The world’s most powerful X-ray source, the wire-array Z-pinch, offers a promising route to producing energy

through controlled thermonuclear fusion

The Z-pinch approach to fusion
Jeremy P Chittenden
FOR MANY years, scientists have been attempting to harness thermonuclear fusion – the process that powers the stars – in a controlled fashion in the laboratory. This is because fusion-based nuclear power is potentially a clean and almost limitless supply of energy. In a fusion reaction, two light nuclei are squeezed together until they react to form a single heavier nucleus. For example, the heavy isotopes of hydrogen – deuterium and tritium – can react to form a helium nucleus and a neutron. The combined mass of the helium and the neutron is less than that of the two hydrogen isotopes, and this mass difference is converted into energy according to Einstein’s famous equation E = mc 2. However, in order to get close enough together to fuse, the nuclei must have sufficient kinetic energy to overcome their mutual elec- Electrical discharges illuminate the surface of the Z generator at Sandia National Laboratories trostatic repulsion. In both stars and the laboratory this is achieved by heating the hydrogen “fuel” until it technique for achieving the desired uniformity is known as becomes an ionized gas or plasma with a temperature of 10– “indirect drive ICF”, where the radiation source is used to 100 million degrees. The key to controllable fusion power is to heat a wall or “hohlraum” that surrounds the capsule to a confine this high-temperature plasma for long enough to temperature of 2–3 million degrees. X-rays from the hohlextract an appreciable amount of energy. raum then provide a highly uniform and distributed radiation Historically, in most fusion experiments the plasma has source, which results in the capsule imploding more symmetbeen confined within complex machines using strong mag- rically than it would if it were illuminated directly with laser netic fields (see “Realizing the potential of fusion energy” by light or bombarded with high-energy particles. Jean Jacquinot Physics World December 1999 p23). The alterOver the last few years, a new approach has been develnative to this steady-state approach is “inertial confinement oped that uses X-rays from a so-called Z-pinch plasma-radifusion” (ICF) in which heating, compression and fusion all ation source to heat the hohlraum. This concept has only occur within a few nanoseconds. recently become viable thanks to spectacular advances in In the ICF concept, intense radiation in the form of laser X-ray power production at Sandia National Laboratories in light, X-rays or energetic particles bombards a spherical the US. This success is due to the high degree of implosion capsule of deuterium and tritium encased in a beryllium or symmetry obtained with plasmas formed from cylindrical plastic “ablater”. Rapid heating and expansion of the ablater cages of fine metallic wires. surface has an effect equivalent to covering the pellet in multiple miniature rockets. The fuel implodes and is compressed Radiation production in dynamic Z-pinches by a factor of 1000–10 000, which causes the temperature to An imploding or “dynamic” Z-pinch is an extremely efficient rise to about 100 million degrees. Fusion then begins, igniting source of X-rays. In a dynamic Z-pinch a large fast-rising and burning up the fuel before it has a chance to expand. electric current is passed through a hollow, cylindrical shell In order to achieve these high levels of compression, the of plasma. The current generates a large magnetic “pinch” irradiation of the target has to be incredibly uniform. One force that causes the plasma to implode with high velocity


1 The dynamic Z-pinch
current z-axis

2 Cylindrical wire arrays
current direction force
magnetic field


magnetic field

magnetic energy

kinetic energy

heat and X-rays

A large rapidly rising current from a generator is applied to a hollow cylindrical shell of plasma (blue). The large azimuthal magnetic field “pinches” the plasma and causes it to implode with high velocity towards the z-axis. Up to 20% of the stored electrical energy is converted into X-rays if the plasma is made from a heavy element.

(a) Wire arrays exploit the fact that there is an attractive force between two parallel wires carrying a current in the same direction. In a cylindrical array, the wires turn into plasma and are pulled towards the z-axis at speeds of up to 106 m s–1. (b) A wire array used in the Z generator at Sandia National Laboratories contains 240 tungsten wires, each 7.5 µm in diameter.

(figure 1). The kinetic energy of the imploding plasma is converted into thermal energy at the cylindrical or z-axis. However, if the plasma is made from a heavy element, such as tungsten, the majority of this thermal energy is emitted as soft X-rays. Indeed, up to 20% of the electrical energy stored in the generator can be converted into X-rays. For many years, however, the X-ray power levels from dynamic Z-pinches were too low to be of interest for ICF. The plasma shell had to be light so that it imploded with a high velocity and thus had a high kinetic energy. It also had to have a uniform structure so that all the material reached the z-axis simultaneously. Researchers were simply unable to create such a plasma using a conventional hollow shell of gas. This lack of uniformity resulted in a relatively long-lasting X-ray pulse with low power. The situation changed with the introduction of the “wirearray Z-pinch load” in which the plasma is formed from a cylindrical array of fine metallic wires (figure 2). The current rapidly heats the wires into the plasma state and causes them to explode. These expanding wire plasmas then coalesce and merge to form something that resembles a cylindrical plasma shell. Although the idea had been around for a number of years, only a small number of wires had been used in order to achieve a low-mass plasma. The early results were not spectacular. It was only when large numbers of fine wires were used in experiments at the 10 MA Saturn generator at Sandia in 1995 that a remarkable effect was observed. By keeping the total mass of the array constant and using larger numbers of finer wires, the Sandia researchers were able to create shorterduration and higher-power radiation pulses (see figure 3). The result was an increase of almost an order of magnitude in X-ray power compared with what had been previously achieved in a Z-pinch. In 1996 the scientists at Sandia converted the world’s most powerful electrical generator (formerly known as the PBFAII) from particle-beam fusion research into a Z-pinch machine. The enigmatically renamed “Z generator” boosted the X-ray power initially to 200 TW and then finally to 280 TW – currently a world record. The final gain in power was achieved using a “nested array” containing two concentric cylindrical arrays of wires. Everything about the Z generator is impressive (figure 4). The 36 capacitor banks store 11 MJ of electrical energy. 40

When this energy is released, it is first concentrated through water-filled transmission lines and then fed via magnetically insulated vacuum transmission lines to the load. The current produced is huge, rising to a maximum of 20 MA in 100 ns and with a peak electrical power of 50 TW. To put this into context, the electrical generating capacity of the entire US national grid is roughly 1 TW. The Z machine generates 50 times this power, but only for 10–7 seconds. The wire arrays typically contain 240 tungsten wires, each 7.5 µm in diameter (roughly 12 times finer than human hair). The entire array is 40 mm in diameter and weighs about 4 mg. It implodes with a velocity of up to 7 × 105 m s–1 (i.e. with approximately 1 MJ of kinetic energy). Imagine something weighing roughly the same as a flea, but with the same kinetic energy as a tank shell. After the implosion, the magnetic field compresses and heats the plasma further. Thus for a brief period the plasma continues to radiate and the total X-ray yield – up to 1.8 MJ in a burst lasting around 5 ns – can exceed the kinetic energy of the implosion. However, the peak X-ray power levels of 280 TW are still not enough to ignite the fuel at the centre of the hohlraum. The challenge is to find ways to improve the symmetry of the implosion and thus increase the X-ray power output even further. However, problems can arise in attempting to observe the plasma in detail. The whole Z machine is designed to maximize the power coupling between the generator and the load. To minimize the load inductance, the current flows though the wire array and returns to earth through a solid cylindrical can that surrounds the array. We would need to make a large hole in the can to diagnose the imploding plasma. However, this hole would spoil the magnetic field and ruin the symmetry of the implosion. Although we can study the radiation emitted after the plasma has imploded, it is virtually impossible to investigate the implosion phase of large machines. Smaller-scale experiments and computer simulations can, however, help us to understand the physics of imploding plasmas. Groups at Imperial College in the UK, the Naval Research Laboratory in the US and the Troitsk Institute of Innovation and Fusion Research in Russia have begun experiments to study imploding wire arrays in an attempt to increase the peak X-ray power. Very few high-current generators provide the access needed to investigate the implosion phase. However, for two and a half years a group of researchPHYSICS WORLD MAY 2000

3 Wire arrays pack a punch

4 Superlative machine

16 wires

X-ray power (TW)


24 wires 48 wires 90 wires
The Z generator at Sandia is the world’s most powerful pulsed-power facility. The machine is 33 m in diameter and is designed to concentrate the electrical energy stored in 36 capacitor banks (red) through water-filled transmission lines (blue). The energy is then fed to the load at the centre via four magnetically insulated transmission lines.


0 –20 0 20 time (ns) Experiments at the Saturn generator at Sandia have demonstrated that more powerful X-ray pulses can be achieved using arrays with a larger number of finer wires.

ers led by Sergey Lebedev of Imperial College has studied the implosion of arrays with up to 64 wires using the 1.4 MA MAGPIE generator at Imperial College. The physics of wire-array implosions In theory the maximum X-ray power is obtained if the wire array acts as a thin and uniform cylindrical shell so that all the plasma reaches the array axis simultaneously. In practice, the uniformity and stability of the implosion are not perfect and the X-ray power is limited by the finite width of the imploding plasma shell as it approaches the axis. Various experiments have been developed to investigate the two mechanisms that are thought to be responsible for broadening the plasma shell: mass injection and the Rayleigh–Taylor instability. Mass injection arises from the reluctance of the wires to expand and form a uniform plasma shell. Rather than uniformly heating the entire array into the plasma state, the current initially forms a light plasma only at the surface of the wires. The majority of the mass remains in cold, dense cores of metallic vapour. The light plasma is easily accelerated and is swept around the much denser cores, thus reaching the array axis far earlier than expected. We have studied the effects of mass injection in detail using a magnetohydrodynamic simulation of an array of eight wires (figure 5a). Our results show that the material from the wire cores gradually streams towards the centre of the array, forming a so-called precursor plasma. Eventually the wire cores run out of material, and the plasma implodes rapidly. The same phenomenon is seen in arrays containing larger numbers of wires. However, if we use a very large number of wires then the gaps between the cores become small. This means that the plasma cannot pass easily between the cores and implodes more uniformly. It is also possible to observe mass injection, radial plasma streams and precursor plasmas in experiments by “backlighting” the low-density plasma using a 400 picosecond optical laser pulse (figure 5b). However, gradients in the refractive index of the plasma cause the laser light to refract – a factor

that limits the technique to plasmas with densities of 1025 electrons per cubic metre or less. We have used a technique known as “X-pinch point projection radiography” to study the high-density region that contains most of the wire material. The method has been developed over the last decade by Dan Kalantar and David Hammer of Cornell University in the US together with Sergei Pikuz and Tania Shelkovenko at the Lebedev Institute in Moscow. In an X-pinch, a large current is passed through a pair of wires that are twisted together to form a cross. The current causes the point at which the wires cross to implode spontaneously, generating a high-temperature plasma that provides an intense source of X-rays. This short-wavelength radiation can penetrate the high-density material in the wire cores. We have found that most of the mass of the array remains in the wire cores (figure 5c). The horizontal striations we observed in radiographs of titanium wires imply that the core consists of an inhomogeneous mixture of liquid and vapour phases. This behaviour is typical of brittle wires, such as tungsten, and may spoil the uniformity of the plasma implosion. In contrast, the cores of ductile wires are more homogeneous. So what insight do these smaller-scale experiments give into the performance of the Z generator? If we use a large number of wires, the magnetic force accelerates the wire array as a whole and the plasma implodes rather like a thin shell. This is because the gaps between the wire cores are small, so the magnetic force cannot “poke” the plasma between the gaps. Our experiments with tungsten wire arrays, however, show that the wire cores are typically 100 µm in size regardless of the initial diameter of the wire used. At the Z generator, however, the gaps between the wire cores are about 500 µm. Our results suggest therefore that mass injection is still an issue at the Z generator and that there remains room for improvement. The Rayleigh–Taylor instability The other phenomenon that is responsible for shell broadening, the Rayleigh–Taylor instability, is a common feature of imploding plasmas. This effect can best be understood by considering why water falls out of a glass when it is turned upside down. The water cannot simply fall straight out of the glass as this would set up a vacuum above it, which would hold the liquid back. However, as the surface of the water is not smooth, air bubbles can rise through it and cause the liquid to fall as a series of “spikes”. This phenomenon is com41

5 Mass injection
a c

6 Rayleigh–Taylor instabilities
a b

1 mm (a) A 2-D magnetohydrodynamic simulation shows the “mass injection” effect, in which plasma gradually builds up on the axis of a wire array. The eight 15 µm diameter aluminium wires are directed into the page. Low-density plasma (shown in blue and green) sweeps around the wire cores (black) and forms a so-called precursor plasma at the centre. (b) Experimentally, we can probe the array using a laser pulse. This side view shows low-density plasma streaming around the wires and the build up of the precursor plasma on the axis. The radial plasma streams are modulated by magnetohydrodynamic instabilities. (c) Shorter-wavelength radiation from an X-pinch device reveals that the majority of the mass of the array remains in high-density wire cores within the low-density plasma. In this case the 100 µm diameter titanium wire cores exhibit a heterogeneous liquid–vapour structure. (a) A side-on view of a model that simulates the imploding wire array as a uniform cylindrical plasma shell. After we apply an initial perturbation, the Rayleigh–Taylor instability grows. The magnetic field behaves like a light fluid and blows bubbles (right) through the plasma. (b) We observed this effect experimentally by paring away some of the wires from the outer edge of the array. The characteristic “bubbles and spikes” structure of the Rayleigh–Taylor instability is seen using a laser probe.



mon to any system where a light fluid – the air in this case – pushes a heavy fluid – such as the water. It also occurs during the deceleration of supernovae remnants by the interstellar medium and during the implosion of an ICF capsule. In the imploding Z-pinch, the plasma is the heavy liquid that is accelerated by the magnetic field, which behaves like a fluid with zero mass. Our magnetohydrodynamic simulation shows that the magnetic field effectively blows bubbles in the plasma (figure 6a). The formation of bubbles increases the width of the plasma shell and the rise time of the X-ray pulse. The time it takes the pulse to reach its maximum value is the time between the first bubble and the last spike reaching the axis. Similar simulations by Darrell Peterson of Los Alamos National Laboratory have successfully reproduced the X-ray radiation pulse observed in the Sandia experiments. In both the simulations and the experiments, the Rayleigh– Taylor instability grows from very small perturbations. The size of these perturbations, however, determines the final amplitude of the instability and thus has a strong and detrimental influence on the radiation power that can be achieved. In experiments, the most obvious source of perturbation is the growth of magnetohydrodynamic instabilities in the plasma around each wire (see figure 5b). Initially these fluctuations behave independently and have little effect. However, when the plasmas from the individual wires merge, these initiate the perturbations for the more dangerous Rayleigh–Taylor instability. One interesting feature is that the amplitude of the Rayleigh–Taylor instability is noticeably smaller when the array contains a larger numbers of wires. This is not surprising if the global perturbation is an average of the magnetohydrodynamic instabilities from each of the wires. This 42

means that the perturbation amplitude should vary as 1/√N, where N is the number of wires. Indeed, our experimental data for arrays containing 8, 16, 32 and 64 wires fit this trend very well. This explains, in part, an empirical observation made at the Saturn generator with aluminium wire arrays. The Sandia team found that they achieved shorter, higher-power radiation pulses using larger numbers of wires. It is very difficult to observe the Rayleigh–Taylor instability directly in a dynamic Z-pinch because we would have to take a cross section through the imploding shell. However we have observed the characteristic bubble and spike structure at MAGPIE by paring away some of the wires around the circumference of the array (figure 6b). Nested wire arrays One concept that was developed to reduce the effect of the Rayleigh–Taylor instability is the nested wire array, in which a second array of wires is placed concentrically within the first. There are thought to be three ways in which these nested wire arrays prevent large instabilities from forming. The first idea assumes that both arrays form thin cylindrical plasma shells. When the first plasma shell collides with the second one, any Rayleigh–Taylor fluctuations that formed annihilate one another. The combined plasma then implodes too rapidly for excessive instabilities to grow. The second explanation is that the inner array of wires remains intact while the plasma from the outer array passes straight through gaps between the inner set of wires. The current then switches to the inner array of wires and the resulting plasma implodes rapidly onto the material from the outer array that has already reached the axis. Again the Rayleigh– Taylor instabilities are smaller than those produced by a single wire array. In the third scenario, the current in the inner array generates a magnetic field between the two arrays. When the outer array implodes, it compresses this magnetic field, which in turn drives the inner array. Thus a magnetic “buffer” prePHYSICS WORLD MAY 2000

vents the plasmas from the two arrays colliding and allows both to reach the axis simultaneously. Each of these three dynamical modes could increase the power production compared with a single array implosion. At Imperial College we have studied nested wire arrays in detail using the MAGPIE generator. By controlling the current in the inner array, we found results that are consistent with the second and third scenarios. While both of these dynamical modes provide substantial increases in the X-ray power compared with a single array, it is currently not clear which mode is best. Similarly, it is not obvious how to optimize experiments using nested wire arrays on the Z generator because we do not know which of the three dynamical modes is in operation. Empirically, however, the use of nested wire arrays has pushed the maximum X-ray power from 200 TW to 280 TW.

fusion uses much larger energies compared with laser fusion. The Z-pinch hohlraum also produces much lower radiation intensities, which means that the capsule implodes more slowly. wire However, the most striking difference array is that a Z-pinch facility can use much larger fuel capsules. This means that the expected energy from the fusion reactions can reach 1.2 GJ – about 80 times greater than the yield expected capsule from laser-driven fusion. The X-1 facility is estimated to cost around $1.2 billion, which is comreturn current parable with the original cost estimate spokes for NIF. While NIF is likely to be the first experiment to demonstrate fusion ignition, X-1 represents the hohlraum next stage of achieving high-energy wall yields and high gain. The large price tag on X-1 is, in part, the price of success. Roughly a third of the construction cost covers the facilities needed A schematic diagram of the load for the X-1 high-yield to handle the radioactive waste profusion experiment proposed by Sandia researchers. The duced by the neutron activation of Z-pinch wire arrays at the top and bottom of the device implode simultaneously and provide the X-ray flux materials close to the fusion target. needed to heat the central hohlraum wall and drive the Future outlook: towards ignition Given the current level of expendifuel capsule until it implodes. How can this powerful plasma-raditure by the US Department of Energy ation source be used to achieve indion the NIF project, it seems unlikely rect-drive ICF? One proposal is the so-called “double-ended” that the X-1 generator will be built in the near future. Instead, vacuum hohlraum (figure 7). X-rays from the two wire-array a more logical next step is to construct an intermediate facility Z-pinch implosions at the top and the bottom of the load heat (aptly named Z–X) to study the physics of Z-pinch capsule the central hohlraum. The hohlraum wall then emits X-rays implosions without the additional costs associated with high uniformly, which cause the capsule to implode. The drawback yield. Either way, the continuing success of the Z-pinch fusion with this scheme is that around 16 MJ of X-rays with a peak concept relies on both the continued technical success of expower of 2000 TW are needed to initiate fusion within a few periments at the Z generator and on obtaining a more detailed nanoseconds. Such high powers are needed because there is understanding of the physics of wire arrays. an intrinsic inefficiency in coupling the X-ray power delivered by the Z-pinches onto the central hohlraum wall. Further reading In an alternative scheme called the “dynamic hohlraum”, J P Chittenden et al. 1999 Plasma formation and implosion structure in wire the imploding plasma itself acts as the hohlraum wall and the array Z-pinches Phys. Rev. Lett. 83 100 fuel capsule is located on the Z-pinch axis. The advantage is C Deeney et al. 1998 Enhancement of X-ray from a Z-pinch using nested wire that the coupling between the X-rays and the target is much arrays Phys. Rev. Lett. 81 4883 better and so less power is needed. However, this approach J H Hammer et al. 1999 High yield inertial confinement fusion target design for has yet to show that it can provide the high degree of sym- a Z-pinch-driven hohlraum Phys. Plasmas 6 2129 S V Lebedev et al. 1999 The dynamics of wire array Z-pinch implosions metry needed to drive a uniform capsule implosion. How large will the next generation of ICF facilities need to Phys. Plasmas 6 2016 be for fusion ignition? The projections are normally based on S V Lebedev et al. 2000 Two different modes of nested wire array Z-pinch the conservative vacuum-hohlraum concept. They assume implosions Phys. Rev. Lett. 84 1708 that the symmetry of the implosion can be maintained and J Lindl 1995 Development of the indirect-drive approach to inertial that the X-ray yield scales with the square of the generator confinement fusion and the target physics basis for ignition and gain current – an empirical scaling law that has previously been Phys. Plasmas 2 3933 demonstrated. A machine called the X-1 generator has been D D Ryutov, M S Derzon and M K Matzen 2000 The physics of fast Z pinches proposed by Sandia scientists to meet these performance Rev. Mod. Phys. 72 167 requirements. (It is named after Cygnus X1, the most power- T W L Sanford et al. 1996 Improved symmetry greatly increases X-ray power ful X-ray source observed in the universe.) The X-1 machine from wire array Z-pinches Phys. Rev. Lett. 77 5063 is designed to operate with a peak current of 60 MA. Such a R B Spielman et al. 1998 Tungsten wire-array Z-pinch experiments at 200 TW high current will produce magnetic forces that will tear the and 2 MJ Phys. Plasmas 5 2105 G Yonas 1998 Fusion and the Z-pinch Scientific American August p40 electrode material apart close to the wire array. How does Z-pinch fusion compare with the laser-fusion approach proposed for the National Ignition Facility (NIF) at Jeremy P Chittenden is in the Plasma Physics Group, Blackett Laboratory, the Lawrence Livermore National Lab in California? Z-pinch Imperial College, Prince Consort Road, London SW7 2BZ, UK

7 Experiment of the future

30 mm


MAY 2000