1. Introduction

KTiOAsO₄ (KTA) is an isomorphic nonlinear optical material belonging to KTiOPO₄ (KTP) family of ferroelectric oxide crystals. Thanks to its extended transparency range in the infrared region together with its appealing nonlinear properties (similar to those of KTP) and availability from several vendors, KTA is a popular choice for use in birefringence phase-matched mid infrared optical parametric devices pumped at 1 µm.

It was previously shown that compared to KTP, KTA not only exhibits improved power handling properties at a wavelength of 1064 nm [1], but also substantially lower optical absorption at 532 nm [2]. This raises the question of whether KTA can be superior to KTP for applications in the green region in terms of a lower linear absorption in the visible range, a higher resistance to color center formation and as a result, a higher long-term stability of conversion efficiency [3–5]. Unfortunately, the use of KTA as a nonlinear medium for second harmonic generation (SHG) involving birefringent type II phase matching is limited by the short wavelength SHG cut-off around 1075 nm [6]. Nevertheless, the ferroelectric nature of KTA allows exploiting the advantages of quasi-phase matching (QPM), i.e the possibility to tailor any nonlinear interaction within the material transparency range utilizing the highest nonlinear coefficient available in the material. However, the potential of KTA for QPM applications in the visible region remains unexplored due to the difficulties in implementing short period QPM structures, associated with the high ionic conductivity of the material, which is one of the dominant factors influencing domain broadening during the periodic poling process [7]. Rosenman et al. overcome this problem by poling KTA crystals at a temperature below 170 K, where KTA ionic conductivity is significantly reduced [8]. However this method requires complex instrumentation, and substantially increased coercive field of the material at low temperatures limits the aperture of the QPM devices. On the other hand, we have recently demonstrated that conductivity-related polarization switching issues can be overcome at room temperature by poling with short electric field pulses, leading to the demonstration of the first optical parametric oscillator based on 1 mm thick PPKTA with a period of 39.5 μm [9].

In this paper we demonstrate that high quality bulk ferroelectric domain gratings with period of 8.49 μm for SHG of 1066 nm can be fabricated in KTA at room temperature taking advantage of periodic poling with short electric field pulses. The fabricated PPKTA crystals were evaluated in a low power continuous wave (CW) SHG setup, exhibiting normalized conversion efficiency of 1.2%/Wcm and effective nonlinear coefficient d_eff = 10.5 pm/V.

2. Experiments and results

For our experiments we have used commercial, flux-grown KTA crystals from two different vendors. All the samples were c-cut with the dimensions 10 × 5 × 1 mm³ along a, b and c crystallographic axes, respectively. We distinguish two different types of KTA crystals: low conductive KTA (LC-KTA) and high conductive KTA (HC-KTA). The conductivity of HC-KTA varied from 8.1 × 10⁻⁵ S/m to 1.1 × 10⁻⁴ S/m, while the conductivity of LC-KTA varied from 5 × 10⁻⁸ S/m to 8.2 × 10⁻⁸ S/m. Most of the as-purchased KTA crystals were found to be multi-domain, with “natural” domains of sizes varying from a few millimeters to tens of micrometers and extending over the whole crystal thickness. Indeed, it is well known that KTA crystals tend to grow in a multi-domain state [10], therefore, we first poled them with planar electrodes to obtain single-domain crystals. This was done by applying several 5 ms long 6 kV/mm square shape electric field pulses both for HC-KTA and for LC-KTA. After pre-poling, a photoresist grating with a period of 8.49 μm was deposited on c- face of each KTA crystal using standard photolithography. The pattern was then covered with an aluminum film. For all the crystals, the metal electrode duty cycle was chosen to be 15% to mitigate the effect of the domain broadening and aiming to obtain a 50% duty-cycle after periodic poling. The crystals were contacted to the external poling circuit using liquid KCl electrodes. The poling area was of dimensions 8 × 3.5 mm² along a- and b- directions, respectively.

In order to reverse the spontaneous polarization in high ionic-conductivity crystals, it is necessary to supply not only an electric field exceeding the coercive field of the material, but also a sufficient amount of charge in order to compensate the ionic current that screens the switching. Moreover, the shape and length of electric field pulse should be designed in such a way that it prevents excessive ion-migration in the crystal, which can lead to material breakdown, at the same time promoting large domain-velocity anisotropy in the in-plane and polar directions. We previously showed that short (millisecond range) electric field pulses are suitable for poling KTA at room temperature with good pattern fidelity for periods around 40 µm [9]. We also found that poling with electric field pulses of symmetric triangular shape mitigates the domain broadening problem for KTP isomorphs [11,12]. Furthermore, compared to square pulses, triangular pulses offer a larger tolerance window regarding the electric-field magnitude that can be utilized for periodic poling [12]. This becomes critical in the KTA case, since in crystals that were multidomain, the former opposite-polarity regions might have slightly different coercive fields compared to those that were originally single-domain. Here, our best poling results were achieved applying symmetric triangular electric field pulses with a duration between 2.5 ms to 5 ms. The pulse magnitude was adjusted depending on the individual crystal conductivity. HC-KTA samples were poled at room temperature applying 5 ms long electric field pulses of magnitude 4.6 kV/mm, while LC-KTA samples were poled applying 2.5 ms long electric field pulses of magnitude 5.4 kV/mm. The total charge transferred through the crystal, including the ionic and the switching current contributions, was 3.3 × 10⁻⁵ C in HC-KTA case, and 7.6 × 10⁻⁶ C in LC-KTA case. Assuming that the spontaneous polarization is the same in both materials, the actual current required for domain switching is about an order of magnitude higher in HC-KTA. Due to high ionic conductivity of KTA it is difficult to distinguish the switching current from the ionic counterpart, therefore we used an alternative poling monitoring method, based on the transverse electro-optic effect [13] together with the in situ second harmonic generation [14]. Figure 1(a) shows the ferroelectric domain structure, revealed by selective chemical etching, on the polar faces of a high conductive periodically poled KTA (HC-PPKTA) sample.

 

Fig. 1 Ferroelectric domain structure, revealed by selective chemical etching, on patterned face of HC-PPKTA (a) and LC-PPKTA (b) crystal. The insets show the domain structure on non-patterned faces of HC-PPKTA and LC-PPKTA crystals respectively.

Download Full Size | PDF

The duty cycle of the inverted domains is around 45% on the patterned face, and 51% on the non-patterned face. Owing to substantially lower ionic conductivity and hence reduced domain broadening, the duty cycle of the reversed domains in low conductive periodically poled KTA (LC-PPKTA) crystals was around 27% on the patterned face, and approximately 36% on the opposite face, as shown in Fig. 1(b). We expect that a slightly larger electrode duty-cycle would have resulted in a domain duty-cycle closer to 50% for LC-PPKTA crystals. Similar results were obtained for other HC-PPKTA and LC-PPKTA samples, therefore these results are representative.

We evaluated the performance of these QPM devices in a SHG setup, pumped by a CW Ti:Sapphire laser (Spectra Physics, model 3900S, pumped by 10W green Millenia Xs system). A waveplate-polarizer arrangement was used to control the polarization and power of the fundamental beam. The pump beam was launched along the crystal x-axis, focused into the crystal to a beam waist of 25 μm (1/e² intensity) by lens with 50 mm focal length, and polarized parallel to the crystal z axis. The crystals were uncoated and their temperature was kept at 21 °C by a Peltier element. The maximum QPM SHG efficiency was obtained at a fundamental wavelength of 1066 nm. Figure 2 shows the conversion efficiencies for HC-PPKTA and LC-PPKTA crystals for different pump powers.

 

Fig. 2 Second harmonic power (open symbols) and conversion efficiency (solid symbols) of HC-PPKTA (squares) and LC-PPKTA (circles) for different pump powers. The solid curves represent quadratic fits in SH power case, and linear fits in efficiency case.

Download Full Size | PDF

From the HC-PPKTA crystal we could obtain 1.15 mW of green light for a fundamental pump power of 347 mW. This crystal shows a normalized conversion efficiency of 1.2%/Wcm, and its effective nonlinear coefficient is d_eff = 10.5 pm/V, which is close to the calculated value 10.3 pm/V from d₃₃ = 16.2 pm/V [10]. Slightly lower values were obtained for the LC-PPKTA crystal (normalized conversion efficiency of 1.03%/Wcm, d_eff = 9.7 pm/V) due to its larger deviation from the ideal 50% ferroelectric-domain duty-cycle. The estimated normalized conversion efficiency and effective nonlinearity values in the ideal domain duty cycle case in this crystal are η_norm = 1.26%/Wcm and d_eff = 10.7 pm/V.

Next, we evaluated the homogeneity of the QPM grating over the crystal aperture by translating the crystal in the two directions perpendicular to the pump beam. The pump power was set to 348 mW. The crystal was scanned in y- and z- directions in steps of 200 μm, covering a total aperture of 3.6 mm by 1 mm, respectively. The normalized conversion efficiency distributions across the crystal aperture for HC-PPKTA and LC-PPKTA are shown in Fig. 3. Note that the drop in power at the edges in the nonpolar direction is due to the fact that a part of the pump beam travels outside the domain grating.

 

Fig. 3 Normalized conversion efficiency distribution across the crystal aperture in HC-PPKTA (a) and LC-PPKTA (b) crystals.

Download Full Size | PDF

The standard deviation across 3 × 1 mm² aperture (in y- and z- directions, representing the poled volume of 8 × 3 × 1 mm³), is 0.23%/Wcm in HC-PPKTA case, while it is 0.14%/Wcm in LC-PPKTA case. The better homogeneity of the QPM grating in LC-PPKTA crystal can be explained by reduced effect of domain broadening, which is a consequence of the relatively low ionic conductivity of the crystal. Additionally, absolute conductivity variation across the crystal is also orders of magnitude lower than in HC-KTA case, and plays less significant role in obtaining homogeneous QPM structure.

Finally, we measured the temperature acceptance bandwidth of our PPKTA crystals. The fundamental wavelength was set to 1066.5 nm, achieving maximum conversion efficiency at 30 °C. The crystal temperature was changed in steps of 0.2 °C between 20 °C to 40 °C and the SH power was recorded at each step. The measurement results for both crystals are shown in Fig. 4(a). The temperature acceptance bandwidth is ΔT_FWHM = 5.6 °C for the HC-PPKTA crystal and ΔT_FWHM = 5.7 °C for LC-PPKTA. These values are similar to those obtained for a high-quality periodically poled KTP of similar length designed for SHG of 1064 nm [15].

 

Fig. 4 Normalized second harmonic power as a function of crystal temperature (a), and temperature tuning of phase-matched wavelength (b) in HC-PPKTA and LC-PPKTA crystals.

Download Full Size | PDF

The phase-matched wavelength dependence on the crystal temperature was also evaluated. In this measurement the pump beam was kept parallel to the grating vector and the crystal temperature was varied in steps of 2 °C between 20 °C to 40 °C. The pump wavelength was tuned to give the highest SH output at each step and was recorded with a spectrum analyzer (Ando AQ-6315A). Figure 4(b) shows the phase-matched pump wavelength as a function of the crystal temperature, measured for HC-PPKTA and LC-PPKTA crystals. The dashed curve represents the theoretical prediction based on the Sellmeier equations derived by Fradkin-Kashi et al. [16], together with temperature correction, optimized for these equations, proposed by Emanuelli and Arie [17]. The large discrepancy between the measured and calculated values in this case may be due to the fact that these equations were optimized for mid-infrared spectral range. The solid curve is calculated using temperature-corrected Sellmeier equations derived by Kato and Umemura [18]. In this case the difference between experimental and calculated values can be explained by small differences in the raw KTA material, which had been already pointed out as a source for relatively large refractive index differences [10,19].

3. Conclusions

In conclusion, we have demonstrated high-quality periodic poling of 1 mm thick KTA crystals with period of 8.49 μm for second harmonic generation of 1066 nm wavelength. The PPKTA crystals were evaluated in a low power CW SHG setup and present an effective nonlinear coefficient as high as d_eff = 10.5 pm/V with good uniformity over the crystal aperture and a normalized conversion efficiency of 1.2%/Wcm. The temperature tuning curves for our samples show discrepancy not only with the existing Sellmeier equations but also among crystals from different vendors, indicating that small differences in material composition may have an impact in the KTA refractive index.