1
Southampton SO17 1BJ, UK
Telephone: +44 1703 593150
e-mail: light@orc.soton.ac.uk
Optoelectronics Research Centre
University of Southampton
Acousto-optic superlattice modulator
using a fiber Bragg grating
W.F.Liu, *P.St.J.Russell and L.Dong
*Optoelectronics Group, School of Physics, University of Bath, Bath BA2 7AY, UK
Abstract
A modulator is reported in which an extensional acoustic wave is launched along a fiber Bragg
grating. The acousto-optic superlattice effect causes an enhancement in reflectivity within a narrow
spectral region on both sides of the Bragg wavelength. For a fixed acoustic propagation direction, the
Doppler shift can be either positive or negative, depending on whether the wavelength of the incident
light lies above or below the Bragg condition. The device can function as a Bragg cell and a tunable
filter.
All-fiber acousto-optic devices have potential uses as frequency shifters, multiplexers, modulators,
and tunable filters. The ease with which these devices can be spliced into systems, and the
consequent low insertion loss, make them an attractive alternative to pigtailed bulk Bragg cells.
Previous designs of intermodal coupler include dual-mode fibers supporting an acoustic flexural
wave [1,2], coupling between the polarization normal modes of a high-birefringence fiber by
means of a torsional acoustic wave [3], and a high-performance device based on a four-port
fused-taper null coupler [4]. All these devices share the requirement that the acoustic wavelength
must match the intermodal beat length, L = 2B/)$, where )$ = *$ - $ * and $ and $ are the
B 2 1 1 2
propagation constants of the modes.
In this Letter we describe a modulator (briefly reported for the first time in Ref. 5) in which a fiber
Bragg grating is excited by an axially propagating extensional acoustic wave. The underlying
principle is acousto-optic superlattice modulation (AOSLM), first proposed in 1986.[6,7] In
AOSLM the counterpropagating optical modes (the Bloch waves [8]) of the fine-pitch Bragg
grating are coupled by a course-pitch acoustic wave, the superposition of the two forming a
superlattice. Coupling is maximum when the inter-Bloch-wave beat period matches the acoustic
wavelength.
The forward-travelling (group velocity in the +z direction) Bloch wave in a fiber Bragg grating
can be closely approximated as a constant superposition of two coupled counterpropagating
guided modes with wave vectors k in the form [8]
k = {±K + h [1 - (26 / h) ] } /2, (1)
2 ½
where K = 2B/7 is the Bragg grating vector and 7 is its physical pitch. Coupling constant 6 and
dephasing parameter h are defined as 6 = Mk /4 and h = 2k - K, where k = Tn /c is the average
o o o o
wave vector in the Bragg grating, n is the modal phase index, and M is the modulation depth of
O
the dielectric constant, i.e., , = n (l + M cos Kz). Reversing the sign of h in Eq. (1) yields the
r O
2
wave vectors of the complementary backward-travelling Bloch wave. The presence of two wave
vectors in each Bloch wave, combined with their strong dispersion with frequency (forming a stop
band in the range -26 <h < 26), yields three different acousto-optic resonance conditions as
opposed to one in normal fiber. Two of these conditions are at approximately the usual Brillouin
frequencies (~10GHz at 1530nm). The third, which concerns us in this Letter, is indicated on the
T - k diagram in Fig. 1. This condition occurs between the two forward (and the two backward)
,(z) & n
2
o
Mn
2
o
' cos[Kz % asin(k
s
z)]
' J
o
(a) cosKz %
j
x
n' 1
J
n
(a)[cos(Kz % nk
s
z)
% (& 1)
n
cos(Kz & nk
s
z)],
2
Fig. 1. Frequency wave-vector diagram for a Bragg
grating. The group velocity is proportional to the
slope of the curves. In case A, a forward-travelling
acoustic wave [(TT , k ) vector shown by the arrow
s s
from b to f ] couples a forward-travelling Bloch
l l
wave (f) into a downshifted backward one (b ). In
l l
case B, forward Bloch wave f is coupled into
2
anomalously downshifted backward Bloch wave b
2
by backward-travelling acoustic wave. The thick
horizontal dashed lines join the (TT, k) points of the
forward and backward fiber modes of each Bloch
wave.
wave vectors of the two
Bloch waves and has an associated frequency shift that can lie between 0 and several hundred
megahertz (for clarity, acoustic frequency shift T is exaggerated in Fig. 1). Note that in
s
one-dimensional photon-photon scattering the two-vector (T, k) must be conserved, i.e., (T ,
out
k ) - (T , k ) = (T , k )~ where the frequency components must all be positive (acoustic
out in in S s
parameters have the subscript ); we ensure this by choosing the input and output two-vectors so
s
as to make T - T = T 0. The group velocity of each Bloch mode is given by *T/*k = ±(c/n )
out in s > o
[1 -(26/h) ] which tends to zero at each stop-band edge and changes sign on opposite sides of
2 ½
the lines k = ±K/2. An intriguing feature of superlattice coupling is that the direction of the
Doppler shift can be reversed. On the upper stopband branch, an acoustic wave travelling into the
incident Bloch mode will produce a frequency-upshifted reflected Bloch mode. On the lower
stop-band branch, however, the
reflected mode will be anomalously
frequency downshifted. As we show
below, these predictions are
confirmed by experiment.
An acoustic wave of average power
P will produce a peak strain of s =
s O
[2P /(EA< )] ~ where E is Young's
s gs
½
modulus, A is the fiber area, and <
gs
is the acoustic group velocity. This
strain field will sinusoidally modulate
both the average index and the pitch
of the grating. Close to the
stop-band edges, where the group
velocity is very small, both of these
effects are significant. Further from
the stop-band edges, however, pitch
modulation dominates. Since the
acoustic wave travels some 10 x
5
more slowly than the light, we
ignore its temporal dependence
(while remembering the Doppler
shift). For a strain field in the form
s(z) = s cos (k z)~ the resulting
O s
relative dielectric constant ,(z) is
given by (2)
c)8
8
2
. )< '
c
2n
o
f
s
<
s
2
%
6
B
2
½
.
cf
s
2n
o
<
s
,
3
Fig. 2. Experimental setup for monitoring the response
of AOSLM. Light from a tunable single-frequency
diode laser (TL) is divided at a fused taper coupler
(FC1), one half going to the fiber Bragg grating (FBG)
and the other to a bulk Bragg cell (BC). The light
reflected from the AOSLM is combined with the
frequency-shifted light from the Bragg cell at a second
coupler (FC2), and the mixed signal is detected at a
square-law detector (D); the redundant light in the
second arm of FC2 is eliminated in an index-matching
cell (IMC). A piezoelectric transducer (PZT) and a
fused-silica horn (SH) are used to excite the AOSLM.
Fig. 3. Reflectivity spectrum of the undistributed
Bragg grating (top) and the spectrum of the
enhancement in reflectivity due to the acoustic wave
(bottom).
here a = Ks /ks. It is clear that a
O
sequence of ghosts of the original
fiber grating forms at spatial
frequencies given by successive
spatial sidebands of K. The
amplitudes of these sidebands are
given, for small argument *a* <<1,
by J (a) . a / (2 n!). The AOSLM
n
n n
resonance occurs when the acoustic
wave vector matches the
wave-vector difference between two
Bloch waves. This leads to the
condition
(3)
where f and < are the acoustic
s s
frequency and the phase velocity and
)< and )8 are the optical frequency
and wavelength shifts from the
Bragg condition of the fiber grating.
The approximate expression on the
right-hand side of Eq. (3) is valid
when the AOSLM condition occurs
far from the Bragg condition. For
bulk silica < is 5760 m/s, and given
s
our experimental value of 6 =
1.3/mm, Eq. (3) shows that AOSLM
resonance will occur, e.g., at NM
from the Bragg wavelength if the
acoustic frequency is 15MHz (NB:
< falls when the acoustic
s
wavelength is comparable with the
fiber radius [9]). It can be shown
that the coupling constant between
the counterpropagating Bloch waves
at the first sideband in Eq. (2) is
given approximately by 6 =
1
6J (Ks /k ) . 6Ks /(2k ). Figure 2
1 0 s 0 s
shows the experimental setup. The
fiber grating was written with a
phase mask in a boron-codoped
germanosilicate fiber (N.A., 0.115;
cutoff, 1300nm). The UV source
was an ArF excimer laser at 193nm.
The grating was 3mm long, with a
4
Fig. 4. RF spectra at the detector for sidelobes
at 88 > 88 (top) and 88 < 88 (bottom). The peak
B B
at 8MHz is caused by beating between the
unshifted and the shifted Bragg grating
reflections. The presence of a band on the
low-frequency side of the 80MHz reference
signal indicates that the AOSLM signal is
frequency upshifted.
bandwidth of 0.7nm, a Bragg wavelength of 1526.5nm, and a coupling constant 6 = 1.3/mm. A
piezoelectric transducer, bonded to a silica horn and driven by a rf signal generator, was used as
the source of acoustic waves. The diameter of the silica horn was tapered from 3mm to 125µm
over a distance of ~7cm. One end of the fiber grating was spliced directly to the horn. To test
whether the light reflected from the grating is upshifted or downshifted, the light was mixed at a
square-law detector with light frequency shifted in a conventional Bragg cell. We found that the
interface loss between the transducer and the silica horn is a very important factor in obtaining
efficient coupling of the longitudinal acoustic wave to the fiber grating. The light was detected
at one of the output ports of 3dB coupler FC2, with the remaining port being immersed in
index-matching oil.
The reflectivity spectrum of the undisturbed Bragg grating and the spectrum of the enhancement
in reflectivity due to the acoustic wave are plotted in Fig. 3. AOSLM is observed on both sides
of the stop band. The acoustic frequency was 8.02MHz, and the electrical drive power to the
transducer was 350mW. The wavelengths of AOSLM reflection were at 1525.3nm on the
short-wavelength side and 1527.6nm on the long-wavelength side. The magnitude of the
wavelength shift )8 from Bragg wavelength is in each case approximately 1.15nm, and the
bandwidth of the AOSLM effect is ~0.2nm. In a series of measurements of wavelength shift )8
as a function of acoustic frequency (4 to 15.2MHz in steps of 1MHz) the agreement between
theory [Eq. (3)] and experiment was better than a few percent. The results showed that AOSLM
could be used to form a tunable filter with a tuning rate of 0.15nm/MHz.
The results of the heterodyne experiment are shown in Fig. 4. A tunable single-frequency laser
was used, and the acoustic drive frequency was 8MHz. On the short-wavelength side of the
Bragg condition the reflected light was
frequency upshifted (conventional
Doppler effect), whereas on the
long-wavelength side it was downshifted
(anomalous Doppler shift) as predicted
by theory.
Since the spatial sidebands produced by
the acoustic wave [Eq. (2)] can be
regarded as weak ghosts of the strong
permanent Bragg grating, the FWHM
bandwidth of the AOSLM effect is
expected to be that of a weak Bragg
grating of the same length, namely, )8
opt
= 1.398 /(BLn ) where L is the fiber
2
o
length and the numerical factor relates to
sinc (1.39) = 1/2. For L = 3mm at 8 =
2
1530nm this equation predicts )8 =
opt
0.23nm, which is in reasonable
agreement with the experimental results.
For small acoustic powers, it can be
shown that the efficiency of conversion
0 can be approximated by 0 . (6 L) =
1
2
[6LJ (Ks /k )] . For the parameters in
1 o s
2
5
our experiment 0 ~ 15%, and using the relationship given in the paragraph above Eq. (2) to relate
acoustic power to s , it can be shown that of the 350mW electrical drive power, only 57mW is
O
actually reaching the Bragg grating as acoustic power. Since the efficiency scales linearly with
the power and as the square of length, there is clearly scope for improving the AOSLM efficiency
by better acoustic transducer design, increasing the grating length, or using thinner fibers. Some
advantages would result if the core was offset from the centre of the fiber and flexural waves were
used.
Substantial improvements are expected with better acoustic transducer design, a longer fiber
grating, and thinner fiber.
References
1. B.Y.Kim, J.N.Blake, H.E.Engan and H.J.Shaw, Opt. Lett. 11, 389 (1986).
2. J.N.Blake, B.Y.Kim, H.E.Engan and H.J.Shaw, Opt. Lett. 12, 281 (1987).
3. M.Berwick, C.N.Pannell, P.St.J.Russell and D.A.Jackson, Electron. Lett. 27, 713 (1991).
4. T.A.Birks, S.G.Farwell, P.St.J.Russell and C.N.Pannell, Opt. Lett. 19, 1964 (1994);
erratum, Opt. Lett. 21, 231 (1996).
5. W.F.Liu, P.St.J.Russell, D.O.Culverhouse and L.Reekie, in Conference on Lasers and
Electro-Optics, Vol. 9 of 1996 OSA Technical Digest Series (Optical Society of America,
Washington, D.C., 1996), pp. 243-244.
6. P.St.J.Russell, Phys. Rev. Lett. 56, 596 (1986).
7. P.St.J.Russell, J. Appl. Phys. 59, 3344 (1986).
8. P.St.J.Russell, J. Mod. Opt. 38, 1599 (1991).
9. H.Kolsky, Stress Waves in Solids (Oxford U. Press, London, 1953), Chap III.
