Development of a new mid-infrared source pumped by an optical parametric chirped-pulse amplifier.
by
Etienne Pelletier
A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy Graduate Department of Physics University of Toronto
Copyright © 2013 by Etienne Pelletier Abstract
Development of a new mid-infrared source pumped by an optical parametric
chirped-pulse amplifier.
Etienne Pelletier
Doctor of Philosophy
Graduate Department of Physics
University of Toronto
2013
The mid-infrared (MIR) system presented in the thesis is based on a sub-100-fs erbium-
doped fiber laser operating at 1.55 m. The output of the laser is split in two, each arm seeding an erbium-doped fiber amplifier. The output of the first amplifier is sent to a grating-based stretcher to be stretched to 50 ps before seeding the optical parametric chirped-pulse amplifier (OPCPA). The output of the second amplifier is coupled to a highly nonlinear fiber to generate the 1 m needed to seed the a neodymium-doped yttrium lithium fluoride (Nd:YLF) system. This work represents the first time this synchronization scheme is used, and the timing jitter between the two arms at the OPCPA is reduced to 333 fs.
The pump laser for the OPCPA is a regenerative amplifier producing 1.6 W followed by a double-pass amplifier, for a final output power of 2.5 W at 1 kHz. Etalons were inserted into the cavity of the regenerative amplifier to stretch the pulses to 50 ps
The OPCPA consists of two potassium titanyl arsenate crystals in a noncollinear configuration. With three passes, the gain is 3.8 106. Using a grating compressor, the pulse duration is reduced to 140 fs, with a power of 300 mW. Because of the reduction of
the timing jitter, the amplitude stability is 1 %, which is a great improvement compare
to existing systems.
To generate ultrafast light in the MIR, an optical parametric amplifier is used, pumped
ii by the output of the OPCPA and seeded with its 3- m idler. Two crystals were tested, both in a single-pass configuration. For the first crystal, a 4-mm thick silver thiogallate, an efficiency of 7.4 % was reached, with 8.76 mW in the signal and 7.2 mW in the idler.
For the second crystal, a 2-mm thick lithium gallium selenide, the efficiency was higher, reaching 10.8 %. The power for the signal was 11.5 mW, and for the idler, 11.11 mW.
Using this new scheme, energies on par with current systems are achieved with much higher efficiencies.
iii Acknowledgements
At first, I thought that the acknowledgement would be the easy part; how hard could it
be to express my gratitude to a few people? Well, harder then I was expecting; so many
people helped me in one way or another and I hope I will do a good job at giving them
the recognition they deserve.
First and foremost, I would like to thank my supervisor, Prof. R.J. Dwayne Miller, for making this opportunity happen. I am not only talking about accepting me as a student in his group and giving me the guidance to achieve my goal, but also for having an extremely contagious enthusiasm, without which none of this would have been possible.
When I first met him, before a talk he was giving at Laval University, I was a disenchanted young man; years before the telecom bubble burst and with the job prospective in Optics still gloomy, grad school seemed like the only option. However, this first meeting changed my mind set and I began to see grad school for what is really was; a chance to work on new and exciting science. The following year, three months into what was supposed to be a two-year master, my transformation was completed and I decided to do a Ph.D., which, even during tough times, I have never regretted.
The meeting with Prof. Miller would have never happen without the generous help of Prof. Nathalie McCarthy, a faculty member at Laval University. At a moment when
I was at a cross-road in my life, she gave appreciated advice and suggestions, without which this thesis might have never happen. And for this, I am extremely grateful to Prof.
McCarthy.
A major breakthrough in my project occurred because of Prof. Miller’s sabbatical in
Italy, in 2008. While in Europe, he found himself visiting the University of Konstanz and met up with a faculty who told him how they could generate almost any color using an erbium-doped fiber laser; this meeting was a turning point in my project. Already by then, we were looking into using nonlinearity in fibers, but it was seen as a bit of a gamble.
However, by having a collaborator, this solution became a no-brainer and we went ahead;
iv the rest of the story can be found in this thesis. For this, I would like to express my most sincere gratitude to Prof. Alfred Leitenstorfer and his group at the University of
Konstanz; without their help, there would be no nonlinear fiber. I would also want to personally thank Alexander Sell, the Ph.D. student (now Doctor) who was my direct connection within the group in Konstanz, for his priceless help with the nonlinear fiber
(i.e. designing/making it) as well as for his numerous advice and tips on erbium-doped
fiber amplifiers.
For the last 7 years, the Miller group became a kind of adoptive family, bringing with it a mix of friendships and professional relations that is hard to describe. I owe a big thanks to all the previous group members, especially those that were there when I first joined.
Their kindness helped ease what could have been an extremely stressful situation; joining a large group of individuals (more than 15 people at the time) is not an easy task. Of all of them, a few deserve to be singled out: Sadia Khan, our former research facilitator, who was a a great help with the administrative business as well as a willing listener; Kresimir
Franjic and Renzhong Hua, who were my first mentors in the lab; Darren Kraemer, who was the trailblazer whose footsteps I followed as well as a constant source of advice and discussion. Over the years, the dynamic of the group changed, as well as my role in it; nonetheless, I would like to acknowledge the current members for their help and support, in particular Alexei Halpin, Christina Mueller, and Philip Johnson. As senior Ph.D. students, we had on more than one occasion discussions about our research, frustrations, and general topics regarding our life as graduate students, which helped me stay sane by keeping things in perspective.
I cannot write acknowledgments without mentioning my family. My brother and my sister who supported me throughout my Ph.D. and never once said "Shouldn’t you be done by now?", which is extremely nice of them. However, I owe more than special thanks to my parents; I would not have made it this far without their help. As a young boy, they successfully engaged me with science through weekend activities, which could be seen as
v the starting point of this long journey. During my undergrad, my parents constantly pushed me to aim for the highest grade that I could possibly achieve. Although it was not always easy (and sometimes annoying), their healthy encouragement taught me to strive for my maximum potential.
And finally, I would like to thank all of my friends who have been there for me during all those years. In particular, all of those who had to endure my complaints: Jean-
Michel Menard, my fellow Quebecer in the Physics department, who was my partner in crime (mostly telling bad jokes) from the beginning; Cristen Adams, another physicist, with whom I walked so many times from the department to the East side, especially after PGSA pub nights; Adrienne Marcotte, with whom I biked to all corners of Toronto;
Martin Parrot, an old friend since my youth, who brought the Canadiens’ fever in Toronto; and finally, Sue Hobson, to whom I owe a special thanks, as she was nice (crazy?) enough to read over my entire thesis, hunting for mistakes and poor grammar, which was not an easy task (i.e. ESL).
vi Contents
1 Introduction 1
1.1 LaserSelectiveChemistry ...... 1
1.2 Mid-infrared optical parametric amplifier ...... 3
1.3 Chirped-pulse amplification ...... 4
1.4 Thiswork-quickoverview...... 5
2 Generation of the seed:
Nonlinear propagation in fibers 7
2.1 Maxwellequations ...... 7
2.2 Spatialmode ...... 11
2.3 Dispersion...... 18
2.4 Nonlinearpropagation ...... 23
2.5 Self-phasemodulation ...... 28
3 Erbium-doped fiber system for signal amplification 35
3.1 Opticalproperties...... 35
3.2 Dispersioninerbium-dopedfibers ...... 38
3.3 Fiberlaser...... 41
3.4 Erbium-dopedfiberamplifier...... 42
4 Highly-nonlinear fiber for all-optical synchronization 50
vii 4.1 Precompensatingfiber ...... 50
4.2 Dispersionoptimizedfiber ...... 52
4.3 Mechanism for wavelength tuning ...... 53
4.4 Result ...... 53
5 Nd:YLF amplification system 58
5.1 Opticalproperties...... 58
5.2 Regenerativeamplifier ...... 63
5.3 Regenerativeamplifierdesign ...... 71
5.4 Multipassamplifier ...... 75
6 Optical parametric amplifier system 78
6.1 Optical parametric amplification ...... 78
6.2 Optical Parametric Chirped Pulsed Amplifier ...... 91
6.3 Stretching-Compression...... 91
7 Optical Parametric Chirped-Pulse Amplifier 100
7.1 OPCPA-Design ...... 100
7.2 OPCPA-Performance ...... 106
8 MIR OPA development 116
8.1 Design...... 116
8.2 Performance...... 120
9 Conclusion 129
9.1 Futurework-improvement ...... 130
9.2 Closingwords ...... 131
viii List of Tables
2.1 Coefficients for the Sellmeier equation for the core, i = 1, and for the
cladding, i =2 (from[16])...... 23
3.1 The parameters for the emission and the absorption cross-section [29]. . . 39
5.1 The beam radius calculated with Paraxia at different elements in the cav-
ity. The two non-zero focal lengths represent the thermal lens...... 72
ix List of Figures
2.1 Graphic solution for equation (2.39) for V = 8 and for (a) l = 0 and (b)
l = 1. In this case, there will be 3 modes associated to l = 0, and two to
l =1.(Adaptedfrom[12]) ...... 16
2.2 Dispersion for a single-mode fiber expressed using (a) the dispersion pa-
rameter and (b) the group velocity dispersion...... 23
2.3 The spectra for different values of nonlinear phase, one the left for a Gaus-
sian pulse and on the right for an hyperbolic-secant pulse...... 30
2.4 Pulse envelope of a N=2 soliton at different positions...... 32
3.1 On the left, the details of the energy levels of the laser transition (adapted
from [26]). A few transitions are highlighted; the arrows indicate if they
show up in the absorption or the fluorescence spectra or both. On the
right, the emission and the absorption spectrum around 1550 nm for the
erbium-doped fiber (adapted from [27])...... 36
3.2 Absorption spectrum for an erbium-doped fiber (adapted from [26]). The
three common pumping bands are shown with the arrows. The level in-
volved and the wavelength associated with it are indicated...... 38
3.3 The second-order dispersion for two different ion concentrations. The con-
centration was set to (a) 4.25 1025 ions/m3 and to (b) 14.2 1025 ions/m3. The arrows indicate the direction of increasing inversion starting with all
the ions in the lower level until the system is fully inverted...... 40
x 3.4 Layout of the fiber laser; WDM,wavelength-division multiplexer, OC, Out-
putcoupler...... 41
3.5 On the left: The retrieved electric field for the output of the fiber laser.
On the right, the retrieved and the measured spectrum...... 42
3.6 Layoutofthefiberamplifier...... 43
3.7 On the left, the output spectrum of the first amplifier with both pump
diodes at their maximum current, 520 mA, as well as the spectrum for the
input. On the right, the spectra for different currents of the first diode. . 44
3.8 The spectra at the output of the first fiber amplifier with three different
currents for the first laser diodes using a fiber (a) before and (b) after. . . 46
3.9 On the left: Retrieved electric field of the output of the first amplifier. On
theright,retrievedandmeasuredspectra...... 47
3.10 On the left: Retrieved electric field at the output of the second amplifier.
The FWHM is 89 fs and 65% of the energy is in the central peak. On the
right,retrievedandmeasuredspectra...... 48
3.11 The spectra, (a) to (d), and the temporal profile, (e) to (h), for different
currents in the first diodes. Also for each current, the spectrum/pulse
shape for the operating current of 436 mA is plotted (dash curve) for
comparaison...... 49
4.1 On the left side, the dispersion curves for the two components of the
HNLF. On the right side, the index profile for the dispersion-optimized
fiber. Typical value for the index change is around 3 % for the increase
and below 0.5 % for the reduction [36, 37]. The ratio of the two diameters
isontheorderof0.5-0.6[37]...... 52
xi 4.2 The spectrum at the output of the HNLF. On the left, the spectrum asso-
ciated with the dispersive wave and on the right, the spectrum associated
with the soliton and the unshifted beam. 3 mW is generated in the dis-
persivewave...... 54
4.3 The spectrum at the output of the HNLF. Only the peaks associated with
the dispersive wave and the red-shifted soliton are shown...... 55
4.4 The spectrum of the dispersive wave for different temperatures of the holder. 56
5.1 On the left: the two levels associated with the main transition and their
sub-sublevels [43]. On the right: The stimulated cross-section for each
polarization for the 4F 4I transition [42]...... 60 3/2 → 11/2 5.2 On the left side, absorption spectrum for Nd:YLF [46]. On the right side,
4 a close-up of the absorption associated with the F5/2 level...... 61 5.3 The detail of the polarization state in the case of (a) the switch close and
(b) the switch open. The dash arrows show the direction of propagation.
TFP, thin-film polarizer, QWP, quarter-wave plate...... 64
5.4 The simulated output intensity of a pulse after going through 160 times in
(a) a single 1-mm etalon, and (b) a combination of a 1-mm etalon and a
0.7-mm etalon. Two different pulse durations were used: 3.5 ps and 10.5 ps. 67
5.5 On the left side: the transmitted pulse shape for angle from 0◦ to 4◦, with a
interval of 0.25◦ between each. On the right side, the integrated intensity
as a function of the incident angle, as well as the phase. The numbers
relatefeaturestoeachother...... 68
5.6 The normalized gain spectrum for low and high gains. In both case, the
sameatomiclinewidthwasused...... 69
5.7 The effect of saturation on a square pulse. The first curve is the input
pulse, whereas the other two are the output for two propagation lengths.
Adaptedfrom[53]...... 71
xii 5.8 Layout of the regenerative amplifier. LH, laser head, HWP, half-wave
plate, QWP, quarter wave-plate, L, lens, TFP, thin-film polarizer, E,
etalons,M,mirrors...... 72
5.9 Autocorrelation trace of the output of the regenerative amplifier with no
etalon inside. The FWHM of the trace is 17 ps resulting in a pulse duration
of12ps,ifaGaussianfitisused...... 73
5.10 Cross-correlations traces for (a) a 1.5-mm etalon, (b) a 1-mm etalon, and
(c) and a 1-mm and a 0.7-mm etalons. The effect of each etalon was
simulated and the results are shown in the dashed line...... 74
5.11 Layout of the multipass amplifier; LH, laser head, HWP, half-wave plate,
L, lens, TFP, thin-film polarizer, M, mirror...... 76
5.12 Cross-correlation traces with the multipass off (0 A) andon(18A). . . . 76
6.1 Orientation of the wavevector with respect to the principal axes of the
crystal...... 84
6.2 Diagram for a plane wave propagating in the X-Z plane. The thicker
arrows indicate the two possible polarizations...... 85
6.3 On the left: The orientations of different fields are shown, as well as the
wavevector and the Poynting vector S. On the right: Because the Poynting
vectors for the two polarizations are different in the crystal, although the
beam is at normal incidence, one of the two polarizations will be bent at
the interface. On the other hand, the wavevectors and the wavefronts for
thetwobeamsarestillparallel...... 87
6.4 Diagram of the different wavevectors in the noncollinear geometry. The
(internal) noncollinear angle, α, is defined in the crystal...... 87
6.5 This diagram illustrates the different angles used to quantify the dispersion
in the idler. For its calculation, it is important to use the external angle,
γout...... 90
xiii 6.6 On the left: The design for the compressor. On the right: The layout for
thestretcher...... 92
6.7 The details of the propagation in the compressor...... 93
6.8 The two pulse trains having similar repetition rates, around 80 MHz; the
difference between the two is 0.1%. On the left, the first four pulses are
shown; at t =0, the pulse from both trains overlap perfectly. On the right,
to amplify the effect of the repetition rate mismatch, the same four pulses
are shown after 500 ns; the overlap is clearly lost. However, for the fourth
pulse in (a), the separation is already 37 ps...... 97
7.1 Phase matching curves at different noncollinear angles for (a) KTA, and
(b) KTP. The noncollinear angle is varied from 0◦ to 4◦, going from left
to right. For KTA, the extra dashed curve is 3.2◦...... 102
7.2 The details for (a) the compressor and (b) the stretcher, with a top view
and a side view. The drawings are not to scale...... 103
7.3 A general layout of the whole OPCPA system. The specifics of each com-
ponent, e.g. Nd:YLF amplifiers, are offered in their respective sections.
The remaining optics are: lenses for beam shaping and transport, half-
and quarter-wave plates (HWP/QWP) for polarization control, and Fara-
day optical isolators (FOI) to protect the nonlinear fiber against leakage
fromtheamplifier...... 106
7.4 The amplified and the unamplified spectrum for (a) the original stretcher,
and (b) the optimized version. Although, the central wavelength differed,
thespectralwidthplottedisthesame...... 109
7.5 The spectra for different noncollinear angles. The angle quoted is the
internalone...... 110
7.6 Layout, to scale, of the KTA amplification stages. Important distances
arealsoindicated...... 111
xiv 7.7 The beam profile after (a) the second crystal, and (b) the compressor. . . 112
7.8 On the left: the retrieved electric field intensity and the retrieved temporal
phase. On the right: the retrieved spectrum and the retrieved phase as
wellasthemeasuredspectrum...... 113
7.9 Cross-correlation trace between the seed and the pump...... 114
8.1 Phase-matching curves for different noncollinear angles - from left to right:
0◦,1◦,2◦,3◦...... 118
8.2 By using the proper focal lengths, the collimated beam diameter for both
dispersiveelementscanbeequal...... 119
8.3 Plot showing the magnification needed for different incident angles, for a
200-lines/mmgrating...... 120
8.4 The layout for the angular dispersion compensation setup. L1 is the 150-
mm ZnSe lens and L2 is the 75-mm CaF2 lens...... 121 8.5 The diagnostic tool for the spatial chirp. The mirror is placed a focal
length away from both the grating and CCD camera...... 122
8.6 The beam profile in the diagnostic tool for (a) the compensated beam,
and (b) the uncompensated one. Note, the images are rotated by 90% to
match the actual input of the OPA. In the inlets, the beam profile at the
entranceisshown...... 122
8.7 LayoutoftheMIROPA...... 123
8.8 Results of the characterization of the signal (a-c) and the idler (d-e) of
the AGS OPA. The vertical structures in both beam profiles arise from a
triggering issue. There is no auto-correlation trace for the idler...... 124
8.9 Results of the characterization of the signal (a-c) and the idler (d-e) for
the LGSe OPA. The vertical structures in both beam profiles arise from a
triggeringissue...... 127
xv Chapter 1
Introduction
The term mid-infrared (MIR) refers to the region of the electromagnetic spectrum going
from 2 m to 20 m, although these boundaries are somewhat artificial and can change depending on the field of research. For chemistry, MIR is extremely important as the frequencies associated with atomic vibrations lie in this region. For many decades already,
MIR spectroscopy has been used to learn more about molecules and their interactions with the environment. More recently, with the advent of ultrafast technology, dynamics and structural changes can be probed in the sub-picosecond time-domain.
This is the motivation behind the work done in this thesis; not just in terms of enabling similar experiments to those already done, but also to push the boundaries for interactions between molecules and electric fields. The system developed here lays the groundwork for a scalable laser system that will deliver hundreds of microjoules in the MIR, enabling strong field control, with laser selective chemistry (LSC) being the ultimate goal.
1.1 Laser Selective Chemistry
Chemistry, at its simplest, is the art of breaking molecular bonds to make new ones. It might sound trivial, but it is not; a great deal of effort is put into finding new catalysts
1 Chapter 1. Introduction 2 to achieve specific chemical reactions.
With the invention of the laser, a dream was born: using this new tool to selectively break bonds [1, 2]. With their narrow bandwidth, high-power continuous-wave lasers could be tuned to a particular stretching mode in a molecule and be used to pump energy into it until it breaks. However, by definition, breaking a bond is a highly anharmonic process. As energy is pumped into the transition, its frequency gets red-shifted with respect to the laser and the interaction is pushed out of resonance.
Using broadband sources or multiple lasers could be an option if it was not for the fact that the energy does not stay in the targeted bond. On the most basic level, a molecule can be described as a series of balls and springs, and vibrations can be analyzed using normal modes. This works perfectly as long as the molecule stays in a low vibrational level. As energy is put into the molecule and higher levels are reached, the harmonic approximation has to break down, otherwise bonds could not be broken. Because of the anharmonicity, different normal modes can be nonlinearly coupled together. Therefore if energy is put into a given mode, it will leak out; this is known as intramolecular vibrational-energy redistribution (IVR) [3].
The behavior of IVR varies depending on the size of molecules [3, 4]. For small molecules, there are only a few modes and the energy can flow back into the initial mode. However this low number of modes makes small molecules a bad model to test laser selective chemistry (LSC). For larger ones, the number of modes is such that the molecule acts as a bath, and the energy flow is irreversible, greatly complicating LSC.
The time scale for IVR in large molecules is sub-picosecond.
One way to try to beat IVR is to use high-energy ultrashort pulses in the MIR.
First, energy could be deposited into a bond before it can leak out. And secondly, the broad spectrum associated with short pulses can be matched to the spectrum of the vibrational potential, reducing the effect of its anharmonicity. High pulse energy is necessary to saturate every transition. Around ten photons are needed to reach the Chapter 1. Introduction 3
top of the potential; the probability to absorb one photon has to be close to unity,
otherwise the absorption of a high number of photons will be highly improbable. For
a molecule like benzene with a cross-section of 2.2 10−19 cm2 [5], this means a photon flux of 10/ (2.2 10−19) photon/cm2 is needed. For a 3- m beam focused down to a 100- m diameter, this results in a pulse energy around 250 J, which at the moment is not
achievable at kilohertz repetition rates.
1.2 Mid-infrared optical parametric amplifier
So far the major limiting factor for LSC is technological. The tools needed to really ask
questions about intramolecular potentials are simply not there. To understand how the
project presented in this thesis could be part of the solution, it is important to look at
how MIR is currently generated.
Currently, most MIR systems are based on Ti:Sapphire amplifiers, as they can easily
produce sub-100-fs pulses with millijoules of energy at 1 kHz. However, their operating
wavelength, 800 nm, is relatively short compared to MIR, putting harsh requirements on
the crystal, especially on their transmission window. To avoid two-photon absorption, it
is important for the nonlinear crystal to be transparent to the second-harmonic of the
pump, which for Ti:Sapphire is in the ultra-violet. For this reason, MIR is generated
using two cascaded nonlinear processes. First, an optical parametric amplifier (OPA) is
used to generate light in the near infrared. Then the resulting signal and idler are mixed
in a nonlinear crystal; for a difference-frequency generation (DFG) process, the output
of this second nonlinear stage can cover a wide region, 3 m to 20 m [6]. Because two
different nonlinear stages are used, the efficiency is low, around 0.5 %, mostly because of
the DFG stage. With this technology, producing 250 J of MIR would therefore require
50 mJ from a Ti:Sapphire, which is far above the performance of current amplifiers.
By using a longer wavelength to start with, the OPA could be used to produce MIR Chapter 1. Introduction 4
directly, which would greatly improve the efficiency. This can be easily done by using an
Erbium-doped fiber laser, operating at 1.55 m. Although this only solves part of the problem: the pump for the OPA still needs to have enough pulse energy.
1.3 Chirped-pulse amplification
Amplifying broadband laser pulses to high power can be challenging. The major problem is with their short pulse duration, even at low pulse energy, the peak intensity can be high enough to induce optical damage. For this reason, chirped-pulse amplifiers (CPAs) were introduced in 1985 [7] and are now widely used. In chirped-pulse amplification, dispersion is introduced in a controlled manner to temporally stretch the pulse and reduce the peak power. After amplification, the pulses can be recompressed, the result being short pulses with a high energy.
However, scaling CPAs to high power is still complicated; the major hurdle is heat management. The excess energy of the pump photon is dissipated as heat in the gain medium, as the pump power increases, the thermal effects become more important. Ther- mal lensing, which degrades the beam quality, and thermal-induced stress, which can lead to catastrophic failure of the crystal, put limits on the amount of pumping power that can be used. Furthermore, for high-gain systems, the bandwidth is seriously reduced because of gain-narrowing.
Using nonlinear optics, in a CPA configuration, is an interesting alternative to amplify broadband pulses to high energy; such systems are known as optical parametric chirped- pulse amplifiers (OPCPAs). Because the extra-energy of the pump photon is released in the form of a light, even in high-power nonlinear systems, heat is not a limiting factor.
Furthermore, because their operating wavelength is easily tunable, OPCPAs can be used with almost any oscillator.
The major disadvantage of OPCPAs is the harsh requirements put on the synchro- Chapter 1. Introduction 5 nization between the different elements. With both CPAs and OPCPAs, a broadband seed is amplified by a narrowband pump, and they are obtained by using two synchro- nized lasers. For CPA, the degree of synchronization can be relaxed compare to the
OPCPA concept as the level of synchronization needed is determined by the upper-level lifetime, which is on the order of a microsecond. On the other hand, for OPCPA, more sophisticated synchronization schemes are needed because the gain is instantaneous; in this case the pulse duration, on the order of picosecond, sets the level of the precision required.
1.4 This work - quick overview
In this thesis, a new scheme to generate MIR light is demonstrated, the goal of which is to improve the efficiency and provide a new pump source with an all-optical synchronization as an enabling element for further scaling of power.
The first and important distinction with regards to currently used MIR systems is that this one is based on a Erbium-doped fiber laser. This choice was made to open up new options for the MIR OPA design, but also because fiber lasers are affordable and reliable, especially compared to Ti:Sapphire systems.
The foundation upon which this new scheme can be built is OPCPA using a potassium- titanyl-arsenate crystal pumped by a neodymium- doped lithium yttrium fluoride (Nd:YLF) amplifier and seeded by the fiber laser. Its importance resides in the simple fact that without the OPCPA, there is no long-wavelength pump for the MIR OPA.
An important contribution of this work, especially compared with previous work from this group, is the use of passive synchronization. The scheme used, which is introduced for the first time in this work, reduces the timing jitter to a usable level and simplifies the everyday use of the OPCPA. The output of the fiber laser is split in two arms, seed and pump, both having a fiber amplifier. For the pump arm, the output of the amplifier Chapter 1. Introduction 6
is coupled into a nonlinear fiber to generate the seed for the Nd:YLF amplifier. Although
the fiber was provided to us by the group of Prof. Leitenstorfer, from University of
Konstanz, a great deal of work went into incorporating it to the previous system and
ensuring the stability of its output.
The pump of the OPCPA is a Nd:YLF regenerative amplifier followed by a Nd:YLF
double-pass amplifier. Although both are well established tools in this group, some care
and effort had to be put in making them compatible with the broadband low-power seed
generated by the nonlinear fiber. The major modification is the introduction of etalons
in the regenerative amplifier cavity to stretch the output pulses and therefore avoiding
optical damage. However, many minor modifications were needed, especially regarding
the temporal overlap of the pump and the seed pulse.
The compressed output of the OPCPA is used to pump the MIR OPA, for which two
crystals were tested: silver thiogallate and lithium gallium selenide. For the proof-of-
principle, the idler of the OPCPA is used to seed the OPA. This new scheme based on
a pump at 1.6 m represent an important contribution of this work and was recently published [8]. Moreover, as lithium gallium selenide is a fairly recent crystal, this work is among the first to investigate its possible use as a replacement for silver thiogallate.
This system provides a robust platform for scaling to average powers and peak powers that will open up LSC as well as numerous other applications requiring high average-peak power pulses in the MIR. Chapter 2
Generation of the seed:
Nonlinear propagation in fibers
Optical fibers play a crucial role on the source side of the optical parametric chirped pulse amplifier. The seeds for both the optical parametric amplifier and the pump laser are derived from a single fiber oscillator, and the light needed to seed the pump laser is generated in nonlinear fiber. Understanding how the electromagnetic field interacts with the fiber is the stepping stone for the next two chapters. In this chapter, the spatial profile of the field and the propagation of a pulse in a fiber will be described in detail.
2.1 Maxwell equations
The propagation and the spatial mode are both described by Maxwell’s equations. This section will lay the ground for the more in-depth discussions that will follows. As fibers are generally non-conducting, σ =0, and non-magnetic, = 0, with no charge density,
7 Chapter 2. Generation of the seed: Nonlinear propagation in fibers 8
ρ =0, Maxwell’s equations can be written as [9]:
∂H (t, r) E (t, r)= , (2.1a) ∇× − 0 ∂t ∂D (t, r) H (t, r)= , (2.1b) ∇× ∂t D (t, r)=0, (2.1c) ∇ H (t, r)=0, (2.1d) ∇ where,
D (t, r)= ǫ0E (t, r)+ P (t, r) . (2.2)
The polarization, P, describes the reaction of the medium to the electric field. For a
weak field, the polarization is a linear function of the electric field. As the field increases,
it is necessary to include a nonlinear correction.
P (t, r)= Pl (t, r)+ Pnl (t, r) , (2.3)
where Pl (t, r) is the linear polarization and Pnl (t, r) the nonlinear polarization. In free space propagation, the nonlinear part is normally omitted as the electric field rarely gets
strong enough. However, for propagation in a fiber, the field is confined to a tight area for
a long distance, and, even for continuous wave lasers, nonlinearity can become important.
Furthermore, for short laser pulses the peak power can be orders of magnitude higher
than the average power. However, even for short pulses in fibers, it is still possible to
treat the nonlinearity as a perturbation and solve Maxwell’s equations, to get the mode
profile and the dispersion using only the linear polarization.
Pulses are often described by separating the overall field in term of an oscillating wave
and an envelope. Although this approximation is in general valid, it may break down for
few-cycle pulses [10]. The pulses described in the thesis are long enough for this so-called
phasor notation to be used. In this case, the fields and the displacement can then be Chapter 2. Generation of the seed: Nonlinear propagation in fibers 9
written as [11]:
1 E (t, r)= E (t, r) e−iω0t + c.c. , (2.4) 2 0
1 − 0 H (t, r)= H (t, r) e iω t + c.c. , (2.5) 2 0
1 − 0 D (t, r)= D (t, r) e iω t + c.c. , (2.6) 2 0 where E0, H0, and D0 are the envelopes for the electric field, magnetic field and displace- ment, respectively, for a wave with a carrier frequency ω0. Using this phasor notation, Maxwell’s equations can be rewritten as:
∂ E (t, r) e−iω0t = H (t, r) e−iω0t , (2.7) ∇× 0 − 0 ∂t 0