- Email: [email protected]
Local structure determination of Mn2+ in CaCl2:Mn2+ by optical spectroscopy
Optical Materials 27 (2005) 1456–1460 www.elsevier.com/locate/optmat
Local structure determination of Mn2+ in CaCl2:Mn2+ by optical spectroscopy D. Castan˜eda, G. Mun˜oz H., U. Caldin˜o* Departamento de Fı´sica, Universidad Auto´noma Metropolitana, Unidad Iztapalapa, P.O. Box 55-534, 09340 Me´xico, D.F., Me´xico Received 26 August 2004; accepted 20 October 2004 Available online 10 December 2004
Abstract Excitation and emission spectra of Mn2+ ions in calcium chloride have been studied using the photoluminescence technique. From a crystal-field analysis of the excitation spectrum is inferred that the Mn2+ ion occupies the Ca2+ cation site forming octahe2+ dral MnCl4 is noticed as a consequence of smaller ionic radius of Mn2+ 6 complexes. A ligand inward relaxation around the Mn than that of host cation. This investigation supports the usefulness of crystal-field spectra for obtaining information on the local structure of impurity ions in ionic lattices. 2004 Elsevier B.V. All rights reserved. PACS: 78.55.m; 78.55.Hx Keywords: Mn2+; CaCl2; Calcium chloride; Photoluminescence; Crystal-field; MnCl4 6
1. Introduction A good characterization of an active ion within an ionic lattice requires to know the local structure around the ion, that is the neighbourhood ions, site symmetry and bond distances, since the optical and magnetic properties of doped crystals depend significantly on the complex formed by the impurity ion and its nearest neighbours. Moreover, the R distance between the impurity ion and its nearest neighbours can be very different from the R0 distance corresponding to the perfect lattice. The R distance has could be determined from a crystal-field analysis of excitation optical spectra [1–4], which has resulted to be the same (within the experimental error) as those previously determined by other techniques, such as the extended X-ray absorption fine structure (EXAFS) [1], and electron paramagnetic resonance (EPR) or electron nuclear double resonance
*
Corresponding author. E-mail address: [email protected] (U. Caldin˜o).
0925-3467/$ - see front matter 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.optmat.2004.10.009
(ENDOR) spectroscopy [5,6] through measuring the isotropic superhyperfine constant. In addition, an accuracy ˚ has been achieved with such an optiof the order 103 A ˚ ) achieved cal procedure in contrast with that (102 A with EXAFS [1], in spite of its continuous improvements in the last years. The goal of the present work is to investigate by means of optical spectroscopy the local structure of Mn2+ ions (site symmetry and bond distances) in CaCl2:Mn2+. The interest in this crystal, with orthorhombic structure, lies in that excitation spectroscopy performed on this phosphor containing Mn2+ ions has revealed that such ions can be excited free of photoconductive process, and hence, their excitation bands have been attributed to Mn2+ internal transitions of the 3d5 configuration [7]. The site symmetry of Mn2+ in CaCl2 could be determined from an analysis of its crystal-field transitions. The Mn2+–Cl distance was obtained from the R-dependence of 10Dq ligand field splitting considering a KRn variation [1–4]. The 10Dq magnitude of Mn2+ in CaCl2:Mn2+ was determined from an analysis of its excitation spectrum
D. Castan˜eda et al. / Optical Materials 27 (2005) 1456–1460
using the procedure developed by Curie et al. [8]. This procedure includes the covalent effects as well as the Racah–Trees or polarization and seniority corrections. Such covalent effects are taken into account through the covalent reduce Racah parameters B 0 and C 0 which represent a measure of the metal–ligand covalency. Thus, in octahedral coordination metal–ligand complexes the simultaneous reduction of the B and C Racah parameters is brought out by the N 4t factor [9]: B0 ¼ BN 4t and C 0 ¼ CN 4t . The Ne normalization factor, which is associated with the Racah parameter reduction in tetrahedral coordination complexes, is related with Nt through the Koide–Pryce covalency parameter e [10]: Ne = Nt(1e)1/2. If such a reduction of B and C or (Nt and Ne) with respect to the free ion values, B = 918 cm1 and C = 3273 cm1 (Nt = Ne = 1) [8] is due to a bonding effect, then the electronic wavefunction within the complex is no longer a pure Mn2+ orbital, but rather a molecular orbital wavefunction, which can be expressed as: jt2g i ¼ N t ðjd xy;yz;xz i kt jLt iÞ
and
jeg i ¼ N e ðjd 3z2r2;x2y2 i ke jLe iÞ;
ð1Þ
where d and L refer to the metal d-orbitals and the pand s-orbital linear combinations, respectively. In this way, Nt and Ne are important bonding parameters, which require to be evaluated in order to obtain some information on the bonding ionicity, since weaker the mixing of metal and ligand functions, Nt and Ne become close to 1.
2. Experimental A colourless single crystal of CaCl2 doped with Mn2+ ions was grown in our laboratory in an evacuated quartz ampoule using the Bridgman technique. Starting materials of very high purity were previously dehydrated by slow heating to 280 C under mechanical vacuum. Due to the quite hygroscopic character of calcium chloride, the optical spectroscopy of CaCl2:Mn2+ was performed with the crystal contained into an evacuated quartz ampoule to avoid adsorption and inclusion of water. The very small oscillator strengths ( 108) of the Mn2+ ion crystal-field transitions make difficult to detect photoluminescence spectra in crystals doped with low concentration of Mn2+. Thus, the manganese concentration in the melt was around 3200 ppm, which was the lowest one so that the Mn2+ luminescence could be detected with our experimental equipment. The lowest concentration possible is employed to prevent dimer formation, as well migration processes and exchange interactions, which can significantly affect the optical spectra. Emission and excitation spectra were obtained with the same experimental equipment described in [7]. The
1457
excitation spectra were not corrected for the photomultiplier response, since we were only concerned with the peak positions and not with the band intensities. Lifetime measurements were performed exciting the crystal at 337 nm, with 10-ns pulses of a PRA LN120C UV laser. The resulting transient fluorescence signal was analysed with a 0.45 m Czerny–Turner monochromator, detected with a cooled Hamamatsu R943–03 photomultiplier tube, and processed by a Hewlett Packard model 54201A digitising oscilloscope.
3. Experimental results: analysis Fig. 1 displays the excitation and emission spectra of Mn2+ ions in CaCl2:Mn2+ at room-temperature (RT). The emission spectrum, obtained under excitation at 414 nm, consists of a broad band peaking at 578 nm, which is associated with the 4T1g(G) ! 6A1g(S) spin-forbidden transition of Mn2+ ions. In fact, the time dependence of this emission is exponential, with a lifetime of about 21 ms, which proves the forbidden character of the transition. The excitation spectrum, monitored at 580 nm, was found to be very similar to those exhibited by Mn2+ ions in octahedral coordination chlorides, such as CdCl2:Mn2+ [11], NH4MnCl3 [12], KMgCl3:Mn2+ [1], KCaCl3:Mn2+ [1] and CsSrCl3:Mn2+ [1]. It consists of several absorption bands, which are associated with Mn2+ transitions from its 6A1g(S) ground state to the 4 T1g(G), 4T2g(G), [4A1g(G), 4Eg(G)], 4T2g(D), 4Eg(D), 4 T1g(P), 4A2g(F), 4T1g(F) and 4T2g(F) levels, since excitation into any of these bands gives rise to the 4 T1g(G) ! 6A1g(S) emission. The same excitation spectra are obtained regardless of the monitored luminescence wavelength, which indicates that all Mn2+ ions are excited via the same excitation bands. The most
Fig. 1. RT emission and excitation spectra of CaCl2:Mn2+. The emission spectrum was obtained after excitation at 414 nm and the excitation one was monitored at 580 nm.
1458
D. Castan˜eda et al. / Optical Materials 27 (2005) 1456–1460
intense excitation band, peaking at 414 nm, and within which was obtained the emission spectrum, corresponds to the 6A1g(S) ! [4A1g(G), 4Eg(G)] transition. The position of this transition does not essentially depend upon the crystal-field strength. The energies of the Mn2+ crystal-field transitions displayed in the excitation spectrum (Fig. 1) were fitted using the procedure developed by Curie et al. [8], in which the adjustable parameters are the crystal-field splitting 10Dq, the covalent reduced Racah parameter B 0 and the Koide–Pryce [10] covalency parameter e. The values for the a Racah–Trees and b seniority corrections and for x ( = C/B = C 0 /B 0 ) were taken from those derived for the free Mn2+ ion (a = 65 cm1, b = 131 cm1 and x = 3.565 [8]). The energy matrices used in the fitting (which are listed in the Appendix A) were those derived by Mehra [13] for the strong-field scheme, but including the corrections mentioned above. A better fitting is achieved taking also a as adjustable parameter, so that if the covalent effect upon the Racah–Trees correction is taken into account, then a is larger in the crystal than for the free ion. The calculated (with best fit a) and experimental energies in cm1 and eV for the Mn2+ excitation transitions displayed in Fig. 1 are given in Table 1. The inclusion of transition energies in eV results proper for readers used to dealing with this energy unit. The energies of the 4 T1g(F) and 4T2g(F) levels were not taken into account in the fitting since they appear overlapped at RT. The values obtained for the parameters 10Dq, B 0 and e, as well as for the normalization factors Nt and Ne, are also included in the Table. Ne is smaller than Nt in concordance with the greater bonding character of the eg-orbitals (Eq. (1)). The 10Dq ligand field splitting obtained (5073 cm1) with a best fit value for a ( = 85.4 cm1) resulted to be very similar to that (5082 cm1) derived with free ion a. A 10Dq value slightly larger (5250 cm1) has been previously reported in CaCl2:Mn2+ (10000 ppm) crystals [7], which can be attributed to the very high concentration of Mn2+ ions in those crystals. The quality of our fitting was estimated using the quantity X defined by:
"
N X ðEjth Ejexp Þ2 X¼ N j¼1
#1=2 ð2Þ
;
where j runs over the N fitted transitions, and Ejth and Ejexp represent the theoretical and experimental energies of the crystal-field transitions labelled by j. X resulted to be 250.6 cm1 with a = 65 cm1 and 210.9 cm1 with a = 85.4 cm1, that is, considering the covalent effect on a the theoretical levels fit even better to the observed levels. This good agreement between experimental and calculated values supports the assignment of the crystal-field transitions made in Fig. 1, and hence, it can be inferred that the Mn2+ occupies the divalent cation site, forming octahedral MnCl4 6 complexes. Moreover, the fact that Nt and Ne resulted to be close to 1 (Table 1) suggests a high ionicity of the MnCl4 complexes formed in 6 CaCl2:Mn2+. The Mn2+–Cl distance in MnCl4 complexes of 6 manganese doped orthorhombic structure crystals such as KMgCl3:Mn2+, RbCaCl3:Mn2+ and CsSrCl3:Mn2+ (ABCl3:Mn2+) has been determined from the Mn2+ excitation spectrum by assuming that the 10Dq ligand field splitting is proportional to R5 [1,14]. In such chloride crystals the Mn2+ ion substitutes the divalent cation (Mg2+, Ca2+ or Sr2+) forming octahedral MnCl4 6 units. Table 2 lists data of 10Dq parameter derived from excitation spectra for these ABCl3:Mn2+ crystals, as well as the Mn2+–Cl distances determined from the relation: 10Dq ¼ KR5
˚ ðunits in cm1 and AÞ;
ð3Þ
using 10Dq and R data of a NH4MnCl3 crystal as a standard [1,14]. The use of NH4MnCl3 data (10Dq = ˚ ) is suitable since R is 5620 cm1 and R = R0 = 2.525 A equal to the R0 host distance, and moreover, such a crystal gives a perfect octahedral environment to Mn2+ [15]. Although the assumption of R5 dependence of ligand field splitting (predicted by the crystal field theory) gives rise to 10Dq values substantially smaller than the experimental ones [2], a very good agreement between Mn2+–Cl distances obtained by EXAFS and optical excitation spectroscopy has been found for KMgCl3: Mn2+ [1] and RbCaCl3:Mn2+ [1] (see Table 2). There-
Table 1 Experimental and fitting energies of the crystal-field transitions of Mn2+ in CaCl2:Mn2+ at RT and values of the fitting characteristic parameters Transitions
Experimental energies
6
(cm1)
(eV)
(cm1)
(eV)
20 321 23 364 24 154 24 154 27 747 28 571 29 913 37 425
2.52 2.90 3.00 3.00 3.44 3.54 3.71 4.64
19 942 23 571 24 063 24 158 28 076 28 576 29 688 37 473
2.47 2.92 2.98 3.00 3.48 3.54 3.68 4.65
4
A1g(S) !
T1g(G) T2g(G) 4 A1g(G) 4 Eg(G) 4 T2g(D) 4 Eg(D) 4 T1g(P) 4 A2g(F) 4
Fitting energies
Fitting parameters
10Dq = 5073 cm1 (0.629 eV) B 0 = 838.3 cm1 e = 0.0208 a = 85.4 cm1 Nt = 0.978 Ne = 0.968
D. Castan˜eda et al. / Optical Materials 27 (2005) 1456–1460
1459
Table 2 R0 host distance, 4T1g(G) first excited state energy, 10Dq ligand field splitting and R Mn2+–Cl distance in ABCl3:Mn2+ and CaCl2:Mn2+ crystals Crystal
4
T1g(G)
10Dq
Host distance ˚) R0 (A
1
(cm )
(eV)
(cm )
(eV)
KMgCl3:Mn2+
2.47
19 440
2.41
5768
0.715
RbCaCl3:Mn2+
2.66
20 020
2.48
5442
0.675
CsSrCl3:Mn2+ CaCl2:Mn2+
2.80 2.70–2.76
20 240 20 321
2.51 2.52
5196 5073
0.644 0.629
a
1
Mn–Cl distance ˚) R (A 2.512 2.51a 2.541 2.53a 2.565 2.577 ± 0.002
Measured from EXAFS experiments [1].
fore, these results confirm the R5 dependence of 10Dq for manganese doped chloride crystals, in which the Mn2+ is forming octahedral MnCl4 6 complexes. Considering that we are able to detect variations in the peak energies with an accuracy of at least ± 20 cm1, ˚ in the value of the then an accuracy of 2 · 103 A Mn2+–Cl distance seems reasonable. In this way, the value of R in the octahedral MnCl4 complexes of 6 CaCl2:Mn2+ was obtained from the Eq. (3) using 10Dq and R data of either any ABCl3:Mn2+ crystal or NH4MnCl3 standard crystal: !1=5 10DqABCl3 :Mn RCaCl2 :Mn ¼ RABCl3 :Mn ð4aÞ 10DqCaCl2 :Mn
or RCaCl2 :Mn ¼ RNH4 MnCl3
10DqNH4 MnCl3 10DqCaCl2 :Mn
!1=5 :
ð4bÞ
˚ The Mn2+–Cl distance results to be 2.577 ± 0.002 A when is used the ligand field splitting value (10Dq = 5073 cm1) derived with best fit a. Practically ˚ ), within the uncertainty, the same value of R ( = 2.576 A is obtained using the ligand field splitting value (10Dq = 5082 cm1) determined with free ion a. It must be pointed out that this procedure for estimating bond distances from excitation spectra cannot be applied to low symmetry crystals like MnCl2, and therefore, it could not be used as standard crystal. This is because the excitation (absorption) bands show an energy splitting due to the low symmetry ligand field, and hence, the 10Dq parameter determined from excitation spectra does not necessarily obey Eq. (3). 4. Final remarks The R0 host distances and the energies of the first excited state 4T1g(G), which is the most sensitive to the ligand field, for the ABCl3:Mn2+ [14] and CaCl2:Mn2+ crystals are also included in Table 2 in order to accomplish some final remarks. Concerning to the value found for the Mn2+–Cl distance in CaCl2 :Mn2+ from Eq. (4) two important facts must be emphasized with aid of the data collected in Table 2:
˚ ) is smaller (i) The Mn2+–Cl distance (2.577 ± 0.002 A ˚ [16]). than the Ca2+–Cl host distance (2.70–2.76 A Such a reduction of R involves a ligand inward relaxation around the Mn2+, as it also occurs in the RbCaCl3:Mn2+ and CsSrCl3:Mn2+ crystals, since ˚ ) than the Mn2+ ions have a smaller ionic radius (0.8 A ˚ ) or Sr2+ that of the host divalent cation, Ca2+ (0.99 A ˚ ). In fact, a larger ligand inward relaxation is (1.12 A observed in CsSrCl3:Mn2+ with respect to the Ca2+ cation matrices (RbCaCl3:Mn2+ and CaCl2:Mn2+) because of the great ionic radius of Sr2+. An opposite situation occurs for the KMgCl3:Mn2+ crystal, in which takes place a ligand outward relaxation because the Mn2+ ionic radius is slightly larger than ˚ ). Thus, the smallest that of the Mg2+ cation (0.66 A relaxation is observed in KMgCl3:Mn2+. (ii) The red shifts experienced by the first 4T1g(G) band on passing from CaCl2 (20321 cm1) to KMgCl3 (19440 cm1) reflect an increase of 10Dq parameter (from 5073 to 5768 cm1), which is associated with ˚ ). a reduction of R (from 2.577 to 2.512 A Therefore, this good correlation of spectroscopic and structural data of the ABCl3:Mn2+ crystals with those obtained for the CaCl2:Mn2+ crystal supports the present procedure for deriving reasonable bond distances from optical data. This procedure is valid in our case, since the bonding in the MnCl4 6 complexes formed in CaCl2:Mn2+ is greatly ionic as proved by the values close to 1 of the Nt and Ne covalency parameters. 5. Conclusions The good agreement between experimental and calculated energies achieved with the crystal-field transition fitting indicates that the Mn2+ occupies the Ca2+ cation site, forming octahedral MnCl4 complexes. 6 The Mn2+–Cl distance, R, determined from the excitation spectrum by fitting the crystal field transitions ˚ ), is smaller than the Ca2+–Cl host dis(2.577 ± 0.002 A ˚ ). Such a reduction of R involves a litance (2.70–2.76 A gand inward relaxation around the Mn2+ ions, which is ˚ ) with a consequence of their smaller ionic radius (0.8 A ˚ ). The present respect that of the Ca2+ cation (0.99 A
1460
D. Castan˜eda et al. / Optical Materials 27 (2005) 1456–1460
investigation provides that from the crystal-field spectrum can be obtained useful information on the local structure of metal impurities placed in ionic lattices as it is our case.
0 4
D1
B T2 ðG; D; FÞ ¼ B @ D4 D5
D4 D2
D5
1
C D4 C A:
D4
D3
D1 ¼ 5xB0 þ ð10 þ xÞB0 ð1 eÞ þ 8B0 ð1 eÞ 10Dq þ 12a þ b;
Acknowledgement The authors acknowledge the financial support by CONACyT under project contract 43016-F and I. Camarillo for his technical assistance. Appendix A
2
D2 ¼ 3ð3 þ xÞB0 þ 2ð2 þ xÞB0 ð1 eÞ þ 14a; D3 ¼ 10B0 þ 2ð4 þ xÞB0 ð1 eÞ þ 4xB0 ð1 eÞ
2
þ A0 e2 þ 10Dq þ 12a þ b; pffiffiffi pffiffiffi D4 ¼ 6B0 ð1 eÞ1=2 2 6a; D5 ¼ ð4 þ xÞB0 ð1 eÞ þ b:
Energy matrices derived by Mehra [13] for the strongfield scheme, including the a Racah–Trees and b seniority corrections.
A0 ¼ AðB0 =BÞ,
A=1 78 400 cm1 [17]
References 4
A1 ðGÞ ¼ 5ð2 þ xÞB0 ð1 eÞ þ 20a;
4
A2 ðFÞ ¼ 3ð2 þ xÞB0 ð1 eÞ þ 4ð4 þ xÞB0 ð1 eÞ þ 12a þ 2b:
4
EðG; DÞ ¼
2
E1 E3
E3 ; E2
E1 ¼ 3ð3 þ xÞB0 þ 2ð2 þ xÞB0 ð1 eÞ þ 14a; 2
E2 ¼ 3ð2 þ xÞB0 ð1 eÞ þ 2ð4 þ xÞB0 ð1 eÞ þ 12a; pffiffiffi pffiffiffi E3 ¼ 2 3B0 ð1 eÞ 4 3a: 0
P1 B 4 T1 ðG; P; FÞ ¼ @ P4
P4 P2
1 P5 C P4 A:
P5
P4
P3 2
P1 ¼ 5xB0 þ ð2 þ xÞB0 ð1 eÞ þ 8B0 ð1 eÞ 10Dq þ 12a þ b; P2 ¼ ð15 þ 5xÞB0 þ ð4 þ 2xÞB0 ð1 eÞ þ 10a þ 2b; P3 ¼ 10B0 þ 2xB0 ð1 eÞ þ 4xB0 ð1 eÞ2 þ A0 e2 þ 10Dq þ 12a þ b; pffiffiffi pffiffiffi 1=2 P4 ¼ 3 2B0 ð1 eÞ 2 2a; P5 ¼ xB0 ð1 eÞ þ 8a þ b:
[1] M.C. Marco de Lucas, F. Rodrı´guez, C. Prieto, M. Verdaguer, H.U. Gu¨del, J. Phys. Chem. Solids 56 (1995) 995. [2] F. Rodrı´guez, M. Moreno, J. Chem. Phys. 84 (1986) 692. [3] F. Rodrı´guez, M. Moreno, A. Tressaud, J.P. Chaminade, Cryst. Latt. Def. Amorph. Mat. 16 (1987) 221. [4] M.C. MarcodeLucas, F. Rodrı´guez, M. Moreno, J. Phys.: Condens. Matter 5 (1993) 1437. [5] M.T. Barriuso, M. Moreno, Phys. Rev. B 29 (1984) 3623. [6] M.T. Barriuso, M. Moreno, Chem. Phys. Lett. 112 (1984) 165. [7] U. Caldin˜o G., J.O. Rubio, Radiat. Effects Def. Solids 127 (1993) 83. [8] D. Curie, C. Barthou, B. Canny, J. Chem. Phys. 61 (1974) 3048. [9] H. Watanabe, Operator Methods in Ligand Field Theories, Prentice-Hall, New York, 1966, p. 146. [10] S. Koide, M.H.L. Pryce, Phil. Mag. 3 (1958) 607. [11] J. Ramı´rez-Serrano, E. Madrigal, F. Ramos, U. Caldin˜o G., J. Lumin. 71 (1997) 169. [12] D. Herna´ndez, F. Rodrı´guez, M. Moreno, H.U. Gu¨del, Physica B 265 (1999) 186. [13] A.K. Mehra, J. Chem. Phys. 48 (1968) 4384. [14] M.C. Marco de Lucas, F. Rodrı´guez, H.U. Gu¨del, N. Furer, J. Lumin. 60–61 (1994) 581. [15] F. Agullo´-Rueda, J.M. Calleja, F. Jaque, J.D. Tornero, Solid State Commun. 60 (1986) 331. [16] R.W.G. Wycoff, Crystal Structures 1, Interscience, New York, 1965, p. 270 ff. [17] L.L. Lohr, J. Chem. Phys. 55 (1971) 27.
Local structure determination of Mn2+ in CaCl2:Mn2+ by optical spectroscopy D. Castan˜eda, G. Mun˜oz H., U. Caldin˜o* Departamento de Fı´sica, Universidad Auto´noma Metropolitana, Unidad Iztapalapa, P.O. Box 55-534, 09340 Me´xico, D.F., Me´xico Received 26 August 2004; accepted 20 October 2004 Available online 10 December 2004
Abstract Excitation and emission spectra of Mn2+ ions in calcium chloride have been studied using the photoluminescence technique. From a crystal-field analysis of the excitation spectrum is inferred that the Mn2+ ion occupies the Ca2+ cation site forming octahe2+ dral MnCl4 is noticed as a consequence of smaller ionic radius of Mn2+ 6 complexes. A ligand inward relaxation around the Mn than that of host cation. This investigation supports the usefulness of crystal-field spectra for obtaining information on the local structure of impurity ions in ionic lattices. 2004 Elsevier B.V. All rights reserved. PACS: 78.55.m; 78.55.Hx Keywords: Mn2+; CaCl2; Calcium chloride; Photoluminescence; Crystal-field; MnCl4 6
1. Introduction A good characterization of an active ion within an ionic lattice requires to know the local structure around the ion, that is the neighbourhood ions, site symmetry and bond distances, since the optical and magnetic properties of doped crystals depend significantly on the complex formed by the impurity ion and its nearest neighbours. Moreover, the R distance between the impurity ion and its nearest neighbours can be very different from the R0 distance corresponding to the perfect lattice. The R distance has could be determined from a crystal-field analysis of excitation optical spectra [1–4], which has resulted to be the same (within the experimental error) as those previously determined by other techniques, such as the extended X-ray absorption fine structure (EXAFS) [1], and electron paramagnetic resonance (EPR) or electron nuclear double resonance
*
Corresponding author. E-mail address: [email protected] (U. Caldin˜o).
0925-3467/$ - see front matter 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.optmat.2004.10.009
(ENDOR) spectroscopy [5,6] through measuring the isotropic superhyperfine constant. In addition, an accuracy ˚ has been achieved with such an optiof the order 103 A ˚ ) achieved cal procedure in contrast with that (102 A with EXAFS [1], in spite of its continuous improvements in the last years. The goal of the present work is to investigate by means of optical spectroscopy the local structure of Mn2+ ions (site symmetry and bond distances) in CaCl2:Mn2+. The interest in this crystal, with orthorhombic structure, lies in that excitation spectroscopy performed on this phosphor containing Mn2+ ions has revealed that such ions can be excited free of photoconductive process, and hence, their excitation bands have been attributed to Mn2+ internal transitions of the 3d5 configuration [7]. The site symmetry of Mn2+ in CaCl2 could be determined from an analysis of its crystal-field transitions. The Mn2+–Cl distance was obtained from the R-dependence of 10Dq ligand field splitting considering a KRn variation [1–4]. The 10Dq magnitude of Mn2+ in CaCl2:Mn2+ was determined from an analysis of its excitation spectrum
D. Castan˜eda et al. / Optical Materials 27 (2005) 1456–1460
using the procedure developed by Curie et al. [8]. This procedure includes the covalent effects as well as the Racah–Trees or polarization and seniority corrections. Such covalent effects are taken into account through the covalent reduce Racah parameters B 0 and C 0 which represent a measure of the metal–ligand covalency. Thus, in octahedral coordination metal–ligand complexes the simultaneous reduction of the B and C Racah parameters is brought out by the N 4t factor [9]: B0 ¼ BN 4t and C 0 ¼ CN 4t . The Ne normalization factor, which is associated with the Racah parameter reduction in tetrahedral coordination complexes, is related with Nt through the Koide–Pryce covalency parameter e [10]: Ne = Nt(1e)1/2. If such a reduction of B and C or (Nt and Ne) with respect to the free ion values, B = 918 cm1 and C = 3273 cm1 (Nt = Ne = 1) [8] is due to a bonding effect, then the electronic wavefunction within the complex is no longer a pure Mn2+ orbital, but rather a molecular orbital wavefunction, which can be expressed as: jt2g i ¼ N t ðjd xy;yz;xz i kt jLt iÞ
and
jeg i ¼ N e ðjd 3z2r2;x2y2 i ke jLe iÞ;
ð1Þ
where d and L refer to the metal d-orbitals and the pand s-orbital linear combinations, respectively. In this way, Nt and Ne are important bonding parameters, which require to be evaluated in order to obtain some information on the bonding ionicity, since weaker the mixing of metal and ligand functions, Nt and Ne become close to 1.
2. Experimental A colourless single crystal of CaCl2 doped with Mn2+ ions was grown in our laboratory in an evacuated quartz ampoule using the Bridgman technique. Starting materials of very high purity were previously dehydrated by slow heating to 280 C under mechanical vacuum. Due to the quite hygroscopic character of calcium chloride, the optical spectroscopy of CaCl2:Mn2+ was performed with the crystal contained into an evacuated quartz ampoule to avoid adsorption and inclusion of water. The very small oscillator strengths ( 108) of the Mn2+ ion crystal-field transitions make difficult to detect photoluminescence spectra in crystals doped with low concentration of Mn2+. Thus, the manganese concentration in the melt was around 3200 ppm, which was the lowest one so that the Mn2+ luminescence could be detected with our experimental equipment. The lowest concentration possible is employed to prevent dimer formation, as well migration processes and exchange interactions, which can significantly affect the optical spectra. Emission and excitation spectra were obtained with the same experimental equipment described in [7]. The
1457
excitation spectra were not corrected for the photomultiplier response, since we were only concerned with the peak positions and not with the band intensities. Lifetime measurements were performed exciting the crystal at 337 nm, with 10-ns pulses of a PRA LN120C UV laser. The resulting transient fluorescence signal was analysed with a 0.45 m Czerny–Turner monochromator, detected with a cooled Hamamatsu R943–03 photomultiplier tube, and processed by a Hewlett Packard model 54201A digitising oscilloscope.
3. Experimental results: analysis Fig. 1 displays the excitation and emission spectra of Mn2+ ions in CaCl2:Mn2+ at room-temperature (RT). The emission spectrum, obtained under excitation at 414 nm, consists of a broad band peaking at 578 nm, which is associated with the 4T1g(G) ! 6A1g(S) spin-forbidden transition of Mn2+ ions. In fact, the time dependence of this emission is exponential, with a lifetime of about 21 ms, which proves the forbidden character of the transition. The excitation spectrum, monitored at 580 nm, was found to be very similar to those exhibited by Mn2+ ions in octahedral coordination chlorides, such as CdCl2:Mn2+ [11], NH4MnCl3 [12], KMgCl3:Mn2+ [1], KCaCl3:Mn2+ [1] and CsSrCl3:Mn2+ [1]. It consists of several absorption bands, which are associated with Mn2+ transitions from its 6A1g(S) ground state to the 4 T1g(G), 4T2g(G), [4A1g(G), 4Eg(G)], 4T2g(D), 4Eg(D), 4 T1g(P), 4A2g(F), 4T1g(F) and 4T2g(F) levels, since excitation into any of these bands gives rise to the 4 T1g(G) ! 6A1g(S) emission. The same excitation spectra are obtained regardless of the monitored luminescence wavelength, which indicates that all Mn2+ ions are excited via the same excitation bands. The most
Fig. 1. RT emission and excitation spectra of CaCl2:Mn2+. The emission spectrum was obtained after excitation at 414 nm and the excitation one was monitored at 580 nm.
1458
D. Castan˜eda et al. / Optical Materials 27 (2005) 1456–1460
intense excitation band, peaking at 414 nm, and within which was obtained the emission spectrum, corresponds to the 6A1g(S) ! [4A1g(G), 4Eg(G)] transition. The position of this transition does not essentially depend upon the crystal-field strength. The energies of the Mn2+ crystal-field transitions displayed in the excitation spectrum (Fig. 1) were fitted using the procedure developed by Curie et al. [8], in which the adjustable parameters are the crystal-field splitting 10Dq, the covalent reduced Racah parameter B 0 and the Koide–Pryce [10] covalency parameter e. The values for the a Racah–Trees and b seniority corrections and for x ( = C/B = C 0 /B 0 ) were taken from those derived for the free Mn2+ ion (a = 65 cm1, b = 131 cm1 and x = 3.565 [8]). The energy matrices used in the fitting (which are listed in the Appendix A) were those derived by Mehra [13] for the strong-field scheme, but including the corrections mentioned above. A better fitting is achieved taking also a as adjustable parameter, so that if the covalent effect upon the Racah–Trees correction is taken into account, then a is larger in the crystal than for the free ion. The calculated (with best fit a) and experimental energies in cm1 and eV for the Mn2+ excitation transitions displayed in Fig. 1 are given in Table 1. The inclusion of transition energies in eV results proper for readers used to dealing with this energy unit. The energies of the 4 T1g(F) and 4T2g(F) levels were not taken into account in the fitting since they appear overlapped at RT. The values obtained for the parameters 10Dq, B 0 and e, as well as for the normalization factors Nt and Ne, are also included in the Table. Ne is smaller than Nt in concordance with the greater bonding character of the eg-orbitals (Eq. (1)). The 10Dq ligand field splitting obtained (5073 cm1) with a best fit value for a ( = 85.4 cm1) resulted to be very similar to that (5082 cm1) derived with free ion a. A 10Dq value slightly larger (5250 cm1) has been previously reported in CaCl2:Mn2+ (10000 ppm) crystals [7], which can be attributed to the very high concentration of Mn2+ ions in those crystals. The quality of our fitting was estimated using the quantity X defined by:
"
N X ðEjth Ejexp Þ2 X¼ N j¼1
#1=2 ð2Þ
;
where j runs over the N fitted transitions, and Ejth and Ejexp represent the theoretical and experimental energies of the crystal-field transitions labelled by j. X resulted to be 250.6 cm1 with a = 65 cm1 and 210.9 cm1 with a = 85.4 cm1, that is, considering the covalent effect on a the theoretical levels fit even better to the observed levels. This good agreement between experimental and calculated values supports the assignment of the crystal-field transitions made in Fig. 1, and hence, it can be inferred that the Mn2+ occupies the divalent cation site, forming octahedral MnCl4 6 complexes. Moreover, the fact that Nt and Ne resulted to be close to 1 (Table 1) suggests a high ionicity of the MnCl4 complexes formed in 6 CaCl2:Mn2+. The Mn2+–Cl distance in MnCl4 complexes of 6 manganese doped orthorhombic structure crystals such as KMgCl3:Mn2+, RbCaCl3:Mn2+ and CsSrCl3:Mn2+ (ABCl3:Mn2+) has been determined from the Mn2+ excitation spectrum by assuming that the 10Dq ligand field splitting is proportional to R5 [1,14]. In such chloride crystals the Mn2+ ion substitutes the divalent cation (Mg2+, Ca2+ or Sr2+) forming octahedral MnCl4 6 units. Table 2 lists data of 10Dq parameter derived from excitation spectra for these ABCl3:Mn2+ crystals, as well as the Mn2+–Cl distances determined from the relation: 10Dq ¼ KR5
˚ ðunits in cm1 and AÞ;
ð3Þ
using 10Dq and R data of a NH4MnCl3 crystal as a standard [1,14]. The use of NH4MnCl3 data (10Dq = ˚ ) is suitable since R is 5620 cm1 and R = R0 = 2.525 A equal to the R0 host distance, and moreover, such a crystal gives a perfect octahedral environment to Mn2+ [15]. Although the assumption of R5 dependence of ligand field splitting (predicted by the crystal field theory) gives rise to 10Dq values substantially smaller than the experimental ones [2], a very good agreement between Mn2+–Cl distances obtained by EXAFS and optical excitation spectroscopy has been found for KMgCl3: Mn2+ [1] and RbCaCl3:Mn2+ [1] (see Table 2). There-
Table 1 Experimental and fitting energies of the crystal-field transitions of Mn2+ in CaCl2:Mn2+ at RT and values of the fitting characteristic parameters Transitions
Experimental energies
6
(cm1)
(eV)
(cm1)
(eV)
20 321 23 364 24 154 24 154 27 747 28 571 29 913 37 425
2.52 2.90 3.00 3.00 3.44 3.54 3.71 4.64
19 942 23 571 24 063 24 158 28 076 28 576 29 688 37 473
2.47 2.92 2.98 3.00 3.48 3.54 3.68 4.65
4
A1g(S) !
T1g(G) T2g(G) 4 A1g(G) 4 Eg(G) 4 T2g(D) 4 Eg(D) 4 T1g(P) 4 A2g(F) 4
Fitting energies
Fitting parameters
10Dq = 5073 cm1 (0.629 eV) B 0 = 838.3 cm1 e = 0.0208 a = 85.4 cm1 Nt = 0.978 Ne = 0.968
D. Castan˜eda et al. / Optical Materials 27 (2005) 1456–1460
1459
Table 2 R0 host distance, 4T1g(G) first excited state energy, 10Dq ligand field splitting and R Mn2+–Cl distance in ABCl3:Mn2+ and CaCl2:Mn2+ crystals Crystal
4
T1g(G)
10Dq
Host distance ˚) R0 (A
1
(cm )
(eV)
(cm )
(eV)
KMgCl3:Mn2+
2.47
19 440
2.41
5768
0.715
RbCaCl3:Mn2+
2.66
20 020
2.48
5442
0.675
CsSrCl3:Mn2+ CaCl2:Mn2+
2.80 2.70–2.76
20 240 20 321
2.51 2.52
5196 5073
0.644 0.629
a
1
Mn–Cl distance ˚) R (A 2.512 2.51a 2.541 2.53a 2.565 2.577 ± 0.002
Measured from EXAFS experiments [1].
fore, these results confirm the R5 dependence of 10Dq for manganese doped chloride crystals, in which the Mn2+ is forming octahedral MnCl4 6 complexes. Considering that we are able to detect variations in the peak energies with an accuracy of at least ± 20 cm1, ˚ in the value of the then an accuracy of 2 · 103 A Mn2+–Cl distance seems reasonable. In this way, the value of R in the octahedral MnCl4 complexes of 6 CaCl2:Mn2+ was obtained from the Eq. (3) using 10Dq and R data of either any ABCl3:Mn2+ crystal or NH4MnCl3 standard crystal: !1=5 10DqABCl3 :Mn RCaCl2 :Mn ¼ RABCl3 :Mn ð4aÞ 10DqCaCl2 :Mn
or RCaCl2 :Mn ¼ RNH4 MnCl3
10DqNH4 MnCl3 10DqCaCl2 :Mn
!1=5 :
ð4bÞ
˚ The Mn2+–Cl distance results to be 2.577 ± 0.002 A when is used the ligand field splitting value (10Dq = 5073 cm1) derived with best fit a. Practically ˚ ), within the uncertainty, the same value of R ( = 2.576 A is obtained using the ligand field splitting value (10Dq = 5082 cm1) determined with free ion a. It must be pointed out that this procedure for estimating bond distances from excitation spectra cannot be applied to low symmetry crystals like MnCl2, and therefore, it could not be used as standard crystal. This is because the excitation (absorption) bands show an energy splitting due to the low symmetry ligand field, and hence, the 10Dq parameter determined from excitation spectra does not necessarily obey Eq. (3). 4. Final remarks The R0 host distances and the energies of the first excited state 4T1g(G), which is the most sensitive to the ligand field, for the ABCl3:Mn2+ [14] and CaCl2:Mn2+ crystals are also included in Table 2 in order to accomplish some final remarks. Concerning to the value found for the Mn2+–Cl distance in CaCl2 :Mn2+ from Eq. (4) two important facts must be emphasized with aid of the data collected in Table 2:
˚ ) is smaller (i) The Mn2+–Cl distance (2.577 ± 0.002 A ˚ [16]). than the Ca2+–Cl host distance (2.70–2.76 A Such a reduction of R involves a ligand inward relaxation around the Mn2+, as it also occurs in the RbCaCl3:Mn2+ and CsSrCl3:Mn2+ crystals, since ˚ ) than the Mn2+ ions have a smaller ionic radius (0.8 A ˚ ) or Sr2+ that of the host divalent cation, Ca2+ (0.99 A ˚ ). In fact, a larger ligand inward relaxation is (1.12 A observed in CsSrCl3:Mn2+ with respect to the Ca2+ cation matrices (RbCaCl3:Mn2+ and CaCl2:Mn2+) because of the great ionic radius of Sr2+. An opposite situation occurs for the KMgCl3:Mn2+ crystal, in which takes place a ligand outward relaxation because the Mn2+ ionic radius is slightly larger than ˚ ). Thus, the smallest that of the Mg2+ cation (0.66 A relaxation is observed in KMgCl3:Mn2+. (ii) The red shifts experienced by the first 4T1g(G) band on passing from CaCl2 (20321 cm1) to KMgCl3 (19440 cm1) reflect an increase of 10Dq parameter (from 5073 to 5768 cm1), which is associated with ˚ ). a reduction of R (from 2.577 to 2.512 A Therefore, this good correlation of spectroscopic and structural data of the ABCl3:Mn2+ crystals with those obtained for the CaCl2:Mn2+ crystal supports the present procedure for deriving reasonable bond distances from optical data. This procedure is valid in our case, since the bonding in the MnCl4 6 complexes formed in CaCl2:Mn2+ is greatly ionic as proved by the values close to 1 of the Nt and Ne covalency parameters. 5. Conclusions The good agreement between experimental and calculated energies achieved with the crystal-field transition fitting indicates that the Mn2+ occupies the Ca2+ cation site, forming octahedral MnCl4 complexes. 6 The Mn2+–Cl distance, R, determined from the excitation spectrum by fitting the crystal field transitions ˚ ), is smaller than the Ca2+–Cl host dis(2.577 ± 0.002 A ˚ ). Such a reduction of R involves a litance (2.70–2.76 A gand inward relaxation around the Mn2+ ions, which is ˚ ) with a consequence of their smaller ionic radius (0.8 A ˚ ). The present respect that of the Ca2+ cation (0.99 A
1460
D. Castan˜eda et al. / Optical Materials 27 (2005) 1456–1460
investigation provides that from the crystal-field spectrum can be obtained useful information on the local structure of metal impurities placed in ionic lattices as it is our case.
0 4
D1
B T2 ðG; D; FÞ ¼ B @ D4 D5
D4 D2
D5
1
C D4 C A:
D4
D3
D1 ¼ 5xB0 þ ð10 þ xÞB0 ð1 eÞ þ 8B0 ð1 eÞ 10Dq þ 12a þ b;
Acknowledgement The authors acknowledge the financial support by CONACyT under project contract 43016-F and I. Camarillo for his technical assistance. Appendix A
2
D2 ¼ 3ð3 þ xÞB0 þ 2ð2 þ xÞB0 ð1 eÞ þ 14a; D3 ¼ 10B0 þ 2ð4 þ xÞB0 ð1 eÞ þ 4xB0 ð1 eÞ
2
þ A0 e2 þ 10Dq þ 12a þ b; pffiffiffi pffiffiffi D4 ¼ 6B0 ð1 eÞ1=2 2 6a; D5 ¼ ð4 þ xÞB0 ð1 eÞ þ b:
Energy matrices derived by Mehra [13] for the strongfield scheme, including the a Racah–Trees and b seniority corrections.
A0 ¼ AðB0 =BÞ,
A=1 78 400 cm1 [17]
References 4
A1 ðGÞ ¼ 5ð2 þ xÞB0 ð1 eÞ þ 20a;
4
A2 ðFÞ ¼ 3ð2 þ xÞB0 ð1 eÞ þ 4ð4 þ xÞB0 ð1 eÞ þ 12a þ 2b:
4
EðG; DÞ ¼
2
E1 E3
E3 ; E2
E1 ¼ 3ð3 þ xÞB0 þ 2ð2 þ xÞB0 ð1 eÞ þ 14a; 2
E2 ¼ 3ð2 þ xÞB0 ð1 eÞ þ 2ð4 þ xÞB0 ð1 eÞ þ 12a; pffiffiffi pffiffiffi E3 ¼ 2 3B0 ð1 eÞ 4 3a: 0
P1 B 4 T1 ðG; P; FÞ ¼ @ P4
P4 P2
1 P5 C P4 A:
P5
P4
P3 2
P1 ¼ 5xB0 þ ð2 þ xÞB0 ð1 eÞ þ 8B0 ð1 eÞ 10Dq þ 12a þ b; P2 ¼ ð15 þ 5xÞB0 þ ð4 þ 2xÞB0 ð1 eÞ þ 10a þ 2b; P3 ¼ 10B0 þ 2xB0 ð1 eÞ þ 4xB0 ð1 eÞ2 þ A0 e2 þ 10Dq þ 12a þ b; pffiffiffi pffiffiffi 1=2 P4 ¼ 3 2B0 ð1 eÞ 2 2a; P5 ¼ xB0 ð1 eÞ þ 8a þ b:
[1] M.C. Marco de Lucas, F. Rodrı´guez, C. Prieto, M. Verdaguer, H.U. Gu¨del, J. Phys. Chem. Solids 56 (1995) 995. [2] F. Rodrı´guez, M. Moreno, J. Chem. Phys. 84 (1986) 692. [3] F. Rodrı´guez, M. Moreno, A. Tressaud, J.P. Chaminade, Cryst. Latt. Def. Amorph. Mat. 16 (1987) 221. [4] M.C. MarcodeLucas, F. Rodrı´guez, M. Moreno, J. Phys.: Condens. Matter 5 (1993) 1437. [5] M.T. Barriuso, M. Moreno, Phys. Rev. B 29 (1984) 3623. [6] M.T. Barriuso, M. Moreno, Chem. Phys. Lett. 112 (1984) 165. [7] U. Caldin˜o G., J.O. Rubio, Radiat. Effects Def. Solids 127 (1993) 83. [8] D. Curie, C. Barthou, B. Canny, J. Chem. Phys. 61 (1974) 3048. [9] H. Watanabe, Operator Methods in Ligand Field Theories, Prentice-Hall, New York, 1966, p. 146. [10] S. Koide, M.H.L. Pryce, Phil. Mag. 3 (1958) 607. [11] J. Ramı´rez-Serrano, E. Madrigal, F. Ramos, U. Caldin˜o G., J. Lumin. 71 (1997) 169. [12] D. Herna´ndez, F. Rodrı´guez, M. Moreno, H.U. Gu¨del, Physica B 265 (1999) 186. [13] A.K. Mehra, J. Chem. Phys. 48 (1968) 4384. [14] M.C. Marco de Lucas, F. Rodrı´guez, H.U. Gu¨del, N. Furer, J. Lumin. 60–61 (1994) 581. [15] F. Agullo´-Rueda, J.M. Calleja, F. Jaque, J.D. Tornero, Solid State Commun. 60 (1986) 331. [16] R.W.G. Wycoff, Crystal Structures 1, Interscience, New York, 1965, p. 270 ff. [17] L.L. Lohr, J. Chem. Phys. 55 (1971) 27.