The Time and Frequency Division (TFD) is located in Boulder, Colorado. The division has 121 staff, including 8 administrative support; 40 NIST scientists; 70 associates (postdoctoral researchers, graduate and undergraduate students, and visiting scientists); and 3 emeritus scientists.1 The annual budget is $22 million, 23 percent of which is sourced from other agencies.
Disseminating accurate and reliable time and frequency information is a core responsibility of NIST and the core responsibility of the TFD. Accurate time and frequency services are critical to areas of the U.S. economy, such as coordination of the phase of electrical power generation for the smart grid, the Global Positioning System (GPS), synchronization of computer networks, time stamping of financial transactions, and national security and research. The TFD, in coordination with the U.S. Naval Observatory, is responsible by law for disseminating time throughout the United States to all stakeholders.
ASSESSMENT OF TECHNICAL PROGRAMS
Time and Frequency Services
NIST disseminates time signals based on the coordinated universal time (UTC) time scale, the internationally generated scale that is coordinated by the International Bureau of Weights and Measures in France. UTC is not available for dissemination, and therefore NIST employs UTC1, a version of UTC that is controlled by two atomic frequency standards, NIST-F1 and NIST-F2. These employ fountains of laser-cooled cesium atoms that are interrogated by microwaves and have fractional uncertainty in their generation of the SI second of 1 part in 1016. NIST-F1 and NIST-F2 (along with the NIST-F2 copy housed at Italy’s National Metrology Bureau) are the world’s most accurate frequency standards.
The time scale UTC1 is generated by an ensemble of about 10 commercial atomic clocks (mostly hydrogen masers). The “tick rate” of this ensemble is calibrated against NIST-F1 and NIST-F2 every few months, permitting UTC1 to be steered to UTC. This system operates effectively, efficiently, and reliably.
Time must be disseminated continuously and in a variety of forms depending on the need. A measure of the value of the service NIST provides in dissemination is the 40 billion synchronization requests it serves each day. The primary methods for disseminating time and frequency are the following:
- For the most stringent requirements, NIST provides frequency uncertainty of 1 part in 1013 and time uncertainty of 1 nanosecond, using specialized black-box devices that NIST makes available to the user. NIST uses telecommunications and GPS satellites to broadcast time and frequency to such devices. These are the only such remote time and frequency measurement
1 Chris Oates, National Institute of Standards and Technology, Physical Measurement Laboratory, “Division 688: Time and Frequency,” presentation to the committee, May 1, 2018.
services in the world at this level of precision and accuracy and are used by about 50 high-tech companies.
- For broad public use, NIST broadcasts time and frequency information by radio on station WWVB, with a frequency stability of 1 part in 1011 and time uncertainty of 1 microsecond. Signals are broadcast at different frequencies for purposes ranging from synchronizing wall clocks and wristwatches to high-precision scientific experiments.
- The Internet Time Service (ITS) synchronizes computers and network devices to NIST time with an accuracy of about 1 millisecond and is used billions of times per day. ITS is the most heavily used network time service in the world and is built into all major computing operating systems (e.g., Windows, Apple, and Linux).
For their respective applications, these three primary methods of time distribution are widely regarded as among the best in the world.
Optical Frequency Metrology and Ion Storage
A New Generation of Frequency Standards
A revolution in metrology is under way, enabled by the invention of the optical frequency comb and the development of trapped ion and atom lattice optical clocks.
NIST’s TFD is a leader in all of these developments. The optical frequency comb was created by John Hall in the Quantum Physics Division (JILA) and Theodor Hänsch in Garching, for which they received the Nobel Prize in 2005. David Wineland of the Ion Storage Group received the Nobel Prize in 2012 for developing the trapped ion technique. There has been major progress in time and frequency metrology since those prizes were awarded.
The optical comb makes it possible to read out optical signals with the same facility that microwave, and lower, frequencies can be manipulated. Its creation opened the road to the development of frequency standards that operate at optical frequencies. The advantage of optical frequencies is that they are 100- to 1,000-times higher than the microwave frequencies of current frequency standards, permitting a corresponding decrease in uncertainty. The advances in high-precision time metrology at the TFD have been rapid.
One of the new generation of optical frequency standards employs ytterbium atoms trapped and cooled in an optical lattice. The ytterbium optical standard has achieved a world-record uncertainty of a few parts in 1019 after several hours of observation. Two independent ytterbium lattice frequency standards were constructed within the TFD and compared, confirming that they agree to better than 1 part in 1018. A second type of standard is based on a single-trapped and cooled aluminum ion. The ion is interrogated by entanglement with a co-trapped beryllium ion using quantum logic protocols pioneered at NIST. A third type of optical standard with comparable uncertainty, based on strontium atoms trapped and cooled in an optical lattice, was created at JILA. These three standards have the lowest uncertainty in the world, and the fact that all three co-exist within the same institution enables unique studies. For example, direct comparison between these frequency standards constitutes one of the most sensitive tests of the invariance of fundamental constants.
The problem of transmitting time signals and comparing frequency standards in this new regime of precision poses a serious challenge. Comparisons in the same laboratory are relatively straightforward, but the problem of comparisons at large distances remains to be solved.
Compact Frequency Combs
Frequency combs are a crucial component of optical clocks, making it possible to compare clocks and to generate outputs at chosen frequencies. They also enable a host of additional applications in precision signal generation and measurement. For example, TFD researchers have applied combs to generate 10 GHz microwave signals with the smallest low-frequency phase noise ever reported. Because sensing slow motion requires low phase noise, this advance is important for radar detection of slowly moving objects. The original combs involved mode-locked lasers. These are bulky and power-hungry laboratory devices that are unsuited for many commercial and military applications. Also, the frequency spacing of their comb lines is too small for some important needs.
TFD researchers have investigated alternative approaches that overcome both problems. One frequency comb alternative uses micro-resonators that are small, require little power, and are well suited to portable operation; some realizations are compatible with integrated silicon photonics. The TFD is among the most active laboratories, both in elucidating the science of micro-resonator frequency combs and in developing them for applications, with an emphasis on those central to time and frequency—for example, chip-scale atomic clocks (CSAC), low-phase noise microwave generation, and precision optical frequency synthesis. In another alternative approach, TFD researchers have developed electro-optic combs (combs generated via strong-phase modulation of a single-frequency input laser) for broad optical bandwidth while maintaining low-phase noise. Recently, they have installed a portable version of their electro-optic comb at the 10-meter Hobby-Eberly telescope in the McDonald Observatory in Texas, where it provides precise astronomical spectrograph calibration with application to the search for exoplanets.
The advances in high-precision time metrology at the TFD have been rapid. Two independent ytterbium lattice frequency standards were constructed within the TFD and compared to confirm that they agree to better than one part in 1018.
Opportunities and Challenges
The definition of the SI second is exactly 9,192,631,770 cycles of the unperturbed ground-state hyperfine transition in the cesium-133 atom. The second can be realized in practice only to the limit of the accuracy of the cesium atomic clocks, currently 1 part in 1016. This is widely regarded as the limit for cesium-based frequency standards, but it is far below the stability and accuracy already demonstrated for optical clocks. To take advantage of the hundred-fold improvement in determination of time and frequency that has been enabled by optical clocks, the SI second must be redefined. There are plans to do this in about a decade, but several significant challenges must first be overcome.
The first challenge is to create and select the best practical optical frequency standard. Several types have already been demonstrated, and others may yet be created. The task of turning these laboratory devices into practical frequency standards is enormous, but work is under way. For instance, the TFD is constructing a portable ytterbium optical lattice clock that can be physically moved to other locations for local clock comparisons with laboratories throughout the world. One can reasonably look forward to advances in making ion and atom-lattice clocks practical.
The problem of transmitting time signals and comparing frequency standards separated by long distances remains a challenge. The second challenge is to develop methods for distributing time over long distances at the new levels of accuracy. Presently, NIST employs the GPS system and two-way microwave satellite time and frequency transfer to synchronize NIST time with UTC. These techniques are suitable for time coordination at the level of 1 part in 1016, but they are not adequate for the future.
The TFD is seriously engaged with the problem. In collaboration with the Applied Physics Division, it has carried out studies using optical fibers to transfer frequency and has demonstrated stability and accuracy at a few parts in 1019 over distances of a few hundred kilometers. (Time transfer has been demonstrated over distances of up to about 10 km with a similar level of degradation.) This accuracy would be sufficient to coordinate UTC at the 1 part in 1018 level that will be required in the future. They have demonstrated similar stability and accuracy of time and frequency transfer using laser beams propagating through free space over distances of a few kilometers. The challenge for extending either fiber- or free-space techniques to intercontinental transfer is formidable. Future experiments with longer distance fiber networks and free space time/frequency transfer are essential for taking advantage of the levels of precision being provided by the new generation of optical frequency standards.
A final challenge to the redefinition of the second is gravity, which is a nuisance, an opportunity, and a dilemma. Near Earth, gravity causes time to change by about 1 part in 1018 per centimeter of altitude, as predicted by General Relativity. The effect is already significant for metrology: corrections for it had to be made in high-precision comparisons of frequency standards in NIST’s own laboratories, some of which are separated by 4 km. For large distances, the problem is severe. Altitude is essentially the distance above the geoid, a hypothetical surface of constant gravitational potential. Because of fluctuations in Earth’s mass distribution, the geoid fluctuates by millimeters or more over periods of a day or less, and by centimeters over periods of longer than a month. The new generation of frequency standards and clocks will be sensitive to these fluctuations.
In preparation for clock comparisons at large distances, NIST carried out a geodetic survey of its campus and its laboratories at JILA (separated by about 4 km) to facilitate comparisons of clocks in those locations.2 The survey achieved a precision that corresponds to a frequency uncertainty at the level of approximately 5 parts in 1018, somewhat larger than the estimated uncertainty of the test clocks. At very large distances, one would expect the geodetic precision to be even lower. Furthermore, as noted above the geoid is known to fluctuate at this level. This implies that ultimately it will not be possible to transfer terrestrial time and frequency information with the accuracy of forthcoming frequency standards. The underlying problem is that at such a level of precision, time is inextricably coupled to the gravitational field. It would be natural to reverse the logic and use frequency standards as a tool for geodesy, a step that is being mentioned increasingly. This offers the possibility of opening new pathways for geodesy. This could have important applications, but it does not solve the problem of time transfer.
The ultimate significance of the new regime of frequency accuracy is that it forces a confrontation between the concepts of space, time, and mass. Clocks cannot be compared without knowing the distribution of nearby mass. The problem could be avoided by requiring that clocks be compared close to the same location. This procedure would necessarily introduce some imprecision. More seriously, it requires giving up the fundamental concept of space, time, and mass as independent physical quantities.
Such a situation has occurred once before. Space and time are now joined by the definition of the speed of light, c. The meter is the distance traveled by light in the time 1/c. Formally, we refer not to space and time but to space-time. With gravity entering the picture, space-time and mass are joined by General Relativity. There is, however, a dramatic difference between the issues of space-time and space-time-gravity: the speed of light, which connects space and time, is the most precisely measured fundamental constant, whereas the gravitational constant, G, which connects mass to space-time, is the least precisely known constant. Consequently, it is difficult to visualize a metrological procedure that involves G. This problem must be resolved in the process of redefining the SI second.
2 N.K. Pavlis and M.A. Weiss, 2017, A re-evaluation of the relativistic redshift on frequency standards at NIST, Boulder, Colorado, USA, Metrologia 54:535.
Atomic Devices and Instrumentation Group
Chip-Scale Atomic Sensors: NIST-on-a-Chip
The Atomic Devices and Instrumentation Group (ADIG) was founded in 2008, with its precursor elements having had a role in developing the chip-scale atomic clock (CSAC) under Defense Advanced Research Projects Agency (DARPA) funding. This stimulated the military and telecommunications industry to develop a new class of small and relatively inexpensive metrological instruments that are described under the rubric of NIST-on-a-chip, a NIST-wide initiative to which the ADIG is a contributor and PML the lead. One class of the proposed chip instruments involves metrological tools based on accurate representations of SI base units such as length, electric current, temperature, luminous intensity, and, of course, time. Common to these applications is a vapor cell under development at ADIG that may be able to provide length (meter), electric current (ampere), temperature (Kelvin), luminous intensity (candela), and time (second). Another innovation within the group involves chip-scale inertial sensors that would take over when GPS becomes unavailable. Such devices have the potential to make high precision widely available without the user needing to turn to NIST to assure calibration. Design goals emphasize miniaturization and portability. Manufacturing would employ micro-fabrication techniques to achieve high-volume production that would decrease costs. The chip-scale magnetometers developed by the ADIG are now widely used in industry.
The ADIG has made important contributions to the vapor cell metrological tool—which provides numerous measurements such as that of length, electric current, and so forth—resident in NIST-on-a-chip.3 The group is also developing a novel chip-scale inertial sensor.
PORTFOLIO OF SCIENTIFIC EXPERTISE
NIST is widely regarded as the world’s leading time standards laboratory. Reasons for this include its long history of atomic clocks—the atomic clock was invented at NIST—and the tradition of carrying out time-keeping activities in an atmosphere of frontier research, a tradition of excellent management that permits the TFD to attract and retain outstanding talent, excellent facilities, and adequate resources. NIST’s TFD has a culture in which routine operations are constantly analyzed and improved. For example, research on the effect of blackbody radiation on the primary frequency standards resulted in a significant reduction in their fluctuations. Furthermore, NIST has a culture of collegiality that encourages a free flow of ideas between the various groups within the TFD, with other divisions throughout the PML, and with the many visiting researchers and students. Ultimately, these visiting researchers and students are NIST’s most effective envoys for spreading the new knowledge and technology to the broad community.
FACILITIES, EQUIPMENT, AND HUMAN RESOURCES
The TFD is primarily located in the new Katharine Blodgett Gebbie Laboratory Building in Boulder. This building has superb facilities. The equipment is judged by the staff to be adequate for the TFD’s needs. Some of the work is located in Wing 5, which is slated for renovation in the coming year
3 For a discussion of atomic vapor cells, see National Institute of Standards and Technology (NIST), “NIST-on-a-Chip: Atomic Vapor,” https://www.nist.gov/pml/nist-chip-atomic-vapor, accessed July 10, 2018.
with a predicted completion date of 2021. Once Wing 5 is renovated, all the facilities will be adequate. PML envisages the renovation as closing the gap between its “world-class lab space” and its 1960s-era space.4
Judged by the presentations, the quality of conversations with staff members, and the scientific output of the TFD, the quality of the staff is outstanding. Further evidence for the quality of scientific expertise is provided by their honors. These include numerous awards from the Institute of Electrical and Electronics Engineers, NIST’s Condon and Astin awards, fellowship in the American Physical Society, and others. Their publications appear in the most prestigious scientific journals and are more numerous and impactful than publications from similar national time and frequency metrology laboratories around the world. The scientific atmosphere, facilities, and staff at NIST attract luminaries, visiting scholars, postdoctoral researchers, and students, who contribute significantly to NIST’s programs.
The human resources appear to be well matched to TFD’s needs.
DISSEMINATION AND TECHNOLOGY TRANSFER
At the top of TFD’s agenda is the dissemination of time and frequency, an enormous responsibility that it meets at the highest level. However, the TFD provides many other services to the community: it performs 500 to 600 calibrations per year—for example, precision phase-noise calibrations at a level unique in the world; it provides calibration equipment and specialized services directly to about 50 industrial firms, which greatly reduces calibration time for industry and reduces the calibration burden on the TFD; and it helps industry to develop new products such as the chip-scale atomic clocks, magnetometers, and phase noise test equipment.
The TFD has an outstanding record of sharing its scientific and technical advances and stimulating new industries for commercial, military, and scientific use. For example, today CSACS are in use for important military applications, and about 100,000 CSACS are used commercially with sales of about $100 million in telecommunications systems, seismic exploration activities, and other areas. In a closely related technical development, chip-scale magnetometers are now widely used by the oil industry. The magnetometers are also being applied in medicine, making magneto-encephalography a practical diagnostic tool that has been commercialized by a number of companies.
In another application, the TFD, working with external collaborators, has developed a compact and portable version of a laser-cooled microwave frequency standard. This was created through partnership with a nearby small company, Spectradynamics. The device was funded in part through a DARPA Small Business Innovation Research grant and is near commercialization.
Opportunities and Challenges
Notwithstanding these remarkable successes, there are some issues related to the commercialization of NIST-on-a-chip. Given that people from NIST, especially in the TFD, are the strongest evangelists for the NIST-on-a-chip concept, and given the success of their vapor-cell-based CSAC and miniature atomic magnetometer efforts, both of which have been commercialized, NIST is arguably in the best position to develop the concept to a level to where industry would take over. While NIST is adept at working the science of this initiative, there are questions regarding how far they are
4 Chris Oates, NIST, Physical Measurement Laboratory, “Division 688Time and Frequency,” presentation to the committee, May 1, 2018.
prepared to do the engineering needed to make the product viable for development. Perhaps more important than solving the scientific problems is the need to realize the control mechanisms and packaging around the science that assure low power consumption and low cost. Much of the work needed might be most appropriate for engineers, particularly those skilled in microelectromechanical systems (MEMS).
NIST-on-a-chip is somewhat new territory for NIST; its success requires a much higher volume product output than previous NIST outputs. NIST will probably need to court companies unlike the low-volume, timing- and frequency-control companies that speckle their past. Companies familiar with MEMS technology will likely be most appropriate for NIST-on-a-chip, and these companies will require some amount of convincing to jump on the NIST-on-a-chip bandwagon. If this requires that NIST take NIST-on-a-chip closer to a product than previously needed, using more MEMS and integrated circuit technology, then they may need to invest in engineering talent. In particular, although the current NIST staff seems happy to do the needed science, they perhaps are not as interested in undertaking the needed engineering. With a new cleanroom fitted with nanofabrication technology, NIST has equipment appropriate to do the engineering itself.
Another approach might be to (continue to) rely on companies or academic institutions proficient in MEMS technology and integrated circuit design to do the necessary engineering. This approach, however, would likely require more support from NIST, not only on technical fronts, but political and financial, as well. Specifically, NIST might need additional funds to direct to technology transfer sites. Alternatively, the local NIST-on-a-chip evangelists might need to focus their convincing arguments on funding agencies that can direct funding to companies interested in commercializing NIST-on-a-chip. This issue is important to consider.
A major emphasis of the TFD is transfer of scientific and technical knowledge through training its many students, postdoctoral researchers, and visiting workers, as well as through publications, talks, seminars, and direct work with collaborators and colleagues. They do this very well. Technology transfer via product development is somewhat secondary to TFD’s main line of activities, despite a number of successes, as narrated above. The review did not include presentations of policies regarding mechanisms for fostering product development through spinning off companies, sponsoring Small Business Innovation Research grants, or policies such as employee leaves to pursue commercialization activities. It is worth making sure that clear policies governing such activities related to technology transfer are in place and communicated to staff.