1. Introduction

Potassium titanyl phosphate (KTiOPO₄ or KTP) crystals have been widely used in a number of nonlinear optical applications due to their excellent physical and optical properties, such as relatively large frequency doubling coefficient, high optical damage threshold, and excellent thermal stability [1]. Particularly, KTP-based wafers have been utilized for frequency conversion in many optical systems [2]. For example, commercially available phase-matched KTP crystals are used to manufacture green laser pointers. Periodically-poled KTP (PPKTP) wafers have intriguing applications in quantum optics for single-photon generation [3] and in terahertz generation through nonlinear wave mixing [4]. Compared to PPKTP, homogeneously poled KTP wafers are of much lower cost making them preferred candidates for frequency conversion.

Integrated optics offers optical waveguide-based platforms to implement compact, diffraction-free photonic systems, which enables various applications on chip level. Due to the strong confinement of light fields, the internal optical intensities of guided light could be strongly enhanced with respect to bulk materials. This feature is of significant advantage for lasing and nonlinear optical applications [5]. A few techniques have been developed to produce waveguides in KTP crystals, including ion exchange, ion implantation, and ultrafast laser writing, and diverse geometries of the guiding structures have been successfully fabricated [6–11]. In most cases, two-dimensionally (2D) confined waveguides (channel, ridge, fiber-cladding-like) are more desirable than one-dimensional (1D) planar structures because the 2D geometry possesses more compact modal fields as well as high flexibility for device production. Recently, diamond blade dicing has emerged as a powerful technique to microstructure the surface of optical materials with high precision and low roughness. Combined with planar 1D waveguide fabrication, for example, ion irradiation [12] or proton exchange [13], it could be utilized to produce high-quality, low-loss ridge waveguides in optical materials, such as Nd:YAG and LiNbO₃ [14–17]. The obtained air/waveguide interfaces in dielectric crystals were found to be of superior quality when compared with those produced by femtosecond laser ablation [18].

In this work, we apply the method of precise diamond blade dicing on a carbon ion irradiated KTP planar waveguide to obtain ridge waveguides for second harmonic generation (SHG). One of significant advantages of ion-irradiated KTP waveguides over ion exchanged ones is the good guidance at both TE and TM polarization [19], which allows second-harmonic generation using orthogonally polarized pump waves. In this work, guided-wave frequency doubling of green light was achieved via type II phase matching under the pump of both CW and pulsed laser of 1064 nm fundamental wavelength.

2. Experimental

The KTP crystal was cut into a wafer with dimension of 10 × 5 × 2 mm³, with the cutting direction (θ = 90° relative to z-axis, φ = 23.5° relative to x-axis) set to satisfy the type II phase matching condition (o₁₀₆₄ + e₁₀₆₄ → e₅₃₂). Its two parallel end-faces (10 × 2 mm²) and one surface (10 × 5 mm²) were polished to optical quality. Figures 1(a) and 1(b) show the schematic plots of the ridge waveguide fabrication.

 

Fig. 1 The schematic diagrams of ridge waveguide fabrication processes: (a) carbon ion irradiation and (b) precise diamond blade dicing; (c) the microscopic photograph from cross section and (d) the SEM image and (e) enlarged view (dashed line enclosed region) of ridge waveguides.

Download Full Size | PDF

First, in order to produce a 1D waveguide layer, the sample was irradiated by 15 MeV C⁵⁺ ions at the fluence of 3 × 10¹⁴ ions/cm², which were accelerated through a 3 MV tandem accelerator at Helmholtz-Zentrum Dresden-Rossendorf. During this process, the incident ion beam was inclined by 7° off the surface normal to minimize channeling effects, and the ion beam current density remained at a low level (<10 nAcm⁻²) to avoid charging and heating of the sample. Afterwards, we applied precise diamond blade dicing (DISCO Corp [20].) to fabricate several air slits on the planar waveguide surface. A resin bond blade was used where the diamonds are bonded by a thermosetting resin. Grit size, bonding strength of the resin and the concentration of diamonds were optimized regarding chipping and wear-out. The blade had a diameter of 51 mm and a thickness of 100 µm. Rotation speed and moving speed were set to 20,000 rpm and 0,1 mm/s respectively. With this processing, the planar waveguide layer was microstructured with parallel air grooves in the transverse direction, forming a number of ridge waveguides. The widths of ridges (i.e., the intervals between two air grooves) were set to be 20, 25 and 15 μm, respectively (see Fig. 1(c) for the microscope images of the end-faces of the ridges). In order to evaluate the dicing quality of the sidewalls of the ridges, the sample was imaged by a SEM [see Figs. 1(d) and 1(e)]. As one can see, the sidewalls produced by diamond blade show only small chipping (typical size below 100 nm) and a roughness of ~2 nm (rms) measured by white-light interferometry, which was significantly superior to that of a femtosecond laser ablated dielectric crystals (roughness ~1-2 μm) [21,22]. This feature enables diamond blade dicing advantageous over other techniques for precise microstructuring of dielectric crystals. In addition, the waveguide planar layer (refractive index changed layer) is measured to be as thick as ~11 μm. The diced grooves were deeper than the planar waveguide, up to a depth of ~30 μm measured from the top of the ridges.

The investigation of the guiding properties were carried out based on a typical end-face coupling arrangement with two linearly polarized lasers at 1064 nm and 532 nm, respectively. The frequency doubling within the waveguides was realized in both cw and pulsed regimes. In the latter case, a Q-switched 1064 nm laser (with pulse duration of ~11 ns, pulse energy of ~80 μJ, frequency of ~5 kHz, and average power of ~480 mW) acted as the fundamental light source. The launched light beam at 1064 nm was focused by a microscope objective (25 × N.A. = 0.4) and coupled into the ridge waveguide. The output light from the waveguide was collected by a second microscope objective. At the end of the light path, a CCD camera, a power meter or a spectrometer could be selected as the detector for different purposes, i.e. imaging guiding modes, measuring light powers, and recording optical spectra. A mirror with high reflectivity at 1064 nm and high transmissivity at 532 nm was positioned between the detector and the out-coupling objective to obtain the transmission of green light only. In order to investigate the microstructural modification of the ion irradiation, we applied a confocal microscope/spectrometer (Horiba/Jobin Yvon HR800) to measure the micro-Raman spectra from the waveguide and bulk, respectively, with a CW laser at 473 nm as the excitation source.

3. Results and discussion

Figure 2 depicts the measured guided modes of KTP ridge waveguides, for both the 1064 nm and the 532 nm light. One may conclude that all excited intensity distributions are dominated by the fundamental mode. For the pump wavelength this assumption is further supported by calculating the overlap integral of these experimental modes with numerical ones obtained from the assumed refractive index profile of the waveguides [see also Fig. 3(e)], which gives an overlap of >90% for all three investigated ridge widths. Furthermore, the confinements of light propagation in these ridge waveguides are fairly good, for both the fundamental and the second harmonic waves.

 

Fig. 2 Distributions of measured TM modes in KTP ridge waveguides for 1064 nm (top row) and 532 nm (bottom row) and different ridge widths (left: 15 μm, middle: 20 μm, right: 25 μm).

Download Full Size | PDF

 

Fig. 3 (a) Stopping power S_e and S_n and induced (negative) refractive index change vs. ion pene-tration depth; (b) micro-Raman spectra obtained from substrate and waveguide regions; (c) recon-structed 2D refractive index profile; (e) calculated fundamental mode (25 μm ridge) at 1064 nm.

Download Full Size | PDF

The energy deposition of the irradiated ions into the KTP crystal lattice plays a crucial role on the waveguide formation. We used the SRIM2013 code (Stopping and Range of Ions in Matter 2013) [23] to calculate the electronic (S_e) as well as the nuclear (S_n) stopping powers of the 15 MeV C⁵⁺ ions into KTP, as shown in Fig. 3(a). The value of S_e varies smoothly from the surface with a peak at ~7.5 μm depth, whilst S_n remains negligible down to a depth of ~8 μm, and then rises rapidly to a maxima at ~10.5 μm. When the depth is larger than ~11 μm, both S_n and S_e become zero, which is consistent with the thickness of the waveguide layer measured under the microscope. According to our previous results, the C⁵⁺ ion irradiation in KTP forms 1D waveguides mainly due to the nuclear damage effect (related to S_n), whereas the electronic damage (related to S_e) induces a low-extent lattice distortion in the waveguide region [19]. The confocal micro-Raman spectroscopy has been proved to be an efficient tool to study lattice damages in small regions of waveguides [24]. Figure 3(b) depicts the comparison of the micro-Raman spectra obtained from the substrate and the waveguide region. The main peak Raman scattering intensities are positioned at frequency shifts of 271 cm⁻¹ and 701 cm⁻¹, respectively, for both waveguide and bulk, whilst the emission intensity in the waveguide is 20% lower than that in the bulk. This indicates that the irradiation has induced lattice damage and disorder in the waveguide region to some extent. Nevertheless, since the two spectra remain highly similar and the main peak positions were not clearly shifted after the irradiation, we believe that the nonlinear optical properties of the KTP crystal are preserved to an acceptable extent in the waveguide region for possible frequency doubling.

Figures 3(a) and 3(c) show the reconstructed 1D and 2D refractive index profiles for the planar and ridge waveguides in KTP, respectively. As demonstrated in Fig. 3(a), the index profile is of typical “optical barrier” type with refractive index, which is well-known for ion-implanted waveguides [12]. The maximum contrast of the (lowered) refractive index (~0.0046) is located at a depth of 10.5 μm, which is in agreement with the main peak in the S_n curve. This points to an only partial amorphisation of the implanted barrier, when compared to heavily He⁺ implanted KTP samples where a maximum index change of ~0.14 was observed [25]. With the 2D index distribution, we calculate the modal profile with the software Rsoft Beam PROP 8.0 [26], which is based on the finite difference beam propagation method (FD-BPM) [27]. The calculated modal profile (TM₀₀) of the 25 μm-wide ridge waveguide at 1064 nm is shown in Fig. 3(e), being in reasonable agreement with the measured intensity distribution (Fig. 2).

The propagation loss has been investigated using teransmission measurements for different ridge waveguides, and loss coefficients at the wavelength of 1064 nm of 1.7, 1.3, and 1.0 dB/cm are found for the 15, 20, and 25 μm-wide ridge waveguides, respectively. For these measurements we took into account Fresnel reflections and in-coupling efficiencies of 52-61% from single-mode fiber to the ridges calculated from the respective overlap integrals.

It is found that, as the width of the ridge increases, the loss of waveguides decreases; similar width-dependent properties were reported before in laser-ablated ridge waveguides [28]. These propagation loss coefficients are relatively low for ion-irradiated waveguides. On the contrary, for laser-ablated waveguides relatively high losses (>5 dB/cm) were found. Such high values may be due to the larger roughness of the ridge sidewalls caused by laser ablation.

Figures 4(a) and 4(b) depict the obtained results for SHG in the KTP ridge waveguides. The polarization of the 1064 nm light, controlled by a half-wave plate, has equal components along TE and TM directions (i.e., polarization under 45° with respect to the z-axis) to enable the most efficient orientation for KTP which is type II phase matching condition. Figures 4(a) and 4(b) present the generated SHG power as a function of the launched fundamental power from the 25 μm-wide KTP ridge waveguide under cw and pulsed laser pump, respectively. The fitted curves apparently demonstrate the nonlinear relationship between the generated second harmonics and the fundamental light through the waveguide. For the cw pump case, the maximum SHG output power is measured to be ~1.08 mW as the fundamental light is ~618.2 mW, leading to a conversion efficiency of ~1.12%W⁻¹. For the pulsed laser situation, the maximum peak power of the second harmonic reaches ~110.9 W, as the 1064 nm pulses have a peak power of ~897.5 W, resulting in a conversion efficiency of ~12.4%. As for other ridges, the data of the maximum SHG output powers and the maximum conversion efficiencies are listed in Table 1, for both cw and pulsed cases. As a trend we observe that, as the ridge width increases, the corresponding maximum conversion efficiency becomes higher.

 

Fig. 4 Second harmonic powers from KTP ridge waveguide as functions of launched powers of 1064 nm light at (a) cw and (b) pulsed laser pump.

Download Full Size | PDF

Table 1. The maximum output SH powers (P_max), the corresponding conversion efficiencies (η_max), and the propagation losses (α) from different waveguides.

View Table

One important factor that may still limit the obtained SHG efficiency is the additional phase mismatch induced by the waveguide dispersion. This is a result of the barrier-type index profile where both effective indices of guided modes are lowered with respect to the bulk values. However, the index decrease for the pump wavelength is slightly larger as the one for the generated green light, which causes the phase matching condition to shift to slightly higher wavelength. With the pump wavelength fixed to 1064nm this finally results in a lowered efficiency for the SHG process.

It is reasonable to mention that the propagation loss of the waveguides (also shown in Table 1) is another crucial factor on SHG efficiency. In this experiment, the 25 μm-wide KTP ridge possesses the best guiding property as well as highest SHG performance. Compared with the 17 MeV O⁵⁺ ion-irradiated KTP planar waveguide reported by Cheng et al. [19], the maximum conversion efficiency for this work is considerably higher (12.4% vs. 11.5%). In addition, the SHG efficiency of laser-ablated KTP ridges with similar geometries was ~1.25%, which is significantly lower than the diamond diced ridge waveguide. This comparison shows the advantage of diamond blade over laser ablation on processing of dielectric crystals, which reflects on the lower propagation losses in the diced ridge waveguides.

4. Summary

We have fabricated low-loss ridge waveguides in KTP nonlinear optical crystal by carbon ion irradiation combined with precise diamond blade dicing. The calculated modal distribution has reasonable agreement with the measured one. The guided-wave SHG of green light based on type II phase matching was realized in the diced ridges, reaching maximum SH powers of 1.08 mW and 110.9 W, and the corresponding conversion efficiencies of ~1.12% W⁻¹ and ~12.4%, for cw and pulsed laser pump, respectively. This work shows the advantage of diamond blade dicing as a powerful technique for fabrication of low-loss waveguides and microstructuring of dielectric crystals to achieve various applications.