Reprinted from Applied Optics 29, 1752-1757 (1990).

Optimization and characterization of a high repetitionrate, high intensity Nd:YLF regenerative amplifier

Muhammad Saeed, Dalwoo Kim, and Louis F. DiMauro

Solid state regenerative amplifiers have proved to be a reliable source for producing stable millijoule pulses asshort as a few picoseconds at repetition rates ranging from a few hertz to several kilohertz. Here we report onthe operation of a cw pumped Nd:YLF regenerative amplifier that uses a convex-concave design to optimizeoutput energy and stability while minimizing the energy density on critical intracavity optical components.The amplifier yields stable 5-mJ 40-ps pulses at 700 Hz and has achieved 1-mJ output at a 3.5-kHz repetitionrate. Results and analysis of the beam profile and contrast ratio for this system are also presented. Ourresults are contrasted to the general design considerations of regenerative amplifiers.

1. IntroductionIn recent years there has been work directed toward

the development of a unique class of solid state opticalamplifiers capable of producing high gain and repeti-tion rates in the kilohertz regime. This class of lasershas become known as regenerative amplifiers. Thepioneering work of Lowdermilk and Murray1 2 laid thefoundations for amplifying picosecond IR pulses tonearly a millijoule with a regenerative amplifier. Intheir scheme, a single pulse from a mode-lockedNd:YAG laser was seeded into a pulsed flashlamppumped Nd:YAG regenerative amplifier, where itmade several passes through the gain medium beforebeing cavity dumped. The single pass gain within theamplifier cavity was kept low allowing the seed pulse tomake -100 passes before reaching the saturation limit.It is the multipass operation of this system whichmakes it distinctive from conventional traveling waveamplifiers. A significant advance in high repetitionrate performance of a regenerative amplifier was madeby Duling et al.3'4 when they demonstrated the cwpumped operation of a Nd:YAG amplifier. Othergroups5-7 have reported on further developments inNd:YAG regenerative amplifiers. Recently, Bado etal.8 have reported on the 500-Hz operation of aNd:YLF regenerative amplifier. As predicted by the-ory, 9 Nd:YLF proved to be more efficient as a gainmedium in a regenerative amplifier as compared toNd:YAG. Its broader bandwidth, weaker thermallensing, stronger natural birefringence, and polarizedoscillations result in shorter pulses and more efficientoperation in the lowest-order Gaussian mode, TEMOO.

The authors are with Brookhaven National Laboratory, Chemis-try Department, Upton, New York 11973.

Received 12 June 1989.0003-6935/91/182527-06$05.00/0.© 1991 Optical Society of America.

We report here on an optimized cavity design for acw pumped Nd:YLF regenerative amplifier system ca-pable of operating at kHz repetition rates. This de-sign has produced 5-mJ, 40-ps, 1.05-,um radiation at700 Hz and operates at repetition rates as high as 3.5kHz. Special emphasis is placed on the analysis of themode quality and contrast ratio of the amplifier's out-put. Moreover, the system provides stable outputwith little day-to-day alignment required and is ideallysuited as a high repetition rate pump source for tun-able amplifiers.

II. General Design ConsiderationsA regenerative amplifier's operational parameters

make it distinct from a conventional single-pass ampli-fier in many aspects of performance. For one, theregenerative amplifier should be viewed as a stable,high-Q resonator capable of producing gain for multi-ple passes of a single seed pulse. Obviously, the choiceof resonator design, gain medium, and mode of opera-tion will greatly influence the ultimate performance ofsuch a system. In fact, regenerative amplifiers thathave been reported on so far have a spectrum of outputcharacteristics and operating conditions. In this sec-tion, we will contrast these various choices by describ-ing the advantages and disadvantages of each.

In principle, the physical and optical properties ofthe amplifying medium determine the output charac-teristics of an amplifier. The pulse width, energy, andthe possible repetition rate depend upon the proper-ties of the material being used like bandwidth, thermalconductivity, radiative lifetime, etc. For high repeti-tion rate operation (>500 Hz), a material with a highthermal conductivity is required. The well-knownproblems associated with the low thermal conductivityof Nd:glass exclude its use in a high repetition ratesystem. However, Nd:YAG is one material that hasbeen used successfully as a gain medium in kHz regen-erative amplifiers.2 4 5 10 A typical output of aNd:YAG regenerative amplifier would be a 90-ps, 1-mJpulse at a 1-kHz repetition rate. However, Nd:YAG

20 June 1991 Vol. 30, No. 18 / APPLIED OPTICS 2527

has some major disadvantages for reliable regenerativeamplifier operation. Its strong thermal birefringenceresults in the unwanted coupling of substantial powerout of the cavity through intracavity polarizing optics.Also, Nd:YAG exhibits strong thermal lensing as afunction of the pump power and repetition rate. Thislensing limits the dynamic range of operation since theNd:YAG rod is effectively a thermally sensitive lens.Thus, the cavity must be optimized for a particularloading of the amplifier, i.e., repetition rate and pumppower. A superior material for use in a high repetitionrate regenerative amplifiers8 is Nd:YLF because of itsbetter optical and thermal properties. Its physicalcharacteristics include a three times larger bandwidthand longer lifetime than Nd:YAG, and a much lowervalue for thermally induced lensing and birefringence.Its stimulated emission cross section is within 30-50%to that of YAG. All these properties lead to an outputwhose temporal pulse width is three times shorter thanYAG, substantially lower losses through the intra-cavity polarizers, a lower lasing threshold and, mostimportantly, an amplifier capable of operating over awide range of thermal loading.

A final consideration for general design of a regener-ative amplifier is the method of pumping, that is, cw orpulsed fashlamps. For high repetition rates, cw oper-ation has many advantages. The most obvious advan-tage is the absence of any pulsed power supplies whichare inherently cumbersome. In fact, for the cwpumped system described here, the repetition rate iscurrently limited to 3.5 kHz by the electrooptic materi-al. This represents a significantly less severe techno-logical limit than high repetition rate flashlamp powersupplies. A more subtle but equally important advan-tage of cw pumping is the mode quality and stability ofthe output. The mode quality of the amplified outputdepends on the time the seed pulse is allowed to remainin the regenerative amplifier to define its transversespatial mode. This time or number of passes is deter-mined by the characteristics of the amplifier which isdirectly dependent upon the buildup time of the Q-switched envelope within the amplifier cavity. Physi-cally, this corresponds to the time of maximum photondensity in the amplifier cavity. The buildup time, Tb,will depend on the pump rate, the higher the gain orinitial population inversion in the amplifier, theshorter the Tb. Consequently, the seed pulse experi-ences fewer passes before reaching saturation. Con-sider the following numerical example. The Q-switchbuildup time is given by,1

Tb t 25 (1)r-1

where T = 2L/c is the round-trip time in the amplifiercavity, 6c is the fractional power loss per round trip dueto cavity losses, and r is defined as the ratio of theinitial population inversion at Q-switching to thethreshold inversion after Q-switching. For a typicalregenerative amplifier cavity T = 10 ns and 6, = 0.1.Consider a pulsed flashlamp pumped system operatingin the high gain limit, r - 200. This yields a value for

BD BS1

:R I I BS2'~~~~~~ o----------__

Jt iM E RL A M2 1

Iml PC WP2 L

.29(0) 3.18(7) 1.91(89) 0.89 (170)

Fig. 1. Block diagram of the regenerative amplifier system. Mland M2 are concave and convex resonator mirrors, PC Pockels cell, Lcylindrical lens, A mode selector, P1 and P2 Brewster angle polariz-ers, WP1 half-wave plate, WP2 quarter-wave plate, BS1 and BS2 5%beam splitters, FR Faraday rotator, BD beam dump and R highreflectors. The e-2 diameters in mm are indicated at various posi-tions (the positions in cm are in parenthesis) in the amplifier. Thevalue of the positions are measured relative to mirror Ml. The valuefor the beam diameters are for a cavity length which gives a 12-ns

round-trip time. See text for details.

Tb - 10 ns or one round trip. By contrast, a cwpumped regenerative amplifier operating at low gain, r

4, has a buildup time Tb - 830 ns or 83 round trips.Clearly, the seed pulse will spend a significantly longertime in the cw pumped amplifier cavity which, in turn,will result in a better defined transverse mode struc-ture. Likewise, the pulse-to-pulse stability will alsobenefit from this scheme, since the long buildup time isless sensitive to fluctuations in the pump power orcavity instabilities. It is important to note that inboth examples the maximum output energy is similar,since the cavity is usually limited by the peak powerdamage threshold. The following sections will exem-plify these advantages as they pertain to our cwpumped Nd:YLF regenerative amplifier system.

III. Experimental

A. ApparatusThe oscillator, as shown in Fig. 1, is a Quantronix

model 496, cw pumped mode-locked Nd:YLF laserproducing 40-ps, 80-nJ, 1.053-Am pulses at 100 MHz.Two 5% beam splitters (BS1 and BS2) are used to seed<1%, 100 pJ, of this output into the regenerativeamplifier. A Faraday rotator (FR), with an extinctionratio of >1000:1, in conjunction with the beam splittersprovides sufficient optical isolation between the oscil-lator and the amplifier. This isolation is necessary toprovide stable operation and protection of the mode-locked laser against amplifier feedback. A half waveplate (WP1) provides control over the polarization ofthe seed pulses. Furthermore, since the seed pulsesare coupled into the amplifier through a polarizer (P1),rotation of the wave plate (WP1) provides control overthe input energy of the seed pulses. It should be notedthat >90% of the oscillator output is available for fre-quency doubling for subsequent synchronous pump-ing of a mode-locked dye laser.

2528 APPLIED OPTICS / Vol. 30, No. 18 / 20 June 1991

en

C

Ld-

I I I II I I I I

TIME ( 50 nsec / division

Fig. 2. Fast photodiode output from <1% resonator leakagethrough mirror Ml during regenerative amplifier operation exclud-ing cavity dumping. Each successive peak is the single trappedpulse changing in amplitude as it returns to mirror Ml. The separa-tion between peaks is 10.5 ns, the round-trip time in the amplifier'sresonator. The temporal envelope is the Q-switch pulse buildup

and decay curve.

The cavity design of the regenerative amplifier is aconvex-concave resonator. This configuration waschosen to provide stable operation with regard to me-chanical and thermal fluctuations and to minimize theenergy density on critical optical components. Theresulting transverse mode distribution along the prop-agation direction is tapered. The maximum beamdiameter is located at the concave mirror while thewaist is located outside the cavity. Optical compo-nents with the lowest damage threshold are thereforelocated at the concave mirror side of the cavity. Thevalues listed in Fig. 1 are the beam diameters at the e-2

intensity values at the various components. The con-cave mirror (Ml) is a high reflector, 99.8% R, with a200-cm radius of curvature. Three 120-cm convexpartial reflectors (M2) with reflectivities of 88, 97, and99% were used to characterize the system. The gainmedium was a wedged-faced 4-mm diam X 104-mmlong Nd:YLF rod housed in a Quantronix model 117single lamp head. An aperture located 22 cm frommirror M2 provides sufficient selectivity for TEMoomode operation. An intracavity cylindrical lens isused to correct for astigmatism and asymmetry distor-tions in the output beam.12 Various cavity lengthswere tested and an optimum length was determined forbest contrast ratio to be that which provided a 10.5-nsround-trip time. A pair of Brewster angle dielectricpolarizers (P1 and P2) operating in the reflective modelimited the cw oscillations to 1.053 ,um while maintain-ing a linear cavity with a good extinction ratio. Thesepolarizers, as well as all optical components, were se-lected for their high damage threshold (few GW/cm2),minimal loss and wavefront distortions. The amplifi-er resonator as described above will produce 16 W of cwpower at 1.05 Am as measured through a 12% outputcoupler. The electrooptic switch consists of a staticquarter wave plate (WP2) and a Medox Electro-Opticsmodel DR85-A Pockels cell (PC) assembly and driver.

Two Pockels cell materials were tested for regenerativeamplifier operation, namely, KD*P and LiNbO3 . Aswill be discussed, the latter was found to be the superi-or material for low single-pass gain regenerative ampli-fiers.

B. OperationThe regenerative amplifier operation can be visual-

ized as four simple steps, (1) trapping a single pulsefrom the mode-lock seed train in the amplifier resona-tor, (2) simultaneously Q-switching the amplifier reso-nator, (3) letting the trapped pulse build up gain byundergoing N round trips, where N > 2, and (4) atsaturation, cavity dumping the amplified pulse. Theabove functions are all controlled by the static quarter-wave plate and the single Pockels cell. The MedoxElectro-Optic Pockels cell and driver13 employs a dou-ble pulse design with a 6 ns optical rise time. The firstquarter-wave transition of the Pockels cell traps theseed pulse and Q-switches the amplifier. The seconddelayed half-wave transition, cavity dumps the ampli-fied pulse. Alternately, two separate Pockels cells canbe used4 but the Medox design minimizes cavity losses.

Lets consider the operation in more detail. Refer-ring to Fig. 1, the input and output beams of theamplifier counterpropagate along the same path anddiverge after the beam splitter, BS2. Each beam mustbe p-polarized to transmit through polarizer P1. Con-versely, the amplifier resonator will only oscillate onthe s-wave as dictated by the reflective mode of thepolarizers P1 and P2. Thus, for a pulse to be trappedin the amplifier, it must have its input polarizationrotate from p- to s-wave. Initially, with no voltage(zero retardation) on the Pockels cell the static quar-ter-wave plate is oriented to produce a 7r/4 retardationin the amplifier cavity. Thus, the cavity Q is frustrat-ed and any seed pulses entering the cavity will undergoa total 7r phase shift in four passes through WP2 andexit the amplifier with no gain. Likewise, with a quar-ter-wave voltage applied to the Pockels cell, the seedpulse will exit after two passes through WP2 and PC.However, a pulse present inside the amplifier duringthe Pockels cell transition from zero to quarterwaveretardation will suffer an additional 7r/4 phase shiftand remain trapped in the amplifier. In addition, thesame transition will Q-switch the amplifier cavity.Amplification of this trapped pulse will continue witheach subsequent pass through the YLF rod until a half-wave voltage (7r/2 retardation) is applied to the Pock-els cell. At that time, the pulse is rotated to a p-waveand exits the amplifier through polarizer P1. Theoptimum time for cavity dumping is equal to the Q-switch buildup time, Tb, which is determined by thecharacteristics of the resonator and pumping rate.

IV. ResultsFigure 2 is a photograph of the leakage light through

the concave mirror, (Ml) sampled by a fast photodiodeduring regenerative amplifier operation without cavitydumping. Each successive peak is the single trappedpulse changing in amplitude (gain or decay) as it re-

20 June 1991 / Vol. 30, No. 18 / APPLIED OPTICS 2529

5

.E 4

, 3

_22

ol | I -- _ I I0.0 1.0 2.0

REPETITION RATE ( kHz)

6

5

43

3 -

2 $

-1 -------- - O3.0 4.0

Fig. 4. Energy per pulse and average power output of the regenera-tive amplifier as a function of repetition rate. The open circles arethe measured energy/pulse and the solid circles are the average

power.

TIME ( 50 nsec /division )

Fig. 3. Fast photodiode output during regenerative amplifier oper-ationincludingcavitydumping. Trace a is output from same photo-diode as in Fig. 2 and trace b is photodiode output of amplifieroutput. Note that at the time of cavity dumping all the energy isextracted from the cavity (trace a) and emerges as a single pulse

(trace b).

turns to mirror Ml in the cavity round-trip time. Thegain profile is identical to that of the Q-switched enve-lope from the same cavity with no seed pulse. Thebuildup time to the point of maximum photon densityfor our system is typically 700 ns or 70 round trips. Itis at this time that the half-wave voltage is applied tothe Pockels cell to initiate cavity dumping. As seen inFig. 3, all the energy is removed from the cavity and asingle pulse emerges along the output direction. Theoutput in Fig. 3(b) is a 40-ps, 3-mJ pulse running at a 1-kHz repetition rate. The pulse-to-pulse stability is±2%. The highest energy that we have achieved is 5mJ per pulse at repetition rates up to 700 Hz, limitedby the damage threshold of the LiNbO3 crystal. Fig-ure 4 is a plot of the energy per pulse and average poweras a function of repetition rate. As is evident from thecurve, there is a roll-off in the energy at 700 Hz. Thisroll off occurs at a value slightly less than the onereported by Bado et al. 8

We used two methods for determining the modequality and stability during operation of the regenera-tive amplifier. These included measuring the beamdiameter at various points outside and inside the cavi-ty as a function of lamp pump power and repetitionrate. To test the mode structure within the cavity, weused Gaussian beam propagation theory14 to calculatethe e 2 radius at various positions in the cavity. Thethermal lensing focal lengths f_ and f for the YLF rodassumed are typically 15 and -6 m, respectively.' 5

The calculated beam diameters agree with the mea-sured values at various positions to within 15%. Infact, as the lamp pump energy is varied over its entiredynamic range, no significant change is produced inthe mode distributions. Likewise, the same insensi-

0

DISTANCE (arb. units I

Fig. 5. One-dimensional slice of the amplifier's output beam profiletaken 6 m from the laser with a CCD camera. The open circles arethe digitized output of the camera and the solid line is the result of anonlinear fit to a TEMOO mode distribution. The result of the fit is

consistent across the 2-D distribution.

tivity is observed over a range of repetition rates from0.2 to 2 kHz. The measured/calculated e 2 diametersof our cavity mode are shown in Fig. 1. The externalfar field mode quality was monitored with an IR-sensi-tive CCD camera. The data were fit using a Gauss-Newton algorithm. The results of one of the fits isshown in Fig. 5. This is a 1-D slice of the beam profiletaken 6 m from the laser. The figure shows excellentagreement between the data and the lowest-orderGaussian mode, TEMoo. To test the validity of thefits, analyses were conducted that include higher-or-der modes; this produced no significant improvementin the fits. The ability to operate the regenerativeamplifier over a broad range of pump powers and repe-tition rates is due to the negligible thermal lensing andbirefringence of Nd:YLF. In Nd:YAG where thermallensing is more significant, care must be taken to avoidhigh energy density buildup within the resonator as

2530 APPLIED OPTICS / Vol. 30, No. 18 / 20 June 1991

I-In

zW

I

the pump power or repetition rate is varied. Conse-quently, there may be a need to readjust the cavityalignment for various operational conditions. Howev-er, the Nd:YLF system runs with constant mode struc-ture and stability over a large dynamic range and forextended intervals of time without the need forlengthy adjustments.

The Q-switching, pulse selection and cavity dump-ing are controlled by a single Pockels cell. Conse-quently, the choice of this electrooptic material willeffect the purity of the output and the maximum repe-tition rate. We have evaluated both LiNbO3 andKD*P crystals for optimum cw regenerative amplifieroperation. The crystal dimensions were 9 X 9 X 25mm. The LiNbO 3 was housed in a dry cell but theKD*P was housed in a specially designed INRAD cellwith a 100-,gm thick film of deuterated index matchingfluid, FC43. This design will minimize thermalblooming problems associated with the fluid. Ourcriteria for judging the materials' performance arebased on the extinction coefficient (contrast ratio),absorption and repetition rate achieved. KD*P is anattractive material because of its higher contrast ratioand relative immunity to acoustic noise at higher repe-tition rates. However, its high residual absorption(-2.5% single pass) overwhelms its benefits. In oursystem, the thermal blooming associated with the highabsorption became a significant problem at 400 Hzwhich resulted in the amplifiers' spatial mode becom-ing unstable. KD*P has not been observed to createsevere thermal problems in pulsed high gain regenera-tive amplifiers running at kHz rates, and at similarenergy outputs.16 This is a consequence of the smallernumber of passes in those amplifiers or the smalleraverage power loading per single output cycle. Theresidual absorption is also a source of degrading modequality. Ultimately, we found LiNbO3 to be the supe-rior material for our regenerative amplifier's overallperformance. Thermal blooming or lensing is negligi-ble in our system since the single-pass absorption is<0.5%. However, the LiNbO3 is limiting our system'smaximum repetition rate due to acoustic ringing.17

The mechanical stress on the LiNbO3 crystal as a re-sult of switching high voltages across it deforms thecrystal. This deformation and subsequent relaxationpropagate through the crystal in the form of acousticwaves. Initially, these waves drastically affected theperformance of our Pockels cell at repetition rates >1kHz. One solution to this problem is to use a largercrystal. Another more practical one is to couple thesewaves out of the crystal by using a material which hassimilar acoustic impedence as LiNbO3 as first reportedby Dawes and Sceats,18 whose technique we followed.The ground electrode is bonded to a lead block toachieve efficient removal of acoustic energy. Thisresulted in our laser producing 1 mJ/pulse at 3.5 kHz.Very recently, there has been a report' 9 of a moreefficient method of damping the acoustic waves inLiNbO3, involving the coupling of the acoustic wavesin a direction transverse to the applied electric fieldand resonator axis.

C~

I-Cl)zwI-

B - - -d-

I I I I I I I I I - >

TIME (20 nsec /division)

Fig. 6. Regenerative amplifier's output with vertical scale expand-ed taken with a fast photodiode. Label A is the amplifier's mainpulse (off scale) and labels B and C are secondary pulses; see text for

description.

V. Discussion

A magnified view of the output of the regenerativeamplifier will show that there are some leading andtrailing secondary pulses around the main pulse. Forexample, in Fig. 6, A is the main pulse, whereas B and Care the secondary pulses. Let us consider the reasonsfor their origin in detail. The finite extinction ratio ofthe polarizers and the limited contrast ratio of thePockels cell crystal allow some energy to leak out of the.cavity through the polarizer during the pulse builduptime. This leakage gives rise to a sequence of prema-ture output pulses (marked B in Fig. 6) separated bythe round-trip time in the amplifier's resonator. Ex-perimental verification was performed by the additionof a pulse selector in the input seed pulse train prior toinjection. This allowed selection of a single seed pulsefor injection into the amplifier. The result showed noimprovement in the contrast ratio of pulses marked Bto the main pulse A. It is worth noting that eventhough two polarizers are installed in the resonator,only one of them is effective in cavity dumping, there-by further decreasing the contrast ratio of the output.In our case, we observed a ratio of 100:1 between themain output and the largest leading pulse. In applica-tions where frequency doubling of the output is de-sired, a quadratic reduction of the leakage would beachieved by the nonlinear crystal. Additional cleanupof the fundamental can be done by using an extracavitypulse selector. A KD*P crystal (for better contrastratio) when used extracavity can be operated at higherrepetition rates than an intracavity one because of thelower average power dissipations external to the cavi-ty.

To understand the origin of pulses C, note that thereis a difference in the lengths of the oscillator and theamplifier cavities. Every pulse makes two passesthrough the amplifier's resonator before exiting.Since the two cavity lengths are not equal, there arealways two pulses present inside the cavity separatedby twice the cavity length difference at the moment thevoltages are switched on the Pockels cell. One of themis approaching the switching device for the first timewhile the other has made one round trip of the cavity.

20 June 1991 / Vol. 30, No. 18 / APPLIED OPTICS 2531

Even if the two cavity lengths are equal, these second-ary pulses are still present since the Pockels cell-polar-izer combination is not capable of completely rejectingthe pulses with wrong polarization. Here one mightnot be able to resolve the secondary pulses. Our stud-ies using KD*P as an intracavity Pockels cell materialdid not result in any substantial improvement in sup-pressing pulses C beyond the limit reached with theLiNbO3 cell. We believe that this is a consequence ofthe limited extinction ratio of our intracavity polariz-ers. Furthermore, the addition of an external polariz-er on the amplifier's output results in no change in thiscontrast ratio, since the pulses C and the main pulse Ahave the same polarization.

A final consideration in trapping the right seed pulseis the finite rise time associated with the high voltagepulse used to switch the Pockels cell crystal. Obvious-ly, this time has to be shorter than the interval betweentwo consecutive mode-locked seed pulses to achievesingle-pulse trapping. For example, in our system theoptical switching time is 6 ns which means that for thePockels cell to switch the polarization of the pulseseffectively, our regenerative amplifier's resonatorlength must be within ±2 ns of the oscillator. If thedifference in resonator lengths is larger than 2 ns, oneor both (depending on the difference) of the secondarypulses would arrive at the Pockels cell when it does nothave an exact quarter-wave or zero voltage. This re-sults in a poor contrast ratio. Since a successive pair ofinput pulses is necessary to produce this secondaryoutput, application of a pulse selector on the input sideof the amplifier greatly enhanced this contrast ratio.We observed that injection of a single seed pulse re-sulted in an improvement in this contrast ratio>1000:1. The only requirement is that the relativetiming of the two Pockels cells must be adjusted toobtain the best possible contrast ratio.

VI. ConclusionWe have reported a Nd:YLF regenerative amplifier

with a concave-convex resonator design that mini-mizes the energy density on critical intracavity opticalcomponents while maintaining excellent performance.This system operates at higher repetition rates (3.5kHz) than previously reported for YAG and YLF re-generative amplifiers. The energy output of our sys-tem is also higher as a consequence of our resonatordesign. The results of our studies on the contrast ratioand mode quality offers a general guideline for regen-erative amplifier design and operation. We have alsotried to contrast the differences in design consider-ations as they pertain to choice of materials and meth-od of pumping. Ultimately, the use of cw pumpedregenerative amplifiers should result in systems capa-ble of repetition rates >5 kHz and provide an excellentpump source for tunable short pulse amplifiers.

We would like to thank Philippe Bado, Martin G.Cohen, Stephen Janes, K. C. Lui, and Sten Tornegardfor helpful discussions concerning this work. This

research was carried out at Brookhaven National Lab-oratory under contract DE-AC02-76CH000016 withthe U.S. Department of Energy and supported, in part,by its Division of Chemical Sciences, Office of BasicEnergy Sciences.

Muhammad Saeed and Dalwoo Kim are on leavefrom the Physics Department of Louisiana State Uni-versity.

References1. W. H. Lowdermilk and J. E. Murray, "The Multipass Amplifier:

Theory and Numerical Analysis," J. Appl. Phys. 51, 2436-2444(1980).

2. W. H. Lowdermilk and J. E. Murray, "Nd:YAG RegenerativeAmplifier," J. Appl. Phys. 51, 3548-3555 (1980).

3. I. N. Duling, P. Bado, S. Williamson, G. Mourou, and T. Baer, "AStable Kilohertz Nd:YAG Regenerative Amplifier," in Techni-cal Digest, Conference on Lasers and Electro-Optics (OpticalSociety of America, Washington, D.C., 1984), paper PD3.

4. I. N. Duling III, T. Norris, T. Sizer II, P. Bado, and G. Mourou,"Kilohertz Synchronous Amplification of 85-Femtosecond Op-tical Pulses," J. Opt. Soc. Am. B 2, 616-618 (1985).

5. Y. J. Chang, C. Veas, and J. B. Hopkins, "Regenerative Amplifi-cations of Temporally Compressed Picosecond Pulses at 2 kHz,"Appl. Phys. Lett. 49, 1758-1760 (1986).

6. D. Strickland and G. Mourou, "Compression of AmplifiedChirped Optical Pulses," Opt. Commun. 56, 219-221 (1985).

7. V. J. Newell, F. W. Deeg, S. R. Greenfield, and M. D. Fayer,"Tunable Subpicosecond Dye Laser Amplified at 1 kHz by aCavity-Dumped, Q-Switched, and Mode-Locked Nd:YAG La-ser," J. Opt. Soc. Am. B 6, 257-263 (1989).

8. P. Bado, M. Bouvier, and J. S. Coe, "Nd:YLF Mode-LockedOscillator and Regenerative Amplifier," Opt. Lett. 12, 319-321(1987).

9. A. L. Harmer, A. Linz, and D. R. Gabbe, "Fluorescence of Nd inLithium Yttrium Fluoride," J. Phys. Chem. Solids 30, 1483-1491 (1969).

10. J. C. Postlewaite, J. B. Meirs, C. C. Reiner, and D. D. Dlott,"Picosecond Nd:YAG Regenerative Amplifier with Acoustoop-tic Injection and Electrooptic VFET Pulse Switchout," IEEE J.Quantum Electron. 24, 411-417 (1988).

11. A. E. Siegman, Lasers (University Press, Edinburgh, 1986).12. H. Vanherzeele, "Thermal Lensing Measurements and Com-

pensation in a Continuous-Wave Mode-Locked Nd:YLF La-ser," Opt. Lett. 13, 369-371 (1988).

13. P. Bado and M. Bouvier, "Multikilohertz Pockels Cell Driver,"Rev. of Sci. Instrum. 56, 1744-1745 (1985).

14. H. Kogelnik and T. Li, "Laser Beams and Resonators," Appl.Opt. 5, 1550-1567 (1966).

15. K. C. Lui, affiliation Quantronix Corporation; private communi-cation.

16. R. Olson, affiliation Continuum Corporation; private communi-cation.

17. R. P. Hilberg and W. R. Hook, "Transient Elastooptic Effectsand Q-Switching Performance in Lithium Niobate and KD*PPockels Cells," Appl. Opt. 9, 1939-1940 (1970).

18. J. M. Dawes and M. G. Sceats, "A High Repetition Rate Pico-Synchronous Nd:YAG Laser," Opt. Commun. 65, 275-278(1988).

19. J. Sweetser, I. A. Walmsley, D. Wang, P. Basseras, and R. J. D.Miller, "Operation of a Nd:YLF Regenerative Amplifier at Fre-quencies Greater than 6 kHz," in Conference on Lasers andElectro-Optics, Vol. 11 (Optical Society of America, Washing-ton, DC, 1989) paper PDO1.

2532 APPLIED OPTICS / Vol. 30, No. 18 / 20 June 1991