Gaussian Beams and the Knife-Edge Measurement

Gaussian Beams:
Gaussian beams propagating in the z-direction may be represented mathematically as:
2
𝑒 𝑖𝑘𝑟 /2𝑞
𝑈 = 𝑈0 𝑒 𝑖(𝑘𝑧−𝜔𝑡) where q = z - 𝑧𝑤𝑎𝑖𝑠𝑡 − 𝑖𝑧𝑅 (1)
𝑞

Here U is the electric field amplitude of the wave, q is known as the ‘complex radius’, 𝑧𝑤𝑎𝑖𝑠𝑡 is a
real constant indicating the position of the beam waist and 𝑧𝑅 known as the ‘Rayleigh range’.
Substituting the expression for q into the equations for U, and assuming that the waist is at the
origin (𝑧𝑤𝑎𝑖𝑠𝑡 = 0), it is straight forward to show that:
𝑈0 𝑖
U= - Amplitude on axis at waist (constant)
𝑧𝑅

0 𝑤
× 𝑤(𝑧) - Amplitude gets smaller as beam spreads

× 𝑒 𝑖(𝑘𝑧−𝜔𝑡) - Wave travelling to positive z

× 𝑒 −𝑖𝛼(𝑧) - Phase change with z

2 /𝑤(𝑧)2
× 𝑒 −𝑟 - Gaussian profile
2 /2𝑅(𝑧)
× 𝑒 𝑖𝑘𝑟 - Curved wavefronts
Where the 1/𝑒 2 radius in the intensity distribution or ‘spot size’ w(z) obeys:

𝑧2 𝜋𝑤02
𝑤(𝑧) = 𝑤0 √(1 + 𝑧 2 ) with 𝑧𝑅 = (2)
𝑅 𝜆

and the radius of curvature of the wavefronts R(z) obeys:

2
𝑧𝑅
R(z) = 𝑧 + (3)
𝑧

and the phase factor (often referred to as Guoy’s phase) takes the form:
𝛼(𝑧) = 𝑡𝑎𝑛−1 (𝑧/𝑧𝑅 ) (4)
The Gaussian beam is therefore narrowest at z = 0 (or z = 𝑧𝑤𝑎𝑖𝑠𝑡 ) where it is said to have a ‘waist’
of 𝑤0 , as illustrated in Fig. 1. Away from the waist the beam spreads with a hyperbolic outline, the
asymptotes of which define the far field beam divergence such that the half-angle at the 1/e2 radius
in the intensity distribution is given by:
𝑤(𝑧) 𝜆
𝜃 = lim = (5)
𝑧→∞ 𝑧 𝜋𝑤0

At the position of the beam waist the wavefront is planar corresponding to an infinite radius of
curvature. Away from the waist, where the beam diverges (or converges) the wavefronts are of
course curved. The Rayleigh range 𝑧𝑅 relates to the distance over which a Gaussian beam can be
1
collimated before it spreads significantly due to diffraction. In particular, the Rayleigh range is the
distance which the beam travels from the waist before the beam diameter increases by √2 or before
the beam area doubles in size. The Rayleigh range marks the approximate dividing line between
the ‘near-field’ or Fresnel and the ‘far-field’ or Fraunhofer regions for a beam propagating out
from a Gaussian waist. Note: text books often refer to the ‘Confocal Parameter’ b = 2zR, so named
because it is equal to the length of a symmetrical confocal mirror cavity into which the beam fits.
It is striking that the entire range of properties of the Gaussian beam depend solely on the size of
the waist 𝑤0 and the wavelength λ. It is therefore possible to fully characterize a Gaussian beam
by determining the size and location of the beam waist. This is readily achieved by making a series
of ‘knife edge’ measurements to determine the spot size as a function of the distance z.

Figure 1: Notation used in describing a Gaussian Beam.

Beam width measurement using the Knife Edge:

Method:
Record the total power in the beam as a knife edge is translated through the beam using a
calibrated translation stage. The power meter records the integral of the Gaussian beam between -
∞ and the position of the knife.
Analysis:
Assume a beam propagating in the z-direction with a Gaussian intensity profile.
2 /𝑤 2 2 /𝑤 2
𝐼(𝑥, 𝑦) = 𝐼0 𝑒 −2𝑥 𝑥 𝑒 −2𝑦 𝑦 (6)
where 𝑤𝑥 and 𝑤𝑦 are the 1/e2 radii of the beam in the x and y directions respectively. I0 is the peak
intensity.
The total power in the beam is:
∞ 2 /𝑤 2 ∞ 2 /𝑤 2 𝜋
𝑃𝑇𝑂𝑇 = 𝐼0 ∫−∞ 𝑒 −2𝑥 𝑥 𝑑𝑥 ∫−∞ 𝑒 −2𝑦 𝑦 𝑑𝑦 = 𝐼 𝑤 𝑤
2 0 𝑥 𝑦
(7)

2
Consider the knife edge being translated in the x-direction as shown in Fig. 2. The transmitted
power is then:

Figure 2: Principle of the knife edge measurement.

𝑋 2 /𝑤 2 ∞ 2 /𝑤 2
P(X) = 𝑃𝑇𝑂𝑇 = 𝐼0 ∫−∞ 𝑒 −2𝑥 𝑥 𝑑𝑥 ∫−∞ 𝑒 −2𝑦 𝑦 𝑑𝑦

𝜋 𝑋 2 /𝑤 2
= 𝑃𝑇𝑂𝑇 − √2 𝐼0 𝑤𝑦 ∫−∞ 𝑒 −2𝑥 𝑥 𝑑𝑥

𝜋 0 2 /𝑤 2 𝑋 2 /𝑤 2
= 𝑃𝑇𝑂𝑇 − √2 𝐼0 𝑤𝑦 [∫−∞ 𝑒 −2𝑥 𝑥 𝑑𝑥 + ∫0 𝑒 −2𝑥 𝑥 𝑑𝑥]

𝜋 𝜋 𝑋 2 /𝑤 2
= 𝑃𝑇𝑂𝑇 − √2 𝐼0 𝑤𝑦 [√8 𝑤𝑥 + ∫0 𝑒 −2𝑥 𝑥 𝑑𝑥]

𝑃𝑇𝑂𝑇 𝜋 𝑋 2 /𝑤 2
= − √2 𝐼0 𝑤𝑦 ∫0 𝑒 −2𝑥 𝑥 𝑑𝑥 (8)
2

Consider now the integral in Eq. 8 which we wish to cast in a standard form. Making the
substitution 𝑢2 = 2𝑥 2 /𝑤𝑥2 so that 𝑑𝑥 = 𝑤𝑥 𝑑𝑢/√2 and making the necessary change to the limits
of the integral leads to:
√2𝑋
𝑃𝑇𝑂𝑇 𝜋 𝑤𝑥 2 𝑤𝑥
P(X) = − √ 2 𝐼0 𝑤𝑦 ∫0 𝑒 −𝑢 𝑑𝑢
2 √2

√2𝑋
𝑃𝑇𝑂𝑇 𝜋 2 𝑤𝑥 2
= − 𝐼0 𝑤𝑦 𝑤𝑥 ∫0 𝑒 −𝑢 𝑑𝑢 (9)
2 4 √𝜋

Using the standard definition of the Error Function listed in the Appendix and the expression in
Eq. 9 for the total power in the beam we arrive at our final result:
𝑃𝑇𝑂𝑇 √2𝑋
P(X) = 2
[1 − 𝑒𝑟𝑓 ( 𝑤 )] (10)
𝑥

3
Fitting:
To fit the data in Origin, use a fitting function of the form:
𝑷𝟏 √2(𝑋−𝐏𝟐)
𝑃𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑 = [1 ± 𝑒𝑟𝑓 ( )] (11)
2 𝐏𝟑

Where P1 corresponds to the power, P3 corresponds to the 1/e2 radius of the Gaussian beam and
the – (+) is chosen when the knife is translated in the positive (negative) direction.
Shorthand method for determining beam widths
Taking a full set of knife edge measurements and fitting as above can be tedious and time
consuming. Therefore, in cases where only the width of the beam, rather that its full profile, is
required (for example, when making measurements of width with distance from a source/a lens in
order to determine the focal point of the beam), it is quicker to use the 90% - 10% method outlined
below.
Method:
Measure the total power in the beam when fully exposed. Translate the knife edge across
the beam and measure the distance between the points at which the power output is 10% and 90%
of this value, X10 and X90 respectively. This distance is X10-90.
Analysis:
According to Eq. 10, at 10% power:
𝑃𝑇𝑂𝑇 √2𝑋10
0.1 𝑃𝑇𝑂𝑇 = [1 − 𝑒𝑟𝑓 ( )] (12)
2 𝑤𝑥

Rearranging gives
√2𝑋10
𝑒𝑟𝑓 ( ) = 0.8 (13)
𝑤𝑥

Using Eq. 13, this can be related to the Gaussian probability so that
√2𝑋10 𝑡10
𝑃(𝑡10 ) = 0.9 where = (14)
𝑤𝑥 √2

Using standard probability tables, this gives

𝑡10 = 1.28 (15)
Therefore from Eq. 14
𝑋10 = 0.64𝑤𝑥 (16)
From the symmetry of the Gaussian function it is straightforward to see that:
𝑋10−90 = 2 × 0.64𝑤𝑥 = 1.28𝑤𝑥 (17)

4
Thus there is a ‘calibration’ from which the 1/e2 radius may be determined by making two
measurements only. A similar treatment could be applied to X80 and X20 etc. However, the 90% -
10% method is favored as it uses points at the extremes of the region of maximum change in the
Error Function (see Fig. 4).
Real Gaussian Beams and M2
Real laser beams will deviate from the ideal Gaussian. A measure of their quality is given by M 2
which is defined such that
M2 = 1 for an ideal Gaussian beam M2 > 1 for a real beam
The definition of M2 :
Consider a Gaussian beam propagating from a source. In the far-field region beyond the
Rayleigh range, from Eq. 5 and Fig. 2:
𝑤(𝑧) 𝜆
𝜃 = lim = (18)
𝑧→∞ 𝑧 𝜋𝑤0

Consider the product of this with the waist 𝑤0 :

𝜆
𝑤0 𝜃 = (19)
𝜋

This is a lower fundamental limit which can only be approached by ideal Gaussian beams. For a
real beam
𝜆
𝑤0 𝜃 = 𝑀2 (20)
𝜋

Therefore, the Rayleigh range of a real beam can be rewritten

𝜋𝑤02
𝑧𝑅,𝑟𝑒𝑎𝑙 = (21)
𝜆𝑀2

Method for measuring M2 :

The setup is as in Fig. 3. Using the 90% - 10% method, the beam width w(z) at least 6
points across the range shown should be measured.
Fitting:
According to Eq. 2 and applying an arbitrary offset to take account of the initial position
of the knife edge, the fitting equation to be used is:
(𝑥−𝑷𝟐)2
𝑤(𝑧)2 = 𝑃1 (1 + ) (22)
𝐏𝟑

where (23)
Using these

5
Figure 3: The set up for determining M2 by measuring the beam width with distance from a lens.

Appendix: The Gaussian Distribution and the Error Function

The Normal or Gaussian distribution function is defined as:
1 2 /2
∅(𝑡) = 𝑒 −𝑡 (24)
√2𝜋

where the pre-factor ensures the correct normalization. Statistical tables give the value of
𝑡
𝑃(𝑡) = ∫−∞ ∅(𝑡)𝑑𝑡 (25)

A perhaps more useful definition includes the width 𝜎 and non-zero mean 𝜇 of the distribution
explicitly :
1 2 /2𝜎 2
𝛷(𝑡) = 𝑒 −(𝑥−𝜇) (26)
𝜎√2𝜋

The full width at half maximum is then related to 𝜎 by :

𝐹𝑊𝐻𝑀 = 2√2𝑙𝑛2𝜎 ≃ 2.35𝜎 (27)

Unfortunately Gaussian beams are defined in yet another way, namely in terms of the 1/e2 radius
or ‘spot size’ 𝑤 :
2𝑃 2 /𝑤 2
𝐼(𝑟) = 𝑒 −2𝑟 (28)
𝜋𝑤 2

where P is the total power in the beam. Comparison of Eqs. 27 and 29 shows that 𝑤 = 2𝜎.
The Error Function is defined as follows:
2 𝑥 −𝑢2
𝑒𝑟𝑓(𝑥) = ∫ 𝑒 𝑑𝑢 (29)
√𝜋 0

and has the properties that:

𝑒𝑟𝑓(∞) = 1 and 𝑒𝑟𝑓(−𝑥) = −𝑒𝑟𝑓(𝑥) (30)

6
The Gaussian probability distribution is related to the error function by:
1 1 𝑥
𝑃(𝑥) = + 𝑒𝑟𝑓 ( ) (31)
2 2 √2

Figure 4: Plot of the Error Function and the corresponding normalized Gaussian distribution
showing both 𝝈 and the ‘spot size’ w.

7
 LASER BEAM PARAMETERS
(KNIFE EDGE)
Some of the important external beam parameters that will be studied are:
1.1 Power distribution within the beam.
1.2 Spot size of the beam.
1.3 Divergence of the beam.
1.4 Coherence
1.4.1 Spatial coherence
1.4.2 Temporal coherence.
1.5 Modes of a laser

1.1 Power Distribution within the Beam

Power distribution within the beam can be studied by a number of techniques viz.
photographic technique, scanning technique etc. we shall, however, discuss a
method which is very simple to perform in the laboratory. Further the spot-size and
the divergence of the beam can also be measured with this method. The method
involves the measurement of the power past a knife-edge which is slowly inserted
in the beam. In order to understand the principle of the method, let us assume that
the laser is oscillating in TEM00 mode so that the spatial distribution of the beam is
Gaussian. Let P0 be the total power in the beam of spot size 2w0. The irradiance
distribution I(x, y) as a function of the Cartesian coordinates (x, y) measured from
the beam center perpendicular to the direction of propagation is given by
2𝑃0 2(𝑥 2 + 𝑦 2 )
𝐼(𝑥, 𝑦) = exp(− ) (1.1)
𝜋𝑤02 𝑤02

The power P transmitted past a knife-edge blocking off all points for which 𝑥 ≤ 𝑎
is, therefore, given by
∞ ∞ 𝑃0 𝑎 √2
P = ∫−∞ ∫𝑎 𝐼(𝑥, 𝑦)𝑑𝑥 𝑑𝑦 = 𝑒𝑟𝑓𝑐 ( ) (1.2)
2 𝑤0

Where a is the depth of knife-edge in the beam. Therefore, the integrated power past
the knife-edge inserted in a Gaussian beam is given by the complementary error

8
function. For other spatial distributions, an integrated power curve can be obtained
from which the power distribution in the beam can be realized. It may be noted that
this method uses information from all the points of the irradiance distribution in the
beam unlike scanning technique.
Experiment 1.1: Determination of the power distribution within the beam.
Equipment: A gas laser, a knife-edge mounted on a micropositioner, a
photodetector of a large linear range, a lens, a bench.
Procedure: The schematic of the experimental set-up is shown Fig. 1.1. The knife-
edge is mounted normal to the beam at any desired plane. A lens is used to focus the
laser beam past the knife-edge on the photo-detector. The lens is of sufficiently large
aperture as to gather all the diffracted light and is placed close to the knife-edge. An
interference filter is mounted in front of the detector. The experiment can be
conducted in a relatively well lit room now.

Fig. 1.1 Study of Power distribution in the Beam

The knife-edge is manually inserted in the beam and corresponding output of the
detector is noted. The output of the detector is plotted as a function of the position
of the knife-edge. This gives a curve which represents one-dimensional integrated
power distribution. Fig. 1.2 shows such a curve when the knife-edge is (i) inserted
and (ii) withdrawn in the beam.

9
 Fig. 1.2 Plot of Relative Power vs Knife-edge Position
One-dimensional integrated power distribution can be then plotted in a straight
forward way.
The irradiance distribution in the beam can be constructed from the integrated power
distribution curve.

1.2 Spot Size of the Beam

Consider the irradiance distribution represented by Eq. (1.1): w0 is the radius at
which irradiance falls to e-2 times its central value, 2w0 is therefore taken as the spot
size.
A Gaussian beam remains Gaussian as it propagates in vacuum or in a homogeneous
medium. The output of a laser oscillating in TEM00 mode is Gaussian, and the spot
size 2w0 refers to the planer wave front. At any other plane the wave front will be
either converging or diverging and will have spot size larger than 2w0. It is
meaningful to measure the spot size for a Gaussian beam only Eq. (1.1) can be cast
into a well-known normalized Gaussian distribution be setting. w0 = 2w’, as
𝐼(𝑥,𝑦) 1 (𝑥 2 + 𝑦2 )
= 𝑒𝑥𝑝 {− } (1.3)
𝑃0 2𝜋𝑤 ′ 2 2𝑤 ′ 2

where w’ is the standard deviation.

Similarly, Eq. (1.2) can be written as

10
 𝑃 1 𝑎
= 𝑒𝑟𝑓𝑐 ( ) (1.4)
𝑃0 2 √2𝑤 ′

(a)

(b)
Fig 1.3
The normalized Gaussian distribution [Eq. (1.3)] and the complementary error
function [eq. (1.4)] are illustrated in Fig. 1.3 (a) and (b) respectively. It is easy to
show that the points for 25% and 75% relative powers are located at distance equal
to the probable error (ep = 0.6745 w’) on either side of the maximum of the gaussian
distribution. Therefore, w’ can be determined from the experimentally obtained
relative power vs knife-edge position curve, the beam spot size 2w0 (=4𝑤 ′ ) then is
easily calculated.

11
Experiment 1.2: To measure the beam-spot size.
Equipment: Same as in experiment (1.1)
Procedure: Following the procedure of experiment (1.1), relative power vs knife-
edge position plot is obtained. In one such experiment following data is obtained.
Location of knife-edge from the laser mirror = 10cm. (take atleast four distance
reading i.e. 30cm, 45cm, 60cm, 75cm).

S.No. Knife-edge position (micrometer reading Power meter reading

on the micro positioner) mm (mw)
1. 15.00 0.
2. 15.20 0.03
3. 15.40 0.14
4. 15.60 0.50
5. 15.80 0.98
6. 16.00 1.44
7. 16.20 1.50

A graph between position and power meter reading is plotted in Fig. 1.4.

Fig. 1.4 Plot of Power Output vs Knife-Edge Position

12
From the graph, it is found that the positions corresponding to 25% and 75% power
output are 15.55 mm and 15.85 mm respectively. This gives a probable error of 0.15
mm. Therefore, w’ = 0.15/0.6745 = 0.22mm. The beam spot size at a plane 10cm
from the laser output mirror is thus 4 x 0.022 = 0.88mm.
1.3 Divergence of the Laser Beam
We shall again assume that the laser is oscillating in TEM 00 mode; its output is a
Gaussian beam. The Gaussian beam has the least divergence and maximum power
contained in the prescribed area of the beam. It can be shown that the radius w(z) of
the beam varies with z (taken along the direction of propagation) as

𝜆𝑧 2 1/2
w(z) = w0 [1 + ( 2) ] (1.5)
𝜋𝑤 0

where 2w0 is the waist diameter (diameter where the wave front is plane) of the beam
and λ is the wavelength of laser radiation.
𝑑𝑤(𝑧)
The gradient [= 𝜃(𝑧)] of the beam radius locus at a distance z.
1

The angle 𝜃(𝑧) varies with z. However when z → ∞, 𝜃(𝑧) = 𝜃(∞) → 𝜃0 , where
𝜆
𝜃0 = (1.7)
𝜋𝑤0

𝜃0 is the half divergence angle of the beam. The determination of 𝜃0 requires the
measurement of w0. Following Eq. (1.5) the waist radius w0 can be expressed in
terms of the measured beam radii w(z1) and w(z2) at planes z1 and z2 respectively as
1 1
𝜋
𝑧2 − 𝑧1 = 𝑤0 [{𝑤 2 (𝑧2 ) − 𝑤02 }2 − {𝑤 2 (𝑧1 ) − 𝑤02 }2 ] = 𝐷 (1.8)
𝜆

where D is the separation between the two planes. The divergence 𝜃0 is therefore
given by
𝜆2 {𝑤(𝑧2 )−𝑤(𝑧1 )}2 1/2 𝑤(𝑧2 )−𝑤(𝑧1 )
𝜃0 = [ + ] = (1.9)
𝜋2 𝑤 2 (𝑧1 ) 𝐷2 𝐷

This expression of 𝜃0 will be used for the experiment of the divergence of the beam.
Another simple method to determine 𝜃0 is to insert a lens of precisely known focal
length f in the beam and measure the radius of focal spot 𝑤𝑓 . The divergence 𝜃0 is
given by

13
 𝑤𝑓
𝜃0 = (1.10)
𝑓

Experiment 1.3: To measure the divergence of a laser beam.

Equipment: Same as in Experiment 1.1
Procedure: Following the procedure described in experiment (1.2), the beam spot
size is measured at two different planes. The separation between the planes is
measured with a meter scale. The divergence angle is then calculated from Eq. (1.9).
The procedure can be repeated by measuring spot sizes at additional planes.
1.4 Coherence
A sinusoidal wave exists in space for all the time, and therefore does exhibit prefect
spatial the temporal coherence. However, both perfect spatial and temporal
coherence are mathematical idealization. A real source emits waves that are limited
both in time and space. The concept of spatial and temporal coherence is an involved
one but it is possible to study them independently. Consider an arbitrary point source
emitting spherical waves. Any two points on the spherical surface (or a plane surface
at larger distance from the point source) will have fixed phase relationship and the
spatial coherence exists.

14