Microsoft's Topological Quantum Computer Explained

Microsoft’s Topological Quantum Computer Explained

Source: Microsoft’s Topological Quantum Computer Explained, Domain of Science, 23:18, uploaded 2024-05-13, playlist index 336.

Domain of Science visits Microsoft’s quantum labs in Copenhagen to explain why a topological qubit might withstand noise better than the qubits used in other designs. The visit has a personal edge. The presenter worked on superconducting nanowires during his Ph.D. and failed to make a useful device. Microsoft is now using the same broad material idea in a much larger attempt to build a useful quantum computer.

Quantum computing and the noise problem

Quantum computers promise useful calculations in areas such as materials science, chemistry, and the simulation of reactions involving large molecules. The source presents this as a theoretical advantage over classical computers, which would need impossible amounts of time or memory for some of those problems. A working machine still has to preserve a delicate quantum state long enough to calculate anything.

For a physicist, noise means random energy entering the system. Radio waves from phones and Wi-Fi, background radiation from Earth or space, and the thermal motion of atoms all count. A quantum computer therefore sits inside a set of defences: a vacuum can with further vacuum cans inside it, shielding against electromagnetic radiation, and temperatures in the millikelvin range. The vacuum removes most moving atoms and helps with thermal insulation. The cold reduces thermal noise. At the centre sits a chip with the quantum devices that perform the calculation.

Most designs build their qubits from a physical property of electrons, atoms, photons, or another system. Microsoft wants to use a topological property as an additional layer of protection. The distinction comes from topology, the area of mathematics that studies which properties survive a smooth change of shape.

Holes, twists, and an energy gap

A mug and a doughnut have different geometry, yet each has one hole. A smooth deformation can turn one into the other whilst the hole remains intact. A loop of paper has two sides. A Möbius strip has one because a half twist joins the two sides. Ripping the paper changes that property, so the two loops belong to different topological classes.

The presenter jiggles the Möbius strip to show what the distinction can do for a quantum device. Jiggling changes its position and exact shape. The strip keeps one side. A sufficiently violent disturbance tears it and changes the topology. The same principle gives a topological qubit an energy range in which ordinary disturbances can change physical details while leaving the encoded topological state intact. Noise can still cross that range when it carries enough energy.

One Microsoft researcher says the topological qubit promises a base physical noise rate 100 to 1000 times better than other designs. The video presents this as a promise from the project rather than as a result it independently establishes. The advantage depends on creating a large energy gap between the protected state and the disturbances that could destroy it.

From clean-room materials to a qubit

The first Microsoft lab grows the material and fabricates the device. A clean room keeps dust away because a single mote can ruin a structure at this scale. The process begins with semiconductor growth, moves through nanofabrication methods borrowed from the silicon industry, and ends with electrical measurements that show how the device behaves. A second lab cools the fabricated devices in a fridge and tests them. Failed measurements point back to changes in the materials or the fabrication process.

The difficult part is the material combination. The device uses layers of different materials grown on top of one another and shaped into a small electrical circuit. Their purity and the exact arrangement of their atoms determine whether the desired quantum state appears. Lauri shows an interface between two materials in which the atoms form an unusually clean lattice. Disorder means holes, cracks, and misplaced atoms. He describes removing that disorder as the major challenge in demonstrating the topological phase.

Every quantum computer needs a controllable quantum bit. A classical bit has the states 0 and 1. A qubit can occupy those states and the quantum states between them. Different approaches choose different physical systems, such as atomic energy levels, a photon’s path, or the spin of an electron. Microsoft builds its qubit from the collective behaviour of many electrons in a nanowire.

The Lego model of a topological qubit

The presenter uses Lego to reduce the device to its working parts. A nanowire contains carefully fabricated layers of material. Gates called quantum dots control whether electrons can move through the wire or remain trapped inside it. The number of trapped electrons supplies two distinguishable states: an even number can represent 0 and an odd number can represent 1. A readout device measures that parity.

This simple design remains exposed to noise. A phone signal, a cosmic ray, or thermal energy could knock an electron away and spoil the state. The topological version adds Majorana zero modes, which the video also calls Majorana particles. They occur in pairs. The device separates the members of a pair towards opposite ends of the nanowire, creating a large energy gap between them.

Noise may disturb one end of the wire. The encoded state remains protected whilst the disturbance fails to reach both Majorana modes at once. Separation matters because modes that sit closer together produce a smaller gap and give noise an easier route to disrupting both. The work in the fabrication lab therefore aims at a particular material structure and device geometry that keep the modes far apart.

The result Microsoft has reached

Microsoft has measured the combined evenness or oddness, called the parity, of one pair of zero modes. Chetan, the researcher interviewed in this section, says the device produces a large signal with low measurement noise, so the readout is unambiguous. The two ends of the wire sit three microns apart. The parity is spread across them, which gives the result its non-local character and supplies the proposed protection from a disturbance that reaches one place at a time.

The measurement remains one part of a larger device. The team needs a second readout on another section of the structure before it can fully use the topological qubit. It then has to connect multiple qubits and make them work together. The video notes that other companies already have chips with hundreds of qubits, while quality and error rates determine how useful a count of qubits becomes.

A particle made from collective behaviour

The final explanation returns to particle physics. The Standard Model contains particles observed in experiments such as those at CERN’s Large Hadron Collider. A Majorana particle belongs to a different category in the source’s account. Physicists have theorised about it, including the property that it would be its own antiparticle, whilst experiments have yet to observe it as an elementary particle.

Microsoft’s device creates a Majorana quasiparticle. The right combination of atoms makes the electrons behave as if another particle were present. The quasiparticle has the relevant particle-like properties, yet those properties emerge from the collective motion of the electrons. An electron hole gives a simpler example. A gap between electrons can move when electrons fill it, so the gap behaves like an anti-electron even though the gap contains no separate particle.

The Majorana quasiparticle extends that idea. Its topological properties belong to the organised behaviour of the electrons, and those properties give the qubit its proposed resistance to noise. The source’s route from topology to hardware therefore passes through a material interface, a nanowire, separated modes, and a parity measurement. The result is an engineered state that behaves like a Majorana particle rather than a newly discovered elementary particle.

Limits

The video describes an early stage of Microsoft’s approach. Its strongest noise comparison, the 100–1000× figure, comes from an interviewed Microsoft researcher and has no supporting measurement in the captions or description. The demonstrated result is a parity readout across one pair of modes. The source says that another readout and the scaling to several connected qubits remain ahead.

Microsoft sponsored the video. The presenter says that he kept control of the final edit, and the sponsored segment promotes Azure Quantum as a free educational and programming tool. That segment does not establish the performance of the hardware. The description lists three references, yet the video does not work through their methods or results, so they remain part of the source’s bibliography here.