Nanosecond-laser-induced damage in potassiumtitanyl phosphate: pure 532 nm pumpingand frequency conversion situations

Frank R. Wagner,1,* Anne Hildenbrand,1,2 Jean-Yves Natoli,1 and Mireille Commandré1

1Institut Fresnel, CNRS, Aix-Marseille Université, Ecole Centrale Marseille,Campus de St Jérôme, 13013 Marseille, France

2Institut Franco-Allemand de Recherches de Saint-Louis (ISL), 5 rue du Général Cassagnou,BP 70034, 68301 Saint-Louis Cedex, France

*Corresponding author: frank.wagner@fresnel.fr

Received 9 May 2011; accepted 23 June 2011;posted 1 July 2011 (Doc. ID 147133); published 28 July 2011

Nanosecond-laser-induced damage measurements in the bulk of KTiOPO4 (KTP) crystals are reportedusing incident 532nm light or using incident 1064nm light, which pumps more or less efficient secondharmonic generation. No damage threshold fatigue effect is observed with pure 532nm irradiation. Thedamage threshold of Z-polarized light is higher than the one for X- or Y-polarized light. During frequencydoubling, the damage threshold was found to be lower than for pure 1064 or 532nm irradiation. Moredata to quantify the cooperative damage mechanism were generated by performing fluence ramp experi-ments with varying conditions and monitoring the conversion efficiency. All damage thresholds plottedagainst the conversion efficiency align close to a characteristic curve. © 2011 Optical Society of AmericaOCIS codes: 140.3330, 160.4330, 230.4320.

1. Introduction

Potassium titanyl phosphate, KTiOPO4 (KTP), is oneof the most used nonlinear crystals for laser appli-cations [1,2]. The main advantages of using KTPare (i) high effective nonlinear coefficient, (ii) largeangular and temperature bandwidth [3,4], and(iii) moderate manufacturing cost. The single pulseinfrared (IR) nanosecond laser damage thresholdis, however, eight times lower than the one forLiB3O5 (LBO) [4–6], and color centers (gray tracks)are created during frequency doubling of 1064nmlight ([7] and references therein). The gray-track phe-nomenon was the subject of an extensive researcheffort ([7,8] and references therein) and more or lessgray-track-resistant KTP crystals are available to-day from different providers. Recent research onKTP is mostly linked to the possibility of creating

periodically poled crystals with even higher non-linear coefficients [9], opening up new applications[10,11]. Still, the main application of this materialremains frequency doubling of Q-switched Nd:YAGlasers from 1064 to 532nm.

Nanosecond-laser-induced damage to KTP crystalsat 1064nm and its dependence on the polarizationand propagation directions has been extensively stu-died recently [4,5]. Strong polarization anisotropy ofthe laser damage threshold has been observed withan enhanced damage threshold for Z-polarized light.This effect has been attributed to a color-center-based laser damage mechanism where the lifetimeof the color centers depends on the polarization ofthe laser light [5]. The damage threshold anisotropyis thus a consequence of the highly anisotropic ionicconductivity [12].

In this article, we present detailed nanosecondlaser damage measurements in the bulk of KTP crys-tals using two types of irradiation: (i) pure 532nmirradiation along the crystalline axes and (ii) 1064nm

0003-6935/11/224509-07$15.00/0© 2011 Optical Society of America

1 August 2011 / Vol. 50, No. 22 / APPLIED OPTICS 4509

irradiation with type II phase matching for secondharmonic generation (SHG). These data will allowthe further development of the laser damage me-chanism model in KTP. Here, as well as in [4,5], la-ser-induced damage is understood as catastrophicdamage initiated by a laser-induced plasma andresulting in microcracks or microvoids.

2. Experimental Setups

The used setups are similar to the basic one de-scribed in [5] and a schematic representation is givenin Fig. 1. We performed two types of measurements:S-on-1 measurements [13], where the laser damageprobability using S pulses per site is reported as afunction of the pump fluence, and R-on-1-type mea-surements, where each test site is damaged by a flu-ence ramp. The R-on-1-type measurements provideapproximate laser damage thresholds if no laserconditioning effect is present and the fatigue effectcan be neglected.

S-on-1 measurements have been carried out using1064 or 532nm pump irradiation. In order to changethe pump wavelength, the setup has been adaptedlike in [6]: A type II KTPSHG crystal and an IR block-ing filter (BF) have been introduced between the lasersource and the attenuator that is built up by the half-wave plate (λ=2) and the polarizer (pol.). The lasersource is a Q-switched Nd:YAG laser with an FWHMpulse duration of 6ns, and the pump light has beenfocused in the center of the uncoated sample.

The S-on-1 damage probability curves wererecorded using 10Hz pulse repetition rate and typi-cally 20 sites per fluence. For these curves, damagedetection was performed by automatically processingimages of backscattered light that was issued bythe fibered lamp (LA) and passed the BF for bothlaser wavelengths. The damage probabilities are pre-sented with 68% confidence error bars [14]. The 1=e2

beam diameter for all damage probability curves isclose to 76 μm (75 μm in the IR and 77 μm at 532nm).

During the R-on-1 measurements, the conversionefficiency was monitored due to the dichroic mirror

(DM) that reflected the generated 532nm light ontophotodetector number 2 (PD2). The internal conver-sion efficiencies are then calculated, taking into ac-count the reflection losses at the crystal interfaces.Laser damage was in this case detected manually byobserving scattered green light on the screen arounddetector 2. For the R-on-1 measurements, differentfocusing lenses were used, resulting in 1=e2 focusdiameters between 23 and 155 μm.

The spatial beam profile in the focal plane wasalways close to a Gaussian beam and the focusingconditions were checked for the absence of aberra-tions that may disturb laser damage measurementsin birefringent crystals [14]. The peak fluence ofeach pulse has been calculated from pulse energymeasurements carried out during the damage tests(using PD1) as well as the transmission of the neu-tral gray filter (NG) and the beam characterizationthat has been performed before the damage test. Thefluence error is estimated to be approximately �10%.

The KTP samples were uncoated flux-grown stan-dard KTP crystals provided by Cristal Laser S.A.,France. We should also mention that measurementsin RbTiOPO4 (RTP), which is very similar to KTP, re-cently showed that the laser damage behavior underthe conditions used for this work does not depend onionic conductivity and absorption for good qualitycrystals [15]. All laser damage discussed in this pa-per appeared in the bulk of the crystals even thoughthe laser beams were practically parallel comparedto the sample thickness of typically 10mm.

3. Pure 532 nm Irradiation

A. Multipulse Fatigue Effect

Figure 2 shows damage probability measure-ments (symbols) recorded in a Y-cut sample usingX-polarized 532nm light and up to 10,000 pulsesper site. The denominations of the principal axesof the orthorhombic crystals are X, Y , and Z, sorted

Fig. 1. (Color online) Schematic of the setup pointing out thedifferent configurations. The setup can be pumped directly withthe 1064nm light or with the generated 532nm light. Only onepump wavelength was used at a time. For S-on-1 measurements,automatic damage detection was used (LA, BF, and CCD). ForR-on-1 type measurements, a DM was added to reflect the gener-ated 532nm light onto PD2 and laser damage was detected by theappearance of scattered green light on the screen.

Fig. 2. (Color online) S-on-1 damage probability curves for Y-cutKTP and X-polarized 532nm light. The measured laser damagethreshold is independent of the number of pulses per site S. Thesolid line is a guide to the eye. The probability error bars delimitintervals corresponding to a confidence of 68%. The fluence errorbar corresponds to a relative uncertainty of �10%.

4510 APPLIED OPTICS / Vol. 50, No. 22 / 1 August 2011

in order of increasing refractive index. As the samplewas Y cut, the light propagation direction was alongthe Y axis of the crystal. The line plot in the graph isa guide to the eye.

Contrary to the observations during 1064nm test-ing [5], the laser damage threshold does not exhibita significant fatigue effect for 532nm irradiation.The S-on-1 laser damage curves for more than tenpulses per site all superimpose on each other, andthe transition zone from zero damage probabilityto 100% damage probability is smaller than the esti-mated fluence error. The deviation of the S-on-1curves for less than ten pulses per site from themajority of curves is probably caused by intensityfluctuations appearing in pulses of identical fluence.

Because of the sharpness of the transition zone,we may use its center as a damage threshold valuefor 532nm measurements. For convenience, we willuse the expression damage threshold to discuss theobserved effects, even if recent advances in laserdamage metrology suggest that one should be verycareful about stating a threshold value for a givencomponent [16]. In fact, in our measurement condi-tions the apparent damage threshold depends on theprobed sample volume. However, as all S-on-1damage curves that are shown in this article areacquired in very similar conditions, we may staywith this simplified view when talking about damagethresholds.

The effect of vanishing fatigue in the damagethreshold is similar to what has been reported onLBO, where the fatigue effect is absent for 355nmradiation [6]. Because of the smaller bandgap, thesame result is already obtained with 532nm irradia-tion in KTP.

B. Polarization Anisotropy

Figure 3(a) shows the 200-on-1 laser damage prob-ability curves recorded in a Y-cut sample usingX- and Z-polarized 532nm light. Figure 3(b) showssimilar results for an X-cut crystal.

For all these conditions, the phase mismatchΔk for 1064 and 532nm radiation is at least11 ; 800 rad=m. The efficiency of frequency conver-sion is thus very low and one can consider thesedamage curves to correspond to damage inducedby pure 532nm radiation. In the absence of fre-quency conversion, we observe a clear polarization-dependent anisotropy of the laser damage threshold,similar to but weaker than what has already beenreported at 1064nm [5]. At 532nm, the damagethreshold of Z-polarized light is approximately 1.6times higher than for X- or Y-polarized light.

The model that has been proposed for the similareffect at 1064nm [5] also holds for the 532nm situa-tions: the presence of a Z-polarized electric fieldenhances the hopping rate of the potassium ions inthe crystal [17], thus decreasing the average lifetimeof the color centers that are stable close to potassiumvacancies. The decreased color center lifetime

inhibits localized energy deposition in the materialand thus increases the laser damage threshold.

We thus suppose that the color centers responsiblefor the gray tracking effect are the laser damageprecursors at both wavelengths. Light-induced colorcenters have been observed during green-induced IRabsorption (GRIIRA) measurements before [9]. Dur-ing 1064nm pumping in situations with high phasemismatch, 532nm photons are generated locally, butthey interfere destructively with the 532nm photonsgenerated at a nearby location. Both wavelengths532 and 1064nm are thus able to generate color cen-ters that act as laser damage precursors.

4. Frequency Conversion Situations

Figure 4 shows 200-on-1 damage probability curvescomparing two situations with frequency conversionto the monowavelength situations. The conditionshave been chosen avoiding Z-polarized light.

The 1064nm situation without conversion hasbeen recorded in an SHG-cut crystal using a polari-zation in the X-Y plane. In this case, the conversionmismatch is huge (1 ; 66 0; 000 rad=m) and the da-mage threshold is approximately 11 J=cm2.

The 532nm situation without conversion has beenrecorded in a Y-cut crystal using X polarization. Inthis case, the conversion mismatch is 62; 300 rad=mand the damage threshold is approximately 7J=cm2.

Fig. 3. (Color online) 200-on-1 laser damage curves at 532nm fordifferent polarizations. Frequency conversion effects are mini-mized by using crystals that are cut along the principal axes.The arrows indicate the approximate laser damage thresholds.

1 August 2011 / Vol. 50, No. 22 / APPLIED OPTICS 4511

The two situations with frequency conversionare symmetric in the sense that the curve“1064nm with conversion” corresponds to a type IISHG, and the curve “532nm with conversion” corre-sponds to an inverse type II SHG (degenerate opticalparametrical generation pumped by 532nm). Bothcurves have been measured in SHG-cut crystalsusing 45° polarization for the IR pump beam andX-Y plane-polarization for the 532nm pump beam.The efficiency of the type II SHG was approximately21%. The efficiency of the inverse process has notbeen measured, but considering that pump beamsof approximately the same waist diameter have beenused, we may assume that the efficiency of the in-verse process is between 20% and 30%. The damagethresholds of both situations are practically identicalat about 4:5J=cm2 of incident radiation.

These results show that the laser damage mech-anism during frequency doubling in KTP is highlycooperative between the two wavelengths: consider-ing a 20% conversion efficiency for the type II SHG,the laser damage threshold is reached by either11 J=cm2 of pure 1064nm light, or by 7 J=cm2 of pure532nm light, or by a combination of only 3:6 J=cm2

of 1064nm light and approximately 1:8 J=cm2 of532nm light. In this last situation, the fluence valuefor the generated 532nm light is roughly two timeshigher than the value obtained simply by multiply-ing the damage threshold (incident pump fluence)with the conversion efficiency. This is the case, be-cause the frequency doubled light has a smallerwaist than the pump light if a Gaussian pump beamis used.

A similar observation of cooperative damage hasbeen reported in 2003 [18]. The authors reported thatthe laser damage threshold of KTP under combined1064 and 532nm light is much lower than undermonowavelength irradiation. However, they couldnot quantify the 532nm damage threshold with themicrosecond pulses and the relatively large beamdiameter that they used. Another hint at cooperative

damage initiation in a material very similar to KTPhas been reported in [19], where the damage thresh-old of Y-cut RTP Pockels cells was found to be re-duced, probably by unwanted type II SHG.

The model, which was put forward by Favre et al.,predicts the generation of color centers by 532nmlight followed by the absorption of IR light, whichleads to damage [18]. This model is very reasonable,as the effect of GRIIRA has been measured in KTP,for example by Wang et al. [9]. Wang et al. also foundevidence for light-induced color centers by studyingthe annealing dynamics.

A more careful qualitative formulation of the mod-el should take into account that not only mixturesdamage the crystal, but also IR light alone (at higherfluence), as well as 532nm light alone. One shouldthus say that the generation of color centers is causedpreferentially by 532nm light and that the finalstep, leading to damage, is predominantly causedby absorption of the IR light.

Further experimental exploration of the coopera-tive nature of the laser damage mechanism hasbeen achieved by using misaligned frequency dou-bling and different beam diameters. Using a type IISHG-cut crystal with a polarization orientation dif-ferent from the optimum 45° orientation leads to aslow but monotonous increase in the frequency-doubled light. In order to avoid the special case ofZ-polarized light, the polarization orientation hasbeen changed from 45° to 90° with respect to theZ axis.

The experiment has been conducted similarly toan R-on-1 experiment, i. e., we started with a smallpumping fluence and increased it gradually. For eachfluence value, the conversion efficiency has beenmeasured up to the appearance of laser damage thathas been detected by scattering of the generatedgreen light. Each fluence step thus corresponds toseveral laser pulses. Figure 5 shows for two situa-tions how the conversion efficiency increased withincident fluence until the damaging fluence wasreached.

Fig. 4. (Color online) 200-on-1 laser damage curves in a type IISHG-cut KTP crystal. The graph compares the situations withoutfrequency conversion (triangles) to two situations with frequencyconversion (squares). The arrows indicate the approximate da-mage thresholds. The damage thresholds for the two situationswith frequency conversion are practically identical.

Fig. 5. (Color online) Conversion efficiency as a function offluence for two different polarization directions. Laser damageoccurred at the last data point of both curves. The fluence value ofthe point before was used as an approximation of the laser damagethreshold.

4512 APPLIED OPTICS / Vol. 50, No. 22 / 1 August 2011

All R-on-1-type measurements were pumped by1064nm light and were performed in SHG-cut crys-tals. For a particular set of measurement conditions,we averaged the values obtained on 4–5 sites in orderto estimate the damage threshold.

The threshold fluences obtained from the R-on-1-type data points (circles and triangles in Fig. 6)are thus not as precise as the damage thresholds ob-tained from S-on-1 damage probability curves (starsin Fig. 6). However, comparing the two types of datapoints in Fig. 6, we see that the approximate R-on-1-type measurements are quite close to the values ob-tained from the 200-on-1 damage probability curvesshown in Fig. 4.

Using given focusing conditions, the R-on-1 mea-surements have been repeated for several IR polar-izations as described above, but we also influencedthe focal volume to obtain varying conversion effi-ciencies. The focal volume has been influenced byusing different focus diameters (24, 75, and 155 μm)and two different sample thicknesses (5 and 10mm).

Figure 6 represents the obtained approximatedamage thresholds as a function of the pulse energyratio r of green light to incident light. This energyratio corresponds to the conversion efficiency forSHG situations.

Regardless of the fact that the data represent alarge number of measurement conditions, all SHGpoints (r < 60%) align closely to a well-defined curve.The curve shows a steep decrease in the damagethreshold for small conversion efficiencies and stabi-lizes for r ¼ 40–50% conversion efficiency at a da-mage threshold much lower that the one for pure IRlight or pure 532nm light. The scattering around thiscurve seems most important for practically pure IRirradiation. This can be understood, considering thatall data points do not correspond to the same numberof pulses and, thus, threshold variations are expectedin this situation due to the fatigue effect [5].

The representation of the data in Fig. 6 is not veryconvenient for quantitatively comparing our mea-surements to a noncooperative damage mechanismwith different but constant sensitivity of thematerialwith respect to the different wavelengths. In order toget a better idea of the degree of cooperativeness, itis useful to consider the literature reports on multi-wavelength laser damage in KDP (KH2PO4) [20,21].In order to compare KTP to KDP or to a noncoopera-tive damage mechanism, we can calculate the γparameter defined by DeMange et al. and plot itas a function of the 1064nm fluence F1064 [20]. Thedefinition of the γ parameter given in [20] isPPDðF532;onlyÞ ¼ PPDðF532 þ γ532=1064F1064Þ, wherePPD is the damage, or pinpoint, density. In our case,we only have information on one PPD, i.e., the PPDthat corresponds to the damage threshold in ourworking conditions. We can thus calculateγ532=1064 ¼ ðT532;only − T532Þ=T1064, where Ts are thefluence values at the damage threshold. Figure 7shows the obtained γ parameters as a function ofthe 1064nm fluence.

In this representation, it is even more evident thatthe data form a systematic curve despite the differ-ent measurement conditions. The only data pointthat is slightly off the curve is the one obtained bydegenerate parametric generation for which the con-version efficiency was estimated (and not measured).

A noncooperative damage mechanism would corre-spond to a horizontal line in Fig. 7 as indicated by thethin dashed line. An example for this type of result isthe γ355=1064 parameter in KDP [20]. On the contrary,as the γ parameter varies from ½ to 10 in KTP (seeFig. 7), the nature of the damage mechanism in thismaterial is strongly cooperative.

Comparing the γ curve of KTP to the γ355=532parameter in KDP, which is the parameter withthe strongest variations in this material, yields tworesults:

First, in KTP γ varies from <1 to >1, whereasin KDP γ < 1 for all conditions. If γ < 1, the short

Fig. 6. (Color online) The approximate laser damage threshold(incident pump fluence) for different conversion efficiencies. Thestar symbols indicate data that have been extracted from the 200-on-1 curves shown in Fig. 4. The uncertainty for the estimatedefficiency of the degenerate optical parametrical generation isindicated by the width of the red background of the black starsymbol. The dashed curve is a guide to the eye.

Fig. 7. (Color online) γ parameter as a function of the 1064nmfluence of the mixture created by frequency conversion. The thickdashed line, a shifted and scaled hyperbola fit, is a guide to theeye. The thin dashed line (γ ¼ 0:64) represents a noncooperativedamage mechanism.

1 August 2011 / Vol. 50, No. 22 / APPLIED OPTICS 4513

wavelength light is more damaging than the longwavelength light; this is the standard case. If γ > 1,the roles are reversed. In KTP, we thus have thesituation where it depends on the fluence ratiowhether green light is more damaging than IR lightor the inverse. In KDP, UV light is always moredamaging than IR light. This indicates that the da-mage mechanism in KTP is much more cooperativethan the one in KDP.

Secondly, the γ curve in KTP is decreasing,whereas the γ curve in KDP is increasing or stable,depending on the analyzed wavelength pair. This in-dicates a fundamental difference in the mechanismsinducing laser damage in these crystals.

5. Summary and Conclusions

We reported nanosecond-laser-induced damage mea-surements in the bulk of KTP crystals. One part ofthe experiments has been carried out with incident532nm light, and for the other part of the experi-ments, incident 1064nm light was used. For bothpump wavelengths, situations with and without fre-quency conversion have been investigated.

For the measurements without frequency conver-sion, we found that no damage threshold fatigue ef-fect is observed with 532nm incident irradiation,which is different from what is observed for 1064nmlight [5].

A common point with 1064nm irradiation and532nm incident irradiation is that the damagethreshold of Z-polarized light is higher than for X-or Y-polarized light. However, this effect is weakercompared to 1064nm light as the damage thresholdis increased by a factor of 1.6 with 532nm light,whereas at 1064nm, the factor was higher than 2.9.

Damage probability curves have also been ac-quired for two frequency conversion situations: thetype II frequency doubling situation where the crys-tal is pumped at 1064nm, and the inverse situationwhere the crystal is pumped at 532nm and 1064nmlight is generated. In both situations, the damagethreshold was found to be lower than for pure 1064or 532nm irradiation. The damage mechanism for si-multaneous 1064 and 532nm irradiations is thusstrongly cooperative between these two wavelengths.

In order to quantify the cooperativeness of the da-mage mechanism, we carried out another series ofexperiments. The damage threshold was estimatedfrom R-on-1-type experiments using varying con-ditions (IR polarization direction, focus diameter,and sample thickness) during which the conversionefficiency has been monitored. All damage thresh-olds (incident fluence) plotted against the conversionefficiency align close to a characteristic curve de-scribing quantitatively the cooperative damage me-chanism. The systematic behavior becomes evenmore evident when plotting the γ parameter as afunction of the 1064nm fluence. The conversion toγ values also allowed for a direct comparison withKDP on which similar studies have been carriedout before. The comparison with KDP points out that

the laser damage mechanisms, although cooperativein both cases, are probably very different in KTPand KDP. Furthermore, the cooperativeness of thedamage mechanism appears to be much strongerin KTP than in KDP.

A simple model for KTP, which has first beenpresented by Favre et al. [18], qualitatively describesthe data. According to the model, the laser damagemechanism in KTP would be based on the preferen-tial generation of color centers by 532nm light,followed by the preferential damage initiation by1064nm light. A quantification of the model basedon the presented measurements and rate equationsshould be possible now.

We acknowledge Hassan Akhouayri for valuablediscussions and correction reading and CristalLaser (France) for collaboration. The work has beenfunded by the Délégation Générale pour l’Armement(DGA) and the Centre National d’Etudes Spatiales(CNES).

References1. BCC Research LLC, http://www.optoiq.com/index/photonics‑

technologies‑applications/lfw‑display/lfw‑article‑display/227060/articles/optoelectronics‑report/volume‑12/issue‑9/features/major‑growth‑predicted‑for‑nonlinear‑materials.html, accessed2011.

2. BCC Research LLC, http://www.bccresearch.com/report/SMC019C.html, accessed 2011.

3. J. D. Bierlein and H. Vanherzeele, “Potassium titanyl phos-phate: properties and new applications,” J. Opt. Soc. Am. B6, 622–633 (1989).

4. H. Yoshida, H. Fujita, M. Nakatsuka, M. Yoshimura,T. Sasaki, T. Kamimura, and K. Yoshida, “Dependences oflaser-induced bulk damage threshold and crack patterns inseveral nonlinear crystals on irradiation direction,” Jpn. J.Appl. Phys. Part 1 45, 766–769 (2006).

5. A. Hildenbrand, F. R. Wagner, H. Akhouayri, J. Y. Natoli,M. Commandré, F. Théodore, and H. Albrecht, “Laser induceddamage investigation at 1064nm in KTiOPO4 crystals and itsanalogy with RbTiOPO4,” Appl. Opt. 48, 4263–4269 (2009).

6. F. R. Wagner, A. Hildenbrand, J.-Y. Natoli, and M.Commandré, “Multiple pulse nanosecond laser induceddamage study in LiB3O5 crystals,” Opt. Express 18, 26791–26798 (2010).

7. G. M. Loiacono, D. N. Loiacono, T. McGee, andM. Babb, “Laserdamage formation in KTiOPO4 and KTiOAsO4 crystals—graytracks,” J. Appl. Phys. 72, 2705–2712 (1992).

8. V. Murk, V. Denks, A. Dudelzak, P. P. Proulx, and V.Vassiltsenko, “Gray tracks in KTiOPO4: mechanism of crea-tion and bleaching,” Nucl. Instrum. Methods Phys. Res. Sect.B 141, 472–476 (1998).

9. S. Wang, V. Pasiskevicius, and F. Laurell, “Dynamics of greenlight-induced infrared absorption in KTiOPO4 and periodi-cally poled KTiOPO4,” J. Appl. Phys. 96, 2023–2028 (2004).

10. P. A. Champert, S. V. Popov, and J. R. Taylor, “3:5W frequency-doubled fiber-based laser source at 772nm,” Appl. Phys. Lett.78, 2420–2421 (2001).

11. J. Chen, A. J. Pearlman, A. Ling, J. Y. Fan, and A. Migdall,“A versatile waveguide source of photon pairs for chip-scalequantum information processing,” Opt. Express 17,6727–6740 (2009).

12. S. Furusawa, H. Hayasi, Y. Ishibashi, A. Miyamoto, andT. Sasaki, “Ionic-conductivity of quasi-one-dimensional

4514 APPLIED OPTICS / Vol. 50, No. 22 / 1 August 2011

superionic conductor KTiOPO4 (KTP) single-crystal,” J. Phys.Soc. Jpn. 62, 183–195 (1993).

13. “Determination of laser-damage threshold of optical surfacespart 2: S-on-1 test,” ISO 11254-2 (International Organizationfor Standardization, 2001).

14. A. Hildenbrand, F. R. Wagner, H. Akhouayri, J.-Y. Natoli,and M. Commandré, “Accurate metrology for laser damagemeasurements in nonlinear crystals,” Opt. Eng. 47, 083603(2008).

15. A. Hildenbrand, F. R. Wagner, J.-Y. Natoli, and M.Commandré, “Nanosecond laser induced damage inRbTiOPO4: the missing influence of crystal quality,” Opt.Express 17, 18263–18270 (2009).

16. L. Lamaignere, S. Bouillet, R. Courchinoux, T. Donval,M. Josse, J. C. Poncetta, and H. Bercegol, “An accurate, repea-table, and well characterized measurement of laser damagedensity of optical materials,” Rev. Sci. Instrum. 78, 103105(2007).

17. X. B. Hu, J. Y. Wang, H. J. Zhang, H. D. Jiang, H. Liu,X. D. Mu, and Y. J. Ding, “Dependence of photochromic

damage on polarization in KTiOPO4 crystals,” J. Cryst.Growth 247, 137–140 (2003).

18. S. Favre, T. C. Sidler, and R. P. Salathe, “High-power long-pulse second harmonic generation and optical damage withfree-running Nd:YAG laser,” IEEE J. Quantum Electron.39, 733–740 (2003).

19. F. R. Wagner, A. Hildenbrand, J.-Y. Natoli, M. Commandré,F. Theodore, and H. Albrecht, “Laser induced damagethreshold of RbTiOPO4: evidence of polarization dependentanisotropy,” Opt. Express 15, 13849–13857 (2007).

20. P. DeMange, R. A. Negres, A. M. Rubenchik, H. B. Radousky,M. D. Feit, and S. G. Demos, “The energy coupling efficiency ofmultiwavelength laser pulses to damage initiating defects indeuterated KH2PO4 nonlinear crystals,” J. Appl. Phys. 103,083122 (2008).

21. S. Reyne, G. Duchateau, J. Y. Natoli, and L. Lamaignere,“Pump–pump experiment in KH2PO4 crystals: coupling twodifferent wavelengths to identify the laser-induced damagemechanisms in the nanosecond regime,” Appl. Phys. Lett.96, 121102 (2010).

1 August 2011 / Vol. 50, No. 22 / APPLIED OPTICS 4515