Strategies to search for two-dimensional materials with long spin qubit coherence time

TL;DR

This study tackles the challenge of finding two-dimensional materials that can host spin qubits with long coherence times by developing a high-throughput workflow based on cluster correlation expansion (CCE) to predict nuclear spin bath-driven decoherence in 2D hosts and their heterostructures. It screens 1173 monolayers, identifying 190 with $T_2 > 1$ ms, and constructs 1554 lattice-matched heterostructures, finding that substrates with low nuclear spin noise can preserve high $T_2$, while decoherence can arise from the noisier component of the stack. The authors derive 2D-specific analytical models to rapidly estimate $T_2$ from structural and isotopic information, enabling expansion of the screening space to 4705 monolayers and validating many predictions with order-2 CCE calculations. The work provides design principles for 2D spin-qubit platforms, highlights the importance of substrate choice in heterostructures, and offers scalable, analytic tools to accelerate AI-assisted discovery of robust spin qubit materials.

Abstract

Two-dimensional (2D) materials that can host qubits with long spin coherence time (T2) have the distinct advantage of integrating easily with existing microelectronic and photonic platforms, making them attractive for designing novel quantum devices with enhanced performance. However, the relative lack of 2D materials as spin qubit hosts, as well as appropriate substrates that can help maintain long T2, necessitates a strategy to search for candidates with robust spin coherence. Here, we develop a high-throughput computational workflow to predict the nuclear spin bath-driven qubit decoherence and T2 in 2D materials and heterostructures. We initially screen 1173 2D materials and find 190 monolayers with T2 > 1 ms, higher than that of naturally-abundant diamond. We then construct 1554 lattice-commensurate heterostructures between high-T2 2D materials and select 3D substrates, and we find that T2 is generally lower in a heterostructure than in the bare 2D host material; however, low-noise substrates (such as CeO2 and CaO) can help maintain high T2. To further accelerate the material screening effort, we derive analytical models that enable rapid predictions of T2 for 2D materials and heterotructures. The models offer a simple, yet quantitative, way to determine the relative contributions to decoherence from the nuclear spin baths of the 2D host and substrate in a heterostructural system. By developing a high-throughput workflow and analytical models, we expand the genome of 2D materials and their spin coherence times for the development of spin qubit platforms.

Figures (5)

• Figure 1: Computational methodology. (a) The workflow for screening 2D materials and heterostructures with long coherence times ($T_2$) adopted in this work. Promising 2D materials with high $T_2$ times are identified initially, followed by the creation and screening of heterostructures. (b) The automated methodology for performing cluster correlation expansion (CCE) simulations to calculate $T_2$ devised in this work.
• Figure 2: 2D materials with high coherence time. (a) Comparison of spin coherence time ($T_2$) for qubits implanted in 2D monolayers (red) and corresponding 3D parent phases (blue). 2D materials can generally accommodate spin qubits with longer coherence time than their 3D parent phase. Structures of monolayers with the highest $T_2$ are shown, namely (b) WS2, (c) Au2Se3O10, (d) Au2Se2O7, and (e) PdSO4.
• Figure 3: Coherence time of heterostructures. (a) Spin coherence time ($T_2$) of heterostructures compared to $T_2$ of the bare 2D host material. The color indicates the substrate material. (b-e) $T_2$ of heterostructures with WS2, Au2Se3O10, Au2Se2O7, and PdSO4. The colored bars represent heterostructures with the substrate indicated along the vertical axis, and the gray bar represents the bare 2D host. (f) $T_2$ of the bare substrate materials, without the 2D material.
• Figure 4: Model of coherence time for 2D materials. (a) Parity plot of the original (light) and revised (dark) models of the spin coherence time ($T_2$) for 2D materials. (b) Statistics from screening various 2D materials databases using the revised model of $T_2$. (c) $T_2$ predicted by the revised model for monolayers across various 2D materials databases. An additional 546 monolayers with $T_1 > 1$ ms are found by screening these databases and verifying with CCE calculations.
• Figure 5: Model of coherence time for heterostructures. (a) The calculated coherence time ($T_2$) of heterostructures, compared to the $T_2$ predicted by the fitted model. (b) Model $T_2$ of heterostructures, calculated using Eq. (\ref{['Eq:Model_T2_HS']}), against $T_2$ of the bare 2D host material. Each curve corresponds to a different substrate $T_2$.