Optical Materials 28 (2006) 395–400 www.elsevier.com/locate/optmat
Optical spectra of Er3+ in Ba2NaNb5O15 single crystals S. Bigotta
a,*
, G. Gorini a, A. Toncelli a, M. Tonelli a, E. Cavalli b, E. Bovero
b
a
b
NEST-CNR, Dipartimento di Fisica, Universita` di Pisa, Largo Pontecorvo 3, 56127 Pisa, Italy Dipartimento di Chimica Generale ed Inorganica, Chimica Analitica e Chimica Fisica, Universita` di Parma, Viale delle Scienze, 43100 Parma, Italy Received 9 August 2004; accepted 21 September 2004 Available online 31 March 2005
Abstract The absorption and luminescence spectra of Er3+-doped Ba2NaNb5O15 single crystals have been measured at 10 and 298 K. The lifetimes of the lower lying excited states, emitting in the visible and NIR regions, have been measured as a function of the temperature. From the oscillator strengths of the absorption transitions, the Judd–Ofelt intensity parameters have been evaluated. The spontaneous transition probabilities, the branching ratios and the radiative lifetimes have been determined for the most important emitting states using the calculated intensity parameters and the results have been compared with the experimental data. The stimulated emission cross section have been estimated for the 1.5 lm transition. These experimental results are discussed in order to evaluate the potentialities of this material as active medium in solid state laser devices. 2005 Elsevier B.V. All rights reserved. PACS: 42.70.Hj; 78.55.Hx Keywords: Rare earth compounds; Barium sodium niobate; Optical spectroscopy; Luminescence; Laser material
1. Introduction Ba2NaNb5O15 (BNN) is a ferroelectric crystal having the tungsten bronze (TB) structure and good electrooptical and non linear optical properties, which make it interesting for technological applications like frequency doubling, optical parametric oscillation, optical storage of information [1,2]. Doping with Nd3+ and Yb3+ has yielded promising active media for selfdoubling solid state laser operation [3–7]. Moreover, the broadband emission properties of these materials can be exploited for both CW wide-tunability oscillators and short pulse generation. The growth of single crystals of BNN is not easy, and it is probably for this reason that the investigations have not yet extended to
*
Corresponding author. Tel.: 39 050 2214536; fax: 39 050 2214333. E-mail address:
[email protected] (S. Bigotta).
0925-3467/$ - see front matter 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.optmat.2004.09.026
other optically active ions [8]. We have developed a low cost flux growth method allowing to obtain rare earth-doped single crystals of BNN suitable for spectroscopy experiments. In this paper we present the results of a study on the optical spectra of Er3+-doped BNN (hereafter Er:BNN). The interest in the development of solid state lasers based on Er3+ doped crystals lies in their possible use in medicine, telecommunication, metrology. It is well known that the characterization of a material by optical spectroscopy techniques is fundamental in order to assess its perspectives of application.
2. Experimental Pale pink crystals of BNN were grown by means of the Ôflux growthÕ technique, using sodium tetraborate as solvent. The doping ions were added as Er2O3 with
S. Bigotta et al. / Optical Materials 28 (2006) 395–400
a nominal concentration of 10% Er/Ba molar ratio and 4% Er/Na (consider that Na+ is a component of the flux). The details of the crystals growth procedure will be presented elsewhere. The average size of the crystals was about 3 · 2 · 2 mm3; their optical quality was good, and their crystallographic orientation was checked by X-Ray Laue technique. Their structure and the doping level were analyzed by means of X-ray diffraction (XRD) associated to selected-area electron diffraction (SAED) technique. The spectroscopic measurements were carried out on optically polished samples. Room temperature Raman spectra were measured on a JobinYvon Ramanor spectrometer followed by an EMI 9884B photomultiplier and a photon counting system. The sample was irradiated with 150 mW of the 458 nm line from an argon laser with a power density of about 2.5 W/cm2. The wavenumber precision was 1 cm1. The absorption spectra were measured with a spectroscopic system equipped with a 300 W halogen lamp fitted with a 0.25 m Spex monochromator as source, and a 1.26 m Spex monochromator with an RCA C31034 photomultiplier or a PbS N.E.P cell to analyze and detect the output radiation. The resolution was 0.1 nm in the visible and 1 nm in the NIR. The luminescence spectra in the visible region (440–830 nm) were measured using the same set-up with the optical pathway properly modified and an Ar+ ion laser as excitation source. The infrared emission (1.5–3 lm) was measured by exciting the sample using an Ar+ ion laser or at 970 nm by means of a diode laser. The samples were mounted onto the cold finger of He-closed cycle cryostat and the measurements carried out at temperatures ranging from 10 to 298 K. The emission was mechanically chopped and focused by a lens on the input slit of a computer controlled monochromator having 0.25 m focal length and detected by a cooled InSb detector. The signal was collected orthogonally to the direction of the exciting beam, a suitable filter and a GlanThompson polarizer (below 2.3 lm) or ZnSe filter at Brewster angle (above 2.3 lm) were put in front of the monochromator. The signal was processed by a lock-in amplifier. The optical response of the system was normalized for both polarizations by use of a black-body source at 3000 K temperature. The resolution was 0.1 nm in the visible, 1 nm in the NIR and 10 nm at 3 lm. The luminescence was detected by means of a Philips XP1002 photomultiplier with S-20 spectral response (visible) or a cooled Hamamatsu R316 with S-1 spectral response (NIR). For the lifetime measurements the setup was similar, but the sample was pumped by either a 10 Hz pulsed Ti:Al2O3 with 30 ns pulse width or its doubled emission, the luminescence was collected from a small portion of the crystal and the signal from the detector was sent to a digital oscilloscope. The system response time was about 1 ls.
3. Structural analysis and absorption spectroscopy Er:BNN is tetragonal, with space group P4bm and ˚ . The abcell parameters [9] a = 12.463 and c = 3.990 A sence of the orthorhombic distortion (space group Cmm2) typical of undoped or slightly doped BNN crystals grown by Czochralski or LHPG techniques [7,10] has been checked by Raman experiments. In Fig. 1 we report the Raman spectrum of Er:BNN at the x(zz)y scattering geometry. The two strong bands peaking at 282 and 649 cm1 are assigned to the m2 (Eg, in Oh notation) and m5 (T2g) vibrational modes of the NbO7 6 octahedral units. They are expected to split into two and three components, respectively, in the case of crystals belonging to the orthorhombic system [11,12]. In the present case the bands are broad but not split, confirming that our crystals are tetragonal. The third m1 (A1g) mode of NbO7 6 in this spectrum gives rise only to a very weak band [13] around 850 cm1. The TB structure can be considered as a framework of NbO7 octahedra 6 linked through the corners, the voids among them describing cavities with pentagonal (A1), square (A2) and triangular (C) sections [1]. The small C sites are mostly empty, whereas the A2 and A1 sites, having 12fold and 15-fold oxygen coordination, are usually occupied by the Na+ and Ba2+ ions, respectively. The site occupancy analysis carried out on the basis of the XRD data has indicated that the Er3+ replaces Na+ in the 10% of the A2 sites of BNN, as in the case of Nd:BNN [9]. This corresponds to a concentration of 3.2 · 1020 ions/cm3. The excess of charge arising from this substitution is compensated mainly by cation vacancies: only the 91% of the A1 sites is in fact occupied by Ba2+. Therefore, each A1 site in the ab plane is surrounded by four A2 sites and one or even two of them have finite probabilities to be empty. The presence of
x(zz)y
Intensity (a. u.)
396
200
400
600
Raman shift
(cm-1)
800
Fig. 1. Polarized Raman spectrum of Er:BNN.
1000
S. Bigotta et al. / Optical Materials 28 (2006) 395–400
cation vacancies in the second coordination sphere affects the local geometry and the electron density, i.e., the crystal field, around the optically active ion. As a consequence, there will be different non equivalent Er3+ ions concurring to the spectroscopic properties of the material. Moreover, SAED studies performed on our samples showed, in agreement with literature data [2,14], that the BNN structure is affected by quasicommensurate structural modulation (with a modulation wave vector about four times the (1 1 0) direction of the fundamental tetragonal TB subcell). This modulation, generated by the cooperative tilting of the NbO7 6 octahedra building the crystal structure, implies, as a matter of fact, a certain degree of disorder in the orientation and/or shape of the coordination polyhedra, and then in the crystal field around the active ions. This will result in the inhomogeneous broadening of the optical features even in the low temperature spectra. The 10 K polarized absorption spectra of Er:BNN are shown in Fig. 2. The transitions from the 4I15/2 ground state to the lowest excited states of Er3+ have been assigned on the basis of the energy level diagram reported in literature [15]. The wavelengths of the observed transitions are reported in Table 1. The full width at half maximum (FWHM) of the components of the multiplets ranges from 20 to 50 cm1, an intermediate value between that typical for ordered crystals (1–10 cm1) and for glasses (100 cm1). It is interesting to note that the p polarized spectrum is in general more intense than the r one. At room temperature (RT) the absorption bands (not shown here for brevity) are further broadened and less intense, and several hot lines appear and often collapse into broad bands. The RT absorption spectra have been analyzed in the framework of the Judd–Ofelt (J–O) theory [16,17]. The experimental oscillator strengths were reliably determined and fitted on the basis of the Judd–Ofelt parameterization scheme after subtraction
397
Table 1 Line wavelengths observed in the 10 K absorption spectrum Assignment 2
4
2
( G, F, H)9/2 4 F3/2 + 4F5/2 4 F7/2 2 H11/2 4 4
S3/2 F9/2
4
I9/2
4
I11/2 I13/2
4
Wavelength (nm) 403.0, 404.5, 406.5, 407.8 440.2, 442.2, 448.6, 450.1, 452.9 483.7, 485.0, 485.8, 486.3, 488.0, 489.0, 493.3 514.4, 516.4, 518.1, 519.9, 521.0, 521.4, 522.2, 523.9, 527.5, 531.3 537.8, 538.8, 540.3, 541.0, 542.6 645.1, 647.6, 648.4, 650.1, 651.8, 653.0, 655.0, 655.8, 660.2 760.3, 777.9, 783.7, 788.1, 792.4, 795.4, 801.9, 806.5, 813.3, 823.3 959.5, 961.1, 964.0, 966.7, 969.5, 972.3 1468.2, 1474.0, 1476.9, 1480.3, 1488.9, 1495.9, 1501.6, 1508.1, 1512.8, 1524.1, 1531.5
of the magnetic dipole contributions for the 4I13/2 4 4 I15/2 and the 4G9/2 + 2K15/2 + 2G7/2 I15/2 transitions [18]. The reduced matrix elements reported by Kaminskii et al. [15] and a mean value (n = 2.33) of the refractive index determined by Singh et al. [2] were employed. The calculated oscillator strengths, the intensity parameters and the root mean square deviation (RMS) are reported in Table 2. The calculated spontaneous emission probabilities, the radiative branching ratios and the radiative lifetimes, were estimated using the calculated intensity parameters and are reported in Table 3. The spontaneous emission probabilities were calculated taking into account the magnetic dipole contributions, using the values reported [19] for LaF3 and correcting for the refractive index. In Table 4 the values of the intensity parameters for Er:BNN are compared with the values obtained for several laser crystals containing Er3+. The table evidences remarkable differences in the values of the X2 parameter, that is connected to the intensity of the 2H11/2 4I15/2 hypersensitive absorption transition and does not significantly affect the emission properties in the infrared. More interesting, even if less evident, could be the differences in the X6 parameter, because it could directly affect the intensities of radiative emissions. The value obtained for BNN is among the
Table 2 Experimental and calculated oscillator strengths (P) of Er:BNN Transition
Barycenter (cm1)
Pexp (106)
Pcalc (106)
4
6605 10237 12427 15255 18425 19145 20472 22298 24549
1.21 0.24 0.37 1.92 0.70 6.46 2.27 1.10 0.64
1.22 0.56 0.34 2.02 0.46 6.43 1.89 0.88 0.69
4
I13/2 I15/2 I11/2 4I15/2 4 I9/2 4I15/2 4 4 F9/2 I15/2 4 4 S3/2 I15/2 2 H11/2 4I15/2 4 4 F7/2 I15/2 4 4 F3/2 + F5/2 4I15/2 (2G4F2H)9/2 4I15/2 4
Fig. 2. 10 K absorption spectrum of Er:BNN.
X2 = 2.07 · 1020 cm2, X4 = 0.84 · 1020 cm2, X6 = 0.63 · 1020 cm2; RMS = 2.47 · 107; %RMS:14.9.
398
S. Bigotta et al. / Optical Materials 28 (2006) 395–400
Table 3 Calculated spontaneous emission probabilities (A), radiative branching ratios (b) and radiative lifetimes (sR) of Er:BNN at room temperature Initial state
Final state
A [s ]
b
2
4
0.06 42 154 436 430 11050 1.19 115 74 918 2183 14 144 136 2637 6 101 268 37 265 200
4.1 · 106 0.003 0.013 0.036 0.035 0.912 0.0004 0.035 0.023 0.279 0.663 0.005 0.049 0.046 0.900 0.016 0.269 0.715 0.196 0.804 1
H11/2
4
S3/2
4
F9/2
4
I9/2
4
I11/2
4
I13/2
S3/2 4 F9/2 4 I9/2 4 I11/2 4 I13/2 4 I15/2 4 F9/2 4 I9/2 4 I11/2 4 I13/2 4 I15/2 4 I9/2 4 I11/2 4 I13/2 4 I15/2 4 I11/2 4 I13/2 4 I15/2 4 I13/2 4 I15/2 4 I15/2
-2
10 K
I 10-3
298 K
83
304
10
-4
0.2
0.4 ms
298 K
0.6
π σ
339 520
530
540
550
560
570
Wavelength (nm) 2644
Fig. 3. 10 and 298 K emission in the 550 nm region.
3042 3110
Table 4 Comparison between the intensity parameters of Er3+ doped in some oxide laser crystals and in BNN
LiNbO3 Ca2Al2SiO7 YAlO3 YAG CaSGG YVO4 GdVO4 BNN
10
sR [ls]
Intensity (a. u.)
1
10 K
X2 (1020 cm2)
X4 (1020 cm2)
X6 (1020 cm2)
Ref.
7.29 3.85 0.95 0.45 12.2 13.45 12.6 2.07
2.24 1.85 0.58 0.98 7.38 2.23 2.5 0.84
1.27 1.0 0.55 0.62 5.1 1.67 0.76 0.63
[20] [21] [22] [23] [24] [25] [26] [this work]
lowest reported, even if it is similar to that of other laser crystals such as YAG and YAlO3.
4. Luminescence spectroscopy and decay kinetics The Er3+ ions in solid lattices give rise to a number of emission channels in the visible and NIR. We have focused our attention on the polarized luminescence properties in the 550 nm, 1.5 and 2.7 lm regions, as they are attractive from a technological point of view. The green 4S3/2 ! 4I15/2 emission is of interest in order to develop compact laser devices for data storage and display applications. It has been measured after laser excitation at 488 nm and it is shown in Fig. 3 with two different crystal orientations, Ekc (p) and E ? c (r). At 10 K we can see that at least 11 lines are located in the 540–560 nm range. As the theoretical number of lines is 8, this result indicates that two or more non
equivalent active centers contribute to this emission system. When the temperature is raised to 298 K, these components broaden up and collapse into a band with maximum at 542.5 nm. Moreover, a rather weak broadband appears in the 515–535 nm range, assigned to the emission from the thermally populated 2H11/2 multiplet, that is separated from 4S3/2 by a minimum gap of only 230 cm1. The decay curves have been measured as a function of the temperature after pulsed excitation at 440 nm. The decay profile at low temperatures is a single exponential, whereas at 298 K it is clearly non exponential at short times (see inset of Fig. 3) but it shows a long exponential tail. This suggests the presence of thermally activated energy transfer processes. The decay times obtained from the fit of the exponential part range from 90 ls to 70 ls at 10 K and 300 K respectively. The experimental room temperature decay time approaches the J–O theoretical lifetime of the 2H11/2 level (83 ls) and is consistent with the thermalization of the 2H11/2 and 4S3/2 states. However, the lifetime of the 4S3/2 (304 ls) level calculated by the J–O theory is longer than the decay measured at 10 K, thus confirming that non radiative processes strongly affect the visible luminescence. Among these, the multiphonon relaxation to the lower lying 4F9/2 state is certainly important, since the energy gap between this and the emitting level, about 2930 cm1, can be bridged by 3–4 phonons, the highest phonon energy present in the Raman spectrum (Fig. 1) being of the order [11,12] of 850 cm1. Further, other non radiative processes are active, such as the crossrelaxation processes [27] 4S3/2 ! 4I9/2; 4I15/2 ! 4I13/2. The 4I13/2 ! 4I15/2 luminescence at 1.5 lm is interesting for telecommunication applications and it is shown in Fig. 4 for both p and r polarizations. The spectra have been measured by exciting the sample at 970 nm. The contribution of different centers results in
S. Bigotta et al. / Optical Materials 28 (2006) 395–400
10 K
6
π Abs. cross section (10-21 cm2)
Intensity (a. u.)
10-1 10 K I 10
-2
10
-3
298 K
5
10 15 ms
5
Em. cross section (10-21 cm2)
σ
298 K
399
4
3
2
1
20
0
1400
1450
1500
1550
1600
1650
1700
1.42 1.44 1.46 1.48 1.50 1.52 1.54 1.56 1.58 1.60 1.62 1.64 1.66
Wavelength (micron)
Wavelength (nm)
a broadband extending from 1450 to 1650 nm at room temperature, and in a structured emission system having at least 15 components at 10 K, much more than the eight lines expected from the Stark splitting of the ground state. In this spectrum we found that the ground multiplet is characterized by an over-all splitting of 600 cm1 and several peaks are split into multiple lines. These results confirm that at least two non equivalent Er3+ ions are involved in the radiative process. The fluorescence lifetimes have been measured as a function of the sample temperature after pulsed excitation at 969 nm. The decay curves are single exponential and yield decay times decreasing from 4.1 ms at 10 K to 3.2 ms at 298 K (see inset of Fig. 4). The rise times observed at the beginning of both decay profiles allow to infer that the 4I13/2 level is mainly populated by relaxation from a relatively long-lived upper state, probably 4 I11/2 (see after). The room temperature decay time is very similar to the calculated lifetime of the emitting level (see Table 3). This points towards a very low efficiency of the non radiative processes, even if it has to be outlined that sometimes the experimental decay time of 4I13/2 is overestimated, probably as a consequence of radiation trapping processes. The effective stimulated emission cross-section for the 4I13/2 level was evaluated on the basis of the luminescence data and the measured decay time by the so-called b–s method [28]. The comparison between polarized absorption and emission cross sections is shown in Fig. 5. The peak emission cross-section, 0.57 · 1020 cm2 at 1538 nm, is comparable to those of other laser hosts [29]. The absorption and emission features strongly overlap up to 1550 nm, whereas at longer wavelengths the reabsorption is considerably lower specially in p polarization. The emission cross-section at 1600 nm is 0.2 · 1020 cm2 and reabsorption is nearly absent. The 1.58–1.64 lm range, evidenced in Fig. 5, can then in principle be considered suitable for tunable laser operation.
Fig. 5. r-polarized absorption (solid) and emission (dashed) crosssections of Er3+ in BNN at RT. The shadowed region is the optimum wavelength extraction range.
The 4I11/2 ! 4I13/2 emission in the 2.75 lm region is shown in Fig. 6. At room temperature the emission system consists of a broadband ranging from 2.6 to 2.9 lm and practically featureless. It is worth noting that the spectra of the other Er hosts used in laser applications are more structured [27,30,31]. The lack of a structure in the emission spectrum suggests the use of the ErBNN as a widely tunable source in the 3 lm region, interesting for applications in medicine and metrology. The decay times of the 4I11/2 level have been measured at 1002 nm (4I11/2 ! 4I15/2) by exciting the crystal at 969 nm. The excitation of the sample at this wavelength gives rise also to luminescence in the visible region, indicating the presence of non linear processes such as excited state absorption or up conversion processes. The decay profiles are single exponential independent of the temperature (see inset of Fig. 6) and the decay times range from 660 ls at 10 K to 490 ls at 298 K.
10 K
10
-1
10 K
I 10-2
Intensity (a. u.)
Fig. 4. 10 and 298 K emission in the 1.5 lm region.
10
298 K
-3
0.0
1.0 ms
298 K
2.0
π σ
2500
2600
2700
2800
2900
3000
Wavelength (nm) Fig. 6. 10 and 298 K emission in the 2.7 lm region.
3100
400
S. Bigotta et al. / Optical Materials 28 (2006) 395–400
The comparison with the calculated lifetime, 3042 ls, confirms that non radiative processes are active. In fact, multiphonon relaxation is very probable [32], since the gap between the 4I11/2 and the next lower lying 4I13/2 level, about 3500 cm1, can be bridged by four high energy phonons, and this is in agreement with the risetime observed in the 4I13/2 decay behavior. 5. Conclusions The absorption and emission spectra of Er:BNN are strongly affected by the presence of various non equivalent optical centers. These mainly arise from the different charge-compensation processes following the substitution of the Er3+ ions for the Na+ ones. As a consequence, at least three, but probably more non equivalent active centers contribute to the optical properties of the material. Moreover, the local disorder connected to the modulation of the crystal structure results in a further inhomogeneous broadening of the spectral features, making the title compound potentially suitable for diode laser pumping and tunable laser operation. The emission transitions interesting for applications have been investigated and the efficiencies of the non radiative processes have been estimated. They confirm that Er:BNN is a promising laser material, especially if we consider the non linear optical properties of the host matrix, which make it particularly attractive in the perspective to develop self doubling laser devices. We are planning experiments to obtain samples with different dopant concentrations in order to find the best conditions for laser operation. Acknowledgment The authors gratefully acknowledge A. Belletti (Universita` di Parma) for his expert technical assistance.
References [1] R.R. Neurgaonkar, W.K. Cory, J. Opt. Soc. Am. B 3 (1986) 274. [2] S. Singh, D.A. Draegert, J.E. Geusic, Phys. Rev. B 2 (1970) 2709. [3] A.A. Kaminskii, V.A. Koptsik, Y.A. Maskaev, I.I. Naumova, L.N. Rashkovich, S.E. Sarkisov, Phys. Stat. Sol. A 28 (1975) K5.
[4] G. Foulon, A. Brenier, M. Ferriol, M.-T. Cohen-Adad, G. Boulon, Chem. Phys. Lett. 249 (1996) 381. [5] J.J. Romero, A. Brenier, L.E. Bausa`, G. Boulon, J. Garcia Sole´, A.A. Kaminskii, Opt. Comm. 191 (2001) 371. [6] K. Lebbou, H. Itagaki, A. Yoshikawa, T. Fukuda, F. CarrilloRomo, G. Boulon, A. Brenier, M.-T. Cohen-Adad, J. Crystal Growth 210 (2000) 655. [7] F. Carrillo-Romo, C. Goutaudier, Y. Guyot, M.-T. Cohen-Adad, G. Boulon, K. Lebbou, A. Yoshikawa, T. Fukuda, Opt. Mat. 16 (2001) 199. [8] K.S. Rao, K.H. Yoon, J. Mater. Sci. 38 (2003) 391. [9] E. Cavalli, G. Calestani, E. Bovero, A. Belletti, A. Migliori, J. Phys.: Condens. Matter 16 (2004) 729. [10] G. Foulon, M. Ferriol, A. Brenier, M.T. Cohen-Adad, M. Boudeulle, G. Boulon, Opt. Mater. 8 (1997) 65. [11] A. Boudou, J. Sapriel, Phys. Rev. B 21 (1980) 61. [12] H.R. Xia, L.H. Hu, C.J. Wang, L.X. Li, S.B. Yue, X.L. Meng, L. Zhu, Z.H. Yang, J.Y. Wang, J. Appl. Phys. 83 (1998) 256. [13] H.R. Xia, L.X. Li, H. Yu, X.L. Meng, L. Zhu, J.L. Hu, Cryst. Res. Technol. 34 (1999) 901. [14] Ph. Labbe´, H. Leligny, B. Raveau, J. Schneck, J.C. Toledano, J. Phys.: Condens. Matter 2 (1989) 25. [15] A.A. Kaminskii, Crystalline Lasers: Physical Processes and Operating Schemes, CRC Press, Boca Raton, Fla, 1996. [16] B.R. Judd, Phys. Rev. 127 (1962) 750. [17] G.S. Ofelt, J. Chem. Phys. 37 (1962) 511. [18] W.T. Carnall, P.R. Fields, K. Rajnak, J. Chem. Phys. 49 (1968) 4412. [19] M.J. Weber, Phys. Rev. 157 (1967) 262. [20] J. Amin, B. Dussardier, T. Schweizer, M. Hempstead, J. Lumin. 69 (1996) 17. [21] B. Simondi-Teisseire, B. Viana, D. Vivien, A.M. Lejus, Phys. Stat. Sol. A 155 (1996) 249. [22] A.A. Kaminskii, V.S. Mironov, A. Kornienko, S.N. Bagaev, G. Boulon, A. Brenier, B. Di Bartolo, Phys. Stat. Sol. A 151 (1995) 231. [23] A.A. Kaminskii, A.G. Petrosyan, G.A. Denisenko, T.I. Butaeva, V.A. Fedorov, S.E. Sarkisov, Phys. Stat. Sol. A 71 (1982) 291. [24] E. Comini, A. Toncelli, M. Tonelli, E. Zannoni, E. Cavalli, A. Speghini, M. Bettinelli, J. Opt. Soc. Am. B 14 (1997) 1938. [25] P. Kabro, J.A. Capobianco, F.S. Ermeneux, R. Moncorg!e, M. Bettinelli, E. Cavalli, J. Appl. Phys. 82 (1997) 3983. [26] C. Bertini, A. Toncelli, M. Tonelli, E. Cavalli, N. Magnani, J. Lumin. 106 (2004) 235. [27] D.S. Knowles, H.P. Jenssen, IEEE J. Quantum Electron. 28 (1992) 1197. [28] B.F. Aull, H.P. Jenssen, IEEE J. Quantum Electron. 18 (1982) 925. [29] S.A. Payne, L.L. Chase, L.K. Smith, W.L. Kway, W.F. Krupke, IEEE J. Quantum Electron. 28 (1992) 2619. [30] F. Auzel, S. Hubert, D. Meichenin, Appl. Phys. Lett. 54 (1989) 681. [31] P.F. Moulton, J.F. Manni, G.A. Rines, IEEE J. Quantum Electron. 24 (1988) 960. [32] J. Koetke, G. Huber, Appl. Phys. B 61 (1995) 151.