Nonlinearity in Fiber Transmission
-
Upload
vanessa-sanchez -
Category
Documents
-
view
19 -
download
0
description
Transcript of Nonlinearity in Fiber Transmission
7/17/2019 Nonlinearity in Fiber Transmission
http://slidepdf.com/reader/full/nonlinearity-in-fiber-transmission 1/5
1232 PROCEEDINGS OF THE IEEE, VOL. 68 , NO.
10,
OCTOBER 1980
1661 E.
G.
Ramon and M. D. Bailey, “Bitaper star couplerswithup
tion, p. 636, Sept. 1978.
to 10 0 fibre channels,” Electron. Lett., vol. 15, no. 15, p. 432,
July 5, 1979.
[
671 T. Ito et al., “Bidirectional tapered fiber star couplers ,” in Proc.
4t h European Conf Optical Communication, p. 318, Sept. 1978.
1681 W. J. Tomlinson, ‘Wavelength multiplexing n multimode optical
fibers,”Appl. Opt.,vol. 16, no. 8, p. 2180, Aug. 19 77.
1691 K. Kobayashi and
M. Seki
“Micro-optic grating multiplexers or
fiber-optic communications,” in Tech. Dig. Topical Meet. Optical
[
701 R. Watanabe and K. Nosu, “Optical demultiplexer using concave
Fiber Communication, p. 54, Mar. 1979.
grating in 0.7 -0. 9 pm wavelength region ” Electron. Lett., vol.
16, no. 3, p. 10 6, Jan. 1980.
[ 711
Y.
Fujii et al., “Opticaldemultiplexerusingasiliconechelette
grating,” IEEE J . QuantumElectron., vol.QE-16, p. 16 5, Feb.
1980.
1721 R. Watanabe et al., “Optical grating mult iplexer n he1.1-1.5
pmwavelengthregion,” Electron. Lett., vol.16,no.3, p. 108,
Jan. 1980.
Nonlinearity in Fiber Transmission
ROGERS
H.
STOLEN
Abstract-Procedures
are
presented for estimating critical powers for
nonlinear optical processes in single-mode
fiber
transmission systems.
Crosstalk due to
Raman gain in
multiplexedsystems
can
appear at
powers of a few mW. The effects of self-phase modulation and stimu-
lated Brillouin scattering can appeararound
100 mW
while ypical
stimulated Raman
hreshold
powers are a few watts.
I .
INTRODUCT ION
PTICAL fibers can exhibit frequency conversion due to
nonlinear processes leading to loss, pulse spreading,
crosstalk, and in some cases, physical damage, The prob-
lems can be particularly severe in very long, low-loss fibers
because thecentralfeature of fibernonlinearities is theex-
change of length for optical intensity
[
11
- [ 3 ] .
Even
so,
critical or threshold powers have
so
far been high enough to
fy neglecting optical nonlinearities in transmission system
design. Recent interest n single-mode submarine-cable sys-
tems may, however, change this situation. Extremely low-loss
fibers
[ 4 ] ,
higher powers
[ 5 ]
and narrow-band lasers
all
move
toward the regime where nonlinear effects become important.
Other developments such as small fiber cores to increase the
minimum dispersion wavelength [6] and the use of germania
glass [71 would also enhance nonlinear effects.
This paper att empts to establish critical powers for various
onlinear processes and to describe their effects n ransmis-
sion systems. Worst case limitsare relatively straightforward
to calculate but, as will be seen, thesepowers can often be
safely exceeded. The nonlinear processes of interest are
timulated Raman cattering which serves to llustra te he
asic concepts nd could lead to crosstalk
in
wavelength-
ultiplexedystems, stimulated Brillouin scattering which
s the most ikelynonlinear loss mechanism, and self-phase
odulation which leads to pulse broadening. Another process,
arametric four-photon r three-wave mixing, could also
cause crosstalk but will only be discussed briefly since hese
ffects would appear at higher powers.
Manuscript received June 3 1980.
The
author is with Bell Laboratories, Holmdel, NJ 0 77 33 .
11.
STIMULATEDAMAN SCATTERING
A .
RamanAmplification
The Raman interact ion makes a fiber an optically pumped
optical amplifier. The process can be viewed as a three-
wave parametricamplifier in which pump and signal (called
the Stokes wave) are optical waves and the idler is a highly
dampedvibrational wave. The equivalent quantummechan-
ical interpretation is one of stimulated scatteringby the Stokes
photons. Raman amplificationepends onhentensity,
which is the pump power
P
divided by the spot size
A ,
the
interaction length L and the Raman gain coefficient g.
Amplification xp
Yl.
In low-loss fibers the ength can be more than a kilometer
which,rom ( l ) , reduces the powernd gain coefficient
necessary for significant Stokes amplification.
It is not necessary to injecta signal to see the effects of
Raman gain since there is always some light present from
spontaneous Raman scattering. If the powernd length
are large enough, this weak signal can be amplified to he
level of pumpdepletion. Because of theexponential ampli-
fication there is onlya small difference n pump powerbe-
tween negligible conversion and near total conversion of
pum p o signal and it is possible to define a threshold or
critical power
[
81.
16A
gL
p
E .
B.
Fiber Parameters
1 ) Raman Gain Coeffic ient: The Raman gain coefficient for
fused silica
as
derived from the spontaneous Raman spectrum
[9] is shown in Fig. 1. The gain sgiven for a pump wave-
length
of
1.0 pm.
To
convert t o a different pump wavelength
the gain coefficient scales by l / X At 1.Opm a shift of
500
cm-’ corresponds approximatelyto
5 nm.
In
general, dopants such as boron, germanium, or phos-
0018-9219/80/1000-1232 00.75 O
1980 IEEE
7/17/2019 Nonlinearity in Fiber Transmission
http://slidepdf.com/reader/full/nonlinearity-in-fiber-transmission 2/5
STOLEN:NONLINEARITY IN FIBER TRANSMISSION
1 ''''''''''1
0
200 400
600
800 IO00
1200 1400
FREOUENCY
SHIFT
(ern- )
Fig. 1 . Raman gain versus frequency hift or used ilica at apump
wavelength of
1.0
pm. The gain coeffic ient cales inversely with
pump wavelength and it is assumed that polarization is maintained.
phorusdonot appreciablymodify the silica gain spectrum
[ 101 particularly in the small percentages used in extremely
low-loss fibers. One should keep in mind, however, that there
are glasses such as pure GeOz which hasa gain coefficient
9.2X that
of
silica [ 11 There are also many silicate glasses
with
g in
curves which look qui te different from fused silica
[ 121 although the peak gains are not very different.
2 Length: Amplification will decrease inongiber
because of pumpabsorptiondue to fiber loss. This is han-
dled by replacing L in (1) by an effective length
[
13
-
€ I
L e f f=
where
I
is theactual engthand
Q
is the linear absorption
coefficient. In the present discussion we are usually inte r-
ested in he limitwhere Leff = l / ~ . orexample, l /a ina
dB/km fiber is 4.34 km.
3 Power:
The power
P
in (1) is the value at the fiber input
but for pulses there is always a question of whether to use
peak or average power in the gain calculation. We will use th e
average power, since n fiber transmission the fibers are long
and the optical pulses are closely spaced so group-velocity dis-
persion will cause relative slippage of pump and Stokes pulses
which averages the gain.
4
Area: We will use the fiber core area
as
the spot size for
single and multimode fiberswith step or graded boundaries.
More carefulalculationsequire computation
of
overlap
integrals
[
11, [21, 9]but he core-size approximation is
adequate within a factor of two. The approximation is very
good in single-mode fibers at cutoff of the second mode and
forstep-indexmultimode fiberswhereenergy
is
uniformly
distributed through the modes by either the initial excitation
or by mode mixing
[
141. In single-mode fibers below cutoff
the effective area is larger than the core size and in graded ndex
fibers the effective area can be smaller than the core size.
5 Polarization: The Raman gain coefficientn Fig. 1
assumes the ideal case in which the fiber maintains polariza-
tion. By maintaining polarization one usually refers to linear
polarization but hisstatement applies to circular and ellip-
tical polarizations as well. In typical single-mode fibers,
which do not preserve polarization, the polarizations of pump
1233
and stokes waves get out
of
stepand he average Raman
gain s reduced by a factor
of
two
[
151
.
While polarization-
maintaining ibers can be made [ 161, t is likely that such
properties would be sacrificed toobtain minimum loss in
fiber ransmission
so
wewill reduce the Raman gain by a
factor
of
two in all estimates of Raman amplification.
C.
Examples
of
Raman Amplif icat ion
Above thetimulated Ramanhresholdower, light is
effectivelyost by downshifting
to
theStokes Raman fre-
quency. An expression for he thresholdpower as defined
by (2) is given in Table I. Here it is assumed that polarization
is not preserved, thespot size can be approximated by the
corearea, and hat he fiber is sufficiently ong to replace
le^
by I/ in (3). As an example, we choose a single-mode
fiber
of
8j tm core diameter, 0.3-dB/km loss at a wavelength
of 1.5 pm, and a length much greater than 1/a. The threshold
power
is
then 1.7 W which is probably well above practical
power levels for communication applications.
Theeffects of Ramanamplification can appear atmuch
lowerowersn wavelength multiplexedystems where
energy can be transferred
to
a onger wavelength channel.
We choose as a criterion for a critical power that the ampli-
fication ofone wavelength by another should be ess than
1 percent which corresponds to 20-dBcrosstalk. This means
that exp g P , L / A) = 1.01 s gP,L/A
=
0.01 rather than 16 as
for hestimulated hreshold. n he worstcase,one is un-
lucky enough to choose a frequency separation corresponding
to t he maximum Raman gain. This case is used for the expres-
sion in Table
I .
For the same example of an 8 j tm core fiber
at h = 1.5 pm the critical power is then 1.1 mW.
This
shows
tha t it will be essential to keep multiplexed wavelengths sep-
arated by more than 500 cm-'
As a f iial example, we consider reducing the Raman thresh-
old power with feedback supplied by reflection from poorly
matched ends. The condit ion for oscillation is that the round-
tripStokes amplificationequals thecombined transmission
and eflection oss. Note hat backwards Raman gain is the
same
as
forward gain. To simplify the calculation weuse
the case of a
3
km,
10
pm core fiber at h =
1
Opm and assume
the loss at both hepump andStokes wavelengths is 1.0
dB/km. The fiber length is less than the absorption length so
(3)
is
used to calculate an effective ength of 2.17 km. If a
reflectivity of 4 percent is arbitrarilychosen for each end,
the round trip transmission
is
4.0
X lo4
so gain equals loss
at exp 2 g P t L / A)= 2500. Threshold power is now 2.9 W
as
comparedo an 11.8-W single-pass threshold in the same
3-km iber and a hreshold of 5.9 W ina similar fiber of
length much greater than
l/a.
111. S T I M U L A T E DRILLOUIN CATTERING
Stimulated Brillouin scattering can have a peak gain 300
times that for stimulated Raman scattering so threshold can
occur at much lower powers
[
171. The interaction is between
light and acoustic waves rather than with high frequency op-
tical phonons and because of wave vector matching considera-
tions [ I ] [31 Brillouin amplificationoccurs only in the
backward direct ion. The frequency shift is given by
2nV
v
=
(4)
where n is the refractive index and V is the velocity
of
lon-
7/17/2019 Nonlinearity in Fiber Transmission
http://slidepdf.com/reader/full/nonlinearity-in-fiber-transmission 3/5
1234
PROCEEDINGS OF THEIEEE,
VOL.
68, NO. 10, OCTOBER
1980
rABLE I
T h r e s h o l an ar i t i c a lo w e r s or st imulated Raman
s c a t t e r i n g , Raman a m p l i f i c a t i o no ra v e l e n g t h
s e p a r a t i o no r r e s p o n d i n go maximum daman g a i n ,
s t i m u l a t e dB r i l l o u i n c a t t e r i n g , a n d e l f - p h a s e modu-
l a t i o n . a v e l e n g t h A a n do r e i a me t e r d a r e n rm
f i b e r
loss
i s i n dB/km and aserbandwidth
b v
i s i n
GHz. D i s t h e u t ya c t o r .T h e s e u mb e r s r e a l c u -
l a t e dnh e limit w h e r eh ei b e r i s much longer
t h a nh eb s o r p t i o ne n g t h ,h ep p r o x i m a t i o nh a t
t h ef f e c t i v er e a
i s
t h eo r er e a ,n ds s u m i n g
t h a to l a r i z a t i o n
i s
n o ta i n t a i n e d .
As
e x a mp l e s ,
p o w e r s r e i v e n o r n 9 rm c o r ed i a m e t e r i b e rw i t h
a loss of .3 dB/kn a t a wave length o 1 . 5
J m
For
t h e r i l l o u i n h r e s h o l d A w was 1.0 GHz andhe uty
f a c t o r was 0 . 5 o r h ec r i t i c a l p o we r r o m e l f- p ha s e
mo d u l a t i o n .
Stimulated Raman
P t
=
5 . 9 ~ 1 0 - ~ 6 ~ 1 5 a t t s 1.1 w
Raman Amplif ication
P C
=
3 . 7 ~ 1 0 - ~ d ’ 1 5a t t s
1.1
mu
S t i m u l a t e da r i l l o u i n
P t =
4.4X10 3d2A25Av
w a t t s 1 9 0
S e l f - 3 h a s eE o d u l a t i o n
P C = 2 . 2 ~ 1 0 - ~ d ~ A S Dw a t t s 2 mu
gitudinal sound waves. For = 1.O pm in fused silica, us =
17.2 GHz.
Brillouin gain can be quite different depending on whether
the pump linewidth
A u , )
is much less or much greater than
the Brillouin linewidth A v B ) . Brillouin linewidths are typi-
cally around 100 MHz so communications systems will operate
in the limit where Au, >>A u B . In this imit the Brillouin
gain coefficient for silica-based glasses becomes
1.73 X
lo-’’
h 2 A
p
=
cm/W
where h is the wavelength in micrometers, Au, i s the pump
bandwidth (FWHM) in GHz and the constant contains factors
such
as
the elasto-optic coefficient, the density and the sound
velocity. In the opposite limit where Av , <<A u B , A u , in (5)
s replaced by A u B . As for Raman amplification, he gain is
a factor of two higher when polarization is maintained and
(5) assumes this ideal situation.
Threshold power for backward stimulated scattering s
[
81
21A
g L
P =
Equations 5 ) and 6 ) are combined to produce the relation
for threshold power in Table I. It is assumed in Table I that
polarization
is
not maintained which reduces the gain by a
factor of two from the value in (5) [ 151. As an example we
again choose a
0.3
dB/km, 8.O-pm core diameter fiber, a wave-
length of 1.5 pm, and Au, = 1.0 GHz. Brillouin threshold
power is then 190 mW.
Use of the pproximation
A U ,> > A y e
conceals several
complications associated with s timulated Brillouin scattering.
For example, the Brillouin linewidth is not a constantbut
varies
[
181 as approximately
u ; ;
the value corresponding to
h
=
1
.O
pm is 38.4 MHz. Doping modifies AuB in two ways.
First is the direct effect on the intrinsic linewidth of the doped
glass which has not been well studied. Second is the indirec t
effect through the change in the sound velocity which from
(4) leads to a change in the Brillouin frequency shift between
core and cladding because of dopingdifferences [ 151 This
shows up as an inhomogeneousbandwidth which couldbe
as
large as 1 GHz. In weakly doped fibers, however, the effects
should be much smaller and need to be included only if the
pump linewidth is small enough to be comparable to he
Brillouin linewidth.
IV.
SELF-PHASE
MODULATION
There is an intensitydependent broadening of the signal
spectrum romhe nonlinear process known
as
self-phase
modulation. This greaterrequency widthhen increases
pulse spreading through group-velocitydispersion. Thepro-
cess can be understood as adifferential phase shift of the
optical carrierbetween thecenter and the tails of a pulse
due to he ntensitydependent refractive index.The ndex
change is extremely small but the corresponding phase modu-
lation can be significant when added up in a long fiber
[
191.
The effect of the combinat ion of self-phase modulation and
group velocity dispersion can be calculatedn straight-
forward way for simple smooth pulses in single-mode fibers.
7/17/2019 Nonlinearity in Fiber Transmission
http://slidepdf.com/reader/full/nonlinearity-in-fiber-transmission 4/5
STOLEN:NONLINEARITY IN FIBER TRANSMISSION
The problem of establishing a critical power in a transmission
system where the pulses are not simple is less straightforward
and we are forced t o make some educated guesses.
For simple Gaussian pulses the ntensity dependent broad-
ening 6w can be related to the spectral width by
6w
=
0.86
AwAGmw (7)
where A -,= is the maximum phase shift in radians and is pro-
portional to the nonlinea r index, peak power, effective length
and inverse area. Real pulses usually have a frequency width
considerablygreater than a imple transform of the pulse
envelope.Thiscould be due to any combinationof phase
or amplitude modulat ion. We will assume that the frequency
width is entirely due to amplitudemodulation so the pulse
containsmicrostructureon a time scale of l/Aw where Aw
is the pectralwidth.For he mplitudemodulated pulse
thebroadening will be proportional
to
the subpulses
so 7)
should still applywithAw again referring to he spectral
width. There is a questionabout what to use for he peak
power. Even if i t were known, using the height of the strongest
spikewould certainly overestimate the broadening. We will
use the height of the pulse envelope as given by the average
power divided by the duty factor.
To calculate the actual pulse spreading the procedure is to
split the fiber into short sections and alternately calculate the
effects of self-phase modulation and group velocity dispersion
[20].
A
simpler approximateapproach would be to assume
that all the broadening takes place in a length l / a and all the
pulse spreading occurs in the remainder of the fiber. Neither
of these approaches is amenable to a quick estimate.
A simpler estimate of a critical power [191 can be obtained
by assuming tha t the system has been designed for maximum
capacity
as
set by linear pulse dispersion.
A
factor
of
two
additional broadening is then a serious problem. This occurs
at = 2.0 which established ritical eak ower of
XA
L
P C = 1 . 2 x
lo-2-w
(8)
where, and A are uni ts of pm and pm2 L is in km, linear
polarization is preserved, and heconstantcontains actors
such as the nonlinear index. In contrast to Raman gain, the
broadening from self-phase modulat ion is only decreased 17
percentwheninearpolarization
is
not maintained in the
fiber [ 191.
The criticalpower for self-phase modulation as defined in
(8) is written n erms of average power in Table I. Here D
is the duty fac tor and polarization is scrambled. The critical
power for the
8
pm core, 0.3 dB/km fiber at 1.5 pm is then
32 mW.
We have not reated such details as chup whichcould in-
crease or decrease the broadening, or the difference between
the positive and negative side of the minimumdispersion
wavelength. On the negative side of the minimum disper-
sion wavelength there hould be intensi tydependent pulse
narrowing
[
21 1 The pulse should first narrow in time and
then broaden again. This will decrease thenet broadening
from group-velocity dispersion.This ffect might be used
to advantage. Forexample, ata wavelength
of
1.6pm, in
a iber for which the minimum dispersion
is
1.3 pm, he
powers required t o cause narrowing of 100-ps pulses are
on the order f a few milliwatts.
1235
V .
FOUR-PHOTON
MIX ING
Four-photon mixing also called three-wave mixing) is a
parametric interaction which can generate new frequencies
oneitherorboth sides of the pump frequency [ 1 - [ 31
.
Forexample,stimulated Raman
or
Brillouin light can mix
with pump light to produce light on the high frequency side
of thepump. These effects equire phase matching which
can be provided by the differing phase velocities of the wave-
guide modes in a multimode fiber [2 2] . In single-mode fibers
the process appears only at small frequency shifts (-1 cm-’)
where coherence lengths are
of
the orderof kilometers [23].
It appears at present tha t this process will not be a serious
limit n iber ransmissionsystems because of the need for
phase matching n single-mode fibers nd the high powers
requiredn multimode fibers. It should be kept in mind,
however, that experiments with high optical powers and multi-
mode ibersalmost always produce spectra which combine
several nonlinear processes and are made even morecompli-
cated by the presence of four-photon mixing.
VI.
CONCLUSION
The foregoing treatment shows that nonlinear effects can
appear n single-mode transmission ystems at powers of a
few milliwatts. The nonlinear processes most ikely to cause
problems are Raman gain inmultiplexedystems, pulse
spreading due o self-phase modulation, and stimulated Bril-
louin scatteringwith very narrow line sources.
Threshold estimates were based on work done in the visible
and it is clear that here is aneed for measurements n the
spectral range beyond 1pm-particularlywith ources that
approximatehepectral and temporal characteristics of
real systems.
We have concentratedon he deleteriousaspects of fiber
nonlinearities but here are also some significantadvantages.
Forexample, fiber Raman lasers are a new-type of laser-
pumped tunable lasers which utilize both the long interaction
lengths
of
low-lossfibers and he broad Raman bandwidths
of glasses [2] , [3 ]. These lasers sharemanyproperties with
dye and F-center lasers and may someday prove to be compact
and inexpensive sources of tunable radiation in the ear infrared .
Raman lasers in their simple single-pass form have been used to
generate a band
of
optical pulses which have found application
in
fiber loss and pulse dispersion measurements 1241
There may even be direct system applications of fiber non-
linearities. For example, the nonlinear pulse narrowing in the
negative dispersion regime could lead to multiplexed systems
where both the minimum dispersion wavelength at 1.3 pm and
the low-loss band around 1.6 pm were used at high capacity.
ACKNOWLEDGMENT
The author would like to thank Tingye Li for helpful com-
ments and suggestions.
REFERENCES
FiberTelecommunications,
S. E. Miller and
A.
G . Chynoweth,
R. H . Sto len , “ qonlinear properties o f optical fibers,” in
optical
E&. NewYork:Academic Press, 1979 .
Ostrowsky,Ed.NewYork:Plenum,1979.
“Fiber Raman lasers,” in
Fiber and Integrated Optics,
D.
B.
K 0
Hill, B.
S.
Kawasaki,
D .
C. Johnson, and Y. Fujii, “Non-
Research
and
Development, B.
Bendow and s.
S.
Mitra, Eds.
linear effects in optical ibers,”n
Fiber Optics-Advances in
New York: Plenum, 1979.
low-loss single-mode fibre at 1.55 bm ”
Electron. L ev .,
vol. 15.
T. Miya, Y. Terunuma,
T.
Hosaka, and
T.
Miyashita,“Ultimate
7/17/2019 Nonlinearity in Fiber Transmission
http://slidepdf.com/reader/full/nonlinearity-in-fiber-transmission 5/5
1236 PROCEEDINGS
OF
THEEEE, VOL. 68, NO. 10 OCTOBER 1980
[ 5 ] S .
Shimada, T. Matsumaoto, T. Ito,
S .
Matsushita,
K
Washio,
pp.
106-108, 1979.
.~
and K. Minemura, “100 km optical-fibre transmission system at
1.32
pm using high-power
C.W.
Nd-Y.A.G. laser,” E l e c t r o n . L e t t . ,
vol. 1 5 pp.
484-486, 1979.
[ 6 ]
L.
G. Cohen, C. Lin, and W. G. French, “Tailoring zero chromatic
dispersion into the
1.5-1.6
pm low-loss spectral region ofsingle-
mode fibers,” E l e c t r o n . L e t t . , vol.
15,
pp.
334-335, 1979.
[ 7 ]
R. Olshansky and G. W. Scherer, “High
GeO
optical waveguides,”
in Rec. O p t i c a l C o m m u n i c a t i o n C o n 5 (Amsterdam, The Nether-
lands), paper
12.5,
Sept.
17, 1979.
[ 8 ]
R. G. Smith,“Opticalpowerhandlingcapacityof ow-lossop-
tical ibers
as
determinedby timulatedRaman ndBrillouin
scattering,”
A p p l . O p t . ,
vol.
1 1
pp.
2489-2494, 1972.
[ 9 ] R.
H. Stolen and E. P. Ippen , “Ram an gain in
g l a s s
optical wave-
[ l o ]
G. E. WalrafenandJ.Stone,“Ramancharacterizationofpure
guides,”
A p p l . Phys. L e t t . , vol. 22,
pp.
276-278, 1973.
and doped fused silica optical fibers,” A p p l . S p e c t r o s c . , vol.
29,
p.
337, 1975.
[
1 1 1
F. L.Galeener,J. C. Mikkelsen, R. H. Geils,and W. J . Mosby,
“TheelativeRaman ross-sections of vitreous SO ,,GeO, ,
[
121 S.
A.Brawerand W.B. White, “Raman spectroscopic nvestiga-
B,O, and P,O,
,” A p p l . Phys. L e t t . ,
vol.
32,
pp.
34-36, 1978.
tionof he tructureof ilicate
g l a sse s
I. Thebinary lkali
silicates,”
J. Ch e m. Phys.,
vol.
63,
pp.
2421-2432, 1975.
[ 131 E. P. Ippen, “Nonlinear effects in optical fibers,” in laser appli-
cations o optics and spectroscopy,
S.
F. Jacobs, M.
0
Scully,
and
M.
Sargent, Eds. Reading, MA: Addison-Wesley, 1975.
[
141 F. Capassoand P. Dip ort o,“Coupled-mode heoryofRaman
amplification n lossless optic al fibers,”
J.
A p p l . Phys., vol.
47 ,
Pp.
1472-1476, 1976.
R. H. Stolen, “Polarization effects n fiber Raman and Brillouin
lasers,” ZEEE J. Quantum Elect ron. , vol. QE-15, pp. 1157-1 160,
1979.
R. H. Stolen, V. Ramaswamy, P.
Kaiser,
and W. Pleibel, “Linear
polarizationn irefringentingle-modeibers,” A p p l .
Phys.
L e t t . , vol.
33,
pp.
699-701, 1978.
E. P. Ippen and R. H. Stolen,“StimulatedBrillouin cattering
in optical fibers,”AppI. Phys.
L e n . , v o l .
21,
pp.
539-541, 1972.
D. Heinman, D.
S.
Hamilton,and R. W. Hellwarth,“Brillouin
scatteringmeas ureme nts n ptica l glasses,”
Phys. Rev. ,
vol.
R. H . Stolen and C. Lin, “Self-phase modulation in silica optical
fibers,”Phys. Rev.,vol.
A17,
pp.
1448-1453, 1978.
playbetween elf-phasemodulationanddispersion or ntense
R. A. Fisher and W. K. Bischel, “Numerical studies of the inter-
plane-wave laser pulses,”
J . A p p l . Phys.,
vol.
46 ,
pp.
4921-4934,
1975.
A. Hasegawa and F. Tappert, “Transmission of stat ionar y non-
linear optical pulses n dispersive dielectric fibers. I. Anomal ous
dispersion,” A p p l .
Phys.
L e t t . , vol.
23,
pp.
142 -144, 1973.
three-wave mixing in silica fiber optical waveguides,”
A p p l . Phys.
R. H. Stolen, J. E. Bjorkholm, and A. Ashkin,“Phase-matched
L e t t . , vol.
24,
pp.
308-310, 1974.
mixing
in
single-modeoptical ibers,” J . A p p l . Phys.,vol.
49 ,
K 0
Hill, D. C. Johnso n, and B.
S.
Kawasaki, “CW three-wave
B 1 9
pp.
6583-6592, 1979.
PP.
5098-5106, 1978.
L.
G.
Cohen a nd C. Lin, “A universal fiber-optic (UFO) measure-
mentsystembasedonanear-IR iberRaman aser,” ZEEE J.
Q u a n t u m E l e c t r o n . ,vol.
QE-14,
pp.
855-859, 1978.
Fiber Transmission osses in High Radiation
GEORGE H. SIGEL, JR .
Abstmct-A
summary is provided on the effects of ionizing radiation
[ 1 ]
[
21. The introduction of index modifying dopants into
on Present Fe n t i o n ptd fiben In
P d C “ h
the P W r focuses
SiOz glass may stronglymodify bondingstrengthsandcon-
on the magnituh and the spectr l and temporpl
characteristics Of
the
figurations and can thereforestrongly mpacton hedefect
radiation-inducedtransmission lossesn rradiated fiben The
most
=diption resistant
d-
of fi rsre identified and possible p p p r o ~ e s production and charge trapping processes [ 3
1
- 5 1 . In addi-
forurthermprovementsn
fiber
performance are suggested. tion, preexisting flaws andtrains are commonn glasses,
so
that they are ypically much more susceptible
to
damage
I.
I NTRODUCTI ON
from ionizingadiation thanheir crystalline counterparts
of
the same composition . Defects in the glasses are normally
bands in
the
near
uv,
visible and
infrared
spectral
result in the formation of lower energy light absorption and
regions. These
so
called “color centers” are particularly serious
in the case of optical fibers since the long optical pathlength
can make the optical link vulnerable to rather low radiation
doses.
The optical damage which is created within the g l a s s matrix
by ionizing radiation is usually not permanent so that recovery
Manuscript received May 22, 1980.
Theauthor
is
wit h he Naval ResearchLaboratory,Washington, DC and annealing
Of
absorption are
Observed
in
most
fibers
at
20375. roomemperature.he recovery process is strongly influenced
N UNDERSTANDING of he tYPiC‘ radiationeffects associated with energy levels within he forbidden gap
observed inoptica l fibers requires some knowledge of
the structure and bonding of the opt ical materials used
in waveguide fabrication.Since optical fibers for hemost
part are manufacturedrom high-purity glasses based on
SiOz, the discussion of radiation damage will focus on these
materials. onizing radiat ionproducesbrokenand dangling
bonds which serve as charge trapping sites
in
the
glass
matrix
U.S.
Government work not protected by
U.S.
copyright