Quantum Electronics  48 (5)  425 – 430  (2018)

© 2018  Kvantovaya Elektronika and Turpion Ltd

https://doi.org/10.1070/QEL16654

On the thermal noise limit of ultrastable optical cavities

N.O. Zhadnov, K.S. Kudeyarov, D.S. Kryuchkov,  
I.A. Semerikov, K.Yu. Khabarova, N.N. Kolachevsky

Abstract.  We consider two methods of lowering thermal noise, attaining precision and stability at a level of 10–18 and better
which limits the attainable frequency stability of lasers stabilised to calls for a more efficient stabilisation of laser radiation fre-
a high-Q external Fabry – Perot cavity. These are cavity lengthen- quency [8].
ing and use of high-Q mechanical materials in the production of its The key element of an ultrastable laser system is a passive
constituent parts. The results of numerical simulation are outlined monolithic high-Q Fabry – Perot cavity with an as high as
for the horizontal and vertical suspension systems of the cavity possible stability of the mirror separation. The frequency of
body and its sensitivity to vibrations in these cases. The progress in the laser light is stabilised to the cavity transmittance peak
the development of ultrastable laser systems using cryogenic silicon with the use of a feedback loop. The most frequently used
cavities is also discussed. method of frequency stabilisation to external cavity mode is
the phase-modulation Pound – Drever – Hall technique [9],
Keywords: ultrastable cavities, thermal noise, vibrational sensitiv- which has proved itself to be advantageous. To date, the best
ity of cavities. fractional instability of the laser frequency over an averaging
time of 1 – 100 s amounts to ~10–17, which was obtained using
1. Introduction a silicon cryogenic cavity [10] and a ULE-glass cavity of
length ~0.5 m [11].
The development of laser sources with a spectral linewidth In this work we consider the use of such cavities in ultrast-
less than 1 Hz is a major avenue in the modern physics of able laser systems. In Sections 2 and 3 we discuss the thermal
high-precision measurements. Initially aimed at solving the noise limit, which limits the cavity stability, and the methods
tasks of precision spectroscopy, this research largely paved for lowering this fundamental limit. The latest results in the
the way to the progress in the area of frequency standards development of ultrastable lasers for an optical frequency
and lent impetus to the development of new areas, such as standard based on strontium atoms are outlined in Sections 4
gravitational wave detection [1], studies of interparticle and 5.
interactions in quantum gases [2], and the development of
femtosecond optical frequency combs [3]. Ultrastable lasers 2. Thermal noise and cavity stability
are employed for the spectroscopy of ultranarrow clock
transitions in optical frequency standards based on atomic The stability of the frequency n of Fabry – Perot cavity eigen-
ensembles [4] and single atoms [5]. On a shorter averaging modes is determined primarily by the stability of its mirror
time than the time of particle ensemble preparation and separation Lt :
measurement, the optical clock stability is completely deter-
mined by the frequency stability of laser radiation. On a lon- Dn DLt , (1)
ger time, the clock stability may be limited by the Dick effect n = - Lt
[6], which is also determined by the level of laser noise. For
a frequency instability of laser radiation of ~10–15, the con- The relative displacement of the cavity mirrors may result
tribution made by the Dick effect may far exceed the funda- from the fluctuations of the cavity body length due to tem-
mental limit imposed by quantum noise [7]. That is why the perature variations or vibrations, but their effect may be
further progress of optical frequency standards aimed at effectively suppressed by isolating the cavity from external
perturbation factors. The fundamental limitation on the sta-
N.O. Zhadnov, K.S. Kudeyarov, D.S. Kryuchkov, I.A. Semerikov, bility of mirror separation is imposed by the thermal motion
K.Yu. Khabarova P.N. Lebedev Physical Institute, Russian Academy of interferometer body particles (thermal noise).
of Sciences, Leninsky prosp. 53, 119991 Moscow, Russia; In formula (1), Lt is not the absolute distance between the
e-mail: [email protected]; geometric centres of two mirrors, which is impossible to stabi-
N.N. Kolachevsky P.N. Lebedev Physical Institute, Russian Academy
of Sciences, Leninsky prosp. 53, 119991 Moscow, Russia; All-Russia
lise with an accuracy of 10–15, but the generalised coordinate,
Research Institute of Physical and Radio Engineering Measurements which characterises the cavity length. It may be shown [12]
(VNIIFTRI), 141570 Mendeleevo, Moscow region, Russia; Russian that the phase shift of a laser beam with a radiation intensity
Quantum Centre, ul. Novaya 100, Skolkovo, 143025 Moscow, Russia profile g(r) in its reflection from the mirror surface distorted
by thermal vibrations is expressed as
Received 21 February 2018; revision received 16 March 2018
Kvantovaya Elektronika  48 (5) 425 – 430 (2018)
Translated by E.N. Ragozin
D f( t) = yS g( r) ku( r, t) d 2 r , (2)
426 N.O. Zhadnov, K.S. Kudeyarov, D.S. Kryuchkov, et al.

Here, the integral is taken over the mirror surface S; r is the expansion coefficient of the material is equal to zero. This
position vector of a point on the mirror surface; u(r, t) is the minimises the responsivity of the interferometer intermir-
displacement of the mirror point with coordinate r from its ror distance to temperature fluctuations. Collected in
unperturbed position; k is the wave vector of the incident Table 1 are the limiting relative instabilities of the mode
radiation aligned with the z axis, which is perpendicular to the frequency calculated for different materials and cavity
mirror surface. By using the phase shift Df(t), it is possible to lengths. One can see that the contribution from the cavity
introduce the effective surface displacement along the z axis body is insignificant for almost all calculated configura-
for one of the mirrors: tions and that the stability is primarily determined by the
thermal noise of the mirrors.
Df( t)
Uz ( t) =
|k |
= yS g( r) uz ( r, t) d 2 r . (3)
3. Lowering of the frequency instability limit
Although the cavity mirror is a complex object with a multi-
caused by thermal noise
tude of degrees of freedom, its effect on the phase of laser Among the ways of lowering the thermal noise, the following
radiation may be characterised by one generalised coordinate method should be mentioned : lowering the amplitude of the
(3). The power spectral density of the thermal fluctuations of thermal noise itself and weakening the effect of thermal noise
the quantity Uz(t) may be calculated using the fluctuation- on the stability of cavity length.
dissipative theorem and the approach introduced in Ref. [13]. The amplitude of thermal vibrations is determined by the
The formulas for the power spectral density of the thermal specific amount of heat contained in a system and its dissipa-
noise of the mirrors, Gm, and the cavity body, Gsp, are given tion power [13]. The amplitude may be lowered by lowering
in Ref. [14]: the operating cavity temperature and using high-Q mechani-
cal materials – crystalline sapphire and silicon, fused silica,
2
4kB T 1 - ssub 1 etc. – for making its components [10, 11, 16]. Also showing
Gm ( f ) =
2p f p Esub w0 Qsub promise for this purpose is a change-over from the commonly
employed high-reflectivity SiO2 /Ta2O5 coatings to the crystal-
2 1 - 2ssub Qsub D , line AlGaAs/GaAs ones, which possess a higher mechanical
# c1 + m (4)
p 1 - ssub Qcoat w0 Q-factor [17].
One way to suppress the effect of thermal fluctuations on
the cavity length is to increase the distance between the cavity
4kB T L 1 ,
Gsp ( f ) = (5) mirrors. First, the relative frequency instability decreases with
2p f 3pR 2 Esp Qsp
increasing length in accordance with formula (6). Second, in a
longer cavity the TEM00 spot radius on the mirrors is longer
where E is the Young modulus; Q is the mechanical Q-factor; and is defined by the expressions:
s is the Poisson coefficient; subscripts sp, sub, and coat apply
1/2
to the materials of the cavity body, the mirror substrates, and w01 = ` l j [ L( R1 - L)] 1/4 , (7)
the reflective coating, respectively; f is the frequency; kB is the p
Boltzmann constant; T is the temperature; D is the thickness
of the reflective coating; w0 is the radius of the radiation beam 1/2
L1/4 R11/2
on the mirror; and L and R are the length and radius of the w02 = ` l j . (8)
p ( R1 - L)1/4
cavity body. The fractional mode frequency instability may
be calculated from the total spectral noise power Gtot( f ) =
2Gm( f ) + Gsp( f ) using the formula [15] Formulas (7) and (8) apply to the mode of a Fabry – Perot
cavity of length L consisting of a plane mirror and a concave
2 ln 2 f Gtot ( f ) spherical mirror with a radius of curvature R1. In accordance
sy = . (6)
L with expression (3), increasing the mode spot on the mirror
contributes to a higher stability of the value of Uz(t) due to a
The working temperature of the cavity is selected at its greater area of thermal vibration averaging over the mirror
‘zero CTE point’, i.e. the temperature at which the thermal surface.

Table  1.  Limiting relative frequency instability of a cavity eigenmode induced by thermal noise.
Relative  Body   Substrate  Coating 
Body/substrate/coating   Cavity  Cavity body 
frequency  contribution   contribution   contribution  
material temperature/K length/mm instability* (%) (%) (%)
ULE glass/ULE glass/ 
297 77.5 8.7 ´ 10–16 0.5 65 34
(SiO2 /Ta2O5)
ULE glass/fused silica/ 
297 77.5 4.9 ´ 10–16 2 1 97
(SiO2 /Ta2O5)

ULE glass/fused silica/ 

297 480 6.8 ´ 10–17 11 1 88
(SiO2 /Ta2O5)

Silicon/silicon/ 
124 77.5 2.2 ´ 10–16 10–3 0.1 99.9
(SiO2 /Ta2O5)
*
The relative frequency instability caused by thermal noise may be lowered to 10–16 by using higher-Q materials, lowering the working temperature,
and lengthening the cavity.
On the thermal noise limit of ultrastable optical cavities 427

4. Finite element method for modelling We denote the susceptibility to vibrations by s and define
suspension systems and determining it in accordance with the expression
the vibrational stability of long cavities
DLt
= sa , (9)
ULE glass, whose ‘zero point’ is close to the room tempera- Lt
ture, is the traditional material for making high-Q cavities.
The spectral linewidths of laser systems stabilised to ULE where a is the acceleration. We set ourselves the task of calcu-
cavities approximately 10 cm in length range down to lating the x-, y-, and z-components of vector s. Also, the sus-
0.5  Hz. A characteristic feature of ULE-glass cavities is an ceptibility pertaining to the mirror rotation will be denoted by
approximately linear frequency drift of about 0.1 – 1 Hz s–1 st and the susceptibility pertaining to the mirror tilt by sr. In
caused by material recrystallisation. The good linearity of this case, s = st + sr.
the drift permits subtracting it and obtaining a relative Among the virtues of the horizontal cavity configuration is
instability of ~10–15 on an averaging time frame of 1 – 10 s its low susceptibility to vertical perturbations when the sup-
[18]. The thermal noise of short cavities becomes a hindrance porting points are located at the so-called Airy points (Fig.  1a).
to a further improvement of frequency stability. The 10–15 Finding these points is a well-known problem in the theory of
instability limit was overcome with the use of long mono- strength of materials: this is precisely how the bearings are
lithic cavities of mass ~20 kg with an intermirror distance of located which support the length standard [22]. The possibility
~0.5 m [11, 19]. It is evident that lengthening a cavity height- that there exists an arrangement such that the displacement
ens its sensitivity to vibrations as well as to fluctuations and and inclination angle of the mirrors turn to zero may be under-
uniformity of the temperature. Design and fabrication of stood from the following reasoning. When the suspension
long monolithic cavities is a challenging scientific and tech- points are close to each other, the cavity edges bend down
nical problem. under gravity and the mirror surfaces turn upwards. When the
For the purposes of spectroscopy of the clock transition in bearings are close to the edges, the central part of the interfer-
a strontium optical clock, which now is under development at ometer body bends down and the mirrors turn downwards.
the VNIIFTRI [20], we designed two laser systems, in which Evidently there should be an intermediate arrangement of the
698-nm lasers are stabilised to 480-mm long external refer- bearing such that the mirror surfaces remain parallel to each
ence cavities of ULE glass in horizontal and vertical configu-
rations. It is planned to make the mirror substrates of fused
silica, which will permit the frequency instability limit induced
by thermal noise to be lowered to 6.8 ´ 10–17 (see Table 1).
The requirement of making two ultrastable systems is caused
by the necessity of comparing their frequencies and determin-
ing the individual characteristics of each of the two systems a
[15]. The development of stabilisation systems with long ULE
cavities brings up the burning question of temperature stabil-
ity and vibrational cavity stability. Each configuration invites
the design of a suspension system least sensitive to external y
vibrations.
The cavity length is affected by two types of mirror defor- z x
mations under external forces: rotations and displacements
Displacement along
(translations) [21]. A quantitative analysis of the elastic defor-
mations of reference cavities performed by the method of the z axis/nm
7.71 b
finite elements [19] turns out to be highly beneficial when
solving the optimisation problem, i.e. the search for support- 5.53
3.34
ing point positions whereby the displacements and tilts of 1.16
cavity mirrors turn to zero simultaneously. Generally, to –1.03
describe the vibrational susceptibility calls for a complete –3.21
dynamic analysis of cavity motion under the action of peri- –5.39
odic perturbing forces at different frequencies, but in our case –7.58
we can restrict ourselves to the solution of the static problem Displacement along
for three reasons. First, of highest significance for our sys- the z axis/nm
tems is low-frequency noise (with frequencies below 100  Hz), 0.88
because it is poorly averaged over the short and medium 0.66
times (1 – 100 s) of interest to us. Second, the higher fre- 0.44
0.21 c
quency noise is quite well suppressed with the use of com-
–0.12
mercially available passive and active vibration isolation
–0.26
systems. Third, the wavelength of a sound wave in the cavity –0.50
material is far greater than its geometric size. This signifies –0.75
that different cavity parts oscillate in phase, and therefore it
would suffice to consider the deformation induced by a con-
Figure 1.  (Colour online) Model of vibration-immune suspension sys-
stant ‘instantaneous’ force. It is noteworthy that the cavity tem for a horizontal ULE cavity: (a) the bearings are located at the Airy
eigenfrequencies usually exceed several kilohertz, which per- points, (b, c) deformations of a horizontal ULE cavity under the force
mits neglecting the resonance nature of vibrations in our of gravity.
consideration.
428 N.O. Zhadnov, K.S. Kudeyarov, D.S. Kryuchkov, et al.

other under the action of a vertical force. Similarly, when the For a vertical arrangement of the cavity, it is fixed at three
upper face of the cavity is the plane of support, its body will equidistant points located on its centre-of-mass plane (Fig. 3).
stretch in the direction perpendicular to the optical axis, and This suspension system provides equal displacements of the
the mirrors will come closer together due to the Poisson coef- upper and lower mirrors under the action of vertical forces
ficient. When the lower face is the support plane, they will and, accordingly, a high immunity to vibrations in this direc-
recede from each other. Therefore, a stable arrangement is tion. Unlike the horizontal interferometer, the body of the
always possible to find by optimising the depth d of position of vertical one is symmetrically cylindric in shape. The cavity
bearings and their distance l from the cavity end face (Fig. 2a). body is 7 cm in diameter and 48 cm in length. The biconical
Due to Hooke’s law, the optimal arrangement is independent of shape of the interferometer body is most expedient as regards
the magnitude of the force that gives rise to elastic deformation. immunity to vibrations, because the greatest part of cavity
mass is concentrated near the suspension points in this con-
figuration (Fig. 3).

a
d
a b
l
l/mm
99.5 Displacement along Displacement along
99.0
98.5 the z axis/nm the z axis/nm
98.0 82.6
27.1
4 19.3 59.0
a /10–9 rad

b 2 11.6 35.4
0 3.9 11.8
–2 –3.8 –11.8
–11.6 –35.4
73.5
–19.3 –59.0
74.0 –82.5
–27.1
74.5
d/mm
75.0
73.5 z
74.0
74.5 d/mm
75.0 x
30
Dz /10–9 mm

c 20
10 Figure 3.  (Colour online) Deformations of (a) the vertical biconical and
0 (b) cylindrical ULE cavities under the action of the same horizontal ac-
celeration equal to 9.8 m s–2. According to calculations, the use of bi-
98.0 conical shape lowers the vibrational susceptibility by a factor of two.
98.5
99.0
We calculated the susceptibility to vibrations of the verti-
99.5 l/mm
cal and horizontal cavities for accelerations applied along dif-
ferent axes (Table 2). The susceptibility to all kinds of pertur-
Figure 2.  Parameters of cavity bearing points optimised in the course of bations does not exceed 3 ´ 10–10/g, with the exception of the
finite-element simulations: (a) depth d of bearing location and their dis-
susceptibility of the vertical cavity to horizontal accelerations.
tance l from the cavity end face; (b, c) results of simulations: mirror tilt
a and mirror displacement Dz on application of the force of gravity in This characteristic should not become a factor that limits the
relation to parameters d and l. Points and grey planes stand for the re- cavity stability, because horizontal accelerations are normally
sults of simulation and approximation. several orders of magnitude weaker than the vertical ones in
laboratory conditions.

The displacement and rotation of the cavity mirrors under 5. Silicon cavity
the constant force of gravity was modelled by the method of
finite-element analysis. Our simulations were made on a spa- Among the highly promising materials for high-Q optical
tial grid consisting of ~105 tetrahedral and prismatic ele- cavities is monocrystalline silicon, which is transparent for
ments. The simulations were performed for different bearing the radiation with a wavelength ranging from 1.1 to 6.7 mm
parameters d and l. The calculated shifts and tilts are shown and has excellent mechanical characteristics. The fractional
in Fig. 2b in relation to bearing coordinates. frequency instability of laser systems locked to silicon cavities
Our simulations for a 480-mm long cavity of square sec- has reached a presently record figure of 4 ´ 10–17 over an
tion yielded the following optimal parameters of bearing posi- averaging time of 1 – 1000 s [10]. Furthermore, in silicon there
tions: l = 98.68 mm, d = 74.46 mm. The gravity-induced cav- are no ageing processes, which are responsible for the fre-
ity deformations with the use of the optimal parameters l and quency drift in ULE cavities, which opens up the possibility
d are plotted in Figs 1b and 1c. of developing laser systems with a small (no greater than 2
On the thermal noise limit of ultrastable optical cavities 429

Table  2.  Susceptibility to vibrations of the vertical and horizontal ULE cavities of length 480 mm. Given are the contributions to the instability
caused by mirror rotations (r) and translations (t) on application of accelerations along the x, y, and z axes (see Figs 1 and 3).
Susceptibility to vibrations (1/g)
ULE cavity
sr x st x sr y st y sr z st z
Vertical 
cylindrical 2 ´ 10–8 2 ´ 10–8 2 ´ 10–8 2 ´ 10–8 8 ´ 10–12 1 ´ 10–10

Horizontal 2 ´ 10–10 1 ´ 10–10 4 ´ 10–11 4 ´ 10–11 2 ´ 10–10 3 ´ 10–10

mHz s–1) frequency drift [23]. With the use of a femtosecond was sufficient for stabilising the spectrally narrow (less than
generator of optical frequencies their stability may be 10 kHz) emission of the fibre laser with a Bragg grating.
imparted to any laser [24]. In the path to achieve the thermal noise limit, various elec-
We pursue research aimed at developing laser systems tronic noise as well as the residual amplitude modulation of
that radiate at a wavelength of 1542 nm and are stabilised to the laser radiation may be a serious problem. To realise the
silicon 7.75-cm long silicon cavities, whose thermal noise limit Pound – Drever – Hall technique, the laser radiation is phase-
due to thermal noise is equal to 2.2 ´ 10–16 at a temperature modulated by an electrooptical modulator (EOM). An inex-
of 124 K corresponding to the ‘zero point’ of silicon (Table 1). act matching of the plane of light polarisation and the extraor-
The use of a wavelength from the C band (near 1.5 mm) opens dinary axis of the EOM results in a parasitic amplitude modu-
up the possibility of transmitting ultrastable signals via fibre lation of the radiation. The modulation depth may vary due
lines over distances longer than 1000 km [25]. The cavity is to temperature variation, which will introduce additional
cooled with the use of our developed cryogenic liquid-nitro- instability into the frequency stabilisation system. As shown
gen system [26]. The cooler vessel is inside a vacuum chamber, in Ref. [27], the active compensation of amplitude modula-
and the heat exchange with the cavity is affected by thermal tion by applying feedback to the EOM permits its contribu-
radiation. To maintain the level of cooler in the vessel, use is tion to the radiation frequency instability to be diminished to
made of a system for the production and automatic transfer a level defined by thermal noise and lower. We also plan to
of liquid nitrogen. Its effect on cavity stability, which is pro- employ this technique in the future.
duced by the vibrations emerging in the boiling of nitrogen, is
eliminated by mechanically decoupling the vessel from the 6. Conclusions
remaining part of the vacuum chamber. At a temperature of
124 K, the measured finesse F * of the silicon cavity for the The research and development of ultrastable laser systems is
TEM00 mode was greater than 580000. The measurement was among the key tasks in the development of optical frequency
made by recording the ring-down of the light transmitted by standards and optical frequency transmission lines as well as
the cavity (Fig. 4). in the generation of radiofrequency fields with a high short-
term stability in the problems of radio photonics. High-Q
optical cavities make it possible to stabilise the output laser
0.6 frequency to a residual relative instability of ~10–17, which is
Cavity transmittance (arb. units)

fundamentally limited by thermal noise. A series of investiga-

t = 46.4 ms,
F * = 584000 tions was carried out to develop laser systems stabilised to
0.4 cavities with a low frequency instability limit imposed by
thermal noise. To achieve a high frequency stability, advan-
tage was taken of two approaches: the use of a long cavity and
0.2 the application of new materials. For the cavity configura-
tions under investigation and development, the fractional fre-
quency instability corresponding to thermal noise was shown
to be at a level of ~10–16. Cavity mount systems were investi-
0
gated and optimised, and the attainable cavity susceptibilities
to acceleration were shown to be equal to ~10–10/g. The sys-
tems under development will be employed in the optical fre-
– 0.2 quency standard based on neutral strontium atoms at the
0 50 100 150
Time/ms VNIIFTRI for interrogating the clock transition at l =
698 nm and for stabilising the radiation frequency of a femto-
Figure 4.  Measurement of cavity finesse F * from the cavity transmit- second optical frequency synthesiser.
tance decay time t upon cutting off laser radiation: F * = 2ptc/(2L).
Acknowledgements.  This work was supported by the Russian
Foundation for Basic Research (Grant No. 16-29-11723).
With the use of the silicon cavity we stabilised the output
frequency of an erbium fibre laser at a wavelength of 1542 nm. References
The optical configuration of the facility was similar to that
  1. Kwee P., Bogan C., Danzmann K., et al. Opt. Express, 20 (10),
outlined in Ref. [18]. A slow feedback signal was delivered to
10617 (2012).
a piezoelectric actuator and a fast feedback signal was applied   2. Chin C., Grimm R., Julienne P., Tiesinga E. Rev. Mod. Phys., 82
to an acoustic modulator. This permitted obtaining a total (2), 1225 (2010).
loop bandwidth of about 100 kHz, which was limited by the   3. Udem T., Reichert J., Holzwarth R., Hänsch T.W. Opt. Lett., 24
PID controller in use. This bandwidth of the feedback loop (13), 881 (1999).
430 N.O. Zhadnov, K.S. Kudeyarov, D.S. Kryuchkov, et al.

  4. Bloom B.J., Nicholson T.L., Williams J.R., et al. Nature, 506

(7486), 71 (2014).
  5. Chou C.W., Hume D.B., Koelemeij J.C.J., et al. Phys. Rev. Lett.,
104, 070802 (2010).
  6. Quessada A., Kovacich R.P., Courtillot I., et al. J. Opt. B:
Quantum Semiclass., 5 (2), 150 (2003).
  7. Itano W.M., Bergquist J.C., et al. Phys. Rev. A, 47 (5), 3554 (1993).
  8. Jiang Y.Y., Ludlow A.D., Lemke N.D., et al. Nat. Photonics, 5
(3), 158 (2011).
  9. Drever R.W.P., Hall J.L., Kowalski F.V., et al. Appl. Phys. B, 31
(2), 97 (1973).
10. Matei D.G., Legero T., Häfner S., et al. Phys. Rev. Lett., 118 (26),
263202 (2017).
11. Häfner S., Falke S., Grebing C., et al. Opt. Lett., 40 (9), 2112
(2015).
12. Gillespie A., Raab F. Phys. Rev. D, 52 (2), 577 (1995).
13. Levin Yu. Phys. Rev. D, 57 (2), 659 (1998).
14. Numata K., Kemery A., Camp J. Phys. Rev. Lett., 93 (25), 250602
(2004).
15. Riehle F. Frequency Standards. Basics and Applications
(Weinheim: Wiley, 2004; Moscow: Fizmatlit, 2009).
16. Storz R., Braxmaier C., Jäck K., et al. Opt. Lett., 23 (13), 1031
(1998).
17. Cole G.D., Zhang W., Martin M.J., et al. Nat. Photonics, 7 (8), 644
(2013).
18. Alnis J., Matveev A., Kolachevsky N., et al. Phys. Rev. A, 77 (5),
53809 (2008).
19. Martin M.J. Quantum Metrology and Many-Body Physics:
Pushing the Frontier of the Optical Lattice Clock (PhD
dissertation, University of Colorado, 2013).
20. Berdasov O.I., Gribov A.Yu., Belotelov G.S., et al. Quantum
Electron., 47 (5), 400 (2017) [ Kvantovaya Elektron., 47 (5), 400
(2017)].
21. Millo J., Magalhães D.V., Mandache C., et al. Phys. Rev. A, 79
(5), 53829 (2009).
22. Phelps F.M. Am. J. Phys., 34 (5), 419 (1966).
23. Hagemann C., Grebing C., Lisdat C., et al. Opt. Lett., 39 (17),
5102 (2014).
24. Hagemann C., Grebing C., Kessler T., et al. IEEE Trans. Instrum.
Meas., 62 (6), 1556 (2013).
25. Droste S., Udem T., Holzwarth R., Hänsch T.W. Comptes Rendus
Physique, 16 (5), 524 (2015).
26. Zhadnov N.O., Masalov A.V., Sorokin V.N., et al. Quantum
Electron., 47 (5), 1 (2017) [ Kvantovaya Elektron., 47 (5), 1 (2017)].
27. Zhang W., Martin M.J., Benko C., et al. Opt. Lett., 39 (7), 1980
(2014).