A CMOS Silicon Spin Qubit Power Scalable Quantum Computers
A CMOS Silicon Spin Qubit
Industrial Manufacturing Makes Scalable Silicon Qubit Arrays Possible: Quantum Computing Revolution
Researchers used advanced commercial semiconductor fabrication processes to produce a two-dimensional array of silicon spin qubits, breaking a major barrier to quantum computer development. This achievement offers up new possibilities for scalable spin qubit technology and is a significant step towards building quantum computers on a scale similar to that of integrated circuits, according to recent papers.
CMOS complementary metal oxide semiconductor
Development of physical qubits with broad scalability and low gate error rate is key to quantum computers. Silicon-based spin qubits have led this industry due to their intrinsic coherence and possibility to leverage CMOS production methods.
Silicon CMOS chips, found in practically all electronic devices, include billions of transistors coupled by nanoscale wires and vias to allow components to communicate with the outside world.
These crucial interconnect technologies have not yet been widely adopted by silicon qubit chip architectures, restricting their scalability to one-dimensional arrays. Early attempts to produce qubits in industrial facilities used “in-line routing” at the gate level to ensure interior gate communication in larger arrays.
Because they require array-wide qubit control, crossbar systems limit yield and homogeneity despite their high scalability.
BEOL backend
A unique connection process with many back-end-of-line (BEOL) layers helped the study team overcome these basic scalability issues. This approach allows signal routing lines to be topographically separated from gate electrodes, like in conventional integrated circuits, enabling a fully extendable architecture. The demonstration equipment utilises three layers to connect with a 5x5 grid of gates that regulate charge sensors and quantum dots. This complicated architecture provides quantum coherent communication to all nearest neighbours for every spin in the two-dimensional array.
Exchange-only (EO) qubits, a spin-qubit modality, are essential to this work. Each EO qubit is encoded with three electron-containing quantum dots. This encoding is ideal for two-dimensional arrays because of its flexibility.
If plunger or exchange gates in the qubit array malfunction due to manufacturing defects or other issues, EO qubits can be formed in linear and right-angle elbow configurations to maximise performance and connectivity. This reconfigurability can drastically minimise manufacturing yield constraints.
The performance indicators are good. The researchers found that all hosted EO spin qubits behaved coherently like single-interconnect linear arrays. Importantly, performance was unrelated to connecting layer count. Blind randomised benchmarking (BRB) was utilised to test single-qubit gate fidelities, and above 99.9% were consistently achieved.
An average leakage rate determined the average single-qubit error rate. The study showed that adding more than one BEOL layer did not affect EO qubit performance. There was no correlation between an exchange gate's BEOL layer and its noise parameter, and exchange coherence in this multilayer BEOL device was similar to that of devices with a single connecting layer.
State management and qubit operations use exchange-based control. Pauli spin blocking simplifies state preparation and measurement (SPAM), and calibrated spin swaps may initialise and measure arbitrary EO triple-quantum-dot qubits by coherently transferring two-spin states across the array's double quantum dots. For single-qubit characterisation, the 2x3 array device supports 10 triple-quantum-dot (TQD) permutations across eight configurations, demonstrating its versatility.
This demonstration is a big advance, but researchers say more has to be done. Even though it can be expanded in two dimensions by adding BEOL layers, the device has manufacturing, material, and electrostatic constraints.
Denser arrays make layer alignment accuracy (overlay) harder, yet extreme-ultraviolet laser lithography can do this. For larger arrays, crosstalk and signal integrity loss due to greater BEOL resistance and interlayer dielectric capacitance are being monitored, even though they weren't found in the three-layer arrangement.
Adjusting larger 2D arrays may be problematic due to quantum-dot-based charge sensors' rapid reduction in sensitivity with distance, requiring other measuring methods.
Despite these challenges, this extended device platform proves that industrial production can produce scalable spin qubit technologies. The ability to build 2D arrays with high-performance reconfigurable qubits brings practical quantum computation closer.
Research on larger, more complicated quantum computing structures can begin. Future study will examine the benefits of complete connectivity for two-qubit operations, scaling to larger sizes, and monitoring multi-qubit performance indicators to improve gate fidelities.
