1. Introduction

Magnetooptical (MO) materials are used in several nonreciprocal applications, for instance in optical isolators [1–3 ], optical circulators [4], polarized microscopy, and spatial light modulators [5]. The magnitude of the MO response is proportional to the propagation length of the light through the MO material; therefore thick or long (> 100 μm) MO materials are needed to obtain sufficient Faraday rotation (FR) or nonreciprocal phase shift to enable optical devices. In particular, the MO response is generally small in the near infrared (NIR) wavelength because of small spin-orbit interactions, and devices are based on thick MO materials or involve geometries where the light passes along the plane of a thin MO film. To enable device scaling, we propose here an MO structure that can provide a FR > 45° within a small thickness (< 1 μm). This structure uses a one dimensional (1D) magnetophotonic crystal (MPC) which is composed of a MO material layer sandwiched between two Bragg mirrors (BM), enhancing linear [6–8 ] and nonlinear [9] MO effects analogous to a Fabry-Perot resonator. In addition, we introduce a MO material, cerium-substituted yttrium iron garnet, CeYIG or Ce_xY_3−xFe₅O₁₂, into this MPC. CeYIG is known as a good MO material showing large FR in the NIR region, but highly cerium-substituted (x > 1.0) garnet has been difficult to grow without secondary phases forming. Hence, an MPC comprising CeYIG has not been reported so far. Recently we showed good MO properties for CeYIG which was prepared on non-garnet substrates by vacuum annealing [10, 11 ] or by growth on a YIG-seed layer [1, 12 ]. In this paper, we demonstrate the enhancement of the FR in a MPC comprising CeYIG. The multilayer structure is designed and analyzed using a matrix approach [6]. Structural properties and the valence state of cerium in the garnet are investigated.

2. MPC preparation

The MPC consisted of a rare-earth substituted gadolinium gallium garnet (SGGG) substrate on which SiO₂/(Ta₂O₅/SiO₂)⁸/CeYIG/(SiO₂/Ta₂O₅)⁸ was grown. First, the bottom BM, SiO₂/(Ta₂O₅/SiO₂)⁸, was fabricated by ion beam evaporation (IBE) at room temperature. The as-deposited SiO₂ and Ta₂O₅ were amorphous and had thicknesses of 224 and 164 nm respectively. The center wavelength of the bottom BM was 1524 nm. The SGGG substrate was chosen only to avoid generation of cracks in the film during annealing and cooling steps caused by thermal mismatch, and other substrates with similar thermal expansion coefficients may also be suitable. After the fabrication of the first BM, a 423 nm thick polycrystalline CeYIG layer was prepared by rf magnetron sputtering (Shimadzu, HSR-551S). The films were deposited at a rf power of 75 W applied to a 4 inch diameter target in 10 mTorr of Ar gas. The composition of the target was Ce_1.0Y_2.5Fe_5.0O_12−δ. The substrates were held at room temperature with water cooling during deposition. The deposition rate was 3.2 nm/min. The as-deposited film was amorphous, then the samples were annealed at various temperatures and for various times in 5 Pa (37 mTorr) of residual air. The composition of the CeYIG was Ce_1.0Y_2.0Fe_5.0O_12−δ, measured by energy dispersive x-ray spectroscopy (EDS, JEOL, JSM-6700F). Figure 1 shows x-ray diffraction (XRD) patterns of CeYIG on a BM annealed at various temperatures for 30 minutes [Fig. 1(a)] and annealed at 800°C for various times [Fig. 1(b)]. A 1 inch diameter substrate was coated in one batch and cut into 5 mm × 5 mm squares, and these were annealed under various conditions. A Cu-Kα source at wavelength 0.1541 nm was used in a ω–2θ thin film geometry XRD measurement. To eliminate the substrate peaks, the samples were tilted by 5° during measurement. Each XRD result is offset along the vertical axis in the figures for clarity.

 

Fig. 1 (a) XRD patterns of the SGGG substrate/SiO₂/(Ta₂O₅/SiO₂)⁸/CeYIG samples annealed at various temperatures T _a (700–900°C) for an anneal time of t _a = 30 minutes. (b) XRD patterns of the samples annealed at T _a = 800°C for various times t _a (1–90 minutes). The thickness of CeYIG was 423 nm. White triangles show the peaks of Ta₂O₅, white circles show hematite, and black squares show garnet, Y₃Fe₅O₁₂.

Download Full Size | PDF

The Ta₂O₅ in the BM was crystallized and showed multiple strong polycrystalline peaks, independent of the garnet crystallization. Crystallized Ta₂O₅ showed a hexagonal structure. The optical loss of the crystallized Ta₂O₅ will be discussed in the next section. The samples annealed at 750–850°C showed polycrystalline peaks which are indexed to the diffraction peaks of YIG. The (420) peak was the best to identify the crystallization of CeYIG because it did not overlap with the Ta₂O₅ peaks. The iron garnet structure is cubic with a large lattice parameter of ~1.2 nm and 8 formula units per unit cell. The film annealed above 900°C showed additional hematite peaks, decreasing the transmissivity of the film drastically. Figure 1(b) showed that an anneal time of 30 minutes was sufficient for crystallization of the CeYIG.

Figure 2 shows x-ray photoelectron spectroscopy (XPS) spectra in the vicinity of the cerium 3d binding energy. Figure 2(a) is for the samples annealed at various temperatures, and Fig. 2(b) is for those annealed for various times. The reference peaks were obtained from [13]. Before XPS measurements, 2 nm thickness of the CeYIG films was etched to remove effects of changes in oxidation state at the surface. The peak position was calibrated with the C1s peak at 284.6 eV to eliminate charging effects. The enhancement of FR introduced by the substitution of cerium is due to the electronic transitions between the Ce³⁺ (dodecahedral sites) and the Fe³⁺ (tetrahedral sites), hence minimization of the concentration of Ce⁴⁺ is important to avoid a decrease of the FR and transmissivity. Ce⁴⁺ is more stable than Ce³⁺ and therefore cerium atoms often form Ce⁴⁺ in oxides, however the CeYIG annealed here in 5 Pa of residual air did not show Ce⁴⁺ regardless of anneal temperature.

 

Fig. 2 (a) XPS spectra of the SGGG substrate/SiO₂/(Ta₂O₅/SiO₂)⁸/CeYIG samples annealed at various temperatures (T _a = 700–900°C) for an anneal time of t _a = 30 minutes. (b) XPS spectra of the samples annealed at T _a = 800°C for various times t _a. Black squares show the peaks for Ce³⁺, red triangles show Ce⁴⁺.

Download Full Size | PDF

Figure 3(a) shows the FR loops of samples annealed at various temperatures for 30 minutes, measured with the field and light perpendicular to the film plane. The wavelength λ = 580 nm was used for characterization of MO response because of the high transparency of CeYIG on BMs. The BMs have a strong photonic stop band in the vicinity of 1550 nm. The error bars indicated the fluctuation of the saturated FR on repeated measurements. The sample annealed at 800°C for 30 minutes showed the largest FR [Fig. 3(b)] of ~1.8 °/μm at an applied field of 2.0 kOe, thus this anneal condition was used for preparation of the MPC. The roughness of CeYIG was ~0.4 nm rms, which is too small to affect transmission of light. Saturation at ~2 kOe corresponds to overcoming the shape anisotropy of the CeYIG layer. This FR was 0.85 times the value of single crystalline CeYIG shown in [14] (~2.1 °/μm at λ = 580 nm) and 1.2 times the value of polycrystalline CeYIG shown in [11] (~1.5 °/μm at λ = 580 nm). The differences might be produced by the effects of interference due to the bottom BM. After annealing the SGGG substrate/SiO₂/(Ta₂O₅/SiO₂)⁸/CeYIG structure, the top BM was fabricated by IBE at room temperature without further anneal. The center wavelength of the top BM was set to 1365 nm in order that the MPC should have a resonant wavelength of around 1550 nm. Figure 4(a) shows a cross-sectional compositional image of a complete MPC taken with a field emission scanning electron microscope (FE-SEM, JEOL, JSM-6700F) using back-scattered electrons. Clear boundaries between the layers were observed and no obvious cracking of the BM was seen, but the top and bottom BMs were asymmetric to tune the resonant wavelength, which might affect the optical response. The measured thicknesses of the Ta₂O₅ and SiO₂ in the top BM were 164 and 224 nm, and that in the bottom BM were 178 and 257 nm, respectively. In the following calculations, these values were used as actual thicknesses.

 

Fig. 3 (a) FR loops versus applied field of the SGGG substrate/SiO₂/(Ta₂O₅/SiO₂)⁸/CeYIG samples annealed at various temperatures (T _a = 700–900°C) for an anneal time of t _a = 30 minutes. The measurement wavelength was λ = 580 nm. (b) FR versus anneal temperature at an applied field of 2 kOe. The error bars show the FR deviation (difference between the maximum and minimum).

Download Full Size | PDF

 

Fig. 4 (a) Cross-sectional compositional image of a fabricated MPC consisting of SGGG substrate SiO₂/(Ta₂O₅/SiO₂)⁸/CeYIG/(SiO₂/Ta₂O₅)⁸. (b) Thin black line shows refractive index profile through the MPC. Thick red line shows amplitude of electric field.

Download Full Size | PDF

Figure 4(b) shows the profile of propagating light calculated as the squared amplitude of the electric field in MPC at a wavelength of 1570 nm. The optical parameters used in this calculation were determined by a fitting approach which is discussed in the next section. The normal incidence linearly polarized light was localized at the CeYIG “defect” layer which breaks the periodicity of the refractive indices. This indicates a long residence time or multiple reflections of light in the CeYIG layer, therefore nonreciprocal effects are expected to be enhanced within this sample.

3. Magnetooptical response of MPC comprising CeYIG

Figure 5 (a) shows the measured and calculated transmission spectra of the fabricated MPC. The transmissivity was measured with a Shimadzu UV-3100PC spectrophotometer. FR was measured with the rotating analyzer method [15] (Neoark, BH-M600VIR-FKR-TU). A magnetic field of 2 kOe was applied to the film during the measurement. The localized mode was observed experimentally at a wavelength of λ = 1570 nm. The lower transmissivity compared with the calculation was due to the layer thickness differences in the two BMs. In Fig. 5(b), an enhanced FR of –2.92° is shown at 1570 nm. Figure 5(c) and 5(d) show the response in more detail. In comparison, a CeYIG film fabricated in the same batch without BMs gave a FR that was 30 times lower, i.e. 0.097°, corresponding to 2.300°/μm, which is comparable with other reports of the FR of polycrystalline CeYIG at this wavelength [11]. However the transmissivity was quite low. This was probably caused by light scattering at the grain boundaries of the crystallized Ta₂O₅ in the bottom BM [16], and thickness distribution within the BMs, both of which also limit the FR enhancement.

 

Fig. 5 (a) and (b) show transmission and FR spectrum of the fabricated MPC in a wide range of wavelength. (c) and (d) are enlarged figures. In all figures, solid lines show the calculation, and the circles show the experiments. Inset of (b) shows the Faraday rotation vs. magnetic field at a wavelength of λ = 1570 nm.

Download Full Size | PDF

To quantify the loss of Ta₂O₅, material parameters such as refractive index n, extinction coefficient κ and the refractive index difference between the refractive indices of the left- and right- circularly polarized light Δn of the entire material stack of the MPC were determined with a fitting method based on a matrix approach. FR and ∆n have a relationship as shown in [15], with FR = –180dΔn/λ, where d is the CeYIG thickness. The calculated results are shown in Fig. 5 as solid lines, in good agreement with experimental data. This fitting analysis showed the quantitative effect of annealing of the bottom BM. The estimated extinction coefficient κ of Ta₂O₅ and of SiO₂ in the bottom BM were 2.85 × 10⁻², and 0, respectively. In contrast, the Ta₂O₅ and SiO₂ in the top BM were zero (less than order 10⁻⁶). Differences in refractive indices on annealing were also obtained. The derived optical parameters are shown in Fig. 6 . The FR calculated for the CeYIG (FR = −0.12°) is in agreement with that measured on a CeYIG film without the BM (FR = −0.10°). Thermal treatment might cause the difference between the refractive indices of Ta₂O₅ and SiO₂ in the top and bottom BMs.

 

Fig. 6 Refractive index spectra of (a) SGGG substrate, (b) Ta₂O₅ in the bottom BM, (c) SiO₂ in the bottom BM, (d) CeYIG, (e) Ta₂O₅ in the top BM, and (f) SiO₂ in the top BM. Only for CeYIG, extinction coefficient, and refractive index difference between the refractive indices of the left- and right- circularly polarized light spectra are shown. The extinction coefficients of SGGG substrate, Ta₂O₅, SiO₂ in the bottom BM, Ta₂O₅, and SiO₂ in the top BM were 0, 2.85 × 10⁻², 0, 0, and 0, respectively. The difference between the extinction coefficients of the left- and right- circularly polarized light of CeYIG was 0. No dispersion was assumed. These parameters were used in the calculated spectra in Fig. 5.

Download Full Size | PDF

The measured FR of the fabricated MPC divided by the thickness of garnet layer was −6.90 °/μm (λ = 1570 nm) and the reported values of the FR of other MPCs working in the NIR region are –7.30 °/μm (λ = 980 nm) [17], –4.11 °/μm (λ = 1149 nm) [18], and –0.19 °/μm (λ = 1552 nm) [19]. In the vicinity of the wavelength of 1550 nm, the obtained FR was 36 times larger than that of [19], however the transmissivity was 21 times lower. Further improvement of the process, for example the suppression of the crystallization of Ta₂O₅ is expected to increase the transmissivity, enabling devices such as an alignment-free thin film optical isolator, a Q-switch for high power laser generating NIR light, and a spatial light modulator for data storage.

4. Conclusion

A 1D MPC comprising CeYIG was fabricated using vacuum annealing which led to crystallized garnet in which the Ce consisted almost entirely of Ce³⁺. The obtained FR was –2.92° at a wavelength of λ = 1570 nm, 30 times greater than that of a CeYIG film without the Bragg mirrors. Matrix approach simulations quantitatively showed the degradation of the optical indices of Ta₂O₅ and SiO₂ in the bottom BM due to annealing. Improvement of the thermal stability of the BM might solve this issue. This report showed a practical growth process to realize a thin film MO structure that can operate in the NIR wavelength regime. This might provide an important milestone for potential device applications.