Piece of the Sun : The Quest for Fusion Energy (9781468310412) Page 15
It didn’t remain a two-horse race for very long, however, with a Japanese contender known as JT-60 joining in April 1985. Japan had noted the fusion results revealed at the Geneva conference in 1958 but decided against an all-out machine building programme. They kept their plasma physics experiments small and in university labs. When the Russian-inspired dash for tokamaks began in the late 1960s Japan decided to take the plunge and built its first tokamak, the JFT-2, which was roughly the same size as Oak Ridge’s Ormak and was completed in 1972. From there, they jumped straight to the giant tokamak class, beginning design work on JT-60 in 1975.
Unlike in Culham and Princeton, the researchers at the new fusion research establishment in Naka did not supervise the construction of the machine themselves. They drew up detailed plans and then handed them over to some of the giants of Japanese engineering, including Hitachi and Toshiba, and left them to get on with it. Although it was a more expensive way of building a fusion reactor, the researchers were spared the stresses and strains of managing a complex engineering project. At Naka there was no unruly rush to demonstrate a half-finished machine just to meet a deadline. Instead the construction companies finished the job in an orderly fashion, tested that it was working, and then handed it over to the researchers. JT-60 was roughly the same size as JET and had a similar D-shaped plasma cross-section. Because of Japanese political sensitivities, it was not equipped to use radio-active tritium so the best it would be able to achieve was an ‘equivalent break-even’ but despite that, in many ways it exceeded the achievements of its rivals in the West.
Russia too, after the success of T-3 in 1968, continued to innovate, building a string of machines from T-4 right up to T-12. T-7 was the first machine to use superconducting magnets. Superconductors, when cooled to very low temperatures, will carry electrical current with no resistance so they allow much more powerful magnets and much longer pulses. This lets researchers explore how plasma behaves in a near steady state, rather than in a short pulse. The size of T-10 was on a par with Princeton’s PLT. Other centres got involved too, such as the Ioffe Institute in Leningrad which built a series of tokamaks.
In the mid 70s, as the US, Europe and Japan began building large tokamaks, Russia too started work on T-15. Although it was not quite as large as the other three giants of that period and was not equipped to use tritium, it was the only one of the four to use superconducting magnets. The intention was to follow T-15 with a dedicated ignition machine, T-20, bigger than TFTR, JET or JT-60. But the researchers’ ambitions were undermined by the crumbling state around them. The Soviet Union in the 1980s was already in a downward spiral. The T-15 team had trouble getting funding and materials and the situation got worse every year. By the time the machine was finally finished in 1988 it was already looking out-of-date and the institute couldn’t afford to buy the liquid helium needed to cool its superconducting magnets. Following the final collapse of the Soviet Union in 1991 the new Russia just didn’t have the resources to run an active fusion programme and T-15 was eventually mothballed.
Back in 1983, the Princeton researchers were getting used to their new machine. The huge scale of the thing compared to earlier tokamaks made it an exciting time. They ran the machine for two shifts each day. Researchers would gather for a planning meeting at 8 a.m. and the first shots would begin at 9 a.m. A roster of thirty-six shots per day was typical. There would be another meeting at 5 p.m. and shots would often run late into the evening but they had to stop at midnight to let the technicians and fire crew go home – they always had to be present for safety reasons. If the team were working late they would send one of their number out to get a carload of pizzas or hoagies – the Philadelphia term for a submarine sandwich.
The Tokamak Fusion Test Reactor at Princeton showing the neutral beam injection system on the left.
(Courtesy of Princeton Plasma Physics Laboratory)
TFTR was bristling with diagnostic instruments and the huge volumes of data these produced demanded a whole new way of working. Smaller machines had essentially been run by one small group of researchers who would plan experiments, carry them out, analyse the results and then do some more. Such an approach on TFTR would waste too much valuable machine time. So different groups were set up with a range of goals so that at any one time some would be preparing experiments, others doing shots and collecting data, and others analysing results of earlier shots. While it was all relatively informal at first, soon there were demands that groups didn’t horde their data but made it available for everyone to study. Competition for time on the machine grew so intense that they had to set up a system of written experimental proposals, five to ten pages long, that were peer-reviewed by other researchers at the lab. Princeton had entered the realm of ‘big science’ and it took some time for its researchers to adjust.
Initially, the researchers were getting very encouraging results. Even though the neutral beam heating systems hadn’t been installed yet, TFTR was producing temperatures in the tens of millions of °C using ohmic heating alone and with respectable confinement times. JET, when it started up in June 1983, made similarly good strides using ohmic heating. The scaling laws had been right that larger machines would lead to better confinement. But later, when heating was applied in both machines, the mood changed. TFTR initially had just neutral beam heating while JET had two heating systems, neutral beams and a radiowave-based technique known as ion cyclotron resonant heating or ICRH. In a tokamak plasma the ions and electrons move in spirals around the magnetic field lines and these spirals have a characteristic frequency. If you send into the plasma a beam of radiowaves at the same frequency, the waves resonate with spiralling particles and pump their energy into the particles, boosting their speed and hence the temperature of the plasma. So JET had radiowave antennas in the walls of the vessel to heat the plasma via ICRH.
But whatever the heating method used, the effect was the same: although heating did lead to higher temperatures, as predicted, it produced instabilities in the plasma which led to reduced confinement time. The loss in one counteracted the gain in the other so the overall effect was not much improvement in overall plasma properties. Projections showed that if the plasmas continued to behave in the same way as heating was increased, neither machine would get to break-even. The warnings about neutral beam heating provided by machines such as PLT had come too late: both Princeton and Culham seemed stuck with designs that would not achieve their goals.
Tokamaks need help to heat plasma to fusion temperatures, usually provided by ohmic heating (friction), radio waves and neutral particle beams.
(Courtesy of EFDA JET)
In February 1982, Fritz Wagner, a physicist at Germany’s fusion lab, the Max Planck Institute for Plasma Physics in Garching near Munich, was carrying out experiments with the lab’s ASDEX tokamak. ASDEX was a medium-sized tokamak and Wagner, who was relatively new to plasma physics, was studying the effect of neutral beam injection on the properties of the plasma – this was before TFTR and JET had started up. He started his shots just heating the plasma ohmically and then turned on the neutral beam and measured what happened. In most of his shots the arrival of the neutral beam produced a jump in temperature and the inevitable dip in density as instabilities caused by the beam made particles escape. But he noticed something strange: if he started out with a slightly higher density of particles and stayed above a certain beam power, when the beam kicked in the density suddenly jumped up instead of down and continued to rise, eventually reaching a state where temperature and density remained high right across the width of the plasma, ending in a steep decline at the plasma edge. This was entirely unlike the usual pattern which showed a maximum of temperature and density in the centre of the plasma and a gradual decline towards the edge. Wagner did more experiments with different starting densities and found that there was no in-between state: high beam power was needed and the density either jumped up or down depending on whether the starting density was above or below a critical value.
/> Wagner was perplexed and spent the weekend checking and rechecking his results to make sure he hadn’t misinterpreted something. His colleagues at the Garching lab were sceptical about it at first. No such effect had ever been predicted by theory or seen at another lab. But Wagner was able to demonstrate the effect reliably on demand so they were forced to take it seriously. If the effect worked on other tokamaks it could be amazingly important because the jump up in density, which was soon dubbed high-mode or H-mode, produced confinement twice as good as the low pressure state (low-mode or L-mode). The wider community of fusion scientists took more persuading. At a fusion conference in Baltimore a few months later he was grilled for hours by a disbelieving audience at an evening session. Until some other tokamak could also demonstrate H-mode, it would remain a curious quirk of ASDEX.
Wagner had to wait two years for another tokamak, Princeton’s Poloidal Divertor Experiment (PDX), to prove him right. Another machine, DIII-D at General Atomics in San Diego, repeated the feat in 1986. Now everyone was interested in H-mode. TFTR and JET, which were both struggling with poor confinement brought on by neutral beam injection, could be saved by H-mode but no one knew if it would work in such big machines. And there was another problem: ASDEX, PDX and DIII-D all had something that the giant tokamaks didn’t have – a divertor – and it seemed that H-mode only worked if you had one.
A divertor is a device in the plasma vessel that aims to reduce the amount of impurities that get into the plasma. Impurities are a problem in a fusion plasma because they leak energy out and make it harder to get to high temperatures. It works like this: if the impurity is a heavy atom, like a metal that has been knocked out of the vessel wall by a stray plasma ion, it will get ionised by collisions with other ions as soon as it strays into the plasma. But while a deuterium atom is fully ionised in a plasma – it has no more orbiting electrons – a metal atom will lose some of its electrons but hold onto others in lower orbitals. It is these remaining electrons that cause the problem. When the metal ions collides with others these electrons get knocked up into higher orbitals and then drop down again emitting a photon which, immune to magnetic fields, will shoot out of the plasma, taking its energy with it.
In early tokamaks, researchers tried to reduce this effect with a device called a limiter. There were different types of limiter but a common one took the form of a flat metal ring, like a large washer, which fits inside the plasma vessel and effectively reduces its diameter at that point. During operation the plasma current has to squeeze through the slightly narrower constriction formed by the limiter. This helps to reduce the plasma diameter and so keeps it away from the walls, and it also scrapes off the outermost layer of plasma where most impurities are likely to be lurking. As the only place where the plasma deliberately touches a solid surface, limiters had to be made of very heat-resistant metals such as tungsten or molybdenum. But when external heating began to be used in tokamaks in the mid 1970s the higher temperature proved too much for metal limiters and they started to become a source of impurities rather than a solution for them. So researchers switched to limiters made of carbon which is very heat-resistant. Even if the carbon did end up as an impurity it would do less damage because, being a light atom, it would probably be fully ionised by the plasma and wouldn’t radiate heat.
Some labs tried to counter the problem of impurities by coating the inside walls of their vessels – usually made of steel – with a thin layer of carbon. The coatings helped but they didn’t last for long, so at some labs they began to cover the inside walls of their vessels with tiles of solid carbon or graphite. Russia produced the first fully carbon-lined tokomak, TM-G, in the early 1980s and after it reported encouraging results others followed suit. By 1988 the interiors of JET, DIII-D and JT-60 were half covered in tiles and total coverage only took a few more years.
Limiters were, however, still proving to be a problem and researchers resurrected the idea of a divertor that Lyman Spitzer had first suggested in 1951 for his stellarators. A divertor takes the meeting point between the outer layer of plasma and a solid surface and puts it in a separate chamber, away from the bulk of the plasma, so that any atoms kicked out of the surface could be whisked away before they polluted the plasma. In some of Spitzer’s stellarators, at a certain point in one of the straight sections, instead of the narrow aperture of a limiter there would be a deep groove going all the way around the vessel poloidally (the short way around). Extra magnets would coax the outermost magnetic field lines – known as the ‘scrape-off layer’ – to divert from the plasma vessel and form a loop into the groove and out again. But inside the groove the field lines would pass through a solid barrier so any ions – deuterium or impurity – following those field lines would be diverted into the groove and then halted by the barrier. Unlike a limiter, this halting of the outermost ions occurs away from the main plasma, where it’s less likely to re-pollute it.
The divertor at the bottom of a tokamak’s plasma vessel removes heat and helium ‘exhaust’ from the plasma, and helps to achieve H-mode.
(Courtesy of EFDA JET)
Divertors didn’t work well in stellarators. The extra fields to divert the scrape-off layer caused such a bump in the magnetic field at that point that it worsened confinement. But in the mid 1970s people tried them again in tokamaks, first in Japan followed by Russia, the UK, the US (PDX) and Germany (ASDEX). In tokamaks it was possible to position the divertors differently: because the plasma moves by spiralling around the plasma vessel – combining toroidal and poloidal motion – a divertor could be fitted as a groove going around the torus the long way, toroidally. In this way the symmetry of the toroidal shape is not spoiled but the scrape-off layer will always pass the divertor once per poloidal circuit. And D-shaped plasma vessels had the perfect place to put a divertor: in the top or bottom corners of the D.
This small group of tokamaks that had divertors seemed to be the only ones in which H-mode worked, but nobody knew why. Both JET and TFTR were desperate to try to reach H-mode to improve their performance, but they were designed before divertors had proved themselves in tokamaks so neither had one. JET at least had the D-shape that could easily accommodate a divertor but installing one would involve an expensive refit. The JET team had a hunch, however: perhaps it was not the divertor itself that was responsible for H-mode but the unusual magnetic configuration with the outer scrape-off layer pulled out into a loop.
In the bulk of the plasma, the field lines loop right around forming closed, concentric magnetic surfaces, like the layers of an onion. The magnetic surfaces in the scrape-off layer are said to be open surfaces because they don’t close the loop around the plasma but veer off into the divertor. There is one surface that marks the boundary between the open and closed magnetic surfaces. Known as the ‘separatrix,’ this surface appears to form a cross – dubbed the x-point – close to the divertor where field lines cross over themselves.
In H-mode, the plasma edge was marked by a steep drop in density and temperature, almost as if something was blocking plasma from escaping. JET researchers wondered whether this ‘transport barrier’ was in some way related to the separatrix, the transition from closed to open magnetic surfaces. The question for JET was whether it could reproduce this magnetic shape with a separatrix and x-point without having a divertor? And if they could, would it produce H-mode? By adjusting the strength of certain key magnets around the tokamak, JET researchers were able to stretch out the plasma vertically and, eventually, produce the desired shape. The x-point was just inside the plasma vessel and the open magnetic field lines simply passed through the wall instead of into a divertor. It was enough to have an attempt at H-mode. They tested JET in this divertor-like mode for the first time in 1986. With a plasma current of 3 MA and heating of 5 MW, the plasma went into H-mode for 2 seconds, reaching a temperature of nearly 80 million °C and holding a high density. Researchers calculated that if they had been using a 50:50 mix of deuterium and tritium they would have produced 1 MW of fus
ion power. So H-mode was possible in a large tokamak. All JET needed now was a divertor.
TFTR, however, was stymied. With its circular vessel cross-section it was difficult to install a divertor and almost impossible to coax its magnetic field into an elongated shape with x-points and open field lines. So instead the Princeton researchers chipped away at the problem in any way they could think of, trying to coax longer confinement times despite the degradation caused by neutral beam heating – and eventually they did make a breakthrough, almost by accident.
A gruff experimentalist called Jim Strachan was doing some routine experiments on TFTR in 1986, trying to produce a heated plasma with very low density. The problem was the tokamak was not playing ball. Ever since they had started coating parts of the vessel interior with carbon – to stop metal from the walls from getting into the plasma – they had encountered a downside of carbon: it likes to absorb things. Carbon will absorb water, oxygen and hydrogen, along with its siblings deuterium and tritium. With all this stuff absorbed into the vessel walls, when you start heating a plasma the heat causes the absorbed atoms to emerge again and contaminate the plasma. Even if it is just deuterium in the walls – the same stuff as the plasma – it means that experimenters had no control over the plasma density because they never knew how much material would emerge from the carbon.
Earlier in 1986 a new carbon limiter had been installed in the TFTR vessel. This wasn’t a narrowing ring at one point in the torus but was instead a sort of ‘bumper’ of carbon tiles along the midline of the outer wall right around the torus. Once installed, the limiter was saturated with oxygen so researchers ran hot deuterium plasmas in the tokamak to oust the oxygen. That worked fine but it left the carbon tiles full of deuterium which would play havoc with Strachan’s attempt to produce low density plasmas. So Strachan started running shot after shot of helium plasma to get rid of the deuterium. Helium is a non-reactive noble gas, so does not get absorbed into the carbon as much. Strachan continued this for days, trying to get the tokamak as clean as possible, and then on 12th June he did a low-density heated deuterium shot – TFTR’s shot number 2204. The density stayed low, the temperature high and, astoundingly, the confinement time – 4.1 seconds – was twice what TFTR had achieved before. What’s more, it produced lots of neutrons – a sign of fusion reactions.