1. INTRODUCTION

We discuss here the advantages of krypton-fluoride (KrF) lasers for implementing inertial fusion, the present status of high-energy KrF laser technology, and the development needed for inertial fusion and its energy application. The next section describes the conditions required for high-performance laser fusion and how a KrF laser would help toward meeting these conditions. This is followed by four sections that describe the advances and paths taken to develop high-energy KrF lasers for inertial fusion. Currently, the Naval Research Laboratory (NRL) Nike laser is the most capable KrF facility for laser target interactions. It provides up to 3 kJ of deep UV light ($\lambda =248\mathrm{nm}$) onto planar targets for studying the hydrodynamics of targets ablatively accelerated to high velocity and for basic studies of high-intensity KrF laser light interacting with a plasma. Nike has deeper UV and is capable of producing the most uniform target illumination of all high-energy laser fusion facilities. The Electra KrF facility is a high-repetition-rate system that has produced up to 700 J at 5 pulses per second. It was built to advance the KrF science and technologies needed for inertial fusion energy (IFE). In the final section we discuss the development of KrF technology needed for its application as a driver for a high-performance laser fusion implosion research facility and the still higher performance required for IFE.

2. INERTIAL FUSION WITH LASERS

The goal of an energy source based on fusion has been pursued for over 50 years. An understandable concern is that practical fusion energy remains a research challenge and may take many more years of research. Fusion has turned out to be a more difficult task than originally hoped. Yet the potential payoff with success is considerable, and a large international effort to develop fusion science and technologies currently exists. The largest effort is in magnetically confined fusion, where the flagship International Thermonuclear Experimental Reactor (ITER) project now underway has the goal of demonstrating fusion energy gains of 10 by the mid 2030s [1]. An alternate and fundamentally different path to fusion is based on inertial confinement, where the thermonuclear fuel is compressed to many times solid density and heated to ignition [2,3]. The high fuel density allows a substantial fraction of the fuel to burn before it expands appreciably. Hence the term “inertial confinement” is used to describe the configuration. Inertial confinement is the basis for thermonuclear weapons, so the concepts are at least as old as those of magnetic fusion. Both “single shot” inertial fusion in the laboratory and the more challenging requirements of IFE had to await development of suitable drivers with sufficient power and energy, plus the capability to deliver that power to fusion targets. Lasers, heavy ion accelerators, and pulse power devices have the potential to meet these requirements.

Figure 1 shows a standard configuration for inertial fusion by central ignition. The pellet’s outer shell is heated to plasma temperatures by the driver, and the resulting ablation pressure drives the shell inward to speeds of $300\mathrm{km}/\mathrm{s}$ or higher. In the optimum configuration for energy gain, the deuterium-tritium (DT) fuel is compressed to high density in a layer surrounding a core that has been heated to ignition. (Here gain is defined as the fusion energy output divided by the input drive energy.) The thermonuclear burn propagates outward from the core to the surrounding fuel. This configuration substantially increases the gain because the energy needed to compress the fuel is much less than that needed to reach ignition temperature. Two different schemes are pursued with lasers: (1) “indirect drive,” where the laser energy is first converted to x rays that illuminate the pellet [4,5], and (2) “direct drive,” where the laser itself illuminates the pellet [6]. Both schemes have their respective advantages and disadvantages. Target gains greater than 60 to 130 are required for the fusion energy application because the available high-energy lasers have modest efficiencies ($\sim 7\%–15\%$) [7,8]. Direct drive has a substantial advantage toward achieving such high gains because it does not suffer the losses from the x-ray conversion step. In this paper, we examine the advantages of the KrF laser for the direct drive.

 

Fig. 1. Inertial fusion concept where (1) a pellet shell is imploded to high velocity, (2) the central portion of deuterium-tritium (DT) fuel is heated to ignition, and (3) the thermonuclear burn propagates out from the ignition spark into the surrounding highly compressed ($\ge 1000\times$ solid) DT fuel.

Download Full Size | PDF

Flash-lamp-pumped Nd:glass lasers pumped by flashlamps are the workhorse driver for almost all large laser fusion facilities. It is capable of delivering the power and energy thought to be required for fusion implosions. The National Ignition Facility (NIF) delivers up to 1.8 MJ of frequency tripled Nd:glass light to indirectly driven targets [9,10]. A similar-sized system called the Laser MegaJoule is under construction in France [11]. NIF was primarily built to demonstrate indirect ignition. Though it has advanced the physics of the indirect drive, it has so far not achieved ignition, and it is uncertain that it can do so. A direct drive is now under consideration for NIF due to the greater amount of energy that can be deposited in the capsule.

The inertial fusion program in the United States is supported by the National Nuclear Security Administration as part of its stockpile stewardship program. A primary goal is to demonstrate and utilize robust inertial fusion and eventually high yields. In the longer term there is the goal of developing inertial fusion as a source of energy, IFE. The latter is a much greater challenge. As we will discuss in this paper, the KrF laser has substantial advantages toward meeting the challenges of both ICF in the laboratory and IFE.

Although it is beyond the scope of this paper to go into a detailed discussion on the laser target interactions involved with laser-driven fusion implosions, Table 1 lists the characteristics of a laser now thought to be needed for high-performance direct drive implosions. It is well accepted in the ICF research community and confirmed by both experiment and simulations that a shorter laser wavelength improves the efficiency of laser coupling to the imploding capsule and also increases the intensity threshold for undesired laser plasma instabilities [3]. It has proven impractical to achieve sufficiently uniform target illumination at high laser energy by making perfect diffraction-limited systems. Both Nd:glass and KrF lasers rely on introducing controlled levels of temporal and spatial incoherence in the laser beams and thereby achieve a time-averaged uniform illumination of targets. Smoothing by spectral dispersion (SSD) is utilized for frequency-tripled Nd:glass lasers [12], while induced spatial incoherence (ISI) [13,14] is utilized for the Nike KrF system and also has been employed on several other high-energy excimer laser systems [15 –18]. Bandwidths ($\mathrm{\Delta }\omega /2\pi$) of a few hundred GHz have proven sufficient to prevent filamentation of the laser light in the coronal plasma. Bandwidths exceeding 1 THz are desirable to reduce laser imprinting on the target that can seed hydrodynamic instability. Very large bandwidths ($>1\mathrm{THz}$) may also reduce parametric laser-plasma instabilities. Zooming involves reducing the laser focal diameter in time to match that of the imploding target [13,19]. Simulations indicate that this significantly improves the laser-target coupling efficiency [20,21]. In addition, zooming should substantially reduce the transfer of energy from beam to beam due to scattering off ion waves in the coronal plasma [22,23]. This cross beam energy transfer (CBET) can cause significant energy losses to the target and also severely complicate achievement of sufficiently uniform energy deposition on the capsule.

Table 1. Laser Requirements for Direct Drive Laser Fusion

View Table | View all tables in this article

Table 2 lists the corresponding characteristics of the frequency tripled Nd:glass laser and the KrF laser. Across the board the KrF laser is superior for direct drive implosions. It has the potential to produce high-gain implosions using substantially less laser energy with less interference from hydrodynamic and laser-plasma instability. So why has it not been utilized? One reason is that there has been much larger investment in Nd:glass lasers for ICF, and it may prove to be adequate to achieve the ICF goals of robust ignition and substantial gains. The other reason is that KrF was thought to be too difficult a technology to tame for the ICF application. In the following we will discuss the development of the KrF technology. Projections from achieved performance indicate that KrF does indeed have the potential to not only meet the requirements for ICF but also the more stringent requirements for IFE.

Table 2. Characteristics of Nd Glass and KrF Lasers for ICF

View Table | View all tables in this article

3. DEVELOPMENT OF KRF LASERS FOR ICF

An ICF capsule requires delivery of megajoule-class energy to a capsule on a few ns timescale. No laser can deliver such energies with a single beamline. NIF for example uses 192 beams to generate the 1.8 MJ on target. The 192 large aperture amplifier and harmonic conversion systems on NIF generate up to 10 kJ of energy on target per beam. A MJ-scale KrF system for ICF would need to utilize multi-kilojoule power amplifiers, probably $>10\mathrm{kJ}$, to keep the number of KrF beamlines reasonable. Discharge-pumped KrF lasers can deliver a few joules of energy at high repetition rates, but limitations on their maximum aperture and pulse length preclude very high energy [25,26]. The only successful technology to achieve multi-kilojoule energy is the electron-beam-pumped KrF laser. Figure 2 shows a typical configuration.

 

Fig. 2. Electron-beam pumped KrF laser.

Download Full Size | PDF

The components include:

The diode structure partially shorts out the repulsive self-electric field of the electron beam, so the electron beam will pinch due to its self-generated magnetic field for high enough currents. There are three solutions to obtain high energy: (1) keep the size of the system and current small enough that the electron beam does not pinch and use many amplifiers; (2) use multiple small diodes to pump a large laser volume; and (3) utilize an external magnetic field larger than the self-generated field to suppress the pinch in a high-current diode. The 800 J Rapier system developed at Lawrence Livermore National Laboratory was a prototype for a system that would use method 1 [27]. The Sprite and Titania KrF systems at the Rutherford Appleton Laboratory [28] and the Ashura and 3 kJ Super Ashura systems at the Electro Technical Institute in Japan used method 2 [29]. The 1 m aperture Aurora amplifier developed at Los Alamos National Laboratory used method 3 and produced 10 kJ of KrF light as an oscillator [30]. The electron-beam-pumped amplifiers used in the Nike and Electra KrF facilities utilize external magnetic fields. As discussed later the use of external magnetic fields (method 3) has provided the most efficient energy transport to the laser gas. However, with very large currents, a transit-time instability appears even with large external magnetic fields [31,32]. As discussed later two cathode configurations were developed and demonstrated on the Nike final amplifier that completely suppress this instability. There are therefore no physics limits on the size and current of magnetically stabilized electron beams to pump KrF lasers. However, there are practical limits to the diode power and amplifier size, which will be discussed in the concluding section of this paper.

A challenge to using KrF lasers for ICF is the mismatch between the high-energy amplifier operation and the several ns FWHM duration pulses needed for ICF. The amplifier needs to be pumped for $>100\mathrm{ns}$ to provide high energy, but the lifetime for the KrF excited state is only a few ns. The amplifier needs a long-duration optical pulse to extract the energy while the target needs a much shorter pulse. Two methods have been explored to meet this target requirement. The first is angular multiplexing, where one divides the few ns duration front-end seed pulse into numerous beams that enter each amplifier stage at different small off-axis angles and cross one another at its center [33 –35]. By introducing suitable time delays in each of these beams, one can ensure that the amplifier sees a train of short contiguous pulses that resembles a single long pulse. After the final stage, the time delays are removed, and the pulses hit the target simultaneously. This system requires many optics and beams, so early KrF research teams explored means to shorten a long-pulse KrF beam’s duration after it was amplified [36]. These schemes involved pumping a gas cell with the long-duration beam and utilizing backward Raman or Brilluoin scattering to transfer KrF laser energy to a shorter duration counter-propagating seed beam. These techniques have worked to a degree but have so far not proven to be efficient on large KrF systems. Angular multiplexing has been used on all successful high-energy KrF systems. In the Sprite facility, short-duration angular multiplexed beams were combined via forward Raman in a gas cell where the energy was extracted by a short-duration seed beam [37]. This system provided a high-power, nearly diffraction-limited 10 ps duration beam for target experiments but did not prove to be efficient when scaled up to the $>1\mathrm{kJ}$ energies provided by the Titania system [28].

4. NIKE LASER

Nike is a 56 beam KrF laser facility that was built to exploit the desirable characteristics of KrF lasers for inertial fusion. It has been used to conduct laser-target experiments and has also acted as a test bed for high-energy KrF development for 20 years [14]. A sampling of recent laser-target interaction experiments is provided in [38 –41]. In this section, we describe the Nike system and its operating parameters. A schematic of the optical system shown in Fig. 3 illustrates the above beam-multiplexing idea, plus the front end that allows the implementation of ISI and temporal pulse shaping. ISI is one of Nike’s most distinguishing characteristics because it provides the most uniform target illumination of all high-energy UV lasers.

 

Fig. 3. Schematic of the Nike optical system. The angularly multiplexed beams are expanded to fill the full aperture of the two electron-beam amplifiers and then are reflected off a concave mirror behind each amplifier (double-pass amplification), which contracts the size of the beams prior to reaching downstream optics, as shown in Fig. 6.

Download Full Size | PDF

A. ISI and Its Implementation on Nike

All large laser systems have many optical elements that introduce phase distortion; at the focal plane, these aberrations normally lead to interference effects that create large and uncontrollable illumination nonuniformities. ISI solves this problem by introducing enough spatial and temporal incoherence to average out that interference, at least on time scales of interest to the target physics. The ISI concept [42] was originally implemented on the NRL Pharos glass laser [43], which operated at bandwidth of $\sim 1\mathrm{THz}$ and hence short coherence time $t_c\sim 0.7\mathrm{ps}$. It used a pair of reflecting echelons to break up the output beam into a two-dimensional (2D) array of multiple beamlets whose widths $d$ are small compared to the full beam width $D$. The beam divergence is characterized by the large number $N=D/d$ of beamlets across the beam; if $N$ is large enough, each beamlet will see much less phase distortion than the full beam, and its individual focal profile can be made nearly diffraction limited.

Because of interference among the beamlets, the instantaneous focal spot intensity is composed of highly nonuniform random speckle whose envelope is the desired distribution. To create the incoherence, the echelon steps impose time delays slightly longer than $t_c$ between each of the beamlets, thereby forcing their phases to become random and independent. This in turn forces the speckle patterns to change randomly within coherence time $t_c$, which is much shorter than the effective hydrodynamic response time. Because all the beamlet phases change independently, all spatial modes of the speckle are randomized each coherence time and are thus all smoothed at the same rate. For an effective averaging time $t_{\mathrm{av}}$, there are $t_{\mathrm{av}}/t_c$ statistically independent patterns, so the root mean square (rms) nonuniformity in each speckle mode reduces to ${(t_c/t_{\mathrm{av}})}^{1/2}$ and approaches zero at sufficiently long averaging times. This is not the case for some other beam smoothing schemes, such as SSD [12], where the phases at nearby locations across the laser aperture cannot be independently randomized and the long wavelength speckle thus remains unsmoothed [44].

The echelon version of ISI was successfully implemented in several Pharos target experiments operating at both 1054 nm and frequency doubled 527 nm wavelengths. The laser target interaction experiments using ISI showed that the laser plasma instability is reduced with the ISI smoothed laser light [43,45 –47]. This was at least partially caused by suppression of laser filamentation in the target’s coronal plasma by the time-average smooth beams for bandwidths greater than $\sim 200\mathrm{GHz}$. For the bandwidths near 1 THz there was evidence of direct suppression of stimulated Brillouin scattering by ISI [45,48]. Bandwidths much larger than 1 THz might directly suppress other laser plasma instabilities.

The original implementation of ISI required that the echelons be placed in the high-energy beams at the end of the Nd:glass laser system. It was not applicable to frequency-tripled 351 nm light because the bandwidth was limited by that set by the tripling crystals. SSD solved the problem of beam smoothing with frequency tripled light but required a phase plate [49] in the high-power beams. Reductions in laser plasma instability were also overserved with SSD beam smoothing [50,51]. Because of this reduction in laser-plasma instability, beam smoothing is utilized for indirect drive research as well as direct drive. Both the original implementation of ISI and SSD require one to utilize a complex optical element located in the high-power beams at the end of the laser. In addition, both require one to exchange that optic in order to change the focal distribution. For the case of the National Ignition Facility one would need to exchange one hundred and ninety-two 30-cm aperture phase plates. As discussed below the characteristics of the KrF laser allowed a much more flexible implementation of ISI that does not require a complex optic placed in the high-energy beams at the end of the laser.

A KrF laser such as Nike allows a more flexible version of ISI, called echelon-free ISI, which places the source of the incoherence in the front end [13,52 –54]. This allows the on-target time-averaged beam distribution to be encoded in the low-energy portion of the system, as shown in Fig. 3. It works because KrF requires no harmonic conversion and because its gaseous medium and lower-intensity pulse trains in the angularly multiplexed beams ensure low nonlinear refraction. Using ISI, Nike has produced smooth flat-topped focal distributions with residual fluctuations of 1% rms in a single beam at 1 THz FWHM bandwidth (coherence time $\sim 0.7\mathrm{ps}$); with 44 overlapped beams, the nonuniformity is too small to measure accurately. The detailed mode spectrum has been measured for a single beam with different averaging times and is in agreement with theory [54]. The Nike bandwidth has been extended to 3 THz (coherence time $\sim 0.35\mathrm{ps}$) for some target experiments by broadening and shaping the input spectrum in the front end.

The ISI concept is illustrated schematically in Fig. 4. A spatially and temporally incoherent light source overfills and uniformly illuminates the front-end (object) aperture; this is currently chosen to be a hard aperture but could also be apodized to create a shaped envelope profile. This object profile is then imaged through the laser system to produce the final focal spot envelope. To avoid imprinting gain nonuniformities onto the focal spot, each amplifier stage is placed at or in the near-field of the object’s (and target’s) Fourier transform (FT) plane. At the FT planes, each object point encodes itself into a bundle of parallel rays whose field angle is determined by that point’s off-axis location. (The shifting interference patterns due to crossing of rays originating from the different incoherent object points create localized coherence zones, which are analogous to the independent beamlets of the echelon ISI.) To avoid cross talk between the angularly multiplexed beams, the field-angle divergences are always much smaller than the angular separations of the beams.

 

Fig. 4. ISI concept, showing amplifiers placed at images of the object’s Fourier transform (FT) plane. The instantaneous speckle and smooth time-averaged profiles are illustrated for the case of a flat-top time-averaged object profile. The focal spot blurring results from random phase aberration, as illustrated by the small additional ray divergences beyond the amplifiers.

Download Full Size | PDF

The near-field requirement on the FT plane locations is easily satisfied at large apertures where the beams are wide and the field angles are small. This ensures that the separation of the ray bundles for different object points remains only a small fraction of the beam width as one moves away from the exact FT plane. For example, the Nike 20 cm amplifier is located $\sim 80\mathrm{m}$ before the FT plane at the 60 cm final stage but is not image relayed to it; in spite of this, target beam tilts and curvature due to the 20 cm gain nonuniformities remain negligible. The near-field requirement is more challenging in the front-end amplifiers where the small apertures and large field angles can introduce small tilts in the target-beam profile. The suppression of these tilts is discussed below in Section 4.C.

Amplifier phase nonuniformities distort the target beam profile by introducing localized refractive perturbations in the field-angle distribution created by the object, as illustrated by the small spread in the field angles beyond the amplifiers shown in Fig. 4. As long as these perturbations remain small compared to the field-angle divergence, their main effect is to blur the image in a reproducible way rather than impose any small-scale nonuniformities. The Nike optical design allows compensation of systematic phase aberrations associated with the angular multiplexing such as astigmatism (see Section 4.F). As discussed later phase perturbations resulting from optical imperfections and air turbulence can be limited to about 12 times diffraction limit (XDL) on target. Nike typically uses a circular flat-top object profile that gives a flat-top angular distribution of 75 XDL diameter. So this is an acceptable effect that softens the edges of the initially top-hat shaped distribution from the front end.

B. ISI Oscillator and Object Aperture

In its initial configuration, the output from a multimode KrF oscillator was passed through a diffuser and focused on the ISI aperture. The spatial incoherence of the oscillator alone was sufficient to fully illuminate the aperture. The holographic diffuser with a flat-top pattern produced a more uniform illumination of the aperture at the expense of energy loss as the aperture was overfilled. This necessitated use of an additional amplifier. This system worked well and provided the needed uniform illumination of the object aperture, which provides the desired flat-topped intensity profile for planar target experiments. In 2003, this system was replaced by a single two-pass amplifier, where amplified spontaneous emission (ASE) from the first pass was spatially filtered to the desired divergence and reflected by a mirror for a second pass. The aperture in the spatial filter was adjusted so that the focused laser beam appropriately overfills a downstream ISI aperture that determines the profile on target. The setup again produces profiles of $\sim 1.3\%$ time-averaged uniformity over a 4 ns pulse length. The observed single-beam time-averaged uniformity from this system in the front end was at the limit set by the laser coherence time ($t_c=0.7\mathrm{ps}$) and a 4 ns averaging time. ${(\mathrm{\Delta }I/I)}_{\mathrm{rms}}={(t_c/t_{\mathrm{avg}})}^{1/2}$. We observe the expected increase or decrease in ${(\mathrm{\Delta }I/I)}_{\mathrm{rms}}$ when the coherence time of KrF light is increased or decreased (to 0.35 ps) by using etalons to adjust the bandwidth.

C. Obtaining a ‘Top Hat’ Focal Profile

The initial amplification is performed by a commercial UV pre-ionized discharge-pumped laser with a $1\mathrm{cm}\times 2\mathrm{cm}$ aperture and 80 cm length. While the Fourier plane of the ISI aperture is relayed into the center of the amplifier, the focal profile can be imprinted with tilts and curvature by gain nonuniformity in the amplifier due to its small aperture and long length. This effect was minimized by using a discharge amplifier that produced a relatively flat-top gain profile both vertically and horizontally. This was sufficient to nearly eliminate curvature, but several percent of tilt typically remained. To suppress linear tilt variations in the profile, a set of beam splitters and mirrors was installed after the amplifier that splits the beam, flipping one both horizontally and vertically relative to the other, and then recombines them. The result effectively cancels the tilt, producing a uniform composite [53], and profiles that are flat to within 1% could be obtained. The larger aperture x-ray pre-ionized discharge amplifiers that follow had sufficient gain uniformity that imaging the Fourier plane into each amplifier’s center was sufficient to obtain uniform-tilt and curvature-free focal distributions. In future systems utilizing ISI, it would be wise to develop and utilize larger aperture (e.g., $2\mathrm{cm}\times 2\mathrm{cm}$) discharge-pumped pre-amplifiers than used on Nike.

D. Pulse Shaping and Focal Zooming

The temporal pulse shape is produced using combinations of polarizers and Pockels cells. Throughout the front end the pulse shape is carried by one target beam while the remainder of the pulse from the oscillator copropagates with an orthogonal polarization and slightly different angle. The discharge-pumped amplifiers thereby see a single 20 ns duration pulse regardless of the duration of the pulse desired on target. The target pulse is separated from the control beam after the final discharge-pumped amplification. This reduces modification of the target pulse shape due to saturation of the amplifiers and also reduces the ASE from the discharge-pumped amplifiers. This system utilizes two Pockels cells in series and provides a contrast ratio of $\sim 5000$. The contrast ratio from the front end 10–15 ns before and after the pulse is effectively infinite due to the short pulse duration of the amplifiers (15–20 ns). The nominal pulse shape used for ablative acceleration and shock experiments is a 4 ns rectangular pulse with various (0.3–4 ns duration) lower power “foot” pulses ahead of it as needed by the experiments. The system is capable of producing pulse lengths as short as 300 ps with adequate energy to saturate the subsequent electron-beam-pumped amplifiers. For these short pulses the control beam is either attenuated or eliminated so that the short pulse gains additional energy due to the few ns lifetime of the KrF excited state.

Incorporating ISI into the optical design also allows for the straightforward implementation of focal zooming, in which the on-target spot size changes with time [19]. This allows for a more efficient coupling of energy to an imploding target and as discussed earlier may be essential for suppressing CBET. The implementation of zooming is shown in Fig. 3. The pulse from the ISI oscillator is split into two beams that pass through two separate pulse shaping and ISI object apertures. The ISI aperture of one is different from the other. The beams are recombined with a beam splitter and form the beam sent to target. This system allows one to zoom up or down the intensity profile on target with arbitrary pulse shapes. For example one can use a longer pulse with a large focal diameter to produce a large spatial scale-length plasma, followed by a shorter duration pulse with a smaller focal diameter to study the laser-plasma interaction at high irradiance. The ICF application may require more than one zoom, and that can easily be achieved by splitting the oscillator beam into more than two beams and adding additional ISI object apertures and pulse shaping systems.

E. X-Ray Pre-Ionized Discharge-Pumped Lasers

The stage of amplification immediately prior to the electron-beam-pumped amplifiers is performed using $4\mathrm{cm}\times 4\mathrm{cm}$ aperture x-ray pre-ionized discharge-pumped module [55] developed as part of the Nike laser program. These amplifiers have short duration gain (16 ns) and can provide up to 2 J energy. The main beams are amplified by one of these amplifiers, spit into three beams and further amplified by an array of three amplifiers. The separate backlighter beams discussed later utilize their own $4\mathrm{cm}\times 4\mathrm{cm}$ amplifier.

F. Angular Multiplexing and Aberration Compensation

Angular multiplexing is used to extract energy from the two electron-beam-pumped amplifiers. A photograph showing some of the time-encoding and decoding optics is shown in Fig. 5. Because the beams are striking a curved optic at non-normal angles, aberrations in the form of astigmatism and coma are generated, with the former being dominant. Astigmatic aberrations generated in each beam are compensated for by setting an appropriate tilt in a dedicated lens in its beam line. In Nike, these compensating lenses are incorporated into the mirror array structure that guides the beams into the 60 cm amplifier [see Fig. 6]. While the appropriate tilt angles can be calculated using optical ray tracing programs, the final adjustments are performed via measurement of the on-target profiles. Years of monitoring have shown that, once the proper angular settings have been established, few changes are required.

 

Fig. 5. Nike laser propagation bay showing the time decoding mirrors as viewed toward the 60 cm amplifier aperture.

Download Full Size | PDF

 

Fig. 6. Top view of the Nike 60 cm final amplifier, showing one row of the input and output optical feed arrays and one of the 56 angularly multiplexed beams, which overlap at the amplifiers. The concave lenses in the input feed array can be independently tilted to correct the astigmatism in each beam.

Download Full Size | PDF

G. Beam Propagation in Air

The propagation of UV radiation at 248 nm can be degraded by propagation in air through absorption by contaminants such as ozone or hydrocarbons. The most relevant section is the beam path from the final amplifier through the de-multiplexing optics to the target chamber. In order to mitigate contaminants, the propagation bay is hermetically isolated from both the amplifier hall and other rooms of the Nike facility. Ozone is formed after each shot, but it is effectively dissipated by the ventilation system over the time before the next shot. With this system the loss due to absorption and scattering in air was measured to be less than 10% for the longest optical paths (110 m) from the final amplifier to the target chamber.

Nonuniformity in the air due to convection or turbulence is another source of aberration in the beam paths. Because the ratio of beam size to scale length of this nonuniformity is small in the low energy portion of the laser, these effects are negligible in the front end. However, beam size in the propagation bay ranges from 15 cm to 60 cm, so care was needed to avoid distortion of the beams by nonuniformity in the optical paths. The air conditioning system was designed to maintain the walls and ceiling of the propagation bay at a uniform temperature while the floor was insulated. With this system, there was little distortion observed for the smaller aperture (15 cm) beam paths in the propagation bay. However there is observable distortion due to air nonuniformity in the double pass through the 60 cm amplifier and the tunnel connecting it to the propagation bay. This aberration is minimized by (1) stirring the gas in the 60-cam amplifier after a shot and waiting at least 20 min for it to settle and (2) maintaining the propagation bay and large amplifier room at as close as possible to the same temperature. With these measures in place, the total aberration on target from the atmospheric aberrations and Nike optical system was as low as 12–14 times the diffraction limit for 248 nm light. We discuss later the quality of the ISI-smoothed focal distributions obtained on target.

H. Electron–Beam-Pumped Amplifiers

Two electron-beam-pumped amplifiers, with $20\mathrm{cm}\times 20\mathrm{cm}$ and $60\mathrm{cm}\times 60\mathrm{cm}$ apertures, serve as the final amplifier stages for the Nike laser. While the smaller of these amplifiers has been a reliable workhorse [56], most of the developmental work has been performed on the larger amplifier, and therefore it will be the focus of the discussion here. A schematic of the layout of the 60 cm amplifier is shown in Fig. 7. A 12-stage, bipolar Marx bank is used to charge four pulse-forming lines (PFLs) to up to 1.2 MV. When the PFLs are fully charged, an external laser is used to trigger four spark gap switches. This transfers the voltage to the cathode ($204\mathrm{cm}\times 64\mathrm{cm}$ area). An anode/cathode (A/K) gap of $\sim 5\mathrm{cm}$ produces an impedance matched beam of $\sim 0.5\mathrm{MeV}$ and $\sim 400\mathrm{kA}$. The laser cells are filled with a krypton–fluorine mix with argon as a buffer. The electron beam passes through two titanium foils. The first (${0.001}^{\prime \prime }$ thick) provides a flat anode plane so that a uniform electric field will be generated between the cathode and ground plane. The second foil (${0.002}^{\prime \prime }$) serves as a barrier between the diode vacuum and the $\sim 1$ atmosphere of gas in the laser cell. Approximately 45% of the electron energy is deposited in the gas [57]. Electron beams are fired from opposite directions into the gas, providing a uniform deposition of energy across the cell. The laser gas mixture ratio is adjusted to provide maximum energy while maintaining a nearly uniform gain over the aperture. A photograph of the 60 cm amplifier is shown in Fig. 8.

 

Fig. 7. Schematic of the Nike 60 cm electron-beam-pumped amplifier.

Download Full Size | PDF

 

Fig. 8. Final amplification stage of the Nike Laser: the 60 cm aperture E-beam-pumped amplifier.

Download Full Size | PDF

When Nike was initially constructed [57], each cathode consisted of a large aluminum plate covered in velvet, which served as the electron emitter. High emission points on the cathode would cause burns in the velvet, limiting lifetime to roughly 200 shots. In addition, electron beams such as this interact with the diode structure and produce radiofrequency (RF) waves along the cathode surface. The RF wave then interacts with the electrons, causing them to decelerate. This interaction, known as the transit-time instability, generated an energy spread of 150 keV [32,58] among the beam electrons.

Development efforts produced two cathode designs that effectively eliminated this instability [58,59]. A schematic of the first cathode is shown in Fig. 9. In this design, a 5 cm thick piece of ceramic (cordierite) honeycomb (300 ppi pore density) is placed 2 mm in front of the velvet emitting surface. Electrons emitted from the velvet enter the capillaries, generating secondary electrons and plasma that electrically connects the velvet to the far side of the honeycomb, which defines the beginning of the A/K gap. The presence of the dielectric in the A/K gap suppresses the generation of RF waves and hence the transit-time instability. In addition, since the explosive emission from the velvet fibers is greatly reduced as compared to the flat cathode, the lifetime of the emitting material is extended. Furthermore, the uniformity of the electron beam is greatly enhanced.

 

Fig. 9. Schematic of the ceramic cathode used in the 60 cm amplifier.

Download Full Size | PDF

Three variations of this cathode were installed in the Nike laser. The first was replaced after 216 shots. Its replacement, which was reduced in dimension to better match the foil aperture into the cell, fired 604 times before structurally failing. The failure produced large amounts of ceramic dust. Inspection of the velvet emitter after the failure revealed no obvious erosion or damage to its surface. While upgrades to improve robustness have been contemplated, e.g., use of a tougher ceramic, the use of ceramic cathodes was suspended in favor of the more durable second design referred to as the patch cathode.

A schematic of the patch cathode is shown in Fig. 10. Here, the cathode surface is slotted along both dimensions, producing an array of 750 velvet emitting patches each $3.4\mathrm{cm}\times 3.4\mathrm{cm}$. Microwave absorbing material is placed at the bottom of each slot. In addition, the slot width depth and pitch are selected so that the phase velocity of the wave associated with the instability is near zero.

 

Fig. 10. Patch cathode used on the Nike 60 cm amplifier.

Download Full Size | PDF

To eliminate high electron beam intensities produced at the edges of the cathode, the A/K gap is graded so that it increases near the radiused frame, effectively eliminating this nonuniformity. In addition, it was found that covering the frame with velvet prevented ablation of the metal, caused by high emission points occurring with both aluminum and steel frames. The patches are individually removable and can be manufactured offline. If necessary, multiple patches can be replaced in minutes without removing the cathode structure. This type of cathode has been in service on the Nike 60 cm amplifier for six and a half years and for over 2300 shots. Full replacement of the patches and frame velvet occurs approximately every two years ($\sim 750–900\mathrm{shots}$).

I. Optics Performance

One of the main design goals of the Nike optical system was to maintain low beam fluences on the components, typically $<0.25\mathrm{J}/{\mathrm{cm}}^2$. While this resulted in larger beam sizes, and hence higher component costs, it removed the necessity of operating in clean room conditions and has resulted in lower replacement rates. In the Nike laser beam paths there are 48 optical elements of dimension 7.6 cm or less and 590 optics with dimensions greater than 10 cm. The smaller optics that are in higher fluence areas (discharge amplifier windows or near the focus of a spatial filter) require cleaning or replacement several times a year. Few of the larger optics exposed to intermediate energies have ever been replaced. Of the optics that define the path between the final amplifier and target, approximately 10% have been replaced over the lifetime of Nike (20 years). The majority of the damage has occurred from specific causes or events, such as hot spots caused by backreflections or anomalies in the optical material. These are exclusive of the laser windows, which are discussed below.

The energy produced by the Nike laser is sensitive to the performance of the optical windows of the 60 cm electron-beam-pumped amplifier. The windows are of dimension $660\mathrm{mm}\times 720\mathrm{mm}\times 51.5\mathrm{mm}$. Of the 10 windows in Nike’s inventory, six are made of Corning 7940 fused silica, while four are made of Corning 7980 fused silica, all of grade 3C/UV Excimer. In the early years of firing the amplifier, the windows were coated on both sides with a fluorine resistant antireflection (AR) coating similar to that developed for the Aurora final amplifier [34,60]. This configuration resulted in the highest output energies (4–5 kJ) [49], but the lifetime of the AR coatings that were exposed to the laser cell was very limited, with performance degrading within 50 shots and complete coating failure, i.e., damage/peeling, within hundreds of shots. Such coating failure renders the windows unusable. The approach was abandoned and operations switched to employing windows AR coated on the external side only. This resulted in a modest ($\sim 10\%$) reduction in the energy. These uncoated windows had longer longevity (several hundred shots) before repolishing of the inner surface is needed. Uncoated windows have been in operation on Nike since 2000. Transmission measurements of the optical path through the amplifier, to the mirror and back through the amplifier again, are performed on a periodic basis. For the measurement, an approximately $2\mathrm{cm}\times 2\mathrm{cm}$ KrF beam is propagated through the system, expanding to about 50% of the full window area aperture and then focused by the 60 cm mirror back through the amplifier and onto a calorimeter. A portion of the original beam is directed by a beam splitter to a calorimeter for calibration of the input energy. Since the 60 cm amplifier mirror has a reflectivity of 89%, the best transmission that single-coated windows can provide is 75%. Typical measured transmission at 248 nm through the double-passed 60 cm with freshly polished windows is 64%–68%.

These windows suffer slow degradation, with typical lifetime ranging from 100 to 250 shots. In time they show deterioration of the inner surface that causes scattering of the laser beam. Used windows have been refurbished in several ways. The first method is simply to have the uncoated side repolished by a vendor who would make the best effort not to damage the AR coating. This produced optical surfaces with no fogging or blemishes. This is costly and time consuming. As a compromise, windows have also been polished by NRL personnel using a polishing compound. While this restored most of the transmission, a slight haze remained on the surface that could not be removed without more vigorous action. This causes scattering of the beams that can result in prepulse on target [61]. Transmission degradation as a function of shot count is similar to that of vendor polishing. In more recent years, cleaning of the windows in situ has been performed, by wiping the windows with lens tissue soaked in deionized water and then ethanol. This is repeated numerous times. The advantage of this method is in cost and time. It is performed whenever the laser cell is opened, typically during maintenance of the pressure foils, and surprisingly has resulted in significant restoration of the windows, often approaching the initial installation transmission. This indicates that much of the transmission loss arises from debris on the window, with the most likely culprit being titanium fluoride generated when the pressure foil is heated during a shot. Typical examples of transmission drop though the amplifier with windows cleaned using the methods described above is shown in Fig. 11. There has been no observed evidence of either UV or x-ray-induced solarization of these windows.

 

Fig. 11. Plot showing degradation of the overall 248 nm transmission of the 60 cm amplifier optics versus number of shots on the windows.

Download Full Size | PDF

The limited lifetime of the Nike windows reflects several factors. Dry fluorine will not attack the synthetic fused silica windows. However, if water is present hydrofluoric acid is formed that can etch the window surface. The electron beam amplifiers are fired several times without fluorine in the gas mixture and pumped to vacuum to remove water after a window is installed. Use of a cryo-pump to help remove water was found to increase the longevity of the windows. Nike uses titanium pressure foils that can react with the fluorine at high enough temperature to form titanium fluoride. Titanium fluoride dust may be slowly contaminating the windows. As discussed later the Electra system uses stainless steel pressure foils and the longevity of windows are much longer, $>\mathrm{10,000}$ shots.

J. Performance of Nike for Target Experiments

Nike is primarily used for basic studies of laser interactions with matter in planar geometry to help develop the physics underpinning for inertial fusion. Nike provides 44 main beams for the target experiments, and an additional 12 beams are available for other applications and diagnostics such as x-ray backlighting of the main target [14]. The two sets of beams have separate temporal pulse shaping systems in the front end. The two basic types of research are (a) studies of high-pressure shocks and the hydrodynamics of ablatively accelerated targets [38,39,41] and (b) studies of laser plasma interactions and instabilities at high intensity [40]. The former requires highly uniform illumination of the target with a flat-topped 0.5–1 mm diameter focal profile with multi-nanosecond pulses. To reach the intensities required for the latter ($\ge {10}^{15}\mathrm{W}/{\mathrm{cm}}^2$), Nike is configured to deliver its energy in shorter pulse (0.3–1 ns) and with smaller focal diameters. It has delivered up to 3 kJ on target in 44 beams using 4 ns pulses and routinely provides $\sim 1.7\mathrm{kJ}$. For the short pulse experiments it delivers about half that energy. We discuss here performance and capabilities of Nike for target experiments.

1. Prepulse on Target due to ASE and Beam Cross Talk

The laser energy arriving on target prior to the intended main pulse needs to be small enough that it does not destroy or modify the target. Nike can provide low enough prepulse for most experiments. As briefly discussed below, the small existing prepulse due to ASE can be an advantage under particular circumstances. A detailed discussion of sources of prepulse as well as measurements of prepulse intensity under various laser conditions on Nike is given in [61]. We give a brief summary of the results of that work here.

On Nike, prepulse occurring within approximately 10 ns of the main pulse is dominated by the pulse shaping and amplification in the front end of the laser prior to beam multiplexing and is present on all the beams. Prepulse arriving to the target at earlier times has its origins in the time and angularly multiplexed electron-beam-pumped amplifiers. This early time prepulse is different for each individual beam and has two sources: ASE and beam-to-beam scatter (BBS) in optics where beams overlap.

Of the two angularly multiplexed amplifiers, the 20 cm is measured to have $\sim 100\times$ stronger contribution to the prepulse intensity on target [61]. The dominance of the 20 cm amplifier is expected from the following. For ASE, intensity produced by a given stage is proportional to the stage gain as well as the gain of subsequent stages. For BBS, intensity originating in the 20 cm is nominally $9\times$ higher than that of the 60 cm because BBS scales inversely with the square of the amplifier aperture. However, it is also dependent on the actual scattering function of the amplifier windows.

While the two amplifiers are on, ASE from them illuminates the optics that collect the individual laser beams. Because of time multiplexing, ASE produced at a particular time while the amplifier is on, and collected by the optics for the beams that have not yet gone through the amplifier, reaches the target prior to the arrival of the laser beams, thus contributing to the prepulse. The contribution to the prepulse is particularly strong at the beginning of the amplifier gain window when the amplifier is not yet being loaded down by the laser beams and all the light collected by the individual beam optics arrives at the target early, albeit at various times. The ASE from the end of the gain window arrives at the target after the laser beams and thus does not contribute to the prepulse. The earliest ASE arrives on target earlier than the laser pulse by the delay between the amplifier turn on and the time the last beam exits the amplifier: approximately 155 ns and 285 ns for the 20 cm and 60 cm amplifiers, respectively.

The primary source of BBS is scattering due to surface roughness of the amplifier windows and mirror. The inner surfaces of the windows are exposed to the harsh environment of fluorine compounds in the amplifiers. Over time they become etched and increasingly scatter the amplified light. (Minimizing presence of water in the laser chamber has been found to prolong the life of amplifier windows on Nike.) Light from a particular beam, designed to be collected by a particular element of the output array, can thus be scattered into the path of another beam. As with the ASE, time multiplexing allows scatter from an earlier beam to enter the path of later beams and thus arrive on target earlier.

BBS is expected to be largely due to small angle scattering. BBS from a particular beam is thus expected to scatter mostly into the paths of beams whose collection optics are nearby. The output arrays of the amplifiers are arranged so that beams that are nearby in angle are also close in time, minimizing the length of the prepulse on target due to BBS. Nevertheless, the earliest prepulse on target due to BBS will arrive on target earlier than the laser pulse by the time difference between the first beam and the last beam. For the 20 cm and 60 cm amplifiers, this time is approximately 120 ns and 240 ns, respectively.

Prepulse has been measured to be $\sim 2\times {10}^{-7}$ from peak for approximately 120 ns prior to the main laser pulse under standard operating conditions (standard amplifier timing, no auxiliary ASE suppression beams, and new amplifier windows). Prepulse fluence on target is approximately $2–3\mathrm{J}/{\mathrm{cm}}^2$. This prepulse level is insufficient to result in significant preheating of a plastic target, where the absorption depth for 248 nm is $\sim 3\mathrm{\mu m}$. It could, however, be significant for targets coated with thin ($\sim 400\AA$), low-heat-capacity metallic coating, such as those used for imprint suppression [41,62]. Since most of the prepulse light is absorbed in the metallic coating, this could result in significant pre-expansion of the coating. A possible benefit to pre-expansion could be increased separation between the laser absorption surface and the ablation surface, decreasing the imprinting of laser nonuniformity onto the target. The minimum prepulse fluence required to affect a target with such a coating would be on the order of the fluence required to vaporize the metallic layer: $\sim 0.2\mathrm{J}/{\mathrm{cm}}^2$ for 400 Å gold. The thin metallic layer is thus expected to be affected by the prepulse since the measured prepulse is an order of magnitude higher than this estimate.

Prepulse measurements on Nike were conducted with ASE suppression by a change in amplifier timing, as well as under the conditions of heavily used amplifier windows. As expected, prepulse has been found to depend strongly on the ASE suppression and window condition of the 20 cm amplifier. Loading the amplifier 20 ns earlier results in a significant decrease in the prepulse level, as seen in Fig. 12. Under this condition of reduced ASE, replacing heavily used 20 cm amplifier windows with newly polished ones result in almost an order of magnitude decrease in prepulse level. The latter underscores the need for timely window replacement if minimizing prepulse is important. The former indicates that use of auxiliary beams from the laser front end to load the amplifier as it is ramping up would significantly decrease the prepulse level on target.

 

Fig. 12. Prepulse intensity on target showing the effect of suppression of 20 cm amplifier ASE. The solid trace is an average of three shots with standard amplifier timing. The dashed trace is an average of two shots with the 20 cm amplifier firing delayed by 20 ns. In the latter case, the amplifier is loaded during its ramp-up, decreasing the ASE.

Download Full Size | PDF

2. High Power Short Pulse Operation

Nike utilizes air paths for the beam propagation in the angularly multiplexed system. The beams after the 60 cm amplifier were “oversized” at $15\mathrm{cm}\times 15\mathrm{cm}$ to avoid both damage to optics and to reduce nonlinear interactions in the air paths. Nonuniformity in the propagating beams can enhance the nonlinear interaction. The near-field beam nonuniformity due to imperfections in the optics and air paths is reduced by the beams’ partial incoherence produced by the ISI smoothing [63]. No measurable nonlinear effects are observed for 4 ns beams operating at 10 GW. However, for the shorter duration 300 ps pulses and 60 GW/beam power levels, there is measureable stimulated rotational Raman scattering (SRRS). With SRRS the laser light excites rotational quantum states of the ambient molecules in the air. For the UV region nitrogen (${\mathrm{N}}_2$) and oxygen (${\mathrm{O}}_2$) have the largest cross sections for this process. As SRRS growth occurs, laser light is absorbed and re-emitted over a slightly broader spectrum and wider range of beam angles, thereby reducing the energy on target, altering the on-target laser spectrum, and degrading the on-target focal profile if the SRRS were severe enough. For short pulse operation, due to the additional power output of the final amplifier, a low level of SRRS growth has been observed in the laser spectrum for the longest path beams. The effect was small for total beam energy ($\sim 3\%–5\%$) and negligible on the overlapped focal spot ($\lesssim 2\%$ broadening of a subset of individual beam spot diameters). As discussed later, future systems with higher power beams would likely need to utilize either vacuum optical paths or paths through a low nonlinear index inert gas such as helium.

3. Focal Distributions on Target

As discussed earlier the Nike front end can provide very uniform flat-topped profiles. We discuss here the focal profiles obtained by overlapping up to 44 of Nike’s ISI smoothed beams onto target. The $15\mathrm{cm}\times 15\mathrm{cm}$ main beams are focused onto targets using 6 m focal length fused silica lens. Measurements of the focal distributions of individual beams at high power indicate that the electron-beam amplifiers do not appreciably affect the focal distributions provided that one waits $\ge 20\min$ between shots for the e-beam heated laser gas to cool and stabilize.

Figure 13 shows an ensemble of on-target beam profiles for a variety of commonly used Nike operations. These images were recorded at low energy by placing a UV fluorescent glass witness sample at the center of the target chamber, overlapping all beams at a common spot, and recording the resulting emission profile with a microscope camera triggered by the laser timing system. The central image shows a case with a standard flat-top region that is used for hydrodynamic or EOS experiments. The 12–14 XDL aberrations in the optical system soften the originally hard edges of the focal distribution produced in the front end. The left image shows a smaller diameter spot commonly used for studies where high irradiance is desired such as studies of laser-plasma instabilities. The noise level in these images reflects the noise level in the CCD or fluorescer rather than the true level engendered by ISI beam smoothing, the images accurately show the 12–14XDL optical performance routinely achieved with optimized beam profiles. This aberration in the optical system places a lower limit of about 200 μm FWMH on the range of usable focal diameters. The third image demonstrates that the imaging properties of the laser system can also be used to project more complicated illuminations. In this case, the pinhole aperture in the front end of the laser has been replaced with a mask consisting of a periodic series of open slots. The image shows that a single beam creates a large-scale rippled pattern at the target plane. These ripples are significantly larger than the time-averaged ISI nonuniformities and allow investigations of the effects of such imposed nonuniformity on laser target interactions.

 

Fig. 13. Sample low-energy focal profiles obtained at the center of the Nike target chamber (top) with horizontal lineouts (bottom). A and B are profiles with all the main beams overlapped but with two different sets of pinhole apertures inserted in the front end. B represents standard operating conditions, while A represents smaller spots used for higher intensity experiments. C shows the profile produced by a single beam when a multi-slit aperture is used (inset). All of these images demonstrate the 12–14XDL performance routinely achieved. The short-scale low-level fluctuations in the images are dominated by the noise in the CCD detector.

Download Full Size | PDF

The above measurement system and the cooled CCD camera systems used to monitor the focal distributions of Nike beams can resolve the 1% rms residual time-averaged nonuniformity in the focal distributions of individual beams. But they cannot resolve the nonuniformity of 44 overlapped beams. However, Nike experiments studying laser imprint in ablatively accelerated targets do show a large reduction in imprint when utilizing many overlapped beams. These results are consistent with the expected $1/{(n)}^{1/2}$ effective reduction in illumination nonuniformity from overlapping $n$ independent ISI-smoothed laser beams (see for example Fig. 7 of [62]).

5. ELECTRA REPETITIVELY PULSED KRF LASER

A repetitively pulsed electron-beam-pumped KrF laser is one of the leading candidates for an IFE laser driver [64 –68]. The Electra laser has the same fundamental electron-beam-pumped architecture as Nike, and it has served as a development platform for laser components and technologies that can meet the fusion energy requirements for repetition rate, efficiency, and durability. A KrF laser system to be used for fusion energy should operate at a repetition rate of $\sim 5$ pulses per second; it should have a wall-plug-to-laser energy efficiency of 6%–7%; and it should operate continuously for up to 2 years before major maintenance is required, resulting in a major component lifetime of $3\times {10}^8\mathrm{pulses}/\mathrm{shots}$. This section reports the progress on the laser components and then discusses the laser performance during repetitive operations and KrF laser physics.

The major components of the Electra KrF main amplifier are shown in Fig. 14. They include a pulsed power system that charges a parallel set of deionized water-filled PFLs; two ${\mathrm{SF}}_6$ spark gap switches that are triggered by a frequency-quadrupled Nd:YAG laser; an electron beam diode that includes the electron beam producing cathode; a 25 μm thick stainless steel foil and a foil support structure called a hibachi; a laser gas cell enclosed by fused silica windows; and a laser gas recirculator. An external magnetic field produced by Helmholtz coils provides both efficient electron beam transport through the hibachi and low scattered deposition in the laser gas. The KrF laser has two opposing electron beam diodes to provide uniform pumping of the laser gas, which is signified by the identical components on the left and right sides of Fig. 14. The KrF laser system is operated both in an oscillator mode with a mirror and output coupler between the laser cell, or as an amplifier with a mirror and input seed laser that is angular multiplexed like Nike.

 

Fig. 14. Major components of an electron beam pumped KrF laser amplifier. Two pulse power systems on each side of the laser cell power two 500 kV 100 kA $100\mathrm{cm}\times 30\mathrm{cm}$ electron beams into the laser gas cell. A recirculator cools the gas for high-repetition-rate operation. Details of the pulse power and electron beam diode are shown in Fig 15.

Download Full Size | PDF

A. Pulsed Power System, PFLs, and ${\mathsf{SF}}_{\mathsf{6}}$ Switches

The first-generation pulsed power system supplies a 500 kV, 110 kA, 140 ns pulse to the cathode, at repetition rates of up to 5 pps. The lifetime of the system is limited by the erosion of the ${\mathrm{SF}}_6$ switch electrodes, which exceeds 150k shots [69]. A simple replacement of the switch electrodes restarts the life cycle. While this system has successfully tested various laser components, it does not meet the requirements for a fusion energy system. A schematic of this system is shown in Fig. 15.

 

Fig. 15. Schematic of Electra’s pulse power and electron beam diode.

Download Full Size | PDF

An all-solid-state system [70] has been developed to replace the spark gap switch-based pulsed power system. The solid-state system uses energy storage capacitors, inductors, magnetic switches, and solid-state closing switches (thyristors). Lifetime tests on thin film capacitors exceeded 300 M shots at 55 pps [71], and individual 4 kV, 7 kA thyristors also exceeded 300 M shots at 20 pps. The magnetic switches, mainly Metglas 2605SA1 and 2605CO, and inductors do not limit the lifetime of the solid-state pulsed power system with proper thermal management and high voltage insulation. The 200 kV, 4.5 kV, 300 ns solid-state system operated continuously at 10 Hz for over 11.5 M shots, and it was limited by the lifetime of the capacitors used in the system. With the appropriate thin film capacitors the system is expected to achieve a lifetime of 300 M shot, and it will become the basis for a future 500 kV upgrade on Electra. A fusion energy size 800 kV all-solid-state system with a 250–400 ns flat-top pulse and a 30–40 ns rise and fall is expected to have a wall-plug-to-flat-top pulse efficiency of 80%–85%.

B. Electron Beam Diode

One of the most critical components of a KrF laser system is the e-beam diode with its cathode and the thin metal foil that serves in Electra’s case both as the anode and the pressure membrane between the e-beam diode vacuum and the laser gas. The ESLI carbon fiber cathode on a carbon base [72] has achieved a lifetime of larger than 300,000 shots, and the tests were limited by Electra’s first-generation pulsed power system. To achieve high-electron-beam transport efficiency through the foil supporting hibachi, the cathode should be patterned into strips. This will maximize electron emission across the hibachi rib opening and minimizes electron beam losses to the hibachi ribs. The cathode strips should be counter-rotated to compensate for the rotation that arises from the interaction of the imposed guiding magnetic field and the self-field of the e-beam [30]. Furthermore, enhanced emission near the edges of the cathode strips should be reduced [73], and metallic areas on the cathode surface should be eliminated.

Electra, operated in a KrF oscillator mode, has achieved a lifetime for a 25 μm thick stainless steel foil of 90,000 shots at 2.5 pps, 50,000 shots at 5 pps, and 300,000 shots over an eight-day period. Minimizing/eliminating voltage reflections in the e-beam diode plays a critical role on the foil lifetime as it prevents localized high-current emission from the anode foil that are precursors for the generation of pin holes in the foil. During repetitive operation, the foil must also be actively cooled since conduction cooling to the hibachi ribs is not sufficient for thin foils and radiation cooling is not effective at foil temperatures below 400°C. Localized forced convection cooling by streaming a fraction of high-velocity laser gas at the foil has kept the Electra foil at operating temperatures around 300°C [74].

Electra achieved an e-beam power deposition efficiency of 75% during the flat-top duration of the pulse with a 25 μm thick Ti foil. This high efficiency required utilization of a patterned cathode that directed the electron beam between the hibachi ribs (see Fig. 16). Deposition efficiency into the laser gas of 80% should be reached for a 800 kV e-beam through a 25 μm thick stainless steel foil [75].

 

Fig. 16. Patterned cathode directs electron beam between hibachi ribs.

Download Full Size | PDF

Both the cathode and foil life were limited by the first-generation pulsed power system. Future lifetime tests of those components are planned with the solid-state pulser.

C. Electra Amplifier Operation

The KrF Electra laser system and performance has been reported elsewhere [76,77]. Briefly, a 1 J laser discharge is split into two beams each with a nominal 25 ns pulse length and used for input into a preamplifier. The electron-beam-pumped pre-amplifier with $10\mathrm{cm}\times 10\mathrm{cm}$ aperture amplifies the laser pulse to over 40 J per pulse. The output of the preamplifier is split from the two beams to six beams using sets of beam splitters. The six laser beams are angularly multiplexed into the Electra main amplifier to provide an output of 522 J. The time-dependent pulse shape of the laser beams is measured throughout the laser system to evaluate performance in laser yield and time-dependent intensity.

D. Optical Performance with Repetitive Laser Operation

Density perturbations within the gas mixture are created when the electron beam deposits its energy in the laser cell. These perturbations can last for several seconds on Electra or several minutes in a closed laser cell on Nike. They must be actively removed in order to operate at a high repetition rate with high beam quality on every subsequent shot during repetition rate operation. The Electra recirculator is a 9000 liter vessel, in which the laser gas flows at a nominal rate of $5–7\mathrm{m}/\mathrm{s}$ in the electron-beam-pumped region of the laser cell. The gas flow pushes the density perturbations downstream, but the expansion of the heated gas creates a density perturbation within the laser gain medium for the next shot. By using acoustic suppressors and thermal management, the perturbations were substantially reduced [78]. If left unchecked the density perturbations manifest into phase aberrations that distort the focal profile too much even with a 50–100 XDL ISI beam. A critical design consideration for a repetition rate target illuminator would be the accumulation of all the causes of density perturbations leading to phase aberrations with the full laser system. Electra has shown that at 5 Hz the 248 nm aberrations can be reduced to about 14 XDL in double pass [79]. The best configuration utilized laminar flow of the laser gas while the pressure foils were cooled by small gas jets in a collaborative design with Georgia Tech [80,81]. This optical performance is close to what is needed for a large aperture amplifier in an IFE system.

E. KrF Laser Amplifier Windows

Optic development is a key requirement to increase durability of electron-beam-pumped KrF lasers. We have shown fused silica does not have bulk losses over 250,000 electron-beam-pumped KrF laser shots [76]. The fused silica surfaces exposed to the laser gas do show signs of degradation within the amplifiers after long periods of use. However, over the initial 30,000 shots of a repetition rate run no change has been observed in fused silica window transmission for the back window. At 90,000 shots continuous operation we observed increase in scatterings representative of a flow pattern on the front fused silica window from operation of the recirculating gas for 10 h operation [76]. The minimal losses within the fused silica during electron-beam-pumped KrF laser operation is in agreement with the observed 2% change in window transmission over 250,000 shots in component testing with identical 248 nm fluence and fluorine concentrations as the KrF amplifier. To attain even longer lifetimes, the window material can change to nonfluorine etching materials such as sapphire, calcium fluoride, and magnesium fluoride. These materials are superior to fused silica in fluorine environments; all have excellent transmission at 248 nm [82]. Magnesium fluoride (${\mathrm{MgF}}_2$) windows have been shown to have billion shot life longevity in modern small KrF discharges lasers.

The experience with Electra indicates that UV grade synthetic fused silica has sufficient durability for 250,000 shot operation in electron-beam-pumped amplifiers. However, for an IFE power plant one would like to operate the amplifier for more than a year without refurbishing or replacing the windows, which corresponds to more than 158 million shots at 5 pulses per second. We believe that fogging from chemical attack can be resolved by using calcium fluoride or magnesium fluoride for the windows. There remain threats from ionizing radiation including scattered electrons and x radiation produced by the electron beams, which left unchecked could solarize window material and thereby produce unacceptable absorption of the laser light [83]. The threat from scattered electrons can be minimized by providing sufficient separation between the electron-beam-pumped gas and the windows. The threat from x radiation is reduced by using the high efficiency diode described earlier, which reduces losses to the foil support structures and thereby reduces the x-ray generation. Experiments indicate that the solarization is significantly reversed by annealing by the KrF light [83,84]. This is an area that needs further research. An Electra-size KrF amplifier utilizing much more durable all-solid-state pulsed power would be a very useful and probably essential tool to make further progress.

F. KrF Laser Efficiency

An important parameter for the energy application is the “wall plug” efficiency of the KrF system. The efficiency of an electron beam amplifier is determined by the efficiency of the pulse power system, the efficiency of transporting the electron beam into the laser gas, and the intrinsic efficiency of producing KrF light from the electron-beam-pumped KrF gas. The deposited electron energy into the mixture of argon, krypton, and fluorine creates excitations and ionization of various atomic and molecular species. A fraction of these species form the nascent KrF molecules. The KrF molecule has a bound excited state and a dissociative ground state, thus creating an automatic population inversion necessary for laser action. For electron-beam-pumped lasers the intrinsic efficiency is defined as the ratio of laser energy output divided by the electron beam energy deposited in the gas. In addition to the kinetics of the electron beam’s interaction with a particular gas mixture, the intrinsic efficiency is dependent on other details of the KrF system such as its gain length and the energy of the input beam. Measurements of the intrinsic efficiencies for electron-beam-pumped KrF lasers have ranged from 10%–14% [85 –87]. Similar intrinsic efficiencies and subsequent laser performance have also been shown for gas mixtures with the addition of helium [88]. The addition of helium increases the thermal conductivity of the gas and dissipates the excess deposited energy, which has advantages in certain repetition rate conditions. Electra during repetition rate operation of 1 Hz and 2.5 Hz with output of over 700 J per shot obtained an intrinsic efficiency of $\ge 9\%$ when operated as an oscillator. Higher intrinsic efficiency would be expected if it were operated as an optimized amplifier.

The Electra amplifier has achieved 75% deposition efficiency of the electron beam into the laser gas. As discussed later, larger amplifiers will use higher voltages where there is less loss to the pressure foil. Simulations indicated deposition efficiencies exceeding 80% should be achievable. Table 3 lists the potential wall-plug efficiency of a large (1 m aperture) electron-beam-pumped amplifier that utilizes efficient pulsed power and electron beam diode system. Simulations indicate 12% intrinsic efficiency should be achievable during the flat-top portion of the pump pulse. The parameters in Table 3 indicate wall-plug efficiencies approaching 8% from the final amplifier are realizable. We would expect net KrF IFE system efficiencies of about 7% due to additional losses in the optical system to target and power needed for auxiliary systems and the balance of the KrF laser. The kinetics of the KrF allows formation efficiencies of $\sim 25\%$. A main limit on intrinsic efficiency is the limited maximum extraction efficiency of $\sim 50\%$ set by typical measured gain to loss ratios of about 10 [89]. It may be possible to substantially increase the intrinsic efficiency if these parameters, particularly the gain to loss ratio, can be better optimized.

Table 3. Estimated Efficiency of an IFE Scale KrF Amplifier

View Table | View all tables in this article

G. KrF Laser Kinetics Modeling

There have been many efforts to model the complex interactions that determine the behavior of electron-beam-pumped amplifiers [90 –97]. One of the most recent development efforts is the Orestes code [97]. The Orestes code KrF kinetics modeling code accurately predicted the experimental observed output of the 756 J Electra KrF oscillator as well as the 522 J performance of the Electra main amplifier in a laser amplifier system configuration. Orestes includes the ionization and excitation arising from electron beam deposition in argon, krypton, and fluorine, the components of the laser gas mixture [98,99]. Orestes also incorporates the generated 24 plasma species from the electron beam deposition and follows them through 146 reactions. It models the time-dependent ASE in three dimensions [100] and has recently been generalized to simulate the transport and pulse-shape evolution in each of the beams that overlap within the angularly multiplexed highly saturated final amplifier stages. With its multibeam capability, the code is now used to aid in the design and evaluate the pulse shaping capabilities of ICF and IFE scale KrF drivers [101]. Therefore, Orestes is a complete KrF modeling code from the initial electron in the gas to the photon leaving the amplifier. The Orestes code has been successful in analyzing data from NRL amplifiers, but also the KrF electron-beam-pumped laser data from around the world. The code is used to aid in the design and evaluate the performance of future ICF and IFE scale KrF beamlines [101].

6. DEVELOPMENT OF KRF FOR ICF AND IFE

The KrF laser looks promising as a driver for both ICF and IFE applications. Use of KrF light is predicted to increases the gain of directly driven targets and to reduce the risk from both hydrodynamic and laser plasma instabilities. Current research also indicates that a KrF laser can provide the wall plug efficiency and high repetition rates needed for the energy application. In this section we discuss the potential and development needed for KrF as a driver for an ICF research facility and for laser fusion energy.

A. KrF as a Driver for ICF

As of this writing the National Ignition facility is operating with target experiments that advance the understanding of both direct and indirect drive. It has the energy thought to be required for ignition and significant gain with direct drive. However it may or may not achieve ignition parameters due to limitations in the NIF design and its frequency-tripled Nd:glass technology. There is nevertheless continued interest in obtaining robust ignition and high yield for both defense and the longer term energy applications. In this section we discuss the status, potential, and developments needed to build a megajoule-class KrF facility as an ICF research platform.

The Nike facility has operated continuously for 20 years and has supported numerous cutting edge experiments in laser matter interactions. It has thereby been demonstrated that a multi-kilojoule KrF laser-target facility can be operated reliably and profitably for ICF research. It has demonstrated the fundamental technologies of highly uniform target illumination, focal zooming, and the pulse shaping needed for planar target experiments. A megajoule-class KrF system would be constructed from multiple Nike-like KrF beamlines. It would be desirable to develop and employ beamlines with higher energy than Nike.

Our kinetics code simulations and experience indicate that the double-sided electron-beam-pumped amplifier configuration utilizing monolithic cathodes can be extended to 20 kJ energies. This would involve a combination of higher pump intensities, higher extraction flux from the preceding amplifier, and increasing the aperture from $60\mathrm{cm}\times 60\mathrm{cm}$ to $60\mathrm{cm}\times 100\mathrm{cm}$. Simulations and design studies indicate that still higher energies are feasible by segmenting the cathodes and going for still larger apertures. The present upper limit on UV windows is about $1\mathrm{m}\times 1\mathrm{m}$ due to a combination of manufacturing capability and the limited strength of current materials. So very large apertures ($>1\mathrm{m}$) and energies ($>30\mathrm{kJ}$) would likely require segmentation of the amplifier windows unless tougher UV materials can be developed. We therefore believe the more practical near-term path forward would be to develop a 20 kJ system where the final amplifier’s pulse lengths (240 ns) and cathode sizes are similar to those employed with Nike.

The Nike system utilizes angular multiplexing and de-multiplexing of the beam lines in air. Nonlinear effects were minimized by using oversized beamlet sizes with loading of the beam transport optics well below $0.5\mathrm{J}/{\mathrm{cm}}^2$ (exclusive of the amplifier optics). This is fine for operation with 4 ns pulses and for higher power operation with 1 ns pulses. However, as discussed earlier, with still shorter pulses and higher powers, measurable nonlinear effects are observed in beamlets in the longer propagation paths. For a larger system we would envision use of either He filled or vacuum propagation paths for the high power beams to reduce or eliminate these nonlinear effects. This would also enable higher beam fluxes and a more compact optical system [102]. Average fluxes of $2–3\mathrm{J}/{\mathrm{cm}}^2$ appear to be reasonable given the damage thresholds of current 248 nm optics.

Figure 17 shows a development path for a MJ-class KrF laser fusion facility. One of the functions of such a facility is to explore and demonstrate robust ignition and high gain with directly driven targets. The higher the laser energy deposited on the target, the more likely one is to fulfill this goal; 1–2 MJ energy is a reasonable target to balance cost against technical risk. However, NRL simulations indicate 0.5 MJ might be sufficient to obtain $60\times$ gain with conventional direct drive implosions and $>100\times$ gain with shock ignited targets [20].

 

Fig. 17. Development path to a KrF ICF driver.

Download Full Size | PDF

B. Scale-Up of Nike

In this section we briefly describe the scale-up from the Nike amplifier and its expected performance based on electron beam deposition and kinetic simulations. The scale-up involves using monolithic cathodes of similar size and current to that employed on Nike but higher voltage to accommodate the longer path (100 cm) through the laser gas. Much higher energy is predicted for this design by (1) utilizing a more efficient E-beam diode developed on Electra that would increase the beam transport efficiency from the $\sim 40\%$ observed on Nike to $\sim 80\%$; (2) increasing the diode voltage and laser aperture; and (3) increasing the input laser flux to more efficiently extract the laser energy.

Figure 18 shows calculated electron beam transport efficiencies into the laser gas with two sided pumping, where the diode separation increased from 60 cm on Nike to 100 cm and the voltage has been increased from 600 to 800 kV. The electron transport simulation code [103] does not account for losses to the hibachi ribs, but as demonstrated on Electra this loss can be made negligible by use of a patterned cathode that emits electron beams between the ribs. Efficiencies above 80% appear to be feasible when using gas mixtures with the lower ratios of krypton to argon. Nike operates in this regime.

 

Fig. 18. Calculated 800 keV electron beam energy deposition efficiency into a 100 cm wide laser cell for 20% Kr/80% Ar and 30% Kr/70% Ar laser gas mixtures (pressure ratio). A 25 μm thick stainless steel foil separates the laser gas from the diode vacuum.

Download Full Size | PDF

Figure 19 shows Orestes simulations for the Nike final amplifier scaled up to a $60\mathrm{cm}\times 100\mathrm{cm}$ laser aperture, pumped lengths $L_P$ of 250 cm and 300 cm, and incident beam intensity of $0.3\mathrm{MW}/{\mathrm{cm}}^2$ (corresponding to 430 J). These simulations also take advantage of the enhanced e-beam deposition discussed above to increase the pump power density to $420\mathrm{kW}/\mathrm{cc}$. In the baseline case (solid $I_{\text{Out}}$ line), $L_P=250\mathrm{cm}$, but the windows remain tilted at the same 14° used on Nike and their inner surface reflectivities remain at 4%; this gives an output energy of 17 kJ in 250 ns. To avoid parasitic feedback loss caused by front window reflection also striking the rear mirror, the second simulation (dotted line) increases the tilts enough to ensure that at least all the reflected light leaves the box in an axial distance $L\le L_P$ before it reaches the rear mirror. The third simulation (short dashes) removes the 4% window reflections, thereby increasing the output energy to 18.5 kJ. Our experience with Nike is that AR coatings have not proven durable enough when exposed to the laser gas in the operating system, so the windows are normally AR coated only on the outside. The final simulation (long dashes) thus retains the 4% reflectivity but increases the pumped length to $L_P=300\mathrm{cm}$, thereby giving an output energy of 21 kJ. Development of AR coatings is desirable and would increase the output energy by an additional $\sim 10\%$ but does not appear to be essential for the ICF application.

 

Fig. 19. Orestes kinetic simulations of a scaled up Nike amplifier with and without AR coatings on the interior surface of the windows.

Download Full Size | PDF

We believe this is near the practical limit for systems consisting of two opposing monolithic cathodes. Larger amplifiers could be constructed using segmented cathodes where several independent cathodes are placed in series along the axis. This leaves short unpumped regions between the cathodes. There are losses in these regions; however simulations indicate that they are acceptable. The advantages to segmented cathodes are that they allow smaller, more manageable pulse power drivers for each cathode element, and they reduce the required external magnetic field due to the reduced current in each cathode. This comes at the price of increased system complexity and the aforementioned losses in the dead regions.

Nike utilizes 56 beams apportioned among 44 main beams with 4 ns inter-beam delays and 12 backlighter beams with 5.2 ns inter-beam delays. Nike’s main beams can provide about 60% of their peak energy when amplifying short ($\ll 4\mathrm{ns}$) beams. If the new system utilized 2.5 ns inter-beam delays, our kinetics simulations indicate nearly 100% of the energy could be extracted in such short pulses since the delays would now be shorter than the excited stated storage times. This would give the facility additional flexibility in the pulse shapes that can be delivered to the target and would be particularly valuable for implementing shock ignition, which requires a short-duration (100–300 ps), high-power final pulse following a longer duration main pulse [20,104].

C. KrF Application to IFE

The requirements for an IFE driver are much more stringent and challenging. The driver must operate at high repetition rates ($\ge 5\mathrm{Hz}$), be very reliable, have sufficient wall-plug efficiency, and have low enough cost that laser fusion power plants’ power production is economically competitive. There are several paths to achieving IFE that were evaluated in a recent review by the National Academy titled “Prospects for Inertial Confinement Fusion Energy Systems.” [105 –107]. This review covered direct and indirect drives with lasers, diode pumped solid state and KrF laser drivers, inertial fusion with heavy ions, and magnetically driven approaches with pulsed power. Here we will discuss the potential and development path of KrF for IFE using directly driven targets.

For a practical fusion power system one wants to minimize the recirculating power that powers the laser and other auxiliary system. A recirculating power of about 25% is achieved when the product of the target gain (${\mathrm{G}}_T$), the energy gain of the lithium blanket that surrounds the target (${\mathrm{A}}_B$), and the laser efficiency ${\eta }_d$ is 10. This regime is considered entrée conditions for a practical inertial fusion reactor. One therefore needs (${\eta }_d\times {\mathrm{G}}_T{\mathrm{A}}_B\ge 10$). The blanket gain is projected to be $\sim 10\%$, so one needs (${\eta }_d\times {\mathrm{G}}_T\ge 9.1$). See Fig. 1.3 of [106]. The net efficiency of an IFE grade KrF laser system is projected to be 7%, so one needs target gains $\ge 130$. High-resolution 2D simulations indicate that one could obtain such target gains with KrF laser energies as low as 0.5 MJ [20] if one utilizes the shock ignition approach to direct drive [104]. This illustrates one of the big advantages of utilizing KrF for IFE: the required laser energy is reduced and that would allow construction of smaller, more modular inertial fusion power plants.

Electra has demonstrated operation at up to 700 Joules and up to 5 Hz for several hours limited primarily by its spark-gap-based pulsed power. Greater than 100,000 shot durability has been demonstrated in the cathodes and pressure foil. An IFE grade system might be expected to operate for several years between any major maintenance. A pulse power system utilizing a combination of silicon and magnetic switches has demonstrated operation for over 100 h at 10 Hz. With further development this technology should be sufficient for the IFE application. Such reliable pulse power could then be utilized to develop and test the other components of a KrF amplifier including the windows pressure foils and cathodes. A cost-effective approach would be to first develop a long-lived Electra-sized system, then scale up to the 20 kJ driver size described in the previous section.

A laser fusion power plant would require numerous other symbiotic technologies in addition to the laser driver. It would require final optics and a reaction chamber system that can withstand for long periods the x-rays, neutrons, and charged particles from the thermonuclear pellet explosions. The frozen targets must be injected into a hot chamber, tracked, and precisely hit with the high-energy laser beams. The targets themselves must be precisely made and have costs consistent with economic power production. Research by the High Average Power Laser (HAPL) program examined these issues and found potential solutions that were in most cases partially tested in the laboratory experiments [68].

Uncertainties remain and it would not be wise to attempt to jump from success with an ICF research facility to a power plant even after development of the needed high-rate driver and targets. There would be need for an intermediate facility that would test the components, materials, and procedures in an operating environment close to that of a laser fusion power plant. In particular one needs a system that can produce a substantial flux of the high-energy (14 MeV) neutrons that carry the bulk of the energy from fusion of deuterium with tritium. These neutrons can produce transmutations of elements in addition to the atomic displacements observed with the lower energy neutrons available from fission reactors. An inertial Fusion Test Facility (FTF) with this purpose and configuration is described in [108,109]. This system employs a 500 kJ KrF driver operating at 5 pulses per second projected to produce 200 MW of fusion thermal power. This system would in principle have the capability of generating net electrical power, but its primary function would be developing the databases and experience needed for follow-on fusion power plants. Use of a KrF driver with inertial fusion would thereby complement the mainline magnetic approach [1]. IFE with KrF just might enable us to reach the goal of practical fusion energy faster and with less development cost.