Skip to content

Benchmarking quantum error-correcting codes on quasi-linear and central-spin processors

MetadataDetails
Publication Date2022-11-11
JournalQuantum Science and Technology
AuthorsRegina Finsterhoelzl, Guido Burkard
InstitutionsUniversity of Konstanz
Citations6
AnalysisFull AI Review Included

Benchmarking Quantum Error-Correcting Codes on Diamond NV-Center Platforms

Section titled “Benchmarking Quantum Error-Correcting Codes on Diamond NV-Center Platforms”
  • Core Focus: Comparative analysis of small Quantum Error-Correcting (QEC) codes transpiled onto superconducting transmon processors and Nitrogen-Vacancy (NV) centers in diamond.
  • Key Finding — Bit-Flip Code: The superconducting processor (SC) achieved superior logical error rates due to faster gates and lower entangling gate overhead (3 CNOTs vs. 7 CNOTs for NV-center).
  • Key Finding — Phase-Flip Code: The NV-center’s central-spin system (CSS) connectivity outperformed the SC platform for multi-qubit controlled operations, requiring significantly fewer CNOT gates (native CCZ / Toffoli gate on the electron spin).
  • Native Gate Advantage: NV-center registers natively support Cn-rotations controlled by the electron spin on any two nuclear spins, providing an intrinsic all-to-all mediated connectivity.
  • Coherence Advantage: NV-center coherence times (T2 Hahn up to 2300 ms for 14N nuclear spin) vastly exceed SC coherence times (~0.1 ms), though gate fidelities and speeds currently favor SC.
  • Application Implication: Small codes benefit from quasi-linear connectivity; multi-qubit controlled codes and flag-qubit stabilizer schemes benefit from CSS connectivity.
  • Methodology: Custom error model combining amplitude-phase damping (E_apd) and depolarizing noise channels, benchmarked against IBM Q Falcon hardware and validated for NV-center via calibrated data from the University of Stuttgart.
ParameterSC (Transmon)NV-center (Electron)NV-center (Nuclear)UnitContext
Operating Temperature (T_op)1.5 × 10-2300300KCryogenic vs. room-temp operation
T1 Relaxation0.15.7≳ 250msEnergy relaxation time
T2 Hahn0.10.40.9 (up to 2300)msSpin-echo coherence time
T2 Extended (Dynamical Decoupling)N/Aup to 1000> 60,000msProtected coherence
Single-Qubit Gate Fidelity (F1)0.99920.99995——Best demonstrated
Two-Qubit Gate Fidelity (F2)0.9940.992——CNOT equivalent
Single-Qubit Gate Time (T_gate)0.00050.001-0.150.001-0.15msNative gate duration
SPAM Error (p_spam)0.02-0.040.02-0.040.02-0.04—State prep & measurement
Qubit Transition Frequency5~GHz~MHz—Electron vs. nuclear spin
Native Two-Qubit GateCNOTC(n)(2πx)Controlled rotation—Platform-specific basis
ConnectivityQuasi-linear / hexagonalCentral-spin (CSS)CSS—Coupling topology
Bit-Flip Code Logical Error Rate0.0230.139——Best result, initial state |0⟩
Phase-Flip Code CNOT Count (SC)35N/AN/AgatesTranspilation overhead
  1. Hardware Selection: IBM Q Falcon processor (SC, hexagonal connectivity) versus NV-center in diamond (CSS connectivity, Stuttgart calibration data).
  2. Noise Model Construction:
    • Amplitude-phase damping channel E_apd parameterized by T1, T2, and gate time Δt.
    • Depolarizing channel E_depol with probability p_depol.
    • Pauli bit-flip channel for SPAM errors.
    • Uncorrelated noise assumption across qubits.
  3. Codes Implemented:
    • Three-qubit bit-flip repetition code (stabilizers Z1Z2, Z2Z3) with post-processing recovery.
    • Three-qubit phase-flip repetition code (stabilizers X1X2, X2X3) with unitary correction.
  4. Qubit Routing / Transpilation:
    • Reduce circuit depth via identities (e.g., CNOT(q0,q1)CNOT(q0,q1) = I).
    • Minimize SWAP operations (each SWAP = 3 CNOTs).
    • Map virtual qubits onto optimal physical layout (linear preferred for SC; electron spin as ancilla for NV-center).
  5. Error Analysis Protocol:
    • Deliberately induce random bit-flip during noise evolution.
    • Calculate average error rate p_av = 1 − ∏N-1i=0(1 − Fi).
    • Benchmark against real SC hardware via syndrome measurement.
  6. Performance Metrics:
    • Logical error rate (infidelity between |ψ⟩_ideal and ρ_out).
    • Logical error threshold scaling: p̄ = C · pe2 for single-error-correcting codes.
  • Applicable Materials: This research requires ultra-high-purity Optical Grade Single Crystal Diamond (SCD) substrates optimized for NV-center formation. 6CCVD’s high-purity SCD (nitrogen concentration < 1 ppb achievable) provides the pristine lattice environment necessary for the long T2 coherence times (up to 2300 ms for 14N nuclear spin) reported in this study. For applications requiring ensemble addressing or multi-NV registers, our Polycrystalline Diamond (PCD) and Boron-Doped (BDD) variants offer cost-effective alternatives.

  • Customization Potential: The NV-center register requires precisely engineered diamond chips with specific crystallographic orientations ([100] or [111]) and ultra-low defect densities. 6CCVD offers custom wafer dimensions up to 125mm (PCD) and 10mm-thick substrates, with thickness control down to 0.1 µm for thin-film SCD applications. For photonic or electrical interfacing, we provide in-house metalization services (Ti, Pt, Au, Pd, W, Cu) to fabricate coplanar waveguides, microwave resonators, and readout structures directly on the diamond surface.

  • Laser Cutting & Dicing: Single NV experiments typically require sub-mm to few-mm diamond membranes (e.g., 3mm × 3mm × 0.5mm for solid immersion lenses or pillar resonators). 6CCVD provides precision laser cutting and dicing to deliver custom-shaped diamond chips matching exact NV-center device geometries.

  • Surface Engineering: Ra < 1nm polishing for SCD and < 5nm for inch-size PCD ensures minimal surface charge noise — critical for NV-center gate fidelities and spectral stability. We can also supply isotopic enrichment consultation (12C enriched diamond) to further extend T2 coherence beyond what was measured in this paper.

  • Engineering Support: Quantum error correction on diamond platforms demands tight integration between material properties (isotopic purity, nitrogen concentration, strain) and qubit performance. 6CCVD’s in-house PhD engineering team can assist with material selection — matching NV density, coherence requirements, and spin-bath engineering to specific QEC architectures (repetition codes, flag-qubit stabilizer schemes, or concatenated codes using multi-NV arrays).

  • Call to Action: For custom specifications, isotopic enrichment consultation, or material selection guidance for your NV-center quantum computing project, visit 6ccvd.com or contact our engineering team directly.

View Original Abstract

Abstract We evaluate the performance of small error-correcting codes, which we tailor to hardware platforms of very different connectivity and coherence: on a superconducting processor based on transmon qubits and a spintronic quantum register consisting of a nitrogen-vacancy center in diamond. Taking the hardware-specific errors and connectivity into account, we investigate the dependence of the resulting logical error rate on the platform features such as the native gates, native connectivity, gate times, and coherence times. Using a standard error model parameterized for the given hardware, we simulate the performance and benchmark these predictions with experimental results when running the code on the superconducting quantum device. The results indicate that for small codes, the quasi-linear layout of the superconducting device is advantageous. Yet, for codes involving multi-qubit controlled operations, the central-spin connectivity of the color centers enables lower error rates.