Modular mixed-species ion-trap networks are a promising framework for scalable quantum information processing, where one species acts as a memory qubit and another as a communication qubit. This architecture requires high-fidelity mixed-species entangling gates to transfer information from communication to memory qubits through their collective motion. We investigate the character of the motional modes of mixed-species ion chains of various lengths and in various trap potentials for entangling operations and sympathetic cooling. We find that the laser power required for high-fidelity entangling gates based on transverse modes is at least an order of magnitude higher than that for gates based on axial modes for ion species with a large mass ratio. We also find that for even moderate mass differences, the transverse modes are much harder to cool than the axial modes regardless of the ion chain configuration. Therefore, transverse modes conventionally used for operations in single-species ion chains may not be well suited for mixed-species chains with widely different masses.
Modular mixed-species ion-trap networks are a promising framework for scalable quantum information processing, where one species acts as a memory qubit and another as a communication qubit. This architecture requires high-fidelity mixed-species entangling gates to transfer information from communication to memory qubits through their collective motion. We investigate the character of the motional modes of mixed-species ion chains of various lengths and in various trap potentials for entangling operations and sympathetic cooling. We find that the laser power required for high-fidelity entangling gates based on transverse modes is at least an order of magnitude higher than that for gates based on axial modes for ion species with a large mass ratio. We also find that for even moderate mass differences, the transverse modes are much harder to cool than the axial modes regardless of the ion chain configuration. Therefore, transverse modes conventionally used for operations in single-species ion chains may not be well suited for mixed-species chains with widely different masses.
Quantum computing promises solutions to some of the world's most important problems that classical computers have failed to address. The trapped-ion-based quantum computing platform has a lot of advantages for doing so: ions are perfectly identical and near-perfectly isolated, feature long coherent times, and allow high-fidelity individual laser-controlled operations. One of the greatest remaining obstacles in trapped-ion-based quantum computing is the issue of scalability. The approach that we take to address this issue is a modular architecture: separate ion traps, each with a manageable number of ions, are interconnected via photonic links. To avoid photon-generated crosstalk between qubits and utilize advantages of different kinds of ions for each role, we use two distinct species - ¹⁷¹Yb⁺ as memory qubits and ¹³⁸Ba⁺ as communication qubits. The qubits based on ¹⁷¹Yb⁺ are defined within the hyperfine "clock" states characterized by a very long coherence time while ¹³⁸Ba⁺ ions feature visible-range wavelength emission lines. Current optical and fiber technologies are more efficient in this range than at shorter wavelengths. We present a theoretical description and experimental demonstration of the key elements of a quantum network based on the mixed-species paradigm. The first one is entanglement between an atomic qubit and the polarization degree of freedom of a pure single photon. We observe a value of the second-order correlation function g⁽²⁾(0) = (8.1 ± 2.3)⨉10⁻⁵ without background subtraction, which is consistent with the lowest reported value in any system. Next, we show mixed-species entangling gates with two ions using the Mølmer-Sørensen and Cirac-Zoller protocols. Finally, we theoretically generalize mixed-species entangling gates to long ion chains and characterize the roles of normal modes there. In addition, we explore sympathetic cooling efficiency in such mixed-species crystals. Besides these developments, we demonstrate new techniques for manipulating states within the D₃⸝₂-manifold of zero-nuclear-spin ions - a part of a protected qubit scheme promising seconds-long coherence times proposed by Aharon et al. in 2013. As a next step, we provide a detailed description of the protocols for three- and four-node networks with mixed species, along with a novel design for the third trap with in-vacuum optics to optimize light collection.
We describe our work on trapping, cooling and detecting mixed ion species for a scalable ion trap quantum information processing architecture. These mixed species chains in linear RF traps may help solve several problems with scaling ion trap quantum computation to large numbers of qubits. Initial temperature measurements of linear Coulomb crystals containing barium and ytterbium ions indicate that the mass difference does not significantly impede sympathetic cooling of normal modes that couple well to the coolant ions (Ba in our case). Average motional occupation numbers are estimated to be 10 to 20 quanta per mode for these well cooled modes for chains with small numbers of ions, consistent with the Doppler limit temperature. For normal modes that do not couple significantly to the coolant atoms, the occupation numbers are significantly higher, of order several thousand. Strategies for better cooling of these modes are discussed. Further, we are working to implement these techniques in microfabricated surface traps in order to exercise greater control over ion chain ordering and positioning.
Owing to their high controllability and connectivity, one-dimensional ion chains are attractive building blocks for near-term quantum computers. However, long ion chains are susceptible to motional heating caused by fluctuating external electric fields. To enable deep computations, this source of noise must be suppressed without inducing qubit decoherence or degrading qubit connectivity. We demonstrate steady-state mid-circuit sympathetic cooling of all computationally-relevant motional modes of a 23-ion chain consisting of \qubit qubit and \coolant coolant ions without using any qubit operations. In a room-temperature system, our cooling scheme preserves the mean phonon occupations of the long-wavelength axial and radial modes near their values following state preparation, while keeping the modes used to implement entangling gates near their ground states throughout a 56-ms circuit. Crucially, we show that our cooling sequence preserves qubit coherence, and evaluate its impact on single- and two-qubit gate operations. Additionally, we demonstrate a parallel qubit reset protocol leveraging the shared radial mode coupling between species. Our sympathetic cooling scheme represents a critical step toward reducing gate errors in individually-addressed long ion chains and establishes a versatile platform for simulating open quantum systems.
A universal quantum computer is an ultimate pursuit, which should possess the ability to perform entangling gate operations in parallel with high quality in a scalable fashion. However, for a long trapped-ion chain, the simultaneous implementation of two-qubit quantum gates is a challenging task due to the drastic increase of the crosstalk resulting from the collective motional modes. In this paper, by utilizing the structure of the localized motional modes in the long mixed-species ion chain, we propose a highly efficient linear iterative scheme to eliminate the crosstalk within the neighboring pairs. We show that this scheme allows the realization of high-fidelity, low-crosstalk, scalable, parallel two-qubit gates with only a few pulse segments for the long chain. In addition, by considering the practical experimental feasibility, we optimize the laser power and improve the robustness against the random static drifts of the motional frequencies.
Abstract Trapped-ion systems are one of the leading platforms for quantum information processing, where a key challenge is to scale up system size while maintaining high-fidelity two-qubit operations. A promising approach is to build high-performance modules interconnected via strong coupling. In particular, axial motional modes offer a practical mechanism to couple the ions in a chain, enabling the preparation of Greenberger–Horne–Zeilinger states with up to 24 ions using global operations, as well as high-fidelity two-qubit gates (96.6%–98.0%) in fully connected five-ion chains. Here, we demonstrate two-qubit quantum logic gates in a 5-ion 40 Ca + chain using axial modes, achieving fidelities exceeding 99% for adjacent pairs and over 98% for arbitrary pairs by carefully tackling dominant error sources. Our results are beneficial to the development of scalable ion-trap quantum processors, quantum simulation and quantum-enhanced metrology.
Harnessing quantum mechanics to solve problems intractable on classical computers is one of the foremost goals of applied physics. Trapped ions make an excellent physical platform for such a quantum computer, as they have demonstrated unparalleled qubit operation fidelities and memory times, but significant challenges remain in scaling up these systems. A promising scalable architecture is a photonically connected network of mixed element trapped ion nodes, each containing a few ions of two species. Poor quality remote entanglement links between nodes can be tolerated, if entanglement between ions in the same node can be achieved with high fidelity. While local gates have been repeatedly demonstrated between ions of the same species, the fidelity of mixed element quantum gates have not previously surpassed the fault tolerant threshold, above which errors can in principle be corrected. In this thesis we demonstrate an entangling operation between 43Ca+, useful for its long lived qubits, and 88Sr+, well suited for remote entanglement, using a novel mechanism requiring only one laser to manipulate both species. This relatively simple scheme, which is independent of either qubit frequency, allows us to achieve Bell state fidelities of F = 99.8(1)%. We assess the operation of the gate further with state and process tomography, which achieve Bell state fidelity F = 98.1% and average gate fidelity F̅ = 99.05(6)% respectively, though they are hampered by experimental drifts during the long data collection. Finally, we use interleaved randomised benchmarking to get the best estimate of our gate errors, finding an average gate fidelity of F̅ = 99.62(3)% for a sequence of 60 interleaved gates. Along with a demonstration of entanglement between two 88Sr+ ions, achieved with fidelity F = 95.9(4)%, this work paves the way towards demonstrating an elementary mixed species quantum network in future experiments.