Author | Zhou Peiying
On January 19, Nature published three papers on major breakthroughs in silicon double quantum computing, and jointly as the cover of the current issue, which is very rare.
For the first time, the researchers completed a breakthrough in silicon double quantum computing fidelity of more than 99 percent, that is, less than one error per 100 operations. Crucially, all three studies exceeded this critical threshold. It makes a viable proposition for a silicon qubit-based quantum computer, and the "last mile" of large silicon quantum processors that are actually manufactured and applied is being opened.
Cover of this issue of Nature
The three papers were from the collaborative team QuTech between Delft University of Technology (TU Delft) and the Dutch Organization for Applied Scientific Research (TNO), the Riken Institute of Science and Chemistry (RIKEN) and the University of New South Wales (UNSW). Among them, there is a Chinese doctor Xue Xiao who undertakes a work in the Dutch QuTech team, who was interviewed by China Science News for the first time.
Thesis URL: https://www.nature.com/articles/d41586-022-00047-0
"What people question, we go to conquer"
Due to their late start, semiconductor quanta (silicon quantum) lag far behind other quantum computing technologies such as ion traps and superconducting materials, but silicon-based quantum chips have an advantage in developing quantum computers.
First, the preparation of semiconductor quantum dots can be compatible with the existing semiconductor chip process, under the support of mature processes, the feasibility of industrialization of subsequent technologies is greatly improved; second, compared with the materials used in other routes, the stability of silicon qubits is better; third, the semiconductor quantum dot system has good scalability, the atomic properties of quantum dots can be artificially regulated, and it is easier to integrate than the general quantum system.
While the advantages of the material are obvious, the technical challenges are equally significant: semiconductor qubits are heavily affected by the spin of the surrounding kernels, and semiconductor qubits face two major problems: decoherence and lack of fidelity.
Starting slowly and technology limited, in this century race, although semiconductors are the most potential quantum track, they have also been questioned.
"Our experimental results are equivalent to breaking this doubt," Xue Xiao said, "proving that silicon-based quanta can also be as good as other platforms!" ”
99% fidelity is actually a threshold: quantum computing to achieve universal calculation is inseparable from quantum error correction, to create a truly practical quantum computer, quantum error correction is an essential technology. To achieve quantum error correction, it is theoretically necessary to ensure that the fidelity of each step in the calculation is higher than 99%. If it can be cracked, it is a great success.
Among them, the fidelity of double qubits has always been a big difficulty. In order to overcome this carp leaping dragon gate fidelity "qualification access", Xue Xiao's team cut from the purification of materials to reduce the effect of nuclear spin and achieve precise control of the interaction between the two qubits.
"What people question, we conquer." Xue Xiao said.
Starting from about 2010, in order to reduce the effects of nuclear spin, a large number of laboratories began to switch from gallium arsenide to silicon, and the silicon directly extracted from nature has three stable isotopes of silicon 28/29/30, of which silicon 28 and silicon 30 are without nuclear spin. Although silicon 29 has nuclear spins, the content accounts for only 5%, which achieves a major leap in materials.
But this does not seem to be enough, how to further reduce the effects of nuclear spin?
Xue Xiao's team thought of purification - removing silicon 29 from nature and upgrading it to a silicon-based material based on silicon 28 with a nuclear spin of 0.
Coherence time is a key metric for quantum computers, which directly limits the maximum number of times a quantum computer can perform quantum operations continuously, and is also a prerequisite for achieving high-fidelity metric sub-operations. Xue Xiao's team finally completed an order of magnitude increase in coherence time: from gallium arsenide to natural silicon, coherence time increased by two orders of magnitude; from natural silicon to purified silicon, coherence time increased by two orders of magnitude, compared with more than a decade ago, the entire quantum coherence time increased by 4 orders of magnitude.
At this point, the order of magnitude of coherence between silicon-based quanta and other routes of quantum computing has finally been equalized.
Another difficulty after overcoming the material level is to control the interaction between the two qubits. The system consisting of electron spins up and down can be used as a qubit that can be controlled by an electronic switch gate depending on the association between the spin and the charge.
Xue Xiao's Dutch team used materials formed from stacks of silicon and silicon germanium alloys to create a dual qubit system, constantly exploring the interaction of electrons, the strength of the coupling, and environmental parameters, achieving precise control of operation.
Their method is to configure the voltage on the red and blue electrodes LP and RP, each attracting an electron, and controlling the state of the electron by regulating the voltage, thus providing a method basis for further improving fidelity.
So, how do you verify the results of your experiments?
The Dutch and Australian teams used the more difficult gate set tomography gate set tomography [1] (GST) verification method, and the Japan-New South Wales team used the random calibration (RB) verification method.
The former can detect fidelity and completely calibrate the errors in the experiment; the latter will not calibrate the specific errors except for telling you the error rate. "This is also an important reason why we use gate set tomography," Xue Xiao said, "knowing where the errors are each time can we further correct them." ”
Next, "Double to Multi"
The independent research results of all three papers show that in this large-scale research and development competition, silicon quantum computing has achieved a key leap from theory to reality:
- The Delft University of Technology team in the Netherlands achieved 99.87% single qubit fidelity and 99.65% double qubit fidelity by using the electron spin of silicon/silicon germanium alloy quantum dots;
- The Riken team in Japan also used a two-electron system of silicon/silicon germanium alloy quantum dots, achieving 99.84% single qubit fidelity and 99.51% double qubit fidelity;
- The team at the University of New South Wales in Australia achieved 99.95% single qubit fidelity and 99.37% double qubit fidelity on a three-two-word bit system composed of electrons and two phosphorus atoms through ion implantation silicon.
Xue Xiao's next step in research is to do the number of qubits. "Only by increasing the number of bits can we get closer to the actual needs of a general-purpose quantum computer."
For him, making the silicon double quantum fidelity from 98% to 99% is a challenge to physical systems; from double quanta to multi-quantum, and even million quantum-level preparation, there is more of an engineering challenge beyond experimentation.
In the laboratory, the step from theory to reality is taken, and the practical prospect of the next step still needs more attention and technical investment from industrial enterprises. At this stage, silicon-based materials that have been in the field of traditional computing for 50 years will usher in its "second spring" in the quantum field, and the superiority of mature technologies will become more and more prominent compared to other quantum platforms.
The track in the field of quantum computing may be shuffled again. The actions of foreign technology giants Intel, Microsoft, IBM, Google, and domestic giants Alibaba, Tencent, Baidu, Huawei, etc. all show that the exponential acceleration brought by quantum to computer computing is about to be realized, and they have invested huge amounts of research and development funds, bet on different quantum computing technologies, and compete to hatch more practical quantum computers to win the race of the century.
"Never stop exploring the quantum realm"
As a paper, Xue Xiao still focuses more on the contribution of the paper itself to the industry.
"I'm very happy, but even more pleased that our paper can bring substantial progress to the field of silicon quantum, and last May I also published a nature, but the significance is not as big as this one."
As the only Chinese in the team, Xue Xiao played an important role in the experimental level of research, but the process was not smooth, and even interrupted for half a year.
In September 2019, Xue Xiao began to select this topic, and the first phase of the experiment once encountered a bottleneck in parameter adjustment, coupled with the severe complexity of the epidemic situation in Europe, which had to end in March 2020.
But he did not give up, and took the initiative to contact Maximilian Russ, another group specializing in theoretical physics, and under the coordination of his mentor Lieven Vandersypen, he went to the unified research group to conduct research and provided help in theoretical models.
The experiment restarted in September 2020, in the Netherlands repeated "lockdown", Xue Xiao offline experiments and online remote control frequently switched, often in the limited laboratory time to debug the equipment, before leaving set up an "all-night" program, to let the experiment operate uninterrupted.
Stressful? Yes. Do you want to do it? want.
In April 2021, the team finally came up with the results of the experiment. Xue Xiao believes that the guidance of his mentor and the experimental environment built by Delft University of Technology for many years have given him great support.
Xue Xiao's mentor Lieven is one of the leading figures in the field of quantum computing, and in 2000 he published a paper on Schauer's algorithm at Nature as the first author, and last year won the highest prize in the Dutch natural sciences. He provided Xue Xiao with as much free research space as possible, and rarely used "meritocracy" to create pressure on him.
Xue Xiao studied at the University of Science and Technology of China as an undergraduate, and during his freshman year, he transferred from engineering mechanics to physics out of interest in the field of quantum. At that time, he often went to listen to the physics class of the school 'Big Bull', and under the infiltration of the cutting-edge technological environment of the University of Science and Technology of China, he gradually became firmly engaged in the road of quantum computing.
After graduation, he studied at the master's level in the semiconductor laboratory of Tsinghua University, but after the technology was limited, the laboratory was shut down, and the prospects were once confused. In order to continue the experiment of silicon quantum computing, he applied to delft University of Technology in the Netherlands for a doctorate, and is currently staying in Delft for postdoctoral research to advance the next step of silicon quantum experiments.
Xue Xiao advised not to over-publicize his work, in fact, for the young doctor, there is no need for him to say, and the research results themselves are the biggest endorsement.