Appendix c

Transfer Matrices

A transfer matrix relates the right- and left-propagating waves at one end of a struc-
ture to the right- and left-propagating waves at the opposite end of the structure.
These matrices form a powerful tool for the analysis of structures consisting of sev-
eral concatenated sections, where each section is described by a transfer matrix.
The transfer matrix for the complete structure is then found as the matrix product of
the individual matrices. In Figure C.I, we show the case for a structure with two
sections. If F 1 is the transfer matrix for section I and F 2 is the transfer matrix for
section 2, we have

( E(2») ( E(I»)
E~2) = F1 E~l) (C.I)

and

( E(3») ( E(2»)
E~3) = F2 E~2) (C.2)

hence,

( E(3») ( E(I»)
E~3) = F2F1 E~l) (C.3)

Note that the order of the matrices is reversed relative to the figure.
To use this method, we need to build up a "library" of matrices for some specific
structures. If a section is simply a homogeneous material of length L, and if the
propagation constant is 13, then

E~2) = exp(-jf3L)EV) (C.4)

E~2) = exp(jf3L)EP) (C.5)

Tunable Laser Diodes and Related Optical Sources, 2nd ed. By Buus, Amann, and Blumenthal 351
ISBN 0-471-20816-7 © 2004 by the Institute of Electrical and Electronics Engineers.
352 TRANSFER MATRICES

£BSlB£
E(1)
.-?-
Section 1 E (2)
.-?-
Section 2 E (3)
.-?-

Figure C.1 Right- and left-propagating waves in a two-section structure.

giving

Fhom(f3L) == (exP(-o-J f3L) 0)

exp(jf3L)
(C.6)

If a "section" is a discontinuity with a field reflection coefficient r when seen from

the left, and -r when seen from the right (see Figure C.2), then

(C.7)

E~2) == -rE~2) + (1 + r)Ep) (C.8)

This is seen by setting E~l) and E~2) equal to zero in tum, and using the continuity
condition at the interface. This leads to the transfer matrix

1/(1 - r) - rl(l - r))

Fdis(r) = ( -rl(l - r) (C.9)
1/(1 - r)

It is important to note that we can use the matrix elements "the other way around,"
and express the reflection in terms of matrix elements:

(F dis) 2 1
r== (C.lO)
(F dis) 22

If the discontinuity is caused by the transition from a material with index n 1 to one
with index n2' then

». -n2
r== - - - (C.ll)
nl + n2

E(1) E(2)
--.r. --.r.
E(1) E(2)
.-L .-L
~~
r -r
Figure C.2 Reflection at a discontinuity.
 TRANSFER MATRICES 353

and using (C.9), the transfer matrix becomes

-(n 1 - n2)/2n2 )
(C.12)
I + (nl - n2)/2n2

We can use the transfer matrix description to analyze a laser with N sections:

(C.13)

In a laser, there are no incoming fields, and it follows from E1j+l = E~l) = 0 that we
must have F 22 = 0, corresponding to an infinite reflectivity. This can be used to de-
rive the oscillation condition. We consider the simplest possible Fabry-Perot laser,
which can be modeled as an amplifying, homogeneous medium placed between two
reflectors, as shown in Figure C.3.
In this case, the whole structure consists of only three sections, and the transfer
matrix is

(C.14)

Carrying out the matrix multiplications and using the oscillation condition (Ffp)22 =
oleads to
(C.15)

Except for the sign, this is the same result as (2.39). The reason for the sign change
is that in the derivation of (2.39) we were looking at the facets from inside the cavi-
ty, whereas in Figure C.3 we are looking at the left facet from outside; consequent-
ly, the sign of the field reflection coefficient r 1 is reversed.
As an additional transfer matrix, we recall the result for a periodic structure from
Section 3.2.2

'a~ ] jK
[ cosh( "IL) - ~ sinh ( "IL) - - sinh( yL)
Y
Fper = (C.16)
jK
- sinh(yL)
y
'a~ sinh( "IL)
[ cosh( "IL) + ~ 1

L
P
Figure C.3 Fabry-Perot laser as two reflectors and a homogeneous gain region of length L.
354 TRANSFER MATRICES

There is an interesting formula for a structure consisting of N identical substruc-

tures, each characterized by the matrix F. It is, of course, possible to use the relation

F tot = FN (C.I7)

but using the Lagrange-Sylvester formula [1] for a 2 by 2 matrix, we have directly

(C.18)

where Al and A2 are the eigenvalues ofF, and 1 is the unit matrix.
The eigenvalues are found from

F ll - A
Det ( F (C.I9)
21

which gives

(C.20)

We can therefore write the eigenvalues as

(C.2I)

where D is the determinant ofF.

We recall that the reflection from a single element is given by (C.10); for a struc-
ture with N identical elements we have

(C.22)

and

(C.23)

This gives

(C.24)

REFERENCE
[1] Barnett, S., Matrices: Methods and Applications, Oxford, U.K., Clarendon Press, 1990.