Quantum queen
Can Michelle Simmons lead Australia to the finish line in the race to build a reliable quantum computer? ELIZABETH FINKEL reports.
AND UNLIKE THE ROBUST BITS of your laptop, quantum bits or qubits are weird and fragile. Trying to corral them is like trying to harness a flock of butterflies.
And then there is the competition: IBM, Google and Microsoft are all in the race.
None of this fazes physicist Michelle Simmons. She is confident the team she leads – the Australian Centre for Quantum Computation and Communication Technology – can deliver the most reliable type of quantum computer ever emerge victorious. In their machine, the quixotic qubit is made of stable silicon.
“I’m not out there to recreate Intel, but I honestly believe our devices will win in the long term,” she says. “They are the most reproducible ones that are out there.” Simmons’ audacity is paying off. This year the Australian Centre garnered A$45 million from the federal government and businesses. Backing the “space race of the century”, telecommunications company Telstra and the Commonwealth Bank each put in A$10 million; the federal government, A$25 million.
The boost should allow the Australian team to shrink their timeline for building a 10-qubit processor from 10 to five years.
They need to work fast: using different types of qubits, MIT and IBM already have a five-qubit processor, Google’s has nine, and Canadian company D-wave controversially boasts 1,000. The Australian team may be behind, but Simmons believes they will win the distance race. And she is not the only one. “I think in the long term, for any number of reasons silicon will be the winner,” says electrical engineer John Randall, president of Texas-based Zyvex labs, an atomic-scale manufacturing company. “Australia can be a big player.”
THE WAY SIMMONS SEES IT, she is tracking a path not unlike the one that conventional computers followed. It took about a decade to advance from the first transistor bit in 1947 to the first silicon chip. The Australian group achieved the first qubit in 2010; if they get to 10 bits in five years, they will be well on track.
Simmons’ team is used to her blithe confidence. “Michelle is just mapping it out step by step,” says lab head Tony Raeside as he takes me on a tour through two floors of glass-walled, state-of-theart rooms of the fabrication centre at University of New South Wales (UNSW). Some of the rooms are brand new – the first fruits of the new funds. The big contraptions, like steel monsters in glass enclosures, are scanning tunneling electron microscopes (STMS). Like a blind person’s fingers scanning braille, their fine tips detect the contours of individual atoms.
These finely tuned electron microscopes allow skilled operators like Simmons to fulfill a vision imagined 30 years ago by Nobel prize-winning physicist Richard Feynman: to sculpt matter atom by atom. They are also the key to making the silicon qubit.
Many remain skeptical about the promise of quantum computing. But things are changing. Two years ago, Simmons was invited to give a tutorial at a satellite conference of the International Electronic Devices Meeting, the premier gathering for the electronics field. The organisers vetted every word of her talk to
“I’M NOT OUT THERE TO RECREATE INTEL, BUT I HONESTLY BELIEVE OUR DEVICES WILL WIN IN THE LONG TERM.”
make sure it didn’t contain anything too mindbending. They needn’t have worried. Her talk was a hit, and last December, they invited her back to give the keynote lecture.
SIMMONS HAS ALWAYS HAD an audacious streak. She tells a story about how, as an eight-year-old, she sat silently week after week watching her father, a high ranking policeman in London, play chess with her elder brother. One day she asked her father if she could play. Her father reluctantly acquiesced and played without paying much attention – until she took his first pawn. By then it was too late. She checkmated him.
Simmons’ mother was a bank manager, and her grandparents included diplomats and members of the military. She describes her family as “take-responsibility kind of people”. The family DNA also includes a sense of adventure. Her father would always tell her, “don’t take the easy route. Do the most challenging thing”.
I am sitting opposite Simmons, now 49, in the Quad café on a sunny wintry day at UNSW. She has rushed in for a snatched lunch and is clad in her signature look of casual black, draped across her tall, solid frame. Her hair is short and practical; she wears no jewellery or make-up. There is a soft, gentle femininity about her. As we talk, I wait for a glimpse of the iron fist that must surely reside inside the velvet glove. Leading a team of brilliant physicists bent on world domination must take some doing.
It was by popular demand that in 2010, 11 years after joining the group, Simmons became their leader, overseeing the entire consortium of 180 researchers from UNSW, University of Queensland, University of Melbourne, Griffith University and Australian National University. The UNSW headquarters hosts four scientific teams, each with its own research leader: Andrew Dzurak, Sven Rogge, Andrea Morello and Simmons. Like mountaineers navigating a maze of crevasses and cliffs, they are trying different paths but are united in their push to scale the peak of silicon-based computing. (See figure.)
“The strength of our centre is that we have three parallel pathways marching forwards. Of course we all love our own children, but we a have respect and regard for the others,” says Dzurak. Leadership here requires the ability to rally and unify the team while belaying your own rope.
Simmons revels in the different strengths and perspectives of her colleagues. “Physicists have very unique ways of seeing the world,” she says. She also enjoys pushing them out of their comfort zones and seeing them scale new heights. The key to her leadership is her clarity of purpose. “I could always see the obvious way to go forward,” she says.
There is the eight-year-old anticipating all the moves ahead. Only today, she is navigating her way through the chessboard of quantum computing.
QUBITS MAY BE WEIRD, but the day-to-day work of building one is very down to earth. The process starts with a commercial computer chip, a pure silicon crystal wafer about 3 millimetres by 10 millimetres, which is placed inside the ultra-high vacuum chamber of the STM. A trickle of hydrogen gas is bled into the chamber, coating the surface of the wafer with a mask of hydrogen one atom thick.
Guided by a scan of the atomic landscape, the tip of the microscope probe becomes a nanoscale etching pen. By varying its voltage, it can be used to poke a single hole through the hydrogen mask or scratch out a line of millions of atoms.
When phosphene gas seeps in, one phosphate atom will parachute into the hole to become the qubit, while others will fill the long scratch to become the wire that measures the qubit’s signal. Next, the crystal wafer is heated to 350 °C for one minute to bond the phosphate atoms. Then the crystal is coaxed to grow over its new components by sprinkling it with a light soot of silicon atoms – a technique known as epitaxy.
Simmons pioneered this two-day, 25-step manufacturing technique. “Her ability to position atoms with this accuracy is unique,” says Klaus Ensslin, who heads the nanophysics lab at the Swiss research institute ETH in Zurich. Learning to master it is tough; typically it takes a student a half a year.
RICHARD FEYNMAN PROPOSED a basic model for a quantum computer in 1982. The Caltech physicist had been part of the tail end of the quantum mechanics revolution that revealed how strange the universe was at the atomic scale. An electron or the nucleus of an atom has a magnetic orientation called “spin” and can exist in one of two spin states: up or down.
But in the quantum world, the spins can also exist in these states at the same time. This phenomenon is termed “superposition”. Even more mind-bending, two electrons could influence each others’ spin even if they were at opposite ends of the universe. They were said to be “entangled”. Eisntein referred to it as “spooky action at a distance”.
It was these two properties, superposition and entanglement, that led Feynman to speculate that a quantum computer would be able to perform a massive number of calculations in parallel.
The bits of a classical computer have a value of either 1 or 0 (because they either pass current or not). But a qubit would also have the value of 1 or 0 simultaneously. The long and short of this quantum logic is that hundreds of qubits are predicted to have the same crunching power as billions of classical bits.
When it comes to problems that stump modern computers, such as finding the prime factors of huge numbers (the basis of encryption) or finding the optimum path between destinations from billions of possible ones, quantum computers would ace it. That’s why banks and companies that deal with vast databases are so keen on the technology.
But for two decades, quantum computing remained stuck on the drawing board. Computing requires that calculations are done many times to correct errors. But that’s a problem for quantum computers because each time you read the result you influence it.
In 1995 several people, including Peter Shor
at Bell Labs in the US, figured out how to solve the error correction problem. Shor had also written an algorithm for a quantum computer to factorise prime numbers. Galvanised by the possibilities, labs around the world dove in to try to build a quantum computer. For qubits, MIT used ions trapped in a vacuum; IBM used tiny loops of superconducting metal; others tried quantum dots of gallium arsenide.
The Australians tried something entirely different: silicon.
THERE IS AN OBVIOUS QUESTION to be asked. If silicon really is the clear winner for reliable quantum computing, then why did Australia end up with it, and not MIT or IBM?
Three things seem to have conspired to make Australia the germination ground: good timing, a core group of visionary physicists, and the exceptionally receptive environment of UNSW.
Australian physicist Bob Clark founded the group in the late 1990s after returning from Oxford where he had helped pioneer low-dimensional physics. Advances in fabricating silicon and gallium arsenide crystals for the semiconductor industry, using extremely low temperatures and strong magnetic fields, were revealing remarkable new states of matter. So-called “electron gases” with novel behaviours lay between the layers of the crystals. Nobel prizes were awarded for those discoveries.
To continue the work at UNSW, Clark established a silicon nanofabrication facility and the National Magnet Lab. His reputation attracted bright young physicists from around the world, including Andrew Dzurak, an Australian who had completed a PHD at Cambridge.
Around 1996, Bruce Kane, a junior scientist from Bell Labs in the US arrived to work on the low-dimensional physics of gallium arsenide crystals. Clark also suggested that he try working with silicon.
Months later, Kane appeared in Clark’s office bearing a hard-back notebook filled with calculations. In his spare time at UNSW, Kane had worked out a design for the basic elements of a quantum computer.
Kane’s idea was entirely different from the other approaches in play. He conceived a way to make a qubit using the computer industry’s standard materials. Kane’s qubit would be a single phosphorous atom embedded in a silicon crystal. Because the phosphorous atom is very close in size to the silicon atom, it should cause minimal disturbance to the silicon crystal. Pure silicon, whose atomic nuclei have zero spin, would provide a noiseless background against which to read the spin of the phosphorous nucleus.
By the time Kane was scribbling in his notebook, it was clear that noise was a limitation of other types of qubits; it was interfering with the ability of the qubit to hold its signal long enough to do some processing – its so-called “coherence” time. While other systems had coherence times of a 1,000th of a second or less, in theory silicon would provide the qubits with entire seconds to carry out its processing.
Clark stayed up all night reading Kane’s paper. By morning it was clear to him this was a work of genius. It was also clear to him that he would move heaven and earth to bring the idea to fruition. The team filed a patent and published a paper in Nature in 1998. Physicists read it and were enthralled. But there weren’t many who were eager to give it a try.
Manipulating single atoms to build the qubit was only the start. No-one knew how to read the spin signal of a single electron or nucleus; the available technologies read signals from a million of them.
“It was a theory on paper but I never thought it would work,” recalls Ensslin.
But there was one person who did. She was just a junior scientist at Cambridge, but she had a reputation as a world leader in fabricating quantum electronic devices. Dzurak, her former Cambridge colleague, had already regaled her with tales of sunny Sydney skies and the blues of Bondi Beach. In 1999, Simmons joined the budding group of UNSW visionaries.
WHEN SIMMONS ARRIVED at UNSW, she was no stranger to daunting problems.
She encountered one of the first as a 16-year-old at a rough inner London comprehensive school. In her final two years, the school decided to roadtest “independent learning”; it was so independent her class had no chemistry teacher. Most of her classmates failed the year. But Simmons unpacked boxes with the textbooks and chemistry experiments and figured out a DIY chemistry course. It was an experience that forged her signature trait: self-reliance. “I still believe the best way you learn is by yourself,” she says.
Years later, that self-reliance got her a dream collaboration with NASA while still working on her PHD. “I love everything space,” she says. Simmons was to test the potential of perfectly symmetrical
“IF WE COULD DO IT, YOU COULD GET SCALE UP BY THE WORLD’S EXISTING COMPUTER INDUSTRY.”