Font Size: SmallerFont Size: DefaultFont Size: Larger
  • 日本語トップ

Fiber-based direct comparison of distant optical lattice clocks at the 10-16 uncertainty

- Real time detection of relativistic time dilation due to an elevation difference of 56 m -

  • Print this page
August 4, 2011

The National Institute of Information and Communications Technology (NICT, President: Dr. Hideo Miyahara), in collaboration with The University of Tokyo (UT, President: Junichi Hamada, Ph.D.), linked the NICT headquarters (Koganei) and the Hongo campus of UT with a 60-km optical fiber, and compared two optical lattice-clock frequencies generated independently. In less than 20 s of signal integration, the relativistic frequency difference caused by the elevation difference of 56 m was clearly observed. After the correction of systematic differences, agreement of the frequencies was observed at the 10-16 level (1 s error in more than 65 million years), confirming the universality of optical lattice clocks as a frequency standard. The result was published online in Applied Physics Express (APEX) on August 4, 2011.

This work is supported by JST-ERATO, JST-CREST, Cabinet Office-FIRST, and MEXT-Photon Frontier Network programs.


Inter-comparison against another clock with same level of accuracy is the only way to evaluate and confirm the reproducibility of state-of-the-art atomic clocks. While the technology to build optical clocks has recently made rapid progress, independently built optical clocks have not been compared at distant locations and their accuracy until now has been unconfigured at the 10-16 level.

New Achievements

NICT has developed an optical lattice clock as well as a fiber-based optical-frequency transfer system. Connecting two lattice clocks separately located at NICT and UT with 60-km optical fiber network, the differential frequency was obtained in short time, indicating the differential time dilation due to a 56-m difference in elevation. Furthermore, the correction of such well-known frequency shifts has resulted in the agreement of the two clock frequencies at the 10-16 level, in other words, no loss or gain of one second in 65 million years. This result is the first case that the reproducibility of clock frequencies at the 10-16 level is experimentally confirmed in short time by using two physically separate clocks. The result also demonstrates the universality of lattice clocks invented in Japan as well as the R&D capability of metrological research in Japan. In addition, the result indicates that NICT has established a method to faithfully transfer the most stable and accurate optical frequency standard to remote locations.

Future Plans

The result will push forward an optical definition of the second in the international system of units. Another one order of improvement will allow us to characterize the gravitational potential from the clock frequency, enabling various applications such as searching for natural resources in the ground. By referring a remote frequency standard using the technique developed here, a frequency reference used for laboratory instruments or a communication network will be quickly calibrated at any time.

* Information on paper publication
  “Direct Comparison of Distant Optical Lattice Clocks at the 10-16 Uncertainty”
Atsushi Yamaguchi, Miho Fujieda, Motohiro Kumagai, Hidekazu Hachisu, Shigeo Nagano, Ying Li, Tetsuya Ido, Tetsushi Takano, Masao Takamoto, and Hidetoshi Katori

Supplementary information

1. Overall design of the fiber-based frequency comparison

Overall design of the fiber-based frequency comparison

We compared two 87Sr lattice clocks at NICT and UT. The wavelength of the clock transition is 698 nm, requiring a wavelength conversion to the telecommunication band (1538 nm) to transfer the clock signal from NICT to UT. A frequency comb on the NICT side connects the two wavelengths. The equidistant spectrum of the frequency comb is tightly locked to the clock laser at 698 nm. A continuous wave laser at 1538 nm is phase-locked to the frequency comb and transferred to UT. Optical fibers normally add phase noise caused by vibrations or temperature variation over the transmission path. On the UT side, the light transferred from NICT is partly reflected back to NICT. This returned light tells how much phase noise the light suffered in the round trip. Flipping the sign of the detected phase noise and dividing by two, the NICT side in advance adds the phase shift for compensation so that the clock signal at NICT is revived at UT without fiber-induced noise. The transferred 1538 nm light is frequency doubled on the UT side and another frequency comb at UT is locked to the 1538 / 2 = 769 nm light. Thus, the lattice clock at NICT is coherently linked to the frequency comb at UT, to which the lattice clock signal at UT is compared.

2. The observed frequency difference and stability

The observed frequency difference and stability

Upper panel: The time record of the frequency difference obtained in every second is shown. The positive difference of 3-4 Hz indicates that the NICT clock generates 3-4 Hz higher frequency than the UT clock. This is mainly due to the relativistic time dilation, namely the time at the NICT headquarters (Koganei) runs faster than that at the Hongo campus of UT due to the smaller gravitational potential.

Lower panel: The graph shows the stability of the fractional frequency difference after the signal integration. As the signal is averaged for 1000 s, the uncertainty of the measurement goes down to 4.5 × 10-16. This means that the frequency difference is determined with the uncertainty of 430 THz × 4.5 × 10-16 = 0.19 Hz by integrating signal for 1000 s (Allan standard deviation). The deviation does not decrease as shown above unless the clock frequency is well stabilized for long time.

3. Total frequency difference after the calibration of systematic shifts

Total frequency difference after the calibration of systematic shifts

Frequency difference after the correction of gravity shift and other minor shifts. The overall average and its uncertainty are evaluated to be 0.04 +/- 0.31 Hz in 430 THz. The accuracy of the two clocks is mutually confirmed to be better than 7 × 10-16. The agreement of the two frequencies generated at physically separated places was for the first time confirmed with uncertainty at the 10-16 level.

Technical Contact

Tetsuya Ido

Space-Time Standards Laboratory
Applied Electromagnetic Research Institute
Tel:+81-42-327- 6527

Media Contact

Sachiko Hirota

Public Relations Department