Applied Surface Science 254 (2007) 1017–1021 www.elsevier.com/locate/apsusc
Au–Cu nanoparticles in silica glass as composite material for photonic applications E. Cattaruzza a,*, G. Battaglin a, F. Gonella a, R. Polloni a, B.F. Scremin b, G. Mattei c, P. Mazzoldi c, C. Sada c a
Physical Chemistry Department, Universita` Ca’ Foscari Venezia, Dorsoduro 2137, I-30123 Venezia, Italy b INFM-CNR, corso Perrone 24, I-16152 Genova, Italy c Physics Department, Universita` degli Studi di Padova, via Marzolo 8, I-35131 Padova, Italy Received 18 May 2007; received in revised form 28 June 2007; accepted 23 July 2007 Available online 27 July 2007
Abstract In the framework of metal nanocluster composite glasses for photonic application, (gold + copper)-containing silica films were synthesized by radiofrequency co-sputtering deposition technique by varying the Au/Cu ratio. To obtain the formation of metallic (alloy) nanoclusters, the deposited samples were annealed in reducing atmosphere at 900 8C. The linear and nonlinear optical properties of the composite glasses were investigated. In particular, the nonlinear ultrafast optical response was measured by means of the Z-scan technique at a wavelength of 527 nm with single 6 ps pulse configuration. Significant refractive and absorptive effects were observed in all the annealed samples. The two main figures of merit related to the performance of materials for optical switching device application were evaluated for all samples, showing interesting values for the Cu-rich composites. # 2007 Elsevier B.V. All rights reserved. Keywords: Composite glasses; Z-scan; Nanoparticles; Optical properties; Sputtering deposition
1. Introduction Silica films containing metal nanoparticles belong to the usually called metal nanocluster composite glasses (MNCGs), the main physical property of which is an enhanced optical Kerr susceptibility that builds up an intensity-dependent refractive index, usually defined as n(I) = n0 + n2I, where n0 and n2 are the linear and nonlinear refractive indices, respectively, and I is the intensity of the light. This feature allows the use of such composites for application in all-optical switching devices, exploiting the dependence on the light intensity of the refractive index [1–4]. A significant intensity-dependent refractive index allows, for example, the switching of a light signal between the two channels of a directional coupler [1] in which one of the two is made of MNCG. The typical switching times are about few picoseconds [1,2], thus making these composites attractive for optical application in the information processing up to the THz range of frequency [3,4]. The sign of n2 is related to * Corresponding author. Tel.: +39 041 2346716; fax: +39 041 2346713. E-mail address:
[email protected] (E. Cattaruzza). 0169-4332/$ – see front matter # 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.apsusc.2007.07.158
phenomena of self-focusing (n2 > 0) or self-defocusing (n2 < 0) of the incident laser beam. The MNCGs exhibit also an intensity-dependent absorption coefficient, defined as a(I) = a0 + bI, where a0 and b are the linear and nonlinear absorption coefficients, respectively. b accounts for the nonlinear processes as induced absorption (b > 0) or saturated absorption (b < 0). Both n2 and b are related to the third-order nonlinear susceptibility x(3) of the MNCG (in turn related to the third-order nonlinear susceptibility of the metal nanoparticles). A useful and quantitative shorthand for assessing the effectiveness of nonlinear optical materials is the figure of merit (FOM) [1]. For the use of a material in the all-optical switching device technology, a nonlinear phase shift of 2p should be possible over one attenuation distance for reasonable device throughput [5]: therefore, depending on the main absorption mechanism, namely linear or nonlinear, a material for switching devices has to satisfy two FOMs for having a 2p phase shift [2]: W¼
jn2 jI max > 1; a0 l
T¼
bl < 1; jn2 j
1018
E. Cattaruzza et al. / Applied Surface Science 254 (2007) 1017–1021
where Imax is the threshold damage intensity for the material, and l is the wavelength of the light signal. Among the most interesting metals to produce embedded nanoparticles in a dielectric matrix are the noble metals. In this paper, (gold + copper)-containing silica films were synthesized by radiofrequency co-sputtering deposition, which allows to easily obtain homogeneous glassy films doped with nanoparticles [6,7], of thickness as large as few microns, suitable for the fabrication of optical waveguides. The Au/Cu ratio was varied to investigate different MNCG compositions, to explore the possibility of controlling the optical properties of the final composite system. After a suitable thermal treatment for inducing the formation of pure metallic, alloy and/or core/shell nanoparticles [8–13], the linear and nonlinear optical properties of the obtained composite systems were investigated by optical absorption spectroscopy and by Z-scan analysis. In particular, by the Z-scan technique both the nonlinear coefficients n2 and b can be directly evaluated, thus determining the two main figures of merit W and T assessing the effectiveness of the MNCG material for all-optical switching technology application.
800 nm range were recorded using a UV–VIS–NIR Jasco V-570 dual-beam spectrophotometer. Transmission electron microscopy (TEM) characterization was performed on cross-sectional samples at CNR-IMM Institute (Bologna, Italy), with a fieldemission gun (FEG) microscope (FEI Tecnai F20 Super Twin) operated at 200 kV. Z-scan measurements were performed by using a ring-cavity, mode-locked, frequency-doubled Nd:phosphate glass laser source (527 nm of wavelength). The laser operates in the TEM00 mode and it supplies trains of about 100 pulses of nearly equal intensity, 5 ns spaced in time, at repetition rate of 1 Hz. The single pulse duration is about 6 ps. To avoid the contribution to the nonlinear refractive index (n2) due to cumulative heating effects induced by consecutive high intensity pulses very close in time [14], we used a pulse-slicer to extract a single pulse for any train. For all the Z-scan measurements, the laser peak intensity I at the focal point was lower than 1 109 W/ cm2. Depending on the sample composition, the damage threshold varied from 2 109 to 5 109 W/cm2. A standard CS2 sample was used as a reference (n2 = 3 1014 cm2/W, at 532 nm, 27 ps of pulse duration) [15].
2. Experimental
3. Results and discussion
Composite dielectric films made of (gold + copper)-doped silica were obtained by co-deposition of silica, copper and gold on fused silica slides, 1 mm thick, with the radiofrequency magnetron sputtering deposition technique. Three 13.56 MHz radiofrequency sources in a neutral Ar atmosphere were used at a pressure of 35 102 Pa. After a four-step cleaning in ultrasonic bath (deionized H2O, trichloroethylene, acetone, isopropyl alcohol), the silica substrates were rf-biased at 20 W for 30 min before deposition to remove surface contaminations (around 15 nm). During the depositions (made at room temperature) the sample holder was rotated at 15 rpm to ensure deposition uniformity. The rf-power to the 2 in. diameter targets was fixed at 250 W for silica, and changed in the 5–13 W and 5– 17 W range for gold and copper, respectively, to vary the Au/Cu ratio. The nominal metal composition was selected to have Au/ Cu (and Cu/Au) close to 0, 1, 2, 3 and 5. For all the depositions, the rf-power to the metal targets was chosen to have a metal volume fraction in the film lower than 10%. A first single deposition of silica was made for 5 min, followed by a 50 min codeposition of silica, gold and copper. As a final step in the preparation, a single silica deposition was performed for 5 min. The film thickness was about 1 mm for all samples, as estimated by a profilometer. To favor the formation of nanoparticles, a thermal annealing in reducing atmosphere was subsequently performed on the deposited samples, in a commercial oven at 900 8C for 2 h in H2(10%)–Ar gas mixture. Rutherford backscattering spectrometry (RBS) measurements were performed at Laboratori Nazionali INFN-Legnaro (Padova), Italy, by using a 4He+ beam at the energy of 2.2 MeV, with the detector placed at 1608. Grazing incidence X-ray diffraction (GIXRD) patterns were recorded with an incidence angle x = 0.88 and a parallel beam geometry with a Panalytical X’Pert Pro MRD diffractometer equipped with a parabolic mirror in front of a Cu Ka X-ray source. Optical absorption (OA) spectra in the 200–
RBS measurements showed that the metal fraction in the composite films is in the range 4–9%. No differences were observed in the RBS signals of gold and copper between the asdeposited and the corresponding annealed samples. The effect of the thermal treatment in reducing atmosphere was evaluated by optical absorption measurements: the optical absorption spectra of the annealed samples are shown in Fig. 1 (only spectra corresponding to five different compositions are reported, vertically shifted for clarity). An absorption band due to the surface plasmon resonance (SPR) of the metallic nanoparticles formed during the annealing is evident in all the spectra. By increasing the Au/Cu ratio on the silica films, the SPR band position shifts in wavelength from the typical value of spherical copper nanoparticles in silica (570 nm) to that of
Fig. 1. Optical absorption spectra of the annealed (Au + Cu)–silica samples.
E. Cattaruzza et al. / Applied Surface Science 254 (2007) 1017–1021
spherical gold nanoparticles in silica (525 nm), suggesting the formation of alloy nanoparticles in the mixed-metal systems. The spectra of the as-deposited samples (not reported) do not exhibit SPR absorption bands, except for that of the pure Au sample, exhibiting a faint SPR band. The reducing thermal treatment is effective in inducing the precipitation of the dispersed and/or oxidized metal atoms. GIXRD analyses confirmed the presence of alloy nanoparticles after the reducing thermal treatment: we found for all the explored compositions the formation of a fcc Au–Cu alloy solid solution (the dominant phase) and in some cases also the ordered or oxidized phases [16]. We also found that the fcc alloy solid solution has a lattice parameter that follows (as a function of the composition) the modified Vegard’s law typical of the bulk Au–Cu alloy. Actually, the phase diagram of the bulk Au–Cu phase [17] shows that for all the compositions a stable and disordered fcc solid solution can be obtained at room temperature, along with three ordered phases for the Au:Cu atomic ratios of 3:1, 1:1, and 1:3. In Fig. 2, the cross-sectional TEM analysis on the annealed sample Au47Cu53 is reported. The low magnification brightfield image is shown in Fig. 2a with the size-histogram in the inset. The distribution is characterized by an average nanoclusters size D = 4.3 nm and a standard deviation s = 3.0 nm (nanoclusters with size up to 15–20 nm are present) and it follows a log-normal function. A typical high-resolution image of an alloy nanocluster is shown in Fig. 2b, where the (1 1 1) and (2 0 0) lattice planes of the fcc structure are visible. Indeed, GIXRD measurements showed that this is the major phase present in the sample, with a smaller contribution of the ordered tetragonal Au50Cu50 one. Considering the long tail of the log-normal distribution, we calculated the volume averaged size hDvoli = hD3i1/3, obtaining a value of about 6.3 nm. This is to be compared to the size extracted from the Scherrer analysis of the GIXRD pattern, which resulted about 7 nm for this sample, therefore in good agreement. From this result and from the peak broadening of the GIXRD pattern in the other samples at different compositions, we conclude that by changing the alloy composition from pure Au to pure Cu the volume averaged size remains almost constant in the region 5–10 nm, allowing a more direct comparison of the nonlinear properties of all the samples. Whereas as-deposited samples did not exhibit nonlinear optical behavior, significantly high optical nonlinearities were measured for the annealed samples, except for the Au/Cu = 1.1 sample, for which we did not observe optical nonlinearity (neither refraction nor absorption). In Fig. 3, the corrected farfield (FF) curves for two samples with symmetrical Au/Cu ratio are shown. The curves were obtained by dividing the recorded FF spectra by the near-field (NF) ones. For certain conditions of irradiance and b/n2 ratio values [15], as in the present case, the relative uncertainty introduced by this approximation is of the order of 10%. All the investigated MNCG films exhibited very close n2 values (Fig. 4), without a significant trend as a function of the MNCG composition. An interesting exception is the pure Au system, for which we found a negative nonlinear refractive value n2 = (6.0 2.0) 1011 cm2/W. The nonlinear absorption coefficient b (Fig. 4) exhibits an interesting behavior
1019
Fig. 2. Cross-sectional TEM analysis on the annealed sample Au47Cu53: low magnification bright-field image with the size-histogram in the inset (a); highresolution image of an alloy nanocluster (b).
as a function of the MNCG composition, starting from positive values for Cu-rich MNCG and changing its sign by increasing the gold content in the alloy. For nanoparticles filling factors p below 10% (as in the present case), the complex third-order nonlinear susceptibility of the composite can be written as ð3Þ ð3Þ ð3Þ xMNCG ¼ p fe2 j f e j2 xNP [18], where xNP is the nonlinear contribution due to the nanoparticles and f e is the local-field factor (or enhancement factor). This complex factor can add a significant phase shift to the MNCG third-order nonlinear ð3Þ ð3Þ susceptibility xMNCG with respect to xNP [19], which – for wavelengths near the SPR – is mainly imaginary with a positive
1020
E. Cattaruzza et al. / Applied Surface Science 254 (2007) 1017–1021
Fig. 4. Values of the nonlinear absorption coefficient b (a) and of the nonlinear refractive index n2 (b) measured in the annealed samples as a function of the nanoparticle composition.
specific trend for the W values with the nanoparticle composition suggests that other properties of the composite, such us the characteristics of the host glass in terms of defects, density and stoichiometry (all of them affect the dielectric properties), should be more deeply explored for improving the
Fig. 3. Corrected far-field (closed aperture) Z-scan curves recorded at 527 nm of wavelength, with a 6 ps single pulse at 1 Hz of repetition rate. The solid line is the curve fit obtained in the Sheik–Bahae frame. The samples have nanoparticle composition Au28Cu72 (a) and Au74Cu26 (b).
sign [18]. Taking into account that in our samples the linear absorption effects are limited [19] (a0/2k n0, where k = 2p/ l: for the cases at issue a0/2k term is always smaller than 0.1), the n2 and b coefficients are simply proportional to the real and ð3Þ the imaginary part of xMNCG , respectively. Knowing the form of ð3Þ the complex function f e, we can estimate the value of xMNCG , and thus of n2 and b. This can in some cases gives rise to a negative sign of n2. The trend shown by the coefficient b as a function of the relative concentration of the two metals can be explained in the same way. Calculation of f e is at present still critical, since it depends on both the dielectric functions of the host matrix and of the embedded alloy nanoparticles, and there are no reliable data for the dielectric functions of AuxCu100x alloys such those in our different samples. As far as the figures of merit are concerned, Fig. 5 shows W and T values as a function of the MNCG composition. To our knowledge, the reported W values are the highest ever obtained for MNCGs, being in some cases larger than 0.1. However, it is clearly evident that – for any composition – the required condition for the W FOM is not satisfied. The absence of a
Fig. 5. Values of the T (a) and of the W (b) figures of merit assessing the effectiveness of the MNCG material for optical switching devices, plotted as a function of the nanoparticle composition.
E. Cattaruzza et al. / Applied Surface Science 254 (2007) 1017–1021
suitability of these composites in terms of the W FOM. On the other hand, the trend exhibited by the coefficient b with the nanoparticle composition is very interesting for possible application of these composites in the switching device technology. As a matter of fact, being the T FOM simply proportional to b, the requirement T < 1 is directly satisfied for any negative values of b, as shown in Fig. 5: this is the case of the samples containing Cu-rich alloy nanoparticles.
1021
MNCGs; however, they are still insufficient for the application of these MNCG materials in all-optical switching technology. Moreover, they did not exhibit a significant trend with respect to the alloy nanoparticle composition. Work is in progress to clarify the role of the host dielectric matrix (in terms of structure and composition) in determining the final values of the figures of merit. References
4. Conclusions In this paper, we studied composite glasses made of Au–Cu alloy nanoclusters embedded in silica obtained by radiofrequency co-sputtering followed by thermal annealing in reducing atmosphere. The Au/Cu atomic ratio covered the whole Au–Cu phase diagram, from pure Au to pure Cu. The annealing in reducing atmosphere induced the formation of Au– Cu metal alloy nanoparticles of different composition, depending on the relative amount of gold and copper inside the as-deposited samples. From the optical point of view, the annealed samples exhibit in the visible range a well-defined surface plasmon resonance absorption: this band is centered at a wavelength that increases monotonically from pure Au to pure Cu. Apart from the AuCu1.1 sample, for which we did not observe optical nonlinearities, significantly high values of the nonlinear refractive index n2 were measured for the annealed samples: the (Au + Cu)-doped silica samples exhibited positive values of n2 in the range (1–5) 1011 cm2/W, without a specific dependence on the composition. Only the sample containing pure Au exhibited a negative n2 value about 6 1011 cm2/W. The nonlinear absorption coefficient b exhibited on the contrary a peculiar trend as a function of the MNCG composition, going from positive to negative by increasing the Au amount in the samples. The two main figures of merit assessing the effectiveness of a material for application in the switching technology field were evaluated for all the MNCG samples. The values of the T FOM show that the composite glasses made by Cu-rich alloy nanoparticles actually satisfy the requirement related to the nonlinear absorption properties. The values of the W FOM, related to the linear absorption properties, are the highest ever detected in
[1] R.F. Haglund Jr., Mater. Sci. Eng. A 253 (1998) 275. [2] M. Perner, P. Bost, U. Lemmer, G. von Plessen, J. Feldmann, U. Becker, M. Mennig, M. Schmitt, H. Schmidt, Phys. Rev. Lett. 78 (1997) 2192. [3] E. Cattaruzza, G. Battaglin, F. Gonella, R. Polloni, G. Mattei, C. Maurizio, P. Mazzoldi, C. Sada, M. Montagna, C. Tosello, M. Ferrari, Philos. Mag. B 82 (2002) 735. [4] F. Gonella, P. Mazzoldi, Metal nanocluster composite glasses, in: H.S. Nalwa (Ed.), Handbook of Nanostructured Materials and Nanotechnology, vol. 4, Academic Press, San Diego, 2000, p. 81. [5] G.I. Stegeman, E.M. Wright, N. Finlayson, R. Zanoni, C.T. Seaton, J. Lightwave Technol. 6 (1988) 953. [6] G. Battaglin, E. Cattaruzza, F. Gonella, R. Polloni, B.F. Scremin, G. Mattei, P. Mazzoldi, C. Sada, Appl. Surf. Sci. 226 (2004) 52. [7] E. Cattaruzza, G. Battaglin, P. Canton, C. de Julia´n Ferna´ndez, M. Ferroni, T. Finotto, C. Maurizio, C. Sada, J. Non-Cryst. Solids 336 (2004) 148. [8] E. Cattaruzza, F. Gonella, Metal nanoclusters by ion implantation, in: H.S. Nalwa (Ed.), Encyclopedia of Nanoscience and Nanotechnology, vol. 5, American Scientific Publishers, Stevenson Ranch, 2004, p. 369. [9] F. Gonella, Nucl. Instrum. Methods B 166–167 (2000) 831. [10] E. Cattaruzza, Nucl. Instrum. Methods B 169 (2000) 141. [11] G. Mattei, Nucl. Instrum. Methods B 191 (2002) 323. [12] G. Mattei, G. De Marchi, C. Maurizio, P. Mazzoldi, C. Sada, V. Bello, G. Battaglin, Phys. Rev. Lett. 90 (2003) 085502. [13] C. Maurizio, Nucl. Instrum. Methods B 218 (2004) 396. [14] G. Battaglin, P. Calvelli, E. Cattaruzza, F. Gonella, R. Polloni, G. Mattei, P. Mazzoldi, Appl. Phys. Lett. 78 (2001) 3953. [15] M. Sheik-Bahae, A.A. Said, T.H. Wei, D.J. Hagan, E.W. Van Stryland, IEEE J. Quantum Electron. 26 (1990) 760. [16] G. Mattei, G. Battaglin, E. Cattaruzza, C. Maurizio, P. Mazzoldi, C. Sada, B.F. Scremin, J. Non-Cryst. Solids 353 (2007) 697. [17] H. Okamoto, D.J. Chakrabarti, D.E. Laughlin, T.B. Massalski, in Phase Diagrams of Binary Gold Alloys, Monograph Series on Alloy Phase Diagrams, ASM International, Metals Park, OH, 1987. [18] F. Hache, D. Ricard, C. Flytzanis, U. Kreibig, Appl. Phys. A 47 (1988) 347. [19] E. Cattaruzza, G. Battaglin, F. Gonella, G. Mattei, P. Mazzoldi, R. Polloni, B.F. Scremin, Appl. Surf. Sci. 247 (2005) 390.
