Ultrahigh-intensity laser: physics of the extreme on a tabletop [report]

G.A. Mourou, C.P. Barty, M.D. Perry
1997 unpublished
produce fields greater than critical and nonlinear quantum electrodynamical effects can be ,.. observed. In many ways, this physical environment of extreme fields, pressure, magnetic field, temperature, and acceleration can. only be found in stellar interiors or close to the horizon of a black hole and it is fascinating to think that an astrophysical environment governed by hydrodynamics, radiation transport, and gravitational interaction could be re-created in university laboratories for an
more » ... oratories for an extremely short time, switching the role of scientists from voyeurs to actors. Also, ultrahigh-intensity lasers will play an important role in the development of new compact, tabletop-size, ultrashort sources of high-energy photons (x ray) and electrons (giga-electronvolt) with attosecond duration. The period of light being 3 fs at 1 p,m, getting into the attosecond regime will require shorter wavelengths, which naturally involves high intensities. For instance, short wavelengths can be produced by nonlinear effects like harmonic generation and produce subfemtosecond pulses. Similarly, the enormous field gradients that can be produced in laser plasma interaction can accelerate electrons over a distance 104 times shorter than what was previously possible and the resulting electron pulses are expected to be well in the subfemtosecond range. These ultrafast sources will have important applications for time resolving the structure of matter at the atomic or molecular level. Short pulses are necessary, considering that the atomic motion is in the 100-fs range. Ultrahigh intensity led to a dramatic extension to the gigs-electron-volt range of laser applications, until then confined to the l-10-keV regime. What is spectacular is that the size of the laser involved is extremely compact (tabletop) and relatively inexpensive. It can deliver high average power and be built and operated by students. Some of these tabletop-laser principles can be easily implemented at a minimum cost on existing large laser systems already built for laser fusion. Lawrence Livermore Laboratory,l CEA-Limeil,l and Osakal have converted their nanosecond lasers to subpicosecond pulse duration, pushing their peak power by 3 orders of magnitude from the terawatt (1012) to 100-1OOO TW.1 They also can be moved to very large structures like synchrotrons (ALS, ESRF) to produce ultrashort, high-energy x rays by Compton scattering, or to accelerator facilities like SLAC to test nonlinear quantum electrodynamics by the interaction of the high-intensity pulses with superrelativistic electrons. Figure 1 shows laser intensity as a function of years. We can see a rapid increase in intensity in the early 1960s, followed by a long plateau at 1015 W/cm2. It took about 20 years, until 1987, for laser power to increase again. It is important to stress the similarity in slopes between the early 1960s and the past ten years, and to remember that it is during this period of very rapid increase in intensities in the 1960s that most of the nonlinear optics phenomena were discovered. In a similar way, the spectacular increase in intensity (four to five orders of magnitude) that we are experiencing today should lead td"exceptional scientific discoveries. Finally, from an educational point of view, this compact laser offers the advantage of bringing some of the research traditionally done on large instruments to human-size setups in university laboratories. Their small scale and large numbers should greatly foster diversification and multidisciplinarity and should attract shdents to scientific disciplines. Although strong-field physics with conventional lasers had already started in the 1960s,2 the scientific revolution started about 10 years ago. At that time, we demonstrated that ultrashort pulses at the 100-fs level could be amplified, without damaging the amplifying media, to the terawatt level, using a technique that we called Chirped Pulse Amplification (CPA).3 Since their inception in 1960, the peak power of lasers has evolved by a succession of leaps of three orders of magnitude. They were produced each time by decreasing the pulse duration accordingly. First the lasers were free runnin g, with a duration in the 10-ps and peak power in the kilowatt range. In 1962, modulation of the laser-cavity quality factor allowed the same .. . . energy to be released in a nanosecond time scale, a thousand times shorter, to produce pulses in the megawatt range. In 1964, locking the longitudinal modes of the laser (mode locking) allowed the' laser pulse duration to be reduced by another factor of a thousand, down to the picosecond level, pushing the peak power a thousand times higher, to the gigawatt level. At this point, the intensities associated with the ultrashort pulses were becoming prohibitively high-i. e., in the GW/cm2 range. At these intensities, the index of refraction becomes linearly dependent on the intensity to vary like n = YZo + n21, where n is the index of refraction, no the index of refraction at low intensity, 722the nonlinear index of refraction, and I the intensity. The result is that, for a beam with a Gaussian radial intensity distribution, the center of the beam sees a larger index of refraction than its sides. The material becomes a positive lens and alters the beam wave-front quality to an unacceptable level, to the point where it can create filaments and irreversible damage to the laser system. The only way to increase the peak power was to increase the beam diameter at the expense of size, low ,repetition rate, and cost. Although the pulse duration kept decreasing, for about 20 years the intensity limitation in laser systems kept the peak power about constant at the gigawatt level for a cm2 beam, until 1985-87, when the technique of CPA was demonstrated. The CPA concept is illustrated in Fig. 2 . The short pulse is first produced by an oscillator. After generation it is not amplified directly, but stretched by a large amount, 103 to 105, from femtosecond to nanosecond, reducing the intensity accordingly. The pulse intensity is now low enough that the stored energy can be safely extracted out of the amplifier, without fear of beam distortions and damage. Once the stored energy is fully extracted, the pulse is recompressed, ideally to its initial value. The key point of the CPA technique is that it decouples pulse fluence (energy/cm2) and pulse intensity (power/cm2). So it reconciles two apparently conflicting needs: to have (a) the highest fluence for efficient energy extraction and (b) minimum intensity to avoid the undesired nonlinear effects. CPA had a dramatic impact. First, we could for the first time use extremely good energy storage media like Nd:glass,3 alexandrite,l Ti:sapphire,~and Cr:LiSAF,l and extract their energy fully. Before CPA, dyes and excimers, which area thousand times inferior in energy storage, were the only choice for ultrashort pulse generation. So at equal size, a CPA laser system, using the superior energy storage media, could produce 103-to-104 times higher peak power than conventional dye or excimer systems. Second, CPA could be easily adapted on very large-scale, expensive lasers already built for laser fusion. By simple beam manipulations, a stretching at the beginning and a compression at the end of the amplifying chain, these expensive systems built to amplify nanosecond pulses to the terawatt level could be converted to amplify femtosecond pulses, as well, and to produce petawatt pulses. rate, 10-103 Hz, bringing the average power of Ti:sapphire-based CPA to the watt level.8z5 This represents an improvement of 2-3 orders of magnitude over conventional amplification systems based on dyes or excimers. Increased repetition rate at constant peak power leads directly to higher experimental utility. With ultrafast CPA systems, for instance, it is possible to apply signal-averaging techniques to high-field laser/matter investigations. At the University of California, San Diego, 4-TW, sub-20-fs FWHM pulses5 have been demonstrated from a Ti:sapphire amplification chain operating at a 1O-HZ repetition rate. This laser is shown in Fig. 4. At ENSTA, France, the University of Tokyo, and the Japanese Atomic Energy Research Institute,5 30-TW, 20-30-fs pulses5 have been obtained. Extension of these concepts to higher peak power and higher repetition rate are straightfonvard and are currently under development. In Japan, the Kansai research establishment of the Japanese Atomic Energy Research Institute is presently designing and testing components for a petawatt (20-J, 20-fs), l-Hz laser which is scheduled for construction Center for Ultrafast Optical Science (CUOS) at in 1999. In Bordeaux, France, and at the NSF the University of Michigan, funds have been ,. allocated and designs completed for the construction a 1-TW (20-mJ, 20-f s), l-kHz laser facilities. get to the ultimate pulse duration, the cavity mirrors (no lenses to avoid dispersive effects) must be broadband enough and the dispersion of the pulse in crystal must be compensatedlo by frequency-chirped mirrors (Krausz and coworkers) or by a sequence of prisms (Murnane and Kapteyn group). Pulses down to 6.5 fs have been produced directly out of the oscillator by U. Keller et al.lo The Petawatt: The Heav-jnveight CPA The I?etawatt Project, to explore the feasibility of the fast ignition concept, was realized at Livermore by one of the authors. It illustrates beautifully the adaptability of CPA. They use a beam of the large NOVA, a Nd:glass laser chain built for nanosecond pulse amplification delivering kilojoules in a few nanoseconds. Although conceptually simple, the adaptation of CPA on this laser chain required the development of sophisticated new technology: in particular, the design of very large diffraction gratings with 75-cm diameter, efficiency greater than 90%, with flatness-better than 1/10, and a good damage threshold. Thus, 1.3 kJ could be extracted in 800 ps and compressed in 430 fs, producing 1.3 PW with >1021 W/cm2 Although strong-field physics with conventional lasers had already started in the 1960s,2 the scientific revolution started about 10 years ago. At that time, we demonstrated that ultrashort pulses at the 100-fs level could be amplified, without damaging the amplifying media, to the terawatt level, using a technique that we called Chirped Pulse Amplification (CPA).3 Since their inception in 1960, the peak power of lasers has evolved by a succession of leaps of three orders of magnitude. They were produced each time by decreasing the pulse duration accordingly. First the lasers were free runnin g, with a duration in the 10-ps and peak power in the kilowatt range. In 1962, modulation of the laser-cavity quality factor allowed the same .. . . energy to be released in a nanosecond time scale, a thousand times shorter, to produce pulses in the megawatt range. In 1964, locking the longitudinal modes of the laser (mode locking) allowed the' laser pulse duration to be reduced by another factor of a thousand, down to the picosecond level, pushing the peak power a thousand times higher, to the gigawatt level. At this point, the intensities associated with the ultrashort pulses were becoming prohibitively high-i. e., in the GW/cm2 range. At these intensities, the index of refraction becomes linearly dependent on the intensity to vary like n = YZo + n21, where n is the index of refraction, no the index of refraction at low intensity, 722the nonlinear index of refraction, and 1 the intensity. The result is that, for a beam with a Gaussian radial intensity distribution, the center of the beam sees a larger index of refraction than its sides. The material becomes a positive lens and alters the beam wave-front quality to an unacceptable level, to the point where it can create filaments and irreversible damage to the laser system. The only way to increase the peak power was to increase the beam diameter at the expense of size, low ,repetition rate, and cost. Although the pulse duration kept decreasing, for about 20 years the intensity limitation in laser systems kept the peak power about constant at the gigawatt level for a cm2 beam, until 1985-87, when the technique of CPA was demonstrated. The CPA concept is illustrated in Fig. 2 . The short pulse is first produced by an oscillator. After generation it is not amplified directly, but stretched by a large amount, 103 to 105, from femtosecond to nanosecond, reducing the intensity accordingly. The pulse intensity is now low enough that the stored energy can be safely extracted out of the amplifier, without fear of beam distortions and damage. Once the stored energy is fully extracted, the pulse is recompressed, ideally to its initial value. The key point of the CPA technique is that it decouples pulse fluence (energy/cm2) and pulse intensity (power/cm2). So it reconciles two apparently conflicting needs: to have (a) the highest fluence for efficient energy extraction and (b) minimum intensity to avoid the undesired nonlinear effects. CPA had a dramatic impact. First, we could for the first time use extremely good energy storage media like Nd:glass,3 alexandrite,l Ti:sapphire,~and Cr:LiSAF,l and extract their energy fully. Before CPA, dyes and excimers, which area thousand times inferior in energy storage, were the only choice for ultrashort pulse generation. So at equal size, a CPA laser system, using the superior energy storage media, could produce 103-to-104 times higher peak power than conventional dye or excimer systems. Second, CPA could be easily adapted on very large-scale, expensive lasers already built for laser fusion. By simple beam manipulations, a stretching at the beginning and a compression at the end of the amplifying chain, these expensive systems built to amplify nanosecond pulses to the terawatt level could be converted to amplify femtosecond pulses, as well, and to produce petawatt pulses. rate, 10-103 Hz, bringing the average power of Ti:sapphire-based CPA to the watt level.8z5 This represents an improvement of 2-3 orders of magnitude over conventional amplification systems based on dyes or excimers. Increased repetition rate at constant peak power leads directly to higher experimental utility. With ultrafast CPA systems, for instance, it is possible to apply signal-averaging techniques to high-field laser/matter investigations. At the University of California, San Diego, 4-TW, sub-20-fs FWHM pulses5 have been demonstrated from a Ti:sapphire amplification chain operating at a 1O-HZ repetition rate. Ultrashort pulses as short as a few optical cycles have been with us for 10 years. But it is the outstanding discovery by W. Sibbett in 1991 of Kerr Lens Mode Locking (KLM)9 with its successive refinementsll that made the generation of pulses shorter than 10 fs routine. These systems will provide a robust source of short seed pulses at the nanojoule level for CPA systems. The concept of the generation of ultrashort pulses by KLM is simple and uses simultaneously the very large bandwidth of Ti:sapphire and its optical Kerr effect, or the intensity-dependent index of refraction, described above. A simplified diagram of a laser cavity is shown in Fig. 3a . .. The cavity is composed of curved mirrors and the amplifying medium (Ti:sapphire). The amplifier is optically pumped, usually with a continuous-wave argon laser. The Ti:sapphire has a double role. The first is to amplify the pulse, to compensate the various losses in the cavity. The second is to be a lens with an intensity-dependent focal length. We have seen previously that the Gaussian beam profile will produce a positive lens due to the intensity-dependent index of refraction. SO the experimentalist adjusts the wave-front radius of curvature to be equal to one of the mirrors for the highest intensity or the shortest pulse duration. In other words, it is where the laser experiences the minimum amount of losses that the laser will run. To get to the ultimate pulse duration, the cavity mirrors (no lenses to avoid dispersive effects) must be broadband enough and the dispersion of the pulse in crystal must be compensatedlo by frequency-chirped mirrors (Krausz and coworkers) or by a sequence of prisms (Murnane and Kapteyn group). Pulses down to 6.5 fs have been produced directly out of the oscillator by U. Keller et al.lo The Petawatt: The Heav-jnveight CPA The I?etawatt Project, to explore the feasibility of the fast ignition concept, was realized at Livermore by one of the authors. It illustrates beautifully the adaptability of CPA. They use a beam of the large NOVA, a Nd:glass laser chain built for nanosecond pulse amplification delivering kilojoules in a few nanoseconds. Although conceptually simple, the adaptation of CPA on this laser chain required the development of sophisticated new technology: in particular, the design of very large diffraction gratings with 75-cm diameter, efficiency greater than 90%, with flatness-better than 1/10, and a good damage threshold. Thus, 1.3 kJ could be extracted in 800 ps and compressed in 430 fs, producing 1.3 PW with >1021 W/cm2
doi:10.2172/303899 fatcat:qgc3pdc22jc3tlxxztfy5aha7m