1. Introduction

Erbium (Er) doped optical materials are of great interest to both researchers and industries for the all-optical information systems, due to the intra-4f emissions of Er³⁺ ions at first telecommunication wavelength of ~1.5 µm which allow to implement active photonic device [1–3]. To this purpose, recent results have shown that Er doped lithium niobate (LiNbO₃) crystals are promising candidates for their proper combination of laser generation and excellent electrooptic and nonlinear properties of the substrate crystals [4-6]. The mostly used commercial LiNbO₃ are congruent crystals (usually known as CLN), i.e., the ratio of Li and Nb ions ([Li]/[Li+Nb]≈48.5%) in the crystals is less than the stoichiometric composition. This Li-deficient structure causes a large number of intrinsic defects, which degrades, to a certain extent, the optical performances of the crystals [7]. More recently, it has been proved that, by increasing the ratio of [Li]/[Li+Nb] to a nearly stoichiometric value (>49.5%), both the electrooptic and the nonlinear properties could be considerably improved in near stoichiometric LiNbO₃ (hereafter abbreviated to SLN), compared with their congruent partners [6–8]. Particularly, it has also been confirmed that the Er³⁺ doped SLN (or co-doped with MgO, namely Er:MgO:SLN) is a promising active gain medium with a superior laser capability in both visible (VIS) and near infrared (NIR) band range over the Er³⁺:CLN due to the reduction of the density of non-stoichiometric defects inside the crystal [9–11].

Besides the choice of the optical material, a second key-factor is the adoption of a guided-wave configuration. Actually, as one of the most basic components of modern photonic networks, optical channel waveguides can confine the light propagation in the two transverse dimensions to order of micrometers, resulting in many enhanced performances for various optical applications [12]. Channel waveguides in CLN wafers have been produced by several methods for active integrated devices, such as titanium [13] or zinc [14] ion indiffusion, proton exchange [15–18], ion implantation [19–22], laser writing [23,24], etc. Among these techniques, the implantation of both light (e.g. H or He) [20,23,26–28] and heavy ions (O or C) [20,21,28,29] (mainly depending on a physical mechanism) has been used to fabricate waveguides in large number of optical materials. The technique exhibits a wide capability of fabricating in versatile substrates, with good reproducibility and controllability, as well as with the possibility of easy combination with other techniques for more complex waveguide designs. For example, reconfigurable optical channel waveguides in LN have been fabricated by combination of O ion implantation and selective white light illumination [30]. It should be pointed out that the conventional chemical methods for CLN waveguide formation are proved to be much less effective for SLN crystals: the indiffusion coefficients of metal ions are much low in SLN because of the reduction of intrinsic defects [31] and the incorporation of protons into SLN replacing original Li ions results in a reduction of [Li]/[Li+Nb] ratio in the waveguides [6,15–18] which causes a degradation in the SLN original properties.

In this work, we report, to our knowledge for the first time, the optical channel waveguides formation in Er:MgO:SLN crystal by using selective low-dose oxygen ion implantation. The guiding properties at wavelength of 632.8 nm as well as the luminescence properties at 1.5 µm of the Er ions inside the waveguide are investigated.

2. Experiments and results

The z-cut Er:MgO:SLN sample (doped with 0.2 mol% Er³⁺ and 1.5 mol% MgO) we use is obtained from Crystalblue Co. Ltd., China, with [Li]/[Li+Nb] ratio of ~49.7%. It is optically polished and cut with size of 7(x)×10(y)×1.2(z) mm³ (z-direction pointing at the crystal c axis). By applying standard lithography technique, one x–y face of the sample is covered with specially designed photoresist stripe mask, which consists of a series of open stripes (with width of 10 µm, uncovered) with separation of 40 µm (covered by 5µm-thick photoresist stripe masks) between the adjacent channels. With this masking, the implantation of O⁺ ions at energy of 3 MeV and dose of 6×10¹⁴ cm^-2 is performed by a 2×1.7 MV tandem accelerator at Peking University. The implantation will induce refractive index changes in the un-protected surface regions (open stripes), whilst does not affect the photoresist-mask-covered regions. In this way, the channel waveguides can be formed in the implantation-exposed stripe regions: see Fig. 1 for the schematic plot of the channel waveguide formation process. Usually for low-dose O implanted optical waveguides, moderate annealing at 200-300°C for 30 min to several hours is suitable for optimization of the waveguide performance: it eliminates the points defects/color centers induced by the implantation but does not destroy the waveguide structures. After the implantation, the sample is annealed at 260°C for 1h in air to reduce the irradiation-induced color centers for an improvement of the waveguide quality. Moreover, the cross sections of the waveguide sample are imaged by a microscope with reflected polarized light (Olympus BX51M, Japan), by which the channel waveguide configuration is clearly observed (see inset of Fig. 1). The trapezoidal shape of the waveguide cross section is due to the wedged geometry of the photoresist masks at their edges. In addition, for comparison, about 1/4 of the sample surface is not covered by any masks, allowing planar waveguide formation with same conditions to the channel one.

 

Fig. 1. Schematic plot of the waveguide fabrication process. The inset shows the microscope image of the transverse cross section of the channel waveguide. The dashed lines denote the location of the channel waveguide.

Download Full Size | PDF

 

Fig. 2. Profile of extraordinary refractive index changes Δn _e induced by the ion implantation as function of the penetration depth of the incident ions inside the crystal (pointing z axis).

Download Full Size | PDF

The dark-mode spectroscopy of the planar waveguide in Er:MgO:SLN is measured by the well-known m-line technique at wavelength of 632.8 nm through a prism coupler (Model 2010, Metricon, USA). According to the dark-line spectrum, we reconstruct the profile of the extraordinary refractive index changes (Δn _e) of the planar waveguide [Fig. 2, solid line] by applying the reflectivity calculation method (RCM) of the planar waveguide [32]. As one can see, the O⁺ ion implantation creates a positive index well (Δn _w≈+0.016) in the near surface regions, by which the waveguide could confine the light propagation in a non-leaky way. Meanwhile, an optical barrier with negative index change (Δn _b≈-0.004) is built at the depth of ~2µm from the surface down to the substrate. Such profile is a typical “well+barrier” distribution, which is similar to those of many ion implanted waveguides [19-22,33,34]. In addition, the peak position of the barrier is in good agreement with the mean project range (ΔR _p≈2µm) of incident 3 MeV O⁺ ions inside Er:MgO:SLN, which means the waveguide dimension could be well determined theoretically. We also estimate the Δn _e profile of the planar waveguide at 1.5µm [Fig. 2, dashed line], via numerical fit based on the Sellmeier equation of the crystal [35].

 

Fig. 3. (a). The 2D n _e distribution of the channel waveguide at the cross section, (b). 3D plot of the calculated modal profile of the quasi-TM₀₀ mode by FD-BPM and (c). measured near-field intensity distribution of quasi-TM₀₀ mode at wavelength of 632.8nm.

Download Full Size | PDF

 

Fig. 4. (a). The 2D n _e distribution of the channel waveguide at the cross section and (b) 3D plot of the calculated modal profile of the quasi-TM₀₀ mode by FD-BPM at wavelength of 1.5µm.

Download Full Size | PDF

Figure 3(a) depicts the estimated the two dimensional (2D) refractive index distribution (at 632.8 nm) of the waveguide cross section, by considering the planar waveguide index map as well as the micro-topography of the channels. With this 2D index profile, we perform a numerical simulation of the light propagation in the channel waveguides through the commercial software BeamPROP [36], which is based on the well-known finite difference beam propagation method (FD-BPM). Figure 3(b) depicts the calculated modal intensity profile of the quasi-transverse magnetic (TM₀₀) mode of the channel waveguides at 632.8 nm. Experimentally, we use an end-face coupling arrangement (similar to Ref. 20, with a polarized He-Ne laser at 632.8 nm) to measure near field intensity distribution of the quasi-TM₀₀ mode [Fig. 3(c)]. By comparing Figs. 3(b) and 3(c), one can conclude that there is a fairly good agreement between the numerical results and the experimental data, which suggests that the estimate of 2D index distribution is good enough for channel waveguide design purpose. We also estimate the 2D n _e profile at 1.5 µm [Fig. 4(a)] in the similar way, and obtain the modal distribution of quasi-TM₀₀ mode [Fig. 4(b)] by FD-BPM. As it can be seen, the numerical simulation suggests that the channel waveguide is still guiding at 1.5 µm.

We use the Fabry-Perot resonance method [37] to measure the propagation loss of the channel waveguides by investigating the power-oscillation effects at the output facet when we gradually heat the sample (up to 5°C). The transmission attenuation is determined to be ~1.7 dB/cm at the wavelength of 632.8 nm.

 

Fig. 4. Confocal room-temperature luminescence emission spectra from (a) the bulk and (b) the channel waveguide volume, corresponding to the ⁴ I _13/2→⁴ I _15/2 transitions

Download Full Size | PDF

Finally, the 1.5-µm wavelength luminescence properties of Er³⁺ ions inside the channel waveguide are investigated by using a fiber-coupled confocal microscope in back-scattering configuration. An 800-nm wavelength fiber-coupled diode at power of 40 mW is used as the excitation source. The input beam is focused into the sample by a long working-distance achromatic infrared microscope objective (×100) with N.A. of 0.9, which promotes the excitation of Er³⁺ ions from their fundamental state (⁴ I _15/2) to the excited state (⁴ I _9/2). The subsequent emitted luminescence at ~1.5 µm (⁴ I _13/2→⁴ I _15/2) is collected by using the same microscope objective and analysed by an InGaAs detector connected to a lock-in amplifier and attached to a fiber-coupled monochromator (SPEX 500M). Figures 5(a) and (b) show the room-temperature micro-luminescence spectra of ⁴ I _13/2→⁴ I _15/2 transitions of Er³⁺ ions from the bulk and the channel waveguide, respectively. As one can see, the emission spectra of the bulk and the channel waveguide, which are both centered at ~1.52 µm, exhibit similar characteristics (in terms of peak positions, widths and relative intensity between sub-stark transitions), and therefore it appears to be reasonable that the main emission characteristics of Er³⁺ ions are well preserved in the waveguides.

3. Conclusions

Optical channel waveguides in Er:MgO:SLN crystal have been successful produced by using selective low-dose oxygen ion implantation with a photoresist stripe mask. The guiding properties are characterized at wavelength of 632.8 nm. Numerical simulation suggests the fabricated waveguides can be useful at telecommunication wavelength of ~1.5 µm. In addition, the luminescence emissions at 1.52 µm of Er³⁺ ions in the waveguides are not affected by the implantation process, which shows potential applications for the Er:MgO:SLN channel waveguides as active integrated photonic devices, e.g., Er doped waveguide amplifiers (EDWA), and waveguide lasers operating at ~1.5 µm.