THE STORY OF THE OPTICAL FIBER*

Ajoy Ghatak and K Thyagarajan

24
Professor K. Thyagarajan is currently a Professor at the Physics Department, IIT Delhi. He
has held Visiting positions at Laboratoire Centrale de Recherches, Thomson-CSF France,
Department of Electrical Engineering, University of Florida, USA, University of Waterloo,
Canada and the City University of Hong Kong. He has published more than 140 research
papers in international journals, has filed five patent applications and is the co-author (with
Professor A. K. Ghatak) of seven books, the latest being , John Wiley,
USA, 2007. In 1998 he was the co-recipient (with Prof B P Pal) of the "
award by Lucent Technologies- Finolex and Voice and Data, India. In 2003 he was
decorated with the title of by the French
Government. He is a and
. His current research interests are in the fields of guided wave quantum optics,
optical fiber amplifiers andnonlinear optical effects inphotonic bandgapstructures.
Fiber Optic Essentials
Fiber Optic Person of the
Year 1997
Officier dans l'ordre des Palmes Acadmiques
Fellowof the Optical Society of America of the Indian National Academy
of Engineers
Professor Ghatak obtained his Ph.D. from Cornell University and has recently retired from
IIT Delhi. He is a recipient of several awards including the CSIR S S Bhatnagar award for
, the 2008 SPIE (USA) Educator award in
recognition of
and the 2003 Optical Society of America Esther Hoffman Beller award in recognition of his
. He received D.Sc.
(Honoris Causa) from University of Burdwan in 2007. He has authored several books
including
the last twocoauthoredwithProfessor Thyagarajan.
outstanding contributions in Physical Sciences
his unparalleled global contributions to the field of fiber optics research, and his
tireless dedication to optics education worldwide and throughout the developing world in particular
outstanding contributions to optical science and engineering education
OPTICS, Einstein & Special Theory of Relativity, Introduction to Fiber Optics, Optical
Electronics
Professor Ajoy Ghatak
Professor K. Thyagarajan
Professor Charles Kao has been awarded half of the 2009 Nobel Prize in Physics for groundbreaking achievements concerning the transmission of
light infibers for optical communication. This is truly a very apt recognitionof anarea which has touched all lives fromthe rich to the poor. In1966,
Charles Kao and George Hockham predicted that if it was possible to produce optical fibers with attenuation less than 20 dB/km, it could compete
effectively with the conventional communication systems; a loss of 20 dBimplies a power loss by a factor of 100. In 1970, Kapron, Keck and Maurer
(at Corning Glass Works in USA) were successful in producing silica fibers with a loss of about 17 dB/km. In 1970 itself, Alferov in Leningrad and
Panish and Hayashi at Bell Labs demonstrated roomtemperature operation of semiconductor lasers; and thus started the revolution in optical fiber
communication. In 1987, David Payne and his collaborators (at University of Southampton) and Emmanuel Desurvire and his collaborators (at
AT&TBell Laboratories) developed Erbiumdoped fiber amplifiers operating at 1550 nm. This was yet another revolutioninthe development of fiber
optic communication systems. Much before this, in 1961 Elias Snitzer took a glass fiber, whose core contained Nd -ions, wrapped the fiber
around a flash lamp and, whensuitable optical feedback was applied, a laser was produced. Thus only one year after the demonstration of the first
ever laser, the first fiber laser was born. The fiber laser nowfinds widespread applications inmany diverse areas including instrategic defense.
Apart from the phenomenal development of optical fiber telecommunications, optical fibers are finding applications as sensors in difficult
environments such as close to heavy electrical equipments, and in distributed measurements tracking strain and temperature variations in civil
structures such as bridges, dams etc. Froma pedagogical point of view, fiber optics provides a mediumfor realizing many beautiful experiments for a
better understanding of basic concepts in optics and other fields such as quantummechanics. One of the most beautiful experiments is probably the
awesome broadband super continuum generation in which a high power laser beam coupled into an optical fiber generates a very broadband
spectrumspanning an extremely wide range of wavelengths including the visible spectrum; this is produced by the non-linear optical effects within
the optical fiber due to the high intensity of the propagating light wave. There are also devices such as the Fiber Bragg Grating (usually abbreviated as
FBG) and many other fiber-optic devices, which find extremely important applications not only inoptical communicationsystems but also insensor
physics. FBG's have a periodic variation of refractive index which is created through the phenomenon of photosensitivity, discovered in Canada in
1978 by Kenneth Hill. The optical fiber is also very extensively used invarious imaging devices.
This article will give a brief history of the evolutionof the field of fiber optics and will also discuss a fewfascinatingexperiments using the optical fiber.
3+
*Based on 6th Atma RamMemorial Lecture (on 25 Oct, 2009 at CGCRI,Kolkata) and 12th D. M.Bose Memorial Lecture (on 4, Jan, 2010 at ISNA,
Kolkata) byProf. Ghatak.
25
optical fibers led to an explosion in
t he appl i cat i on of l aser s i n
communication. The backbone
network that is providing this
capability is based on optical fibers
crisscrossing the entire earth under
the seas, over land, and across
mountains. Today more than 10
terabits of information can be
transmitted per second through one
hair thin optical fiber; this amount of
i nf ormat i on i s equi val ent t o
simultaneous transmission of about
150 million telephone calls this is
certainly one of the very important
technological achievements of the
20 century. In addition, devices
like Fiber Bragg Gratings, fiber
amplifiers, and fiber lasers are now
f i ndi ng ext remel y i mport ant
th
1. Introduction
Optics is today responsible for many
r e vol ut i ons i n s c i e nc e a nd
technology. This has been primarily
brought about by the invention of the
laser in 1960 and subsequent
development in realizing extremely
wide varietyof lasers. One of the most
interesting applications of lasers with
its direct impact on our lives has
been in communication. Use of
e l e c t r o m a g n e t i c wa v e s i n
communication is quite old and the
development of the laser gave
communication engineers with a
source of electromagnetic waves with
extremely high frequency compared
tomicrowaves and millimeter waves.
The development of low loss
1
A nice historical account on the development of the optical fiber has been given in Ref. He1 some of the dates given above are as given in
References He1 and Ga1.
Important milestones
1841
1842
1854
1880
1926
1930
1954
1960
1961
1966
1970
1971
1975
1975
1976
1978
1987
1988
1996
2009
Daniel Colladondemonstrates (inGeneva) light guidinginwater jets.
Jaques Babinet demonstrates (inParis) light guidinginwater jets andalsoinbent glass rods.
JohnTyndall demonstrates light guidinginwater jets, duplicatingbut not acknowledgingBabinet.
Alexander GrahamBell invents Photophone inWashington.
C.W. Hansell outlines the principles of fiber-optic imagingbundle.
HeinrichLamm, a medical student inMunich, first assembleda bundle of transparent
vanHeel inNetherlands andHopkins andKapanyinUKsuggest a claddingwill improve transmissioncharacteristics.
Maimanfabricates the first laser.
Snitzer publishes the theory of single mode fibers and also fabricates the first fiber laser (barium crown glass doped
withNd ions).
Kao and Hockham predict that if it was possible to produce optical fibers with attenuation less than 20 dB/km, it could
compete effectivelywiththe conventional communicationsystems.
Kapron, KeckandMaurer (at CorningGlass inUSA) were successful inproducingsilicafibers witha loss of about 17 dB/km.
Alferov in Leningrad and Panish & Hayashi at Bell Labs demonstrated room temperature operation of Semiconductor
Lasers.
Continuous-wave semiconductor laser operatingat roomtemperature commerciallyavailable.
Payne and Gamblingshowverysmall pulse dispersionat 1.27 m.
Bell Labs test parabolic indexfiber-optic communicationsystemtransmitting45 Mbits/s.
NTT(Japan) transmits 32 Mbits/s through 53 kmof gradedindexfibers at 1.3 m.
Payne, Mears & Reekie (at University of Southampton) and Desurvire, Becker, and Simpson (at AT&T Bell Laboratories).
developEDFAs (ErbiumDopedFiberAmplifiers) operatingat 1.55 m.
First transatlantic fiber cable, usingsingle mode fibers was made operative at 1.3 m.
Fujitsu, NTT Labs and Bell Labs independently report sending over 1 Tbit/s through one single mode fiber using
WDM techniques.
Half of the 2009 Nobel Prize in Physics awarded to Charles Kao for
.
fibers totransmit animage.
3+
m
m
m
m
groundbreaking achievements concerning the transmission
of light infibers for optical communication
applications in many diverse areas
from defense to sensor physics.
At t he he ar t of an opt i c al
communication system is the optical
fiber that acts as the transmission
channel carrying the light beam
loaded with information. The light
beam gets guided through the
optical fiber due to the phenomenon
of total internal reflection (often
abbreviated as TIR). Figure 1 shows
an optical fiber, which consists of a
(cylindrical) central dielectric core (of
refractive index cladded by a
material of slightly lower refractive
index (< ). The cladding is usually
ultra-pure silica and the core is silica
doped with germanium; doping with
germanium results in an increase in
n
n n
1
2 1
2. The Optical Fiber
26
Although the phenomenon of total
internal reflection has been known
for hundreds of years, the first
experimental demonstration of
guidance of a light beam using total
internal reflection was carried out by
Daniel Colladon in 1841 and also by
Jacques Babinet in 1842. This was
done by sending a light beam in a
water jet as shown in Fig. 3; light
undergoes total internal reflection at
the water-air interface and travels
along the curved path of water
emanating from an illuminated
vessel.
the refractive index. Now, for a light ray entering the fiber, if the angle of
incidence (at the core-cladding interface) is greater than the corresponding
critical angle, the ray will undergo TIRat that interface. Further, because of the
cylindrical symmetry in the fiber structure, this ray will suffer TIRat the lower
interface also and therefore get guided through the core by repeated total
internal reflections. Even for a bent fiber, light guidance can occur through
multiple total internal reflections.
Figure 1 : (top) A glass fiber consists of a cylindrical central core cladded by a material of
slightly lower refractive index. (bottom) Light rays incident on the core-cladding interface
at an angle greater than the critical angle are trapped inside the core of the fiber. The
diameter of thecladdingis almost always 125 m. For multimodefibers, thecorediameters
are usually inthe range of 25-50 m. For single-mode fibers, the core diameters are usually
between5 and 10 m.
m
m
m
Figure 2 shows a photographof anoptical fiber carrying a light beam. The fiber
is visible due to the phenomenon of Rayleigh scattering, which scatters a tiny
part of the light propagating through the fiber. Rayleigh scattering is the same
phenomenonthat is responsible for the blue color of the skyandthe redcolor of
the risingor the settingsun.
Figure 2 : A step index multimode fiber illuminated by HeNe laser with bright output light
spot. The light coming out of the side of the optical fiber is primarily due to Rayleigh
scattering. [The fiber was produced at the fiber drawing facility at CGCRI, Kolkata; figure
courtesyDr. Shyamal BhadraandMs. Atasi Pal].
Figure 3 : Light guidance through a water
jet demonstrating the phenomenon of
total internal reflection; this was first
demonstratedbyDaniel Colladonin1841.
Figure 4 : A cold light source using optical
fiber
If a large number of optical fibers are
assembled together, it forms what is
known as a bundle. If the fibers are
not aligned, i.e. they are all jumbled
up, the bundle is said to form an
incoherent bundle. Such incoherent
bundles can be used to transmit light
from one point to another along a
flexible path into regions where
normal access is not available.
27
Incoherent bundles are also used in
illuminationsuchas intraffic lights or
road signs or even for lighting in real
estate applications wherein the light
source is removed from relatively
inaccessible areas and fibers are used
to guide light from a lamp. They can
also be used as cold light sources (i.e.,
light sources giving only light and no
heat) by cutting off the heat radiation
using a filter at the input to the fiber
bundle (see Fig. 4). The light
emerging from the bundle is also free
from UV radiation and is suitable for
illumination of paintings etc. in
museums. Another very interesting
application is using such bundles to
bring sunlight into working areas
which can give light without heat and
other harmful radiations. There are
many companies which offer such a
source. Decorative applications of
fiber optic illuminators are also very
interesting. Fiber optic illuminators are attractive since they provide safety
from electrical hazards and also project less load on air conditioning due to
removal of heat component fromlight.
If the optical fibers are alignedproperly, i.e., if the relative positions of the fibers
in the input andoutput ends are the same, the fibers are saidto forma coherent
bundle. Now, if a particular fiber is illuminated at one of its ends, then there
will be a bright spot at the other endof the same fiber at the same position; thus
a coherent bundle can transmit an image fromone endto another. Perhaps the
most important application of a coherent bundle is in a fiber optic endoscope
where it can be put inside a human body and the interior of the body can be
viewed fromoutside; for illuminating the portion that is to be seen, the bundle
is enclosed in a sheath of fibers which carry light fromoutside to the interior of
the body (see Fig. 5). A state-of-the-art fiberscope can have more than 10000
fibers, which would form a bundle of about 1 mm in diameter capable of
resolvingobjects 70 macross.
The idea of using light waves for communication can be traced to as far back as
1880 when Alexander Graham Bell invented the photophone shortly after he
invented the telephone in 1876. In this remarkable experiment, speech was
transmitted by modulating a light beam, which traveled through air to the
receiver. The transmitter consisted of a flexible reflecting diaphragm which
could be activated by sound and which was illuminated by sunlight. The
reflected beam was received by a parabolic reflector placed at a distance (see
). The parabolic reflector concentrated the light on a photo conducting
selenium cell, which forms a part of a circuit with a battery and a receiving
earphone. Sound waves present in the vicinity of the diaphragm vibrate the
diaphragm, which leads to a consequent variation of the light reflected by the
m
Fig. 6
3. Some Historical Remarks
Figure 5:(a) An optical fiber medical probe
called an endoscope enables doctors to
examine inner parts of human body (b) A
stomach ulcer as seen through an
endoscope.
(a)
(b)
Figure 6 : The diagram of the Photophone; this has been taken from Alexander Graham
Bell's 1880 paper "On the Production and Reproduction of Sound by Light",
, Third Series, vol. XX, #118, pp. 305 324, October 1880. In this system,
sunlight was modulated by a diaphragmand transmitted through a distance of about 200
meters in air to a receiver containing a selenium cell connected to the earphone.
American
Journal of Sciences
2
Actually, according to reports (published in June 2002), an Italian immigrant Antonio Meucci, was the inventor of telephone; indeed Meucci
demonstrated his teletrfono in NewYork in 1860. Alexander GrahamBell took out his patent 16 years later. This has apparently been recognized
byUSCongress.
28
diaphragm. The variation of the light
falling on the selenium cell changes
the electrical conductivity of the cell,
which in turn changes the current
in the electrical circuit. This
changing current reproduces the
sound on the earphone. This was
the first experiment on optical
communication. To quote from Ref.
Ma2:
"...
...".
After this beautiful experiment by
Alexander Graham Bell, not much
work was carried out in the field of
optical communications. This is
because of the fact that there was no
suitable light source available that
could reliably be used as the
information carrier. The advent of
lasers in 1960 (see Fig. 7)
immediately triggered a great deal of
investigations aimed at examining
the possibility of building optical
a na l o g ue s o f c o nve nt i o na l
communication systems. The very
f i r s t s uc h mo de r n o p t i c a l
c ommuni c at i on exper i ment s
involved laser beam transmission
through the atmosphere. However, it
was soon realized that laser beams
could not be sent in open atmosphere
through reasonably long distances to
carry signals unlike, for example,
mi crowave or radi o syst ems
operating at longer wavelengths.
This is due to the fact that a light beam
(of wavelengthabout 1 m) is severely
attenuated and distorted owing to
scattering and absorption by the
atmosphere. Thus for reliable long
distance lightwave communication
under terrestrial environments it
In 1880 he (Graham Bell)
produced his "photophone" which to the
end of his life, he insisted was the
greatest invention I have ever made,
greater than the telephone Unlike the
telephone it hadno commercial value.
m
would be necessary to provide a transmission medium that can protect the
signal carryinglight beamfromthe vagaries of the terrestrial atmosphere.
Figure 7: (a) Theodore HaroldMaimanof Hughes ResearchLabinCalifornia fabricatedthe
first laser onMay17, 1960 (b) Thefirst laser was aRubylaser
In1966, KaoandHockham[Ref. Ka1] made anextremelyimportant suggestion;
they said that optical fibers based on silica glass could provide the necessary
transmission medium if metallic and other impurities could be removed from
silica. Toquote fromthe 1966paper of KaoandHockham[Ref. Ka1]:
We maymentionhere that losses inoptical fiber are specifiedinterms of the unit
dB/km. It is definedas
Here and are the input and output powers corresponding to an optical
fiber of length (km). In 1966, when the paper by Kao and Hockham was
published, the most transparent glass available had extremely high losses (of
more than about 1000 dB/km implying a power loss by a factor of 100 in
traversing through only 20 meters of the fiber); the high loss was primarily due
to trace amount of impurities present in the glass. Obviously this loss is too
highevenfor short distances suchas afewhundredmeters.
The 1966 paper of Kao and Hockham triggered the beginning of serious
research in removing traces of impurities present in the glass, which resultedin
realization of low loss optical fibers. In 1970, Kapron, Keck and Maurer (at
Corning Glass in USA) were successful in producing silica fibers with a loss of
about 17 dB/km at the Helium Neon laser wavelength of 633 nm (Ref. Ka2).
Further, in 1970 itself, Alferov in Leningrad and Panish & Hayashi at Bell Labs
demonstrated room temperature operation of Semiconductor Lasers; and
thus started the revolution in optical communication. By 1985, with
improvement in technology of purifying glass, silica fibers were routinely
produced with extremely low losses: less than 0.25 dB/km which corresponds
to a transmissionof more than94 %of the incident power after traversing1 km
of the optical fiber. Because of such low losses, the distance between two
consecutive repeaters (used for amplifying and reshaping the attenuated
signals) could be as large as 250km.
Theoretical and experimental studies indicate that a cladded glass fiber with a core
diameter of about and an overall diameter of about 1000 represents a possible
practical optical waveguide with important potential as a new form of communication
medium. The refractive index of the core needs to be about 1% higher than that of
cladding. However, the attenuation should be around 20 dB/km which is much higher
thanthe lower limit of loss figure imposedbyfundamental mechanisms.
P P
L
l l
0 0
in out
10
10
(dB/km) log
(km)
P
in
L P
out
a

=


......(1)
Figure 8 : Typical wavelength dependence
of loss for a silica fiber. The peaks in the
attenuation curve in the wavelength
regions 1.25 m and 1.40 m are due to the
presence of minute amount of water and
other impurities. Notice that the lowest loss
occurs at 1550nm[adaptedfromRef. Mi1].
m m
Figure 8 shows a typical dependence
of fiber as a
function of wavelength of a typical
silica optical fiber. As can be seen,
there are two windows at which loss
attains its minimum value in silica
fibers. The first window is around
1300 nm (with a typical loss
coefficient of less than 1 dB/km)
where fortunately (as we will see
later) the material dispersion is
negligible. However, the loss attains
its absolute minimum value of about
0.2 dB/kmaround 1550 nm. The latter
window has become extremely
important in view of the availability
of erbiumdoped fiber amplifiers. The
losses in optical fibers are caused due
to various mechanisms such as
Rayleigh scattering, absorption due
tometallic impurities andwater inthe
fiber, and due to intrinsic absorption
by the silica molecule itself. The
Rayleighscatteringloss varies as ,
i.e., shorter wavelengths scatter more
than longer wavelengths; here
represents the free space wavelength.
This is the reason why the loss
coefficient decreases up to about
1550 nm. The two absorption peaks
around 1240 nm and 1380 nm are
primarily due to traces of OH ions
and t races of met al l i c i ons.
attenuation coefficient a
l
0
-
For example, even1 ppm(part per million) of ironcancause a loss of about 0.68
dB/km at 1100 nm. Similarly a concentration of 1 ppm of OH ion can cause a
loss of 4 dB/km at 1380 nm. This shows the level of purity that is required to
achieve low-loss optical fibers. If these impurities are completely removed, the
two absorption peaks will disappear (see Fig. 9) andwe will have very lowloss
inthe entire range of wavelengthstarting from1250 nmto 1650 nm. Ina typical
commercially available fiber the loss is about 0.29 dB/km, 0.19 dB/km and 0.21
dB/kmat = 1310nm, 1550nmand 1625nmrespectively. Such fibers open up
a bandwidth of more than 50 THz for communication. Beyond = 1600nm,
the increase in the loss coefficient is due to the absorption of infrared light by
silica molecules. This is an intrinsic property of silica and no amount of
purificationcanremove this infraredabsorptiontail.
-
l
l
0
0
29
4
0
1
l
Figure9: Usingsophisticatedtechniques, it is possibletoremovethetraceamount of water
and other impurities. The loss is less than 0.4 dB/kmin the entire wavelength range from
1250nmto1600nm. Thediagramcorresponds tothefiber fabricatedbySterliteIndustries
at Aurangabad, Indiaand is courtesyMr S. Bhatiaof SterliteIndustries.
When a light pulse propagates through an optical fiber it suffers from
attenuation due to various mechanisms and also the pulse broadens in time
leading to what is termed as pulse dispersion. Apart from this, due to the high
intensity of light present in the fiber, nonlinear optical effects come into play.
Attenuation, pulse dispersionandnonlinear effects represent three of the most
important characteristics that determine the information transmission
capacity of optical fibers. Obviously, lower the attenuation (and similarly,
lower the dispersion and smaller the nonlinear effects) the greater will be the
required repeater spacing and therefore higher the information carrying
capacity and lower will be the cost of the communication system. In order to
understand the dispersion phenomenon and how to manage it, it is necessary
to understand the modes of the waveguide which we plan to do in the next
section.
Figure 1 hows an optical fiber, which consists of a (cylindrical) central
dielectric core (of refractive index ) cladded by a material of slightly lower
refractive index
s
n
1
n n
2 1
(< ). The corresponding refractive index distribution (in the
transverse direction) is givenby:
4. Modes Of AStepIndexFiber
1
2
( ) 0 core
cladding
n r n r a
n r a
= < <
= >
......(2)
propagation constant can assume
only discrete values. Since
represents the propagation constant
along the -direction, this implies that
the propagating plane waves by
which we can represent the modes
propagate making some discrete
angles only. Thus although ray
analysis does not restrict the ray
angles to be discrete, modal analysis
does predict discrete angles of
propagation of light through the
optical fiber.
b
b
z
where and (< ) represent
respectively the refractive indices of
core and cladding and represents
the radius of the core. We define a
parameter through the following
equation
When << 1 (as is indeed true for
silicafibers), we maywrite
The necessity of a cladded fiber
rather than a bare fiber i.e., without a
cladding, was felt because of the fact
that for transmissionof light fromone
place to another, the fiber must be
supported, andsupportingstructures
may considerably distort the fiber
thereby affecting the guidance of the
light wave. This can be avoided by
choosinga sufficientlythickcladding.
Further, in a fiber bundle, in the
absence of the cladding, light can leak
through from one fiber to another
leading to possible cross talk among
the information carried between two
different fibers. The idea of adding a
second layer of glass (namely, the
c l a d d i n g ) c a m e i n 1 9 5 5
independently from Hopkins and
Kapany in the UK and van Heel in
Holland. However, during that time
the use of optical fibers was mainly in
image transmission rather than in
communications. Indeed, the early
pioneering works in fiber optics (in
the 50's) byHopkins, KapanyandVan
Heel led to the use of the fiber in
optical devices (see, e.g. Ref. Ka3).
In most practical fibers the refractive
index depends only on the cylindrical
coordinate and therefore it is
convenient to use the cylindrical
system of coordinates and it is
possible to write the solutions of
Maxwell's equations in the form
n n n
a
r
(r, ,z)
1 2 1
D
D
f
30
where is the angular frequency,
( , ) represents the transverse field
profile and is known as the
pr o pa g a t i o n c o ns t a nt . The
propagation of the mode is assumed
to be along the -direction. The
quantity is similar to for a plane
wave given by . Equation (5)
defines the modes of the system. Since
( , ) depends only on the transverse
coordinates r and ,
A waveguide (like an optical fiber) is
characterized by a finite number of
modes which are guided by the
waveguide each mode is described
by a definite transverse field
distribution ( , ) corresponding to a
definite (discrete) value of . The
precise forms of ( , ) (and the
correspondi ng val ues of t he
propagation constant ) are obtained
bysolvingMaxwell's equations see,
e.g. References Gh1 andSn1.
Now, while discussing step index
fibers, we had considered light
propagation inside the fiber as a set of
many rays bouncing back andforth at
the core-cladding interface [see Fig. 1]
and the angle could take possible
values from 0 (corresponding to a ray
propagating parallel to the -axis) to
cos ( ) (corresponding to a ray
incident at the critical angle on the
core-cladding interface). However
usi ng rays t o descri be l i ght
propagation is an approximation
and the correct description is in
terms of waves obtained as solutions
to the Maxwell's equations. As
mentioned earlier, the solutions
give us the modes of propagation
t hr ough t he f i be r a nd t he
w
y f
b
b
y f
f
y f
b
y f
b
r
z
k
e
r
the modes represent
transverse field configurations that do
not change as they propagate through the
optical fiber except for a phase change.
r
r
all
z
n / n
i t kz ( ) w -
-1
2 1
2
1
2
2
2
1
2n
n n -
D ......(3)
1
2 1
2
2 1
n
n n
n
n n -

-
D ......(4)
( ) ( )
) (
, , , ,
z t i
e r t z r
b w
f y f
-
= Y ......(5)
Figure 10 : The optical fiber is fixed in a
groove in a quartz block and polished from
the side so as to come close to the core-
cladding boundary. When light is coupled
into the fiber core, then the light beam
emerges from the prism at specific angles.
Each emerging beam corresponds to a
specific mode. The experiment beautifully
demonstrates the discrete nature of the
modes and is a very nice example to
understand discretization of energy states
inquantummechanics.
A very beaut i f ul experi ment
demonstrating the discrete nature of
the angles of propagation is shown in
Fig. 10. In this experiment, an optical
fiber is fixedwithina groove ina silica
block and polished from the side so
that the light propagatingthroughthe
core can be accessed. For this the
polished surface is at a distance of
about 1 m from the core-cladding
boundary. Onthe polishedsurface we
place a glass prism whose refractive
index is larger than that of the core of
the fiber. When light is coupled into
the fiber then due to frustrated total
internal reflection, light propagating
through the fiber core couples out of
the prismwhich can be projected on a
m
For lying in the above range, the
fields (associated with the mode) are
oscillatoryinthe core anddecayinthe
claddingand assumes onlydiscrete
values; these are known as the
(discrete) of the system.
......(12)
For such values, the fields are
oscillatory even in the cladding and
can assume a continuum of values.
These are known as the
In most fiber optic communication
systems, one uses single mode
fibers; as such, from now on we will
consider only single mode fibers.
For a single mode step index fiber, a
convenient empirical formula for
the propagation constant of the
fundamental mode is givenby
......(13)
where 1.1428 and 0.996. The
t ransverse f i el d di st ri but i on
associated with the fundamental
mode of a single mode fiber is an
extremely important quantity and it
determines various important
parameters like splice loss at joints
between fibers, launching efficiencies
from sources, bending loss etc. For a
step index fiber one has analytical
expression for the fundamental field
distribution in terms of Bessel
functions (see, e.g., Ref. Gh1, My1,
Sn1). For most single mode fibers
with a general transverse refractive
index profile, the fundamental mode
field distributions can be well
approxi mat ed by a Gaussi an
function, which may be written in the
form
b
b
b
2
2
guided modes
radiation
modes.
A B
(b) Radiationmodes:
5. Single Mode Fiber

screen. This phenomenon is similar to
tunneling in quantum mechanics and
is a purely wave phenomenon. We
will find that at the output the light
emerges only along discrete angles
corresponding to discrete values of
Eachemergingbeamcorresponds toa
different mode of the fiber. So if the
fiber shown in the figure supports
three guided modes, then the output
would have three beams emerging at
t hree di st i nct angl es. I f t he
wavelength of the incident light beam
is changed, then the output angles
would also change due to the change
in the propagation constant of the
modes. Indeed, in the experiment
shown in Fig. 10, if we measure the
angles of the emerging beams, we can
obtain the corresponding angles
within the fiber core using Snell's law
and we can then obtain the discrete
values of the propagation constant
byusingthe followingformula:
For a step index fiber defined by
Eq.(2), we define a dimensionless
parameter defined by the following
equation
is the free space wavelength of the
light beamand is defined by Eq. (4).
The parameter (whichalso depends
on the operating wavelength is
k n o wn a s t h e wa ve g ui de
parameter and is an extremely
important quantity characterizing an
optical fiber. For a stepindexfiber [see
Eq. (2)] if
V<2.4045 ......(8)
there is only one allowed discrete
value of andthe fiber is referredtoas
a single mode fiber. For a given fiber,
the wavelength for which =2.4045 is
known as the and is
denoted by and the fiber will be
single modedfor < .
q.
b
q
b
D
b
l
l l
V
V
cut-off wavelength
C
0 C
......(7)
)
l
l
0
0
31
0 1
cos k n b q = ......(6)
2 2
1 2 1
0 0
2 2
2 V a n n a n
p p
l l
= - = D
For an arbitrary cylindrically
symmetric profile having a refractive
index that decreases monotonically
from a value on the axis to a
constant value beyond the core-
claddinginterface (see Fig. 11) the
solutions of Maxwell's equations can
be dividedintotwodistinct classes:
......(11)
n
n
r = a
1
2
(a) Guidedmodes:
As an example, we consider a step
index fiber with =1.447, =0.003 and
= 4.2 m giving =2.958/ , where
is measured in m. Thus the fiber
would be single moded for < 1.23
m. In this example, the cutoff
wavelengthis =1.23 m.
For V 10 the number of modes (for a
step-index fiber) is approximately
/2 and the fiber is said to be
multimodedfiber.
The phase velocityandgroupvelocity
of amode are givenby
where
is referred to as the effective index of
the mode and = /c represents the
free space propagationconstant.
n
a V
V
k
2
0
0
0
c
0
D
m l
l m
l
m
l m
w
2

0
eff
n
k
b

......(10)
Figure 11 : A cylindrically symmetric
refractive index profile having a refractive
index that decreases monotonically from a
value on the axis to a constant value
beyondthecore-claddinginterface
n n
r =a.
1 2
2
2 2
2 1 2
0
n n
k
b
< <
2
2
0
2
2
2
2 2
1 2
( ) ;
k
n
B
b V A
n n V
b
-

= -

-
1.5 < 2.5 < V
( )
2
2
,
r
w
r A e ......(14) y f
-
=
n r ( )
n
1
Core
Cladding
n
2
a
( )
p
eff
c
v
n
w
b w
= = and
1
g
d
v
d
b
w
-

=


......(9)
2
2
2
0
2
n
k
<
b
0 <
l m
Dl
l
l
l
l
l
l
0
0
0
0
0
0
0
0
is measured in mand the quantity
inside the square brackets is
dimensionless. Thus represents
t he ma t e r i a l di s pe r s i on i n
picoseconds per kilometer length of
the fiber per nanometer spectral
width of the source. At a particular
wavelength, the value of is a
characteristic of the material and is
(almost) the same for silica fibers.
When is negative, it implies that
the longer wavelengths travel faster;
this is referred to as normal group
velocity dispersion (GVD). Similarly,
a positive value of implies that
shorter wavelengths travel faster; this
is referred to as anomalous GVD.
The spectral width of an LED
operating around =825 nm is about
20 nm; at this wavelength for pure
silica ps/km.nm. Thus a
pulse will broaden by 1.7 ns per
kilometer of fiber. It is interesting
to note that for operation around
= 1300 nm, where
ps/km.nm, the resulting material
dispersion is only about 50 ps per
kilometer of the fiber. The very small
value of is due to the fact that the
group velocity is approximately
constant around = 1270 nm. Indeed
the wavelength 1270 nm is
usually referred to as the zero
material dispersion wavelength, and
it is because of such low material
di s pe r s i on t hat t he opt i c al
communication systems shifted their
operationtoaround 1300nm.
Optical communication systems in
operation today use LDs (Laser
Diodes) with 1550 nm having a
spectral width of about 2 nm. At this
wavelength the material dispersion
coefficient is 21.5 (ps/km-nm) and the
material dispersion (corresponding
to a 2 nmspectral width of the source)
wouldbe 43ps/km.
D
D
all
D
D
D
D
D
m
m
m
m
m
m
m
84.2
2.4

where is referred to as the spot size

of the mode field pattern and 2 is
called the mode field diameter
(MFD). MFD is a very important
characteristic of a single mode optical
fiber. For a step index fiber one has
the following empirical expression
for (see Ref. Ma1):
where is the core radius. As an
example, for the step index fiber
considered earlier (n = 1.447,
m) and operating at 1300nm
we have V giving
Note that the spot size is larger
than the core radius of the fiber;
this is due to the penetration of the
modal field into the cladding of the
fiber. The same fiber will have a
value of 1.908 at =1550nm giving
a value of the spot size
. The standard single mode
fiber designated as G.652 fiber for
operation at 1310 nm has an MFDof
9.2 0.4 m and an MFD of
10.4 0.8 mat 1550nm.
In digital communication systems,
information to be sent is first coded in
the form of pulses and then these
pulses of light are transmitted from
the transmitter to the receiver where
the information is decoded. Larger
the number of pulses that can be sent
per unit time and still be resolvable at
the receiver end, larger would be the
transmissioncapacityof the system. A
pulse of light sent into a fiber
broadens in time as it propagates
through the fiber; this phenomenon is
knownas pulse dispersionandoccurs
primarily because of the following
mechanisms:
w
w
w
a
a
w
V
Thus,
in general, the spot size increases with
wavelength
2
0
D= 0.003,
= 4.2 m
l
m
m
2.28 4.8m
5.5m
m.
m
6. Pulse Dispersion In Optical
Fibers
32
1. I n mul t i mode f i be r s ,
dispersion is caused by different rays
taking different times to propagate
through a given length of the fiber. In
the language of wave optics, this is
known as
because it arises due to different
modes traveling with different group
velocities.
2. Any given light source emits
over a range of wavelengths and,
because of the dependence of
refractive index on wavelength,
different wavelengths take different
amounts of time to propagate along
the same path. This is known as
material dispersion and obviously, it
is present in both single mode and
multimode fibers.
3. In single-mode fibers since
there is only one mode and there is no
intermodal dispersion. However,
apart from material dispersion, we
have what is known as waveguide
dispersion. Physically, this arises due
to the fact that the spot size (of the
f undament al mode) depends
explicitlyonthe wavelength.
Since we will consider only single
mode fibers, we will restrict our
discussions to only material and
waveguide dispersion.
Every source of light has a certain
wavelength spread which is often
referred to as the
. AnLEDwouldhave a spectral
width of about 25 nm and a typical
laser diode (LD) operating at 1300 nm
would have a spectral width of about
2 nm or less. The pulse broadening
(due to wavelength dependence of
the group velocity) is given in terms
of the material dispersion coefficient
(which is measured in ps/km-nm)
and is given by [see, e.g., Ref. Gh2,
Chapter 10]:
intermodal dispersion
spectral width of the
source
D
m
Material dispersion

- =
2
0
2
2
0
0
4
3
10
l
l
l d
n d
D
m
ps/km.nm
......(16)
+
-
+
-
5 ......(15) . 2 8 . 0 V
;
879 . 2 619 . 1
65 . 0
6 2 / 3
+ +
V V a
w
1550 nm window, fiber designs can
be modified to shift the zero
dispersionwavelengthto the 1550 nm
wavelength window. Such fibers are
referredto as dispersionshiftedfibers
(with zero dispersion around 1550
nm) or non-zero dispersion shifted
fi bers (wi t h fi ni t e but smal l
dispersion around 1550 nm). With
proper fiber refractive index profile
design it is also possible to have flat
dispersion spectrum leading to
dispersion flattened designs. Figure
12 gives the spectral variations of
dispersion in a standard SMF (with
zero dispersion close to 1310 nm) and
indispersionshiftedfibers (DSF).
It appears that when an optical fiber is
operated at the zero dispersion
wavelength, the pulses will not suffer
any dispersion at all. In fact, zero
dispersion only signifies that the
second order dispersive effects are
absent. In this case, the next higher
order dispersion namely third order
dispersion characterized by /
will become the dominating term in
determining the dispersion. Thus in
the absence of second order
dispersion, we canwrite for dispersion
sufferedbyapulse as
where = / represents the
dispersion slope at zero dispersion
wavelength and is measured in units
of ps/km-nm . Third order dispersion
becomes important when operating
cl ose t o t he zero- di spersi on
wavelength. In the presence of only
third order dispersion, the pulse does
not remain symmetric. Table I lists
values of and for some standard
fibers at 1550nm.
d d
S dD d
D S
3 3
2
b w
l
0
Dispersioninsingle mode fibers
In the case of a single mode optical fiber, the effective index (= / ) of the
mode would vary with wavelength even if the core and cladding media were
assumed to be dispersionless (i.e., the refractive indices of core and cladding
are assumed to be independent of wavelength). This dependence of effective
index onwavelength is due to the wave guidance mechanismandis referredto
as . Waveguide dispersioncanbe understoodfromthe fact
that even if we neglect the wavelength dependence of the core and cladding
refractive indices, as the wavelength changes, the effective index of the mode
changes [see, e.g., Eq. (13)]. It is this dependence of on the wavelength that
leads towaveguide dispersion.
Thus the total dispersion in the case of a single mode optical fiber can be
attributed to two types of dispersion namely material dispersion and
waveguide dispersion. It canbe shownthat the total dispersioncoefficient is
given to a good accuracy by the sum of material (D ) and waveguide (D )
dispersion coefficients. The material contribution is given by Eq. (16) while the
waveguide contributionfor astep-indexfiber is givenby
Asimple empirical expressionfor waveguide dispersionfor stepindex fibers is
givenby(see Ref. Ma1):
where is measuredinnanometers.
Inthe single-mode regime, the quantitywithinthe square brackets inEq. (17) is
usually positive; hence the waveguide dispersion is negative. Since the sign of
material dispersion depends on the operating wavelength region, it is possible
that the two effects namely, material and waveguide dispersions cancel each
other at a certain wavelength. Such a wavelength, which is a very important
parameter of single-mode fibers, is referred to as the zero-dispersion
wavelength ( ). For typical step index fibers the zero dispersion wavelength
falls inthe 1310 nmwavelengthwindow. Since the lowest loss inanoptical fiber
occurs at awavelengthof 1550nmandoptical amplifiers are available inthe
n k
waveguide dispersion
n
D
eff
eff
w
ZD
b
l
l
0
m
0
33
2
2
2
0
( )
w
n d bV
D V
c dV l
D
=-


......(17)
ps/km.nm 834 . 2 549 . 0 080 . 0 10
3
2 7
0
2
) (

- +
D
- = V
n
D
w
l
......(18)
Figure 12 : The wavelength dependence of total dispersion in a conventional single mode
fiber (CSF) andadispersionshiftedfiber (DSF)
( )
0
2
0
2 l
l
t
d
dD L D
= D
......(19)
T
o
t
a
l

D
i
s
p
e
r
s
i
o
n

(
p
s
/
K
m
.
n
m

)
l m
0
( m)
34
Fiber type (ps/km-nm) (ps/km. nm ) D S
2
StandardSMF(G.652) 17 0.058
LEAF(Corning) 4.2 0.085
Truewave-Reducedslope (OFS) 4.5 0.045
TeraLight(Alcatel) 8.0 0.057
Dispersion and maximum bit
rate insingle mode fibers
Dispersion compensating
fibers
In a digital communication system
employing light pulses, pulse
broadening would result in an
overlap of adjacent pulses, resulting
in inter symbol interference leading
to errors in detection. Apart from
this, since the energy in the pulse gets
reduced within the time slot, the
corresponding signal to noise ratio
(SNR) will decrease. One can offset
this by increasing the power in the
pulses. This additional power
requirement is termed as
. Increased dispersion
would imply increased power
penalty.
In order to keep the interference
between adjacent bits below a
specified level, the root mean square
width of the dispersed pulse needs to
be kept below a certain fraction of the
bit period. An estimate of the
maximum bit rate for a given link
length and dispersion coefficient
operatingat 1550nmis givenby
where is measured in Gbps, in
ps/km.nm and in km. Thus for a bit
rate of 2.5 Gb/s the maximumallowed
dispersion ( . ) is approximately
30,400 ps/nm while for a bit rate of 10
Gb/ s t he maxi mum al l owed
dispersionis 1900ps/nm.
There already exist millions of
kilometers of conventional single
mode fibers in the underground
dispersion
power penalty
B
L D
B D
L
D L
ducts operating at 1310 nm and these fibers have very low dispersion at the
operating wavelength. One could significantly increase the transmission
capacity of these systemby operating these fibers at 1550 nm(where the loss is
extremelysmall) andwe canhave the addedadvantage of usingEDFA(Erbium
Doped Fiber Amplifiers) for optical amplification in this wavelength range.
However, if we operate the conventional single mode fibers at 1550 nm, we will
have a significant residual dispersion of about 20 ps/km.nm. Such a large
dispersion would result in significant decrease in the information carrying
capacity of the communication system. On the other hand, replacing the
existing conventional single mode fibers by lower dispersion fibers would
involve huge costs. As such, in recent years there has been considerable
amount of workinupgradationof the installed1310 nmoptimizedoptical fiber
links for operation at 1550 nm. This is achievedby developing fibers with very
large negative dispersion coefficients, a few hundred meters to a few
kilometers of which can be used to compensate for dispersion accumulated
over tens of kilometers of the fiber inthe link.
ps/nm Gb 10 9 . 1
2 5 2
< DL B ......(20)
Table I : Values of dispersion and dispersion slope for some standard fibers at 1550 nm (after Ref. Li1)
Figure 13 : A short length of a DCF can be
used in conjunction with the conventional
single mode fiber (CSF) so as to have small
dispersionvalueat theendof thelink.
Figure 14 : The wavelength variation of
group velocity for a typical dispersion
compensating fiber and a typical
convent i onal si ngl e mode f i ber.
Calculations courtesy Dr Parthasarathi
Palai.
By changing the refractive index profile, we can alter the waveguide
dispersion and hence the total dispersion. Indeed, it is possible to have
specially designed fibers whose dispersion coefficient ( ) is large and
negative at 1550 nm. These types of fibers are known as dispersion
compensatingfibers (DCF's). Ashort lengthof DCFcanbe usedin conjunction
with the 1310 nm optimized fiber link so as to have small total dispersion
value at the end of the link. (see Fig. 13). In order to understand this
phenomenon, we have plottedinFig. 14(as asolidcurve) atypical wavelength
D
variation of the group velocity (of
the fundamental mode) for a
conventional single mode fiber (CSF)
with zero dispersion around 1300 nm
wavelength. As can be seen from the
figure, attains a maximum value at
the zero dispersion wavelength and
on either side it monotonically
decreases with wavelength. Thus, if
the central wavelength of the pulse is
around 1550 nm then the red
components of the pulse (i.e., longer
wavelengths) will travel slower than
the blue components (i.e., smaller
wavelengths) of the pulse. Because of
this the pulse will get broadened.
Now, after propagating through a
CSF for a certain length , the pulse is
allowed to propagate through a
length of the DCF in which the
group velocity varies as shown by
the dashed curve in Fig. 14. The red
components (i.e. longer wavelengths)
will now travel faster than the blue
components and the pulse will tend
to reshape itself into its original form.
Indeed if the lengths of the two fibers
( and ) are suchthat
then the pulse emanating from the
second fiber will be almost identical
to the pulse entering the first fiber as
showninFig. 13.
There are a number of different fiber
designs with optimized refractive
index profiles which have extremely
large negative dispersion coefficient
and a small length of which can
compensate for the accumulated
dispersion of a link. Some of the
important ones include depressed
clad designs, W-type fiber designs,
dual core coaxial designs etc. [see, eg.,
Refs. Au1, Th1and. Go1].
v
v
L
L
v
L L
g
g
1
2
g
1 2
35
Figure 15 shows the layout of a typical WDM (Wavelength Division
Multiplexed) optical communication system in which typically more than 200
closely spaced wavelengths (in the wavelength region 1530-1565 nm) are
simultaneously propagated in a single optical fiber; if each channel is assumed
to carry 5 Gbits/s we will have total transmission of over 1 terabits/s. As shown
in the diagram, the loss is periodically compensated by using erbium doped
fiber amplifiers and the dispersion is periodically compensated by using
dispersioncompensatingfibers.
1 1 2 2
0 D L D L + =
Figure 15 : A typical WDM fiber optic system with each wavelength carrying an
independent channel. The capacity of each channel can be ~10 to 40 gigabits/s and if we
have25to100channels, it will result inatotal capacityof thelink~1terabit/s
Since the wavelength dependence of dispersion of the link fiber and the
dispersion compensating fiber are in general different, in general the DCF
wouldcompensate for dispersion only at the design wavelength. However in a
wavelength division multiplexed system it is necessary to compensate for the
accumulated dispersion of all the wavelength channels simultaneously. For
this to happen the relative dispersion slope (RDS) which is the ratio of the
dispersionslope ( ) anddispersioncoefficient ( ) of the twofibers evaluatedat
the wavelength must be equal. Typically RDS of G.652 fiber is about 0.0034
nm while that of large effective area (LEAF) fiber fromCorningis about 0.0202
nm . DCFs withsimilar RDSvalues are commerciallyavailable.
In a normal optical fiber, the refractive index of the core and the cladding do not
change along the length of the fiber. If we introduce a periodic modulation of
refractive index along the length in the core of the optical fiber, then such a device
is referred to as a Fiber Bragg Grating (usually abbreviated as FBG); the spatial
S D
?
n
-1
-1
7. Fiber Bragg Gratings (FBG)
3
We also have what are knownas long period gratings (usually abbreviated as LPG) which couple light fromthe core into the cladding. The spatial period
of refractive index modulation in LPGs is of the order of a fewhundreds of micrometers; these also have interesting characteristics and find interesting
applications. Inthis paper we will confine ourselves to FBG's whichhave spatial period of refractive index modulationof the order of a micrometer.
L = 2d
d
n - Dn
0
n
0
+ Dn
Figure 16 : Reflection froma periodic structure consisting of alternate layers of refractive
indices (n + n) and (n - n), each of thickness d. Thus the periodicity of the grating is
Thereflectivityis maximumwhenthewavelengthequals 4n d; this wavelength is
knownas theBraggwavelength anddenotedby
0 0
0
D
l
D
L(=2 ). d
B
As the wavelength deviates from the
Bragg wavel engt h, t he waves
reflected from the layers will not add
constructively and thus the reflection
would decrease; thus the overall
reflection from such a multiple layer
medium would be very strongly
wavelength dependent (see Figs. 17
and18).
periodof the modulationis about half a micrometer . The periodic refractive index
modulation couples light from a forward propagating mode to a mode
propagating in the backward direction and thus acting like a mirror. The
reflectivity is high only for a very narrow range of wavelengths and thus it is a
highly wavelength selective mirror. FBG's exhibit very interesting spectral
properties andfindmanyapplications infiber optic communications andsensing.
In order to understand the principle of working of an FBG, we consider a
medium consisting of alternate layers of high and low refractive indices of
and of equal thickness [see Fig. 16]. Thus the spatial period of
the refractive indexvariationis givenby
= 2 ......(22)
Nowif << , andif we choose the thickness of eachlayer tobe
then the reflected waves arising out of individual reflections from the various
interfaces wouldall be inphase andshouldinterfere constructivelyresultingin
a strong reflection. Thus although each reflection may be weak, with a large
number of layers, there would be a large number of reflections and if they all
add constructively, the total reflection can become significant. Thus for strong
reflection at a chosen (free space) wavelength the period of the refractive
indexvariationshouldbe
This is referred to as the Bragg condition and is very similar to the Bragg
diffraction of X-rays from various atomic layers. Equation (24) corresponds to
the Bragg condition for normal incidence. The quantity is referred to as the
Braggwavelength.
As an example, we consider a periodic medium comprising of alternate layers
of refractive indices 1.51 and 1.49, i.e., = 1.5 and = 0.01. If we require a
strong reflectivity at = =1550 then the required periodicity is
3
n + n n n d
d
n n
,
n n
0 0
0
0
D -D
L
D
l
l
D
l l
B
B
B
D
36
0 0 0
4 4( ) 4( )
d
n n n n n
l l l
=
+ D - D
......(23)
0
2
B
n
l
L = 2d = ......(24)
4
Such multilayer coatings are used to achieve high reflectivity mirrors for applications in lasers. Since they are made of only dielectric media as
compared to normal mirrors, which are realized using metal coatings, these mirrors exhibit very low reflection losses and can also be made highly
wavelength selective.
Figure 17 : When light from a broad spectrum source is incident on the FBG (Fiber Bragg
Grating), then a narrow band of wavelengths is reflected strongly; the other wavelengths
get transmittedthroughthegrating.
Figure 18 : The spectrum of the wave
reflected by an FBG; solid line shows the
calculated spectrum and the open circles
show the experimentally measured values
of the FBG fabricated at CGCRI, Kolkata.
[Figure courtesy Dr S Bhadra and Dr S.
Bandyopadhyayof CGCRI, Kolkata.]
Nature produces beautiful colors
us i ng t hi s phe no me no n o f
interference. Morpho butterflies and
many varieties of beetles show
brilliant colors that change with the
orientation of the view (see ).
Such colors are referred to as
iridescent colors. The wings of these
Fig. 19
insects have naturally occurring
multiple stacks of layers; interference
of light reflected from the multiple
layers is responsible for their colors.
The core of an optical fiber is usually
silica doped with germanium; the
5500D
2 1.5
L =

1833 D
Thus exposing a germanium doped
silica fiber to the interference
pattern formed between two UV
beams leads to the formation of a
periodic refractive index variation
in the core of the fiber; the variation
of the refractive index of the core can
be writtenas
where
Since light propagates with an
effective refractive index given by
withinthe fiber, insteadof Eq. (24), we
will have the followingequation:
=2
The given period and the effective
index of the fiber would decide the
Bragg wavelength at which the
reflection would be maximum.
When broadband light or a set of
wavelengths are incident on an
FBG, then only the wavelength
cor r espondi ng t o t he Br agg
wavel ength wi l l get strongl y
reflected; the other wavelengths just
get transmitted to the output (see
Figs. 17 and 18). The peak reflectivity
and the bandwidth (spectral
width over which the reflectivity is
high) of a fiber Bragggratingof length
are approximatelygivenby
=tanh ......(28)
and
Here is the coupling constant
defined by I/ where I
represents the overlap integral
between the fiber modes and the
grating region (typically about 0.7 or
so). As an example, let us consider a
grating made in a fiber with 3 m,
NA = 0.1 and 1.45. The fiber is
single moded at 850 nm and the
fundamental mode has an effective
indexof 1.4517. If we wishtostrongly
n
n
R
L
R L
n
a
n
eff
B eff
l L
Dl
k
k
k= l
= m
=
2
pD
B
2
......(27)
37
Figure 19: Thewings of thesebutterflies havenaturallyoccurringmultiplestacks of layers;
interference of light reflected from the multiple layers is responsible for their beautiful
colours. Ref.: byJ.C.
GonzatoandB. Pont in
http://www.labri.fr/perso/gonzato/Articles/GONZATO_Butterfly_WSCG2004.pdf
Aphenomenological representationof iridescent colors inbutterflywings
doping results in a slightly higher refractive index. Now, when a germanium-
doped silica core fiber is exposed to ultraviolet radiation (with wavelength
around 0.24 m), the refractive index of the germanium-doped region
increases; this is due to the phenomenon known as which was
discovered by Kenneth Hill in 1978. The refractive index increase can be as
large as 0.001 in the core of the fiber. If the fiber is exposed to a pair of
interfering UV beams (see Fig. 20), then we would obtain an interference
pattern. In regions of constructive interference, the refractive index would
increase while inregions of destructive interference, the refractive indexwould
remainunchanged. Since the fringe widthwoulddependonthe angle between
the interfering beams, the period of the grating can be controlled by choosing
the angle betweenthe interferingbeams.
m
photosensitivity
Figure 20 : A Fiber Bragg Grating is produced by allowing two laser beams to produce an
interferencepattern. In regions of constructive interference, the refractive index would
increase while in regions of destructive interference, the refractive index would remain
unchanged.
1
cos n n n Kz = + D ......(25)
2
K
p
=
L
......(26)
2
0 B
B 0 B

1
2
nL
n L
D D
+


......(29)
reflect a wavelength of 850 nm, the
correspondi ng spat i al peri od
requiredwouldbe
Note that the required spatial period
is less than the optical wavelength. If
the grating has a length of 1 cm, then
the required value of for 20% peak
reflectivitywouldbe
The corresponding
bandwidth is Note the
extremely small bandwidth of the
reflectedlight.
From Eqs. (28) and (29) it also follows
that one can design gratings having
same peak reflectivity but different
bandwidths by appropriate choice of
the peak index modulation and
grating length. The bandwidth can be
increasedor reducedby decreasing or
increasing the grating length while
keepingthe product constant.
Figure 18 shows a measuredreflection
spectrum of an FBG. The peak
wavelength of reflection is about 1551
nmandthe wavelengthregionof high
reflectivity is about 0.2 nm wide
showing the strong wavelength
sel ect i vi t y of such grat i ngs.
Wavelengths not falling within this
bandwidth get transmitted through
the grating.
Fiber Bragg gratings find many
applications in telecommunication
and s ens i ng. Thes e i nc l ude
applications in fiber grating sensors,
add/drop multiplexers, to provide
external feedback for laser diode
wavelength locking, dispersion
compensation etc. Here we discuss
some of the important applications.
FBGs have a great potential for
applications as sensors for sensing
D
Dl
D
n
nL
assuming =1.0. I
0.025 . nm
Some applications of FBGs
FBGbasedsensors
38
mechanical strain, temperature, acceleration etc. Since the Bragg wavelength
(wavelength of maximum reflectivity) depends on both the refractive index of
the fiber as well as on the period of the grating, any external parameter that
changes anyof these wouldresult ina change inthe reflectedwavelength. Thus
by measuring the , the external perturbations
affecting the grating could be sensed. This is the basic principle of their
applicationinsensing.
changes in the reflected wavelength
m 293 . 0
2
m
l
= L
eff
n
B
Figure 21 : Measured variation of the Bragg wavelength with strain applied to an FBG. The
inset shows the reflection spectrumfromthe grating without and with applied strain. The
measurements werecarriedout at IITDelhi.
Figure 21 shows a measured variation of Bragg wavelength as a function of
longitudinal strain (pulling) of the grating. The grating shown has a strain
sensitivityof 1.3pm/ (picometer per microstrain) at 1550nm. This means that
if a one meter long fiber is pulled by a length of 1 m, this would result in a
strain of 1 which would result in a wavelength shift of 1.3 pmof the reflected
wavelength.
Since the refractive index as well as the period of the grating change with
change in temperature, temperature change of an FBG would also result in a
change in the peak reflected wavelength. The temperature sensitivity of FBGs
is typically about 6 pm/ C i.e., a change of temperature of 1 C will result in a
change of the reflectedwavelengthby6pm.
me
m
me
o o
From the above discussions it can be seen that the changes in the peak
wavelength are indeed very small and hence special techniques are needed for
sensingsuchsmall changes. Since the changes inreflectedwavelengthcouldbe
due to strain as well as temperature changes, techniques to deconvolve
(separate) the changes in temperature and strain need to be implemented for
precise sensing.
Figure 22 : Multiplexed FBG sensors. Changes in , and will indicate strains in the
threecorrespondingregions
l l l
1 2 3
5
1
ln 2.4 10
2 1
B
R
n
LI R
l
p
-

+
D =


-

Among the various issues in
connection with FBG sensors, is the
problem of separating the changes
brought about by temperature and
strain. There is considerable work
going on to achieve this. Fiber optic
sensing technology is rapidly
advancing and FBGs are now being
tested for structural monitoring. In
the future it may be possible to
multiplex perhaps 100 sensors on a
single fiber which gives enormous
capabilitytostructural monitoring.
As compared to the other types of
fiber optic sensors, FBGs have some
distinct advantages:
They are immune to intensity
fluctuations of the source since the
sensor signal is in the form of shift of
peak reflection wavelength rather
thanchanges inintensityof the light.
The Bragg wavelength is almost a
l i near f unct i on of st r ai n or
temperature.
Fiber grating sensors can be
written directly on the fiber and are
mass producible.
Multiplexing of FBGs is relatively
easy and thus they can be configured
i nt o quasi - di st ri but ed sensor
assemblies.
I n many t el ecommuni cat i on
networks carrying wavelength
division multiplexed signals, it is
required to add or drop a wavelength
channel. In such a case we need to
have devices which can choose the
desired wavelength and filter it out of
the link or let us add signal
correspondi ng t o t he chosen
wavelength. Figure 25 shows a typical
configuration of an add/drop
multiplexer basedonanFBG. AnFBG
which reflects light at the desired
wavelength that needs to be dropped
is placed between two optical
?
?
?
?
Add/dropmultiplexer
One of the important attributes of FBG sensors is that they can be multiplexed
into a single fiber. Figure 22 shows a typical arrangement of a multiplexed
sensing arrangement in which a number of FBGs with different Bragg
wavelengths are fabricatedat different points alongthe lengthof a single mode
fiber. Light from a broad band source is coupled into the fiber and light at
different wavelengths get reflected from the individual gratings and are
analyzed by the detection circuit. The wavelengths of the FBGs are so chosen
that they do not overlap with each other and fall within the band of the source.
By measuring the changes in the Bragg wavelength of individual FBGs, the
strains or temperature changes at each of the location of the FBG can be
independentlymeasured.
39
Sensor 1
Sensor 2
Splice
housing box
Figure 23: FBGbasedtemperature sensor
system on 400 KV power conductor at
Subhashgram substation (near Kolkata)
of Powergrid Corporation of India.
Photographs courtesy: Dr. Kamal
Dasgupta and Dr. Tarun Gangopadhyay,
CGCRI, Kolkata.
Figure 23 shows an installation of an FBGbased temperature sensor systemon
400 KV power conductor at Subhashgram substation (near Kolkata) of
Powergrid Corporation of India. Since FBGs are based on dielectrics, they are
totallyimmune toelectromagnetic interference.
Preliminary experiments are underway to use FBGbased sensors in real world
environments suchas inbridges, other civil structures etc. Figure 24(a) shows a
350 meter long 5 span bridge in Norway wherein 32 FBG sensors were laid for
strainmonitoring. Figure 24(b) shows the response of the grating sensor as a 50
ton lorry is driven across the bridge at 30 km/h speed. The measurements
carried out using the FBG sensors were found to be consistent with the
measurements usingother conventional sensors.
Figure 24 : (a) FBG
sensors on a bridge in
Norway.
(b) The response of the
grating sensor as a 50
ton lorry is driven
across the bridge at 30
km/hspeed.
since the period of the grating varies
al ong t he l engt h, t he Bragg
wavelength also varies along the
position in the grating (see Fig. 26).
When light propagates through such
a grating, different wavelength
components present in the incident
wave will get reflected at different
positions along the grating; this will
l ead t o di fferent wavel engt h
components having different time
delays to return to the input end. By
using an appropriately chirped FBG
one can indeed compensate for the
differential delay of different
wavelengths accumulated while
propagating through an optical fiber
link.
Because of t he vari ous l oss
me c h a n i s ms i n a n o p t i c a l
communi cat i on syst em, i t i s
necessary to periodically amplify the
signal. Optical amplifiers are devices
that amplify the incoming optical
signals in the optical domain itself
without any conversion to the
electrical domain; these amplifiers
have revolutionized long distance
f i ber opt i c communi cat i ons.
Compared to electronic regenerators,
optical amplifiers have the following
advantages:
They do not needany highspeed
electronic circuitry.
They are transparent to bit rate
andformat andmost importantly
They can amplify multiple
opt i c al s i gnal s at di f f er ent
wavelengths simultaneously.
Thus their development has ushered
in the tremendous growth of
communication capacity using
wavelength division multiplexing
( WDM) i n wh i c h mul t i pl e
wavelengths carrying independent
?
?
?
8. Erbi um Doped Fi ber
Amplifier (EDFA)
circulators. (The circulator is a device in which light incident from port 1 gets
transmittedtoport 2andlight incident fromport 2gets transmittedfromport 3.)
40
Figure 25 : An add-drop multiplexer based on FBG.
From among the incoming wavelength channels of the DWDM transmission
system, the wavelengthmatching the FBGwavelengthwill get reflectedandbe
routed to the dropped port. All other wavelength channels proceed in the
forward direction along the link. At the same time signal at the same
wavelength can be added using the second circulator with the same grating
reflectingthe addedchannel into the link. Since anFBGreflects onlythe chosen
wavelength and transmits all other channels, it is possible to add and drop
multiple channels by using multiple FBGs at the desired wavelengths instead
of a single FBG. The gratings used in add drop multiplexers need to have high
reflectivity so that there is no residual signal leading to cross talk among the
droppedandaddedc
When pulses of light propagate through an optical fiber link they suffer from
pulse broadening due to the dependence of the velocity of propagation on the
wavelength. This dispersion needs to be compensated in the link and this can
be achieved either through the use of dispersion compensating fibers or with
the helpof chirpedFBGs.
hannels.
Dispersioncompensation
Figure 26 : A chirped FBG used as a dispersion compensator
In a uniform FBG, the period of the refractive index modulation is constant
along the length of the grating. If the period of the grating varies along its
length, this is referred to as a . In such chirped FBGs chirped fiber Bragg grating
signals are propagated through the
same single mode fiber, thus
multiplying the capacity of the link.
Of course compared to electronic
regenerators, they have some
drawbacks t oo: t hey do not
c o mpe ns a t e f o r di s pe r s i o n
accumulated in the link and they also
addnoise tothe optical signal.
41
Optical amplification by EDFAis based on the process of stimulated emission,
which is the basic principle behind laser operation. In fact a laser without any
optical feedback is just an optical amplifier. Figure 28 shows the three lowest
lying energy levels of erbium ion located within silica glass. Light from a
semiconductor laser at 980 nm (called as the pump laser) excites the erbium
ions from the ground state to the level marked E , i. e., erbium atoms in the
groundstate absorbthe 980 nmradiationandget excitedtothe level markedE .
The level E is a short-livedenergy level andthe ions fromthis level jumpdown
to the level marked E after a fewmicroseconds. The lifetime of level E is much
longer and is about 12 milliseconds. Hence ions brought to level E stay there
for a significantly longer time. Thus by pumping hard enough, the population
of ions inthe level E canbe made larger thanthe populationof level E andthus
achieve population inversion between levels E and E . In such a situation, if a
light beamat a wavelengthcorrespondingto the energydifference ( ) falls
on the collection, it will get amplified by the process of stimulated emission.
For erbiumions, the energy difference ( - ) is approximately 1.28 x 10 J and
the corresponding wavelength falls in the 1550 nm band and thus it is an ideal
amplifier for signals inthe 1550nmwindow.
3
3
3
2 2
2
2 1
1 2
2 1
2 1
E E
E E
-19
Figure 27 : Shows a typical fiber optic
systemwith optical amplifiers as a booster,
in-lineandpre-amplifier
Optical amplifiers can be used at
many points in a communication link.
Figure 27 shows some typical
examples. Booster amplifier is usedto
boost the power of the transmitter
before launching into the fiber link.
The increased transmitter power can
be used to go farther in the link. The
preamplifier placed just before the
receiver is used to increase the
receiver sensitivity (the minimum
power required by the receiver to
function properly). Inline amplifiers
are used at intermediate points in the
l i nk t o over c ome t he f i ber
transmissionandother losses. Optical
amplifiers can also be used for
overcoming splitter losses for
example for distributionof CATV.
There are currently three main types
of optical amplifiers namely the
(EDFA),
the (RFA) and
the
(SOA). Today most optical fiber
communication systems use EDFAs
due to their advantages in terms of
bandwidth, high power output and
erbium doped fiber amplifier
Raman fiber amplifier
semiconductor optical amplifier
noise characteristics. As such, in this
section, we will briefly discuss the
characteristics of EDFAs.
Figure28: Thethreelowest lyingenergybands of erbiuminsilicamatrix. Thepumplaser at
980 nmexcites the erbiumions fromthe ground level E to the level marked E fromwhich
they make a non-radiative transition to the level E . The level E is a metastable level and
population inversion between the levels E and E is responsible for the amplification of
signals inthe1550nmband.
1 3
2 2
2 1
Figure 29: Aschematic of anEDFAconsisting of the 980nmpumplaser, a WDMcoupler and
a short piece of erbium doped fiber (EDF). The tap couplers are used to monitor the input
andoutput fromtheamplifier andtheisolator prevents anyreflectedlight toenter theEDFA.
WDM
Tap
Tap photodiode
EDF
Weak Signal
~1550nm
Amplified signal
Isolator
980 nm pump
Laser Diode
Amplified
signal
start to reduce. This happens because
as the pump propagates through the
doped fiber it gets absorbed and thus
its power reduces. After propagating
a certain distance its power is too
small to create population inversion
and thus after this length, the doped
fiber would start to absorb the signal
rather than amplify it. Figure 30
shows the variation of gain with the
length of the doped fiber for different
pump powers. Hence for a given
pump power there is an optimum
length of the doped fiber to achieve
maximumgain.
For a given length of the doped fiber,
as the pump power increases, we
expect the gain to increase. At the
same time, as the pump power
increases it creates more and more
population inversion and once all
erbium ions in the fiber are excited
there would be no more erbium ions
available and hence the gain would
saturate. Similarly for a given pump
power, the gain remains constant for
small signal powers and as the input
signal power increases, the gain starts
to reduce. This is called signal
saturation and is a characteristic of all
amplifiers. Figure 31 shows the
variation of gain with the output
signal power showingsaturation.
Now, in the case of erbium ions
located within silica glass, due to
interactions between neighboring
atoms, the energy levels are not sharp
levels but are broadened, i.e., ions can
have energies over a range of values
and this implies that as they jump
from the higher level to the lower
level their wavelengths can have a
range of values. Hence the system is
capable of absorbing or emitting
over a band of wavelength and
consequently of amplifying optical
signals over abandof wavelengths.
Figure 29 is a schematic of an EDFA
which consists of a short piece (~ 20
meter in length) of erbium doped
fiber (EDF), a single mode fiber doped
witherbium(typically with100 to 500
parts per million- ppm) in the core
and which is pumped by a 980 nm
pump laser through a wavelength
di vi si on mul ti pl exi ng (WDM)
c oupl er. The WDM c oupl er
multiplexes (combines) light of
wavelength980 nmand1550 nmfrom
two different input fibers to a single
output fiber. The 980 nm pump light
is absorbed by the erbium ions to
create population inversion between
levels and . Thus, incoming
signals in the 1550 nm wavelength
region get amplified as they
propagate through the population
inverted doped fiber. The tap
couplers are couplers which tap a
very small fraction of the light from
the input and from the output to
enable one to measure the signal
power entering and exiting the
amplifier. These values are used to
control the amplifier for constant gain
or constant output power operation.
The isolator is a device that allows
light to propagate only along one
direction. The isolator is placed to
prevent any reflected light from
E E
2 1
42
enteri ng the ampl i fi er whi ch
otherwise can get destabilized and
start tooscillate like alaser.
There are three important parameters
of anyoptical amplifier: gainandgain
spectrum, saturation behavior and
noise. In the following sections we
will discuss these.
The gain of an EDFAis defined as the
ratio of the output signal power ( )
tothe input signal power ( ):
The gain depends on the doping
concentration and doping profile of
the erbium doped fiber, the length of
the fiber and the pump power.
Typical gain values of an EDFA are
about 20 to 30 dB, i.e. the output
power is about hundred or thousand
times the input power. The gain
providedbythe amplifier depends on
the erbium doping in the doped fiber,
the length of the fiber and the pump
power. These parameters are usually
optimized for achieving the desired
gaincharacteristics.
P
P
out
in
Gainandgainspectrum

=
in
out
P
P
dB G log 10 ) ( ......(30)
Figure 30 : Variation of gain with the length
of EDFAfor different values of pumppowers.
For agivenpumppower thereis anoptimum
lengthfor achievingmaximumgain.
For a given input pump power, as we
increase the length of the doped fiber,
the gain would first increase and then
after reaching a maximum would
Figure 31 : Variation of gain with output
signal power showingsignal saturation.
Figure 32 shows typical measured
gain spectra of an EDFA for different
input signal powers. As can be seen,
EDFAs can provide amplifications of
gr eat er t han 20 dB ( power
amplification by a factor greater than
100) over the entire band of 40 nm
from 1525 nm to about 1565 nm. This
wavelength band is referred to as the
Cband(conventional band) andis the
most common wavelength band of
operation. With proper amplifier
optimization, EDFAs canalsoamplify
signals in the wavelength range of
1570 to 1610 nm; this band of
wavelengths is referred to as the L-
band (long wavelength band). The C-
band and L-band amplifiers together
can be used to simultaneously
amplify 160 different wavelength
channels. Such systems are now
commerciallyavailable.
It can be seen from Fig. 32 that
although EDFAs can provide gains
over an entire band of 40 nm, the gain
is not flat, i. e., the gaindepends onthe
signal wavelength. This is especially
true for small input signal powers.
Thus if multiple wavelength signals
with same power are input into the
amplifier, then their output powers
will be different. In a communication
system employing a chain of
amplifiers, a differential signal gain
a mo n g t h e va r i o u s s i g n a l
43
wavelengths (channels) from each amplifier will result in a significant
difference in signal power levels and hence in the signal to noise ratio (SNR)
among the various channels. In fact, signals for which the gain in the amplifier
is greater than the loss suffered in the link, will keep on increasing in power
level while those channels for which the amplifier gain is less than the loss
suffered will keep on reducing in power. The former channels will finally
saturate the amplifiers and will also lead to increased nonlinear effects in the
link while the latter will have reduced SNR leading to increased errors in
detection. Thus such a differential amplifier gain is not desirable in a
communication system and it is very important to have gain flattened
amplifiers.
There are basically two main techniques for gain flattening: one uses external
wavelengthfilters toflattenthe gainwhile the other one relies onmodifyingthe
amplifying fiber properties to flatten the gain. We will briefly discuss only the
first technique. In gain flattening using external filters, the output of the
amplifier is passed through a special wavelength filter whose transmission
characteristic is exactly the inverse of the gain spectrum of the amplifier (see
Fig. 33). Thus channels which have experienced greater gain in the amplifier
will suffer greater transmission loss as they propagate through the filter while
channels which experience smaller gain will suffer smaller loss. By
appropriately tailoring the filter transmission profile it is possible to flatten the
gainspectrumof the amplifier.
Gainflattening of EDFAs
Figure 33 : Principle of gain flattening in EDFAs. The filter transmission profile is exactly
oppositetothegainprofileof theamplifier resultingingainflattening.
Filters withspecific transmissionprofiles canbe designedandfabricatedusing
various techniques. These include thinfilminterference filters andfilters based
on short period or long period fiber gratings (LPG). Typical gain flatness of
better than 1 dB can be achieved and gain flattened EDFAs are commercially
available.
The wavelengthvariationof gainof anEDFAdepends onthe fiber properties as
well as the doping level and where the erbium ions are actually located within
the fiber. It is possible to achieve significant gain flattening by an appropriate
choice of fiber type and also by controlling the doping of erbium in the fiber.
EDFAs realized using such fibers can lead to lower cost since the number of
components used in the amplifier reduces. Such a technique is referred to as
intrinsic gainflattening.
Figure 32 : Measured variation of gain with
wavelength for different input signal
powers. As theinput signal power increases,
the gain decreases. Also note that the gain is
different at different wavelengths. The
measurements werecarriedout at IITDelhi.
44
for laser oscillation namely compensation of loss by the gain provided by
population inversion. This would result in an output laser beam from both the
ends of the coupler. The wavelength of emission is usually determined by the
wavelength satisfying the maximum gain and minimum loss; this is around
1530 nm for erbium doped fibers. Figure 35 shows the output from a erbium
doped fiber laser as the pumping in increased. Just before starting to lase the
9. The Fiber Laser
In a fiber laser, the active medium
providing optical amplification is an
optical fiber with its core doped with
suitable dopants. When such a fiber is
pumped by a suitable light source
(usually another laser) then we can
achieve populationinversionbetween
two of the energy levels thus leading
to optical amplification. By providing
an optical feedback using mirrors at
the end of the amplifying fiber, it is
possible to achieve laser oscillation
leading to a fiber laser. Indeed in 1961,
Elias Snitzer wrapped a flas
ions clad in a
lower index glass) and when suitable
feedback was applied, the first fiber
laser was born. Thus, the fiber laser
was fabricated within a year of the
demonstrationof the first ever laser by
Theodore Maiman.
hlamp
around a glass fiber (having a 300 m
core doped with Nd
3+
Fi gure 34 : A typi cal f i ber l aser
configuration
Figure 34 shows a schematic of a fiber
laser. As anexample let us consider an
erbiumdopedfiber laser. Light froma
pump laser emitting at 980 nm is
coupled into a short length of erbium
doped fiber using a wavelength
di vi si on mul ti pl exi ng (WDM)
coupler. The WDM coupler at the
output removes any unused pump
laser power. If the two ends of the
signal ports of the WDM coupler are
cut properly and not treated with any
coatings etc., they can reflect about 4
percent of the incident light. Since the
gains providedby erbiumions is very
large, this small reflectivity is
sufficient to satisfy the condition
Figure 35 : Output spectrum of a fiber laser below threshold and above threshold. The
measurements werecarriedout at IITDelhi.
pump power is insufficient to
overcome the losses in the cavity and
thus the output is only amplified
spontaneous emission(lower curve in
Fig. 35). As we increase the pump
power, the erbium doped fiber starts
to lase and the spikes correspond to
t he var i ous wavel engt hs of
oscillation; the ends of the fiber act as
the resonator.
In case it is required to have the laser
oscillate at a specific wavelength
within the gain bandwidth of the
erbium ion, then this can be achieved
by using a fiber Bragg grating at
one end of the laser. Since an FBG
reflects a particular wavelength, this
wavelength would have much higher
feedback into the laser cavity and
thus would suffer much lower loss
than other wavelengths. This would
ensure that the fiber laser oscillates at
the frequency as determined by the
FBG.
Fiber lasers possess many interesting
advantages vis a vis other laser
systems. In particular, since the laser
beam is confined to a very small cross
sectional area within the core of the
fiber, large pump intensities can be
achieved even with small pump
powers and thus leading to lower
pump power thresholds. Since both
the pump and the laser beam are
propagating within the fiber they
overlapverywell andthis alsoadds to
increased efficiency of the laser.
Efficiencies of 80 percent are possible.
Since the fiber guides the pumpbeam,
one can use very long length cavities
wi t hout bot heri ng about t he
divergence of the pump laser beam.
Since the ratio of surface area to
volume of fiber laser is very large, it
Figure 36 : A double clad fiber design to
achievehighpower lasers
does not suffer from thermal
problems and heat dissipation is
much easier. The output beam is of
very good quality since it emerges as
the fundamental mode of the fiber.
Alsosince the components inthe laser
are made up of fibers which are all
spliced, there are no mechanical
perturbation problems such as in
bulklasers withseparate mirrors.
With conventional fibers with doped
single mode core and a cladding, the
laser power is restricted to about 1
Watt. To achieve higher output
powers, fiber lasers use double clad
fibers as the amplifying medium (see
Fig. 36). In this fiber, the central core
guides the laser wavelength and is
single moded at this wavelength. The
inner cladding is surrounded by an
outer cladding and this region acts as
a multimoded guide for the pump
wavelength. This radius of the inner
cladding is large and so is the
refractive index difference between
the inner cladding and the outer
cladding. This ensures that power
from large area diode lasers can be
launched into the fiber efficiently. At
the same time since the laser
wavelength is propagating as a
fundamental mode in the inner core,
the laser output would be single
m o d e d . T h e p u m p p o we r
propagating in the inner cladding
propagates in the form of different
rays (or modes). If the cladding is
circular in cross section, then it is
possible that some of the rays
propagating in the cladding never
have an opportunity to cross the core
and this portion of the pump would
never be used in creating inversion
and thus l eadi ng to reduced
conversion efficiencies. In order that
all the rays corresponding to the
pumppower propagatinginthe inner
cladding of the fiber crosses the core,
45
the inner cladding is made non-circular (see Fig. 36). This type of designs can
leadtoverymuchincreasedpumpconversionefficiencies.
Figure 37 (a) shows a fiber laser construction with pump power (at 975 nm)
coupled through one end of the doped fiber and a dichroic mirror (which
transmits the pump wavelength but reflects the laser wavelength) and Fig.
37(b) shows the corresponding laser output with the input pump power. With
cladding pumped geometry continuous wave fiber laser (Ytterbium doped
fiber) emitting more than 1000 Watts of power has been realized. Such high
Figure37: (a) AnYbdopedfiber laser
pumped by a 975 nmlaser diode. (b)
Variation of the output laser power
withinput pumppower [After Up1].
power fiber lasers are expected to find wide applications in various industries
such as for cutting, hole drilling etc. Fiber lasers should make these
applications quite convenient and practical. Using various mode locking
techniques, pulse widths as short as a few hundred femtoseconds have been
realized. Such lasers find applications in various scientific and technological
areas.
Figure 38 : Increase of fiber laser power
with year. Note the steep rise after 2002
[After Li2].
Figure 39:Extremely highpower fiber laser
emitting a continuous wave power output
of morethan1kW. [After Je1].
achievable is about 200 fs. Thus fiber
lasers are very interesting candidates
for generation of ultra short pulses of
a few hundred femto-seconds
duration. Such femto-second fiber
lasers are very interesting as they are
light weight andverycompact, do not
need any alignments and integrated
with nonlinear crystals can be used to
generate ultra-short pulses in the
visible regionof the spectrum.
When a light wave is incident in a
medium, apart from Rayleigh
scattering, in the scattered light one
also observes very weak light at
frequencies slightly smaller and
slightly larger than the incident
frequency. This scatteringmechanism
is referred to as Raman scattering and
was discoveredby Sir C. V. Raman for
which he received the Nobel Prize in
1930. This scattering is extremely
weak and only about one photon out
of 10 million photons gets Raman
scattered. Figure 42(a) is a schematic
of Raman scattering in which the
scattered photon has a slightly lower
energy and hence lower frequency
(larger wavelength) than the incident
photon. This scattered light is also
referred to as Stokes line. In this
process a small well defined value of
energy from the incident photon is
given away to the molecule and thus
the scattered photon has a slightly
smaller energy than the incident
photon. At the same time the
molecule finds itself in an excited
state after the interaction process.
The difference in energy between the
incident photon and the scattered
photon is a characteristic of the
molecule since the energylevels of the
molecule are well defined. A
measurement of the difference in
energy between the incident and the
Raman scattered photon can be used
Ramanscatteringinfibers
Usi ng components devel oped
s p e c i f i c a l l y f o r t h e
telecommunication application of
fiber optics, fiber lasers have seen an
explosive growth in terms of high
output powers, ultrashort pulses and
extensive wavelength region. With
developments in large mode area
optical fibers, photonic crystal fibers
etc. the field is continuing to grow.
Figure 38 show the growth of fiber
46
lasers during the last 15 years
showingasteeprise after 2002.
It is indeed possible to generate
extremely high powers from fiber
lasers. Figure 39 shows the output
from a Ytterbium doped large core
fiber laser with a continuous wave
output power of 1.36 kW. There are
efforts to achieve even larger powers
fromfiber lasers andsuch high power
f i ber l asers are expect ed t o
revolutionize the area of applications
of lasers inindustries.
Figure 40 shows a photograph
demonst rat i ng breakdown i n
atmospheric air by a pulsed fiber
laser. With such high optical powers,
optical fibers start toexhibit nonlinear
effects which tend to degrade the
performance of the fiber laser. Since
the nonlinear effects are proportional
to the intensity of the propagating
radiation one way to reduce the
nonlinear effects is to increase the
mode area so that for a given power
the intensity would be less. Using
photonic microstructure fibers it is
possible to achieve single mode
operation over a large wavelength
range and also achieve large mode
areas. Figure 41 shows typical
examples of large mode area fibers
with a microstructure cladding.
These fibers have a very small
numerical aperture of less than 0.03
and core diameters larger than 60 m
leading to mode field diameters of
about 50 m.
Mode locking is a technique used to
generate ultra-short pulses of light;
the broader the gainbandwidthof the
laser, the shorter the achievable pulse
duration. With a gain bandwidth of
40 nm at 1550 nm, the corresponding
spectral bandwidth of an erbium
dopedfiber laser is about 5 THz andif
modes over this bandwidth are mode
locked then the pulse duration
m
m
Figure 40 : Breakdown of air due to the
intense electric field produced at the focus
of a fiber laser output by a converging lens.
[After Ga2].
Fi g ur e 41 : La r g e mode a r e a
microstructure fibers for high power laser
applications.
undergoes Ramanscattering fromthe
molecules of the glass fiber and this is
expected to give rise to light
appearing at higher wavelengths.
Indeed Fig. 43 shows a typical
spontaneous Raman spectrum from
an optical fiber pumped by radiation
at 1450 nm. As can be seen the
scattered radiation occupies a large
band and the peak of the scattered
radiation lies at about 100 nm away
from the pump wavelength. Indeed
Raman scattering in silica leads to a
Raman shift of between 13 and 14
THz which corresponds to about 100
nm at the wavelength of 1550 nm.
Also notice that the Raman spectrum
from the optical fiber is not sharp but
is broadened.
When we launch a high power light
beaminto an optical fiber, we observe
appearance of Raman scattered light
at the end of the fiber. This is referred
to as spontaneous Raman scattering.
If in addition to the strong pumplight
we also launch a weak light beam
(referred to as the signal beam) with
its wavelength lying within the band
of spontaneous Raman scattering,
then it leads to what is referred to as
stimulated Raman scattering. In this
c as e, t he pump and s i gnal
wavelengths are coherently coupled
by the Raman scattering process and
the scattered radiation is coherent
with the incident signal radiation
much like stimulated emission that
occurs in the case of a laser. The
coherent nature of the process implies
that the scattered light and the
incident light are identical in all
respects such as direction of
propagation, state of polarization,
frequency etc. It is this process that is
used to build Raman fiber amplifiers.
Since the spontaneous Raman
scattering spectrum is broad, the
corresponding gain spectrum of the
in the identification of the molecular species and finds wide applications in
many diverse fields such as material characterization, pollution monitoring,
explosives identification, medical applications etc.
In another process the incident photon may find the molecule in the excited
state and when it gets scattered may take up that energy and get scattered to a
47
Figure 42 (a) : Stokes line generation in Raman scattering
Figure 42(b): Anti-Stokes line generation in Raman scattering
Figure 43 : Ramanscatteringfromanoptical fiber. The lower figure shows the spectrumof
theRamanscatteredlight whentheinput light is at 1450nm.
photonof higher energyleavingthe molecule inthe lower energystate (see Fig.
42(b)). The light scattered this way is termed as anti-Stokes line. Since the
probabilityof findingmolecules is higher inthe lower energystate as compared
to the higher energy state, the Stokes emission (into a lower energy photon) is
muchmore intense thanthe anti-Stokes emission(intohigher energyphoton).
Now let us consider an experiment in which we send a strong light beam at a
wavelength of 1450 nm through a long (~ 10 km) optical fiber. The light beam
Raman amplifier is also very broad.
The other interestingfeature is that no
matter what the wavelength of the
pump light is, the fiber can act like an
amplifier in the wavelength range
corresponding to the spontaneous
Raman scattering spectrum. Hence if
we needto amplify signals in the 1310
nm(~ 229 THz) windowthenwe need
to choose a pump wavelength of
about 1240 nm(~ 242 THz) which will
give a peak Raman scattering at a
wavelength of 1310 nmand thus such
a pump will lead to amplification of
signals at 1310 nm. Similarly if we
need to amplify signals in the 1550 (~
194 THz) window then we need to
choose a pump wavelength of about
1450 nm (~ 207 THz); in each case the
pump frequency is about 13 THz
more than the signal frequency.
Notice that unlike EDFA which
operated only in specific wavelength
bands, Raman amplifier can operate
at anywavelengthregion.
48
region, they allow us to expand the operation region of fiber optic
communication systems to other bands wherein EDFAs do not operate. Apart
fromthis, Raman amplifiers can also be used to extend the operation of optical
fiber communication systems in C-band in terms of increased bit rates or
increased distance between repeaters. Raman fiber amplifiers have helped
expand the operating wavelength band of optical amplifiers to the entire low
loss window of transmission of silica based optical fibers. Combining Raman
fiber amplifiers with EDFAs to form hybrid amplifiers can lead to additional
benefits in terms of better noise figure and extended reach of optical fiber
communicationsystems.
Consider a light beam having a power of 100 mW propagating through an
optical fiber having an effective mode area of 50 m . The corresponding
optical intensityis 2 x 10 W/m . At suchhighintensities, the nonlinear effects in
optical fibers start to influence the propagation of the light beam and can
significantly influence the capacity of a WDM optical fiber communication
system. The most important nonlinear effects that affect optical fiber
communication systems include self phase modulation (SPM), Cross phase
modulation (XPM) and four wave mixing (FWM). Stimulated Raman
scattering (SRS) and Stimulate Brillouin scattering (SBS) are also important
nonlinear phenomena. In this section, we will discuss mainly SPM, XPM and
FWMwhichaffect pulse propagationthroughoptical fibers.
The lowest order nonlinearity present in an optical fiber is the third order
nonlinearity. Thus in an optical fiber, the polarization generated consists of a
linear andanonlinear term:
where and represent the linear and third order susceptibility of the
medium (silica) and represents the electric field of the propagating light
wave/pulse. The nonlinearity due to results in an intensity dependent
refractive indexgivenby
where = 3 and is the refractive index of the medium at low
intensities. It is this intensity dependent refractive index that gives rise to self
phase modulation(SPM).
Due to intensity dependent refractive index, the propagation constant of a
mode becomes intensitydependent andcanbe writtenas
where
represent the nonlinear coefficient and the nonlinear mode effective area
respectively and is the propagationconstant of the mode at lowpowers. If we
assume the mode to be described by a Gaussian function, then where
is the Gaussian mode spot size. Note that the nonlinear coefficient of the
m
b =b+g
b
p
g
2
9 2
(3)
(3)
E
n= n +n I
n n
P
A = w
w
0 2
2
NL
eff 0
0
0
2
10. Nonlinear Fiber Optics &SupercontinuumGeneration
Self phase modulation(SPM)
c c
c
c / e
(3)
4c
0 0
n
......(32)
......(33)
Figure 44 : Measured on-off gain spectrum
of a Raman f i ber ampl i f i er. The
measurements werecarriedout at IITDelhi.
Figure 44 shows a measured on-off
gain spectrum of a backward
pumped Raman fiber amplifier
corresponding to an input pump
power of 750 mW at a wavelength of
1453 nmand an input signal power of
0.14 mW over a 25 km long span of
single mode fiber. On-off gains of
greater than 12 dB are easily
achievable.
Since Raman fiber amplifiers can
operate at any signal wavelength
3 ) 3 (
0 0
E E P c e c e + =
......(31)
0 2
;
eff
k n
A
g =
( )
2
2
4
( )
2
( )
eff
r rdr
A
r rdr
y
p
y
=

......(34)
chirped pulse generated due to SPM.
The output chirped pulse with the
same temporal width has a larger
frequency spectrum. These new
frequencies have been generated by
the nonlinear process.
The chirping due to nonlinearity
without any corresponding increase
in pulse width leads to increased
spectral broadening of the pulse. This
spectral broadening coupled with the
dispersion in the fiber leads to
modified dispersive propagation of
the pulse in the presence of
nonlinearity. Inthe normal dispersion
region the chirping due to dispersion
is to downshift the leading edge and
upshift the trailing edge of the pulse.
This is of the same sign as that due to
SPM. Thus in the normal dispersion
regime (wavelength less than the zero
dispersion wavelength) the chirping
due to dispersion and nonlinearity
add. Thus at high powers, where the
nonlinear effects are not negligible,
the pulse will suffer additional
dispersion as compared to the
dispersion of the same pulse at low
powers. On the other hand, in the
anomal ous di spersi on regi on
(wavelength greater than the zero
dispersion wavelength), the chirping
due to dispersion is opposite to that
due to nonlinearity and thus in this
wavelength region, nonlinearity and
dispersion induced chirpings can
partially or even totally cancel each
other. When total cancellation takes
place, the pulse neither broadens in
time nor in its spectrum and such a
pulse is called a . Such solitons
can hence be used for dispersionless
propagation of pulses to realize very
highbit rate systems.
Assumi ng onl y second order
dispersion and nonlinearity, the
amplitude ( , ) of the electric fieldof
soliton
A z t
c
(3)
fiber depends on the effective area of the mode; the larger the effective mode
area, the smaller are the nonlinear effects. Table II gives the mode effective area
of some commonfiber types.
If represents the attenuation coefficient of the optical fiber, then the power
propagating through the fiber decreases exponentially as ( ) = where
is the input power. In such a case, the phase shift sufferedby an optical beamin
propagatingthroughalength of the optical fiber is givenby
where
is called the effective length of the fiber. If >>1 then ~ 1/ and if <<1
then ~ . For single mode fibers operating at 1550 nm, ~ 0.25 dB/km and
thus ~ for <<20kmand ~20kmfor >>20km.
Since the propagation constant of the mode depends on the power carried
by the mode, the phase of the emergent wave depends on its power and
hence this is referredtoas (SPM).
For a light pulse in Eq. (33) becomes time dependent and this leads to an
additional time dependent phase apart from . Thus the output pulse is
chirpedandthe instantaneous frequencyof the output pulse is givenby
The leading edge of the pulse corresponds to the positive values of / and
the trailing edge to negative values of / . Thus in the presence of SPM, the
leading edge gets downshifted in frequency while the trailing edge gets
upshifted in frequency. The frequency at the center of the pulse remains
unchanged from . Figure 45 shows an input unchirped and the output
a
a a a
a
b
F
w
w
P z P e P
L
L L L
L L
L L L L L
self phase modulation
P
t
dP dt
dP dt
0 0
eff
eff
eff eff
NL
0
0
0
0
0
-az
49
Fiber type Effectivearea ( m ) m
2
Single mode fiber (SMF) G652 85 m
DispersionShiftedFiber (DSF) 46 m
NonZeroDSF(NZDSF) 52 m (D> 0), 56 m (D< 0) and73 m
Dispersioncompensatingfiber (DCF) 23 m (D< 0)
Photonic crystal fiber/holeyfiber 3 m
m
2
2
2 2 2
2
2

m
m m m
m
m
Table II Mode effective area of typical commercially available fibers
eff
L
NL
L P L dz
0
0
g b b + = = F

......(35)
( )
a
a L
eff
e
L
-
-
=
1
......(36)
( )
dt
dP
eff
L
eff
L P t
dt
d
t
0
0 0 0
) ( g w g w w - = - =
......(37)
Figure 45 : An input unchirped pulse gets chirped due to self phase modulation as it
propagates throughanoptical fiber.
frequency is the result of mixing of
two waves at frequency and one
wave at frequency . Four wave
mixing results in the generation of
newfrequencies and is detrimental to
optical fiber communication systems.
At the same time, it can be shown
that FWM can lead to optical
amplification through the nonlinear
interaction. If we consider two closely
lying frequencies and
then FWM leads to the generation
of a new frequency
During this process the frequency
gets amplified. For maximum
amplification of the signal the
wavelengths must be close to the
zero dispersion wavelength of the
f i ber. Fi ber opt i c paramet ri c
amplifiers (FOPA) based on such an
effect are beingstudiedextensively.
Four wave mixing is also used in
wavelength conversion devices
wherein bit streams at a specific
wavelength need to be converted to
the same bit streams at a different
wavelength in optical networks. Thus
input bit streams at get converted
to bit streams at due to the four
wave mixingprocess.
Four wave mixing causes cross talk
among the channels and in order to
avoi d f our wave mi xi ng t he
wavelength of operation should not
be too close to the zero dispersion
wavelength. Thus in wavelength
division multiplexed systems, the
fibers have a finite non-zero value of
dispersion and this leads to much
reduced cross talk due to four
wave mixing in optical fiber
communicationsystems.
Supercontinuum (SC) generation is
the phenomenon in which a nearly
continuous spectrally broadened
output (bandwidth >1000 nm) is
w
w
w
w w w Dw,
w w - Dw.
w
w
w
i
p
s
S
S
i
P S P
i P
= +
=
Supercontinuumgeneration
an incident pulse can be shown to
satisfy the following equation [see
e.g., Gh1]:
where / and = - /v .
Equation (38) is referred to as the
nonlinear Schrdinger equation and
describes the propagation of a pulse
in a mediumpossessing second order
dispersion and nonlinearity. The
solution of the above equation give us
sol i tons whi ch are descri bed
mathematicallyby:
Equation (39) shows that the peak
power required to form a soliton is
related to the pulse width and
dispersion coefficient As an
example, for a soliton pulse at 1550
nm with full width at half
maximumof = 10ps propagating in
a fiber with = 2.4 W km and =
2 ps/km.nm and the required peak
power will be P =33mW.
A heuristic derivation of the power
required to form a soliton by
cancellation of chirping due to
dispersion and nonlinearity can be
foundinRef. Gh1.
Even if the cancellation between
dispersive and nonlinear chirping is
not perfect, the nonlinear effects in an
optical fiber lead to reduced pulse
broadening in the anomalous
dispersion region. Thus the net
dispersion suffered by the pulse
decreases as the power increases. This
fact needs to be kept in mind while
designing dispersion compensation
schemes.
Let us consider an input containing
lightwaves at two frequencies and
; the subscripts p and s refer to
b = b w
c
t
w
w
2 g
P
S
d d T t z
D.
D
2 2
(3)
1 1
f
0
XPMandFWM
50
where c.c. represents complex
conjugate, and represent the
amplitudes of the pump and signal
waves and and are the
propagation constants at the pump
and signal frequencies respectively.
Substituting for the total electric field
(sum of the pump and signal electric
fields) in the nonlinear polarization,
we find that apart from the self phase
modulation terms, there are terms
proportional to the following
expressions:
The first term gives nonlinear
polarization at frequency and this
depends on the intensity of the wave
at frequency . Similarly the second
term gives nonlinear polarization at
frequency and this depends on the
intensity of the wave at frequency .
These two terms are referred to as
cross phase modulation terms; this is
because for example, the intensity of
the pump leads to a change in the
propagationof the wave at .
The last two terms are at new
frequencies:
=2 ......(42)
and
=2 ......(43)
These two terms lead to what is
referred to as four wave mixing
(FWM) since, for example, the
A A
p s
p s
s
p
p
s
s
b b
w
w
w
w
w
w w -w
w w -w
i P S
j S P
0
/2 0
0
2
( , ) sech
i P z
P
A z t P T e
g
g
b
-

=


......(39)
) (
2
6
z
s
t
s
i
s p
e A A
b w -
) ( 2
6
z
p
t
p
i
p s
e A A
b w -
) ) 2 ( ) 2 ({
* 2
3
z
s p
t
s p
i
s p
e A A
b b w w - - -
) ) 2 ( ) 2 ({
* 2
3
z
p s
t
p s
i
p s
e A A
b b w w - - -
2
2
2
2
2
A A
i i A A
z T
b
g

= -

......(38)
pump and signal respectively. Let us
write for the electric fields at the two
frequencies as
( ) c c e A E
z
s
t
s
i
s s
.
2
1
) (
+ =
- b w
......(41)

+ =
-
c c e A E
z
p
t
p
i
p p
.
2
1 ) ( b w
......(40)
profiles have been used in the
literature to achieve broadband SC.
Some studies have used dispersion
decreasi ng f i bers, di spersi on
flattened fibers, while others have
us ed a c ons t ant anomal ous
dispersion fiber followed by a normal
dispersion fiber. Figure 46 shows the
input and output broadened spectra
obtained by passing a light pulse
through a photonic crystal fiber [Ref.
Ra1]. Figure 47 shows a photograph
of the generation of supercontinuum
usingaphotonic crystal fiber.
Ad1. M Adams,
, John Wiley,
Chichester, (1981).
Gh1. A. Ghatak andK. Thyagarajan,
,
Cambri dge Uni versi t y Press,
Cambridge (1998).
Gh2. A. Ghatak, , 4 Edition,
Tata McGraw Hill, New Delhi (2009);
also published by McGraw Hill, New
York(2010).
He1. J. Hecht, , Oxford
(1999)
Ka1. N.S. Kapany,
, Academic
Press, NewYork(1967).
Ma1. D. Marcuse,
, Van Nostrand Reinhold
Company, 1972.
My1. D. K. Mynbaev and L.L.
Scheiner,
, Prentice Hall, USA(2001).
Pa1. B. P. Pal (Ed.),
,
Academic Press, Amsterdam(2006).
Sn1. A. W. Snyder and J. D. Love,
, Chapman&
Hall, London(1983).
Introduction to
Optical Waveguides
I nt roduc t i on t o Fi be r Opt i c s
Optics
City of Light
Fibre Optics:
Principles & Applications
Light Transmission
Optics
Fiber-Optic Communications
Technology
Guided Wave
Optical Components and Devices
Optical Waveguide Theory
th
References & Suggested
Readings
Textbooks inFiber Optics
produced through nonlinear effects on high peak power picosecond and
femtosecond pulses. Such broadened spectra find applications in
spectroscopy, optical coherence tomography, WDM source for optical
communication by slicing the spectrum etc. Supercontinuum generation in an
optical fiber is a very convenient technique since the intensity levels can be
maintained high over long interaction lengths by choosing small mode areas
andthe dispersionprofile of the fiber canbe appropriatelydesignedbyvarying
the transverse refractive index profile of the fiber. The spectral broadening that
takes place in the fiber is attributed to a combination of various third order
effects such as SPM, XPM, FWM, Raman scattering. Since dispersion plays a
significant role in the temporal evolution of the pulse, different dispersion
51
Figure 46 : Supercontinuumgeneration by an optical fiber. The figure shows the input and
output spectra; theoutput spectrumcovers theentirerangeof wavelengths fromabout 400
nmto1600nm. [After Ra1].
Figure 47 : Supercontinuumgeneration by 350 fs laser pulses (at 1060 nmwavelength) as
it passes through a Photonic Crystal Fiber. Notice the color of light changing as the pulses
propagate through the fiber. The output is passed through a prismto disperse the different
wavelengths generatedwithinthe fiber. The fiber was fabricatedat CGCRI, Kolkata andthe
experiment was carriedout byDr. Shyamal BhadraandMr. JohnMcCarthyinProfessor Ajoy
Kar's Laboratory at Heriot Watt University inEdinburgh. Photographcourtesy Dr. Shyamal
Bhadraof CGCRI, Kolkata.
Other References
Au1. J.L. Auguste, R. Jindal, J.M.
Blondy, Clapeau, J. Marcou, B.
Dussardier, G. Monnom, D. B.
Os t r ows ky, B. P. Pal and K.
Thyagarajan , ,
1689, (2000).
Ga1. W. A. Gambling,
, Ninth
W.E.S. Turner Memorial Lecture,
Glass Technology Vol.27 (6), 179,
(1986).
Ga2 Galvanauskas, A., Ming-Yuan
Cheng, Kai-Chung Hou, and Kai-
Hsiu Liao, (2007) High Peak Power
Pulse Amplification in Large-Core
Yb-Doped Fiber Amplifiers, IEEE J.
Sel. topics Quant. Electron., , 559
Go1. I. C. Goyal, R.K. Varshney and
A. K. Ghatak,
, Optical Engineering, , pp
977-980, (2003).
Electron. Lett.
Glass, light, and
the information revolution
Design of a Small
Residual Dispersion Fiber and a
C o r r e s p o n d i n g D i s p e r s i o n
Compensating Fiber for DWDM
Systems
Vol. 36
13
42
52
Je1 Jeong, Y., J. K. Sahu, D. N. Payne,
J. Nilsson, (2004) Ytterbium-doped
large-core fiber laser with 1.36 kW,
continuous-wave output power,
Optics Express, , pp. 6088-6092
Ka1. C. K. Kao and G. A. Hockham,
Dielectric-fibre surface waveguides
for optical frequencies, Proc. IEE,
Vol.113(7), 1151, (1966).
Ka2. F. P. Kapron, D. B. Keck and R.
D. Maurer, Radiation losses in glass
optical waveguides, Appl. Phys. Lett.,
Vol.17, 423(1970).
Ka3: N. S. Kapany, Fiber Optics,
Academic Press, NY(1967)
Li1. M.J. Li, Recent progress in fiber
dispersion compensators, Proc
ECOC 2001, Amsterdam, Paper
ThM1.1, Vol. 4, pp486-489,(2001)
Li2 Limpert J, Fabian Roser, Sandro
Klingebiel, Thomas Schreiber,
Christian Wirth, Thomas Peschel,
Ramona Eberhardt, and Andreas
Tunnermann, (2007) The rising
power of fiber lasers and amplifiers,
IEEE J Selected Topics in Quantum
Electronics, , 537
12
13
Mi 1. T. Miya, Y. Terunama, T.
Hosaka, and T. Miyashita, An
ultimate low loss single mode fiber at
1.55 mm, Electron.Letts. vol. 15, 106
(1979).
Ma2. D. J. H. Maclean,
, JohnWiley, Chichester, 1996.
Ra1 J. K. Ranka, R. S. Windeler, andA.
J . St ent z, Vi si bl e cont i nuum
g e n e r a t i o n i n a i r - s i l i c a
microstructure optical fibers with
anomalous dispersionat 800 nm, Opt.
Lett. , 25--27(2000).
Th1. K. Thyagarajan, R.Varshney, P.
Palai, A. Ghatak and I. C. Goyal,
, Photon. Tech.
Letts., , 1510, (1996).
Up1. Upadhyaya B.N., A. Kuruvilla,
S. Kher, M.R. Shenoy, K. Thyagarajan
and T.P.S. Nathan, Development and
Analysis of a Highly-Efficient 10W
Single-Mode Yb-doped CW Fiber
Laser, Proceedings ICOL 2005,
Dehradun, India.
Optical Line
Systems
A
no ve l de s i gn o f a di s pe r s i o n
compensating fiber
25
Vol. 8
IPA ELECTIONS AND TAKING OVER BY THE NEW EXECUTIVE COMMITTEE
OF IPA
The biennial elections of IPA Office Bearers for the term 2010-2012 were conducted by the returning officers, Prof. R.
Varma, IIT Bombay, Mumbai and Dr. B. N. Jagatap, BARC, Mumbai. The election results were declared on June 3, 2010
andthe followingmembers of IPAexecutive Committee have beendeclaredelected:
Dr. S. Kailas, BARC, Mumbai
Prof. M.K.Sanyal, SINP, Kolkata
Prof. S. S. Major, IITBombay, Mumbai
Dr. D. Kanjilal, IUAC, NewDelhi
Prof. Ashok Mody,
, Mumbai
Prof. Jahan Singh, Shobit University,
Meerut
A joint meeting of the outgoing and incoming executive
committees of IPAtook place on Wednesday, August 4, 2010
at 1430 hrs the in New VIP Hall of the BARC Training School
Hostel, Anushakti nagar, Mumbai - 400 094 during which the
newexecutive committee tookover the charge.
President -
Vice- President-
General secretary-
Joint Secretary-
Treasurer -
Member (North Zone) -
Dixit Road BMC Junior
College of Science
Outgoing President Prof. Bikash Sinha (centre)
with (L to R) Prof. S. S. Major, Shri A.M. Nakra,
Prof. Ashok Mody and Dr. S. Kailas