Stimulated Raman Scattering in a

Plasma channel
Ankur Nath, Bhushan Thakur

Abstract

Stimulated Raman scattering (SRS) in a plasma as a successful candidate

for generating intense short pulses has been a big interest of research among
the scientific community. However, like other parametric instabilities, Raman
scattering also obeys certain cut-off conditions, when laser pulses are injected
inside a stationary plasma channel. In this report, we first briefly talk about
parametric instabilities, and then we proceed to derive the fundamental theory
of stimulated Raman scattering. We highlight the essential pre-requisites to
amplify a seed laser pulse, compare the energy gain processes for both back-
scattering (BSRS) and forward-scattering and solve coupled wave equation to
obtain the dispersion relation and growth rate for BSRS. The theory work is
then complimented with Particle-In-Cell simulations, which demonstrates how
the pump laser loses energy to the scattered wave under appropriate conditions
as they counter-propagate in a plasma channel. Furthermore, we investigate
the effect of temperature on Raman instability. In the end, we analyse our
results and report amplified pulse intensities.

1
Contents
1 Introduction 3
1.1 Parametric Instabilities - a brief discussion . . . . . . . . . . . . . . . 3
1.2 Three wave coupling - threshold condition . . . . . . . . . . . . . . . 4

2 Theory of Stimulated Raman scattering 5

2.1 Coupled wave equations . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.2 Dispersion relation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.3 The growth rate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

3 Simulation 11

4 Results and Conclusion 12

2
1 Introduction
The generation of ultra-intense laser pulses has been a huge demand owing to its
massive applications in various sectors, just as building compact plasma-based accel-
erators ([1], [2], [3], [4]), medicine ([5]), Inertial-confinement fusion etc.([6], [7], [8],
[9], [10]). The standard procedure to obtain a high intensity short pulse involes the
process of stretching and amplification (Chirped pulse amplification or CPA), as was
first demonstrated by Strickland et. al [11]. However, a nobel approach to generate
ultra-short pulses of large intensities was suggested by G. Shvets [12], by exploiting
the non-linear three-wave coupling process in a plasma medium, which allows a short
seed pulse of low intensity to be fed into the plasma and make it grow to intensities
subsequently highly than the drive laser pulse. The fundamental phenomena which
amplifies the seed pulse is called the Raman Back scattering that involves the in-
teraction between counter propagating electromagnetic waves and the plasma wave
(Langmuir wave). This has gained considerable experimental attention over the years,
owing to its advantages over thermal damages which is an essential caveat one faces
while following the stretching method.
Raman scattering is one of the famous parametric instabilities, which is very
common in plasma systems with laser pulses injected in it. Parametric instabilities
aren’t unique to plasmas only, rather it is universal to any system showing oscillatory
behaviour due to change in its ”parameters”. To understand, we take an oscillatory
system subjected to a huge amplitude to begin with, and let’s say the force was
longitudinal to initiate the oscillation. Then, it is observed that the oscillating system
no longer obeys oscillating only in the longitudinal plane, rather, the system also starts
showing the transverse pendulum mode. This happens because now the pendulum
mode and the longitudinal mode are coupled, one dies, another grows, and thus the
system becomes unstable.
A corollary could be drawn from this example to understand the potential agents
that could lead to excitation of waves in a plasma. A plasma supports a lot of modes,
and the frequencies and wavenumbers are dependent on parameters like density, par-
ticle velocity, temperature etc. Turns out, if a laser pulse of frequency is greater than
2ωpe , (natural electron plasma frequency), then the laser may decay into a photon
(Raman scattered light), and a plasmon (electron-plasma wave)[13].

1.1 Parametric Instabilities - a brief discussion

There are many reports which show extensive discuss parametric instabilities in plas-
mas, including magnetized and unmagnetized plasmas, inhomogeneous and uniform
plasmas, for linear and non-linear electrostatic and electromagnetic waves. Generally,
parametric excitations require a minimal set of common characteristics
• Matching frequency

3
 • Theshold frequency

• Frequency locking.
The matching and frequency locking conditions follow from the intrinsic non-linearity
of the multiple frequency system, and can be viewed either as coming from conserva-
tion of energy and momentum or as coming from the resonance [14].

1.2 Three wave coupling - threshold condition

Let’s understand the setup of Raman scattering in detail. A laser that has frequency
higher than ωpe will be able to propagate through a plasma channel (pump). What
happens next depends on how large the pump frequency is. Assuming that the pump
frequency ω0 is larger than 2ωp we observe Raman scattering: ω0 gets scattered to ωs
(scattered) + ωe (plasmon)
ω0 = ωs + ωe (1)

The wave-numbers can be expressed as

k0 = ks + ke (2)

The scattered wave and the pump wave have frequencies closeby, which generates
beat frequencies equal to the difference ω0 − ωs . Inside plasma, electron respond
resonately only when beat frequency becomes equal to ωpe (rest of the oscillations die
off with time).
ω0 = ωs + ωpe (3)

k0 = ks + kpe (4)

It is worth mentioning that the scattered wave, and the beat wave, all of it is born
from the density fluctuations present inside the plasma, called noise. If we already
supply a short laser pulse (probe/seed) that has the parameters matched exactly with
the scattered wave, we can exploit Raman scattering to amplify this seed pulse.
The dispersion relations for the pump and scattered electromagnetic (EM) wave.

ω 2 = ωp2 + c2 k 2 (5)

And for the Langmuir wave in a cold plasma, will have

ω = ωp (6)

As discussed before, since Raman scattering is a parametric instability, it needs

to satisfy frequency matching. So the scattered wave needs to have a frequency,

ωs = ω0 − ωp (7)

4
Using this condition, in the EM wave dispersion relation, we get the wave propagation
vector ks for the scattered wave as

ω02 − 2ω0 ωp
ks2 = (8)
c2

Since ks < 0 is non physical, this means the threshold condition is,

ω02 − 2ω0 ωp = 0 (9)

Which gives us the condition

ω0 = 2ωp (10)

This is the fundamental SRS cut-off condition.

√
If ωp > ω0 /2, and since ω ∝ n, where n is the density, it means, the plasma
density np > nc /4, where the suffice represent the plasma and the critical plasma
density affected by the frequency ω0 of the pump laser. This results in the scattered
ωs = ω0 − ωp < ωp , and
(ω0 − ωp )2 − ωp2
ks2 =
c2
(from the EM wave dispersion relation) becomes

ks < 0

This means the wave propagation vector is no longer real, and thus we can claim
that SRS happens only if np ≤ nc /4.

2 Theory of Stimulated Raman scattering

2.1 Coupled wave equations
In this section, we derive the coupled wave equations describing the pump, scattered
wave, and the role of Langmuir operator in their dependencies. The scattered wave
amplitude is significantly low than the pump laser, and vector potential corresponding
to the scattered wave can be treated as a perturbation [15]. From Maxwell’s equations
we have
B=∇×A (11)
4π 1 ∂E
∇×B= J+ (12)
c c ∂t
and we have the generalised expression

1 ∂A
E=− − ∇ϕ (13)
c ∂t

5
Using the expression of B and E in eq.(12), and choosing the Gauge ∇ · A = 0 we
get,
1 ∂2 4π 1∂
!
− ∇ 2
A = J − (∇ϕ) (14)
c2 ∂t2 c c ∂t
This the equation to take care of the pump laser and its subsequent scattering as it
propagates in plasma.
Now, we try to look at the geometry of the problem. We divide the current density
into longitudinal and transverse components as J = Jl + Jt .
From Poisson’s equation,
∇2 ϕ = −4πρ (15)

which we can use in the conservation of charge equation

∂ρ
+∇·J=0 (16)
∂t

and get this modified equation as

!
∂
∇· ∇ϕ − 4πJ = 0
∂t

. We use the longitudinal and transverse of current density now, and using the
condition ∇ · Jt = 0, we get
∂
∇ϕ = 4πJl (17)
∂t
Using this result in eq.(14), we get

1 ∂2 4π
!

2 2
− ∇2 A = Jt (18)
c ∂t c

Thus, the vector potential is connected with the transverse component of the current
density only. We take the condition A · ∇ne = 0. We have Jt = −ene ue , where ue
is the oscillatory velocity of plasma electrons as a response to the EM wave. We can
similarly divide ue = ul + ut . If ut ≪ c, and given ut = eA
mc
, then we can write

∂ut e ∂A
=−
∂t mc ∂t

also, from the momentum equation,

∂ut −eEt
= (19)
∂t m

These are the useful relations that we can use for the transverse velocity ut . Now,

6
using the expression of Jt and ut in it, and putting them in eq.(12), we get

4πe2 ne
!
∂2
− c 2 2
∇ A = − A (20)
∂t2 m

This is the wave equation for EM waves inside plasma.

The pump laser vector potential can be considered undergoing a perturbation.

A = AL + Ã

and
n = n0 + n˜e

Where n0 is the equilibrium plasma density. We linearise eq.(12) and only keep the
first-order terms.
4πe2 ne
!
∂2
− c 2 2
∇ + ω 2
pe à = − AL (21)
∂t2 m
This is the equation that relates the scattered wave à with the pump laser AL .
To account for the density perturbation due to the electron-plasma wave (EPW)
- Langmuir wave, we involve the first two-moments of Vlasov equation.

∂ne
+ ∇ · (ne ue ) = 0 (22)
∂t

and
∂ue −e ue × B ∇Pe
 
+ (ue · ∇)ue = E+ − (23)
∂t m c ne m
the velocity, as mentioned before for current density, can be divided into the longitu-
dinal and transverse components, as

eA
ue = ul +
mc

. We use this in the momentum equation above, and also use the vector identity

v2
(v · ∇)v = ∇ − v × (∇ × v)
2

in the second term of eq.(23). This gives us

u2e
(ue · ∇)ue = ∇ − ue × (∇ × ue )
2

The term (∇ × ue ) can be expanded

eA
 
(∇ × ue ) = ∇ × ul +
mc

7
 e e
(∇ × A) =
= B
mc mc
Also, we use the generalised E expression in the right hand side. Thus, we have the
whole eq.(23)
!
∂ul e ∂A u2 e e e ∂A e ∇Pe
+ +∇ e − (ue × B) = ∇ϕ + − (ue × B) − (24)
∂t mc ∂t 2 mc m mc ∂t mc ne m

Cancelling terms that are the same both sides, we reduce this equation to

1
2
∂ul e eA ∇Pe

= ∇ϕ − ∇ ul + − (25)
∂t m 2 mc ne m

To reduce the pressure term in a usual format, we take the equation of state as

Pe
=C (26)
n3e

For the sake of convenience, we use ul = u.

 2
We linearise the term u + eA mc
using A = AL + Ã and consider the fact that
u ⊥ AL .
2
eA e e e2 

2
u+ = u2 + 2 u · AL + 2 u · Ã + 2 2 AL 2 + 2AL · Ã + Ã
mc mc mc mc

Now, we get rid of the second order terms and also use the fact that ∇A2L = 0 (plane
pump wave shouldn’t have a variation in its amplitude). Also, by differentiating the
equation of state, we can get the first order term as δPe = 3Pn0e0 n˜e . Recalling that
thermal velocity can be written as ve2 = mn
Pe0
0
from thermodynamics, we can obtain

∂ũ e e2 3v 2
= ∇ϕ̃ − 2 2 ∇(AL · Ã) − e ∇n˜e (27)
∂t m mc n0

We linearise the first moment of Vlasov equation now. We use the gauge ∇ · A = 0,
and obtain
∂ n˜e
+ n0 ∇ · ũ = 0 (28)
∂t
We differentiate eq.(28) with respect to time, and take a divergence of eq.(27).
Also, we use the Poisson’s equation (eq.15) to obtain the correlation between the
EPW that is generated and its influence on number density ( ponderomotive force
field). !
∂2 n0 e2 2
+ ω 2
pe − 3v e
2 2
∇ ˜
ne = ∇ (AL · Ã) (29)
∂t2 m2 c2
The particle density suffers fluctuations due to the pump and the scattered EM wave,
and the equation above captures this picture, with the Langmuir wave operator on
the left-hand side of the equation.

8
2.2 Dispersion relation
We re-write the two equations again

4πe2 ne
!
∂2
− c 2 2
∇ + ω 2
pe à = − AL
∂t2 m
! (30)
∂2 n0 e2 2
+ ωpe − 3ve ∇ n˜e = 2 2 ∇ (AL · Ã)
2 2 2
∂t2 mc

To convert this into the usual frequency and wave-vector domain, we conduct a fourier
transform on these two equations. We take the pump laser wave as

AL = A0 cos(k0 · x − ω0 t)

We use the Euler’s form cos θ = [exp(iθ) + exp(−iθ)]/2 (since for Fourier analysis we
need the exponential form.) We consider perturbed quantities n˜e and à to have their
respective frequency and wave-vector as ω and k. We thus obtain,

4πe2 A0
(ω 2 − c2 k 2 − ωpe
2
)Ã(k, ω) = [n˜e (k − k0 , ω − ω0 ) + n˜e (k + k0 , ω + ω0 )]
2m
n0 e2 k 2
(ω 2 − ωek
2
)n˜e (k, ω) = A0 · [Ã(k − k0 , ω − ω0 ) + Ã(k + k0 , ω + ω0 )]
2m2 c2
(31)

Here, ωek = (ωpe 2

+3ve2 k 2 )1/2 is the Langmuir frequency (Bohm-Gross wave frequency).
The expressions of down-shifted and upshifted scattered wave amplitude à can be
found by downshifting and upshifting k, ω in the first equation. Writing the expression
(ω 2 − c2 k 2 − ωpe
2
) = D(k, ω), then

4πe2 A0
Ã(k − k0 , ω − ω0 ) = n˜e (k, ω)
2m D(ω − ω0 , k − k0 )
(32)
4πe2 A0
Ã(k + k0 , ω + ω0 ) = n˜e (k, ω)
2m D(ω + ω0 , k + k0 )

These expressions takes care of the fact that oscillation ω ∼ ωpe , so terms with ω ±2ω0
and k ± 2k0 , are too far from the resonance, hence the corresponding oscillations will
quickly die off. This was mentioned previously in the introduction section. So, we
ignore all these terms while obtaining the upshifted and downshifted scattered waves
functions.
Using the above expressions for à in the second equation of eq.(31), we obtain

k 2 ωpe
2 2
1 1
" #
vos
ω 2 − ωek
2
= + (33)
4 D(ω − ω0 , k − k0 ) D(ω + ω0 , k + k0 )

eA20
Here, vos
2
= m 2 c2
is the oscillatory velocity.

9
2.3 The growth rate
Recalling
k0 = ks + ke (34)

Where k0 is the pump wave-vector, ks is the scattered wave-vector, and ke represents

the Langmuir wave. When we try to look at Raman instability, we aim to see at what
conditions the plasma electrons respond the greatest. So, greater the magnitude of
ke , greater should be the corresponding growth rate. By vector-algebra, we have
q
k= k02 + ks2 − 2k0 ks cos(θ) (35)

Clearly, if θ = π, the Langmuir wave has the largest value. We do a magnitude

comparison of Langmuir wave in backward and forward scattering,

• Backward - SRS
Because the wave-vector of the scattered wave is reversed to the pump laser, so
we have
ke = k0 − (−ks )
ω0 −ωpe
Now, ks = c
, and since ωpe ≪ ω0 , so

ke ∼ 2k0

• Forward - SRS Now, the scattered wave-vector ks is in the same direction as

the pump wave k0 , so
ke = k0 − ks
ωpe
ke ∼
c
, which is ≪ k0 .
So, for back-scattering case, while dealing with dispersion relation, we can con-
sider the upshifted terms as too far to be resonant with the EPW. This leads
us to write the dispersion relation as
2 2 2
ωpe k vos
(ω 2 − ωek
2
)[(ω − ω0 )2 − ωpe
2
− c2 (k − k0 )2 ] = (36)
4

We take ω = ωek + δω, δω ≪ ωek . Also, when the left-hand side is very close
to zero, its only the δω terms that will show up solely to count for the non-zero
value from the right-hand side of the equation.
So, with the resonance condition

(ωek
2
− ω0 )2 − (k − k0 )2 c2 − ωpe
2
=0

10
 we obtain the relation

ωpe2 k 2 vos
2
4ωek (ωek − ω0 )(δω)2 =
4

Since ωek is smaller than ω0 , we are supposed to face imaginary roots while
solving for δω. Choosing δω = iγ, we get γ as
" 2
#1/2
kvos ωpe
γ= (37)
4 ωek (ω0 − ωek )

Here, γ is called named as Growth rate of the stability. Using the resonance
condition, the maximum k can be evaluated

2ωpe
s
ω0
k = k0 + 1− (38)
c ω0

So, k goes from the largest value 2k0 to k0 , as the plasma density increases to
n ∼ nc /4, where nc is the density we can affect via laser frequency ω0

3 Simulation

(a) Density distribution in the simulation

box. (b) Seed laser

Figure 1: Global simulation input for density and seed pulse for all cases of varying
thermal temperature considered below.

To observe stimulated Raman scattering in a plasma channel, we conducted a

simulation with counter-propagating long pump laser pulse and short seed pulse being
injected to a region of plasma. We considered a pump laser with a wavelength of 800
nm and a pulse duration of 6 ps, and a seed pulse with an wavelength of 843 nm
that satifies the mathcing condition and has a pulse duration of 33 fs. Both of
these Gaussian pulses have dimensionless vector potential a0 = mc
eA
2 = 0.0216 and an

intensity of 1.0 × 1015 W/cm2 . A plasma slab of 2 mm is considered in normalised

11
plasma time-period units, particle density normalised to 4.5 × 1018 cm−3 [16]. We
take advantage of the code ZPIC 1 for simulations. We only consider 1-dimensional
simulations to study SRS. We restrict ourselves to 10 particles per cell in the box
owing to heavy computational costs.
First, we fix the slab to be cold and stationary, and record the amplification for
the seed pulse. Subsequently, we introduce temperature of 10, 50 and 100 eV in the
plasma to study its influence on the amplification. The results are provided in the
next section.

4 Results and Conclusion

We look at the amplified seed pulse, the growing E field with time, the dispersion
relations and the field energy varies with time for four different setups with varying
temperatures.

Thermal velocity E-field of Probe (normalized units) Intensity (W/cm2 ) Amplification

0 1.441 1.3 × 1016 13
0.006 (10 eV) 1.167 0.85 × 1016 8.5
0.014 (50 eV) 0.9618 0.58 × 1016 5.8
0.020 (100 eV) 0.7667 0.37 × 1016 3.7

Table 1: Simulation parameters for varying thermal velocities. The electric field is
given in normalized units (a0 × ω0 ), intensity corresponds to the amplified laser wave-
length of 843 nm, and amplification factor is dimensionless. Here the amplification is
ratio is calculated based on the initial seed pulse intensity of 1 × 1015 W/cm2

(a) 0 eV (b) 10 eV

(c) 50 eV (d) 100 eV

Figure 2: Amplified seed pulses for different thermal plasma setups

1
[Link]

12
 (a) 0 eV (b) 10 eV

(c) 50 eV (d) 100 eV

Figure 3: Amplification of E-field during SRS with time

(a) 0 eV (b) 10 eV

(c) 50 eV (d) 100 eV

Figure 4: Dispersion relation ω(k) plots

13
 (a) 0 eV (b) 10 eV

(c) 50 eV (d) 100 eV

Figure 5: The growth and saturation of field energy due to SRS instability

We observe that the seed pulses shown in fig.2 gets lesser and lesser amplified,
as temperature increases. The influence of increasing temperature also affects the
amount of time taken by the field energy to saturate in fig.5, which indicates a slower
growth rate of Raman instability. We detect the linear phase of the instability at the
early times of pulse-probe interaction, inside the plasma.
So, we conclude our report with a note of successful detection of SRS on a seed
pulse with frequency close to the pump laser frequency. It was expected from the
theoretical derivation that the growth rate might be lower as the pump laser encoun-
ters a thermal plasma, and the simulation results indicates likewise. To understand
the detailed physics of how temperature is leading to lower growth of the seed pulse
with more clarity, we propose to conduct higher-dimensional simulations in future,
with varying parametric values.

14
References
[1] A. Andreev, “Generation and Application of Ultrahigh Laser Filds.,” New Nova
Science, 2002.
[2] C. Deutsch, A. Bret, M. C. Firpo, L. Gremillet, E. Lefebvre, and A. Lifschitz,
“Onset of coherent electromagnetic structures in the relativistic electron beam
deuterium–tritium fuel interaction of fast ignition concern,” Laser and Particle
Beams, vol. 26, pp. 157–165, 2008.
[3] D. Giulietti, M. Galimberti, A. Giulietti, L. A. Gizzi, L. Labate, and P. Tomassini,
“The laser-matter interaction meets high energy physics: Laser-plasma acceler-
ators and bright x- and gamma-ray sources,” Laser and Particle Beams, vol. 23,
pp. 309–314, 2005.
[4] V. A. Gorbunov, S. V. Papernyi, V. F. Petrov, and V. R. Startsev, “Time
compression of pulses in the course of stimulated brillouin scattering in gases,”
Kvantovaya Elektronika, vol. 10, pp. 1386–1395, 1983.
[5] N. G. Borisenko et al., “Physical processes in laser interaction with porous
low-density materials,” Laser and Particle Beams, vol. 26, pp. 537–543, 2008.
[6] D. C. Eder, P. Amendt, and S. C. Wilks, “Optical-field-ionized plasma x-ray
lasers,” Physical Review A, vol. 45, pp. 6761–6772, 1992.
[7] S. Eliezer, M. Murakami, and J. M. M. Val, “Equation of state and opti-
mum compression in inertial fusion energy,” Laser and Particle Beams, vol. 25,
pp. 585–592, 2007.
[8] H. Hora, “New aspects for fusion energy using inertial confinement,” Laser and
Particle Beams, vol. 25, pp. 37–45, 2007.
[9] H. Hora, “New aspects for fusion energy using inertial confinement,” Laser and
Particle Beams, vol. 25, pp. 327–328, 2007.
[10] M. H. Key et al., “Hot electron production and heating by hot electrons in fast
ignitor research,” Physics of Plasmas, vol. 5, no. 5, pp. 1966–1972, 1998. doi:
10.1063/1.872867.
[11] D. Strickland and G. Mourou, “Compression of amplified chirped optical pulses,”
Optics Communications, vol. 56, no. 3, pp. 219–221, 1985, issn: 0030-4018.
[12] G. Shvets, N. J. Fisch, A. Pukhov, and J. Meyer-ter-Vehn, “Superradiant am-
plification of an ultrashort laser pulse in a plasma by a counterpropagating
pump,” Phys. Rev. Lett., vol. 81, pp. 4879–4882, 22 Nov. 1998.
[13] A. Sharma and I. Kourakis, “Laser pulse compression and amplification via
raman backscattering in plasma,” Laser and Particle Beams, vol. 27, no. 4,
pp. 579–585, 2009.

15
[14] C. S. Liu, V. K. Tripathi, and B. Eliasson, High-Power Laser-Plasma Interac-
tion. Cambridge University Press, 2019.
[15] W. L. Kruer, The Physics of Laser–Plasma Interactions, 1st ed. CRC Press,
2003. doi: 10.1201/9781003003243.
[16] R. M. G. M. Trines et al., “Simulations of efficient Raman amplification into
the multipetawatt regime,” Nature Physics, vol. 7, no. 1, pp. 87–92, Jan. 2011.
doi: 10.1038/nphys1793.

16