hardware.co · AI hardware for on-prem LLM deploymentsVendor-neutral · Quote onlyBuilt to spec
ContactGet a quote →
GATE-BASED · DIGITAL

Trapped Ion Quantum Computers

Trapped-ion quantum computers confine individual charged atoms in electromagnetic fields and use laser or microwave pulses to perform quantum operations. Ions serve as near-perfect qubits: they are identical by nature, have extremely long coherence times, and offer native all-to-all connectivity within a trap. IonQ and Quantinuum lead commercial deployment. The primary limitation is gate speed — trapped-ion gates are 10–100× slower than superconducting gates.

GATE SPEED

Single-qubit gates: ~1–10 µs

CONNECTIVITY

Natively all-to-all within a trap chain…

ENVIRONMENT

Ultra-high vacuum (~10⁻¹¹ torr)

MS gatei=2i=7LINEAR PAUL TRAP — ALL-TO-ALL CONNECTIVITY

How Trapped Ion Qubits Work

Trapped-ion quantum computers use individual atomic ions as qubits. Ions are electrically charged atoms and are confined in electromagnetic traps — most commonly a linear Paul trap, which uses oscillating radiofrequency electric fields to create a pseudopotential well. The ions form a linear crystal held in place by the balance between the trapping potential and their mutual Coulomb repulsion. IonQ uses Ytterbium-171 ions; Quantinuum uses Barium-133 or Ytterbium-171.

Quantum information is encoded in two internal electronic states of each ion, typically hyperfine clock states that are extremely stable against magnetic field fluctuations. These states serve as |0⟩ and |1⟩ with coherence times often exceeding one second — orders of magnitude longer than superconducting qubits. Single-qubit gates are implemented by addressing individual ions with focused laser beams or microwave fields tuned to the qubit transition frequency.

Two-qubit gates exploit the shared motional modes of the ion crystal. When a laser is applied to ion A in a way that imparts momentum, the resulting phonon (vibration) propagates through the crystal and creates a correlated interaction with ion B. The Mølmer–Sørensen gate and related protocols use this phonon bus to generate maximally entangling two-qubit gates between any pair of ions in the chain — providing native all-to-all connectivity within the trap. This is a fundamental advantage over nearest-neighbor architectures.

Physical Implementation

Qubit Medium

Individual atomic ions (Yb-171 or Ba-137) suspended in linear Paul traps at ultra-high vacuum; hyperfine ground states as qubit states

Control Mechanism

Individually addressed laser pulses or microwave fields for single-qubit gates; Mølmer–Sørensen or similar phonon-mediated interactions for two-qubit gates

Operating Environment

Ultra-high vacuum (~10⁻¹¹ torr); trap housing at room temperature or mild cryogenic temperature (4K for some designs); no dilution refrigerator required

Readout Mechanism

State-dependent fluorescence: a readout laser causes |1⟩-state ions to scatter many photons (bright); |0⟩-state ions remain dark. Detected by CCD cameras or photomultiplier tubes. Readout fidelity exceeds 99.9% in leading systems.

Connectivity and Gate Operations

Connectivity

Native all-to-all connectivity within a single ion chain. Any two ions in the trap can be entangled directly via the shared phonon bus without SWAP overhead. This eliminates a major source of gate overhead common in nearest-neighbor architectures. Modular scaling via photonic interconnects between separate chains is in development.

Native Gate Set

Single-qubit: arbitrary SU(2) rotation (Rx, Ry, Rz). Two-qubit: Mølmer–Sørensen (XX or MS gate), which is equivalent to an all-to-all maximally entangling operation. Circuit compilers decompose standard gate sets to these native operations. CNOT and CZ are derived from the MS gate.

Gate Speed

Single-qubit gates: ~1–10 µs; two-qubit gates: ~100 µs–1 ms (significantly slower than superconducting)

ALL-TO-ALL

Potential Advantages

  • ✓Highest published two-qubit gate fidelities of any commercial platform (~99.9%+ for Quantinuum H2)
  • ✓Native all-to-all connectivity eliminates SWAP overhead; any two qubits can be directly entangled
  • ✓Long coherence times (T1/T2 of seconds to minutes) enable very deep circuits
  • ✓Ions are physically identical — no frequency crowding, manufacturing variation, or drift between units
  • ✓Demonstrated algorithmic qubit (AQ) metrics: IonQ Forte reported 35 AQ, Quantinuum H2 up to 56 AQ
  • ✓Photon-based quantum networking naturally compatible with ion-photon entanglement protocols

Limitations & Challenges

  • —Gate speeds are significantly slower than superconducting: two-qubit gates take 100 µs–1 ms vs. ~100 ns for superconducting
  • —Chain length is limited by heating from background electric field noise and mode crowding; current practical limit ~20–50 ions per chain
  • —Scaling to hundreds of chains requires photonic interconnects between modules — a technically challenging unsolved problem
  • —Optical and laser control systems are complex, large, and expensive; miniaturization is an active engineering challenge
  • —Circuit throughput (circuits per second) is substantially lower than superconducting due to slow gates and ion cooling overhead
  • —No commercially demonstrated path to the millions of physical qubits needed for large fault-tolerant computations

Scaling Considerations

The dominant scaling path is modular: multiple short ion chains (10–50 ions each) are connected via photon-mediated quantum links, building up a quantum network of traps. IonQ's "Barium" roadmap and Quantinuum's multi-zone architectures use this approach. Some designs use integrated photonics to route light to individual ions in a chip-scale trap. The key challenge is generating high-rate, high-fidelity entanglement between distant modules via photonic links. Quantum error correction on trapped-ion platforms has been demonstrated with high-fidelity gates; the challenge is throughput — QEC requires many fast sequential operations.

Coherence: T1 and T2 coherence times measured in seconds to minutes — orders of magnitude longer than superconducting qubits. Error rates are consistently the lowest across all commercial platforms.

Available QPUs

View all →

IonQ Forte

Cloud Access
Trapped Ion

Qubits

32

2Q Fidelity

99.50%

aws-braketazure-quantumionq-cloud
View →

IonQ Forte Enterprise

Private Access
Trapped Ion

Qubits

35

#AQ

35

ionq-cloud
View →

Quantinuum H2-1

Reservation
Trapped Ion

Qubits

56

2Q Fidelity

99.90%

azure-quantumquantinuum-nexus
View →

Leading Companies

IonQ→Quantinuum→Oxford IonicsUniversal Quantum

Workload Suitability

Full guide →
Quantum ChemistryStrong Fit

Simulating molecular electronic structure to compute ground-state energies, reac…

Materials Science & DiscoveryStrong Fit

Simulating solid-state physics phenomena including high-temperature superconduct…

Combinatorial OptimizationPossible

Solving NP-hard combinatorial problems including vehicle routing, supply chain s…

Financial ModelingPossible

Monte Carlo simulation acceleration via quantum amplitude estimation, derivative…

Quantum Machine LearningPossible

Training and inference on quantum neural networks (QNNs), kernel methods using q…

Cryptanalysis & Post-Quantum CryptographyExperimental

Shor's algorithm can factor large integers and solve discrete logarithms exponen…

WELL SUITED FOR

  • ›Deep quantum chemistry circuits requiring high gate fidelity
  • ›Quantum error correction code demonstrations (high fidelity gates reduce overhead)
  • ›Algorithms requiring all-to-all connectivity (quantum simulation, portfolio optimization)
  • ›Precision metrology and sensing applications
  • ›Quantum network node applications leveraging ion-photon entanglement

LESS SUITED FOR

  • —High-throughput sampling requiring many circuit executions per second (slow gate speed bottleneck)
  • —Algorithms requiring millions of physical qubits in the near term
  • —Applications where classical simulation speed is the constraint rather than qubit count

Frequently Asked Questions

How do laser beams entangle ions that are micrometers apart?›

Ions in a trap share collective vibrational modes (phonons). A laser tuned slightly off resonance from the qubit transition drives "sideband" interactions that couple the qubit state to these vibrations. By carefully designing the laser pulse, the phonon bus mediates a two-qubit phase operation that entangles any two ions in the chain — regardless of their distance from each other.

What does "algorithmic qubit" or #AQ mean?›

The Algorithmic Qubit (#AQ) metric, developed by IonQ, measures the number of qubits that can run a fully connected random circuit at above-threshold fidelity. It captures both qubit count and gate fidelity in a single number. #AQ 35 means 35 qubits executing a fully entangled circuit with meaningful output fidelity — more useful than raw qubit count alone.

Why can't trapped-ion systems just add more ions to scale up?›

As the ion chain grows longer, the number of vibrational modes grows with it, and their frequencies crowd together. This makes it increasingly difficult to address a single motional mode without off-resonantly exciting neighboring modes. Background electric field noise also heats the ions proportionally. Beyond roughly 50–100 ions, the two-qubit gate fidelity degrades significantly, making longer single chains impractical.

Are trapped-ion computers available in the cloud?›

Yes. IonQ systems are accessible via Amazon Braket, Microsoft Azure Quantum, and IonQ Cloud. Quantinuum H-Series systems are available via Microsoft Azure Quantum and Quantinuum's own Nexus portal.

How do trapped-ion systems compare to superconducting for fault-tolerant computing?›

Trapped-ion gates have much higher fidelity, which reduces the physical qubit overhead for error correction. However, trapped-ion gates are 1000× slower, limiting the number of error correction rounds per second. Both architectures are pursuing fault-tolerant operation; the trade-off between fidelity and speed will determine which achieves useful fault tolerance first.

Key References