Study of Number-Phase Uncertainty Relation

using Polarization Domain Interferometry

A thesis Submitted
in Partial Fulfilment of the Requirements
for the Degree of

MASTER OF SCIENCE

By

Rohan Thomas Mathew

Under the Guidance of

Dr. Ashok Mohapatra

Associate Professor

School of Physical Sciences

National Institute of Science, Education and Research (NISER), Bhubaneswar

to the

School of Physical Sciences

National Institute of Science Education and Research
Bhubaneswar

May, 2024
To My Parents And My Sister,

For being there by my side.

 Declaration

Ihereby declare that I am the sole author of this thesis in partial fulfillment
of the requirements for a postgraduatedegree from National Institute of Science
Education and Research (NISER). Iauthorize NISER to lend this thesis to other
institutions or individuals for the purpose of scholarly research.

Signature of the Student

Date: 15 l05/Q024

The thesis work reported in the thesis entitled "Study of Number-Phase Un

certainty Relation using Polarization Domain Interferometry" was car
ried out under my supervision, in the School of Physical Sciences at NISER,
Bhubaneswar, India.

Signature of the thesis supervisor

School: PHYSTAL SCtENCES

Date: 15|o5|2.024

ii
 Acknowledgement

To begin with, I would like to thank my supervisor Dr. Ashok K Mohapatra for
suggesting me this topic and exposing me to the world of quantum and non- linear
optics. I am extremely grateful for his guidance throughout my project and the
many enlightening discussions we had related to this topic. His patience with my
mistakes and his understanding nature have helped me have more honest and real
conversations on my understanding of this topic.

I would like to thank my lab members Dr. Surya Narayan Sahoo, Nandini Mon-
dal, Sujit Garain and Tausif Alam, Purvesh Garge. My project would have been
incomplete without the help of Dr. Surya, who had helped me throughout and was
a mentor to me for the project, I am grateful for all the insightful conversations we
had and the constant support he has given me. I would also like to thank all the
other members of the lab who equally participated in my journey of completing
the project.

I would also like to offer my gratitude to the institute NISER, for all the courses
which has laid the foundations of my understanding of physics. I am also very
grateful for all the facilities provided be it in the lab or campus which led to the
smooth functioning of the project. I also thank the scientific officers of the teach-
ing lab for giving me the permission to borrow all the optical components.

To end this, I would like to thank my family for their support and love. I would
also like to thank Diya for her sincere support throughout and for helping me proof
read my thesis. I also thank my friends who helped and guided me throughout my
NISER journey.

iii
 Abstract

The number-phase uncertainty relation is a fundamental concept in quantum me-

chanics, describing the inherent limitations in simultaneously measuring the num-
ber of photons in a quantum system and the phase corresponding to the optical
state of light. In this dissertation, I have initially proposed an experiment to ob-
tain the number-phase uncertainty relation using phase shift between polarization
modes. A half-wave plate and polarizing beam-splitter with a poor extinction ratio
is used to prepare a superposition state of two polarization modes. A variable-
phase retarder made out of combination of wave-plates creates the phase shift be-
tween the modes as a function of which we measure the polarization quadratures
using balanced detection. Using a number difference operator Ô, we obtain the
mean and uncertainty in photocurrent difference in the two detectors. It is found
that the phase uncertainty depends on the relative phase θ, whereas the uncer-
√
tainty in ∆Ô is independent of the phase. Also ∆θ ∝ 1/ n̄, where n̄ is the mean
photon number of the incident coherent state. But due to lack of waveplates, a
phase shifter made using a combination of polarizing beam splitter essentially gives
a phase shift between both the interfering beams. The new experimental setup as
shown in 4.2 is almost a direct mapping of the setup given in [12]. Since the power
of the signal in the setup ≈ 6µW , the times scale at which the mean photon num-
ber of the signal is roughly one (n̄ ≈ 1) is T ≈ 10−14 s =⇒ Frequency ≈ 1014 Hz.
Hence in our case we have to use a detector that should work in this frequency
regime, which is impossible. This experiment performed is a basic tool to learn
the operation techniques of handling the equipment, optical components, the data
acquisition techniques and to understand the noise characteristics of the detector
as well as the other sources of noise so that we can operate it at the appropriate
power level that can probe the number phase uncertainty relation.

iv
Contents
Acknowledgement iii
Abstract iv
Page
1 Introduction 1
1.1 Theoretical Background . . . . . . . . . . . . . . . . . . . . . . . . 3
1.1.1 Number Phase Uncertainty . . . . . . . . . . . . . . . . . . 3
1.2 Coherent State: phase space portrait . . . . . . . . . . . . . . . . . 4
1.3 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2 Literature Review 10
2.1 Jones Matrix for Waveplates . . . . . . . . . . . . . . . . . . . . . . 10
2.1.1 The Jones matrix for the following waveplates . . . . . . . . 11
2.2 Combination of Waveplates as a Phase Shifter . . . . . . . . . . . . 12
2.3 Combination of Polarizing Beam Displacers as a Phase Shifter . . . 13
3 Experimental Proposals and Calculations 19
3.1 Initial Proposal of the Experiment . . . . . . . . . . . . . . . . . . . 19
3.1.1 Step 1: State Preparation . . . . . . . . . . . . . . . . . . . 20
3.1.2 Step 2: Transformation due to the polarizing beam splitter
PBS1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
3.1.3 Step 3: Transformation from Phase Shifter element Jps (θ) . 22
3.1.4 Step 4: Half wave Plate . . . . . . . . . . . . . . . . . . . . 22
3.1.5 Step 5: Balanced Homodyne Detection . . . . . . . . . . . . 23
3.1.6 Mean and Variance . . . . . . . . . . . . . . . . . . . . . . . 24
3.1.7 Noise dependence on parameters . . . . . . . . . . . . . . . 29
3.2 Advantages of this experimental proposal . . . . . . . . . . . . . . . 31
3.3 Shortcomings to the initial experimental proposal . . . . . . . . . . 31
4 Experimental Setup and Procedure 33
4.1 Steps for interferometery . . . . . . . . . . . . . . . . . . . . . . . . 35
4.1.1 Step 1: Fixing State polarization by polarizing beam splitter
PBS1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
4.1.2 Step 2: Rotating H0 to control the power of both beams . . 36
4.1.3 Step 3: Setting up of the interferometer . . . . . . . . . . . . 37
4.1.4 Step 4: Balanced Homodyne Detection . . . . . . . . . . . . 39
4.2 Data Acquisition Technique and Observations . . . . . . . . . . . . 41
4.2.1 Data acquisition technique . . . . . . . . . . . . . . . . . . . 42
5 Observations 46
6 Results and Analysis 48

v
 CONTENTS

6.1 Noise Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50

7 Conclusions and Future prospects 57

vi
List of Figures
1.1 A classical wave given by E = E0 cos(ωt + θ). Figure adapated from
[8] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.2 Phase space portrait of a coherent state. Figure adapted from [1]. . 6
1.3 The experimental setup described in the paper. Figure directly
taken from [12]. In this setup, PBS stands for a polarizing beam
splitter, λ/2 refers to a half-wave plate, PZT represents a piezoelec-
tric translator, and ND denotes a neutral-density filter. . . . . . . . 8
2.1 Combination of waveplates acting as an effective phase shifter . . . 12
2.2 A +45◦ polarized light beam is divided into two beams and later
recombined by a pair of Polarization Beam Displacers (PBDs). . . 13
2.3 Inserting a half-wave plate in a polarization interferometer serves
to balance the path lengths of the two arms. The relative phase ϕ
between the two arms can be fine-tuned by tilting one of the PBDs. 15
2.4 A complete polarization interferometer . . . . . . . . . . . . . . . . 17
3.1 Schematic for measuring the number-phase uncertainty relation us-
ing phase shift between polarization modes. A half-wave plate H0
and polarizing beam-splitter PBS1 with a poor extinction ratio is
used to prepare a superposition state of two polarization modes.
A variable-phase retarder made out of combination of wave-plates
creates the phase shift between the modes as a function of which
we measure the polarization quadratures using balanced detection
using a high extinction ratio polarization component PBS2 . . . . . . 19
3.2 Experimental Setup in a simpler notation. . . . . . . . . . . . . . . 20
3.3 Noise in the balanced detection ∆Ô as a function of phase shift for
different mean photon numbers . . . . . . . . . . . . . . . . . . . . 30
4.1 Experimental Setup using the Polarizing Beam Displacers. . . . . . 33
4.2 Experimental Setup. . . . . . . . . . . . . . . . . . . . . . . . . . . 34
4.3 Part of setup for fixing the polarization state and power. . . . . . . 35
4.4 The setup which we use to adjust the step 1 and 2. . . . . . . . . . 36
4.5 Interferometer part of the setup . . . . . . . . . . . . . . . . . . . . 37
4.6 BD40 Calcite Beam Displacers with 4mm separation. Figure taken
from [9] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
4.7 Phase shifter element of the setup made using the PBD’s and HWP.
The motor is used to control the Z806 actuator. . . . . . . . . . . . 39
4.8 Balanced Homodyne detection . . . . . . . . . . . . . . . . . . . . . 39
4.9 Balanced detection setup using the PBD and detector. . . . . . . . 41
4.10 Flow chart for the data acquisition process . . . . . . . . . . . . . . 42
4.11 KDC101 - K-Cube Brushed DC Servo Motor Controller. Figure
taken from [10]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
4.12 Details of the sequence set in the Kinesis software to move the Z806
actuator using KCube DC Motor Controller . . . . . . . . . . . . . 44

vii
 LIST OF FIGURES

5.1 Flow chart for the data acquisition process for θ′ = 0◦ . . . . . . . . 47

6.1 Plot for signal amplitude variation for different angles from 0◦ to
20◦ . The arrow shows the increase in the voltage amplitude of the
signal in this cycle. . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
6.2 Peak to peak amplitude as a function of θ′ . . . . . . . . . . . . . . 50
6.3 Noise amplitude for θ′ = 0◦ . . . . . . . . . . . . . . . . . . . . . . 52
6.4 Spectral noise amplitude for θ′ = 0◦ obtained after taking the FFT. 54
6.5 Spectral noise amplitude vs frequency. . . . . . . . . . . . . . . . . 55
7.1 Figure adapted from [12] showing the ideal regime to probe the
number phase uncertainty and the regime we are working currently. 58
7.2 Setup using a non-linear optical medium like Rubidium atom vapor. 59

viii
List of Tables
5.1 Peak amplitude values for different angles from 0◦ to 20◦ . . . . . . 46
5.2 Power values for LO and Signal. . . . . . . . . . . . . . . . . . . . . 47

ix
Chapter 1

Introduction

The number-phase uncertainty relation is a fundamental concept in quantum me-

chanics, describing the inherent limitations of simultaneously measuring the num-
ber of particles (or photons) in a quantum system and the phase of the corre-

sponding wavefunction. It is a manifestation of Heisenberg’s uncertainty principle,

which states that there is a fundamental limit to the precision with which certain
pairs of properties, such as position and momentum or, in this case, number and

phase, can be simultaneously known. The initial motivation was obtained from
the paper, where they have used optical homodyne tomography[12] to ascertain

the quantum state of a coherent state characterized by a low mean photon num-
ber. This technique utilizes the observed distributions of electric-field quadrature
amplitudes to deduce the Wigner function and subsequently derive the density

matrix. The obtained state information is then employed to compute the uncer-
tainty product associated with the number and phase[7], along with calculating
the expectation value of the commutator involving the number and phase opera-

tors.[2] Here, they have used a pulsed laser source, but in our experiment, we will
be using a continuous-wave laser.

1
 1 Introduction

We try to design an experiment to obtain the number-phase uncertainty relation

using phase shift between polarization modes. The benefit of using polarization
modes is that we can get rid of the noise due to vibrations, acoustic noise, etc.
These noises are difficult to remove separately. Since both the interfering beams

are traveling along the same path, the effect of vibrations will be the same for
both. So the common mode noise will be subtracted upon balanced detection.
A half-wave plate and polarizing beam splitter with a poor extinction ratio are

used to prepare a superposition state of two polarization modes. A variable-phase

retarder made out of combination of wave-plates creates the phase shift between
the modes as a function of which we measure the polarization quadrature[13] using

balanced detection using a high extinction ratio polarization component PBS.

Fundamental limit on knowledge in its best known form:

∆x∆p ≥ ℏ/2 ℏ ∼ 10−34 (1.1)

Now from the position - momentum uncertainty,

∆x → 0 ∆p → ∞ (1.2)

So when we improve measurement of position, we lose information about the mo-

mentum. This is important on microscopic scale where the minimum uncertainty

wavepacket is where the equality hold.

ℏ
∆x∆p = (1.3)
2
ℏ
∆p = (1.4)
∆x

2
 1 Introduction

1.1 Theoretical Background

1.1.1 Number Phase Uncertainty

Figure 1.1: A classical wave given by E = E0 cos(ωt + θ). Figure adapated from [8]

Consider a classical field given by E = E0 cos(ωt + θ). We see that there is some
noise in the arrival of the wave at any particular point, say when the amplitude
goes to zero. We see that there is a spread in the field at this point. This noise in

the arrival time has a one to one mapping with the noise in phase, as both time
and phase are dynamical parameters related by the equation:

ω∆t = ∆θ (1.5)

where ω is the frequency of the single mode field, ∆t is the uncertainty in time
and ∆θ is the uncertainity in phase.

Now from Heisenberg’s uncertainty principle

ℏ
∆E∆t ≥ (1.6)
2

3
 1 Introduction

Substituting the expression for ∆t from Equation 1.5, we get

∆θ ℏ
∆E ≥ (1.7)
ω 2

Now for a single mode field the energy is given by

E = nℏω (1.8)

The uncertainty in energy (∆E) for fixed ω is given by

∆E = ∆nℏω (1.9)

where ∆n is the uncertainty in the photon number.

Substituting the expressions of 1.9 in 1.7, we get

∆nℏω∆θ ℏ
≥ (1.10)
ω 2

Upon cancellation of terms we get the number phase uncertainty relation given
by
1
∆n∆θ ≥ (1.11)
2

1.2 Coherent State: phase space portrait

A coherent state in quantum mechanics is a particular type of quantum state that

exhibits properties similar to classical waves, such as a well-defined amplitude

and phase. In the context of quantum optics, coherent states are often used to
describe the behavior of light. Mathematically, a coherent state |α⟩ is defined as

an eigenstate of the annihilation operator â with eigenvalue α:

â|α⟩ = α|α⟩ (1.12)

4
 1 Introduction

where the mean photon number n̄ = |α|2 .

Coherent states are important because they have several unique properties:

1. They are eigenstates of the annihilation operator, which means they are
stable over time and do not spread out like other quantum states.

2. They exhibit Poissonian photon statistics, which is characteristic of classical

light sources.

3. They have a well-defined phase and amplitude, making them useful for co-
herent manipulations in quantum optics experiments.

4. They satisfy the minimum number phase uncertainty condition.

5. From an experimental point of view, coherent states are easily produced

using a laser.

In quantum mechanics, the coherent state phase space portrait is a representation

of the state in phase space, where each point in the phase space corresponds to
a possible state of the system. It is defined using the quadrature components X1

and X2 where X1 and X2 are defined as:

1
â + â†


X1 = (1.13)
2
1
â − â†


X2 = (1.14)
2i

For a coherent state the ∆X1 = ∆X2 = 12 . Hence ∆X1 ∆X2 = 1

4

The coherent state is a displaced vacuum state. Figure 1.2 shows the phase space
representation of a coherent state defined by α = |α|eiθ .

5
 1 Introduction

Figure 1.2: Phase space portrait of a coherent state. Figure adapted from [1].

Using geometry we can get for small values of ∆θ

1
sin(∆θ) ≈ ∆θ = (1.15)
2r

where, r = |α|. Hence

1
∆θ = (1.16)
2|α|
Now for a coherent state
√
∆n = n̄ = |α| (1.17)

From Eq 1.16 and Eq 1.17 we get

1
∆n∆θ = (1.18)
2

Hence the coherent state satisfies the minimum uncertainty relation.

When we know ∆n, then we can experimentally measure ∆θ.

6
 1 Introduction

1.3 Motivation

The initial motivation was obtained from the paper, where they have used opti-
cal homodyne tomography[12] to ascertain the quantum state of a coherent state
characterized by a low mean photon number. This technique utilizes the observed

distributions of electric-field quadrature amplitudes to deduce the Wigner func-

tion and subsequently derive the density matrix. The obtained state information
is then employed to compute the uncertainty product associated with the number

and phase[7], along with calculating the expectation value of the commutator in-
volving the number and phase operators.[2] Here, they have used a pulsed laser
source, but in our experiment, we will be using a continuous-wave laser. In the

experiment done by [12], they have used a PZT to generate the path difference
which creates a change in the phase between the two beams.. In our proposed
experiment we have created a phase difference by using a phase shifter between

the polarization modes. The experimental setup used in [12] is shown in Figure
1.3

7
 1 Introduction

Figure 1.3: The experimental setup described in the paper. Figure directly taken
from [12]. In this setup, PBS stands for a polarizing beam splitter, λ/2 refers to a
half-wave plate, PZT represents a piezoelectric translator, and ND denotes a
neutral-density filter.

In the paper, experimental measurement of number - phase uncertainty relation

was done by using a Mach-Zhender interferometer (MZI). They used a mode-
locked, Q-switched, cavity-dumped Nd:YAG laser, that generated 300-ps pulses

at 1064 nm with a repetition rate of 420 Hz and a pulse-energy stability of ±3%.

The beam pulse was split into the signal and the local oscillator using a polarizing
beam splitter (PBS1) to create the two arms of the MZI. The signal is highly

attenuated by using ND filters and the mean photon number of the signal pulse
ranged from zero to around 10. The local oscillator pulses contained around 2×106
photons and it given a relative phase shift by pushing the mirror using a piezo-

electric translator. The signal and LO having orthogonal polarizations are then
temporally and spatially interfered using the PBS2. A balanced homodyne detec-
tion is done to measure the quadrature amplitudes and the photo-diode outputs

8
 1 Introduction

are subtracted to obtain the photo-electron difference number between the detec-

tor outputs. The balanced detection setup remove the classical noise of the LO,
enabling measurements at the shot-noise level(SNL).

The new setup as shown in 4.2 is almost a direct mapping of the setup given in
[12]. But the power of the signal in my setup ≈ 6µW . Hence the timesscale where
the mean photon number of the signal is roughly one (n̄ ≈ 1) is

T ≈ 10−14 s =⇒ Frequency ≈ 1014 Hz (1.19)

Hence in our case we have to use a detector that can work in this frequency regime,

which is impossible. So right now this experiment performed is a basic tool of what
needs to be done in the future, where we operate it at a power level by which we
can probe the number phase uncertainty relation.

9
Chapter 2

Literature Review

2.1 Jones Matrix for Waveplates

A waveplate or retarder is an optical tool that alters the polarization state of light
waves as they traverse it. There are two prevalent varieties of waveplates:

1. Half-wave plate: It adjusts the orientation of linearly polarized light.

2. Quarter-wave plate: This variant transforms linearly polarized light into

circularly polarized light and the reverse. Moreover, a quarter-wave plate
has the capability to produce elliptical polarization.

Waveplates are made of a birefringent material (like mica or quartz), whose index
of refraction changes depending on the direction in which light passes through

it. A waveplate’s behavior, regardless of its type half-wave, quarter-wave, etc. is

decided by the wavelength of the light, the thickness of the crystal used, and the

index of refraction’s fluctuation. It is possible to make a controlled phase shift

between a light wave’s two polarization components and change its polarization
by carefully selecting the connection between these parameters.

The retarder, or waveplate, enhances the phase by + ϕ2 along the fast (x) axis and
diminishes the phase by − ϕ2 along the slow (y) axis. The correlation between the

incident light beam components Ex , Ey , and the resultant light beam components

10
 2 Literature Review

Ex′ , Ey′ is as follows:

Ex′ = ei(+ 2 ) Ex
φ
(2.1)

Ey′ = ei(− 2 ) Ey
φ
(2.2)

We can write Eq:(2.1,2.2) in Jones formalism as:

 ′  " i(+ φ ) # 
′ Ex e 2 0 Ex
J = = (2.3)
Ey′ 0 ei(− 2 )
φ
Ey

From (18), the Jones matrix for the retarder or the phase shifter can be obtained

as:
 
1 0
Jps (φ) = i(−φ) (2.4)
0 e

This is by ignoring the global phase ei(+ 2 ) , where the total phase shift between
φ

the two Ex and Ey is φ .

2.1.1 The Jones matrix for the following waveplates

Linear polarizer
 
cos2 θ sin θ cos θ
(2.5)
sin θ cos θ sin2 θ

Half wave plate

 
cos 2θ sin 2θ
(2.6)
sin 2θ − cos 2θ

11
 2 Literature Review

Quarter wave plate

cos2 θ + i sin2 θ (1 − i) sin θ cos θ

 
(2.7)
(1 − i) sin θ cos θ sin2 θ + i cos2 θ

2.2 Combination of Waveplates as a Phase Shifter

Figure 2.1: Combination of waveplates acting as an effective phase shifter

Using the Jones matrices for the waveplates as given in the 2.1, we get

Jps (θ) = HWP(22.5◦ ) × QWP(0◦ ) × HWP(θ) × QWP(0◦ ) × HWP(22.5◦ ) (2.8)

     
1 1 1 1 0 cos 2θ sin 2θ 1 0 1 1
= (2.9)
2 1 −1 0 i sin 2θ − cos 2θ 0 i 1 −1
   
1 1 i cos 2θ sin 2θ 1 1
= (2.10)
2 1 −i sin 2θ − cos 2θ i −i
  
1 1 i cos 2θ + i sin 2θ cos 2θ − i sin 2θ
= (2.11)
2 1 −i sin 2θ − i cos 2θ sin 2θ − i cos 2θ
 
1 cos 2θ + i sin 2θ + i sin 2θ + cos 2θ cos 2θ − i sin 2θ + i sin 2θ − cos 2θ
=
2 cos 2θ + i sin 2θ − i sin 2θ − cos 2θ cos 2θ − i sin 2θ − i sin 2θ − cos 2θ
(2.12)
 
1 cos 2θ + i sin 2θ 0
= (2.13)
2 0 cos 2θ − i sin 2θ

12
 2 Literature Review

 
1 ei2θ 0
= (2.14)
2 0 e−i2θ
ei2θ 1
 
0
Jps (θ) = (2.15)
2 0 e−i4θ
(2.16)

2.3 Combination of Polarizing Beam Displacers

as a Phase Shifter

Figure 2.2: A +45◦ polarized light beam is divided into two beams and later
recombined by a pair of Polarization Beam Displacers (PBDs).

Considering the configuration of two PBDs as illustrated in Figure 2.2, the initial

PBD1 separates the incoming beam (assumed to be polarized at +45◦ ) into dis-
tinct horizontal and vertical components. For the upper beam, PBD1 functions
as a vertical polarizer, thus its Jones matrix is denoted as JV . Simultaneously,

PBD1 acts as a horizontal polarizer, represented by the Jones matrix JH , for the
lower beam. Subsequently, the second PBD2 recombines the beams, effectively
combining the Jones matrices for the two paths.
     
0 0 1 0 1 0
J = JV + JH = + = =1 (2.17)
0 1 0 0 0 1

13
 2 Literature Review

The resulting matrix is the identity matrix, indicating that this sequence of PBDs

doesn’t alter the beam. However, in Figure 2.2, the two polarization compo-
nents undergo differing path lengths within the system. Consequently, there exists
a phase difference between the horizontal and vertical polarizations, inducing a

phase shift denoted as ϕ on the vertical polarization concerning the horizontal po-
larization. The effective Jones matrix corresponding to this scenario is expressed
by the following matrix.

 
eiϕ 0
Jϕ = (2.18)
0 1

      
0 0 eiϕ 0 1 0 eiϕ 0
J = JV + J ϕ JH = + = (2.19)
0 1 0 1 0 0 0 1
Figure 2.2 illustrates a form of polarization interferometer where two polarizations
undergo separation, phase shifting, and subsequent recombination. The presence

of a relative phase shift between these polarizations implies that the output po-
larization typically differs from the input polarization. However, the polarization
interferometer depicted in Figure 2.2 possesses a limitation. While it performs

effectively with a continuous wave (CW) laser source like a helium-neon laser, its
functionality is constrained with other sources. Consider introducing a very brief
pulse of light into this interferometer; the pulse divides into two pulses traversing

the interferometer. Should the path length difference between the arms exceed the
pulse duration, the pulses arrive at the second PBD at distinct times, leading to
non-coinciding pulses and consequently, no interference. Moreover, even in the ab-

sence of pulses, the coherence length of the light source must be taken into account
for interference to occur. For two light waves to interfere, they must demonstrate

14
 2 Literature Review

coherence concerning each other. If the path length difference surpasses the co-

herence length, interference fails to manifest.

To enable interference observation with sources having short coherence lengths,

adjustments to the polarization interferometer depicted in Figure 2.2 are necessary

to balance the arm lengths. This can be accomplished through the design presented
in Figure 2.3. Here, a half-wave plate positioned with its fast axis at an angle of

θ = 45◦ from the horizontal is inserted between the two PBDs. This plate flips
the polarizations of the beams, establishing symmetry in the arms’ behavior and
ensuring equal path lengths. Usually, the arm lengths are not perfectly matched.

To alter the relative phase shift ϕ between the arms, one can tilt one of the PBDs,
slightly modifying the optical path lengths. It’s crucial to recognize that the
relative phase shift ϕ is proportionate to the tilt angle of PBD2 in Figure 2.3,

rather than being equivalent to the tilt angle itself.

Figure 2.3: Inserting a half-wave plate in a polarization interferometer serves to

balance the path lengths of the two arms. The relative phase ϕ between the two
arms can be fine-tuned by tilting one of the PBDs.

The Jones matrix for the interferometer depicted in Figure 2.3 can be derived

15
 2 Literature Review

through the following process. The upper beam initially encounters a vertical

polarizer, then proceeds through a half-wave plate that rotates the polarization to
horizontal, and finally undergoes a phase shift. Conversely, the lower beam involves
a horizontal polarizer followed by a half-wave plate that rotates its polarization to

vertical. Subsequently, these two beams are combined.

J = Jϕ Jλ/2θ=45◦ JV + Jλ/2θ=45◦ JH (2.20)

 iϕ      
e 0 0 1 0 0 0 1 1 0
= + (2.21)
0 1 1 0 0 1 1 0 0 0
     
0 eiϕ 0 0 0 eiϕ
= + = (2.22)
0 0 1 0 1 0

When the phase shift is set to zero, the interferometer behaves precisely like the

half-wave plate positioned at its center. Consequently, incident light that is hor-
izontally polarized transitions to vertical polarization, and vice versa. This com-
ponent forms the crux of our experiment. The polarization interferometer, as

previously elucidated, upholds the beam’s intensity due to energy conservation

principles. All input power is transformed into output power as the beam enters
the interferometer, splits into two, and subsequently recombines. To manipulate

the beam’s intensity, a polarizer must be employed.

16
 2 Literature Review

Figure 2.4: A complete polarization interferometer

By offering distinct paths for output power, the interferometer enables modulation

of each output beam’s intensity. In our setup, a quarter-wave plate is positioned

before PBD3 for component splitting.

To demonstrate intensity modulation by this interferometer, let’s calculate the

effective Jones matrix for the +45◦ polarized output beam. Beginning from the
left of Figure 2.4, the input beam traverses the polarization interferometer as

described in equation 1.16, and then passes through a polarizer oriented at +45◦ .
The Jones matrix for this setup is provided by:
    
1 1 1 0 eiϕ 1 1 eiϕ
J= = (2.23)
2 1 1 1 0 2 1 eiϕ

For an input beam polarized at +45◦ , the output beam’s polarization vector is
given by:

ε = Jε+45 (2.24)
   
1 1 eiϕ 1 1
= iϕ √ (2.25)
2 1 e 2 1

17
 2 Literature Review

 
1 1 + eiϕ
= √ iϕ (2.26)
2 2 1+e
 
1 iϕ

1 1
= 1+e √ (2.27)
2 2 1
1
1 + eiϕ ε45


= (2.28)
2

As expected, the output beam is seen to be polarized at +45◦ , which is multiplied

by the vector’s square magnitude:
2
1
1 + eiϕ


I = Ii (2.29)
2
Ii
1 + eiϕ 1 + e−iϕ



= (2.30)
4
Ii
2 + eiϕ + e−iϕ


= (2.31)
4
Ii
= (1 + cos ϕ) (2.32)
2

When the phase is set to ϕ = 0, resulting in I = Ii , constructive interference occurs,

and all light is passing out through the +45◦ port of PBD3 . Conversely, with

ϕ = π and I = 0, destructive interference happens, and for energy conservation,

light must emerge from the −45◦ port instead.

18
Chapter 3

Experimental Proposals and Cal-

culations
3.1 Initial Proposal of the Experiment

The first proposal for the experiment for measuring the number-phase uncertainty
relation using phase shift between polarization modes is shown below.

Figure 3.1: Schematic for measuring the number-phase uncertainty relation using
phase shift between polarization modes. A half-wave plate H0 and polarizing
beam-splitter PBS1 with a poor extinction ratio is used to prepare a superposition
state of two polarization modes. A variable-phase retarder made out of combination
of wave-plates creates the phase shift between the modes as a function of which we
measure the polarization quadratures using balanced detection using a high extinction
ratio polarization component PBS2 .

For convenience, I am naming the modes a1 to a9 as a to i. The above figure can

be drawn in a more simplified form as

19
 3 Experimental Proposals and Calculations

Figure 3.2: Experimental Setup in a simpler notation.

3.1.1 Step 1: State Preparation

• Incident State = |αH ⟩

• Intensity of beam |α|2 and Extinction Ratio ( ε )

Let the action of half-wave plate H0 be such that the polarization gets oriented at
π
an angle 4
wrt. the horizontal. So the input state will now be = |α+ ⟩

|α+ ⟩ = D̂(α)|0⟩+ (3.1)

 
1
where |0⟩+ = √ (|0⟩H ⊗ |0⟩V
2
† ∗ â
D̂+ (α) = eαâ+ −α +
(3.2)

1
where â+ = √ (âH + âV )
2
1
1  
√ αâH † − α∗ âH √ αâ†V − α∗ âV
D̂+ (α) =e 2 e 2 |0⟩H |0⟩V (3.3)
 
α α
= √ ⊗ √ (3.4)
2 H 2 V

20
 3 Experimental Proposals and Calculations

In our diagram, the input port is a, and we can assume that a coherent vacuum

state |α = 0⟩ is input through b port.[3] The input basis state can be written as
= |0⟩a,H |0⟩a,V |0⟩b,H |0⟩b,V
Input in our given diagram |α⟩+,a and |0⟩+,b
   
α α
| in ⟩a,b = D̂b,V (0)D̂b,H (0)D̂a,V √ D̂a,H √ |0⟩a,H |0⟩a,V |0⟩b,H |0⟩b,V (3.5)
2 2

The above equation shows the complete description of the input state with oper-

ators âH , âV , b̂H , b̂V

3.1.2 Step 2: Transformation due to the polarizing beam

splitter PBS1

â†H −→ ĉ†H â†V −→ dˆ†V (3.6)

b̂†H −→ dˆ†H b̂†V −→ ĉ†V (3.7)

On doing the transformations in Eq(3.5) we get

 
α
| out ⟩c,d = D̂d,H (0)D̂c,V (0)D̂d,V (0)D̂c,H √ |0⟩c,H |0⟩c,V |0⟩d,H |0⟩d,V (3.8)
2

α
| out ⟩c,d = √ |0⟩c,V |0⟩d,H |0⟩d,V (3.9)
2 c,H

Now, a part of the mode that has a horizontal polarization component will contain

some residual vertical component due to the non-ideal behavior of the PBS (polar-
izing beam splitter). The factor that determines the amount of this non-ideality
in the PBS is called the extinction ratio ( ε ).

21
 3 Experimental Proposals and Calculations

Now the output at port c, would look like

     
α εα α εα
| out⟩c = D̂c,H √ D̂c,V √ |0⟩c,H |0⟩c,V = √ √ (3.10)
2 2 2 c,H 2 c,V

In general, there could be a phase shift between the two polarization components,

which we have ignored now as later we would use an arbitrary retarder.

3.1.3 Step 3: Transformation from Phase Shifter element

Jps (θ)

Ignoring the Global Phase in phase retarder matrix calculated in Eq:(2.16) we

get
 
1 0
Jˆps (θ) = (3.11)
0 e−i4θ
The phase retarder can be thought of as adding the extra phase −4θ to the vertical
component of the state. Thus the resultant state in mode e after the phase retarder

is given as follows:
   
α εα α −i4θ εα
Jˆp,s (θ) √ √ = √ e √ (3.12)
2 c,H 2 c,V 2 c,H 2 c,V
−i4θ
 
α e εα
| out ⟩e = √ √ (3.13)
2 e,H 2 e,V

3.1.4 Step 4: Half wave Plate

The operators, upon transformation by a half-wave plate, transform as

â†f,H + â†f,V â†f,H − â†f,V

â†e,H → √ â†e,V −→ √ (3.14)
2 2

Now the input state at e mode is given by

e−i4θ εα
   
α −i4θ εα α
| input ⟩e = √ ⊗e √ = √ ⊗ √ (3.15)
2 e,H 2 e,V 2 e,H 2 e,V

22
 3 Experimental Proposals and Calculations

Using Operator Transformations

â†f,H + â†f,V
" ! #
α∗ âf,H + âf,V

α
√ √ −√ √
e−i4θ εα
 
α 2 2 2 2
√ ⊗ √ =e
2 e,H 2 e,V
† †
" ! #
e−i4θ εα âf,H − âf,V e−i4θ εα âf,H − âf,V

√ √ − √ √
2 2 2 2
e |0⟩f,H |0⟩f,V
(3.16)

The transformation of the displacement operator is given as follows

e−i4θ α e−i4θ α
α α    
| out ⟩f = D̂f,H D̂f,V D̂f,H D̂f,V − |0⟩f,H |0⟩f,V
2 2 2 2
(3.17)

Now using the property of Displacement Operators

∗)
D̂(α + β) = D̂(α)D̂(β)e−i Im(αβ (3.18)

Now state just before PBS2 , | out ⟩f is given by

 
−i4θ
∗ 
α e−i4θ εα i Im α2 · − e 2 aεα
 
| out ⟩f = D̂f,H + e
2 2
 
−i4θ
∗ 
α e−i4θ εα i Im α2 · e 2 εα
 
D̂f,V − e |0⟩f,H |0⟩f,V (3.19)
2 2
α  α 
−i4θ −i4θ
= | out ⟩f = D̂f,H (1 + e ε) · D̂f,V (1 − e ε) |0⟩f,H |0⟩f,V
2 2
(3.20)

The exponential constants get cancelled.

3.1.5 Step 5: Balanced Homodyne Detection

Transformation due to the polarizing beam splitter PBS2

fˆH† −→ ĥ†H fˆV† −→ î†V (3.21)

†
ĝH −→ î†H ĝV† −→ ĥ†V (3.22)

23
 3 Experimental Proposals and Calculations

Using the transformations, we then have the following transformed state:

α  α 
−i4θ −i4θ
| out ⟩h,i = D̂h,H (1 + e ε) · D̂i,v (1 − e ε) |0⟩h,H |0⟩i,V |0⟩i,H |0⟩h,V
2 2

(3.23)
(3.24)
α −i4θ
E α −i4θ
E
| out ⟩h,i = (1 + e ε) (1 − e ε) (3.25)
2 h,H 2 i,V
| {z }| {z }
α1 , α2

= |α1 ⟩h,H |α2 ⟩i,V (3.26)

3.1.6 Mean and Variance

In a standard phase-shift measurement using a Mach-Zehnder interferometer, the

shift θ is determined by subtracting the intensities obtained from detectors D1 and

D2 . This difference is proportional to the number difference operator of the two
modes. Now we have taken operators (ĥ, ĥ† ) to represent the beam directed to D1

and (î, î† ) of that towards D2 . Let the photo-current in detector D1 be Ih and in
detector D2 be Ii . The photocurrent Ih will have components from horizontal(H)
and the vertical(V ) modes. But along the h mode, only the horizontal component

is present, and along the i mode, we have only the vertical component. So we
can write Ih and Ii in terms of their respective number operators, but with the
horizontal and vertical components of the operators acting on the states, respec-

tively.
D E D E
¯
Ih = ĥ† ĥ ¯
Ii = î† î (3.27)

24
 3 Experimental Proposals and Calculations

So now, the mean of photocurrent in detector D1 is given by

D   E
†
⟨Ih ⟩ = α1 |h,H ĥ ĥ | α1 (3.28)
h,H

Now, since only the horizontal polarization accounts for the photocurrent in h
mode
D   E
⟨Ih ⟩ = α1 |h,H ĥ†H ĥH | α1 (3.29)
h,H

= |α1 |2 (3.30)

similarly for i mode, only the vertical polarization accounts for the photocur-
rent.
D   E
⟨Ii ⟩ = α2 |i,V î†V îV | α2 (3.31)
i,V

= |α2 |2 (3.32)

Since each detector receives only one of the modes of polarization, the polarization
index was ignored for simplicity. Now the mean of the square of the photocurrent

is given by,

Ih2 = α1 Ih2 α1 (3.33)

  
= ⟨α1 | ĥ†H ĥH ĥ†H ĥH |α1 ⟩ (3.34)
D   E
= α1 ĥ†H 1 + ĥ†H ĥH hH α1 (3.35)
D E
† †2 2
= α1 ĥH ĥH + ĥH ĥH α1 (3.36)

= |α1 |2 + |α1 |4 (3.37)

25
 3 Experimental Proposals and Calculations

Now the variance in the photocurrent is given by,

q
21
σ (Ih ) = ⟨Ih2 ⟩ − ⟨Ih ⟩2 = |α1 |2 + |α1 |4 − |α1 |4 = |α1 | (3.38)

Similarly,
q
21
σ (Ii ) = ⟨Ii2 ⟩ − ⟨Ii ⟩2 = |α2 |2 + |α2 |4 − |α2 |4 = |α2 | (3.39)

Calculation using difference photocurrent (Ô)

In each port of the detector, there is only one polarization mode. Hence, we

can drop the polarization indices for simplicity. The operator for the differential
detection between the two ports is given by the number difference operator [5]

Ô = ĥ† ĥ − î† î (3.40)

The expectation value of the differential current is given by the following:

D D   E E
⟨Ô⟩ = α2 |i α1 |h ĥ† ĥ − î† î | α1 | α2 (3.41)
h i

= ⟨Ih ⟩ − ⟨Ii ⟩ (3.42)

= |α1 |2 − |α2 |2 (3.43)

Now |α1 | and |α2 | can be calculated as follows:

α |α|
1 + εe−i4θ =


|α1 | = |1 + ε cos 4θ − iε sin 4θ| (3.44)
2 2
|α| p
= (1 + ε cos 4θ)2 + (ε sin 4θ)2 (3.45)
2
|α| p
= 1 + ε2 cos2 4θ + ε2 sin2 4θ + 2ε cos 4θ (3.46)
2
|α| √
|α1 | = 1 + ε2 + 2ε cos 4θ (3.47)
2
α |α|
1 − εe−i4θ =


|α2 | = |1 − ε cos 4θ + iε sin 4θ| (3.48)
2 2

26
 3 Experimental Proposals and Calculations

|α| p
= 1 + ε2 cos2 4θ − 2ε cos 4θ + ε2 sin2 4θ (3.49)
2
|α| √
|α2 | = 1 + ε2 − 2ε cos 4θ (3.50)
2

Mean of differential current operator (Ô )

 D E 
⟨Ô⟩ = α2 |i,V α1 |h,H (Ih − Ii ) | α1 | α2 (3.51)
h,H i,V

= |α1 |2 − |α2 |2 (3.52)

|α|2
|α|2
1 + ε2 + 2ε cos 4θ − 1 + ε2 − 2ε cos 4θ


= (3.53)
4 4
2
|α|
⟨Ô⟩ = (4ε cos 4θ) = |α|2 ε cos 4θ (3.54)
4

Now the second-order moment

D E    2  
2 † †
Ô = α2 |i α1 |h ĥ ĥ − î î | α1 | α2 (3.55)
h i
      
2  2
† † † † † †
= α2 |i α1 |h ĥ ĥ + î î − ĥ ĥî î − î îĥ ĥ | α1 | α2 (3.56)
h i
D D h    i E E
† † † † † † † †
= α2 i α1 |h ĥ ĥĥ ĥ + î îî î − ĥ ĥî î − î îĥ ĥ | α1 α2 (3.57)
h i

= |α1 |2 + |α1 |4 + |α2 |2 + |α2 |4

D D h i E E
† † † †
− α2 |i α1 |h ĥ ĥî î + î îĥ ĥ | α1 | α2 (3.58)
h i

Now we know,

ĥ† ĥî† î = n̂h n̂i (3.59)

î† îĥ† ĥ = n̂i n̂h (3.60)

If n̂h , n̂i commute then [n̂h , n̂i ] = 0

⇒ n̂n n̂i = n̂i n̂h (3.61)

27
 3 Experimental Proposals and Calculations

⇒ ĥ† ĥî† î = î† îĥ† ĥ (3.62)

so
D E D D h i E E
2 2 4 2 4 † †
Ô = |α1 | + |α1 | + |α2 | + |α2 | − α2 |i α1 |h 2 ĥ ĥî î | α1 | α2 (3.63)
h i

The last term signifies the correlated detections[11], i.e., the joint detection of
photons in the two ports.
D E hD D E Ei
2 2 4 2 4 † †
Ô = |α1 | + |α1 | + |α2 | + |α2 | −2 α2 |i α1 |h ĥ ĥî î | α1 | α2
| {z } h i
ψ
(3.64)
hD D E Ei
=ψ−2 α2 i α1 |h α1∗ ĥît α2 | α1 α2 (3.65)
h
h D E ii
= ψ − 2α1∗ α2 ⟨α2 ⟩i α1 |h ît ĥ | α1 |α2 ⟩i (3.66)
h
h i
Since ĥ, î+ = 0 (3.67)
D E
Ô2 = ψ − 2α1∗ α2 α2∗ α1 = |α1 |2 + |α1 |4 + |α2 |2 + |α2 |4 − 2 |α1 |2 |α2 |2
(3.68)

Now the uncertainty in the differential current is given by

rD E
∆(Ô) = Ô2 − ⟨Ô⟩2 (3.69)
q

2
= |α1 |2 + |α1 |4 + |α2 |2 + |α2 |4 − 2 |α1 |2 |α2 |2 − |α1 |2 − |α2 |2 (3.70)
2 1/2
4
i
= |α1 |2 + |α1 |4 + |α2 |2 + |α2 |4 − 2 |α1 |2 |α2 |2 − α1 |4 − α2 + 2 |α1 |2 |α2


(3.71)
1/2
∆(Ô) = |α1 |2 + |α2 |2

(3.72)

28
 3 Experimental Proposals and Calculations

The variance is square of the standard deviation: σ(Ô) = |α1 |2 + |α2 |2 . Substi-
 

tuting the values of |α1 | and |α2 | into Eq(3.72), we get:

|α|
1/2
∆(Ô) = √ 1 + ε2 (3.73)
2

3.1.7 Noise dependence on parameters

By using the calculus of error propagation, the uncertainty of the phase measure-
ments is given by
∆Ô
∆θ = (3.74)
∂⟨Ô⟩
∂θ

From Eq(3.54) we evaluate

(α) 1/2
∆Ô √
2
(1 + ε2 )
∆θ = = (3.75)
4 |α|2 ε sin 4θ


∂⟨Ô⟩
∂θ
1/2 1/2
(1 + ε2 ) (1 + ε2 )
∆θ = √ = √ √ (3.76)
4 2ε|α|| sin 4θ| 4 2ε n̄| sin 4θ|

where n̄ = |α|2 .

29
 3 Experimental Proposals and Calculations

|α| =0.1 |α| =1

 
0.001
0.06 0.001 0.6

0.04 0.4

0.2

∆d]θ0=π/4
0.02
∆d]θ0=π/4

Δ(O) Δ(O)
0.00 0.0

−0.02 −0.2

−0.04 −0.4

−0.06 −0.6

π π π π π π 3π π 5π 3π π 9π 5π 11π 3π 13π 7π 15π

0 8 4
3π
2
5π 3π
8 π 9π 5π 11π 3π 13π 7π 15π 2π 0 8 4 8 2 8 4 8 π 8 4 8 2 8 4 8
2π
8 8 4 8 4 8 2 8 4 8
Phase Shift φ Phase Shift φ

(a) |α| = 0.1 (b) |α| = 1

|α| =10 |α| =100

80 
0.001
6 0.001
60
4
40
2
∆d]θ0=π/4

∆d]θ0=π/4

20

Δ(O) 0 ∆(O) 0

−2 −20

−4 −40

−60
−6
−80
π π 3π π 5π 3π π 9π 5π 11π 3π 13π 7π 15π π π π π
0 8 4 8 2 8 4 8 π 8 4 8 2 8 4 8
2π 0 8 4
3π
2
5π 3π
8 π 9π 5π 11π 3π 13π 7π 15π 2π
8 8 4 8 4 8 2 8 4 8
Phase Shift φ Phase Shift φ

(c) |α| = 10 (d) |α| = 100

Figure 3.3: Noise in the balanced detection ∆Ô as a function of phase shift for
different mean photon numbers

From the plots we see that the noise or the uncertainty in ∆Ô is independent of
the phase shift. That is the width remains constant, and only the mean value
√
fluctuates with phase. From Eq(3.1.7) we see that ∆θ ∝ 1/ n̄. Clearly, the

phase uncertainty depends on the relative phase θ and we obtain the minimum
uncertainty.[5]
Till now we have taken the calculation for a coherent state for which the answer is

30
 3 Experimental Proposals and Calculations

known[4, 6], later, we will go to a squeezed state and measure the phase dependence

separately.
p
(1 + ε2 ) π
∆θmin = √ √ for odd values of (3.77)
4 2ε n̄ 8

3.2 Advantages of this experimental proposal

1. Experimental advantage to get rid of the classical phase noise that includes
vibrations, acoustic noise, etc.

2. Both the interfering beams are traveling along the same path, the effect of
vibrations will be the same for both and only the common mode noise will

be present which gets subtracted upon balanced detection.

3. Ease of alignment of the optical beams.

3.3 Shortcomings to the initial experimental pro-

posal

In the previous semester the above mentioned experiment was proposed to deter-
mine the number phase uncertainty principle. But there were a few setbacks faced

after the initial proposal of the experiment.

1. Firstly the problem was lack of availability of half wave plate in the lab. For

this setup we need a specific zero order half wave plate. Initially the order
was placed but there was a delay in the delivery of the product.

2. In this proposed setup we require 5 half wave plates and 2 quarter wave
plates. But only 1 HWP and 1 QWP were available in lab of the specification.

3. The lack of availability of Polarizing Beam splitters was also another major

31
 3 Experimental Proposals and Calculations

limitation. I had only 1 available PBS, while I required two for the set up.

But this could be sorted by using a polarized beam displacer instead of PBS,
although the path length won’t remain the same.

32
Chapter 4

Experimental Setup and Procedure

Figure 4.1: Experimental Setup using the Polarizing Beam Displacers.

The new setup as shown in 4.2 is almost a direct mapping of the setup given in [12].
We use a He-Ne laser which was borrowed from the open lab for this setup. The

setup is divided into three parts which is the state preparation, the phase shifter
and then the balanced detection. The main difference in this setup compared to
the previously proposed one is that the phase shifter is now made of the Polarizing

33
 4 Experimental Setup and Procedure

Beam Displacer combination. Also in the balanced detection the PBS is replaced

by the PBD with good extinction ratio of ϵ = 105 : 1.

Figure 4.2: Experimental Setup.

34
 4 Experimental Setup and Procedure

4.1 Steps for interferometery

4.1.1 Step 1: Fixing State polarization by polarizing beam
splitter PBS1

Figure 4.3: Part of setup for fixing the polarization state and power.

1. The first step was to align the 633nm He-Ne laser light using two mirrors.
For this I placed mirrors M1 and M2 and also kept two apertures one close to
M2 and one far along the same line. I closed both the apertures and tried to

align the beam after reflecting from M1 and M2 such that the beam passes
through the aperture. This helps in fixing the height of the beam as well as
to make sure it goes in a straight path. I moved the knobs of M1 to align

the beam about the aperture close to M2 . Same way, I moved the knobs of
the mirror M2 to align the beam about the aperture kept far.

2. An important thing to note is that he He-Ne Laser is an unpolarized source

because it has many cavity modes. Hence we have to fix the polarization
using a PBS.

35
 4 Experimental Setup and Procedure

3. The next step is to put the PBS with an extinction ratio of about ϵ1 = 103 : 1,

in the path of the beam. The PBS transmits the horizontal component and
reflects the vertical component. This fixes the polarization modes for the
next step.

4.1.2 Step 2: Rotating H0 to control the power of both

beams

1. In this step we put a half wave plate using which we control the relative
intensity between the signal and the local oscillator.

2. The rotation of the HWP (θ′ ) is the most important feature which we use to
change the relative intensity between the two polarization modes.

Figure 4.4: The setup which we use to adjust the step 1 and 2.

36
 4 Experimental Setup and Procedure

4.1.3 Step 3: Setting up of the interferometer

Figure 4.5: Interferometer part of the setup

1. The next step is to align the Polarizing Beam Displacers. In the introduction,
I have discussed in detail the working of the PBD.

Figure 4.6: BD40 Calcite Beam Displacers with 4mm separation. Figure taken
from [9]

2. In this experiment the PBD’s which I used is BD40 as shown in 4.6. It

separates the unpolarized light into two orthogonally polarized beams with

a separation of about 4.3 mm at 633 nm wavelength. The length (L) is 4.1

37
 4 Experimental Setup and Procedure

cm, whereas the clear aperture is 10 mm x 10 mm.

3. The beams are divided into signal and local oscillator by using PBD1, giving
it a one to one mapping with the experiment done by Smithey et al.(1993)[12].

4. The first PBD is aligned taking into account the separation of both beams.
Then the HWP was aligned such that both the beams pass through it.

5. H1 is kept at 45◦ to flip the polarization of both the beams.

6. The PBD2 is mounted on a mirror mount which has a motorized DC actuator

(Z806 - 6 mm) to rotate the knob that controls the horizontal deviation. We
align the PBD2 such that only one beam passes through after passing through

it. Now we rotate the H1 such that once the alignment is done fixing the
angle, if we rotate it further we will get both the beams back.

7. This Z806 actuator is connected to the K-Cube Brushed DC Servo Motor

Controller which can be controlled using the kinesis software. Using the
motor we can move the actuator upto 6 mm.

8. PBD2 is tilted by ϕ to create a phase shift between both the beams.

9. The amplitude of the state d using Malus law is

|d⟩ ∝ sin(2θ′ )|H⟩ + cos(2θ′ )eiϕ |V ⟩ (4.1)

where |H⟩ and |V ⟩ is the horizontal and vertical polarization modes of the
signal and the LO.

38
 4 Experimental Setup and Procedure

Figure 4.7: Phase shifter element of the setup made using the PBD’s and HWP.
The motor is used to control the Z806 actuator.

4.1.4 Step 4: Balanced Homodyne Detection

Figure 4.8: Balanced Homodyne detection

1. First step is to put the PBD which splits the incoming two beams again. So
in total we get four beams when we rotate the PBD. Since I was short of

39
 4 Experimental Setup and Procedure

a half wave plate, I used a quarted wave plate to balance both the beams

before it goes into the PBD polarization interferometer setup.

2. After aligning the PBD such that both the beams pass through, essentially

its four beams where the extra two is because of the splitting of the two
original beams.

3. The state |d⟩ is projected on to the diagonal basis using Q1

4. Since now they are balanced by the QWP they convert to right circular and

left circular and therefore will get an orthogonal difference in polarization

between both the beams. Each of this beam now will again split into two
beams with two orthogonal components.

5. ND filters is used to reduce the power of the beam so that it doesn’t saturate
the signal.

6. Now we place and align the mirrors such that only one beam gets reflected
and the other beam directly falls on one of the port of the detector. The

reflected beam is then directed to the other port of the detector by using
another mirror as shown below.

7. Using PBD3 we do the balance detection and the amplitude of the resul-
tant state after interference is ∝ sin(4θ′ )sin(ϕ). The interference pattern is

observed in the oscilloscope connected to the detector.

40
 4 Experimental Setup and Procedure

Figure 4.9: Balanced detection setup using the PBD and detector.

4.2 Data Acquisition Technique and Observations

The following preliminary steps are done before the data acquisition.

1. The detector is connected to the oscilloscope and we observe the signal which

is obtained from the differential current from the photo-detector.

2. To balance the detector, we start by rotating the HWP1 such that we get

around equal intensity of both signal and LO after passing through PBD1 .

3. Now we block one beam and then rotate the HWP2 such that we get equal

intensity of both beams after the PBD2 .

4. We then observe the signal and try to make it to zero level in the oscilloscope

by rotating the PBD3 . Once that is made zero, we rotate the HWP1 to change
the relative intensities and see that the signal still coincides with the zero

level.

5. We then remove the blocker and try to reduce the signal as close to zero by

41
 4 Experimental Setup and Procedure

rotating the Q1 only. We do this for a few iterations until we get the best

balancing.

6. When the PBD’s are properly aligned, turning the knob of any of the PBD’s

of the phase shifter element will cause an interference patter in both the
beams. This means that while tilting any of the PBD’s the path length
changes and there is effectively a phase difference and hence we get interfer-

ence.

4.2.1 Data acquisition technique

Figure 4.10: Flow chart for the data acquisition process

Step: 1

1. We rotate the half wave plate H0 to control the relative intensities between
the polarization of signal and LO. They both interfere in the polarization

space after PBD3 .

2. Firstly, the two beams are superposing after PBD2 . One reason why I would
prefer this arrangement is because you can make all the light go through the
same path after PBD2 .

3. Once all light comes through the PBS the intensity there won’t change and
there will be superposition between H and V modes of the signal and LO

42
 4 Experimental Setup and Procedure

respectively.

Step: 2

1. The PBD2 is mounted on a mirror mount which has a motorized DC actuator

(Z806 - 6 mm) to rotate the knob that controls the horizontal deviation.

2. The Z806 actuator is connected to KDC101 - K-Cube Brushed DC Servo

Motor Controller as shown in figure below.

Figure 4.11: KDC101 - K-Cube Brushed DC Servo Motor Controller. Figure taken
from [10].

3. By moving the actuator we can tilt the PBD2 and this will create additional

path difference which in turn will cause phase shift. The Z806 actuator can
be moved up to 6 mm.

Step: 3

1. DC Servo Motor is controlled using Kinesis software.

2. The Z806 actuator can be moved up to 6 mm, and we continuously move it

using the motor.

43
 4 Experimental Setup and Procedure

3. We set the sequence in the Kinesis software which sets the range of the

motion. The actuator in our case was moved from 4.2 mm to 4.4 mm and
back.

4. This cycle it repeated 10 times before the sequence is stopped. The details
of the sequence is attached below.

Figure 4.12: Details of the sequence set in the Kinesis software to move the Z806
actuator using KCube DC Motor Controller

Step: 4

1. We set the sequence in the Kinesis software to move the actuator from 4.2
mm to 4.4 mm and back.

2. This corresponds to varying the optical path length by moving the actuator
by 0.2 mm to get 5 oscillations in one cycle.

44
 4 Experimental Setup and Procedure

3. The optical path difference can be calculated as:

∆=5×λ (4.2)

∆ = 5 × 633 × 10−9 (4.3)

∆ = 3.165µm (4.4)

Step: 5

1. For every angle θ′ for H0 we run the sequence and then we observe the
interference pattern in the oscilloscope.

2. The data was taken in steps of 4 degrees in rotation of the H0 from 0◦ to

46◦ .

3. The oscilloscope data for few of the angles are shown below.

(a) Oscillsocope data for 0◦ (b) Oscillsocope data for 20◦

45
Chapter 5

Observations
The data was obtained for 13 set of half wave plate angles 0◦ to 46◦ . We see two
sets of cycles in the relative intensity where we started from a point in which both
the intensity of the beams are equal to where one of the beam intensity is much

higher compared to the other beam. It changed again to being equal and then the
reverse. Two set of data cycles were hence taken. Below, we see how the peak to
peak amplitude changes with the angle.

θ′ (degrees) Amplitude (V)

0 0.06274
2 0.14241
4 0.3211
8 0.60857
12 0.85687
16 1.03803
20 1.12018
28 1.04484
32 0.90448
36 0.68544
40 0.39951
44 0.08756
46 0.0321

Table 5.1: Peak amplitude values for different angles from 0◦ to 20◦

From the oscilloscope data we get five oscillations but we isolate the middle three
oscillations for further analysis and sine fit is taken for that dataset. The technical
reason is because, when we move the actuator initially it accelerates to reach a

46
 5 Observations

constant velocity and then in decelerates to reverse the cycle. So it has a trape-

zoidal curve. Hence the first and the last cycles are regions where it accelerates and
decelerates respectively. So we avoid those regions by removing them to reduce
the error.

Figure 5.1: Flow chart for the data acquisition process for θ′ = 0◦

Power measurements

θ′ Power of LO Power of signal Total power

Epsilon Alpha
(degrees) (mW) (mW) (mW)
0 3.14 0.00643 1:488.34 3.14 1.772
8 2.84 0.263 1:10.8 3.1 1.76
20 1.75 1.28 1:1.37 3.01 1.735
32 0.587 2.39 1:0.25 2.98 1.726
46 0.017 2.95 1:0.0057 2.96 1.72

Table 5.2: Power values for LO and Signal.

While taking the power measurements the power was fluctuating rapidly. These
might be because of fluctuation in the external temperature.

47
Chapter 6

Results and Analysis

After isolating the middle three cycles, the sine fit is done for all the datasets.
Important results to be noted are given as follows:

• By varying θ′ , we control the signal to LO power ratio (ϵ′ ). The ϵ′ is a

function of θ′

• The total power is proportional to |α′ |2 .

• Using this we calculate the mean photon number of the signal and LO as
n̄ = |α′ |2 .

• At 0◦ the total power is 3.14 mW and ϵ′ = 1 : 500. This is the range in

which we have to operate, where the mean photon number of the signal is
much lower than LO.

Plot for signal amplitude variation for different angles from 0◦ to 20◦ is shown
below. The arrow shows the increase in the voltage amplitude of the signal in this

cycle.

48
 6 Results and Analysis

Figure 6.1: Plot for signal amplitude variation for different angles from 0◦ to 20◦ .
The arrow shows the increase in the voltage amplitude of the signal in this cycle.

1. The output has a sinusoidal variation by varying the phase (ϕ) in time. The
amplitude of the output signal varies as a function of phase (ϕ) and we

observe this result.

2. The regime in which we are working is where the signal amplitude is less,

that is when ϵ′ is the lowest. This regime corresponds to when the signal has
much lower photon number when compared to LO. At 0◦ we see that the ϵ′
is the lowest and we observe the signal amplitude to be lowest. But as the

θ′ increases the signal amplitude also increases.

49
 6 Results and Analysis

Figure 6.2: Peak to peak amplitude as a function of θ′

From 5.1 we see that the peak amplitude varies as a function of cos(θ′ ).

6.1 Noise Results

The physical limit at which we take the data is in the regime where the signal has
very low mean photon number. We are changing the phase shift by varying it in
the time scale. The ideal way of measuring the noise is to go to the limit where

the amplitude is very small, then many datasets are taken for a particular phase
and the statistics are calculated at every point. In my experiment I was unable to
trigger the same due to technical difficulties. So the following method was used to

50
 6 Results and Analysis

understand the noise characteristics instead:

• After isolating only the three oscillations from the original data the sine fit
was done. Now the noise can then be obtained by taking the residual for the

sine fit of the output signal.

• This noise as a function of time then as a proxy or a hand waving argument,

can be said to be similar to taking multiple datasets. This gives an idea as

to how the noise changes because in this regime where we are far off from the
ideal regime.I will be proceeding with this assumption for further analysis.

• The noise amplitude after taking the residual for the output signal of θ′ = 0◦
is plotted and shown in Figure 6.3.

51
 6 Results and Analysis

Figure 6.3: Noise amplitude for θ′ = 0◦

The next step is to take the Fast Fourier Transform (FFT) of the noise in the time
domain for the signal processing. This is done mainly because:

1. Frequency Analysis: Noise often contains information at various frequen-

cies. By taking the FFT, you can analyze the frequency content of the noise
and identify any dominant frequencies or patterns.

2. Filtering: In many cases, noise interferes with the signal of interest. By

analyzing the noise in the frequency domain, you can design filters (like

bandpass or notch filters) to remove specific frequency components of the

noise while preserving the signal.

52
 6 Results and Analysis

3. Characterization: FFT helps in characterizing the noise. For example,

white noise has a flat frequency spectrum, while other types of noise may
have distinct spectral characteristics that can be identified through FFT
analysis.

4. Statistical Analysis: FFT can be used to study the statistical properties

of noise, such as its power spectral density (PSD), which describes how the

power of the noise is distributed across different frequencies.

5. System Analysis: In systems where noise plays a significant role, such

as communication systems or electronic circuits, analyzing the noise in the

frequency domain can provide insights into how the system behaves under
different conditions.

In this case the FFT was done for the residual noise using a rectangular window.
Rectangular simply means that the FFT computes the transform of the entire input

signal without any additional weighting or filtering. This can result in spectral
leakage and other artifacts, especially when analyzing signals with sharp transitions
or non-periodic components.

Taking the FFT for the noise signal at θ′ = 0◦ , we get the following Figure 6.4

53
 6 Results and Analysis

Figure 6.4: Spectral noise amplitude for θ′ = 0◦ obtained after taking the FFT.

The major contribution of the low frequency noise could be due to the non-linear

angular motion of the PBD phase shifter. If it was linear we would get a perfect
sine curve, but since it not linear it would slightly deviate from sine curve. Other
reasons for the low frequency noise could be because of the fluctuation of laser

power. This might be because of the laser current fluctuations, cavity fluctuations
or temperature deviations.

Next step is to remove the low frequency noise so that we can get rid of the
noise from the known sources and then obtain spectral amplitude of the noise
for the higher frequencies. The average of the magnitude of noise in the spectral

domain is plotted as shown in Figure 6.5

54
 6 Results and Analysis

Figure 6.5: Spectral noise amplitude vs frequency.

We observe that as θ′ increases the spectral noise amplitude also increases. This
√
is because as θ′ increases the ϵ′ also increase. Now since ∆n = n̄, as n̄ increases
the ∆n also increases. The spectral noise amplitude here corresponds to the noise

in the mean photon number.

From this graph we can question whether we are in the regime where we can probe

the number phase uncertainty relation, because ideally we should have taken the
data at the limit where the mean photon number in one beam (LO) is very large
and the other (signal) one very small. We see that when the amplitude of os-

55
 6 Results and Analysis

cillation is small the noise corresponding to it is also small. We should take the

data in the regime where the amplitude is almost close to zero because now when
the interferometry is performed with the balanced detection, the intensity noise
will be subtracted away and only the phase noise remains, as both of them are

complimentary.

56
Chapter 7

Conclusions and Future prospects

To observe the number phase uncertainty relation we need a variation in ∆θ of
the order of π. Now from the number phase uncertainty principle

∆n∆θ = 1/2 (7.1)

1
=⇒ ∆n = (7.2)
2π
√
∆n = n̄ =⇒ n̄ ≈ O(1) (7.3)

Taking the timescale to be of 100 ns for 1 photon to be observed, we have 107

photons in a second. The power level associated with this timescale is in the order
of 0.1pW to 10pW.

Hence we are about 106 orders of magnitude away from the actual regime to probe
the number phase uncertainty. Ideally we should go to much lower time scales like
in nano seconds to measure the noise. But we should know what to expect, how

the noise should vary and what the classical noises would be.

The Figure 7.1 shows the clear shows the difference between the two regimes that
we require to operate. The marked using the dotted lines represent the ideal
regime at which the number phase uncertainty can be probed. In this region, a

small change in n̄ can cause a significant change in the phase uncertainty. In the
current experimental setup we are only operating in the regime where the curve is
flat. So no matter how much the n̄ changes there will be no significant change in

57
 7 Conclusions and Future prospects

the phase uncertainty.

Figure 7.1: Figure adapted from [12] showing the ideal regime to probe the
number phase uncertainty and the regime we are working currently.

Right now the experiment performed is a basic tool of what needs to be done in the
future, but we are not at the power level at which we can probe the number phase

uncertainty relation. This basic had to be done to learn the operation techniques
of handling the equipment and optical components. This was also performed to
learn the data acquisition techniques and to understand the noise characteristics

of the detector as well as the other sources of noise. There is a saying that “First
remove all the possible sources of classical noise and what remains is quantum”.
So this exercise is a way to identify what are the possible sources of noise which

need not be attributed to quantum limit. The next task would be to perform the
experiment in the low power limit where we expect to observe the number phase
uncertainty relation.

58
 7 Conclusions and Future prospects

Other future prospects of this experiment are:

1. To put ND filters and take data at pW level and using a detector in the ns
timescale.

2. To vary the phase of the HWP between the two PBD’s and take measure-
ments as function of photon number.

3. To setup the previously proposed experiment using phase shifter as waveplate

combination and observe the uncertainty in phase, and see how it varies with

the mean photon number.

4. To reconstruct the Wigner function from the quadrature components. Then

infer the coherence property of the state and the number phase uncertainty.

5. Using a non-linear optical medium like Rubidium atom vapor to generate a

squeezed state and to check how the coherence properties of the state change.

Figure 7.2: Setup using a non-linear optical medium like Rubidium atom vapor.

59
Bibliography
[1] Sergio Bartalucci, VI Vysotskii, and MV Vysotskyy. “Correlated states and
nuclear reactions: An experimental test with low energy beams”. In: Physical
Review Accelerators and Beams 22.5 (2019), p. 054503.
[2] M Beck et al. “Experimental determination of number–phase uncertainty
relations”. In: Optics letters 18.15 (1993), pp. 1259–1261.
[3] Max Born and Emil Wolf. Principles of optics: electromagnetic theory of
propagation, interference and diffraction of light. Elsevier, 2013.
[4] P Carruthers and MM Nieto. “Coherent states and the number-phase un-
certainty relation”. In: Physical review letters 14.11 (1965), p. 387.
[5] Christopher C Gerry and Peter L Knight. Introductory quantum optics. Cam-
bridge university press, 2023.
[6] Roman Jackiw. “Minimum Uncertainty Product, Number-Phase Uncertainty
Product, and Coherent States”. In: Journal of Mathematical Physics 9.3
(1968), pp. 339–346.
[7] Robert Lynch. “Fluctuation of the Barnett-Pegg phase operator in a coherent
state”. In: Physical Review A 41.5 (1990), p. 2841.
[8] MS Windows NT Kernel Description. https://sitn.hms.harvard.edu/
flash/2016/quest-gravitational-waves-pushes-boundaries-quantum-
optics/. Accessed: 2024-05-15.
[9] MS Windows NT Kernel Description. https://www.thorlabs.com/thorproduct.
cfm?partnumber=BD40. Accessed: 2024-05-15.
[10] MS Windows NT Kernel Description. https://www.thorlabs.com/thorproduct.
cfm?partnumber=KDC101. Accessed: 2024-05-15.
[11] Uwe Schilling. Measurements in Quantum Optics. Friedrich-Alexander-Universitaet
Erlangen-Nuernberg (Germany), 2011.
[12] DT Smithey et al. “Measurement of the Wigner distribution and the density
matrix of a light mode using optical homodyne tomography: Application to
squeezed states and the vacuum”. In: Physical review letters 70.9 (1993),
p. 1244.
[13] K Vogel and H Risken. “Determination of quasiprobability distributions in
terms of probability distributions for the rotated quadrature phase”. In:
Physical Review A 40.5 (1989), p. 2847.

60