II. EXPERIMENTAL

Au/TiO₂ films were prepared by the multitarget sputtering system. The Au (99.9%) and TiO₂ (99.9%) targets of 3 inches in diameter were independently manipulated. The Au and TiO₂ were deposited alternatively onto synthetic quartz (SiO₂) glass (0.5 mm in thickness) or GaAs (0.35 mm in thickness) substrates. The substrates were not intentionally heated. The concentration of Au in the films was controlled by adjusting the deposition time. The radio frequency powers applied for Au and TiO₂ were 25 and 200 W, respectively. The sputtering was carried out in the gas mixture of 50% Ar and 50% O₂ (total gas pressure of 2.0 Pa). The heat treatment of the as-deposited films was carried out at 600, 700, 800, 900, and 1000 °C in air to control the size of the Au particles. The typical heat treatment time was 1 h.

X-ray diffraction (XRD) measurements were performed to investigate the crystallinity of the films and the mean diameter, d, (crystallite size) of Au particles in the films. The mean diameter was estimated using Scherrer’s equation. The nature of the chemical bond of the films was investigated by x-ray photoelectron spectroscopy (XPS) using a Mg Kα x-ray source. The morphology of the films and the diameter of the Au particles were examined directly by cross-sectional transmission electron microscopy (TEM) and the electron diffraction in the TEM. The composition of the films on GaAs substrate was determined by wavelength dispersive electron probe microanalysis.

Optical absorption spectra of the Au/TiO₂ films were obtained by a double beam spectrophotometer at room temperature. The nonlinear refraction and absorption coefficients of the films were measured using the standard Z-scan technique with the optical parametric amplifier (OPA) pumped by the Ti:sapphire regenerative amplifier. The typical pulse energy and the pulse duration were 1 μJ and less than 150 fsec, respectively. In the Z-scan measurement, we used fused silica as the standard material. The value of χ⁽³⁾ for fused silica was evaluated to be 1.3 × 10⁻¹⁴ esu from fitting with the data in Refs. 8 and 9. In the femtosecond pump-probe experiment, the OPA pumped by the Ti:sapphire regenerative amplifier was used. The wavelengths of the pump and probe laser beams were near the LSPR peak of the films. The typical pulse energy and the pulse duration were 1 μJ and less than 200 fsec, respectively.

III. RESULTS AND DISCUSSION

Figures 1 and 2, respectively, show the cross-sectional TEM images of the as-deposited Au/TiO₂ film and the film heat-treated at 1000 °C for 1 h together with the electron diffraction pattern of a selected area. The film was prepared by alternate deposition of Au and TiO₂ 30 times. The concentration of Au in the film was 29.6 at% (Au/Au+Ti). The thickness of the as-deposited and heat-treated Au/TiO₂ film shown in Figs. 1 and 2 were 250 and 190 nm, respectively. In Fig. 1, 30-layer structure is clearly seen and the Au particles are clearly recognized as dark contrast spots with diameters ranging from 2 to 6 nm in the layer. After the heat treatment, the film became thinner and the Au particles became larger. It is noteworthy that the particles grew and penetrated into the synthetic quartz (SiO₂) glass substrate. It seems that the nanometer-sized Au particles were grown catalytically in the presence of TiO₂. The electron diffraction pattern of the selected area of the as-deposited and the heat-treated film show diffraction rings attributed to the Au metal. The broad diffraction rings shown in Fig. 1 indicate that the size of Au particles is small in the as-deposited film. On the other hand, in Fig. 2, the spotty diffraction rings that could be indexed as the reflections of the Au metal indicate the larger crystallite size. Thus, the size of the Au particles in the films could be controlled by the heat treatment in air.

FIG. 1.

Cross-sectional TEM image of the as-deposited Au/TiO₂ film (29.6 at% of Au) together with the electron diffraction pattern of a selected area.

FIG. 2.

Cross-sectional TEM image of the Au/TiO₂ film (29.6 at% of Au) heat-treated at 1000 °C for 1 h together with the electron diffraction pattern of a selected area.

Figures 3(a) and 3(b) show the particle size distribution obtained from the analysis of the TEM micrographs shown in Figs. 1 and 2, respectively. A part of the micrograph field including 200 (Fig. 1) and 32 (Fig. 2) particles were randomly selected to analyze the size. On the one hand, the mean diameter of Au particles in the as-deposited film was estimated to be 3.6 nm. On the other hand, the mean diameter of Au particles in the heat-treated film was estimated to be 29.6 nm. The mean diameter of the Au particles in the heat-treated film was one order of magnitude larger than that in the as-deposited film.

FIG. 3.

Particle size distribution of Au particles obtained from the analysis of TEM micrograph shown in (a) Fig. 1 and (b) Fig. 2.

Figure 4 shows the XRD patterns of the as-deposited Au/TiO₂ film and the film heat-treated at 600, 700, 800, 900, and 1000 °C in air for 1 h. The TEM images of the as-deposited film and the film heat treated at 1000 °C are shown in Figs. 1 and 2, respectively. In Fig. 4, the as-deposited film shows only a halo pattern of amorphous SiO₂ glass substrate in a range of 2θ of 18°–25°. On increasing the heat treatment temperature, the intensity of the peaks at 38.1° and 44.3° increases, which are, respectively, assigned as (111) and (200) reflection lines of cubic Au. The crystal structure of the TiO₂ matrix changed from anatase to rutile at the heat treatment temperature above 900 °C. Beside the case of the as-deposited Au/TiO₂ film, the mean diameter (crystallite size) of Au particles in the films shown in Fig. 4 was estimated from the broadening of the (111) line. Figure 5 shows the relation between the mean diameter of Au particles in the films shown in Fig. 4 and the heat treatment temperature. The mean diameter of the Au particles increases from 4.9 nm at 600 °C to 33.7 nm at 900 °C and it becomes constant. It is noteworthy that the Au particles grow in the film even at the temperature of 600 °C. These results indicated that the size of Au particles could be controlled by changing the heat treatment under the melting point of bulk Au (1064.4 °C).

FIG. 4.

X-ray diffraction (XRD) patterns (●, Au; Δ, anatase; ●, rutile) of the as-deposited Au/TiO₂ film and the film heat-treated at 600, 700, 800, 900, and 1000 °C in air for 1 h.

FIG. 5.

Relation between the mean diameter of Au particle from the XRD shown in Fig. 4 and the heat-treatment temperature of the films.

XPS experiments were performed to elucidate the chemical state of the Au particles in the Au/TiO₂ film (shown in Fig. 1). Figure 6 shows the Au 4f electron spectrum of the film. The spin doublet of Au 4f, i.e., Au 4f_5/2 and Au 4f_7/2, is clearly observed. The binding energy of the experimental peak position of the 4f_7/2 was almost the same as that of the standard Au metal as indicated by an arrow in Fig. 6. Thus, the Au particles in the film were likely to be metallic Au. The electron spectrum of the Ti 2p and O 1s for the Au/TiO₂ film exhibited the TiO₂ structure. These results confirm that Au metal and TiO₂ are successfully formed by the multitarget sputtering method.

FIG. 6.

Au 4f electron spectrum of the as-deposited Au/TiO₂ film shown in Fig. 1.

Figure 7 shows the optical absorption spectra of the Au/TiO₂ films, and Fig. 4 shows the XRD spectra of the Au/TiO₂ films. In Fig. 7, the absorption peaks due to the LSPR of the Au particles are observed at the wavelength of 658–668 nm. The LSPR peak of the Au/TiO₂ films was shifted to more than 130 nm longer wavelength in comparison with that of the already reported Au/SiO₂ films.Reference Tanahashi, Manabe, Tohda, Sasaki and Nakamura⁶ This is because the refractive index of TiO₂ is larger than that of SiO₂. The redshift is explained in the literatureReference Kreibig and von Fragstein¹⁰ by using the following equation: , where is the real part of the complex dielectric constant of the particles, is the imaginary part, n is the refractive index of the matrices, and ω _sp is the LSPR peak frequency. In the case of the as-deposited film, the absorption band of the LSPR was weak. On the other hand, on increasing the heat treatment temperature from 600 to 800 °C, the LSPR absorbance increased and the LSPR peak position shifted to longer wavelength slightly, indicating that the Au particles became large.Reference Kreibig¹¹ When the heat treatment temperatures of the films was 900 and 1000 °C, the absorbance of the films decreased compared with that of the films heat treated at 700 and 800 °C. As mentioned earlier, the crystal structure of the TiO₂ matrix changed from anatase to rutile and the thickness of the film decreased at the heat treatment temperature above 900 °C. It seems that the crystal structure of TiO₂ matrix influenced the optical absorption spectra of the films. In Fig. 7, the broad background absorption, especially in the shorter wavelength (around 500 nm) region of the peak, is due to the band-to-band transition of the Au particles.

FIG. 7.

Optical absorption spectra of the as-deposited Au/TiO₂ film and the film heat-treated at 600, 700, 800, 900, and 1000 °C (shown in Fig.4) in air for 1 h.

The Z-scan technique is described in detail in Refs. Reference Sheik, Said, Wei, Hagan and Van Stryland12 and Reference Cheung and Gayen13. By a theoretical fit of the experimental data for both the closed-aperture and open-aperture transmittance, the values of Re [χ⁽³⁾] and Im [χ⁽³⁾] can be obtained. In this Z-scan experimental condition, the Rayleigh range of the beam was about 1.0 mm, which is greater than the thickness of either the Au/TiO₂ films or SiO₂ substrate. Figures 8(a) and 8(b), respectively, show the typical close- and open-aperture Z-scan normalized transmittance of the Au/TiO₂ film heat treated at 800 °C for 1 h (shown in Figs. 4 and 7). In Figs. 8(a) and 8(b), the normalized transmittance is defined as the ratio of the transmittance of the beam through the aperture in the nonlinear regime to that in the linear regime. In the film, the average diameter of the Au particles was estimated to be 16.8 nm from XRD and the peak position of the LSPR was 668 nm. In the measurement shown in Figs. 8(a) and 8(b), the wavelength of the laser beam was 584 nm, which was more than 80 nm shorter than that of LSPR peak of the film. In Fig. 8(a), the curve is characterized by a prefocal valley followed by a postfocal peak, indicating that the Re [χ⁽³⁾] of the film is positive. On the other hand, in Fig. 8(b), the enhanced transmission peak at the focus is indicative of the absorption saturation. The values of Re [χ⁽³⁾] and Im [χ⁽³⁾] were calculated to be 9.1 × 10⁻⁸ and −1.8 × 10⁻⁸ esu, respectively. Because of the absorption saturation, the sign of Im [χ⁽³⁾] was negative. These values were normalized by the absorption coefficient, α, (1.88 × 10⁵ cm⁻¹) of the film at the corresponding wavelength of χ⁽³⁾ measurement. The values of the figure of merit (FOM), Re [χ⁽³⁾]/α and |Im [χ⁽³⁾]|/α, of the film were calculated to be 4.8 × 10⁻¹³ and 9.5 × 10⁻¹⁴ esu·cm, respectively. When the wavelength of the laser beam (660 nm) was close to that of the LSPR peak (668 nm), the values of Re [χ⁽³⁾] and Im [χ⁽³⁾] were 2.5 × 10⁻⁷ esu (FOM, 9.7 × 10⁻¹³ esu·cm) and −3.6 × 10⁻⁷ esu (FOM, 1.4 × 10⁻¹² esu·cm), respectively. The FOM of 1.4 × 10⁻¹² esu·cm for the Au/TiO₂ film was one order of magnitude larger than that of 2.2 × 10⁻¹³ esu·cm for the Au/SiO₂ film already reported.Reference Tanahashi, Manabe, Tohda, Sasaki and Nakamura⁶ This is mainly due to the local field enhancement. In this Z-scan experiment, the FOM value of Im [χ⁽³⁾] at 793 nm was calculated to be 3.2 × 10⁻¹³ esu·cm and the Re [χ⁽³⁾] could not be measured correctly because the signal due to the Re[χ⁽³⁾] was too small. The sign of Im [χ⁽³⁾] at 793 nm was also negative. In the case of Ag nanocrystal–glass composites, the dispersion curves of Re [χ⁽³⁾] and Im [χ⁽³⁾] around the LSPR was reported in the literature.Reference Hamanaka, Nakamura, Hayashi and Omi¹⁴ In accordance with the results,Reference Hamanaka, Nakamura, Hayashi and Omi¹⁴ the |Im [χ⁽³⁾]|/α of the Au/TiO₂ film was resonantly increased near the LSPR peak position of the film and the sign of Im [χ⁽³⁾] was negative around the LSPR.

FIG. 8.

(a) Typical close and (b) open-aperture Z-scan normalized transmittance of the Au/TiO₂ film heat-treated at 800 °C for 1 h shown in Figs. 4 and 7. The solid line indicates the theoretical fit. The wave length of the laser beam was 584 nm.

Figure 9 shows the relation between the |Im [χ⁽³⁾]|/α and the average particle diameter of Au particles in the Au/TiO₂ films shown in Fig. 5. For the measurement, the laser beam (660 nm) was close to the LSPR peak positions (658–668 nm) of the films. In Fig. 9, the film with the average Au particle diameter of 16.8 nm shows larger |Im [χ⁽³⁾]|/α than the film with the average Au particle diameter of 4.9 nm. This is probably because the local field factor increases with increase in the diameter of the Au particles in the range of 3–30 nm.Reference Tanahashi, Manabe, Tohda, Sasaki and Nakamura⁶ However, the film with the Au particles of larger average diameter (33.7 nm) shows smaller |Im [χ⁽³⁾]|/α. This is probably because the size distribution of the Au particles with larger diameter of the film is large.

FIG. 9.

Relation between the absolute values of the Im [χ⁽³⁾]/α and the mean particle diameter of Au particles in the Au/TiO₂ films shown in Fig. 5.

The differential transmission ΔT/T of the pump-probe delays for the Au/TiO₂ film with average Au particle diameter of 33.7 nm is shown in Fig. 10. In this pump-probe experiment, the wavelength of the laser (the input power of 262.4 μJ·cm⁻²) was adjusted to the LSPR peak position (654 nm) of the film. In Fig. 10, the temporal behavior is decomposed into two decay components, the decay times of the fast and slow components were estimated to be 2.1 ps and more than 100 ps, respectively. The origin of the fast decay time is associated with the thermal equilibrium process between the electrons and lattice in the Au nanoparticles. On the other hand, the slow decay time comes from the cooling process by a thermal diffusion from the Au nanoparticles to the TiO₂ matrix.Reference Inouye, Tanaka, Tanahashi and Hirao¹⁵

FIG. 10.

Time evolution of ΔT/T of the pump-probe delays for the Au/TiO₂ film with the mean Au particle diameter of 33.7 nm.