General rights Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.

Users may download and print one copy of any publication from the public portal for the purpose of private study or research.

You may not further distribute the material or use it for any profit-making activity or commercial gain

You may freely distribute the URL identifying the publication in the public portal If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.

Downloaded from orbit.dtu.dk on: Aug 30, 2020

Supercontinuum laser for broadband spectroscopy using upconversion

Huot, Laurent

Publication date:2018

Document VersionPublisher's PDF, also known as Version of record

Link back to DTU Orbit

Citation (APA):Huot, L. (2018). Supercontinuum laser for broadband spectroscopy using upconversion. Technical University ofDenmark.

Supercontinuum laser for broadband spectroscopyusing upconversion

Submitted in partial fulfillment of the requirements for the degree of

Doctor of Philosophy

by

Laurent Huot

September, 2018

DECLARATION

I hereby declare that the thesis entitled “Supercontinuum laser for broadband

spectroscopy using upconversion” submitted by me, for the award of the degree of

Doctor of Philosophy to the Technical University of Denmark is a record of bonafide

work carried out by me under the supervision of Dr. Christian Pedersen, DTU Fotonik,

Technical University of Denmark, Frederiksborgvej 399 Bygning 108, rum S08 4000

Roskilde.

I further declare that the work reported in this thesis has not been submitted

and will not be submitted, either in part or in full, for the award of any other degree or

diploma in this institute or any other institute or university.

Roskilde 18.09.2018

Laurent Huot

CERTIFICATE

This is to certify that the thesis entitled “Supercontinuum laser for broadband

spectroscopy using upconversion” submitted by Mr. LAURENT HUOT, DTU Fotonik,

Technical University of Denmark, Frederiksborgvej 399 Building 108, 4000 Roskilde

for the award of the degree of Doctor of Philosophy, is a record of bonafide work carried

out by him under my supervision.

The contents of this report have not been submitted and will not be submitted

either in part or in full, for the award of any other degree or diploma in this institute or

any other institute or university. The thesis fulfills the requirements and regulations of

the University and in my opinion meets the necessary standards for submission.

Roskilde 18.09.2018

Signature of the Supervisor

(Dr. Christian Pedersen)

ABSTRACT

This thesis describes an experimentally based, and application oriented investiga-

tion of the possibilities offered by the combination of upconversion detection methods

developed at DTU Fotonik with mid-infrared supercontinuum sources built at NKT

Photonics.

The first introductory chapter provides a brief overview of prior work in the field of

upconversion detection and presents the motivations and goals of the Ph.D. project.

Chapters 2 and 3 cover the relevant physics of upconversion detection and supercon-

tinuum generation as support for subsequent discussions and analyses. These chapters

also include construction notes and practical design aspects of pulsed upconversion se-

tups using supercontinuum illumination.

Chapter 4 focuses on the experimental work achieved during this project. The exper-

iments are based on synchronous broadband upconversion of mid-infrared (MIR) pulses

from a supercontinuum source and include both imaging and spectroscopy setups. The

experimental results are presented in the order in which they were realized.

The thesis is concluded in chapter 5 with a summary of the main results of this

project followed by a discussion about the possible improvements that could be brought

to synchronous pulsed upconversion in the future as well as its strong potential for many

key applications.

Keywords: Mid-infrared supercontinuum, Upconversion detection, Infrared imaging,

Spectroscopy, Nonlinear frequency conversion.

i

RESUME

Denne afhandling beskriver en eksperimentel undersøgelse af opkonversionsdetek-

tion/afbilding udviklet ved DTU Fotonik, implementeret med mid-infrarøde supercon-

tinuumkilder konstueret ved NKT Photonics.

Det første indledende kapitel giver et kort overblik over tidligere arbejder inden

for opkonversionsdetektion/billeddannelse og afstikker mal og rammer for dette Ph.D

arbejde.

Kapitel 2 og 3 præsenterer den relevante fysik bag henholdsvis opkonversionsde-

tektering og superkontinuumgenerering og bruges ved de efterfølgende diskussioner og

analyser. Disse 2 kapitler indeholder ogsa konstruktionsnoter og praktiske designaspek-

ter anvendt i dette Ph.D arbejde.

Kapitel 4 fokuserer pa det eksperimentelle arbejde, der blev udført i projektperi-

oden. Eksperimenterne er baseret pa synkroniseret bredbands-opkonversion af mid-

infrarøde (MIR) pulser genereret af en supercontinuum kilde. Det eksperimentelle

arbejde inkluderer bade opstillinger til billeddannelse og spektroskopi. Resultaterne

præsenteres i den rækkefølge, hvori de blev realiseret.

Afhandlingen afsluttes med et resume af hovedresultaterne samt en diskussion af

mulige forbedringer, der kunne implementeres i fremtiden, samt metodens store poten-

tiale inden for mange vigtige anvendelser.

ii

PREFACE

This thesis presents the outcome of a Ph.D. study performed from 2015 to 2018 at

the Technical University of Denmark at the Department for Photonics Engineering in

the Optical Sensor Technology Group in collaboration with the company NKT Pho-

tonics. Work relating to the construction of supercontinuum and fiber laser sources

was performed at NKT Photonics while the upconversion experiments were done in the

laboratory facilities of the Risø Campus of the Technical University of Denmark. The

thesis represents original work done by me unless it is stated otherwise. It is a mix

of published and unpublished work, which will be specified in the appropriate sections

when relevant. This research was conducted as part of the Mid-TECH project which

has received funding from the European Union’s Horizon 2020 research and innovation

programme under Grant Agreement No. 642661.

Acknowledgements

With immense pleasure and deep sense of gratitude, I wish to express my sincere

thanks to my supervisors and co-supervisors Dr. Christian Pedersen, Dr. Peter

Tidemand-Lichtenberg, from DTU and Dr. Peter Moselund and Dr. Lasse Leick

from NKT Photonics without whose motivation and continuous encouragement, this

research could not have been successfully completed. Additionally, I would like to ac-

knowledge the support rendered by my colleagues from NKT Photonics and the Optical

Sensor Technology group at the Technical University of Denmark.

Place: Roskilde

Date: 15/09/2018 Laurent Huot

iii

Mid-TECH ITN project

The Mid-TECH project is a Marie Skłodowska-Curie Actions Innovative Training

Network (ITN), providing training for 15 Early Stage Researchers (ESRs) working to-

wards a doctoral degree. The ESR group is composed of 4 female and 11 male fellows.

The project brings together a consortium comprising 6 academic institutions and 2 com-

panies, and is coordinated by the Technical University of Denmark (DTU). The Mid-

TECH program aims at combining novel mid-infrared (mid-IR) light sources, mid-IR

upconversion detection and mid-IR applications.

The mid-IR wavelength range is an emerging and important new research frontier.

Its general importance relates to a multitude of mid-IR industrial and biomedical sensor

applications. Chemical fingerprints of most complex molecules such as those found in

food, human tissue or combustion compounds all have vibrational absorption features in

the mid-IR, thus identifiable through mid-IR spectroscopy. Incidentally, the fundamen-

tal absorption bands of gas molecules are also located in the mid-IR which enables the

development of novel instrumentation for mid-IR gas spectroscopy for measurement

of small concentrations. This is relevant for applications like “leak-tests” or remote

sensing of greenhouse gases.

The main obstacle for the exploitation of the mid-IR optical range has been a lack

of efficient mid-IR light sources and sensitive mid-IR detectors. In Mid-TECH we have

gathered the best European academic and industrial partners to show that in a combined

effort, both technological shortcomings can be overcome, paving the way for novel

instrumentation for industry and society.

Mid-TECH aims to achieve an ambitious set of objectives both in relation to science

and training:

• Train a group of 15 highly skilled ESRs forming a new generation of networked

scientists

• Develop and discover new technologies for the mid-IR

• Technology transfer from university to industry

• Promote new innovation and entrepreneurial behaviour

iv

• Demonstrate novel instrumentation for society and industry

In the Mid-TECH program we deploy the methodology needed for a given mid-IR ap-

plication, e.g. spectral imaging of cancerous tissue. Instrument designs (measurement

principle and specifications) will emerge, including two key elements: a mid-IR light

source and a mid-IR detector. From the European academic and industrial scene we

have matched specialised “top-notch” partners for each of the 3 research-oriented tasks,

i.e. the application, the light source and the detection. This approach has led to the

formation of a consortium of partners with excellence in each of the three fields, com-

plementary in skills but all sharing the same common objective.

Participants:

• LUNDS UNIVERSITET Sweden

• THE UNIVERSITY OF EXETER United Kingdom

• FORSCHUNGSVERBUND BERLIN EV Germany

• HUMBOLDT-UNIVERSITAET ZU BERLIN Germany

• FUNDACIO INSTITUT DE CIENCIES FOTONIQUES Spain

• NKT PHOTONICS A/S Denmark

• RADIANT LIGHT SL Spain

Partner organizations:

• HALDOR TOPSOE AS Denmark

• FOSS ANALYTICAL AS Denmark

• EAGLEYARD PHOTONICS GmbH Germany

• QUANTIOX GmbH Germany

• ROYAL COLLEGE OF SURGEONS IN IRELAND Ireland

• IRSEE Aps Denmark

v

The Mid-TECH project has received funding from the European Union’s Horizon 2020

research and innovation program under Grant Agreement No. 642661. The Mid-TECH

consortium comprises 8 European project partners, academic as well as industrial.

http://www.midtech-itn.eu/

vi

LIST OF PUBLICATIONS

First author publications

1. Journal paper

Laurent Huot, Peter M. Moselund, Peter Tidemand-Lichtenberg, Lasse Leick,

and Christian Pedersen, ”Upconversion imaging using an all-fiber supercontin-

uum source,” Opt. Lett. 41, 2466-2469 (2016)

2. Conference paper

Laurent Huot, Peter M. Moselund, Lasse Leick, Peter Tidemand-Lichtenberg,

Christian Pedersen, ”Broadband upconversion imaging around 4 µm using an

all-fiber supercontinuum source,” Proc. SPIE 10088, Nonlinear Frequency Gen-

eration and Conversion: Materials and Devices XVI, 100880J (20 February 2017)

3. Journal paper

Laurent Huot, Peter M. Moselund, Peter Tidemand-Lichtenberg, and Christian

Pedersen, ”Electronically delay-tuned upconversion cross-correlator for charac-

terization of mid-infrared pulses,” Opt. Lett. 43, 2881-2884 (2018)

4. Conference presentation

Laurent Huot, P. M. Moselund, P. Tidemand-Lichtenberg, and C. Pedersen, ”Char-

acterization of infrared pulses using upconversion,” in High-Brightness Sources

and Light-driven Interactions, OSA Technical Digest (online) (Optical Society of

America, 2018), paper MT1C.3.

5. Journal paper (submitted)

Laurent Huot, Peter M. Moselund, Peter Tidemand-Lichtenberg, and Christian

Pedersen, ”Pulsed upconversion imaging of mid-infrared supercontinuum light

using an electronically synchronized pump laser,” Appl. Opt.

vii

Co-author publications

1. Conference paper

I. D. Lindsay, S. Valle, J. Ward, G. Stevens, M. Farries, L. Huot, C. Brooks, P. M.

Moselund, R. M. Vinella, M. Abdalla, D. de Gaspari, R. M. von Wurtemberg, S.

Smuk, H. Martijn, J. Nallala, N. Stone, C. Barta, R. Hasal, U. Moller, O. Bang, S.

Sujecki, A. Seddon, ”Towards supercontinuum-driven hyperspectral microscopy

in the mid-infrared,” Proc. SPIE 9703, Optical Biopsy XIV: Toward Real-Time

Spectroscopic Imaging and Diagnosis, 970304 (7 March 2016);

2. Conference paper

Peter M. Moselund, Laurent Huot, Chris D. Brooks, ”All-fiber mid-IR supercon-

tinuum: a powerful new tool for IR-spectroscopy,” Proc. SPIE 9703, Optical

Biopsy XIV: Toward Real-Time Spectroscopic Imaging and Diagnosis, 97030B

(7 March 2016);

3. Journal paper

Michael Hermes, R. Brandstrup Morrish, L. Huot, L. Meng, S. Junaid, J. Tomko,

G. R. Lloyd, W. T. Masselink, P. Tidemand-Lichtenberg, C. Pedersen, F. Palombo

and N. Stone, ”Mid-IR hyperspectral imaging for label-free histopathology and

cytology,” Journal of Optics 20, number 2, 023002, (2018)

4. Conference paper

Peter M. Moselund, Patrick Bowen, Laurent Huot, Joanna Carthy, Ross Powell,

Lucy Hooper, ”Compact low-power mid-IR supercontinuum for sensing applica-

tions,” Proc. SPIE 10540, Quantum Sensing and Nano Electronics and Photonics

XV, 105402I (26 January 2018)

5. Journal paper

Christian R. Petersen, Peter M. Moselund, Laurent Huot, Lucy Hooper, Ole Bang,

Towards a table-top synchrotron based on supercontinuum generation, Infrared

Physics and Technology, Volume 91, June 2018, Pages 182-186, ISSN 1350-

4495, https://doi.org/10.1016/j.infrared.2018.04.008.

6. Conference presentation

M. Hermes et al., ”Towards rapid high-resolution mid-IR imaging for molecular

viii

spectral histopathological diagnosis of oesophageal cancers,” 2017 Conference

on Lasers and Electro-Optics Europe and European Quantum Electronics Con-

ference (CLEO/Europe-EQEC), Munich, 2017, pp. 1-1. doi: 10.1109/CLEOE-

EQEC.2017.8087789

ix

x

TABLE OF CONTENTS

ABSTRACT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . i

RESUME . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ii

PREFACE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iii

MID-TECH ITN PROJECT . . . . . . . . . . . . . . . . . . . . . . . . . . . iv

LIST OF PUBLICATIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii

1 Introduction 1

1.1 Historical background . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

1.2 Prior work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

1.3 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

1.4 Project goals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

2 MIR SC generation 7

2.1 History of SC generation . . . . . . . . . . . . . . . . . . . . . . . . . . 7

2.2 SC generation mechanisms . . . . . . . . . . . . . . . . . . . . . . . . . 9

2.3 Architecture of a MIR SC source . . . . . . . . . . . . . . . . . . . . . . 11

2.3.1 Reliability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

3 Upconversion detection 13

3.1 History of parametric frequency conversion . . . . . . . . . . . . . . . . 13

3.2 Sum frequency generation . . . . . . . . . . . . . . . . . . . . . . . . . . 14

3.3 Non-collinear phase-matching for broadband MIR upconversion . . . . . 15

3.4 Architecture of upconversion imaging setup . . . . . . . . . . . . . . . . 18

3.5 Using off-axis parabolic mirrors for upconversion . . . . . . . . . . . . . 20

4 Applications of pulsed upconversion of MIR SC 23

xi

4.1 Pulsed upconversion imaging of SC light using a single 1550 nm source

both for generating the SC and pumping the upconversion process . . . . 24

4.1.1 Upconversion imaging of 2 µm to 2.6 µm in bulk lithium niobate . 24

4.1.2 Upconversion around 4 µm using AGS crystal . . . . . . . . . . . 31

4.2 Pulsed upconversion of SC using an electronically synchronized MOPA

laser for pumping the upconversion process . . . . . . . . . . . . . . . . 36

4.2.1 Electronically delay-tuned cross-correlator . . . . . . . . . . . . . 38

4.2.2 Pulsed upconversion imaging of mid-infrared supercontinuum light

using an electronically synchronized pump laser . . . . . . . . . . 47

5 Outlook and conclusion 57

5.1 Outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57

5.2 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57

REFERENCES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59

Appendices

Appendix A List of Terms and Abbreviations 69

Appendix B List of figures 71

xii

xiii

xiv

CHAPTER 1

Introduction

This first introductory chapter provides a brief overview of prior work in the field of

upconversion detection and presents the motivations and goals of the Ph.D. project.

Chapters 2 and 3 cover the relevant physics of supercontinuum (SC) generation and

ucponversion detection and include construction notes and practical design aspects of

their implementation. These chapters support the discussions and analysis throughout

chapter 4.

Chapter 4 focuses on the experimental work achieved during this project. The exper-

iments are based on synchronous broadband upconversion of mid-infrared (MIR) pulses

from a SC source and include both imaging and spectroscopy setups. The experimental

results are presented in the order in which they were conducted.

The thesis is concluded in chapter 5 with a summary of the main results of this

project followed by a discussion about the possible improvements that could be brought

to synchronous pulsed upconversion in the future as well as its strong potential for many

key applications.

1

1.1 Historical background

The discovery of infrared (IR) radiation by William Herschel in 1800 [1] marked the

start of over two centuries of inventions and innovations in the field of IR detection.

Initially, IR detection was achieved using traditional thermometers. The 19th century

witnessed the invention of the first thermocouples, thermopiles and bolometers [2; 3].

The discovery of the photoconductive effect led to the appearance of the IR first photon

detectors during the early 20th century. These have, from 1930 onwards, dominated the

development of IR technologies [4]. Today, the mid-infrared (MIR) presents a strong

potential for a wide range of applications in many important fields like biomedical imag-

ing, environmental monitoring, food analysis, combustion analysis, civil and defense

applications [5; 6; 7]. This is because fundamental absorption lines of gases as well as

vibrational spectra of many complex compounds have distinct spectroscopic features

in this wavelength range. However, the development of these applications has been

slowed down because the development of fast, low-noise detectors in this wavelength

range remains a major challenge today. Many present-day mid-IR detectors are based

on low band-gap semiconductor materials like indium antimonide (InSb) or mercury

cadmium telluride (HgCdTe). Alternatively, micro-bolometer arrays and thermopiles

are used. All these detectors suffer from inherent dark noise and require cooling to

perform optimally. These detection systems are usually very expensive and have slow

response times that make them unsuitable for many applications. Additionally, their

requirement for either Stirling or liquid nitrogen cooling makes their integration into

compact reliable systems significantly more difficult [4].

Some MIR spectroscopy and hyperspectral imaging applications rely on Fourier-

Transform Infrared Spectroscopy (FTIR) techniques. However, the complexity and cost

associated with FTIR can be prohibitive for many applications. Specifically, FTIR relies

on very accurate scanning of a mirror over relatively long distances with nanometer

accuracy. This can be a big challenge in industrial settings where mechanical vibrations

are often present, limiting wider use in many industries.

With the increasing availability of high intensity lasers and nonlinear crystals, fre-

quency upconversion constitutes an appealing alternative to traditional detectors. As

shown in [8], upconversion detection circumvents the difficulties of mid-IR detection

2

by translating the mid-IR signal to the near-IR wavelength region by nonlinear para-

metric sum frequency mixing in a χ2 medium [9]. The upconverted signal can then be

detected using an affordable silicon-based detector operating in the near-IR region. A

schematic representation of such a device is presented in fig. 1.1.

Fig. 1.1 Schematic representation of the upconversion detection process. Reproducedfrom [8]

These sensors are known to work with much better noise and time performance than

their mid-IR counterparts and rarely require active cooling. The potential increase in

detectivity of upconversion detection is illustrated in fig. 1.2. Moreover, silicon based

sensors are widely available and affordable.

Fig. 1.2 Overview of detector technology available for different wavelengths and theirspecific detectivities. Reproduced from [10]

3

1.2 Prior work

Frequency upconversion of IR images was already under investigation in the 1960s and

1970s [11]. However, this field was virtually abandoned 30 years ago, probably due to

prohibitively low quantum efficiencies (QE). Since then, thanks to the advances in the

development of high brightness light sources and high quality nonlinear crystals, this

technology has received renewed interest from researchers [12].

Recently, DTU demonstrated a QE of 2.10−4 for incoherent image upconversion

with 200 x 1000 pixel elements from the red to blue spectral region [13]. These results

demonstrate the potential of the technology [14]. It was also shown that such a device

could be made into a field-deployable, rugged image upconversion device, which can

be attached directly to a regular charge-coupled device (CCD) camera [8]. In recent

years CW upconversion has mainly been used for applications such as upconversion

spectroscopy and QE of up to 45% have been reported [15]. This is an improvement of

six orders of magnitude over the QE achieved in early CW IR image upconversion ex-

periments [13]. The physics behind the operation of upconversion detectors is covered

in chapter 3.

In parallel, many developments in MIR sources have been made over the past

decades, and today, a vast array of sources in the MIR are commercially available.

These sources include Globars, Optical Parametric Oscillators (OPO), Quantum Cas-

cade Lasers (QCL), synchrotron illumination and most recently MIR SC sources which

will be at the center of our attention in this study. MIR SC light sources are broadband

fiber-based sources that exhibit excellent beam quality, high spectral brightness and can

be made to be compact, robust and affordable. This makes them very attractive for

many practical applications. Figure 1.3 compares the spectra and brightness of various

SC sources against synchrotron and Globar illumination. Throughout this work, mainly

fluoride fiber based SC has been studied. The properties, theory and construction of

MIR SC sources will be covered in detail in chapter 2.

In this project, novel upconversion methods were explored using SC lasers. Various

imaging and spectroscopy were conducted to investigate the advantages and disadvan-

tages of SC illumination for MIR upconversion detection.

4

Fig. 1.3 Brightness of silica, fluoride, and chalcogenide fiber-based SC sources com-pared to a synchrotron, the sun (5778 K black-body), and a Globar (1500 K black-body).This figure was reproduced from [16]

1.3 Motivation

The upconversion process, being a nonlinear parametric process, greatly benefits from

being implemented in a pulsed manner. The high peak power of the conversion pump

pulses ensures the upconversion of the IR signal pulses and suppresses the need for

the power enhancement that intra-cavity systems are known to provide [12]. A single-

pass setup with synchronous pumping [17] relaxes the demanding constraints for the

nonlinear material to have high nonlinearity and low transmission losses. This allows

the use of bulk crystals which can easily be manufactured in sizes of the order of several

mm. Their larger aperture permits the use of a large beam diameter in the nonlinear

crystal which increases spatial resolution of imaging systems[12].

Figure 1.4 shows the spectral ranges within which phase matching can be achieved

in various nonlinear crystals. This goes to show that SC sources located further in the

IR spectrum may be upconverted using a similar method exploiting crystals like AgGaS

for example. Additionally, thermal noise being proportional to the average pump power,

pulsed upconversion systems are far less sensitive to thermal noise than continuous

wave upconversion setups since they generally operate at much lower average pump

powers [18].

Therefore, combination of novel high brightness mid-IR SC lasers and ultra-sensitive

upconversion detection opens a unique possibility for combining two leading edge

technologies pointing towards extremely fast and/or extremely sensitive imaging and

5

Fig. 1.4 Overview of the different nonlinear crystals used in parametric light sourcesreported in literature. Note the difference in y-axis on the two plots. Reproduced from[19] and [20], left and right, respectively.

spectroscopy in the mid-IR range. Fast spectral data acquisition can be used for high

throughput industrial monitoring or improved signal-to-noise for detection of faint sig-

nals using averaging of data.

1.4 Project goals

The overall goal of the project was to investigate the possibilities of using MIR SC

illumination for upconversion imaging and spectroscopy. Three major sub-goals were

identified and will be presented in the subsequent chapters of this work:

• Building and testing of a MIR SC source. This part of the work was completed at

NKT Photonics

• Designing and building an upconversion setup implementing the MIR SC source

• Testing of various imaging and spectroscopy setups based on the upconversion

setup

6

CHAPTER 2

MIR SC generation

This chapter presents the information pertaining to MIR SC generation that is relevant to

this project. After a brief account of the history of SC generation, the physical processes

at play will be introduced. The following sections deal with the architecture of fiber

based MIR SC sources, their performance and some of the practical aspects of building

and working with such devices.

2.1 History of SC generation

The term SC generation refers to the process in which a narrow band optical pump pulse

undergoes significant spectral broadening through the combined effect of a number of

nonlinear optical effects in the propagation medium, to yield a broadband spectrally

continuous output.

SC generation was first observed in bulk borosilicate glass in 1970 [21]. It has since

been widely studied and has found many applications in fields such as spectroscopy,

pulse compression, frequency metrology, optical coherence tomography and confocal

microscopy. SC generation in optical fibers in particular has received a lot of attention

due to the unique advantages offered by their long optical interaction lengths, high non-

linearity, and potential applications in optical telecommunications. It was first demon-

strated in multimode fiber in 1974 [22] and then in single-mode fiber in 1978 [23].

In the late 1990s, the emergence of photonic crystal fibers (PCF) spurred great in-

terest throughout the scientific community. Indeed, the wide range of design parame-

ters of PCF fibers allows them to be tailored for single-mode propagation over broad

wavelength ranges, enhanced modal confinement and thus, elevated nonlinearity. Ad-

ditionally, their group velocity dispersion (GVD) could be engineered and optimized.

Thus, PCFs led to the birth of the ”modern” SC as we know it and enabled high bright-

7

ness single-mode ultra-broadband SC sources [24]. Today, visible light supercontinuum

sources are available commercially, for example NKT’s SuperK Extreme shown in Fig.

2.1.

(a)(b)

Fig. 2.1 (a)NKT’s flagship SC source, the SuperK Extreme. (b) Spectrum of a SuperKExtreme EXW 12 source.

SC generation in PCF has been studied extensively and various pumping schemes

have been investigated. The mode-locked ytterbium (Yb) fiber lasers at 1060 nm are

often used to generate light in the visible region [25; 26]. However, the use of a mode-

locked laser has one primary drawback, which is the lack of average power scalability

due to a fixed repetition rate, often in the tens of Mhz range, dictated by the cavity

roundtrip time. SC systems based on microchip laser and master-oscillator power am-

plifier (MOPA) type pumps on the other hand enable power scaling through tuning of

the repetition rate. Based on this method, a 5.3 W SC ranging from 0.8 to 3 µm, gen-

erated in highly nonlinear silica fiber using nanosecond diode pulses amplified with a

multistage amplifier, was demonstrated in [27].

Due to its IR absorption limit, SC generated in silica fibers typically does not extend

beyond 3 µm [28]. In order to generate SC in the mid-IR, optical fibers with low loss in

the mid-IR windows, such as fluoride, chalcogenide and tellurite are required. We will

hereinafter focus on SC generated in ZBLAN (ZrF4−BaF2−LaF3−AlF3−NaF )

fibers.

ZBLAN fibers are a type of fluoride glass fibers which have received a lot of at-

tention for SC generation in the mid-IR. Although, ZBLAN has a lower nonlinearity

than that of silica, ZBLAN glasses are superior to other soft glasses for high-power

SC generation in the mid-IR. This is due to their much lower background loss, their

8

comparatively high stability which allows them to be drawn into fibers, and a wide

transparency window in the mid-IR region [29] as shown in Fig. 2.2.

Fig. 2.2 Optical losses from three kinds of fluoride fibres compared to Si02. Reproducedfrom [29]

The first SC generated in ZBLAN fibers extended from 1.8 µm to 3.4 µm, with a

total average power of 5 mW [30]. Since then, ZBLAN SC have greatly improved.

For example, high power MIR SC generation in a single-mode ZBLAN fiber with up

to 21.8 W average output power from 1.9µm to beyond 3.8 µm was reported in [31].

In [32], a fully packaged all-fiber turn-key low noise 4.8 W 1.8-4.2 µm ZBLAN SC

was demonstrated. In the past years, the technology has finally matured sufficiently for

the first commercial turn-key MIR SC sources to start becoming available, for example

NKT’s MIR SuperK source shown in Fig. 2.3.

2.2 SC generation mechanisms

SC sources in the IR have also been demonstrated in various pump and fiber config-

urations. Two approaches to generating SC are widely used: pumping a short length

of nonlinear fiber using sub-picosecond pulses with high peak power, or using CW or

quasi-CW with lower peak power to pump longer lengths of fiber. Fig. 2.4 provides a

summary of the dominant mechanisms for the different pulse and dispersion regimes as

described in [17].

9

(a) (b)

Fig. 2.3 (a)NKT’s new MIR SuperK source (b) Typical spectrum of the NKT MIRSuperK Extreme source.

Throughout this project, the supercontinuum was pumped with ns pulses at tele-

com wavelengths in the anomalous GVD regime of the fiber. This allows us to take

advantage of the high maturity and availability of high power fiber amplifiers and fiber

coupled laser diode pumps.

In this pumping regime, the pulses first undergo modulation instability (MI) dynam-

ics. MI is a phenomenon whereby deviations from a periodic waveform are reinforced

by nonlinearity, leading to the generation of spectral sidebands and the eventual breakup

of the waveform into a train of pulses. The ns pump pulses are thus broken down into

shorter duration sub-pulses. The long wavelength generation is then due to the Raman

induced soliton self frequency shift (SSFS) which causes the sub-pulses to redshift as

they propagate through the fibers [33; 34].

Fig. 2.4 Summary of dominant mechanisms for SC generation in different pulse anddispersion regimes. Reproduced from [35]

10

It is worth noting that MI arises spontaneously from noise and is thus an inher-

ently random process and destroys any temporal coherence the pump pulse might have

possessed. As a consequence, supercontinuum sources pumped with ns pulses in the

anomalous GVD regime generally exhibit significant pulse to pulse fluctuation in their

power spectral density and temporal shape [33; 34].

2.3 Architecture of a MIR SC source

For the purpose of this project, we have developed a fully packaged turn-key mid-IR SC

source based on amplified gain-switched diodes. A rough sketch of the system build-up

can be seen in Fig. 2.5.

Fig. 2.5 Architecture of the MIR SC used throughout this project

In order to take advantage of the high maturity and low cost of telecom components,

we based our system around an off-the-shelf 1550 nm amplified gain-switched diode

laser. This laser emits ns pulses at a repetition rate of 40 kHz and requires an external

triggering signal.

During propagation in silica, the pulses are broken up into sub-pulses and start red-

shifting until they reach a wavelength slightly beyond 2 µm. A section of thulium-doped

fiber was used to absorb the majority of the remaining light at 1550 nm in order to pro-

vide additional gain at 2 µm. Finally, the light was then coupled into a single-mode

ZBLAN fiber to broaden the spectrum to cover approximately 1.8 µm 4.2 µm.

Additionally, a 10% fiber splitter was added after the pump laser. The portion of

the pump light that was split off will be used in chapter 4 to pump the upconversion

process in our first experiments. The length of the fibers was adjusted so that both the

supercontinuum pulses and the extracted pump light propagate through approximately

the same length of fiber and are emitted in temporal synchronism.

11

An image of the packaged system can be seen in Fig. 2.6. Two fibers can be seen

exiting the front panel of the case. The black armoured fiber carries the supercontinuum

while the yellow fiber carries the tapped 1550 nm pump.

Fig. 2.6 MIR SC source built for this project

2.3.1 Reliability

Today, the construction of robust high brightness ZBLAN based SC sources remains

challenging. Handling ZBLAN fibers in particular can prove difficult due to its brit-

tleness compared to standard silica fibers. Additionally, a serious problem with these

sources is the connection between the silica and fluoride fiber. Indeed, the melting tem-

perature for both fibers is vastly different and thus prohibits the use of standard fiber

splicing methods. The use of mechanical gluing methods and fusion arc splicers have

been demonstrated [36; 37; 38], and great effort was made throughout this project on

improving the reliability of ZBLAN based SC sources.

A consequence of the low reliability of ZBLAN fiber connections was that the SC

source used for this project broke down and was repaired multiple times. As a result,

the average power and spectrum of the SC changed throughout the sections of chapter

4.

12

CHAPTER 3

Upconversion detection

The third chapter of this thesis starts by presenting the history and basic theory of up-

conversion detection. The following sections of this chapter will put emphasis on the

theory of non-collinear phase matching and the practical aspects of implementing a

synchronous pulsed upconversion setup using SC illumination.

3.1 History of parametric frequency conversion

While some nonlinear optical effects have been studied as early as 1941 [39], the scien-

tific field of nonlinear optics is often considered to have been spurred by the demonstra-

tion of the first working laser in 1960 [40]. Indeed, second-harmonic generation was

demonstrated in 1961 [41] and many nonlinear optical effects were discovered in quick

succession in the following years. These include two-photon absorption, stimulated

Raman scattering, phase matching in parametric processes, reflected harmonic beams,

third harmonic generation, and anti-Stokes frequency mixing. A comprehensive theo-

retical framework for nonlinear optics was constituted as early as 1965 [42].

In 1967, Midwinter and Warner [43] demonstrated 1% quantum efficiency for up-

conversion of the 1.7072-µ Hg emission line, mixing with a Q-switched ruby laser in a

1 cm long lithium niobate bulk crystal. Temperature sweeping of the phase-match con-

dition was used for measuring the Hg-lamp spectrum and the upconversion technology

was discussed for the use as an infrared detector. As mentioned in chapter 1, the field

of upconversion detection was virtually abandoned for several decades. Thanks to the

advances in the development of high brightness light sources and high quality nonlinear

crystals, this technology has attracted a lot of attention in recent years [12; 8; 14; 15].

13

3.2 Sum frequency generation

Sum-frequency generation (SFG) is a second order nonlinear optical process in which

two photons at frequencies ω1 and ω2 interact inside a material exhibiting χ(2) nonlin-

earity. Under the right conditions of phase-matching, the two photons are annihilated

while a new photon at a frequency of ω3 = ω1 + ω2 is generated simultaneously as

illustrated in Fig. 3.1.

(a)

(b)

Fig. 3.1 Sum-frequency generation. (a) Geometry of the interaction. (b) Energy leveldescription

The relation between the frequencies of the input photons and output photons illus-

trates the conservation of energy in the SFG process. The energy conservation equation

can be written in terms of wavelengths in free space according to eq. 3.1.

1

λ3=

1

λ1+

1

λ2(3.1)

Let us consider a sum-frequency experiment in which two fields of intensity I1 and

I2 at frequencies ω1 and ω2 and k-vectors ~k1 and ~k2 are overlapped inside a material

exhibiting χ(2) nonlinearity. The fields are considered to be plane waves and collinear.

The medium is assumed to be lossless and the slowly varying amplitude approximation

is considered valid [9]. It can be shown that the intensity I3 of the generated field at

frequency ω3 = ω1 + ω2 with the k-vector ~k3 is

I3 =8d2effω

23I1I2

n1n2n3ε0c2L2sinc2

(∆kL

2

), (3.2)

where deff is the effective nonlinearity n1, n2 and n3 are the refractive indices of

the medium at frequencies ω1, ω2 and ω3 respectively, L is the length of the crystal and

14

∆k = k1 +k2−k3 is the phase mismatch [9]. From eq. 3.2 it is clear that the efficiency

of the sum-frequency generation process decreases dramatically as |∆k|L increases.

3.3 Non-collinear phase-matching for broadband MIR upconversion

In the case of MIR upconversion, one of the input fields is the MIR signal while the other

is the pump of the upconversion process. The newly generated field is the upconverted

field. This method enables the detection of very weak MIR signals and even single

photons [8]. From this section onward, symbols with the subscripts 1, 2 and 3 refer to

the MIR signal, pump field and upconverted field respectively. As shown in the previous

section, for a given pump beam k-vector, the MIR k-vectors that will lead to efficient

upconversion are those that satisfy:

~∆k = ~k1 + ~k2 − ~k3 = 0 (3.3)

Solving the phase-matching condition becomes analogous to the plane geometry

problem shown in Fig. 3.2. Non-collinear phase-matching is a particularly interesting

method for achieving broadband upconversion. By tuning the angles of the input fields

and the rotation of the nonlinear crystal, a vast range of wavelengths can be upconverted

with a single crystal. The calculation of the non-collinear phase-matching condition is

described in detail in [44], and we will, in this section, provide a short summary of the

resolution method.

Fig. 3.2 Geometric representation of the phase matching condition

Figure 3.3 illustrates the coordinates and parameters used throughout the phase-

match calculations. In the following calculations, ui is the internal angle between the ~ki

vector projected onto the xz-plane, and the direction of the pump laser field. The cor-

responding angle in the yz-plane is called vi . The pump laser field inside the crystal is

15

considered parallel to the z-axis. ρi describes internal angles in the crystal. The crystal

is considered to be uniaxial birefringent. For the purpose of this demonstration, type

I phase-matching will be considered, though the calculation of phase-matched wave-

lengths and angles remains almost identical for type II phase-matching. The cutting

angle of the crystal, relative to the cristallographic axis c, is denoted θc, while the ro-

tation angle of the nonlinear crystal is called ρc . Both the cutting angle and crystal

rotation are assumed to be in the xz-plane.

Fig. 3.3 Illustration of the coordinate system and parameters used in the phase-matchcalculations. ui is the internal angle between the ~ki-vector projected to the xz-plane,and the direction of the pump laser field. The corresponding angle in the yz-plane iscalled vi . The pump laser field inside the crystal is considered parallel to the z-axis.

The longitudinal and transverse phase-mismatch for type I phase-matching can then

be derived as shown in eq. 3.5 and 3.6. The material’s ordinary refractive index and

extraordinary angular dependent refractive index are denoted no and ne respectively.

cos(ρi) =1√

tan(ui)2 + tan(vi)2 + 1(3.4)

∆kz = 2π

(ne(λ3, ρ3)

λ3· cos(ρ3)−

no(λ2)

λ2− no(λ1, ρ1)

λ1· cos(ρ1)

)(3.5)

∆kT = 2π

(ne(λ3, ρ3)

λ3· sin(ρ3)−

no(λ1, ρ1)

λ1· sin(ρ1)

)(3.6)

The refractive indices are then derived using the Sellmeier equations combined with

eq. 3.7 which describes the index ellipsoid of the material [45].

1

n2e(θi, λi)

=sin2(θi)

n2e(λi)

+cos2(θi)

n2o(λi)

(3.7)

16

The final relation required for the full three-dimensional resolution of non-collinear

phase-matching equations is eq. 3.8 which is obtained from the conservation of trans-

verse momentum:

sin(v1)

sin(u1)=sin(v3)

sin(u3)(3.8)

Solving equations for energy conservation 3.1, longitudinal and transverse phase-

matching conditions 3.5, 3.6, and using 3.8 that the input signal and the upconverted

field is in the same plane, it is now possible to find corresponding phase-matched wave-

lengths and directions of propagation.

Figure 3.4 shows an example of the angular distribution of phase-matched wave-

lengths for type I upconversion in LiNbO3 at different crystal rotation angles. The

pump field wavelength is 1064 nm and the crystal is at room temperature. It is notewor-

thy that the phase-matched wavelength range can be tuned by rotating the crystal.

Fig. 3.4 Phase-matched wavelength as a function of incidence angle calculated for dif-ferent crystal orientations of LiNbO3 in type I configuration pumped with 1064 nm atroom temperature.

Figure 3.5 shows the angular distribution of wavelengths across the input and output

fields of view. The solutions to the phase-matching condition generally form concentric

rings, the diameter of which varies with wavelength. The width of these rings is set by

their acceptance angles.

17

(a) (b)

Fig. 3.5 Non-collinear phase-matched wavelengths as a function of angle at the input(a) and output (b) of a 48◦ cut bulk LiNbO3 crystal pumped at 1064 nm.

Non-collinear phase-matching is thus an interesting method for increasing the total

operational wavelength range of an upconversion device. In the next section, we will

show how this property can be used to increase the field of view in an upconversion

imaging setup [44; 46].

3.4 Architecture of upconversion imaging setup

Figure 3.6 illustrates the basic architecture of an upconversion imaging setup. It is based

on the design of a 4f setup. An MIR object field Eobject is first converted to frequency

space by the action of a lens L1 of focal length f1 used in a 2f configuration. In the

Fourier plane, it interacts with a Gaussian pump field inside a nonlinear crystal accord-

ing to the previously described phase-matching condition. This leads to the generation

of the upconverted field. This field is then collected by another Fourier transforming

lens L2 of focal length f2. At a given input wavelength λ1, at the image focal plane of

L2, we obtain the upconverted intensity field Iimage according to eq. 3.9.

Iimage(x, y) =8π2d2effL

2λ21n1n2n3ε0cλ43

· PGauss · sinc2(

∆(x, y)kL

2

)∣∣∣∣Eobject(λ1f1λ3f2

x,−λ1f1λ3f2

y

)~

(2πw2

0

(λ3f2)2e− 2(x2+y2)π2w2

0(λ3f2)

2

)∣∣∣∣2(3.9)

ε0 is the vacuum permeability, c is the speed of light in vacuum, w0 is the radius of

the Gaussian pump beam and PGauss its peak power.

18

Fig. 3.6 Schematic representation of an upconversion imaging setup. An object field isfocused to the Fourier plane inside a nonlinear crystal where it interacts with a Gaussianpump field.

Due to the phase-matching condition, for each input wavelength λ1 from the input

signal, a ring shaped portion of the object is upconverted. In the case of broadband illu-

mination, these adjacent rings increase the effective field of view of the system without

affecting the resolution of the system [47].

In the limit where the beam radius w0 of the Gaussian pump field becomes suffi-

ciently large, and assuming the phase-matching condition is met, a perfect upconverted

replica of the original image, in the new spectral region, can be obtained, scaled with a

factor M = −λ3f2λ1f1

as shown in eq. 3.10. This typically causes barrel distortion in the

upconverted images. A correction algorithm for this type of distortion is presented in

chapter 4.

Iimage ∝ Iobject

(λ1f1λ3f2

x,−λ1f1λ3f2

y

)(3.10)

In reality, as illustrated in Fig. 3.6, the Gaussian pump beam has a finite size. As

shown in eq. 3.9, the diameter of the pump beam defines the size of the point spread

function, and by extension the resolution of the imaging process. In practice, the size

of the Gaussian pump is limited by the clear aperture of the nonlinear crystal as well

as the available pump peak power needed to yield acceptable upconversion efficiency.

Therefore, as mentioned in chapter 1, the upconversion imaging greatly benefits from

being implemented in a pulsed manner in bulk crystals.

19

3.5 Using off-axis parabolic mirrors for upconversion

In order for the upconversion to take place, the pump beam and the broadband MIR

signal must be spatially superimposed inside the nonlinear crystal. However, combining

both beams can prove challenging due to broadband MIR optical components being

quite rare. Throughout this project, an off-axis parabolic mirror (OAPM) with hole

drilled through it was used. A 3 mm diameter hole was drilled in a standard OAPM

along the axis of the MIR focused beam. The pump beam is passed through it and

aligned to be coaxial to the MIR beam. A schematic representation of this setup is

given in figure 3.7.

Fig. 3.7 Schematic representation of the drilled OAPM solution.

This is a simple and inexpensive method of combining the broadband IR signal with

the pump beam. The presence of the hole causes a part of the image to be obscured.

These mirrors have a uniform high reflectance in the mid-IR, eliminating chromatic

aberration and offer excellent collimation, and point-focusing performances. However,

they introduce different types of aberration [48]. When used for upconversion imag-

ing, they cause a lateral keystone distortion due to the variation in effective reflected

focal length at different points of the mirror. This effect can be reduced, using OAPMs

20

designed for incidence angles as close to normal incidence as possible. A rescaling

algorithm for this type of distortion is presented in chapter 4.

Additionally, in the upconversion imaging systems described in this work, the OAPM

also converts the object into the Fourier plane in a 2f setup. It is important to note that

the focal length of OAPMs is not constant across its surface. Therefore, the sample or

target has to be tilted a couple of degrees with respect to the collimated SC beam. The

angular position of the target must be optimized to achieve the sharpest possible image

across the entire field of view. The angle being shallow, and the targets generally being

thin, this tilting of the object plane does not cause any measurable shadowing effects in

the object.

21

22

CHAPTER 4

Applications of pulsed upconversion of MIR SC

This chapter describes the experiments that were conducted throughout this project on

synchronous upconversion of MIR supercontinuum. The experiments are presented in

the order in which they were completed.

The first section of this chapter presents the first two iterations of single-pass pulsed

upconversion imaging systems. These systems have the special feature of relying on

the same pump laser to synchronously generate both the broadband MIR signal pulse

and the pump pulse for the nonlinear upconversion process. This both eliminates the

need for a second laser and ensures perfect temporal overlap of the SC and conversion

pump pulses.

The second half of this chapter introduces an electronically synchronized 1064 nm

MOPA used as the pump for the upconversion process and the advantages and perfor-

mance of such a device are presented in detail.

23

4.1 Pulsed upconversion imaging of SC light using a single 1550 nm

source both for generating the SC and pumping the upconversion pro-

cess

In this section, a 1550 nm erbium fiber-laser delivering 3.5 ns pulses at 40 kHz is used

as a pump source for both generating the SC and for pumping the upconversion process.

The pulse is separated into two arms with a 90/10 fiber coupler. The higher power pulse

is propagated through a combination of standard single-mode silica fibers and a section

of passive thulium-doped fiber, which are used to generate a SC. The lower power pulse

is propagated through a standard single-mode silica fiber of the same length as the high

power arm so that the two pulses will be synchronized at the outputs of the source.

4.1.1 Upconversion imaging of 2 µm to 2.6 µm in bulk lithium niobate

In this first experiment, a mid-IR imaging system was built by combining a MIR SC

source emitting between 1.8 µm and 2.6 µm with upconversion detection. The infrared

signal is used to probe a sample and is mixed with a synchronized 1550 nm laser pulse

inside a bulk lithium niobate (LiNbO3) crystal. The signal is thus upconverted to the

860 nm to 970 nm range and acquired on a standard silicon CCD array at a rate of 22

frames per second. In our implementation, spatial features in the sample plane as small

as 55 µm could be resolved.

4.1.1.1 Experimental setup

The synchronized SC and mixing pulses are delivered to the setup through two separate

fibers as indicated in Fig. 4.1. The SC was generated by propagating a high power

1550 nm 3.5 ns pulse through a combination of standard single-mode silica fibers and

a section of passive thulium-doped fiber used to generate a SC ranging from 1.8 µm to

2.6 µm with an average power of approximately 200 mW [49]. No ZBLAN fiber was

used for this first experiment. 10% of the 1550 nm pump power is extracted using a

90/10 fiber coupler. These pulses are propagated through a standard single-mode silica

fiber of the same length as the high SC arm so that the two pulses will be synchronized

at the outputs of the source.

The raw spectrum of the SC can be seen in Fig. 4.2. It is of interest to note the

24

Fig. 4.1 Top view schematic representation of the pulsed upconversion imaging setup.The use of OAPMs make for a mostly achromatic setup. The broadband signal pulse iscombined with the 1550 nm pump pulse through a hole drilled through an OAPM.

presence of a strong peak at 2.38 µm which in the following measurements will show

up as a bright ring in the images.

Both the SC and the mixing laser are randomly polarized. The SC light is first fil-

tered with a 2000 nm long-pass filter. Indeed, for the crystal cut angle and rotation

angle used in the experiment, SFG between wavelength components close to 1800 nm

of the SC could occur and cause parasitic light around 900 nm in the image. By using a

long pass filter, the range of SC wavelengths that could potentially cause this parasitic

light are removed. After this filter, the SC spans from 2000 nm to 2600 nm. The SC

is then collimated to a 16 mm diameter beam using a 50.8 mm reflected focal length

30◦ off axis parabolic mirror (OAPM) and is shone through the sample plane contain-

ing a US Air Force 1963 resolution test target. The sample plane is transformed by a

25.4 mm focal length 30◦ OAPM to create a two-dimensional Fourier transform of the

sample’s transmission in the nonlinear crystal as described in chapter 3.4. The pump

arm is collimated to a 2.2 mm diameter beam and is shone through a 3 mm diameter

hole drilled through the OAPM according to the description in chapter 3.5. The beams

are superimposed inside a 5 mm x 5 mm x 10 mm LiNbO3 crystal cut at 48◦. The

crystal is undoped, of congruent composition, and is kept at room temperature. The

crystal is placed on a rotation stage to allow for angle tuning of the phase-matching

condition. The sum-frequency generation process relies on type I phase-matching. The

phase-matching scheme is illustrated in Fig. 4.3. As a result of the nonlinear process,

the IR signal gets converted to the 860 nm to 970 nm range as dictated by conservation

of energy. As described in chapter 3, the IR signal, pump and upconverted signal wave-

25

Fig. 4.2 Spectrum of the SC. It can be noted that the strong peak at 2.38 µm provides arecognizable spectral signature that will be used to investigate the distribution of wave-lengths across the image plane.

lengths λ1, λ2 and λ3 respectively are linked by the relation of energy conservation eq

4.1:

1

λ3=

1

λ1+

1

λ2(4.1)

The near-IR upconverted light exiting the crystal is then propagated through a sim-

ple Keplerian beam expander. The first lens is a 30 mm focal length plano-convex

placed at 30 mm of the Fourier plane. The second lens is a 45 mm focal length plano-

convex lens. The near-IR signal is imaged on an affordable near-IR CCD camera

equipped with a 16 mm focal length objective lens. This last lens has its object fo-

cal plane coinciding with the image plane out of the beam expander, and thus performs

the inverse Fourier transform of the near-IR signal and restitutes the spatial information

of the sample object.

The quantum efficiency of the phase-matched light traveling along the center of the

nonlinear crystal is calculated to be 1.2 × 10−5 [9]. The large diameter of the pump

beam is here the main parameter reducing the efficiency. Additionally, the setup does

not implement any polarization control, which decreases the efficiency by a factor of

26

Fig. 4.3 Type I phase-matching upconversion scheme. The two input beams are po-larized along the crystals ordinary axis and the upconverted signal is polarized alongthe extraordinary direction. ~k1, ~k2 and ~k3 represent the wavevectors of the input signal,pump, and upconverted signal respectively.

four. However, the high number of available IR signal photons allows the easy detection

of an image on the CCD camera. It may be noted that other crystals might offer more

efficient upconversion. Silver gallium sulfide (AgGaS2) in particular has an effective

nonlinear susceptibility of 16 pm/V [50] for this process, which would translate to an

order of magnitude increase in efficiency compared to LiNbO3 the effective nonlinearity

of which is -4.1 pm/V [51].

4.1.1.2 Results

First, images were acquired without the resolution target. As can be seen in Fig. 4.4, a

consequence of the angular dependent phase-matching condition is that different spec-

tral components of the object light are upconverted at different angles. A bright spectral

component in the object light will result in a bright ring in the image. The phase-

matching condition is highly dependent on the cut angle, rotation and temperature of

the crystal. This is the property that can be exploited to reconstruct hyperspectral im-

ages by angle or temperature tuning of the crystal or by translating the sample [52].

We note that the spatial distribution of spectral components in the upconverted im-

age is in good agreement with the emission spectrum of the SC source. Figures 4.4(a)-

4.4(c) show the acquired upconverted images in the absence of a resolution target for

three different angular positions of the crystal. The bright peak at 2.38 µm in the spec-

trum of the SC directly translates to a bright ring in the upconverted image with a radial

position which depends on the crystal rotation. The lower spectral brightness region of

the SC spectrum is upconverted as a darker area mostly towards the center of the image.

Figure 4.4.(d) shows the theoretical distribution of wavelengths across a diameter of the

27

image. The horizontal dotted line represents the peak at 2.38 µm of the spectrum of the

SC.

(a) (b) (c)

(d)

Fig. 4.4 (a), (b) and (c) are acquired images of upconverted image of object light whenno sample is present for different angular positions of the crystal corresponding to in-ternal angles of θc = 45.4◦, θc = 46.1◦ and θc = 48.7◦ respectively. The radii of thebright rings are measured to be respectively 0.1 mm, 1.3 mm and 1.9 mm. Figure (d)displays the theoretical distribution of wavelengths in the image plane. A dotted linemarks the intensity peak of the SC. Additionally, the dark circle that can be observedon the right hand side in figures (b) and (c) is due to the hole in the OAPM.

To further investigate the correlation between the spatial distribution of wavelengths

in the upconverted image and the theoretical distribution anticipated by non-collinear

phase-matching, a fiber-coupled spectrometer was used to probe the spectral content

across the image plane along a diameter of the image. The experimental values are

confronted with the theoretical curve in Fig. 4.5. We can observe good correlation

between the experimental and theoretical data. The fact that we can easily link the

position on the detector to the corresponding infrared object wavelength makes this

source suitable for spectroscopy, and hyperspectral imaging applications.

28

Fig. 4.5 Spectral components across a diameter of the image for an angular position ofthe crystal corresponding to an internal angle of θc = 46.5◦. The dotted line representsthe experimental data and the full line represents the theoretical values.

A US Air Force 1963 resolution target is placed in the sample plane. Figure 4.6

shows upconverted images that were acquired in real time at a rate of 22 frames per

second.

We can observe some lateral field distortion which is due to the use of OAPMs as

discussed in chapter 3.5. Additionally, the brighter outer ring due to the peak at 2.38

µm in the spectrum of the SC is still present in these upconverted images. The image

formed is 4.2 mm in diameter and the imaging resolution was measured as the smallest

resolvable line of the resolution target. Resolutions down to 55 µm were measured in

the center of the image at a wavelength of approximately 2.2 µm.

We notice that the vertical lines of the resolution target appear blurred when com-

pared to the horizontal lines. The LiNbO3 crystal being placed at a slight angle with

respect to the pump beam, multiple reflections of the input beams inside the LiNbO3

crystal would result in multiple images being formed with a slight horizontal offset from

one another thus causing the image to be blurred along one direction. Alternatively, we

suspect that the aberrations introduced by the OAPMs could result in the vertical lines

appearing more blurry than the horizontal lines.

29

Fig. 4.6 Acquired upconverted image of a part of an US Air Force resolution test target.The angle between the crystal axis and the pump beam is 46.5◦ . Images were acquiredin real-time at a rate of 22 images per second.

4.1.1.3 Summary

In conclusion, we have demonstrated the first ever published pulsed upconversion imag-

ing system using a SC light source [53]. A complex infrared signal in the 2 µm to 2.6

µm range was thus successfully upconverted to the 860 nm to 970 nm range and ac-

quired on an affordable near-IR CCD camera. Additionally, the use of the high bright-

ness SC and the synchronous pulsed upconversion scheme greatly expands the choice

of usable crystals. In the present case, the use of a large pump beam diameter inside

a bulk LiNbO3 crystal has allowed us to obtain images with good spatial resolution.

As a consequence of the non-collinear phase-matching condition, spectral components

are upconverted in different parts of the image, which makes this system an interesting

choice for spectroscopy and hyperspectral imaging applications. Finally, this particular

system has the advantage of being mostly achromatic. By choosing the nonlinear crys-

tal appropriately, we can rely on a very similar system to perform pulsed upconversion

imaging using novel long wavelength SC sources even further in the infrared, as will be

demonstrated in the following section.

30

4.1.2 Upconversion around 4 µm using AGS crystal

In this experiment, an infrared signal of 100 mW average power with a spectrum ex-

tending up to 4.5 µm is passed through a sample and then focused into a bulk AgGaS2

crystal and subsequently mixed with a synchronous mixing signal at 1550 nm extracted

from the pump laser of the SC. Through SFG, an upconverted signal ranging from 1030

nm to 1155 nm is generated and acquired using an InGaAs camera.

4.1.2.1 Experimental setup

The experimental setup used is similar to the one presented in the previous section. A

schematic representation of the setup is provided in Fig. 4.7.

Fig. 4.7 Top view schematic representation of the pulsed upconversion imaging setup.The use of OAPMs makes for a mostly achromatic setup. The broadband signal pulseis combined with the 1550 nm pump pulse through a hole drilled through an OAPM.

The source used in section 4.1 was modified by connecting a segment of ZBLAN

fiber at the output of the supercontinuum arm. This causes the supercontinuum to extend

up to 4.5 µm and the spectrum of the new supercontinuum can be seen in Fig. 4.8. An

equal length of standard silica fiber was added to the 1550 nm tap arm to ensure that the

pulses are synchronized at the output of the source.

The synchronized SC and mixing pulses are delivered to the setup as indicated in

Fig. 4.7. Both the SC and the mixing laser are still randomly polarized. The SC light

is first filtered with a 2 µm long-pass filter. The filtering is necessary to prevent direct

31

Fig. 4.8 Spectrum of the SC source. The intensity is represented on a linear scale.

detection of the shorter wavelengths of the SC on the InGaAs camera. After this filter,

the SC spans from 2 µm to 4.5 µm. The SC is then collimated and focused down into

the nonlinear crystal using the same cofiguration as in section 4.1.1. The pump beam

configuration is also identical to section 4.1.1. The beams are superimposed inside a 5

mm x 5 mm x 10 mm AgGaS2 crystal cut at 54◦.

The sum-frequency generation process relies on type II phase-matching. The phase-

matching scheme is illustrated in Fig. 4.9. The MIR signal gets converted to the 1030

nm to 1155 nm range.

The upconverted plane waves exiting the crystal are then propagated through a sim-

ple 50 mm focal length plano-convex lens placed at 50 mm from the Fourier plane.

In the focal image plane of this lens, the upconverted image is acquired on an Peltier-

cooled InGaAs camera.

Fig. 4.9 Type II phase-matching upconversion scheme. ~k1, ~k2 and ~k3 represent thewavevectors of the input signal, pump, and upconverted signal respectively. The pumpis polarized along the ordinary direction while the input signal and upconverted signalare polarized along the extraordinary direction

32

The quantum efficiency of the phase-matched light traveling along the center of the

nonlinear crystal is calculated to be 1.0× 10−4 [9], which is a significant increase over

the efficiency calculated in section 4.1.1. This is due to the higher effective nonlinearity

of AgGaS2 over LiNbO3.

First, images were acquired without the resolution target. As detailed in chapter 3.3,

a wide range of wavelengths is upconverted through non-collinear upconversion which

results in an increased field of view. The raw upconverted image can be observed in

Fig. 4.10.

Fig. 4.10 Raw upconverted image of SC illumination with 75 ms exposure time. Due tothe phase-matched nature of the upconversion process, the upconverted wavelengths aredistributed radially across the image plane. The dark circle in the center of the image isdue to the hole in the mixing OAPM.

Next, a US Air Force 1951 resolution target is placed in the sample plane. Figure

4.11 show an upconverted image that was acquired with 75 ms exposure time. We can

observe some lateral field distortion which is due to the use of OAPMs as discussed in

chapter 3.5. The smallest resolvable feature is 12.7 lines/mm which gives us a resolution

of 79 µm . If we consider that resolution is mainly limited by the diameter of the pump

beam which acts as a soft aperture in the Fourier plane, based on the theory presented

in section 3.4, we calculate that the theoretical resolution is approximately 55 µm. The

experimental and theoretical resolution are of the same order of magnitude, and the

difference can be accounted for by the misalignment and aberrations of the optics in

the setup. While the resolution can be tailored by changing the magnification of the

optics, the number of resolvable elements is limited by the ratio of the area of the

33

point-spread function over the total field of view, and this is thus more relevant for the

characterization of upconversion imaging devices. The use of a pulsed source allows

the use of 2.2 mm 1/e2 diameter mixing pump diameter in the Fourier plane which

thus leads to enhanced spatial resolution. Additionally, the broad spectral range of the

source allows for a wide field of view of 14.9 mm in diameter. Taking into account

the 3 mm diameter hole in the center of the image, this gives us approximately 32000

resolvable elements. It should be noted that the number of resolvable elements is in an

inverse square relation to the (MIR) wavelength, thus impacting the resolution of any

MIR imaging device.

Fig. 4.11 Upconverted transmission image of a USAF 1951 resolution test target with75 ms exposure time. The smallest resolvable feature is 12.7 lines/mm. The imagecounts approximately 32000 resolvable elements. The deformation of the features nearthe center of the image is due to the hole in the mixing OAPM.

Finally, we place a sample cut out from a polypropylene drinking cup in the sam-

ple plane. The upconverted image gives us access to both the spatial information of

the sample as well as information about the absorption spectrum of the material. On

Fig. 4.12(a) we notice a dark ring that corresponds to an absorbing spectral feature of

polypropylene. This feature corresponds to the stretching vibration bands of the CH2

and CH3 groups of the polymer as is shown in Fig. 4.12(b).

4.1.2.2 Summary

In conclusion, we have demonstrated a pulsed upconversion imaging system using a SC

light source. A MIR signal in the 3 µm to 4.5 µm range was successfully upconverted

34

(a) (b)

Fig. 4.12 (a) Upconverted transmission image of a polypropylene drinking cup. Wecan observe both the spatial features of the characters printed on the cup as well assome spectral absorption features. The dark ring is due to strong absorption peaks ofCH2 and CH3 bonds. (b) Infrared absorption spectrum of polypropylene. The blue arearepresents the probed wavelength region

.

to the 1030 nm to 1155 nm range and acquired on a Peltier-cooled InGaAs camera with

an acquisition time of 75 ms. The non-collinear phase-matching of a broad wavelength

range was shown to increase the field of view of the imaging system. Additionally,

the use of a synchronous pulsed upconversion scheme in a bulk crystal enables the use

of a large pump beam diameter and has allowed us to obtain images with good spatial

resolution.

The two systems presented in this section demonstrate the potential of pulsed upcon-

version imaging using supercontinuum illumination for potential applications like, for

example, on-line industrial vision applications, fast gas sensing and bio-medical imag-

ing. Finally, the system being mostly achromatic, it can easily be modified to operate at

different wavelengths and with a variety of different crystals.

Both systems relied on the same pump laser to synchronously generate both the

broadband MIR signal pulse and the pump pulse for the nonlinear mixing process.

While this eliminates the need for a second laser and ensures perfect temporal over-

lap of the SC and conversion pump pulses, the fixed 1550 nm pump wavelength limits

the range of MIR wavelengths that can be detected with silicon based detectors and thus

required the use of an InGaAs camera in our second experiment.

The following sections introduce the use of an electronically synchronized 1064 nm

MOPA used as the pump for the upconversion process instead of the 1550 nm tap.

35

4.2 Pulsed upconversion of SC using an electronically synchronized MOPA

laser for pumping the upconversion process

In this section, the pump laser for the upconversion process is replaced by an external

laser source. This laser is based on a YDFA MOPA architecture seeded by a 1064 nm

gain-switched fiber-coupled DFB diode [54] that was built for this project and can be

seen in Fig. 4.13. The repetition rate, and pulse duration of this source can be adjusted

electronically. The output of this source is linearly polarized.

Fig. 4.13 Fully packaged YDFA MOPA source.

The pulses from this laser are electronically synchronized with the supercontinuum

pulses with the help of a T560 4-channel compact digital delay and pulse generator

from Highland Technology shown in Fig. 4.14.

Two experiments based on this method were carried out. Firstly, the temporal accu-

racy of the synchronization between pump and supercontinuum pulses was investigated.

An electronically delay-tuned cross-correlator was built to perform time-resolved spec-

tral characterization of the ns mid-infrared SC pulses.

Secondly, the synchronized 1064 nm MOPA pump was used in an upconversion

setup similar to those of section 4.1. The system’s noise is characterized and we present

a simple algorithm for correcting for the image distortion caused by the use of off axis

parabolic mirrors.

36

Fig. 4.14 T560 4-channel compact digital delay and pulse generator from HighlandTechnology

37

4.2.1 Electronically delay-tuned cross-correlator

In this section, a novel method for the characterization of mid-infrared pulses is pre-

sented. A cross-correlator system with no moving parts combining ultra-broadband

pulsed upconversion detection with fast active electronic delay tuning was built to per-

form time-resolved spectral characterization of 1.6 ns mid-infrared SC pulses. Full

wavelength/time spectrograms were acquired in steps of 20 ps over a range that can in

theory extend to microseconds, in a matter of seconds, with 48 ps temporal resolution

and 22 cm−1 spectral resolution in the 2700 nm to 4300 nm range.

This work proves the potential of the use of electronic delay tuning instead of me-

chanical delay for applications like cross-correlators and laser spectroscopy where their

fast precise tunability and long delay ranges are a strong asset.

As mentioned in chapter 1, the use of pulsed mid-infrared sources like broadband

SC sources and tunable quantum cascade lasers is becoming ever more common in a

variety of spectrometric optical systems for detection, identification, and quantification

of chemical species [7; 5; 6]. It is however critical to characterize their temporal profile

as well as the spectral variation within the duration of a pulse to ensure the best possible

accuracy, resolution and repeatability of such measurements.

Currently, there are no commercially available time-resolved spectrum analyzers in

the MIR. Indeed, traditional IR detection based on low bang-gap semiconductor ma-

terials, thermopiles or micro-bolometers are usually expensive and have slow response

times that make them unsuited for the temporal characterization of pulsed lasers. These

limitations can be avoided with upconversion.

Synchronous pulsed upconversion has been demonstrated and successfully used for

imaging and spectroscopy applications [55; 17]. The addition of a tunable delay line

in a pulsed upconversion setup enables the investigation of the cross-correlation of the

temporal profiles of the pulses. However, the complexity of the synchronization scheme

varies with the duration of the synchronized pulses and the required temporal precision.

Mechanical delay-tuning [56; 57; 58] can prove very impractical for measuring ns

pulses which require long delay arms. Comparatively, electronic delay tuning can prove

very advantageous. It was demonstrated in [59] to perform time-resolved spectral char-

acterization of QCL sources with a synchronization jitter of 2.5 ns and a temporal res-

olution of 25 ns limited by the pulse duration of the Q-switched Nd:YAG pump laser

38

that was used for the experiment. Additionally, gain-switched diode seeded master-

oscillator power amplifier (MOPA) systems have been successfully applied to electron-

ically synchronize nonlinear frequency conversion experiments in [56] and offer laser

sources with pulse durations and temporal jitter in the ps range.

In this section, we present a novel method of performing time-resolved characteriza-

tion of ns MIR supercontinuum pulses. In our setup, a digital delay and pulse generator

provides electronic triggering, synchronization and delay tuning between the supercon-

tinuum source and a gain-switched diode seeded MOPA. With this method, we were

able to measure full wavelength/time spectrograms of the SC in a matter of seconds

with a temporal resolution orders of magnitude lower than observed in previous similar

work [59].

4.2.1.1 Experimental setup

The experiment is based on the setup from [55] and is represented in Fig. 4.15. In this

experiment, light from a MIR SC source is upconverted in a synchronous fashion. The

externally triggered SC emits ns pulses at 40 kHz repetition rate and its spectrum ranges

from 2 µm to 4.5 µm [35]. The average power of the full SC is 80 mW. The supercon-

tinuum is colliamted and focused into the nonlinear crystal with the same optics as in

section 4.1 and the same LiNbO3 crystal was used.

The sum frequency generation is performed according to type I phase-matching. We

superimpose the focused SC in the nonlinear crystal with a linearly polarized 1064 nm

MOPA laser used as the pump for the upconversion process. The direction of the po-

larization is adjusted using a half-wave plate to match the ordinary axis of the LiNbO3

crystal. This 945 mW average power laser is based on a MOPA architecture seeded by

a gain-switched DFB diode [54]. The duration of the MOPA pulses was measured to be

40 ps FWHM with a standard autocorrelation method [60].

The short duration of the pump pulses with respect to the SC pulses makes them

adequate for probing the spectral content of the SC at many discrete times within the

pulse. The pump light is first collimated to a 1.4 mm diameter beam with a 7.5 mm

focal length aspheric lens (L1) and then focused with a 100 mm plano-convex lens (L2)

so as to spatially overlap the SC focus inside the nonlinear crystal.

39

Fig. 4.15 Schematic representation of the experimental setup. L1 : 7.5 aspheric lens,L2 : 100 mm plano-convex lens, L3 : 60 mm achromat, OAPM 1 : 2 inch focal length30◦ gold coated OAPM, OAPM 2 : 1 inch focal length 30◦ gold coated OAPM, LPF : 2µm long-pass filter, SPF : 1 µm short-pass filter

In addition to the beams having good spatial overlap inside the nonlinear crystal,

the experiment requires precise control over the temporal overlap between the pulses

of both light sources. A 160 kHz master clock triggers the 1064 nm pump diode and

the delay and pulse generator. This device will in turn generate one synchronized and

delay-tunable trigger pulse for every four trigger pulses it receives from the 160 kHz

clock signal. The newly generated 40 kHz trigger signal is used to trigger the SC source.

The temporal overlap between the SC pulses and pump pulses can thus be tuned in steps

of 20 ps over a range of multiple microseconds.

The pump light transmitted through the crystal is filtered out using a polarizer and a

1 µm short-pass filter and the upconverted light is collected with a 60 mm focal length

achromat (L3) which collimates it onto a reflective grating blazed for 750 nm with 1200

lines/mm at an incidence angle of approximately 50◦. The diffracted light is collected

on a regular silicon-based camera equipped with a 50 mm focal length camera objective.

40

4.2.1.2 Results

We first proceeded to characterize the timing jitter of the system. In order to do so,

we characterized the jitter of each source with respect to their trigger by using a fast

InGaAs photodiode placed at the position of the nonlinear crystal and a digital oscil-

loscope. Over the course of approximately one minute, we acquired the histogram of

the time interval errors between the 50% peak height points of the rising front of the

measured pulses relative to the trigger signal used to trigger both the source and the

digital oscilloscope measurement. The histograms are presented in Fig. 4.16.

Fig. 4.16 (a) jitter histogram of 1064 nm MOPA pump laser with respect to a triggerpulse, (b) jitter histogram of SC with respect to a trigger pulse

The standard deviation of the 1064 nm histogram and SC histogram are measured to

be 41 ps and 44 ps respectively. These measurements are dominated by the timing jitter

of the delay and pulse generator which is specified by the manufacturer to be typically

35 ps and also include the measurement jitter of the oscilloscope specified to be 4 ps.

By assuming that the jitter mechanisms are independent we obtain that the jitter of the

1064 nm pump and SC sources are 21 ps and 26 ps respectively. The total temporal jitter

in the upconversion process includes the jitter of both sources and the pulse generator

and is estimated to be approximately 48 ps RMS.

Next, we investigated the stability of the delay control between two pulses over a

period of 30 minutes. In order to do so, both light sources were set to illuminate a

single InGaAs photodiode connected to a digital oscilloscope in a way that both pulses

can be visualized on the same trace. The delay between the two pulses was set to 2

41

ns. The delay is measured as the time difference between the moments when the rising

fronts of a SC pulse and its corresponding pump pulse each reach 50% of their peak

value. Each point is averaged over 100 measurements to exclude the effect of the 48 ps

timing jitter. Figure 4.17 displays the variation of the delay between the SC and pump

pulses over a period of 30 minutes measured at 30 second intervals. The average delay

between the SC and pump pulses varied by approximately 45 ps. Experimentally, we

observed that this is mainly due to changes in the temporal shape of the SC pulse related

to temperature fluctuations of the system. This variation is negligible over the course of

the < 30 s measurement times presented subsequently in this section.

Fig. 4.17 Variation of the delay between the SC pulses and the 1064 nm laser pulsesover the course of 30 minutes

The upconverted light is diffracted by the grating and collected on a standard silicon-

based camera with an integration time of 0.4 ms. Figure 4.18(a,b) shows the raw camera

signal obtained without (Fig. 4.18(a)) and with (Fig. 4.18(b)) a 50 µm thick polystyrene

film in the beam path of the SC. The spectral features of the upconverted light appear as

bright and dark pixels on the camera in a fashion similar to a traditional grating spec-

trometer. The wavelength scale of the spectrometer setup is calibrated against an FTIR

measurement of the transmission of polystyrene by using the vibrational C-H absorp-

tion lines at 3304 nm and 3420 nm (see Fig.4.18(c)). We can, first of all, notice that the

upconverted wavenumber band is limited by the spectral bandwidth of the upconversion

process covering the 2700 nm to 4300 nm wavelength range [9].

42

Fig. 4.18 (a) Raw camera image of the upconverted full SC light, (b) raw camera imageof the upconverted SC light transmitted through a 50 µm polystyrene film, (c) Compar-ison of the transmission spectra of a 50 µm thick polystyrene film obtained with FTIR(red) and upconversion spectroscopy (blue)

Figure 4.18(c) compares the transmission spectrum of a polystyrene film measured

in our experiment with a spectrum obtained with FTIR. The plotted spectra are repre-

sented in terms of transmittance which is defined as the ratio of transmitted power over

incident power. This figure is thus unaffected by the wavelength dependence of the effi-

ciency of the upconversion process. While the FTIR spectrum was originally measured

with a spectral resolution of 4 cm−1 the data plotted in Fig. 4.18(c) was modified so as

to simulate a 22 cm−1 spectral resolution. For this spectral resolution, the upconverted

spectrum (blue) and the FTIR spectrum (red) are in good agreement. This indicates that

the spectral resolution of our system is approximately 22 cm−1. Additionally, oscilla-

tions of the absorption lines can be observed on the transmission spectrum obtained in

our experiment. These oscillations are due to interference effects arising from multiple

reflections of the SC light on the surfaces from the polystyrene film.

Figure 4.19 shows the relative variation of the efficiency of the upconversion process

with respect to the input wavelength. This plot was obtained by tracing the ratio between

43

the intensity of the light collected by the camera with a reference FTIR measurement of

the power spectral density of the SC source. We notice that the upconversion process

is more efficient at shorter wavelengths than longer wavelengths. This is due to the

fact that the efficiency of the conversion process is strongly dependent on the overlap

between the signal and the pump. In this experiment, both the signal and the pump are

spatially coherent sources. Therefore, the effective overlap of their beam-waists inside

the nonlinear crystal is significantly larger for the shorter collinearly upconverted wave-

lengths compared to the longer wavelengths which satisfy the phase-matching condition

for a larger incidence angle of the SC. The efficiency profile is consistent with [61] with

the exception of the small dip in efficiency at 2870 nm which is due to the presence of

the hole in OAPM 2.

Fig. 4.19 Variation of the efficiency of the upconversion process as a function of inputwavelength.

Lastly, we performed multiple spectra acquisitions of the upconverted SC light

while shifting the delay time between the trigger signals of the SC and the pump by 20

ps for each measurement in order to obtain a full wavelength/time spectrogram. Spectra

were thus acquired for 300 delay values spread over 6 ns. The camera integration time

for one full spectrum was 0.4 ms, and each spectrum was averaged over 50 consecutive

measurements. Therefore, each spectrum represents the average spectrum of 800 con-

secutive SC pulses, thus removing the pulse to pulse fluctuations of the ns SC source

[35]. Each spectrum was also corrected for the wavelength dependence of the efficiency

of the upconversion process (Fig. 4.19). The electronic delay line allowed the entire

44

measurement process to be automated and the acquisition of the full spectrogram took

less than 30 seconds. Figure 4.20 shows the variation of the upconverted wavelength

range of the SC pulses as a function of time.

Fig. 4.20 (a) Comparison of the spectra at 2.4 ns (black dotted), at 3.1 ns (red dot/dash)and at 3.86 ns (blue dashed), (b) spectrogram of the SC pulses, (c) Temporal profile ofthe SC pulse.

We first note that the spectrum is mostly uniform along the duration of a pulse and

that these pulses are 1.6 ns FWHM. At a given wavelength, the upconverted signal as

a function of delay is the temporal cross-correlation of the temporal profiles of the IR

signal and the pump. Since the pulse lengths (Fig. 4.20(c)) of the SC in this experiment

are significantly longer than the 40 ps FWHM pump pulses, the cross-correlation pre-

serves the temporal profile of the IR signal with good fidelity and thus does not require

any deconvolution processing.

45

4.2.1.3 Summary

In this section, we have demonstrated a system capable of time-resolved spectral char-

acterization of broadband MIR pulses based on electronically controlled delay tuning.

The system yielded full spectrograms of ns SC in 20 ps steps over a delay range of

6 ns in less than 30s. 48 ps temporal resolution and approximately 22 cm−1 spectral

resolution in the 2700 nm to 4300 nm wavelength were measured. The digital delay

and pulse generator allows for fully automated active delay tuning of the IR pulse with

respect to the pump pulse with a precision of tens of ps over delays that can be as long

as multiple µs without the need for mechanical delay lines. This system could easily

be adapted to perform time-resolved spectral characterization of QCL and Q-switched

lasers. It could also further be improved to perform single pulse measurements in order

to study pulse to pulse noise dynamics. The active electronic synchronization scheme is

extremely versatile as it can be used to simultaneously control and automate any number

of processes like the emission of optical pulses and time gated detection among others,

and we believe that it could be a powerful tool for frequency mixing applications, laser

spectroscopy, and range-resolved sensing.

46

4.2.2 Pulsed upconversion imaging of mid-infrared supercontinuum light using an

electronically synchronized pump laser

In this section, a novel method for synchronized imaging upconversion in the MIR

wavelength range is presented. The 1064 nm MOPA source pump laser is electronically

adjusted in pulse duration and repetition rate to match the output from a 40 kHz, 1.6

ns pulses MIR SC light source followed by upconversion imaging to the near-infrared

captured by a sensitive CCD camera. The system’s noise is characterized and we present

a simple algorithm for correcting the image distortion caused by the use of off axis

parabolic mirrors.

An imaging system based on this principle was previously demonstrated in section

4.1.2 and relied on a single 1550 nm laser source to pump both the SC extending up to

4.5 µm and the upconversion process. Unfortunately, with these wavelengths the sum-

frequency signal ranged from 1030 nm to 1155 nm and required the use of an InGaAs

based detector to acquire the upconverted signal. In section 4.2.1, active electronic

delay tuning has been shown to be a simple and effective method for synchronizing

optical pulses for upconversion. The temporal stability of the system was a mere 48

ps (jitter) and with low drift over time, thus no active locking mechanism was needed

when operated in the ns regime. Besides facilitating time resolved measurements, this

method gives the operator the freedom to independently choose the wavelengths of the

pump and infrared signal wavelengths to best suit the application.

In this section, we exploit this feature by synchronizing a 1064 nm MOPA with

electronically adjustable pulse duration and repetition rate to match a 40 Khz super-

continuum laser with 1.6 ns pulse duration. As a consequence, the upconverted wave-

lengths end up in the attractive 650 nm to 850 nm wavelength range where standard

silicon CCD cameras can be used. The proposed architecture is general in nature and

can be implemented with different illumination sources. We deploy in this section a

simple achromatic upconversion imaging setup using off-axis parabolic mirrors. The

performance of the system is characterized in terms of imaging quality and noise.

4.2.2.1 Experimental setup

The experiment is based on the setup from 4.2.1 and is represented in Fig. 4.21(a). The

externally triggered SC emits unpolarized 1.6 ns pulses at 40 kHz repetition rate and its

47

spectrum ranges from 1.8 µm to 4.2 µm [35]. The spectrum of the SC is shown in Fig.

4.22. Note that compared to the previous experiment, the spectrum and average power

of the SC have changed due to the source breaking down and having been reserviced.

The SC light is filtered through a 2 µm long pass filter (LPF1). The average power

of the full SC between 2 µm and 4.2 µm is 24 mW. The light is then collimated and

focused into the nonlinear crystal using the same optics as in previous secitons. The

same LiNbO3 crystal as in previous sections is used. It is placed at an incidence angle

of approximately 0.2◦.

(a)

(b)

Fig. 4.21 (a) Schematic representation of the experimental setup. L1 : 7.5 asphericlens, L2 : 50 mm plano-convex lens, OAPM 1 : 50.8 mm parent focal length 30◦ goldcoated OAPM, OAPM 2 : 25.4 mm parent focal length 30◦ gold coated OAPM, LPF1 :2 µm long-pass filter, LPF2 : 650 nm long-pass filter SPF : 850 nm short-pass filter.(b) Clear optical path USAF 1951 resolution test target.

48

The SFG is achieved in the same way as in section 4.2.1. The duration of the

MOPA pulses is set to 1.6 ns FWHM to match the duration of the supercontinuum

pulses. The pump light is first collimated to a 1.4 mm diameter beam using a 7.5 mm

focal length aspheric lens (L1) and then spatially overlapped with the SC focus inside

the nonlinear crystal by passing it through a hole drilled in OAPM2 according to section

3.5. The temporal overlap between the SC and pump pulses is ensured by the Highland

technologies T560 digital delay and pulse generator [62] in a similar fashion to section

4.2.1.

Fig. 4.22 Spectral radiant flux of the SC source

After the crystal, the light is filtered through a polarizer, transmitting the upcon-

verted signal while blocking the pump laser, a 850 nm short-pass filter (SPF) and a 650

nm long pass filter (LPF2) and the upconverted light is collected with a 50 mm focal

length plano convex lens (L2) and acquired on an Andor Technologies LucaEM S 658M

camera with a silicon based CCD sensor. Figure 4.21a(b) shows the clear optical path

USAF 1951 resolution test target used to assess the imaging quality.

49

4.2.2.2 Results

Figure 4.23(a) is a raw camera image obtained when no target is inserted. The dark

circle is due the the hole in OAPM2.

The distribution of wavelengths in the upconverted image is dictated by non-collinear

phasematching [44]. The theoretical distribution of wavelengths can be seen in Fig.

4.23(b), 4.23(c).

(a)

(b) (c)

Fig. 4.23 (a) Raw camera image when the resolution target is removed from the objectplane. The white dashed circle delimits the area obscured due to the hole in OAPM2. (b)Distribution of wavelengths below the phase-matching inflexion point. (c) Distributionof wavelengths above the phase-matching inflexion point.

The shortest wavelengths are in the center of the image while the longer wavelengths

appear as rings of varying radius. It is important to note that the wavelength range used

in this experiment spans beyond the inflexion point in the phase-matching condition,

50

which causes multiple wavelengths to be upconverted in the same position in the image

plane [63]. Additionally, the very bright ring on the outer edge of the image is due to the

peak around 3.6 µm in the spectrum of the SC that can be seen in Fig. 4.22 combined

with an increased spectral acceptance bandwidth close to the inflexion point.

Figure 4.24(a) shows the raw upconverted image that the system can generate. The

images hold barrel distortion due to the wavelength dependent angular magnification

in non-collinear phase-matching [12; 44]. Additionally, the figures also bear horizon-

tal keystone distortion (features on the right of the image appear more magnified than

features on the left) due to the use of the off-axis parabolic mirror. As demonstrated in

section 3.4, for large pump beam diameters, the upconverted image is a replica of the

original image scaled with a factor M as expressed in eq. (4.2).

M = −λ3f2λ1f1

, (4.2)

where λ3 and λ1 are the wavelengths of the upconverted and input beams respec-

tively, f1 is the local focal length of OAPM2 and f2 is the focal length of lens L2 [64].

The scaled image is further convolved with the point spread function set by the spot

size of the pump laser [12]. However, the focal length f1 of the OAPM varies across the

surface of the mirror according to the equation of a paraboloid. Thus, for every beam

section reflected by the mirror with an angle θ with respect to the axis of symmetry of

the paraboloid, the focal length f1 can be expressed simply as in eq. (4.3), where fparent

is the parent focal length of OAPM2 [48].

f1(θ) = − 2fparent1 + cos(θ)

(4.3)

The spatial distribution of wavelengths in both the input MIR signal and the up-

converted image is calculated according to non-collinear phase-matching and thus all

the parameters of the scaling factor M are known. Figure 4.24(b) shows the the image

after it has been rescaled by the scaling factor 1/M. The dashed line marks the limit

within which the image correction algorithm is applicable. Within this boundary the

resized image is in good overall agreement with the target and any discrepancies can be

attributed to misalignments. Outside of the dashed boundary we notice deformations

of the image due to multiple wavelengths above and below the inflexion point of the

51

crystal being upconverted at the same phasematch angles.

(a)

(b)

Fig. 4.24 (a) Raw camera image of the resolution target. (b) Distortion corrected andscaled image of the resolution target.

We also measured the resolution limit close to the center of the rescaled image.

Figure 4.25 shows a pixel-value cross-section of the image along the features of group

3 of the resolution target. Element 5 is resolved while element 6 is not according to

the Rayleigh criterion. Thus, the resolution limit lies between 78.8 µm and 70.2 µm.

The theoretical resolution limit was calculated to be 56 µm according to the size of the

pump beam. The difference between the theoretical and experimental value is within

expectations and can be largely attributed to aberration introduced by the use of OAPMs

and uncertainty of the pump beam size.

52

Fig. 4.25 Pixel-value cross-section along a a vertical line across group 3 in the cor-rected image. The numbers above the features denote the corresponding number of theelements within the group

It has been shown that upconversion imaging and spectroscopy can be prone to

thermal noise [18]. We investigated the variation of the intensity noise in each pixel

as a function of integration time. For a given integration time, 100 camera frames

are acquired first when both the SC and pump beams are blocked, then when only the

blocking the SC, and finally when both beams are active. Statistics are made on the

variation of readout values for each pixel. These measurements were performed for

integration times ranging from 0.5 ms to 150 ms.

The noise introduced by the camera is constant throughout the image and does not

vary with integration time as it is dominated by the sensor readout noise. When only the

pump laser is switched on, a small portion of the 1064 nm light leaks through the filters

and gets detected by the camera as a small bright spot in the center of the obscured

part of the image. No additional noise is introduced in the image and, in particular, no

thermal noise is observed, even at the longest integration times. This is due to the low

average power of the pump system compared to continuous-wave pumped intra-cavity

upconversion systems which typically operate at much higher average power [18]. Fi-

nally, when the sources are switched on, the noise contribution of the upconverted light

is simply added to the camera readout noise as both noise sources are independent.

53

Figure 4.26 shows the variance of the pixel counts as a function of integration time

for 5 different wavelengths in the upconverted image. For all wavelengths, the variance

increases linearly with time which indicates that the noise is limited by shot noise from

the signal. The slope of the plots is proportional to the brightness of the upconverted

signal at the corresponding wavelengths which is a product of of the spectral brightness

of the SC source combined with the angular dependent efficiency of the upconversion

process. Note that the variance evolves linearly even at the shortest integration times

and that the pulse to pulse noise of the SC source does not appear to have any influence

at these integration times. Indeed, with a repetition rate of 40 kHz, even at the shortest

integration time of 0.5 ms, the images are averaged over 20 SC pulses which is enough

to make the influence of pulse to pulse fluctuations negligible compared to shot noise

from the signal. Last but not least, for an integration time of 0.5 ms, the noise variance

of the upconverted signal is approximately equal to the camera readout noise variance

and thus at this integration time the signal to noise ratio is close to 1.

Fig. 4.26 Evolution of the total intensity variance at given wavelengths as a function ofcamera integration time.

54

4.2.2.3 Summary

In this section, a pulsed upconversion imaging system relying on active electronic syn-

chronization was presented. MIR pulses from a SC source were electronically synchro-

nized with pulses from 1064 nm MOPA pump source and upconverted in bulk lithium

niobate in a single-pass imaging setup. This detection scheme was shown to be shot

noise limited mainly due to the pulsed nature of the pump, minimizing the thermal

noise while conserving high quantum efficiency. A simple rescaling algorithm cor-

recting for distortion induced by the use of OAPMs was demonstrated and the spatial

resolution of the system was measured to be better than 78.8 µm at a wavelength of 2.5

µm. This method could be adapted for use with different MIR light sources like quan-

tum cascade lasers and the versatility of the electronic synchronization could prove to

be an invaluable tool for frequency conversion applications across the MIR.

55

56

CHAPTER 5

Outlook and conclusion

5.1 Outlook

Based on the results obtained throughout this project, this section presents suggestions

on how MIR SC illumination and synchronized pulsed upconversion might lead to fu-

ture research and applications.

Pulsed upconversion using an electronically synchronized MOPA pump laser holds

a lot of potential for many applications. Besides facilitating time-resolved measure-

ments, this method gives the operator the freedom to independently choose the wave-

length of the pump and those of the IR signal wavelengths to best suit the applications.

This principle may be applied to upconversion of QCL illumination which typically

occurs at longer wavelengths than SC to unlock novel spectroscopy and hyperspectral

imaging applications. The setup used in section 4.2.1 may be adapted to perform time-

resolved spectral characterization of QCL and Q-switched lasers. It could also further

be improved to perform single pulse measurements in order to study pulse to pulse

noise dynamics of SC sources. It could be of interest to extend the spectrum of the

supercontinuum beyond 4.5 µm by using chalcogenide fibers. Combined with pulsed

upconversion, this could lead to novel spectroscopy and hyperspectral imaging applica-

tions in the fingerprint region. The reliability and robustness of MIR SC sources could

even further be improved to facilitate their implementation.

5.2 Conclusion

The aim of this project was to investigate the advantages offered by the combination of

a MIR SC source from NKT Photonics with upconversion detection technology devel-

opped at DTU Fotonik. A MIR SC was built and successfully implemented in various

57

novel imaging and spectroscopy setups. Pulsed upconversion of imaging of SC was

demonstrated to yield large fields of view and good spatial resolutions. It was shown

that an electronically synchronizing a gain-switched diode seeded MOPA laser with the

MIR signal pulses was a simple and effective way to achieve pulsed upconversion.

In conclusion, this thesis demonstrates the strong potential of upconversion detec-

tion systems implementing MIR SC illumination and that a range of novel applications

of such systems is open for future investigations.

58

REFERENCES

[1] William Herschel. Experiments on the refrangibility of the invisible rays of the

sun. by william herschel, ll. d. f. r. s. Philosophical Transactions of the Royal

Society of London, 90:284–292, 1800.

[2] E. Scott Barr. Historical survey of the early development of the infrared spectral

region. American Journal of Physics, 28(1):42–54, 1960.

[3] E.Scott Barr. The infrared pioneers—ii. macedonio melloni. Infrared Physics,

2(2):67 – 74, 1962.

[4] Antoni Rogalski. Infrared Detectors. CRC Press, 2011.

[5] Robert F. Curl, Federico Capasso, Claire Gmachl, Anatoliy A. Kosterev, Barry

McManus, Rafał Lewicki, Michael Pusharsky, Gerard Wysocki, and Frank K.

Tittel. Quantum cascade lasers in chemical physics. Chemical Physics Letters,

487(1):1 – 18, 2010.

[6] Angela B. Seddon, Trevor M. Benson, Slawomir Sujecki, Nabil Abdel-Moneim,

Zhuoqi Tang, David Furniss, Lukasz Sojka, Nick Stone ad Nallala Jayakrupakar,

Gavin Rhys Lloyd, Ian Lindsay, Jon Ward, Mark Farries, Peter M. Moselund,

Bruce Napier, Samir Lamrini, Uffe Møller, Irnis Kubat, Christian R. Petersen, and

Ole Bang. Towards the mid-infrared optical biopsy. Proc.SPIE, 9703:970302,

2016.

[7] Jakob Kilgus, Kristina Duswald, Gregor Langer, and Markus Brandstetter.

Mid-infrared standoff spectroscopy using a supercontinuum laser with compact

fabry–perot filter spectrometers. Applied Spectroscopy, 72(4):634–642, 2018.

PMID: 29164925.

59

[8] Jeppe Seidelin Dam, Peter Tidemand-Lichtenberg, and Christian Pedersen. Room

temperature mid-ir single photon spectral imaging. Nature Photonics, 6:788–793,

2012.

[9] Robert W. Boyd. Nonlinear Optics third edition. Elsevier, 2008.

[10] Optical detection systems - technical note. 2015.

[11] J. E. Midwinter. Image conversion from 1.6 µ to the visible in lithium niobate.

Applied Physics Letters, 12(3):68–70, 1968.

[12] Christian Pedersen, Emir Karamehmedovic, Jeppe Seidelin Dam, and Peter

Tidemand-Lichtenberg. Enhanced 2d-image upconversion using solid-state lasers.

Opt. Express, 17(23):20885–20890, Nov 2009.

[13] Jeppe Seidelin Dam, Christian Pedersen, and Peter Tidemand-Lichtenberg. High-

resolution two-dimensional image upconversion of incoherent light. Opt. Lett.,

35(22):3796–3798, Nov 2010.

[14] Steven Baldelli. High-resolution two-dimensional image upconversion of inco-

herent light. Nature Photonics, 5:75, February 2011.

[15] Lichun Meng, Andreas Fix, Martin Wirth, Lasse Høgstedt, Peter Tidemand-

Lichtenberg, Christian Pedersen, and Peter John Rodrigo. Upconversion de-

tector for range-resolved dial measurement of atmospheric ch4. Opt. Express,

26(4):3850–3860, Feb 2018.

[16] Christian R. Petersen, Peter M. Moselund, Laurent Huot, Lucy Hooper, and Ole

Bang. Towards a table-top synchrotron based on supercontinuum generation. In-

frared Physics and Technology, 91:182 – 186, 2018.

[17] Sune Dupont, Peter M. Moselund, Lasse Leick, Jacob Ramsay, and Søren R. Kei-

ding. Up-conversion of a megahertz mid-ir supercontinuum. J. Opt. Soc. Am. B,

30(10):2570–2575, Oct 2013.

[18] Ajanta Barh, Peter Tidemand-Lichtenberg, and Christian Pedersen. Thermal noise

in mid-infrared broadband upconversion detectors. Opt. Express, 26(3):3249–

3259, Feb 2018.

60

[19] Malcolm H. Dunn and Majid Ebrahimzadeh. Parametric generation of tunable

light from continuous-wave to femtosecond pulses. Science, 286(5444):1513–

1517, 1999.

[20] Valentin Petrov. Frequency down-conversion of solid-state laser sources to the

mid-infrared spectral range using non-oxide nonlinear crystals. Progress in Quan-

tum Electronics, 42:1 – 106, 2015.

[21] R. R. Alfano and S. L. Shapiro. Emission in the region 4000 to 7000 a via four-

photon coupling in glass. Phys. Rev. Lett., 24:584–587, Mar 1970.

[22] Chinlon Lin and R. H. Stolen. New nanosecond continuum for excited-state spec-

troscopy. Applied Physics Letters, 28(4):216–218, 1976.

[23] C. Lin, V. T. Nguyen, and W. G. French. Wideband near-i.r. continuum (0.7-2.1

µm) generated in low-loss optical fibres. Electronics Letters, 14(25):822–823,

December 1978.

[24] Jinendra K. Ranka, Robert S. Windeler, and Andrew J. Stentz. Visible continuum

generation in air–silica microstructure optical fibers with anomalous dispersion at

800 nm. Opt. Lett., 25(1):25–27, Jan 2000.

[25] T. Schreiber, J. Limpert, H. Zellmer, A. Tunnermann, and K.P. Hansen. High

average power supercontinuum generation in photonic crystal fibers. Optics Com-

munications, 228(1):71 – 78, 2003.

[26] A. B. Rulkov, M. Y. Vyatkin, S. V. Popov, J. R. Taylor, and V. P. Gapontsev. High

brightness picosecond all-fiber generation in 525–1800nm range with picosecond

yb pumping. Opt. Express, 13(2):377–381, Jan 2005.

[27] C. Xia, M. Kumar, M. Cheng, O. P. Kulkarni, M. N. Islam, A. Galvanauskas, F. L.

Terry, M. J. Freeman, D. A. Nolan, and W. A. Wood. Supercontinuum genera-

tion in silica fibers by amplified nanosecond laser diode pulses. IEEE Journal of

Selected Topics in Quantum Electronics, 13(3):789–797, May 2007.

[28] T. Izawa, N. Shibata, and A. Takeda. Optical attenuation in pure and doped fused

silica in the ir wavelength region. Applied Physics Letters, 31(1):33–35, 1977.

61

[29] Jacques Lucas. Fluoride glasses for modern optics. Journal of Fluorine Chemistry,

72(2):177 – 181, 1995. Proceedings of the International Conference on Fluorine

Chemistry ’94 Kyoto.

[30] C. L. Hagen, J. W. Walewski, and S. T. Sanders. Generation of a continuum

extending to the midinfrared by pumping zblan fiber with an ultrafast 1550-nm

source. IEEE Photonics Technology Letters, 18(1):91–93, Jan 2006.

[31] Kun Liu, Jiang Liu, Hongxing Shi, Fangzhou Tan, and Pu Wang. High power

mid-infrared supercontinuum generation in a single-mode zblan fiber with up to

21.8 w average output power. Opt. Express, 22(20):24384–24391, Oct 2014.

[32] Chris D. Brooks Peter M. Moselund, Laurent Huot. All-fiber mid-ir supercontin-

uum: a powerful new tool for ir-spectroscopy. Proc.SPIE, 9703:9703 – 9703 – 6,

2016.

[33] John M. Dudley, Goery Genty, and Stephane Coen. Supercontinuum generation

in photonic crystal fiber. Rev. Mod. Phys., 78:1135–1184, Oct 2006.

[34] Goery Genty, Stephane Coen, and John M. Dudley. Fiber supercontinuum sources

(invited). J. Opt. Soc. Am. B, 24(8):1771–1785, Aug 2007.

[35] Vinay V. Alexander, Ojas P. Kulkarni, Malay Kumar, Chenan Xia, Mohammed N.

Islam, Fred L. Terry, Michael J. Welsh, Kevin Ke, Michael J. Freeman, Man-

ickam Neelakandan, and Allan Chan. Modulation instability initiated high power

all-fiber supercontinuum lasers and their applications. Optical Fiber Technology,

18(5):349 – 374, 2012. Fiber Supercontinuum sources and their applications.

[36] M. M. Kozak, W. Kowalsky, and R. Caspary. Low-loss glue splicing method to

join silica and fluoride fibres. Electronics Letters, 41(16):897–899, Aug 2005.

[37] Ruifeng Zhao Chunhui Qi Reinhard Caspary Mah Siew Kien Shuisheng Jian

Li Pei, Xiaowei Dong. Low loss splicing method to join silica and fluoride fibers.

Proc.SPIE, 6781:6781 – 6781 – 6, 2007.

[38] R. Al-Mahrous, R. Caspary, and W. Kowalsky. A glue splicing method to join

silica and fluoride fibers with low attenuation. Journal of Lightwave Technology,

32(9):1669–1673, May 2014.

62

[39] Gilbert N. Lewis, David Lipkin, and Theodore T. Magel. Reversible photochem-

ical processes in rigid media. a study of the phosphorescent state. Journal of the

American Chemical Society, 63(11):3005–3018, 1941.

[40] T. H. Maiman. Stimulated optical radiation in ruby. Nature, 187:493, August

1960.

[41] P. A. Franken, A. E. Hill, C. W. Peters, and G. Weinreich. Generation of optical

harmonics. Phys. Rev. Lett., 7:118–119, Aug 1961.

[42] Nicolaas Bloembergen. Nonlinear Optics. WORLD SCIENTIFIC, 4th edition,

1996.

[43] J. E. Midwinter and J. Warner. Up-conversion of near infrared to visible radiation

in lithium-meta-niobate. Journal of Applied Physics, 38(2):519–523, 1967.

[44] P. Tidemand-Lichtenberg, J. S. Dam, H. V. Andersen, L. Høgstedt, and C. Peder-

sen. Mid-infrared upconversion spectroscopy. J. Opt. Soc. Am. B, 33(11):D28–

D35, Nov 2016.

[45] David E. Zelmon, David L. Small, and Dieter Jundt. Infrared corrected sell-

meier coefficients for congruently grown lithium niobate and 5 mol.% magnesium

oxide–doped lithium niobate. J. Opt. Soc. Am. B, 14(12):3319–3322, Dec 1997.

[46] A. J. Torregrosa, H. Maestre, and J. Capmany. Intra-cavity upconversion to 631

nm of images illuminated by an eye-safe ase source at 1550 nm. Opt. Lett.,

40(22):5315–5318, Nov 2015.

[47] Peter John Rodrigo Peter Tidemand-Lichtenberg Morgan Mathez, Christian Ped-

ersen. Investigation of mid-ir picosecond image upconversion. Proc.SPIE,

10088:10088 – 10088 – 7, 2017.

[48] James E. Howard. Imaging properties of off-axis parabolic mirrors. Appl. Opt.,

18(15):2714–2722, Aug 1979.

[49] Vinay V. Alexander, Ojas P. Kulkarni, Malay Kumar, Chenan Xia, Mohammed N.

Islam, Fred L. Terry Jr., Michael J. Welsh, Kevin Ke, Michael J. Freeman, Man-

ickam Neelakandan, and Allan Chan. Modulation instability initiated high power

63

all-fiber supercontinuum lasers and their applications. Optical Fiber Technology,

18(5):349–374, 2012.

[50] P. Kupecek, C. Schwartz, and D. Chemla. Silver thiogallate (aggas2) - part 1:

Nonlinear optical properties. IEEE Journal of Quantum Electronics, 10(7):540–

545, July 1974.

[51] Ichiro Shoji, Takashi Kondo, Ayako Kitamoto, Masayuki Shirane, and Ryoichi

Ito. Absolute scale of second-order nonlinear-optical coefficients. J. Opt. Soc.

Am. B, 14(9):2268–2294, Sep 1997.

[52] Louis Martinus Kehlet, Nicolai Sanders, Peter Tidemand-Lichtenberg, Jeppe Sei-

delin Dam, and Christian Pedersen. Infrared hyperspectral upconversion imaging

using spatial object translation. Opt. Express, 23(26):34023–34028, Dec 2015.

[53] Laurent Huot, Peter Morten Moselund, Peter Tidemand-Lichtenberg, Lasse Leick,

and Christian Pedersen. Upconversion imaging using an all-fiber supercontinuum

source. Opt. Lett., 41(11):2466–2469, Jun 2016.

[54] Sebastian Kanzelmeyer, Hakan Sayinc, Thomas Theeg, Maik Frede, Joerg Neu-

mann, and Dietmar Kracht. All-fiber based amplification of 40 ps pulses from a

gain-switched laser diode. Opt. Express, 19(3):1854–1859, Jan 2011.

[55] Laurent Huot, Peter M. Moselund, Lasse Leick, Peter Tidemand-Lichtenberg, and

Christian Pedersen. Broadband upconversion imaging around 4 mum using an

all-fiber supercontinuum source. Proc.SPIE, 10088:10088 – 10088 – 7, 2017.

[56] Laura Abrardi and Thomas Feurer. Electronic synchronization of gain-switched

laser diode seeded fiber amplifiers. Proc.SPIE, 8237:8237 – 8237 – 12, 2012.

[57] Jacob Ramsay, Sune Dupont, Mikkel Johansen, Lars Rishøj, Karsten Rottwitt,

Peter Morten Moselund, and Søren Rud Keiding. Generation of infrared super-

continuum radiation: spatial mode dispersion and higher-order mode propagation

in zblan step-index fibers. Opt. Express, 21(9):10764–10771, May 2013.

[58] Sebastian Wolf, Tobias Trendle, Jens Kiessling, Johannes Herbst, Karsten Buse,

and Frank Kuhnemann. Self-gated mid-infrared short pulse upconversion detec-

tion for gas sensing. Opt. Express, 25(20):24459–24468, Oct 2017.

64

[59] Jean-Michel Melkonian, Myriam Raybaut, Antoine Godard, Johan Petit, and

Michel Lefebvre. Time-resolved spectral characterization of a pulsed external-

cavity quantum cascade laser. Proc.SPIE, 8546:854607, 2012.

[60] J. A. Armstrong. Measurement of picosecond laser pulse widths. Applied Physics

Letters, 10(1):16–18, 1967.

[61] Sebastian Wolf, Jens Kiessling, Michael Kunz, Gregor Popko, Karsten Buse,

and Frank Kuhnemann. Upconversion-enabled array spectrometer for the

mid-infrared, featuring kilohertz spectra acquisition rates. Opt. Express,

25(13):14504–14515, Jun 2017.

[62] Laurent Huot, Peter Morten Moselund, Peter Tidemand-Lichtenberg, and Chris-

tian Pedersen. Electronically delay-tuned upconversion cross-correlator for char-

acterization of mid-infrared pulses. Opt. Lett., 43(12):2881–2884, Jun 2018.

[63] Ajanta Barh, Christian Pedersen, and Peter Tidemand-Lichtenberg. Ultra-

broadband mid-wave-ir upconversion detection. Opt. Lett., 42(8):1504–1507, Apr

2017.

[64] Jeppe Seidelin Dam, Christian Pedersen, and Peter Tidemand-Lichtenberg. The-

ory for upconversion of incoherent images. Opt. Express, 20(2):1475–1482, Jan

2012.

65

66

Appendices

67

Appendix A

List of Terms and Abbreviations

IR Infrared

MIR Mid Infrared

OAPM Off-Axis Parabolic Mirror

SC Supercontinuum

CW Continuous Wave

NIR Near-Infrared

OPO Optical Parametric Oscillator

QCL Quantum Cascade Laser

YDFA Ytterbium-Doped Fiber Amplifier

EDFA Erbium-Doped Fiber Amplifer

RMS Root Mean Square

FWHM Full Width at Half Maximum

DFB Distributed Feedback

PCF Photonic Crystal Fiber

MI Modulation Instability

SSFS Raman-induced Soliton Self Frequency Shift

CCD Charged-Coupled Device

QE Quantum Efficiency

GVD Group Velocity Dispersion

MOPA Master Oscillator Power Amplifier

69

ZBLAN ZrF4 −BaF2 − LaF3 − AlF3 −NaF

SFG Sum Frequency Generation

FTIR Fourier-Transform Infrared spectroscopy

70

Appendix B

List of figures

1.1 Schematic representation of the upconversion detection process. Repro-duced from [8] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

1.2 Overview of detector technology available for different wavelengthsand their specific detectivities. Reproduced from [10] . . . . . . . . . . 3

1.3 Brightness of silica, fluoride, and chalcogenide fiber-based SC sourcescompared to a synchrotron, the sun (5778 K black-body), and a Globar(1500 K black-body). This figure was reproduced from [16] . . . . . . . 5

1.4 Overview of the different nonlinear crystals used in parametric lightsources reported in literature. Note the difference in y-axis on the twoplots. Reproduced from [19] and [20], left and right, respectively. . . . . 6

2.1 (a)NKT’s flagship SC source, the SuperK Extreme. (b) Spectrum of aSuperK Extreme EXW 12 source. . . . . . . . . . . . . . . . . . . . . 8

2.2 Optical losses from three kinds of fluoride fibres compared to Si02. Re-produced from [29] . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

2.3 (a)NKT’s new MIR SuperK source (b) Typical spectrum of the NKTMIR SuperK Extreme source. . . . . . . . . . . . . . . . . . . . . . . . 10

2.4 Summary of dominant mechanisms for SC generation in different pulseand dispersion regimes. Reproduced from [35] . . . . . . . . . . . . . . 10

2.5 Architecture of the MIR SC used throughout this project . . . . . . . . 11

2.6 MIR SC source built for this project . . . . . . . . . . . . . . . . . . . 12

3.1 Sum-frequency generation. (a) Geometry of the interaction. (b) Energylevel description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

71

3.2 Geometric representation of the phase matching condition . . . . . . . . 15

3.3 Illustration of the coordinate system and parameters used in the phase-match calculations. ui is the internal angle between the ~ki-vector pro-jected to the xz-plane, and the direction of the pump laser field. Thecorresponding angle in the yz-plane is called vi . The pump laser fieldinside the crystal is considered parallel to the z-axis. . . . . . . . . . . . 16

3.4 Phase-matched wavelength as a function of incidence angle calculatedfor different crystal orientations ofLiNbO3 in type I configuration pumpedwith 1064 nm at room temperature. . . . . . . . . . . . . . . . . . . . . 17

3.5 Non-collinear phase-matched wavelengths as a function of angle at theinput (a) and output (b) of a 48◦ cut bulk LiNbO3 crystal pumped at1064 nm. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

3.6 Schematic representation of an upconversion imaging setup. An objectfield is focused to the Fourier plane inside a nonlinear crystal where itinteracts with a Gaussian pump field. . . . . . . . . . . . . . . . . . . . 19

3.7 Schematic representation of the drilled OAPM solution. . . . . . . . . . 20

4.1 Top view schematic representation of the pulsed upconversion imag-ing setup. The use of OAPMs make for a mostly achromatic setup.The broadband signal pulse is combined with the 1550 nm pump pulsethrough a hole drilled through an OAPM. . . . . . . . . . . . . . . . . 25

4.2 Spectrum of the SC. It can be noted that the strong peak at 2.38 µm pro-vides a recognizable spectral signature that will be used to investigatethe distribution of wavelengths across the image plane. . . . . . . . . . 26

4.3 Type I phase-matching upconversion scheme. The two input beams arepolarized along the crystals ordinary axis and the upconverted signal ispolarized along the extraordinary direction. ~k1, ~k2 and ~k3 represent thewavevectors of the input signal, pump, and upconverted signal respec-tively. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

4.4 (a), (b) and (c) are acquired images of upconverted image of objectlight when no sample is present for different angular positions of thecrystal corresponding to internal angles of θc = 45.4◦, θc = 46.1◦ andθc = 48.7◦ respectively. The radii of the bright rings are measured tobe respectively 0.1 mm, 1.3 mm and 1.9 mm. Figure (d) displays thetheoretical distribution of wavelengths in the image plane. A dotted linemarks the intensity peak of the SC. Additionally, the dark circle thatcan be observed on the right hand side in figures (b) and (c) is due tothe hole in the OAPM. . . . . . . . . . . . . . . . . . . . . . . . . . . 28

72

4.5 Spectral components across a diameter of the image for an angular posi-tion of the crystal corresponding to an internal angle of θc = 46.5◦. Thedotted line represents the experimental data and the full line representsthe theoretical values. . . . . . . . . . . . . . . . . . . . . . . . . . . . 29

4.6 Acquired upconverted image of a part of an US Air Force resolution testtarget. The angle between the crystal axis and the pump beam is 46.5◦ .Images were acquired in real-time at a rate of 22 images per second. . . 30

4.7 Top view schematic representation of the pulsed upconversion imag-ing setup. The use of OAPMs makes for a mostly achromatic setup.The broadband signal pulse is combined with the 1550 nm pump pulsethrough a hole drilled through an OAPM. . . . . . . . . . . . . . . . . 31

4.8 Spectrum of the SC source. The intensity is represented on a linear scale. 32

4.9 Type II phase-matching upconversion scheme. ~k1, ~k2 and ~k3 representthe wavevectors of the input signal, pump, and upconverted signal re-spectively. The pump is polarized along the ordinary direction whilethe input signal and upconverted signal are polarized along the extraor-dinary direction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32

4.10 Raw upconverted image of SC illumination with 75 ms exposure time.Due to the phase-matched nature of the upconversion process, the up-converted wavelengths are distributed radially across the image plane.The dark circle in the center of the image is due to the hole in the mixingOAPM. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33

4.11 Upconverted transmission image of a USAF 1951 resolution test tar-get with 75 ms exposure time. The smallest resolvable feature is 12.7lines/mm. The image counts approximately 32000 resolvable elements.The deformation of the features near the center of the image is due tothe hole in the mixing OAPM. . . . . . . . . . . . . . . . . . . . . . . 34

4.12 (a) Upconverted transmission image of a polypropylene drinking cup.We can observe both the spatial features of the characters printed onthe cup as well as some spectral absorption features. The dark ring isdue to strong absorption peaks of CH2 and CH3 bonds. (b) Infraredabsorption spectrum of polypropylene. The blue area represents theprobed wavelength region . . . . . . . . . . . . . . . . . . . . . . . . . 35

4.13 Fully packaged YDFA MOPA source. . . . . . . . . . . . . . . . . . . 36

4.14 T560 4-channel compact digital delay and pulse generator from High-land Technology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37

73

4.15 Schematic representation of the experimental setup. L1 : 7.5 asphericlens, L2 : 100 mm plano-convex lens, L3 : 60 mm achromat, OAPM 1: 2 inch focal length 30◦ gold coated OAPM, OAPM 2 : 1 inch focallength 30◦ gold coated OAPM, LPF : 2 µm long-pass filter, SPF : 1 µmshort-pass filter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40

4.16 (a) jitter histogram of 1064 nm MOPA pump laser with respect to atrigger pulse, (b) jitter histogram of SC with respect to a trigger pulse . . 41

4.17 Variation of the delay between the SC pulses and the 1064 nm laserpulses over the course of 30 minutes . . . . . . . . . . . . . . . . . . . 42

4.18 (a) Raw camera image of the upconverted full SC light, (b) raw cam-era image of the upconverted SC light transmitted through a 50 µmpolystyrene film, (c) Comparison of the transmission spectra of a 50µm thick polystyrene film obtained with FTIR (red) and upconversionspectroscopy (blue) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43

4.19 Variation of the efficiency of the upconversion process as a function ofinput wavelength. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44

4.20 (a) Comparison of the spectra at 2.4 ns (black dotted), at 3.1 ns (reddot/dash) and at 3.86 ns (blue dashed), (b) spectrogram of the SC pulses,(c) Temporal profile of the SC pulse. . . . . . . . . . . . . . . . . . . . 45

4.21 (a) Schematic representation of the experimental setup. L1 : 7.5 as-pheric lens, L2 : 50 mm plano-convex lens, OAPM 1 : 50.8 mm parentfocal length 30◦ gold coated OAPM, OAPM 2 : 25.4 mm parent focallength 30◦ gold coated OAPM, LPF1 : 2 µm long-pass filter, LPF2 :650 nm long-pass filter SPF : 850 nm short-pass filter. (b) Clear opticalpath USAF 1951 resolution test target. . . . . . . . . . . . . . . . . . . 48

4.22 Spectral radiant flux of the SC source . . . . . . . . . . . . . . . . . . . 494.23 (a) Raw camera image when the resolution target is removed from the

object plane. The white dashed circle delimits the area obscured due tothe hole in OAPM2. (b) Distribution of wavelengths below the phase-matching inflexion point. (c) Distribution of wavelengths above thephase-matching inflexion point. . . . . . . . . . . . . . . . . . . . . . . 50

4.24 (a) Raw camera image of the resolution target. (b) Distortion correctedand scaled image of the resolution target. . . . . . . . . . . . . . . . . . 52

4.25 Pixel-value cross-section along a a vertical line across group 3 in thecorrected image. The numbers above the features denote the corre-sponding number of the elements within the group . . . . . . . . . . . . 53

4.26 Evolution of the total intensity variance at given wavelengths as a func-tion of camera integration time. . . . . . . . . . . . . . . . . . . . . . . 54

74