What is a spin qubit?

Quantum computing is an emerging technology that encodes and processes information using the physics that governs the universe at its smallest scales. Quantum has the potential to solve certain problems beyond what’s possible using even today’s best computing technology. However, building a machine capable of realizing this promise has long proved challenging. Today, several architectures have emerged offering different routes to running quantum algorithms at scale.

The core unit of quantum computing is the quantum bit, or qubit. These are physical objects like atoms, electrons, or superconducting circuits whose innate properties are quantized, meaning they assume discrete (rather than continuous) values onto which we can encode the zeros and ones of computation. Being quantum mechanical, these values can also enter superposition — a linear combination of zero and one — and entangle with one another, so the information is distributed across multiple qubits at once. A quantum computer must prepare and manipulate the qubit states, and at the end of the computation, measure the array to a final bit string.

IBM currently researches two hardware architectures for quantum computing. The first and more mature of the two is based on the superconducting transmon qubit, which is made from superconducting electrical circuits — circuits that transmit electrical current without resistance and can assume discrete energy states. The second and newer approach is based on spin qubits, which encode and manipulate information stored in the spin state of individual electrons. HRL Laboratories, an IBM company, has been pursuing spin qubits.

What is spin?

Spin is an intrinsic quantum property with no classical analogue, although some of its mathematics look similar to those of a spinning ball. Electron spins can either point up or down, taking values of positive one-half or negative one-half. When measured, quantum spins can never take values other than those options, breaking the direct analogy to a spinning ball, which might go faster or slower, or rotate about a different axis than the one you’re measuring. Spin can present in many ways in atoms, molecules, and materials: Multiple electrons working together can take on higher spin states, or even exhibit spin from the absence of an electron (a “hole”). A spin-based quantum computer encodes information in selected states of one or more spins.

Using a spin qubit first requires trapping one or more individual electrons in structures called quantum dots. HRL’s quantum dots utilize a layer of silicon sandwiched between layers of silicon-germanium, which confine the electron in that silicon layer in a two-dimensional plane. We then apply a voltage to a microscopic metal gate fabricated on top of this “heterostructure,” trapping an electron directly below the gate. There are other ways to trap electrons, using different quantum dot structures or even using implanted atoms.

How can spin represent qubits?

There are several types of spin qubit architectures. These are the two most popular:

1. In the Loss-DiVincenzo spin qubit, a single quantum dot traps a single electron whose state is controlled by microwave pulses. The direction that the spin points (up or down) stores the value of the qubit. Neighboring qubits are entangled by causing the electrons’ quantum wavefunctions to briefly overlap using voltage pulses on the gates, producing an “exchange” interaction.
2. Exchange-only qubits use three quantum dots (and therefore three electrons) to represent a single qubit. This type of qubit can be controlled entirely through electron-electron exchange interactions (no microwaves), with baseband voltage pulses similar to those of conventional electronics, though many of them.

HRL has developed a spin qubit architecture based on the exchange-only qubit, most recently demonstrating a complete system with 54 quantum dots, configurable to host as many as 18 qubits, on which we implemented a simple quantum-error-detecting code. Like superconducting quantum processors, spin qubits must be kept cold in a dilution refrigerator.

What are the comparative advantages of spin qubits?

Spin qubits have several properties that make them promising candidates for quantum computing. They have high gate fidelities, and can be fabricated using advanced semiconductor processes, making them amenable to scaling toward the large qubit counts required by many quantum use cases. Spin qubits are relatively fast — certainly faster than ion traps or neutral atoms. Considering measurement speed, though, they are not as fast as superconducting quantum processors.

Perhaps most importantly, since exchange-only qubits rely on only voltage pulses for applying gate operations, these qubits only require relatively simple control signals. In the same paper where HRL showed basic error correction, the quantum processing unit was controlled entirely with a custom CMOS chip operating at 4 K, near the coldest plate of the dilution refrigerator. This greatly reduced the number of necessary connections from room temperature, enabling future systems to grow larger more quickly. This will be a greater challenge for qubits that rely on microwaves for control, since that hardware generally has to stay at room temperature and be connected to qubits with long coaxial cables.

Spin qubit-based quantum computers don’t have as many qubits as superconducting ones at present, and there are still challenges to scaling. We are working to make our qubits more uniform and improve their fabrication yield using the facilities at Anderon. We also must reduce the amount of power required by our controller and further consolidate the wiring required as we scale up. We have research underway to tackle these challenges and more.

There are comparative advantages of pursuing spin qubits and superconducting qubits in tandem. Both are based on semiconductor fabrication technology, so advances on fabricating one platform could potentially benefit the other. Additionally, given the similarities of their underlying hardware platform, we are excited for the potential to explore hybrid systems employing both flavors of qubit architecture.

Ultimately, we expect that superconducting qubits will take us into the era of fault tolerance before spin qubits do, beginning with IBM Quantum Starling and continuing with IBM Quantum Blue Jay. However, IBM is also pushing forward with spin qubits to better understand how they might contribute to future, larger or less expensive systems. We are excited to have spin qubits as part of our portfolio as we work to bring useful quantum computing to the world.