Skip to main content

Full text of "Terahertz frequency standard based on three-photon coherent population trapping"

See other formats

Terahertz frequency standard based on three-photon coherent 

population trapping 

C. Champenoisl*] G. Hagel, M. Houssin, M. Knoop, C. Zumsteg, and F. Vedel 
Physique des Interactions loniques et Moleculaires (CNRS UMR 6633), 
Universite de Provence, Centre de Saint Jerome, 
Case C21, 13397 Marseille Cedex 20, France 
(Dated: February 1, 2008) 


A scheme for a THz frequency standard based on three-photon coherent population trapping in 
stored ions is proposed. Assuming the propagation directions of the three lasers obey the phase 
matching condition, we show that stability of few 10~^^ at one second can be reached with a 
precision limited by power broadening to 10^^^ in the less favorable case. The referenced THz 
signal can be propagated over long distances, the useful information being carried by the relative 
frequency of the three optical photons. 

PACS numbers: 06.30.Ft, 32.80.Qk, 32.80.Pj 


Thanks to the progress made in the generation of continuous THz radiation, Terahertz 
spectroscopy is strongly gaining in interest in different domains, in particular for imaging in 
astronomy and biology and for material analysis [l!]. In this paper we propose a frequency 
standard in the Terahertz domain, based on a three-photon dark resonance where the three 
wavelengths belong to the visible domain. Atomic frequency standards in the optical domain 
can now reach performances in the 10~^^ range, thanks to laser-cooling of the atoms or single 
ions and to progress made in the stabilization of the clock lasers {2!, y] . The proposed THz 
frequency standard aims at a lower relative precision of about 10^^^, which can be realized 
in a robust and easily achievable experimental configuration and with a large atomic sample, 
allowing frequency stabilities of a few parts in 10^^ at one second. 

In the scheme described here, the THz reference signal is produced by coherent popula- 

tion trapping fCPT) in a three-photon process [4!] involving visible lasers in a Doppler-free 
configuration ^. For instance, this process can occur in several ions commonly used for 
optical metrology or quantum computation. For these atoms (Ca^, Sr"*", Ba^, Hg"^), the 
narrow dark line resulting from CPT corresponds to the magnetic dipole transition between 
the nD3/2 and nD^i2 metastable states, whose frequency lies in the THz range. This electric- 
dipole forbidden line has been studied and probed directly in former experiments 
but in a configuration sensitive to the Doppler effect. The proposed experiment can be 
seen as analogous to two-photon CPT microwave cesium clock [qI where a trade-off is made 
between robustness and clock precision. Two-photon CPT is also considered for an optical 

clock based on neutral ®^Sr 

10| and a three-photon coherent process has been proposed as 

an optical frequency standard in alkaline earth atoms in 

In this paper, we first introduce the three-photon coherent process resulting in the sharp 
line that can serve as a frequency standard. This process is largely discussed in j4| and 
we mention only the points useful for metrological application. Then, the expected perfor- 
mances of such a standard are studied to show a possible stability better than 8 x 10^^^ at 
one second. 

The scheme we propose can apply to any atomic system composed of four electronic levels 
which are coupled by laser fields, according to the A^-shaped scheme depicted in Fig. [1] and 
where states IS"), \D) and \Q) are (meta)stable while state \P) is short-lived and decays 
radiatively into IS*) and \D). This level configuration is realized, for instance, in alkaline- 
earth atoms with hyperfine structure and in alkali-like ions with a metastable d-orbital, such 


as Hg"*", Ba+, Sr"*", or Ca+. In this manuscript we focus on this last case where the levels can 
be identified with the states 15) = 151/2), \P) = I-P1/2)) \D) = 1-03/2), and \Q) = \D^/2). The 
dark line can be observed by collecting the photon signal on the \S) \P) transition. Here, 
all the transitions are electric-dipole allowed except |5i/2) ^ I-D5/2) (named \S) ^ \Q) in 
Fig. [T]), which is an electric quadrupole transition with a linewidth of the order of one Hertz. 
The magnetic-dipole transition I-D3/2) ^ I-D5/2) (|-D) ^ \Q)) considered as the frequency 
standard has a spontaneous emission rate of the order of 10~^ s~^ which can be neglected. 

FIG. 1: Level scheme: The states \D), \P), \S) coupled by laser couplings ^Iji and ^Ib form 
a A- configuration, state \S) couples weakly to the metastable state \Q) by Q]v. The wavy lines 
indicate the radiative decay. Parameters and possible atomic species are discussed in the text. 

For our purpose, the evolution of the internal degrees of freedom of a motionless atom can 
be understood in the dressed state picture, where the non-coupled eigenstates are defined 

Ho = hAn\D){D\ + hAB\S){S\ + - Aw)\Q){Q\ (1) 

with detunings defined as = ub—ujps, Ar = ur — upd, and Aw = ojw — ojqs {B,R,W 
stands respectively for the \S) ^ |P), \D) ^ \P) and \S) ^ \Q) transitions, ux is 
the laser frequency on the transition labelled X and uu is the frequency of the atomic 
transition |/) ^ |J)). ^r^b,w are the corresponding Rabi frequencies characterizing the 
laser couplings. The radiative processes taken into account couple \P) to states \S) and \D) 
by the decay rate ■jp and branching ratio Pps/Ppd {Pps + Ppd = !)• The radiative decay of 
states \Q) and \D), whose lifetime is of the order of 1 s is neglected in the analytical model 
we present below. It should be noted that the dipole and the quadrupole couplings differ 


by a few orders of magnitude. Nevertheless, as the process described in this manuscript is 
of interest when state \Q) is weakly coupled to the A-scheme (15*), |P), \D), coupled by the 
two strong lasers) it is thus experimentally feasible with standard laser sources. 

Considering that \Q) is weakly coupled to \S), the subsystem (15), \Q)) can be diago- 
nalised and solved analytically to first order in aw = ^w/'^'^w ^ 1- The new eigenstates 
are then 

\Sq) = Af {\S) + awlQ)) (2) 
\Qs) = Af i\Q) - aw\S)) 

(where A/" is the normalization factor) with eigenfrequencies light-shifted by ±aw^w f"^- 
\Qs) being coupled to \P) by the Rabi frequency —aw^B, the dressed states configuration 
ends up in a A-scheme (formed by \Qs),\P) and \D), see Figj2]), well known to give rise to 
coherent population trapping into a dark state when the metastable states \Qs) and \D) 
are degenerated when dressed by the laser photons This condition is fulfilled on the 
light-shifted three-photon resonance condition 

A,ff = Ar + Aw-Ab + aw^wf^ = 0. (3) 

Detuned from the two-photon resonance condition {i. e. A^ — Ab ^ 0, see j4| for justifica- 
tion), the atomic system is then pumped into the dark state 

\<ilDark)=Af'i£\D) + \Qs)) (4) 

with S = aw^s/^R and normalization factor A/"'. 

The sharp line produced in the fluorescence spectra by pumping into the dark state can 
be used to assure A^ -|- A^y — + avy^vy/2 = 0. Experimentally, this is realised by locking 
one of the three lasers on the dark line. For the following laser frequencies, it translates into 

Provided the light-shift aw^w/'^ is made negligible or controlled, the frequency combination 
ur + ojw — ojb can be referenced to the magnetic-dipole transition frequency ojqd which lies 
between 1.8 THz for Ca+ and 451 THz for Hg"*" (see table [11). 

The stability of the dark state requires the reduction of the relative phase diffusion of the 
three lasers to a diffusion rate smaller than the dark line width. This can be achieved by 

FIG. 2: Part of the dressed state picture relevant for the three-photon resonance condition, giving 
rise to the coherent dark state. The straight hnes stands for laser couplings and the wavy lines for 
radiative decay (see text for details). 

TABLE I: Wavelength involved in the three-photon process and frequency of the reference transition 

fons B R W ^3/2-^2 

Ca+ 397 nm 866 nm 729 nm 1.82 THz 

Sr+ 422 nm 1092 nm 674 nm 8.40 THz 

Ba+ 493 nm 650 nm 1761 nm 24.0 THz 

Hg+ 194 nm 10.67 fim 282 nm 451 THz 

reducing the spectral width of each laser or, as a less stringent condition, by phase-locking 
the three lasers on the same reference like a frequency comb ISj. The fundamental limit for 
the linewidth of the coherent dark line, defined by the lifetime of the metastable states, may 
not be reached, mainly because of power broadening, which is known to be one of the main 
limiting factor in two-photon CPT clock [l^ . For evaluating the dark line width, the system 
can be approximated by the A-scheme depicted in Figj2] where the decay from \P) to \Qs) is 
(^wIpPps- Because of spontaneous emission from \P) to l^g), the decoherence rate between 
\P) and \Qs) or \D) is equal to 7p/2. An analytical expression of the linewidth of coherent 
dark line can be found in ]3] for very general cases. Such an expression is simplified when, 
for <^ 1 and Oi^Vl?^ <^ Qj^ one has PpdQ"^^ ^ Pps^%, which is obeyed for realistic 
working parameters. Then, if the decoherence sources like the laser relative phase diffusion 


are neglected, the dark linewidth (FWHM) can be approximated by 

r = "fi (Q\ 

ipjl + (3 - ^) -4^^ 

with 7a = 'Jp{Ppd + c^wPps)- When collected on the \P) \S) transition, the fluorescence 
rate emitted around the dark resonance A^jj = (Eq. [3D scales with 

'5 = 7pt/-5— • (7) 
"/J Ppd 

According to these last two equations, the linewidth can be tuned continuously by Qp and 

in practice, results from a tradeoff between the desired signal over noise ratio S/N = ^r]S/2 
(with 7] the detection efficiency, 1/2 indicating we work at half maximum and if we assume 
no background) and signal linewidth A > Tf.ff. Both parameters control the stability of the 
frequency standard, as can be clearly seen in the Allan deviation 

which quantifies the stability versus the integration time r, being the cycle time. Increas- 
ing the S/N ratio can be realised without power broadening the transition by increasing 
the number of ions emitting the fluorescence signal. For ions in radiofrequency trap, this 
option is rarely used because of the Doppler effect due to the micromotion (the radiofre- 


quency driven motion), which increases with the number of ions [16 
three-photon resonance, there is a (first order) Doppler- free configuration 
by the phase matching condition 

In the case of a 

51] which is defined 

Ak = ka - kB + kw = 0. (9) 

The driven micromotion can not be laser cooled and its contribution to the second-order 
Doppler effect can not be nulled. Assuming the thermal macromotion (motion in the pseudo- 
potential created by the RF-field ) is laser cooled, the relative shift induced by second order 

Doppler effect can be estimated 17| from the number Ni of trapped ions, their mass M in 

atomic units and the secular frequency of oscillation z/5 (MHz): 

A/Z52//0 = -3.0 X l{)-~'\usN,/Mf\ (10) 

In a centimeter-size trap, where Ni = 10^ and us = 0.1 MHz are typical values, this relative 
shift is —1.2 X 10~^^ for calcium ions, the lightest of the species considered here. For this 

ion, the resulting frequency shift on the magnetic dipole transition is then 2.2 Hz and a 
broadening of the same order is expected. Tolerating such a shift, the S/N ratio behaves 

like yriNiS/2, if one assumes that the signal increases linearly with the number of ions, 
which is a good approximation for laser cooled ions. 

As a starting point and considering this broadening, let's assume a reachable precision of 
10 Hz, then the dark line can be power broadened to that value for maximum signal. The 
expected stability can then be estimated by assuming a cycling time of 1 s (long enough for 
coherence to build) and a detection efficiency of 10~^. We obtain 

Ca+, 2 X for Sr+, 1 x lO-^V-^/^ fo^ Ba+ and 4 x IQ-^'t-^/^ for Hg+ (where the 

transition stands in the optical domain). The THz radiation can be produced by photomixing 
the 5-radiation with the one resulting from the frequency sum of the R- and 14^-radiations. 
The generation of the THz signal from three visible laser beams allows propagation and 
dissemination of the frequency standard over long distances, despite the high absorption 
coefficient known for THz radiation in air or in fibers. 

The accuracy of the proposed standard depends on all the effects that can induce a 
frequency shift of the dark line from the ideal equation ur + uw — ^^b = ^qd- Beside the 
Doppler effect, the first effect to deal with is the one due to the method itself. The laser 
coupling between \S) and \Q), needed to create the A-scheme generating the dark state, 
gives rise to a light-shift aw^wf^ (Eq. E]). Nevertheless, values as small as aw < 10~^ and 
flw < 10 kHz are sufficient to efficiently create this coupling, and to keep the induced light 
shift smaller than 1 Hz, without any effect on the collected signal (see Eq. [Tj). By definition, 
the dark state is not light-shifted by the direct coupling mentioned on Fig I2] but it is shifted 
jy far off resonance coupling between one of the three lasers with the other two transitions 

18l | . According to the level configuration of the considered ions, the largest shift comes from 
the coupling between I-D5/2) and the IP3/2) state lying close to I-P1/2), induced by the laser 
labelled R coupling I-D3/2) to \Pi/2)- Among the considered ions, Ca"*" has the smallest fine 
structure splitting and will therefore show the biggest shift which is estimated to be lower 
than 0.1 Hz for Qr below 1 MHz. This effect is clearly not a limitation here. 

The other effects limiting the accuracy and precision of a possible THz-standard are 
the DC-Stark shift, the quadrupole shift caused by the coupling between the electric field 
gradient and the electronic quadrupole of the I-D3/2) and I-D5/2) states and the Zeeman 
shift. The DC-Stark effect depends on the polarizability of the I-D3/2) and I-D5/2) states. As 


mentioned in uM for Ca"*" and in 

20[ | for Sr"*", there is an uncertainty on these polarisabihties 

for some species. Nevertheless, from available atomic transition data, one can deduce a 
compensation between the polarisabihties of these two states which bring the absolute value 
of the order of few 10~^^ J/(V/m)^ to a differential value of the order of few 10~^^ J/(V/m)^, 
which translates into \Sfs\ — lO^^i?^ Hz with E in V/cm. At room temperature, the biggest 
contribution comes from the isotropic blackbody field radiated by the vessel. This field 
can be taken into account by its mean-squared value averaged over the whole spectrum: 
(Els) = 831.92(T/300)^ (V/m)^ ^ which results in a DC-Stark shift below 0.01 Hz. 
The quadrupole shift can be written like [22 1 

AfQ{n, J, mj) = ^^^^^j^^^^in, J)AU/h (11) 

with Q{n,J) the electric quadrupole of state \n,J), A = [ff^V /2dx^) defining the local 
electric field gradient and H a geometrical factor depending on the shape and orientation 
of the electric field (H = 2 if the field is quadrupolar with its symetry axis along the 
quantization axis). If one assume that the \n, Dj) states are pure d orbitals, the quadrupole 
varies like (2J - 1)/(2J + 2) and Gfn, 3/2) = (7/lO)0(n, 5/2). Calculations for all the ions 
mentioned and for J = 3/2,5/2 23|] show that the quadrupole 6(n, J) rank from 0.5 to 4 
eog. For the states with the smallest Zeeman shift (mj = ±1/2), the frequency shift for one 
ion is (5/q = (ll/40)9(n, 5/2)AH//i. If the highest value for 9(n, 5/2) is used, reaches 
1 Hz for AH = 150 V/cm^. This value is the order of magnitude of the gradient of the 
trapping field and of the Coulomb field in the considered regime and the quadrupole shift 
can not be neglected here. 

As for the Zeeman effect, a small magnetic field B is needed to split the Zeeman sub levels 
chosen for the transition I-D3/2, ^^j"^) I-D5/2, ^^j"^)- The first order Zeeman shift is minimal 
for the rn'j'^ = ±1/2 —>■ rrb'j'^ = ±1/2 transition, and is then equal to ±0.28 MHz/Gauss. 
A 2 mG field is then necessary to split these two transitions by 1 kHz. The Zeeman shift 
can be nulled by adding the frequency of these two transitions, assuming that the magnetic 
field is constant. The second order Zeeman effect is negligible here. In the regime chosen 
for our estimation {Ni = 10^, and us = 0.1 MHz) and for perfect compensation of first order 
Zeeman shift at the Hz level, the magnetic field must be stable and homogenous on a mm- 
size to better than 3.6 /iC. Such a stabihty requires magnetic shielding but is experimentally 


From this overview of the systematic shifts expected for an ion cloud, it appears that the 
important effects for the magnetic-dipole transition come from the second order Doppler 
shift, the quadrupole shift and the Zeeman shift, which hmit the precision of the proposed 
THz standard to the Hz level, depending on the size of this cloud. 

In conclusion, we show that a THz frequency standard can be realised from three-photon 
CPT in trapped ion clouds. A stability in the 10~^^r~^/^ range can be expected, depending 
on the chosen species. Such level of stability is reached to the detriment of precision which 
is mainly reduced by power broadening, and to a smaller extent by second order Doppler 
effect, quadrupole shift and first order Zeeman shift. 

The authors would like to thank P. Dube for his comments on the manuscript. 

* |caroline. 
[1] K. Kawase, Optics and Photonics News 15, no. 10, 34 (2004). 

[2] W. H. Oskay, S. A. Diddams, E. A. Donley, T. M. Fortier, T. P. Heavner, L. Hollberg, W. M. 
Itano, S. R. Jefferts, M. J. Delaney, K. Kim, et al., Physical Review Letters 97, 020801 (2006). 

[3] M. M. Boyd, A. D. Ludlow, S. Blatt, S. M. Foreman, T. Ido, T. Zelevinsky, and J. Ye, Physical 
Review Letters 98, 083002 (2007). 

[4] C. Champenois, G. Morigi, and J. Eschner, Physical Review A 74, 053404 (2006). 

[5] G.Grynberg, F.Biraben, M. Bassini, and B. Cagnac, Phys. Rev. Lett. 37, 283 (1976). 

[6] G. Werth, Possible precision far-infrared spectroscopy on trapped ions (Summer school "Ap- 
plied Laser Spectroscopy", San Miniato, 1989). 

[7] A. A. Madej, J. D. Sankey, G. R. Hanes, K. J. Siemsen, and A. R. W. McKellar, Phys. Rev. 
A 45, 1742 (1992). 

[8] K. J. Siemsen, A. A. Madej, J. D. Sankey, B. G. Whitford, R. F. Clark, and G. Magerl, IEEE 

frequency control symposium p. 76 (1992). 
[9] T. Zanon, S. Guerandel, E. de Clercq, D. Holleville, N. Dimarcq, and A. Clairon, Phys. Rev. 
Lett. 94, 193002 (2005). 
[10] R. Santra, E. Arimondo, T. Ido, C. Greene, and J. Ye, Phys. Rev. Lett. 94, 173002 (2005). 
[11] T. Hong, C. Cramer, W. Nagourney, and E. N. Fortson, Phys. Rev. Lett. 94, 050801 (2005). 
[12] E. Arimondo, in Progress in Optics, edited by E. Wolf (Elsevier Science, 1996), vol. XXXV, 


chap. V, p. 258. 

[13] T. Udem, J. Reichert, R. Holzwarth, and T. W. Hansch, Optics Letters 24, 881 (1999). 
[14] S. Knappe, R Schwindt, V. Shah, L. HoUberg, and J. Kitching, Optics Express 13, 1249 

[15] T. Zanon-Willette, Ph.D. thesis, Universite Paris VI (2005). 
[16] H. Dehmelt, Adv. At. Mol. Rhys. 5, 109 (1969). 

[17] D. Wineland, W. Itano, J. Bergquist, J. Bollinger, F. Diedrich, and S. Gilbert, in Proc. 4th 
symp on Freq. Standard and Metrology, edited by A. Demarchi (Ancona, Italy, 1988). 

[18] T. Zanon-Willette, A. D. Ludlow, S. Blatt, M. M. Boyd, E. Arimondo, and J. Ye, Physical 
Review Letters 97, 233001 (2006). 

[19] C. Champenois, M. Houssin, C. Lisowski, M. Knoop, M. Vedel, and F. Vedel, Rhys. Lett. A 
331, 298 (2004). 

[20] A. A. Madej, J. E. Bernard, P. Dube, L. Marmet, and R. S. Windeler, Physical Review A 70, 
012507 (2004). 

[21] W. Itano, I. Lewis, and D. Wineland, Phys. Rev. A 25, 1233 (1982). 
[22] W. Itano, J. Res. Natl. Inst. Stand. Technol. 105, 829 (2000). 
[23] W. M. Itano, Physical Review A 73, 022510 (2006).