Journal of Luminescence
Article in Press,Corrected Proof - Note to users

/ / This Document /
/ / / SummaryPlus / /
/ / / Full Text + Links / /
/ / ·Full Size Images / /
/ / / PDF (808 K) / /
/ / Actions /
/ / / Cited By /
/ / / Save as Citation Alert /
/ / / E-mail Article /
/ / / Export Citation /


doi:10.1016/j.jlumin.2006.08.031
Copyright © 2006 Elsevier B.V. All rights reserved.

Oxysulfide optical ceramics doped by Nd3+ for one micron lasing

Yu.V. Orlovskiia, , , T.T. Basieva, K.K. Pukhova, M.V. Polyachenkovaa, P.P. Fedorova, O.K. Alimova, b, E.I. Gorokhovac, V.A. Demidenkoc, O.A. Khristichc and R.M. Zakalyukind
aLaser Materials and Technology Research Center, General Physics Institute, RAS, 38 Vavilov Street, Bld. D, Moscow 119991, Russia
bInstitute of Nuclear Physics, Uzbekistan Academy of Sciences, Tashkent, Ulugbek 702132, Usbekistan
cFederal State Unitary Organization Research and Technological Institute of Optical Materials, All-Russia Scientific Center, S.I. Vavilov State Optical Institute, 36-1, Babushkin Street, St. Petersburg, 193171, Russia
dInstitute of Crystallography, RAS, Leninsky prospect 59, Moscow 119333, Russia
Available online 18 September 2006.

Abstract

The spectroscopic characteristics of translucent oxysulfide optical ceramics doped by Nd3+, which can be perspective for 1-μm lasing with high quantum efficiency is considered. Analysis of multiphonon relaxation (MR) rate of the low 4I11/2 laser level in the ceramics is provided using nonlinear theory of MR for multifrequency model of lattice vibration. MR rate dependence for the transitions with equal number of phonons p=4 on the extent of the phonon spectrum and Nd3+ to the nearest ligands distance R0 is established experimentally and analyzed. For the initial 4F3/2 laser level in the Gd2O2S:Nd3+ and La2O2S:Nd3+ ceramic samples an analysis of the energy transfer kinetics is provided. Different stages of energy transfer—direct energy transfer and a stationary migration-controlled stage are observed and analyzed. Dipole–dipole mechanism of energy transfer is established and microparameters of energy transfer and migration are determined.

Keywords: Gd2O2S:Nd3+; La2O2S:Nd3+; LaF3:Nd3+; SrF2:Nd3+; PbF2:Nd3+; Optical ceramics; Crystals; Optical properties; Multiphonon relaxation (MR) rate dependence; Nonlinear theory of MR; Single-frequency model; Multi-frequency model; Rare-earth–nearest ligands distance; Direct energy transfer; Energy migration; Dipole–dipole interaction; One micron lasing

PACS classification codes: 71.10.−w; 78.20.Ci; 78.47.+p; 78.55−m; 78.55.Qr; 42.70.Hj

Article Outline

1. Introduction

2. Experimental technique

3. Optical properties

4. Multiphonon relaxation

5. Energy transfer

6. Conclusion

Acknowledgements

References

1. Introduction

The rare-earth doped oxysulfides (R2O2S) are effective phosphors. They were studied in the form of phosphorescent powders and used for television tubes, X-ray medical technique and other devices [1]. High melting temperature and considerable decomposition is the drawback for the R2O2S single crystal growing process. Anyway, the La2O2S:Nd3+ single crystals were grown and studied as new high-gain laser materials [2].

R2O2S crystallize in trigonal system, space group P3-m1, Z=1 [3]. Its crystal structure corresponds to the A-R2O3 structural type of cerium subgroup rare-earth (RE) oxides, with substitution of oxygen by sulfur in the crystallographic position. Cation coordination number is seven, with four atoms of oxygen and three atoms of sulfur. Three oxygen and three sulfur atoms form a regular Archimedian antiprism, the oxygen bottom of antiprism is capped by fourth oxygen atom.

Nowadays, various types of oxide ceramic lasers were developed. There are numerous publications about fabrication and investigation of laser ceramics based on TR-ion-doped yttrium oxide, YAG (YAG:Nd3+, YAG:Cr4+), Y3ScxAl5−xO12:Nd3+, YGdO3:Nd3+, Lu2O3:Nd3+, and Sc2O3:Yb3+[4] and [5]. Ceramic lasers have several advantages. Among them are an ease of fabrication, low cost, fabrication of large size elements, large concentration and homogeneous distribution of dopants. The R2O2S was studied recently as fluorescent ceramics [6], [7], [8] and [9].

Here we present the spectroscopic characteristics of oxysulfide translucent optical ceramics which can be perspective for 1-μm lasing with high quantum efficiency.

2. Experimental technique

Translucent oxysulfide ceramics of Gd2O2S and La2O2S doped by Nd3+ have been prepared by hot pressing process. Initial R2O2S powders with neodymium oxide sub-micron-sized particles were used as starting materials. According to scanning electron microscopy data the ceramics is formed from elongated grains of 30–50μm width.

Transmission spectra of the samples were measured at 77 and 300K by a single-channel spectrometer using incandescent halogen lamp. A CW laser diode operating at 812nm was used as a pump source for the measurements of fluorescence spectra. A double grating high spectral resolution DFS-12 spectrometer with 600 grooves/mm grating was used for spectral measurement. The transmitted and fluorescent signal was detected by PMT-79 and PMT-83, respectively, and was recorded using an ADC and a PC.

Fluorescence kinetics decay of the 4F3/2 manifold and time-resolved fluorescence spectra of the optical samples for the transitions from the 4F3/2 manifold were measured using tunable LiF: F2→F2+ color center laser pumped by YAG:Nd laser (tp=10ns, f=12Hz). High throughput MDR-2 monochromator and a PMT-83 were used for fluorescence measurements. A Tektronix TDS 3032B digital oscilloscope was employed for fluorescence kinetics measurements. Time-resolved fluorescence spectra were measured using standard gated Boxcar averager (PAR 162/164) with variable time gate tgate and gate delay td.

The fluorescence decay curves of the high-lying strongly quenched 4G7/2 level in the Gd2O2S:Nd3+ (0.5wt%) and La2O2S:Nd3+ (1wt%) optical ceramics and in the PbF2:Nd3+ (1mol%) crystal were measured at the visible 4G7/2→4I13/2 transition at room and liquid nitrogen temperatures under pulsed copper vapor laser excitation at 511nm (tp=10ns; f=10kHz) and a time-correlated single-photon counting detection technique. The time-resolved fluorescence spectra at the same transition under pulsed copper vapor laser excitation at room temperature were measured for identification of Nd3+ fluorescence using the same technique and a special program of our own for multichannel amplitude analyzer (MCA, EG&G Ortec) operation.

3. Optical properties

Let us first consider the optical properties of Gd2O2S:Nd3+ (0.1wt%) oxysulfide optical ceramics. Transmission spectrum for the sample with thickness d=0.175cm at the 4I9/2→4F3/2 transition was measured at 77K and absorption spectrum was calculated (Fig. 1a). The notations near the spectral lines denote the transitions between the crystal-field (CF) levels of corresponding manifolds. The primed numbers denote the CF levels of the excited manifold and those without prime—the CF levels of the ground manifold. Counting is from bottom to top. Corresponding fluorescence spectrum at the 4F3/2→4I9/2 transition (Fig. 1b) allows to determine the CF (Stark) splitting of the 4I9/2 and 4F3/2 manifolds (Fig. 2).


(96K)

Fig. 1.Absorption spectrum of the Gd2O2S: Nd3+ (0.1wt%) oxysulfide optical ceramics measured at the 4I9/2→4F3/2 transition of the Nd3+ ion—a; and fluorescence spectrum at the 4F3/2–4I9/2 transition under 812nm CW laser excitation—b. Both spectra are at 77K.

(41K)

Fig. 2.Identified crystal-field (Stark) levels diagram of the 4I9/2, 4I11/2, 4F3/2, 4G5/2,2G7/2 and 4G7/2 manifolds of the Nd3+ ion in the Gd2O2S:Nd3+ (0.1wt%) oxysulfide optical ceramics.

Absorption spectra of the 4I9/2→4G7/2 and 4I9/2→2G7/2; 4G5/2 transitions at 77 and 300K (Fig. 3) allow to separate the 2G7/2 and 4G5/2 manifolds. A spin-forbidden 4I9/2→2G7/2 transition has lower intensity than the spin-allowed 4I9/2→4G5/2 transition and takes up higher position on the energy scale. Thus the minimal energy gap ΔEmin=1517cm−1 is in between the lowest CF level of the 4G7/2 manifold and the upper CF level of the 2G7/2 manifold. The absorption spectral lines measured in the Gd2O2S:Nd3+ (0.1%) optical ceramics are well pronounced as opposed to those in the La2O2S:Nd3+ (1%) optical ceramics (not presented). The later may be concerned with the inhomogeneous broadening of the CF transition spectral lines at larger concentration of Nd3+ in the La2O2S ceramics.


(26K)

Fig. 3.Absorption spectra of the Gd2O2S:Nd3+ (0.1wt%) oxysulfide optical ceramics at the 4I9/2→4G7/2, 4I9/2→2G7/2, and 4I9/2→4G5/2 transitions at 77K (gray curve) and 300K (black curve).

The fluorescence spectrum at the 4F3/2→4I11/2 transition was measured with high spectral resolution in Gd2O2S:Nd3+ (0.1wt%) (Fig. 4a) and La2O2S:Nd3+ (1wt%) (Fig. 4b) at 77 and 300K using CW diode laser excitation. The positions of the three lowest CF levels of the 4I11/2 manifold are found in the Gd2O2S:Nd3+ ceramics (Fig. 2). The fluorescence spectral lines measured at the 4F3/2→4I11/2 CF level transitions in the Gd2O2S:Nd3+ (0.1wt%) optical ceramics at 77 and 300K are much more narrow (Fig. 4a) comparing to those in the La2O2S:Nd3+ (1wt%) optical ceramics (Fig. 4b), because of the same reasons as in the case of absorption spectra.


(131K)

Fig. 4.Fluorescence spectra of the 4F3/2→4I11/2 transition in Gd2O2S:Nd3+ (0.1wt%)—a and La2O2S:Nd3+ (1wt%)—b under 812nm laser excitation at 77 and 300K.

4. Multiphonon relaxation

The 4I11/2 state can be potentially a bottleneck during 1-μm lasing and gives rise to transient absorption at laser wavelength. So, the faster the rate of multiphonon relaxation (MR), faster the depletion of the 4I11/2 level and the smaller the losses [10].

It is a well-known fact that the energy gaps of the 4G7/2→4G5/2; 2G7/2 and the 4I11/2→4I9/2 transitions of Nd3+ are nearly the same and in oxide and fluoride crystals are mostly determined by MR. One can estimate the MR rate of the 4I11/2→4I9/2 transition from the rate of the 4G7/2→4G5/2; 2G7/2 transition. For correct estimation the results of the nonlinear theory of MR relaxation can be used [11], [12], [13], [14], [15], [16] and [17]. In the frame of nonlinear theory of MR the total p-phonon transition probability WJ′→J(p) between the two J manifolds can be written in a form similar to the Judd–Ofelt expression for an electro-dipole radiative transition probability between the same manifolds [18] and [19]:

/ (1)

Here depends on parameters of static CF, (LSJ||U(k)||L′S′J′) is the reduced matrix element of the unit tensor operator U(k) of rank k that specifies the J′→J transition (the values of (LSJ||U(k)||L′S′J′)2 are tabulated [20]), ΩJ′J=ΔEJ′J/ the frequency, and ΔEJ′J the energy gap between lowest Stark level of the manifold J′ and highest Stark level of manifold J. The spectral function J(p)(Ω)is

/ (2)

where the correlation function of the displacements

/ (3)

In Eq. (3) the symbol … denotes averaging over the thermal lattice vibrations, R0 is the equilibrium distance between the RE ion and the nearest ligand, u=uL-uRE, uL and uRE are the ligand and RE ion displacements from their equilibrium positions, respectively. It should be mentioned that J(p)(Ω) depends on R0 and the characteristics of the phonon subsystem, only.

The combined electronic factor accounts for both the point–charge and the exchange–charge interaction between the RE ion and the nearest ligands and can be presented as

/ (4)

with

/ (5)
/ (6)

In Eqs. (5) and (6)z is the number of the anions nearest to the RE ion, l the orbital angular momentum of optical electrons (l=3 for 4f electrons) and is the 3j-symbol. In Eq. (5)is a parameter of point–charge CF:

/ (7)

where e is the electron charge, q the effective charge of a ligand, and the mean value of the kth power of the radius ξ of optical 4f electron.

In Eq. (6)is a parameter of exchange–charge CF:

/ (8)

Here Gs, Gσ, Gπ=5–10 are the CF fitting parameters in the frame of exchange–charge model [21] which can be determined from the fitting of the calculated and measured Stark structure of RE manifolds; γk=2−k(k+1)/12; Sν (R0)=S0νexp(−ανR0) are the overlap integrals of the 4f-electron wave functions with the wave functions of the external electronic shells of the ligands. The parameters αν and Sν0 are determined from the R dependence of the overlap integrals Sν(R). The function Sν(R) is calculated on the basis of the known Hartree–Fock radial wave functions for RE ions and wave functions of the ligands. For important practical cases (chlorine, sulfur, fluorine and oxygen surrounding the trivalent RE ions) this is the pσ, pπ, and s orbitals of the ligands; and Φkp can be easily calculated from expression given in Ref. [14].

The greatest difficulties appear in the course of calculation of the spectral density J(p)(Ω). The simplest model of lattice vibration is the single-phonon frequency model (model of “effective phonon”).

For the single-phonon frequency model the spectral density is determined as [11]

/ (9)

Here p=ΩJ′J/ωeff,

/ (10)

is a parameter of the single-phonon frequency model, and

/ (11)

is the population of a phonon mode of frequency ω at temperature T described by the Bose–Einstein distribution. Thus, for the single-phonon frequency model we have

/ (12)

with

/ (13)

The phonon factor η can be roughly estimated using the following Eq. [15]:

/ (14)

where c is the velocity of light, the maximum phonon frequency of the crystal lattice (in cm−1), and M the reduced mass of the atoms involved in the vibrations and can be calculated as [15]

/ (15)

where Mcat=(Ca, Sr, La, Pb,…); Manion=(O,F,S,Cl,…). Hence, values of ηest in the range of 10−3–10−4 are expected for the considered crystals.

Below we consider the more realistic (multifrequency) model of lattice vibration than the single-frequency model of “effective phonon”. At present there are a few data about the inelastic neutron scattering by crystals. The neutron scattering data analysis allows to extract the density of phonon states (DPS) ρ(ω). We shall employ the DPS data for calculation of spectral density J(Ω). In approximation,

/ (16)

We can derive from Eq. (2) that

/ (17)

where the “multiphonon density”

/ (18)

is the convolution of the p functions of

ρT(ω)=ρ(ω)[n(ω,T)+1]. / (19)

We emphasize that in Eqs. (16), (17), (18) and (19) the normalized DPS is employed, so that .

Using Eq. (18) we can write Eq. (1) in the form

/ (20a)

or, in view of expression (13), as

/ (20b)

(Here and ΔEJ′J are in cm−1).

Thus, total MR rate is given by the expression

/ (21)

where all combinations of phonon frequencies make contributions to the p-phonon process and summation is made for all p. This multifrequency model was used for the theoretical analysis of experimental data on MR rates at T=0K in RE-doped CaF2, SrF2, PbF2, and BaF2 crystals [17]. (Here and above it is supposed that the CF splitting (ΔεCF) of the low-lying J-manifold is small (ΔεCF≤ωmax)). It resolves the contradiction between the “energy gap” law and the dependence of the MR rate WMR on the number of phonon p(WJ′→J(p)). Now, the energy gap always can be bridged by an integer number of phonons p by taking an appropriate combinations of phonon energies within a phonon spectrum of a crystal, which are compatible with the law of conservation of energy:

(ω1+ω2+…+ωp)=ΔEJ′J. / (22)

But p with maximum contribution can be determined as well. The outlined theory implies the validity of the continuous approximation for dependencies ω(k) of phonon frequencies ω on phonon wave vector k. In general, one must take into account the confinement effects that can lead to impracticability of Eq. (22) in view of discrete character of frequencies for a grain of small size. Let us discuss the validity of continuous approximation for our samples. As indicated above, the considered ceramics is formed from grains of 30–50μm width. The density of the Gd2O2S and La2O2S single crystals are equal to ρsingl.=7.337 and 6.281g/cm3, respectively. The measured density of respective ceramics is equal to that of a crystal with the precision of a measurement. The porosity (ρsingl.–ρceramics)/ρsingl. of the considered samples is much less than the unity. (Here ρceramics is ceramics density). Taking the grain density equal to single crystal density ρsingl. we obtain that the 30μm size grains contains about N=1014 atoms for both Gd2O2S and La2O2S ceramics. As discussed above, the value of maximum phonon frequencies ωmax for Gd2O2S and La2O2S is about 440 and 400cm−1, respectively. The average frequency difference Δω can be roughly estimated as Δω=ωmax/3N. Using the foregoing data we obtain that Δω is of the order of 10−12cm−1 both for the Gd2O2S and La2O2S grains. This means that the continuous approximation is indubitably applicable to the considered cases and the outlined nonlinear theory of MR can be used for an analysis of the MR processes in considered ceramics without any modification.

According to the theory for the specific ion in specific crystal the only difference in the MR rates for different transitions between the two manifolds comes from the terms (2J′+1)−1 and (LSJ||U(k)||L′S′J′)2. Thus, the MR rate of the 4I11/2→4I9/2 transition can be recalculated from the measured MR rate of the 4G7/2→4G5/2; 2G7/2 transition by Eq. (1). Such calculations using point–charge model of RE-nearest ligands interaction are made below in this paper for Gd2O2S:Nd3+, La2O2S:Nd3+, and LaF3:Nd3+.

Also, it is interesting to compare the measured MR rates in oxysulfide ceramics with those in fluoride crystals with low-phonon spectra (LaF3, SrF2, and PbF2) to verify regularities of MR predicted by the theory. MR rate depends on many parameters etc. (see Eqs. (1), (5), (6), (7), (8) and (10) and (21)). The best plan for researcher is to fix all the parameters except one of primary interest to watch the dependence of the MR rate on just one parameter. For example, the dependence of the MR rate on ΔE and p was studied by many authors and it is a well- known fact that an increase of the number of phonons p by one decreases the MR rate by one–two orders of magnitude ([17] and the references therein). The number of phonons p is the strongest parameter which determines the order of magnitude of the MR rate. In a single-frequency model the minimal number of phonons p participating in a nonradiative transition is pmin=ΔEmin/ωmax. We try to fix p and watch the dependence of the MR rate on solid state matrix considering the same the 4G7/2→4G5/2; 2G7/2 transition in different crystals and optical ceramic samples with slightly different extent of phonon spectra. In that way we also fix the reduced matrix elements of the unit tensor operator U(k) of electronic transitions and have close values of ΔEmin. The maximum phonon frequencies ωmax in the SrF2 and PbF2 crystals can be calculated from neutron scattering data [22] and [23]. For SrF2 it is about 380cm−1. For PbF2 it is about 340cm−1 and closely agrees with maximum phonon frequency obtained from Raman spectra measured in Ref. [24]. To the best of our knowledge there is no neutron scattering data in the literature for the LaF3 and the Gd2O2S and La2O2S crystals, but taking into account the result for SrF2 and PbF2, the maximal phonon frequencies for the LaF3 crystal and the Gd2O2S and La2O2S ceramic samples were determined from their Raman spectra. As will be shown below the sulfur ligands do not give significant contribution to the MR rate. And according to Ref. [25] sulfur atom motion in the La2O2S crystal are not Raman active, which to some extent confirms our assumption. Raman spectra of the Gd2O2S and La2O2S optical ceramics doped by Nd3+ were measured using highly sensitive Raman spectrometer with photon counting (Fig. 5) and the values of maximum phonon frequencies obtained are about 440 and 400cm−1, respectively.


(15K)