Digital Predistortion of Optical Fields for Fast and High-Fidelity Entangling Gates in Trapped-Ion Qubits

1. Problem Statement

High-fidelity quantum gates in trapped-ion processors rely on precise classical control signals. However, the analog hardware used to generate these signals—specifically Acousto-Optic Modulators (AOMs) in laser-driven systems—introduces nonlinear distortions.

• The Issue: When AOMs are driven at high RF power to maximize diffraction efficiency (and thus gate speed), they exhibit a compressive nonlinear amplitude response.
• The Consequence: This nonlinearity generates spurious intermodulation products (IM). In spectrally dense multi-tone waveforms used for entangling gates (e.g., the Cardioid gate), these IM products can fall directly on motional sideband frequencies.
• The Impact: These spurious tones drive unintended spin-motion dynamics, causing phase-space trajectory non-closure. This creates a practical trade-off: operators must either reduce drive power (slowing down gates and reducing throughput) to avoid distortion or accept lower gate fidelities.

2. Methodology

The authors adapt Digital Predistortion (DPD), a feed-forward linearization technique common in RF communications, to the optical drive chain of a trapped-ion quantum processor.

• System Characterization:
  • The team characterized the static nonlinear amplitude and phase response of the AOM used to modulate the 674 nm laser beam for $^{88}\text{Sr}^+$ ions.
  • They measured the optical amplitude envelope $y(A)$ versus the normalized RF input amplitude $A$.
  • They also measured the amplitude-dependent phase shift $\phi(A)$ via heterodyne detection, though the primary correction applied was amplitude-only.
• The Correction Algorithm:
  • A polynomial model ($K=8$) was fitted to the measured amplitude transfer function $y(A)$.
  • The numerical inverse of this function, $f^{-1}$, was computed.
  • Feed-forward Correction: Before generating the gate waveform, the desired ideal envelope $x_{\text{ideal}}(t)$ is passed through the inverse function to create a predistorted input $x_{\text{DPD}}(t) = f^{-1}(x_{\text{ideal}}(t))$.
  • When this predistorted signal drives the AOM, the nonlinearity of the hardware cancels the predistortion, resulting in a linear optical output $y(t) \approx x_{\text{ideal}}(t)$.
• Experimental Setup:
  • Platform: Two-ion $^{88}\text{Sr}^+$ crystal in a linear Paul trap.
  • Gate: Cardioid(1,2) entangling gate using four tones (two blue-detuned, two red-detuned) to mediate interaction via the axial stretch mode.
  • Metrics: Performance was benchmarked via heterodyne spectral analysis (measuring gate-tone-to-IM power ratios) and direct Bell-state fidelity measurements (parity-fringe contrast).

3. Key Contributions

• First Demonstration of Nonlinear DPD for Entangling Gates: While amplitude linearization and phase predistortion have been used previously for single-qubit gates or instrumentation, this is the first demonstration of nonlinear predistortion specifically improving two-qubit entangling gate fidelity.
• Device and Platform Agnosticism: The "calibrate-and-invert" methodology requires only a measured transfer function and its numerical inverse. It is applicable to any nonlinear element in a quantum control chain, including RF power amplifiers in microwave-driven gates and electro-optic modulators.
• Quantitative Extension of Linear Range: The method successfully extends the effective linear operating range of the AOM, allowing for higher optical throughput without fidelity degradation.

4. Key Results

• Spectral Suppression:
  • Within the correctable range (up to a normalized amplitude $A_{\text{corr}} \approx 0.566$), DPD suppressed the dominant intermodulation tones (IM3) by 3–5 dB relative to the gate tones.
  • This suppression was consistent across different drive amplitudes and matched numerical simulations incorporating the measured transfer functions.
• Gate Fidelity Improvement:
  • Direct measurements of Bell-state fidelity showed that DPD consistently outperformed uncorrected operation.
  • At similar gate rates, DPD yielded a ~5.5 percentage point increase in fidelity (e.g., from ~0.91 to ~0.965).
  • Parity-fringe measurements confirmed that the improvement was driven by better phase-space closure (reduced residual spin-motion entanglement).
• Throughput and Efficiency Gains:
  • DPD shifted the onset of nonlinearity-limited performance to higher diffraction efficiencies.
  • At an estimated gate-error threshold of $10^{-3}$, the usable diffraction efficiency was doubled (increasing from ~0.037 to ~0.080).
  • This allows for faster gate speeds (higher Rabi frequencies) while maintaining high fidelity.

5. Significance and Future Outlook

• Scalability: As quantum processors scale to more qubits and utilize increasingly complex, dense multi-tone waveforms, hardware nonlinearities will generate more intermodulation products. DPD provides a scalable software-defined solution to maintain control fidelity without requiring perfect analog hardware.
• Speed vs. Fidelity Trade-off: This technique breaks the traditional trade-off between gate speed and fidelity. It enables the use of higher drive powers (faster gates) without the penalty of spurious tones degrading the quantum state.
• Future Work: The authors note that the current implementation corrects only amplitude nonlinearity. Future iterations could incorporate phase predistortion and memory-polynomial models to address history-dependent distortions (memory effects) and further improve performance at high modulation bandwidths.

In conclusion, this work demonstrates that digital predistortion is a powerful, low-overhead tool for enhancing the performance of trapped-ion quantum processors, directly addressing a critical bottleneck in the analog control chain.