JOURNAL OF APPLIED PHYSICS 109, 113108 (2011)
Luminescence properties and optical dephasing in a glass-ceramic containing sodium-niobate nanocrystals E. Almeida,1 L. de S. Menezes,1,a) Cid B. de Arau´jo,1 and A. A. Lipovskii2 1
Departamento de Fı´sica, Universidade Federal de Pernambuco, 50670-901 Recife, Pernambuco, Brazil 2 St. Petersburg State Polytechnical University, 29 Polytechnicheskaja, St. Petersburg 195251, Russia
(Received 25 February 2011; accepted 26 April 2011; published online 10 June 2011) Photoluminescence (PL) and degenerate four-wave-mixing (DFWM) experiments were performed in a silica–niobic composite containing NaNbO3 nanocrystals. The PL results indicate the presence of in-gap states attributed to excitons in the nanocrystals and defect centers. The luminescence of the samples becomes more intense at low temperatures, indicating that nonradiative relaxations dominate the dynamics of the in-gap states. The DFWM experiments allowed for measurements of the homogeneous relaxation time, (20 6 3) fs, of the third-order polarization at room temperature. The main contributions to the dynamics of the electronic response are attributed to the trapping of C 2011 American Institute of electrons in the in-gap states and to carrier and phonon scattering. V Physics. [doi:10.1063/1.3596518]
I. INTRODUCTION
Glass-ceramics (GC), i.e., glasses containing some amount of micro- or nanocrystals, are very attractive materials for photonic applications such as microchip lasers, displays, and optical waveguides/fibers. In the past, a large effort was dedicated to developing methods of fabricating transparent GC, as is reviewed in Refs. 1–9 and references therein. The works were justified because, in principle, the fabrication of GC is simple and might allow flexibility for controlling their physical properties through the size distribution of the inclusions, and through the ratio between the volume occupied by the micro-/nanocrystals and the volume of the sample, the so-called filling fraction, f. Also, GC might present new physical properties as compared to those of glasses or crystals having the same composition. For example, depending on the size of the nanocrystals, they might exhibit quantum confinement effects.1,10 Furthermore, the mismatch between the dielectric functions of the host and of the nanocrystal can enhance the linear and nonlinear (NL) optical properties of the GC due to the increased local electromagnetic field inside and in the vicinity of the nanocrystals’ surface. However, the nanocrystals might give origin to undesirable light scattering when their size and/or the average spacing between them and/or their aggregates are comparable to the light wavelength. Another problem is the presence of electron traps due to surface defects. Thus, a careful balance between the competing effects must be taken into account when designing a new GC. GC containing nanocrystals of the ABO3 group (for example, LiNbO3, BaTiO3, NaNbO3) are attracting large interest because these crystals are widely used in electro-optical and NL optical applications, such as enhanced second harmonic generators.11–18 Indeed, GC containing sodium niobate a)
Author to whom correspondence should be addressed. Electronic mail:
[email protected].
0021-8979/2011/109(11)/113108/5/$30.00
(NaNbO3) nanocrystals (GC-SNN) proved to be an interesting material for all-optical switching and optical limiting applications.15–18 The fabrication of GC-SNN is reported in Refs. 9 and 14, in which a detailed investigation of the structural transformations from the vitreous state to the GC state is described and the conditions under which to prepare samples with controlled properties are identified. The NL optical response of the GC-SNN was characterized at 532 and 1064 nm with 80 ps pulses, and it was found that the crystalline phase enhances its nonlinear refractive index n2 and the twophoton absorption (TPA) coefficient a2 .16 The temporal response of the nonlinearity was investigated in Refs. 15 and 16 using 100 fs pulses at 800 nm, and a response faster than the excitation pulses was revealed. The figure of merit, n2 =a2 k, where k is the laser wavelength, indicated the large potential of the GC-SNN for all-optical switching. The optical limiting behavior of the GC-SNN was investigated using nanosecond pulses in the visible and infrared regions.17,18 In these works, the nanosecond TPA coefficient is substantially larger than the picosecond TPA coefficient determined in Ref. 16, and this result was understood by taking into account contributions due to absorption by free-carriers and states with energies smaller than the nanocrystals’ bandgap. In particular, we recall the large dependence of a2 as a function of f when the laser wavelength was 580 nm, which clearly indicates possible resonance with in-gap states. The aim of the present work is to investigate states having energy smaller than the energy bandgap and their contribution to the optical response of the GC-SNN. Photoluminescence (PL) excited by nanosecond lasers and degenerate four-wave mixing (DFWM) with partially coherent nanosecond pulses were the techniques used. Luminescence originating from states with lifetimes of 500 ns was observed. The ultrafast dephasing of the third-order susceptibility was studied, and a homogeneous relaxation time of 20 fs was measured for wavelengths resonant with exciton states. The results indicate the presence of in-gap states
109, 113108-1
C 2011 American Institute of Physics V
Author complimentary copy. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp
113108-2
Almeida et al.
associated with excitons, trapping states associated with defects, and the possible influence of the starting compounds used in the fabrication of the samples. II. EXPERIMENTAL DETAILS
The samples were prepared as described in Refs. 9 and 14. The glass with a molar % composition of 35SiO2– 31Nb2O5–19Na2O–11K2O–2CdO–2B2O3 was heat-treated at 610 C for different time intervals. The nucleation of NaNbO3 nanocrystals during the heat-treatment procedure was confirmed through x-ray analysis, as shown in Fig. 1(a). Samples treated for longer times presented higher values of f, but after 69 h of treatment, saturation of f was observed; the size of the nanocrystals was verified to be independent of the heat-treatment time, as previously confirmed by small angle x-ray scattering.14–16 Figure 1(b) shows a representative size distribution of the nanocrystals as determined using a transmission electron microscope (TEM); the inset of Fig. 1(b) shows a TEM image of the sample that has f ¼ 0.37. The log-normal curve adjusted to the histogram indicates that the average size of the nanocrystals is 12 nm.
J. Appl. Phys. 109, 113108 (2011)
PL experiments were performed using the samples heattreated for 2 h (f ¼ 0) and 206 h (f ¼ 0.37). The excitation was made with the third harmonic of the Nd:YAG laser (355 nm, 5 ns, 5 Hz), and the PL signal was analyzed using a 0.25 ˚ ) attached to a m spectrometer (spectral resolution: 25 A 1P28 A photomultiplier. The samples were mounted in a cryostat, and their temperature could be varied from 300 to 17 K. The PL signals were averaged by a boxcar and processed using a personal computer. For the dephasing time measurements, the sample heat treated for 206 h (f ¼ 0.37) was chosen because it has the highest NL optical susceptibility16 and because of our interest in investigating the influence of the NaNbO3 nanocrystals on the NL response. DFWM experiments were performed at room temperature, using a broad band dye laser (BBQ dye,19 2 103 M in a 1:1 ethanol/toluene solution) pumped by the third harmonic of a Nd:YAG laser (355 nm, 5 ns, 2 Hz). The broadband dye laser delivers ultraviolet light centered at 386 nm (3.21 eV) with a bandwidth Dk 2.6 nm, which corresponds to a coherence time sC 170 fs, according to sC ¼ 4k2 ln2=cpDk. III. RESULTS AND DISCUSSION A. Photoluminescence properties
FIG. 1. (a) X-ray diffractograms for samples heat treated for 2 h (lower curve) and 206 h (upper curve). The peaks labeled with an asterisk are characteristic of quasicubic crystalline NaNbO3. (b) Representative size distribution of the NaNbO3 nanocrystals in the glass-ceramic. Inset: TEM image of the sample with f ¼ 0.37.
The absorption spectra of the two samples with f ¼ 0 and f ¼ 0.37 at room temperature, as well as the temperature dependence of the energy bandgap Eg , are shown in Fig. 2. A large transparency window extending from the blue to the infrared region is observed for both samples. The small linear absorption coefficient a0 in the transparent region indicates that the amount of in-gap states is small and that clusters of the original starting components are in small amounts or have an indirect energy gap. The value of Eg was calculated assuming the dependence of a0 on Eg , given by a20 / hx Eg . Note that Eg for the sample with f ¼ 0.37 goes from 3.20 eV (300 K) to 3.32 eV (17 K), probably due to the thermal compression of the sample. Figure 3 shows the PL spectra of the two samples for excitation in the ultraviolet range (355 nm, 3.51 eV, 5 ns). Comparing the results of Figs. 3(a) and 3(b), we conclude that the two PL bands centered at 430 nm and 540 nm, observed only in the sample having f ¼ 0.37, are due to the presence of the NaNbO3 nanocrystals. The signal at 430 nm is attributed to direct exciton recombination, and the broad bandwidth, the Stokes shift, and the low quantum efficiency indicate that the emission at 540 nm originates from trap states. Note that the emission at 540 nm is more intense at low temperatures, indicating that nonradiative processes are suppressed when the temperature is lowered. The emission band centered at 680 nm is attributed mainly to the presence of the oxygen dangling bond : SiO* (the nonbridging oxygen hole center [NBOHC]).20 This defect is often present in multicomponent glasses. However, we cannot exclude the contribution from CdO clusters, whose bandgap is 2.07 eV.21 This emission band was further investigated in an independent experiment, in which we excited the samples at 575 nm (2.16 eV), off-resonance with nanocrystals’ states. The obtained spectrum is similar to the one
Author complimentary copy. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp
113108-3
Almeida et al.
J. Appl. Phys. 109, 113108 (2011)
FIG. 2. (Color online) (a) The linear absorption coefficient a0 at different temperatures for the GC-SNN with f ¼ 0.37. (b) Temperature dependence of the band-gap energy. Squares: sample with f ¼ 0. Circles: sample with f ¼ 0.37.
FIG. 3. (Color online) Photoluminescence spectra (not intensity-correlated) for (a) the sample with f ¼ 0.37 and (b) the sample with f ¼ 0 (excitation at 355 nm in both cases).
presented in Ref. 21, in which the spectrum of CdO is investigated. The band at 680 nm is not observed in the PL spectrum of sample f ¼ 0.37 when pumped at 355 nm (3.51 eV) because the absorption coefficient of the NaNbO3 nanocrystals at 3.51 eV is much larger than that of the host matrix, whose bandgap is 3.57 eV, thus strongly absorbing the excitation photons. Figure 4 shows the temporal evolution of the PL band at 540 nm, excited by 355 nm, for the sample with f ¼ 0.37. Note that the PL signal presents a fast decay of 25 ns followed by a slow decay (500 ns) for temperatures smaller than 60 K. The longer decay times at low temperatures are attributed to long-living in-gap states. The strong peak common to all temperatures is related to the impulsive response of the detection system to the excitation pulse and determines the temporal resolution for this kind of measurement. In contrast, the temporal decay of the emission centered at 430 nm is temperature independent because the probability of nonradiative relaxation is low due to the large energy bandgap.
Third-order experiments performed at 532 nm and 1064 nm with samples having 0 < f < 0.4 provided values for n2 varying from 0.2 1014 to 0.8 1014 cm2/W (excitation at 532 nm) and þ0.2 1014 to þ0.5 1014 W/cm2 (excitation at 1064 nm).16 Moreover, previous results for excitation at 800 nm in a Kerr-gate setup have shown that the time response of the samples is faster than 100 fs (which is the experimental temporal resolution, limited by the 100 fs laser pulse duration).15,16 However, it is desirable to operate
B. Wave-mixing experiment with incoherent light
The relaxation rate of the third-order polarization induced in the GC-SNN is another important parameter when evaluating the potential of GC-SNN for all-optical switching using photon energies smaller than the energy gap.
FIG. 4. (Color online) Temporal behavior of the luminescence at 525 nm (excitation wavelength: 355 nm) for the sample with f ¼ 0.37.
Author complimentary copy. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp
113108-4
Almeida et al.
ultrafast switches based on materials with a large nonlinearity that would require a small pump intensity. A larger nonlinearity can be achieved by exciting the sample near resonance with exciton states, and thus it becomes important to know the time response of the third-order nonlinearity for light excitation near the bandgap region. Therefore, the light source used was a pulsed dye laser operating in the near UV region. DFWM experiments were performed using a broadband dye laser centered at 386 nm, with a coherence time sC 170 fs. The central wavelength was chosen to be near resonance with the inhomogeneously broadened exciton transitions of the NaNbO3 nanocrystals; accordingly, the photon energies were smaller than the energy bandgap of the glass matrix. The DFWM technique with incoherent light is well known, and it has been useful in investigating different physical processes in the condensed matter phase. As an example, we refer to Ref. 22, in which the ultrafast excitonic dephasing of polyaniline has been measured and identified as the result of the different degrees of torsion of adjacent benzene rings along the molecular structure, affecting the electronic transfer integral. In Ref. 23, the populational decay times of the nonfluorescent electronic levels of an organic saturable absorber were measured using an interferometric variant of the incoherent four-wave mixing technique. Finally, the ultrafast orientational dynamics of mesoionic molecules in solution have been recorded by using a Kerr gate technique with incoherent light.24 The DFWM setup used here is similar to the one presented in Ref. 25. The linearly polarized broadband laser beam is split into two beams with wavevectors k1 and k2 , which have equal intensities. The k1 beam is delayed with respect to the other beam; the delay time s is negative when the k1 beam arrives first in the sample. The polarization of the k2 beam could be controlled using a k=2 plate in order to avoid the formation of a thermal grating. The two beams are then focused in the sample by a 20 cm focal length lens, crossing in a small angle of 10 mrad. Depending on the light wavelength used, absorptive and/or refractive index gratings can be created in the overlap region of the beams. Then, photons are exchanged between the incident beams and create new beams that are diffracted along well-defined directions. The diffracted beams propagating along the k3 ¼ 2k1 k2 and k4 ¼ 2k2 k1 directions were collected by optical fibers coupled to photomultipliers, and the electric signals were sent to a boxcar and a computer. When the inhomogeneous broadening of the excited states is dominant over the homogeneous one, the intensity profiles of the k3 and k4 signals, as a function of s, can be approximated by Gaussian curves.25 The maxima of the diffracted signals occur at different relative delays of the input beams, and the peaks’ separation sS can be related to T2 , the homogeneous u½ð2uþ1=uÞð4u2 þ41=u2 Þ expðu2 Þ erfcðuÞ ; relaxation time, by TsS2 ¼ 1ð2u1=uÞ expðu2 ÞerfcðuÞ where u 1=ðcT2 Þ is the ratio between the homogenous and inhomogeneous Ð 1linewidths, and the error function is defined by erfcðuÞ u expðx2 Þdx.25 A symmetrical signal profile indicates that T2 is smaller than the light correlation time. Furthermore, although the auto-correlation signal has a width
J. Appl. Phys. 109, 113108 (2011)
dependent on sC , the time resolution of the experiment is determined by sS , which can be measured with good accuracy. We adopted the analysis procedure introduced in Ref. 25. The intensity profiles of the k3 and k4 beams as a function of s were symmetric using either parallel or perpendicular polarizations of the incident beams. This indicates that thermal effects are negligible, and hence the measurements were made using parallel polarizations in order to obtain a better signal-to-noise ratio. Each data point shown in Fig. 5 represents an average over five laser shots. The signals’ profiles could be fitted using Gaussian functions, and the separation between their peaks gives the homogeneous relaxation time T2 ¼ (20 6 3) fs, which corresponds to the homogenous linewidth C ¼ 2h/T2 ¼ 66 meV. The error in the measurement of T2 is estimated in another experiment, splitting one of the diffracted beams into two beams that are sent to the two different detection channels and running the apparatus for different values of s. Thus, the separation between the two signals as a function of s gives the uncertainty in the measurement of 6 3 fs. The homogeneous relaxation time is given by P 1=T2 ¼ 1=2T1 þ i 1=Ti , where T1 is the population lifetime and Ti* takes into account the contribution of dephasing mechanisms such as (i) carrier-carrier scattering, (ii) carrierphonon coupling, and (iii) carrier trapping on the nanocrystals’ surface.26 Unfortunately, under the present conditions, we cannot make a quantitative evaluation of the individual mechanisms contributing to T2 . Of course, the phonon scattering mechanism is very important at room temperature. However, the contribution of the carrier-phonon coupling to the nonlinearity could not be identified because at lower temperatures the value of Eg changes according to Fig. 2(b), and the dye laser central wavelength could not be tuned. Therefore, the contributions of carrier scattering and trap states for T2 can be evaluated only in a qualitative way. Because we have no information with which to determine the exciton Bohr radius aB , we considered aB 6 nm, as for the III-V and II-VI semiconductors with Eg > 1.8 eV.26,27 Thus, quantum confinement effects could be present in our system, and if that is the case it will contribute to the enhancement of the dephasing rates,
FIG. 5. (Color online) Self-diffracted signals in the k3 (circles) and k4 (squares) directions. The solid lines represent the best fitting of the Gaussian curves. The separation between the peaks is ss ¼ (20 6 3) fs.
Author complimentary copy. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp
113108-5
Almeida et al.
given that the electrons’ wavefunctions “see” the nanocrystals’ walls. Carrier-carrier scattering appears as another important factor, because the laser intensity used (11 MW/cm2) is enough to create more than one electron–hole pair in each NaNbO3 nanocrystal. Another important process that might contribute to T2 is the fast trapping of electrons in defects on the nanocrystals’ surface. The presence of such defects, assigned as dangling bonds on the nanocrystal-matrix interface, was confirmed by the PL experiments, but it is not possible to separate this contribution from the carriers’ scattering contribution. Therefore, two important conclusions have come out of the experiments reported here. First, we concluded that NL optical experiments with GC-SNN, in the visible range, might be affected by in-gap states associated with matrix defects (NBOHCs) and nanocrystal surface defects. For instance, the TPA coefficients measured in the experiment described in Ref. 18 were enhanced by one-photon resonance with in-gap states, although the presence of these states is not observed in the linear absorption spectra. Second, by selecting the laser wavelength to be in the exciton spectral region, we measured a 20 fs relaxation time for the NL polarization. This result indicates that for the used wavelength, the NL process exploited was essentially nonresonant, and thus it demonstrates the possibility of ultrafast all-optical switching in the blue region in which not many materials were identified. It is important to note that the large diffracted signal with a good signalto-noise ratio shown in Fig. 5 was obtained using low intensities (107 W/cm2). IV. SUMMARY
In the present experiments, we studied relaxation processes related to the population of in-gap states and the optical dephasing related to the third-order polarization in niobic-silicate glass-ceramics. Photoluminescence experiments showed that emission from states located in the gap of the NaNbO3 nanocrystals and emissions due to the original constituents of the glass provide an important contribution to the samples’ luminescence at low temperatures. Self-diffraction experiments with partially coherent light allowed the measurement of the NL polarization dephasing time related to the sodium niobate nanocrystals. The small value of T2 ¼ (20 6 3) fs indicates that this material might be very useful for all-optical switching in the blue region.
J. Appl. Phys. 109, 113108 (2011)
ACKNOWLEDGMENTS
The authors acknowledge the Brazilian agencies Conselho Nacional de Desenvolvimento Cientı´fico e Tecnolo´gico (CNPq) and Fundac¸a˜o de Amparo a` Cieˆncia do Estado de Pernambuco (FACEPE), and the Russian National Program “Development of High School Potential.” We also thank B.J.P. da Silva for cutting and polishing the samples, Dr. Marco Sacilotti for helping with the low temperature system, and A.M.B. Silva for measuring the size distribution of the NaNbO3 nanocrystals. 1
G. H. Beall and D. A. Duke, J. Mater. Sci. 4, 340 (1969). N. F. Borrelli and M. M. Leyton, J. Non-Cryst. Solids 6, 197 (1971). 3 A. V. Belyakov and A. N. Sukhozhak, Glass Ceram. 52, 14 (1995). 4 R. Apetz and M. P. B. van Bruggen, J. Am Ceram. Soc. 86, 480 (2003). 5 M. Yamane and Y. Asahara, Glasses for Photonics (Cambridge University Press, Cambridge, England, 2000). 6 J. Dong, K. Ueda, H. Yagi, and A. A. Kaminskii, Opt. Rev. 15, 57 (2008). 7 P. A. Tick, N. F. Borrelli, and I. M. Reaney, Opt. Mater. 15, 81 (2000). 8 J. Dong, K. I. Ueda, and A. A. Kaminskii, Opt. Express 15, 14516 (2007). 9 A. A. Lipovskii, D. K. Tagantsev, B. V. Tatarintsev, and A. A. Vetrov. J. Non-Cryst. Solids 318, 268 (2003). 10 P. N. Prasad, Nanophotonics (Wiley, New York, 2004). 11 E. V. Rodriguez, C. B. de Arau´jo, A. M. Brito-Silva, V. I. Ivanenko, and A. A. Lipovskii, Chem. Phys. Lett. 467, 335 (2009). 12 S. K. Mishra, N. Choudhury, S. L. Chaplot, P. S. R. Krishna, and R. Mittal, Phys. Rev. B 76, 024110 (2007). 13 A. Malakho, E. Fargin, M. Lahaye, B. Lazoryak, V. Morozov, G. V. Tendeloo, V. Rodriguez, and F. Adamietz, J. Appl. Phys. 100, 063103 (2006). 14 A. A. Zhilin, G. T. Petrovsky, V. V. Golubkov, A. A. Lipovskii, D. K. Tagantsev, B. V. Tatarintsev, and M. P. Shepilov, J. Non-Cryst. Solids 345–346, 182 (2004). 15 C. A. C. Bosco, E. L. Falca˜o-Filho, G. S. Maciel, L. H. Acioli, C. B. de Arau´jo, A. A. Lipovskii, and D. K. Tagantsev, J. Appl. Phys. 94, 6223 (2003). 16 E. L. Falca˜o-Filho, C. A. C. Bosco, G. S. Maciel, L. H. Acioli, C. B. de Arau´jo, A. A. Lipovskii, and D. K. Tagantsev, Phys. Rev. B 69, 134204 (2004). 17 G. S. Maciel, N. Rakov, C. B. de Arau´jo, A. A. Lipovskii, and D. Tagantsev, Appl. Phys. Lett. 79, 584 (2001). 18 T. R. Oliveira, L. de S. Menezes, C. B. de Arau´jo, and A. A. Lipovskii, Phys. Rev. B 76, 134207 (2007). 19 See http://www.exciton.com/catalog/newcat.htm for details. 20 L. Skuja, J. Non-Cryst. Solids 179, 51 (1994). 21 S. Zou, V. V. Volkov, and Z. L. Wang, Chem. Mater. 11, 3037 (1999). 22 L. de S. Menezes, C. B. de Arau´jo, E. H. L. Azevedo, and W. M. de Azevedo, Chem. Phys. Lett. 377, 647 (2003). 23 V. Kozich, L. de S. Menezes, and C. B. de Arau´jo, Opt. Lett. 26, 262 (2001). 24 L. de S. Menezes, C. B. de Arau´jo, M. A. R. C. Alencar, P. F. AthaydeFilho, J. Miller, and A. M. Simas, Chem Phys. Lett. 347, 163 (2001). 25 T. Kobayashi, A. Terasaki, T. Hattori, and K. Kurokawa, Appl. Phys. B 47, 107 (1988). 26 U. Woggon, Optical Properties of Semiconductor Quantum Dots (Springer, Berlin, 1997). 27 M. Fox, Optical Properties of Solids (Oxford University Press, Oxford, England, 2001). 2
Author complimentary copy. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp
