A quasi-continuous-wave deep ultraviolet laser source ... - IEEE Xplore

1160

IEEE JOURNAL OF QUANTUM ELECTRONICS, VOL. 39, NO. 9, SEPTEMBER 2003

A Quasi-Continuous-Wave Deep Ultraviolet Laser Source Kevin F. Wall, Joseph S. Smucz, Bhabana Pati, Yelena Isyanova, Peter F. Moulton, Senior Member, IEEE, and Jeffrey G. Manni, Member, IEEE

Abstract—We have developed a quasi-CW, deep ultraviolet source producing 250 mW of 205-nm radiation. The source consists of a 100-MHz, mode-locked, 15-W, 1047-nm master-oscillator/power-amplifier, a synchronously pumped optical parametric oscillator, and three nonlinear conversion stages. Index Terms—Deep-UV, harmonic generation, solid-state, synchronously pumped optical parametric oscillator (OPO).

I. INTRODUCTION

T

HE DESIRE of the semiconductor industry to comply with Moore’s law (the doubling of transistors per integrated circuit every 18 months) using photolithography has spurred the development of shorter-wavelength sources. Currently, the state-of-the-art photolithographic methods use pulsed excimer lasers at deep ultraviolet (DUV) wavelengths of 193 and 157 nm and repetition rates of 1 Hz or greater. While pulsed sources are adequate for most applications, for some applications (such as mask generation) it is desirable to have a DUV source that is either continuous wave (CW) or a quasi-CW, the latter with a high enough repetition rate that it can be considered as a CW source for the process. With Q-switched and pulsed solid-state laser sources, a variety of methods have been used to generate DUV wavelengths. Several approaches have involved the use of tunable solid-state lasers. One employs fourth-harmonic generation from a Q-switched alexandrite laser at 772 nm, by generation of the third harmonic, and sum-frequency mixing of the third harmonic with the fundamental to produce 193-nm radiation [1]. With a repetition rate of 20–25 Hz, this technique has generated 25 mW (1 mJ/pulse) with efficiencies of 0.5%.1 In another scheme, the 205–230-nm fourth harmonic from ) a gain-switched, nanosecond-pulse Ti:sapphire (Ti: laser was mixed with the output of an optical parametric oscillator (OPO) in the 1700–2500-nm region, in the crystal ) to generate as much as 1.8 mJ at 195.4 nm, at LBO ( a 10-Hz pulse rate [2]. At higher rates, a commercial product2

Manuscript received December 23, 2002; revised May 8, 2003. K. F. Wall, B. Pati, Y. Isyanova, and P. F. Moulton are with Q-Peak, Inc., Bedford, MA 01730 USA (e-mail: [email protected]). J. S. Smucz was with with Q-Peak, Inc., Bedford, MA 01730 USA. He is now with Coherent-DEOS, Bloomfield, CT 06002 USA. J. G. Manni is with JGM Associates, Inc., Burlington, MA 01830 USA (e-mail: [email protected]). Digital Object Identifier 10.1109/JQE.2003.816088 1A commercial version of this (PAL/PRO-UV) is available from Light Age, Inc. (Somerset, NJ).

employs the fourth harmonic of a 5-kHz, laser-pumped of power at 193 nm. Ti:sapphire laser to generate Fourth-harmonic generation of a Q-switched Cr:LiSAF laser has generated 40 mJ of energy at 215 nm, with a 2-Hz pulse rate [3]. Finally, a frequency-tripled alexandrite laser has been color center laser in sum frequency mixed with a LiF: ) to produce tunable UV radiation from CLBO ( 194 to 204 nm [4]. Fixed-wavelength solid-state lasers have also been the basis for DUV sources. Fifth-harmonic generation of a Q-switched Nd:YAG laser using CLBO has demonstrated 200-mJ pulses at efficiency at a 10-Hz repetition rate [5]. 213 nm with , Mead et al. Operating at pulse repetition rates of demonstrated the generation of 193-nm light by sum frequency mixing the fifth harmonic of a Nd:YAG laser with the output of a periodically poled lithium niobate OPO [6]. Umemura and Kato demonstrated that wavelengths as low as 185 nm could be generated using CLBO to mix the fifth harmonic of Nd:YAG laser ) OPO [7]. A commercial with the output of KTP ( of 193-nm energy at a 10–20-Hz product3 generates rate by mixing the Nd:YAG fifth harmonic with an OPO. Fifth-harmonic generation of an Nd:YAG laser continued to advance with the demonstration of 0.63 W with an efficiency of 7% at a repetition rate of 1 kHz [8]. Even higher powers were demonstrated by Yap et al. where 4.0 W of 213-nm radiation were generated at 100-Hz repetition rates [9]. By mixing 2 W of the fifth harmonic with the output of a potassium titanyl arsenate (KTA) OPO, the same group demonstrated 200 mW of average power at 192 nm, at a 100-Hz repetition rate [10]. A slightly more complicated geometry has used a Q-switched, 5-kHz, single-frequency Nd:YLF laser to generate the third harmonic [11]–[13]. Sum-frequency generation with the third harmonic and the output of a single-frequency, Ti:sapphire laser generated 237–243-nm radiation. When mixed with the Nd:YLF fundamental wavelength of 1047 nm, this yielded 1.5 W of 196.3-nm radiation. A remarkably efficient way of generating 193-nm radiation has been demonstrated with a system consisting of a distributed feedback (DFB) diode laser, a lithium niobate modulator and a fiber amplifier operating a 1547 nm [14]. Using a combination of LBO, beta-barium borate (BBO), and CLBO in five harmonic stages, the authors demonstrated 3 mW of power at a 1-kHz repetition rate with a conversion efficiency from the IR of 7.0%. Harmonic conversion of femtosecond pulses has been demonstrated by Petrov et al. [15]. Using a femtosecond 2The

Indigo DUV laser from Positive Light (Los Gatos, CA). Actinix, Soquel, CA.

3From

0018-9197/03$17.00 © 2003 IEEE

WALL et al.: A QUASI-CW DEEP ULTRAVIOLET LASER SOURCE

Ti: regenerative amplifier and the idler pulses of a traveling-wave parametric amplifier, the authors used sum frequency generation to yield wavelengths as short as 175 nm. True CW generation of the fourth harmonic of a Nd:YAG laser has been demonstrated by Oka et al., where 1.5 W of 266-nm was generated using an external cavity with an input of 2.9 W of 532-nm radiation [16]–[18]. A similar approach was used by Zanger et al., where 1.4 W of 266-nm radiation was generated with an input of 5 W of 532-nm radiation [19]. CW UV generation at shorter wavelengths has been reported by Fujii laser was doubled using et al., where a CW 746-nm Ti: an external cavity to 373-nm [20]. The 746- and 373-nm radiation were then sum frequency mixed in a second external cavity that was doubly resonant to yield 50 mW of CW 252-nm radiation. Although not all solid state, CW generation near 194 nm has been demonstrated by Watanabe et al. [21] and Berkeland et al. [22]. Mode-locked lasers sources have also been frequency converted to the UV. Nebel and Beigang demonstrated external frelaser to 210 nm quency conversion of a mode-locked Ti: with 10 mW of time-averaged power [23], [24]. Tidwell et al. demonstrated efficient conversion of a mode-locked Nd:YLF laser to the fourth harmonic by using a resonant ring geometry [25]. Kohler et al. demonstrated fifth-harmonic conversion of a laser to 1.2 W at 213 nm using 27.4-W, mode-locked Nd: CLBO [26]. In this paper, we report on the development of a quasi-CW, deep ultraviolet laser source that produces of 205-nm radiation. The source consists of a 100-MHz mode-locked, 15-W, 1047-nm, master-oscillator/power-amplifier, a synchronously pumped optical parametric oscillator, and three nonlinear conversion stages. In the following, we will describe the general approach that we followed and the rationale for choosing this approach. Next, we will describe the amplification of a mode-locked Nd:YLF laser to 15 W of time-averaged power with high beam quality. Conversion of the 1047-nm radiation to the fourth harmonic (262-nm) will be discussed. The design and operation of a synchronously pumped LBO OPO will be described in detail and the results of sum-frequency mixing the signal wavelength of the OPO and the fourth harmonic will be presented.

II. APPROACH The approach that we chose to generate quasi-CW deep UV wavelengths was to start with a near-IR, 1047-nm, mode-locked, Nd:YLF laser. True CW generation of deep UV radiation has the disadvantage of not having sufficient intensity for efficient harmonic conversion unless actively stabilized resonant cavities are used. A mode-locked source at a high repetition rate has the advantage of being effectively CW for many processes while possessing sufficient intensity to allow efficient nonlinear conversion. As near-IR laser technology is relatively mature, high average powers with good beam quality can be readily achieved in a variety of Nd-doped laser media. Nd:YLF was chosen as the laser media as the pulsewidths of mode-locked Nd:YLF lasers are generally shorter than other Nd hosts and high-gain,

1161

Fig. 1. Design schematic. The indicated wavelengths are in nanometers. The final generated wavelength is 205 nm. TABLE I NONLINEAR CRYSTAL DATA FOR OUR DESIGN

high-beam-quality Nd:YLF amplifiers are available [27]. Another approach that is possible is to start with a Ti:sapphire laser operating at 820-nm, frequency double this to 410 nm, and use BBO to quadruple the Ti:sapphire laser to 205 nm. While this approach is relatively straightforward, for an all solid-state system one still needs to start with a frequency doubled near-IR pump source to pump the Ti:sapphire crystal; effectively a total of four stages of frequency conversion. Also, two-photon absorption of BBO at 205 nm may not compatible with quasi-CW harmonic generation, as the highest deep-UV power that we are aware of is 0.5 W at 213 nm generated with a Q-switched, 7-kHz repetition rate Nd:YAG laser [28]. The next consideration in our approach was to minimize the number of frequency conversion stages (particularly sum-frequency mixing) and to use noncritical phase matching (NCPM) wherever possible. NCPM allows the use of long interactions lengths for efficient conversion of low intensity sources. Fig. 1 shows the approach taken in our work and Table I summarizes the nonlinear crystal data for our design. Our design begins by frequency-doubling the 1047-nm master-oscillator/power-amplifier using NCPM in LBO (SHG1). The resulting 524-nm radiation is split, with one beam used to pump an OPO and the second doubled again using a critically phase matched CLBO crystal (SHG2). The signal wavelength of a NCPM LBO OPO at 942-nm is combined with the output of the fourth harmonic by means of a dichroic mirror and sum-frequency mixed in a second CLBO crystal using NCPM to generate 205-nm radiation. To summarize the design, there are four nonlinear conversion stages, of which three are noncritically phase matched. An important consideration for the practical use of UV radiation generated by the use of mode-locked pulses is the bandwidth which, due to the high dispersion of optics in the UV, can result in complex achromatic optical design. If we assume that the IR source has a time-bandwidth product of no greater than 0.6 and a pulsewidth of 5 ps, the bandwidth is 438 pm. When

1162


Fig. 2. Design layout. SHG: second-harmonic generation. OPO: optical parametric oscillator. 4HG: fourth harmonic generation. SFM: sum frequency mixing.

frequency doubled, the resulting pulses will have widths that shorter in the high conversion regime. This results in a are bandwidth of 155 pm again, assuming a time-bandwidth product of 0.6. For the fourth harmonic the bandwidth is 55 pm, and for the final mixing stage the bandwidth is 48 pm. This should be viewed as an upper limit for the bandwidth, as it is unlikely that the high conversion regime pulse compression will apply to the stages after the second-harmonic stage. The implementation of the design shown in Fig. 1 is shown schematically in Fig. 2. As the OPO is synchronously pumped, its cavity length equals that of the 1047-nm master oscillator. A delay line is used to synchronize the pulses from the fourth harmonic leg with those of the OPO leg in the final mixing stage. III. THE NEAR-IR SOURCE The starting point for our near-IR source is a Time-Bandwidth Products (TBWP), Inc. (Zurich, Switzerland) oscillator that is a modified version of their standard GE-100 product. The GE-100 oscillator is a diode-pumped, passively mode-locked Nd:YLF laser producing 5-ps pulses with a repetition rate of 100 MHz at a wavelength of 1047 nm. The primary modification of the oscillator was the use of a higher power pump diode laser to increase the output power from the standard 100 mW level to 700 mW. The output beam is not perfectly Gaussian, with correlation constants of 0.95 and 0.90 for the horizontal and vertical axes, respectively. We measured the beam quality using a Co-101 and found it to be herent ModeMaster and a Spiricon 1.08 and 1.06 for the horizontal and vertical axes, respectively. We also measured the pulsewidth of the TBWP oscillator using an autocorrelator and assuming a Gaussian pulse shape, we determined that the pulsewidth was 4.5 ps. To amplify the output power of the mode-locked oscillator, we used multipass slab (MPS) gain modules [27]. In the MPS

Fig. 3. MPS amplifier. Efficient extraction is achieved by multipassing the signal beam through the sheet of gain formed by two diode laser bars.

module, a 2.8-cm-long Nd:YLF crystal is transversely pumped by a pair of 1-cm-long diode-laser bars, as shown in Fig. 3. The diode-laser bars are coupled to the gain element through a single cylindrical lens attached directly to each bar package. These lenses minimize the divergence of the pump light in the plane perpendicular to the linear emitter. The diode lasers are offset on opposite sides of the Nd:YLF crystal to create a sheet of gain in the crystal. The pump faces of the crystal have segmented dielectric coatings (anti-reflection/high reflection) to allow double-pass pump absorption. Power is extracted efficiently and high gain per pass is obtained by passing the signal laser beam five times though the gain sheet that is produced in the slab. By proper choice of the mode size, one can extract essentially all of the power in mode. The relatively uniform pump power density in a the crystal minimizes excess heating and loss due to upconver-


Fig. 4. Beam propagation model used to design the preamplifier and power amplifier. The solid line represents the beam size in the horizontal plane and the dashed line, the vertical. The lightly dashed lines at the position of the preamplifier and amplifier represent the desired beam sizes in the both dimensions. The solid vertical lines represent the positions of lenses.

sion. The MPS geometry is unique in providing high gain while avoiding many of the difficulties encountered in end-pumped laser media, such as: 1) achieving good overlap between the or low-order laser mode and 2) avoiding pump and a rod fracture at high pump powers. Since YLF is a birefringent material, stress-induced birefringence (which can cause deleterious depolarization effects) is a second-order effect. Not shown in the figure are heat sinks, which remove heat from the large faces of the Nd:YLF slab. In addition to effective heat removal, the properties of YLF lead to very weak bulk thermal lensing. With Nd:YLF pumped with two 20-W bars, we observed negligible thermal lensing in the vertical direction (i.e., perpendicular to the pump beam) and about 1 diopter of lensing in the horizontal direction (i.e., parallel to the pump beam). The surface stress, which ultimately limits the amount of pump power, for a slab, can be lower than for a rod geometry if the aspect ratio of the slab is greater than 2 [29]. For our present design, the aspect ratio is 2.4. We designed a double-pass pre-amplifier using the measured output beam characteristics of the mode-locked oscillator. A power amplifier further amplified the output of the pre-amplifier. The overall amplifier design is based upon paraxial propas a paramagation of Gaussian beams. The model uses eter to account for nonideal beams. Fig. 4 shows the calculated propagation of the mode-locked oscillator output through beam shaping optics, a double-pass pre-amplifier, and a power amplifier. The solid line represents the beam size in the horizontal plane and the dashed line, the vertical. The lightly dashed rectangles at the position of the pre-amplifier and power amplifier represent the desired beam sizes in the both dimensions. The solid vertical lines represent the positions of lenses. Fig. 5 shows a schematic of the layout of the amplifier system. The first three lenses, V1, H1, and H2, form a beam waist that matches the height of the gain sheet in the amplifier modules in the vertical dimension (perpendicular to the large faces of the Nd:YLF slab) and which is about four times larger in the horizontal dimension. S1 and S2 form a pair of relay lenses that pro-

1163

vide space for a Faraday rotator and relay the beam waist formed by lenses V1, H1, and H2, to the center of the Nd:YLF pre-amplifier slab. The beam is then relayed from the center of the pre-amplifier back to the center of the pre-amplifier using lens S3 and mirror M1. Lens H3 is a weak (2-m focal length) cylindrical lens to partially compensate for the thermal lensing of the pre-amplifier. After the second set of five passes through the pre-amplifier gain module, the beam is rejected by the Faraday isolator and sent to the power amplifier gain module. In this case, lenses S2 and S3 form a pair of relay lenses that re-image the beam waist from the center of the pre-amplifier to the center of the power amplifier. After a set of five passes through the power amplifier, a cylindrical lens is used to compensate for astigmatism and circularize the output beam. The output power of the double-pass pre-amplifier is shown in Fig. 6. The input signal power was fixed at 0.45 W and the amplifier pump power was varied. At 30 W of pump power, the amplifier output power is linear with the pump power indicating that the amplifier is becoming saturated. Typically, the output power of the double-pass pre-amplifier is 6.7 W at 40 W of pump power for an overall signal gain of 15. The output of the pre-amplifier was further amplified to 15 – 16 W in the power amplifier when the latter was pumped with 40 W of diode power. The signal beam was transmitted through the power amplifier for five passes as the input intensity is on ). A the order of the Nd:YLF saturation intensity (2 near-field beam profile, prior to cylindrical lens V2, is shown in Fig. 7. The angle between the semi-major axis and horizontal originates with the output of the oscillator. We measured the beam quality to be 1.24 and 1.14 for the horizontal and vertical axes, respectively. The peak power of our 1047-nm source is 30 kW, assuming an average power of 15 W and 5-ps pulses at a repetition rate of 100 MHz. IV. SECOND-HARMONIC GENERATION We frequency-doubled the 1047-nm MOPA system using Type I NCPM in a 20-mm long lithium triborate (LBO) crystal. The crystal was placed in a windowless oven and heated to a temperature of 170 . The crystal temperature was stabilized . The 1047-nm fundamental beam was focussed to into the LBO crystal with a spherical lens. From the measured beam parameters of the 1.047-nm beam, the calculated beam ) at the focus of the lens are 84 and 72 for the radii ( (horizontal) and (vertical) dimensions, respectively. We used a single spherical lens to recollimate the output beam from the LBO crystal. With 15.0 W of fundamental power, 7.5 W of 524-nm power was generated. A beam profile of the second-harmonic beam is shown in Fig. 8, where the vertical and horizontal axes are swapped. This figure shows a small “satellite” which originates in the fundamental beam. Maximizing the power of the MOPA tends to increase the size of the satellite peak. Careful adjustment of the horizontal alignment of the power amplifier stage minimizes the size of the satellite. The penalty for this optimization is that the MOPA output power drops from 16 to 15 W. When propof the 524-nm beam is 1.03 and 1.07 erly optimized, the for the horizontal and vertical axes, respectively. From the mea-

1164


Fig. 5. Top view of the physical setup for the preamplifier and power amplifier. Elements whose labels begin with V and H are cylindrical lenses in the vertical and horizontal planes, respectively. Elements labeled S are spherical lenses.

Fig. 6. Preamplifier output power versus pump power. The input signal intensity was fixed at 0.45 W and the pump power was varied.

Fig. 8. Second-harmonic beam profile. The vertical and horizontal axes are swapped in this figure. The lines through the false-color representation of the beam profile indicate the positions of the vertical and horizontal cross sections shown at the left and right, respectively.

smaller than the fundamental beam, which compares well with the ratio of second harmonic to fundamental beam radii expected for second-harmonic generation in the limit of low conversion (0.707). We measured the 524-nm pulsewidth using an autocorrelator. The autocorrelator signal is shown in Fig. 9 and shows a symmetric Gaussian-like autocorrelation. Assuming Gaussian pulse profiles, the calculated pulsewidth is 4.8 ps for the second harmonic. This should be compared with 4.5 ps measured for the 1047-nm fundamental. This may be indicative of a small amount of pulse stretching due to group velocity mismatch between the fundamental and second harmonic. Fig. 7.

Power amplifier output beam profile.

sured beam parameters, we have calculated the 524-nm beam , for the horizontal waist in the LBO crystal to be 57 and 49 and vertical dimensions, respectively. These radii are 0.7 times

V. THE SYNCHRONOUSLY PUMPED OPO Using 2.5 W from the 524-nm beam, we synchronously pumped a LBO OPO. The 20-mm-long LBO crystal was to mounted in a windowless oven and heated to 165


1165

Fig. 9. Second-harmonic autocorrelation signal. The calculated second-harmonic pulsewidth is 4.8 ps assuming Gaussian pulse profiles.

achieve noncritical phase matching. The LBO crystal was anti-reflection coated at 950 and 524 nm, and to a lesser extent at the idler wavelength of 1166 nm. The OPO cavity was a linear, standing-wave cavity that was chosen to facilitate alignment rather than minimize intracavity loss. All of the cavity mirrors were coated to be highly reflective at the signal wavelength and transmissive for the pump and idler wavelengths. This yielded single-pass pumping of the LBO crystal and a singly resonant OPO. The LBO crystal was positioned at a waist formed between two 15-cm radii mirrors and the remainder of the cavity was formed with flat mirrors. The angle between the arm of the OPO containing the LBO crystal and the next leg of the OPO cavity was 30 to accommodate the size of the LBO oven. For the radii of the curved mirrors that we used, the cavity was stabilized for this angle by the inclusion of a vertical cylindrical lens in the OPO arm. The calcuradius formed by the two concave mirrors is 25 lated for the and axes, respectively. The output couand 24 pler had a transmission of 30% at 950 nm and was mounted on a PZT and translation stage. The pump beam was focused to a for the and axes, respeccalculated radius of 28 and 27 tively. Fig. 10 is a false-color representation of the OPO output beam profile. At the particular position of the CCD camera, the ratio of the beam radii was 0.978. The correlation coefficients for Gaussian fits to the profiles shown below and to the left of the figure are 0.983 and 0.967 for the and axes, respectively. With 2.5 Wof pump power the maximum output power of the OPO was 500 W. The output power versus input pump power of the OPO is shown in Fig. 11 for a wavelength of 970 nm. Only the signal power is measured as the idler is transmitted through the cavity mirrors. The slope efficiency and threshold calculated from a linear fit to the data are 29% and 0.67 W, respectively. The optical to optical efficiency of the OPO is 16% for conversion of the pump to signal and 29% if the idler power is included. With an output coupler transmission of 9.5%, the maximum output power of the signal wave of the OPO was 680 mW. We measured the pulsewidth of the OPO with an autocorrelator and the resultant signal is shown in Fig. 12. Assuming

Fig. 10. Synchronously pumped LBO OPO beam profile. The lines through the false-color representation of the beam profile indicate the positions of the vertical and horizontal cross sections shown at the left and right, respectively.

Fig. 11. Synchronously pumped LBO OPO output power versus input pump power. Only the signal power is measured and the wavelength is 970 nm. The solid line is a linear fit to the data (solid circles).

Fig. 12. Synchronously pumped LBO OPO autocorrelation signal. The calculated pulsewidth is 3.8 ps assuming Gaussian pulse profiles.

Gaussian pulse profiles, the calculated pulsewidth is 3.9 ps. When the OPO output power was maximized, the spectrum

1166


Fig. 13. Signal-wave interferogram of the synchronously pumped OPO. Ten interferograms taken 1-s apart are shown. Most of the interferogram are unresolved using an etalon with a free spectral range of 4.2 nm.

Fig. 14. Wavelength of the synchronously pumped LBO OPO as a function of cavity detuning. The tuning rate of the OPO was 0.9 = .

nm m

as displayed by a Burleigh pulsed wavemeter could not be resolved properly. Fig. 13 shows ten interferograms taken 1-s apart when an output coupler with a transmission of 2.4% was used. Occasionally the spectral width could be resolved by the wavemeter (interferogram #5), but most of the time the spectral width of the OPO exceeded the free spectral range of the etalon used in the Burleigh pulsed wavemeter (4.2 nm). For a 3.9-ps pulse, the measured time-bandwidth product is 5.2, which is more than ten times the transform limit. When the cavity length was detuned from the optimum length, the spectrum narrowed to 600 pm. Misalignment of the cavity also reduced the linewidth of the OPO. As the cavity length is detuned the center wavelength of the OPO shifts as illustrated in Fig. 14. The change in wavelength with cavity . The detuning is roughly linear with a slope of 0.9 output power of the OPO as a function of cavity detuning is shown in Fig. 15. The assumption is that self-phase modulation due to high intracavity intensity causes broadening of the linewidth [30] and reduction in the intracavity intensity due to misalignment or cavity length detuning (both of which

Fig. 15. Output power of the synchronously pumped OPO as a function of cavity detuning.

can reduce the intracavity intensity) reduces the self-phase modulation. To reduce the linewidth of the synchronously pumped OPO, we added group velocity dispersion (GVD) compensating prisms to the cavity. Two isosceles Brewster SF10 prisms were inserted into the long arm of the OPO prior to the output coupler. The distance between the prisms was 25 cm and the insertion depth was varied from 1–6 mm. The GVD compensating prisms helped reduce the linewidth (occasionally to 390 pm), but when the OPO was aligned for maximum output power, the spectrum was similar to that shown in Fig. 13 and could not be resolved. For the sum-frequency mixing stage that followed, the acceptance bandwidth was 252 pm-cm for an OPO wavelength of 950 nm. With a 4-mm-long crystal, this implies that the OPO bandwidth cannot exceed 630 pm. The exact bandwidth determines the OPO center-wavelength stability requirement. If, for example, the OPO bandwidth were 630 pm, the center wavelength would need to be stabilized to approximately one tenth of this bandwidth (or better) or a cavity length stabilization of 0.07 . If the linewidth of the OPO were considerably smaller than 630 pm, the requirements on cavity stability would be relaxed . to 0.7 To lower the intracavity fluence, and the effects of self-phase modulation, we reduced the reflectivity of the output coupler. With a 55% reflectivity output coupler, the spectrum was still not guaranteed to be a narrow single peak. The narrowest linewidth obtained under these circumstances was 400 pm. To reduce the OPO linewidth further, we placed a coated (80% reflectivity), 200- m-thick, fused-silica etalon into the OPO cavity. The etalon immediately narrowed and stabilized the output wavelength of the OPO. The spectral width of the OPO emission was reduced to 300 pm. The wavelength of the OPO observed with the etalon in place was stable over hours, in contrast to the stability when there was no etalon in the OPO cavity (minutes). Finally, trying to change the OPO wavelength by changing the cavity length with the PZT had no effect. Tilting the etalon also had no effect on the wavelength of the OPO. The reduction in OPO output power with the etalon was 20% (400–500 mW) and the output wavelength of the OPO was 942.36 nm.


Fig. 16. Fourth harmonic generation efficiency. The linear efficiency versus input intensity indicates that we are not saturating the conversion efficiency for these input intensities.

VI. FOURTH-HARMONIC GENERATION For both the fourth-harmonic stage and the final sum-frequency mixing stage, we used CLBO as the nonlinear crystal. CLBO is a relatively hygroscopic crystal whose hydrates will form a stress-induced lens in the crystal [31]. Further exposure to water vapor leads to fogging/cracking. Earlier work with CLBO indicated that heating the crystals to a temperature of 150 would eliminate these problems even in the presence of ambient humidity []. We found that our CLBO crystals fogged even when heated to temperatures of 160 and exposed to ambient humidity over the course of weeks. We designed heaters that could maintain our CLBO crystals and allow purging with dry niat temperatures above 160 trogen gas and used windows to seal the ovens. Under these conditions, the CLBO crystals could be maintained indefinitely without lensing, cracking, or fogging. The CLBO crystal was from Kogakugiken (Atsugi, Japan) and the uncoated crystal for fourth-harmonic generation had dimensions of 5 5 10 mm. The angle of the crystal cut was , . The oven . The input window for the oven temperature was set to was fused silica and was anti-reflection (AR) coated for 524 nm. The output window was and was broadband AR coated for 193–248 nm. (The broadband UV AR coating has low reflection loss at 262 nm as well.) With 4.6 W of 524-nm power incident on the fourth harmonic CLBO crystal, we generated 660 mW of 262-nm light. Fig. 16 is a plot of the fourth-harmonic efficiency as a function of input intensity. The linear efficiency versus input intensity indicates that the conversion efficiency was not saturated for these input intensities. The light was focussed into the CLBO crystal using a cylindrical lens to focus tightly in the insensitive direction of the crystal. From beam-propagation calculations, the estimated for the horizontal spot radii at the focus were 1053 and 18 and vertical axes, respectively. Attempts to improve upon the efficiency included shortening the cylindrical focussing lens focal length and adding a cylindrical telescope prior to the focussing lens. None of these efforts resulted in increased output power. VII. THE FINAL MIXING STAGE The final nonlinear stage in the system mixed the output of the OPO with the fourth- harmonic radiation to generate

1167

Fig. 17. Phase-matching angle of sum frequency mixing of the OPO wavelength with 262-nm radiation. The curves are calculated for different temperatures and the solid circle is our experimental result. The data point corresponds to our observation of noncritical phase matching at a temperature of 175 C.

205-nm radiation. Calculations showing that NCPM would be possible in CLBO were the basis for the design of the mixing stage. Fig. 17 plots our calculations, showing the phase-matching angle for sum-frequency mixing of 262-nm radiation in CLBO as a function of the OPO wavelength, with crystal temperature as a parameter. We used Sellmeier coefficients from the work of Umemura et al. [33] . In the experimental work, before the radiation was combined with a dichroic beam-splitter, both the OPO and 262-nm beams were individually expanded with 2 telescopes and focused with single spherical plano-convex lenses. At the position that corresponds with the center of the CLBO oven, the beam radii, measured by transmission through pinholes, were 80 and 38 for the 942- and 262-nm radiation, respectively. The delay line was adjusted such that the 942- and 262-nm pulses overlapped at the focus of the beams in the CLBO crystal. windows that The oven for the CLBO crystal had were broadband AR coated for 193–248 nm. The crystal for this stage was also from Kogukugiken and was 5 5 10 and . We mm in size, uncoated, and cut at chose a crystal length that exceeded the bandwidth acceptance (resulting in lower efficiency) to try to avoid surface damage to the crystal by lowering the intensity at the input and output faces while having high intensity in the center of the crystal. In our initial experiments, based upon the calculated noncritical phase matching angle, critical phase matching was achieved and the exterior angle that at a CLBO temperature of 175 the incident beam made with the optic axis was estimated to be 85 or an internal angle of 86.7 . Fig. 17 shows this point compared to the calculated phase-matching condition, showing a general, but not exact agreement with the calculations. Under these conditions, 225 mW of 205-nm radiation was generated. However, the beam quality was poor. The beam appeared to be a single spot 30 cm from the CLBO crystal but broke into two spots in the vertical plane at a distance of 2 m. It was observed that if the 262-nm power was reduced, the beam quality improved. In later experiments, we were able 3 nm) and to change the OPO wavelength slightly (by achieved true NCPM. The beam quality was greatly improved

1168


and was roughly Gaussian in both directions (albeit with a slightly distorted beam profile). While as much as 250 mW of 205-nm power was generated, the stability over time was poor. We observed that the power would decrease over the course of several minutes. Adjusting the angle of the nonlinear crystal would restore the output power, which suggests that heating of the CLBO crystal due to absorption of either the 262- or 205-nm light caused phase mismatch in the harmonic generation. Further experiments to elucidate the cause of the power decrease were not conducted. Additional work is needed to obtain reliable long-term UV generation with CLBO. VIII. CONCLUSIONS We have demonstrated that using multi-pass slab gain modules, a 100-MHz, mode-locked, 1047-nm Nd:YLF oscillator with an output power of between 600–700 mW can be amplified with high beam quality to 15 W. Harmonic conversion of the 1047-nm radiation can be accomplished with 50% conversion efficiency using noncritically phase matched LBO. Using 2.5 W of the 524-nm radiation, we synchronously pumped a noncritically phase matched LBO OPO to produce 400 mW of narrow-band (300 pm) radiation at 942 nm. Using 4.6 W of 524 radiation, we were able to generate 600 mW of 262-nm radiation with 14% efficiency. Mixing of the 262- and 942-nm light in a noncritically phase-matched CLBO crystal resulted in 250 mW of 205–nm radiation with 42% conversion of the 262-nm light to 205 nm. The overall efficiency of the system in converting 1047-nm radiation to quasi-CW, 205 nm was 2%. ACKNOWLEDGMENT The authors acknowledge the support of Etec, an Applied Materials Company, and NIST. REFERENCES [1] W.-B Yan, T. F. Steckroat, R. A. Frost, J. C. Walling, and D. F. Heller, “An all-solid-state, deep-UV laser source at 193 nm,” in Conf. Lasers and Electro-Optics, vol. 11 , 1997 OSA Tech. Dig. Series, Washington, DC, 1997, p. 485. [2] G. A. Rines, H. H. Zenzie, R. A. Schwarz, Y. Isyanova, and P. F. Moulton, “Nonlinear conversion of Ti:sapphire laser wavelengths,” IEEE J. Select. Topics Quantum Electron., vol. 1, pp. 50–57, Apr. 1995. [3] H. H. Zenzie and Y. Isyanova, “High-energy, high-efficiency harmonic generation from a Cr:LiSAF laser system,” Opt. Lett., vol. 20, pp. 169–171, Jan. 1995. [4] S. B. Mirov, V. V. Fedorov, B. Boczar, R. Frost, and B. Pryor, “All-solidstate laser system tunable in deep ultraviolet based on sum-frequency generation in CLBO,” Opt. Commun., vol. 198, pp. 403–406, Nov. 2001. [5] Y. K. Yap, M. Inagaki, S. Nakajima, Y. Mori, and T. Sasaki, “High-power fourth- and fifth harmonic generation of a Nd:YAG laser by means of CsLiB O ,” Opt. Lett., vol. 21, pp. 1348–1350, Sept. 1996. [6] R. Mead and C. E. Hamilton, “Solid-state lasers for 193 nm photolithography,” in Proc. 3rd Int. Symp. 193 nm Lithography, Jun. 1997, pp. 28–30. [7] N. Umemura and K. Kato, “Ultraviolet generation tunable to 0.185 m in CsLiB O ,” Appl. Opt., vol. 36, pp. 6794–6796, Sept. 1997. [8] U. Stamm, W. Zschocke, T. Schroder, N. Deutsch, and D. Basting, “High efficiency UV-conversion of a 1 kHz diode-pumped Nd:YAG laser system,” in OSA Trends in Optics and Photonics, C. R. Pollack and W. R. Bosenberg, Eds. Washington, DC: Optical Society of America, 1997, vol. 10, Advanced Solid-State Lasers, pp. 7–9. [9] Y. K. Yap, Y. Mori, S. Haramura, A. Taguchi, T. Sasaki, K. Deki, Y. Ohsako, and M. Horigutchi, “High power all-solid-state ultraviolet laser by CLBO crystal,” in OSA Trends in Optics and Photonics, C. R. Pollack and W. R. Bosenberg, Eds. Washington, DC: Optical Society of America, 1997, vol. 10, Advanced Solid-State Lasers, pp. 10–13.

[10] K. Deki, J. Sakuma, Y. Ohsako, N. Kitatochi, T. Yokota, M. Horiguchi, Y. Mori, and T. Sasaki, “200 mW 192 nm generation using CsLiB O crystal,” in Postdeadline Papers, Conf. Lasers and Electro-optics ’98, Washington, DC, May 3–8, 1998, Paper CPD-4. [11] A. Finch, J. Sakuma, Y. Ohsako, K. Deki, M. Horiguchi, and T. Yokota, “All-solid-state, tunable, high power UV generation by sum-frequencymixing in CLBO,” in OSA Trends in Optics and Photonics, M. M. Fejer, H. Injeyan, and U. Keller, Eds. Washington, DC: Optical Society of America, 1999, vol. 26, Advanced Solid-State Lasers, pp. 70–73. [12] J. Sakuma, A. Finch, Y. Ohsako, K. Deki, M. Yoshino, M. Horiguchi, T. Yokota, Y. Mori, and T. Sasaki, “All-solid-state, 1-W, 5-kHz laser source below 200 nm,” in OSA Trends in Optics and Photonics, M. M. Fejer, H. Injeyan, and U. Keller, Eds. Washington, DC: Optical Society of America, 1999, vol. 26, Advanced Solid-State Lasers, pp. 89–92. [13] J. Sakuma, K. Deki, A. Finch, Y. Ohsako, and T. Yokota, “All-solid-state, high-power, deep-UV laser system based on cascaded sum-frequency mixing in CsLiB O crystals,” Appl. Opt., vol. 39, pp. 5505–5511, Oct. 2000. [14] T. Ohtsuki, H. Kitano, H. Kawai, and S. Owa, “Efficient 193 nm generation by eighth harmonic of Er -doped fiber amplifier,” in Proc. Conf. Lasers and Electro-Optics, OSA Tech. Dig. Series, Washington, DC, 1997, postdeadline paper CPD 9, pp. 17–18. [15] V. Petrov, F. Noack, F. Rotermund, M. Tanaka, and Y. Okada, “Sum-frequency generation of femtosecond pulses in CsLiB O ,” Appl. Opt., vol. 39, pp. 5076–5079, Sept. 2000. [16] M. Oka, N. Eguchi, H. Masuda, and S. Kubota, “All solid-state continuous wave 0.1 W ultraviolet laser,” in Proc. Conf. Lasers and ElectroOptics, 1992, vol. 12, OSA Tech. Dig. Series, Washington, DC, 1992, pp. 374–375. [17] M. Oka, N. Eguchi, L. Y. Liu, W. Wiechmann, and S. Kubota, “1-W cw 255-nm radiation from an external resonant cavity using a novel voicecoil-motor actuator,” in Conf. Lasers and Electro-Optics, 1992, OSA Tech. Dig. Series, Washington, DC, 1994, paper CThM1. [18] M. Oka, L. Y. Liu, W. Wiechmann, N. Eguchi, and S. Kubota, “All solidstate continuous wave frequency-quadrupled Nd:YAG laser,” IEEE J. Select. Topics Quantum Electron., vol. 1, pp. 859–866, Sept. 1995. [19] E. Zanger, R. Mueller, B. Liu, M. Koetteritzsch, and W. Gries, “CW diode-pumped all solid-state laser at 266 nm,” in Proc. Conf. Lasers and Electro-Optics, OSA Tech. Dig. Series, Washington, DC, 1999, pp. 88–89. [20] T. Fujii, H. Kumagai, K. Midorikawa, and M. Obara, “Development of a high-power deep-ultraviolet continuous-wave coherent light source for laser cooling of silicon atoms,” Opt. Lett., vol. 25, pp. 1457–1459, Oct. 2000. [21] M. Watanabe, K. Hayakawa, H. Imajo, R. Ohmukai, and S. Urabe, “Sum-frequency generation near 194 nm with an external cavity by simultaneous enhancement of frequency-stabilized fundamental lasers,” Jpn. J. Appl. Phys., vol. 33, pp. 1599–1602, 1994. [22] D. J. Berkeland, F. C. Cruz, and J. C. Berquist, “Sum-frequency generation of continuous-wave light at 194 nm,” Appl. Opt., vol. 36, pp. 4159–4162, 1997. [23] A. Nebel and R. Beigang, “External frequency conversion of cw modelocked Ti:Al O laser radiation,” Opt. Lett., vol. 16, pp. 1729–1731, Nov. 1991. , “Tunable picosecond pulses below 200 nm by external fre[24] quency conversion of cw modelocked Ti:Al O laser radiation,” Opt. Commun., vol. 94, pp. 369–372, Dec. 1992. [25] S. C. Tidwell, J. F. Seamans, D. D. Lowenthal, G. Matone, and G. Giordano, “Efficient high-power UV generation by use of a resonant ring driven by a cw mode-locked IR laser,” Opt. Lett., vol. 18, pp. 1517–1519, Sept. 1993. [26] B. Kohler, T. Andres, A. Nebel, and R. Wallenstein, “High-power, highrepetition-rate fourth and fifth harmonic generation of a cw mode-locked Nd:YVO laser,” in Proc. Conf. Lasers and Electro-optics, OSA Tech. Dig. Series, Washington, DC, 2000, pp. 142–143. [27] K. J. Snell, D. Lee, K. F. Wall, and P. F. Moulton, “Diode-pumped, high-power CW and modelocked Nd:YLF lasers,” in OSA Trends in Optics and Photonics, H. Injeyan, U. Keller, and C. Marshall, Eds. Washington, DC, 2000, vol. 34, Advanced Solid-State Lasers, p. 55. [28] H. Masuda, H. Kikuchi, H. Mori, K. Kaneko, M. Oka, S. Kubata, and J. Alexander, “Single frequency 0.5 W generation at 213 nm from an injection-seeded, diode-pumped, high-repetition-rate, Q-switched Nd:YAG laser,” in OSA Trends in Optics and Photonics, C. R. Pollack and W. R. Bosenberg, Eds. Washington, DC: Optical Society of America, 1997, vol. 26, Advanced Solid-State Lasers, pp. 2–6. [29] W. Koechner, Solid-State Laser Engineering, 3rd ed. New York: Springer-Verlag, 1992, p. 423.


[30] S. D. Butterworth, S. Girard, and D. C. Hanna, “High-power, broadly tunable all-solid-state synchronously pumped lithium triborate optical parametric oscillator,” J. Opt. Soc. Amer., vol. 12, pp. 2158–2167, Nov. 1995. [31] A. Taguchi, A. Miyamoto, Y. Mori, S. Haramura, T. Inoue, K. Nishijima, Y. Kagebayashi, H. Sakai, Y. K. Yap, and T. Sasaki, “Effects of moisture on CLBO,” in OSA Trends in Optics and Photonics, C. R. Pollack and W. R. Bosenberg, Eds. Washington, DC: Optical Society of America, 1997, vol. 26, Advanced Solid-State Lasers, pp. 19–23. [32] Y. K. Yap, T. Inoue, H. Sakai, Y. Kagebayashi, Y. Mori, T. Sasaki, K. Deki, and M. Horigutchi, “Long-term operation of CsLiB O at elevated crystal temperature,” Opt. Lett., vol. 23, pp. 34–36, Jan. 1998. [33] N. Umemura, K. Yoshida, T. Kamimura, Y. Mori, T. Sasaki, and K. Kato, “New data on the phase-matching properties of CsLiB O ,” in OSA Trends in Optics and Photonics, M. M. Fejer, H. Injeyan, and U. Keller, Eds. Washington, DC: Optical Society of America, 1999, vol. 26, Advanced Solid-State Lasers, pp. 715–719.

Kevin F. Wall was born in Worcester, MA, in 1954. He received the B.S. degree in physics from Worcester Polytechnic Institute, Worcester, MA, in 1976, the M.S. degree from Rensselaer Polytechnic Institute, Troy, NY, in 1980, and the Ph.D. degree in applied physics from Yale University, New Have, CT, in 1986. From 1979 to 1980, he was with General Electric Research and Development Center, Schenectady, NY where he was engaged in Mössbauer studies of amorphous magnetic materials. From 1986 to 1992, he was with MIT Lincoln Laboratory, Lexington, MA, where he worked on the development of tunable solid-state laser materials, particularly, Ti:Al2O3. From 1992 to 1997, he was with Micracor, Acton, MA, where he worked on the development of microchip lasers for telecommunications applications. In 1997, he joined the research division of Schwartz Electro-Optics (now Q-Peak, Inc.), Bedford, MA, where he currently works on diode-pumped solid-state lasers. His interests include diode-pumped solid-state lasers, nonlinear optics including parametric oscillators, and computer modeling of temperature and stress distributions in laser media. Dr .Wall is a member of the Optical Society of America.

Joseph S. Smucz received the B.S. degree in optics from the University of Rochester, Rochester, NY, in 1987, and the M.S. and Ph.D. degrees in physics from the University of Massachusetts in 1995 and 1998, respectively. His dissertations involved the development of an ultra-short pulsed submillimeter-wave laser. He was with Science and Engineering Services, Inc., from 1998 to 2000, working on a number of diode-pumped solid-state laser and lidar systems. From 2000 to 2002, he was with Q-Peak, Inc., Bedford, MA, where he worked on several laser-development programs including the one reported in this paper. He is currently with Coherent-DEOS, Bloomfield, CT, as a Senior Development Engineer working on CO2 laser-development projects. Dr. Smucz is a member of the Optical Society of America and the International Society for Optical Engineering.

Bhabana Pati received the M.Sc. degree from Hyderabad Central University, India, in 1992 and the Ph.D. degree from Michigan Technological University, Houghton, in 1998. In 1998, she joined Q-Peak, Inc., Bedford, MA, as a Senior Engineer, where she works on a variety of diode-pumped solid-state-lasers. Dr. Pati is a member of the Optical Society of America.

1169

Yelena Isyanova received the M.S degree from the Leningrad Electrotechnical Institute, St.-Petersburg, Russia, and the Ph. D. degree from the Institute of Physics of the Belorussian Academy of Science, Minsk (former USSR), both in physics. From 1972 to 1991, she was a Principal Research Engineer at the Laser Division of Leningrad Optics and Mechanics Association, Leningrad, Russia, where she worked on research, development, and prototype production of solid-state and tunable lasers and nonlinear optical devices. In 1992, she joined the Research Division of Schwartz Electro-Optics (now Q-Peak, Inc.), where she is presently a Principal Research Scientist and Project Manager at Q-Peak, Inc. Her interests include diode-pumped solid-state lasers, techniques for spatial and temporal control of laser parameters, optical parametric oscillators, and other nonlinear devices. She has published over 30 papers in scientific journals on lasers and electrooptics and has been granted 18 patents in the former USSR. Dr. Isyanova is a member of the Optical Society of America.

Peter F. Moulton (S’74-M’74–SM’84) received the A.B degree in physics from Harvard College, Cambridge, MA, in 1968 and the M.S. and Ph.D. degrees from the Department of Electrical Engineering and Computer Science, Massachusetts Institute of Technology, Cambridge, in 1971 and 1975, respectively. He spent a postdoctorate year during 1975 at MIT Lincoln Laboratory, Lexington, MA, and became a Staff Member in 1976. His work at Lincoln Laboratory included high-resolution infrared spectroscopic measurements of molecules, development of lasers for remote sensing, and research and development of tunable and high-efficiency solid-state lasers, including invention of the Ti:sapphire laser. In 1985, he became Vice President and General Manager of the Research Division of Schwartz Electro-Optics, Bedford, MA, and was engaged in the research and development of a new solid state laser materials and systems. When the Research Division was spun out as separate company, Q-Peak, he served as Chief Technology Officer and Vice-President, a position he continues to hold. His areas of research interest have included Ti:sapphire lasers, high-power diode-pumped lasers, mid-infrared solid state lasers, nonlinear optics including parametric oscillators, medical applications of lasers, and lidar/ladar systems. Dr. Moulton has served as Co-Chair of CLEO and Chairman of the CLEO Steering Committee. He was responsible for the initial organization of the Optical Society of America (OSA)/IEEE Topical Meeting on Advanced Solid State Photonics, and has served on the Boards of IEEE/LEOS and the OSA. He was awarded the R.W. Wood Prize from the OSA and the William Streifer Scientific Achievement Award from IEEE/LEOS, both in 1997. He is a Fellow of the Optical Society of America and a member of the U.S. National Academy of Engineering.

Jeffrey G. Manni (M’87) received the B.S. degrees in chemistry and physics from Massachussetts Institute of Technology, Cambridge, in 1977 and 1978, respectively, the M.S. degree in applied physics from Stanford University, Stanford, CA, in 1980, and the M.B.A. degree from Northeastern University, Boston, MA, in 1987. He has worked as a Field Service and Applications Eengineer at Quanta-Ray, Inc., and as a Staff Laser Physicist at Candela Laser Corporation, Raytheon Laser Products, Schwartz Electro-Optics, and Spire Corporation. He formed JGM Associates, Inc., Burlington, MA, in 1988, and now provides contract laser R&D services to clients. His areas of research interest include high-power semiconductor diode and diode-pumped solid-state lasers, nonlinear optics, optical parametric oscillators, and applications thereof. Mr. Manni is a member of the Optical Society of America and SPIE.