Systematic Analysis of the Crystal Chemistry and Eu3+ Spectroscopy along the Series of Double Perovskites Ca2LnSbO6 (Ln = La, Eu, Gd, Lu, and Y).

Eu3+ (1 mol %)-doped Ca2LnSbO6 (replacing Ln3+; Ln = Lu, Y, Gd, and La) and Ca2EuSbO6 were synthesized and structurally characterized by means of X-ray powder diffraction. The Eu3+ luminescence spectroscopy of the doped samples and of Ca2EuSbO6 has been carefully investigated upon collection of the excitation/emission spectra and luminescence decay curves of the main excited states. Surprisingly, apart from the dominant red emission from 5D0, all the doped samples show an uncommon blue and green emission contribution from 5DJ (J = 1, 2, and 3). This is made possible thanks to both multiphonon and cross-relaxation mechanism inefficiencies. However, the emission from 5D3 is more efficient and the decay kinetics of the 5DJ (J = 0, 1, and 2) levels is slower in the case of Y- and Lu-based doped samples. This evidence can find a possible explanation in the crystal chemistry of this family of double perovskites: our structural investigation suggests an uneven distribution of the Eu3+ dopant ions in Ca2YSbO6 and Ca2LuSbO6 hosts of the general A2BB'O6 formula. The luminescent center is mainly located in the A crystal site, and on average, the Eu-Eu distances are longer than in the case of the Gd- and La-based matrix. These longer distances can further reduce the efficiency of the cross-relaxation mechanism and, consequently, the radiative transitions are more efficient. The slower depopulation of Eu3+ 5D2 and 5D1 levels in Ca2YSbO6 and Ca2LuSbO6 hosts is reflected in the longer rise observed in the 5D1 and 5D0 decay curves, respectively. Finally, in Ca2EuSbO6, the high Eu3+ concentration gives rise to an efficient cross-relaxation within the subset of the lanthanide ions so that no emission from 5DJ (J = 1, 2, and 3) is possible and the 5D0 decay kinetics is faster than for the doped samples.

Introduction

Rare earth double perovskite materials with the general formula A2BB′O6 are characterized by interesting magnetic and dielectric properties.[1−3] The main structural motif of these compounds consists of a network of alternating BO6 and B′O6 octahedra, with A-atoms occupying the 12-coordinated interstitial spaces between octahedra. Depending on the nature and size of the other elements, the rare earth ions can occupy the A-site or B-site. In the Ca2LnRuO6 (Ln = La–Lu) system,[4] which crystallizes in the monoclinic P21/n space group, the Ca2+ and Ln3+ cations are partially disordered in the A-site and B-site positions of the A2BB′O6 double perovskite, and the Ru(V) cations are located at the B′-site; therefore, the general formula of these compounds is (Ca2–Ln)(Ln1–Ca)RuO6. The abundance of Ln3+ located at the B-site varies with its cationic radius: the larger Ln cations tend to occupy the A-site, whereas the smaller Ln cations tend to enter the B-site. Similar crystal chemistry is expected for antimonates with double perovskite materials and Ca2LnSbO6 formula. Although a systematic study on their crystal chemistry is still missing in the literature, two components of this family (Ca2LaSbO6 and Ca2YSbO6) have been effectively employed as hosts of luminescent trivalent lanthanide ions. In particular, Ca2LaSbO6, which can be obtained with Eu3+ up to 80% substituting La3+, has been considered a useful red phosphor.[5,6] Another efficient red phosphor can be obtained by doping Ca2YSbO6 with Eu3+ ion. The codoping with Bi3+ has been reported to enhance the intensity of the red emission.[7] Considering the crystal chemistry of the host, Y3+ ions are supposed to occupy only the centrosymmetric B site. Nevertheless, Eu3+ should be located in a noncentrosymmetric crystal site, since the 5D0 → 7F2 band dominates the luminescence spectrum.[8] Since Ca2+ is located in the A-site (C1 point symmetry), the authors reasonably assumed a Ca2+/Eu3+ substitution and the presence of a charge compensation mechanism. Finally, Ca2YSbO6 is also an effective host for other luminescent ions, such as trivalent Sm, Dy, Ho, and Er.[9]

Due to the lack of a comprehensive study on the crystal chemistry of the Ca2LnSbO6 family, we have found it interesting to undertake a structural study on Ca2LaSbO6, Ca2GdSbO6, Ca2LuSbO6, and Ca2YSbO6 doped with 1 mol % Eu3+, and neat Ca2EuSbO6, by means of X-ray diffraction. The effects of the different nature of the hosts on Eu3+ luminescence spectroscopy have been also discussed and some structural details have been revisited. This study, focusing on the structural/spectroscopic relationship, reveals the presence of unusual spectroscopic features of Eu3+ when introduced as an impurity in these antimonate hosts.

Experimental Section

Materials and Synthesis

Crystalline samples of 1 mol % Eu3+-doped Ca2LnSbO6 (replacing Ln3+; Ln = Lu, Y, Gd, and La) and Ca2EuSbO6 were prepared by solid-state reaction in and air atmosphere. CaCO3 (>99%), Sb2O5 (99.995%), Ln2O3 (Ln = Y and La, 99.99%; Ln = Lu and Gd, 99.9%), and Eu2O3 (99.99%) were thoroughly mixed and pressed into pellets under a pressure of 10 tons. The samples underwent two heat treatments: the first one at 600 °C for 6 h to eliminate carbonates and the second one at 1400 °C for 24 h with a slow cooldown of 3 °C/min. Intermediate grindings were performed to improve the homogeneity of the materials.

Structural Investigation

X-ray diffraction (XRD) patterns were measured with a Thermo ARL X’TRA powder diffractometer, operating in the Bragg–Brentano geometry and equipped with a Cu-anode X-ray source (Kα, λ =1.5418 Å), using a Peltier Si(Li)-cooled solid-state detector. The patterns were collected with a scan rate of 0.002°/s in the 18–120° 2θ range. Polycrystalline antimonate samples were ground in a mortar and then put in a side-loading sample holder for data collection.

The General Structure Analysis System (GSAS) program was employed for the Rietveld refinement calculations.[10] The instrumental X-ray peak profile functions and the sample displacement (SHFT variable) were determined by Rietveld refinement of the diffraction pattern of the LaB6 powder standard reference material (NIST 660C).

The reference structural model exploited in the Rietveld calculation was the one pertaining to the isostructural perovskite-like Ca3TeO6 determined in a study by Hottentot and Loopstra[11] in which Sb has been located in the place of Te and Ca, and Ln ions shared the two crystallographic positions of Ca.. The following structural refinement strategy has been performed: (i) refinement of the background functions (shifted Chebyschev), scale factor, and cell parameters; (ii) refinement of the occupation factors (OFs) of Ca and Ln in the two available crystal sites; (iii) refinement of the fractional atomic coordinates for Ca/Ln in the 4e crystal site (site A); (iv) refinement of the fractional atomic coordinates for the oxygen atoms; (v) refinement of the isotropic thermal parameter (Uiso) for Ca, Ln, and Sb ions; (vi) refinement of the isotropic thermal parameter (Uiso) for oxygen atoms (we do not observe significant improvement of the refinement employing anisotropic thermal parameters); (vii) global refinement of all structural variables mentioned above. In each step of the Rietveld calculation, also the GU, GV, GW, LX, LY, and asym profile terms of the pseudo-Voigt profile function no. 2, included in the GSAS program, were refined. Due to its very low concentration (1 mol %), the presence or absence of Eu3+ within the structural model is expected to not change, in a significant way, the result of the Rietveld refinements. For this reason, for the sake of simplicity, the presence of Eu3+ is not considered during the structural calculation. Crystal data such as atomic fractional coordinates, OFs, and Uiso for 1% Eu3+-doped Ca2LnSbO6 and Ca2EuSbO6 are reported in the Supporting Information (Tables S1–S6), along with other relevant powder diffraction data (see the Powder Diffraction Data section).

Spectroscopic Investigation

Room-temperature luminescence spectra and decay curves were measured with a Fluorolog 3 (Horiba-Jobin Yvon) spectrofluorometer, equipped with a Xe lamp, a double excitation monochromator, a single emission monochromator (mod. HR320), and a photomultiplier in photon counting mode for the detection of the emitted signal. All the spectra were corrected for the spectral response of the setup.

Results and Discussion

Structural Determination of 1% Eu3+-Doped Ca2LnSbO6 (Ln = Lu, Y, Gd, and La) and Ca2EuSbO6

In Figure , the picture of the crystal structure of the investigated double perovskite is shown.

Figure 1

Picture of the crystal structure, along the a axis, of the double perovskite of the general formula A2BB′O6. The location of the cationic sites A, B, and B′ is shown. Red spheres represent oxygen atoms.

The observed and fitted XRD patterns of doped Ca2LnSbO6 (Ln = Lu, Gd, and La chosen as representative samples) are shown in Figure .

Figure 2

Observed (crosses) and refined (continuous red line) powder patterns of (a) 1% Eu3+-doped Ca2LuSbO6, (b) 1% Eu3+-doped Ca2GdSbO6, and (c) 1% Eu3+-doped Ca2LaSbO6. The observed–refined curves are shown at the bottom of each plot. The same plots for 1% Eu3+-doped Ca2YSbO6 and Ca2EuSbO6 are reported in Figure S1.

Inspection of this figure shows that there is good agreement between the observed and refined powder patterns. The refined lattice parameters and OFs are given in Tables and 2, respectively, and some selected bond distances are listed in Table .

Table 1

Refined Lattice Parameters for the 1% Eu3+-Doped Ca2LnSbO6 Family (Ln = Lu, Y, Gd, and La) and for Ca2EuSbO6

	cell parameters (Å, °, Å3)
Ln in Ca2LnSbO6 host	a	b	c	β	V
Lu	5.5711(1)	5.7530(1)	7.9958(3)	89.913(2)	253.27(2)
Y	5.5888(1)	5.8021(1)	8.0494(3)	89.970(4)	261.01(1)
Gd	5.5884(2)	5.8466(3)	8.0817(1)	89.753(5)	264.06(3)
Eu	5.5947(2)	5.8522(2)	8.0887(2)	90.255(3)	264.83(2)
La	5.6830(1)	5.8795(1)	8.1707(1)	89.913(4)	273.01(2)

Table 2

Occupation Factors of Ln3+ and Ca2+ in the Different Crystal Sites of the Ca2LnSbO6 Host

		OF	ionic radius (Å)a
host	cation	site A (CN 8)/site B (CN 6)	CN 6	CN 8
Ca2LuSbO6	Lu3+	0.032(2)/0.936(3)	0.85	0.97
	Ca2+	0.968(2)/0.064(3)	1.00	1.12
Ca2YSbO6	Y3+	0.130(2)/0.872(2)	0.892	1.015
	Ca2+	0.870(2)/0.128(2)	1.00	1.12
Ca2GdSbO6	Gd3+	0.375(6)/0.250(5)	0.94	1.06
	Ca2+	0.625(6)/0.750(5)	1.00	1.12
Ca2EuSbO6	Eu3+	0.414(3)/0.171(2)	0.95	1.07
	Ca2+	0.586(3)/0.829(2)	1.00	1.12
Ca2LaSbO6	La3+	0.490(2)/0.020(2)	1.06	1.18
	Ca2+	0.510(2)/0.980(2)	1.00	1.12

Data taken from ref (12).

Table 3

Average M–O Bond Distances along the Ca2LnSbO6 Host Family

	average bond distance (Å)
Ln in Ca2LnSbO6 host	Sb–O	Ca/Ln(1)–O (site B; CN 6)	Ca/Ln(2)–O (site A; CN 8)
Lu	2.03(1)	2.18(1)	2.54(2)
Y	1.96(1)	2.28(1)	2.56(2)
Gd	1.99(1)	2.33(2)	2.55(3)
Eu	2.00(1)	2.33(2)	2.55(2)
La	2.01(1)	2.34(1)	2.61(2)

As expected, according to Vegard’s law, the cell size increases as the ionic radius of the trivalent lanthanide ions increases (Table ). Ca2+ and Ln3+ cations are partially disordered in the A-site (C1 point symmetry) and B-site (C point symmetry) positions of the A2BB′O6 double perovskite and the Ca/Ln distribution over these two available crystal sites is strongly dependent on the nature of the Ln ion. When small Lu and Y are considered, the trivalent ion shows a strong preference for the site with coordination number (CN) 6 (Table , site B). In the cases of Gd and Eu, the Ca/Ln distribution is almost homogeneous over the two crystal sites. Finally, the La3+ ion prefers to occupy the crystal site with CN = 8, while site B (CN = 6) is almost fully occupied by Ca2+ (Table ). Obviously, the degree of Ca/Ln disorder is mainly dependent on the difference between their ionic radii. The bigger the difference is, the smaller the disorder is. As discussed before, this is particularly true for Ln ions smaller than Ca2+, e.g., the Lu ion is mainly located in the crystal site B. Moreover, it is interesting to note that on average, the Sb–O distances (around 2 Å) are not significantly affected by the nature of the trivalent ion and the Ca/Ln–O distances deviate from 2.55 Å only for Ln = La (Table ). On the other hand, in the case of the crystal site B (CN = 6), the average Ca/Ln–O distance is around 2.33 Å for Ln = Gd, Eu, and La, while it is shorter for Y (2.28 Å) and, in particular, for Lu (2.18 Å) (Table and Figure S2). This behavior is quite similar to the one observed for the Ca2LnRuO6 (Ln = La–Lu) double perovskite family, already discussed above.[4]

We also point out some discrepancy with the crystal data present in the current literature. In the case of the Ca2LaSbO6 host, while the La ion has been calculated to exclusively occupy site A, by Yin et al.,[6] we found the presence of a small percentage of La3+ (2%) in the crystal site B. The most important discrepancy concerns the Ca2YSbO6 host. In the literature, Y3+ ions are supposed to occupy only the centrosymmetric octahedral B site,[7] while we detect the presence of this ion also in site A (OF = 0.130, Table ). Taking into account the substitution of Ca2+/Ln3+ ions in the crystal lattice by the luminescent Eu3+ (see the discussion below), this finding should have a significant impact. In fact, according to our conclusions, we can assume that Eu3+ could occupy both crystal sites A and B without the necessity of involving charge compensation mechanisms.

Luminescence of Ca2EuSbO6 and 1% Eu3+-Doped Ca2LuSbO6, Ca2YSbO6, Ca2GdSbO6, and Ca2LaSbO6

The normalized RT excitation spectra of 1% Eu3+-doped Ca2LnSbO6 (Ln = Lu, Y, Gd, and La) and Ca2EuSbO6 are shown in Figure . The spectra have been normalized to the Imax of the 7F0 → 5L6 transition.

Figure 3

Room-temperature luminescence excitation spectra of 1% Eu3+-doped Ca2LnSbO6 (Ln = Lu, Y, Gd, and La) and Ca2EuSbO6.

All compounds show various intense Eu3+ excitation peaks [around 362 nm (7F0 → 5D4 transition); 394 nm (7F0 → 5L6 transition); 415 nm (7F0 → 5D3 transition); 464 nm (7F0 → 5D2 transition); 526 nm (7F0 → 5D1 transition); 534 nm (7F1 → 5D1 transition)]. Also, an O → Eu charge transfer (CT) band below 300 nm is detected (not shown).

Upon excitation at 394 nm, we have obtained the luminescence emission spectra shown in Figure (the Lu-, Eu-, and La-based compounds are chosen as representative samples). All the emission spectra (Figures –7) have been normalized to the Imax of the 5D0 → 7F2 transition.

Figure 4

Room-temperature luminescence emission spectra of 1% Eu3+-doped Ca2LnSbO6 (Ln = Lu and La) and Ca2EuSbO6 upon excitation at 394 nm. Similar spectra of the 1% Eu3+-doped Ca2LnSbO6 (Ln = Y, Gd) are reported in Figure S3.

Figure 7

Details of the spectra of the 1% Eu3+-doped Ca2LnSbO6 and Ca2EuSbO6 phosphors upon excitation at 394 nm. (a) 400–500 nm range, bands originating in the 5D3 and 5D2 levels; (b) 500–570 nm range, bands originating in the 5D2 and 5D1 levels.

The peaks above 570 nm are mainly assigned to emission bands originating from Eu3+ 5D0 excited state. A close inspection of the peak around 580 nm (17,241 cm–1, 5D0 → 7F0 transition) shows significant differences between the different compounds (Figure and Table ).

Figure 5

Details of the 5D0 → 7F0 emission band upon excitation at 394 nm.

Table 4

Peak Position and Full Width Half Maximun (FWHM) of the 5D0 → 7F0 Transition in the Investigated Samples

host cation	peak position (nm)	peak position (cm–1)	FWHM (cm–1)
Lu	579.6	17,253	31.4
Y	579.3	17,262	36.2
Gd	578.9	17,275	42.4
Eu	578.8	17,276	39.5
La	578.6	17,281	38.9

We note that the 0-0 peak energies decrease along the lanthanide series (La → Lu). On the other hand, the full width half maximum (FWHM) is slightly larger for Ca2GdSbO6, Ca2EuSbO6, and Ca2LaSbO6 (close to 40 cm–1) than for Ca2YSbO6 and Ca2LuSbO6 hosts (36 and 31 cm–1, respectively). The Eu3+ ion, which is supposed to substitute for the Ln3+ one, can be located in the two available crystal sites (sites A and B, Table ). In principle, due to the presence of only one Stark level both for 5D0 and 7F0, there should be a one-to-one correspondence between the number of the 0-0 emission bands and the number of emitting crystal sites. Seemingly, even though only one 0-0 component is detected for all materials, the broader peak could possibly be due to the presence of two overlapping 0-0 bands corresponding to two emitting Eu3+ crystal sites. This seems to be more evident in the case of Gd-, Eu-, and La-based compounds, where the 0-0 feature is broader. This statement is confirmed by the crystal chemistry for the following reasons: (i) the Eu3+ ion can occupy both available crystal sites in Ca2EuSbO6 (Table ); (ii) in the case of Ca2LnSbO6 hosts (with Ln = Gd, Eu, and La), the similarity of the Ca(Ln)–O bond distances for each site suggests the presence of Eu in both available positions (Table ). On the other hand, in light of its short Ca(Ln)–O distances (2.28 Å for Y and 2.18 Å for Lu, Table and Figure S2), the crystal site B appears too small to accommodate Eu3+ in the case of Ca2YSbO6 and Ca2LuSbO6 matrices. In fact, the sum of the ionic radii of Eu and O (2.35 Å; CN = 6) is significantly higher than the aforementioned Ca(Ln)–O distances. In these hosts, the probable predominant occupation of site A by the Eu3+ ion is supported by the smaller FWHM of the 0-0 peak (Table ). As far as the 0-0 peak position (transition energy) is concerned, it is well known that it can provide information about the covalency of the donor atom–Eu3+ bonds. This feature, which is strictly related to the nephelauxetic effect, would predict a decrease in the 0-0 transition energy upon an increase in the donor atom–Eu3+ bond covalency.[13,14] Nevertheless, since the factors affecting the 5D0 energy of Eu3+ are many and still a subject of debate in the literature,[15] we prefer, in the present contribution, to not discuss further this aspect. At this stage, it is useful to point out that the electric dipole transitions in the emission spectrum of Eu3+ located in the centrosymmetric crystal site B are highly forbidden and only the magnetic dipole-allowed 5D0 → 7F1 transition should be detectable. Nevertheless, in related antimonate materials, we demonstrated that the presence of cationic disorder (Ca and Ln, in the present case) induces the removal of the local inversion symmetry in the case of ions formally occupying centrosymmetric sites, from a crystallographic point of view.[16] Therefore, also the Eu3+ emission from the cationic site B in Ca2LnSbO6 can occur through a forced electric dipole mechanism and, therefore, also the 5D0 → 7F (J = 0, 2, 3, and 4) transition can be detected. The different luminescence emission features of Eu3+ [i.e., 5D0 → 7F (J = 1 and 2); Figure ] are likely to be related to a different occupation of the crystal sites by Eu3+ along the Ca2LnSbO6 family.

Figure 6

Emission upon 394 nm excitation: (a) 5D0 → 7F1 transition for 1% Eu3+-doped Ca2LaSbO6 and Ca2LuSbO6; (b) 5D0 → 7F2 emission band for 1% Eu3+-doped Ca2LnSbO6.

In the case of Eu3+-doped Ca2LuSbO6 and Ca2YSbO6, the components of the emission manifolds are sharper and located at different values of wavelengths compared to the ones of the samples having Ca2GdSbO6 and Ca2LaSbO6 as hosts. Furthermore, the more complex emission pattern and the broader emission peaks are compatible with a multisite emission in the case of Ca2GdSbO6, Ca2EuSbO6, and Ca2LaSbO6. In particular, the emission profile of the 5D0 → 7F2 transition is reasonably similar in all the materials under investigation (Figure b); this agrees with the fact that for the 0-2 hypersensitive transition, only the noncentrosymmetric sites significantly contribute to the emission intensity. On the other hand, the behavior of the emission profile is much more complex in the case of the 5D0 → 7F1 magnetic dipole-allowed transition; as shown in Figure a, the 0-1 band has very different shapes for Ca2LaSbO6 and Ca2LuSbO6. This is due to the fact that in the former host (representative of a large Ln ion), Eu3+ is also located in centrosymmetric sites that give their contribution to the emission intensity, together with the noncentrosymmetric ones, but with different crystal field splitting, giving rise to broader features.

In the case of Eu3+-doped samples, and not of neat Ca2EuSbO6, luminescence from the high-energy levels 5D3, 5D2, and 5D1 has been also detected (Figure ) upon excitation at 394 nm. i.e., in the 5L6 level.[17] This is made possible by the relatively low energy vibrational modes of the antimonate double perovskite host (ν∼ ≤ 800 cm–1)[18] that makes multiphonon relaxation among the 5D levels not fully efficient, given the values of the energy gaps between these levels (2549–2592 cm–1 for 5D3-5D2, 2511–2515 cm–1 for 5D2-5D1, and 1724–1750 cm–1 for 5D1-5D0). The situation is similar to the one reported many years ago for YVO4:Eu3+ (zircon phase),[19] where the maximum phonon energy is around 900 cm–1,[20] while in the case of the fluoride β-NaYF4 host, having dominant energy vibrational modes located between 300 and 400 cm–1,[21] emission can be observed also from the 5L6 level and even higher-energy ones.[22] This is not possible in oxide-based hosts. We also note that the 5D1-5D0 energy gap (∼1700 cm–1) cannot be efficiently bridged by multiphonon relaxation due to a selection rule that occurs if the two involved levels have J = 0 and J = 1.[23,24]

The observed behavior can be explained on the basis of cross-relaxation processes, leading to the depopulation of the high-energy 5D (J = 1, 2, and 3) levels at high Eu3+ concentrations.[25] These processes are identified as energy transfer mechanisms involving the 7F0 ground level or the thermally populated first excited level (7F1 and 7F2 above 7F0). The almost resonant transitions are:

The mismatches are relatively small so that the cross-relaxation processes are presumably almost resonant. It is well known that the energy transfer probabilities significantly decrease as the intracenter distances increase; this agrees with a fast cascade depopulation of 5D3, 5D2, and 5D1 to 5D0 in fully concentrated Ca2EuSbO6. For this reason, only 5D0 is emissive in the neat material, where the shortest Eu3+–Eu3+ distance is equal to only 3.27(2) Å and energy transfer can be efficient.

The decay curves of the 5D3 level were found to be too fast to be measured with present equipment. On the other hand, the temporal evolution of the emission intensity after pulsed excitation at 394 nm was measured for the 5D0 level for all samples and for 5D2 and 5D1 for 1% Eu3+-doped Ca2LnSbO6 (Ln = Lu, Y, Gd, and La) (see Figures –10).

Figure 8

Decay curve of the 5D2 level upon 394 nm excitation for 1% Eu3+-doped Ca2LnSbO6.

Figure 10

Decay curve of the 5D0 level upon 394 nm excitation for 1% Eu3+-doped Ca2LnSbO6.

In the case of the 5D2 level, upon excitation at 394 nm, the decay curves for the doped samples show an extremely short buildup of the emission intensity, followed by a decay that is exponential for Ln = Lu and Y and non-exponential for Ln = Gd and La. The decay times are about 0.15 ms for the former materials and 0.04–0.08 ms for the latter (e-folding time), indicating that the cross-relaxation is more efficient for Ln = Gd and La (Figure ).

As for the temporal evolution of the 5D1 emission, it is characterized by a clear rise that is clearly longer for Ln = Lu and Y than for Gd and La. The rise time cannot be properly evaluated but is in the region of tens of microseconds. This is followed by a nearly exponential decay with rates that are in the regions of 0.17 ms for Ln = Lu and Y and 0.11–0.13 ms for Gd and La (Figure ).

Figure 9

Decay curve of the 5D1 level upon 394 nm excitation for 1% Eu3+-doped Ca2LnSbO6.

Finally, in the case of the 5D0 level, upon pulsed excitation at 394 nm, a clear rise is observed for the doped materials due to feeding from the upper levels. This rise is approximately in the region of 0.10–0.30 ms, being longer for Ln = Lu and Y than for Ln = Gd and La. The decay is nearly exponential, with 5D0 lifetimes of about 1.02 ms for Ln = Lu and Y and 0.83 ms for Gd and La, although longer components seem to be present in the long time tail of the decay curve (Figure ). As for Ca2EuSbO6, an exponential decay is observed, with a decay constant of 0.19 ms (Figure ).

The short and exponential 5D0 decay in the neat Eu compound is clearly due to the presence of migration in this level in the Eu3+ subset of ions, until a killer center is reached and nonradiative relaxation occurs. This behavior has been reported many times in the literature for neat crystals containing Eu3+ (for instance, see Kellendonk and Blasse’s study[26]). As noted above, the absence of a buildup time clearly indicates that the population of 5D0 from the higher lying level is fast, as expected for cross-relaxation in the neat material.

On the other hand, the observed rise times of 5D0 and 5D1 appear to be close to the decay times of the level lying immediately above. This is compatible with their sequential population from the level above through cross-relaxation, in agreement with the rate equation model proposed by Berdowski and Blasse and for Eu3+ in NaGdTiO4.[27]

In general, the results obtained upon pulsed excitation appear to be different for the doped samples with Ln = Lu and Y with respect to the ones with Ln = Gd and La. In parallel, the emission intensity from the 5D3 level is significantly lower for Gd- and La-based hosts (Figure a). This peculiar behavior can find a tentative explanation if we consider a more efficient energy transfer mechanism by cross-relaxation between Eu3+ ions, where the Eu–Eu distances are, on average, shorter. In this context, the 5D3 level is more efficiently depopulated and the lifetimes of 5D0, 5D1, and 5D2 levels are shorter. This is the case of Ca2LaSbO6 and Ca2GdSbO6 hosts, in which both available A and B crystal sites are supposed to be occupied by Eu3+ and the shortest Eu–Eu distance is around 3.29(1) Å (in Ca2GdSbO6), corresponding to the smallest distance between sites A and B. Otherwise, in Ca2LuSbO6, where the luminescent ion is supposed to selectively occupy site A, the shortest possible Eu–Eu distance is larger than 3.29 Å [3.90(1) Å (the distance between two adjacent sites A)]. Nevertheless, in view of the complex crystal structure of the hosts and the structural disorder possibly inducing the breaking of the local inversion symmetry of B sites,[16] other factors can contribute to determine the observed spectroscopic features.

It is interesting to note that due to the different contributions of the emission stemming from 5D3, 5D2, 5D1 (in the blue and green spectral regions), and 5D0 (mainly in the red), the final emission color can be tuned. The point 4 in the CIE diagram (Figure ) (corresponding to an almost pure red color in Ca2EuSbO6) can be moved toward the green region (points 3 and 5) thanks to the presence of a significant green component (the 5D1 → 7F1 band around 535 nm) in Ca2GdSbO6 and Ca2LaSbO6 hosts. A significant blue component (410–430 nm, corresponding to the 5D3 → 7F1,2 transitions) has been observed for the Ca2LuSbO6 and Ca2YSbO6 hosts so that points 1 and 2 are closer to the blue spectral region (Figure ).

Figure 11

CIE coordinate diagram of 1% Eu3+-doped Ca2LnSbO6 upon excitation at 394 nm.

Conclusions

Eu3+ (1 mol %)-doped Ca2LnSbO6 (Ln = Lu, Y, Gd, and La) samples show unusual spectroscopic features, which are connected to the low-energy vibrational modes of the antimonate double perovskite and to the different site distribution of the luminescent Eu3+ dopant ions. In particular, the color of the emitted light can be tuned by simply acting on the nature of Ln ions in the host. The typical red emission of Ca2EuSbO6 can be slightly shifted toward the green and blue spectral regions when 99% of Eu is replaced by La (or Gd) and by Lu (or Y), respectively. This is made possible thanks to the inefficient multiphonon relaxation process among 5D levels in the doped samples. In this way, we observe emission in the 460–570 nm region (from 5D1 and 5D2) in all the Eu3+-doped samples and emission in the 400–450 nm region (from the 5D3 level) mainly in the Eu3+-doped Ca2LuSbO6 and Ca2YSbO6 materials. In general, the emission from the 5D (J = 0–3) levels of Eu3+ is more efficient in the Y- and Lu-based hosts and a possible explanation for this behavior can be found in the crystal chemistry of these materials. The structural investigation on the host matrix suggests a different occupation of the cationic sites by the Eu3+ dopant, which should be preferentially located in the A site when Ln = Y or Lu, while they can occupy both available crystal sites (A and B) in the case of Ca2LaSbO6 and Ca2GdSbO6 hosts.

Although the cross-relaxation mechanism involving Eu3+ ions (a nonradiative energy transfer pathway) is weakly active in all the doped samples, it is expected to be more efficient when the Eu–Eu distances are shorter. This is the case of the La- and Gd-based compounds, where the shortest possible Eu–Eu distance is around 3.29 Å vs 3.90 Å in the case of Ca2LuSbO6. As a consequence, the emission intensity from the 5D3 level is significantly lower and the decay of the 5D0, 5D1, and 5D2 levels is faster for Gd- and La-based hosts. Finally, as a consequence of the drastic increase in Eu3+ concentration, in the case of Ca2EuSbO6, an efficient cross-relaxation mechanism involving 5D levels takes place and no emission from 5D levels (J = 1, 2, and 3) is possible. In addition, a fast migration of the energy toward killer centers is responsible for a drastic decrease in 5D0 excited state lifetime.

This contribution clearly shows how important is the detailed knowledge of the host crystal chemistry and of the nonradiative mechanisms taking place to control the luminescence features of an optical material.