Types of quantum computer
There are at least four main approaches towards building a quantum computer processor that have attracted major commercial investment. These are known as silicon spin qubits, superconducting qubits, ion trap qubits and topological qubits. Each brings specific challenges.
Silicon spin qubits: can operate at 1.5 kelvin
Spin qubits in silicon are excellent candidates for scalable quantum information processing, because they can be made in their billions using the same tools and infrastructure that the global computer industry already uses to make silicon chips for today’s computers and smart phones. The physical properties of silicon also mean that the qubits can be very stable and accurate. Now, thanks to Andrew Dzurak’s team, they can be operated at substantially warmer temperatures (on the quantum scale) than competing technologies.
Silicon spin qubits are the focus of several teams at UNSW. Companies using this approach include Intel, Silicon Quantum Computing, Quantum Motion and HRL Laboratories.
Superconducting qubits: require near absolute zero temperatures
These qubits are created from a loop of superconducting material, typically aluminium, combined with thin insulating barriers through which pairs of electrons can tunnel. This approach has produced the most advanced quantum computers to date, including a 53-qubit machine developed by Google which was used last October to demonstrate ‘quantum supremacy’ – meaning the quantum machine, for the first time, outperformed a large supercomputer. The technology needed to control each qubit means that scaled up machines are expected to be very large. Required operating temperatures are one-tenth of a degree above absolute zero (or -273 degrees Celsius).
Companies using this technology include Google, IBM, Intel, Quantum Circuits, DWave, and Rigetti.
Ion trap qubits: require lasers to cool atoms close to absolute zero
This approach uses one of the electrons in an ion (of calcium, for instance) to create a qubit in two states, defined either by the electron’s orbital state or its interaction with the atom’s nucleus. The ions are suspended and moved around in free space using electric and magnetic fields, which makes them easy to isolate from interference, giving them considerable stability. They were the first type of qubit to be studied more than two decades ago and there are now prototype systems with up to 20 qubits in use. On the downside, the qubits have to be held in ultra-high vacuums to prevent them interacting with other atoms, and they need to be kept cool and controlled by sophisticated lasers. Optimal operating temperatures are below a thousandth of a degree Celsius above absolute zero.
The main company using this approach is IonQ.
Topological qubits: require near absolute zero temperatures
These qubits are typically made from thin wires or sheets of a semiconductor coupled to a superconducting circuit to produce exotic states of matter known as Majorana fermions. Such a qubit has not yet been demonstrated, but it is predicted that they will be far more immune to errors than other types, which is their main advantage. Like superconducting qubits, however, topological qubits need to be cooled to temperatures around one-tenth of a degree above absolute zero.
Microsoft is the main company investing in this technology.
Timeline of the development of silicon CMOS quantum computing
1994: Peter Shor from Bell Labs (USA) shows that a quantum computer would be able to decrypt Public Key Encrypted codes (at the heart of modern secure communications) exponentially faster than today’s supercomputers. This triggers massive interest in quantum computing worldwide, mainly from government agencies.
1998: UNSW paper outlines concept for silicon-based quantum computing
Bruce Kane, a postdoctoral researcher at UNSW, publishes a paper in Nature outlining the concept for a silicon-based quantum computer, in which the qubits are defined by single phosphorus atoms in an otherwise ultra-pure silicon chip. This is the first such scheme in silicon – the material used for all modern-day microprocessors. Kane’s paper attracts great interest because: (a) silicon is ‘industrially relevant’; (b) silicon electron ‘spins’ have very long ‘coherence times’ and, hence, low error rates. This paper has now generated over 2,800 citations.
2000: UNSW research team established
Bob Clark establishes the Australian Research Council Special Research Centre for Quantum Computer Technology (CQCT), headquartered at UNSW, with the task of developing the technologies needed to build a quantum computer. The centre expanded in 2003 to become an ARC Centre of Excellence, and since 2010 has been led by Michelle Simmons (UNSW). It has more than 150 researchers in Australia, with major collaborations worldwide. In 2000, Andrew Dzurak (UNSW), a founding investigator in CQCT, begins his development of silicon device technologies for building a silicon quantum computer.
2007: A single electron device
Dzurak’s group develops a variation of a silicon Metal-Oxide-Semiconductor Field-Effect Transistor (MOSFET) device that can be reduced to the level of a single electron. These devices can be manufactured using complementary metal-oxide-semiconductor (CMOS) technology, from which most silicon chips for smart phones and computers are made. The device is central to quantum computing breakthroughs by Dzurak and colleagues to this day.
Andrea Morello and Dzurak publish in Nature a paper describing the first step of measuring a silicon qubit.
2012: A qubit writer
Another paper in Nature by the Morello group, “A single-atom electron spin qubit in silicon”, describes the crucial step of writing information on an electron to operate the first silicon qubit.
2014: A silicon CMOS quantum bit
Dzurak, Menno Veldhorst and Henry Yang patent the concept of a gate-addressable quantum bit based on the modified silicon MOSFET transistor.
2014: A high accuracy quantum dot qubit in silicon
A paper in Nature Nanotechnology by Dzurak’s group, with lead author Veldhorst, describes an addressable quantum dot qubit with fault-tolerant control-fidelity, which is a high accuracy qubit based on the modified CMOS transistor developed by Dzurak’s team in 2007.
2015: A two-qubit logic gate in silicon
Dzurak’s group (again with lead author Veldhorst) publishes paper in Nature titled “A two-qubit logic gate in silicon”. This is the first ever demonstration of a calculation between two qubits in silicon, meaning that all of the logic requirements for silicon quantum computing are now in place.
2017: Design for a full-scale silicon quantum computer chip
Dzurak’s group publishes a paper in Nature Communications titled “Silicon CMOS architecture for a spin-based quantum computer”. This sets out a detailed design of how computer chip manufacturers can use their existing fabrication plants build a quantum chip with millions of qubits.
2019: Most accurate electron qubit in silicon
Dzurak’s group (with lead author Yang) publishes a paper in Nature Electronics demonstrating the most accurate electron spin qubit ever, based on his silicon CMOS technology, with an error rate of only 0.04%. Such low rates are vital for full-scale quantum computers, since otherwise errors quickly accumulate and destroy any calculations.
2019: First accuracy measurement of a two-qubit logic gate in silicon
Dzurak’s group (with lead author Wister Huang) publishes paper in Nature titled “Fidelity benchmarks for two-qubit gates in silicon”, which provides the first measure of accuracy for silicon two-qubit logic. The results show that silicon is now near the threshold of accuracy needed for fault-tolerant quantum computing.
2020: Silicon CMOS qubit breaks the one Kelvin barrier
Dzurak’s group (with lead author Yang) publishes apaper in Nature titled “Silicon quantum processor unit cell operation above one Kelvin”. This is a temperature at which most other qubit types cannot operate. It also means that far cheaper cooling systems can be used, valued at thousands, rather than millions, of dollars.
Qubits are the basic units of information in a quantum computer, equivalent to the ‘bits’ in everyday computers. Like a bit, a qubit stores a binary code of a ‘0’ or a ‘1’, but unlike a conventional bit, a qubit can also be in a ‘superposition’ of both ‘0’ and ‘1’ at the same time. Every time you add one qubit to a quantum computer, the amount of information it can store and compute with doubles. This gives quantum computers an exponential increase in processing power (or speed) compared with conventional computers for a wide range of calculations (or algorithms).
Quantum superposition is a basic principle of quantum mechanics, which means that quantum systems can exist in multiple states at the same time, until you try and measure the system. A qubit is a system that has only two basic states, equivalent to ‘0’ or ‘1’, which can exist in a superposition of those, until you measure it, after which it is forced to randomly choose which one of the two states it is in, via a process known as ‘wavefunction collapse’. This principle of quantum superposition is what gives rise to the concept of Schrödinger’s cat, which can be thought of as simultaneously alive and dead, until you look inside the closed box it exists in.
Quantum coherence is related to the fact that all objects have wave-like properties. In the presence of noise, the wave-like properties get washed out, over a time known as the coherence time. To be useful for computing, you must be able to complete calculations on the qubits in a quantum computer within this coherence time, otherwise the information gets scrambled.
Quantum dots are devices or particles which are so small that they can behave like ‘artificial atoms’, and the electrons inside the quantum dot exist in specifically allowed quantum states. They can be used to create qubits.
Quantum tunnelling is the term used to describe a quantum function that has no equivalent in classical physics. Because of the wave-like properties of subatomic particles they are able to pass through seemingly impenetrable barriers even when they don’t have the energy to do so. What’s more, they can do so instantaneously. It sounds weird, but is well understood and is a key part of many human endeavours, including electronics and some types of microscopy.
Spin is a quantum property of subatomic particles, such as electrons, which is related to the particle’s angular momentum, analogous to its rotation. The spin of an electron also produces a tiny magnetic field, which will either line up in the same direction as (spin ‘up’) or opposite to (spin ‘down’) an externally applied magnetic field. An electron spin is an ideal qubit, whereby the spin up represents a ‘1’ and a spin down a ‘0’.
Silicon chips are used as computer memory or processing units in computers, smart phones, and other electrical items. They are typically about one centimetre square, and have tiny electrical integrated circuits inscribed on them, typically with billions of CMOS transistors, or memory cells. Silicon is a chemical element that is the second most abundant in the earth’s crust.
Silicon CMOS (or complementary metal-oxide-semiconductor) is the technology used to make and manufacture the transistors used for information storage and processing on silicon chips. Every mobile phone has one or more silicon CMOS chips inside it.
Silicon CMOS qubit is a qubit made using standard CMOS manufacturing technology, in which the spin of one or more electrons is used to encode the quantum information (‘0’ or ‘1’). The electrons that make up the qubit are electrically confined inside a silicon quantum dot, which is just a modified conventional transistor, as found on everyday silicon chips.