Greg R. Hays,* Erhard W. Gaul, Mikael D. Martinez, and Todd DitmireTexas Center for High Intensity Laser Science, Department of Physics, University of Texas at Austin,
1 University Station, C1510, Austin, Texas 78712, USA
Spectral gain narrowing is one of the primary diffi-culties in developing sub-100 fs high-energy laser sys-tems with traditional high-energy laser amplifiers,such as Nd:glass. The technique of chirped-pulse am-plification (CPA) [1] has been applied to various laserhost materials in efforts to generate high peak powerlaser systems. One approach is to use Ti:sapphire asthe laser medium to achieve a joule class laser withextremely short pulses ��20 fs� [2]. However, theshort fluorescence lifetime and small aperture sizemakes realizing higher energies difficult. A recentadvancement in CPA design has been the develop-ment of optical parametric chirped-pulse amplifica-tion (OPCPA) [3,4]. OPCPA is interesting since thegain is extremely high and uniform over a broad spec-trum. Several groups have successfully combinedOPCPA front ends with conventional energy storageamplifiers in order to reduce spectral gain narrowing[5,6]. In these systems, the standard regenerativeamplifier is replaced by a broad-spectrum parametricamplifier that generates a few tens of millijoules.Broad spectra can be maintained in this way, and
some spectral enhancement can be achieved throughstrong gain saturation. However, OPCPA’s key con-straint is the pump laser. Parametric gain is stronglynonlinear and critically dependent on the pump in-tensity. Small pump intensity variations will resultin drastic temporal and subsequent spectral shapingin the amplified seed pulse. OPCPA pump laser in-tensities therefore must be well formatted in bothtemporal and spatial profiles. Since high-energy,high-fidelity custom pump lasers are not readilyavailable, OPCPA amplification is generally limitedto subjoule energies.
In other systems, the availability of large-apertureNd:glass has enabled construction of CPA lasers withenergies in excess of 500 joules and peak powers up to1 PW [7]. However, commonly available phosphatelaser glasses are mainly limited by their finite spec-tral bandwidth (�29 nm FWHM). To circumvent thisproblem, it is possible to mix various laser glasseswith different emission spectra in order to increase theoverall gain bandwidth [8–10]. The most straightfor-ward approach is to use Nd:phosphate and Nd:silicateglasses. These two media have peak emission spectraseparated by approximately 7 nm, each with �25 nmFWHM of bandwidth. The combination of the twoglasses broadens the gain spectrum, enabling shorter
pulses. However, only a handful of these Nd:glassesare commercially available.
During the 1970s, Lawrence Livermore NationalLaboratories conducted an exhaustive evaluation ofseveral hundred laser glasses. From these investiga-tions we have identified two broad-spectrum laserglasses that could be used in a mixed-glass architec-ture to realize gain spectra higher than currentlyachieved with Nd:phosphate and Nd:silicate com-binations. These glasses are designated as K-824Nd:tantalum�silicate and L-65 Nd:aluminate glass[11]. Neither glass was ever made commercially avail-able; however, sufficient spectral and technical dataare available to perform a comprehensive evaluation.In this paper we present a study of a mixed-glasslaser system that combines commercially availableNd:phosphate and each of these new glasses. Thegoal of this study was to ascertain how broad of anamplified spectrum could be achieved beyond thehundred joule level. To find the optimum glass, wesacrifice the need for high peak gain cross section,and we shift the emphasis to finding broader fluores-cence spectra with fluorescence peaks further awayfrom the Nd:phosphate peak. We find that with theaddition of the new laser glasses in a heterogeneousgain media system, near 100 fs pulses with a gain of104 and 150 fs pulses with a combined gain of 108 areachievable. Energetics and spectral modeling suggestthat this novel mixed glass, combined with OPCPAarchitecture, could be used to achieve in excess of 1exawatt peak power.
Most neodymium-doped laser glasses available to-day were developed for use in high-energy nanosecond-scale laser systems. In these cases the merit functionwas dominated by peak gain cross section and lownonlinear coefficient, not total spectral bandwidth. Thefirst glass developed was Nd:silicate. It has a peakspectral emission around 1061 nm and a gain crosssection of approximately 2.4 � 10�20 cm2. Later phos-phate glasses were used with peak emissions at1054 nm and higher gain cross sections of �3 �10�20 cm2. During the development period of theseglasses, many variant mixtures were processed andanalyzed in small non-optical-quality batches formetrological characteristics. Most of these glasses
were catalogued and never further developed. Wehave isolated two laser glasses that have significantlybroader emission spectra than the currently com-mercially available phosphate and silicate glasses.The two glasses examined are K-824 Nd:tantalum�silicate and L-65 Nd:aluminate. Both were producedby the National Bureau of Standards. Sufficient quan-tities were produced in order to perform Judd–Ofeltmeasurements of the spectral intensities [12,13]. How-ever, no physical or mechanical properties were de-tailed. Table 1 lists both the measured and calculatedoptical properties pertinent to this investigation.
Unique to the K-824 tantalum�silicate glass is a30% molar fraction of tantalum oxide �Ta2O5�. Allsilicate glasses examined in the catalog that pos-sessed Ta2O5 in their chemistry demonstrated signif-icantly broader emission spectra (�30 nm FWHM).The K-824 glass contained the highest molar fractionof them all, and had the broadest emission linewidth.In contrast, the aluminate glass is dominated by alu-minum oxide �Al2O3� in its chemistry (50% molar frac-tion).
APG-1 is a common phosphate glass availablefrom Schott Glass Technologies that exhibits a27.8 nm bandwidth centered at 1053.9 nm. It wasdeveloped for high-average power applications and isused in our calculations. We also considered Q-246silicate glass from Kigre, Incorporated. This com-mercially available glass emits at 1061 nm witha 28.5 nm FWHM bandwidth. Typical of silicateglasses, Q-246 has an extended red spectral footthat is broader in total bandwidth than phosphateglass. This is shown more clearly in Fig. 1. In con-trast to Q-246, both the tantalum�silicate and alu-minate glasses exhibit much broader emissionspectra, 38.5 nm and 41.2 nm, respectively. Further-more, both glass spectra are centered further to thered side of the spectrum. This is more appealing be-cause, when combined in a heterogeneous amplifier,the net available gain bandwidth is broader. Anotherkey point of interest in high-gain scenarios is therelative shape of the spectra. At 80% of the peakemission cross-section spectrum, Q-246 is approxi-mately 14 nm wide. However, the tantalum�silicateand aluminate glasses are both 19 nm wide. This is
Table 1. Optical Properties of Neodymium-Doped Laser Glassesa
the spectral region where most of the bandwidth willsurvive after high amplification ��104�.
There are a few important optical limitations of thenew glasses. First, the gain cross section in aluminateis significantly smaller at 1.8 � 10�28 cm2 than othersilicate glasses. This then requires more passesthrough the gain medium to accumulate the neces-sary gain. In addition, the estimated nonlinear coef-ficients in both the tantalum�silicate and aluminateglasses (3.44 � 10�13 and 2.92 � 10�13 esu, respec-tively) are greater than, by a factor of 2, Q-246 andare a factor of 3 larger than APG-1. This will inher-ently affect the nonlinear phase accumulated in eachamplifier (B-integral). High-fluence saturation of analuminate or tantalum�silicate amplifier will have tobe carefully considered.
The goal of this investigation was to examine theenergetics versus gain narrowing effects in a mixed-glass heterogeneous amplifier chain. We consideredthe mixing of APG-1 phosphate glass with eitherNd:tantalum�silicate or the Nd:aluminate. The pri-mary focus was to ascertain the total gain and thebest gain ratio between the mixed-glass amplifiers inorder to maintain sufficient bandwidth for 100 fspulses.
We employed a one-dimensional small signal gainamplification code in order to simplify the model. Itis difficult to achieve saturation in glass amplifierssince the saturation fluence is generally large(Table 1). This fact, coupled to high nonlinear coef-ficients and impurity inclusions in the material,makes the effects of gain saturation small in com-parison with spectral shaping from gain spectra.Pulses must exceed 5 J�cm2 in these materials be-fore saturation effects become prevalent. Further,for chirp factors �100 ps�nm, nonlinear phase ac-cumulation becomes significant at fluences above2 J�cm2 for the bandwidths considered here. There-fore gain saturation was not considered in these sim-ulations. Using the measured and published spectralgain curves for APG-1 and the other glasses, awavelength-dependent gain spectrum for the mixedmedia was generated. To vary the balance in gainbetween the two glasses, the method used by Ross
et al. [10] was utilized. The output spectrum from themixed-glass amplifier depends only on the ratio fromthe exponents in gain factor shown in Eq. (1),
1 � �
��
�phos
�
ln G
ln Gphos. (1)
Here �phos and � are the gain cross sections, and Gphosand G are the gain factors of Nd:phosphate and thealternate glasses, respectively. The subscript � isused to identify either tantalum�silicate or alumi-nate glass. The parameter �, which ranges from 0 to1, varies the net gain between the two media. When� is 1, the gain is entirely in the phosphate amplifier.Oppositely, when it is 0, the gain is entirely done bythe alternate glass. Amplified spectra were evaluatedat each gain order of magnitude. The net gain be-tween the two glasses was kept constant while � wasvaried from 0 to 1. The gain factor �, represented by
gain � exp���phos� � ��1 � ����,
�n ln 10
�phos� � ��1 � ��, (2)
is the product of the excited-atom population andtotal length of amplifier material. It indicates howhard each glass should be pumped for a given gainlength. Here n, which was varied from 0 to 8, repre-sents the exponential peak gain order of magnitude.
The amount of bandwidth required for mixed-glassamplification is determined by the relative separa-tion of the peak emissions of the two glasses. Most ofthe spectral amplification by each glass is accom-plished in a region that lies within 20% of its peakgain cross section. In the mixed-glass system, theblue side of phosphate and the red side of the silicateglass defines this space. For the two novel glasses,this region is approximately 30 nm. Therefore, seed-ing with a spectrum that is broader than 30 nm isrequired.
Initially a variety of seed beams and spectra fromvarious front-end amplification systems were consid-ered. Ultimately a Ti:sapphire oscillator followed byan OPCPA preamplifier chain was determined mostsuitable. Commercially available Ti:sapphire oscilla-tors operating beyond 1 micron can have sufficientbandwidth to generate 100 fs pulses. Parametric am-plification is then effective because of the high gainthat can be achieved with relatively small materialpath length. Most importantly, a strongly saturatedparametric amplifier will spectrally broaden the seedpulse. Several groups have recently demonstratedbroad bandwidth pulse compression of a spectrallybroadened OPCPA pulse [14,15]. Furthermore, pre-pulse contrast �107 has been achieved by appropri-ately tailoring the geometric and temporal windowsof the pump [16,17].
The pulse used in this simulation to seed themixed-glass amplifiers was modeled as a 16 nm
Fig. 1. Stimulated emission cross sections of neodymium-dopedlaser glasses.
FWHM Gaussian spectrum. The pulse width fromthe Fourier transform limit is 100 fs. The pulse wasfirst preamplified using a one-dimensional paramet-ric amplification code [18] to simulate a spectrallysaturated broadened pulse with 42 nm FWHM ofbandwidth. Once the preamplified spectral profilewas generated, it could be shifted in wavelength tofacilitate seeding various wavelength centers in aneffort to broaden the amplified output spectrum. The42 nm seed spectrum represents amplification of theentire 16 nm FWHM Gaussian pulse. No spectralclipping is assumed from the stretcher. We performedsimulations with narrower bandwidth seed pulses,and determined this was the optimum bandwidth.
A variety of wavelength centers were used for theOPCPA spectrally broadened pulse. APG-1 has a
peak gain centered at 1053.9 nm, tantalum�silicateat 1064.5 nm, and aluminate at 1067 nm. Also, thebandpass limit of the pulse stretcher further estab-lishes the finite bandwidth of the seed. Thereforeseed spectra with centers ranging from 1053 nm to1080 nm were examined during this simulation.APG-1 has the largest gain cross section, so it willcontribute most to the spectral shaping in the ampli-fier chain. As a result, it was found that the mosteffective wavelength at which to seed was 1070 nm.Here the strong gain pulling from the APG-1 could bebest utilized while still accessing the lower gain avail-able on the red side of the spectrum from either thetantalum�silicate or aluminate. Figure 2 illustratesthe seed centered at 1070 nm being amplified by amixed-glass amplifier. Here APG-1 is mixed with ei-
Fig. 2. Simulated spectral gain narrowing and the conjugate Fourier transform pulsewidths from mixing APG-1 Nd:phosphate glass witheither K-824 Nd:silicate or L-65 Nd:aluminate glass. Gain balance factor ��� is set to 0.5 in all graphs: (a) amplified pulse spectra withtotal gain of 101, (b) Fourier transform pulsewidths for 101, (c) amplified pulse spectra with total gain of 104, (d) Fourier transformpulsewidths for 104, (e) amplified pulse spectra with total gain of 108, and (f) Fourier transform pulsewidths for 108.
ther tantalum�silicate or aluminate, amplifying theseed by factors of 101, 104, and 108. Also shown foreach gain order of magnitude is the conjugate Fouriertransform limit pulsewidth. Seeding at longer wave-lengths ��1070 nm� became less efficient since therewas less bandwidth available at the shorter wave-lengths within the gain bandwidth of APG-1.
Throughout the simulation at all amplification lev-els, the optimum value for the gain balance factor �ranged primarily from 0.375 to 0.5. Therefore theproper gain balance tended to rely on higher gain fromthe phosphate glass. However, at � equal to 0.375, thegain is near equal in both materials. Additionally dif-ferent values of � at the same gain order of magnitudeyielded similar amplified bandwidths. This was pri-marily attributed to gain pulling of the spectra fromthe longer to shorter wavelengths. Hence the same netamplified bandwidth survived at different wavelengthcenters with different values of �.
The point at which the new glasses become mostattractive is when the combined glass gain exceeds103. Figure 3 illustrates that the broader spectraglasses mixed with a phosphate amplifier show amarked improvement in amplified bandwidth overthe standard silicate glass. When seeding the mixed-glass amplifier containing tantalum�silicate with aspectrum centered at �1065 nm, the optimum valuesof � ranged from 0.25 to 0.5. Here amplified band-widths were sufficient to generate 100 fs pulses. Inthis case, more gain is needed from the tantalum�silicate to maintain the longer wavelengths. How-ever, by shifting the seed spectrum to �1065 nm,100 fs pulses could still be achieved with an increasedgain of 104. Further � shifted to an optimum valuerange of 0.5 to 0.625 relying on more gain from phos-phate. With the higher nonlinear coefficient coupledwith the ample supply of phosphate glass, this is aclear systematic improvement. At this gain level thealuminate glass performed very similarly to tanta-lum�silicate with minutely more amplified band-width. This occurred primarily from its broader gainspectrum.
When the gain in the heterogeneous amplifiers wasincreased to 108, both glasses were able to support
sufficient bandwidth to produce 150 fs pulses. Con-versely, a mixture with a standard silicate glass nar-rowed the spectrum dramatically more. At this levelof amplification, gain narrowing is the dominant ef-fect regardless of seed wavelength. With tantalum�silicate, amplified pulses were generally �150 fs and� was optimum at �0.5. Shifting the seed wavelengthled to only small improvement. When seeding at1057 nm, the Fourier transform of the amplifiedbandwidth yielded 172 fs. Seeding at 1070 nm, thetransform produced a 165 fs pulse. The benefits of thebroader spectrum aluminate glass also become moreobvious at this level of amplification. Aluminate-amplified bandwidths were generally �10% broader,producing a best pulsewidth of 152 fs pulse whenseeded at 1070 nm. Seeding at 1057 nm produced a161 fs pulse. However, the optimum � value wasshifted to 0.375, thus requiring more gain from alu-minate glass than the phosphate.
Although both glasses performed similarly in theiramplified spectral content, overall, tantalum�silicateshowed higher merits. Even though it does not ex-actly match the spectral content of the aluminate, ithas an equivalent gain cross section to standard sil-icate glasses. This dramatically reduces the pumpinglevel and�or the total amount of material that needsto be passed to extract the same energy. Tantalum�silicate equaled aluminate with nearly the sameamount of amplified bandwidth maintained through-out all gain levels. Conversely, tantalum�silicatedoes possess a larger nonlinear coefficient that willcontribute to larger B-integral accumulation.
Acquisition of these new glasses poses the mostdifficult challenge. Currently, it would be difficult toprocure these new glasses in any form other than alarge aperture rod ��5 cm�. With that caveat aside,implementation of these glasses into present dayhigh-energy CPA glass lasers could be relatively sim-ple. An OPCPA front end can easily generate �50 mJwith a significantly gain broadened spectrum. Next, a�5 cm aperture tantalum�silicate rod could be mul-tipassed to extract a gain �102. The final energywould then be extracted and gain-balanced frommore affordable large-aperture Nd:phosphate slabamplifiers. Energies could easily exceed several hun-dreds of joules with sufficient bandwidth to generate100 fs pulses. With current technologies this could bea straightforward path to multipetawatt lasers.
If large-aperture slabs ��10 cm� of these glassescould be attained, the design of a mixed-glass exa-watt laser system is possible. This laser could beconstructed with further development of currenttechnologies, incorporating OPCPA, mixed-glass am-plification, and tiled gratings in the pulse compres-sor [19]. We offer here the conceptual design of aNd:glass laser system with an integrated peakpower of one exawatt from eight beamlines compris-ing one National Ignition Facility (NIF) amplifierbundle (Fig. 4).
Pulse generation would come from a 100 fs, mode-locked oscillator producing 16 nm FWHM Gaussian
Fig. 3. Fourier transform limit pulsewidths as a function of peakspectral gain order of magnitudes [n from Eq. (2)].
spectrum at 1.065 �m. In order to avoid significantnonlinear phase accumulation in the final amplifier,this pulse is stretched to 8 ns using 1740 lines�mmdiffraction gratings with a bandpass of �40 nm. Pre-amplification to 1 joule would be accomplished in amultistage broadband OPCPA to produce a super-saturated 40 nm FWHM spectrum [20]. Mixedglasses would be used for the final amplification witha � value of 0.5 and a combined gain of 1.7 � 104.Using multipassed 9.4 cm diameter slab amplifierswith Nd:tantalum�silicate glass, the pulse would beamplified to �50 joules. Finally, the pulse is injectedinto a NIF-style, 40 cm Nd:phosphate glass amplifierto increase the energy to �17 kJ [21].
We performed a numerical simulation of this sys-tem, which included gain saturation and nonlinearphase accumulation. Saturation effects were in-cluded here since the final amplifiers achieve an out-put fluence of �10 J�cm2. The small signal gainmodel predicted a transform-limited pulsewidth of115 fs. In contrast, inclusion of gain saturation re-duced the simulated bandwidth and produced atransform-limited pulsewidth of 135 fs. This is pri-marily attributed to square pulse distortion, andcould be precompensated before glass amplification.The model also predicts an accumulated B-integral of0.9 from the glass amplification. The majority of thenonlinear phase arises from the phosphate amplifier�B � 0.85�, which experiences the highest fluence andlongest optical path length.
Ultimately, the final technological barrier would becompression grating diameter. At high energy the
damage threshold will dominate grating size overdiffractive dispersion from the increased bandwidth.However, using tiled multilayer dielectric gratingtechnology, a two-grating single-pass compressorcould be constructed. Allowing for a 1 J�cm2 beamfluence exiting the compressor, a pair of tiled gratingswith 7.5 and 10.5 m2 of area, respectively, with �80%fill factor will be required to compress the pulse. Thefirst grating would be comprised of a 3 � 5 tiled arrayof 0.5 m � 1 m gratings. The second grating horizon-tal diameter would increase to a 3 � 7 tiled array. Tocompress the pulse with a 500 ps�nm chirp in a sin-gle pass, the nominal grating separation would be13.4 meters. The exiting beam diameter would be130 � 170 cm with an approximate lateral chirp of10 mm�nm. The lateral chirp will cause the pulsefront to tilt in the far field; however, it is small incomparison to the overall size of the beam. With�90% transmission from the compressor, each beam-line should compress 15 kJ to 120 fs, producing a125 PW peak power. The eight 125 PW beamlinescould combine to deliver 1 exawatt in peak power.
In conclusion, we have shown that ultrafast high-energy pulses can be achieved with a high gain inglass. Implementation of novel broad-spectrum glassin a mixed-glass architecture can decrease amplifiedpulse widths to the 150 fs regime with up to 8 ordersof magnitude in amplification. This could clearly pavethe way for the next generation multipetawatt andeven exawatt laser. We would like to acknowledgefunding from the U.S. Department of Energy NNSAcontract DE-FL52-03NA00156.
Fig. 4. (Color online) Conceptual schematic of one of eight beamlines comprising an exawatt laser system.
References1. D. Strickland and G. Mourou, “Compression of amplified
chirped optical pulses,” Opt. Commun. 55, 447–449 (1985).2. J. P. Zhou, C. P. Huang, M. M. Murnane, and H. C. Kapteyn,
“Amplification of 26-fs, 2-TW pulses near the gain-narrowinglimit in Ti-sapphire,” Opt. Lett. 20, 64–66 (1995).
3. I. Jovanovic, B. J. Comaskey, C. A. Ebbers, R. A. Bonner, D. M.Pennington, and E. C. Morse, “Optical parametric chirped-pulse amplifier as an alternative to Ti:sapphire regenerativeamplifiers,” Appl. Opt. 41, 2923–2929 (2002).
4. I. N. Ross, P. Matousek, M. Towrie, A. J. Langley, and J. L.Collier, “The prospects for ultrashort pulse duration and ultra-high intensity using optical parametric chirped pulse amplifi-ers,” Opt. Commun. 144, 125–133 (1997).
5. J. Collier, C. Hernandez-Gomez, I. N. Ross, P. Matousek, C. N.Danson, and J. Walczak, “Evaluation of an ultrabroadbandhigh-gain amplification technique for chirped pulse amplifica-tion facilities,” Appl. Opt. 38, 7486–7493 (1999).
6. I. Jovanovic, C. A. Ebbers, and C. P. J. Barty, “Hybrid chirped-pulse amplification,” Opt. Lett. 27, 1622–1624 (2002).
7. M. D. Perry and G. Mourou, “Terawatt to petawatt subpico-second lasers,” Science 264, 917–924 (1994).
8. N. Blanchot, C. Rouyer, C. Sauteret, and A. Migus, “Amplifi-cation of sub-100-TW femtosecond pulses by shifted amplifyingNd-glass amplifiers: theory and experiments,” Opt. Lett. 20,395–397 (1995).
9. R. Danielius, A. Piskarskas, D. Podenas, and A. Varanavicius,“Broad-band Nd glass regenerative amplifier with combinedactive medium,” Opt. Commun. 84, 343–345 (1991).
10. I. N. Ross, M. Trentelman, and C. N. Danson, “Optimization ofa chirped-pulse amplification Nd:glass laser,” Appl. Opt. 36,9348–9358 (1997).
11. S. E. Stokowski, R. A. Saroyan, and M. J. Weber, “Laser glassNd-doped glass spectroscopic and physical properties, rev. 2,Vol. 1, 2” (Lawrence Livermore National Laboratory, 1981).
12. B. R. Judd, “Optical absoption intensities of rare earth ions,”Phys. Rev. B 127, 750 (1962).
13. G. S. Olfelt, “Intensities of crystal spectra of rare-earth ions,”J. Chem. Phys. 37, 511 (1962).
14. O. V. Chekhlov, J. L. Collier, I. N. Ross, P. K. Bates, M. Notley,C. Hernandez-Gomez, W. Shaikh, C. N. Danson, D. Neely, P.Matousek, and S. Hancock, “35 J broadband femtosecond op-tical parametric chirped pulse amplification system,” Opt.Lett. 31, 3665–3667 (2006).
15. F. Tavella, A. Marcinkevicius, and F. Krausz, “90 mJ para-metric chirped pulse amplification of 10 fs pulses,” Opt. Ex-press 14, 12822–12827 (2006).
16. I. Jovanovic, C. G. Brown, C. A. Ebbers, C. P. J. Barty, N.Forget, and C. Le Blanc, “Generation of high-contrast milli-joule pulses by optical parametric chirped-pulse amplificationin periodically poled KTiOPO4,” Opt. Lett. 30, 1036–1038(2005).
17. F. Tavella, K. Schmid, N. Ishii, A. Marcinkevicius, L. Veisz,and F. Krausz, “High-dynamic range pulse-contrast measure-ments of a broadband optical parametric chirped-pulse ampli-fier,” App. Phys. B 753–756 (2005).
18. R. L. Sutherland, Handbook of Nonlinear Optics (Marcel Dek-ker, 1996).
19. T. J. Kessler, J. Bunkenburg, H. Huang, A. Kozlov, and D. D.Meyerhofer, “Demonstration of coherent addition of multiplegratings for high-energy chirped-pulse-amplified lasers,” Opt.Lett. 29, 635–637 (2004).
20. E. W. Gaul, T. Ditmire, M. D. Martinez, S. Douglas, D. Gorski,G. R. Hays, W. Henderson, A. Erlandson, J. Caird, C. Ebbers,I. Jovanovic, and W. Molander, “Design of the Texas PetawattLaser,” in Conference on Lasers and Electro-Optics (OpticalSociety of America, 2005).
21. G. H. Miller, E. I. Moses, and C. R. Wuest, “The NationalIgnition Facility,” Opt. Eng. 43, 2841–2853 (2004).
