^7F1	21 700	500	^7F1	36 050	810
^7F2	22 500	1300	^7F2	36 540	1300
^7F3	23 170	1970	^7F3	37 310	2070
^7F4	24 160	2960	^7F4	38 430	3190
^7F5	25 240	4040	^7F5	39 290	4050
^7F6	26 280	5080	^7F6	40 600	5360

---

A detailed observation on the excitation spectrum of the Eu²⁺-related emission depicted in Fig. 5 reveals the presence of another band located at 27 770 cm⁻¹ (360 nm). This remarkable feature has already been observed in Eu²⁺-activated isostructural compounds such as CsCdBr₃ ^25,26 and CsMgBr₃.¹⁷ 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, ε_trig(7,2+,CsMgI₃) = 4310 cm⁻¹.

3.4. CsCaI₃

CsCaI₃:0.1% Eu²⁺ 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 Eu²⁺-activated CsCaI₃ 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⁻¹ (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 Γ_10K = 640 cm⁻¹. The emission can be clearly identified by the 4f⁶5d¹(t_2g) → 4f⁷(⁸S_7/2) transition of Eu²⁺ substituting for Ca²⁺ ions under the assumption of an approximate octahedral site symmetry. The difference in ionic radii (r(Eu²⁺) = 1.17 Å and r(Ca²⁺) = 1.00 Å)²⁹ explains the observed asymmetry of the Gaussian band.

Fig. 6 Emission (blue line, E_ex = 25 640 cm⁻¹) and excitation spectrum (black line, E_em = 22 470 cm⁻¹) of CsCaI₃:0.1% Eu²⁺ at 10 K. Inset: CsCaI₃:0.1% Eu²⁺ showing blue luminescence upon UV irradiation at room temperature.

A distinct fine structure is observed in the excitation spectra arising due to the different ⁷F_J states from the 4f⁶ core. As there are only two broad bands observed in the excitation spectrum, the assumption of an octahedral environment for Eu²⁺ seems legitimate, giving rise to a t_2g and an e_g state. Similar to CsMgI₃:Eu²⁺ (see Fig. 5), the fine structure is also observable in the e_g state. This indicates, again, a weak exchange interaction between the 4f⁶ electrons and the 5d electron.³² The positions of the different ⁷F_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 Eu³⁺.³³

Table 3 Positions of the 4f⁶(⁷F_J)5d¹ states in CsCaI₃:0.1% Eu²⁺ and energy differences ΔE_J–0 of the 4f⁶(⁷F_J)5d¹ (J = 1–6) states relative to the 4f⁶(⁷F₀)5d¹ state

4f^6(^7FJ)5d^1(t2g)			4f^6(^7FJ)5d^1(eg)
State ^7FJ	Position/cm^−1	ΔEJ–0/cm^−1	State ^7FJ	Position/cm^−1	ΔEJ–0/cm^−1
^7F0	23 680	0	^7F0	34 970	0
^7F1	24 350	670	^7F1	35 600	630
^7F2	24 940	1260	^7F2	36 230	1260
^7F3	26 050	2370	^7F3	36 990	2020
^7F4	27 230	3550	^7F4	37 860	2890
^7F5	27 790	4110	^7F5	38 980	4010
^7F6	29 050	5370	^7F6	39 810	4840

---

Due to the well-resolved fine structure, it is also possible to determine the Stokes shift accurately, which is ΔS(2+,CsCaI₃) = 1310 cm⁻¹. The octahedral crystal field splitting, 10 Dq, can be also determined and is given by ε_cfs(7,2+,CsCaI₃) = 11 050 cm⁻¹ using the nomenclature used by Dorenbos.¹⁰

3.5. CsSrI₃

Similar to Eu²⁺-activated CsCaI₃, CsSrI₃:0.1% Eu²⁺ 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 CsCaI₃, Eu²⁺-activated CsSrI₃ also shows efficient radioluminescence and is thus a promising candidate for scintillation materials.^20,21 The emission band arising from the 4f⁶5d¹ → 4f⁷(⁸S_7/2) transition is an almost perfect Gaussian band with a maximum at 22 010 cm⁻¹ (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 CsCaI₃ and given by Γ_10K = 710 cm⁻¹. However, the slight asymmetry cannot be explained by a mismatch between the ionic radii of Eu²⁺ and Sr²⁺ in this case, as they are almost equal (r(Eu²⁺) = 1.17 Å, r(Sr²⁺) = 1.18 Å).²⁹ Moreover, it turns out that the emission is slightly redshifted to the 5d–4f emission of Eu²⁺ in CsCaI₃, which does not correspond to the expectation when considering the decrease of the ionic radii from Ca²⁺ to Sr²⁺ and in turn the weaker crystal field splitting.

Fig. 7 Emission spectrum (cyan line, E_ex = 25 000 cm⁻¹) and excitation spectrum (black line, E_em = 21 980 cm⁻¹) of CsSrI₃:0.1% Eu²⁺ at 10 K. Note the splitting of the anticipated “t_2g” state into the trigonal field states a_1g and e_g. Inset: CsSrI₃:0.1% Eu²⁺ showing blue luminescence upon UV irradiation at room temperature.

As for the other iodides, the fine structure due to the different ⁷F_J levels is also observed in CsSrI₃ in all excitation bands (see Table 4 for their positions). The agreement with the values reported for Eu³⁺ is also satisfactory.³³ The Stokes shift determined thereof reads ΔS(2+,CsSrI₃) = 1090 cm⁻¹.

Table 4 Positions of the 4f⁶(⁷F_J)5d¹ states in CsSrI₃:Eu²⁺ (0.1%) and energy differences ΔE_J–0 of the 4f⁶(⁷F_J)5d¹ (J = 1–6) states relative to the 4f⁶(⁷F₀)5d¹ state

4f^6(^7FJ)5d^1(“t2g”)			4f^6(^7FJ)5d^1(eg)
State ^7FJ	Position/cm^−1	ΔEJ–0/cm^−1	State ^7FJ	Position/cm^−1	ΔEJ–0/cm^−1
^7F0	23 100	0	^7F0	35 300	0
^7F1	23 610	510	^7F1	35 980	680
^7F2	24 350	1250	^7F2	36 610	1310
^7F3	25 070	1970	^7F3	37 410	2110
^7F4	26 190	3090	^7F4	38 180	2880
^7F5	27 300	4200	^7F5	39 260	3960
^7 F 6	27 930	4830	^7 F 6	40 440	5140

---

However, similar to the case of CsMgI₃:0.1% Eu²⁺ (cf.Fig. 5), another isolated band can be observed at 29 320 cm⁻¹ (341 nm). This indicates a similarity of the site symmetry for the Eu²⁺ ions in CsSrI₃ compared to CsMgI₃. The larger disposal of the [SrI₆]⁻ octahedra with respect to CsCaI₃ (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 Eu²⁺ ions in CsSrI₃, 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 CsSrI₃ given by ε_trig(7,2+,CsSrI₃) = 3960 cm⁻¹.

3.6. Temperature-dependent emission

The temperature dependence of the 5d–4f emission of Eu²⁺ in the series CsMI₃ (M = Mg, Ca, Sr) was also investigated. The respective emission spectra are depicted in Fig. 8–10. In all three compounds, Eu²⁺ shows an intense luminescence even at room temperature and a slight blueshift with increasing temperature is observed. In the case of CsMgI₃:0.1% Eu²⁺ (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

Fig. 8 Temperature-dependent emission spectra (E_ex = 24 390 cm⁻¹) of CsMgI₃:0.1% Eu²⁺.

Fig. 9 Temperature-dependent emission spectra (E_ex = 25 640 cm⁻¹) of CsCaI₃:0.1% Eu²⁺.

Fig. 10 Temperature-dependent emission spectra (E_ex = 25 000 cm⁻¹) of CsSrI₃:0.1% Eu²⁺.

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 Eu²⁺ in CsCaI₃ and CsSrI₃ decrease to only 88% and 85% of the initial value when heated to 300 K, whereas in the case of CsMgI₃:0.1% Eu²⁺, it decreases to 71%. This indicates that the quenching temperature T_1/2, at which the integrated intensity 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 Eu²⁺ in these compounds and very low phonon energies in iodides (below 200 cm⁻¹)³⁴ make non-radiative transitions very inefficient and a weak electron–phonon coupling should be expected. This behavior is clearly observed in Fig. 11.

Fig. 11 Evolution of the integrated intensities of the 5d–4f emission of Eu²⁺ ions doped in the iodides CsMI₃ (M = Mg, Ca, and Sr).

3.7. Decay times

The decay curves of the luminescence signal of Eu²⁺ in CsMI₃ (M = Mg, Ca, Sr) are depicted in Fig. 12. They were recorded upon excitation into the “t_2g” states (λ_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 Eu²⁺ 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.

Fig. 12 Semi-log plot of the decay curves of CsMI₃:0.1% Eu²⁺ obtained upon excitation into the “t_2g” state (λ_ex = 355 nm) at room temperature. The solid black lines indicate exponential fits to the respective data.

Table 5 Decay times and radial integrals |〈5d|r|4f〉| of CsMI₃:0.1% Eu²⁺ (M = Mg, Ca, and Sr) determined at room temperature

Compound	Decay time/ns	Radial integral |〈5d|r|4f〉|/Å
CsMgI3:Eu^2+	667	1.03
CsCaI3:Eu^2	875	0.50
CsSrI3:Eu^2	707	0.70

---

From the decay times, it is possible to obtain direct information about the electric dipole matrix element μ_5d→4f, since

μ_5d→4f = −e·|〈5d|r|4f〉| (1)

where e is the elementary charge. The matrix element 〈5d|r|4f〉 represents the radial integral between these two states and is related to the decay time by³⁵where A denotes the total transition probability per unit time, E_em(7,2+,A) is the emission energy of Eu²⁺ in compound A in cm⁻¹ and χ is a correction factor that includes the refraction index n of the host medium, For CsCaI₃, a refraction index of 2.0 has been deduced from band structure calculations,¹⁹ whereas for CsSrI₃, a refraction index of 1.79 was calculated.³⁶ Due to the similarity in the structures of CsSrI₃ and CsMgI₃, a refraction index of 1.79 was also assumed for CsMgI₃. With these values, it is possible to estimate the radial integrals for the 5d–4f transition for Eu²⁺ in the iodides CsMI₃ (M = Mg, Ca, Sr), as summarized in Table 5.

4. Discussion

4.1. Photoluminescence of Eu²⁺ doped in CsMI₃ (M = Mg, Ca, Sr)

Eu²⁺ shows very intense luminescence in the series CsMI₃ (M = Mg, Ca, Sr), which goes from blue in CsCaI₃ and CsSrI₃ to yellow in CsMgI₃. 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 ⁷F_J states (J = 0,…,6) arising from the 4f⁶ core in the excited 4f⁶5d¹ 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 Eu²⁺ is also known to exhibit well-resolved excitation spectra with such a fine structure, include Sr₃(PO₄)₂,³⁷ Ba₂B₅O₉Br³⁸ 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 Eu²⁺ 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 CsMI₃ (M = Mg, Ca, Sr).

Table 6 Characteristic energies for Eu²⁺ extracted from the luminescence spectra of CsMI₃:0.1% Eu²⁺. All values are in cm⁻¹ and were recorded at 10 K. The addition (7,2+,A) was omitted here

CsMI3			E em	D	ΔS	Γ 10K
CsMgI3	21 200	35 240	17 330	12 800	3870	2430
CsCaI3	23 680	34 970	22 370	10 320	1310	640
CsSrI3	23 100	35 300	22 010	10 900	1090	710

---

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 CsMgI₃ and CsSrI₃.

We also attempted to analyze the emission energies E_em(7,2+,A) and the positions of the ⁷F₀ levels within the “t_2g” and e_g states, denoted as and , respectively, with respect to the difference in the ionic radii between Eu²⁺ and the respective alkaline earth ion M²⁺ (M = Mg, Ca, Sr). The plots are illustrated in Fig. 13.

Fig. 13 Variation of the emission energies E_em(7,2+,A) and the energies of the ⁷F₀ level within the hypothetical “t_2g” and e_g states, and , with the difference in Shannon radii ΔR between Eu²⁺ and the alkaline earth ion in CsMI₃ (M = Mg, Ca, and Sr).²⁹

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 Eu²⁺ 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 Eu²⁺-activated CsMgI₃ and CsSrI₃ that both the energy of the “t_2g” states and the emission energy are lower than those in the case of CsCaI₃. 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 Eu²⁺ and I⁻, 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 Eu²⁺ and the 5p electrons of I⁻ ions is negligibly small giving rise to the relatively constant behavior.

A more detailed view in Fig. 13 already reveals the special role that CsCaI₃ plays within this series. In fact, the emission energy and the energy of the ⁷F₀ levels within the t_2g state go through a maximum, whereas the respective energy of the ⁷F₀ 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 CsCaI₃, 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 [EuI₆]⁴⁻ octahedra increases from CsCaI₃ over CsSrI₃ to CsMgI₃. In the last compound, the octahedra are face-sharing (see Fig. 2). This goes along with an increasing trigonal distortion of the [EuI₆]⁴⁻ octahedra to D_3d symmetry and the fact that the Eu²⁺ ions experience stronger repulsive forces due to the alkaline earth ions in the second coordination sphere. This is not the case for CsCaI₃. The subsequent symmetry reduction to the real site symmetry C_2h in CsSrI₃ 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 Eu²⁺ in CsMgI₃ and CsSrI₃ should be lower than those in the case of CsCaI₃. These aspects are well reproduced by the spectra (see Fig. 13) and are, in fact, the reason for the strange behavior of CsSrI₃:Eu²⁺ showing a slightly redshifted emission with respect to CsCaI₃:Eu²⁺. In the latter Eu²⁺-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 Eu²⁺ in iodidoperovskites. Especially in the case of CsMgI₃:Eu²⁺, the emission is extraordinarily redshifted. We have reported a similar behavior for CsMgBr₃:Eu²⁺.¹⁷ Both compounds exhibit a high anisotropy and a pseudo-one-dimensional characteristic along the hexagonal c axis. Together with the small size of the Mg²⁺ 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 CsSrI₃:Eu²⁺ with respect to CsMgI₃:Eu²⁺ and the observed splitting in the excitation spectra, we assign the extraordinary redshift for CsMgI₃:Eu²⁺ to the effect of the trigonal distortion induced by the tilts of the [EuI₆]⁴⁻ 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 red-emitting 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, ε^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 ε_trig(7,2+,A), whereas the resulting e_g state is increased by 1/3 ε_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 is the average energy of the excitation band assigned to the e_g state in O_h symmetry and and 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.

Table 7 Deduced values of the undistorted octahedral crystal field splitting, ε^undist._cfs(7,2+,A) = 10 Dq for CsMI₃:0.1% Eu²⁺ (M = Mg, Ca, and Sr)

Compound	ε ^undist.cfs(7,2+,A)/cm^−1
a No trigonal distortion in this case.
CsMgI3:Eu^2+	11 300
CsCaI3:Eu^2+^a	11 050^a
CsSrI3:Eu^2+	9600

---

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⁻ is a larger ion and more polarizable and, thus, a softer ligand. It should be noted, however, that for CsMgBr₃:Eu²⁺, the crystal field splitting was directly derived without considering trigonal distortion.¹⁷ Therefore, the corrected value for 10 Dq in the case of CsMgBr₃:Eu²⁺ will be smaller (13 230 cm⁻¹). 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 halides³¹ making our interpretation plausible.

4.2. Energetic order of trigonal crystal field states in CsMgI₃:Eu²⁺ and CsSrI₃:Eu²⁺

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 Schäffer⁴⁰ 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 Eu²⁺, the latter were neglected in the AOM approximation. In the case of CsMgI₃:Eu²⁺ and CsSrI₃:Eu²⁺, the trigonal distortion of the [EuI₆]⁴⁻ octahedra evolves along the C₃ axis, thus leading to the disposal of the ligands with respect to the normal z axis by a small polar angle Θ (see Fig. 14). The strong tilting pattern of the respective octahedra in the structures of CsMgI₃ and CsSrI₃ (see Fig. 2 and 4) induces a stronger repulsion between the ligands, which makes the trigonal distortion also plausible from an electrostatic perspective.

Fig. 14 Numbering of the ligands and their respective angular positions within the equatorial plane (Φ) and relative to the normal z axis (Θ) of the [EuI₆]⁴⁻ octahedron in CsMgI₃ and CsSrI₃.

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 Eu²⁺ ion is placed into the origin of the coordinate system, the positions of the ligands can be parametrized by the disposal Θ of the Eu–I bond with respect to the z axis and their azimuthal angle Φ relative to the x axis. The numbering of the ligands as denoted in Fig. 14 leads to the parametrization shown in Table 8.

Table 8 Parametrization of the polar angle Θ and the azimuthal angle Φ of the ligands as denoted in Fig. 13

	L1	L2	L3	L4	L5	L6
Φ	π/4	0	π/2	π	3π/2	5π/4
Θ	θ	π/2 − θ	π/2 + θ	π/2 + θ	π/2 − θ	π + θ

---

The distortion angle Θ is not known quantitatively, thus, it will be denoted by a small angle θ. 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 × 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 CsMgI₃ and CsSrI₃, some other assumptions have still to be taken into account. Within the AOM approximation, only σ- and π-like interactions are considered and parametrized by the interaction terms e_σ and e_π. Moreover, for small angles θ, the functions sin θ and cos θ 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:

ε_xz = 4e_π + θ²(9e_σ − 11e_π) (6)

ε_yz = 4e_π + θ²(9e_σ − 11e_π) (7)

ε_xy = 4e_π − 2e_πθ² (8)

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

ε^AOM_trig(7,2+) = 9θ²(e_σ − e_π) (9)

Since I⁻ ions are weak π donors, it may be assumed that e_σ ≫ e_π. 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 [MI₆]⁴⁻ octahedra (M = Mg, Ca, and Sr) increases in the series CsCaI₃ < CsSrI₃ < CsMgI₃, the repulsion between the I⁻ ligands also increases, and so does θ. In the case of CsCaI₃:Eu²⁺, θ 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_π. This agrees with the well-known result from ligand field theory.⁴⁰ In the cases of CsMgI₃:Eu²⁺ and CsSrI₃:Eu²⁺, however, the angle θ has to be taken into account. The [MgI₆]⁴⁻ octahedra in CsMgI₃ are tilted to a much larger degree than the respective octahedra in CsSrI₃. Therefore, the splitting due to the trigonal distortion should be lower in the latter compound, which is indeed observed (see Table 9).

Table 9 Experimental values for the trigonal crystal field splitting of the “t_2g” manifold, as deduced from the excitation spectra in CsMI₃:Eu²⁺ (M = Mg, Ca, and Sr)

Compound	ε trig(7,2+,A)/cm^−1
CsMgI3:Eu^2+	4310
CsCaI3:Eu^2+	—
CsSrI3:Eu^2+	3960

---

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, , should decrease with increasing θ. This is exactly the observed behavior. CsCaI₃:Eu²⁺ provides the limiting case θ → 0, thus leading to a maximum for in this series.

Fig. 15 finally represents a summarizing schematic overview of the evolution of the 5d states of Eu²⁺ in CsMI₃ (M = Mg, Ca, and Sr). It shows the irregular behavior of the emission energies illustratively. The centroid shift, ε_c(7,2+,A), should only be regarded qualitatively and was intentionally not determined for iodidoperovskites. Since a strong degree of tilt for the [MI₆]⁴⁻ octahedra is not reasonable in terms of Pauling's rules⁴² 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 bromidoperovskites¹⁷ does not seem very reasonable. We will discuss the effect of higher covalence of the Eu–I bond in CsMgI₃:Eu²⁺ and CsSrI₃:Eu²⁺ in terms of the decay times in Section 4.4.

Fig. 15 Schematic energy level scheme for the 5d states of Eu²⁺ and the resulting luminescence in CsMI₃:Eu²⁺ (M = Mg, Ca, and Sr). The centroid shift as drawn was determined from the average position of the “t_2g” state and should only be considered qualitatively.

4.3. Temperature-dependent emission

All Eu²⁺-activated iodidoperovskites CsMI₃ (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 CsCaI₃:Eu²⁺ and CsSrI₃:Eu²⁺ 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⁻¹).³⁴ The 5d–4f emission of Eu²⁺ in CsCaI₃ and CsSrI₃ exhibits a blueshift with higher temperatures, which arises due to the thermal population of the higher ⁷F_J levels within the lowest excited 5d state.

Interestingly, the Stokes shift ΔS(2+,CsMgI₃) and the FWHM Γ at 10 K are both very large in the case of CsMgI₃:Eu²⁺ (see Table 6). Moreover, the emission is highly asymmetric. This behavior is already familiar from CsMgBr₃:Eu²⁺.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 Eu²⁺ is typically characterized by Stokes shifts in the range of 4000–10 000 cm⁻¹, large FWHMs (4000–6000 cm⁻¹), shortened decay times and low quenching temperatures T_1/2.⁴³ However, the large covalence of the Eu–I bond in CsMgI₃ will shift the 5d states to low energies. Due to that fact and the large crystal field splitting due to the small Mg²⁺ 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 ⁷F_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 Eu²⁺ in CsMgI₃. 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 Mg²⁺ and Eu²⁺ (ΔR = 0.45 Ų⁹) and the relatively short Mg–I bond length (see Table 1). Upon excitation into the diffuse 5d orbitals of Eu²⁺, 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 Ce³⁺-activated elpasolite CsNaLuCl₆ 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 CsSrI₃:Eu²⁺, however, the Stokes shift ΔS(2+,CsSrI₃) and the FWHM Γ at 10 K are very low (see Table 6), although the structure of the host compound is very similar to that of CsMgI₃. Moreover, a lower Stokes shift of 1090 cm⁻¹ and 1420 cm⁻¹,¹⁷ respectively, is observed compared to CsSrBr₃:Eu²⁺. 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 CsCaBr₃:Eu²⁺ (ΔS(2+,CsCaBr₃) = 1080 cm⁻¹ (ref. 17)) with CsCaI₃:Eu²⁺ (ΔS(2+,CsCaI₃) = 1310 cm⁻¹, 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 1²⁴). Since Eu²⁺ ions have approximately the same ionic radius as Sr²⁺ 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.

4.4. Decay times and radial integrals

The measured decay times and deduced values of the radial integrals (see Table 5) of Eu²⁺ in CsMI₃ (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.

Fig. 16 Dependence of the decay times τ (●, dashed curve) and radial integrals |〈5d|r|4f〉| (▲, dotted curve) of Eu²⁺ upon the size of the alkaline earth site in CsMI₃ (M = Mg, Ca, and Sr).

The behavior of the decay times of Eu²⁺ with varying differences in ionic radii of Eu²⁺ and the alkaline earth ion is very similar to the variation of the emission energies and thus, the energies indicate that the “t_2g” state is the emissive state. As has already been discussed above, Eu²⁺-activated CsSrI₃ and CsMgI₃ show an exceptional behavior also in the decay times and conclusively, also in the radial integrals. In fact, the decay times of Eu²⁺ 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 4f⁷ states of Eu²⁺. Thus, the decay times of CsMI₃:Eu²⁺ (M = Mg, Sr) are lowered compared to the decay time of Eu²⁺ in CsCaI₃.

The decay times detected for CsCaI₃:Eu²⁺ and CsSrI₃:Eu²⁺ 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 CsMgI₃:Eu²⁺, the decay time is strikingly similar to the value found for CsMgBr₃:Eu²⁺.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 Eu²⁺ in CsMgBr₃ and CsMgI₃ and thus, to a similar decay time.

The radial integrals are also in a reasonable range. They are significantly larger for CsMgI₃:Eu²⁺ and CsSrI₃:Eu²⁺ than for CsCaI₃:Eu²⁺. 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 CsMgI₃:Eu²⁺, the small size of the Mg²⁺ site still increases the value of the radial integral and thus, lowers the decay time.

In the cases of CsCaI₃:Eu²⁺ and CsSrI₃:Eu²⁺, 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 Eu²⁺ ions.

5. Conclusions

The photoluminescence properties of Eu²⁺-activated CsMI₃ (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 Eu²⁺. All emission spectra are characterized by an asymmetric Gaussian-shaped band that can be assigned to 4f⁶(⁷F_J)5d¹ → 4f⁷(⁸S_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 Eu²⁺ ions. It can be assigned to the different ⁷F_J levels arising from 4f⁶ 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 bromides¹⁷ and is otherwise only known for the alkali bromides and iodides.³¹ In the cases of Eu²⁺-activated CsMgI₃ and CsSrI₃, the excitation spectra reveal a non-negligible trigonal distortion of the [EuI₆]⁴⁻ 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 Eu²⁺ in CsSrI₃ with respect to Eu²⁺-activated CsCaI₃ that would not be anticipated assuming a simple octahedral symmetry. The asymmetry of the emission of Eu²⁺ in CsMgI₃ and CsSrI₃ 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 Eu²⁺ 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 Eu²⁺-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 CsMgI₃:Eu²⁺. 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 CsMgI₃:Eu²⁺, similar to the decay times detected for the respective bromides.¹⁷ This is in accordance with the expectations. The decay times of CsMgI₃:Eu²⁺ and CsSrI₃:Eu²⁺ 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 CsCaI₃:Eu²⁺. In general, the materials investigated in this work allow an extremely detailed insight into the Eu²⁺ structure–luminescence relationship due to the high resolution of the spectra and may, thus, serve as a textbook example in this context.

Acknowledgements

We are grateful to Prof. Dr Richard Dronskowski and Prof. Dr Andrei Tchougréeff from RWTH Aachen for their enormous help with the angular overlap model considerations.

Notes and references

1. T. Kobayasi, S. Mroczkowski, J. F. Owen and L. H. Brixner, J. Lumin., 1980, 21, 247.
2. A. Ellens, A. Meijerink and G. Blasse, J. Lumin., 1994, 59, 293.
3. D. K. Sardar, W. A. Sibley and R. Alcala, J. Lumin., 1982, 27, 401.
4. C. Braun, M. Seibald, S. L. Börger, O. Oeckler, T. D. Boyko, A. Moewes, G. Miehe, A. Tücks and W. Schnick, Chem. – Eur. J., 2010, 16, 9646.
5. Y. Q. Li, J. E. J. v. Steen, J. W. H. v. Krevel, G. Botty, A. C. Delsing, F. J. DiSalvo, G. D. With and H. T. Hintzen, J. Alloys Compd., 2006, 417, 273.
6. V. Bachmann, C. Ronda, O. Oeckler, W. Schnick and A. Meijerink, Chem. Mater., 2009, 21, 316.
7. V. Bachmann, T. Jüstel, A. Meijerink, C. Ronda and P. J. Schmidt, J. Lumin., 2006, 121, 441.
8. N. Yamashita, O. Harada and K. Nakamura, Jpn. J. Appl. Phys., 1995, 34, 5539.
9. J. O. Rubio, J. Phys. Chem. Solids, 1991, 52, 101.
10. P. Dorenbos, J. Lumin., 2003, 104, 239.
11. G. Blasse, Phys. Status Solidi B, 1973, 55, K131.
12. C. Wickleder, J. Alloys Compd., 2000, 300, 193.
13. D. H. Gahane, N. S. Kokode, B. M. Bahirwar and S. V. Moharil, Phys. Procedia, 2012, 29, 42.
14. A. Y. Grippa, N. V. Rebrova, T. E. Gorbacheva, V. Y. Pedash, N. N. Kosinov, V. L. Cherginets, V. A. Tarasov, O. A. Tarasenko and A. V. Lopin, J. Cryst. Growth, 2013, 371, 112.
15. V. L. Cherginets, A. Y. Grippa, T. P. Rebrova, Y. N. Datsko, T. V. Ponomarenko, N. V. Rebrova, N. N. Kosinov and O. V. Zelenskaya, Funct. Mater., 2012, 19, 429.
16. V. L. Cherginets, N. V. Rebrova, A. Y. Grippa, Y. N. Datsko, T. V. Ponomarenko, V. Y. Pedash, N. N. Kosinov, V. A. Tarasov, O. V. Zelenskaya, I. M. Zenya and A. V. Lopin, Mater. Chem. Phys., 2014, 143, 1296.
17. M. Suta, P. Larsen, F. Lavoie-Cardinal and C. Wickleder, J. Lumin., 2014, 149, 35.
18. M. Zhuravleva, B. Blalock, K. Yang, M. Koschan and C. L. Melcher, J. Cryst. Growth, 2012, 352, 115.
19. M. Tyagi, M. Zhuravleva and C. L. Melcher, J. Appl. Phys., 2013, 113, 203504.
20. K. Yang, M. Zhuravleva and C. L. Melcher, Phys. Status Solidi RRL, 2011, 5, 43.
21. H. Wei, M. Zhuravleva, K. Yang and C. L. Melcher, J. Cryst. Growth, 2013, 384, 27.
22. M. D. Taylor and C. P. Carter, J. Inorg. Nucl. Chem., 1962, 24, 387.
23. G. L. McPherson, A. M. McPherson and J. L. Atwood, J. Phys. Chem. Solids, 1980, 41, 495.
24. G. Schilling and G. Meyer, Z. Anorg. Allg. Chem., 1996, 622, 759.
25. S. García-Revilla and R. Valiente, J. Phys.: Condens. Matter, 2006, 18, 11139.
26. S. García-Revilla, F. Rodríguez and R. Valiente, J. Lumin., 2008, 128, 937.
27. P. Dorenbos, J. Phys.: Condens. Matter, 2003, 15, 575.
28. P. Dorenbos, J. Phys.: Condens. Matter, 2003, 15, 4797.
29. R. D. Shannon, Acta Crystallogr., Sect. A: Cryst. Phys., Diffr., Theor. Gen. Crystallogr., 1976, 32, 751.
30. H. J. Seo, W. S. Zhang, T. Tsuboi, S. H. Doh, W. G. Lee, H. D. Kang and K. W. Jang, J. Alloys Compd., 2002, 344, 268.
31. J. Hernandez, W. K. Cory and J. Rubio, J. Chem. Phys., 1980, 72, 198.
32. F. M. Ryan, W. Lehmann, D. W. Feldman and J. Murphy, J. Electrochem. Soc., 1974, 121, 1475.
33. W. T. Carnall, G. L. Goodman, K. Rajnak and R. S. Rana, J. Chem. Phys., 1989, 90, 3443.
34. K. Wakamura, Phys. Rev. B: Condens. Matter Mater. Phys., 1997, 56, 11593.
35. S. H. M. Poort, A. Meijerink and G. Blasse, J. Phys. Chem. Solids, 1997, 58, 1451.
36. G. Shwetha and V. Kanchana, Phys. Rev. B: Condens. Matter Mater. Phys., 2012, 86, 115209.
37. S. H. M. Poort, J. W. H. v. Krevel, R. Stomphorst, A. P. Vink and G. Blasse, J. Solid State Chem., 1996, 122, 432.
38. A. Meijerink and G. Blasse, J. Lumin., 1989, 43, 283.
39. S. H. M. Poort, W. P. Blokpoel and G. Blasse, Chem. Mater., 1995, 7, 1547.
40. C. E. Schäffer, Pure Appl. Chem., 1970, 24, 361.
41. C. E. Schäffer, Struct. Bonding, 1968, 5, 68.
42. L. Pauling, J. Am. Chem. Soc., 1929, 51, 1010.
43. P. Dorenbos, J. Phys.: Condens. Matter, 2003, 15, 2645.
44. R. Valiente, F. Rodríguez, J. González, H. U. Güdel, R. Martín-Rodríguez, L. Nataf, M. N. Sanz-Ortiz and K. Krämer, Chem. Phys. Lett., 2008, 481, 149.

---