28 Jul 2021 by Margaret Harris
Collected at: https://physicsworld.com/a/cosmic-ray-threat-to-quantum-computing-greater-than-previously-thought/?utm_medium=email&utm_source=iop&utm_term=&utm_campaign=14290-50702&utm_content=Title%3A%20Cosmic-ray%20threat%20to%20quantum%20computing%20greater%20than%20previously%20thought%20-%20Editors_pick&Campaign+Owner=
Quantum computers may need a redesign to protect them from background radiation, say physicists. After earlier experiments showed that cosmic rays can severely disrupt the operation of superconducting quantum bits (qubits), an international team led by Robert McDermott of the University of Wisconsin-Madison, US, has now concluded that a leading error-correction method is unlikely to fix the problem on its own. Instead, McDermott and colleagues suggest that a combination of shielding and changes in qubit chip design may be required to keep errors at a manageable level.
Cosmic rays have created headaches in classical computing for decades. When these energetic particles fly in from space and strike a silicon computer chip, one or more bits in the chip may change state, or flip, in ways that programmers never intended. If these errors go uncorrected, damaging glitches may result – including, in one case, injuries to passengers on a Qantas flight after a bit-flip error fed incorrect data to the aeroplane’s instruments.
Surface code error correction
For quantum computers, the problem is more complicated since qubit states can flip in two directions (representing the X and Z axes) rather than one. Nevertheless, a form of error correction known as a two-dimensional surface code should, in principle, be able to handle qubit flips as long as the quantum processor meets certain requirements.
Surface code error correction works by encoding information in a flat sheet of superconducting qubits, each of which is connected to its nearest neighbours. If the error rates of individual qubit operations are low enough, it should be possible to use some of these qubits to identify and correct errors in neighbouring qubits via multi-qubit operations. The other requirement is that errors cannot be correlated – in other words, an error that affects one qubit cannot affect its neighbours at the same time.
Unfortunately, McDermott’s team discovered that errors caused by cosmic rays and gamma rays from background radiation do not meet this second condition. “We basically are finding multiple mechanisms for correlated errors,” Chris Wilen, a PhD student at Wisconsin and the lead author of a new study about the research, tells Physics World.
Quasiparticle poisoning
To study these correlated errors and quantify their effects, McDermott and colleagues constructed a chip containing two pairs of qubits: one pair separated by 340 μm, the other by 640 μm. While performing quantum operations on this four-qubit system, the physicists observed numerous simultaneous jumps in the charge induced on the paired qubits. When they modelled these bursts of charge using a standard particle-physics toolkit, they determined that the correlated jumps stem from collisions between the chip and a mixture of gamma rays and cosmic rays.
The probability of correlated jumps was highest for the qubit pair with the smallest physical separation, indicating that spacing qubits further apart reduces the direct effects of energetic particles striking the chip. However, the group also encountered a thornier problem: the energy released in these strikes ultimately gets transferred to the qubit substrate in the form of phonons, which are vibrations in a material and can lead to the creation of quasiparticles. As these phonons spread, they produce other kinds of correlated errors, and these errors affect the entire millimetre-scale chip. This phenomenon is known as quasiparticle poisoning, and Wilen says it “could be really damaging for error correction” unless it can be mitigated.READ MORECosmic challenge: protecting supercomputers from an extraterrestrial threat
Writing in Nature, the researchers suggest two possible solutions. One is to protect the quantum processor by shielding it with lead and shifting it to an underground site, as is already done for dark matter and neutrino detection experiments that are especially sensitive to radiation. Another is to reduce the sensitivity of the qubits by, for example, adding materials to the chip that can trap quasiparticles or funnel them away from the qubit substrate. “It’s a roadblock that we’re going to get over,” Wilen says, adding that the Wisconsin group plans to explore several of these mitigation strategies in the future.