Shimadzu.com
The world’s most accurate atomic clocks have a structural flaw that their precision figures never mention: they blink. Every optical lattice clock operating in every metrology laboratory on earth spends a substantial fraction of its time doing nothing at all — loading atoms, cooling them, preparing their quantum states — while the clock laser it is supposed to be stabilizing drifts unmonitored and uncorrected. The clock measures time with extraordinary accuracy during the measurement window. Outside that window, it is just an expensive laser sitting in a vacuum chamber. The phenomenon has a formal name — dead time in atomic clocks — and it has real consequences for clock performance.
A research team from the University of Tokyo’s Graduate School of Engineering, RIKEN’s Center for Advanced Photonics, and JEOL Ltd. published a demonstration Wednesday that eliminates this gap. In a paper published in Nature Communications, researchers Koki Nishida, Ryoto Takeuchi, Shigenori Tsuji, Shoichi Okaba, and Professor Hidetoshi Katori showed that a continuous stream of laser-cooled strontium atoms — carried along a moving “atomic conveyor belt” — can be interrogated with full clock-transition precision while atoms are simultaneously being prepared upstream and read out downstream. The result: a 1.2 Hz spectral linewidth approaching the theoretical Fourier limit, with the clock laser never once left without atoms to monitor.
Dead Time Was Not an Oversight — It Was Unavoidable Until Now
The measurement gap in optical lattice clocks has a formal name — dead time — and a well-documented effect on performance: the Dick effect. Named for G. John Dick, who identified it in 1987, the Dick effect describes how the frequency noise of the clock laser aliases into the clock’s output during the periodic gaps between interrogations. The more dead time per cycle, the worse the aliasing and the more the clock’s stability degrades from its theoretical ceiling.
The problem is not small. As Katori himself wrote in a 2011 review in Nature Photonics, the Dick effect had become “the major obstacle in achieving higher stability” in optical lattice clocks — precisely because these clocks’ quantum projection noise (the shot-noise-equivalent floor for atom-based measurements) is so low that the laser-drift-during-dead-time noise now dominates. A conventional approach to suppressing it requires either a very long laser coherence time (demanding an enormous, vibration-isolated optical cavity) or two interleaved atom ensembles timed in anti-phase — approaches that have yielded excellent laboratory results but add complexity and bulk.
A December 2025 paper in Physical Review Letters demonstrated a zero-dead-time clock using two interleaved ensembles of strontium-87 atoms, achieving fractional frequency instability at 10⁻¹⁹ at one day — a ninefold improvement over a single-ensemble design. The Katori group’s approach takes the problem from a different angle: rather than using two static ensembles in an established lattice, it uses a single continuous flow of atoms through a single lattice that is always moving. The two strategies solve the same problem through different architectures, with different implications for compactness and field deployment.
How the Conveyor Belt Works
The architecture the Katori group demonstrated is, in its core principle, a division of labor in space rather than in time. Where a conventional optical lattice clock performs atom loading, quantum state preparation, clock laser interrogation, and state detection sequentially in the same location, this design performs all four operations simultaneously in different locations along a flowing atom stream.
Atoms cooled in a magneto-optical trap are loaded into a moving optical lattice — a pattern of laser light traveling at 16 mm/s (0.63 in/s) that acts as a conveyor belt, carrying atoms from one end of the apparatus to the other. As each batch of atoms travels along the belt, it passes through three spatially separated zones: in the first zone, atoms are prepared in the correct quantum state; in the second zone, the clock laser interrogates them for 0.75 seconds; in the third zone, the population is read out by fluorescence imaging. A fresh batch of atoms is always entering the first zone while a previously interrogated batch is being read out in the third — and the clock laser is never left without atoms in its interrogation window.
The engineering problem that had previously made this architecture impractical was Doppler broadening. Shining a clock laser along the same axis as the direction of atom motion — what the paper calls “longitudinal excitation” — would ordinarily produce catastrophic spectral smearing, because atoms at different positions in the moving lattice have different velocities relative to the laser, and therefore see different Doppler-shifted frequencies. The solution lies in a quantum mechanical regime called Lamb-Dicke confinement.
When atoms are trapped tightly in a lattice potential well whose spatial extent is much smaller than the wavelength of the clock laser light, they cannot accumulate a Doppler shift — the physics of tight confinement suppresses the motional coupling — the physics of tight confinement suppresses the motional coupling. The team exploited this: because atoms in the moving optical lattice remain in the Lamb-Dicke regime even while the lattice itself moves, the longitudinal clock laser sees each atom as effectively stationary relative to the laser wavelength, and the spectral line stays sharp.
A localized magnetic field region along the conveyor defines the excitation zone. The strontium clock transition used — a “forbidden” transition normally inaccessible to a laser — is opened by the magnetic field mixing quantum states in a narrow spatial strip 12 mm (0.47 in) long. Upstream of this strip, atoms are being prepared. Downstream, they are being imaged. Only in the strip itself is the clock laser driving transitions.
The direct result of this spatial arrangement is that the clock laser is continuously monitoring atoms throughout the measurement period, correcting its own frequency drift at every instant — with no dead time at all.
Spatial Rabi Fringes and the 1.2 Hz Linewidth
Because atoms spend a predictable, fixed time in the 12 mm (0.47 in) excitation zone — 0.75 seconds at 16 mm/s (0.63 in/s) — the normally time-domain Rabi oscillation pattern (the sinusoidal population cycling that characterizes a clock transition being driven on resonance) maps directly onto a spatial distribution. An atom that has traveled 6 mm (0.24 in) into the excitation zone has been interacting with the clock laser for 0.375 seconds; an atom at 12 mm (0.47 in) has been interacting for 0.75 seconds. The resulting periodic variation in excited-state population is visible as a spatial fringe pattern in an image of the atom cloud — an oscillation in space rather than in time, but containing the same phase information as a conventional Rabi measurement.
Reading the spectral linewidth from this spatial pattern, the team measured a full-width at half-maximum of 1.2 Hz. The Fourier limit for an interaction time of 0.75 seconds — the theoretical minimum linewidth set by the quantum mechanical uncertainty principle applied to finite-duration spectroscopy — is approximately 1.33 Hz. The 1.2 Hz result is below even this simple estimate, confirming that the moving-atom architecture introduces no additional spectral degradation beyond what a static pulsed clock would produce for the same interaction time.
At the strontium clock transition frequency of approximately 429 terahertz (corresponding to a laser wavelength of 698 nm), a 1.2 Hz linewidth corresponds to a quality factor Q of approximately 3.6 × 10¹⁴ — the ratio of center frequency to linewidth that determines the clock’s fundamental ability to resolve tiny frequency shifts. This is not a precision record; many existing static optical lattice clocks achieve comparable or higher Q factors. The significance of the result is not the Q factor per se but that it was achieved from a continuously flowing atom stream with no measurement interruption.
Why Continuous Operation Matters for the 2030 SI Second Redefinition
The global metrology community has been working toward a formal redefinition of the SI second since the late 2010s, with the Bureau International des Poids et Mesures’ Consultative Committee for Time and Frequency (CCTF) setting 2030 as the preferred target date at the 2022 General Conference on Weights and Measures (CGPM). The 2026 CGPM meeting, expected later this year, is anticipated to present a formal proposal with a clear roadmap; a backup scenario would postpone redefinition to 2034.
What is less often reported is that the CCTF roadmap includes specific operational requirements for optical clocks that go beyond raw accuracy. Among them: the ability to compare independent realizations of the second at the 10⁻¹⁸ level continuously, and to contribute to International Atomic Time (TAI) calibration on an ongoing basis rather than in episodic measurement campaigns. A clock that runs continuously and never surrenders monitoring of its laser is, structurally, far better positioned to meet this operational requirement than one that cycles between measurement and dead time.
The Katori group’s result removes one of the concrete technical barriers — not just to making clocks more accurate, but to making them continuously operable in a way that the CCTF roadmap actually requires. A pulsed optical lattice clock, however precise in its measurement window, cannot contribute to TAI calibration at the continuous level that a zero-dead-time architecture enables.
RIKEN and Shimadzu Corporation signed a collaboration memorandum with the BIPM on May 12, 2026, in Kawasaki, Japan, to collaborate on research and demonstration testing of optical frequency standards as candidates for the second’s redefinition. Katori, who serves as team director of RIKEN’s Space-Time Engineering Research Team, was present at the signing alongside RIKEN president Dr. Makoto Gonokami, Shimadzu CTO Takahiro Nishimoto, and BIPM director Dr. Annette Koo. The National Metrology Institute of Japan (NMIJ) is also involved in the collaboration, which remains open to other national metrology institutes.
From Lab Instrument to Field Sensor: What Zero Dead Time Changes
The research program Katori has led at the University of Tokyo for more than two decades has consistently pursued two parallel goals: extreme precision and practical deployability. The Aetherclock OC020, a commercial strontium optical lattice clock launched by Shimadzu Corporation on March 5, 2025, represents the current state of the deployability track. Priced at ¥500,000,000 (approximately $3,156,000 USD), the instrument measures 114 cm (44.9 in) wide by 108.3 cm (42.6 in) high by 65 cm (25.6 in) deep — roughly the size of a large household appliance — and achieves fractional frequency uncertainty at the 10⁻¹⁸ level. Shimadzu has set a target of selling 10 units over three years in domestic and overseas markets.
The dead-time problem is not merely a stability issue for laboratory clocks. It is also an infrastructure problem for field deployment: suppressing the Dick effect in a pulsed clock has historically required extremely large, thermally isolated laser cavities to keep the laser frequency stable during measurement gaps. A clock that continuously monitors its laser can, in principle, use simpler stabilization hardware — which means a smaller instrument, a lighter instrument, an instrument more plausibly carried to a mountainside or deployed on a tectonic monitoring station.
Katori’s BIPM keynote in November 2025, given at the BIPM’s 150th Anniversary Scientific Conference, described the long-term vision: clocks as relativistic sensors capable of detecting gravitational potential differences in real environments. He cited a clock-based measurement of ground uplift following the 2011 Tōhoku earthquake as an example of the kind of geophysical monitoring that becomes possible when optical clocks achieve field-deployable precision.
The mechanism is relativistic geodesy: because Einstein’s general theory of relativity requires that time passes more slowly in stronger gravitational fields, a clock at lower elevation ticks measurably slower than an identical clock at higher elevation. An optical clock sensitive to 10⁻¹⁸ fractional uncertainty can detect the gravitational time dilation corresponding to a height difference of roughly 1 cm (0.4 in). Used continuously, such a clock could monitor slow tectonic deformation, groundwater table changes, or volcanic inflation — all as shifts in the clock’s tick rate relative to a reference, with no satellite needed and no survey crew required.
What the Atom Conveyor Cannot Yet Do — and What Comes Next
The August 6 paper is an architectural demonstration, not a complete operational optical lattice clock. The current experiment uses strontium-88, a bosonic isotope less commonly used in the highest-accuracy clocks (which favor fermionic strontium-87 for its better-controlled systematic shifts). The 1.2 Hz linewidth result confirms that the continuous-flow architecture is spectrally compatible with clock-level interrogation, but the team has not yet demonstrated a full systematic uncertainty evaluation at the 10⁻¹⁸ level from a flowing-atom system — the step required before any continuous-flow result can contribute to TAI calibration.
The paper traces its lineage through two earlier milestones from the same group: a 2021 theoretical proposal by Katori in Applied Physics Express outlining spectroscopy in a moving lattice using longitudinal Ramsey techniques, and a 2024 paper demonstrating continuous ultracold atom beam generation using crossed moving lattices. The August 6 result combines both developments into a working demonstration of near-Fourier-limited spectroscopy from a continuous atom source — the first time clock-quality linewidths have been achieved from a never-interrupted atom flow in an optical lattice.
What comes next, per the UTokyo press release, is extending the demonstration to full systematic uncertainty evaluation, miniaturizing the platform, and eventually running continuous field demonstrations of relativistic geodesy. The Japan Science and Technology Agency (JST) is funding this work through its Future Society Creation Program under a large-scale grant titled “Construction of a Spacetime Information Infrastructure using Cloud Optical Lattice Clocks” (JPMJMI18A1), with Katori as principal investigator. JEOL Ltd. has been a contributing industrial partner, and the commercial trajectory runs through Shimadzu. The full scope of the program is described at the Katori Project website.
For the global timing community watching the 2030 SI second redefinition timeline, the August 6 result is not a precision record to be added to a leaderboard. It is a demonstration that the operational architecture required for continuous timekeeping — the ability to run a clock the way a power plant runs a generator, without stopping — is achievable in an optical lattice framework. The clock has learned not to blink.
Frequently Asked QuestionsWhat is the “dead-time problem” in optical atomic clocks, and why does it matter?
Every conventional optical lattice clock operates in cycles: load atoms, cool them, prepare their quantum state, probe them with the clock laser, read out the result, and repeat. Only the probing step actually compares the atoms to the laser — during everything else, the laser runs uncorrected and drifts freely. That gap is called dead time, and it causes a well-characterized stability degradation known as the Dick effect: the laser’s high-frequency noise aliases into the clock’s output during the gaps, limiting how stable the clock can be over time. Suppressing dead time has historically required either two interleaved atom ensembles (complex and bulky) or an ultra-stable laser cavity (expensive and room-sized). The Katori group’s continuous-flow architecture removes dead time entirely by running preparation, interrogation, and detection simultaneously in different locations along a moving atom stream.
When is the SI second expected to be redefined, and what does continuous clock operation have to do with it?
The global metrology community, coordinated by the Bureau International des Poids et Mesures’ Consultative Committee for Time and Frequency (CCTF), is targeting the 2030 General Conference on Weights and Measures (CGPM) as the preferred date for replacing the 1967 cesium microwave definition of the second with an optical frequency standard. The CCTF roadmap requires not only that candidate optical clocks achieve 10⁻¹⁸ fractional uncertainty — which they already do — but also that they can contribute to International Atomic Time (TAI) calibration continuously and comparably across laboratories. A clock architecture that runs without measurement gaps is fundamentally better suited to meeting that operational requirement than one that cycles between interrogation and preparation. The August 6 paper demonstrates that continuous interrogation is architecturally achievable in an optical lattice framework — removing a technical barrier that could otherwise push the redefinition to the 2034 backup scenario.
What can a continuously operating optical clock do that a pulsed optical clock cannot?
Beyond the stability improvement from eliminating the Dick effect, a clock that never stops measuring can serve as a continuous gravitational sensor. Through relativistic geodesy — the application of Einstein’s general relativity to clock-rate comparisons — an optical clock sensitive to 10⁻¹⁸ fractional uncertainty can detect height differences of roughly 1 cm (0.4 in) as changes in its tick rate relative to a reference clock. Running continuously, such a clock can track slow ground deformation from tectonic stress accumulation, monitor aquifer depletion as mass redistribution changes the local gravitational field, or detect pre-eruptive volcanic inflation. These applications require continuous uninterrupted monitoring — the same requirement that makes zero-dead-time architecture essential for both the best possible timekeeping and the broadest geophysical sensing.
Can this design be turned into a compact field instrument?
Not yet, but that is the explicit goal of the research program. The commercial Shimadzu Aetherclock OC020 — a static pulsed optical lattice clock launched in March 2025 at ¥500,000,000 (approximately $3,156,000 USD) — represents the current compact form factor: roughly the size of a large refrigerator. The conveyor-belt architecture demonstrated in this paper opens a path to simpler laser stabilization (no need for the large cavities that compensate for laser drift during dead time), which in turn reduces the instrument’s size and environmental sensitivity. The UTokyo-RIKEN group has stated that field-deployable continuous optical clocks are the long-term target, with the August 6 result as a foundational architectural demonstration on the road to that goal.
The paper, “Spatially defined Rabi spectroscopy for uninterrupted optical clock interrogation,” by Koki Nishida, Ryoto Takeuchi, Shigenori Tsuji, Shoichi Okaba, and Hidetoshi Katori, was published in Nature Communications on August 6, 2026. DOI: 10.1038/s41467-026-76017-1