1. Introduction

In recent years there has been an increasing interest on the study of different materials for applications based in optical signal processing systems, which require a large nonlinearity and an ultra-fast response time [1, 2]. Different materials have been synthesized and studied for this purpose, among them nanocomposites with different geometries. Given the nanometric nature of some of the dimensions of the composites, their response to visible light can be considered as that of an effective medium. An interesting feature is the possibility of tailoring the nonlinear response of such effective media through manipulation of the physical properties [3], such as: composition, density, and the shape of the nanostructures [4].

The possibility of manipulating the optical properties of the nanostructured materials makes them of a great interest for applications in optical information processing systems [5]. Those applications include optical switches and other photonic devices with geometries based in waveguide form [2, 6]. Thereby, in order to achieve the integration of these nanomaterials for this purpose, fabrication of devices with precise control of their physical properties, represent a technological challenge. Since this is of interest for the optoelectronics and photonics industries [7,8], many methods for the production of nanostructured materials, either physical or chemical, have been implemented and evaluated [3,9–11].

Among the different techniques explored for producing nanostructures, the Atomic Layer Deposition (ALD) has proved to be a reliable technique for producing nanolayers with well controlled physical and chemical properties. In addition, the ALD technique [12], allows the obtention of ultrathin multilayer films composed of several bilayers with precise thickness control on nanometric scale [13]. The optical, electronic and physical properties of these materials will be a function of the layer thicknesses and growth temperature, as well as of the appropriate selection of materials [14,15]

Some studies of the third order nonlinear response of this kind of nanolayered materials have focused on the enhancement of ${\chi }_{\mathit{eff}}^{(3)}$ of the composite through the manipulation of the effective dieletric constant ∊_eff, related to the effective refractive index $n_{\mathit{eff}}=\sqrt{{∊}_{\mathit{eff}}}$, of two ultrathin materials [16]. In this case, ∊_eff (using a E parallel to the plane of the layer) is given by 1/∊_eff = f_a/∊_a + f_b/∊_b, with ∊_a and ∊_b the dielectric constants for materials a and b respectively, and f_a and f_b the corresponding volume fractions. For such material, the effective third-order nonlinear susceptibility can be written as:

Nevertheless, for E perpendicular to the plane of the layers, ∊_eff = f_a∊_a + f_b∊_b and ${\chi }_{\mathit{eff}}^{(3)}$ is given by [17]:

From here, it is easily seen that the nonlinear optical responses can be tuned through manipulation of the volume fraction of the nanolayers, the dielectric constants, and the interaction of E with the plane of layers.

On the other hand, there have been many studies of the optical response of ZnO, a good candidate for implementing different optical functions [11], including the fabrication of waveguides. This is due to its high linear refractive index n₀, which results in a strong confining factor, their large nonlinear refractive index n₂ and luminescence properties. Some studies [7] include the third-harmonic generation (THG) in ZnO − Al₂O₃ bilayers synthesized by ALD, and its dependence on the layers widths. They found an increase of the THG in ZnO/Al₂O₃ nanolaminates through the control of the crystal structure of the ZnO without affecting the material volume [7].

In this work, we study the absorptive and refractive contributions to the nonlinear response, given by the nonlinear refractive index n₂ (related to Re χ⁽³⁾(−ω; ω, −ω, ω)), and the two-photon absorption (TPA) coefficient β (related to Im χ⁽³⁾(−ω; ω, ω, −ω)), of samples based on multiple Al₂O₃/ZnO bilayers fabricated by ALD in silica. The absorptive and refractive contributions to the nonlinearity of the samples were studied by means of the z-scan technique in the near-infrared regime. A study of the nonlinear parameters β and n₂ is made for different Al₂O₃/ZnO ratios in the nanolaminated stack. Finally, we discuss the possible applications of this materials in optical signal processing by means of figures of merit W and T.

2. Sample preparation and linear optical properties

Ultrathin multilayer films based on Al₂O₃/ZnO bilayer stacks were deposited on Si (100) and SiO₂ substrates in order to study their linear and nonlinear optical properties. We fabricated the samples by ALD technique using a Beneq TFS 200 ALD viscous-flow reactor at 200 °C. Trimethylaluminum (TMA) (Strem 93 − 1360) and Diethylzinc (DEZ) (Strem 93 − 3030) were used as the aluminum and zinc sources, respectively. For this process, deionized water was used as the oxidant agent, all the precursor were used at room temperature. UHP N₂ (< 10⁶ ppm O₂) was used as carrier/purging gas purified by means of a Centorr 2G − 100 − SS − 120 gettering furnace.

The scheme in Fig. 1(a) depicts the pulse sequence of ALD cycles used to grow the Al₂O₃/ZnO ultrathin multilayer stacks. The complete ALD cycle for both precursors with the exposure dose and purge time, respectively, is also shown in Fig. 1(a). In the scheme, the parameter d corresponds to the number of repetitions in the recipe, necessary to obtain the nominal total programmed thickness (∼ 500 nm). In this arrangement, the Al₂O₃ layers have a thickness Δ_x (nm), while Δ_y (nm) corresponds to the ZnO layers. Both, Δ_x and Δ_y, were varied for different samples.

 

Fig. 1 a) The sequence of ALD cycles for multi-layer fabrication samples, and b) the geometry of the resulting materials.

Download Full Size | PDF

A set of 5 samples, labeled M_Δx,Δy, with different number of cycles were prepared. The total thickness of the samples was kept at ∼ 500 nm in order to keep their absorption coefficients values sufficiently low to get a useful throughput and to avoid any damage to the samples. Additionally, this thickness value allows efficient light coupling when the samples are used to produce slab or channel waveguides. Figure 1(b) shows the multilayer scheme of the bilayers M_Δx,Δy, deposited in SiO₂ substrate for the nonlinear measurements.

Figure 2 shows a cross-sectional TEM image of a Al₂O₃/ZnO nanolaminate, in this case (Al₂O₃/ZnO)₅/SiO₂/Si, in which Al₂O₃ and ZnO layers were made by 100 and 50 ALD cycles, respectively. At the bottom of Fig. 2(b) the Si wafer is clearly observed, as well as a native SiO₂ thin layer. As is demonstrated on Fig. 2(a), well defined nanolaminated structures have been grown, composed by gray layers (Al₂O₃) and darker ones (ZnO) with very flat interfaces between the different metal oxides layers. The single layer thickness obtained was approximately 10 nm for Al₂O₃ and 10 nm for ZnO, in agreement with the growth rate per cycle calculated for both Al₂O₃ and ZnO. It is easy to discern that the Al₂O₃ layer is amorphous, while the ZnO layer presents crystalline domains (Fig. 2(b)). The above result indicates that the ALD method, has a high accuracy in the thickness control when the multilayers are grown on flat substrates.

 

Fig. 2 Nanolaminated were observed by TEM using a JEOL JEM2010 microscope at 200 kV. Sample was prepared through focus ion beam (FIB) on a JEOL FIB-4500 SEM.

Download Full Size | PDF

2.1. Linear characterization

2.1.1. Absorption spectra

In order to characterize the linear optical properties of the resulting samples, the optical density spectra of the materials were taken using a Hitachi 7000 spectrophotometer. Figure 3 shows the optical density spectra for all samples M_Δx,Δy. Also the Fig. 3 shows a high transparency in the visible region of all samples produced, and periodic oscillations in the curves, due to interference effects related to the thickness and refractive index of the stack. In addition, the samples show the typical absorption edge towards the UV, that starts below 400 nm, for thin films composites containing ZnO.

 

Fig. 3 Absorbance spectra of samples M_Δx,Δy. The values of Δ_x and Δ_y are in nm.

Download Full Size | PDF

2.1.2. Ellipsometry measurements

It is important to measure the linear refractive index in order to observe the dependence of ∊_eff with the manipulation of the amount of Al₂O₃ and ZnO and its correlation with the nonlinear optical parameters of the materials. Briefly, mixing 2 different materials in a nanolaminate structure will produce intermediate refractive index values that will go from 1.65, for Al₂O₃, to 2 for ZnO (at 632 nm). To calculate this, we use the variable angle spectroscopic ellipsometry technique that can give us information of the total multilayer stack thickness and its refractive index across the optical spectrum [8]. Hence, we use a J. A. Woollam M − 2000 ellipsometer to measure the amplitude and phase difference between reflection coefficients (r_p and r_s) from λ = 350 nm to 1000 nm at different angles (θ = 45°, 55°, 65° and 75°). With the results obtained, we use the Tauc-Lorentz model in order to get a fit, and extract the refractive index for all samples.

The dispersion curves of the samples fabricated on Si are shown in Fig. 4. The refractive index curves of samples M_1,Δy (Δ_x = 1 nm and Δ_y is varying), are presented in Fig. 4(a), while Fig. 4(b), exhibits the refraction index curves of samples M_Δx,10 (Δ_x is varying and Δ_y = 10 nm is fixed). The refractive index of the samples shown in Fig. 4(a) exhibit an abrupt change going from samples M_1,2 to samples M_1,5 and M_1,10. This is probably due to the significant change in ZnO content. A quasi-linear control of the refractive index could be observed in Fig. 4(b) due to the changing amount of Al₂O₃, as expected. These results are consistent with those previously reported in the literature, where a modulation of the linear refraction index has been demonstrated in samples consisting of Al₂O₃/ZnO alternating layers of different thickness [7, 18], having overall values of ∼ 500 nm and ∼ 100 nm, respectively.

 

Fig. 4 Refractive index for a) M_1,Δy and b) M_Δx,10 samples.

Download Full Size | PDF

The total thickness, amount of the materials deposited on the substrate, the aspect ratio and the number of interfaces are shown in the Table 1.

Table 1. The total thickness, amount of the materials deposited on the substrate, the aspect ratio and the number of interfaces of the multilayer M_Δx,Δy samples.

View Table | View all tables in this article

3. Nonlinear optical experiments

3.1. Experimental setup

The nonlinear response of the samples was studied using the z-scan technique with fs pulses. This technique allows the resolution of the refractive and absorptive contributions to the nonlinear response, as well as to the determination of the sign of both contributions [19]. Briefly, in this technique the nonlinear sample is scanned across the focal plane of a lens, where a spatially Gaussian pulse is tightly focused. In this sense, the transmittance through an aperture located in the far field is measured as the nonlinear sample is scanned across the focal plane of the lens. Hence, the apertured detector, is sensitive to changes in both nonlinear absorption and refraction (closed aperture z-scan). Nevertheless, we can measure the nonlinear absorption through a slightly different open-aperture configuration, where the whole transmitted beam is detected (open-aperture Z-scan), which then will be sensitive to only nonlinear absorption. Thereby, is possible to discern the nonlinear optical responses by a simple division of the closed-aperture by the open-aperture transmittance traces [20].

In the setup employed, illustrated in Fig. 5, the light source is a 76 MHz repetition rate Ti:sapphire laser with 100 fs duration pulses at 800 nm wavelength. The incident laser power was controlled by a half-wave plate followed by a polarizer prism and monitored by detector D1. The sample was mounted in a stage that moves along the beam propagation direction (z axis). The beam waist at z = 0 was 47 μm. Pre-focal and post-focal positions correspond to z < 0 and z > 0, respectively. After crossing the sample, the laser beam passes through a circular aperture, placed in the far-field region, being detected by detector D2.

 

Fig. 5 Z-scan setup employed in the experiments,BS is a beam splitter, L1 is a lens with focal length of 20 cm, D1 is the detector whose monitors the input laser power, while the D2 is de detector to monitoring the closed aperture z-scan signal, respectively.

Download Full Size | PDF

The experimental data for the closed to open aperture ratio can be used to extract the nonlinear refractive index n₂ by measuring the peak to valley transmittance change ΔT_pv, using ΔT_pv = 0.406kL_effn₂I₀, where k = 2π/λ, L_eff = [1 − exp (−α₀L)]/α₀, L is the sample length, α₀ is the linear absorption coefficient and I₀ is the laser intensity at the focal plane. In the same way, the open aperture z-scan data (for S = 1), can be used to get the two-photon absorption coefficient β, through $T(z=0)=1-\beta I_0{L}_{\mathit{eff}}/\sqrt{8}$ [21].

3.2. Results and discussion

Figure 6 displays experimental open and closed-aperture z-scan results for two representative samples: M_1,2 and M_1,5, the circles represent the experimental data and the continuous lines are the theoretical fits to them. Figures 6(a) and 6(b) show the open and closed z-scan data for M_1,2, while Fig. 6(c) and Fig. 6(d) the open and closed z-scan data for M_1,5, respectively. Notice that in Fig. 6(a), there is not discernible nonlinear absorption, since there is no changes in the normalized transmittance. On the other hand, Fig. 6(c), shows the typical two photon absorption profile, where a minimum is discernible in the focal plane. For the closed-aperture z-scan results in Fig. 6(b) and Fig. 6(a), it is easy to discern the typical signature of a positive n₂, a pre-focal minimum followed by a post-focal maximum. The irradiance used for all these measurements was varied from 75.8 to 341 MW/cm² (see Table 2).

 

Fig. 6 Open and closed-aperture z-scan results for the samples M_1,2 a) open, b) closed; and M_1,5 c) open, and d) closed. For these experiments, the irradiances used were I₀ = 152 MW /cm² and I₀ = 341 MW /cm², respectively.

Download Full Size | PDF

The nonlinear parameters β and n₂ for all samples are displayed in the Fig. 7. These values have been obtained by fitting the theory, according to [19, 21], to the experimental open and closed z-scan data. The black points in Fig. 7(a), are the values for the two photon absorption coefficient β for each sample, while the black points in Fig. 7(b), are the n₂ obtained values for each sample.

 

Fig. 7 Correlation between the linear and nonlinear optical parameters. The Fig (a) show the linear absorption α at 2ω and the nonlinear optical absorption coefficient β. The Fig (b) show the linear optical index n₀ at ω and the nonlinear refractive index n₂.

Download Full Size | PDF

The nonlinear optical parameters have been obtained far away of resonance, nevertheless the linear optical absorption is slightly higher at the two photon regime, i.e. at twice the laser frequency, so that TPA process is possible. This is confirmed by the open z-scan data obtained, where the two photon absorption effect is observed (Fig. 6). Due to this, we have added the value of α(2ω) in Fig. 7(a), in order to check if there is a correlation between them. The results indeed suggest the small differences in linear absorption at 2ω generate a change in the non-linear absorption at ω in the samples studied, as expected. In the same way we have plotted in Fig. 7(b) the data of the linear refractive index n₀(ω) together with the measured n₂ values. The results show a similar behaviour, that there is a correlation between the linear and nonlinear refraction coefficients n₀ and n₂, thereby to ∊_eff for both samples as well. Even though in the z-scan geometry employed, the electric field vector is parallel to plane of the layers, we still expect an enhancement in the ∊_eff and thereby in n₂ as well. If the multilayers were to be used as waveguides we could expect in principle a further enhancement in ${\chi }_{\mathit{eff}}^{(3)}$.

Comparing our results with the literature, the nonlinear refractive index for Al₂O₃ bulk has been calculated and its value is around n₂ = 1.33 × 10⁻⁹ cm²/GW [22], and on the other hand Lin et al [23] have obtained a n₂ = 2.65 × 10⁻² cm² /GW value for ZnO thin films samples. Both values of the nonlinear refractive index are lower than the one we have determined for the M_1,10 bilayer sample.

Regarding the physics behind the nonlinear response, intra-pulse thermal effects are effectively discarded, due to to relatively long rise time of the temperature change induced by a single fs pulse, given by τ_rise = ω₀/v_s, where ω₀ is the beam waist and v_s is the sound velocity in the sample. However, the high repetition rate of the pulses (76MHz) could in principle imply the possibility of pulse to pulse thermal effects [24–26], but, given the off-resonance characteristics of the interaction, heating by linear absorption can be effectively ruled out.

In order to evaluate the potential of the material for application in all-optical switching devices, it is useful to calculate the figures of merit devised to assess the potential of different materials [27]. The first figure of merit W is given by:

In our case, we use the experimental data taken at the highest irradiance employed (where I₀ < I_sat) in order to estimate Δn_max, and hence a lower bound for W. Although we were not able to measure the actual β value for M_1,2, we can calculate the smaller β value than our apparatus can detect in order to give a possible value to T. To the case of the other samples, we use the values of n₂ and β obtained, to calculate in turn upper bounds for the values of W and T. The calculated W and T values are shown in Table 2.

At this point, we have obtained the limit values for the W and T figures of merit. Although we were not able to reach the I_sat (which could cause an enhancement in W value through Δn_max), the W values obtained from all samples are close to 1, even could be higher to this value. In the case of the T values, the β coefficient calculated for the sample M_1,2 suggest a value closer to 1, since the β coefficient could not be obtained with the irradiances used, while for M_1,10 sample, the T value is marginally close to 1 with β coefficient obtained.

Table 2. Nonlinear optical parameters n₂ and β, and figures of merit W and T, evaluated under the irradiances showed for samples M_Δx,Δy.

View Table | View all tables in this article

4. Conclusions

In summary, this work reports the fabrication by ALD of Al₂O₃/ZnO multilayer samples. The nonlinear optical parameters n₂ and β were evaluated for the samples with differences in the thickness of the Al₂O₃ and ZnO layers. The samples present a large positive nonlinear refractive index, that can be manipulated through the fabrication parameters. Comparing the results obtained, we note that to the samples with a higher content of ZnO with respect to Al₂O₃ amount, shows an enhancement in the nonlinear refractive index measured. Although we do not seem to reach irradiance values close to a saturation irradiance I_sat, and considering the large n₂ and low β values obtained, we claim that composites with Al₂O₃/ZnO multilayer could have a good potential for application in all-optical signal processing devices.