How do superconducting qubits and transmons work?

What is a superconducting qubit?

An arXiv review explains that a qubit is a physical system whose selected quantum states encode information [6]. In a superconducting circuit, those states belong to an engineered electrical structure rather than to an individual atom, according to the review [6]. Research in npj Quantum Information describes transmons built from superconducting films, a capacitor, and a Josephson junction [2]. The arXiv review explains that the circuit has a ground state and excited states, with the lowest pair conventionally labeled zero and one [6]. As that review describes, a prepared superposition combines those states with a measurable relative phase [6].

The arXiv review explains that measuring a qubit produces a classical outcome, unlike the full quantum state before measurement [6]. According to that review, computation depends on preparing states, applying controlled operations, and measuring outcomes across repeated runs [6]. The npj Quantum Information study describes coupling transmons to microwave drive and readout circuits [2]. Its devices are fabricated objects, and the study compares how material and geometry changes affect their behavior [2]. The general idea of quantum information is distinct from the practical question of whether a particular chip can preserve it long enough, as the review's discussion of noise makes clear [6].

Why does a Josephson junction matter?

A Josephson junction connects superconductors across a thin barrier, according to the arXiv review [6]. In the junction construction described by npj Quantum Information, aluminum electrodes are separated by aluminum oxide [2]. The junction's nonlinear response changes the circuit's energy-level spacings, as the original transmon analysis explains [1]. Unlike an ideal harmonic oscillator with evenly spaced energy levels, a transmon has uneven spacings that allow a pulse to address the chosen low-energy transition more selectively [1].

The transmon analysis explains that a real transmon has more than the pair of levels used as a computational qubit [1]. The original arXiv paper describes retaining sufficient anharmonicity for selective control while suppressing sensitivity to charge fluctuations [1]. As the arXiv review notes, a control pulse can still populate an unwanted higher level, known as leakage [6]. The junction supplies the useful nonlinearity, but the whole circuit and its control pulses determine how well that nonlinearity can be used, according to the review [6].

What makes a transmon qubit different?

The transmon is a design derived from the Cooper pair box, as the original arXiv analysis explains [1]. Its large shunting capacitance lowers charging energy relative to Josephson energy, reducing the circuit's response to unwanted changes in offset charge [1]. The analysis predicts an exponential reduction in charge dispersion as that energy ratio grows, alongside a much slower reduction in anharmonicity [1]. As the authors explain, the design trades some level separation for substantially less charge sensitivity [1].

The trade is not a guarantee that all noise disappears, as the original arXiv paper discusses multiple noise channels [1]. Research in npj Quantum Information identifies material interfaces as important loss channels in its devices [2]. npj Quantum Information describes fabricating transmons, coupling them to readout circuitry, and wire-bonding the chips into sample boxes [2]. The arXiv review therefore considers both the ideal energy levels and the ways a physical device exchanges energy with its surroundings [6]. Transmon describes a circuit family rather than a single material or fixed performance specification, as the different materials in npj Quantum Information illustrate [2].

How do quantum computers control and read transmons?

According to the arXiv review, microwave pulses drive transitions and change the state of a transmon [6]. EPJ Quantum Technology describes drive lines for control and readout resonators for measurement [3]. Coupling between circuit elements also provides a way to implement operations involving multiple qubits, as demonstrated in a Nature processor experiment [7]. The arXiv review describes calibrating pulses in amplitude, duration, and frequency for the device, and measuring the resulting operation accuracy [6].

Reading a state requires coupling the qubit to a measurement system, which can itself influence the circuit, according to the arXiv review [6]. The npj Quantum Information experiment used transmons coupled to readout resonators and compared their relaxation under different conditions [2]. The review discusses the tension between isolating a qubit to preserve its state and providing controlled access to it [6]. The separate cryogenic wiring article examines heat and cabling tradeoffs, while this explanation focuses on the chip and its quantum states.

Which superconducting materials are used?

Aluminum and niobium are established choices for circuit films, and tantalum and titanium nitride have also been investigated, as examples in npj Quantum Information and arXiv show [2][4]. In npj Quantum Information, researchers fabricated transmons with tantalum, niobium, and aluminum base films using related designs and compared their coherence [2]. Nature Communications describes aluminum and aluminum oxide junctions on tantalum base films, so a tantalum transmon need not have a junction made entirely of tantalum [5].

A study posted on arXiv used titanium nitride for the shunting capacitors and reported relaxation and dephasing times up to approximately 60 microseconds in its devices [4]. The researchers linked their comparison with aluminum capacitors to evidence for interface-related loss [4]. As the Nature Communications experiment shows, comparisons depend on fabrication and geometry, not only the name of the superconducting material [5]. Its authors report that surface preparation and fabrication changes affected the resulting circuits [5].

Comparing results also requires attention to the measurement protocol, as the different procedures in npj Quantum Information illustrate [2]. The experiment reports that relaxation times varied across chips and fluctuated over time on the same chip [2]. A peak result cannot represent every qubit, and a relaxation measurement is not interchangeable with a gate-error measurement, as the arXiv review explains [6]. The npj Quantum Information experiment also reported lower coherence on its larger chip than on its single-qubit samples, illustrating the distinction between an isolated device result and system-level performance [2].

What limits qubit coherence and why are chips cold?

The arXiv review describes coherence in terms of the persistence of quantum states needed for an operation [6]. As that review explains, relaxation time measures loss of excitation, while dephasing describes loss of phase information [6]. In npj Quantum Information, one tantalum device reached a best measured relaxation time of 503 microseconds, with fluctuations and lower values in other devices [2]. That best measurement is a result for a particular experiment, not a lifetime promised by tantalum itself, as the paper's range of results shows [2]. According to the arXiv review, coherence is only one contributor to operation quality [6].

Microscopic defects can act as two-level systems and interact with the qubit's electric field, according to the arXiv review [6]. The titanium nitride study on arXiv presents evidence for such defects at or near interfaces [4]. EPJ Quantum Technology explains that dilution refrigerators reach millikelvin temperatures using helium isotopes, helping initialize qubits in their ground states and suppress thermal excitation [3]. Nature Communications reports measurements between 9 and 20 millikelvin for its tantalum devices [5]. Cooling alone does not remove every loss mechanism, as the material-interface studies in npj Quantum Information demonstrate [2].