0% found this document useful (0 votes)
36 views4 pages

Theoretical Analysis of Linewidth Narrowing in BFL

The document analyzes linewidth narrowing in Brillouin fiber ring lasers theoretically. It shows that the phase noise of the pump laser is transferred to the Stokes wave but is strongly reduced and smoothed by acoustic damping and cavity feedback. A simple analytical relation is derived connecting the full width at half maximum of the Stokes linewidth to the pump laser linewidth. Simulations based on the three-wave model of stimulated Brillouin scattering confirm that the Stokes phase fluctuations are strongly correlated to but weaker than the pump phase noise.
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
36 views4 pages

Theoretical Analysis of Linewidth Narrowing in BFL

The document analyzes linewidth narrowing in Brillouin fiber ring lasers theoretically. It shows that the phase noise of the pump laser is transferred to the Stokes wave but is strongly reduced and smoothed by acoustic damping and cavity feedback. A simple analytical relation is derived connecting the full width at half maximum of the Stokes linewidth to the pump laser linewidth. Simulations based on the three-wave model of stimulated Brillouin scattering confirm that the Stokes phase fluctuations are strongly correlated to but weaker than the pump phase noise.
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 4

PHYSICAL REVIEW A, VOLUME 62, 023803

Linewidth narrowing in Brillouin lasers: Theoretical analysis


Alexis Debut, Stéphane Randoux, and Jaouad Zemmouri
Laboratoire de Physique des Lasers, Atomes et Molécules, UMR 8523, Centre d’Études et de Recherches Lasers et Applications,
Université des Sciences et Technologies de Lille, F-59655 Villeneuve d’Ascq Cedex, France
共Received 26 August 1999; revised manuscript received 10 March 2000; published 14 July 2000兲
The linewidth narrowing observed in Brillouin fiber ring lasers is studied within the framework of the usual
three-wave model of stimulated Brillouin scattering. We show that the phase noise of the pump laser is
transferred to the emitted Stokes wave after being strongly reduced and smoothed under the combined influ-
ence of the acoustic damping and the cavity feedback. We then derive a simple analytical relation connecting
the full width at half maximum of the Stokes linewidth to that of the pump laser.

PACS number共s兲: 42.55.⫺f, 42.65.Es, 42.25.Kb

The Brillouin fiber ring laser is a highly coherent light damping rate of the acoustic wave. f ( ␨ , ␶ ) is a Langevin
source. Experiments have shown that the linewidth of the noise term describing the random thermal fluctuations of
Stokes radiation emitted by this laser can be several orders of density 共i.e., spontaneous scattering兲 occurring inside the fi-
magnitude narrower than that of the incident beam used to ber 关6兴. This noise is a zero-mean Gaussian process and it is
pump it 关1兴. Although this effect is commonly used in many ␦ correlated both in space and time. In the chosen normal-
applications such as gyroscopes 关2兴 or temperature sensors ization, f is of the order of 10⫺6 whereas the amplitude of the
关3兴, it is not clearly understood. In fact, the questions related optical fields is of the order of unity ( 兩 ␧ p 兩 , 兩 ␧ s 兩 ⯝1). Despite
to the noise properties or to the temporal coherence of Bril- its weak relative importance, spontaneous scattering plays a
louin lasers have not yet been attentively examined from the determining role in Brillouin generators by inducing a sto-
theoretical point of view. Brillouin lasers and conventional chastic dynamics 关4兴. In Brillouin lasers, spontaneous scat-
lasers have common noise sources. For instance, fluctuations tering is usually considered as the effect that is responsible
of the pumping mechanism or of the cavity length alter the for the initiation of the SBS process.
spectral purity of the emitted wave in both systems. How- As in Refs. 关5兴 and 关7兴, we study a Brillouin fiber ring
ever, contrary to lasers with population inversion, it is not laser in which pump recoupling is avoided by an intracavity
spontaneous emission but spontaneous scattering that funda- isolator. The boundary conditions characterizing this system
mentally limits the degree of monochromaticity of the Stokes read
radiation. The aim of this paper is to analyze the role of the
different noise sources existing in Brillouin fiber ring lasers ␧ p 共 ␨ ⫽0,␶ 兲 ⫽ ␮ e i␾0(␶), 共2a兲
and to characterize their influence on the temporal coherence
of the laser. In particular, we will show that the phase noise ␧ s 共 ␨ ⫽1,␶ 兲 ⫽Re i ␦ s ␧ s 共 ␨ ⫽0,␶ 兲 . 共2b兲
of the pump laser is the predominant noise source and we
will study in detail the connection between the linewidth of R is the amplitude feedback parameter and ␦ s is a detuning
the Brillouin laser and that of its pump laser. term expressing the fact that the cavity is not necessarily
Our theoretical study enters within the framework of the resonant. In an all-fiber laser, the fluctuations of the fiber
usual three-wave model of stimulated Brillouin scattering length can be reduced by stabilization techniques and the
共SBS兲. This model reproduces properly most of the dynami- fluctuations of ␦ s are then weak. Moreover, the use of cou-
cal behaviors experimentally observed in Brillouin genera- plers offers a great stability of the power feedback parameter
tors 关4兴 or lasers 关5兴. It reads R 2 . This is not the case for Brillouin fiber lasers operating in
low-finesse resonators. In these systems including an aerial
⳵ ␶ ␧ p ⫹ ⳵ ␨ ␧ p ⫽⫺gB␧ s , 共1a兲 arm, the relative fluctuations of R 2 can be of the order of
20% and the fluctuations of the resonator length can exceed
⳵ ␶ ␧ s ⫺ ⳵ ␨ ␧ s ⫽gB * ␧ p , 共1b兲 the optical wavelength. The variations of ␦ s are then greater
than 2 ␲ and mode hops are observed 关7兴. All these fluctua-
共 1/␤ A 兲 ⳵ ␶ B⫹B⫽␧ p ␧ s* ⫹ f 共 ␨ , ␶ 兲 , 共1c兲 tions in the cavity characteristics arise from mechanical and
thermal noise and their bandwidth is typically of the order of
where ␧ p , ␧ p , and B represent, respectively, the complex 100 Hz. These technical noise sources are responsible for
amplitudes of the pump, Stokes, and acoustic waves. These slow drifts of the frequency of the Brillouin laser but they do
equations in dimensionless form have been obtained by ne- not contribute to the intrinsic linewidth of the emitted Stokes
glecting the weak attenuation of the fiber. The time ␶ is radiation.
normalized to the transit time of the light inside the fiber. ␨ is In Eq. 共2a兲, ␮ is a dimensionless pump parameter and ␾ 0
the space coordinate that is normalized to the fiber length. represents the phase of the pump laser. For well-stabilized
The fields ␧ p and ␧ s are measured in units of the maximum single-mode lasers usually used to pump Brillouin fiber la-
pump field available at the entrance end of the fiber. g is the sers, the amplitude noise can be neglected so that only the
SBS coupling constant and ␤ A represents the normalized phase is a randomly fluctuating quantity. In the phase diffu-

1050-2947/2000/62共2兲/023803共4兲/$15.00 62 023803-1 ©2000 The American Physical Society


DEBUT, RANDOUX, AND ZEMMOURI PHYSICAL REVIEW A 62 023803

sion model commonly used to describe the field emitted by a


single-mode laser, ␮ is time independent whereas the phase
␾ 0 ( ␶ ) performs a random walk governed by the stochastic
Langevin equation

d ␾ 0共 ␶ 兲
⫽q 共 ␶ 兲 , 共3兲
d␶

in which q( ␶ ) is a ␦ -correlated Gaussian noise of zero mean


关8兴. The field spectrum of the pump laser is then a Lorentzian
with a full width at half-maximum 共FWHM兲 ⌬ ␯ p in units of
cavity free spectral range 共FSR兲 关9兴.
In most of the experiments, the linewidth of the pump
laser is of the order of a few tens of kHz. Apart from spon-
taneous scattering, the phase noise of the pump laser is the
noise source presenting the wider spectrum. We are now
going to analyze its influence on the noise characteristics of
the emitted Stokes radiation. To that effect, we will ignore
the influence of the low-frequency noise sources previously
mentioned. As they only induce slow variations of R and ␦ s , FIG. 1. Spatial profiles and temporal evolutions of the various
we can consider that these two parameters do not drift on the phases and amplitudes 共see the text兲. The parameters used in the
short time scale on which we analyze the influence of the simulations are those describing the laser of Ref. 关12兴 (g
pump phase noise. In particular, we will assume that the ⫽6.04, ␤ A ⫽10.93, R⫽0.36, ␮ ⫽0.7).
cavity is resonant for the Stokes wave ( ␦ s ⫽0). Frequency
pulling effects that are discussed in Ref. 关7兴 do not play an 共2兲 are integrated in the presence of phase noise generated by
important role and can thus be ignored. In our analysis, we Eq. 共3兲. The numerical procedure used is based on the
will also neglect the influence of the weak Langevin term method of characteristics described in Ref. 关11兴. As shown in
f ( ␨ , ␶ ) describing spontaneous scattering. As with spontane- Figs. 1共a兲 and 1共b兲, the phase of the acoustic wave nearly
ous emission in a conventional laser, this term will be re- follows the same temporal evolution as the phase of the
sponsible for the existence of a lower limit to the Stokes pump laser. The phase fluctuations of the Stokes wave are
linewidth. We will show at the end of this paper that this strongly correlated to that of the pump laser but they are also
limit ranges in the sub-Hertz domain. Let us finally mention much weaker. In these conditions, the variable ␪ ( ␨ ⫽0,␶ )
that we only consider situations in which the FSR of the ring oscillates slightly around 0 关Fig. 1共b兲兴. The trigonometric
laser is comparable to the width of the Brillouin gain curve. functions appearing in Eqs. 共4兲 can thus be expanded to the
The intensity of the backscattered Stokes wave is then time lowest order in ␪ . The equations for the amplitudes then
independent 关10兴 and the corresponding operating regime is become independent from that governing the evolution of the
called the ‘‘Brillouin mirror’’ 关5兴. phases. Since the amplitude ␮ of the incident pump field is
The first step of our theoretical analysis consists in trans- time independent, we can cancel all the time derivatives ap-
forming the complex amplitudes to modulus-phase form. We pearing in Eqs. 共4a兲, 共4b兲, and 共4c兲 and the field amplitudes
then obtain then depend only on ␨ . This is confirmed by numerical simu-
lations that show that the phase fluctuations of the pump
⳵ ␶ A p ⫹ ⳵ ␨ A p ⫽⫺gA a A s cos␪ , 共4a兲 source do not modify significantly the amplitudes of the
three waves involved in the interaction. They always remain
⳵ ␶ A s ⫺ ⳵ ␨ A s ⫽gA a A p cos ␪ , 共4b兲 very close to the stationary longitudinal profiles presented in
Fig. 1共c兲. With these approximations, the equations govern-
共 1/␤ A 兲 ⳵ ␶ A a ⫹A a ⫽A p A s cos ␪ , 共4c兲 ing the spatiotemporal evolution of the phases finally be-
come
⳵ ␶ ␾ p ⫹ ⳵ ␨ ␾ p ⫽⫺g 共 A a A s /A p 兲 sin ␪ , 共4d兲
⳵ ␶ ␾ p ⫹ ⳵ ␨ ␾ p ⫽⫺gA s2 共 ␨ 兲 ␪ , 共5a兲
⳵ ␶ ␾ s ⫺ ⳵ ␨ ␾ s ⫽⫺g 共 A a A p /A s 兲 sin ␪ , 共4e兲
⳵ ␶ ␾ s ⫺ ⳵ ␨ ␾ s ⫽⫺gA 2p 共 ␨ 兲 ␪ , 共5b兲
共 1/␤ A 兲 ⳵ ␶ ␾ a ⫽⫺ 共 A p A s /A a 兲 sin ␪ , 共4f兲
⳵ ␶ ␾ a ⫽⫺ ␤ A ␪ . 共5c兲
where ␪ ( ␨ , ␶ )⫽ ␾ s ( ␨ , ␶ )⫹ ␾ a ( ␨ , ␶ )⫺ ␾ p ( ␨ , ␶ ). A i and ␾ i (i
⫽p,s,a) represent, respectively, the amplitudes and phases The terms appearing in the right-hand sides of Eqs. 共5a兲 and
of the pump, Stokes, and acoustic waves. The SBS interac- 共5b兲 are weak and comparable. However they do not play the
tion is submitted to resonance and phase matching conditions same role in the two equations. Equation 共5a兲 must satisfy
imposing that ␪ is necessarily a weakly fluctuating variable. the boundary condition ␾ p ( ␨ ⫽0,␶ )⫽ ␾ 0 ( ␶ ). ␾ 0 ( ␶ ) acts as a
As the cavity is resonant, its average value is zero. This is source term inducing strong spatiotemporal variations of ␾ p
illustrated by numerical simulations in which Eqs. 共4兲 and 关see Figs. 1共a兲and 1共d兲兴. In these conditions, the term gA s2 ␪

023803-2
LINEWIDTH NARROWING IN BRILLOUIN LASERS: . . . PHYSICAL REVIEW A 62 023803

is only perturbative and our last approximation consists in phase to diffuse over one radian on the average 关13兴, is much
neglecting its influence. This means that the phase of the greater for the Brillouin laser than for the pump laser.
pump wave remains undisturbed in the interaction so that the If pump depletion is taken into account, S̃ 0 ( ␯ ) and ␾ ˜ 0( ␯ )
solution of Eq. 共5a兲 simply reads are no longer connected by a term that can be factorized into
two parts. This term can be calculated by taking into account
␾ p共 ␨ , ␶ 兲 ⫽ ␾ 0共 ␶ ⫺ ␨ 兲 . 共6兲 the expression of A 2p ( ␨ ) given in Ref. 关12兴 but contrary to
Eq. 共8兲, the averaging effect does not clearly appear in its
On the other hand, no source term appears in the boundary complicated analytical expression. However this term shows
condition verified by the Stokes phase 关 ␾ s ( ␨ ⫽1,␶ )⫽ ␾ s ( ␨ that the amplitude of the pump fluctuations is reduced by the
⫽0,␶ ) 兴 and the term gA 2p ␪ is the source inducing the spa- factor K not only near the threshold but whatever the value
tiotemporal variations of ␾ s . of ␮ . As we are now going to see, this is the most important
By taking into account the previous result 关Eq. 共6兲兴 that consideration for the evaluation of the Stokes linewidth.
has been verified by numerical simulations, our problem is The quantities of interest for the determination of the line-
reduced to the resolution of Eqs. 共5b兲 and 共5c兲 in the pres- widths are the variances ␴ 2p ( ␶ )⫽ 具 关 ␾ 0 ( ␶ )⫺ ␾ 0 (0) 兴 2 典 and
ence of the boundary condition characterizing the cavity
resonance 关 ␾ s ( ␨ ⫽1,␶ )⫽ ␾ s ( ␨ ⫽0,␶ ) 兴 . It can be solved ana- ␴ s2 ( ␶ )⫽ 具 关 ␾ s (0,␶ )⫺ ␾ s (0,0) 兴 2 典 关9兴. The phase noise being
lytically for any value of ␮ and therefore for any shape of not a stationary process, the brackets denote an ensemble
the pump profile A p ( ␨ ). However, for the sake of simplicity average performed at the time ␶ . In our phase diffusing
and in order to get some insight on the physical mechanisms model, ␴ 2p ( ␶ ) is a linear function of time 关 ␴ 2p ( ␶ )
responsible for the effects presented in Fig. 1, we are first ⫽2 ␲ ⌬ ␯ p ␶ 兴 that measures the evolution of the dispersion of
going to assume that the Brillouin laser does not operate very the values taken by the pump phase. As the phase fluctua-
far from its threshold. In these conditions, the pump deple- tions of the Stokes wave are much weaker than that of the
tion effect can be neglected and the function A 2p can be ap- pump wave, their dispersion is also much weaker. In fact, ␴ 2p
proximated to (⫺ln R)/g 关12兴. Owing to the spatial periodic- and ␴ s2 are simply connected through the relation ␴ 2p ⫽K 2 ␴ s2 .
ity imposed by the boundary condition ␾ s ( ␨ ⫽1,␶ )⫽ ␾ s ( ␨ Whatever the incident pump power, the FWHM of the
⫽0,␶ ), the function ␾ s ( ␨ , ␶ ) can be decomposed as Stokes spectrum is therefore given by
⫹⬁
⌬␯p
␾ s共 ␨ , ␶ 兲 ⫽ 兺 S n 共 ␶ 兲 e ik n ␨ 共7兲 ⌬ ␯ s⫽ . 共9兲
n⫽⫺⬁ K2

with k n ⫽2 ␲ n and S 쐓n ( ␶ )⫽S ⫺n ( ␶ ). The problem is then


By returning to the physical variables 关12兴, one can easily
solved by Fourier analysis and by using the orthonormality
show that K⫽1⫹ ␥ A /⌫ c , where ␥ A and ⌫ c , respectively,
condition
represent the damping rate of the acoustic wave and the cav-

冕 1 ity loss rate. ␥ A is equal to ␲ ⌬ ␯ B , where ⌬ ␯ B represents the


e ik n ␨ e ⫺ik m ␨ d ␨ ⫽ ␦ nm , FWHM of the Brillouin gain curve. ⌫ c is equal to
0 ⫺c ln R/nL where c/n is the light velocity in the fiber of
length L. Equation 共9兲 represents the main result of this paper
it can be shown that and the previous expression of the coefficient K clearly
shows that the narrowing effect observed in Brillouin lasers
⫺ln R e ⫺i ␲␯ sin ␲␯ is due to the combined influence of the acoustic damping and
S̃ 0 共 ␯ 兲 ⫽ ␾
˜ 0 共 ␯ 兲 , 共8兲
␤ A ⫺ln R⫹i2 ␲␯ ␲␯ of the cavity feedback.
The analytical result given by Eq. 共9兲 has been verified by
where ␾ ˜ 0 ( ␯ ) and S̃ 0 ( ␯ ), respectively, represent the Fourier the numerical integration of Eqs. 共4兲 and 共2兲 over an en-
transforms of ␾ 0 ( ␶ ) and S 0 ( ␶ ). Equation 共8兲 shows that the semble of 50 000 realizations of the random process ␾ 0 ( ␶ ).
incident phase noise is filtered by the association of two Such a statistical treatment permits us to determine the vari-
linear systems 关14兴. The first one is a low-pass filter that ances ␴ 2p and ␴ s2 at different times. The data thus obtained
reduces the amplitude of the pump fluctuations by a factor are then fitted to straight lines 共see Fig. 2兲 and the ratio
K⫽( ␤ A ⫺ln R)/(⫺ln R) over a bandwidth equal to ( ␤ A between the two slopes gives the numerical value of the co-
⫺ln R)/2␲ in units of cavity FSR. The second one is a sys- efficient K 2 . For parameter values characterizing a Brillouin
tem that smoothes the fluctuations by averaging them over a laser operating in a low-finesse resonator 关7,11,12兴, the value
time interval ⌬ ␶ ⫽1. As shown in Fig. 1共d兲, the function of K 2 given by the analytical relation is 137 and that deduced
␾ s ( ␨ ⫽0,␶ ) can be well approximated to S 0 ( ␶ ) and although from numerical simulations presented in Fig. 2 is 144. The
it has been established by neglecting the pump depletion, Eq. slight relative difference between the two values arises from
共8兲 describes qualitatively well the behavior reported in Fig. the fact that the function ␾ s ( ␨ , ␶ ) has been approximated to
1共a兲. The fluctuations of the Stokes phase are indeed much S 0 ( ␶ ) and that the influence of the terms S n ( ␶ ) (n⫽0) has
weaker than the phase fluctuations of the pump laser and the been neglected.
high frequencies found in ␾ 0 ( ␶ ) do not appear in ␾ s (0,␶ ). If the linewidth of the pump laser is of the order of several
The optical coherence time, defined as the time it takes to the tens of kHz, a Brillouin laser operating in a low-finesse cav-

023803-3
DEBUT, RANDOUX, AND ZEMMOURI PHYSICAL REVIEW A 62 023803

approximation that consists in neglecting spontaneous scat-


tering must be questioned. Before analyzing the importance
of this effect, let us first recall that spontaneous emission can
be seen as a process that randomly perturbs the amplitude
and the phase of the optical field emitted by a conventional
laser 关15兴. By analogy with that description, spontaneous
scattering can be considered as an effect perturbating the
amplitude and the phase of the acoustic wave. In particular,
it will lead to a diffusion of the phase of the acoustic wave.
The influence of this diffusion process will become notice-
able when it will be comparable to the diffusion process
induced by the pump laser. As the phase of the acoustic
wave nearly follows the same evolution than the phase of the
pump laser 关see Figs. 1共a兲 and 1共b兲兴, spontaneous scattering
cannot be neglected if q( ␶ ) is of the same order of magni-
tude as the Langevin term f ( ␨ , ␶ ). The pump spectrum of
Fig. 2共b兲 has been obtained for a value of q( ␶ ) of the order
of 10⫺2 ( 具 q 2 典 ⯝10⫺4 ) and its dimensionless FWHM is 0.002.
This corresponds to a pump linewidth of 40 kHz for a Bril-
louin laser characterized by a FSR of 20 MHz. As f ( ␨ , ␶ ) is
of the order of 10⫺6 , spontaneous scattering is then effec-
tively negligible. This is no longer the case for pump line-
FIG. 2. Same parameters as in Fig. 1. Temporal evolution of the widths as narrow as 10 Hz and Stokes linewidths that fall in
variance of the phase noise 共diffusion process兲 and corresponding
the sub-Hertz domain. Obviously, the previous analysis just
normalized Lorentzian spectrum of the 共a兲,共b兲 incident pump field;
allows us to give an idea of the order of magnitude of the
共c兲,共d兲 Stokes field.
effect. Additional studies must be undertaken to precisely
ity will emit a Stokes radiation with a linewidth of several characterize the influence of spontaneous scattering on the
hundreds of Hz. In an all-fiber ring resonator, a finesse linewidth of the Brillouin laser.
greater than 100 can be easily achieved and the ratio ␥ A /⌫ c
is then typically of the order of 100. This means that the The Centre d’Etudes et de Recherches Lasers et Applica-
Stokes linewidth is at least 104 times narrower than the line- tions 共CERLA兲 is supported by the Ministère Chargé de la
width of the pump laser. This order of magnitude is effec- Recherche, the Région Nord/Pas de Calais, and the Fonds
tively the one that has been experimentally measured in that Européen de Développement Economique des Régions. This
kind of laser 关1兴. In these conditions, the linewidth of the work was partially supported by the European contract ‘‘In-
Stokes radiation becomes of the order of a few Hz and the tereg II Nord-Pas de Calais/Kent.’’

关1兴 S.P. Smith, F. Zarinetchi, and S. Ezekiel, Opt. Lett. 16, 393 Optics 共Cambridge University Press, Cambridge, 1995兲.
共1991兲. 关9兴 A. Yariv, Quantum Electronics, 3rd ed. 共Wiley, New York,
关2兴 F. Zarinetchi, S.P. Smith, and S. Ezekiel, Opt. Lett. 16, 229 1988兲.
共1991兲. 关10兴 S. Randoux, V. Lecoeuche, B. Ségard, and J. Zemmouri, Phys.
关3兴 P.A. Nicati, K. Toyama, S. Huang, and H.J. Shaw, Opt. Lett. Rev. A 51, R4345 共1995兲.
18, 2123 共1993兲. 关11兴 V. Babin, A. Mocofanescu, V.I. Vlad, and M.J. Damzen, J.
关4兴 A.L. Gaeta and R.W. Boyd, Phys. Rev. A 44, 3205 共1991兲. Opt. Soc. Am. B 16, 155 共1999兲.
关5兴 C. Montes, A. Mamhoud, and E. Picholle, Phys. Rev. A 49, 关12兴 S. Randoux, V. Lecoeuche, B. Ségard, and J. Zemmouri, Phys.
1344 共1994兲. Rev. A 52, 2327 共1995兲.
关6兴 R.W. Boyd, K. Rzażewski, and P. Narum, Phys. Rev. A 42, 关13兴 M.P. van Exter, S.J.M. Kuppens, and J.P. Woerdman, IEEE J.
5514 共1990兲. Quantum Electron. QE28, 580 共1992兲.
关7兴 V. Lecoeuche, S. Randoux, B. Ségard, and J. Zemmouri, Phys. 关14兴 A. Papoulis, Signal Analysis 共McGraw-Hill, New York, 1984兲.
Rev. A 53, 2822 共1996兲. 关15兴 M.P. van Exter, W.A. Hamel, and J.P. Woerdman, Phys. Rev.
关8兴 L. Mandel and E. Wolf, Optical Coherence and Quantum A 43, 6241 共1991兲.

023803-4

You might also like