Today, silicon microchips underlie every aspect of digital computing. Now Moore’s Law is stuttering, and the world’s supercomputer builders are confronting an energy crisis. But those big gains using silicon seem to have ended, with the high-end Intel Core i7 chips, for instance, have been on computer store shelves for nearly a decade. And as supercomputers grow bigger, so too does their energy consumption. The world’s fastest known supercomputer today, China’s 34-petaflop Tianhe-2, consumes some 18 megawatts of power. That’s roughly the amount of electricity drawn instantaneously by 14,000 average U.S. households. Similarly, Summit the world’s most powerful computer – a 200-petaflop behemoth at the Oak Ridge National Laboratory in Tennessee, needs more than 17,000 litres of water every minute to keep it running safely. Even with the latest cooling technology, which uses water to remove waste heat, it’s tricky for engineers to keep the processor at the right operating temperature. The next race is for exascale supercomputers, capable of 1,000 petaflops—1 million trillion floating-point operations per second—or greater. Some estimates to reach exascale supercomputing are in the hundreds of megawatts.
By 2040, the world’s computers may need more electricity than our global energy production can deliver, according to a report released by the Semiconductor Industry Association and the Semiconductor Research Corporation in 2015. Therefore researchers are searching for the innovative ways to stop such devices guzzling so much energy. A computer based on superconducting logic and cryogenic memory can help solve these issues.
In 1911, a Dutch scientist discovered a class of materials which, at temperatures near absolute zero, could conduct electricity with no resistance and therefore zero loss of power. These materials called Superconductors have unique properties including, Zero resistance to direct current; Extremely high current carrying density; Extremely low resistance at high frequencies; Extremely low signal dispersion; High sensitivity to magnetic field; Exclusion of externally applied magnetic field; Rapid single flux quantum transfer; and Close to speed of light signal transmission. In theory, the system results in almost zero resistance and would require just a fraction of the energy of traditional computers, from one-fortieth to one-thousandth depending on the estimate.
The superconducting computer could be one of the most radical solutions to an ever-increasing energy demand. The concept rests on sending electric currents through supercooled circuits made of superconducting materials. After the second world war, the United States, the former Soviet Union, Japan and some European countries tried to build large-scale, cryogenically cooled circuits with low electric resistance.
Superconducting supercomputers was first inspired by the electrical engineer Dudley Buck. He built a novel superconducting switch he named the cryotron. The device works by switching a material between its superconducting state—where electrons couple up and flow as a “supercurrent,” with no resistance at all—and its normal state, where electrons flow with some resistance. A number of superconducting metallic elements and alloys reach that state when they are cooled below a critical temperature near absolute zero. Once the material becomes superconducting, a sufficiently strong magnetic field can drive the material back to its normal state.
Buck developed a digital switch by coiling a tiny “control” wire around a “gate” wire, and plunged the pair into liquid helium. When current ran through the control, the magnetic field it created pushed the superconducting gate into its normal resistive state. When the control current was turned off, the gate became superconducting again. Buck thought miniature cryotrons could be used to fashion powerful, fast, and energy-efficient digital computers. But liquid-helium temperatures made cryotrons challenging to work with, and the time required for materials to transition from a superconducting to a resistive state limited switching speeds.
IBM’s superconducting supercomputer program ran for more than 10 years, at a cost of about US $250 million in today’s dollars. It mainly pursued Josephson junctions made from lead alloy and lead oxide. Late in the project, engineers switched to a niobium oxide barrier, sandwiched between a lead alloy and a niobium film, an arrangement that produced more-reliable devices.
Josephson junction is switch based on the Josephson effect. Josephson predicted that if the insulating barrier between two superconductors is thin enough, a supercurrent of paired electrons could flow across with zero resistance, as if the barrier were not there at all.
Inspired by IBM’s project, Japan’s industrial ministry, MITI, launched a superconducting computer effort in 1981. The research partnership, which included Fujitsu, Hitachi, and NEC, lasted for eight years and produced an actual working Josephson-junction computer—the ETL-JC1. It was a tiny, 4-bit machine, with just 1,000 bits of RAM, but it could actually run a program. In the end, however, MITI came to share IBM’s opinion about the prospect of scaling up the technology, and the project was abandoned.
In 1983, Bell Telephone Laboratories researchers formed Josephson junctions out of niobium separated by thin aluminum oxide layers. The new superconducting switches were extraordinarily reliable and could be fabricated using a simplified patterning process much in the same way silicon microchips were.
Then in 1985, researchers at Moscow State University proposed a new kind of digital superconducting logic. Originally dubbed resistive, then renamed “rapid” single-flux-quantum logic, or RSFQ, it took advantage of the fact that a Josephson junction in a loop of superconducting material can emit minuscule voltage pulses. Integrated over time, they take on only a quantized, integer multiple of a tiny value called the flux quantum, measured in microvolt-picoseconds.
China is latest entry whose aim is to have a prototype of the machine up and running as early as 2022, according to a programme quietly launched by the Chinese Academy of Sciences (CAS) in November 2017 last year with a budget estimated to be as much as one billion yuan.
Supercomputers remain indispensable for the maintenance of a nuclear deterrent and the design of nuclear weapons through “virtual nuclear tests”. Supercomputers have helped Russia and China develop and deploy an entirely new generation of nuclear weapons, again without testing. Exascale computers are also required by Intelligence agencies like NSA and GCHQ for counter terrorism operations. They are also essential for Cybersecurity.
US Intelligence launched Superconducting Computer project C3
In 2014, The Intelligence Advanced Research Projects Activity (IARPA) signed research contracts with IBM, Raytheon-BBN, and Northrop Grumman Corp. to support the program called C3, or Cryogenic Computing Complexity.
The promise of superconducting processing is to take computing far beyond limits of the currently used complementary metal oxide semiconductor (CMOS) technology. If the technology proves to be effective and possible to manufacture at low enough cost, it will greatly reduce power and cooling requirements of supercomputers and data centers as well as the amount of space required to house cooling infrastructure.
“The power, space, and cooling requirements for current supercomputers based on complementary metal oxide semiconductor (CMOS) technology are becoming unmanageable,” Marc Manheimer, C3 program manager, said in a statement.
Superconducting circuits make quantum computing possible. Quantum computing takes advantage of the special property of subatomic particles called qubits to be in more than one state at once, which theoretically can deliver much faster computing than the current binary paradigm.
Besides the spy computer research program, there are multiple ongoing R&D projects around superconducting computing in academia and the public sector, including efforts by researchers at MIT, University of California Santa Barbara, Google, and National Aeronautics and Space Administration.
Some superconducting circuits have been clocked at 770 gigahertz, according to a report by MIT News. For comparison, Intel’s fastest processor to date, Core i7-4790K, clocks at a maximum of 4.40 GHz.
Tianhe-2, the Chinese system currently considered the world’s most powerful supercomputer, runs on 2.2 GHz processors, albeit it runs on more than 3.1 million processor cores. The system requires nearly 18 MW of power.
Superconducting circuits have no electrical resistance and thus produce no heat. This is achieved by cooling the material down to a point where the atoms stop moving, allowing electrons to pass without bumping into them and producing heat as a result.
A recently published paper by MIT researchers describes superconducting circuits made of niobium nitride that operate at minus 257 degrees Celsius. They are cooled to that temperature by liquid helium.
These circuits need about 1 percent of the energy a conventional chip needs.
IARPA’s program is looking well beyond exascale. C3 administrators expect to have the technology needed to demonstrate a small superconducting processor, and in five years a “small-scale working model of a superconducting computer.”
C3’s initial focus was on the fundamental components. This first phase was planned with the aim to demonstrate core components of a computer system: a set of key 64-bit logic circuits capable of running at a 10-GHz clock rate and cryogenic memory arrays with capacities up to about 250 megabytes. If this effort is successful, a second, two-year phase will integrate these components into a working cryogenic computer of as-yet-unspecified size. If that prototype is deemed promising, Manheimer estimates it should be possible to create a true superconducting supercomputer in another 5 to 10 years.
Manheimer projects that a superconducting supercomputer produced in a follow-up to C3 could run at 100 petaflops and consume 200 kilowatts, including the cryocooling. It would be 1/20 the size of Titan, currently the fastest supercomputer in the United States, but deliver more than five times the performance for 1/40 of the power.
China building a superconducting computer
Chinese scientists are building a low-energy and high-performance superconducting computer, aiming to complete a prototype as early as 2022 according to a media report. Scientists at the Chinese Academy of Sciences launched the project worth one-billion yuan ($145 million) last November, the South China Morning Post reported on Sunday.
The unprecedented machine will have central processing units running at a frequency of 770 gigahertz or higher. By contrast, the fastest existing commercial processor runs at just 5 Ghz, according to the report. The machine will also help find a way to reduce the massive energy consumption of supercomputers. As supercomputers grow bigger, so does their appetite.
China’s Sunway TaihuLight, the world’s second-fastest supercomputer behind the Summit in the US, requires 15 megawatts of power to run for one year, according to a researcher involved in the R&D process.
Li Xiaowei, executive deputy director of the State Key Laboratory of Computer Architecture, is well acquainted with the project and said the main motivation to build a superconducting computer is to cut energy demands of future high-performance computers.
CAS president Bai Chunli said superconducting digital circuits and superconducting computers will help China achieve parity in integrated circuit technology.
Chinese scientists have already made a number of breakthroughs in applying superconducting technology to computers. They have developed new integrated circuits with superconducting material in labs and tested an industrial process that would enable the production of relatively low cost, sophisticated superconducting chips at mass scale. They have also nearly finished designing the architecture for the computer’s systems.
In 2017, Chinese researchers realised mass production of computer chips with 10,000 superconducting junctions, according to the academy’s website. That compares to the more than 800,000 junctions a joint research team at Stony Brook University and MIT squeezed into a chip. But most fabrication works reported so far were in small quantities in laboratories, not scaled up for factory production.
Zheng Dongning, leader of the superconductor thin films and devices group in the National Laboratory for Superconductivity at the Institute of Physics in Beijing, said that if 10,000-junction chips could be mass produced with low defect rates, they could be used as building blocks for the construction of a superconducting computer.
If these efforts are successful, the Chinese military would be able to accelerate research and development for new thermonuclear weapons, stealth jets and next-generation submarines with central processing units running at the frequency of 770 gigahertz or higher. By contrast, the existing fastest commercial processor runs at just 5Ghz.
The advance would give Chinese companies an upper hand in the global competition to make energy-saving data centres essential to processing the big data needed for artificial intelligence applications, according to Chinese researchers in supercomputer technology.
CAS president Bai Chunli said the technology could help China challenge the US’ dominance of computers and chips.
“The integrated circuit industry is the core of the information technology industry … that supports economic and social development and safeguards national security,” Bai said in May during a visit to the Shanghai Institute of Microsystem and Information Technology, a major facility for developing superconducting computers.
“Superconducting digital circuits and superconducting computers … will help China cut corners and overtake [other countries] in integrated circuit technology,” he was quoted as saying on the institute’s website.
But the project is high-risk. Critics have questioned whether it is wise to put so much money and resources into a theoretical computer design that is yet to be realised, given that similar attempts by other countries have ended in failure.
Emerging field of “superconducting spintronics” could lead to a new generation of green supercomputers
Niladri Banerjee a senior lecturer in physics at Loughborough University, UK, explains, “One solution could lie in the new field of “superconducting spintronics”, which marries superconducting electronics with room-temperature spintronics. The first part of this union – superconducting electronics – relies on materials such as niobium in which current flows with no resistance and hence no heat loss. The overall energy loss, or “dissipation”, from a supercomputer built entirely from such materials would be much reduced. Unfortunately, we don’t yet know how to make the components of superconducting circuits as small as those found in conventional electronic circuits, roughly tens of nanometres in size.”
The second half of the partnership – room-temperature spintronics – exploits the spin as well as the charge of electrons to store and process information. The spin – or intrinsic angular momentum – of an electron can point up (↑) or down (↓), which offers a way to efficiently store, process and manipulate information as 0s and 1s. One way we can do this is using magnets a few atomic layers thick, made from cobalt, nickel or iron. The resulting current of spins lets us do interesting things like switch the magnetization using spin-polarized currents, in which all the spins point in the same direction.
For superconducting spintronics to work, however, you cannot use conventional Cooper pairs, which are responsible for the remarkable ability of electrons to flow without losing any energy. Consisting of two electrons with spins pointing in opposite directions (↑↓ or ↓↑), their overall spin is zero, making them useless for spintronics. In 2001, however, three theoretical physicists – Sebastian Bergeret, Anatoly Volkov and Konstantin Efetov from the Ruhr University in Bochum, Germany – predicted the existence of “exotic” spin-polarized Cooper pairs, in which both spins point in the same direction (↑↑ or ↓↓). As long as these pairs can survive inside a ferromagnet, it ought to be possible to exploit both their spins and their superconductivity. Spin-polarized Cooper pairs offered the prospect of information-processing devices that are small, lose hardly any energy and have interesting functionalities.
It took physicists almost a decade, however, to firmly establish the existence of these Cooper pairs. The breakthrough came in a series of experiments carried out in 2010 by Mark Blamire and colleagues at the University of Cambridge and Norman Birge’s group at Michigan State University (Science 329 59 and Phys. Rev. Lett. 104 137002). The trouble is, it proved tricky to generate these spin-polarized Cooper pairs. They only form in artificial thin-film heterostructures of superconductors and two ferromagnets whose magnetic moments are perpendicular to each other.
Recently, however, researchers including Niladri Banerjee at Loughborough University, Cambridge and the Norwegian University of Science and Technology have been able to hugely simplify the structure needed to generate spin-polarized Cooper pairs, opening the door to practical applications (Phys. Rev. B 97 184521). Instead of having to delicately align several magnets to form the bridge between superconductivity and spintronics, we have found that similar effects are possible using a single magnet and the subtle relativistic effect of “spin-orbit coupling”, which links the electron’s spin with its motion around the nucleus of an atom.
“What we did was to take a standard superconductor such as niobium and deposit on its surface an atomically thin layer of platinum, followed by a layer of ferromagnetic cobalt and finally another layer of platinum. With the thicknesses of the platinum and cobalt layers carefully selected, we looked at the impact of tilting the magnetization of cobalt with respect to the film plane. When we applied an external magnetic field so that the magnetization of the cobalt layer was fully in the plane, we were surprised to find that the temperature at which the niobium starts superconducting fell dramatically compared with when there was no magnetic field. Although we knew that applying an external magnetic field reduces the superconducting transition temperature, the drop we saw was far bigger than expected.
The bottom line is that, thanks to spin-orbit coupling (in the platinum), we can control superconductivity (in the niobium) by adjusting the magnetization direction of a single magnet (the cobalt). The spin-orbit coupling lets us generate exotic Cooper pairs in a controllable way simply by adjusting the direction of a single ferromagnet. In essence, our stacked structure of platinum and cobalt has given us a spin-orbit-coupled ferromagnet. This effective communication between superconductivity and magnetism is like forming a bridge between the two phenomena, dramatically simplifying the structures needed to make useful circuit components in superconducting spintronics.
This form of superconducting spintronics, driven by spin-orbit coupling, makes it much easier to build the components needed for a fully functioning and practical superconducting spintronic circuit, taking us one step closer to more efficient supercomputers. However, our work goes far beyond practical applications. It shows that three very exciting phenomena can coexist in this niobium–cobalt–platinum system: superconductivity, magnetism and spin-orbit coupling. Indeed, under specific conditions, entirely new phases of matter can emerge, including – in our case – a novel form of “magnetic superconductivity” (usually magnetism kills any superconducting behaviour). Such phases are not possible to generate in nature and exist only in these kinds of artificially engineered structures.
What we now need is a device that can exploit our exotic Cooper pairs. One possibility would be to build a superconducting version of a “spin transfer torque” (STT) device – a conventional spintronics component in which the magnetization of a ferromagnet can be flipped using a spin-polarized current.
Such devices are already being used for STT magnetic random-access memory (STT-MRAM) chips. They are not only cheaper, use less energy and can store more information than conventional memory chips, but also let us precisely switch the magnet of one device without disrupting the magnetic alignment of nearby devices, which is always a danger when you switch a magnet using an external magnetic field.
In the spintronics version of STT, we could switch the magnetization of a nanomagnet by transferring spin angular momentum from the spin-polarized current. Building an STT device that uses superconducting spin-polarized currents might not be easy, but it would massively improve the energy efficiency of such devices. While it is hard to predict how information technology will evolve, I envisage today’s supercomputers, which exploit only the charge of the electron, one day becoming obsolete – replaced by a new generation of superconducting spintronics supercomputers. Green supercomputers will then have finally arrived.