Lecture notes – Lasers

Lecturer – Prof Chris Dunsby

October 2024

Acknowledgements

This course is firmly based on the PowerPoint slides of Paul French and I’m very grateful for him sharing his
lecture material.

1 Absorption, spontaneous emission and stimulated emission

• laser gain media comprise an assembly of active species (atoms/ions/molecules/complexes etc.) with
approximately identical energy level systems (see homogenous and inhomogenous broadening later
in course)
• photons interact with populations of the active species in different energy states
• we will consider three different light-matter interactions: absorption, spontaneous emission and
stimulated emission

1.1 Spontaneous emission

Spontaneous emission is a random process where an atom/ion/molecule in an excited state decays via the
emission of a photon. The photon is emitted in a random direction.

E2, N2

γ21
E1, N1

Luminescing
gain medium

Pump energy

Figure 1.1: Top; energy level diagram for a pumped system where E2 has decay
rate γrad to E1. Bottom; crude schematic of system.
Page 1
In the gain medium shown in Figure 1.1, pump energy from an external source is used to drive gain medium
species from E1 into E2. The actual physical mechanisms that can be used for this pumping depend on the
gain medium and are discussed later in the course.

Here

En is the energy of the nth electronic level of the active species – dimensions [energy]

Nn is the number density of active species in the nth energy level – dimensions [length]-3

The gain species in the excited E2 state are always being drained by two processes: spontaneous emission of
photons and non-radiative decay (no photons emitted). Non-radiative decay can be due to a number of
processes depending on the type of gain species. In a gas laser this could be e.g. due to inelastic collisions
with other atoms/molecules or collisions of the gas with the walls of its container. In a solid-state laser this
could be due to phonon emission (energy lost to lattice vibrations).

Overall, the total decay rate γ21 is described by

 21 =  rad +  nrad

with

rad = radiative decay probability per unit time (rate at which photons are emitted spontaneously)

nrad = nonradiative decay rate (rate at which species loses energy and changes energy state from E2
to E1 without the emission of a photon)

γ21 = total decay rate from E2 to E1

where all of these are decay probabilities per unit time (rates) for a single atom/ion/molecule and have
dimensions [time]-1.

The quantum efficiency  of the laser medium is defined as

 rad
=
 rad +  nrad

and is the probability of a photon being emitted for each excitation of a laser species into its excited state.
For laser media typically  nrad  rad , i.e.  ~ 1 .

So the downward transition rate from E2 to E1 is given by the equation

dN 2 dN1
Equation 1.1 − = =  21 N 2  A21 N 2
dt dt

A21 is the Einstein ‘A’ coefficient that is equal to the radiative decay rate, i.e. A21 =  rad , and if  ~ 1 then
 21  A21 .

By solving Equation 1.1, the rate of decay from E2 to E1 can be described by the following formula

N2 ( t ) = N2t =0 exp ( − 21t ) ,

Page 2
where N 2t = 0 is the number density of laser species in the excited state at t = 0. The intensity of the
spontaneous emission as a function of time, I(t), is proportional to the rate of decay i.e.

dN 2
I (t )  rate of decay  − =  21 N 2 (t )
dt
and therefore

I (t )   21 N2t =0 exp ( − 21t ) = I t =0 exp ( − 21t ) = I t =0 exp ( −t /  )

where here I t =0 is the spontaneous emission intensity at time t = 0 and τ is the lifetime of E2 with
dimensions [time].

So, if the pump energy to a gain medium is switched off instantaneously, the intensity of the emitted
spontaneous light will have an exponential decay profile of lifetime τ, where τ is the time taken for the signal
to decay to 1/e of the initial level.

1.2 Absorption – stimulated upward transition

Absorption is the process where a photon incident on an atom/ion/molecule is absorbed through the
promotion of an orbital electron from one energy state to a higher energy state. Absorption can be considered
to be a stimulated process – it is stimulated by the arrival of the photon at the atom/ion/molecule.

E2, N2

W12
 E1, N1

Figure 1.2: Energy level diagram for absorption (stimulated upward transition).

Figure 1.2 shows four photons incident on a gain medium where two of the incident photons are absorbed
during their passage through it. The absorption results in the promotion of two electrons from E1 into E2.
Absorption can be considered to be a stimulated upwards transition from E1 to E2.

We can therefore write the stimulated upward transition rate as

dN 2 dN
Equation 1.2 = − 1 = N1W12 = N1  b12 .
dt dt
Here

W12 is the absorption probability per unit time (rate) for a single atom/ion/molecule in the ground E1
state with dimensions [time]-1. W12 depends on the radiation energy density ρ in the gain medium at
the transition frequency.

ρ is the energy density of the incident radiation at the transition frequency with dimensions
[energy].[length]-3.

Page 3
 b12 describes the probability of an absorption event for a single atom/ion/molecule as the upward
transition rate per unit energy density of incident radiation and so has dimensions
[length]3.[energy]-1.[time]-1, where W12 =  b12 .

1.3 Stimulated emission

Stimulated emission is a process where a photon incident on an atom/ion/molecule in an excited state causes
the emission of a second photon with identical frequency and direction as the first photon that causes the
atom/ion/molecule to become de-excited to a lower energy state.

E2, N2
W21


E1, N1

Figure 1.3: Energy level diagram for stimulated emission.

Figure 1.3 shows two photons incident on a gain medium where both photons induce stimulated emission
resulting in a total of four photons being emitted from the gain medium.

W21 is the probability per unit time (rate) for a single atom/ion/molecule in the excited E2 state to undergo
stimulated emission with dimensions [time]-1. W12 depends on the radiation energy density  in the gain
medium at the transition frequency.

We can therefore write the stimulated downward emission rate as

dN 2 dN1
Equation 1.3 − = = N 2W21 = N 2  b21
dt dt

Similarly to the case for absorption, W21 =  b21 . The parameter b21 describes the probability of stimulated
emission for a single atom/ion/molecule per unit energy density of incident radiation and so has dimensions
[length]3.[energy]-1.[time]-1.

We will prove below that the probability of a photon stimulating a downward electron transition from E2 to
E1 (stimulated emission) is the same as the probability of absorption (a photon stimulating an electron
transition upward from E1 to E2), i.e. that b12 = b21.

1.4 Overall rate equation for a 2-level system

The overall rate equation for a 2-level system including absorption, spontaneous emission and stimulated
emission is therefore

dN 2 dN1
Equation 1.4 − = = N2 A21 + N 2 b21 − N1b12
dt dt spontaneous emission
stimulated emission absorption

= N2 ( A21 + b21 ) − N1b12

Page 4
as illustrated in Figure 1.4.

E2, N2
W21 A21

 W12
E1, N1

Figure 1.4: Energy level diagram for a 2-level system including (from left to right)
absorption, stimulated emission and spontaneous emission.

1.5 Thermal equilibrium

Let’s consider a 2-level gain medium that is in thermal equilibrium with its surroundings, i.e. a black body
radiation field. First, as we are considering the system at steady state (equilibrium) then we must have

dN 2 dN1
Equation 1.5 = =0
dt dt
and so from Equation 1.4 we can write

N2 ( A21 + s ( ) B21 ) = N1s ( ) B12 .

Here, because we want to consider broadband blackbody radiation, we have switched from radiation density
 with dimensions [energy].[length]-3 to the spectral radiation density s() with dimensions
[energy].[length]-3.[time]. B21 and B12 have dimensions [length]3.[energy]-1.[time]-2 and are the spectral
density versions of b21 and b12. This equation can then be rearranged to give an expression for the spectral
radiation energy density

N 2 A21 A21
 s ( ) = = .
N1 B12 − N 2 B21  N1 
  B12 − B21
 N2 
Second, we can make use of the Boltzmann factor describing the occupation ratio of E2 to E1 that is expected
due to the ambient temperature T to write

N2  E − E1   h 
= exp  − 2  = exp  − 
N1  kBT   kBT 
where

h is Planck’s constant = 6.6310-34 J.s

kB is Boltzmann’s constant = 1.3810-23 J.K-1

T is temperature in Kelvins

υ is the optical frequency corresponding to the transition E2 →E1

hυ is the energy corresponding to the transition E2 →E1 and equals E2 – E1

Page 5
The Boltzmann factor tells us the probability of thermal excitation of the upper E2 state at a given
temperature. Substituting this result into our expression for ρs(υ) gives

A21
Equation 1.6  s ( ) = .
 h 
exp   B12 − B21
 k BT 
This expression tells us the expected spectral energy density for a given temperature. We can then compare
this result with Planck’s formula for the distribution of blackbody radiation at temperature T

8 n3h 3
Equation 1.7  s ( ) =
  h  
c3  exp   − 1
  kBT  

where

c is the speed of light in vacuum

n is refractive index

Equation 1.6 and Equation 1.7 can only both be true if the following equations hold:

A21 8 n3h 3
Equation 1.8 B12 = B21 and = .
B21 c3

Therefore, we have shown that the probability of absorption (B12, stimulated upward transition) is equal to
that of stimulated emission (B21). We also have an expression linking A21 to B21 in terms of known values.

Problem 1.1: For a gain medium operating at a wavelength of 500 nm at room temperature, calculate
the fraction of laser species in the upper state.

1.6 Gain line width

Real lasers have a finite gain line width: they can absorb radiation over a range of wavelengths and
stimulated emission can occur over a range of wavelengths. It is therefore necessary to include this gain line
width in our model of the gain medium. We therefore introduce a normalised Lorentzian gain line shape
function g(υ) where

 g
g ( ) = .
   g  
2

2  (0 −  ) +   
2

  2  
 

This function is shown in Figure 1.5. The area under the curve is normalised to unity and therefore g(υ) has
dimensions [frequency]-1 = [time]. The gain is centred at frequency υ0 and has a full width at half maximum
(FWHM) of Δυg.

Page 6
 g()

g


0

Figure 1.5: Lorentzian line shape

We can now rewrite the overall rate equation for a 2-level system (Equation 1.4) as

dN 2 dN1
Equation 1.9 − = = N 2 A21 + N 2  g ( ) B21 − N1  g ( ) B12 ,
dt dt

Where  is the radiation density of monochromatic light at frequency . Alternatively, this can also be
written as

dN2 dN1
− = = N 2 A21 +  g ( ) B21 ( N 2 − N1 ) = N 2 A21 +  g ( ) B21N
dt dt
net rate of stimulated photon
emission rate per unit vol

where N = N 2 − N1 is the population inversion per unit volume.

1.7 Gain coefficient and gain cross section

If we consider a monochromatic beam of frequency υ and intensity I incident on a volume of gain medium
Adz, see Figure 1.6, the increase in power of the beam is given by

AdI =  g ( ) B21N h Adz

photon volume
net stimulated photon emission energy
rate per unit volume

I ( ) dz I ( ) + dI ( )

Figure 1.6: Increase in beam intensity through thickness dz of gain medium.

So making use of Equation 1.8 we get

A21c3
dI =  g ( ) N h dz ,
8 n3h 3
c nI
and then using I =  , i.e. that  = , (see handout 2 slide 5) we get
n c

Page 7
 A21c 2
dI = I g ( ) N dz .
8 n2 2
We can see from this equation that we only obtain an increase in intensity when ΔN > 0. Remember that
N = N 2 − N1 is the population inversion density, i.e. the difference in gain species number density between
states E2 and E1.

Let’s now consider a gain medium of length z. By dividing both sides of this equation by I and integrating
we get
I ( z) z
dI A21c 2
z=0 I 0 8 n2 2 g ( ) Ndz
=
I

Where here I z =0 is the incident intensity at z = 0. Therefore

A21c 2
ln  I ( z ) − ln  I z =0
 = g ( ) N z
8 n2 2
 ( )

and so then

I ( z ) = I z =0 e
 ( ) z
.

where the gain coefficient as a function of frequency () is

A21c 2 N
Equation 1.10  ( ) = g ( )
8 n2 2
and  has dimensions [length]-1.

(υ) describes the gain per unit length of gain medium and depends on the normalised gain line shape
function g(υ), see Figure 1.5. It can only be positive if N  0 , i.e. gain in the laser medium is only possible
if there is a population inversion where N2 > N1.

It is sometimes useful to write Equation 1.10 as

Equation 1.11  ( ) =  ( )N

where we’ve introduce the gain cross-section () with dimensions [length]2. Therefore

A21c 2 g ( )
Equation 1.12  ( ) = .
8 n 2 2

Problem 1.2: Verify the dimensions of  are [length]2 and therefore that  ( ) =  ( )N is
dimensionally correct.

Problem 1.3: Consider a 2-level system where N2 = N/2, with N = N1 + N2. What is the intensity
transmission of this medium?

Page 8
1.8 Condition for laser action

Cavity mirrors

R1 R2

Gain medium
Pump energy

Figure 1.7: Simple laser cavity schematic showing an excited laser gain medium
and feedback provided by two cavity mirrors.

Figure 1.7 shows a simple laser cavity consisting of two mirrors with intensity reflectivities R1 and R2, and a
gain medium of length L with gain coefficient . The gain medium is pumped with a source of energy that
promotes laser species into the excited E2 state.

We now want to derive a threshold condition for laser operation by considering the gain required to exactly
equal the cavity losses. First, a beam passing through thickness z of gain medium experiences a gain of
I ( z ) = I 0e z . In round trip, the total gain factor is therefore e 2 L . In any real laser gain medium there are
also scattering losses that can be described by a scattering coefficient s with dimensions [length]-1. So the
−s2L
total round trip cavity transmission factor due to scattering is e . The radiation in the cavity also
experiences losses when reflecting from the two cavity mirrors and the round-trip transmission factor due to
( − s )2 L
the mirrors is R1R2. Therefore, the net cavity round trip gain is R1R2e and for laser action to occur we
must have

R1R2e( −s)2 L  1 .

We can rearrange this to find an expression for the threshold gain, i.e. the minimum gain coefficient in the
gain medium in order that the gain from a double pass of the beam through the gain medium will balance the
cavity losses

1
Equation 1.13  thresh = s − ln ( R1 R2 ) .
2L
If we combine Equation 1.13 with Equation 1.10 we then get

A21c 2 N thresh 1
 thresh = g ( ) =  ( ) N thresh = s − ln ( R1 R2 )
8 n 2 2
2L
where ΔNthresh is the threshold population inversion for laser action. So then we can rearrange to find an
expression for the threshold population inversion

Page 9
 Equation 1.14

8 n 2 2  1  1  1 
N thresh = s− ln ( R1 R2 )  = s− ln ( R1 R2 ) 
A21c g ( ) 
2
2L   ( )  2L 

In summary, in order to get laser action it is necessary for the population inversion to reach a threshold value
ΔNthresh: below this value laser action is not possible as the cavity losses are greater than the gain.

1.9 Is gain possible in a 2-level laser medium?

In this section we will use the equations we’ve derived above to consider the case where a 2-level gain
medium is optically pumped from E1 to E2 by light at the transition frequency that is incident on the gain
medium. We will show that it is not possible for a 2-level gain medium to provide gain because there is no
way to produce a population inversion.

E2, N2
Incident W21 A21
pump
radiation  W12
E1, N1

Figure 1.8: Diagram of a 2-level laser medium

From Equation 1.9 we have the following rate equation for a 2-level system

dN 2
− = N 2 A21 + N 2  g ( ) B21 − N1  g ( ) B12
dt

dN 2
− = N 2 A21 + N 2W21 − N1W12 = N 2 A21 + N Wst
dt

where  g ( ) B21 = W21 =  g ( ) B12 = W12 = Wst since B21 = B12. Remembering that N = N 2 − N1 and
N = N1 + N2 we have N = 2 N2 − N then we can write

dN 2
= − N 2  A21 + 2Wst  + NWst
dt

and so

dN 2  b
= − N 2 ( t ) a + b = −a  N 2 ( t ) − 
dt  a

where a = A21 + 2Wst and b = NWst are both constant in time. We can then rearrange and integrate both
sides

N2 (t ) 1 t
N 2t =0  b
dN 2 = −a  dt
 N2 (t ) − 
0

 a
Page 10
where N 2t =0 is the number density of laser species in the upper state at t = 0. This then gives

 b 
  N2 (t ) − a  
ln     = −at .
  N t =0 −  
b
  2 
a  

and therefore

 b b
N 2 ( t ) =  N 2t =0 −  exp ( −at ) +
 a a

so

 
 exp ( − ( A21 + 2Wst ) t ) +
NWst NWst
Equation 1.15 N2 ( t ) =  N2t =0 − .
 A21 + 2Wst  A21 + 2Wst

Now we consider the case of strong pumping where the radiation energy density in the gain medium ρ is
large and hence  g ( ) B21 = W21 = Wst A21 . We then get

 N N
N 2 ( t ) =  N 2t =0 −  exp ( −2Wst t ) + .
 2 2

N2(t)

N/2

N2t = 0
t

Figure 1.9: Population inversion as a function of time for a strongly pumped 2-

level gain medium.

The result is plotted in Figure 1.9 – it doesn’t matter how long you excite the gain medium for, you never
N
achieve N2 > N/2. In the steady state, i.e. as t → , and with strong pumping then N 2 → .
2

Remembering that N = 2 N 2 − N we can see that as t → , N → 0 .

Also recall Equation 1.10 that describes the gain coefficient  as a function of frequency, i.e.
A21c 2 N
 ( ) = g ( ) . Therefore, as t → , then  ( ) → 0 because N → 0 , i.e. a beam propagating
8 n2 2
through the gain medium experiences no gain and the gain medium has an intensity transmission of unity.

Page 11
In summary, in a 2-level system ΔN always remains negative, i.e. a population inversion is never achieved
even with strong pumping and after an infinite period of time. This is due to the fact that as you approach
N2 = N/2 then there is an equal probability of the pump radiation being absorbed as it causing stimulated
emission.

Page 12
2 Multi-level systems

In the previous section we showed that a 2-level system cannot provide gain and therefore laser action is not
possible in a 2-level system. Real gain media and lasers therefore need to be multi-level systems and in
general they can be approximated as either 3 or 4-level systems, see Figure 2.1.

E3, N3, 3 E3, N3, 3

fast 32 fast 32
P13 E2, N2, 2 P03 E2, N2, 2
W21 21 W21 21

W12 W12
E1, N1 E1, N1, 1
fast 10
E0, N0

a) Three level laser b) Four level laser

Figure 2.1: Generalised energy level diagrams for a) 3 and b) 4-level gain media

E3

E2

2 >> 3

E1
 E3

E2

a) Ruby laser ~ 3 level system

E1
E0

b) Nd:YAG ~ 4 level

Figure 2.2: Energy level schematics for a) ruby and b) Nd:YAG.

Figure 2.2 shows the energy level diagrams for ruby and Nd:YAG. Transitions are possible between any two
energy levels but population inversion and gain are only possible for special cases. Pump level (E3) can be a
manifold of higher levels with rapid relaxation to laser upper level (E2).

There are some important differences between 3 and 4-level systems and these will be explored in the
following sections.

Page 13
2.1 Ideal 3-level gain medium

The energy level diagram for a 3-level gain medium is shown in Figure 2.1a).

If a 3-level gain medium is not excited, then all species are in the ground E1 state. Therefore the gain
medium absorbs light at the laser wavelength if it is not excited.

We can write the rate equations for the three energy levels as follows

Equation 2.1

dN3
a) = P13 N1 −  32 N3
dt

dN 2
b) =  32 N3 − N 2 21 − W21 N 2 + W12 N1 =  32 N3 − N 2 21 − Wst N
dt

dN1
c) = − P13 N1 + N 2 21 + W21 N 2 − W12 N1 = − P13 N1 + N 2 21 + Wst N
dt

with N = N1 + N2 + N3

Here we’ve introduced P13 which is the rate at which the gain species are pumped from E1 to E3 via whatever
pumping mechanism is being employed. It has dimensions [time]-1 and depends on the amount of pump
energy being put into the laser system.

In an ideal 3-level laser system we will assume that the decay rate from E3 to E2 is infinitely fast, i.e.
 32 →  , and therefore that the population of E3 tends to zero, i.e. N3 → 0 .

Under this assumption, the rate at which gain species are pumped into E3 must equal the rate at which they
decay to E2, i.e. the decay from E3 to E2 is immediate and

P13 N1 =  32 N3 .

dN1 dN 2
In steady state we have that = 0 and = 0 so then we obtain from Equation 2.1c) that
dt dt

0 = − P13 N1 + N2 21 + Wst N

and then

Equation 2.2 P 13 N1 − N2 21 = Wst N .

2.2 Ideal 3-level gain medium for small radiation energy density

Consider the situation where the radiation density in the gain medium ρ is small hence Wst → 0 . Also, we
must have N1 + N2 + N3 = N = constant and as we are assuming that N3 → 0 we have N1 + N 2 = N . So
then

P 13 N1 = N2 21 = ( N − N1 )  21 ,

which can be rearranged to give

Page 14
  21 N P N
Equation 2.3 N1 = and N 2 = 13 ,
P13 +  21 P13 +  21

and therefore

N ( P13 −  21 )
Equation 2.4 N 3 =level = .
P13 +  21
0

We’ve introduced the subscript  = 0 here to remind us that this formula is valid for zero radiation in the gain
medium. In order for the gain medium to provide gain we must have ΔN > 0 otherwise absorption by laser
species in the ground state will be larger than the gain provided by stimulated emission. So in order to have
gain we must have P13   21 , i.e. the pumping rate must be greater than the spontaneous decay rate of the
upper E2 state. Unlike the 2-level system, a population inversion can be achieved in a 3-level system and
therefore the gain medium can provide gain.

The pump power absorbed by the system per unit volume (absorbed pump power density) is h13 P13 N1 and
by using Equation 2.3 this gives us an expression for the absorbed pump power density in terms of P13 and
physical constants

  N 
Absorbed pump power density = h13 P13 N1 = h13 P13  21 .
pump photon rate photons  P13 +  21 
energy absorbed
per unit volume

Using the fact that P13 =  21 when ΔN = 0 then

h13 21 N
Minimum absorbed pump power density required for population inversion =
2

2.3 Gain saturation in an ideal 3-level system

In the section above we considered the 3-level gain medium close to population inversion where Wst is small.
Here we consider the general case for any energy density of laser radiation in the gain medium and derive a
general expression for ΔN. Starting from Equation 2.1(c) and considering the steady state we can write

dN1
= 0 = − P13 N1 + N 2 21 + W21 N 2 − W12 N1
dt

and then P13 N1 + W12 N1 = N 2 21 + W21 N 2 and so

N 2 P13 + Wst
= .
N1  21 + Wst

We must have N1 + N 2 + N3 = N and as we are assuming that N3 → 0 we have N1 + N 2 = N . In order to

solve for ΔN we first find expressions for N1 and N2 in terms of just N and N2/N1

 1   N 2 / N1 
N1 = N   and N 2 = N1 ( N 2 / N1 ) = N  .
 1 + N 2 / N1   1 + N 2 / N1 

Therefore
Page 15
  N 2 / N1   1   N 2 / N1 − 1 
N = N2 − N1 = N  − N = N .
 1 + N 2 / N1   1 + N 2 / N1   1 + N 2 / N1 
Substituting in the expression for N2/N1 gives

 P13 + Wst 
  + W −1   P −    21 + Wst  P13 −  21
N = N  21 st
 = N  13 21  =N .
 1 + P13 + Wst    21 + Wst  P13 +  21 + 2Wst  P13 +  21 + 2Wst
  +W 
 21 st 

This can be rewritten as

 P −  1
N = N  13 21  .
 P13 +  21  1 + 2Wst
P13 +  21

P13 +  21
We can see that ΔN drops to half its peak value when 2Wst = P13 +  21 and so we define Wsat = .
2
Then we obtain

 P −  1
Equation 2.5 N = N  13 21  .
 P13 +  21  1 + Wst
Wsat

A21 8 n3h 3
Remembering that Wst =  g ( ) B21 ,
n
= 3 and  = I then
B21 c c

A21c3 IA21c 2 g ( ) I ( )
Equation 2.6 Wst =  g ( ) = =
8 n3h 3 8 n 2 h 3 h

A21c 2 g ( )
where we make use of our earlier definition of the gain cross section  ( ) = (Equation 1.12).
8 n 2 2
So therefore Wst  I and Equation 2.5 becomes

 P −  1 1
N = N  13 21  = N 3 =level .
 P13 +  21  1 + I
0
I
1+
I sat I sat

I sat ( )
Here we have used Wsat = where Isat is the saturation intensity at which ΔN drops to half its peak
h
h P13 +  21
value. Isat = Wsat = h which depends on . Remember that  ( )  g ( ) (Equation 1.12)
 ( ) 2 ( )
and so Isat is therefore lowest at the peak of the gain. Using Equation 1.11, we can then find an expression for
the gain coefficient of the 3-level gain medium

Page 16
 N 3 =level
0

 P13 −  21  1  03 level
Equation 2.7  ( ) =  ( )N =  ( ) N   = .
 P13 +  21  1 + I 1+
I
 03 level I sat I sat

Here 03 level =  ( )N3 =level

0 is the small signal gain coefficient for the 3-level gain medium, i.e. the gain

coefficient of the gain medium for incident intensities that are sufficiently small that they do not significantly
perturb the population of E2 or E1.

2.4 Round trip gain and gain clamping in a laser cavity

End
point
2( − s ) L
I je R1R2 Starting
point
I j e(
Gain medium  −s)L
Ij

Gain  Output
Loss s
I j e(
2( − s ) L  −s)L
I je R1 R1

R2 R1
L
Figure 2.3: Diagram showing how the intra cavity intensity is modified as it
completes a full round trip.

In a laser cavity the optical field passes forwards and backwards through the gain medium as shown in
Figure 2.3. The initial intensity for the jth round trip is Ij and at the end of the round trip the intensity is then
2( − s ) L
I je R1R2 . So we can write

I j +1 = I j exp ( 2 ( − s ) L ) R1R2 = GI j

where G is the round trip cavity gain. In order for the laser to reach threshold then we require G = 1, i.e.

exp ( 2 ( thresh − s ) L ) R1R2 = G = 1

Now consider what happens above the threshold for laser action. Can G > 1? This can occur for short time
periods, but in the steady state we require G = 1, otherwise the output from the laser must be increasing over
time and this would be inconsistent with the system being in a steady state!

Above threshold in steady state the output of a laser must be constant and for this to be the case we
require G = 1. This is called gain clamping.

As already shown in section 1.8, we can rearrange the expression above to obtain threshold gain medium
gain coefficient for laser action thresh

Page 17
 ln ( R1 R2 )
Equation 2.8  thresh = s − =  ( ) N thresh
2L

where we also define the threshold population inversion required for laser action ΔNthresh.

2.5 Gain clamping in a laser cavity

Below threshold, the optical energy density in the laser cavity at the laser frequency is low and the
probability of stimulated emission is low. The main route through which the gain medium loses energy is via
spontaneous emission. Therefore increasing the pump power increases the population inversion and increases
the gain available in the laser medium.

Once threshold is reached, the optical energy density in the gain medium rapidly increases due to stimulated
emission.

Above threshold, increasing the pump power increases the optical energy density in the gain medium which
increases the rate of stimulated emission. The increased rate of stimulated emission depopulates the excited
state at exactly the same rate that the additional pump power is promoting laser species to the excited state.
Overall the net effect is that the gain remains clamped at unity in steady state.

In a real laser spontaneous emission is isotropic, i.e. it is emitted equally in all directions into 4 steradians.
Therefore, spontaneous emission is not very likely to be coupled into a mode of the laser cavity, i.e. it is
much less likely to be reflected off a cavity mirror and amplified than a photon produced by stimulated
emission. Stimulated emission is directional, i.e. the emitted photon travels in exactly the same direction as
the stimulating photon. So the stimulated photons remain within the laser mode and so are therefore much
more likely to cause stimulated emission themselves.

2.6 Laser cavity modes

A simple laser consists of a gain medium inside an optical cavity consisting of two highly reflective mirrors,
see Figure 2.4. The field inside the cavity must form a standing wave and therefore only certain frequencies
are allowed within the cavity.

-s

R2 R1

c
E()
2d

Figure 2.4: Top – labelled diagram of cavity. Middle – cavity standing wave.
Bottom – comb of allowed frequencies in the cavity.

Page 18
The round trip path length must equal an integer number of wavelengths, i.e. 2d = j , where d is the
optical path length between the cavity mirrors and j is an integer. Converting this expression to frequency
yields

jc cj
2d = j  = and therefore  =
 2d
The spacing in frequency between the cavity modes is then

c
 =
2d

c
Therefore, the laser oscillation is confined to discrete frequencies  j = j .
2d

Frequencies allowed by cavity modes

G()
G =1
Gain
profile


j-1 j j+1 j+2

Figure 2.5: Allowed cavity mode frequencies plotted within overall gain profile.

Therefore, within the gain profile of the gain medium, there will be a finite number of specific laser modes
(frequencies) that experience gain, see Figure 2.5.

The laser mode with the highest gain will produce laser action first. The first mode to lase will extract the
gain from the laser medium, clamp the population inversion to ΔNthresh and clamp the gain G = 1. No other
modes will be able to lase as they are all below threshold. (In non-ideal lasers this is not the case, see e.g.
spatial and spectral hole burning.)

2.7 Non-ideal 3-level system

E3, N3, 3
31 32
P13 E2, N2, 2
W21 21

P31 W12
E1, N1

Figure 2.6: Energy level diagram for a non-ideal 3-level laser system
Page 19
The energy level diagram for a non-ideal 3-level gain medium is shown in Figure 2.6. Here there are
additional unwanted transitions: the pumping mechanism can also deplete the E3 state at rate P31; and the E3
state can decay directly to E1 at rate γ31.

If we define

 21  32
= and  =
 32  32 +  31

then it is possible to show that

 (1 −  )  P13 −  21 
N  =0 = N   .
 (1 + 2  )  P13 +  21 

It is necessary to have β < 1 in order to obtain ΔN > 0 and the condition for achieving a population inversion
(ΔN = 0) is

  21 
P13    .
 (1 −  )  

For an ideal 3-level laser  → 0 and  →1 .

2.8 Ideal 4-level gain medium

The energy level diagram for an ideal 4-level gain medium is shown in Figure 2.2b) and shows the additional
E0 energy level beneath the lower level of the laser transition (E2→E1).

The key difference between a 3-level gain medium and a 4-level gain medium is that a 4-level gain
medium presents no absorption at the laser wavelength when it is in the ground (E0) state.

We can write the rate equations for the four energy levels as follows

Equation 2.9

dN 3
a) = P03 N 0 −  32 N3
dt

dN 2
b) =  32 N3 − N 2 21 − W21 N 2 + W12 N1 =  32 N3 − N 2 21 − Wst N
dt

dN1
c) = − 01 N1 + N 2 21 + W21 N 2 − W12 N1 = − 10 N1 + N 2 21 + Wst N
dt

dN 0
d) = − P03 N 0 +  10 N1
dt

with N = N0 + N1 + N 2 + N3

Page 20
As for the ideal 3-level system we will assume that  32 →  , that N3 → 0 , that P03 N0 =  32 N3 , and
dN 0 dN dN 2
therefore that N = N0 + N1 + N 2 . In steady state we have = 0, 1 = 0 and =0.
dt dt dt

2.9 Ideal 4-level gain medium for small radiation energy density

The condition for obtaining a population inversion is ΔN > 0. In a laser cavity below inversion of the gain
medium there are few photons in the gain medium and hence ρ and therefore Wst are both small. Using these
assumptions from Equation 2.9(b) we therefore get

dN 2
=  32 N3 − N 2 21 = 0 = P03 N 0 − N 2 21
dt
and therefore that

N 0  21
Equation 2.10 = .
N 2 P03

Similarly, from Equation 2.9c) and Equation 2.9d) we get

dN1 dN 0
= − 10 N1 + N 2 21 = 0 and = − P03 N 0 +  10 N1 = 0
dt dt
and so from the equation for dN1/dt we have

N1  21
Equation 2.11 = .
N 2  10

Then using Equation 2.10 and Equation 2.11 we can begin to work towards a useful expression for ΔN. First

N N        + P + P  
N = N0 + N1 + N2 = N2  0 + 1 + 1 = N2  21 + 21 + 1 = N2  21 10 21 03 03 10  .
 N2 N2   P03  10   P03 10 
Therefore

 P03 10 
N2 = N  
  21 10 +  21P03 + P03 10 
and using Equation 2.11 gives

 P03 21 
N1 = N  .
  21 10 +  21P03 + P03 10 
Thus

NP03 (  10 −  21 )
Equation 2.12 N 4=level = .
P03 10 + P03 21 +  21 10
0

If the spontaneous decay rate from E1 to E0 is faster than the spontaneous decay rate of the laser transition
from E2 to E1, i.e. that  10   21 , then always get a population inversion in a 4-level system if P03 > 0.

Page 21
In an ideal 4 level system then  10 →  and so N1 → 0 and N → N 2 . Therefore, we have the condition
that  10  21 and so then

NP03 (  10 −  21 ) NP03 10 NP03

Equation 2.13 N  =0 =  =
4 level
.
P03 (  21 +  10 ) +  21 10 P03 10 +  21 10 P03 +  21

Compare this equation to the one obtained for a 3 level system (Equation 2.4) where

N 3 =level =N
( P13 −  21 )
. It is clear that in a 4-level system there is no threshold pump rate that is required
P13 +  21
0

in order to obtain a population inversion.

2.10 Gain saturation in a 4-level system

It is possible (see problem sheet 3) to obtain a general expression for ΔN in the general case when significant
laser radiation is present in the cavity and the result is

NP03 (  10 −  21 ) A
N = N 2 − N1 = =
Wst ( 2 P03 +  10 ) + P03 21 + P03 10 +  21 10  W C
B 1 + st 
C B  B 

dN 0 dN1 dN 2
assuming steady state = = = 0 , N3 → 0 and N = N0 + N1 + N 2 .
dt dt dt

As for the 3-level system we define saturation of the population inversion when ΔN reaches half of its peak
value and so introduce

B P03 10 + P03 21 +  21 10

Wsat = = .
C 2 P03 +  10

Then

NP03 (  10 −  21 ) 1
N = . .
P03 10 + P03 21 +  21 10 1 + Wst
Wsat

For 4-level systems we define Isat as

h h  P03 10 + P03 21 +  21 10 

Isat = Wsat =  
 ( )  ( )  2 P03 +  10 
and so then

NP03 (  10 −  21 ) 1
Equation 2.14 N = .
P03 10 + P03 21 +  21 10 1 + I
N 4 =level
0
I sat

and therefore

Page 22
  04 level ( )
Equation 2.15  ( ) =  ( ) N =
I
1+
I sat

where

 P03 (  10 −  21 ) 
 04 level ( ) =  ( ) N  .

 03 10
P + P 
03 21 +  
21 10 

N 4 =level
0

 04 level is the small signal gain coefficient for the 4-level gain medium, i.e. the gain coefficient of the gain
medium for intensities that are sufficiently small that they do not significantly perturb the population of E2.

In an ideal 4-level laser then  10  21 and so

 P03 
 04 level ( )   ( ) N  
 P03 +  21 
and

h P03 +  21
I sat ( )  . .
 ( )  P03 
2   +1
  10 
If we make the additional assumption that the decay rate from E1 is much larger than the pumping rate, i.e.
 10 P03 , then we get

h ( P03 +  21 )
I sat ( ) 
 ( )

P13 +  21
3 level
Compare this to the expression for Isat in a 3-level system Isat = h , which is half that for the 4-
2 ( )
level system, i.e. the gain in a 3-level system saturates more easily than in a 4-level system.

In a practical 4-level laser system then the pumping rate is much smaller than the spontaneous decay rate of
the laser transition, i.e. P03  21 , and so in this case

 P03 
 04 level ( )   ( ) N   =  ( ) NP03 21
  21 
and

 21 h
Isat  h = .
 ( )  ( ) 21

Page 23
2.11 Non-ideal 4-level gain media

In the section above we consider the condition for gain in a 4-level gain medium under steady state and we
derived Equation 2.12

 NP03 ( 10 −  21 ) 
N  =0 = N 2 − N1 =  
  21 10 +  21 P03 + P03 10 

from which we can see that it is possible to achieve N  0 if  10   21 and P03 > 0.

If  10   21 then a 4-level laser becomes self-terminating, i.e. it can provide gain for a short period of time
until the E1 energy level fills up. This is because as E1 fills then E0 is depleted and it is no longer possible for
the pumping mechanism to promote laser species from E0 to E3. Also, with the gain medium in E1 the gain
medium will absorb radiation at the transition frequency in the cavity.

Therefore, continuous wave (c.w.) laser operation is not possible in a 4-level gain medium if  10   21 .
1
However pulsed operation is possible and pulses with durations  p ~ can be achieved.
 21

E3, N3, 3
32
P30 E2, N2, 2
P03 W21 21
31

30 20
W12
E1, N1, 1
10
E0, N0

Figure 2.7: Energy level diagram for a non-ideal 4-level gain medium

The energy level diagram for a non-ideal 4-level laser system is shown in Figure 2.7. Here there are
additional unwanted transitions:

• the pumping mechanism can also deplete the E3 state at rate P30
• the E3 state can decay directly to E0 at rate γ30
• the E3 state can decay directly to E1 at rate γ31
• the E2 state can decay directly to E0 at rate γ20

We define

  21  31 ( 21 +  20 )   32
 = +  and  =
  10  32 10   32 +  31 +  30

and it is then possible to show (not derived here) that the population inversion in a non-ideal 4-level gain
medium becomes
Page 24
 N (1 −  )  P03
N  =0 =
 2 21 
1 +  +  P +  + 
  32  03 20 21

A population inversion is achieved if   1 and P03  0 .

For an ideal 4-level gain medium need:  32 →  , i.e. that N3 → 0 ;  10  21 , i.e. that the spontaneous
decay rate out of E1 is much greater than the spontaneous decay rate into E1; and for  31 ,  30 ,  20 → 0 .
Therefore  → 0 and  →1 .

2.12 Comparison of population inversion required for laser action in 3 and 4-level lasers

Remember from section 2.4 that for steady state laser operation we require a net cavity round-trip gain G = 1
and that the threshold gain and population inversions to achieve this were given in Equation 2.8

ln ( R1 R2 )
 thresh =  ( ) N thresh = s − .
2L

N  NP03 
N 4 =level = 
 P03 +  21 
0

Nthresh
Normalised
pump rate P

N 3 =level =
( P13 −  21 ) N
P13 +  21
0

-N

Figure 2.8: plot of theoretical population inversions for ideal 3 and 4-level gain
media as a function of pump rate. The solid black curves show how ΔN changes
when the gain medium is not in a cavity. The dashed red line shows how ΔN is
clamped when the gain medium is in a cavity.

Laser cavities with higher losses, e.g. a higher fraction of output coupling or scattering within the gain
medium, require a higher steady state ΔN and  to achieve the threshold for laser output.

Page 25