PaperlayerView exact revision on arXiv

Source-linked AI summary

Trapped-Ion Quantum Computing: Progress and Challenges

Colin D. Bruzewicz, John Chiaverini, Robert McConnell, Jeremy M. Sage

arXiv:1904.04178v1quant-phphysics.atom-ph

TL;DR

Trapped-ion QC has demonstrated the core ingredients of quantum computation, but scaling from few-ion experiments to useful large systems remains unresolved. This review synthesizes control methods, decoherence and error challenges, scaling architectures, and near-term demonstrations, highlighting high-fidelity gates alongside slower operations and implementation constraints.

  • Problem

    The field must determine how to increase trapped-ion QC beyond few-ion systems while mitigating decoherence, control errors, and architectural scaling challenges.

  • Method

    The paper reviews trapped-ion qubits, gate methods, decoherence and error mechanisms, scaling technologies, architectures, and near-term experiments.

  • Results

    99.9% two-qubit fidelity has been demonstrated with a geometric-phase gate, while microwave-gradient gates reached 99.7% and 98.5% fidelity on millisecond timescales.

  • Takeaways & Limitations

    Trapped ions combine high-fidelity operations with long-standing scaling challenges involving motional heating, gate speed, connectivity, and error correction.

Abstract

from arXiv · show

Trapped ions are among the most promising systems for practical quantum computing (QC). The basic requirements for universal QC have all been demonstrated with ions and quantum algorithms using few-ion-qubit systems have been implemented. We review the state of the field, covering the basics of how trapped ions are used for QC and their strengths and limitations as qubits. In addition, we discuss what is being done, and what may be required, to increase the scale of trapped ion quantum computers while mitigating decoherence and control errors. Finally, we explore the outlook for trapped-ion QC. In particular, we discuss near-term applications, considerations impacting the design of future systems of trapped ions, and experiments and demonstrations that may further inform these considerations.

I. INTRODUCTION

Trapped ions satisfy the main requirements for quantum computation and support high-fidelity control, preparation, and readout, but practical systems must overcome scaling and speed limitations. This review surveys trapped-ion QC progress, outstanding challenges, and near-term directions.

  • Achievements and challenges: Single-qubit gates, two-qubit gates, state preparation, and readout have all exceeded fidelities required by high-threshold fault-tolerant codes.These capabilities were demonstrated before the field achieved large controlled registers.
  • Scope and motivation: The review examines how trapped-ion QC can progress from high-fidelity few-ion demonstrations toward systems containing hundreds or more ions.It emphasizes control, loading, detection, scaling technologies, near-term experiments, and long-term outlook.
  • Limitations: The largest fully controlled trapped-ion register had only 20 ions despite the main DiVincenzo criteria having been essentially satisfied since 2004.The review identifies increasing the number of simultaneously trapped ions as a chief unresolved challenge.
  • Strengths: 50 s hyperfine-qubit coherence has been achieved without dynamical decoupling, extending to 600 s with dynamical decoupling.These are effectively T2 times limited by technical dephasing rather than fundamental state lifetimes.
  • Strengths: Single-qubit rotations reached 99.9999% fidelity, while two-qubit entangling gates reached 99.9% for hyperfine qubits and 99.6% for optical qubits.Only superconducting qubits achieved comparable two-qubit performance according to the review.
  • Strengths: Laser-based readout exceeded 99.99% fidelity in less than 200 µs, and combined preparation and readout reached 99.93% fidelity.The review reports inferred state-preparation errors of 2 × 10^-4 from the combined measurement.
  • Limitations: Trapped-ion gates are slower than those in superconducting qubits despite trapped ions’ high coherence-to-gate-time ratio.The review estimates approximately 10 days and 100 days to factor 1024-bit and 2048-bit numbers, respectively, under optimistic parameters.

E. Considerations for Scaling a Trapped-Ion Quantum Computer

Scaling trapped-ion QC requires architectures that preserve control, fidelity, coherence, connectivity, and physical qubit retention as systems grow. The review considers modular and monolithic approaches while emphasizing unresolved questions and newly emerging challenges.

  • Definition of scalability: A scalable QC system must increase qubit numbers over orders of magnitude while maintaining control, high-fidelity gates, long coherence, and practical resource growth.The review notes that no QC technology currently achieves scalability in this full sense.
  • Architectural approaches: Modularity combines independently built and tested subsystems with defined functions and compatibility across modules.Maintaining entanglement across modules may introduce challenges specific to quantum systems.
  • Architectural approaches: Monolithic integration combines functions into a single component such as a microfabricated chip and complements modular system design.Examples include on-chip light-delivery waveguides and detectors.
  • Error correction: Error correction encodes information in logical qubits composed of multiple physical qubits, creating substantial resource overhead.The review identifies error correction as a key capability needed for scaling.
  • Control and connectivity: Scalable architectures must support large ion populations, high-fidelity operations on any ion, and low crosstalk during scaling.They also require sufficient connectivity for generating entanglement throughout the processor.
  • System stability: Ion loss, caused by background-gas collisions or experimental imperfections, must be physically managed or treated as a detectable amplitude-damping error.Suitable codes can correct loss when it is detectable within the architecture.
  • Control infrastructure: Scalable addressing and measurement must replace the large collections of bulk optics and external voltage supplies used in most current experiments.The review frames the sheer number of optical and electrical components as a scaling challenge.
  • Open questions: The review cautions that approaches remain uncertain and that new challenges will appear as trapped-ion systems expand from a few qubits to hundreds or thousands.Its discussion focuses on currently known obstacles rather than asserting a settled scaling route.

A. Trapping Individual Ions

Paul traps localize ions using time-dependent or combined electromagnetic fields, with point and linear configurations providing the main QC architectures. Trap geometry, micromotion, fabrication, and loading choices shape scalability and control.

  • Trap principles: Paul traps use oscillating electric fields to create a ponderomotive confining pseudopotential in two or three dimensions.Penning traps instead combine static electric confinement axially with magnetic confinement radially.
  • Trap principles: RF trapping produces secular harmonic motion plus higher-frequency micromotion, with stability constrained by the applied RF voltage and frequency.The secular frequency is typically somewhat less than half the RF drive frequency.
  • Paul-trap configurations: Point traps provide three-dimensional RF confinement, whereas linear traps combine two-dimensional RF confinement with static axial confinement.Multiple ions in point traps generally experience excess micromotion because only one RF null exists.
  • Paul-trap configurations: RF Paul trap geometries range from ring-and-endcap and four-rod designs to planar surface-electrode traps and segmented multizone arrays.Segmentation enables trapping in multiple axial zones.
  • Ion registers: Linear traps can hold multiple ions along the RF null, forming ion chains whose spacing is set by trap forces and Coulomb repulsion.For same-charge ions, spacing is independent of mass, allowing multispecies crystals to share identical spacing in linear traps.
  • Ion registers: Arrays of single-ion point traps are one proposed architecture for avoiding the excess micromotion caused by multiple ions in a point trap.Linear traps offer an alternative by holding multiple ions in one chain.
  • Fabrication: Microfabricated traps introduced smaller and more precisely defined structures, while multilayer lithography enabled increasingly complex segmented arrays.Some current designs contain hundreds of separate electrode segments.
  • Fabrication: Surface-electrode traps place all electrodes in one plane and form RF-null trapping minima above the surface.Surface-electrode Penning traps have been explored but are not widely used for ion-based QC.

B. Internal States: Qubit Levels

Trapped-ion QC offers several qubit choices from long-lived electronic states, each trading coherence, field sensitivity, detection complexity, and control requirements. Hyperfine and optical qubits provide particularly useful combinations of coherence and scalability properties.

  • Zeeman Qubits: Zeeman qubits offer essentially infinite lifetimes and simple level structures but are highly sensitive to magnetic-field variations.Magnetic shielding has enabled 300 ms coherence, extended to 2.1 s with dynamical decoupling, while residual thermal fluctuations remain limiting.
  • Qubit choices: Qubit states can be selected from Zeeman, hyperfine, optical, or fine-structure levels, with splittings ranging from megahertz to terahertz.Typical splittings are 1–10 MHz for Zeeman, 1–10 GHz for hyperfine, 100–1000 THz for optical, and 1–10 THz for fine-structure qubits.
  • Hyperfine Qubits: Hyperfine qubits combine long lifetimes with reduced magnetic-field sensitivity and simpler detection than Zeeman qubits, at the cost of more complicated level structures.Clock-type hyperfine states are first-order field-insensitive at zero field, while finite-field FOFI qubits retain practical operation at nonzero field.
  • Hyperfine Qubits: FOFI qubits provide extremely low first-order sensitivity to field fluctuations while operating at nonzero magnetic fields.Coherence times of minutes have been demonstrated in standard Ramsey measurements without dynamical decoupling; technical drift and oscillator instability remain limitations.
  • Optical Qubits: Optical qubits use metastable states for high-efficiency detection and favor scalable photonic integration, but require narrow-linewidth lasers and careful phase control.Control lasers may need linewidths around 1 Hz, while experiments have achieved more than 0.2 s coherence; red-to-near-IR control wavelengths support integrated routing technologies.

4. Fine-Structure Qubits

Fine-structure qubits use long-lived D-manifold states coupled through the shared motional modes of trapped-ion chains. Motional decoherence, especially anomalous heating, constrains gate fidelity and complicates trap miniaturization, although cooling and surface treatments mitigate it.

  • 4. Fine-Structure Qubits: Fine-structure qubits encode information in the D3/2 and D5/2 manifolds, with terahertz splittings and second-scale lifetimes.Quantum logic can use Raman transitions through P levels or potentially direct terahertz control.
  • C. Motional States: Shared vibrational normal modes act as quantum buses that couple the internal electronic states of separate ions.For N ions, the trap has 3N normal modes, and laser interactions can change internal and motional states together through red and blue sidebands.
  • C. Motional States: The Lamb-Dicke regime, η√n + 1 ≪ 1, restricts motional transitions and supports tractable, high-fidelity dynamics.The Lamb-Dicke parameter depends on optical wavevector, motional-mode orientation, ion mass, and mode frequency.
  • 2. Anomalous Motional Heating: Anomalous motional heating is a practical limit to multiqubit gate fidelity and a hindrance to miniaturizing traps for higher-frequency logic.Its electric-field-noise source is unresolved and heating increases approximately as d^-4 with ion-electrode distance d.
  • 1. Motional State Decoherence: 10^-6–10^-5 fractional trap-frequency fluctuations occur over tens to hundreds of seconds, creating an engineering challenge for large arrays and fault-tolerant gates.Low-frequency fluctuations can change motional-mode phase evolution and cause decoherence without heating.
  • 2. Anomalous Motional Heating: Cooling electrodes to approximately 4 K reduces anomalous motional heating by about two orders of magnitude, while several surface treatments also reduce electric-field noise.For d = 40–80 µm, in situ ion milling or operation below about 10 K can produce heating rates compatible with high-threshold fault-tolerant QC.

III. TRAPPED ION QUBIT CONTROL

Trapped-ion control combines internal-state preparation, motional cooling, universal gate decompositions, and state detection. Laser, Raman, microwave, and magnetic-gradient approaches offer different trade-offs among fidelity, linewidth, addressability, and scalability.

  • A. State Preparation: Trapped-ion QC requires repeated high-fidelity internal-state preparation and often motional-state preparation before quantum operations.Doppler cooling rapidly reaches the millikelvin scale but leaves ions distributed across several motional states at typical ∼1 MHz trap frequencies.
  • A. State Preparation: EIT cooling has cooled chains of up to 18 ions to phonon occupations of about 0.01–0.02 while addressing multiple motional modes.Performance was limited by laser-polarization purity, and the technique has also been applied to mixed-species chains.
  • Types of Gates: Universal trapped-ion control combines arbitrary single-qubit rotations with at least one entangling two-qubit gate, commonly implemented through shared motional modes.Arbitrary gates can be decomposed into a smaller universal gate set, and nearly any entangling gate suffices with additional single-qubit rotations.
  • 1. Types of Gates: Optical, Raman, Microwave: Hyperfine single-qubit gates can use microwaves or Raman transitions, whereas optical qubits can be driven with a resonant laser.Raman gates use two detuned beams whose difference frequency is controlled with acousto-optic modulators and stable RF or microwave synthesizers.
  • 1. Types of Gates: Optical, Raman, Microwave: Optical-qubit gates require near-hertz laser linewidths because optical phase noise and path-length fluctuations limit achievable coherence.Longer-lifetime optical transitions impose even tighter demands on optical control resources.
  • 1. Types of Gates: Optical, Raman, Microwave: Microwave control is easy for a single ion but its centimeter-scale wavelength makes individual addressing difficult, motivating integrated wires and magnetic gradients.Magnetic gradients enable spectroscopic addressing and microwave-mediated coupling to motion, although crosstalk remains an issue.

2. Single Qubit Gates

Single-qubit gates in trapped ions combine very high fidelity with microsecond-to-nanosecond operation times, although faster operation commonly trades accuracy for speed. The section also introduces two-qubit gate mechanisms and their fidelity, speed, and control limitations.

  • Single-qubit gate performance: Optical single-qubit gates reach 99.995% fidelity in a few microseconds, with ion selectivity and low crosstalk as notable strengths.Their ultimate fidelity is limited by the excited-state T1 time, approximately 1 s for commonly used quadrupole transitions.
  • Single-qubit gate performance: 99.9999% fidelity was achieved for microwave single-ion gates in 12 µs, while Raman gates reached about 99.993% in 7.5 µs.Raman fidelity was limited by off-resonant scattering.
  • Single-qubit gate performance: Single-qubit gate speed and fidelity generally trade off: sub-2-µs gates were demonstrated with errors of approximately 2 × 10^-4.Ultrafast demonstrations reached below 50 ps with 99% fidelity, while another microwave demonstration used gates below 20 ns without reporting fidelity.
  • Multi-qubit gate mechanisms: Multi-qubit gates use shared motional modes as a quantum bus, and single-qubit rotations can transform the controlled-phase CZ gate into a CNOT.The CZ gate transfers population through motional sidebands, while the MS and geometric-phase gates use related motion-mediated interactions.
  • Multi-qubit gate performance: Hyperfine MS gates reached 99.91% fidelity in 30 µs, while geometric-phase gates reached 99.9% in 100 µs but are not applicable to FOFI qubits.Microwave-mediated two-qubit gates reached 99.7% with AC gradients and 98.5% with static gradients, but both required millisecond timescales.
  • Multi-qubit gate limitations: Two-qubit gate errors remain at least an order of magnitude higher than single-qubit errors and arise from frequency and amplitude drifts, magnetic-field drifts, and off-resonant scattering.Ultrafast entangling gates achieved 2–20 µs operation times but no higher than 76% fidelity, with multiple motional modes becoming relevant above trap frequencies.

4. Gate Characterization: Tomography, Benchmarking, and Calibration

The paper reviews complementary methods for characterizing trapped-ion gate errors, from fidelity and tomography to randomized benchmarking, gate set tomography, and robust phase estimation. These methods differ in which errors they expose and in their experimental overhead and sensitivity to slow drifts.

  • Characterization methods: Fidelity compresses agreement between goal and experimental states into one number but does not fully describe gate errors.For pure states, the paper defines fidelity as F = |⟨ψg|ψe⟩|2.
  • Tomography: Quantum process tomography characterizes an N-qubit operation by determining its effects across 4^N input states.It is presented as a method for fully characterizing a quantum process.
  • Randomized benchmarking: Randomized benchmarking repeats variable-length random gate sequences and uses success probability versus sequence length to amplify gate errors relative to SPaM errors.It extracts stochastic errors relevant to gate fidelity and became a standard characterization tool.
  • Gate set tomography: Gate set tomography applies optimized sequences with exponentially increasing repetition counts to amplify and characterize possible errors across a gate set.Its rigorous bounds and experimental overhead are uncertain in the presence of non-Markovian noise such as slow magnetic-field and laser-frequency drifts.
  • Robust phase estimation: Robust phase estimation extracts a limited set of systematic gate errors in a Heisenberg-limited manner and supports experimental calibration.For single-qubit rotations, relevant parameters include phase and rotation angle.
  • Practical considerations: Trapped ions are well suited to quantum-characterization experiments because small gate errors merit accurate analysis and ion identicality helps calibrations remain useful over long times.Their generally slow gates also make complete characterization methods more time-consuming.

C. State Detection

Trapped-ion state detection uses state-dependent fluorescence and photon-count thresholds to distinguish bright and dark states accurately. Scaling readout to many ions and fault-tolerant operation requires spatial discrimination, faster analysis, improved detectors, and methods that avoid fluorescence-induced decoherence.

  • Fluorescence readout: State-dependent fluorescence distinguishes bright ions that scatter many photons from dark ions that scatter very few.Collected photons are analyzed to infer the ion state.
  • Fluorescence readout: 99.97% bright-state classification and 99.99% dark-state classification follow from a 7-photon threshold when λb = 20 and λd = 1.The threshold separates Poissonian bright- and dark-state photon-count distributions.
  • Scaling to multiple ions: 99% fidelity per ion was achieved for a 53-ion chain using spatially resolved EMCCD readout in 300 µs.A four-ion experiment reached 99.99% fidelity in approximately 400 µs with greater ion spacing.
  • Faster readout: 99.9–99.99% accuracy was achieved with average readout times of hundreds of microseconds using photon-arrival tracking and maximum-likelihood analysis.Approximately 10 µs readout is possible with reduced fidelities around 99%.
  • Decoherence and alternative readout: Fluorescence photons can cause projective measurement of nearby ions, creating decoherence during error-correction measurements performed while other ions remain in superposition.Quantum-logic readout transfers information to an auxiliary species whose far-detuned fluorescence avoids decohering the logical ion.
  • Detection hardware: Superconducting nanowire single-photon detectors have reached efficiencies as high as ∼0.8 for ultraviolet photons from ions.Conventional PMTs, EMCCDs, and APDs typically have detector efficiencies of 0.2–0.4 at relevant wavelengths.

IV. METHODOLOGIES FOR PRACTICAL TRAPPED-ION QUANTUM COMPUTING

Scaling trapped-ion computers requires controlling and measuring many ions while preserving few-ion performance. Modular 1D, 2D, and QCCD architectures address long-chain gate limitations through ion transport, functional zoning, and improved connectivity, but introduce transport and trap-complexity challenges.

  • Scaling requirements: Scaling requires more than trapping additional ions: large systems must control and measure many qubits while maintaining high performance.The paper frames this as the central practical challenge beyond few-ion demonstrations.
  • Linear arrays: Two-qubit gate speed generally decreases as 1D chains grow because collective-mode coupling weakens and ion spacing increases.Larger chains also increase spectral crosstalk and motional heating susceptibility.
  • Modular architectures: Modular ion chains keep modules small enough for high-fidelity, high-speed operations, shifting the scaling problem to moving information between modules.Ion transport and reconfiguration are proposed ways to connect modules.
  • Two-dimensional arrays and QCCD: 2D arrays can distribute information between arbitrary qubits with fewer distance-independent split-and-join operations than reordering or entangling-gate approaches.Ion transport is typically faster and lower-error than a two-ion gate or reordering operation.
  • Two-dimensional arrays and QCCD: 2D architectures can separate memory, interaction, measurement, and loading regions, allowing different array areas to serve different functions.Memory zones can be spatially separated from potential decoherence sources.
  • Two-dimensional arrays and QCCD: Fast transport must limit motional excitation because hotter ion chains significantly degrade multi-qubit gate fidelity.Cooling mitigates excitation but adds time cost, creating a speed–motional-heating trade-off.
  • Two-dimensional arrays and QCCD: 2D transport requires segmented electrodes and junctions, increasing trap complexity, although surface-electrode traps have demonstrated Y and X junctions.Complex 2D array building blocks have also been demonstrated in several geometries.

3. Photonic Interconnects

Photonic interconnects create heralded entanglement between communication ions in separated modules by interfering their emitted photons and detecting coincident outputs. They offer distance-independent connectivity in principle, but generation rates remain limited by collection, coupling, repetition, and cavity-engineering challenges.

  • Remote entanglement: Remote entanglement excites one communication ion per module, interferes the emitted photons, and uses detector coincidences to herald inter-module entanglement.The process can be retried until successful.
  • Remote entanglement: ΓRE = γrep(ηcηd)^2/2 gives the remote-entanglement generation rate under unity excitation probability.The rate depends on repetition rate, photon collection efficiency, and detector efficiency.
  • Optical efficiency: Typical single-system photon collection efficiencies are 0.02–0.04, while coupling collected photons into a single-mode fiber for remote entanglement is typically ∼0.1 efficient.Mode matching at the beamsplitter adds an additional requirement beyond photon collection.
  • Rate limitations: ∼5 Hz is the highest-achieved trapped-ion remote-entanglement generation rate using standard methods, limiting practical quantum-computer operation speed.Spatial multiplexing is proposed to make multiple remote-entanglement attempts in parallel.
  • Cavity-enhanced collection: Ion-cavity coupling could increase photon collection, but high-finesse blue-to-UV cavities remain difficult because of mirror reflectivity, degradation in ultrahigh vacuum, and dielectric charging.These difficulties motivate continued research because of potential impacts on interconnects and state measurement.
  • Architectural role: Photonic interconnects can connect modules within one trap or separate vacuum chambers, with practical connection speed largely independent of physical distance.Scaling to many modules may require a many-port optical switch.

1. Decoherence-Free Subspaces and Composite-Pulse Control

The paper reviews strategies for reducing trapped-ion errors and scaling toward fault-tolerant computation, including decoherence-resistant encoding, dynamical decoupling, error correction, and dual-species architectures.

  • Decoherence-Free Subspaces: Decoherence-free subspaces encode virtual qubits across multiple physical qubits to suppress shared decoherence sources.Their effectiveness relies on nearby qubits experiencing similar noise.
  • Decoherence-Free Subspaces: A two-ion 9Be+ implementation improved coherence time by approximately 50× under intentionally applied fluctuating AC-Stark-shift noise.The comparison was against a single-qubit state under the same applied noise.
  • Decoherence-Free Subspaces: Universal quantum computation has been shown theoretically for decoherence-free-subspace qubits and experimentally through a universal gate set on 40Ca+ ions.The experiments demonstrated gates acting within the encoded subspace.
  • Composite-Pulse Control: Dynamical decoupling has enabled trapped-ion two-qubit gates with a current best fidelity of 99.9%.The technique counteracts dephasing during the gate.
  • Error Correction: Quantum error correction remains necessary for algorithms requiring roughly 10^10 operations, because accumulated physical errors remain too large for useful high-fidelity execution.Reducing physical error rates also reduces the resource overhead needed for a target logical error rate.
  • Dual-Species Ion Systems: Dual-species ions mitigate decoherence from measurement, remote entanglement, and sympathetic cooling because scattered photons are far-detuned from the other species.Experiments demonstrated dual-species gates, error-correction primitives, GHZ states, and photonic interconnect primitives, with reported fidelities ranging from 60% to above 86%.

V. INTEGRATED TECHNOLOGY FOR CONTROL OF TRAPPED IONS

Scaling trapped-ion computers requires hardware that controls and measures many ions while balancing scalability against performance. Chip-scale surface-electrode traps are highlighted as a flexible platform for modular fabrication and integrated control technologies.

  • Integrated Technology: Practical trapped-ion computers will require low-error hardware for controlling and measuring large ion populations.The paper emphasizes exploring tradeoffs between scalability and performance even in small systems.
  • Chip-Scale Ion Traps: Surface-electrode traps use patterned metal electrodes on substrates such as sapphire, quartz, or silicon to confine ions above the chip.Microfabrication supports arbitrary electrode patterns and multiple metal layers for signal routing.
  • Chip-Scale Ion Traps: Surface-electrode traps can be fabricated on wafers up to 12 inches in diameter, while also supporting modular construction by tiling smaller chips.The paper presents wafer-scale and tiled-chip approaches as distinct scaling options.
  • Chip-Scale Ion Traps: Chip-scale traps provide a platform for integrating scalable ion control, measurement, classical signal processing, and computing beneath the electrodes.This integration is identified as a potentially transformative advantage of the architecture.

B. Integrated Photonics for Light Delivery

Integrated photonics is proposed to deliver many precisely positioned laser beams to surface-electrode ion arrays with low crosstalk and stable registration. Its scalability is tempered by wavelength-dependent device design, optical loss, and detector-integration challenges.

  • Motivation: The number of laser beams requiring precise delivery grows with the size of an ion array and roughly doubles for dual-species systems.Trapped-ion control typically requires about five wavelengths before accounting for the dual-species increase.
  • Integrated Photonics: Integrated waveguides route light on chip through a high-index core surrounded by lower-index cladding, using fabrication methods compatible with surface-electrode traps.Waveguide geometry can support single-mode propagation and polarization maintenance.
  • Integrated Photonics: Vertical grating couplers diffract waveguide light out of the chip toward ions through openings in the trap electrodes.Emission angle depends on grating period, material indices, and wavelength.
  • Integrated Photonics: Integrated waveguides and grating couplers could deliver high-intensity light to 2D ion arrays with low crosstalk and little free-space propagation.Lithographic registration may improve beam-position and phase stability relative to free-space optics.
  • Challenges: Different wavelengths may require distinct waveguides and grating couplers because device bandwidth, modal behavior, and emission angle are wavelength-dependent.This complicates multi-wavelength delivery for full ion control.
  • Challenges: Rayleigh-scattering loss increases as 1/λ^4, making low-loss operation especially difficult at near-UV wavelengths.Suitable materials must combine UV transparency with low fabrication roughness.
  • Challenges: Integrated waveguides inherently add loss relative to free-space optics, although efficient grating couplers and tight focusing may recover useful intensity.Input and output coupling contribute to the loss budget.
  • Integrated Detection: Integrated detectors and collection optics are being pursued for high-speed, high-fidelity ion measurement that is less susceptible to stray light.SNSPDs offer higher detection efficiency and lower dark counts, while APDs offer room-temperature CMOS-compatible operation; integrated fluorescence readout has not yet been demonstrated.

D. Integrated Electronics

Integrated electronics could improve the scalability and responsiveness of trapped-ion systems, but their effects on heating, power dissipation, magnetic fields, and decoherence remain to be established. Scaling also requires choices about ion species and control wavelengths that balance trapping, cooling, detection, and gate-performance tradeoffs.

  • Integrated Electronics: Ion-motion control increasingly requires many remotely generated electrode voltages, motivating integration to reduce wiring complexity and noise susceptibility.Digital-to-analog converters are typically housed in remote electronics racks, with approximately one converter per electrode.
  • Integrated Electronics: 80 commercially available DAC channels have been connected in vacuum to a surface-electrode trap through printed-circuit-board traces, with Ca+ trapping and transport demonstrated.
  • Integrated Electronics: Monolithic CMOS integration has enabled stable Sr+ trapping and a surface-electrode trap with 16 integrated DAC channels fabricated in a 180-nm process.
  • Integrated Electronics: Integrated electronics could support detector pulse processing, on-chip analog and digital processing, reduced error-correction latency, signal routing, and magnetic-field generation.These possibilities depend on demonstrating that integrated circuits do not introduce significant deleterious effects.
  • Choice of Ion Species: Ion-species selection creates tradeoffs because mass affects trap voltages, power dissipation, optical-force strength, sympathetic cooling, and two-qubit gate speed.Heavier ions require larger voltages for comparable secular frequencies and generally have slower motion-based two-qubit gates at comparable optical power.
  • Choice of Ion Species: Control wavelength and electronic structure affect detection efficiency, available qubit transitions, scattering channels, repumping requirements, and the number of required laser wavelengths.Blue and near-UV light generally supports high detection efficiency, while metastable D or F levels enable optical qubits but introduce additional scattering and repumping considerations.

2. Choice of Qubit and Gate Type

Trapped-ion QC platforms must choose between hyperfine Raman qubits and optical qubits, balancing memory coherence, state preparation, detection, wavelengths, gate speed, power, and scattering errors. The preferred gate type depends on ion species and the desired speed–power tradeoff, with additional limitations from recoil and scattering.

  • Qubit Choices: Hyperfine Raman qubits offer longer decay and coherence times, whereas optical qubits provide simpler state preparation and detection with more favorable laser wavelengths.
  • Single-Qubit Gates: Longer-wavelength transitions and smaller beams reduce required single-qubit optical power, supporting integrated photonic technologies for scalable designs.The comparison applies to both Raman and direct optical single-qubit excitation under the stated error and beam assumptions.
  • Single-Qubit Gates: At longer gate times direct optical gates require less power, while at shorter gate times Raman gates require less power; the crossover depends on ion species and target error.
  • Two-Qubit Gates: For two-qubit gates, motion excitation reduces the Rabi frequency by the Lamb-Dicke parameter η, increasing the required power by 1/η2 relative to comparable single-qubit Raman gates.The same 1/η2 power increase applies to direct optical two-qubit gates because their Rabi frequency is proportional to optical electric-field amplitude.
  • Two-Qubit Gates: Heavy ions require proportionally more power for two-qubit gates, and the crossover gate time increases by a factor related to η2 and the quadratic gate-time dependence.
  • Error Mechanisms: The gate comparisons include spontaneous-emission errors but exclude recoil, which can significantly affect lighter ions in two-qubit gates.Low-lying D levels also impose a Raman-gate error floor near 10−4 for considered ions, except Yb+, where the limit is a couple orders smaller.
  • Alternative Gates: Magnetic-field-gradient gates avoid photon-scattering errors but require faster gates and solutions to addressing and power-dissipation challenges to exceed analyzed laser-based fidelities.

3. Choice of System Temperature

Trapped-ion system design involves balancing temperature-related noise, gate speed, optical and electrical power, qubit lifetime, and species compatibility. The review identifies ion and qubit choices that trade these constraints differently for scalable and portable systems.

  • Temperature and motional heating: Room-temperature traps can support quantum operations because laser cooling reduces ion motion while internal electronic qubits remain effectively isolated from trap heat.Doppler cooling reaches approximately 1 mK and resolved-sideband cooling approximately 20 µK, although scalable systems still face temperature-related electrode and vacuum constraints.
  • Temperature and motional heating: Higher trap frequencies reduce motional-heating-induced two-qubit gate errors because measured heating scales approximately as ω^-2 to ω^-2.5.The heating rate depends on electric-field noise, ion charge and mass, and trap frequency; gate error is directly proportional to heating for sufficiently slow heating.
  • Gate implementation trade-offs: For gate durations near 10^-5 s, optical gates can reduce optical-power requirements, whereas Raman gates require less power when two-qubit gates reach the microsecond scale or below.Very fast-gate extrapolations may fail because operation outside the Lamb-Dicke regime and off-resonant motional excitation introduce additional nonidealities.
  • Gate implementation trade-offs: Optical gates are advantageous for integrated photonics because their longer red and infrared wavelengths generally experience lower waveguide scattering loss than blue and ultraviolet Raman wavelengths.This supports parallel operations over large arrays, while Raman implementations may require high optical power at wavelengths with higher loss.
  • Qubit and ion selection: Hyperfine qubits are preferred when low memory error is paramount, while optical and fine-structure qubits remain limited by metastable-state lifetimes.Zeeman qubits are a potential alternative when sufficient magnetic shielding is available.
  • Qubit and ion selection: Medium-weight ions offer a power compromise for portable applications, balancing the high voltages and optical powers of heavy ions against the ultraviolet wavelengths of light ions.Ca+ is presented as a flexible general-purpose choice, while closer-mass dual-species pairs can be preferable for sympathetic cooling or ancilla functions.

C. Future Experiments to Enable Practical Trapped-Ion Quantum Computers

The review identifies experiments needed to assess whether trapped-ion systems can become practical at larger scale. Priorities include mitigating motional heating, developing faster and more robust control, characterizing noise, demonstrating fault-tolerant correction, and testing integrated hardware and architectures.

  • Understanding and mitigating anomalous motional heating: Motional heating can limit multi-ion gate fidelities and may become dominant as quantum-control precision improves unless electric-field noise is understood and mitigated.The review highlights anomalous motional heating as a key target for future experiments.
  • Development of new techniques for robust ion control: High-fidelity ion control has been demonstrated only in few-ion systems and remains insufficient to eliminate quantum-error-correction overhead during scaling.The review therefore calls for techniques that maintain robust control as system size increases by orders of magnitude.
  • Techniques for faster gates: Two-qubit gates are a principal speed bottleneck, motivating fast, high-fidelity control methods that do not depend solely on high control-field intensity.Candidate approaches may operate outside the Lamb-Dicke regime, where off-resonant motional and carrier excitation must be addressed.
  • Noise characterization: Noise characterization is necessary because the effectiveness of error-mitigation and error-correction strategies depends sensitively on whether their assumed noise models match trapped-ion noise.The review notes that memory and control noise currently limits coherence times and gate fidelities.
  • Fault-tolerant error correction: A fault-tolerant logical qubit must be demonstrated to assess practical prospects, using physical ion qubits to detect and correct realistic errors while reducing logical error rates.Error correction introduces substantial qubit-count and gate-count overhead through logical encoding.
  • Integrated hardware: Integrated control and measurement hardware must be designed, built, and tested in small trapped-ion systems because integration may improve performance but could also introduce new problems.These experiments are intended to determine the long-term potential of integrated approaches for scaling.
  • Experiments to inform architectural analysis: Architecture experiments are needed because insufficient data currently determine which architectural primitives, parameters, or combinations are best for practical trapped-ion systems.For linear ion chains, longer chains may slow and degrade two-qubit gates but reduce the number of split/join and transport operations, implying a potentially optimal module length.
Loading 1904.04178v1…