1. Introduction

KTiOPO₄ (KTP) is widely used as a gain material in nonlinear frequency conversion schemes [1]. Moreover, KTP’s ability to be periodically structured allows one to take advantage of quasi-phase matching (QPM) and thereby to build highly efficient nonlinear frequency conversion devices [2]. It has been found that doping KTP with Rb-ions improves the poling properties of KTP and allows large aperture [3] and high aspect ratio crystals [4]. Even though there is a wide range of literature available covering the laser induced damage threshold (LIDT) of KTP under various conditions [5–11], none of the studies covers Rb-doped KTP (RKTP). Additionally, since most studies were undertaken with regards to birefringent phase matching and the LIDT of KTP is itself anisotropic [7], there is only little data available for the case of QPM where all interacting fields are polarized along the crystallographic z-axis. In order to receive reliable data for the design of highly-efficient, high-energy conversion schemes we made investigation of the LIDT for KTP and RKTP for the common pump wavelength of 1.064 µm as well as in the 2 µm region. This wavelength is suitable for sensing applications [12] as well as a pump source for parametric down-conversion in nonlinear materials such as ZGP or CdSe [13] required for applications in standoff detection [14]. Antireflection coatings for some materials have been shown to increase the LIDT considerably [15]. Motivated by this, we also investigated the influence of anti-reflection coatings on the LIDT at 1.064 µm.

The LIDT depends on sample preparation procedures and the excitation conditions. For consistency we used the same procedure for optical-grade polishing of all the sample surfaces. The excitation conditions were those dictated by the application of large aperture periodically poled KTP (PPKTP) crystals in a high-energy master-oscillator power amplifier (MOPA) chain operating around 2 µm and pumped by a 1.064 µm, single-longitudinal mode, diode-pumped master-oscillator power amplifier. For optimization of such MOPA chains the most relevant LIDT values are obtained in the so-called S-on-1 measurement procedure detailed in the appropriate ISO standard [16] and also utilized in other LIDT measurements [5–7]. In short, the S-on-1 procedure prescribes the application of S pulses at a given energy fluence and a nominal repetition rate of the laser on a single spot of the sample, followed by translation of the excitation spot by a distance ensuring that the previous excitation spot is not influencing the results at a new position. Owing to the experimental evidence that the LIDT was always lowest at the surface of our crystals, in this work we investigated only the surface LIDT.

2. Experimental setup

The setups for the LIDT measurements at 1 µm and 2 µm are depicted in Fig. 1 and Fig. 2 respectively. For the measurements at 1.064 µm, a diode pumped, Nd:YAG, seeded master oscillator power amplifier system was employed providing ~8 ns long, Gaussian temporally shaped, single longitudinal mode pulses at a repetition rate of 100 Hz. The beam was polarized along the z/c-axis of the crystal. The power delivered to the sample was controlled by a wave-plate/thin-film polarizer arrangement. A small percentage of the pump power was reflected in order to monitor the power incident on the sample. The beam was focused with a f = 200 mm spherical lens as well as a f = 150 mm cylindrical lens to account for astigmatism in the beam. The beam waist radius (1/e²) was determined with the 90-10 travelling knife-edge technique and was measured to be approximately 40 µm and 50 µm in the x and y directions, respectively. The M² value of the beam was measured to be approximately 3.2 for both transversal directions.

 

Fig. 1 Experimental setup for 1 µm LIDT measurements.

Download Full Size | PDF

 

Fig. 2 Experimental setup for 2 µm LIDT measurements.

Download Full Size | PDF

For the 2 µm measurements, the above described Nd:YAG laser was employed as a pump source for a degenerate OPO. The OPO consisted of a dielectric mirror as input coupler, a periodically poled RKTP crystal and a volume Bragg grating (VBG) as output coupler. The use of the VBG as an output coupler narrowed the output spectrum and improved the output power stability. Similar to the 1 µm experiment, the 2 µm beam was also polarized along the z/c-axis of the crystal. The RKTP crystal was AR-coated for both 1 µm and 2 µm and was periodically poled with a period of Λ = 38.86 µm, resulting in a signal/idler wavelength of 2.128 µm. The un-depleted pump was dumped and the 2 µm output was reflected towards two CaF₂ spherical lenses with focal lengths of f = 100 mm and f = 50 mm. The beam waist radius was measured to be approximately 40 µm and 50 µm in the x and y directions respectively. The M² value of the 2 µm output beam was measured to be approximately 8 and 4.2 for the x and y directions respectively.

The dependence of the LIDT on the number of pulses, S, was investigated in detail in [17]. It was found that the damage threshold at 1 µm in bulk KTP decreases until S reaches a value of about 200 and then stays approximately stable and independent of S. This is a, so called, quasi-fatigue effect. Therefore in the S-on-1 configuration, we choose an S value of 200 pulses. Additionally, this number is frequently used in literature and allows (to some extent) a comparison to literature values and is suitable to represent realistic applications and not just single shot experiments.

In both setups an electrically controlled shutter was used to allow an opening time of 2 seconds corresponding to 200 pulses. The scattered light was detected by an InGaAs photo-diode (Hamamatsu) in the case of 1.064 µm light and PbSe photoconductive detector (Hamamatsu) in the case of 2 µm radiation. In addition, the crystals were inspected under a microscope after the damage experiments.

The samples were polished to optical finish on both x-faces using industry-standard Logitech equipment and employing SF1 Syton chemo-mechanical polishing suspension. All samples were cleaned with isopropyl alcohol and flushed with dry air prior to LIDT testing. The samples were orientated such that the beam waist was positioned on and perpendicular to the front/entry crystal x-face. The samples where left at room temperature.

Laser induced surface damage of materials typically results in some debris spreading out from the initial damage site. The areas covered with the debris have a much lower LIDT value than the clean surface. Therefore it is necessary to choose an appropriate spacing between the damage sites. It was found that for fluences around 10 J/cm², a spacing of 300 µm was sufficient. The required spacing for higher fluences was also determined and the spacing was increased accordingly.

Finally, as a definition for the applied fluence, we make use of the peak fluence definition given by $F=\frac{E_p}{A}=2\frac{E_p}{\pi {\omega }_0^2}$, where $E_p$ is the pulse energy, ω₀ is the beam waist radius and A is the effective beam area.

3. Results and discussion

In some initial experiments we tried to cause bulk damage with a collimated pump beam as well as with a beam focused into the crystal. It was reported elsewhere [7] that a collimated beam with a wavelength of 1064 nm damages the bulk of KTP prior to the (uncoated) surface. However, we were not able to obtain isolated bulk damage in any of our samples. In the cases where the bulk of the crystal was damaged, the damage actually started at the back surface and propagated further into the bulk. This may be due to the presence of some SHG, with the 532 nm light lowering the LIDT of the back surface in comparison to that of the front surface [6]. Previous work has found that the 1-on-1 bulk damage threshold of KTP, with beams polarized parallel to the z-axis is around five times larger than the bulk damage threshold with beams polarized perpendicular to this axis, with 1-on-1 LIDT values > 45 J/cm² [11].

The LIDT is defined as that fluence value where $P\left(F\right)\approx 0$, where $P\left(F\right)$ is the damage probability at the fluence F. If one has n available sites for damage with k of those sites damaging then an estimate of the damage probability for a particular fluence is given by $P\left(F\right)=k/n$. Figure 3 shows the measured data sets for un-coated KTP and RKTP at 1 µm. Here we present three measurement sets for KTP and three for RKTP. As can be seen from these data sets, and using the definition of the LIDT, we measured a surface LIDT value of ~12 J/cm² for KTP and ~10 J/cm² for RKTP samples. The damage morphology is typical of that caused by nanosecond pulses, with crater formation and melting attributed to damage arising from thermal and acoustic shock wave effects [18].

 

Fig. 3 Multiple LIDT measurements for KTP and RKTP at 1 µm (left); typical surface damage with 200 pulses (right).

Download Full Size | PDF

Numerical modelling of laser damage mechanisms and behavior for different pulse-regimes have been well studied [19]. In general, laser induced surface damage by long pulses (~ns) is regarded to arise from thermal related damage to structural defects already present on the sample [20]. We made use of a theoretical model that was developed by H. Krol, et al [21] that builds on the idea that damage initiates due to the presence of nano-defects which could create electronic states below the bandgap energy. Using the model fitting given by Eq. (6) in [21], which assumes that the damage threshold for an ensemble of nano-defects is normally distributed, one can obtain fitting model parameters, such as the defect density = d, the mean damage threshold = T₀ and the threshold standard deviation = ΔT. For the particular data set in Fig. 4 showing the damage probability as a function of fluence, we obtain values: d = 4 × 10⁴ /cm², T₀ = 12 J/cm², ΔT = 1 J/cm².

 

Fig. 4 Damage probability model fit applied to an uncoated KTP (left) and AR-coated RKTP (right) at 1 µm.

Download Full Size | PDF

The error-bars in Fig. 4 are calculated using the method derived in [22]. Essentially, this method relies on the counting problem with n independent trials (sites tested for damage at a given fluence) with k trials resulting in damaged sites. The probability distribution of the possible outcomes in such experiment would be described by a binomial distribution. This distribution is normalized, and the area under the distribution is integrated to yield a confidence value. As in [22], we choose a confidence value of 68% corresponding to a 2σ width of the distribution.

LIDT measurements for AR coated RKTP samples are also shown in Fig. 4. A higher LIDT value was measured in comparison to the uncoated samples, with a surface LIDT of 18 J/cm². The fitting parameters for the AR coating were determined to be: d = 8 × 10⁴ /cm², T₀ = 18.5 J/cm² and ΔT ≈ 1 J/cm². According to the model, the AR coating had a higher defect density value compared to the un-coated samples. Therefore the higher LIDT value in this case may be attributed to the larger band gap of the AR coating material in comparison to the bandgap of RKTP.

Results for the 2 µm LIDT for KTP and RKTP are presented in Fig. 5. The data set contains two measurement sets for KTP and two for RKTP. As can be seen from the graph, we measured a surface LIDT of 10 J/cm² for both KTP and RKTP. The damage probabilities at different fluences are similar to those measured in the 1 µm case. The damage morphology was similar to the 1 µm result.

 

Fig. 5 Data set for KTP and RKTP at 2 µm (left); Model fit to a single 2 µm RKTP data set (right).

Download Full Size | PDF

We again applied the model fit to the measured data, shown here as an example is the curve for RKTP (Fig. 5). The model yields parameter values of: d = 3 × 10⁴ /cm², T₀ = 10.5 J/cm², ΔT = 1.1 J/cm². The defect density value is very similar to that found in the 1 µm case. This is to be expected since all samples were polished by the same equipment and in the same polishing run.

The experimental results show that the LIDT at 2 µm is essentially the same as at 1 µm. This might be somewhat surprising, considering that at the wavelength of 2 µm it would take absorption of 6 photons to cover the bandgap in KTP. This assumes the widely used optical damage physical picture which contains two steps: electron transfer from the valence band into a quasi-continuum of states close to the conduction band and subsequent heating by free-carrier absorption. The free carrier absorption coefficient is proportional to the free carrier concentration and the free carrier absorption cross-section, which, in turn, scales as ω⁻² for optical frequencies above the plasma frequency, which is a typically valid approximation [23]. Thus for generating the same heat power for the same optical fluence as at 1 µm one would need 4-times lower free-carrier concentration for 2 µm pumping. This difference of 4-times is way too small to give the same LIDT at 2 µm if 6-photon absorption indeed were the main mechanism generating free carriers. A more plausible scenario for the heat generation and damage both at 1 µm and 2 µm is then by linear absorption and non-radiative relaxation from surface defect states.

4. Conclusions

In conclusion we presented surface S-on-1 LIDT measurements on KTP and RKTP at 1 µm and 2 µm. We measured the LIDT in the QPM configuration, with all fields polarized along the z-axis of the crystal. We summarize the main results in Table 1.

Table 1. Surface LIDTs and fitting parameter values for KTP, RKTP and AR-coated RKTP.

View Table

For 1 µm, KTP was found to have a slightly higher damage threshold than RKTP with values of 12 J/cm² and 10 J/cm², respectively, however with regard to the overall accuracy of the LIDT experiment the values can still be considered to be identical. The damage morphology was typical of surfaces bombarded with ns pulses. AR coatings were applied to some of the RKTP samples by a commercial vendor and LIDT testing of these samples yielded a surface LIDT value of 18 J/cm². For 2 µm measurements, KTP and RKTP were found to have a surface LIDT of 10 J/cm². The similarities between the 1 µm and 2 µm results suggest that surface damage is more dependent on surface quality at this wavelength.

As mentioned, for this QPM configuration no isolated bulk damage was observed. The large surface LIDT values found for both crystals are beneficial for the construction of high-energy, frequency conversion schemes which utilize the QPM configuration.