- Email: [email protected]
Optical and nonlinear optical properties of copper nanocomposite glasses annealed near the glass softening temperature
Available online at www.sciencedirect.com
Optics Communications 281 (2008) 2933–2937 www.elsevier.com/locate/optcom
Optical and nonlinear optical properties of copper nanocomposite glasses annealed near the glass softening temperature B. Karthikeyan a,*, M. Anija a, C.S. Suchand Sandeep a, T.M. Muhammad Nadeer b,1, Reji Philip a a
Light and Matter Physics Group, Raman Research Institute, C.V. Raman Avenue, Sadashivanagar, Bangalore 560 080, India b International School of Photonics, Cochin University of Science and Technology, Cochin 682 022, India Received 17 August 2007; received in revised form 7 December 2007; accepted 8 January 2008
Abstract Copper nanocomposite glasses have been prepared by the ion-exchange method, and annealed at different temperatures up to and above the glass softening temperature. The absorption spectra, fluorescence spectra, and nonlinear optical transmission of the samples at 532 nm for nanosecond laser pulses, have been investigated. The optical and nonlinear optical properties of the glasses are found to be distinctly different below and above the glass softening temperature. For instance, thermal annealing up to the glass softening temperature makes the samples behave like saturable absorbers, while annealing at higher temperatures makes them behave like optical limiters. Such flexibility in controlling the optical nonlinearity in these materials makes them potential candidates for photonic applications. Ó 2008 Elsevier B.V. All rights reserved. PACS: 61.46.+w; 78.67.Bf; 42.50.p Keywords: Metal nanocomposite; Nonlinear absorption; Optical limiting; Saturable absorption; Surface plasmon resonance
1. Introduction Metal nanocluster composite glasses (MNCGs) are potential candidates for photonic nanodevice fabrication. For instance, their large and fast optical nonlinearity can find applications in optical switching devices. Nanocomposite glasses can be prepared through the sol–gel [1], melt-quench [2], ion implantation [3,4] and ion-exchange [5] methods. Among these, ion-exchange is one of the oldest and most successful techniques. Noble metal clusters (copper, silver and gold) have gained a lot of attention *
Corresponding author. Fax: +91 (0)80 23610492. E-mail addresses: [email protected] (B. Karthikeyan), [email protected] (R. Philip). 1 Present address: Amal Jyothi College of Engineering, Kanjirappally 686 518, India. 0030-4018/$ - see front matter Ó 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.optcom.2008.01.032
because of their optical nonlinearities, which are substantial due to the surface plasmon resonances (SPR) lying in the visible region. The dispersive third order nonlinear susceptibility vð3Þ m of metal nanoclusters will be very strong when excited near the SPR, due to the local field enhancement effect. The use of MNCGs as ultrafast optical switches is known before [6], and there are previous reports on their near and off-resonant nonlinear optical properties at various cluster sizes [3,7]. Thermal annealing is a common route used for changing the particle size. In the present work, we have prepared copper nanocomposite glasses by the ion-exchange method, which are annealed at different temperatures up to and above the glass softening temperature (600 °C). The absorption and fluorescence spectra of the samples have been measured. The nonlinear optical transmission of nanosecond laser pulses through the samples also has been measured at the
2934
B. Karthikeyan et al. / Optics Communications 281 (2008) 2933–2937
wavelength of 532 nm. The nonlinearity is found to originate mostly from the Kerr nonlinearity and free carrier absorption occurring in the samples. In general, the samples are found to behave as saturable absorbers below the glass softening temperature, and optical limiters above the glass softening temperature. Thus the glass softening temperature turns out to be a critical control parameter for optical nonlinearity in these materials. This observation may be generally true for glass hosts embedded with nonlinear media, in which case it is important information for fabricating photonic devices with controlled and predesigned optical nonlinearities. 2. Experimental In our experiment, metal ions are incorporated into sodalime glass through the thermal ion-exchange method. Commercial soda lime glasses are initially cleaned using an ultrasonic cleaner. The glass slides are then immersed in a molten salt bath containing CuSO4 and Na2SO4 in the 1:3 wt. ratio. The ion-exchange process is allowed to take place at 500–560 °C for a period of 1 min. The copper ions in the molten salt bath penetrate and occupy the sites left by the Na+ ions which are the glass modifiers in the host matrix. The prepared glass slides are air annealed at 500, 550, 600 and 650 °C (Cu-500, Cu-550, Cu-600 and Cu-650, respectively) for 5 h and then furnace cooled. Transmission electron micrographs (TEM) of the as-prepared glass, obtained using a Teenai F30 machine, are shown in Fig. 1. The pictures reveal that the as-prepared Cu nanoclusters are about 3–4 nm in diameter. Optical absorption spectra were recorded using a dual beam Perkin Elmer spectrophotometer. Steady state fluorescence spectra were recorded using a Perkin Elmer spectrofluorometer, with a xenon arc lamp source and a photomultiplier detection system. Samples were excited at 340 nm and the fluorescence spectra were recorded from 550 to 660 nm.
To measure the optical nonlinearity, open aperture zscan measurements were done at 532 nm using 7 ns laser pulses from a frequency-doubled Nd:YAG laser (Quanta Ray-Spectra Physics). In the z-scan the laser beam is focused using a lens, and the sample is translated along the beam axis (z-axis) through the focal region over a distance several times that of the diffraction length. At each position z the sample sees a different laser intensity, and the position dependent (ie, intensity-dependent) transmission is measured using an energy meter placed after the sample. Laser pulses are fired at a repetition rate of 1 Hz, and the data acquisition is automated. The low repetition rate is chosen for avoiding heating-up of the samples during measurement. The pulse energy reaching the sample is approximately 60 lJ. 3. Results and discussion The optical absorption spectra of the Cu nanocomposite glasses are shown in Fig. 2. All spectra show the surface plasmon resonance (SPR) around 560 nm. Under the dipole approximation, Mie theory gives the extinction cross-section of spherical nanoparticles as [8] rext ðxÞ ¼ 9p
emðRÞ ðxÞ x 3=2 e c d ½emðIÞ ðxÞ þ 2ed 2 þ e2mðRÞ ðxÞ
ð1Þ
where p is the volume fraction, and x is the applied optical frequency. The SPR maximum occurs at em ðlÞðxÞþ 2ed ¼ 0 where em ðxÞ ¼ emðRÞ ðxÞ þ iemðlÞ ðxÞ is the frequency dependent, complex dielectric constant of the metallic particle, and ed is the dielectric constant of the host matrix which is assumed to be real and nondispersive. The SPR is superposed over the interband absorption that gains strength in the short-wavelength region in the measured spectral range. The net absorption increases with annealing temperature up to the glass softening temperature, after which it is found to decrease. In fact the absorption of
0.6
Absorbance (a.u)
Laser excitation
0.4
Cu-600 AP
Cu-550
0.2 Cu-500
Cu-650
500
600
700
800
Wavelength (nm)
Fig. 1. TEM micrographs of as-prepared Cu nanocomposite glass (labels A–H represent the Cu nanoparticles).
Fig. 2. Optical absorption spectra of annealed Cu nanocomposite glasses. Annealing was done at various temperatures for 5 h. Glass softening temperature is 600 °C. AP means as-prepared glass.
B. Karthikeyan et al. / Optics Communications 281 (2008) 2933–2937 150
2935 Cu 550
1.15 Cu 500
Cu-600
1.08 1.06
normalized transmittance
Emission Intensity (a.u)
1.10
100
Cu-650 50
Cu-500
1.02 1.00
575
0.98
1.1
1.0
1.0
0.9
0.7
0.7 600
625
650
-10000 -5000
Cu 650 0
5000 10000 -10000 -5000
0
5000 10000
where Ein is the input laser pulse energy and x(z) is the beam radius [9]. The intensity can then be obtained by dividing the fluence with the laser pulsewidth. At lower intensities the sample shows a saturable absorption behavior, which changes to an optical limiting type behavior at higher intensities. Fig. 5 shows the open aperture z-scan
b
Fig. 5. Open aperture z-scan plots of Cu nanocomposite glasses, annealed at the temperatures shown. Solid curves are numerical fits to the experimental data obtained using Eq. (6).
curves obtained for the annealed glasses. A clear absorption saturation behavior is found for the 500 °C annealed glass, but it changes to an optical limiting behavior for those annealed at 600 °C and 650 °C. At the intermediate temperature of 550 °C, an in-between behavior is observed in the form of humps flanking the central valley. The normalized transmission of the samples, as a function of input laser intensity, is shown in Fig. 6. To explain the optical nonlinearity results, first it may be noted that the optical nonlinearity will be enhanced when excited near the SPR, due to the local field enhancement. If metal particles of dielectric constant em are distributed uniformly and randomly in the dielectric host having a dielectric constant ed, the local field is given by El ¼
3ed Eo em þ 2ed
ð3Þ
where Eo is the applied optical field. Near the SPR em + 2ed is a minimum so that El is a maximum. The corresponding nonlinear polarization is given by [10]
1.10
1.05
1.08
Cu 550
1.15 Cu 500
1.05
1.06
1.00
1.00
0.95
0.95
0.90
0.90
0.85 0
5000 10000
z (microns)
1x1012
1x1013 2
input intensity (W/m )
normalized transmittance
1.10
0.85 -10000 -5000
0.5
z (microns)
Cu-650 is a little lower than that of the as-prepared sample. Fig. 3 shows the fluorescence emission spectra of the samples. The fluorescence amplitude is enhanced up to the annealing temperature of 600 °C, after which it is found to drop. Fig. 4a shows the open aperture z-scan curve obtained for the as-prepared glass. In Fig. 4b, the normalized transmittance is plotted as a function of the input intensity. This plot is calculated from the z-scan data considering the fact that for a Gaussian beam, each z position corresponds to an input fluence of pffiffiffiffiffiffiffiffi EðzÞ ¼ 4 ln 2Ein =p3=2 xðzÞ2 ð2Þ
a
0.6
0.6 Cu 600
675
Fig. 3. Emission spectra of the as-prepared and annealed glasses. Wavelength of excitation is 340 nm.
normalized transmittance
0.8
0.8
Wavelength (nm)
1.10
1.00
0.95
0.9
As Prepared 0 550
1.04
1.05
1.04
1.05
1.02
1.00
1.00
0.95
0.98
1.1
1.0
1.0
0.9
0.9
0.8
0.8
0.7
0.7
0.6
0.6 Cu 600
Cu 650
1x1012
Fig. 4. Open aperture z-scan of the as-prepared samples (a), and normalized transmittance plotted as a function of input laser pulse intensity (b). Data for (b) is obtained from (a) using Eq. (2). Solid curves are numerical fits to the experimental data obtained using Eq. (6).
1x1013
1x1012
1x1013
0.5
2
input intensity (W/m )
Fig. 6. Intensity dependent transmission of the copper nanocomposite glasses, plotted using data extracted from Fig. 5.
2936
B. Karthikeyan et al. / Optics Communications 281 (2008) 2933–2937
P NLS
3ed ¼ 3p em þ 2ed
4
vmð3Þ E3o
ð4Þ
which scales as the fourth power of the local field factor. Flytzanis and coworkers [11] have shown that there are three major electronic contributions to the Kerr nonlinearity in metal nanoclusters. Recalling that the optical properties of noble metal clusters are primarily determined by the electrons occupying their ‘‘d” and conducð3Þ tion bands, the first of these, vintra , is derived from the intraband conduction electrons. It is electric dipole in nature, originates totally due to the confinement of the free electrons, and is strongly size dependent. The second term ð3Þ vinter originates from interband transitions between the dbands and the conduction band, which is size and shape independent down to very small sizes. The third and most ð3Þ important contribution vhot arises from hot electrons, which are conduction electrons that are easily elevated to temperatures of several hundred degrees by photoexcitation, as their specific heats are very small. Calculations [11] and experiments [12] show that the hot electron and interband contributions are mainly imaginary in nature, and much stronger than the intraband contribution. Within the few picoseconds taken by hot electrons (after laser excitation) to thermalize with the lattice, the Fermi–Dirac electron distribution will be modified, since part of the one-electron levels below the Fermi level is emptied and part of those above is occupied. This leads to a modification of the dielectric constant em, that results in a transient re-distribution of the equilibrium plasmon band. As a result the absorption around the peak is reduced and that at the wings is increased. This reduction in absorption will be manifest as a transient absorption saturation which is generally referred to as ‘‘plasmon band bleach” [13,14] in the literature. For copper the SPR peak is at 560 nm, which is not far from the present excitation wavelength of 532 nm. Therefore in the present case, we can expect the Kerr nonlinearity to result in absorption saturation around the SPR, which will be reflected in the z-scan curves. In fact the glass annealed at 500 °C shows this saturation behavior exclusively, as can be seen from Figs. 5 and 6. Saturation can be seen in the as-prepared glass also, at the lower incident intensities (Fig. 4). However, when the annealing temperature is increased there is a clear change in the sign of the nonlinear transmission, as above the glass softening temperature the absorption saturation increasingly gives way to an optical limiting behavior. At 650 °C the sample behaves like a standard optical limiter. We explain this variation in nonlinear absorption with annealing temperature qualitatively as follows. The as-prepared sample contains nanoparticles of approximately 3–4 nm diameter, and the associated SPR is relatively weak. When this sample is annealed the particles grow to a larger size showing enhanced SPR and interband absorption. However when the annealing temperature exceeds the glass softening temperature the glass matrix becomes soft,
triggering a fragmentation of the particles generating small-sized particles with lower SPR amplitudes. The nonlinear behavior is closely related to the SPR amplitude: when SPR is strong there is a concomitant enhancement in the local field and the Kerr nonlinearity, resulting in a pronounced absorption saturation behavior. When the SPR is weak interband transitions gain relative prominence, which are followed by the excitation of the conduction electrons (free carrier absorption – FCA) at higher intensities. This two-step excitation process is a reverse saturable absorption (RSA) phenomenon leading to increased transient absorption. Therefore we see only the optical limiting type of behavior in the high-temperature annealed (small particle size) samples. In fact earlier we have reported similar exclusive optical limiting behavior in small Au nanoparticles in which SPR was absent [15]. The nonlinear behavior of the present samples in the whole intensity range of study can be modeled by defining an effective nonlinear absorption coefficient a(I), given by a0 aðIÞ ¼ þ bI ð5Þ 1 þ ðIIs Þ where a0 is the unsaturated linear absorption coefficient at the wavelength of excitation, and Is is the saturation intensity (intensity at which the linear absorption drops to half its original value). bI = rN is the FCA coefficient, where r is the FCA cross-section and N(I) is the generated intensity-dependent free carrier density. For calculating the output laser intensity for a given input intensity, first we numerically evaluate the output intensity from the sample for each input intensity by solving the propagation equation dI I ¼ a0 = 1 þ þ bI I ð6Þ dz0 Is using the fourth order Runge–Kutta method. Input intensities for the gaussian beam for each sample position in the z-scan are calculated from the input energy, laser pulsewidth and irradiation area. Here z0 indicates the propagation distance within the sample. The normalized transmittance is then calculated by dividing the output intensity with the input intensity and normalizing it with the linear transmittance. As seen from Figs. 4–6, there is reasonable agreement between the experimental data and numerical simulation, despite the scattering in data points for Cu-500 and Cu-550. The numerically estimated values of b and Is are presented in Table 1.
Table 1 Numerically estimated values of b and Is for the Cu nanocomposite glasses S. no
Sample
b (1012 [m/W])
Is (1013 [W/m2])
1 2 3 4 5
Cu-500 Cu-550 Cu-600 Cu-650 As-prepared
5 ± 1.5 22 ± 1.5 84 ± 16 100 ± 20 70 ± 10
5 ± 0.5 1.3 ± 0.1 0.4 ± 0.01 0.3 ± 0.05 0.6 ± 0.02
B. Karthikeyan et al. / Optics Communications 281 (2008) 2933–2937
It is significant that in addition to changes in the nonlinearity, changes in the fluorescence spectra also occur within the temperature range of 500–600 °C. The fluorescence spectra shown in Fig. 3 reveal an emission maximum situated around 605 nm. In the past Mooradian [16] has observed fluorescence from bulk copper, peaked around 600 nm, due to radiative recombination of s–p conduction band electrons below the Fermi level with the holes in the d bands. In the present case also, emission occurs due to interband transitions. According to Darugar et al. [17] the emission intensity from bigger copper nanoparticles is relatively higher than that from smaller copper nanoparticles. In our case emission intensity increases up to 600 °C annealing, but it decreases for 650 °C annealed glasses. Therefore it is an additional indication that the particle size is decreasing when the sample is annealed near and above the glass softening temperature. 4. Conclusion In summary, we have prepared Cu nanocomposite glasses which are annealed at different temperatures, and studied their optical and nonlinear optical properties for annealing below and above the glass softening temperature. The absorption spectra and fluorescence spectra indicate that maximum particle size is attained near the glass softening temperature. At higher annealing temperatures the particles seem to fragment, forming smaller nanoparticles. A saturable absorption behavior is found for glasses annealed below the glass softening temperature, while an optical limiting behavior is found for glasses annealed near and above the glass softening temperature. This change of sign in the nonlinearity is explained principally on the basis of the dependence of surface plasmon resonance amplitude on the annealing temperature. The fact that the sign of the nonlinearity can be controlled by the annealing temperature is an
2937
important aspect when these materials are considered for use in photonic applications. Acknowledgements The authors wish to thank K.N. Vasudha for recording the optical absorption spectra of the samples. Muhammad Nadeer thanks Prof. V.P.N. Nampoori for his encouragement during the course of this work. References [1] G. Mitrikas, C.C. Trapalis, G. Kordas, J. Non-Cryst. Solids 286 (2001) 41. [2] H. Itoigawa, T. Kamiyama, Y. Nakamura, J. Non-Cryst. Solids 220 (1997) 210. [3] Y. Takeda, V.T. Gritsyna, N. Umeda, C.G. Lee, N. Kishimoto, Nucl. Instrum. Meth. B 148 (1999) 1029. [4] Y. Takeda, O.A. Plaksin, J. Lu, N. Kishimoto, Vacuum 80 (2006) 776. [5] S. Bera, P. Gangopadhyay, K.G.M. Nair, B.K. Panigrahi, S.V. Narasimhan, J. Electron Spectrosc. Relat. Phenom. 152 (2006) 91. [6] P.P. Kiran, B.N.S. Bhaktha, D.N. Rao, G. De, J. Appl. Phys. 96 (2004) 6717. [7] A.I. Ryasnyanskiy, B. Palpant, S. Debrus, U. Pal, A. Stepanov, J. Lumin. 127 (2007) 181. [8] U. Kreibig, M. Vollmer, Optical Properties of Metal Clusters, Springer-Verlag, Berlin, 1995. [9] R.L. Sutherland, Handbook of Nonlinear Optics, Marcel Dekker, New York, 1996. [10] D. Ricard, P. Roussignol, C. Flytzanis, Opt. Lett. 10 (1985) 511. [11] F. Hache, D. Ricard, C. Flytzanis, U. Kreibig, Appl. Phys. A 47 (1998) 347. [12] G.L. Eesley, Phys. Rev. B 33 (1986) 2144. [13] S.L. Logunov, T.S. Ahmadi, M.A. El-Sayed, J.T. Khoury, R.L. Whetten, J. Phys. Chem. B 101 (1997) 3713. [14] P.V. Kamat, M. Flumiani, G.V. Hartland, J. Phys. Chem. B 102 (1998) 3123. [15] J. Thomas, M. Anija, J. Cyriac, T. Pradeep, R. Philip, Chem. Phys. Lett. 403 (2005) 308. [16] A. Mooradian, Phys. Rev. Lett. 22 (1969) 185. [17] Q. Darugar, W. Qian, M.A. El Sayed, M. Pileni, J. Phys. Chem. B 110 (2006) 143.
Optics Communications 281 (2008) 2933–2937 www.elsevier.com/locate/optcom
Optical and nonlinear optical properties of copper nanocomposite glasses annealed near the glass softening temperature B. Karthikeyan a,*, M. Anija a, C.S. Suchand Sandeep a, T.M. Muhammad Nadeer b,1, Reji Philip a a
Light and Matter Physics Group, Raman Research Institute, C.V. Raman Avenue, Sadashivanagar, Bangalore 560 080, India b International School of Photonics, Cochin University of Science and Technology, Cochin 682 022, India Received 17 August 2007; received in revised form 7 December 2007; accepted 8 January 2008
Abstract Copper nanocomposite glasses have been prepared by the ion-exchange method, and annealed at different temperatures up to and above the glass softening temperature. The absorption spectra, fluorescence spectra, and nonlinear optical transmission of the samples at 532 nm for nanosecond laser pulses, have been investigated. The optical and nonlinear optical properties of the glasses are found to be distinctly different below and above the glass softening temperature. For instance, thermal annealing up to the glass softening temperature makes the samples behave like saturable absorbers, while annealing at higher temperatures makes them behave like optical limiters. Such flexibility in controlling the optical nonlinearity in these materials makes them potential candidates for photonic applications. Ó 2008 Elsevier B.V. All rights reserved. PACS: 61.46.+w; 78.67.Bf; 42.50.p Keywords: Metal nanocomposite; Nonlinear absorption; Optical limiting; Saturable absorption; Surface plasmon resonance
1. Introduction Metal nanocluster composite glasses (MNCGs) are potential candidates for photonic nanodevice fabrication. For instance, their large and fast optical nonlinearity can find applications in optical switching devices. Nanocomposite glasses can be prepared through the sol–gel [1], melt-quench [2], ion implantation [3,4] and ion-exchange [5] methods. Among these, ion-exchange is one of the oldest and most successful techniques. Noble metal clusters (copper, silver and gold) have gained a lot of attention *
Corresponding author. Fax: +91 (0)80 23610492. E-mail addresses: [email protected] (B. Karthikeyan), [email protected] (R. Philip). 1 Present address: Amal Jyothi College of Engineering, Kanjirappally 686 518, India. 0030-4018/$ - see front matter Ó 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.optcom.2008.01.032
because of their optical nonlinearities, which are substantial due to the surface plasmon resonances (SPR) lying in the visible region. The dispersive third order nonlinear susceptibility vð3Þ m of metal nanoclusters will be very strong when excited near the SPR, due to the local field enhancement effect. The use of MNCGs as ultrafast optical switches is known before [6], and there are previous reports on their near and off-resonant nonlinear optical properties at various cluster sizes [3,7]. Thermal annealing is a common route used for changing the particle size. In the present work, we have prepared copper nanocomposite glasses by the ion-exchange method, which are annealed at different temperatures up to and above the glass softening temperature (600 °C). The absorption and fluorescence spectra of the samples have been measured. The nonlinear optical transmission of nanosecond laser pulses through the samples also has been measured at the
2934
B. Karthikeyan et al. / Optics Communications 281 (2008) 2933–2937
wavelength of 532 nm. The nonlinearity is found to originate mostly from the Kerr nonlinearity and free carrier absorption occurring in the samples. In general, the samples are found to behave as saturable absorbers below the glass softening temperature, and optical limiters above the glass softening temperature. Thus the glass softening temperature turns out to be a critical control parameter for optical nonlinearity in these materials. This observation may be generally true for glass hosts embedded with nonlinear media, in which case it is important information for fabricating photonic devices with controlled and predesigned optical nonlinearities. 2. Experimental In our experiment, metal ions are incorporated into sodalime glass through the thermal ion-exchange method. Commercial soda lime glasses are initially cleaned using an ultrasonic cleaner. The glass slides are then immersed in a molten salt bath containing CuSO4 and Na2SO4 in the 1:3 wt. ratio. The ion-exchange process is allowed to take place at 500–560 °C for a period of 1 min. The copper ions in the molten salt bath penetrate and occupy the sites left by the Na+ ions which are the glass modifiers in the host matrix. The prepared glass slides are air annealed at 500, 550, 600 and 650 °C (Cu-500, Cu-550, Cu-600 and Cu-650, respectively) for 5 h and then furnace cooled. Transmission electron micrographs (TEM) of the as-prepared glass, obtained using a Teenai F30 machine, are shown in Fig. 1. The pictures reveal that the as-prepared Cu nanoclusters are about 3–4 nm in diameter. Optical absorption spectra were recorded using a dual beam Perkin Elmer spectrophotometer. Steady state fluorescence spectra were recorded using a Perkin Elmer spectrofluorometer, with a xenon arc lamp source and a photomultiplier detection system. Samples were excited at 340 nm and the fluorescence spectra were recorded from 550 to 660 nm.
To measure the optical nonlinearity, open aperture zscan measurements were done at 532 nm using 7 ns laser pulses from a frequency-doubled Nd:YAG laser (Quanta Ray-Spectra Physics). In the z-scan the laser beam is focused using a lens, and the sample is translated along the beam axis (z-axis) through the focal region over a distance several times that of the diffraction length. At each position z the sample sees a different laser intensity, and the position dependent (ie, intensity-dependent) transmission is measured using an energy meter placed after the sample. Laser pulses are fired at a repetition rate of 1 Hz, and the data acquisition is automated. The low repetition rate is chosen for avoiding heating-up of the samples during measurement. The pulse energy reaching the sample is approximately 60 lJ. 3. Results and discussion The optical absorption spectra of the Cu nanocomposite glasses are shown in Fig. 2. All spectra show the surface plasmon resonance (SPR) around 560 nm. Under the dipole approximation, Mie theory gives the extinction cross-section of spherical nanoparticles as [8] rext ðxÞ ¼ 9p
emðRÞ ðxÞ x 3=2 e c d ½emðIÞ ðxÞ þ 2ed 2 þ e2mðRÞ ðxÞ
ð1Þ
where p is the volume fraction, and x is the applied optical frequency. The SPR maximum occurs at em ðlÞðxÞþ 2ed ¼ 0 where em ðxÞ ¼ emðRÞ ðxÞ þ iemðlÞ ðxÞ is the frequency dependent, complex dielectric constant of the metallic particle, and ed is the dielectric constant of the host matrix which is assumed to be real and nondispersive. The SPR is superposed over the interband absorption that gains strength in the short-wavelength region in the measured spectral range. The net absorption increases with annealing temperature up to the glass softening temperature, after which it is found to decrease. In fact the absorption of
0.6
Absorbance (a.u)
Laser excitation
0.4
Cu-600 AP
Cu-550
0.2 Cu-500
Cu-650
500
600
700
800
Wavelength (nm)
Fig. 1. TEM micrographs of as-prepared Cu nanocomposite glass (labels A–H represent the Cu nanoparticles).
Fig. 2. Optical absorption spectra of annealed Cu nanocomposite glasses. Annealing was done at various temperatures for 5 h. Glass softening temperature is 600 °C. AP means as-prepared glass.
B. Karthikeyan et al. / Optics Communications 281 (2008) 2933–2937 150
2935 Cu 550
1.15 Cu 500
Cu-600
1.08 1.06
normalized transmittance
Emission Intensity (a.u)
1.10
100
Cu-650 50
Cu-500
1.02 1.00
575
0.98
1.1
1.0
1.0
0.9
0.7
0.7 600
625
650
-10000 -5000
Cu 650 0
5000 10000 -10000 -5000
0
5000 10000
where Ein is the input laser pulse energy and x(z) is the beam radius [9]. The intensity can then be obtained by dividing the fluence with the laser pulsewidth. At lower intensities the sample shows a saturable absorption behavior, which changes to an optical limiting type behavior at higher intensities. Fig. 5 shows the open aperture z-scan
b
Fig. 5. Open aperture z-scan plots of Cu nanocomposite glasses, annealed at the temperatures shown. Solid curves are numerical fits to the experimental data obtained using Eq. (6).
curves obtained for the annealed glasses. A clear absorption saturation behavior is found for the 500 °C annealed glass, but it changes to an optical limiting behavior for those annealed at 600 °C and 650 °C. At the intermediate temperature of 550 °C, an in-between behavior is observed in the form of humps flanking the central valley. The normalized transmission of the samples, as a function of input laser intensity, is shown in Fig. 6. To explain the optical nonlinearity results, first it may be noted that the optical nonlinearity will be enhanced when excited near the SPR, due to the local field enhancement. If metal particles of dielectric constant em are distributed uniformly and randomly in the dielectric host having a dielectric constant ed, the local field is given by El ¼
3ed Eo em þ 2ed
ð3Þ
where Eo is the applied optical field. Near the SPR em + 2ed is a minimum so that El is a maximum. The corresponding nonlinear polarization is given by [10]
1.10
1.05
1.08
Cu 550
1.15 Cu 500
1.05
1.06
1.00
1.00
0.95
0.95
0.90
0.90
0.85 0
5000 10000
z (microns)
1x1012
1x1013 2
input intensity (W/m )
normalized transmittance
1.10
0.85 -10000 -5000
0.5
z (microns)
Cu-650 is a little lower than that of the as-prepared sample. Fig. 3 shows the fluorescence emission spectra of the samples. The fluorescence amplitude is enhanced up to the annealing temperature of 600 °C, after which it is found to drop. Fig. 4a shows the open aperture z-scan curve obtained for the as-prepared glass. In Fig. 4b, the normalized transmittance is plotted as a function of the input intensity. This plot is calculated from the z-scan data considering the fact that for a Gaussian beam, each z position corresponds to an input fluence of pffiffiffiffiffiffiffiffi EðzÞ ¼ 4 ln 2Ein =p3=2 xðzÞ2 ð2Þ
a
0.6
0.6 Cu 600
675
Fig. 3. Emission spectra of the as-prepared and annealed glasses. Wavelength of excitation is 340 nm.
normalized transmittance
0.8
0.8
Wavelength (nm)
1.10
1.00
0.95
0.9
As Prepared 0 550
1.04
1.05
1.04
1.05
1.02
1.00
1.00
0.95
0.98
1.1
1.0
1.0
0.9
0.9
0.8
0.8
0.7
0.7
0.6
0.6 Cu 600
Cu 650
1x1012
Fig. 4. Open aperture z-scan of the as-prepared samples (a), and normalized transmittance plotted as a function of input laser pulse intensity (b). Data for (b) is obtained from (a) using Eq. (2). Solid curves are numerical fits to the experimental data obtained using Eq. (6).
1x1013
1x1012
1x1013
0.5
2
input intensity (W/m )
Fig. 6. Intensity dependent transmission of the copper nanocomposite glasses, plotted using data extracted from Fig. 5.
2936
B. Karthikeyan et al. / Optics Communications 281 (2008) 2933–2937
P NLS
3ed ¼ 3p em þ 2ed
4
vmð3Þ E3o
ð4Þ
which scales as the fourth power of the local field factor. Flytzanis and coworkers [11] have shown that there are three major electronic contributions to the Kerr nonlinearity in metal nanoclusters. Recalling that the optical properties of noble metal clusters are primarily determined by the electrons occupying their ‘‘d” and conducð3Þ tion bands, the first of these, vintra , is derived from the intraband conduction electrons. It is electric dipole in nature, originates totally due to the confinement of the free electrons, and is strongly size dependent. The second term ð3Þ vinter originates from interband transitions between the dbands and the conduction band, which is size and shape independent down to very small sizes. The third and most ð3Þ important contribution vhot arises from hot electrons, which are conduction electrons that are easily elevated to temperatures of several hundred degrees by photoexcitation, as their specific heats are very small. Calculations [11] and experiments [12] show that the hot electron and interband contributions are mainly imaginary in nature, and much stronger than the intraband contribution. Within the few picoseconds taken by hot electrons (after laser excitation) to thermalize with the lattice, the Fermi–Dirac electron distribution will be modified, since part of the one-electron levels below the Fermi level is emptied and part of those above is occupied. This leads to a modification of the dielectric constant em, that results in a transient re-distribution of the equilibrium plasmon band. As a result the absorption around the peak is reduced and that at the wings is increased. This reduction in absorption will be manifest as a transient absorption saturation which is generally referred to as ‘‘plasmon band bleach” [13,14] in the literature. For copper the SPR peak is at 560 nm, which is not far from the present excitation wavelength of 532 nm. Therefore in the present case, we can expect the Kerr nonlinearity to result in absorption saturation around the SPR, which will be reflected in the z-scan curves. In fact the glass annealed at 500 °C shows this saturation behavior exclusively, as can be seen from Figs. 5 and 6. Saturation can be seen in the as-prepared glass also, at the lower incident intensities (Fig. 4). However, when the annealing temperature is increased there is a clear change in the sign of the nonlinear transmission, as above the glass softening temperature the absorption saturation increasingly gives way to an optical limiting behavior. At 650 °C the sample behaves like a standard optical limiter. We explain this variation in nonlinear absorption with annealing temperature qualitatively as follows. The as-prepared sample contains nanoparticles of approximately 3–4 nm diameter, and the associated SPR is relatively weak. When this sample is annealed the particles grow to a larger size showing enhanced SPR and interband absorption. However when the annealing temperature exceeds the glass softening temperature the glass matrix becomes soft,
triggering a fragmentation of the particles generating small-sized particles with lower SPR amplitudes. The nonlinear behavior is closely related to the SPR amplitude: when SPR is strong there is a concomitant enhancement in the local field and the Kerr nonlinearity, resulting in a pronounced absorption saturation behavior. When the SPR is weak interband transitions gain relative prominence, which are followed by the excitation of the conduction electrons (free carrier absorption – FCA) at higher intensities. This two-step excitation process is a reverse saturable absorption (RSA) phenomenon leading to increased transient absorption. Therefore we see only the optical limiting type of behavior in the high-temperature annealed (small particle size) samples. In fact earlier we have reported similar exclusive optical limiting behavior in small Au nanoparticles in which SPR was absent [15]. The nonlinear behavior of the present samples in the whole intensity range of study can be modeled by defining an effective nonlinear absorption coefficient a(I), given by a0 aðIÞ ¼ þ bI ð5Þ 1 þ ðIIs Þ where a0 is the unsaturated linear absorption coefficient at the wavelength of excitation, and Is is the saturation intensity (intensity at which the linear absorption drops to half its original value). bI = rN is the FCA coefficient, where r is the FCA cross-section and N(I) is the generated intensity-dependent free carrier density. For calculating the output laser intensity for a given input intensity, first we numerically evaluate the output intensity from the sample for each input intensity by solving the propagation equation dI I ¼ a0 = 1 þ þ bI I ð6Þ dz0 Is using the fourth order Runge–Kutta method. Input intensities for the gaussian beam for each sample position in the z-scan are calculated from the input energy, laser pulsewidth and irradiation area. Here z0 indicates the propagation distance within the sample. The normalized transmittance is then calculated by dividing the output intensity with the input intensity and normalizing it with the linear transmittance. As seen from Figs. 4–6, there is reasonable agreement between the experimental data and numerical simulation, despite the scattering in data points for Cu-500 and Cu-550. The numerically estimated values of b and Is are presented in Table 1.
Table 1 Numerically estimated values of b and Is for the Cu nanocomposite glasses S. no
Sample
b (1012 [m/W])
Is (1013 [W/m2])
1 2 3 4 5
Cu-500 Cu-550 Cu-600 Cu-650 As-prepared
5 ± 1.5 22 ± 1.5 84 ± 16 100 ± 20 70 ± 10
5 ± 0.5 1.3 ± 0.1 0.4 ± 0.01 0.3 ± 0.05 0.6 ± 0.02
B. Karthikeyan et al. / Optics Communications 281 (2008) 2933–2937
It is significant that in addition to changes in the nonlinearity, changes in the fluorescence spectra also occur within the temperature range of 500–600 °C. The fluorescence spectra shown in Fig. 3 reveal an emission maximum situated around 605 nm. In the past Mooradian [16] has observed fluorescence from bulk copper, peaked around 600 nm, due to radiative recombination of s–p conduction band electrons below the Fermi level with the holes in the d bands. In the present case also, emission occurs due to interband transitions. According to Darugar et al. [17] the emission intensity from bigger copper nanoparticles is relatively higher than that from smaller copper nanoparticles. In our case emission intensity increases up to 600 °C annealing, but it decreases for 650 °C annealed glasses. Therefore it is an additional indication that the particle size is decreasing when the sample is annealed near and above the glass softening temperature. 4. Conclusion In summary, we have prepared Cu nanocomposite glasses which are annealed at different temperatures, and studied their optical and nonlinear optical properties for annealing below and above the glass softening temperature. The absorption spectra and fluorescence spectra indicate that maximum particle size is attained near the glass softening temperature. At higher annealing temperatures the particles seem to fragment, forming smaller nanoparticles. A saturable absorption behavior is found for glasses annealed below the glass softening temperature, while an optical limiting behavior is found for glasses annealed near and above the glass softening temperature. This change of sign in the nonlinearity is explained principally on the basis of the dependence of surface plasmon resonance amplitude on the annealing temperature. The fact that the sign of the nonlinearity can be controlled by the annealing temperature is an
2937
important aspect when these materials are considered for use in photonic applications. Acknowledgements The authors wish to thank K.N. Vasudha for recording the optical absorption spectra of the samples. Muhammad Nadeer thanks Prof. V.P.N. Nampoori for his encouragement during the course of this work. References [1] G. Mitrikas, C.C. Trapalis, G. Kordas, J. Non-Cryst. Solids 286 (2001) 41. [2] H. Itoigawa, T. Kamiyama, Y. Nakamura, J. Non-Cryst. Solids 220 (1997) 210. [3] Y. Takeda, V.T. Gritsyna, N. Umeda, C.G. Lee, N. Kishimoto, Nucl. Instrum. Meth. B 148 (1999) 1029. [4] Y. Takeda, O.A. Plaksin, J. Lu, N. Kishimoto, Vacuum 80 (2006) 776. [5] S. Bera, P. Gangopadhyay, K.G.M. Nair, B.K. Panigrahi, S.V. Narasimhan, J. Electron Spectrosc. Relat. Phenom. 152 (2006) 91. [6] P.P. Kiran, B.N.S. Bhaktha, D.N. Rao, G. De, J. Appl. Phys. 96 (2004) 6717. [7] A.I. Ryasnyanskiy, B. Palpant, S. Debrus, U. Pal, A. Stepanov, J. Lumin. 127 (2007) 181. [8] U. Kreibig, M. Vollmer, Optical Properties of Metal Clusters, Springer-Verlag, Berlin, 1995. [9] R.L. Sutherland, Handbook of Nonlinear Optics, Marcel Dekker, New York, 1996. [10] D. Ricard, P. Roussignol, C. Flytzanis, Opt. Lett. 10 (1985) 511. [11] F. Hache, D. Ricard, C. Flytzanis, U. Kreibig, Appl. Phys. A 47 (1998) 347. [12] G.L. Eesley, Phys. Rev. B 33 (1986) 2144. [13] S.L. Logunov, T.S. Ahmadi, M.A. El-Sayed, J.T. Khoury, R.L. Whetten, J. Phys. Chem. B 101 (1997) 3713. [14] P.V. Kamat, M. Flumiani, G.V. Hartland, J. Phys. Chem. B 102 (1998) 3123. [15] J. Thomas, M. Anija, J. Cyriac, T. Pradeep, R. Philip, Chem. Phys. Lett. 403 (2005) 308. [16] A. Mooradian, Phys. Rev. Lett. 22 (1969) 185. [17] Q. Darugar, W. Qian, M.A. El Sayed, M. Pileni, J. Phys. Chem. B 110 (2006) 143.