The cat in the bag? Taking a look under the hood of qubit architectures

Related Vendors

Leibniz-Rechenzentrum (LRZ) der Bayerischen Akademie der Wissenschaften

Wuxi InfiMotion Technology Co., Ltd.

Qubits open up new possibilities in data processing. However, their high sensitivity to external influences—decoherence—limits computing power. Fortunately, not all qubits are the same. New architectures promise more robustness.

Anna Kobylinska and Filipe Pereia Martins work for McKinley Denali, Inc., USA.

Qubits, the fundamental computing units of a quantum processor, have the bizarre ability to take on a third state known as superposition, in addition to the states 0 and 1, allowing them to capture multiple values simultaneously. This behavior forms the basis for the computational advantages of a quantum computer. The biggest challenge remains the error-proneness of qubits as information storage. However, different qubit implementations behave differently. Each architecture has its own strengths and technical challenges.

In Germany, the Helmholtz Association is among those researching new qubit platforms. These include Majorana qubits (FZJ), phase-slip qubits, and molecular qubits (KIT, DLR).

In the lament of superconducting qubits

Superconducting qubits combine high-frequency technology with quantum mechanics. The core component is the Josephson junction—a nonlinear tunneling connection that generates discrete energy levels, similar to an artificial atom. Unlike classical LC oscillators, it allows the creation of discrete energy levels that behave similarly to the states of an artificial atom.

Control is achieved via precise microwave pulses (4–10 GHz). Superconducting qubits can execute quantum gates in the nanosecond range, but they are short-lived.

From an engineering perspective, the greatest challenge is to minimize environmental influences. Even tiny changes in the Earth's magnetic field, thermal noise, ground vibrations, and electromagnetic fields accelerate decoherence. Therefore, superconducting qubits are operated in highly efficient cryosystems at a few millikelvin.

Superconducting qubits can be integrated on silicon wafers, but due to the requirements for advanced cryotechnology, only a few companies can currently produce such systems in significant quantities. The main players include IBM, Google, Rigetti Computing, IQM, D-Wave, and Quantinuum (a merger of Honeywell Quantum Solutions and Cambridge Quantum).

"Cat qubits" from Alice&Bob in Paris

Conventional superconducting qubits (e.g., transmon qubits) are sensitive to bit-flip errors (the X errors) and phase-flip errors (the Z errors). A European startup named Alice&Bob aims to realize a fault-tolerant architecture with the so-called "cat qubits," which requires fewer physical qubits per logical qubit than common approaches from companies like IBM or Google.

Named in reference to Schrödinger's cat thought experiment, these qubits differ from conventional qubits by their inherent error resistance to one of the central problems in quantum computing: bit-flip errors.

Cat qubits represent a specific physical architectural variant within the family of superconducting qubits. They are based on the coherent superposition of two macroscopic quantum states in a nonlinear superconducting resonator.

The coherent states of a cat qubit are stabilized by a nonlinear potential landscape that creates a high energy barrier between them, thus suppressing bit-flip errors.

A cat qubit consists of a superconducting circuit with a nonlinear element. The latter is usually a Josephson junction (JJ). This creates an anharmonic energy level structure that allows for precise control of the qubit states.

These superconducting circuits generate a coherent superposition of two opposite states in phase space. These two states can still be precisely controlled and read out with standard quantum operations. However, they are separated by an energy barrier, preventing easy transitions between them—which curtails bit-flip errors.

Cat qubits use multiple components to maintain their coherent states.

Initially, a superconducting inductance (L) combined with a nonlinear superconducting element, the Josephson junction—the nonlinear core of the circuit—creates an anharmonic energy level structure. The superconducting inductance is a loss-free coil made of superconducting material such as niobium or aluminum, used for energy storage.

A Josephson junction (JJ) is a superconducting tunnel connection with nonlinear inductance. It consists of two thin layers of superconducting material separated by a non-superconducting insulating layer (such as an oxide layer or a normal conducting metal). The barrier is thin enough that electrons can quantum mechanically tunnel through without encountering electrical resistance.

In a normal conductor, electric current flows when a voltage is applied. In a Josephson junction, a superconducting current (the so-called Josephson current) can flow even without voltage—solely due to the quantum coherence between the two superconducting electrodes.

The nonlinear sinusoidal dependence of the Josephson current on the phase difference of the superconducting wave functions is one of the central properties of the Josephson junction. This nonlinearity enables the targeted creation of discrete and coherent quantum states in superconducting circuits, which manifest as coherent states in phase space. This behavior can be modeled using the nonlinear Schrödinger equation.

Through capacitive coupling (C), the system interacts with the control and measurement technology and can be specifically excited or read out via a microwave resonator. This element is a superconducting cavity or transmission line resonator used for the manipulation and readout of qubit states.

Unlike conventional superconducting qubits (e.g., transmon qubits), where bit-flip errors frequently occur, the coherent states of a cat qubit are trapped in a stable potential landscape in phase space. This results in intrinsic suppression of bit-flip errors, reducing the need for complex error correction methods. Phase-flip errors (Z errors) remain a challenge.

The momentum of silicon spin qubits

Silicon spin qubits use the spin of a single electron trapped in a tiny semiconductor structure called a quantum dot. These qubits can be manufactured like miniaturized transistors, and their spin states can be manipulated using electric or magnetic fields.

Since they usually have lower excitation energies compared to superconducting qubits, they are less sensitive to thermal noise. However, their realization requires a highly pure starting material (for example, 99.9999% high-purity silicon, abbreviated as "^28Si"). With less than 1 part per million (ppm) of the problematic isotope silicon-29 (^29Si), ^28Si is primarily used in quantum computing as a substrate for qubits.

Silicon spin qubits are considered very space-efficient as they can connect to the classical CMOS manufacturing process. With successful optimization, thousands or millions of spin qubits can be implemented on a single chip.

The great strength of silicon spin qubits is their space-saving potential: with successful optimization, thousands or even millions of these qubits could be integrated onto a single chip. This prospect of high scalability attracts renowned players like Intel, HRL Laboratories, and the Dutch QuTech, who are familiar with CMOS processes and have extensive manufacturing capabilities. Universities, including the University of New South Wales (UNSW), are also intensively researching the spin-based architecture. Thanks to long coherence times in isotopically pure silicon, this technology could represent an important step toward widely deployable quantum processors.

Trapped in an ion trap

Ion trap qubits utilize individual electrically charged atoms (ions, e.g., of ytterbium or barium) trapped in an electromagnetic field and can be controlled by laser pulses.

A typical ion trap system consists of laser fields that cool the ion and control its quantum states, RF or DC electrodes that fix the ion in the vacuum, and optical systems that read out the qubit.

In contrast to superconducting or spin-based qubits, ions are natural quantum objects with inherently well-defined quantum states. This gives them exceptionally long coherence times of several minutes or even hours and enables extremely precise gate operations. However, they can only be driven slowly (in the microsecond range) and present scalability challenges.

In addition to Quantinuum, pioneers of this technology include IonQ, a spin-off of the University of Maryland; Oxford Ionics; and Alpine Quantum Technologies, a spin-off from the University of Innsbruck (Germany) by Rainer Blatt and Thomas Monz, two leading quantum physicists.

AQT is working on developing modular ion trap quantum computers that can be operated in standard data centers—without the need for extreme cooling or laboratory-sized vacuum chambers. Research at the University of Innsbruck and the Institute for Quantum Optics and Quantum Information (IQOQI) of the Austrian Academy of Sciences has played a central role in the development of this technology.

At the Leibniz Computing Center (LRZ) of the Bavarian Academy (Germany) of Sciences, a 20-qubit ion trap quantum computer from Alpine Quantum Technologies is available for novel research tasks. The system was procured by the Leibniz Computing Center and the Munich Quantum Valley with funding from Bavarian state ministries as part of the Hightech Agenda Bayern. With this modular machine in a 19-inch architecture, AQT became the first hardware provider to install and commission an ion-trap quantum computer in a data center, according to Dr. Thomas Monz, founder and CEO of AQT.

"This collaboration strengthens MQV's resources, enhances its capacity for innovation, enables groundbreaking discoveries, and reinforces MQV's position as a leading initiative in the field of quantum computing," commented Prof. Dr. Joachim Ullrich, Director General MQV.

The LRZ intends to advance the development of the Munich Quantum Software Stack in close collaboration with AQT and MQV to "provide users with a robust, end-to-end hybrid quantum-HPC computing resource for scientific activities," reveals Laura Schulz, Head of Quantum Computing and Technology at the LRZ.

Photonic qubits

Photonic qubits use the quantum mechanical properties of light particles (photons) to store and process quantum information. Photonic qubits can be easily networked and are room-temperature capable, as photons do not interact directly with their environment and therefore exhibit very long coherence times.

Quantum information can be encoded in various degrees of freedom of photons, be it polarization states (horizontal/vertical or right-/left-circular), time-bin encoding (early vs. late arrival time), or path encoding (different light paths in interferometers).

In Germany, research on photonic qubits is being conducted by TRUMPF subsidiary Q.ANT from Stuttgart (Germany), Fraunhofer IOF in Jena (germany), Max Planck Institute for Quantum Optics (MPQ, Garching), Karlsruhe Institute of Technology (KIT), QuNET (German quantum communication network, funded by the BMBF), and LMU Munich & Humboldt University of Berlin.

On an international level, companies such as PsiQuantum (USA/UK), Canadian Xanadu, and British ORCA Computing are active in this field, alongside research institutions at MIT, Harvard, Caltech, and others.

Topological qubits

Topological qubits are a promising yet experimental type of qubits based on the exotic properties of quasiparticles (anyons). Their main advantage over other qubit technologies is intrinsic error resistance. Topological qubits encode quantum information in the system's global properties, which are not easily affected by local disturbances.

As a basis for qubit manipulations, anyons with their non-trivial exchange statistics are used here. A particularly interesting type of these particles are non-abelian anyons—their quantum states depend not only on their current position but also on their exchange path, the so-called braiding (the twisting of quasiparticle paths).

In a topological system, one can "encircle" or "twist" two anyons. The order of the movements determines the final state of the system—similar to a logic gate in classical computers. Since these qubit states are not locally stored but distributed across the entire system, they are robust against disturbances and decoherence.

Research on topological qubits includes efforts by Microsoft (StationQ, QuArC) and Nokia Bell Labs, a research facility of the Finnish telecommunications company by the same name. Redmond relies on Majorana fermions in hybrid semiconductor-superconductor structures, while Nokia focuses on the Nobel Prize-winning discovery of the fractional quantum Hall effect.

Molecular qubits

Molecular qubits encode quantum information in the electron spin of paramagnetic metal complexes (e.g., with copper, vanadium, or lanthanides) and/or nuclear spin within a molecule. The quantum information in electron spin-based molecular qubits is stored and read using magnetic fields or microwaves. Nuclear spin-based molecular qubits are controlled using NMR (Nuclear Magnetic Resonance) or ESR (Electron Spin Resonance) techniques.

A key advantage of molecular qubits is the ability to optimize properties such as coherence time and coupling through targeted molecular synthesis. Additionally, these qubits can be combined with other qubit technologies. They can be integrated into solid-state structures or superconducting systems, among others. In Germany, research in this area is being conducted by institutions including the University of Stuttgart (Germany) and the Fraunhofer IAF.

Conclusion

Through their bizarre quantum properties, qubits open up completely new perspectives for research and industry. However, not all qubits are the same: each architecture sets its own priorities and brings peculiar challenges on the way to large-scale industrial quantum computing.

Superconducting qubits excel with extremely fast switching operations but suffer from comparatively short coherence times. Silicon spin qubits are compatible with the established semiconductor industry and could facilitate the transition from the laboratory to industrial manufacturing processes. In an ion trap, qubit states remain stable for a particularly long time, promising high reliability. Photonic qubits, on the other hand, are easy to network and can potentially operate even at room temperature.

In the long run, other, today completely unthinkable, qubit platforms could prevail. The race for the best qubit architectures is certainly in full swing. (mbf)