JFEdraft07232019 submitted

Retrospective of the ARPA-E ALPHA fusion program C. L. Nehl, 1 R. J. Umstattd, 2,* W. R. Regan, 2,** S. C. Hsu, 2, † P. B. Mc Grath 2 1: Booz Allen Hamilton, Mc Lean, VA 22102 2: Advanced Research Projects Agency–Energy, U.S. Department of Energy, Washington, DC 20585 * R. J. Umstattd is presently affiliated with Fusion Industry Association, Gaithersburg, MD 20879 ** W. R. Regan is presently affiliated with X, the Moonshot Factory, Mountain View, CA 94043 † Corresponding author email: [email protected] Introduction: In 2014 the Advanced Research Projects Agency—Energy (ARPA-E) of the U.S. Department of Energy (DOE) launched a new research program on low-cost approaches to fusion-energy development[1]. The “Accelerating Low-Cost Plasma Heating and Assembly” (ALPHA) program set out to enable more rapid progress towards fusion energy by establishing a wider range of technological options that could be pursued with smaller, lower-cost experiments, short development and construction times, and high experimental throughput. Mainstream fusion research generally refers to magnetically or inertially confined fusion, both of which require expensive facilities for reasons briefly described below and explored in more detail in several books[2][3]. ALPHA focused on magneto-inertial fusion (MIF), a class of pulsed fusion approaches with fuel densities in between those of magnetic and inertial fusion[4], [5] [6]. This paper presents a brief background on the origins of the ALPHA program, the results achieved by ALPHA-funded teams, and a look ahead to potential next steps for low-cost fusion development. Origins ARPA-E’s mission is to develop transformational new energy technologies[7]. While DOE has pursued fusion energy as a potentially transformational opportunity for decades, ARPA-E had not supported any work in fusion prior to the ALPHA program. This was in part due to a perception that fusion was inherently the realm of “big science” and that ARPA-E, which runs relatively small, targeted, short-term programs across a wide spectrum of energy technologies—did not have a role to play in that development. In launching ALPHA, ARPA-E sought to change this dynamic and bring new players into the field – both in terms of the kinds of teams doing fusion development (e.g., smaller groups and private startups), and in terms of the sources of funding (e.g., private investors). The ALPHA program was also a way for ARPA-E to address a longer-term problem in energy development with a targeted program. The motivation and timing for the ALPHA program were driven by three major factors: 1) analysis suggesting the potential for lower-cost pathways with fuel densitines between those of the mainstream approaches of magnetic confinement fusion (MCF) and inertial confinement fusion (ICF) [8], [9], 2) significant experimental results from magnetized inertial confinement fusion [10] and from the Magnetized Liner Inertial Fusion (Mag LIF)[11] program that supported this analysis, and 3) growing private sector investment in fusion[12][13], opening an opportunity for new approaches if they can achieve performance gains at costs compatible with private investors. 1: Potential for lower-cost pathways: The overwhelming majority of fusion research funding is currently devoted to major programs in MCF (principally the ITER collaboration and supporting plasma science in conventional tokamaks), and in ICF (principally laser-driven systems such as the National Ignition Facility, NIF, in the U.S.). ITER and NIF are each multi-billion dollar facilities, and the costs are driven in large part by performance requirements that are intrinsic to their respective approaches. ITER, which will operate as a long-pulse device with an ion density of approximately 10 14 cm -3 , requires an exceptionally large vacuum vessel and magnet set to contain a plasma of sufficient size to meet and exceed Lawson conditions, and the costs of the vacuum vessel and magnets are correspondingly large[14]. NIF, a pulsed device that compresses targets to ion densities greater than 10 26 cm -3 , requires exceptionally high power and power density to overcome the thermal losses (hundreds of TW peak power), and the cost of a MJ- class laser and optics systems to deliver sufficient energy in a sufficiently short time to the target system drives high costs for the machine[15]. ITER and NIF are the leading facilities within MCF and ICF, respectively. For the purposes of burning plasma research (as in ITER) or ignition and propagating burn (as in NIF), these “big science” projects have arguably the lowest scientific risk for achieving their respective goals. However, there are a wide range of alternative approaches spanning the full range of parameter space for fusion plasmas, including many that lie near the middle of the ten-plus orders of magnitude in ion density between MCF and ICF[8]. In fact, there are a number of analyses suggesting that some of these intermediate-density approaches with very high magnetic fields (megagauss or higher) may be able to achieve Lawson conditions at significantly lower costs than the mainline MCF or ICF approaches. The reasoning behind these analyses varies – from an optimal balance between the minimum size/energy of a plasma against the minimum power to overcome thermal losses[8], to an optimum magnetic field in the megagauss range for sizing plasma and pulsed power components[9], to the power density of the fusion core matched to practical reactor scaling [16], but they each suggest that the space in between ITER and NIF may be less costly to explore than the MCF and ICF extremes. 2. Mag LIF experiments constituting proof-of-concept for MIF: The analyses referenced above have developed over decades, but there has been relatively little exploration of the concepts that might fall in this range, and thus little experimental data or validated models to offer more detailed support. However, in 2014, Sandia completed their first integrated shots of the Mag LIF experiments, which used the Z-machine to implode a pre-heated and magnetized cylindrical D plasma target [17]. The experiments reached peak ion densities exceeding 10 22 cm -3 and multi-ke V temperatures, producing significant DD neutron yields from thermonuclear fusion[11]. These results – which came very early in the first campaigns, and have subsequently been exceeded – represent the first significant experimental evidence to support the claim that intermediate-density, magnetized fusion approaches could be significantly lower in cost that MCF or ICF [18]. The Mag LIF experiments were performed on a multi- purpose pulsed power machine that is more than an order of magnitude lower in cost than the single- purpose ICF machine NIF[19]. While the Mag LIF results do not represent a new record in fusion yield, the very fact that these experiments produced high yield in early experiments on a non-purpose built, relatively low-cost machine suggests that this is an area of fusion research that warrants further exploration. 3. Increased private interest in fusion: At the same time that these scientific developments were taking place, there was also a growing movement of private investors taking increased interest, and devoting significant private resources, to fusion development. In the years leading up to the ALPHA program, hundreds of millions of dollars were invested into private fusion companies, led by Tri Alpha Energy (now TAE Technologies) in the U.S., Tokamak Energy in the U.K., and General Fusion in Canada[12], [13]. Acknowledging the extremely high technical risks and long timelines associated with fusion, the interest and appetite for private investors to participate in fusion development signals an opportunity to bring in new players and expand the field. A central thesis of the ALPHA program was that we could expand the field if we could offer more options for fusion development that could be developed at funding levels compatible with private investment (i.e., tens to hundreds of millions of dollars for R&D, not several billion for scientific proof of concept). When combined with (1) compelling arguments for low-cost pathways and (2) experimental evidence supporting those arguments, ARPA-E determined that this could offer transformational opportunity to change the trajectory for the field. Based on this combination of factors, ARPA-E launched the ALPHA program to explore “intermediate- density” fusion approaches with peak ion densities ranging from 10 18 -10 23 cm -3 [1]. This is a range that includes a diversity of approaches, but all share the common attributes of a magnetized plasma (a “target”) that must be compressed in a pulsed fashion (using a “driver”) to reach high density and temperature. Some examples include magneto-inertial fusion (MIF, sometimes called magnetized target fusion, MTF), which utilize an imploding conductive liner to compress a fusion plasma, and stabilized dense Z-pinches, which use direct pulsed power to assemble, compress, and heat a column of fusion fuel. The focus on intermediate-density approaches reflected the opportunity for low-development cost in a range that was relatively under-explored as compared to MCF and ICF approaches. Beyond the focus on approaches in the intermediate ion density range, the ALPHA program set specific goals for the cost (<$0.05/MJ delivered driver energy, measured over full driver life), engineering gain (>5 for product of driver efficiency and projected fusion gain), and shot rate (hundreds of shots in ALPHA program, path to >1 Hz operation) of the proposed plasma systems, all with the purpose of achieving rapid experimental progress in the near term, and enabling economical fusion power reactors in the long term[1]. These constraints ruled out many destructive experimental approaches that, such as the use of explosives for compression. Recent progress in pulsed power technology, such as the continued development of wide bandgap devices for high current/high voltage solid-state switches[20][21], [22] and linear transformer drivers (LTD)[23], [24] offer promise that pulsed fusion approaches can achieve high efficiency, low cost, and high repetition rate. ALPHA teams were permitted to use “legacy” pulsed power machines to demonstrate performance, but each had to justify that the current and voltage levels, and the required timescales and profiles for discharge could be compatible with eventual efficient operation at high repetition rate (e.g., 1 Hz). Out of this competitive solicitation, a portfolio consisting of nine teams was selected for award in the ALPHA program. There was a diversity of approaches within the program, including pulsed magnetic compression, MIF with piston-driven liquid liner compression, MIF with high velocity plasma jet compression, and stabilized Z-pinches. There were also projects in the portfolio performing applied Figure 1: List of lead organizations and approaches selected in the ALPHA program. scientific studies relevant to the densities, magnetic fields, and plasma/liner interface environments encountered in this range of fusion parameter space. The program also included some exploratory efforts on new components that could be broadly enabling for new fusion concepts. For the purposes of discussion, we group those projects as “Integrated Concept” teams, which developed integrated plasma and compression systems to produce thermonuclear fusion plasmas; “Driver” teams that developed technologies for liner compression systems that could be applied to MIF concepts (but did not integrate the drivers with plasma targets during the ALPHA program); “Applied Science” teams that performed experimental and simulation studies to better inform intermediate density fusion plasmas and MIF; and “Exploratory Concepts” that developed novel plasma configurations and driver components. (Note that these groupings were not categories of the FOA, but are rather post-hoc descriptions of the general thrusts within the program. There is some overlap within these groupings, insofar as all teams performed some level of exploratory work and applied science.) The following section describes the goals, progress, and status for each of the individual projects. ALPHA projects and results i. Integrated Concept Teams University of Washington/Lawrence Livermore National Lab: Sheared-flow Z-Pinch for Fusion The University of Washington (UW), along with its partner Lawrence Livermore National Laboratory (LLNL), developed a variant of the Z-pinch that exploits sheared flow in the axial direction to mitigate the m=0 and m=1 instabilities that plague Z-pinch plasmas. The concept builds upon prior work in the ZAP and ZAP-HD experiments which demonstrated that a Z-pinch initiated from a high velocity plasma gun experiences a shear flow from r=0 at the center of the Z-pinch axis to r=R at the plasma edge[25], [26][27]. In those experiments, it was shown that at sufficiently high velocities (typically observed as ~10% of the Alfvén velocity times k, the axial Figure 2: Results from the University of Washington Sheared flow Z-pinch. Top: Signal observed on the scintillator detector, which shows neutron producting during the stable (quiescent) period. Bottom: Normalized magnetic field fluctuation amplitude for the m=1 mode, as measured a multiple locations, which shows that the plasma is relative stable for about 5 u S. Note that the quiescent period aligns to the time in whch neutrons were detected. Figure adapted from PRL 122, 135001 (2019). wave number) from the plasma gun, the shear in the plasma was able to suppress the growth of sausage and kink modes in a Z-pinch over a stable period about 700X the expected instability growth time in a non-sheared Z-pinch, at densities of 10 16 -10 17 cm -3 and temperatures of 50-80 e V[26]. Under the ALPHA program, the UW/LLNL team sought to determine if this shear stabilization mechanism scales to fusion conditions, specifically by pushing the current to 100’s of k A, thereby increasing the density to ~10 17 cm - 3 and temperature to ~1-2 ke V. As shown in figure 2, and as summarized in a recent paper, the team was able to demonstrate experimentally that the sheared-flow Z-pinch at high currents (approximately 200 k A) exhibited stability for 5-20 μs, several orders of magnitude longer than the characteristic growth time for sausage and kink instabilities[28]. At these currents, plasmas of 20% Deuterium/80% Hydrogen reached densities of 10 17 cm -3 and temperatures estimated at 500 e V-1 ke V, and reproducibly generated neutron yields >10 5 for 5-μs periods, and observed a scaling of neutron emission with the square of the deuterium ion number density, which suggests thermonuclear origin[28]. The UW/LLNL team also performed extensive MHD and PIC simulations of the sheared-flow Z-pinch system. As the system pushes to higher currents, temperatures, and densities, the plasma will approach kinetic conditions, and at the outset of the project it remained an open question as to whether the shear stabilization demonstrated in the ZAP and ZAP-HD experiments would hold in the kinetic regime. PIC simulations from LLNL suggest that the shear stabilization mechanism will remain effective at the high currents projected for fusion conditions. [29] [30] This is a valuable addition to the existing literature on sheared flow Z-pinch stabilization. To review it was theoretically predicted that the kink (m=1) mode would be stabilized when when the flow shear (V Z /r) exceeds 0.1k V A (0.1 times the axial wave number times the Alfvén wave velocity) [25]. This prediction was later verified experimentally, which set the stage for further development of the sheared-flow Z-pinch[31]. Fully kinetic PIC simulations have also shown the suppression of instabilities in sheared flow stabilized Z-pinch plasmas at scales ranging from current experiments up to reactor-scale[29]. Based upon the promising results of the research under the ALPHA program, the team from UW launched a new company, Zap Energy, and has won a follow-on award from ARPA-E to push the sheared-flow Z-pinch to higher currents, possibly necessitating improved materials and designs for high- current-density electrodes, and refinement of timing and current profiles for plasma initiation and for stabilization at increased densities and temperatures[32]. Helion Energy – Magnetic Compression of Field Reversed Configuration (FRC) Targets for Fusion Helion Energy is developing an MIF concept for the compression of an FRC plasma using high power pulsed magnetic coils. The concept builds upon prior work at UW and at MSNW LLC, and utilizes the dynamic formation of two FRCs, accelerated towards each other and merged in a central chamber[33]– [35]. A high power magnet coil surrounding the central chamber compresses and heats the FRC. In prior experiments, such as the “Grande” experiment at MSNW, the merged and compressed FRCs reached high temperatures and densities. Based upon empirical scaling relationships from the LSX experiments at UW in the 1990’s[35], [36], the Helion team has projected that fusion conditions are achievable in a relatively low-cost machine with increased trapped flux in the FRCs, and increased peak B-field from compression [37]. However, these projections from the relatively low density, steady state conditions of LSX must extrapolate to several orders of magnitude higher in ion density, as well as significant increases in other experimental parameters, and there is limited theoretical or simulation-based understanding of the FRC to confidently make those projections. The ALPHA-supported research sought to increase the trapped flux in the FRCs by a factor of 2 and then compress it to a peak magnetic field of 20 T, providing experimental data in the higher density regime, as well as an experimental basis for performance projections for compressed FRC targets in fusion conditions. The experiments (in keeping with prior nomenclature, the updated machine was named “Venti”), proved to be challenging, particularly with the mechanical structure and pulsed power system for the central compression coil (designed for 10 T, but ultimately operated at 8 T), and for keeping the highly compessed FRCs on-axis in the relatively small radius of the central chamber. Even with the aggressive experimental goals, the team was able to conduct over 900 FRC compression shots, and the team observed DD fusion neutrons. As the team reported to the an independent review team (JASON) in 2018, Helion’s integrated system achieved a density of 8 x 10 16 ions/cm 3 , a final magnetic field of 8 T, a final radius of 6 cm, and an energy confinement time at maximum compression of 4 x 10 -5 s [38] While this data is encouraging, further experimental measurements of the plasma parameters are necessary to validate these claims. Significantly, the team believes they showed that the micro-scale confinement and macro-scale stability scale as expected. Separate from the ALPHA award, Helion is also pursuing increased performance for a larger scale compressed FRC system. Beyond their technical progress, Helion has also successfully secured private investment (much of it prior to their ALPHA award). Figure 3: An early depiction of the Helion approach. Image reproduced from Nucl. Fusion 51 (2011) 053008 Magneto-Inertial Fusion Technologies, Inc. (MIFTI)/University of California, San Diego (UCSD)/University of Nevado, Reno (UNR): Staged Z-Pinch Target For Fusion MIFTI, along with partner UCSD, is developing the “Staged Z-Pinch” MIF concept, which delivers high current pulsed power to an annular shell of high atomic number gas (e.g., Ar or Kr), which then compresses at high velocity on a cylindrical target of magnetized D-D fuel[39][40] [41] [42]. The geometry of the Staged Z-Pinch is very similar to Sandia’s Mag LIF, in that both are cylindrical magnetized plasma targets, and are compressed radially by a high-Z liner. However, there are important differences for MIFTI’s Staged Z-Pinch concept. First, there is no laser pre-heat of the MIFTI target, instead the pre- compression heating in the plasma target is provided by the initial shock of the imploding high-Z liner at the interface with the (stagnant) low-Z fuel target. Another key difference is that the liner in the Staged Z-Pinch is an annular gas puff, as opposed to a solid metal liner as in Mag LIF. The gas puff allows for more rapid experimentation, as each shot does not require replacement of the liner hardware, and in the course of the ALPHA project, MIFTI routinely complete 10’s of shots per day (albeit at current levels >10x lower than Mag LIF shots on Sandia’s Z-machine). The Staged Z-Pinch concept has been described in a number of simulation-based studies, and was explored in limited experimental work at UC Irvine in the 1990’s. The goals for the ALPHA project were to demonstrate the Staged Z-Pinch at fusion conditions utilizing the 2 TW (up to 1.2 MA) Zebra pulsed power machine at the Nevada Terawatt Facility at The University of Nevada Reno[43]. Hundreds of shots were completed, exploring a range of parameters for initial magnetization in the target, ion density (both in the target and in the liner), and using different liner species (principally Ar and Kr). Through the course of several campaigns on Zebra, the MIFTI team was able to produce consistent and repeatable shots with neutron yields exceeding 10 9 , with top-performing shots (Kr liner imploding on target with initial axial B-field of 10 k G) exceeding 10 10 neutrons[44][41] . The neutron yields appeared to have an isotropic and repeatable distribution suggestive of a thermonuclear origin, although direct measurements of temperature and neutron spectrum are necessary to fully establish that the yield is predominantly thermonuclear in origin[45]. It is worth noting, that the neutron time-of-flight diagnostics showed a signal with the timing and magnitude that would be expected for D-T neutrons. If this can be verified, it would indicate secondary fusion events from tritium produced in the D-D fuel, as had been shown on Mag LIF experiments at Sandia. In addition to the Zebra experiments, the team also completed a series of shots on the Cobra pulsed power machine at Cornell University for increased diagnostic access, especially for imagery to assess the stability of the inner surface of the liner during the implosion. These experiments were not able to utilize D fuel, and thus did not offer insights on fusion performance. Most of the simulations for the Staged Z Pinch were performed in MACH2, and as noted in the literature, there are disagreements over the extent of heating seen in MACH2 simulations of the Staged Z Pinch. In particular, simulations in Figure 4: Schematic of a liner-on-target Z-pinch, reproduced from Phys. Plasmas 26, 032708 (2019). MHRDR show significantly lower levels of heating during the implosion, and suggest that the Staged Z Pinch will not extend to fusion breakeven[45]. The MIFTI/UCSD team also performed 1D HYDRA results, which were consistent with the MACH2 results showing shock heating followed by adiabatic compression, but additional work in 2D and 3D simulations will be required to properly assess the disparity between different codes. The latest analysis suggests that the major code discrepancies arise due to the differences between Lagrangian versus Eulerian numerical methodolgies. Additional experiments with improved diagnostics for time-resolved data on temperature and density, and exploration of different implosion initial conditions, are needed to improve understanding of the Staged Z-Pinch[45] and its potential for scaling to net-gain fusion. ii. Drivers Los Alamos National Laboratory/Hyper V Technologies: Plasma Liners For Fusion Los Alamos National Laboratory, along with Hyper V Technologies and other partners, are developing a new MIF driver technology that is non-destructive, and should allow for more rapid experimentation and progress toward economical fusion power[46]. The team designed, built, tested, and deployed multiple plasma guns to produce hypersonic jets that merge to create a section of an imploding plasma liner, to support the development of the plasma-jet-driven magneto-inertial-fusion (PJMIF) concept [47][48] Because the guns are located several meters away from the fusion burn region (i.e., they constitute a “standoff driver”), the plasma gun components should be protected from damage during repeated experiments. By project completion the team should better understand the behavior of plasma liners as they implode in order to demonstrate the validity of this driver design, optimize the precision and performance of the plasma guns, and obtain experimental data in a 36-gun experiment on ram-pressure scaling and liner uniformity critical to progress toward an economical fusion reactor. Figure 5: Total plasma pressure in the staged Z-pinch as a function of time and radius near stagnation for Ar and Kr plasma shells compressing a deuterium target. Plots made with MACH2, a single-fluid radiation-MHD code. Figure reproduced from Phys. Plasmas 26, 052706 (2019). The project team designed, built, and tested seven state-of-the-art coaxial plasma guns, and used them to merge up to seven hypersonic plasma jets to form a section of a spherically imploding plasma liner [49] [47]. The team assessed two key scientific issues of plasma-liner formation via merging plasma jets: (i) shock heating leading to a degradation in the sonic Mach number of the merged jets, which would cause overly large spreading in the subsequently formed plasma liner leading to low calculated 1D energy gain [50] for the PJMIF concept, and (ii) degree of uniformity for the liner formed by discrete jet merging. For (i), ion shock heating was measured in two- and three-jet merging experiments [51] , which benchmarked simulations showing that the liner-average Mach number remained above approximately 10. For (ii), the first six- jet merging experiments were quite imbalanced due to large mass imbalance among the six jets[49]. An upgrade to the gas-valve design allowed for mass balance across seven jets of better than 2%. The more balanced jets led to the formation of a