EUROPHYSICS LETTERS 15 May 2001

Europhys. Lett., 54 (4), pp. 456–460 (2001)

Electromagnetically induced two-photon transparency

in rubidium atoms
D. Wang 1 , J. Y. Gao 1 , J. H. Xu 2 , G. C. La Rocca 3,2 and F. Bassani 2
1
Department of Physics, Jilin University - Changchun, PRC
2
INFM, Scuola Normale Superiore - Piazza dei Cavalieri 7, 56126 Pisa, Italy
3
Dipartimento di Fisica, Università di Salerno - 84081 Baronissi (Sa), Italy

(received 16 November 2000; accepted in final form 7 March 2001)

PACS. 42.50.-p – Quantum optics.

PACS. [Link] – Effects of atomic coherence on propagation, absorption, and amplification
of light.
PACS. [Link] – Strong-field excitation of optical transitions in quantum systems; multi-
photon processes; dynamic Stark shift.

Abstract. – We present an experimental demonstration of electromagnetically induced two-

photon transparency (EITT) in room temperature rubidium vapor. The 8S1/2 to 5P1/2 flu-
orescence is used to monitor the 5S1/2 (F = 3) to 8S1/2 (F = 3) two-photon absorption near
resonance with the intermediate state 5P3/2 . A controlling pump laser beam is employed to
coherently couple the 5P3/2 and 5D5/2 states, thus producing two dressed intermediate states
which give rise to destructive interference in the two-photon transition. An induced two-photon
transparency of about 80% has been obtained at resonance; our experimental findings are in
good agreement with the general theory of Agarwal et al. (Phys. Rev. Lett., 77 (1996) 1039),
when the appropriate spectroscopic parameters are used.

The phenomenon of electromagnetically induced transparency (EIT) based on coherent

population trapping has attracted much interest after the discovery of the “black line” in
sodium vapor in 1976 [1]. In recent years, many theoretical and experimental studies on
exploiting atomic quantum coherence have been carried out [2]. Quantum interference in
the transition probability amplitude may decrease or increase the absorption near resonance,
leading to EIT or electromagnetically induced absorption (EIA) [3], respectively. Many appli-
cations of EIT (or EIA) have been considered, for instance: lasing without inversion [4], elec-
tromagnetically induced focussing and defocussing [5], subrecoil laser cooling due to velocity-
selective coherent population trapping (VSCPT) [6], a dramatic reduction of the speed of light
pulses in ultracold [7] as well as room temperature [8] atomic vapors.
Besides the effects on one-photon absorption, electromagnetically induced two-photon
transparency (EITT) in two-photon transition near resonance with an intermediate state has
been proposed by Agarwal and Harshawardhan [9]. A first evidence of EITT has very recently
been reported by Gao et al. [10] in sodium vapor achieving a 60% transparency at resonance.
We here present an experiment of EITT in room temperature rubidium atoms demonstrating
c EDP Sciences


D. Wang et al.: Electromagnetically induced two-photon transparency etc. 457

Fig. 1 – Scheme of the relevant energy levels of 85 Rb. The γ’s and ∆’s are the corresponding spon-
taneous decay rates and detunings, respectively.

a narrow and deep window reaching 80% transparency at resonance. We show that our ex-
perimental findings compare very well with theory [9] when the known spectroscopic values
of the relevant parameters are used, and prove that the EITT is due to coherent interference
by experimentally monitoring at the same time the EIT on the intermediate state.
We show in fig. 1 the 85 Rb energy levels involved in the present experiment: the 5S1/2 (F =
3) to 8S1/2 (F = 3) two-photon transition is studied near resonance with the intermediate
state 5P3/2 , which is coherently coupled by the pump control beam to the 5D5/2 level. Our
experimental setup is shown in fig. 2. Two laser beams ω1 and ω2 are used for the 5S-
8S two-photon excitation: these two lasers are, respectively, a single-mode diode laser with
wavelength of 780 nm and a single-mode dye laser with wavelength of 616 nm, and they both
have a linewidth of about 1 MHz. They are counterpropagating so as to reduce the Doppler
broadening and they are circularly polarized with opposite polarizations with respect to fixed
axes to satisfy the two-photon transition selection rule. A control pump laser beam ω couples
the 5P3/2 and the 5D5/2 levels; it is from a single-mode diode laser with wavelength of 776 nm
and linewidth of 1 MHz, and it is copropagating with the ω2 beam, but with the opposite
circular polarization. Thus, the Doppler broadening for the 5S-5D transition due to the
beams ω and ω1 is nearly cancelled, whereas the residual Doppler broadening of the 5S-8S
transition is relatively large, due to the significant frequency difference of beams ω1 and ω2 .
Two lenses with equal focal length of 50 cm are used to focus the two diode lasers ω and ω1 ,
while the beam ω2 from the single-mode dye laser is not focussed because its beam diameter
is sufficiently small. On the sample, the light intensity of the coupling beam ω is of the order
of a few W/cm2 , while the probe beams ω1 and ω2 are about a hundred times weaker. The
rubidium vapor is contained in a 5 cm long cylindrical cell made of Pyrex glass. The Rb
density in the cell is about 1011 cm−3 at room temperature. No buffer gases are added in
the cell to avoid additional dephasing due to collisions. During the experiment, the Rb cell
is always kept at room temperature and shielded by µ metal to reduce the influence of the
Earth and any other stray magnetic fields.
The physics of EITT can be simply described as follows [9]. The coherent coupling of
the 5P to the 5D level due to the control pump beam gives rise to a doublet of dressed
states which serve as intermediate states for the 5S-8S two-photon transition. The quantum
interference between the two nearly resonant contributions of such intermediate states leads,
458 EUROPHYSICS LETTERS

Fig. 2 – Experimental setup.

for appropriate values of the detunings, to a cancellation in the total two-photon transition
probability, and thus to a pump-induced transparency window within the 5S-8S two-photon
absorption line. Partly due to the large residual Doppler broadening, a direct measurement of
the weak two-photon absorption is very difficult; therefore, we use the fluorescence from the
8S to the 5P1/2 level as a measure of the 5S-8S two-photon absorption. At the same time,
either with the pump beam on or off, the absorption of the ω1 beam nearly resonant with the
one-photon 5S-5P3/2 transition is also recorded.
By scanning the probe beam ω1 around the 5S-5P3/2 transition at a fixed frequency of
the other probe beam ω2 around 616 nm, the fluorescence induced by the 5S-8S two-photon
transition is detected by a photomultiplier through a monochromator (Jobin-Yvon 1000) at
the wavelength of 607 nm corresponding to the decay from the 8S to the 5P1/2 level. At the
same time, the absorption of the ω1 beam is detected by a sensitive photodiode. Both the
ω1 absorption and the two-photon–induced fluorescence are recorded by a digital oscilloscope
and stored by a computer. The peak value of the two-photon–induced fluorescence from each
ω1 scan is then stored for many values of the frequency of the ω2 beam, which is varied step
by step over the range of 1 GHz around 616 nm. The 5S1/2 (F = 3)-8S1/2 (F = 3) two-photon
absorption spectra thus obtained in the absence of the pump show that the intermediate-state
resonance is dominated by the contribution of the 5P3/2 (F = 4) level, separated by about
120 MHz from its nearest hyperfine component. Moreover, we have verified that the two-
photon signal when the polarization of one of the two probe beams is reversed is suppressed
by more than an order of magnitude.
The EITT effect on the 5S-8S transition has been measured with the same procedure in the
presence of the pump control beam ω near 776 nm, tuned to the 5P3/2 (F = 4)-5D5/2 (F = 5)
transition. Typical 5S-8S two-photon absorption spectra are shown in fig. 3 and fig. 4, which
correspond, respectively, to a resonant coupling beam (∆ = 0) and to a red detuned coupling
beam (∆ = 60 MHz). In fig. 3, it is demonstrated how a narrow window of 80% transparency
is opened at the center of the two-photon absorption line even for a resonant coupling beam
having a relatively low Raby frequency (about 12 MHz, the largest available with our setup). In
the case of a detuned control beam ω, we also achieve a 60% transparency at the corresponding
detuning of the probe beam ω1 , as is evident in fig. 4. The numerical simulations, also shown
in the figures, have been made by solving the steady-state density matrix equations of a four-
level system [9] using independently estimated values of the relevant parameters (see table I);
D. Wang et al.: Electromagnetically induced two-photon transparency etc. 459

Fig. 3 Fig. 4

Fig. 3 – Spectra of two-photon–induced fluorescence as a function of the ω1 probe beam detuning

from the 5S1/2 (F = 3)-5P3/2 (F = 4) intermediate-state resonance: the squares are measured with
the pump beam off, the full dots with the pump beam on and at resonance with the 5P3/2 (F = 4)-
5D5/2 (F = 5) coupling transition. The estimated errors are about 10 MHz for the detuning and 5%
for the fluorescence intensity. The dashed and full lines are the respective theoretical calculations.

Fig. 4 – Spectra of two-photon induced fluorescence as a function of the ω1 probe beam detuning
from the 5S1/2 (F = 3)-5P3/2 (F = 4) intermediate-state resonance: the squares are measured with the
pump beam off, the full dots with the pump beam on and red-detuned (∆ = 60 MHz) with respect to
the 5P3/2 (F = 4)-5D5/2 (F = 5) coupling transition. The estimated errors are about 10 MHz for the
detuning and 5% for the fluorescence intensity. The dashed and full lines are the respective theoretical
calculations.

the agreement between the theory and the data is very good. In the experiment, several
four-level systems corresponding to different values of the magnetic quantum number M may
contribute to establish EITT; however, due to optical pumping the one with the highest M
value might be dominant (i.e., the four-level system involving the states 5S1/2 (F = 3, M = 3),
5P3/2 (F = 4, M = 4), 5D5/2 (F = 5, M = 5) and 8S1/2 (F = 3, M = 3)). The absorption of
the ω1 beam, measured at the same time as the two-photon–induced fluorescence, exhibits a
well-developed dip corresponding to the usual EIT “black line”. This observation proves the
creation of the dressed state doublet due to the coherent coupling of the 5P and 5D levels
induced by the control beam ω, and also rules out a simple explanation of the two-photon
transparency in terms of two-step processes affected by a depletion of the population of the
bare 5P3/2 intermediate state due to pumping to the 5D level. A discussion of the dependence
of EITT on the linewidth of the coupling beam ω, on its detuning and its intensity, will be
described elsewhere, in a more detailed presentation including also the study of the 87 Rb
isotope and the effect of an external magnetic field.
In summary, we have experimentally demonstrated electromagnetically induced two-photon

Table I – Rabi frequencies Ω’s at the sample and decay rates γ’s (the values for the latter include the
laser linewidth and the transit time broadening); all quantities are in MHz.

Beam ω1 ω2 ω
Ω
3
3
12
γ 8 3 3
460 EUROPHYSICS LETTERS

transparency (EITT) in room temperature rubidium atoms. A transparency of 80% at res-

onance has been obtained at a relatively low-control beam intensity. Our findings are well
explained by a four-level density matrix calculation employing realistic parameters. The fact
that electromagnetically induced transparency takes place also for two-photon processes is
particularly significant as it allows to widen the scope of EIT to one-photon parity-forbidden
transitions.

∗∗∗

It is a pleasure to thank G. Alzetta, E. Arimondo and M. Inguscio for stimulating

discussions, and G. S. Agarwal for useful comments. DW and JYG thank Scuola Normale
Superiore for hospitality. Financial support from NNSF of China and from INFM (PAIS
project of Section A) is also gratefully acknowledged.

REFERENCES

[1] Alzetta G., Gozzini A., Moi L. and Orriols G., Nuovo Cimento B, 36 (1976) 5; Arimondo
E. and Orriols G., Lett. Nuovo Cimento, 17 (1976) 333.
[2] Harris S., Physics Today, 50 (1997) 36; Arimondo E., Progress in Optics XXXV, edited by
E. Wolf (Elsevier Science, Amsterdam) 1996, p. 257, and references therein.
[3] Akulshin A. M., Barreiro S. and Lezama A., Phys. Rev. A, 57 (1998) 2996.
[4] Scully M. O., Phys. Rep., 219 (1992) 191; Gao J. Y. et al., Opt. Commun., 93 (1992) 323.
[5] Moseley R. R., Shepherd S., Fulton D. J., Sinclair B. D. and Dunn M. H., Phys. Rev.
Lett., 74 (1995) 670.
[6] Aspect A., Arimondo E., Kaiser R., Vansteenkiste N. and Cohen-Tannoudji C., Phys.
Rev. Lett., 61 (1988) 826.
[7] Hau L. V., Harris S. E., Dutton Z. and Behroozi C., Nature, 397 (1999) 594.
[8] Kash M. M. et al., Phys. Rev. Lett., 8 (1999) 5229; Budker D., Kimabll D. F., Rochester
S. M. and Yashchuk V. V., Phys. Rev. Lett., 83 (1999) 1767.
[9] Agarwal G. S. and Harshawardhan W., Phys. Rev. Lett., 77 (1996) 1039.
[10] Gao J. Y., Yang S. H., Wang D., Guo X. Z., Chen K. X., Jiang Y. and Zhao B., Phys.
Rev. A, 61 (2000) 23401.