Course 7

Circular fibers
PHS8205
Maksim Skorobogatiy
 Circular fibers
It can be demonstrated from a general symmetry consideration that due to rotational symmetry around the fiber axis, harmonic
solution for the modal fields of a circularly symmetric fiber can be presented in the cylindrical coordinate system in the form:

Fβ (ω ),m ( ρ , θ , z , t ) Fβ (ω ),m ( ρ ) exp(imθ ) exp(i ( β z − ωt ))

F( ρ ) = (E( ρ ), H ( ρ )) is a six-component electromagnetic field vector which is only a function of distance ρ.

ρ θ
O

Schematic of a simple step-index fiber and definition of the cylindrical

coordinate system with the origin coinciding with the fiber center.
Fields in a circular fiber

ρˆ ⋅ Eρ ( ρ )
( ρ ,θ )
ρ
Typically, there is also a Z
θ component of the fields,
θˆ ⋅ Eθ ( ρ )
O which is significantly smaller
than the transverse fields in
the limit of large core sizes.
zˆ ⋅ Ez ( ρ )
 VIDEOS
• Fiber fabrication (9:23)
• How it’s made fiber optics (4:46)
 Helmholtz equation,
case of the
infinite uniform dielectric
PHS8205
Maksim Skorobogatiy
 Helmholtz equation,
infinite uniform dielectric
Harmonic solutions of Maxwell’s equations in the case of an infinite uniform dielectric with dielectric constant ε are described by Helmholtz
equation. To derive this equation, we first rewrite Maxwell’s equations in terms of the longitudinal and transverse fields:

r, t ) ( Et (r ), E z (r )) exp( −iωt ), H(=

E(= r, t ) ( Ht (r ), H z (r )) exp( −iωt )
Maxwell equation for harmonic fields can then be written as:

z� × 1 ∂B 1 ∂D
∇ × E= − ∇ × H=
𝐳𝐳� ⋅ c ∂t c ∂t

∂Et (r ) ω ∂H t (r ) ω
− ∇t Ez (r ) = −i ( zˆ × H t (r ) ) ; − ∇t H z (r ) = iε ( zˆ × Et (r ) )
∂z c ∂z c
ω ω
zˆ ∇t × Et (r ) =i H z (r ) ; zˆ ∇t × H t (r ) = − iε Ez (r )
c c
 Helmholtz equation,
infinite uniform dielectric
Considering, furthermore, that the waves guided in the OZ direction have exp(iβz) dependence :

E(r, t ) ( Et ,β (rt ), E z ,β (rt )) exp(i β z − i=

= ωt ), H(r, t ) (Ht ,β (rt ), H z ,β (rt )) exp(i β z − iωt )
Then we can rewrite the first two equations from the previous slide as:

∂Et (r ) ∂H t (r )
− ∇t Ez (r ) = −ik0 ( zˆ × H t (r ) ) ; − ∇t H z (r ) = iε k0 ( zˆ × Et (r ) )
∂z ∂z

i β Et , β (rt ) − ∇t Ez , β (rt ) = −ik0 ( zˆ × H t , β (rt ) ) ; i β H t , β (rt ) − ∇t H z , β (rt ) = iε k0 ( zˆ × Et , β (rt ) )

Which can be further solved for the transverse fields in terms of the longitudinal fields:
β k0 β k0
Et , β (r=
t) i 2 ∇t Ez , β (rt ) − i 2 zˆ × ∇t H z , β (rt ) ; H t , β (r=
t) i 2 ∇t H z , β (rt ) + iε 2 zˆ × ∇t Ez , β (rt )
kt kt kt kt
 Helmholtz equation,
infinite uniform dielectric
Finally, the substitution of the transverse fields (in blue) into the Maxwell equations (in red) results in two Helmholtz equations for
the longitudinal (directed along the OZ axis) electric and magnetic field components:
β k0 β k0
Et , β (r=
t)
ˆ
i 2 ∇t Ez , β (rt ) − i 2 z × ∇t H z , β (rt ) ; H t , β (r=
t) i 2 ∇t H z , β (rt ) + iε 2 zˆ × ∇t Ez , β (rt )
kt kt kt kt

zˆ ∇t × Et (r ) =ik0 H z (r ) ; zˆ ∇t × H t (r ) = − iε k0 Ez (r )

∇t2 H z , β (rt ) + kt2 H z , β (rt ) =

0 ; ∇t2 Ez , β (rt ) + kt2 Ez , β (rt ) =
0
∂2 ∂2
k0= ω c ; k = ε k − β ; ∇=
t
2 2
0 2
2
+ 2 2
t
∂x ∂y
Waves of circular symmetry
(Bessel beams)
propagating in the infinite
uniform dielectric
PHS8205
Maksim Skorobogatiy
 Helmholtz equation in cylindrical coordinates
We now use a representation of the Laplace operator in the cylindrical coordinate system to find circular symmetric solutions of
the Helmholtz equation. The form of the Laplace operator operating on a scalar function f is given by:
2 2 2
∂ f ∂ f 1 ∂  ∂f  1 ∂ f
∇t 2 f ( ρ ,θ , z ) = 2 + 2 = ρ +
∂x ∂y ρ ∂ρ  ∂ρ  ρ 2 ∂ 2θ
Assuming that longitudinal electromagnetic field components have the form:

ezβ ,m ( ρ ) ⋅ exp(imθ ) ; H z , β (rt ) =

Ez , β (rt ) = hzβ ,m ( ρ ) ⋅ exp(imθ )

∇t2 Ez , β (rt ) + kt2 Ez , β (rt ) =

0 ; ∇t2 H z , β (rt ) + kt2 H z , β (rt ) =
0

k0 = ω c
1 ∂  ∂f ( ρ )   2 m 2  2
ε 2
β 2

ρ  +  kt − 2  f ( ρ ) =
k
0 t
= k 0 −
ρ ∂ρ  ∂ρ   ρ  f ( ρ ) = ezβ ,m ( ρ ) or hzβ ,m ( ρ )
 Solutions of Helmholtz equation
in cylindrical coordinates
k0 = ω c
1 ∂  ∂f ( ρ )   2 m  2

ρ  +  kt − 2  f ( ρ ) =
0 kt2 ε k02 − β 2
=
ρ ∂ρ  ∂ρ   ρ  f ( ρ ) = ezβ ,m ( ρ ) or hzβ ,m ( ρ )
Solutions of this equation are generally known as cylindrical Bessel functions (linear combinations of Bessel functions can have
their proper names, such as Hankel functions, etc.). For a differential equation of the second order, there exist two linearly
independent solutions. For the problem of propagation of guided waves, the two linearly independent solutions can be chosen in
the form of a Jm Bessel function and in the form of a Hankel function of the first kind Hm(1). The choice of these functions is
motivated by their specific asymptotic behaviors at infinity and in the vicinity of the center of the coordinate system. Namely:
Analytical at O Outgoing wave Incoming wave
( kt ρ )
m
2  π 1
J m ( kt ρ )  m
ρ →0 2 m !
; J m ( kt ρ ) 
π kt ρ
− ( 2m + 1) 
cos  kt ρ=
4 2π kt ρ
( exp ( ikt ρ + ...) + exp ( −ikt ρ + ...) )
ρ →∞

Divergent at O
 2m (m − 1)! Outgoing wave only
( )
−m
 −i k ρ ;m>0
π t
2  π
H m ( kt ρ )  
(1)
; m ( kt ρ ) 
H (1) exp  ikt ρ − i ( 2m + 1) 
ρ →0 2
 log(k ρ ) ; m = 0
ρ →∞ π kt ρ  4
 π t
 Solutions of Helmholtz
= E(r, t ) (E t ,β (rt ), Ez , β (rt )) exp(i β z − iωt )
= H (r, t ) (H (rt ), H z , β (rt )) exp(i β z − iωt )
equation in cylindrical t ,β

coordinates E (r ) = e z ,β t
β ,m
z ( ρ ) exp(imθ )
H z , β (rt ) = hzβ ,m ( ρ ) exp(imθ )
The general form of a solution of
Maxwell equations inside of a uniform
ezβ ,m ( ρ ) = AE ⋅ J m (kt ρ ) + BE ⋅ H m(1) (kt ρ )
dielectric region that respects circular
symmetry can be written in the hzβ ,m ( ρ ) = AH ⋅ J m (kt ρ ) + BH ⋅ H m(1) (kt ρ )
following form:

β k0
Et , β (r=
t) i 2
∇t Ez , β (rt ) − i 2
zˆ × ∇t H z , β (rt )
kt kt
( E(r, t ), H(r, t ) )
β k0
H t , β (r=
t) i 2
∇t H z , β (rt ) + iε 2
zˆ × ∇t Ez , β (rt )
k0 = ω c
kt kt
kt2 ε k02 − β 2
=
Et , β (rt ) = etβ ,m ( ρ ) exp(imθ )
H t , β (rt ) = htβ ,m ( ρ ) exp(imθ )
 Solutions of Helmholtz equation
in cylindrical coordinates
Finally, using a representation of the longitudinal
β ,m β ,m
E=
β , m (r , t ) (et ( ρ ), ez ( ρ )) ⋅ exp(imθ ) ⋅ exp(i β z − iωt ) fields in terms of the various Bessel functions, one can
compute analytically all the components of the
H=
β , m (r , t ) (htβ ,m ( ρ ), hzβ ,m ( ρ )) ⋅ exp(imθ ) ⋅ exp(i β z − iωt ) electromagnetic fields that satisfy the Maxwell
equation and respect circular symmetry:

 β ′   β ′(1)   k0 m   k0 m (1) 
i
 k m t J ( k ρ )   i k H m ( kt ρ )   − k 2 ρ J m ( kt ρ )   − k 2 ρ H m ( kt ρ ) 
 t   t   t   t 
β ,m
 eρ ( ρ )   β m   β m (1)   k0   k0 (1) 
 β ,m  −
 k2 ρ m t  J ( k ρ ) −
 k2 ρ m t  H ( k ρ )  −i J m
′ ( k t ρ )   −i H m
′ ( k t ρ ) 
 eθ ( ρ )   t   t   kt   kt 
 ez ( ρ ) 
β , m  J m ( kt ρ )   H m ( kt ρ )
(1)
  0   0 
 β ,m  = AE ⋅   + BE ⋅   + AH ⋅  + BH ⋅ 
h
 ρ ( ρ )  k m
 ε 0 J (k ρ )  k m
 ε 0 H (k ρ )  (1) β
 i J ′ (k ρ )  β
 i H ′ (k ρ ) 
(1)
 h β ,m ( ρ )   kt ρ 2 m t
  kt2 ρ m t   kt m t   kt m t 
 θβ ,m         
 h (ρ )  k k β m − β m
 z   iε 0
J m′ (kt ρ )   iε 0
H m′ (kt ρ ) 
(1) − J m ( kt ρ )  H m(1) (kt ρ ) 
 kt   kt   kt ρ 2   kt ρ2 
       
 0   0   J m ( k t ρ )   H (1)
m ( k t ρ ) 
 Circular fibers
Boundary conditions
PHS8205
Maksim Skorobogatiy
 Radiation boundary conditions
Outside the fiber, there are
only outgoing fields because a
passive waveguide can only
lose energy. Therefore,
outside the fiber only Hm(1)
At the origin O, modal
solutions are possible.
fields have to be finite.
Hence, in the fiber
region containing the
origin, only Jm
solutions are possible.

In the intermediate
Analytical at O Outgoing wave only
layers, both the
( kt ρ )
m
outgoing Hm(1) and J m ( kt ρ )  m 2  π
the incoming Jm ρ →0 2 m ! m ( kt ρ ) 
H (1) exp  ikt ρ − i ( 2m + 1) 
ρ →∞ π kt ρ  4
waves can co-exist
 Fields in the cladding, 
 i
β
ktcl
J ′
m ( k t
cl
ρ )



 i
β
ktcl
H ′
m
(1)
( k t
cl
ρ )



radiating  eρclad ( ρ ) 


 −
β m
J ( k cl
ρ )





 −
β m (1) cl 
H m ( kt ρ ) 


 ( kt )  ( kt )
m t
 clad  cl 2 ρ cl 2 ρ

boundary condition  eθ ( ρ ) 
 ezclad ( ρ )  cl 

J m (ktcl ρ )


 + B ⋅ cl

H m
(1)
( k t
cl
ρ )


+
 clad = AE ⋅
Outside the fiber, there are  hρ ( ρ )   k0 m  E
 k0 m (1) cl 
ε J m ( kt ρ )  ε H m ( kt ρ ) 
cl cl cl
only outgoing fields because  h clad ( ρ ) 
 ( kt )  ( kt )
cl 2 ρ cl 2 ρ
 θclad   
a passive waveguide can  h (ρ )     
 z 0 k
 iε cl 0 J ′ (k cl ρ )  k
 iε cl 0 H ′(1) (k cl ρ ) 
only lose energy. Therefore,
 ktcl m t
  ktcl
m t

outside a waveguide, only    
Hm(1) solutions are possible.  0   0 
 k0 m   k0 m (1) cl 
 − J ( k cl
ρ )   − H m ( kt ρ ) 
 ( kt )  ( kt )
2 m t
cl ρ cl 2 ρ
 
   
k
 −i 0 J m′ (kt ρ )  cl k
 −i 0 H m′ (kt ρ )  (1) cl
cl
O  kt   ktcl 
   
0 0
+ AHcl ⋅   + B cl ⋅  
 β  H
 β 
 i cl
J ′
m ( k t
cl
ρ )   i cl
H ′
m
(1)
( k t
cl
ρ ) 
 k   k 
εcr
t t
0  β m   β m (1) cl 
 − cl 2 J m (kt ρ )   − cl 2 H m (kt ρ ) 
cl

 ( kt ) ρ  ( kt ) ρ
εcl 
 J m ( k t
cl
ρ ) 

 
 H (1)
m ( k t
cl
ρ ) 


  β   β 
Fields in the core, ′ ρ ′ ρ
(1)
 i J m ( k t
cr
)   i H m ( k t
cr
) 
ktcr ktcr
   
 β m   β m (1) cr 
no divergence condition  eρcore ( ρ )  ( ρ ) H m ( kt ρ ) 
cr
 − J k   −
 ( kt )  ( kt )
m t
 core  cr 2 ρ cr 2 ρ
 eθ ( ρ )   
 ezcore ( ρ )     
cr 
J m (ktcr ρ )  + B ⋅ cr
H (1)
m ( k t
cr
ρ ) +
 core = AE ⋅
 hρ ( ρ )   k0 m  E
 k0 m (1) cr 
ε J m ( kt ρ )  ε H m ( kt ρ ) 
cr cr cr
 h core ( ρ ) 
 ( kt )  ( kt )
cr 2 ρ cr 2 ρ
 θcore   
 h (ρ )   0  
At the origin O, modal  z  k
 iε cr 0 J ′ (k cr ρ )  k
 iε cr 0 H ′(1) (k cr ρ ) 
fields have to be finite.  kt cl m t
  ktcr
m t

Hence, in the fiber    
 0   0 
region containing the
 k0 m   k0 m (1) cr 
origin, only Jm  − J ( k cr
ρ )   − H m ( kt ρ ) 
 ( kt )  ( kt )
2 m t
cr ρ cr 2 ρ
solutions are possible.  
   
k
 −i 0 J m′ (kt ρ )  cr k
 −i 0 H m′ (kt ρ )  (1) cr

O  kt cr
  ktcr 
   
0 0
+ AHcr ⋅   + B cr ⋅  
 β  H
 β 
 i cr
J ′
m ( k t
cr
ρ )   i cr
H ′
m
(1)
( k t
cr
ρ ) 
 k   k 
εcr
t t
 β m 0  β m (1) cr 
− J ( k cr
ρ )   − H m ( kt ρ ) 
 ( kt )  ( kt )
m t

εcl
cr 2 ρ cr 2 ρ
 
   
 J m ( k t
cr
ρ )   H (1)
m ( k t
cr
ρ ) 
Field continuity at the interfaces
 eρclad ( ρ ) 
 clad 
e
 θ ( ρ ) 
 ezclad ( ρ ) 
 clad 
 hρ ( ρ ) 
 h clad ( ρ ) 
 θclad
 h (ρ ) 
 E&M field continuity conditions
 z 
 eρcore ( ρ )   ε clad eρclad (a )   ε core eρcore (a ) 
a  core
( ρ )
  clad   core 
 eθ (a )   eθ (a ) 
e
 θ 
 ezcore ( ρ ) 
 core   ezclad (a )   ezcore (a ) 
 hρ ( ρ )   clad  =  core 
 core
O  hθ ( ρ ) 
  hρ (a )   hρ (a) 
 h core ( ρ ) 
 z   hclad (a )   h core (a ) 
 θclad   θcore 
 h (a)   h (a) 
 z   z 
εcr
εcl
 E&M field continuity conditions
Field continuity at the interfaces for components  to the interface
 eθclad (a )   eθcore (a ) 
 clad   core 
 ez (a )  =  ez (a ) 
 β m (1) cl   k0   β m   k0 
 −i k cl H m′ (kt a )   −i k cr J m′ (kt a ) 
(1) cl cr
 − cl 2 a H m (kt a )   − cr 2 a J m (kt a ) 
cr

 ( kt )   t   ( kt )   t 

BEcl ⋅ 
(1)
H m ( kt a )
cl  
 + BHcl ⋅ 
0  
 = AEcr ⋅  J m ( kt a )
cr  
 + AHcr ⋅ 
0 

 hθclad (a )   hθcore (a ) 
 cl k0
 iε cl H m′ (kt a ) 
  −
β m (1)
H (k a) cl  
 iε
cr k0
′
  −
β m
J m ( kt a ) 
cr
 clad   core 
 hz (a)   hz (a) 
(1) cl
 ( k cl )2 a m t  J ( k cr
a )   ( kt ) a 
cr 2
cl m t
 k t   t
  k t   
 0   H (1)
( k cl
a )   0   J ( k cr
a ) 
   m t     m t 

Eigenvector
 β m (1) cl k β m k 
 − cl 2 H m ( kt a ) −i cl0 H m′(1) (ktcl a ) J m (ktcr a ) i cr0 J m′ (ktcr a ) 
 ( kt ) (k )
a kt cr 2 a kt
t   BEcl 
   cl 
 H m(1) (ktcl a ) 0 − J m (ktcr a ) 0   BH  = 0
 cl k0 β m (1) cl k β m   AEcr 
 iε cl H m′ (kt a ) −iε cr cl0 J m′ (ktcr a )
(1) cl
− H m ( kt a ) J ( k cr
a )   cr 
(k ) ( ktcr ) a
m t
kt cl 2 a kt 2
 t   AH 
 
 0 H m(1) (ktcl a ) 0 − J m (ktcr a ) 
k0 = ω c

( k=
)t
cl 2
ε cl k02 − β 2 det(M(β))=0

( k=)
t
cr 2
ε cr k02 − β 2 Nonlinear eigen value problem
 Circular Fibers
Transfer Matrix Theory
PHS8205
Maksim Skorobogatiy
 Tangential field components
inside of a region n: ρn-1<ρ<ρn
 β m   β m (1) n   k0   k0 
 − J ( k n
ρ )   − H m ( kt ρ )   − i J m′ (ktn ρ )   −i H m′(1) (ktn ρ ) 
 ( kt )  ( kt )
m t
n 2 ρ n 2 ρ ktn
ktn
 eθ ( ρ ) 
n
     
 n       0   0 
 ez ( ρ )  = An ⋅  J m (ktn ρ ) ρ
(1)
H ( k n
)
 + B n
⋅  m t  + A n
⋅  + BHn ⋅ 
 hθn ( ρ )   − β m J (k n ρ )  β m
E E H
 n k0   n k0  − H m ( kt ρ ) 
(1) n
 n   iε J ′ ( k n
ρ )   iε H ′ (1)
( k n
ρ )   ( k n )2 ρ m t   (k ) ρ
n 2

 hz ( ρ ) 
n m t n m t
 k t   k t   t
  t

 0   0   J m ( kt ρ )
n   H m ( kt ρ )
(1) n 
       
ρn
 eθn ( ρ )  . . . .   AEn   AEn 
 eθn ( ρ )   n       n
 n
e ( ρ )
  ez ( ρ= ) . . . .   BEn  
⋅ n= M n ( ρ , β ) ⋅ n
BE 
 z 
ρn-1  hθ ( ρ ) 
n  hθn ( ρ )  . . . .   AH   AH 
 n   n     n   n 
 hz ( ρ )   hz ( ρ )  . . . .   BH   BH 
O
ρ εn
 Field continuity at the circular interface n
 eθn +1 ( ρ n )   AEn +1   AEn   eθn ( ρ n ) 
 eθn +1 ( ρ n )   n +1   n 
 n +1   n
 ez ( ρ n )   ez ( ρ n ) 
 n +1  BE  BE 
e ( ρ )  
n +1 ( ρ n , β ) ⋅ = M n ( ρn , β ) ⋅ n
 z n 
M
=
 hθn +1 ( ρ n ) 
 n +1 
 hθn +1 ( ρ n )   AH 
n +1  AH   hθn ( ρ n ) 
h
 z ( ρ )
n   n +1   n +1   n   n 
 eθn ( ρ n )   hz ( ρ n )   BH   BH   hz ( ρ n ) 
 n 
e (
 z n  ρ )
 hθn ( ρ n ) 
 n   AEn +1   AEn   AEn 
 hz ( ρ n )   n +1   n  n
O ε ε n n+1  BE 
=
 AH 
n +1
M n +1 ( ρ n , β ) M=
−1
n ( ρn , β ) ⋅
 BE 
 AH 
n

Tn ,n +1 ( β ) n
BE 
 AH 
 n +1   n   n 
 BH  𝑇𝑇𝑛𝑛,𝑛𝑛+1 𝛽𝛽  BH   BH 

ρn 4x4 transfer matrix that relates field expansion coefficients in regions n and n+1
Relation between field expansion coefficients in the
first and the last layers
 AEN   AEN −1   AEN − 2 
 N  N −1   N −2 
= ε
N  BE 
 AH 
N N −1,N ( β )
T=  BE 
 AH 
N −1
TN −1,N ( β ) TN −=
2,N −1 ( β )
 BE 
 AHN − 2 
 N   N −1   N − 2 
 BH   BH   BH 
 AE1   AE1 
 1  1
O ε1 N −1,N ( β ) TN − 2,N −1 ( β ) ...T1,2 ( β )
T=  BE  
Ttot ( β ) 1
BE 
 AH 
1  AH 
 1   1 
𝑇𝑇𝑡𝑡𝑡𝑡𝑡𝑡 𝛽𝛽  BH   BH 

4x4 total transfer matrix that relates field expansion coefficients in the first and last regions
 Relation between field expansion coefficients in the
first and the last layers
Outside the fiber, there are

εN only outgoing fields because

a passive waveguide can  0   AE 
1
only lose energy. Therefore,
 N  
 BE  = T ( β )  0 
outside of a waveguide, only
Hm(1) solutions are possible.
Hence AEN=AHN=0
 0  tot
 AH1 
 N  
ε1  BH   0 
At the origin O, modal
fields have to be finite.
Hence, in the fiber
region containing the Bx Ax
origin, only Jm solutions Outgoing wave only Analytical at O
are possible. Therefore 2  π ( kt ρ )
m

BE1=BH1=0 H (1)
m ( k t ρ ) 
ρ →∞ π k ρ
exp ik
 t ρ − i ( 2 m + 1)  J m ( kt ρ )  m
t  4  ρ →0 2 m !
 Finding propagation constant β of a fiber mode
 0   AE1   Ttot1,1 ( β ) AE1 + Ttot1,3 ( β ) AH1 
 N    2,1 1 
εN ( β ) + Ttot ( β ) AH 
1 2,3
 BE   0  T
 tot A
tot ( β )
E
= T=
 0   AH1   Ttot3,1 ( β ) AE1 + Ttot3,3 ( β ) AH1 
 N    4,1 
1 
 BH   0  T
 tot ( β ) AE
1
+ Ttot ( β ) AH 
4,3

 Ttot1,1 ( β ) Ttot1,3 ( β )   AE1   Ttot1,1 ( β ) Ttot1,3 ( β ) 

ε1  3,1  ⋅  1=

 Ttot ( β ) Ttot ( β )   AH 
3,3  0 ⇒ det  3,1
 Ttot ( β ) Ttot ( β ) 
3,3 =0


 BEN   Ttot2,1 ( β ) Ttot2,3 ( β )   AE1 

 N  =  4,1  ⋅  1 
 BH   Ttot ( β ) Ttot ( β )   AH 
4,3

Nonlinear (with respect to b) eigenvalue problem Corresponding eigenvector

 Course 7/8
Core/clad fibers
(step-index fibers)
PHS8205
Maksim Skorobogatiy
 The fundamental mode of a step-index fiber
Using the continuity of the E&M components parallel to the core/clad interface, we get the following characteristic equation for
the propagation constant of a step-index fiber det(M(β))=0:

ε   
 core J m′ (ktcore a )
−
H m′(1) (ktclad a )
⋅
J m′ (ktcore a )
−
H m′(1) (ktclad a )
−m
 ε clad ( ktcore a ) J m (ktcore a ) ( ktclad a ) H m(1) (ktclad a )   ( ktcore a ) J m (ktcore a ) ( ktclad a ) H m(1) (ktclad a ) 
   
2
ε eff
ε clad (( k t
core
a) − (k
−2
t
clad
a) )=
−2 2
0

This equation can be solved analytically in many asymptotic cases, including the long and short wavelength limits, as well as in
the vicinity of the modal cut-off frequencies.

k0 = ω c

( k=)
t
clad 2
ε clad k02 − β 2

( k=)
t
corr 2
ε core k02 − β 2
 Fundamental
The long wavelength limit:
mode of a step index fiber
k0 a ω a c → 0 In the long wavelength limit ω->0, the characteristic equation can be expanded with
respect to the small parameters ξ and δ:

Definitions: ε J m′ (ktcore a ) H m′(1) (ktclad a ) 

 core
− ⋅
∆ε= ε core − ε clad ; δε= ε eff − ε clad  ε clad ( ktcore a ) J m (ktcore a ) ( ktclad a ) H m(1) (ktclad a ) 
 

ω0 (
c a ∆ε ) 
 core
J m′ (ktcore a )
−
H m′(1) (ktclad a ) 

 ( kt a ) J m (ktcore a ) ( ktclad a ) H m(1) (ktclad a ) 
−m 2
ε eff
ε clad (( k t
core
a) − (k
−2
t
clad
a) )=
−2 2
0
 

Principal nondimentional parameters:

( −1)
n 2n+m m m+2
ξ k0 a ∆=
ε ω ω0 → 0 x 1 x 1 x
= ∑ n=0 n !(n + m)!  2  + O ( x m+4 )
∞
J m ( x) = =   −  
x →0 m !  2  ( m + 1) !  2 
δ = δε eff ∆ε → 0
m −1 ( m − n − 1) !  x 
2n−m
 2i  x  i
H m(1) (=
x) J m ( x) 1 +  ln + γ   − ∑ n =0   +
x →0
 π  2  π n! 2
(β k0 ) = ε eff= ε clad + δε eff= ε clad + δ ⋅ ∆ε
2
n −1  1 1 1 1 
( −1) 1 + + ... +  + 1 + + ... + 
n + m    x 
2n+m
i  2 n  2
∑
∞
aktcore k0 a ε core − ε eff = ξ 1 − δ π n =0 n !(n + m)!
 
2
ak=
t
clad
k0 a ε clad −=
ε eff ξ i δ
 Dispersion relation of a step-index fiber in the
long wavelength limit
∆ε  A 
+ B + (ω ω0 ) C + O((ω ω0 ) ) 
2 4
δε eff = ε eff − ε clad = exp  −
( ω ω0 )  
 ( ω ω0 )
2 2

δε eff
c  ε  neff = ε clad + δε eff ≈ nclad +
ω
= 0 ;=A 2 1 + core  2nclad
a ∆ε  ε clad 
B=A 8 + 2 ( ln 2 − γ ) ; C =A 192
γ = 0.577215665...

Dispersion relation of the fundamental m=1 mode of a step-

index fiber with ncore =1.5, nclad=1, a=43 mm. The solid curve
is calculated using exact transfer matrix theory, while the
dashed curve is computed using analytical approximation.
Inset: the ratio between the values of the modal refractive
index using analytical and exact results.
Dispersion of the fundamental HE11 mode
1 ( neff ω )  A 
2 7
∂ 2 A2 ∆ε  ω0 
D= ≈   exp  − 

( )
2
c ∂ω 2 cω0 nclad ω   ω ω 0 

∆ε 1 exp ( A / 8 )
max(
= D(ω )) D(ωD max ) ≈ 2.55
nclad cω0 A A
c  ε core 
ωD max ≈ 0.469 ⋅ Aω0 ; ω0 = ;A=2 1 + 
a ∆ε  ε clad 

Group velocity dispersion of the fundamental m=1 mode, with ncore

=1.1, nclad=1, a=105 mm. Fiber is designed to operate near the
characteristic frequency of 1THz. The solid curve is calculated using
exact transfer matrix theory, while the dashed curve is computed
using analytical approximation.
Losses of the fundamental HE11 mode

 
2 A r

( 4 2
)
α ≈ α clad + α core 2r + α clad ( r − 2r − 1) exp  − 
(ω ω0r )  (ω ω0r ) 
4 2

c  ε core r

ω0 = ; A =
r r
2 1 + r  ; r = r
ncore r
nclad
a ∆ε r  ε clad 

Losses of the fundamental m=1 mode, with ncore =1.5, nclad=1, a=43
mm. Fiber is designed to operate near the characteristic frequency of
1 THz. The solid curve is calculated using exact transfer matrix
theory, while the dashed curve is computed using analytical
approximation. Here, we assume that losses of the fiber core (100
dB/m) are smaller than those of the fiber cladding (200 dB/m). As a
result, modal losses at lower frequencies are more pronounced than
those at higher frequencies.
 Higher order modes and their dispersion
relations near cutoff frequencies
ε J m′ (ktcore a ) H m′(1) (ktclad a ) 
 core
− ⋅
 ε clad ( ktcore a ) J m (ktcore a ) ( ktclad a ) H m(1) (ktclad a ) 
 
 
 core
J m′ (ktcore a )
−
H m′(1) (ktclad a )

 ( kt a ) J m (ktcore a ) ( ktclad a ) H m(1) (ktclad a ) 
 
−m 2
ε eff
ε clad (( k t
core
a) − (k
−2
t
clad
a) )=
−2 2
0

At the modal cutoff frequency ωcutoff : ξ c = ωcutoff ω0 ; η = ε core ε clad


= m 0= (TE and TM ) : J 0 (ξ c ) 0
m = 1 ( HE ) : no cutoff, mode exists at any ω
ε eff= ε clad ⇒ δ=
c δε eff ∆ε= 0  11

β = k0 nclad = m 1 ( HE1 n +1 and EH1n ), n >= 1: J1 (ξ c ) 0


 m ≥ 2 ( EH ) : J m (ξ c ) = 0
ak= t
clad
ak0 ε clad −=
ε eff 0
 ξ c J m (ξ c )
aktcore

= ak0 ε core − ε eff= ak0 ∆ε= ωcutoff ω0 m ≥ 2 ( HE ) : J m−1 (ξc ) = (1 + η ) ( m − 1)

 Higher order modes and their
Modes of a circular step-index fiber. Fiber under study is a rod-in-the-air step-index fiber
with ncore =1.5, nclad=1, a=150 mm. Dispersion relations of the fiber modes with angular

dispersion relations near cutoff

frequencies
ξ c = ωcutoff ω0 ; η = ε core ε clad

= m 0= (TE and TM ) : J 0 (ξ c ) 0
m = 1 ( HE ) : no cutoff, mode exists at any ω
momenta m=0,1,2,3.

 11

= m 1 ( HE1 n +1 and EH1n ), n >= 1: J1 (ξ c ) 0


m ≥ 2 ( EH ) : J m (ξ c ) = 0
 ξ c J m (ξ c )
m ≥ 2 ( HE ) : =(1 + η ) ( m − 1)
 J m −1 (ξ c )

EHmn modes have higher refractive indices than HEm n+1 modes, while
their dispersion relations are close to each other. EH modes have a
dominant electric field component Eθ parallel to the interface
(continuous through interface), while HE modes have a dominant field
component Eρ perpendicular to the interface (discontinuous through
interface). Hence, HE modal fields are more present in the fiber
cladding than EH fields, and consequently, HE modes have lower
effective RIs.
 Field distribution in m=0 modes
In the core
 β ′   k0 m 
BE= BH= 0  k i J m ( k t ρ )   − k 2 ρ J m ( kt ρ ) 
 t   t 
 eρβ ,m ( ρ )   β m   k0 
𝜷𝜷,𝟎𝟎
𝒆𝒆𝜽𝜽  β ,m  −
 k2 ρ m t  J ( k ρ )  −i J m
′ ( k t ρ )  𝜷𝜷,𝟎𝟎
𝒉𝒉𝜽𝜽
𝜷𝜷,𝟎𝟎  e ( ρ )   t   k  𝜷𝜷,𝟎𝟎
𝒉𝒉𝝆𝝆 θ t
𝒆𝒆𝝆𝝆
 ezβ ,m ( ρ )   J m ( kt ρ )   0 
 β ,m  = AE ⋅   + AH ⋅ 
h
 ρ ( ρ )  k m
 ε 0 J (k ρ )  β
 i J ′ (k ρ ) 
 h (ρ ) 
β , m  kt2 ρ m t   kt m t 
 θβ ,m     
 h (ρ )  k  − β m J (k ρ ) 
 z core  iε 0 J ′ (k ρ ) 
 kt
m t
  kt2 ρ m t 
   
 0   J m ( k t ρ ) 
TE0n TM0n

TE TM
 eρβ ,0 ( ρ )   0   eρβ ,0 ( ρ )   −i β kt ⋅ J1 (kt ρ ) 
 β ,0     β ,0   
 eθ ( ρ )   i k 0 k t ⋅ J 1 ( k t ρ )   eθ ( ρ )   0 
 ezβ ,0 ( ρ )   0   ez ( ρ ) 
β , 0  J 0 ( kt ρ ) 
 β ,0  = AH ⋅   β ,0  = AE ⋅  
h
 ρ ( ρ )   − i β k t ⋅ J 1 ( k t ρ )  h
 ρ ( ρ )   0 
 h (ρ ) 
β ,0  0   h β ,0 ( ρ )   −iε k k ⋅ J (k ρ ) 
 θβ ,0     β ,0
θ
  0 t 1 t

 h (ρ )  J ( k ρ )  h (ρ )   0
 z core  0 t   z core  
 Field distribution in m≠0 modes
EH HE

𝜷𝜷,𝟏𝟏 𝜷𝜷,𝟏𝟏
𝜷𝜷,𝟏𝟏 𝒆𝒆𝜽𝜽 𝜷𝜷,𝟏𝟏 𝒉𝒉𝜽𝜽
𝒉𝒉𝝆𝝆 𝒆𝒆𝝆𝝆
In the core
 β ′   k0 m 
BE= BH= 0 i
 k m t J ( k ρ )  −
 k2 ρ m t  J ( k ρ )
 t   t 
 eρβ ,m ( ρ )   β m   k0 
 β ,m   − k 2 ρ J m ( kt ρ )   −i J m′ (kt ρ ) 
kt
e
 θ ( ρ )   t   
 ez ( ρ ) 
β , m  J m ( kt ρ )   0 
+  β ,m  = AE ⋅ 
k m
 + AH ⋅
β
 +
 hρ ( ρ )   ε 0 J (k ρ )   i J ′ (k ρ ) 
 h β ,m ( ρ )   kt2 ρ m t   kt m t 
 θβ ,m     
 h (ρ )  k − β m
 z core  iε 0 J ′ (k ρ )  J m ( kt ρ ) 
𝜷𝜷,𝟏𝟏
𝒉𝒉𝜽𝜽  kt
m t
  kt ρ2  𝜷𝜷,𝟏𝟏
𝒆𝒆𝜽𝜽
𝜷𝜷,𝟏𝟏
𝒆𝒆𝝆𝝆     𝜷𝜷,𝟏𝟏
𝒉𝒉𝝆𝝆
 0   J m ( k t ρ ) 

TE – like mode TM – like mode

 ρˆ ⋅ Eρ ( ρ )

Linear polarized modes ρ

θ
( ρ ,θ )

O θˆ ⋅ Eθ ( ρ )
Up till now, we have considered pure modes having optical angular momentum
m and angular dependence exp(imθ). In practice, most of the laser sources are of
linear polarization, meaning that the electric field of the emitted light is directed
along a predominant direction (say Y direction). Linear polarization can be
considered as a mix of two states with angular momenta m=1 and m=-1.

(Ex,Ey,Ez)=(0,E,0) (Eρ,Eθ,Ez)=(Esin(θ),-Ecos(θ),0)
ρ
Y
In Cartesian θ In Cylindrical
coordinate coordinate
X system system

(Eρ,Eθ,Ez)=(Esin(θ),-Ecos(θ),0)=exp(iθ).E/2.(-i, -1,0)+ exp(-iθ).E/2.(i, -1,0)

m=1 m=-1
 Linear polarized modes
To construct linearly polarized modes using the modes of pure angular momentum m, one adds (or subtracts, depending on
polarization) the fields of the two degenerate modes with angular momenta m and –m. In the case of low refractive index contrast
fibers, the convention is to label the linearly polarized (LP) modes as LPm-1,n , where m is the modal optical angular momentum.

1
F lin pol X
m,β ∝ ( Fm , β ( ρ ) exp ( imθ ) − F− m , β ( ρ ) exp ( −imθ ) )
2
i
Fmlin, βpol Y ∝ − ( Fm , β ( ρ ) exp ( imθ ) + F− m , β ( ρ ) exp ( −imθ ) )
2
Here, F indicates a 6-component electromagnetic field vector of a fiber mode. To derive expressions for the fields of linearly
polarized modes, one has to know the fields of a mode with angular momentum –m. Using the expressions derived earlier for the
E&M field of the circular symmetric modes, we get the following relation between the fields of the m and –m modes:

H m ( ρ , θ , z ) = exp(imθ ) exp(ik z z ) ( hρm ( ρ ), hθm ( ρ ), hzm ( ρ ) ) , z ) exp(−imθ ) exp(ik z z ) ( hρm ( ρ ), −hθm ( ρ ), hzm ( ρ ) )
H − m ( ρ ,θ =
Em ( ρ , θ , z ) = exp(imθ ) exp(ik z z ) ( eρm ( ρ ), eθm ( ρ ), ezm ( ρ ) ) , z ) exp(−imθ ) exp(ik z z ) ( −eρm ( ρ ), eθm ( ρ ), −ezm ( ρ ) )
E− m ( ρ ,θ =
 Linear polarized modes
For example, using the degenerate modes of angular momenta 1 and –1, the fields of the two linearly polarized modes are given by
adding or subtracting the fields of the pure angular momentum states. For example, for the Y polarization, in the cylindrical
coordinate system, the linearly polarized modes are:
i (Ex,Ey,Ez)=(0,E,0)
Flin pol Y
m 1, β ∝ − ( F1, β ( ρ ) exp ( iθ ) + F−1, β ( ρ ) exp ( −iθ ) ) =
2 In the Cartesian
coordinate system
  eρ ( ρ ) 
β ,1 β ,1
 −eρ ( ρ )    eρ ( ρ ) sin(θ ) 
β ,1

  β ,1   β ,1    β ,1 
  eθ ( ρ )   eθ ( ρ )    −ieθ ( ρ ) cos(θ ) 
  β ,1
ρ   β ,1
ρ    β ,1 
i e ( ) − e ( ) ez ( ρ ) sin(θ )
−   β ,1
z
 exp ( iθ ) +  β ,1
z
 exp ( −iθ )  =  β ,1 
2   hρ ( ρ )   hρ ( ρ )    −ihρ ( ρ ) cos(θ )  (Er,Eθ,Ez)=(Esin(θ),-Ecos(θ),0)
  h β ,1 ( ρ )   −h β ,1 ( ρ )    h β ,1 ( ρ ) sin(θ )  In the Cylindrical coordinate
 θ   θ
   θ  system
  h β ,1 ( ρ )   h β ,1 ( ρ )    −ih β ,1 ( ρ ) cos(θ ) 
 z   z    z 
 Y ρˆ ⋅ Eρ ( ρ )

ρ X
Linearly polarized modes O
θ
θˆ ⋅ Eθ ( ρ )

Let’s verify that we indeed have a Y polarized mode

in a sense that |Ey|>>|Ex| ~0, in the long wavelength limit
and for low RI contrast waveguides
Eβx ( ρ , θ ) = Eβρ ( ρ , θ ) cos (θ ) − Eβθ ( ρ , θ ) sin (θ ) =

=eρ ( ρ ) sin(θ ) cos (θ ) + ieθ ( ρ ) cos(θ ) sin (θ ) =

β ,1 β ,1 ( eρ
β ,1
( ρ ) + ieθ (ρ ))
β ,1

sin ( 2θ )
2
Eβy ( ρ , θ ) = Eβρ ( ρ , θ ) sin (θ ) + Eβθ ( ρ , θ ) cos (θ ) =

β ,1 2 β ,1
=eρ ( ρ ) sin(θ ) − ieθ ( ρ ) cos(θ ) =
2 ( eρ
β ,1
( ρ ) − ieθ
β ,1
( ρ ) ) −
( eρ
β ,1
( ρ ) + ieθ (ρ ))
β ,1

cos ( 2θ )
2 2
Linearly polarized modes
~1 +
~cos(2θ)

Low RI contrast waveguide ~sin(2θ)

High RI contrast waveguide

~sin(θ)