- Email: [email protected]
Time-resolved fluorescence line narrowing spectroscopy and fluorescence quenching in Nd3+-doped fluoroarsenate glasses
Optical Materials 25 (2004) 193–200 www.elsevier.com/locate/optmat
Time-resolved fluorescence line narrowing spectroscopy and fluorescence quenching in Nd3þ -doped fluoroarsenate glasses L.M. Lacha a, R. Balda
a,b,c,*
, J. Fern andez
a,b,c
, J.L. Adam
d
a
d
Dpto. Fisica Aplicada I, Escuela Superior de Ingenieros, Alda. Urquijo s/n 48013 Bilbao, Spain b Centro Mixto CSIC-UPV/EHU, P Manuel de Lardiz abal 3, 20018 Donostia, Spain c Donostia International Physics Center (DIPC), P Manuel de Lardiz abal 3, 20018 Donostia, Spain Laboratoire des Verres & C eramiques, Universit ede Rennes 1, UMR-CNRS 6512, Campus de Beaulieu, 35042 Rennes Cedex, France Received 21 October 2002; accepted 5 May 2003
Abstract The spectral migration of energy among the Nd3þ ions in fluoroarsenate glasses of the Na4 As2 O7 , BaF2 , YF3 system with different Nd3þ concentrations (0.5, 2, 3, 4, and 5 mol%) has been investigated by using time-resolved fluorescence line narrowing (TRFLN) spectroscopy. The features of the time-resolved fluorescence line narrowing 4 F3=2 fi 4 I9=2 emission spectra obtained under resonant excitation at different wavelengths along the 4 I9=2 fi 4 F3=2 transition as a function of Nd3þ concentration reveal the existence of energy migration between discrete regions of the inhomogeneously broadened spectral profile. The analysis of the time evolution of the 4 F3=2 fi 4 I9=2 narrowed emission shows that the electronic mechanism responsible for the ion–ion interaction can be identified as a dipole–dipole energy transfer process. From the thermal behavior of lifetimes of the 4 F3=2 state a T 3 dependence for the non-radiative Nd–Nd relaxation processes has been found in the 10–80 K temperature range for Nd3þ concentrations higher than 0.5 mol% which is in agreement with a two-site non-resonant process. 2003 Elsevier B.V. All rights reserved. Keywords: Energy transfer; Glasses; Time-resolved laser spectroscopy; Optical materials; Rare-earth
1. Introduction The absorption and emission spectra of rareearth ions in glasses are characterized by the inhomogeneous broadening resulting from the
* Corresponding author. Address: Dpto. Fisica Aplicada I, Escuela Superior de Ingenieros, Alda. Urquijo s/n 48013 Bilbao, Spain. Fax: +34-94-601-4178. E-mail address: [email protected] (R. Balda).
distribution of the crystal fields at the variety of rare-earth-ion sites in the amorphous solid [1]. However, the lines can be narrowed by optical siteselective methods such as laser-induced fluorescence line narrowing (FLN) [2,3]. These techniques are useful to obtain a detailed information about the local field, and ion–ion and ion–host interaction processes. When the activator ion concentration in glass becomes high enough, ions interact and ion–ion energy transfer occurs. Due to the inherent disorder of glass, ions in nearby sites may
0925-3467/$ - see front matter 2003 Elsevier B.V. All rights reserved. doi:10.1016/S0925-3467(03)00269-6
194
L.M. Lacha et al. / Optical Materials 25 (2004) 193–200
be in physically different environments with greatly varying spectroscopic properties. Therefore, in addition to causing a spatial migration of energy, the transfer may also produce spectral diffusion within the inhomogeneously broadened spectral profile [2]. The migration of the electron excitation over the inhomogeneous profile (spectral migration) determines the effectiveness of the generation (amplification) of the stimulated emission [1]. One of the kinetic methods of migration investigation consists in analysing time-resolved luminescence spectra after selective excitation [1,3–7]. The timeresolved fluorescence line narrowing (TRFLN) technique provides us with a way of measuring optical energy propagation from the initially excited subset of ions to other elements of the inhomogeneously broadened line. Previous works [1,4,5] on Nd-doped glasses utilized a non-resonant pumping scheme to excite the 4 F3=2 state which is responsible for the principal Nd3þ fluorescence. The introduction of tunable lasers for resonant excitation in the near infrared made it possible to perform measurements on TRFLN under resonant excitation of the 4 F3=2 state of Nd3þ in order to study the structure of inhomogeneously-broadened bands and migration of excitation energy among Nd3þ ions [3,8,9]. It is worthwhile mentioning that TRFLN techniques have only been used in a few studies on spectral diffusion among Nd3þ ions in glasses [1,3–5,9]; among these, there are only a few works performed in a resonant pumping scheme [3,8–11] to excite the 4 F3=2 state of Nd3þ ions. Recently, new glasses were discovered in the Na4 As2 O7 , BaF2 , YF3 system which are parent materials with fluorophosphate glasses of the NaPO3 , BaF2 , YF3 system. As in the case of fluorophosphate glasses, large amounts of active rare-earth ions can be incorporated in replacement of yttrium. Whereas optical properties of some active ions have already been investigated in fluorophosphate glasses [12–14], the behavior of rareearth ions in fluoroarsenate glasses is practically unknown. In a previous work we have characterized the optical properties of Nd3þ ions in these glasses [15]. In this work, we investigate the existence of energy transfer among Nd3þ ions in fluoroarsenate glasses of the Na4 As2 O7 , BaF2 , YF3
system with different Nd3þ concentrations (0.5, 2, 3, 4, and 5 mol%) by using TRFLN spectroscopy and observing the emission characteristics of the system as a function of time, concentration, and excitation wavelength.
2. Experimental The fluoroarsenate glasses used in this work were prepared at the Laboratoire de Verres et Ceramiques of the Rennes University (France). Glasses were synthesized in the (40Na4 As2 O7 – 30BaF2 –30YF3 ) ternary system and doped with 0.5, 2, 3, 4, and 5 mol% of NdF3 which correspond to 0.47 · 1020 , 1.87 · 1020 , 2.80 · 1020 , 3.74 · 1020 , and 4.7 · 1020 cm3 respectively. Details on the entire glass forming diagram and synthesis can be found in Ref. [15]. The samples temperature was varied between 4.2 and 300 K in a continuous flow cryostat. Timeresolved resonant fluorescence line-narrowed emission measurements were obtained by exciting the samples with a Ti-sapphire laser, pumped by a pulsed frequency doubled Nd:YAG laser (9 ns pulse width and 1 cm1 linewidth). The fluorescence was analysed with a 0.27 m Spex monochromator and the signal was detected with a Hamamatsu R5108 photomultiplier provided with a gating circuit designed to enable gate control from an external applied TTL level control signal. Data were processed by an EGG-PAR boxcar integrator. Lifetime measurements were performed with a narrow band (0.08 cm1 linewidth) tunable dye laser of 9 ns pulse width and the emission detected with a Hamamatsu R5108 photomultiplier.
3. Results and discussion 3.1. TRFLN spectra We have investigated the existence of energy transfer among Nd3þ ions in this glass by using TRFLN spectroscopy and observing the emission characteristics of the system as a function of time, concentration, and excitation wavelength. TRFLN
L.M. Lacha et al. / Optical Materials 25 (2004) 193–200
spectra for the 4 F3=2 fi 4 I9=2 transition were performed for all concentrations at 4.2 K by using different resonant excitation wavelengths in the low energy Stark component of the 4 I9=2 fi 4 F3=2 absorption band at different time delays after the laser pulse. As an example, Fig. 1 shows the normalized 4 F3=2 fi 4 I9=2 spectra at 4.2 K for four concentrations 0.5, 2, 4, and 5 mol% of Nd3þ ions obtained at two different time delays after the laser
2 5% 1
0 4
Intensity(arb.units)
4% 2
0 4 2% 2
0 4 0.5 % 2
0 860
880
900
920
940
Emission Wavelength (nm) Fig. 1. Time-resolved fluorescence line-narrowed spectra of the 4 F3=2 fi 4 I9=2 transition obtained at 1 ls (solid line) and 700 ls (dashed line) after the laser pulse by exciting at 874 nm for the samples doped with 0.5, 2, 4, and 5 mol%. Measurements correspond to 4.2 K.
195
pulse, 1 and 700 ls, by exciting at 874 nm. The spectra present two components. The one in the high energy side is a simple FLN line (linewidth 8 cm1 ) corresponding to the emission to the lowest component of the 4 I9=2 multiplet, which is the ground state of the ion. The position of this narrow line is essentially determined by the wavelength of the pumping radiation. The broad emission arises from the non-narrowed inhomogeneous line. As time delay increases the relative intensity of the narrow line and the broad (nonselected) emission changes, and the later becomes stronger, which indicates the existence of energy transfer between discrete regions of the inhomogeneous broadened profile. This effect increases with Nd3þ concentration. As can be observed, when Nd3þ concentration increases the broad nonselected emission becomes more intense. At higher concentrations the transfer process appears at shorter time delays, but in all cases energy transfer produces a relative increase of the broad emission with respect to the narrow band. The 4 F3=2 fi 4 I9=2 emission spectra performed by exciting at different wavelengths along the low energy Stark component of the 4 I9=2 fi 4 F3=2 absorption band show that the broad emission decreases as the excitation energy decreases, which indicates that excitation energy can migrate mainly in one direction. As an example Fig. 2 shows the TRFLN spectra for the sample doped with 2 mol% of Nd3þ obtained at three different excitation wavelengths at 1 ls after the laser pulse. The same behavior is observed in all samples. The time evolution of laser-induced resonant line-narrowed fluorescence is produced by a combination of radiative decay and non-radiative transfer to other nearby ions. Subsequent fluorescence from the acceptor ions replicate the inhomogeneously broadened equilibrium emission profile, showing that transfer occurs not only to resonant sites but to the full range of sites within the inhomogeneous profile. In this case, a quantitative measure of the transfer is provided by the ratio of the intensity in the narrow line to the total intensity of the fluorescence in the inhomogeneous band. Neglecting the dispersion in the radiative decay rate, and using the F€ oster formula for dipole–dipole energy transfer, one can write for the
196
L.M. Lacha et al. / Optical Materials 25 (2004) 193–200
The macroscopic parameter cðEL Þ has the meaning of an integral characteristic, which reflects the average rate of excitation transfer from donors to the ensemble of spectrally non-equivalent acceptors [6]. This analysis deals with migration induced changes in an individual luminescence band originating from transitions between a pair of Stark sublevels [6]. However, in the case of Nd3þ ions, the luminescence spectra represent a superposition of different Stark components. At low temperature, where only the lowest Stark sublevels of the ground and metastable states are populated, the spectra show a narrow resonant Stark component and some broad emission due to transitions to the other Stark components of the ground state, whereas at higher temperatures the spectral pattern becomes more complicated. As a consequence, though at high temperatures spectral migration can be experimentally observed, the analysis in terms of Eq. (1) which considers a twolevel scheme becomes more complicated. We have analyzed the TRFLN spectra of the 4 F3=2 fi 4 I9=2 transition obtained at different time delays between 1 and 700 ls according to Eq. (1) at 4.2 K. As an example, Fig. 3 shows the results for all samples for an excitation wavelength of 874 nm. As can be observed a linear dependence of the
λexc= 880nm 2
0
Intensity(arb.units)
4
λexc= 874nm
2
0 4
λexc= 871nm
2
4.5
880
900
920
940
Excitation Wavelength (nm) Fig. 2. Time-resolved fluorescence line-narrowed spectra of the 4 F3=2 fi 4 I9=2 transition obtained at 1 ls after the laser pulse by exciting at three different wavelengths for the sample doped with 2 mol%. Measurements correspond to 4.2 K.
Ln [(IB /IN)+1]
860
5%
T=4.2 K
4.0 0
3.5
4%
3.0
3%
2.5
2%
2.0 0.5% 1.5
relationship between the integral intensities of the broad background emission IB and the narrow luminescence component IN [6]: IB Ln þ 1 ¼ cðEL Þt1=2 IN
ð1Þ
0
10
20
t1/2
30
( µs1/2 )
Fig. 3. Analysis of the time evolution of the TRFLN 4 F3=2 fi 4 I9=2 emission by means of Eq. (1) for all samples. Symbols correspond to experimental data and the solid lines are fits to Eq. (1). Data correspond to 4.2 K.
L.M. Lacha et al. / Optical Materials 25 (2004) 193–200
3.5
T=4.2 K
2%
871 nm
Ln [(IB /IN )+1]
3.0 2.5
874 nm
700 µ s 500 µs 300 µs 100 µs
T=4.2 K
4
Ln [(IB /IN)+1]
Ln½ðIB =IN Þ þ 1 function on t1=2 was found, which indicates that a dipole–dipole interaction mechanism among the Nd3þ ions dominates in this time regime. The values of the average transfer rate indicate that energy transfer among Nd3þ ions is weak at concentrations up to 2 mol% at low temperature, and increases with increasing concentration. The c values for this excitation wavelength were found to be 1.7 ± 0.6, 19 ± 0.8, 24 ± 0.3, 30 ± 0.7, and 37 ± 0.3 s1=2 for the samples doped with 0.5, 2, 3, 4, and 5 mol% respectively. The average energy transfer rate found in these fluoroarsenate glasses is a 40% lower than the one found in oxide glasses [11] for the same concentration of Nd3þ ions. The analysis of the TRFLN spectra obtained under excitation at longer wavelengths along the low energy Stark component of the 4 I9=2 fi 4 F3=2 absorption band shows that the transfer rate depends on the excitation energy. As an example, Fig. 4 shows the analysis of the TRFLN spectra for the sample doped with 2 mol% of Nd3þ at three different excitation energies. As can be observed the value of the average transfer rate increases from 0.2 ± 0.4 to 36 ± 1.1 s1=2 when the excitation energy increases from 11 389.5 cm1 (878 nm) to
197
3
2 2
3
4
Nd 3+
5
6
(10 20 cm -3)
Fig. 5. Concentration dependence of the quantity LnðIB =IN þ 1Þ for Nd3þ ions. Symbols correspond to experimental data and the solid lines are linear fits. Data were obtained at different time delays after the laser pulse by exciting at 874 nm.
11 481.05 cm1 (871 nm). This rise in the energy migration rate is due to the increasing number of possible acceptors. The concentration dependence of the spectral migration rate can be obtained from the TRFLN spectra taken on different concentration samples after the same time delay. The increase of the intensity of the broad band relative to that of the narrow component shows the concentration induced growth of the migration transfer rate. Fig. 5 shows the concentration dependence of Ln½ðIB =IN Þ þ 1 obtained at 4.2 K at 100, 300, 500, and 700 ls after the laser pulse for the samples doped with 2, 3, 4, and 5 mol% of Nd3þ . The observed linear behavior is again consistent with a dipole–dipole energy transfer mechanism.
2.0 878 nm
1.5 1.0 0
10
20
30
t1/2 (µs1/2 ) Fig. 4. Analysis of the time evolution of the TRFLN 4 F3=2 fi 4 I9=2 emission by means of Eq. (1) for the sample doped with 2 mol% of Nd3þ for three different excitation wavelengths. Symbols correspond to experimental data and the solid lines are fits to Eq. (1). Data correspond to 4.2 K.
3.2. Temperature dependence concentration quenching of Nd3þ fluorescence The decays of the 4 F3=2 fi 4 I9=2 transition were found to be single exponential at all temperatures and concentrations. As an example, Fig. 6 shows the room temperature logarithmic plot of the experimental decays for the samples doped with 0.5, 3, and 5 mol%, obtaining by exciting at 580 nm and collecting the luminescence at 890 nm. The temperature dependence of the lifetimes were
198
L.M. Lacha et al. / Optical Materials 25 (2004) 193–200
400
-1
λexc= 580nm
0.5%
Lifetime (µs)
2%
0.5%
-2
300
3% 4%
200
-3
5%
100 0
50
100
150
200
250
300
Ln [I(t) (arb.units)]
Temperature (K) -1
Fig. 7. Temperature dependence of the 4 F3=2 lifetime of Nd3þ for different concentrations. Lifetimes were obtained by exciting at 580 nm.
3%
-2
-3
-2
5%
-3 -4 -5 -6 0
250
500
750
1000
Time ( µs) Fig. 6. Logarithmic plot of the fluorescence decays of the 4 F3=2 state for samples doped with 0.5, 3, and 5 mol% of Nd3þ . The decays were obtained by exciting at 580 nm and collecting the luminescence at 890 nm. Measurements correspond to 295 K.
obtained in the 4.2–295 K temperature range for five different concentrations 0.5, 2, 3, 4, and 5 mol% of Nd3þ in fluoroarsenate glass. Fig. 7 shows the lifetime values as a function of temperature. As can be seen in this figure, as concentration rises the
experimental lifetime decreases even at helium temperature, indicating that Nd–Nd relaxation processes are present at concentrations higher than 0.5 mol%. This thermal behavior also suggests the presence of quite a strong thermal mechanism between 10 and 80 K at concentrations higher than 0.5 mol%. In a previous work [16] some of the authors have investigated the temperature dependent concentration quenching of Nd3þ fluorescence in three pure fluoride glasses which shows a strong thermal quenching between 10 and 90 K where the non-radiative rates were found to increase with temperature as T 3 . This is in agreement with a two-site non-resonant process in the shortwavelength regime as has been shown by Holstein et al. [17]. In this process, the phonon emission and absorption take place at different sites, the ion– phonon interaction acts twice, and the site–site Hamiltonian once. This phonon-assisted excitation transfer between different states of similar ions could be understood by considering a small spread in the energy values of 4 F3=2 transitions to the J multiplicity, due to the inherent disorder of the glass structure. The lifetime of the 4 F3=2 state of Nd3þ ion in the doped glasses should be governed by the sum of probabilities for several competing processes such as radiative decay, non-radiative decay by multiphonon emission, and energy transfer to other
L.M. Lacha et al. / Optical Materials 25 (2004) 193–200
[WNd-Nd(T) - WNd-Nd(0)]1/3
15 5% 12
4%
9
3% 6
0
20
40
60
80
100
Temperature (K) Fig. 8. Plot of the non-radiative Nd–Nd relaxation rates in the 10–80 K temperature range. Symbols represent the experimental values, and solid lines are the fit to a T 3 dependence.
Nd3þ ions. In this case the multiphonon relaxation rate for the 4 F3=2 state is expected to be small because of the high energy gap to the next lowerlying J manifold (5500 cm1 ) and the typical values of phonon energies involved ( 6 878 cm1 ) [18]. If one assumes that the purely radiative lifetimes of 4 F3=2 levels are independent of Nd3þ concentration and temperature, and disregards multiphonon relaxation processes, the variations of the experimental lifetimes can be related with the non-radiative Nd–Nd relaxation processes by the simple relation: 1 s1 exp ¼ sR þ WNd–Nd ðT Þ
ð2Þ
where sexp is the measured lifetime and sR represents the radiative lifetime. Fig. 8 shows the plot of ½WNd–Nd ðT Þ WNd–Nd 1=3 ð0Þ in the 10–80 K temperature range for the samples doped with 3, 4, and 5 mol% of Nd3þ . It is worth noticing the good fit of WNd–Nd to a T 3 dependence. This result holds for all concentrations higher than 0.5 mol%.
4. Conclusions (i) Resonant time-resolved fluorescence line-narrowing spectroscopy shows the existence of energy migration between discrete regions of
199
the inhomogeneous broadened 4 F3=2 spectral profile. The analysis of the 4 F3=2 fi 4 I9=2 emission spectra obtained at low temperature as a function of time, concentration, and excitation wavelength shows that the electronic mechanism responsible for the Nd3þ –Nd3þ interaction can be interpreted in terms of a dipole–dipole energy transfer. (ii) The emission lifetimes from the 4 F3=2 state are single exponentials at all temperatures and concentrations. As the concentration rises the decay remains single exponential but a decrease in the lifetime is observed even at helium temperature indicating that relaxation via fast Nd–Nd diffusion processes occurs. (iii) The non-radiative Nd–Nd relaxation rates of the 4 F3=2 fi 4 I9=2 transition were found to increase with temperature as T 3 in the 10–80 K temperature range for concentrations higher than 0.5 mol%. This behavior, also found in pure fluoride matrices, is in agreement with a two-site non-resonant process. Acknowledgements This work has been supported by the Spanish Government (Ref.: MAT2000-1135 and BFM20000352) and the Basque Country University (Ref.: UPV13525/2001). References [1] L.A. Riseberg, Phys. Rev. A 7 (1973) 671. [2] M.J. Weber, in: W.M. Yen, P.M. Selzer (Eds.), Laser Spectroscopy of Solids, vol. 49, Springer, Berlin, 1981. [3] S.A. Brawer, M.J. Weber, Appl. Phys. Lett. 35 (1979) 31. [4] L.D. Merkle, R.C. Powell, E.E. Freed, J. Lumin. 24/25 (1981) 755. [5] L.A. Riseberg, Solid State Commun. 11 (1972) 469. [6] T.T. Basiev, V.A. Malyshev, A.K. Prhevuskii, in: A.A. Kaplyanskii, R.M. Macfarlane (Eds.), Spectroscopy of Solids Containing Rare Earth Ions, North-Holland, Amsterdam, 1987. [7] W.M. Yen, in: A.A. Kaplyanskii, R.M. Macfarlane (Eds.), Spectroscopy of Solids Containing Rare Earth Ions, North-Holland, Amsterdam, 1987. [8] M.M. Broer, D.L. Huber, W.M. Yen, W.K. Zwicker, Phys. Rev. Lett. 49 (1982) 394. [9] T.T. Basiev, Yu.K. VoronÕko, S.B. Mirov, A.M. Prokhorov, JETP Lett. 29 (1979) 639.
200
L.M. Lacha et al. / Optical Materials 25 (2004) 193–200
[10] R. Balda, J. Fern andez, M. Sanz, A. de Pablos, J.M. FdezNavarro, J. Mugnier, Phys. Rev. B 61 (2000) 3384. [11] R. Balda, M. Sanz, J. Fernandez, J.M. Fdez-Navarro, J. Opt. Soc. Am. B 17 (2000) 1671. [12] N. Duhamel-Henry, J.L. Adam, B. Jacquier, C. Linares, Opt. Mater. 6 (1996) 197. [13] R. Balda, J. Fernandez, J.L. Adam, M.A. Arriandiaga, Phys. Rev. B 54 (1996) 12076. [14] R. Balda, J. Fern andez, I. Saez de Ocariz, J.L. Adam, A. Mendioroz, E. Montoya, Opt. Mater. 13 (1999) 159.
[15] J.L. Adam, R. Balda, I. Melsco€et, F. Smektala, L.M. Lacha, J. Fernandez, J. Non-Cryst. Solids 256–257 (1999) 390. [16] R. Balda, J. Fernandez, A. Mendioroz, J.L. Adam, B. Boulard, J. Phys.: Condens. Matter 6 (1994) 913. [17] T. Holstein, S.K. Lyo, M. Orbach, in: W.M. Yen, P.M. Selzer (Eds.), Laser Spectroscopy of Solids, vol. 49, Springer, Berlin, 1981. [18] R.A. Nyquist, R.O. Kagel, Infrared Spectra of Inorganic Compounds, Academic Press, New York, 1971.
Time-resolved fluorescence line narrowing spectroscopy and fluorescence quenching in Nd3þ -doped fluoroarsenate glasses L.M. Lacha a, R. Balda
a,b,c,*
, J. Fern andez
a,b,c
, J.L. Adam
d
a
d
Dpto. Fisica Aplicada I, Escuela Superior de Ingenieros, Alda. Urquijo s/n 48013 Bilbao, Spain b Centro Mixto CSIC-UPV/EHU, P Manuel de Lardiz abal 3, 20018 Donostia, Spain c Donostia International Physics Center (DIPC), P Manuel de Lardiz abal 3, 20018 Donostia, Spain Laboratoire des Verres & C eramiques, Universit ede Rennes 1, UMR-CNRS 6512, Campus de Beaulieu, 35042 Rennes Cedex, France Received 21 October 2002; accepted 5 May 2003
Abstract The spectral migration of energy among the Nd3þ ions in fluoroarsenate glasses of the Na4 As2 O7 , BaF2 , YF3 system with different Nd3þ concentrations (0.5, 2, 3, 4, and 5 mol%) has been investigated by using time-resolved fluorescence line narrowing (TRFLN) spectroscopy. The features of the time-resolved fluorescence line narrowing 4 F3=2 fi 4 I9=2 emission spectra obtained under resonant excitation at different wavelengths along the 4 I9=2 fi 4 F3=2 transition as a function of Nd3þ concentration reveal the existence of energy migration between discrete regions of the inhomogeneously broadened spectral profile. The analysis of the time evolution of the 4 F3=2 fi 4 I9=2 narrowed emission shows that the electronic mechanism responsible for the ion–ion interaction can be identified as a dipole–dipole energy transfer process. From the thermal behavior of lifetimes of the 4 F3=2 state a T 3 dependence for the non-radiative Nd–Nd relaxation processes has been found in the 10–80 K temperature range for Nd3þ concentrations higher than 0.5 mol% which is in agreement with a two-site non-resonant process. 2003 Elsevier B.V. All rights reserved. Keywords: Energy transfer; Glasses; Time-resolved laser spectroscopy; Optical materials; Rare-earth
1. Introduction The absorption and emission spectra of rareearth ions in glasses are characterized by the inhomogeneous broadening resulting from the
* Corresponding author. Address: Dpto. Fisica Aplicada I, Escuela Superior de Ingenieros, Alda. Urquijo s/n 48013 Bilbao, Spain. Fax: +34-94-601-4178. E-mail address: [email protected] (R. Balda).
distribution of the crystal fields at the variety of rare-earth-ion sites in the amorphous solid [1]. However, the lines can be narrowed by optical siteselective methods such as laser-induced fluorescence line narrowing (FLN) [2,3]. These techniques are useful to obtain a detailed information about the local field, and ion–ion and ion–host interaction processes. When the activator ion concentration in glass becomes high enough, ions interact and ion–ion energy transfer occurs. Due to the inherent disorder of glass, ions in nearby sites may
0925-3467/$ - see front matter 2003 Elsevier B.V. All rights reserved. doi:10.1016/S0925-3467(03)00269-6
194
L.M. Lacha et al. / Optical Materials 25 (2004) 193–200
be in physically different environments with greatly varying spectroscopic properties. Therefore, in addition to causing a spatial migration of energy, the transfer may also produce spectral diffusion within the inhomogeneously broadened spectral profile [2]. The migration of the electron excitation over the inhomogeneous profile (spectral migration) determines the effectiveness of the generation (amplification) of the stimulated emission [1]. One of the kinetic methods of migration investigation consists in analysing time-resolved luminescence spectra after selective excitation [1,3–7]. The timeresolved fluorescence line narrowing (TRFLN) technique provides us with a way of measuring optical energy propagation from the initially excited subset of ions to other elements of the inhomogeneously broadened line. Previous works [1,4,5] on Nd-doped glasses utilized a non-resonant pumping scheme to excite the 4 F3=2 state which is responsible for the principal Nd3þ fluorescence. The introduction of tunable lasers for resonant excitation in the near infrared made it possible to perform measurements on TRFLN under resonant excitation of the 4 F3=2 state of Nd3þ in order to study the structure of inhomogeneously-broadened bands and migration of excitation energy among Nd3þ ions [3,8,9]. It is worthwhile mentioning that TRFLN techniques have only been used in a few studies on spectral diffusion among Nd3þ ions in glasses [1,3–5,9]; among these, there are only a few works performed in a resonant pumping scheme [3,8–11] to excite the 4 F3=2 state of Nd3þ ions. Recently, new glasses were discovered in the Na4 As2 O7 , BaF2 , YF3 system which are parent materials with fluorophosphate glasses of the NaPO3 , BaF2 , YF3 system. As in the case of fluorophosphate glasses, large amounts of active rare-earth ions can be incorporated in replacement of yttrium. Whereas optical properties of some active ions have already been investigated in fluorophosphate glasses [12–14], the behavior of rareearth ions in fluoroarsenate glasses is practically unknown. In a previous work we have characterized the optical properties of Nd3þ ions in these glasses [15]. In this work, we investigate the existence of energy transfer among Nd3þ ions in fluoroarsenate glasses of the Na4 As2 O7 , BaF2 , YF3
system with different Nd3þ concentrations (0.5, 2, 3, 4, and 5 mol%) by using TRFLN spectroscopy and observing the emission characteristics of the system as a function of time, concentration, and excitation wavelength.
2. Experimental The fluoroarsenate glasses used in this work were prepared at the Laboratoire de Verres et Ceramiques of the Rennes University (France). Glasses were synthesized in the (40Na4 As2 O7 – 30BaF2 –30YF3 ) ternary system and doped with 0.5, 2, 3, 4, and 5 mol% of NdF3 which correspond to 0.47 · 1020 , 1.87 · 1020 , 2.80 · 1020 , 3.74 · 1020 , and 4.7 · 1020 cm3 respectively. Details on the entire glass forming diagram and synthesis can be found in Ref. [15]. The samples temperature was varied between 4.2 and 300 K in a continuous flow cryostat. Timeresolved resonant fluorescence line-narrowed emission measurements were obtained by exciting the samples with a Ti-sapphire laser, pumped by a pulsed frequency doubled Nd:YAG laser (9 ns pulse width and 1 cm1 linewidth). The fluorescence was analysed with a 0.27 m Spex monochromator and the signal was detected with a Hamamatsu R5108 photomultiplier provided with a gating circuit designed to enable gate control from an external applied TTL level control signal. Data were processed by an EGG-PAR boxcar integrator. Lifetime measurements were performed with a narrow band (0.08 cm1 linewidth) tunable dye laser of 9 ns pulse width and the emission detected with a Hamamatsu R5108 photomultiplier.
3. Results and discussion 3.1. TRFLN spectra We have investigated the existence of energy transfer among Nd3þ ions in this glass by using TRFLN spectroscopy and observing the emission characteristics of the system as a function of time, concentration, and excitation wavelength. TRFLN
L.M. Lacha et al. / Optical Materials 25 (2004) 193–200
spectra for the 4 F3=2 fi 4 I9=2 transition were performed for all concentrations at 4.2 K by using different resonant excitation wavelengths in the low energy Stark component of the 4 I9=2 fi 4 F3=2 absorption band at different time delays after the laser pulse. As an example, Fig. 1 shows the normalized 4 F3=2 fi 4 I9=2 spectra at 4.2 K for four concentrations 0.5, 2, 4, and 5 mol% of Nd3þ ions obtained at two different time delays after the laser
2 5% 1
0 4
Intensity(arb.units)
4% 2
0 4 2% 2
0 4 0.5 % 2
0 860
880
900
920
940
Emission Wavelength (nm) Fig. 1. Time-resolved fluorescence line-narrowed spectra of the 4 F3=2 fi 4 I9=2 transition obtained at 1 ls (solid line) and 700 ls (dashed line) after the laser pulse by exciting at 874 nm for the samples doped with 0.5, 2, 4, and 5 mol%. Measurements correspond to 4.2 K.
195
pulse, 1 and 700 ls, by exciting at 874 nm. The spectra present two components. The one in the high energy side is a simple FLN line (linewidth 8 cm1 ) corresponding to the emission to the lowest component of the 4 I9=2 multiplet, which is the ground state of the ion. The position of this narrow line is essentially determined by the wavelength of the pumping radiation. The broad emission arises from the non-narrowed inhomogeneous line. As time delay increases the relative intensity of the narrow line and the broad (nonselected) emission changes, and the later becomes stronger, which indicates the existence of energy transfer between discrete regions of the inhomogeneous broadened profile. This effect increases with Nd3þ concentration. As can be observed, when Nd3þ concentration increases the broad nonselected emission becomes more intense. At higher concentrations the transfer process appears at shorter time delays, but in all cases energy transfer produces a relative increase of the broad emission with respect to the narrow band. The 4 F3=2 fi 4 I9=2 emission spectra performed by exciting at different wavelengths along the low energy Stark component of the 4 I9=2 fi 4 F3=2 absorption band show that the broad emission decreases as the excitation energy decreases, which indicates that excitation energy can migrate mainly in one direction. As an example Fig. 2 shows the TRFLN spectra for the sample doped with 2 mol% of Nd3þ obtained at three different excitation wavelengths at 1 ls after the laser pulse. The same behavior is observed in all samples. The time evolution of laser-induced resonant line-narrowed fluorescence is produced by a combination of radiative decay and non-radiative transfer to other nearby ions. Subsequent fluorescence from the acceptor ions replicate the inhomogeneously broadened equilibrium emission profile, showing that transfer occurs not only to resonant sites but to the full range of sites within the inhomogeneous profile. In this case, a quantitative measure of the transfer is provided by the ratio of the intensity in the narrow line to the total intensity of the fluorescence in the inhomogeneous band. Neglecting the dispersion in the radiative decay rate, and using the F€ oster formula for dipole–dipole energy transfer, one can write for the
196
L.M. Lacha et al. / Optical Materials 25 (2004) 193–200
The macroscopic parameter cðEL Þ has the meaning of an integral characteristic, which reflects the average rate of excitation transfer from donors to the ensemble of spectrally non-equivalent acceptors [6]. This analysis deals with migration induced changes in an individual luminescence band originating from transitions between a pair of Stark sublevels [6]. However, in the case of Nd3þ ions, the luminescence spectra represent a superposition of different Stark components. At low temperature, where only the lowest Stark sublevels of the ground and metastable states are populated, the spectra show a narrow resonant Stark component and some broad emission due to transitions to the other Stark components of the ground state, whereas at higher temperatures the spectral pattern becomes more complicated. As a consequence, though at high temperatures spectral migration can be experimentally observed, the analysis in terms of Eq. (1) which considers a twolevel scheme becomes more complicated. We have analyzed the TRFLN spectra of the 4 F3=2 fi 4 I9=2 transition obtained at different time delays between 1 and 700 ls according to Eq. (1) at 4.2 K. As an example, Fig. 3 shows the results for all samples for an excitation wavelength of 874 nm. As can be observed a linear dependence of the
λexc= 880nm 2
0
Intensity(arb.units)
4
λexc= 874nm
2
0 4
λexc= 871nm
2
4.5
880
900
920
940
Excitation Wavelength (nm) Fig. 2. Time-resolved fluorescence line-narrowed spectra of the 4 F3=2 fi 4 I9=2 transition obtained at 1 ls after the laser pulse by exciting at three different wavelengths for the sample doped with 2 mol%. Measurements correspond to 4.2 K.
Ln [(IB /IN)+1]
860
5%
T=4.2 K
4.0 0
3.5
4%
3.0
3%
2.5
2%
2.0 0.5% 1.5
relationship between the integral intensities of the broad background emission IB and the narrow luminescence component IN [6]: IB Ln þ 1 ¼ cðEL Þt1=2 IN
ð1Þ
0
10
20
t1/2
30
( µs1/2 )
Fig. 3. Analysis of the time evolution of the TRFLN 4 F3=2 fi 4 I9=2 emission by means of Eq. (1) for all samples. Symbols correspond to experimental data and the solid lines are fits to Eq. (1). Data correspond to 4.2 K.
L.M. Lacha et al. / Optical Materials 25 (2004) 193–200
3.5
T=4.2 K
2%
871 nm
Ln [(IB /IN )+1]
3.0 2.5
874 nm
700 µ s 500 µs 300 µs 100 µs
T=4.2 K
4
Ln [(IB /IN)+1]
Ln½ðIB =IN Þ þ 1 function on t1=2 was found, which indicates that a dipole–dipole interaction mechanism among the Nd3þ ions dominates in this time regime. The values of the average transfer rate indicate that energy transfer among Nd3þ ions is weak at concentrations up to 2 mol% at low temperature, and increases with increasing concentration. The c values for this excitation wavelength were found to be 1.7 ± 0.6, 19 ± 0.8, 24 ± 0.3, 30 ± 0.7, and 37 ± 0.3 s1=2 for the samples doped with 0.5, 2, 3, 4, and 5 mol% respectively. The average energy transfer rate found in these fluoroarsenate glasses is a 40% lower than the one found in oxide glasses [11] for the same concentration of Nd3þ ions. The analysis of the TRFLN spectra obtained under excitation at longer wavelengths along the low energy Stark component of the 4 I9=2 fi 4 F3=2 absorption band shows that the transfer rate depends on the excitation energy. As an example, Fig. 4 shows the analysis of the TRFLN spectra for the sample doped with 2 mol% of Nd3þ at three different excitation energies. As can be observed the value of the average transfer rate increases from 0.2 ± 0.4 to 36 ± 1.1 s1=2 when the excitation energy increases from 11 389.5 cm1 (878 nm) to
197
3
2 2
3
4
Nd 3+
5
6
(10 20 cm -3)
Fig. 5. Concentration dependence of the quantity LnðIB =IN þ 1Þ for Nd3þ ions. Symbols correspond to experimental data and the solid lines are linear fits. Data were obtained at different time delays after the laser pulse by exciting at 874 nm.
11 481.05 cm1 (871 nm). This rise in the energy migration rate is due to the increasing number of possible acceptors. The concentration dependence of the spectral migration rate can be obtained from the TRFLN spectra taken on different concentration samples after the same time delay. The increase of the intensity of the broad band relative to that of the narrow component shows the concentration induced growth of the migration transfer rate. Fig. 5 shows the concentration dependence of Ln½ðIB =IN Þ þ 1 obtained at 4.2 K at 100, 300, 500, and 700 ls after the laser pulse for the samples doped with 2, 3, 4, and 5 mol% of Nd3þ . The observed linear behavior is again consistent with a dipole–dipole energy transfer mechanism.
2.0 878 nm
1.5 1.0 0
10
20
30
t1/2 (µs1/2 ) Fig. 4. Analysis of the time evolution of the TRFLN 4 F3=2 fi 4 I9=2 emission by means of Eq. (1) for the sample doped with 2 mol% of Nd3þ for three different excitation wavelengths. Symbols correspond to experimental data and the solid lines are fits to Eq. (1). Data correspond to 4.2 K.
3.2. Temperature dependence concentration quenching of Nd3þ fluorescence The decays of the 4 F3=2 fi 4 I9=2 transition were found to be single exponential at all temperatures and concentrations. As an example, Fig. 6 shows the room temperature logarithmic plot of the experimental decays for the samples doped with 0.5, 3, and 5 mol%, obtaining by exciting at 580 nm and collecting the luminescence at 890 nm. The temperature dependence of the lifetimes were
198
L.M. Lacha et al. / Optical Materials 25 (2004) 193–200
400
-1
λexc= 580nm
0.5%
Lifetime (µs)
2%
0.5%
-2
300
3% 4%
200
-3
5%
100 0
50
100
150
200
250
300
Ln [I(t) (arb.units)]
Temperature (K) -1
Fig. 7. Temperature dependence of the 4 F3=2 lifetime of Nd3þ for different concentrations. Lifetimes were obtained by exciting at 580 nm.
3%
-2
-3
-2
5%
-3 -4 -5 -6 0
250
500
750
1000
Time ( µs) Fig. 6. Logarithmic plot of the fluorescence decays of the 4 F3=2 state for samples doped with 0.5, 3, and 5 mol% of Nd3þ . The decays were obtained by exciting at 580 nm and collecting the luminescence at 890 nm. Measurements correspond to 295 K.
obtained in the 4.2–295 K temperature range for five different concentrations 0.5, 2, 3, 4, and 5 mol% of Nd3þ in fluoroarsenate glass. Fig. 7 shows the lifetime values as a function of temperature. As can be seen in this figure, as concentration rises the
experimental lifetime decreases even at helium temperature, indicating that Nd–Nd relaxation processes are present at concentrations higher than 0.5 mol%. This thermal behavior also suggests the presence of quite a strong thermal mechanism between 10 and 80 K at concentrations higher than 0.5 mol%. In a previous work [16] some of the authors have investigated the temperature dependent concentration quenching of Nd3þ fluorescence in three pure fluoride glasses which shows a strong thermal quenching between 10 and 90 K where the non-radiative rates were found to increase with temperature as T 3 . This is in agreement with a two-site non-resonant process in the shortwavelength regime as has been shown by Holstein et al. [17]. In this process, the phonon emission and absorption take place at different sites, the ion– phonon interaction acts twice, and the site–site Hamiltonian once. This phonon-assisted excitation transfer between different states of similar ions could be understood by considering a small spread in the energy values of 4 F3=2 transitions to the J multiplicity, due to the inherent disorder of the glass structure. The lifetime of the 4 F3=2 state of Nd3þ ion in the doped glasses should be governed by the sum of probabilities for several competing processes such as radiative decay, non-radiative decay by multiphonon emission, and energy transfer to other
L.M. Lacha et al. / Optical Materials 25 (2004) 193–200
[WNd-Nd(T) - WNd-Nd(0)]1/3
15 5% 12
4%
9
3% 6
0
20
40
60
80
100
Temperature (K) Fig. 8. Plot of the non-radiative Nd–Nd relaxation rates in the 10–80 K temperature range. Symbols represent the experimental values, and solid lines are the fit to a T 3 dependence.
Nd3þ ions. In this case the multiphonon relaxation rate for the 4 F3=2 state is expected to be small because of the high energy gap to the next lowerlying J manifold (5500 cm1 ) and the typical values of phonon energies involved ( 6 878 cm1 ) [18]. If one assumes that the purely radiative lifetimes of 4 F3=2 levels are independent of Nd3þ concentration and temperature, and disregards multiphonon relaxation processes, the variations of the experimental lifetimes can be related with the non-radiative Nd–Nd relaxation processes by the simple relation: 1 s1 exp ¼ sR þ WNd–Nd ðT Þ
ð2Þ
where sexp is the measured lifetime and sR represents the radiative lifetime. Fig. 8 shows the plot of ½WNd–Nd ðT Þ WNd–Nd 1=3 ð0Þ in the 10–80 K temperature range for the samples doped with 3, 4, and 5 mol% of Nd3þ . It is worth noticing the good fit of WNd–Nd to a T 3 dependence. This result holds for all concentrations higher than 0.5 mol%.
4. Conclusions (i) Resonant time-resolved fluorescence line-narrowing spectroscopy shows the existence of energy migration between discrete regions of
199
the inhomogeneous broadened 4 F3=2 spectral profile. The analysis of the 4 F3=2 fi 4 I9=2 emission spectra obtained at low temperature as a function of time, concentration, and excitation wavelength shows that the electronic mechanism responsible for the Nd3þ –Nd3þ interaction can be interpreted in terms of a dipole–dipole energy transfer. (ii) The emission lifetimes from the 4 F3=2 state are single exponentials at all temperatures and concentrations. As the concentration rises the decay remains single exponential but a decrease in the lifetime is observed even at helium temperature indicating that relaxation via fast Nd–Nd diffusion processes occurs. (iii) The non-radiative Nd–Nd relaxation rates of the 4 F3=2 fi 4 I9=2 transition were found to increase with temperature as T 3 in the 10–80 K temperature range for concentrations higher than 0.5 mol%. This behavior, also found in pure fluoride matrices, is in agreement with a two-site non-resonant process. Acknowledgements This work has been supported by the Spanish Government (Ref.: MAT2000-1135 and BFM20000352) and the Basque Country University (Ref.: UPV13525/2001). References [1] L.A. Riseberg, Phys. Rev. A 7 (1973) 671. [2] M.J. Weber, in: W.M. Yen, P.M. Selzer (Eds.), Laser Spectroscopy of Solids, vol. 49, Springer, Berlin, 1981. [3] S.A. Brawer, M.J. Weber, Appl. Phys. Lett. 35 (1979) 31. [4] L.D. Merkle, R.C. Powell, E.E. Freed, J. Lumin. 24/25 (1981) 755. [5] L.A. Riseberg, Solid State Commun. 11 (1972) 469. [6] T.T. Basiev, V.A. Malyshev, A.K. Prhevuskii, in: A.A. Kaplyanskii, R.M. Macfarlane (Eds.), Spectroscopy of Solids Containing Rare Earth Ions, North-Holland, Amsterdam, 1987. [7] W.M. Yen, in: A.A. Kaplyanskii, R.M. Macfarlane (Eds.), Spectroscopy of Solids Containing Rare Earth Ions, North-Holland, Amsterdam, 1987. [8] M.M. Broer, D.L. Huber, W.M. Yen, W.K. Zwicker, Phys. Rev. Lett. 49 (1982) 394. [9] T.T. Basiev, Yu.K. VoronÕko, S.B. Mirov, A.M. Prokhorov, JETP Lett. 29 (1979) 639.
200
L.M. Lacha et al. / Optical Materials 25 (2004) 193–200
[10] R. Balda, J. Fern andez, M. Sanz, A. de Pablos, J.M. FdezNavarro, J. Mugnier, Phys. Rev. B 61 (2000) 3384. [11] R. Balda, M. Sanz, J. Fernandez, J.M. Fdez-Navarro, J. Opt. Soc. Am. B 17 (2000) 1671. [12] N. Duhamel-Henry, J.L. Adam, B. Jacquier, C. Linares, Opt. Mater. 6 (1996) 197. [13] R. Balda, J. Fernandez, J.L. Adam, M.A. Arriandiaga, Phys. Rev. B 54 (1996) 12076. [14] R. Balda, J. Fern andez, I. Saez de Ocariz, J.L. Adam, A. Mendioroz, E. Montoya, Opt. Mater. 13 (1999) 159.
[15] J.L. Adam, R. Balda, I. Melsco€et, F. Smektala, L.M. Lacha, J. Fernandez, J. Non-Cryst. Solids 256–257 (1999) 390. [16] R. Balda, J. Fernandez, A. Mendioroz, J.L. Adam, B. Boulard, J. Phys.: Condens. Matter 6 (1994) 913. [17] T. Holstein, S.K. Lyo, M. Orbach, in: W.M. Yen, P.M. Selzer (Eds.), Laser Spectroscopy of Solids, vol. 49, Springer, Berlin, 1981. [18] R.A. Nyquist, R.O. Kagel, Infrared Spectra of Inorganic Compounds, Academic Press, New York, 1971.