Minutes of the CCL-CCTF Frequency Standards Working Group Meeting
BIPM, 10th – 11th September 2012
Agenda:
1.
2.
3.
4.
5.
6.
7.
8.
9.
Welcome to attendees.
Approval of the agenda, appointment of rapporteur(s).
Consideration of returns from NMIs to WG questionnaire.
Consideration of additions/updates to the revised single list of frequencies including
secondary representations of the second and frequencies for the realisation of the definition of
the metre.
Report on key comparison CCL-K11.
On optical frequency ratios.
Status of the publication of the list on the web and in Metrologia
Preparation of WG recommendations to CCL and CCTF.
Other business.
1.
Welcome to attendees.
Dr Riehle opened the meeting by welcoming the delegates, the observers and the BIPM
representatives to the meeting.
The delegates and observers briefly introduced themselves.
2.
Approval of the agenda, appointment of rapporteur(s).
Dr Riehle presented the proposed agenda for the meeting (CCL-CCTF/12-01), which was
approved by the delegates.
Dr Margolis was appointed as rapporteur for the meeting.
3.
Consideration of returns from NMIs to WG questionnaire.
Dr Robertsson presented an overview of the summary of the responses to the WG
questionnaire. Several proposals had been made for updates to existing values and/or
uncertainties in the list, as well as for the inclusion of new values and for new secondary
representations of the second. Dr Robertsson had developed a series of spreadsheets as a
starting point for the discussion, one for each wavelength to be considered. These
spreadsheets included preliminary calculations based on the results from each laboratory,
including different possibilities for weighting and expansion of uncertainty.
4.
Consideration of additions/updates to the revised single list of frequencies including
secondary representations of the second and frequencies for the realisation of the
definition of the metre.
6.8 GHz ground-state hyperfine transition in 87Rb
Discussion of this secondary representation began with a presentation from Dr Petit of BIPM,
based partly on a presentation he gave at CPEM 2012. The LNE-SYRTE group had carried
out 13 evaluations of their Rb fountain, which were published in June 2012. These were
1
submitted for review to the CCTF WGPFS, in the same way as for a new primary frequency
standard. Three methods were used to estimate the frequency of the Rb standard:
a) Local comparisons to the SYRTE Cs fountains (suggest frequency is lower than
recommended value by 1.48×1015).
b) Comparison to TT(BIPM11) (suggests frequency is lower than recommended value by
1.67×1015 but uncertainty is difficult to estimate).
c) Comparison to the best estimate of the worldwide ensemble of primary frequency
standards (suggests frequency is lower than recommended value by 1.67×1015 , and
gives less dispersed values than method b).
Dr Arias reminded the delegates that there should be two steps in the evaluation of secondary
representations of the second. In the first step, the FSWG decides on the frequency value and
uncertainty for each secondary representation of the second included in the list of
recommended frequencies. In the second step, reports should be submitted to the WGPFS
(expected soon to become the WG on primary and secondary frequency standards). In the first
instance the secondary standards would not be used for steering TAI; instead, the reporting
mechanism provides a means of validating the value and uncertainty in the list.
Since more data had been reported to BIPM since the analysis presented, Dr Petit agreed to
process the additional data and to provide an updated value to the working group. Further
discussion of this frequency value was therefore postponed until the second day of the
meeting. However the additional data was found to make very little difference to the best
estimate of the Rb frequency value (1.74×1015 lower than the recommended value rather than
1.67×1015).
After discussion, it was agreed to use a weighted mean of the results published by SYRTE
and to expand the uncertainty by a factor of three taking into account that the value is based
on measurements from a single institute (although it was noted that a measurement campaign
was currently underway at NPL). Dr Petit was thanked for his detailed analysis that provides
additional confidence in this value and uncertainty.
The recommended frequency to be put forward to the CCTF was therefore
f87Rb = 6 834 682 610.904 312 Hz
with an estimated relative standard uncertainty of 1.3×1015. This radiation is already
endorsed by the CIPM as a secondary representation of the second.
1H,
1S–2S two-photon transition (243 nm)
Measurements of this transition frequency have only been reported by a single group (at
MPQ). Previously the published uncertainty was expanded by nearly a factor of ten, because
it did not reflect the scatter of the observed data. Two new measurements had been carried
out, one using the SYRTE transportable Cs fountain, the other using an optical link to PTB.
However the second was not yet published.
After some discussion about whether the expansion factor could be reduced, a sub-group was
delegated to review the MPQ publications and to report back on the second day of the
meeting with a recommendation for the frequency value and uncertainty that should be used.
This sub-group, after reviewing the published work, considered that there were some
questions about systematic frequency shifts and parameter sensitivities arising in the 1997 and
2000 measurements, discussed by the MPQ group themselves although not explicitly in
refereed publications. The 2010 measurement with the SYRTE fountain used a new laser
2
system which eliminated previous problems with laser instabilities and reduced the scatter of
the data. A proper measurement of the atom velocity distribution was also carried out in this
case. Based on the lower scatter and the more complete uncertainty analysis the sub-group
therefore had a higher level of confidence in the most recent data.
It was therefore agreed to use only the most recent published measurement and to expand the
uncertainty by a factor of three since there was only a measurement from one laboratory. It
was noted that there were no significant discrepancies between the values obtained over the
years, giving extra confidence that the reduced expansion factor was reasonable.
The recommended frequency to be put forward to the CCTF was therefore
f1H = 1 233 030 706 593 518 Hz
with an estimated relative standard uncertainty of 1.2×1014. It was noted that, since the 1S –
2S transition is a two-photon transition, this frequency corresponds to half of the energy
difference between the 1S and 2S states.
171Yb+,
6s 2S1/2 (F=0, mF=0) – 5d 2D3/2 (F=2,mF=0) transition (436 nm)
New measurements of this transition frequency had been reported by both NPL and PTB
since the last meeting of the FSWG. These are in good agreement but the uncertainty of the
NPL measurement (the first of this transition at NPL) is significantly larger than that of the
PTB value. PTB have made an extensive series of measurements over more than one decade.
All are consistent within their uncertainties, but PTB’s more recent (and more accurate)
values appear to be more than 1 Hz below the previous recommended value. It was therefore
agreed that the value and uncertainty of this transition frequency should be updated in the list.
It was agreed to use a weighted mean of the values from the two laboratories. Since the
uncertainty of the NPL value was more than ten times larger than that of the PTB value, it was
also agreed to expand the uncertainty of the weighted mean by a factor of three. It was,
however, noted that two Yb+ standards had been compared at PTB and agreed at the parts in
1016 level.
The recommended frequency to be put forward to the CCTF was therefore
f171Yb+ (quadrupole) = 688 358 979 309 307.1 Hz
with an estimated relative standard uncertainty of 3×1015. This radiation is already endorsed
by the CIPM as a secondary representation of the second.
171Yb+, 2S
1/2
(F=0, mF=0) – 2F7/2 (F=3, mF=0) transition (467 nm)
The 2009 recommended value for this frequency was based only on measurements carried out
at NPL, with the uncertainty expanded by a factor of three. Since then much more accurate
measurements have been carried out by both PTB and NPL, and these are in good agreement.
It was therefore agreed that the value and uncertainty of this transition frequency should be
updated in the list.
It was agreed to use the weighted mean of all available values (including old data) and to
enlarge the uncertainty by a factor of two since the values are from only two laboratories.
The recommended frequency to be put forward to the CCTF was therefore
f171Yb+ (octupole) = 642 121 496 772 645.6 Hz
with an estimated relative standard uncertainty of 1.3×1015.
3
It was agreed that this radiation should be put forward to the CCTF as a secondary
representation of the second.
171Yb
neutral atom, 6s2 1S0 – 6s6p 3P0 transition (578 nm)
Accurate measurements of this transition frequency are now available from both NIST and
NMIJ (accepted for publication) and are in good agreement. A measurement has also been
made at KRISS, but this value is not yet published. It was noted that a number of other groups
are also developing ytterbium lattice clocks, for example in Italy, Germany, China and Japan.
It was agreed to use the weighted mean of the published values and to enlarge the uncertainty
by a factor of two since these values are from only two laboratories.
The recommended frequency to be put forward to the CCTF was therefore
f171Yb = 518 295 836 590 865.0 Hz
with an estimated relative standard uncertainty of 2.7×1015.
It was agreed that this radiation should be put forward to the CCTF as a secondary
representation of the second.
88Sr+,
5s 2S1/2 – 4d 2D5/2 transition (674 nm)
A new measurement of this transition frequency had recently been made at NRC
(measurement campaign in February – May 2012), and had been accepted for publication in
PRL, subject to corrections (not affecting the frequency value). The main uncertainty was in
the link to the SI second, which was realised through GPS common view transfer. The
measured frequency is consistent with the previous NPL measurement.
It was agreed to use the weighted mean of the NRC and NPL values and to enlarge the
uncertainty by a factor of two since values are only available from two laboratories.
The recommended frequency to be put forward to the CCTF was therefore
f88Sr+ = 444 779 044 095 485.3 Hz
with an estimated relative standard uncertainty of 4.0×1015. This radiation is already
endorsed by the CIPM as a secondary representation of the second.
87Sr
neutral atom, 5s2 1S0 – 5s5p 3P0 transition (698 nm)
Three new measurements of this transition frequency had been reported, from NICT, PTB and
LNE-SYRTE. However the LNE-SYRTE measurement had not yet been published in a peerreviewed journal and so could not be used to contribute to an updated value according to the
FSWG rules.
It was agreed to use the weighted mean of the published values. No expansion factor would be
required according to the usual procedures of the FSWG. However it was agreed that it would
be unwise to reduce the previous uncertainty quoted in the list, bearing in mind that the latest
PTB and NICT values differ by 1 Hz, and that recent measurements at PTB (not yet accepted
for publication) suggest that the calculated value for the blackbody Stark shift (used by all
groups) may have to be corrected.
The recommended frequency to be put forward to the CCTF was therefore
f87Sr = 429 228 004 229 873.4 Hz
4
with an estimated relative standard uncertainty of 1.0×1015. This radiation is already
endorsed by the CIPM as a secondary representation of the second.
40Ca+,
4s 2S1/2 – 3d 2D5/2 transition (729 nm)
New frequency measurements had been reported from NICT and Wuhan, to add to the
previous results from Innsbruck and NICT. The new NICT measurement has an uncertainty of
2.9×1015, but disagrees with the Wuhan measurement by 1.3×1014 (5.3 Hz). The Innsbruck
value agrees with the Wuhan measurement. NICT have also done an optical frequency ratio
measurement between their Ca+ standard and Sr, which confirms the result of the absolute
frequency measurement. A direct comparison of the Wuhan and NICT Ca+ standards has
recently been carried out via GPS, and a difference of several Hz was observed, and so it does
appear that the frequencies of the Ca+ standards are different.
Despite the discrepancies between the different values, the overall scatter of the values was
8 Hz, smaller than the previously recommended uncertainty. It was therefore agreed that the
frequency value and uncertainty should be updated to reflect this additional uncertainty.
However an unweighted mean of the available values would have to be used since some or all
of the estimated uncertainties could not be trusted. After some discussion, an expansion factor
of two for the uncertainty was agreed upon.
The recommended frequency to be put forward to the CCTF was therefore
f40Ca+ = 411 042 129 776 395 Hz
with an estimated relative standard uncertainty of 1.5×1014.
27Al+ 1S
0
– 3P0 transition (267 nm)
Although NIST had not made a specific request for the Al+ standard to be included either in
the list of recommended frequencies or as a secondary representation of the second, it was felt
by a number of delegates that it should not be left out. Dr Riehle consulted Dr Oates by
telephone to ask what the NIST position was. After consulting his colleagues, Dr Oates
confirmed that NIST would be happy for Al+ to be included in the list, and would not be
against it being recommended as a secondary representation of the second.
The NIST group has made one direct absolute frequency measurement of the Al+ clock
transition, but with relatively high uncertainty. Their more accurate measurement is obtained
from an Al+/Hg+ optical frequency ratio measurement, combined with the Hg+ absolute
frequency measurement. It was noted that NIST has also directly compared two Al+ optical
clocks, observing agreement at the 1.7×1017 level (statistical uncertainty 7×1018).
It was agreed to use the weighted mean of the two NIST absolute frequency values and to
expand the uncertainty by a factor of three because the frequency value was derived from
measurements from a single laboratory.
The recommended frequency to be put forward to the CCTF was therefore
f27Al + = 1 121 015 393 207 857.3 Hz
with an estimated relative standard uncertainty of 1.9×1015.
It was agreed that this radiation should be put forward to the CCTF as a secondary
representation of the second.
It was also noted that the procedure followed for the Al+ frequency meant that the Hg+
frequency value and uncertainty should also be updated for consistency.
5
199Hg+,
5d106s 2S1/2 (F=0) – 5d96s2 2D5/2 (F=2) transition (282 nm)
The value used by the NIST group for the Hg+ frequency in the Al+ frequency measurement
was taken from their 2007 Applied Physics B publication on Hg+. This includes all available
absolute frequency ratio measurements, spanning the period 2000 – 2006. It was therefore
agreed to use this value and to expand the uncertainty by a factor of three because all
frequency measurements were from the same laboratory.
The recommended frequency to be put forward to the CCTF was therefore
f199Hg+ = 1 064 721 609 899 145.3 Hz
with an estimated relative standard uncertainty of 1.9×1015.
199Hg
neutral atom, 6s2 1S0 – 6s6p 3P0 transition (266 nm)
A new measurement of this transition frequency had been made by SYRTE. This was the first
carried out in a lattice configuration, and had an uncertainty of 6.4 Hz, four orders of
magnitude better than their initial measurement.
It was agreed that this value should be included in the list of recommended frequencies, but
that at this early stage the uncertainty was too high for it to be recommended as a secondary
representation of the second. A factor of three expansion factor for the uncertainty was
agreed, since there was only a single measurement available.
The recommended frequency to be put forward to the CCTF was therefore
f199Hg = 1 128 575 290 808 162 Hz
with an estimated relative standard uncertainty of 1.7×1014.
5.
Report on key comparison CCL-K11.
Dr Gill gave a short presentation incorporating the conclusions of a report from Dr Matus,
who was unable to attend the meeting.
A gradual divergence in the approach being adopted by different laboratories was noted.
Some are continuing to operate their iodine-stabilised lasers within the operating parameters
specified in the MeP, whilst others are moving towards using the lasers as independent
frequency standards which are calibrated using frequency combs. In the second case,
participants choose not to correct the laser frequency for operational parameters, relying
instead on the stability of the setups. This can give smaller uncertainties. Both approaches are
acceptable, and the CCL-K11 key comparison is supposed to show the equivalence between
the measurement capabilities of different NMIs.
The work undertaken by a sub-group to revise the CCL-K11 protocol to include comb-based
calibrations of laser frequencies was reviewed. The draft protocol was available on the FSWG
document site, although there were still some typographical errors to be corrected. Most of the
discussion centred around the situation involving calibrations using a comb referenced to a
suitable frequency reference, with uncertainty significantly better than a few parts in 1011, and
whether self-confirmation via peer-reviewed publication was sufficient for demonstration of
capability for key comparison and associated CMC statements. It was considered that the
peer-reviewed publication should demonstrate a high accuracy frequency measurement of a
particular system, preferably one that had also been measured by a different NMI.
6
Dr Madej gave an example of validation of a comb system at NIST Gaithersburg. This uses a
GPS-disciplined oscillator as a frequency reference, which was validated by measuring the
frequency of the same laser system at Gaithersburg, at NRC, and finally at Gaithersburg
again. The reproducibility was better than 1 kHz, demonstrating measurement capability at
this level, and the results were summarized in a peer-reviewed paper.
It was agreed that the point discussing peer-reviewed publication should be modified to
suggest examples of the type of evidence that could be presented to demonstrate measurement
capability. The activity to revise the protocol would then be considered completed.
Dr Madej, Dr Merimaa and Dr Gill confirmed that their institutes were happy to continue as
node laboratories for the key comparison. Dr Gill would check the situation regarding BEV.
Dr Hong stated that NMIJ would be prepared to continue, but would be happy to hand over
the role if another laboratory wanted to take over. Dr Warrington pointed out that there was
provision in the protocol for a laboratory that is not a node to act as a host, although nobody
has yet taken this up as an option. This might help to reduce the load on node laboratories.
6.
On optical frequency ratios.
At the last meeting of the FSWG, a sub-group headed by NIST was formed to look at how to
include extra information available from optical frequency ratio measurements. A
presentation had been prepared by Dr Oates, but since he had got stuck in Chicago on his way
to the meeting this was presented by Dr Riehle.
A number of optical frequency standards now have reported uncertainty budgets lower than
those of primary frequency standards. To move forward it is therefore necessary to make
optical-optical comparisons, which can be classified into two types: intra-species comparisons
(e.g. Sr – Sr or Al+ – Al+) and inter-species comparisons (e.g. Sr – Yb or Hg+ – Al+). The
suggestion was that a matrix of frequency ratio measurements (including Cs) could be
appended to the end of the reported absolute frequencies. As the frequency ratios accumulate
it will be possible to make interesting consistency tests. It will also be possible to create
“synthesised frequencies” to fill in gaps in the frequency matrix. In the longer term there will
be an issue of how to use these (hopefully overdetermined) matrices to perform meaningful
statistical tests on the standards, to enable the international community to decide which
systems can serve as secondary representations of the second and which could be a basis for a
redefinition of the second. Dr Riehle pointed out that the problem is similar in some respects
to analysis of matrices from iodine-stabilised lasers to work out frequency differences,
although there may be problems with numerical stability because of the large numbers
involved.
Dr Madej and Dr Bize both expressed their support for retaining Cs in the matrix. Dr Bize
favoured a scheme in which original ratios are kept rather than synthetic frequencies.
Dr Margolis gave a short presentation on a proposed EMRP project “International timescales
with optical clocks”. If successful, this would involve a coordinated programme of
intercomparisons between high accuracy optical atomic clocks in Europe, and an analysis of
the resulting frequency ratio matrix using methods similar to that used in the CODATA leastsquares adjustment of the fundamental physical constants. The involvement of the
international community was encouraged, particularly with regards to contributing additional
frequency ratio measurements to the matrix analysis. The experience gained within this
project would be expected to provide input into future meetings of the FSWG.
7
7.
Status of the publication of the list on the web and in Metrologia
Dr Riehle gave a presentation on the current status of the publication of the list of
recommended frequencies. The values agreed in 2009 have been made available on the BIPM
website by Dr Miles (Metrologia).
Due to the large number of modifications to the list, and the new function of the list, it was
previously agreed that a complete list should be published in Metrologia, in a form similar to
that of the last publication in 2003. Dr Miles had indicated that she would like to receive such
a manuscript. Much of the paper has been drafted but since some parts took much longer than
anticipated, it was decided to postpone the submission of the paper until after this meeting. It
was expected that a final manuscript can be submitted by the end of 2012.
Dr Madej pointed out that source data is not cited for all values currently on the website.
Recently modified values simply refer to reports of the CCTF. Dr Riehle confirmed that the
intention was to include the original source data, and asked WG members to check the list and
send him any information that is missing.
8.
Preparation of WG recommendations to CCL and CCTF.
It was agreed that a report would be prepared by Dr Riehle and Dr Gill for the CCTF on
13th/14th September. This would include information about the activities of the FSWG,
including meetings, publication of the list of recommended frequency values, a discussion
forum on the inclusion of optical frequency ratios, and development of a protocol for
traceability of the metre directly from caesium clocks. Recommendations would be made to
the CCTF for updates to and new entries in the list of recommended frequencies, as agreed
under agenda item 4. Recommendations would be made for three new secondary
representations of the second (171Yb+ E3, 171Yb, 27Al+). No new radiations would be
recommended for realisation of the definition of the metre. The report would also include
information about future activities, i.e. publication of the list of frequencies in Metrologia.
9.
Other business.
Mr Erard, president of the CCTF, gave a short report about the new CCTF strategic planning
working group and its activities. This is chaired by the president of the CCTF and the
membership is the chairs of the CCTF working groups, the director of the BIPM time
department, plus any additional experts proposed by the chair. Three meetings of this working
group have been held (December 2010 at INRIM, May 2011 at EFTF in San Francisco, April
2012 at EFTF in Gothenburg), and it has two main activities: to propose a strategy plan to the
CIPM and to review the activities of the time department at BIPM. Currently it is preparing a
strategic planning document following the directives given by the CGPM in November 2011.
This should be ready by the end of January 2013. Mr Erard gave an overview of the contents
of the draft document, which includes sections on the configuration of the CCTF working
groups, stakeholders and their level of involvement, required key comparisons and resource
implications, etc., as well as a future scan covering the period 2013 – 2023.
It was agreed to schedule a short meeting of the FSWG at the joint EFTF/IFCS conference to
be held in Prague from 21st – 25th July 2013, since a substantial fraction of the working group
members are likely to be attending the conference.
It was agreed that the minutes of this meeting would be circulated to all present once they had
been reviewed by the FSWG co-chairs.
8