Division home page

Technical Highlights

• Ultra-Narrow-Linewidth Optical Frequency Standard. R. Rafac, B. Young, W. Itano, D. Wineland, and J. Bergquist of the Division and J. Beall of EEEL have frequency locked an ultra-narrow-linewidth laser to an ultraviolet transition in a single ¹⁹⁹Hg⁺ ion to produce the highest-Q, narrowest-linewidth optical-frequency standard ever demonstrated. They project that the uncertainties for such standards should approach 1 x 10^-18, orders of magnitude better than the best present microwave frequency standards.

  The laser linewidth is initially narrowed by frequency locking it to a Fabry-Perot etalon of high-finesse, and high stability is achieved through exceptionally good isolation from building vibrations and acoustic noise. This line-narrowed (linewidth of ~0.2 Hz) laser is frequency doubled and then used to probe the 282 nm (1.06 x 10¹⁵ Hz) electric-quadrupole transition in a single ¹⁹⁹Hg⁺ ion stored in a cryogenic radio-frequency ion trap. The measured linewidth is 6.7 Hz resulting in a Q  1.6 x 10¹⁴.

  Systems with such exceptional Q are extremely attractive as frequency standards, since large Q translates to small uncertainty in locating the line center of the resonance. Of course, systematic frequency shifts caused by non-ideal experimental conditions must be considered, but while these limit the performance in the present experiments to an uncertainty of order 10^-14, they appear to be controllable at much lower levels. The optical-frequency output of this standard has been linked to the microwave region using the new optical-frequency-synthesis methods described below. (J. Bergquist).

• Improved Calcium Optical Frequency Standard. Substantial improvements in the performance of NIST's calcium optical-frequency standard have recently been made by C. Oates and L. Hollberg of the Division, together with E. Curtis, a graduate student at the University of Colorado. In particular, the stability (shown in Fig. 1) of the standard has been improved to _y() = 4 x 10^-15 at 1 s. For optical frequency standards, this is second only to the mercury ion standard described above. This standard has played a key role in recent optical-frequency measurements using frequency combs (described below), since without this standard, the stability performance of the mercury-ion standard could not have been measured, and the ability of the combs to translate high stability in the optical region to lower frequency could not have been evaluated.
  

  Figure 1. Noise performance of the laser-cooled calcium frequency standard. Curve (a) shows the raw frequency fluctuations, while curve  (b) shows the stability _y() as a function of . The dotted line represents the noise floor for the system used for obtaining the measurement.

  Calcium continues to be an attractive optical frequency standard at many laboratories in the world, since the 657 nm transition is relatively insensitive to external electric and magnetic fields, and because the entire system (trapping, laser cooling, and state probing) is operated with diode lasers. While the system is already laser cooled, the temperature of the atoms might be further reduced using second-stage cooling schemes to achieve a projected stability as low as _y() = 2 x 10^-16 -1/2. (C. Oates)

• Measurement of an Optical Frequency in a Mercury Ion. Division staff members have recently reduced the uncertainty in the measurement of the 282 nm optical-clock transition of a single positive mercury ion by a factor of 1000. The measurement was made by K. Vogel, S. Diddams, C. Oates, E. Curtis, R. Rafac, J. Bergquist, R. Fox, D. Lee, and L. Hollberg of the Division. This transition and optical transitions in other ions appear to be extremely promising as primary frequency standards, because their systematic frequency shifts can be exceedingly small (as low as 1 part in 10¹⁸) and their Q's (frequency divided by linewidth) are so enormous. The Boulder group recently measured a Q exceeding 10¹⁴.

  The measurement of the Hg⁺ line was made relative to a known optical transition in calcium using a new femtosecond-laser frequency comb. The comb concept, developed by T. Udem, T. Hänsch, and their colleagues at the Max Planck Institute in Germany and J. Hall and collaborators at JILA, provides a large set of equally spaced frequencies that can be used, almost like a ruler, to make measurements of very large frequency differences, in this case a difference of 76 THz. The uncertainty of the new measurement is limited, not by the measurement method, but rather by the uncertainty of 120 Hz for the calcium frequency, measured at PTB using conventional frequency-synthesis methods. A further advance in measurement of the mercury transition is described in the following highlight. (J. Bergquist, L. Hollberg).

• Connecting Optical Frequency Standards to the Microwave Region. S. Diddams, D. Lee, K. Vogel, and L. Hollberg of the Division, along with T. Udem of the Max Planck Institute in Germany, have built a high-repetition-rate mode-locked laser that has been used to produce evenly spaced optical-frequency combs for generating a microwave-frequency output from optical-frequency standards. A picture of the laser system is shown in Fig. 2. The output of their laser, operating at a repetition rate of 1 GHz, is fed to highly nonlinear microstructure optical fiber that substantially broadens the frequency comb to cover more than an octave spanning the visible portion of the spectrum. The comb extends from less than 300 THz (1000 nm) to more than 600 THz (500 nm). To demonstrate generation of a microwave output from an optical standard, they stabilized a line in the comb to the Hg⁺ optical frequency standard described above. They were then able to make frequency measurements at the 1 GHz repetition frequency as well as at any other optical frequency spanned by the comb. The stability of the microwave output could not be characterized, since its short-term stability is substantially better than that of any oscillator cu rrently available within the Division. However, they were able to make measurements against the very-high-stability calcium frequency standard operating at 657 nm. The relative fractional-frequency stability observed in these experiments was _y() = 8 x 10^-15 -1/2, an order of magnitude better than the stability of the best quartz oscillators.
  

  Figure 2. Femtosecond mode-locked laser used for generating frequency combs. The system is contained on a table area of 0.6 m x 1.2 m. Future development will certainly decrease the size of the system.

  While this same system was also referenced to a microwave frequency to make measurements of both the calcium and mercury-ion frequencies relative to the cesium-fountain frequency standard (see Fig. 3), it is the generation of a microwave output that is so significant to the future of primary frequency standards. It has long been recognized that line Q provides a very good figure of merit for assessing the performance of frequency standards, and that optical standards, having such high Qs (better than 10¹⁴ for the mercury ion), should outperform their microwave counterparts (best Q of ~10¹⁰) by a wide margin. What has been missing until now is a means for counting cycles and making measurements at lower frequencies. Now that the extraordinary performance of the optical frequency standards (such as the mercury-ion standard) can be translated to lower frequency, they can be seen as a new generation of frequency standards that should substantially outperform even the cesium-fountain frequency standard. (S. Diddams).

  

  Figure 3. Frequency measurements of 1/2 the frequency of the 282 nm (1.06 x 10^-15 Hz) ¹⁹⁹Hg⁺ transition made using the femtosecond comb system. The measurements were made relative to the microwave frequency of the primary cesium frequency standard. The uncertainty of the measurement of the frequency of this transition is ~3 x 10^-15, limited by the uncertainty of the frequency of the cesium transition and the stability of the hydrogen maser used as an intermediate reference in the experiments. Systematic effects in the Hg⁺ ion have not yet been evaluated, but are expected to be < 1 Hz.

Quantum Entanglement of Four Particles. In the cover story for the March 16, 2000 issue of Nature magazine, members of the Ion Storage Group describe the first successful quantum entanglement of four particles, an important step in demonstrating a quantum-processing system that possesses the requisite characteristics for scaling to larger computing systems. Such entangled states explicitly demonstrate the non-local character of quantum theory, and have been suggested for use in high-resolution spectroscopy, quantum communication, cryptography, and computation. The project team, led by C. Monroe and D. Wineland of the Time and Frequency Division included W. Itano, C. Sackett, D. Kielpinski, B. King, C. Langer, V. Meyer, C. Myatt, M. Rowe, and Q. Turchette. The entanglement used a recently proposed technique applicable to trapped ions. Coupling between the ions, stored in a lithographically fabricated trap (see Fig. 4), is provided by the Coulomb interaction through their collective motional degrees of freedom, but actual motional excitation is minimized. Entanglement was achieved using a single laser pulse, and the same one-step method can in principle be applied to any number of ions. (D. Wineland).

Figure 4. Microcircuit rf trap used for the four-ion experiments. A blowup of the trapping region is shown at the bottom.

• Quantum Memory Using Trapped Ions. D. Kielpinski, V. Meyer, M. Rowe, C. Sackett W. Itano, C. Monroe, and D. Wineland have recently demonstrated operation of a decoherence-free quantum memory using trapped ions. A quantum memory stores information in superposition states of a collection of two-level systems called qubits. Quantum computation works by operating on information in the form of such superpositions, and robust quantum memories are therefore essential to realizing the potentials gains of quantum computing. However, interaction of a quantum memory with its environment destroys the stored information, a process called decoherence. Many proposed quantum memories decohere via an environment that has the same coupling to each qubit. In the trapped-ion demonstration, information from an arbitrary qubit stored in a single ion is encoded into a decoherence-free subspace of two ions. The decoherence-free subspace states are invariant under the coupling to the environment, protecting the encoded information. The experiments on this memory concept involved measurement of the storage time under both ambient conditions and under interaction with an engineered noisy environment. They found that encoding the qubit information into the decoherence-free subspace increases the storage time by up to an order of magnitude. (D. Wineland).

• Test of Bell's Inequalities Using Trapped Ions. In experiments using trapped, laser-cooled ions, M. Rowe, D. Kielpinski, V. Meyer, C. Sackett, W. Itano, C. Monroe, and D. Wineland of the Division have demonstrated violation of Bell's inequalities. These mathematical inequalities provide a basis for experimental tests whose results can distinguish between quantum mechanics and local realistic theories. Many experiments have been done that are consistent with quantum mechanics and inconsistent with local realism. Because these conclusions have generated considerable debate, experiments are still being refined in order to overcome "loopholes" that might affect the results. This is the first violation of Bell's inequalities with massive particles (⁹Be⁺ ions) obtained by use of a complete set of Bell measurements. In addition, the high detection efficiency of the experiments eliminates the detection loophole for the first time. All previous experiments have had detection efficiencies low enough to allow the possibility for the subensemble of detected events to agree with quantum mechanics even though the entire ensemble satisfies Bell's inequalities.

  The experiment prepares a pair of two-level atomic ions in a repeatable configuration. Next a laser field is applied to the particles; the classical manipulation variables are the phases of this field at each ion's position. Finally, upon application of a detection laser beam, the classical property measured is the number of scattered photons emanating from the particles. The Bell signal B was constructed using the results for four sets of phase parameters. Analyses of the photon count distributions indicate that the Bell's signal was B = 2.25 ± 0.03, a result that clearly exceeds 2, the maximum value allowed by local realistic theories of nature. (D. Wineland).

• Improved Evaluation of NIST-F1. S. Jefferts and D. Meekhof have recently reduced the uncertainty of NIST-F1 to 1.7 x 10^-15. An important aspect of their improved evaluation of the standard is their measurement of the spin-exchange frequency shift, which results from cesium-cesium collisions within the clouds of atoms that traverse the standard. Figure 5 shows the relative frequency shift as a function of the clock signal, which is proportional to the density of atoms launched through the clock. The large bunching of points near 100 mV indicates the typical region of operation where the bias due to the spin-exchange shift is about -2 x 10^-15. The uncertainty in the shift determined through extrapolation using this curve is 1.4 x 10^-15. These measurements have been limited by noise in the quartz local oscillator, and substantial improvement should be achieved with the impending arrival of an oscillator with much higher short-term stability. A report on the evaluation of NIST-F1 has been submitted to Metrologia and 3 formal evaluations of the standard have been submitted to the BIPM. (S. Jefferts and D. Meekhof)
  

  Figure 5. Measurements of the frequency of NIST-F1 as a function of signal amplitude. The amplitude is varied by changing the density of atoms launched through the standard. It is the increased spin-exchange collision shift that gives rise to the observed variation, and this type of plot is used to obtain a frequency correction. The shift is measured to be 2 x 10^-15 ± (1.4 x 10^-15) relative to the nominal operating point.

• Stick-Slip Motion of a Stressed Coulomb Crystal. T. Mitchell, J. Bollinger, and W. Itano of the Division, in collaboration with D. Dubin of the University of California at San Diego, have recently identified earthquake-like disturbances in what had previously been considered to be stable nonneutral-plasma crystals. Such a crystal, confined in a Penning trap with its rotation locked to a rotating electric field, has potential applications in atomic frequency standards and quantum logic. The value of such crystals lies in their high-degree of order and stability, so it is important to learn how to control these unexpected motions.

  The experiments demonstrated that the application of a small torque produces sudden angular jumps or 'slips' of the crystal orientation spaced by intervals where the crystal orientation is phase-locked or ‘stuck’ relative to the rotating field. The angular slips follow power-law frequency-versus-amplitude spectra, where the power-law exponent depends on the amount of applied torque and is therefore not universal. Positive correlation is measured between the waiting time between slips and the amplitude of the following slip. (J. Bollinger).

• Comparison of NIST-7 and NIST-F1 Using a Special Time Scale. A post-processed time scale, involving an ensemble of 5 hydrogen masers, has been developed by T. Parker to serve as reference for comparing primary frequency standards. During the last several years, this time scale has been used to evaluate the relative frequencies of NIST-7, an optically pumped cesium-beam standard, and NIST-F1, the cesium-fountain frequency standard. When combined with frequency comparison data attained through various satellite based methods, this time scale also provides for absolute comparisons of the frequencies of NIST primary frequency standards with those of other countries. It is particularly important to note that this scale allows comparison of frequency standards not operated at exactly the same time. The stability of this time scale, called AT1E, is less than _y() = 1 x 10^-15 for averaging times from 1 to 100 days, with a minimum near 3 x 10^-16 at about 20 days.

  Figure 6 shows comparisons of the performance of 6 different primary frequency standards with the scale (and thus with each other) over a period of more than 1200 days. These comparisons indicate that the frequencies of NIST-7 and NIST-F1 have maintained agreement within their measured uncertainties over the last 2 years, thus adding confidence to the methods used for evaluating the uncertainties of the standards. Having completed this overlap of operation with good results, NIST-7 will be taken out of operation.

  

  Figure 6. Comparisons of 6 different standards using a special time scale composed of 5 hydrogen masers. The standards are from NIST, the Physikalisch Technische Bundesanstalt (PTB), and the Laboratoire Primaire du Temps et Frequences (LPTF). The cesium-beam standards are LPTF-JPO, NIST-7, and PTB-CS2. The fountain standards are LPTF-FO1, NIST-F1, and PTB-CSF1. The European results are obtained from satellite frequency comparisons and data obtained from the BIPM.

  The figure also compares the NIST standards with several of the other best standards in the world. These are the standards at the Physikalisch Technische Bundesanstalt (PTB) in Germany and the Laboratoire Primaire du Temps et Frequences (LPTF) in Paris. The measurements show remarkably good agreement among standards of quite different designs indicating that the measurement of frequency at these levels is in very good shape. The more recent comparisons of NIST and PTB standards have been done using a two-way time-transfer link between these two labs. This link supports comparisons at an uncertainty of less than 1 x 10^-15, more than a factor of 2 better than other comparisons, which are made using the common-view Global Positioning System method. The PTB and NIST fountain standards are seen to agree within their uncertainties. (T. Parker)

• Atomic Clock for the International Space Station. The Division is collaborating with the Jet Propulsion Laboratory, the University of Colorado, the Harvard-Smithsonian Center for Astrophysics, and the Politecnico di Torino in the development of a Primary Atomic Reference Clock in Space (PARCS), a NASA-funded program to put a laser-cooled cesium atomic clock in space. Following a successful NASA review of the program in 1999, this collaborating group has initiated the development of preliminary component designs and fabrication of prototypes for some of the components. The objectives of the PARCS mission, scheduled to fly in early 2005, are (1) to test relativity theory by measuring the net (Doppler-plus-gravitational) frequency shift of the cesium clock, (2) to perform a test of local position invariance by comparing the frequency of the cesium clock with that of a hydrogen maser that is part of the PARCS package, (3) to improve upon the realization of the second by taking advantage of the long atom-observation times provided by the microgravity environment, and (4) to study the performance of GPS satellites through comparison of the received GPS clock signals with the PARCS clock.

  Staff members at NIST and the University of Colorado are serving as the Co-Principal Investigators for the experiment, and the Project Scientist and Project Manager are from the Jet Propulsion Laboratory where the space hardware is being developed. A hydrogen maser, originally developed (but not flown) for an earlier space mission by the Harvard-Smithsonian Center for Astrophysics, will serve as both the local oscillator for the cesium clock and reference for the test of local position invariance. The Politecnico di Torino, which has expertise on the microwave components of atomic clocks, is providing input to the design of the microwave cavity for the cesium clock. A block diagram of the PARCS systems is shown in Fig. 7. (D. Sullivan).

  

  Figure 7. Block diagram of the PARCS experiment. Ground components are shaded and labeled. Frequency comparisons between the space and ground clocks are made using a GPS carrier-phase common-view method. The computer controller steers the microwave synthesizer to the cesium resonance. This steering signal in fact represents the frequency difference between the output of the local oscillator (hydrogen maser) and the cesium reference and is used to test local position invariance.

• Improvements in GPS Carrier-Phase Time Transfer. L. Nelson and J. Levine have recently completed a frequency comparison between primary standards at NIST and PTB using both the two-way method and the GPS carrier-phase method. The results indicate a substantial improvement in reliability and noise level for the GPS method. They further showed that the methods agree to about 3 x 10^-15 (for several days of averaging), allowing for comparison of standards to this level. The agreement between the NIST and PTB standards is well within this limit. The noise level for the comparison of transfer methods over a period of 48 days is shown in Fig. 8. This graph shows time comparisons and reflects a slight frequency offset between NIST and PTB time scales. Frequency comparisons are obtained by considering specific time intervals.
  

  Figure 8. Comparison of UTC(NIST) and CS2(PTB) using both the GPS carrier-phase (CP) time-transfer method and the two-way satellite time transfer (TWSTT) method. The measurements span a period of 48 days (days are labeled as the Modified Julian Date). The slope of the curve reflects the small frequency offset between the NIST time scale and the PTB standard. Since the NIST primary standard is regularly compared to UTC(NIST), these data provide the means for comparing the frequencies of the primary standards at the two institutions.

  The present analysis of the GPS carrier-phase data is limited by data frequency steps, which must be carefully reconnected to form the desired time series. Furthermore, there are gaps in the data from the GPS reference stations that limit the length of the time series that can be processed at one time. Thus, elimination or reduction of these problems could further improve the performance of the method. The analysis limitation could be improved through better analysis methods, and studies are now being done using a new analysis package from the University of Berne. (J. Levine).

• Determination of the Relativistic Red Shift in Boulder. In a recently completed project, N. Pavlis of Raytheon Corporation and M. Weiss of the Division have estimated the relativistic red shift correction due to gravity with an uncertainty of 2 parts in 10¹⁷. An accurate estimate of this shift is an important factor in the list of systematic frequency shifts that must be determined to evaluate NIST-F1, the NIST cesium-fountain frequency standard, in fact, this is the largest frequency bias for this standard. This shift will become even more important in the future as yet more accurate frequency standards are developed.

  Three different methods were used to arrive at their estimate. These were (1) a 1998 global gravitational model produced by NASA and the National Imagery and Mapping Agency; (2) a 1999 regional, high-resolution geoid model generated by Smith and Milbert; and (3) recent measurements made by the National Geodetic Survey of a reference marker on the NIST site. Through a critical analysis of these three methods they estimate that the frequency correction for NIST-F1 in its current location is -1798.93 x 10^-16, with an estimated uncertainty of 0.2 x 10^-16. This is well below the NIST-F1 uncertainty of 1.5 x 10^-15, so at this time the red-shift correction is not of particular concern. It is worth noting that when the Division moves this standard to a different floor of the building, which should happen in about a year, the correction will have to be changed. (M. Weiss).

• Transverse Cooling for Fountain Frequency Standards. In a collaborative theoretical effort, A.V. Taichenachev, A.M. Tumaikin and V.I. Yudin of Novosibirsk State University and L Hollberg of the Division have developed a concept for two-dimensional sideband Raman cooling and Zeeman state preparation in an optical lattice. The development of this concept was driven by the need to reduce transverse velocities of laser-cooled atoms in primary cesium-fountain frequency standards. The approach suggested in their recent paper appears to be both simpler and more effective than previously proposed methods and has two significant advantages: the method only requires laser beams transverse to the atomic fountain axis, and the cooling process simultaneously pumps atoms into the desired (m=0) ground-state energy level. The theory predicts that 95 % of the atoms can be cooled to transverse temperatures of about 100 nK. An experimental apparatus designed to test the effectiveness of this cooling method is now under development within the Division.

  Further transverse cooling will provide for significant reduction in the uncertainty of NIST's cesium-fountain frequency standard. The largest uncertainty in this standard is the spin-exchange frequency shift, the magnitude of which depends on the density of atoms averaged over the fountain trajectory. In this standard, the magnitude of the transverse velocities of the launched atoms (cooled to about 1.5 µK) results in a loss of approximately 90 % of the atoms before they reach the detection region. This means that most of these atoms contribute to an increased spin-exchange frequency shift without contributing to the output signal. With better transverse cooling, a larger fraction of the atoms will reach the detection region and the launch density can be decreased leading to a lower spin-exchange shift while retaining a good signal-to-noise ratio. (L. Hollberg).

• Comparing Frequency Standards Through an Optical-Fiber Network. In a joint program among the City of Boulder, the Department of Commerce (DOC), the University of Colorado (UC), and the National Center for Atmospheric Research, an optical-fiber network has been installed connecting various organizations dispersed throughout the city. The network acronym is BRAN, meaning Boulder Research and Administrative Network. A number of network fibers have been assigned to NIST. Of particular relevance is the assignment of a dark pair (no included optoelectronic interfaces) of fibers connecting optical systems in the Time and Frequency Division to systems within the Quantum Physics Division on the UC campus. The distance (along the fiber) between these two sites is approximately 3.5 km. Exceptional microwave and optical frequency standards are located in these two Divisions providing the opportunity to study the performance of this direct fiber connection. The long list of potential studies that can be done using this network includes: (1) comparison of microwave atomic clocks and frequency standards and transfer of absolute time, (2) transfer of optical atomic-clock signals at optical frequencies, (3) accurate transfer of microwave frequencies, (4) calibration and transfer of laser wavelengths at wavelengths important to wave-division multiplexing, (5) dissemination of optical frequencies used as references for measuring length, and (6) distribution of high-accuracy pulse trains used for time-domain measurements.

  In preliminary experiments, J. Kitching and L. Hollberg of the Division demonstrated high-short-term-stability transfer of frequency across the network. They first modulated the output of a 1.3 µm laser at 2.3 GHz using a source locked to a hydrogen maser in the NIST time scale. This was transferred to the University (over BRAN) and then back to the Division. Comparison with the original signal produces the st ability results shown in Fig. 9. The measurement-system noise floor, shown as the dotted line in the figure, limited the results of the measurements. The uncompensated diurnal variations of the time delay between the two sites were about 140 ps, and these can be controlled well below this level. (L. Hollberg).

  Figure 9. Stability (Allan deviation) of the round-trip (7 km) link between the Division and the University as a function of averaging time. The dashed line is an estimate of the noise floor of the measurement system, so it is clear that the noise in the link is lower than that indicated.

Compact Cesium Frequency Standard Driven by a Diode Laser. J. Kitching and L. Hollberg, in collaboration with guest researchers Robert Wynands and Svenja Knappe, have developed a small, 4.6 GHz frequency standard using a vertical-cavity surface-emitting laser (VCSEL) that pumps a cesium atomic vapor in a small cell structure. Design information developed in this program indicates that the package for the standard can be as small as 10 mm x 10 mm x 20 mm, and that the system might operate at a power as low as 100 mW. A block diagram of this device is shown in Fig. 10. This standard has potential applications in areas such as wireless telecommunications where good synchronization is needed to assure efficient data transfer between network nodes. In these systems, it is particularly important to maintain reference timing at each node, even when external synchronization is lost. Thus, the industry is searching for comp act, high-stability oscillators that can meet this "holdover" requirement. The timing uncertainty of these devices is currently better than 10 µs over 1 day, which meets the industry requirements. Current research efforts are aimed at understanding the fundamental physics of operation of these standards and further improvement of their performance. (L. Hollberg).

Figure 10. Block diagram of the compact cesium standard

• Enhancements to the NIST Frequency Measurement Service. M. Lombardi and A. Novick have completed a set of enhancements to the Frequency Measurement Service resulting in substantial improvement in measurement uncertainty and flexibility. The measurement uncertainty has been reduced to 2 x 10^-13 (2) for a 24 hour averaging period, and the system can now measure any frequency from 1 Hz to 120 MHz in 1 Hz increments. Up to 5 devices can be calibrated simultaneously. Su bscribers to the service receive monthly calibration reports compliant with ISO Guides 25 and 17025 and the ANSI Z-540 standard. A wide range of high-level calibration laboratories in industrial and government organizations use this service, which provides continuous frequency traceability to NIST.

  To further simplify the problem of achieving frequency traceability, a database of frequency comparisons between the NIST time scale and individual GPS satellites has been developed and is updated daily on the Division web site. While it is generally acknowledged that the frequency delivered by these satellites is quite accurate, the mode of operation of GPS does not allow for a clear specification of frequency accuracy that can be used for legal traceability. This database provides a very simple means for a calibration laboratory to achieve measurement traceability to NIST. (M. Lombardi).

A New Web Site for Official Time. With assistance from M. Douma, an independent contractor, M. Lombardi and A. Novick of the Division have developed a web site (http://time.gov), which provides official United States time with an uncertainty of less than 1 second in a format that can be widely appreciated by non-technical users. The time distributed by this site is considered traceable to both NIST and the U.S. Naval Observatory, although NIST operates the site for both agencies. The format (Fig. 11) allows the user to select any U.S. time zone from a map and then initiate operation of a Java clock that runs in the local time zone. During October, this site received 3.7 million hits from more than 400,000 distinct IP addresses. (A. Novick) New Web Page for the Time and Frequency Division. The home page for the Division has been redesigned and expanded through the efforts of A. Novick and M. Lombardi of the Division. The site, at http://www.boulder.nist.gov/timefreq, features an easy-to-use interface with a sidebar menu that is common to every page. New and expanded content includes: a time-scale data archive, a section on the new cesium-fountain frequency standard, a section on GPS carrier-phase time transfer, an expansion of the sections on the radio broadcast services, and an expansion of the sections on computer time synchronization. Current usage of this web site averages 164,000 page views per month. (M. Lombardi).

Figure 11. The two key pages for the new time.gov web site.

• Further Expansion of the Network Time Service. J. Levine of the Division has been managing an ever-increasing activity with the Division's Network Time Service. Over the last year, the volume of daily hits on this time service have increased by about a factor of 4 to nearly 4 x 10⁷ per day. During the short period of Y2K rollover, the service level was nearly 2.4 x 10⁸ requests. To meet this increasing demand, he is installing new servers in Virginia and California to complement the 12 servers already in the system.

  A rapid rise in commercial interest has been associated with this service. Two companies have established cooperative arrangements with NIST and interactions with a third are being discussed. Six servers have been configured by NIST and delivered to one of these companies to become part of a commercial authenticated time service. In cooperation with another of these companies, real-time tests between Boulder and San Jose were conducted of remote synchronization of a time server. The third interaction, still in the paper design stage, involves yet another network for authenticated-time delivery. (J. Levine).

• Replacement of Antennas at WWVH. During the last year, the Division initiated the process of replacing all of the HF antennas for radio station WWVH in Kauai. The salt-water environment has been causing substantial corrosion of the present antennas, which are now nearly 30 years old. Despite an aggressive program of regular cleaning and repainting, the WWVH antennas are becoming structurally weakened, resulting in serious risk to those who must regularly scale the antennas for maintenance. D. Okayama and D. Patterson at the station and W. Hanson in Boulder are leading work on this project.

  The antennas are being replaced with fiberglass whip antennas, which were developed especially for marine environments. Two broadband antennas are now being installed to replace units first put into service in 1971. These new antennas are hinged and can be lowered to the ground for maintenance, eliminating the need for further tower scaling. It is estimated that the capital investment in these new antennas will be recovered (in the form of reduced maintenance costs) in approximately 7 years. (W. Hanson).

• Phase and Amplitude Noise Measurements Between 10 GHz and 100 GHz. D. Howe and F. Walls have developed a system for making high-resolution phase-and-amplitude-noise measurements in the frequency range from 10 GHz to 100 GHz to support the characterization of high-performance radars, which use digital methods for processing return radar signals. The performance of these radar systems is critically dependent on noise in the systems. The new measurement system uses the two-channel cross-correlation method. The system reference for the measurements is provided by any of the lines in a comb of extremely stable reference frequencies produced by the combination of an ultra-stable, cooled-sapphire resonator (oscillator) and a set of low-noise regenerative dividers. The stability of the reference signals is in fact the key to making these measurements. To date there have been no high-stability reference sources for measurements across this frequency region. (D. Howe).