1. Introduction

As one of the most well-known ferroelectric crystals, lithium niobate (LiNbO₃) crystals have obtained wide applications in many branches of modern photonics owing to the combination of many excellent properties [1,2]. Particularly, LiNbO₃ also serves as one of the most popular media for constructing integrated optical devices, e.g., electrooptic (EO) modulators, optical amplifiers, and integrated laser gain/conversion components [3]. The two dimensional (2D) optical waveguides (in either channel or ridge configurations) are the necessary platforms to realize the versatile functions of compact devices in modern photonic systems. Compared with channel waveguides, the ridge guiding structures usually possess much stronger lateral confinement of the light fields because the refractive index contrast of the cladding and waveguide in the transverse dimension is higher in ridge cases (the cladding is air with index of 1), which suggests a reduction of bending losses and component sizes of the photonic devices [4]. A few techniques have been applied to produce channel waveguides in LiNbO₃ crystals, including Ti or Zn indiffusion [3,5], proton exchange [6], ion implantation/irradiation [7–9], ultrafast laser writing [10], and optical inductions (e.g., by spatial solitons) [11]. Ion implantation has been considered to be one of the most universal techniques to produce waveguides in numerous optical materials [12,13]. By selecting different implantation parameters (e.g., ion species, energies, and fluences), one can construct guiding structures with diverse index profiles [7]. One of the specific features of ion implanted LiNbO₃ waveguides is that it is of more compact spatial distribution at the transverse cross section but with considerably large index contrast with the surroundings (particularly compared with Ti-indiffused one). In addition, ion beams can also be applied to fabricate bulk-quality LiNbO₃ thin films/membranes [7,14–16]. Nevertheless, to fabricate a ridge guiding structure one usually requires combination of two techniques: one is to form waveguide layers and another is to etch the selected regions for ridges. In practice, both dry etching (e.g., Ar ion beam sputtering) and chemical wet etching have been applied to remove the patterned regions of LiNbO₃ crystals to form ridges [17,18]. For example, Hartung et al. developed a method to form Zn-doped LiNbO₃ ridge waveguide by combination of liquid phase exitaxy (LPE) and ion beam enhanced etching (IBEE) [19]. Rabiei et al. fabricated a ridge waveguide on an ion sliced LiNbO₃ thin film [20]. For ferroelectric z-cut LiNbO₃ wafers, the immersion in hydrofluoric (HF) and nitric acid (HNO₃) can etch the –z domains in a considerably high rate (several microns per hour), whilst does not attack the + z domains [15]. By taking this feature into account, Barry et al. [21] applied electric field poling to realize spatially selective domain inversion on a z-cut LiNbO₃ wafer. After immersion in the acid etchant, the ridge stripes were formed owing to the different etching rate of + z and –z regions. With this patterned substrate they fabricated successfully ridge waveguides in the + z domain ribs by subsequent ion implantation, proton exchange or Ti indiffusion. However, such processing preserved simultaneously the guiding layers in the –z domain regions, which usually was not expected in the practical applications. In addition, this method requires electric field poling, which complicates the fabrication process. In this paper, we propose a new but simple method to produce ridge waveguides in z-cut LiNbO₃ wafers by combination of proton implantation and chemical wet etching.

Previous works have shown that the proton implanted LiNbO₃ waveguides are nearly “crystalline”, in which the excellent features of the bulks, such as EO, nonlinear optical, luminescent, and photorefractive properties, have been well preserved [22–25]. Since LiNbO₃ is also a thermo-optic (TO) material, the determination of the TO coefficient $\left(\frac{\text{d}n}{\text{d}T}\right)$ of the LiNbO₃ waveguides is not only intriguing for construction temperature-controlled integrated elements, but also helpful to examining the quadratic EO responses of the LiNbO₃ waveguides. In this work, we investigate the TO properties of the formed waveguides, which is to our knowledge the first time for ion implanted LiNbO₃ waveguides.

2. Experiments in details

The z-cut congruent LiNbO₃ wafers are cut to be with dimensions of 10(x) × 10(y) × 1.5(z) mm³ and optically polished. Figure 1 shows the schematic plots of the ridge waveguide fabrication process. A Cr film with thickness of 48 nm is deposited by sputtering on the surface of the negative z face (-z domains) [Fig. 1(a)]. The protons at energies of (475 + 500) keV and fluences of (3.6 + 6) × 10¹⁶ cm⁻² are implanted into LiNbO₃ wafer with this Cr-film mask, forming low-index optical barriers at the end of ions’ track (~3.5 μm beneath the sample surface) via the nuclear energy deposition [Fig. 1(b)]. After implantation, straight stripes with width of 10μm are defined by lithography technique with photomask and etched by a cerium sulfate solution. The remaining photoresist was removed by acetone [Fig. 1(c)]. Then the sample is immerged into an etchant solution [60 ml HF (40%), 39 ml HNO₃ (90%) and 5 ml ethanol (98%)] at 28°C. After 8 hours, the regions without the protection of Cr-stripes are removed by the acid owing to the selective etching of -z domains. The vertical depth of the etched regions is ~15 μm. Therefore the ridge waveguides are fabricated beneath the Cr stripes [Fig. 1(d)]. After the removal of Cr stripes, the waveguide sample is annealed at 400°C for 30 min in air to improve the guiding properties [Fig. 1(e)]. The inset photograph shows the microscope image of the ridge waveguide cross section. As it is indicated, the shape of the transverse cross section is approximately trapezoidal rather than rectangular, which is attributed to the etching behavior of the LiNbO₃ in such acid.

 

Fig. 1 Schematic plots of the ridge waveguide fabrication process. The inset shows the microscope image of the ridge waveguide cross section.

Download Full Size | PDF

To determine the refractive index changes in the waveguide regions, a planar waveguide sample is prepared by using proton implantation under same conditions. The dark-mode spectra of the guiding layer are measured by a prism coupler (Metricon 2010, USA). We reconstruct the refractive index profile of the one dimensional (1D) planar waveguide by using reflectivity calculation method (RCM), which has been proved to be particularly successful to ion implanted waveguides [26]. With this 1D profile, we construct the 2D index distribution at the cross section of the ridge waveguide by considering the waveguide shape. The light propagation in the waveguides has been simulated by a commercial software BeamProP© based on the finite difference beam propagation method (FD-BPM) [27]. With this numerical simulation we obtain the modal profiles of the guiding modes. The experimental characterization includes the measurements of the near-field optical intensity of the guided light and the propagation losses of the waveguides, which are both realized in an end-face coupling optical system. The wavelength of the light used in this work is ~633 nm (from a polarized He-Ne laser). The attenuation of the waveguides is measured by using Fabry-Perot method [28], according to the periodic power-oscillation curves of the coupled-out beam when we gradually heat the sample up to an increment of ~5°C.

The TO coefficients of the waveguides are measured and compared with those of the bulk by using prism coupling method [29]. This technique has been successfully applied to determine the TO coefficients of a SBN thin-film waveguide [29]. During the measurement, the sample is put on a heater which is controlled by a high-accuracy digital power controller. The sample temperature is monitored by a thermal detector adhered to the surface of the LiNbO₃ wafer. A heatconducting silicon wafer is used as an adhesive between heater, detector and the sample to achieve good thermal contact. In this work, the changes of refractive index are recorded when the sample is heated up to 120°C from room temperature.

3. Results and discussion

The ion implantation generates defects and damages inside the crystal, which is considered to be of great importance to the waveguide formation and modifications of the substrate. For light-ion-implanted LiNbO₃ crystals, the most-damaged regions are usually located at the end of ion range, which is mainly caused by the nuclear energy deposition of incident ions on the original lattices. Such damaged region is the so-called “optical barrier” with reduced refractive index, which confines the light propagation together with the surface cladding (air). Figure 2 shows the defect concentration (n _da) (calculated by the method in [30]) and the depth distribution of the primary displacement (n _dpa) (obtained by SRIM 2008 calculation [31]) in the as-implanted LiNbO₃ crystals induced by the proton beams. As one can see, the incident protons generate a damaged layer at depth of ~3.5 μm beneath the sample surface, i.e., in the barrier region. For this damaged layer the maximum n _da ≈0.11 at n _dpa ≈0.08 dpa. With such defect concentration it is expected that the etching rate in LiNbO₃ is still very low because the effect in the acid from the ion beam enhanced etching is almost negligible [32]. This is also confirmed by the microscope image of the ridge waveguide cross section [inset of Fig. 1], which does not show any side etching effect in the barrier regions.

 

Fig. 2 The defect concentration n _da (solid line) and relative displacement of the original atoms n _dpa (dashed line) for the as-implanted LiNbO₃ implanted by protons at energies of (475 + 500) keV and fluences of (3.6 + 6) × 10¹⁶ cm⁻².

Download Full Size | PDF

Figure 3(a) depicts the extraordinary refractive index (n _e) profile of the planar waveguide (i.e., the vertical index distribution of the ridge waveguide) after the thermal annealing treatment. By using this 1D profile, we have constructed the 2D index mapping [Fig. 3(b)] of the ridge waveguide cross section, taking the shape of the formed ridges into account. Figure 3(c) shows the calculated modal profile of the quazi-TM₀₀ mode (in 3D plot), which is in good agreement with the experimentally measured one (near-field intensity distribution of the output light) by the end-face coupling arrangement [Fig. 3(d)]. This suggests that our simulation is reasonable, which can be used for designing more complicated guiding devices.

 

Fig. 3 The refractive index profile of the 1D planar waveguide (a) and 2D ridge waveguide (b); the calculated modal profile (c) and the measured intensity distribution of the TM₀₀ mode in 3D plots.

Download Full Size | PDF

The propagation losses of the post-annealed ridge waveguides (i.e., at 400°C for 30 min) are determined to be as low as ~0.9 dB/cm at 633 nm. Further reduction of the attenuation values can be realized by optimizing the annealing conditions, or/and adjusting the dimension parameters of the ridge structures.

Figure 4 shows the temperature-dependent n _o and n _e alternations for the waveguides and bulk, respectively. The average values of the TO coefficient are calculated by the linear fit of the experimental data. The TO coefficients in the guiding regions related to n _o and n _e are $\frac{\text{d}{n}_o}{\text{d}T}$≈2.0 × 10⁻⁵ K⁻¹ and $\frac{\text{d}{n}_e}{\text{d}T}$≈6.4 × 10⁻⁵ K⁻¹, respectively, which are both approximately same to the bulk values ($\frac{\text{d}{n}_o}{\text{d}T}$≈2.1 × 10⁻⁵ K⁻¹ and $\frac{\text{d}{n}_e}{\text{d}T}$≈6.0 × 10⁻⁵ K⁻¹). It should be pointed out that the obtained TO coefficient values in the bulk are with a good agreement from the literature [33]. Through this comparison, we find that there is no obvious change of the TO coefficients in the proton implanted LiNbO₃ ridge waveguides, which suggests the TO features are well preserved in the guiding structures. This, in turn, implies the proton implanted LiNbO₃ waveguides are with bulk-quality for possible TO photonic applications.

 

Fig. 4 The measured temperature dependence of the a) n _o and b) n _e in the waveguide and bulk.

Download Full Size | PDF

It should be pointed out that the defect concentration plays an important role for both the ridge waveguide formation and the TO effects. The proton implantation does not create enough high defects in the whole implanted regions, because the in the waveguide, the electronic damage could be neglected (owing to the small values of electronic stopping power S_e) whilst the nuclear damage in the barriers is not enough high for enhanced etching or slicing in the acid. In addition, the TO properties have been well preserved in the waveguides also owing to the very slight modifications of the original lattices by the protons. We have also found that, for heavy ion implanted LiNbO₃ waveguides, the TO coefficients are considerably modified when S_e is above 2.2keV/nm, whilst keep the same with those of the bulks in cases of S_e<2.2keV/nm. In this sense, there may be a fine bridge between the formation of ridge waveguides and the maintenance of the TO effects of the bulks by using implantation of protons instead of other ions.

4. Summary

We presented a simple method to fabricate ridge waveguides in LiNbO₃ crystals at the surface with –z domains by combination of proton implantation and selective wet etching. The formed ridge waveguides with well-defined guided modes and relatively low propagation losses exhibited acceptable guiding properties. In addition, we found that the TO properties are well preserved in the waveguides with respect to the bulks, which suggests potential applications of the formed LiNbO₃ waveguides as integrated TO photonic elements.