Effect of dispersion on the operation of a KTP electro-optic Q switch

Effect of dispersion on theoperation of a KTP electro-optic Q switch

Ti Chuang, Alan D. Hays, and Horacio R. Verdun

The effect of dispersion on the operation of a KTP electro-optic Q switch was investigated. A KTP crystalin the normal orientation has been successfully operated as a Q switch. The experiment was performedwith a laser resonator containing a Nd:YLF crystal end pumped with a cw diode-laser array. A pulselength of approximately 100 ns was obtained at a repetition rate of 1 KHz. Temperature tuning was usedto eliminate static phase retardation. The effect of dispersion was found to affect Q switching.Wavelength stabilization was performed to counteract the effect of dispersion. These measures allowedus to achieve Q switching action.

Key words: Electro-optic Q switch, KTP Q switch, dispersion of KTP.

Potassium titanyl phosphate (KTiOPO4), or KTP, hasfound an increasing role as an efficient nonlinearcrystal for various frequency-conversion applications,such as second-harmonic generation and optical para-metric oscillation for laser sources, with wavelengthsnear 1 m.1 This is mainly due to KTP's highnonlinear optical coefficients, large angular band-width, high optical-damage threshold, and good ther-mal stability of phase-matching properties. Also,KTP is nonhygroscopic. However, one of its proper-ties, that is, its large linear electro-optic (E-O) coeffi-cients,1 has not been well explored for device applica-tions compared with KTP's other properties. Thisproperty will make KTP a good candidate for E-Omodulators or Pockels cells. Although KD*P andLiNbO3 are widely used nowadays as Pockels cells forQ switching and cavity dumping, they both haveshortcomings. LiNbO3 has a low half-wave voltagewhen it is transversely biased, and it is nonhygro-scopic. The latter property makes it possible for anintracavity LiNbO3 Pockels cell to be directly antire-flection (AR) coated on its faces, rather than having touse coated windows, to reduce the interface loss.A major drawback of an intracavity LiNbO3 Pockelscell is its relatively low optical-damage threshold,which limits its use in high-peak-power laser systems.KD*P, on the other hand, has a higher optical-damage threshold than LiNbO3. However, the KD*P

The authors are with Fibertek Inc., 510 Herndon Parkway,Herndon, Virginia 22070.

Received 8 March 1994; revised manuscript received 1 July 1994.0003-6935/94/368355-06$06.00/0.c 1994 Optical Society of America.

Pockels cell suffers from other drawbacks. It re-quires relatively high half-wave voltage and is hygro-scopic, which forces an intracavity KD*P Pockels cellto use index-matching materials and AR-coated win-dows on its interfaces, increasing intracavity loss.A KTP Pockels cell will retain the merits of bothLiNbO3 and KD*P Pockels cells, yet overcome theproblems with both cells. The KTP Pockels cell isalso free from the piezoelectric effect, which degradesthe performance of LiNbO3 in many applications.Therefore there is interest in developing KTP Pockelscells for use in laser systems. The use of the KTPPockels cell as a cavity dumper,2 as well as the resultsof a study of the use of a KTP crystal in thermallyinsensitive orientations as a Pockels cell,3 have beenpreviously reported. Recently we reported an appli-cation of a KTP crystal as an E-O Q switch used in thenormal orientation.4 In this paper we address theeffect of dispersion of a KTP crystal on the Q switch-ing operation and detail the experimental realizationof Q switching in the presence of dispersion.

KTP is a biaxial that belongs to the mm2 pointgroup (an orthorhombic crystal system). The crystal-lographic axes a, b, and c coincide with the opticalaxes x, y, and z, in which z(c) is the polar axis. Themeasured E-O coefficients of KTP can be found inRef. 1. Figure 1 depicts the crystal orientations,dimensions, and the crystallographic axes for theKTP sample used in this paper. For this configura-tion the phase retardation is given by2

2rrr = rF =- l~n -nx) + -n riV (1)

20 December 1994 / Vol. 33, No. 36 / APPLIED OPTICS 8355

( ) d /jtl ~~Z (c)

j4-d IL••~~Y (b)_ * d_3 k~~~~~x (a)

dim: d x I x d = 2.5 mm x 5 mm x 2.5 mm

Fig. 1. KTP Q switch layout, dimensions (dim), and crystallineorientation. Transverse high voltage (HV) was applied along the zaxis. The oven for temperature control is not shown.

for light propagating along the y axis with a length and a transversely applied voltage V = Ed. In Eq.(1), rj1 is the effective E-O coefficient and is given byrc = r33 - (n/n,)3 r13 , in which r33 and r13 are E-Ocoefficient tensor elements of KTP. The secondterm of Eq. (1) is the dynamic (voltage-related) phaseretardation and is responsible for the E-O modulation.The first term of Eq. (1) is a static phase retardation,which has to be compensated in order for the secondterm to be effective. The static phase retardation,when the temperature (T) and the wavelength (X)dependence are included, can be written as

rsl = rO + rT + rX, (2)

2rrFo = - A-lo(nzo - n.o), (3)

2sr atrT = - A al AT(nzo - nxo)

2r10 2nz

- A l T

an, AT,aT/

2mr = an, _ an, AX (5)

where ro is the phase retardation at the referencetemperature (usually at room temperature) and thereference wavelength and rT and rF are the tempera-ture- and wavelength-dependent terms, respectively.Equation (5) represents the contribution from thedispersion an/dA. In Eqs. (2)-(5) parameters with asubscript 0 are values at the reference temperatureand wavelength. In the case in which there is nowavelength change, the contribution from dispersiondisappears, and rI and Fo + rT can be rearranged intoa simple form:

2'TF = - - 1O(nZ0x

an, an,Al T AT

- n1+10 Anz - nxo T

(6)

It can be seen from Eq. (6) that, with a propertemperature, the static phase retardation can betuned to be a multiple of 2wr so that it has no influenceon the modulation, leaving the total phase retarda-

tion [Eq. (1)] to be entirely determined by the appliedvoltage. When the wavelength change occurs, thedispersion will exert a considerable effect on the totalphase retardation, even though in this situation thetemperature tuning may still compensate some partsof the static phase retardation.

Unlike cavity dumping, in which a KTP Pockels celldoes not influence the laser-oscillation wavelength, aKTP Pockels cell inside a laser resonator that acts asa Q switch does influence the laser-oscillation wave-length. This influence is primarily due to the effectof dispersion of the KTP crystal. This means thattemperature tuning as well as wavelength limitingare required for compensating fully for the staticphase retardation. In our experiment we used anintracavity 6talon to limit the wavelength shift. Atthe same time we temperature tuned the KTP Qswitch. The operating temperature for the KTPO Qswitch at a fixed wavelength was determined experi-mentally as described below.

The KTP crystal used in this work had dimensionsof 2.5 mm x 5 mm x 2.5 mm (x, y, z). It washydrothermally grown by Litton Airtron and was cutnormal to the principal crystal axes. Figure 1 de-picts the dimensions and the crystallographic axes ofthe KTP sample. Its two surfaces along the y axiswere AR coated at 1064 nm (note that in this work thelaser operated at 1047 nm). On the each of twosurfaces normal to the z axis a gold thin film wassputtered as the electrode for the applied high voltage.The laser light propagated along the y axis with itspolarization at 450 away from the x axis. To deter-mine the operating temperature for the KTP Qswitch at which the static phase retardation (thermalpart) can be compensated, a measurement was car-ried out as shown in Fig. 2. In this setup thethin-film polarizer (TFP, coated at 1064 nm), posi-tioned at the Brewster angle with respect to the laserlight's (X = 1047 nm) propagating direction, acted asa polarization discriminator. The photodiode PD1measured the intensity of the transmitted laser light(defined as mr light), and the photodiode PD2 mea-sured the intensity of the reflected laser light (definedas or light). Without the KTP in place, the Calcite

KP PD2 ()

TFPPD1 ()

L4i

Power meter

ND filters Pinholei..................

- E...................

KTIwith o

3ftI............

oven

. .†. M3 i. mirror

Calcite 1. ChopperPolarizer ITT

Laser................................................. ... .. . . .Beam splitter

Laser wavelength: 1047 nmFig. 2. Setup for determining the operating temperature for theKTP Q switch: ND, neutral density; PD, photodiode. A pinholewas used to collimate the laser light.

8356 APPLIED OPTICS / Vol. 33, No. 36 / 20 December 1994

..... .

...,............ .................................

I

polarizer was oriented such that PD1 read maximumsignals and PD2 read minimum signals. The out-puts of PD1 and PD2 were sent to a Tektronix 2440digital oscilloscope for signal averaging. The triggerwas provided by the mechanical chopper shown inFig. 2. The purpose of chopping light was to permitthe signal averaging to be performed to have a goodsignal-to-noise ratio. When the KTP was placedinside the oven and properly oriented, the tempera-ture of the oven was varied and the readouts of PD1and PD2 were recorded as a function of the tempera-ture. During the entire measurement, the laserpower was maintained at a constant level as moni-tored with a power meter, also indicated in Fig. 2.The result of this measurement is given in Fig. 3.In Fig. 3 a temperature cycle of 19 K at the 1047-nmwavelength of a Nd:YLF laser operating in a cw modeis shown. The operating temperature for the KTP Qswitch was chosen to be 319 K as a result of themeasurement. The measured polarization ratio wasgreater than 250:1 and was limited by the TFP. TheKTP oven consisted of two resistors for heating and atemperature-sensing integrated circuit chip (AnalogDevices AD590) and was equipped with a feedbackloop to achieve the desirable temperature stability.The temperature stability was better than 0.1 K, andthe temperature uniformity across the crystal wasbetter than 0.01 K.

As mentioned above, the static phase retardation ofthe KTP Pockels cell consists of two parts: thermaland dispersion. The thermal part can be compen-sated by the temperature tuning, as has been demon-strated above. The dispersion part, however, canprevent the KTP Pockels cell from working as an E-OQ switch.In the case in which the KTP Pockels cell isused as a cavity dumper, the laser wavelength is setby the laser oscillator; dispersion has no effect on thecavity dumping. In the case in which the KTPPockels cell is used as a Q switch, the KTP Pockelscell becomes an integral part of the laser oscillator.The dispersion will shift the laser wavelength to a

300

250

C,)

C1

0)C2

200

150

100

50

0

305 310 315 320 325 330 335

Temperature (Kelvin)Fig. 3. Intensity variation of r and r laser lights as a function ofthe KTP Q switch operating temperature. The r and Cr laserlights are defined in the text. Pol., polarization.

value such that the resonator loss that is due to thephase retardation is minimized. This impact is bet-ter demonstrated by Figs. 4-6. Results given inFigs. 4-6 were obtained with the laser-resonatorsetup shown in Fig. 7, with the following modifica-tions: (1) the 6talon was removed, and (2) a TFP andtwo photodiodes were positioned behind the concavehighly reflective (HR) mirror, where a leakage of laserlight was available for detection. The arrangementof the TFP and photodiodes was identical to thatshown in Fig. 2 for detecting rr laser light and r laserlight. One of the two surfaces of the Nd:YLF lasercrystal was fabricated at the Brewster angle, which,together with crystal's orientation, defined the laserpolarization.

Figure 4 shows the laser wavelength shift and therelative laser power change as the quarter-wave plate(QWP) was rotated. Figure 5 shows the intensityvariation of Tr laser light and u laser light as a functionof the QWP rotation. These intensities were normal-ized to the laser output power so that the intensityvariation could indisputably indicate the change ofthe polarization states of the laser. In obtaining theresults shown in Figs. 4 and 5, we took measures toensure that (1) the 00 rotation of the QWP was theposition where the optical axis of the QWP wasparallel to the laser polarization, (2) the temperatureof the KTP oven was set and kept at 319 K, and (3) theKTP Q switch was properly oriented. It is clear fromFig. 4 that as the QWP was rotated to introduce aphase retardation to change the polarization of thelaser light, the dispersion of the KTP Q switch shiftedthe laser wavelength so that the change of staticphase retardation of the KTP crystal counteractedthe phase retardation introduced by the QWP, main-taining laser oscillation, although at a lower powerlevel because of the laser gain value change associatedwith the wavelength shift. At 450 rotation the laserstill lased, in sharp contrast with a normal case inwhich a hold off usually occurs. One example ofsuch a normal case is the KD*P Q switch, which doesnot possess a static phase retardation. It is interest-

1047.6

-E 1047.5c

s 1047.40)C(Da> 1047.3co

L 1047.2a)

CZ- 1047.1

1047.0

800

700 E

a)600

0a-

500 a)

-400 a,

300 a)

200

-90 -45 0 45 90Quarter-Wave Plate Rotation (deg)

Fig. 4. Laser wavelength shift and the relative laser-power changeversus the QWP angular position.


DN/ HR mirror\% 50 cm C.c.

X/4 plate

KTP Q Switch

Etalon

Nd:YLF

Beam-shaping optics

-90 -45 0Quarter-Wave Plal

45 90:e Rotation (deg)

Fig. 7. Configuration of the laser resonator used to test the KTPQ switch.: C.C., concave; OC, output coupler.

Fig. 5. Change of the polarization state of the laser light versusthe angular position of the QWP.

ing to observe the flatness of the two curves in Fig. 4within a certain range of QWIP rotation. This flat-ness indicated that within this range the differencebetween resonator gain and losses remains fairlyconstant, even when the laser wavelength is shiftingaway from the fluorescence peak of the Nd:YLFcrystal. This observation needs further study.

The change of the polarization state of the laserlight is shown in Fig. 5. It is complementary to Fig.4 and is helpful for understanding the impact of thedispersion. Figure 6 illustrates the wavelength shiftthat is due to the dispersion. In Fig. 5 there are twoparticular rotations of the QWP that deserve atten-tion. The first one is at +45° rotation, where therewas virtually no rr laser light. This indicated thatthe laser polarization behind the concave HR mirrorwas perpendicular to that of the laser light before itpassed through the KTP Q switch. This could onlyhappen if the KTP Q switch introduced a 900 phasechange in addition to that already introduced by the

4Q000

30000

C.' 24000(n

a)

4000

0o

1046.5 1047.0 1047.5 1048.0

Laser Wavelength (nm)Fig. 6. Laser wavelengths at different QWP angular positions.The intensities are not normalized.

QWP at +45° rotation. The second rotation is at-45°, at which, quite contrary to the +45° rotation,the phase change introduced by the KTP Q switchcanceled the phase change introduced by the QWP.We are currently developing a model that will allow usto understand more quantitatively the effect of thedispersion and the physical process of the dispersionto influence laser performance.

The Q switching action of the KTP Q switch wasrealized in a laser-resonator setup,5 shown in Fig. 7.The detailed resonator parameters can be found inRef. 5. This laser resonator was basically composedof the following elements: a trapezoid-shapedNd:YLF laser crystal, whose c axis was long its base; aflat output coupler with 80% reflection at 1.064 [um; aconcave mirror that was HR at 1.064 im, with r = 50cm; an AR-coated (at 1.047 jim) QWP; the KTPPockels cell, which acted as an E-O Q switch, and asold thin talon made of quartz. The geometriclength of the resonator between the HR mirror andthe output computer was 42 cm. The pumpingsource for this resonator was a 15-W cw diode lasermanufactured by Spectra Diode Labs (Model SDL-3450-S). The incident power to the Nd:YLF crystalwas 11 W. The solid 6talon had a thickness of 0.5mm and a coating of 72% reflection at 1.047 [im atboth faces. The parameters provide a free spectralrange of 0.73 nm and a finesse of 9.5 for the 6talon.The 6talon was introduced to counter the effect ofdispersion of the KTP Q switch. To operate theKTP Q switch, we tuned the laser wavelength bytilting the 6talon to 1047.2 nm while the QWP was setat 00 rotation. Then the QWP was rotated by 450 toprovide the hold off for Q switching before a high-voltage (HV) pulse was applied to the KTP Q switch.In this experiment the introduction of the QWP wassimply for practical purposes. With the QWP, thelaser could be optimized before the Q-switching tookplace. Of course one can reverse the timing se-quence of the Q switch's HV pulse and completelyeliminate the QWP. The choice of X = 1047.2 nm


0.5

CZ

cod

.

-C)

N

CZ

E0

0.4

0.3

0.2

0.1

Oc80%

0.0

can be explained by the results shown in Fig. 4,where, at 1047.2 nm, the derivative is maximum.

The performance of the laser with the KTP Qswitch is illustrated in Fig. 8. The top trace is theHV pulse applied to the KTP Q switch, and thebottom trace is the Q switched laser output pulse.The laser-pulse buildup time was 400 ns. Thelaser-pulse width was approximately 100 ns. The Qswitching repetition rate was 1 KHz and was limitedby the HV power supply used in this paper. Thepulse energy was 0.12 mJ. For a comparison we givea result obtained earlier with the same resonator butwith an acousto-optical Q switch and without anintracavity talon. Operated at a 1-KHz repetitionrate, this laser generated pulses with a 52-ns pulsewidth, with a few hundred nanoseconds of builduptime and energy of 0.74 mJ per pulse. This compari-son suggests that the long pulse width and the longpulse-buildup time obtained with the KTP Q switchcould be mostly attributed to the relatively large lossproduced by the talon and the QWP. Figure 9shows the laser wavelength with and without Qswitching. It is evident from Fig. 9 that there was nowavelength shift. The best results of the Q switch-ing were obtained at an ac quarter-wave voltage of1.79 kV, which was different from the value of 1.56kV that we calculated with the following numbers1 :r33 = 35 pm/V, r13 = 8.8 pm/V, n = 1.7404, n =1.8302, d = 2.5 mm, = 5 mm, and the second term ofEq. (1). We are currently investigating the source ofthe discrepancy. Some green light was observedduring Q switching. This was the consequence ofthe frequency doubling of 1.047-jim laser light insidethe KTP Q switch, even though the switch was notcut for second-harmonic generation. An interesting,yet puzzling, observation was made during the experi-ment depicted in Fig. 2. At the operating tempera-ture of 319 K, a QWP was positioned between theTFP and the KTP, and the optical axis of the QWPwas properly oriented such that PD1 and PD2 regis-tered the same intensities. Then a dc quarter-wavevoltage was applied to the KTP. The intensities

400

.a0)a

300 -

200 -_

100 -

0

800

HV pulseL.

L pulse

1 2 I I I 01200 1600 2000 2400 2800

Time (ns)Fig. 8. Temporal performance of the laser: top trace, HV pulseapplied to the Q switch; bottom trace, laser output pulse.

annnn

. _a)

Q4LVUU -

20000 -

1Q000 -

O

104 6.0 1046.5 1047.0. I I . I . I

1047.5 1048.01

1048.5

Laser Wavelength (nm)Fig. 9. Laser wavelength when the talon was used to limit thewavelength shift: top trace, laser wavelength with Q switching;bottom trace, laser wavelength without Q switching and with theQWP at 0° rotation.

detected with PD1 and PD2 changed with a timedelay of approximately 10 s after the voltage wasapplied. When the dc voltage was disconnected fromthe KTP, it took even longer than 10 s for theintensities to return to their original state of equalintensities. This phenomenon perhaps was due tosome electrolytic process that reduced the internalelectric field throughout most of the cell volume.No damage of the KTP crystal as a result of theapplied HV (both ac and dc) was observed.

In conclusion, we have successfully demonstratedQ switching action with a KTP E-O Q switch in aNd:YLF laser. The effect of dispersion of the laserwas observed and was neutralized by the use ofwavelength stabilization. These results are prelimi-nary; there is still room for improvement. Improve-ment includes the reduction of insertion losses thatresult from the KTP Q switch and the etalon. Withits properties of high damage threshold, lower half-wave voltage, nonhygroscopicity, etc., the KTP Qswitch will find its way into applications in whichthese properties are indispensable. Improvement ofthis new device and its use in a high-peak power laserresonator where an ordinary LiNbO 3 Q switch wouldbe damaged is currently being contemplated.

This work was supported by U.S. Army contractDAAB07-92-C-K760.

Note added at proof A paper presented at the1994 topical meeting on compact blue-green lasers byTaira and Kobayashi6 described the application of aKTP crystal as both a frequency doubler (Type II) anda Q switch in a diode-pumped Nd:YVO4 laser. TheKTP crystal was oriented for Type II phase matchingat 1.064 jim. In their paper no wavelength-selectivecomponent was used, indicating that the effect ofdispersion of the KTP crystal was not observed anddid not cause any difficulty in their paper. Thisperhaps was due to the relatively narrower laser gaincurve of Nd:YVO4 crystals compared with Nd:YLF


Top: with QS

1 t

Bottom: without QS

I

crystals. The wavelength shift could greatly reducethe gain of the Nd:YV0 4 laser, hence stopping thelaser oscillation.

References1. J. D. Bierlein and H. Vanherzeele, "Potassium titanyl phosphate:

properties and new applications," J. Opt. Soc. Am. B 6, 622-633(1989) and references therein.

2. X. D. Wang, P. Bass6ras, R. J. D. Miller, and H. Vanherzeele,"Investigation of KTiOPO4 as an electro-optic amplitude modu-lator," Appl. Phys. Lett. 59, 519-521 (1991).

3. C. A. Ebbers, "Thermally insensitive, single crystal, KTP Qswitch for kilowatt, average power lasers," in Conference onLasers and Electro-Optics, Vol. 12 of 1993 OSA Technical

Digest Series (Optical Society of America, Washington, D.C.,1993), pp. 166-168.

4. Ti Chuang, A. D. Hays, and H. R. Verduin, "Application of KTPas an electro-optic Q switch," in Advanced Solid-State Lasers,T. Y. Fan and B. Chai, eds., Vol. 20 of OSA Proceedings Series(Optical Society of America, Washington, D.C., 1994), p. 75.

5. L. R. Marshall, A. Kaz, and H. R. Verdn, "Power scaling andwavelength conversion of cw diode-pumped lasers," in Ad-vanced Solid-State Lasers, A. A. Pinto and T. Y. Fan, eds., Vol.15 of OSA Proceedings Series (Optical Society of America,Washington, D.C., 1993), pp. 78-80.

6. T. Taira and T. Kobayashi, "Frequency doubled and Q switchedNd:YVO4 laser," in Compact Blue-Green Lasers, Vol. 1 of 1994OSA Technical Digest Series (Optical Society of America, Wash-ington, D.C., 1994), pp. 104-107.


Documents

Effect of dispersion on the operation of a KTP electro-optic Q switch