Partially Coherent Light

Le Thuy Mai Anh

1. Introduction:
Ideally, light is considered coherent, i.e., the angular velocity ω and angular wavenumber
light k in the expression of light as electromagnetic wave are constant.
In this assumption, the phase relationship of electric field between any positions in the light
path and between any time is definite.
However, the radiation from a normal light source can only be coherent for a period or
distance. The electric fields at time t and t+ Δt will have a definite phase relationship if
Δt ≪ τ c and almost never have any phase relationship for Δt ≫ τ c . We say the field is
coherent for times τ c. The time τ c is called the coherence time of the radiation.
The electric field having coherence time τ c can be written as the superposition of waves with
frequencies in a region Δ ν around the frequency ν=ν 0 , where
1
Δν
τc
Besides temporal coherence discussed above, another type of coherence is spatial
coherence characterized by the distance over which the radiation remains coherent. The
corresponding coherence distance is known as τ c c .

2. Coherence properties:
To study the coherence properties of the radiation field, we will neglect the polarization
effects first. Additionally, the following phenomenon and quantities are discussed in optical
system having two light sources. Practical sources often emit radiation over a narrow band
of frequencies Δ ν about a mean frequency ν 0 such that Δ ν ≪ ν 0. These sources are quasi-
monochromatic sources. Ideally, Δ ν=0 corresponding to monochromatic sources.
To describe the nature of light relating to coherence, we use complex degree of coherence
γ 12. The visibility (or contrast) of the pattern obtained from interference of partially coherent
light is derived from
2 ( I 1 I 2 )1 /2
v= ∨γ 12∨¿
I1 + I 2
where I 1 and I 2 are the intensity of light originating from the two light sources P1 and P2 at
the measuring point.
With complete coherent light |γ 12|=1, and with partially coherent light 0<|γ 12|<1 . Hence, the
complete coherent light gives highest visibility.
When light source is 2D object, the radiation that was incoherent when it started form the
source can becomes partially coherent as it propagates as stated in van Cittert–Zernike
theorem. This can be understood by physical intuition that each point of the source
contributes to each set of points in any subsequent plane.[1] The expression of complex
degree of coherence μ12 between two points on the plane is much more complicated
depending on coordinates of points, dimension of source and main frequency of source.
In experiment, μ12 can be obtained by measuring the variation of the visibility as a function
of the distance from source to detector. Then the angular size of source can be derived from
measured μ12.
3. Polarization:
Polarized wave has the tip of electric field moves following a line or circumference of a circle
or an ellipse. Unpolarized light has state of polarization changes in a random manner with
time. In general, light is partially polarized which can be expressed as the combination of
polarized and unpolarized waves. Monochromatic light is the completely polarized wave,
while quasi-monochromatic light is partially polarized wave. Generally, polarization
properties of light also affect the intensity of pattern.

4. Generation of partially coherent light:

There are two main strategies for generating partially coherent light mentioned in [2]:
The first strategy is to transform spatially incoherent light into a partially coherent beam
based on van Cittert-Zernike theorem. The van Cittert-Zernike theorem states that the
normalized Fourier transform of the intensity distribution of a spatially incoherent source is
equivalent to the complex degree of spatial coherence of the partially coherent wavefront.
Therefore, we can manipulate the spatial coherence structure of beam by controlling the
intensity distribution of the incoherent source.
The second mentioned strategy is to superpose fully coherent but mutually uncorrelated
modes into a single partially coherent beam. This strategy is based on coherent-mode
representation theory.

5. Application:
Even though highly coherent light produces pattern of high contrast, it also meets some
negative effects such as speckle effects in optical imaging. In order to mitigate those negative
effects while still maintain desirable features of light, the coherence of light can be reduced
appropriately. Partially coherent beams are widely applied in holography, interferometry,
optical communication, optical imaging, etc. One of the applications we focus on is phase
imaging.

6. Partial coherence in phase imaging:

6.1. Introduction:
Phase imaging works based on the fact that objects may not absorb light but cause
measurable phase delays. In fact, optical phase changes too rapidly to be directly measured
by cameras. Therefore, there need to be methods to convert phase information into
intensity information. One of those methods is using optical system with a complex transfer
function. Complex transfer functions will result in phase contrast at the imaging plane
instead of the desired phase of object. Experimentally, phase contrast is obtained by defocus
means, then some different methods can be used to retrieve phase from the phase contrast.

Measured (intensity) pattern is generated by structure of object such as refractive index,

thickness, density, etc. and by absorption of object. If the absorption effects and phase
information can be separated, the resulting phase contrast is said to be linearly related to
the actual phase, and the phase imaging is called quantitative phase imaging. In such case,
phase contrast will relate directly to object’s physical properties.
Coherent light’s intensity can be expressed with 2 components in the form of a complex
function, while partially coherent light’s intensity needs 4 components to expressed
resulting in a 4D function.

As discussed earlier in section 1, partially coherent light does not have definite phase
relationship, or well-defined phase, like in coherent case. Hence, a definition of phase of
partially coherent light is needed and specified as “a statistical ensemble of phases at every
point in space”.[3] In experiment, the measured phase map shows the phase changes
caused by object’s physical properties, which has well-defined phase. This result can be
interpreted as the phase map would be obtained if the object had been illuminated by
monochromatic, coherent light of the mean wavelength of the partially coherent light used.

6.2. Methods:
Previously, the retrieved phase is emphasized to be phase contrast, or phase difference,
instead of absolute phase of the beam. To attain the latter phase, propagation-based
intensity methods are used with the assumption that coherent and partially coherent light’s
propagation transfer function are alike. The propagation of partially coherent light can only
be described uniquely if we know its full 4D coherent functions, but the phase map we need
is 2D quantity only, so we avoid measuring full 4D quantities.

One of the methods is using Transport of Intensity Equation (TIE). This method aims to
recover phase directly from the measurement of the derivative of intensity with respect to
the propagation direction based on
∂I
∂z
=−( )λ
2π
∇⋅ I ∇φ

This derivative can be estimated from two images taken at different defocus distances. TIE
offers a direct and linear solution for phase which can be computed in real-time. When there
are more than two images taken, the result of phase can be corrected for higher order
nonlinearities. However, the method meets inaccuracies from non-linearity with large
defocus or presence of noise.

Another method is iterative method, based on Gerchberg-Saxton algorithm, where an in-

focus image and a Fourier domain image obtained at defocus position. Phase retrieval is
then treated as a non-linear convex optimization problem. The algorithm iteratively bounces
back and forth between two domains, with some imposed physical constraints, to update
the estimation of complex expression, i.e., amplitude and phase of image using measured
data or previous step’s result. Unlike TIE, small defocus is not required. In contrary, larger
defocus distances lead to better diffraction contrast, hence better estimation. Although the
problem is not convex, with appropriate constraints and multiple measurement at different
distances can give correct and unique solution.

The above methods’ properties are discussed for coherent source. However, with partially
coherent illumination it is more complicated to draw conclusion. Intuitively, coherence
effects increase with larger propagation distance, so the error is expected to increase more
steeply with iterative method comparing to TIE method.
References:
[1] A. K. Ghatak, Contemporary Optics, 1st ed., Springer, 1978.
[2] J. Yu, X. Zhu, F. Wang, Y. Chen, and Y. Cai, “Research progress on manipulating spatial
coherence structure of light beam and its applications,” Prog. Quantum Electron. 91, 100486
(2023), [Link]
[3] L. Waller, “Phase imaging with partially coherent light,”, University of California, Berkeley,
CA, USA, 94702.