Chirped Pulse Transform
Chirped Pulse Transform
Chirped Pulse Transform
David J. Gibson
March 17, 2006
A linearly-chirped electric field has a frequency that changes linearly with time
dω
ω(t) = ω0 + t = ω0 + bt
dt
A Gaussian pulse with a frequency chirp can therefore be represented by
2 2 2
E(t) = E0 e−at e−iω(t)t = E0 e−at −i(ω0 +bt)t
= E0 e−(a+ib)t −iω0 t
To find the spectrum of this pulse we take the Fourier transform, defined as
Z ∞
1
f˜(ω) = F[f (t)] ≡ √ f (t)eiωt dt
2π −∞
Plugging in the electric field above,
Z ∞
1 2
Ẽ(ω) = √ E0 e−(a+ib)t −iω0 t eiωt dt
2π −∞
Z ∞
1 2
=√ E0 e−[(a+ib)t −i(ω−ω0 )t] dt
2π −∞
To solve this integral, we can complete the square in the argument of the exponent:
1
Now, we change variables
√ i(ω − ω0 )
u≡ a + ib t − √
2 a + ib
du √
= a + ib
dt
So we have i(ω−ω0 )
−(ω−ω0 )2
Z ∞−=[ √ ]
1 2 a+ib 2 du
Ẽ(ω) = √ E0 e 4(a+ib) e−u √
2π i(ω−ω0 )
−∞−=[ 2√a+ib ] a + ib
h i 2
which is a complex integral over the line =[u] = −< 2(ω−ω )
√ 0 . Because the integrand, e−u , is everywhere
a+ib
analytic, we can use Cauchy’s Theorem to move the integral to the real axis. Cauchy’s Theorem states that
the integral of a complex function that is analytic in the domain D over the piecewise-smooth closed contour
C in D is 0: I
f (z) dz = 0
c
so in our case, tracing out a box,
I Z p Z p+iq Z −p+iq Z −p
2 2 2 2 2
e−u du = e−u du + e−u du + e−u du + e−u du = 0
c −p p p+iq −p+iq
Looking at the integrals over the line <[u] = p, we can make the substitution
u = p + ix
du
=i
dx
and so the integral becomes
Z p+iq Z q Z q
2 2 2 2
e−u du = ie−(p+ix) dx = i e−p e−2ipx ex dx
p 0 0
because the second and third terms are finite, and the first term goes to zero. Because the integrand goes
to zero, we know
Z p+iq
2
lim e−u du = 0
p→∞ p
with a similar result for the line <[u] = −p. Now our contour integral becomes
Z p Z p+iq Z −p+iq Z −p
2 2 2
−u2
0 = lim e−u du + e−u du + e−u du + e du
p→∞ −p p p+iq −p+iq
Z ∞ Z −∞+iq
−u2 2
= e du + e−u du
−∞ ∞+iq
Z ∞ Z∞+iq
2 2
e−u du = e−u du
−∞ −∞+iq
and so we can rewrite our electric field spectrum as an integral on the real axis
Z ∞
1 −(ω−ω0 )2 2
Ẽ(ω) = √ √ E0 e 4(a+ib) e−u du
2π a + ib −∞
2
What remains now is to evaluate the integral. We start with
Z ∞ sZ
∞ Z ∞
2
−x 2
e dx = e −x dx e−x2 dx
−∞ −∞ −∞
sZ
∞ Z ∞
= e−x2 dx e−y2 dy
−∞ −∞
sZ
∞ Z ∞
= e−(x2 +y2 ) dx dy
−∞ −∞
1 −(ω−ω0 )2
Ẽ(ω) = √ √ E0 e 4(a+ib)
2 a + ib
and the spectral intensity is
1 −(ω−ω0 )2 1 −(ω−ω0 )2
∗
I(ω) = Ẽ(ω)Ẽ (ω) = √ √ E0 e 4(a+ib) √ √ E0 e 4(a−ib)
2 a + ib 2 a − ib
1 −(ω−ω0 )2 −(ω−ω0 )2
= √ E02 e 4(a+ib) + 4(a−ib)
2 a2 + b2
−a(ω−ω0 )2
1
= √ E02 e 2(a2 +b2 )
2 a2 + b2
We can write the FWHM pulse length and spectrum for these Gaussian intensities:
s
−ρ ∆x
2 1 ln 2
e 2 = → ∆x = 2
2 ρ
and therefore
√
√
r r h r h
2 ln 2 b 2
i 2 ln 2 b 2
i
∆tFWHM = ∆ωFWHM = 2 2 ln 2 a 1 + a ∆νFWHM = a 1+ a
a π