The Effect of Proton Bombardment on Semiconductor
Saturable Absorber Structures
by
Juliet Tara Gopinath
B.S., Electrical Engineering
University of Minnesota, 1998
Submitted to the Department of Electrical Engineering and Computer Science in Partial
Fulfillment of the Requirements for the Degree of
Master of Science
at the
Massachusetts Institute of Technology
September 2000
@ Massachusetts Institute of Technology, 2000. All rights reserved.
Signature of A uthor......................................................
Department of Electrical Engineering and Computer Science
September 1, 2000
Certified by...........................................................ich.....
Erich Ippen
Elihu Thompson Professor of Electrical Engineering
-~~~ 2Thesm Supervisor
Certified by.............................
.. ...............................
Arthur C. Smith
Chairman, Department Committee on Graduate Students
. . . . .....
MASSACHUSETTS INSTITUTE
OF TECHNOLOGY
OCT 2 3 2000
LIBRARIES
BARKER
The Effect of Proton Bombardment on
Semiconductor Saturable Absorbers
by
Juliet Tara Gopinath
Submitted to the Department of Electrical Engineering and Computer Science
on September 1, 2000 in partial fulfillment of the requirements for the
Degree of Master of Science
Abstract
Carrier lifetime reduction resulting from proton bombardment of InGaAs/InP-based semiconductor saturable absorbers was studied experimentally, using a standard degenerate, cross-polarized
pump-probe technique. Proton bombardment reduced carrier lifetimes by as much as a factor of
40 at low optical excitation densities. For high fluences, significant induced absorption was
observed. The recovery of this excited state absorption did not show as significant a dependence
on the level of proton bombardment. It is possible that the cause of this induced absorption - carriers outside the InGaAs quantum wells, highly excited carriers, or those trapped in satellite valleys - is not sensitive to the effects of bombardment. Also, the bombardment-created defects may
saturate at such high fluences. The detrimental side-effects of proton bombardment - reduced
modulation depth and increased non-saturable loss - have been shown to be mitigated with a short
post-growth anneal. Finally, modelocking was demonstrated with the proton-bombarded samples
in an erbium fiber laser.
Thesis Supervisor: Erich P. Ippen
Title: Elihu Thompson Professor of Electrical Engineering
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
3
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
4
Acknowledgements
I am grateful to Professor Ippen for his guidance and supervision. Without his insights on experimental techniques and intuition about the physics studied in this experiment, this thesis would not
have been possible. I enjoy working in his group, the research, and his warm personality and
good sense of humor. I thank Professor Haus for also stimulating me and keeping me on my toes.
I am indebted to Erik Thoen for teaching me how to perform the measurements in this thesis and
helping me through many of the experimental bugs. He was (and still is) never too busy to spare
a minute to discuss data, despite an extremely hectic schedule. He seems to have a supply of infinite patience and was always willing to bail me out with my endless problems with the OPO
(Optical Parametric Oscillator) even from home! I thank Elisabeth Marley Koontz for growing
the structures studied, and for her help and insights with the experiments, although she still calls
me a "kid"! Also, I thank both Erik and Elisabeth for their friendship. Without them, this thesis
would not have been possible!
Dan Ripin was often able to give me good advice and gave much needed encouragement. He was
always willing to explain difficult concepts with clarity and to lend a hand in lab, and has also
been quite a walking encyclopedia of Boston-area restaurants! Matt Grein has encouraged me
along the way, and taught me all I know about fiber lasers to date. He has been generous with his
time and explanations. Leaf Jiang has also been helpful - whether it be with computer problems
or lab equipment. I thank my officemates, Pei-Lin Hsiung (I owe her many a dinner), Rohit
Prasankumar, and John Fini for all their camaraderie (and wondering where I was that week I took
off for vacation without telling them)! I thank the rest of the group for their friendliness and help:
Peter Rakich, Milos Popovic, Pat Chou, Charles Yu, Christine Manolatou, Jalal Khan, Mike
Watts, J. P. Laine, Dr. Yijang Chen, Dr. Franz Kartner, and the Fujimoto contingent of Seong-Ho
Cho, Tony Ko, Costas Pitris, Andrew Kowalevicz, Kathleen Saunders, Ravi Ghanta, Dr. Ingmar
Hartl, Dr. Christian Chudoba, Dr. Xing-De Li, and Dr Kaoru Minoshima. From Professor Leslie
Kolodziej ski's group, Dr. Gale Petrich has also been helpful (and also one of few here who under-
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
5
stands the meaning of cold weather). I'd also like to thank the three group secretaries Cindy
Kopf, Mary Aldridge, and Donna Gale for all their help and interesting gossip.
Dr. Markus Joschko and Dr. Patrick Langlois did much pump-probe that has helped us to understand these structures better. I thank Dr. Joe Donelly for many insightful discussions about proton
bombardment and annealing. Bob Bailey at Lincoln Labs also provided help and resources. I
thank Elisabeth L. Shaw for help with the CMSE facilities and Peter O'Brien at Lincoln Labs for
coating deposition.
I am grateful for the support of a National Science Foundation fellowship.
I thank Professor James Leger, Professor Mostafa Kaveh, Professor C. C. Huang (University of
Minnesota) and Professor Gadi Eisensten (Technion, Israel Institute of Technology) for their
guidance, advice, and encouragement as well as the research opportunities they provided.
I thank Professor Marcus Thompson, my viola teacher here at MIT, for wonderful lessons and
musical opportunities that have preserved my sanity while working on this thesis. Thanks to my
Grieg string quartet for also providing stress relief and putting up with this project! I am grateful
to all my friends for their support and for putting up with all the complaints that went along with
this thesis.
I thank my parents and my little sister, Charlotte, for their support and love. Without their inspiration and help, I don't think that I'd be here at MIT. I am very excited to be finishing this thesis.
I'm still wondering whether it is a dream. Thank you all for making it real!
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
6
Table of Contents
Acknowledgements
5
Chapter 1 Introduction
9
Chapter 2 Semiconductor Saturable Absorbers
11
2.1
Review of Modelocking Techniques.........................................11
2.2
Saturable Absorber Theory......................................................
16
2.3
Types of Saturable Absorbers..................................................
24
2.4
Absorber Lifetime Reduction Techniques...............................25
2.5
Growth of Semiconductor Saturable Absorbers......................28
2.6
Optical Characterization of Saturable Absorbers....................34
Chapter 3 Experimental Design
36
3.1
Pump-Probe Theory..................................................................
36
3.2
L aser Sources.............................................................................
37
3.3
Experim ental Setup.................................................................
40
Chapter 4 Experimental Results
4.1
43
Pump-Probe of Non-Proton-Bombarded Absorbers..................43
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
7
4.2
Pump-Probe of Proton-Bombarded Absorbers............................58
4.3
Pump-Probe of Proton-Bombarded InP Structures.....................73
4.4
Laser Results from Proton-Bombarded Absorbers.....................77
Chapter 5 Conclusions and Future Work
80
5.1
C onclusions.................................................................................
80
5.2
Future Work.................................................................................
81
83
References
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
8
Chapter 1: Introduction
Saturable absorbers have been used for passive modelocking for many years. With the recent
explosion in telecommunications, their use for the generation of short pulses at 1500 nm is of particular interest. Semiconductor saturable absorbers often exhibit a bitemporal response, consisting of fast and slow components. The fast time constant is on the order of 100 fs to several
picoseconds, and results from intraband dynamics. The slow time constant, due to recombination,
is on the order of picoseconds to several nanoseconds, depending on the materials system and
growth parameters. For a saturable absorber to be useful at high repetition rates and for pico- and
femtosecond pulse generation, it is important to reduce the long recovery time constant.
Lifetime reduction in a semiconductor can be accomplished by the introduction of defect states.
Much research has been done on the use of low-temperature growth and ion bombardment for this
purpose. The effect of proton bombardment on semiconductor saturable absorbers at 1500 nm is
the subject of this thesis. To the best of my knowledge, there have not been previous studies of the
effects of proton bombardment on this particular type of saturable absorber. In addition, experiments have not been conducted at 1.5 jm, the telecommunications wavelength, or extensively at
high fluence, important for the operating conditions of lasers.
Time resolved spectroscopy is ideal for studies of absorber recovery times. I performed measurements, using a cross-polarized collinear degenerate time-resolved pump-probe technique. With
150 fs pulses from an Optical Parametric Oscillator (OPO), tunable from 1400 to 1600 nm, I
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
9
achieved a time resolution of 300 fs and minimum detectable signal levels of 10-3. Proton-bombarded absorbers that were characterized were then tested in a 17 MHz erbium fiber laser.
In Chapter 2, the motivation and theory of modelocking with saturable absorbers is discussed. A
brief review of modelocking techniques is presented, and then a discussion of saturable absorber
theory follows. Previous work on saturable absorber lifetime reduction is presented. Finally, a
description of the devices investigated and the fabrication process is presented. Chapter 3 contains a description of the experiment design and theory. The characteristics and limitations of the
laser system used for the pump-probe measurements are discussed. Details of the experimental
setup are described. In Chapter 4, experimental results are presented. At low fluences (fluences
less than the absorber saturation fluence), proton bombardment can reduce lifetimes up to a factor
of 40. However, at high fluences (fluences greater than the absorber saturation fluence), significant induced absorption is observed. The recovery of this induced absorption does not show a significant dependence on the level of proton bombardment. It is possible that the cause of this
induced absorption - carriers outside the InGaAs quantum wells, highly excited carriers, or those
trapped in satellite valleys - is not sensitive to the effects of bombardment. Also, the bombardment-created defects may saturate at such high fluences. The detrimental side-effects of proton
bombardment - reduced modulation depth and increased non-saturable loss - are shown to be mitigated with a short post-growth anneal. Finally, the results of tests in a linear erbium fiber laser
are presented. All proton-bombarded absorbers studied modelocked the laser, and for two of
them, self-starting modelocking was observed. In Chapter 5, conclusions and future work are discussed.
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
10
Chapter 2: Semiconductor Saturable
Absorbers
This chapter contains a brief review of modelocking techniques, to motivate interest in developing
more effective semiconductor saturable absorbers. The parameters that can be adjusted in saturable absorber design are discussed; specifically, the importance of absorber lifetime reduction is
addressed. Two different techniques for lifetime reduction are presented: ion bombardment and
low-temperature epitaxial growth. The fabrication process for the absorbers studied in this thesis
is described, along with a short description of the optical characterization necessary for accurate
growths.
2.1 Review of Modelocking Techniques
With the current revolution in Internet and communications, the demand for bandwidth doubles
every 100 days, creating a demand for higher repetition rate, shorter pulse sources. Ultrashort
pulses also have important applications in spectroscopy, high resolution imaging technology, and
surgery. In the past thirty years, the generation of short pulses has been revolutionized: from
pulses of 10- 11s to pulses of femtosecond length, a decrease of four orders of magnitude. The first
sub-picosecond pulses where produced in 1974 [23], with a dye laser. Semiconductor diode
lasers and color center lasers were modelocked in a similar fashion, using slow saturable absorber
modelocking. However, advances in ultrashort pulse generation in solid-state lasers (below pico-
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
I1I
second pulses) only picked up pace in the 1980's, with the use of fast, reactive nonlinearities for
modelocking. Currently, 5 fs pulses, about two optical cycles long, have been produced from a
Titanium:sapphire (Ti:sapphire) laser [48] [56], using Kerr lens modelocking (KLM).
0-11
101
Nd:glass
Nd:YAG
Nd:YLF
S-P Dye
Dye
1 01
Diode
CW Dye
Color
Center
Cr:YAG
Cr:LiS(C)AF
Er:fiber
CPM
Cr:forsterite
Nd:fiber
10
14
Compression
Ti:sapphire
I
I
I
I
I
I
I
1965
1970
1975
1980
1985
1990
1995
1
FIGURE 2.1 History of ultrashort pulse generation. The achieved pulse width versus year for several different
types of lasers systems. Figure from [23].
In a modelocked laser, the phases of a group of longitudinal modes (frequencies) are "locked"
together. To produce modelocked operation, a variety of techniques, both active and passive, can
be employed to favor a pulsed state over continuous-wave operation. All techniques involve the
introduction of an intensity-dependent loss into the system.
Amplitude modulation, developed thirty-five years ago, is one method of short pulse generation.
An amplitude modulator can be placed in a laser cavity, providing amplitude filtering at the cavity
round-trip frequency. A pulse arriving at maximum transmission experiences minimum loss; thus
shorter pulses see less loss than longer ones. However, shorter pulses are shortened progressively
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
12
less and less by the modulator [23]. Therefore, active modelocking is not suitable for the production of ultrashort pulses.
Shorter pulses can be generated with passive modelocking techniques in which the laser self-modulates itself. Most of these passive techniques fall under the general heading of saturable
absorber modelocking. These methods have been demonstrated in both solid-state and fiber
lasers, stimulated by the advent of broadband gain media in the last decade.
Pulses can be generated with so-called "real" saturable absorbers, absorbers that rely on material
excitation to produce intensity-dependent loss. As intensity increases, the transmission of these
"real" absorbers increases, thus favoring shorter pulses with higher peak intensities. Semiconductor saturable absorbers, the subject of this thesis, can be used in several different ways to modelock lasers. In so-called "slow" saturable absorber modelocking, the leading edge of the pulse is
shaped by the saturable absorber response. However, the saturable absorber recovery time is not
fast enough to shape the trailing edge of the pulse. Instead, the trailing edge of the pulse is shaped
by the gain recovery of the laser medium itself. With this technique, pulses much shorter than the
recovery time of the absorber can be produced [23].
Pulses can also be generated with "fast" saturable absorber modelocking. With this method, the
absorber itself shapes both the leading and the trailing edges of the pulse. Thus, one is limited in
"fast" saturable absorber modelocking by the recovery time of the absorber. Since real saturable
absorbers rely on material excitation to produce their intensity-dependent loss, the response time
is dominated by the recovery of the medium. This usually occurs on a picosecond or nanosecond
scale, depending on the materials system and fabrication techniques. Thus, semiconductor saturable absorbers are often used in "slow" saturable absorber modelocking.
Fast saturable absorber modelocking is often implemented with so-called "artificial" absorbers.
Artificial saturable absorbers rely on the index nonlinearities of transparent media, often the gain
medium itself. In artificial saturable absorbers, changes in the refractive index with intensity are
converted to amplitude modulation. Because saturable absorption action is produced through
non-resonant optical nonlinearities, it is fast. Currently, the shortest pulses out of a solid-state
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
13
laser to date, 5 fs from a Ti:sapphire, are produced by Kerr lens modelocking, an artificial
absorber technique [48] [56].
With the Kerr nonlinearity (the nonlinear change of index of refraction with intensity), loss modulation can be introduced into the system by adding an aperture that allows more intense pulses a
greater transmission. An actual aperture can be introduced into the system (hard aperture KLM)
or the gain medium can play the role of an aperture when it varies spatially (soft aperture KLM).
However, in Kerr lens modelocking, the laser cavity is typically operated near the edge of the stability regime, thus requiring systems to be critically aligned within several hundred microns.
Polarization additive-pulse modelocking (APM), also uses the Kerr nonlinearity as an artificial
absorber. It is often used in fiber lasers, because the nonlinearity is weak, but the interaction
length is long. The cavity contains a Kerr medium, so that the pulse in this cavity experiences
self-phase modulation, causing an intensity-dependent phase shift, which can result in intensitydependent polarization rotation.
With the addition of polarizers, this can be converted into an
intensity-dependent loss, or self-amplitude modulation (SAM) [23]. This technique has mainly
been applied to fiber lasers, with pulses as short as 77 fs generated from Er doped systems [18].
However, it is difficult to build high-repetition rate APM sources because long interaction lengths
are needed to produce the polarization rotation.
One drawback of artificial saturable absorbers is that they are weak, a direct consequence of reliance on a non-resonant nonlinearity of the gain media (incidentally, the fact that gives them speed
too). Thus, self-starting is a problem of artificial absorber systems, as extremely high peak intensities are required for the nonlinearity to take effect. Real semiconductor absorbers can be used to
self-start artificial absorber systems, as they have much lower saturation fluences than artificial
absorbers. They also relax the constraint on critical cavity alignment required for KLM [29] [56].
Self-starting 6.5 fs pulses from a Kerr lens modelocked Ti:sapphire laser have been achieved with
an intracavity saturable absorber [28].
Pulses can also be generated by soliton modelocking, in which a pulse is shaped by the balance
between dispersion (causing temporal broadening) and self-phase modulation (causing spectral
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
14
broadening). In both slow and fast saturable absorber modelocking, a relatively short net-gain
window exists. However, in soliton modelocking, it has been shown that the net gain window can
remain open for up to ten times longer than the actual pulse [32], leaving room for instabilities to
grow. The soliton can shed to the continuum, causing instabilities and eventual pulse breakup.
The addition of a saturable absorbers can filter out this continuum, increasing overall system stability [31].
loss
loss
loss
gi gain
gain
FIGURE 2.2 The three passively modelocking models. From left to right: (a) passive modelocking with a slow
saturable absorber and dynamic gain saturation (b) fast saturable absorber modelocking and (c) soliton
modelocking. Figure from [32].
In summary, there are many techniques, active and passive, that can be used to generate short
pulses from lasers. Passive methods are capable of producing shorter pulse widths than active
techniques, since active techniques are inherently limited by the speed of electronics. Semiconductor saturable absorbers (so-called "real" absorbers) are one way to produce short pulses passively. Although the shortest pulses to date have been produced with Kerr lens modelocking, this
technique is difficult to start. Semiconductor saturable absorbers do not suffer from these starting
problems, and in addition, can be used to stabilize techniques such as Kerr lens modelocking and
soliton modelocking.
They are important for optical applications such as all-optical switching.
In a world in which devices are shrinking and increasing in speed, saturable absorbers are an
important component.
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
15
2.2 Saturable Absorber Theory
The history of saturable absorbers begins about the time the solid-state laser was invented. In
1965, Mocker and Collins used dye and color filter saturable absorbers to modelock a ruby laser
[47]. Twenty years ago, semiconductor nonlinearities were used to modelock a semiconductor
laser with a reduced recovery time, induced by aging [22]. Fifteen years ago, a multiple quantum
well structure was used to modelock a semiconductor diode laser [52].
The first solid state laser
modelocked with an intracavity saturable absorber was a color-center laser in 1989, producing
275 fs pulses [25]. Since then, saturable absorbers have been used to passively modelock many
solid-state lasers, including Nd:YLF, Nd:YAG, Nd:LSB, Nd: YVO 4 , Cr 4 +:YAG and Cr:fosterite
[29]. They have also been used to modelock fiber lasers, achieving repetition rates three times
higher than those achieved by APM [36]. The first fiber laser modelocked with a transmissive saturable absorber was reported in 1991 [64].
Later, a semiconductor saturable absorber deposited
on a distributed Bragg reflector was used to generate sub-picosecond pulses from a linear fiber
cavity [43] [44].
With the advent of bandgap engineering and innovations in materials processing, today, there is
much freedom in absorber design, and devices can be tailored for specific applications. Semiconductor saturable absorbers provide a compact bulk nonlinearity that does not require critical alignment, and is easily integrable into compact optical systems. As pulses become progressively
shorter, there is also growing interest in developing broad-band saturable absorbers.
Semiconductor saturable absorbers modelock or assist in modelocking lasers, because, like every
other modelocking technique, they introduce an intensity-dependent loss. States with higher peak
intensity (i.e. shorter pulses) are favored. Semiconductor saturable absorber mirrors are basically
nonlinear mirrors whose operation relies on carrier excitation. For absorbers used in a reflective
geometry, the basic design is a layered Bragg stack (DBR), on top of which is absorbing material,
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
16
either in the form of quantum wells or bulk sections. This absorbing material produces the nonlinear reflectivity response of the mirror. Light at a designed wavelength excites carriers from the
valence to the conduction band, and is absorbed. As the incident light becomes increasingly
intense, more and more carriers are excited. Eventually, all of the available states in the conduction band become full. At this point, the absorber is "bleached", and light is no longer absorbed
and instead, fully reflected from the DBR.
The basic design parameters of a general saturable absorber are: saturation intensity, Isat; saturation fluence, Fsat; modulation depth or saturable loss; and recovery time. These parameters determine the startup modelocking conditions, the resilience of the system against Q-switching and
other instabilities, and the shortest pulse width achievable.
A saturable absorber has both nonsaturable and saturable loss. Loss that has no power dependence is nonsaturable. In the ideal absorber, the non-saturable loss would be zero. Low-gain
lasers, in particular, cannot tolerate much non-saturable loss. The more non-saturable loss in a
system, the closer to threshold it becomes, which increases susceptibility to instabilities. In contrast to non-saturable loss, saturable loss is intensity-dependent loss. The modulation depth of an
absorber is the total amount of saturable loss of the absorber: the amount of loss that can be saturated by a high-intensity pulse. The modulation depth,
AR
~2col, of an absorber is related to the
pulse width produced as follows [29]:
n
TP
T
a
(1)
where n varies, depending on the type of modelocking used [20] [21] [31]. Thus, shorter pulses
can be achieved with higher saturable losses.
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
17
The saturation intensity of the absorber is important for initial pulse creation in laser systems.
Noise fluctuations in the laser system lead to pulse formation. In this regime, the continuouswave (cw) intensity of the light incident on the absorber determines the saturation behavior of the
absorber. The absorber should be barely saturated (if it was saturated, pulses would not be
favored) and act as a fast absorber, with a recovery time much faster than the noise fluctuations.
(When the laser modelocks, and the pulses become short relative to the absorber recovery time,
the absorber acts as a "slow" saturable absorber.) In the cw regime, the saturation intensity is
given as [32]:
Isat =
(2)
GATA
where ho is the photon energy, TA the absorption cross-section and TA, the absorber recovery
time. The absorption cross-section, defined to be the absorption probability per incident photon
flux, or the absorption coefficient per absorption center, is a material parameter [3]. From the
absorption cross-section, we can define an absorption coefficient of the material to be:
a = NDcA
(3)
where ND is the density of the absorber atoms, or density of states (semiconductors). In the cw
state, the absorption coefficient saturates as
(X = Ls +
0
(4)
1+ Isat
where a1ns is the nonsaturable loss of the absorber, ao is the saturable loss of the absorber, and Isat
is the cw saturation intensity of the absorber [33].
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
18
IdR/dII
0
I-a
Intensity on Absorber
FIGURE 2.3 Nonlinear reflectivity of a saturable absorber mirror due to absorption bleaching from
Reflectivity is denoted by R, and intensity by 1.
cw intensity.
The modelocking buildup time will be proportional to the slope of the increase in reflectivity vs.
the change in intensity at I ~ Icw~0, as shown in Figure 2.3. Under certain approximations [32],
we can write:
Tbuildup(X dR I I
;w,
(5)
where dR/dI represents the change in absorber reflectivity vs. cavity intensity. The larger the
slope of the saturable absorption, the smaller the intensity fluctuation required to produce a reflectivity change in the absorber itself. This means that as saturation intensity decreases, modelocking time decreases also.
However, if the laser is in a modelocked, or pulsed regime, the intensity is a function of time.
Once the pulses become significantly shorter than the absorber recovery time, the absorber
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
19
becomes a "slow" saturable absorber, assisting to shape one edge of the pulse. Until this condition is true, equations (2) - (5) hold. When the absorber becomes a "slow" absorber, it is the saturation fluence rather than the saturation intensity that is important. The saturation fluence is
written as [4]:
hv
Fsat = 'YA
(6)
The fluence incident on the absorber, F, is given as:
F(t) =
I(t)dt
(7)
The absorption for a slow saturable absorber, is then given as [33]:
X=
F
ns +
(8)
1+ -
Fsat
From the previous discussion, it is clear that the saturation intensity and fluence of an absorber are
important for the startup of modelocking. These parameters can also help prevent the growth of
instabilities. However, there is a fine balance between the shortest possible mode-locking buildup
time, and prevention of instabilities. If the saturation intensity becomes too small, Q-switching
takes over. Because the saturation intensity is inversely proportional to the absorber recovery
time, shorter lifetimes increase the saturation intensity and system resilience to Q-switching.
Q-switching results in the non-equilibrium buildup of a very intense pulse, that ends when the
gain is depleted. The process can repeat itself when the gain recovers. Q-switching mechanisms
can be produced via a shutter that blocks an end mirror cyclically or an intensity-dependent loss
with a large saturation energy [35]. A Q-switched envelope can modulate a modelocked pulse
train; this is called Q-switched modelocking. The pulses in this state are also quite intense, and
can damage saturable absorbers. The state is quite common in startup conditions of saturable
absorber modelocked lasers [19]. Sometimes, a small disturbance of the laser cavity can drive a
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
20
cw-modelocked laser into a Q-switched modelocked state. This state is undesirable for high repetition rate applications that require constant pulse energy.
cw - Q - switching
cw - running
"single"
mode
o
0
time
time
self-starting mode locking
Q - switched mode locking
cw - mode locking
multi
mode
0m
time
time
FIGURE 2.4 Different regimes of laser operation with an intracavity saturable absorber. Figure from [32].
To prevent cw Q-switching, the condition following must be satisfied [30] (fast saturable
absorber):
I< r
dIl
(9)
where r represents the number of times above threshold that the laser is pumped, TR is the cavity
round trip time, and t 2 is the upper state lifetime of the laser. Thus, factors in the suppression of
Q-switching include: high absorber saturation intensity (i.e. a small slope, dR/dI), laser operation
many times above threshold (large r), low repetition rates (large cavity round-trip times), and
small upper state lifetimes, t 2 [32]. The right side of the equation is shows the saturation of the
gain in the cavity per round trip. The left-hand side describes the loss reduction per round trip due
to the bleaching of the saturable absorber. If the gain cannot respond fast enough to the loss
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
21
reductions induced by the absorber, intensity in the cavity is no longer constant, and increases
with increasing bleaching of the absorber, leading to Q-switching in a stable or unstable form. If
the recovery time of the absorber is greater than or equal to the cavity round-trip time, pure cw Qswitching is favored over modelocked operation [32]. Thus, as the repetition rate of a system
increases, the tendency to Q-switch becomes stronger, unless absorber recovery time decreases
also.
If an absorber is employed whose recovery time is much less than the cavity round trip time, then
usually equations (5) and (9) are satisfied, allowing short pulse formation. However, in this case,
there is an additional stability requirement against Q-switched modelocking. In the following
derivations, it is assumed that the pulse width, rp is shorter than the absorber recovery time, tAThe saturation is now determined by the saturation fluence, Fsat' defined in equation (8). Now, the
loss reduction per round trip is due to the saturation of the absorber by pulses, not from the cw
intensity. Obviously, the absorber saturates much further in this limit. The condition for stability
against Q-switched modelocking becomes [32]:
dR F< r
dF
(10)
T2
This condition can be fulfilled by choosing F, the incident pulse fluence, to be much greater than
Fsat- This will also produce maximum modulation depth. Again, however, there is an upper
bound to F, given by the onset of multiple pulsing. At a fluence far beyond Fsat' the reflectivity is
no longer strongly affected by the pulse energy. Shorter pulses see a reduced averaged gain also,
since eventually, the gain-bandwidth of the laser becomes a limiting factor. Finally, at high
enough fluences, given these two factors, a multiple pulse state becomes favored [32].
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
22
LL
.
IdR/dFl
.
Multiple Pulsing
F pulse >> F,sa
.0
Fsat
Energy Fluence on Absorber
F
FIGURE 2.5 Nonlinear reflectivity change of a saturable absorber mirror due to absorption bleaching with short
pulses.
The saturation fluence and intensity of the absorber are generally material parameters. The incident spot size is usually adjusted to fulfill the conditions described for stability against Q-switching and multiple pulsing. In addition, proper placement of the absorbing layers at the peak or null
of the standing wave of the electric field in the absorber can produce variation of the saturation
intensity and energy.
The final parameter left to adjust is the absorber recovery time, the subject of this thesis. Because
semiconductor absorbers rely on carrier excitation, they exhibit a bitemporal response. The fast
response, on the order of several hundred femtoseconds to a few picoseconds, is produced by the
thermalization and carrier-carrier scattering, intraband processes. The slow response is due to
interband dynamics, the recombination of carriers. Depending on the materials system and
growth parameters, this can range from picoseconds to a few nanoseconds. This response time
will determine what the dominant modelocking mechanism in the laser system will be: slow saturable absorber modelocking, fast saturable absorber modelocking, soliton or dispersion-managed
soliton modelocking, or Kerr lens modelocking. Thus, the absorber response time will aid, or
fully determine (depending on the modelocking type), the pulse width, the system stability against
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
23
Q-switching and other instabilities, and the modelocking self-starting conditions. As intracavity
power increases, modelocking build up time decreases. It also decreases with increasing absorber
recovery times and decreasing saturation energy.
Pulse widths are partially or solely determined by the response time, depending on the type of
modelocking employed. By varying growth conditions and temperatures, or through post-growth
ion or proton bombardment, this recovery time can be tailored for the application. As data rates
continue to increase, higher repetition rate and shorter pulse systems are becoming important,
thus motivating an interest in absorber lifetime reduction.
2.3 Types of Saturable Absorbers
There are several types of saturable absorbers. Absorber structures used in a reflective geometry
are placed on a bottom mirror, and a Fabry-Perot cavity is formed between the top surface and the
bottom mirror. If the Fabry Perot is operated at antiresonance, an antiresonant saturable absorber
is formed; if the Fabry-Perot is operated at resonance, a resonant saturable absorber results.
If the field intensity inside the absorber is less than the incident intensity, the device is an antiresonant device. In this case, the resulting devices are broad-band and have minimal group velocity
dispersion [32]. An uncoated device, with a Fresnel reflection from its top surface, is one special
case of an antiresonant device. For low-gain laser systems, such as Cr 4 +:YAG, the small modulation depth associated with such absorbers can be desirable. Another example of an antiresonant
structure is an anti-reflection coated device, such that the field intensity inside the absorber is
equal the to the incident intensity. In the other limit, the top surface of an absorber could be
coated with a high reflector, greater than 95% [32] [58].
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
24
At antiresonance, the intensity inside the device is decreased by a factor , compared to the incidenc intensity [33].
1-Rt
(1 + JRtRbexp(-2i
(11)
2
0 d))
where ao is the absorption coefficient of the absorber, d is the absorber thickness, and Rt and Rb
are the top and bottom mirror reflectivities. The saturation fluence decreases by a factor 1/p, and
the saturation intensity increases by this same factor.
For higher modulation depths and lower saturation fluences, it is possible to design resonant
devices. In these devices, the field inside the structure is enhanced, compared to the incident field.
This can be accomplished with a resonant coating on the top of the device. The drawback of resonant devices is that they are not as broadband; however, for applications requiring high modulation depths, they can be useful [59].
The type of coatings added to the top of the saturable absorber structure determine if the absorber
is resonant or antiresonant. This adds yet another parameter of freedom to the design of semiconductor saturable absorbers. By tailoring the field intensities inside the device, the saturation fluence, intensity, and modulation depth can be adjusted.
2.4 Absorber Lifetime Reduction Techniques
Semiconductor saturable absorbers consist of a bitemporal response: an ultrafast response, due to
the intraband carrier dynamics, and a slow response, dominated by the interband carrier dynamics.
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
25
The slow response time is determined by the carrier recombination time, which in pure materials
is on the order of several nanoseconds. This recovery time makes design of ultrafast optical
devices difficult. Reduction of this response time can be accomplished by introducing defect
states into the material. The lifetime can also be reduced by tightly focusing onto the sample,
causing carriers to diffuse out of the interaction region at a speed faster than the device recovery
time [53]. However, a small spot size causes the absorber to saturate at low powers.
Efforts to reduce recovery times via defect introduction have a long history. The first semiconductor diode laser to be modelocked with semiconductor nonlinearities had a reduced recovery
time induced by aging [22]. Shortly afterwards, proton bombardment was used for lifetime reduction [62]. The first solid state laser to be modelocked, in 1989, was modelocked with a protonbombarded saturable absorber [25]. Defect states can be introduced in material through strain,
low-temperature molecular beam-epitaxy (LT-MBE) growth, or through post-growth ion or neutron bombardment. These defect states serve as recombination sites, thus decreasing the recovery
time of devices. For lifetime reduction, LT-MBE growth and ion bombardment are the two methods. Both techniques have been used in demonstrations of passive modelocking [32] [41]. The
techniques both introduce defects, smearing the bandedge by the introduction of near-bandedge
states. High defect densities can cause changes in saturation fluence [33]. The saturation intensity of the absorbers is affected also, as it depends on absorber recovery time. From the previous
section, recall that shorter recovery times can stabilize against Q-switching, but are not helpful for
self-starting. Disadvantages of the lifetime reduction techniques include increased non-saturable
loss and reduced modulation depth. There is still discussion as to the origin of the increased nonsaturable loss: introduction of mid-gap traps is one possible explanation [41]. Carriers may be
captured by these trap states, then re-excited, and finally recombine. If the trap states created have
transitions that are difficult to saturate, they contribute to the non-saturable loss, and reduced
modulation depth. The increased nonsaturable loss and reduced nonlinearity of the devices can be
mitigated with a post-process anneal (for both LT-MBE and ion bombardment). In the case of LTMBE, dopants can also alleviate these detrimental side effects. The sharpness of the bandedge
can be restored with annealing and or the addition of dopants [54]. The physics behind the benefits of annealing are not well understood [9]; however, it may allow the defects to find the lowest
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
26
energy stable states. Some of the trap states are probably converted into recombination centers
[54], and annealing may remove some of the shallow states, improving stability [51].
The effects of low-temperature (LT) epitaxial growth on the recovery times of semiconductor saturable absorbers has been studied extensively. In conventional MBE systems, the substrate is
heated to -500 or -600 0 C respectively for GaAs and InP, producing high optical and electrical
quality material [16]. Low-temperature growth causes the material quality to decrease, consequently reducing the lifetime. This method has been studied with various materials systems
including: GaAs[45] [50], InGaAs/ InAlAs [27] [57], InGaAs/InP [16], and AlAs [37].
All of
these materials contain arsenic, and low-temperature growth causes As-rich conditions, inducing
excess-As caused defects in the form of precipitates, interstitials, microclusters, and vacancies. In
low-temperature GaAs and InGasAs, beryllium doping, as well as annealing, has been shown to
mitigate non-saturable losses and decreases in modulation depth resulting from LT growth. It is
thought that beryllium doping changes the type of trap created. The traps created with this dopant
have a transition that is easier to saturate than that of the traps resulting from non-beryllium doped
LT-GaAs or LT-InGaAs growth [17] [27]. Subpicosecond relaxation times have been observed in
LT GaAs, LT InAlAs/InP and lifetimes have been reduced to 2.5 ps for LT InGaAs/InP [16].
Ion implantation is a commonly used technique in semiconductor processing. Unlike LT epitaxial
growth, it is a relatively simple post-growth step and is a more mature processing procedure [14].
In semiconductor saturable absorbers, a variety of ions have been used for lifetime reduction: protons [51], arsenic [42], oxygen [42], gold [5], helium [11], and nickel [6], just to name a few.
There is a nonlinear relationship between damage profile and ion dose. Penetration depth
depends on the material, crystalline orientation, and weight of the ion used for bombardment [46].
As dose is increased and annealing times decreased, recombination lifetimes decrease. As in LTgrowth (in which temperature, materials, dopants and annealing must be optimized), the right recipe of ion type, dosage, energy and annealing time must be found. Some groups claim that
heavier ions create more stable defect centers [6], but heavy ions at high energies create undesir-
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
27
able amorphous layers [42]. When these layers form, lifetime reduction saturates and lifetimes
may even begin to increase with higher bombardment levels [8] [38]. Lighter ions avoid this
problem, and so, higher energies can be used. This allows deeper penetration depth without sacrificing device nonlinearity [42]. Lifetimes of 500-600 fs have been achieved in proton-bombarded
GaAs [8] and lifetimes of 200 fs have been achieved in As implanted GaAs [14]. Lifetimes of
-100 fs have been achieved in proton-bombarded InP [39].
2.5 Growth of Semiconductor Saturable Absorbers
The semiconductor saturable absorbers studied in this thesis were grown by gas-source molecular
beam epitaxy (GSMBE) by Elisabeth Marley Koontz, in Professor Leslie Kolodziejski's lab at
MIT. Initially, a distributed Bragg reflector (DBR) was grown on a semiconductor substrate,
either by GSMBE or purchased from an metal-organic vapor deposition (MOCVD) facility. Some
of the mirrors grown by GSMBE were tunable from 1.42 - 1.58 gm: this was a result of uneven
sample heating, causing variation in deposition rate and layer thickness. On top of this mirror,
absorber and spacer layers were grown by GSMBE. In some cases, additional InP was overgrown by GSMBE. A portion of the absorbers were sent out for proton bombardment. They were
bombarded with different energies and doses of protons depending on the thickness of the specific
device. Finally, the anti-reflection coatings were deposited by electron-beam evaporation at Lincoln Lab with the help of Chris Cook and Peter O'Brian.
The saturation parameters of the devices were designed by calculating an electric field profile
using matrix and interface techniques [3] [58]. Figure 2.6 illustrates this calculation for 1.54 pim.
As the field distribution will change as a function of wavelength, it important to design for a specific center wavelength. On the left axis, the refractive index of the structures as function of position is shown. The square of the electric field as a function of position is shown on the right axis.
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
28
A
22 pairs
-12
InGaAs
-8
-
U
4
0
5.0
5.5
6.0
6.5
7.0
z (Am)
FIGURE 2.6 Schematic of an anti-reflection coated structure with six InGaAs quantum wells at the peak of the
electric field. The refractive index and magnitude squared of the electric field (Q=1.54rim) are plotted as a
function of distance from the GaAs substrate-DBR interface.
The structure consists of 22 layer pairs of GaAs/AlAs, grown on a GaAs substrate. This produces
a high reflector at 1.55 gm. On top of this, is a V/2 layer of InP. Six InGaAs quantum wells, 100
A wide, are placed at the peak of the electric field in the InP. They have a bandedge of ~1.58 pm.
On top of the InP, is an anti-reflection coating of a V/4 layer of A12 0 3 . This particular device,
with a fairly large modulation depth of 6.2% and a saturable loss of 5.2% [59], was designed for
implementation in high-gain fiber laser systems. {For lower gain systems, a smaller modulation
depth can be achieved by eliminating the anti-reflection coating. Then, a -30% Fresnel reflection
from its top surface exists, due to the air-semiconductor index mismatch [58].1 The large modulation depth and low saturation fluence of this device was achieved by using a comparatively large
number of quantum wells, placed at the peak of the electric field, with a long wavelength bandedge.
The square of the electric field is shown in Figure 2.6, not the intensity. To convert to intensity,
the square of the electric field would be multiplied by a term proportional to the refractive index
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
29
of the materials. The difference in refractive index between the air (n- 1) and the semiconductor
(n-3) may lead to confusion. The square of the electric field is significantly smaller in the structure. However, the intensity inside the structure is comparable to the incident intensity, as the
index of refraction is much higher in the device.
Figure 2.7 illustrates another structure studied in this thesis. It is similar to the previous structure,
except, that the quantum wells are replaced with 1000 A of quasi-bulk InGaAs. For more absorption, thicker layers of bulk material could be used. Depending on the thickness of these bulk layers, quantum confinement effects may or may not be present. The structure shown below has very
similar characteristics to the previous structure with six quantum wells discussed above.
4
22 pairs
InP
<0
X
0 3
-12
InGaAs
I-8
%6W
>
20L
-0
n
-
5.0
5.5
6.0
6.5
7.0
z (Pm)
FIGURE 2.7 Schematic of an anti-reflection coated semiconductor saturable absorber containing a quasi-bulk
layer region. The refractive index and magnitude squared of the electric field (X = 1.54 gm) are plotted as a
function of distance from the GaAs substrate-DBR interface. Figure from [58].
Both of the structures shown so far have a high modulation depth, as the absorption regions are
placed at the peak of the electric field. Shown below, in Figure 2.8, is an absorber also studied in
this thesis, but with significantly different properties. There are only two quantum wells, with a
bandedge at 1.45 gm, and a well width of -45 A. They are placed at the null of the field, so con-
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
30
sequently, the modulation depth is much smaller than the previous structures (-0.5% for a nonanti-reflection coated structure). The structure studied was placed on a tunable mirror from 1.42
to 1.58 gm. In contrast to the two previous structures, this structure was designed to modelock a
low-gain solid-state Cr 4 +:YAG system, which cannot tolerate much loss.
4
22 pairs
InGaAs
12
a
03
N
2
4-
'I
0
7
I
5.0
-
I
5.5
I
6.0
z (pm)
1
6.5
7.0
0
FIGURE 2.8 Schematic of an anti-reflection coated structure containing two quantum wells. The refractive
index and magnitude squared of the electric field (X = 1.54 gm) are plotted as a function of distance from the
GaAs substrate-DBR interface. Figure from [58].
One additional device is shown in Figure 2.9. The absorbing region is removed, leaving a structure of indium phosphide, on a DBR, with an anti-reflection coating. Although this is not a saturable absorber, it was used for studying the contribution of the InP to the time-dependent
reflectivity. In addition, its induced absorption properties make it useful as an intensity limiter in
lasers, favoring pulses of equal amplitude in harmonically modelocked systems, and stabilizing
against Q-switched cw modelocking [60] [61].
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
31
4
-1"
22 pairs
X
1.2
3-
0.8
InP_
N
0. 4
0.0
II
5.0
I
5.5
I
11.0
z (Pm)
I
i
11.5
12.0
FIGURE 2.9 Schematic of intensity-limiting device consisting of a DBR, a layer of InP, and an anti-reflection
coating at 1.54 tm. Devices with different InP layer thicknesses were investigated. Figure from [58].
The design of these structures allows many parameters to be adjusted. The width of the quantum
wells, and the bandwidth of the DBR allow the center wavelength of the device to be varied. All
structures studied in this thesis were grown on a DBR with reflectivity of greater than 99.9%, over
a bandwidth of 100 nm, either centered at 1.55 gm or 1.58 gm [35]. The number and placement
of the quantum wells or quasi-bulk sections allow the saturation fluence and modulation depth of
the devices to be adjusted. Post-growth optical coatings allow adjustment of the saturation fluence and modulation depth also.
A single quarter-wave layer of A12 0 3 was chosen as an anti-reflection coating for the protonbombarded absorbers to ensure maximum experimental flexibility. Since the coating is amorphous, absorbers could be annealed or re-bombarded if necessary, without coating damage.
Shown in Figure 2.10, is a plot of reflectivity versus wavelength for the coating deposited on a
GaAs wafer.
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
32
0.08
0.06
0.04
; 0.02
0.
/opdab r 4udieVrc I ngs/00a prl arcoaiaaI
0.001300
1400
1500
1600
----1
1700
Wavelength (nm)
FIGURE 2.10 Reflectivity vs. wavelength of a single \/4 A12 0 3 anti-reflection coating deposited on GaAs. The
coating run was done at Lincoln Labs, with an electron-beam evaporation system.
One novel feature of the absorber design is the GaAs-InP interface, as a large lattice mismatch
exists between the two materials. For pure material, it is important to grow lattice matched structures. If InP-based material were used for the mirror, a lattice matched system could be designed.
However, a DBR based on this materials system would need twice as many layer pairs to achieve
the reflectivity of a GaAs/AlAs structure [35]. This is because the index contrast in the InP materials system is not as great as the GaAs/AlAs system. Currently, the bandwidth of these devices is
somewhat limited by the mirror. The bandwidth could be improved by using a silver mirror, and
wafer bonding epitaxial grown layers on top. However, the beauty of the structures studied is that
they are monolithically integrable, important for commercial applications.
Defects are created at the GaAs-InP interface because of the lattice mismatch. Pure material has a
recovery time of several nanoseconds. The defects are misfit dislocations and act as non-radiative
recombination centers [35]. They shorten the recombination time (a beneficial side-effect) and
smear the bandedge of the material, thus broadening the photoluminescence. Since it is possible
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
33
to operate within the absorption edge of the InGaAs (and below the bandedge for InP), these
defects are not a drawback [58]. The recovery time varies with the positioning of the quantum
wells from the interface. The further the quantum wells from the interface, the longer the recovery time. This is probably an indication of the material strain, and will be discussed in more detail
in Chapter 4, Section 4.1.
2.6 Optical Characterization of Saturable Absorbers
It is important to know the precise relationship between MBE growth parameters, and the optical
properties of the structures. The position and thickness of the absorber layers must be controlled
properly. To obtain the designed absorption edge, room-temperature photoluminescence was used
to characterize the structures. Transmission as a function of wavelength was also used to determine the sharpness and depth of the absorption edge. From these measurements, the bandedge is
smeared by approximately -60 nm. The step of the absorption is also fairly constant across a
range of -150 nm [58]. Thus, since the bandwidth of the DBR is approximately 100 nm, with
proper design, the absorption can be constant across the entire mirror bandwidth.
In Figure 2.11, a typical photoluminescence measurement and DBR reflectivity are shown. It is
important to have the proper relationship between absorption spectrum and mirror bandwidth in a
successful saturable absorber design. Figure 2.11 shows that the bandwidth of the DBR is -100
nm, with a reflectivity greater than 99.9%, centered at 1.55 gm. This is not the ideal sample, since
in order to use the full mirror bandwidth, the absorption bandedge should coincide with the center
wavelength of the mirror.
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
34
1.OBR
PL
U
0)
'I-
C)
0.60.4OA-
0
02/Masters The ;is/Rna figresP&dbr
0.0 "
1.3
1. __________
1.4
1.5
1.6
di
1.7
1.8
Wavelength (am)
FIGURE 2.11 Transmission as a function of wavelength for a typical GaAs/AlAs DBR. For comparison, the
normalized photoluminescence intensity of a quantum well structure deposited on GaAs is plotted. Figure from
[58].
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
35
Chapter 3: Experimental Design
The experimental design and implementation of the pump-probe system used to study the recovery time of the absorbers is described. Resolution limits are discussed, as well as the motivation
for the particular experimental setup.
3.1 Pump-Probe Theory
The recovery time of the semiconductor saturable absorbers was studied through femtosecond
time-resolved spectroscopy, using a measurement technique called pump-probe. In this technique, a powerful light pulse, the "pump" interacts with a sample at tO. The sample gradually
recovers its equilibrium state at some later time, 1 l. The relaxation to equilibrium can be quantified with a second weaker pulse, a "probe" pulse. The delayed probe pulse measures the change
in reflectivity, AR, that the pump produces. By repeating this measurement at different delays,
one can map out AR as a function of time [7]. This time dependent reflectivity change is determined by the time dependent carrier dynamics, which are important for ultrafast pulse generation.
The temporal resolution of this technique is obviously limited by the duration of pump and probe
pulses. Thus, there is a balance between temporal and spectral resolution to be sought, as the
pump or probe must select out a specific spectral feature. Because the bandedge is smeared by
-60 nm in the saturable absorbers studied in this thesis, this is not of great concern.
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
36
The change in reflectivity, AR, is measured as a function of delay, td, of the probe pulse from to,
the time at which the pump pulse hits the sample. Since electronic detection is slow, the signal
detected will have the following form [7].
AR(td) =f
Iprobe(td- t)Ipump(t)f(t)dt
where f(t) is the time dependent response of the sample;
'probe,
(12)
the intensity of the probe pulse;
and IpMP, the intensity of the pump pulse. The time dependent sample response, f(t), is given by
the following equation [15]:
f(t) = ais,(t)+ aP 1 (t) + a 2 P 2 (t) +
...
(13)
where the amplitude factors are given by aj, and the instantaneous response is represent by the
Dirac 8 function. Successive excited-state populations are represented with Pi(t), with relaxation
times ti. Both electron and hole densities contribute to Pi(t) when they have similar relaxation
times [15]. So, in order to extract the sample response, one must deconvolve its response from the
cross-correlation. If the time scales of the signal response are much larger than the pulse crosscorrelation width, the cross-correlation can be approximated as a 6-function, and then AR is
directly proportional to the response of the sample.
3.2 Laser Sources
The pump-probe in this thesis was done in the 1.5 gm wavelength range, using a commercial
Optical Parametric Oscillator (OPO). Transform limited pulses of approximately 150 fs were
used. The OPO is synchronously pumped by a 2 W pulsed Titanium:Sapphire laser, which in turn
is pumped by a 20 W argon-ion laser. The argon laser is continuous-wave with multiple lines
between 460-514 nm, with the strongest two at 488 nm and 514 nm. The Titanium:Sapphire laser
produces greater than 2 W of output power when pumped with ~14 W of Argon. It is a pulsed
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
37
laser, producing 100 fs transform-limited pulses, through Kerr lens modelocking. A typical spectra is shown in Figure 3.1.
I
I
-40-50-60C
cc~
-70-8078 0
/poab/e
Ldietpmppibdsbs/-23,00)
800
ek23tiw:.at
820
840
Wavelength (nm)
FIGURE 3.1 Spectra of Spectra-Physics Titanium:Sapphire in ~100 fs pulsed operation.
The OPO produces ~110-150 fs pulses, with signal tunable from 1.4 to 1.6 pm, and idler from 1.7
to 2.0 gm. Time resolution of pump-probe is limited by the length of the pulses. The OPO is a
good source for pump-probe, but suffers from stability problems, and an inefficient parametric
conversion process. Because the argon laser is noisy, the system requires a three hour warm-up
time before work can be begun. Despite the drawbacks listed above for the OPO, it is a flexible
and versatile source, suitable for the pump-probe experiments in this thesis. Figures 3.2 and 3.3
show spectra and autocorrelation from the OPO, respectively.
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
38
I
-50-
I
I
I
-60 -
IM
cc
-801/pd ab/94uliet/pipp-obacis
-901480
1500
1520
1540
1560
Wavelength (nm)
d
,3DaXb23opdat
1580
1600
FIGURE 3.2 Typical spectra of Spectra-Physics Optical Parametric Oscillator.
1
-AC
-
0.1
0)
-
Gaussian (124 fs)
-Sech (108 fs)
0.01
4.'
0.
E
1E-3obm/bei2t3e
/opd a/euuIfe
ar4pnpAat
1E-4
-400
-200
0
200
400
Time Delay (fs)
FIGURE 3.3 Autocorrelation of Optical Parametric Oscillator pulses. The pulse shape is in between a Gaussian
and a Sech.
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
39
There are a lack of good ultrafast tunable solid-state high power laser sources at 1.5 gm, the telecommunications wavelength, as alternatives to the OPO systems. Color-center lasers are high
power tunable sources at 1.5 gm, but the crystals are sensitive to room light and must be cryogenically cooled. Recently, there has been much interest in Cr 4 +:YAG, a room temperature alternative to color-center lasers. Unfortunately, it is a low gain system, and as it is a relatively new
material, there are still issues with crystal reliability. However, it is a broadly tunable solid-state
source, with potential for femtosecond operation, with proper dispersion compensation. With an
ultrafast Cr 4 +:YAG system, superior temporal resolution to the current OPO system could be
achieved.
3.3 Experiment Setup
To measure the recovery time of the semiconductor saturable absorbers, a time-resolved pumpprobe technique was used. The experimental setup is shown in Figure 3.4.
Delay Stage
/
1
17-1
I
Lpr
I
Chopper
C.
E
PBS
Detector
Y2
L.
/
150 fs OPO
Isolator
pulse
AOM
n
PBS
r;;zci
Probe
- U
V2
IL
Pump
r1K
-.
Saturable
Absorber
Delay r
FIGURE 3.4 Schematic of pump-probe setup.
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
40
The 150 fs pulses, at 82 MHz, from the OPO were split into two paths, the pump and the probe, by
an 70/30 beam splitter. They are cross-polarized with several polarizing beam splitters (PBS's).
The cross-polarization helps to eliminate interference effects at zero delay. In addition, the frequency of the probe pulse is shifted by 210 MHz with a TeO 2 AOM (20% diffraction efficiency),
to suppress fringing artifacts further. There is a variable time delay (up to 400 ps) between pump
and probe. The setup is optimized for fluence, thus pump and probe are kept collinear, in order to
strike the sample at normal incidence. Since these structures are used in laser cavities at normal
incidence, the collinear setup also allows accurate reproduction of conditions inside laser cavities.
Pump and probe travel through an aspheric lens, to the sample. They are reflected off the sample,
and then off a beam splitter to a slow (MHz) Germanium photodetector, consisting of a photodiode and a simple amplifier. The signal on the photodetector is sent to a lockin, and then to a
computer. The pump is rejected through a PBS before the detector. The pump is chopped, thus
only probe photons that have nonlinearly interacted with the pump are detected.
An isolator was placed immediately after the OPO to eliminate back reflections. The PBS's were
used in transmission whenever possible, to ensure maximum rejection of -30 dB. Several PBS's
were used in pump and probe paths to ensure pure polarization. Half-wave plates in each path
before the PBS's allowed power adjustments: however, for attenuating factors greater than 10,
external glass attenuators were used to preserve the polarization states.
Measurements were taken with a pump to probe power ratio of 10:1. Pump and probe spot sizes
were in a ratio of approximately 1:2 before the sample focusing lens, ensuring a measurement
with a smaller probe spot than pump on the sample. Knife edge scanning measurements, before
the focusing lens, were used to determine spot sizes impinging upon the focusing lens. Typical
spot sizes (radii) for probe and pump, before the sample focusing lens, were 1 mm and 0.5 mm,
respectively. Two different aspheric focusing lens were used: 6.24 mm and 11 mm. Spot sizes
incident on the sample were calculated, using Gaussian beam focusing approximations. These
calculations have an error of about +/- 20%, due to the short focal length of the lenses used. The
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
41
spot sizes (radii) incident on the sample were calculated to be approximately 10 pm for the 11 mm
lens, and 6.12 prm for the 6.24 mm lens. Fluences incident on the sample ranged from 5 to 800 J/
cm 2 . (Fluences can be calculated from incident power as follows. F =
',
where R is the repe-
Rnw
tition rate; P, the incident power; and o, the spot size).
The signal resolution limit of the system is 10-3 and the time resolution limit, -300 fs, determined
by the width of a typical cross-correlation (shown in Figure 3.5). The signal sensitivity could be
improved by implementing balanced detection, a simple modification. However, for the structures studied, a signal resolution limit of 10-3 was adequate. If the setup were to be optimized for
maximum signal resolution (10-6), a non-collinear, electronically double chopped, cross-polarized
pump-probe system could be designed. However, this setup would not be suitable for the high
fluence measurements performed in this thesis.
0.16-
r\ --
0.12-
Cro ss-correlation
-
Se ch (134 fs)
Ga uss (155 fs)
uC 0.08-
0.040.00 -400
/eAbpola bjuil et/p Lnppro besbrs/Web2 2ac.da
-200
0
200
400
Time Delay (fs)
FIGURE 3.5 Cross-correlation of pump and probe pulses used for the experiment. The cross-correlation is
longer than the autocorrelation because the probe pulse is slightly broadened by the AOM.
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
42
Chapter 4: Experimental Results
The results of pump-probe on proton-bombarded semiconductor saturable absorbers is presented
in this chapter. The proton-bombarded absorbers are compared with identical structures that are
not bombarded. Both anti-reflection and non-anti-reflection coated structures are studied. The
results of an annealing study, mitigating the detrimental side-effects of bombardment, are discussed. Results from tests in a 17 MHz erbium fiber laser are presented.
4.1 Pump-Probe of Non-Proton-Bombarded Structures
This thesis focuses on the structure shown in Figure 2.6, with six quantum wells placed at the
peak of the electric field, with a bandedge of 1.58 Rm. Shown in Figure 4.1, is a single beam measurement of the reflectivity of an anti-reflection coated device as a function of fluence. Initially,
the reflectivity increases as a function of fluence, illustrating the saturable absorber's function: an
intensity-dependent loss element, favoring higher peak intensities. The reflectivity increases as a
function of fluence, initially, because of spectral hole burning, or the state filling effect. The
reflectivity starts to roll over at high fluence due to the effects of induced absorption, both twophoton absorption and free-carrier absorption. Pump-probe traces were taken, as a function of
fluence, to study the carrier dynamics of this specific saturable absorber. Each pump-probe trace
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
43
corresponds to a specific place on this curve. This curve can be extracted from pump-probe traces
at zero delay, but a single beam measurement has less potential for experimental error.
0.940.93
0.92
0.91%0.90
0.89e 0.88E 0.870.86
0.85
rsb
.
0 .8 4
0.1
. .- .
1
rvcte, &wArlr
. .1 . ....1 if
10
.I
I. . . -.
100
Energy Density (gJ/cm 2)
FIGURE 4.1 Saturation measurement for an anti-reflection coated structure with 6 quantum wells. The
measurement was done at 1.54 gm, using a variety of lenses to achieve the large dynamic range. The saturation
2
fluence was 3.7 pJ/cm . Figure from [58].
Following are a series of pump-probe traces on this structure taken at low to moderate fluences; i.
e. at or below the saturation fluence of the absorber (before the reflectivity roll-off). Pump-probe
measurements were initially performed extensively at 1.54 gm, the telecommunications wavelength. Data is plotted in terms of change of reflectivity. Measurements are calibrated by taking a
ratio of the raw data to the signal produced by the probe itself, with the pump blocked. A 10-20%
background produced by leaked pump light was also subtracted, as it scales linearly with pump
power.
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
44
0.0300.025
0.0200.015
0.0100.005 -
Increasing fluence
0.000 -
/opolabM4uiesunpprcbebrsA,&jan itdt e
*
I
0
4000
I
I
12000
16000
I
8000
I
20000
Time Delay (fs)
FIGURE 4.2. Pump-probe of a non-proton-bombarded non-anti-reflection coated structure at 1.54 gm. The
curves are taken with increasing fluence.
0.0300.0250.0200.0150.010 -
0.005
Increasing filuence
0.000
/cpMlt/ajuIe~purnpptoe/tbr&oqai ltt/de
0
100000
50000
150000
200000
Time Delay (fs)
FIGURE 4.3. A zoomed-out version of Figure 4.2.
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
45
The response of the absorber consists of two components: an ultrafast component due to intraband dynamics and a long component due to interband dynamics. The ultrafast components in the
data were fitted using a convolution fitting algorithm. The slow components were fitted with single exponentials as the slow components are much longer than the pulse cross-correlation, allowing approximation of the pulse as a Dirac 8 function.
The ultrafast dynamics can be seen in Figure 4.2. Initially, one sees a strong bleaching signal due
to spectral hole burning. The number of carriers in the conduction band is increased within a limited energy range, and the number of holes in the valence band, reduced by the excitation, creating
a spectral "hole" in the absorption spectrum. The carriers thermalize forming Fermi distributions,
via carrier-carrier scattering, filling the spectral hole formed. This occurs in less than 100 fs,
beyond the experimental time resolution [13]. (A short pulse source at 1.5 jim like Cr 4 +:YAG
could explore this regime). Excess energy is disspated via carrier-phonon scattering, cooling the
carrier distribution to the lattice temperature on the order of -1 ps [26]. The slow time constant
following, seen clearly in Figure 4.3, is approximately 40 ps. This slow time constant shows no
power dependence and is due to carrier recombination.
As discussed in Section 2.5, this recom-
bination time is shorter than the one ns associated with lattice-matched layers deposited on InP
[1]. This is due to the presence of defects in the quantum wells, caused by the large lattice mismatch at the GaAs DBR-InP interface.
The device recovery time is affected by the distance of the quantum wells absorber sections from
the GaAs DBR-InP interface. In the structure studied above (see Figure 2.6), the six quantum
wells are placed at the peak of the electric field and are -110 nm from the GaAs DBR-InP interface. The recombination time of a structure, similar to that shown in Figure 2.8, with two quantum wells placed -220 nm from the GaAs-InP interface was measured. (The only difference
between this structure and the one shown in Figure 2.8 is the width of the quantum wells: 66 A
rather than 45 A and the lack of an anti-reflection coating). This structure was found to have a
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
46
time constant of 100 ps, almost double that of the six-quantum well structure. A comparison is
shown in Figure 4.4.
1.00.8-
-
0.6-
4)
-
Structure with 6 QW's 110 nm fr. interface
1.54 [tm
Structure with 2 QW's 220 nm fr. interface
1.47
m
0.4-
N
E
0.2-
0
Z
0.0- I
0
/opo
50000
abWe/ju fiet/pump probe /sbrs/00 apr1f. dat
' I
100000
'
I
I
'
200100
150000
Time Delay (fs)
FIGURE 4.4 Comparison of recovery times of two absorbers with absorbing regions at different places from the
GaAs-DBR/InP interface. The pump-probe has been performed at different wavelengths on the structures because
the bandedge of each structure was different: -1.46 gm and -1.54 gm respectively. Normalized data has been
plotted, because to the difference in number of the quantum wells and their placement at either the peak or null of
the electric field, the modulation depths of each structure were quite different.
Returning to the discussion of the six quantum well structure (Figure 2.6), pump-probe data at
higher fluences was taken, on both non-anti-reflection coated and anti-reflection coated structures. (An anti-reflection coating lowers the saturation fluence of the structures because more of
the incident light is coupled into the devices.) Data is shown in Figures 4.5, 4.7, 4.8 and 4.9. At
higher fluences, the effects of induced absorption (negative signals) become noticeable.
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
47
0.030-
Incident pump/probe powers
-7.5/0.75
mW
0.025-
40/4
mW
----75/7.5
mW
0.0200.015 0.010-
"
--
...
0.005
0 .00 0 -/oke4ju
0
lIe/puprobesbrs/00ar23154/gh.dat
4000
8000
12000
16000
20000
Time Delay (fs)
FIGURE 4.5 Higher fluence data taken at 1.54 im on a non-anti-reflection non-bombarded coated six quantum
well structure using a 11 mm focusing lens.
As the fluence is increased, the instantaneous spectral hole-burning/bleaching peak begins to
decrease. This is due to induced absorption. With increasing fluence, two-photon absorption
(TPA) and TPA-induced free-carrier absorption become apparent [40] [61]. At high fluences (the
75/7.5 mW curve in Figure 4.5), the bleaching signal actually increases with increasing delays, for
a few picoseconds. This is due to carrier cooling. As the highly excited carriers cool, they fill up
the lower lying states, causing the quasi-Fermi level to rise, and resulting in increased bleaching
[26]. Two-photon-induced free-carrier absorption depends quadratically on the pump fluence,
whereas, two-photon absorption (absorption of a probe photon induced by the presence of the
pump) depends only linearly on pump fluence. Thus, at lower fluences, the effect of two-photon
absorption is dominant, but at higher fluences, TPA-induced free-carrier absorption takes over.
This relationship is shown in Figure 4.6.
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
48
a
I
o 4.5x10
3
4.0x10
3
I
-
I
.
U-
1.56gm
gm
1.56
;&=
=
X
.4
.4
3.5x10-3
- - -- -linear fit
quadratic fit
,'
3.0x10-3
-
G) 2.5x10-
TPA .
(D 2.0x10-3
o
1.5x10-'0) 1.0x10--3
C
-m
FCA
,-'
.40
,.'
5.0x10'0.00.0
40.0
20.0
100.0
80.0
60.0
120.0
Pump fluence (gJ/cm 2)
FIGURE 4.6 The relationship between two-photon absorption and free-carrier absorption. This data was
extracted from measurements on the structure in Figure 2.8. Figure is from [40].
0.04ONO-*
0.02-
illao
0.00-
..
WM&
flo
am
-
-
-a
-0.02Pump/Probe Incident Po wers
-0.04-0.06-
-
-0.08-
--
25/2.5 mW
mW
-40/4
75/7.5 mW
/o'ae/juleVpumpprobesb rsOO apr25 154abk/de. dat c11aens 1.54pm
OOapr251 s4sbrvam bhf
-0.100
4000
8000
12000
16000
20000
Time Delay (fs)
FIGURE 4.7 Pump-probe traces for an anti-reflection coated sample, using a 6.24 mm focusing lens at 1.54 [im.
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
49
In Figure 4.7, data from an anti-reflection coated structure is shown. The combination of a tighter
focussing lens (6.24 mm vs. 11 mm) and the anti-reflection coating enables higher fluence characterization. At higher fluence, the effects due to induced absorption increase. The one picosecond
carrier cooling time constant is still apparent. However, in this figure, the appearance of a new
component of the absorber response is noted. From the 40/4 and 75/7.5 mW traces, an induced
absorption signal with a time constant of -4-10 ps is apparent. Both the InGaAs quantum wells
and the InP contribute to this signal; the weighting of each component is currently unclear. This
0.64
component could be due to carriers excited out of the wells [63] (the InGaAs-InP barrier is
eV: so two-photon absorption in the quantum wells could excite carriers out of the well) or carriers excited high into the bands and scattered into satellite valleys [2] [10]. It is unlikely that this
to
component is caused by a phonon bottleneck, in which there are no longer phonons available
assist carrier cooling. The time constant does not appear to become significantly longer with
increasing fluence, as one would expect in the case of a phonon bottleneck [49]. The beginnings
of this induced absorption component can be seen in the long scans at moderate fluence. Shown
in Figure 4.8, is data for non-anti-reflection coated structures, a zoomed-out look at the data in
Figure 4.5.
0.0300.0251
0.020 -
I
C
Incident pump/probe powers
7.5/0.75 mW
-40/4
mW
mw
75/7.5
-
0.0150.010-
0.005
0
30000
60000
90000
120000 150000
Time Delay (fs)
Fabry-Perot
FIGURE 4.8 A zoomed-out version of Figure 4.5. The hiccup in the 7.5/0.75 mW trace is due to a
from an external glass attenuator.
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
50
When initial analysis was done, it appeared that the long time constant was increasing with fluence, reaching 60 ps, from 40 ps. After more careful fits were done, it became clear that this was
not correct. The data, starting with the 40/4 and 75/7.5 mW traces, is better fitted with a dual
exponential, one that is growing (representing the influence of the 4-10 ps induced absorption signal), and one decaying, with the familiar 40 ps recovery time. Figure 4.9 shows long scans of the
data taken at the highest fluence. It is clear, even in the 75 mW trace, that a bleaching signal is
competing with an induced absorption signal.
0.04-
0.02-
losw'b " M
0.00-0.02Pump/Probe Incident Pov ers
- 10/1 mW
-15/1.5 mW
25/2.5 mW
mW
-40/4
mW
-75/7.5
-0.04-0.06.
-0.08-
/opolabeuliet'pumpprobetbrsooapr25 154abk/de.dat c10lens 1.54pm
0apr251 54sbrvarn bhf
-0.100
40000
20000
60000
80000
Time Delay (fs)
FIGURE 4.9 A zoomed-out version of Figure 4.7.
In summary, there are many competing effects that determine a semiconductor saturable
absorber's nonlinear time-dependent response. These include: spectral hole burning, carrier-carrier scattering, two-photon absorption, free-carrier absorption, and perhaps intervalley scattering,
carrier capture outside the quantum well sections of the absorber, or a phonon bottleneck effect, in
addition to interband recombination.
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
51
In traces dominated by bleaching signals and also, in those dominated by induced absorption,
there is a small but persistent step seen at the end of the traces (100-200 ps after zero delay). This
apparently results from sample heating, and scales linearly with increasing pump power. Since
the pump-probe setup can only measure signals up to 400 ps, the heating component of the sample
response appears as a step in the data.
To determine the origin of the induced absorption, pump-probe measurements at similar fluences
were performed on a half-wave layer of indium phosphide, on the same DBR. Comparisons
between the induced absorption in the absorber and the induced absorption from the pure indium
phosphide sample are shown below. The signals from the absorber and the indium phosphide
have similar time constants (with the indium phosphide having a slightly longer time constant),
but the comparison does not provide the full answer. The two-photon absorption coefficients in
InP and InGaAs are similar. However, in the structure studied, there are about equal amounts of
InGaAs and InP overlapping with the field, due to the large number of quantum wells (six). The
induced absorption signal probably has contributions from both the InP and the InGaAs.
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
52
0.20.0-0.2
-
-0.4-0.6
-
-0.8
-
Pump/probe incident powers/structure
75/7.5 mW saturable absorb er
-75/7.5 mw X/2 InP
100/10 mW V2 InP
-
/eopo'aju iet/pumpprobehb rs00apr23non bomb(
00jun4/00jun4dOOapr23154e 1.54 pmc 11) I ens ar
-1.0I 0
0
I
4
8000
4000
I
I
16000
12000
'
I
20000
Time Delay (fs)
FIGURE 4.10 Comparison of pump-probe traces from /2 InP and saturable absorber structures (consisting of
/2 InP embedded with six InGaAs quantum wells). The structures measured were anti-reflection coated. Data is
normalized for the sake of comparison and taken at 1.54 Rm.
0.2
-
0.0
-
-0.2
-
-0.4
-
-0.6
Pump/probe incident powers/structur e
75/7.5 mW saturable absort er
-=--75/7.5 mw V2 InP
100/10 mW X2 InP
-
-
-0.8 -
/'opa'a bjiet/pnpprobe/sbrs/00apr23 nbomb/
1.54 lm 110 lens ar
-1.0-
00jun4c/O0jun4d/0apr23154e
'I
0
I
20000
I
I
40000
60000
I
I
80000
100000
Time Delay (fs)
FIGURE 4.11 A zoomed-out version of Figure 4.10.
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
53
These fluence-dependent time-resolved reflectivity measurements were repeated for wavelengths
ranging from 1.48 to 1.58 gm. By wavelength tuning, one is varying the saturation fluence of the
absorber and exploring the band structure. As the wavelength becomes progressively shorter, one
is probing progressively higher and higher into the bands. Because there are more states available, our bleaching signals increase. The bandedge of the structure studied in this section was
1.58 gm. Because the bandedge of this device was smeared by -60 nm, and the degeneracy point
of the OPO was 1620 nm, it was difficult to do pump-probe truly below band on these devices.
Instead, pump-probe of InP on DBR's was done (see Figure 2.9 and Section 4.3). In Figures 4.12,
4.13, 4.14, 4.15, 4.16, and 4.17, wavelength-dependent data for different fluences on a non-bombarded non-anti-reflection coated saturable absorber (Figure 2.6) is shown. The data follows our
predictions. Although it is not possible to go truly below band, the induced absorption signals
definitely increase as the wavelength becomes longer. In the long time delay traces, the dual
exponential (the growing exponential due to free-carrier absorption and two-photon absorption,
and the decaying exponential due to recombination) can be seen.
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
54
0.07-
-
II
0.06-
1.48
1.50
1.52
-
-- --
'K
--
em.
- -1.54
-1.58
0.050.04-
-I
0.030.02-
- --
-b
-
0.01/O0mar231 54f/O0inar231 58'/00mr23150b/tflma,241 48c/O0nar241 52c
0.000
3000
6000 9000 12000 15000 18000
Time Delay (fs)
FIGURE 4.12 Wavelength dependent data on a non-anitreflection coated structure using an 11 mm focusing lens.
The incident pump and probe power were 7.5/0.75 mW respectively.
0.070.060.05-
0.04-
1.48gm
1.50 gm
1.52 gm
1.54 gm
-- 1.58 gm
--
0.030.02-
"M-
0.010.00 -M
0
80000
40000
Time Delay (fs)
120000
/00mar23154fs/00mar231 58ds/OOnar23150bIs/cMmar24148cls/00nar241 52d s
FIGURE 4.13 A zoomed-out version of Figure 4.12.
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
55
it
0.030-
I
%a
f
'I
0.025-
I',
0.020--M
-
1.48
1.50
1.52
1.54
gm
tm
tm
gm
-- 1.58 gm
0.015
0.010-
,
--aft
.
-W
s
- --
M'
0.005/opol ab/e/j ul e/pum probe/s brs/0omar23154g
/ oOm~ar23158b/0omar2315oc/oomar24148UOOmar241,
0.000
3000
C
32b
6000 9000 12000 15000 18000
Time Delay (fs)
FIGURE 4.14 Wavelength dependent data on a non-anti-reflection coated structure. An 11 mm focusing lens
was used, and the incident pump/probe powers were 40/4 mW respectively.
0.030
I
0.025-
--
0.0200.015-
-
NN--1.58
1.48gm
1.50 pm
1.52 m
-1.54 gm
pim
0.0100.0050.000-t
0
cpolab'e/jdi/pumprobe'sbrs/OOmar23154g s
OOnrar23158bisO0mar231 9cls/OOmar24148UstOOma i4152b Is
30000
60000
90000
120000 150000
Time Delay (fs)
FIGURE 4.15 A zoomed-out version of Figure 4.14.
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
56
0.0300.0250.020
0.0150.010-
-11.48 gm
- 1.50m
--- 1.52 ptm
--- 1.54 tm
-
-- 1.58 ptm
0.005/opolab/jule/pumprobefsbrs/OOmar23154h
/00"ar231 aIOOmar23150d/OOnar24148aOOmar24 152a
0.0000
3 00 6000
9000 12000 15000 18000
Time Delay (fs)
FIGURE 4.16 Wavelength dependent data for a non-anti-reflection coated structure. An 11 mm focusing lens
was used, and the incident pump/probe powers were 75/7.5 mW.
0.0300.0251.48 jIm
-- 1.50 pm
1.52 jim
-- -1.54 tm
-1.58 tm
0.020.
--
0.015-
..
0.0100.005
cpol ab'ej di /pumprobetsbrs/00mar2315
/omar231 aasaOOmer2315Odist/OOnar24148astOOmar24152aIs
0.0000
3
S30000
600
60000
- 9
90000
2 00
-
1
120000 150000
Time Delay (fs)
FIGURE 4.17 A zoomed-out version of Figure 4.16.
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
57
4.2 Pump-Probe of Proton-Bombarded Absorbers
The previous section discussed. the time response of non-proton-bombarded absorbers. In this
section, the response from proton-bombarded absorbers will compared to the response from the
non-bombarded absorbers. A range of absorber structures was sent out for proton bombardment,
and later, some of these structures were anti-reflection coated. First, the results of bombardment
on the structure discussed in Section 4.1 (see Figure 2.6) will be presented, and then compared
with the results of bombarded InP/mirror samples.
The structures, with six quantum wells at the peak of the electric field, were bombarded with 40
keV protons, with three different doses: 1013 protons/cm 2 , 1014 protons/cm 2 , and 1015 protons/
cm 2 . Pump-probe measurements were performed on all three structures. At low fluences (less
than the absorber saturation fluence), the proton-bombarded samples showed significant reduction
of recombination times. The sample with 1013 protons/cm 2 had a recovery time of 12 ps; the 1014
protons/cm 2 sample, a recovery time of 3 ps; and the 1015 protons/cm 2 sample, a recovery of 1 ps.
The highest dosed sample shows a factor of 40 reduction in lifetime compared to an non-bombarded sample, at low fluences! No amorphous layers appeared to be created in the proton-bombarded samples. This is one advantage to using light ions, such as protons for bombardment. In
Figure 4.18, a comparison at low fluence between non-bombarded and bombarded structures is
shown. Data is normalized for the sake of comparison. None of the structures were anti-reflection coated.
For each of the proton-bombarded samples, similar characterization to that of the non-bombarded
sample was carried out. In general, the ultrafast dynamics are fairly similar between bombarded
and non-bombarded samples. The thermalization of the carriers is not measurable, so conclusions
cannot be made about the effect of bombardment on this process. Carrier cooling occurs on a
time scale of ~1 ps in the non-bombarded sample. In the bombarded samples, it shortens slightly,
to ~ 0.6 -0.7 ps in the 1015 protons/cm 2 dosed sample.
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
58
I
I
1.00.80.6N
E
-
non-bombarded
10
.4
- -6 10
0~-
otons/cM
protons/cM2
0.20.0 III
0
100000
80000
60000
40000
20000
4
opolae/juIet/pumpprote/sbrs/2-14-OO/OOe>1
T ime Delay (fs)
FIGURE 4.18 Comparison of the time response of samples with different proton doses, compared to a nonbombarded sample. The pump-probe was done at 1.54 tm on non-anti-reflection coated samples. Data is
normalized for the sake of comparison.
2
Shown in Figures 4.19 and 4.20, are low fluence traces for the 1014 protons/cm and the 1015 pro-
tons/cm 2 samples. Data is taken for non-anti-reflection coated structures. In Figure 4.20, data at
2
fairly low fluence, slightly below and above the saturation fluence, for the 1015 protons/cm
absorber is shown. Note the lack of ultrafast dynamics. This is because the absorber recovers too
quickly for these dynamics to be apparent. Also, the modulation depth is considerably lower than
that of the non-bombarded sample (see Figure 4.5 and 4.8).
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
59
-~
-
-
-~--~w
I
-
0.035-
I
i
0.0300.025
-
Incident pump/probe powers
7.5 mW
--
0.020-
mW
-2.5
40mW
75 mW
0.015 -
0.010r~ 41ft41-k
0.0050.000
-M
6000
3000
0
1
I
*
F
-
L
12000 15000 18000
9000
Time Delay (fs)
/opolab/e ulettbrs/3-25.YOOmar25 daasets- a,b,c,d
2
.
FIGURE 4.19 Pump-probe measurements at low fluence for a sample bombarded with the 1014 protons/cm
coated
The measurements were performed at 1.54 pm, with an 11 mm focusing lens, on a non-anti-reflection
structure.
0.020-
0.015-
0.010 -
Incident pump/probe fluer nces
--- 100 mW
75 mW
Is'
40 mW
7.5 mW
0.005-
0.000
I
0
2000
I
I
-%"
4000
6000
8000
/p
Time Delay (fs)
oa tijL
10000 12000
et4,umppro besbrs/4-9-0 0100a pr915 4
2
15
.
FIGURE 4.20 Pump-probe measurements at low fluence for a sample bombarded with the 10 protons/cm
coated
non-anti-reflection
a
on
lens,
focusing
mm
11
an
with
The measurements were performed at 1.54 Rm,
structure.
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
60
Pump-probe on these samples at higher fluences was done, using a combination of a more tightly
focusing lens, 6.24 mm, instead of the 11 mm lens used for the measurement above, and antireflection coated samples. Data is shown in Figures 4.21 and 4.22. Induced absorption becomes
apparent, and ultrafast dynamics appear in the 1015 protons/cm 2 absorber. The most interesting
feature of these higher fluence traces is the effect of the proton bombardment on the induced
absorption signals. At lower fluences (below or slightly above the saturation fluence), measurements showed a lifetime reduction by one and a half orders of magnitude, for the 1015 protons/
cm 2 absorber. However, the induced absorption components do not follow this trend. At the
highest pump/probe powers, (on the anti-reflection coated structures with the 6.12 mm focusing
lens), the non-bombarded structure has an induced absorption component with a time constant of
-4-10 ps. The 1015 protons/cm 2 absorber has an induced absorption component with a recovery
time of approximately 2 ps. From the low fluence traces, it appears that the bombarded absorbers
would be recover before an effect like this manifested itself. Obviously, there is another mechanism at work that has not been discussed, as of yet. There are several possibilities. Perhaps the
damage profile in InP is different from that of InGaAs, and the induced absorption signal is dominated by InP. Another explanation is that the induced absorption signal is coming mainly from the
InGaAs, with some contributions from the InP. However, as InP and GaAs have similar crystal
structures, it is unlikely that this is the cause [34]. Perhaps the cause of the induced absorption
recovery - carriers outside the wells, highly excited carriers, or those trapped in satellite valleys is not sensitive to the effect of proton bombardment. Perhaps the defects created by the bombardment do not affect states high in the bands, so highly excited carriers behavior is not as affected by
bombardment. It could be that the time constant is somewhat shorter in the bombarded samples
than the non-bombarded samples because there are fewer carriers to begin with, to participate in
high fluence events such as induced absorption or that the bombardment-created defect states are
saturated. Note, that although the induced absorption is not much affected by the bombardment,
even at these fluences, the bleaching signals still are. Proton bombardment appears to have a dramatic effect on the interband dynamics, and not much of an effect on the intraband dynamics
(such as cooling and induced absorption).
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
61
0.040.02-r
0.00-0.02-
-
b
-
0
40
%%n
Incident pump/probe fluence
- 10/ 1 m W
- 15/1 .5 mW
--- 25/2.5 mW
40/4 mW
- 75/7.5 mW
--
-0.04-0.06bpolab'e4
-0.08-
uLiet4umpprobsbrs/4-23-0O00apr231014bomb/graph
1
9000 12000 15000 18000
3000 6000
Time Delay (fs)
2
FIGURE 4.21 High fluence data for the 1014 protons/cm anti-reflection coated structure. Data was taken at
1.54 gm with a 6.24 mm lens.
0.02-
P~.
t.
0.00-
I
-0.02-
Incident pump/probe powers
mW
-10/1
-- -15/1.5 mW
- - -25/2.5 mW
40/4 mW
mW
-75/7.5
-0.04-0.06-
r231015bho mbrap~h 1
Jo pOlabA4 ull etlo umppro be/sbrs/-23-flIOAp
be/sbrs/4-23-00/nOan r231015bo mb/nmr-A 1
rida WPA ull etth um-
3
0
-
-
3000 6000
hb
U 9000 12000 15000 18000
'1-
-
Time Delay (fs)
2
FIGURE 4.22 High fluence data for the 1015 protons/cm anti-reflection coated structure. Data was taken at 1.54
gm with a 6.24 mm lens.
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
62
For the sake of comparison, the high fluence data for both non-bombarded and bombarded samples has been plotted in Figures 4.23, 4.24, 4.25, 4.26, 4.27, and 4.28. The data has been normalized for comparison purposes. Data is taken at 1.54 gm on anti-reflection coated samples with a
6.24 mm lens, as a function of fluence.
0.020
fit
0.015
NVI
%A
non-boifbarded
protons/cm2
-10
10" protons/cm
10" protons/cm 2
- -
V.-
0.010
-
0.005
0.000
C:Users\IM
0
5000
sOrsthess\nd
10000
fgtres\GDapi231i ftrM
15000
1540r
20000
Time Delay (fs)
FIGURE 4.23 Data on anti-reflection coated samples, taken with a 6.24 mm focusing lens. The wavelength was
1.54 gm, and the incident pump/probe powers 15/1.5 mW.
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
63
0.020
0.015
non-bombzarded
2
10 proton s/cm
-
-
proton s/cm 2
-. -10
0.010-
5
101 proton s/cm
--
2
0.005
0.000
I Mw VW V
AwAAL
C:\Users\Juliet\Ntsirstheds\Fina
0
20000
figues\Mapi23hghtcarmbombl540rnmr
I
I
I
40000
60000
80000
U
1
100000
Time Delay (fs)
FIGURE 4.24 A zoomed-out plot of Figure 4.23.
0.60.4
0.2-
0.0-0.2-
4
--
protons/cm
0101
non-bombarded
N
2
13
-0.4-
/
E
0 -0.6-
10
14
-10
Z
protons/cm
15
protons/cm
2
2
-0.8
-1.0-
/pol abetjti tI Lmppro be/sbrs/423.OQOOap r2N ghcom bo mbgraph 2
0
5000
10000
15000
20000
Time Delay (fs)
FIGURE 4.25 Data on anti-reflection coated samples, taken with a 6.24 mm focusing lens. The wavelength was
1.54 gm, and the incident pump/probe powers 40/4 mW. Data is normalized for comparison purposes.
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
64
0.60.40.2
0.0~0 -0.2N
-0.40
Z
-
non-bombarded
-103 protons/cm2
-
10 protons/cm
-10" protons/cm 2
2
14
-0.6
-0.8-1.0"
/b pol able/jul lettpuLppro be/sbrs/4-23-OQ'O00ap r23hi ghfcombo mb/graph 2
0
20000
60000
40000
80000
100000
Time Delay (fs)
FIGURE 4.26 A zoomed-out version of Figure 4.25.
0.0-0.2-0.4N
a0
Z
-
-0.6-
---
-
non-bombarded
-
10' protons/cm
14
2
10" protons/cm
15
10b protons/c
2
2
-0.8-1.0/opol ab/adju iet/pumpprobe/sbrs/4-23
-
0
5000
10000
00/00ap r23h gffcomnbomb/graph
1
I
15000
20000
Time Delay (fs)
FIGURE 4.27 Data on anti-reflection coated samples, taken with a 6.24 mm focusing lens. The wavelength was
1.54 Rm, and the incident pump/probe powers 75/7.5 mW. Data is normalized for comparison purposes.
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
65
0.
-
-V-
,
-0.2
-0.4
- non-bombarded
2
13
10 protons/cm
-
--
-
.
E-
10"14protons/cm2
101protons/cm 2
z -0.8
-1.0
/opolabl/jU iet/pumpprobe/sbrs/42300/00ap r23h ghf combomb/graph 1
0
20000
40000
60000
80000
100000
Time Delay (fs)
FIGURE 4.28 A zoomed-out version of Figure 4.27.
Wavelength-dependent measurements of the bombarded samples were also performed. Wavelength-dependent results as a function of fluence were very similar to what was observed in the
non-bombarded sample. The effect of proton bombardment is to smear out the bandedge. However, since the bandedge of the devices is fairly smeared to begin with (-60 nm) and it is not possible to perform measurements truly below band with the OPO, the effect of this additional
smearing was not clearly observed in the data. Shown in Figures 4.29, 4.30, 4.31, 4.32, 4.33, and
4.34, are wavelength-dependent measurements for the bombarded structures dosed with 1014 pro2
2
tons/cm 2 and 1015 protons/cm as a function of fluence. For the 1015 protons/cm sample, note
that the sample behavior does not appear to be wavelength-dependent. Bleaching signals do not
appear to increase with shorter wavelengths: this could be an indication of bandedge smearing.
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
66
0.0350.030I
0.0250.020-
10
0.015-I-
0.0100.005
gm
Rm
jm
jm
jm
54cls/
I 58cls15 2cls/lS0clsl 48cls c22O lens 7.5 mW 1014 bomb
/e/poahiliet/mniprbe/sbrsA0Omar251
Ito
-
- 1.48
-1.50
-1.52
1.54
1.58
i
0.000)
5000
I
*
*
10000
15000
U
U
-
20000
25000
30000
Time Delay (fs)
2
FIGURE 4.29 Pump-probe measurements of a 1014 protons/cm sample. The sample was not anti-reflection
coated, and the fluence was for 7.5/0.75 mW for pump and probe powers respectively.
0.020-
0.015-
0.010'6
0.005-
. . . 1.48 gm
-1.50 gm
--1.52 gm
- -1.54 gm
,- - 1.58 gm
I
%
0.000
0
5000
10000 15000 20000 25000 30000
robe2sbrs/0n
blpp
a425154bms/
Time Delay (fs) /ebpabjLisi/14pL
158b~152bs15*,Is/148bds; c220 lens 40 mWl1 d'bomb
2
FIGURE 4.30 Pump-probe measurements of a 1014 protons/cm sample. The sample was not anti-reflection
coated, and the fluence was for 40/4 mW for pump and probe powers respectively.
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
67
0.0150.012
0.009
IE . 1.48 gm
< 0.1.50
gm
0.006
1.52 pm
1.54 gm
--%'Ito
..
0.003
0.000
0
5000
25000
20000
15000
10000
30000
a
le*umobsrs/00maQ25154
TimeTim Delay
c220 lens 75 mW1 d"bomb
IS)158als/152als/150als/148als
Deay (fs)
paj
2
FIGURE 4.31 Pump-probe measurements of a 1014 protons/cm sample. The sample was not anti-reflection
coated, and the fluence was for 75/7.5 mW for pump and probe powers respectively.
0.024 0.0200.0160.012
1.50 gm
" 1. 54 gm
1.58 gm
-
0.008S
-
----
0.004 -'
0.000
4000
/opolab/euleYpumpprobsbrsO
n~n ar
7.5 mW 10" c22
0lems
5000
6000
7000
8000
apr/154d/15/15Od
9000
10000
Time Delay (fs)
2
FIGURE 4.32 Pump-probe measurements of a 1015 protons/cm sample. The sample was not anti-reflection
coated, and the fluence was for 7.5/0.75 mW for pump and probe powers respectively.
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
68
0.0160.0140.0120.010M
0.0080.006M
-
0.004'
-
1.50 gm
1.54 pm
- 1.58 gm
0.002.
/opolb/ julie pumpprobeibrEfOO apr'154C/1158d/1
0.000.
4000
1
5000
2
6000
9000
8000
7000
10000
Time Delay (fs)
2
FIGURE 4.33 Pump-probe measurements of a 1015 protons/cm sample. The sample was not anti-reflection
coated, and the fluence was for 40/4 mW for pump and probe powers respectively.
0.012
0.0100.008;
0.006 -
0.004-
',
-
.-
1.50 pm
-1.54
gm
1.58 gm
0.0020.000
4000
5000
6000
Time Delay (fs)
7000
8000
9000
10000
/opolab/edulepumpprob'sbrs0apr9/1 54b/158b/150b
75mW10c220lens
nonar
2
FIGURE 4.34 Pump-probe measurements of a 1015 protons/cm sample. The sample was not anti-reflection
respectively.
powers
coated, and the fluence was for 75/7.5 mW for pump and probe
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
69
In the proton-bombarded samples, a decrease in modulation depth with increasing bombardment
was apparent, consistent with the side-effects of bombardment documented in the literature (see
Section 2.5). The saturation fluence of the absorbers did not appear to affected by the bombardment. A plot of modulation depth vs. bombardment dose, for 40 keV protons, is shown in Figure
4.35.
3.50
0
3.0
.
2.54O-
~2.0
.0 1.59a
0
0.50ijuL18campmodopj
nonarsamples c220ens 1.54 pm
0.00.0
5.0x101 4
1.Oxl1
5
Bombardment Dosage (protons/cm2) [40 keV]
FIGURE 4.35 Plot of modulation depth vs. bombardment dosage for 40 keV protons. Modulation depths for
non-anti-reflection coated structures at 1.54 gm are plotted. Samples plotted are: non-bombarded, 1013 protons/
cm 2 , 1014 protons/cm 2 , and 1015 protons/cm 2.
Along with a decrease in modulation depth, there is an increase in non-saturable loss, with
increasing bombardment levels. However, it is fairly small (on the order of a few percent). Figure
4.36 illustrates the bombardment-dependence of the non-saturable loss.
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
70
5
%0 4
C3-O
2Q
01
z
0
0.0
5.0 x1 0 1
1.0x1 01
Dosage (protons/cm ) [40 keV]
C:AU s rs\Ju iet\Masi rs the si s'Fina
IIgure sOO aug 6non satloss
sbrvnon ar
FIGURE 4.36 Plot of non-saturable loss vs. bombardment dosage for 40 keV protons. Modulation depths for
non-anti-reflection coated structures at 1.54 gm are plotted. Samples plotted are: non-bombarded, 1013 protons/
cm 2 , 1014 protons/cm 2 , and 1015 protons/cm 2 .
The detrimental side-effects of decreased modulation depth and increased non-saturable loss can
be alleviated through short anneals. Samples were annealed in a rapid thermal annealer (RTA), in
a nitrogen atmosphere, at Lincoln Labs. Fifteen to thirty second anneals were performed at 200
0C,
and temperatures were increased in 25 0 C increments. The annealing time is not as critical as
the annealing temperature (a 15 s anneals vs. a 30 s anneal at a fixed temperature should be fairly
similar [9]). After each anneal, pump-probe was performed, to measure annealing-induced
changes. No changes in samples were detected until a temperature of 325 0C was reached. At
this temperature, a 20 s anneal of a 40 keV 1015 protons/cm 2 sample, increased the modulation
depth to 2.05%, and reduced the non-saturable loss to ~4%. The time constant increased slightly
to -1.6 ps, but this recovery time is still shorter than the 3 ps produced with 1014 protons/cm 2
dosage. In Figure 4.38, fluence-dependent measurements on this annealed sample are shown.
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
71
0.021-
0.018
5/0.5
mW
0.015
25/2.5 mW
0.012
-75/7.5 mW
0.009
0.006
niom
1-
0.003
0.000
\\S3 rab'c\Users\Jultet\Masters thesis\Fina figures\00augl 6ran5325sbr6101 5
0
4000
2000
6000
8000
10000
Time Delay (fs)
FIGURE 4.37 Pump-probe traces as a function of fluence at 1.54 pm, for an non-anti-reflection coated sample.
An 11 mm focussing lens was used, and the sample was bombarded with 40 keV protons, at a dose of 1015
protons/cm 2 . The sample was annealed for 20 s at 325 'C.
Proton bombardment of the structure shown in Figure 2.7 was also done. Lifetime reduction
appeared to be very similar. In addition, structures overgrown with 512 of indium phosphide
were bombarded. Overgrowths were performed to increase the amount of two-photon absorption
present in the samples, as it can prevent instabilities such as Q-switched modelocking [61]. Lifetime reduction occurred in an analogous manner. Plotted in Figure 4.38, is a comparison of lifetime for these overgrown structures. The dosage scheme is detailed in Figure 4.39. A more
elaborate bombardment scheme was necessary because the samples are now five times the thickness of the absorbers shown in Figures 2.6 and 2.7. Dosage schemes were designed to produce a
uniform damage profile, from the quantum wells to the sample surface. Bombardment schemes
were based on phenomenological data [9].
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
72
I
I
I
0.008
0.006
0.004-
0.002-
%0
0.000 -"""
-0.002
-0.004
-
-non-bombarded
-Dosage scheme 1
-0.06-Dosage scheme 2
-
-0.008.
mar25cl2sbrvonarcompom.op 75rmWcO2201ens 1.54pmsbr
*
0
Dosage scheme 3
30000
60000
90000
nonar 512
120000
150000
Time Delay (fs)
FIGURE 4.38 Pump-probe traces on a structure similar to that in Figure 2.6 with 5/2 of indium phosphide
overgrown. Measurements were made at 1.54 pm, with an 11 mm lens on non-anti-reflection coated samples with
an 11 mm lens. The lifetime for a non-bombarded sample was 35 ps, for dosage scheme 1 18 ps, for dosage
scheme 2 5.5 ps, and for dosage scheme 3, 1.3 ps.
4.3 Pump-Probe of Proton-Bombarded InP Structures
As discussed in the previous section, for several reasons, it was not possible to do pump-probe
measurements below-band on the saturable absorber structures. Instead, measurements were
made of pure InP structures (see Figure 2.9) of differing thickness, as a function of bombardment.
Similar reductions in time constant were found as a function of bombardment, however, it was not
possible to compare samples to an non-bombarded sample, as a non-anti-reflection coated sample
did not exist. Following, are pump-probe results from three different doses of bombardment on a
5/2 piece of InP. The measurements were not done on /2 InP (the thickness of InP in the saturable absorber structures), due to the extremely small signals from the non-anti-reflection coated
structures. anti-reflection bombarded V/2 InP structures were not available for measurements.
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
73
Two-photon absorption is proportional to the intensity, the interaction length, and the two-photon
absorption coefficient. Because of this, thicker samples produced greater signals. Since the samples were thicker (5W/2 vs. /2), multiple doses of different energy protons were used to ensure a
uniform damage profile. Below, is a table of the three different bombardment schemes used on
the 512 indium phosphide pieces.
Dosage Scheme
Energy (keV)
Dose(proton/cm 2 )
200
1013
150
4x 10
100
2x 1012
40
2
12
200
1014
150
4x 1013
100
2 x 1013
40
3
1x 10
12
1x 10
13
200
1015
150
4 x 10
14
100
2 x 10
14
40
1 x 10
14
FIGURE 4.39 Proton bombardment schemes for 512 InP samples.
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
74
0.0-0.2
CD
cc
0
-
-0.4-
aI-
-
-
-0.6-
E -0.8 0
z
-1.0-
Dosage scheme 3
Dosage scheme 2
Dosage scheme 1
ab/uliet/pumpprobeheb
/e/bpol
00feb17aAeb1
19combombirpar
9b((0eb19f c1 10 lensnonar, 25 mW 1.54 gm
20000 40000 60000 80000 100000
0
Time Delay (fs)
FIGURE 4.40 Pump-probe measurements as a function of dosage scheme for 512 InP. Measurements were
performed at 1.54 pm on non-anti-reflection coated samples, with pump/probe fluences of 25/2.5 mW
respectively. A 6.24 mm sample focussing lens was used.
0.0-
~-6
~
~
ee*~
'P
-0.2 -
1
I,
/
0
I
-0.4-
I
Ii
3
I
-0.6N
E -0.80
z
-1.0-
scheme 3
-Dosage
Dosage scheme 2
Dosage scheme 1
/eopolab'ieleYpumpprobtbrs.V0eb1 9combombirpar
Web17b/00fab19a0Ifeb19e c110 lens nonar, 50 mW 1.54 pm
I
0
I
I
4000
I
20000 40000 60000 80000 100000
Time Delay (fs)
FIGURE 4.41 Pump-probe measurements as a function of dosage scheme for 5/2 InP. Measurements were
performed at 1.54 gm on non-anti-reflection coated samples, with pump/probe fluences of 50/5 mW
respectively.A 6.24 mm sample focussing lens was used.
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
75
Dual exponential fits of the data were done. The lowest dosed sample (scheme 1) had time constants of 3 and 36 ps. Dosage scheme 2 had recovery times of 2-3 ps, and 12 ps, and dosage
scheme 3, recovery times of 1.5 ps and 10 ps. A non-bombarded sample had a recovery time of
60-90 ps. The instantaneous signal is due to a combination of two-photon and TPA-induced freecarrier absorption. At lower fluences, two-photon absorption dominates; at higher fluences, TPAinduced free-carrier absorption takes off. The short time constant probably represents the cooling
of carriers. It decreases slightly as dosage increases, similar to the behavior observed in the bombarded saturable absorber samples. This may be due to the fact that there are simply fewer high
energy carriers, due to the increase in non-saturable loss, and the trap states. The long time constant reflects recombination of the carriers.
For an indium phosphide sample, single beam saturation measurements can also be performed.
The following curve was generated from the structure in Figure 2.9. Although this structure has
quantum wells, they have a bandedge of -1450 nm, and since the measurement was performed at
1.54 gm, only the indium phosphide contributes to the signal.
1.01
1.00
-
!
pa%
0.99
.
0.98
S0.97
0.960
0.95
0.941
0.1
1
10
Energy Density
100
1000
(gJ/cm2)
FIGURE 4.42 Saturation measurements of an indium phosphide structure. Figure from [59].
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
76
The roll-off in this curve is caused by two-photon and TPA-induced free-carrier absorption. It is
possible to achieve extremely short recovery times in non-bombarded InP, with proper operating
conditions. If two-photon absorption is to be used as an intensity limiter, it is better to operate at a
point on this curve, where the effects of free-carrier absorption are negligible, allowing an "instantaneous" recovery time. A demonstration of intensity-limiting was carried out in an actively harmonically modelocked laser, with the structure shown in Figure 2.9. The laser experiment was
performed at fluences, producing a modulation depth of 0.5 - 1.5% in this structure [60]. At this
modulation depth, the absorber recovers on the order of the pulse width.
0.000
-
-0.002-0.004-0.006-0.008
-0.01 0
2000
/(oaejldprqr~/bi*90Yop~ratb
4000
6000
8000
10000
Time Delay (fs)
FIGURE 4.43 Instantaneous recovery time of TPA, used as an intensity limiter in a high-repetition rate system.
Pump-probe was performed at 1.55 gm on an anti-reflection coated structure of indium phosphide, 21/2 thick.
An 11 mm lens was used.
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
77
4.4 Laser Results from Proton-Bombarded Absorbers
Anti-reflection proton-bombarded saturable absorber samples were tested in a linear erbium fiber
laser, with a 17 MHz repetition rate. With adequate pumping, a multiple pulse regime was
entered. However, at certain high pump powers, more pulses did not continue to develop: instead
an instability appeared. It was thought that the instability regime resulted from the 40 ps recovery
time of the non-bombarded absorbers. Thus, the proton-bombarded absorbers were tested in this
laser. The instability regime was not eliminated, but the pulses did get shorter, with higher levels
of bombardment. A schematic of the laser is shown in Figure 4.44. Self-starting modelocking
was observed for the absorbers with the 1013 protons/cm 2 and 1014 protons/cm 2 . Modelocking
also occurred with the 1015 protons/cm 2 absorber, but it was not self-starting.
980 nm
Pump
Lens
DM
105
15% Output
Coupler
EDF
SMF
SMF
Saturable X1/4 X/2 Collimator
A bsorber
FIGURE 4.44 Schematic of 17 MHz Er fiber laser used to test the proton-bombarded absorbers. The laser
consisted of: 0.33 m of flexcor, 1.39 m erbium, and 4.34 m of single-mode fiber.
Eight-hundred fs pulses were produced with saturable absorbers of the structures shown in Figures 2.6 and 2.7. Shown in Figures 4.45 and 4.46, are a spectrum and autocorrelation of a singlepulse state produced from the laser with a proton-bombarded (1014 protons/cm 2 ) absorber, of the
structure shown in Figure 2.6.
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
78
-40-45
E
-50-
0 -55CL
-60
00070214.dat
-G
1570
390 le ns D-0.007ps sbrv 10"
ar
1
1580
1590
1600
1610
Wavelength (nm)
FIGURE 4.45 Spectrum of erbium fiber laser in a single pulse state modelocked with an anti-reflection coated
proton-bombarded absorber, bombarded with 1014 protons/cm 2 at 40 keV.
-AC
1
C)
-
Gaussian Fit (822 fs)
Sech Fit (770 fs)
-
0.1
E
I
N
I
I
0.01
0
z
1E-3
1k
-I
-2000
000 70215.dat
-1000
d390 len s D -0.007ps
0
2
sbrv 1
ar
1000
2000
Time Delay (fs)
FIGURE 4.46 Autocorrelations of erbium fiber laser in a single pulse state modelocked with an anti-reflection
coated proton-bombarded absorber, bombarded with 1014 protons/cm 2 at 40 keV.
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
79
Chapter 5: Conclusions and Future Work
5.1 Conclusions
Proton bombardment can reduce the lifetime of semiconductor saturable absorber mirrors by a
factor of forty, at low fluences, fluences at the saturation fluence of the absorber. However, the
lifetime of the absorber is not reduced by such a dramatic amount at the higher fluences, which
may occur during the startup of many lasers,. Induced absorption, that occurs at these high fluences, is not affected much by the bombardment. The mechanism that causes the induced absorption is obviously not as sensitive to proton bombardment as the state filling effect seen at lower
fluences. The induced absorption is probably caused by carriers outside the InGaAs quantum
wells, highly excited carriers, or those trapped in satellite valleys. At such high fluences, it is also
possible that the bombardment-created defects are saturated. Obviously, the high fluence regime
for these particular saturable absorbers, bombarded or non-bombarded, is not a desirable operating point. However, high fluence conditions occur during the start-up of lasers, and in Q-switched
modelocking; thus understanding absorber dynamics under these conditions is important. Finally,
proton-bombarded samples were tested in a laser, and modelocked operation was observed for all
three proton doses studied. All absorbers in this thesis modelocked an erbium fiber laser, with no
significant change in the threshold conditions. From this, we conclude that the detrimental
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
80
effects of increased non-saturable loss and reduced modulation depth did not appear to be too
severe. Low-gain laser systems, however, are much more sensitive to small changes in nonsaturable loss or modulation depth. However, it has been shown that annealing can reduce these detrimental side-effects.
In conclusion, proton-bombarded absorbers can dramatically reduce the lifetime of semiconductor saturable absorbers, when they are operated under "low" fluence conditions. High fluence
conditions are undesirable modes of operation for these semiconductor saturable absorbers: bombarded or not. Proton bombardment is a relatively simple post-growth processing step, demonstrated on a monolithically integrable device in this thesis. In addition, to proton bombardment,
two other techniques for lifetime reduction were briefly presented. Depending on the constraints
on lifetime, placement of the absorbing layers at different distances from a defect-riddled interface can reduce or increase the lifetime. If two-photon absorption is used as an intensity limiter,
biasing the absorber in the correct regime can lead to "instantaneous" recovery times.
5.2 Future Work
The real application of saturable absorbers with faster recovery times is in high repetition rate systems. In the scope of this thesis, there was not time to build such a system. Absorbers with 40 ps
recovery times would start to limit the repetition rate of systems running at higher than 15 GHz.
Er:Yb waveguide laser systems [59] show promise for potentially high repetition rate systems, as
do actively harmonically modelocked fiber systems. The results of low-temperature growth and
ion bombardment of these particular structures could be compared; as there is not an extensive
comparison of the two methods in the literature. Proton-bombarded structures could be optimized
for low-gain systems, to aid with self-starting or noise cleanup.
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
81
This thesis investigates one of many parameters that can be varied in saturable absorber design.
Future designs of new saturable absorbers could include a much more broad-band absorber structure that remains monolithically integrable. By inserting varying thicknesses of quantum wells,
one can create a broad-band absorption region. Developing a broad-band mirror that is integrable
poses more of a challenge. Currently, the broadest band saturable absorbers are fabricated by
wafer bonding an epitaxially grown absorbing region to a silver mirror [12]. Growing absorbers
on top of double-chirped mirrors, which compensate for second and third order dispersion in laser
cavities, is another possibility. Absorbers with greater modulation depth would also probably further noise cleanup investigations.
With an ultrafast Cr 4 +:YAG system, it would be possible to make pump-probe measurements with
more time resolution. This would enable a more careful study of ultrafast dynamics in these
structures. Index measurements using a heterodyne technique could also be performed. These
measurements would allow further understanding of modelocked laser dynamics.
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
82
References
[1] G. P. Agrawal and N. K. Dutta. Long Wavelength SemiconductorLasers. Van Nostrand:
Princeton, 1986.
[2] P. C. Becker, H. L. Fragnito, C. H. Brito Cruz, J. Shah, R. L. Fork, J. E. Cunningham, J. E.
Henry, and C. V. Shank. Femtosecond Intervalley Scattering in GaAs. Applied Physics Letters
53, 2089-2091 (1988).
[3] I. P. Bilinsky. Novel Saturable Absorber Materials and Devices for Laser Modelocking. MIT
Doctoral Thesis. February 1999.
[4] L. R. Brovelli, U. Keller, and T. H. Chiu. Design and Operation of Antiresonant Fabry-Perot
Saturable Semiconductor Absorbers for Mode-Locked Solid-State Lasers. Journal of the Optical Society of America B 12, 311-322 (1995).
[5] S. Collin, J. Ramos, J. Lopez, J. Mangeney, and N. Stelmakh. Optical Sampling by Ultra-fast
High-contrast Saturable Absorber Created by Heavy Ion Irradiation. 1999 Digest Conference
on Lasers and Electro-Optics 35, 1667-1668 (1999).
[6] E. L. Delpon, J. L. Oudar, N. Bouche, R. Raj, A. Shen, N. Stelmakh, and J. M. Lourtioz.
Ultrafast Excitonic Saturable Absorption in Ion-implanted InGaAs/InAlAs Multiple Quantum
Wells. Applied Physics Letters 72,759-761 (1998).
[7] J. C. Diels and W. Rudolph. UltrashortLaser Pulse Phenomena. Academic Press: Boston,
1996.
[8] F. E. Doany, D. Grischkowsky, and C. C. Chi. Carrier Lifetime Versus Ion-implantation Dose
in Silicon on Sapphire. Applied Physics Letters 50, 460-463 (1987).
[9] J. P. Donnelly. Personal Communication. March 2000.
[10] D. J. Dougherty, S. B. Fleischer, E. L. Warlick, J. L. House, G. S. Petrich, L. A. Kolodziejski, and E. P. Ippen. Ultrafast Carrier Dynamics and Intervalley Scattering in ZnSe. Applied
Physics Letters 71, 3144-3146 (1997).
[11] P. M. Downey and B. Schwartz. Picosecond Response in 3He+ Bombarded InP Photoconductors. Applied Physics Letters 44, 207-209 (1984).
[12] R. Fluck, I. D. Jung, G. Zhang, F. X. Kartner, and U. Keller. Broadband Saturable Absorber
for 10-fs Pulse Generation. Optics Letters 21, 743-745 (1996).
[13] A. Grabmaier, M. Schofthaler, A. Hangleiter, C. Kazmierski, M. Blez, and A. Ougazzaden.
Carrier Transport Limited Bandwidth of 1.55 pm Quantum-Well Lasers. Applied Physics Letters 62, 10-13 (1993).
[14] F. Ganikhanov, G. Lin, W. Chen, C. S. Chang, and C. Pan. Subpicosecond Carrier Lifetimes
in Arsenic Ion-implanted GaAs. Applied Physics Letters 67, 3465-3467 (1995).
[15] P. Grenier and F. F. Whitaker. Subband Gap Dynamics in Low-temperature Grown GaAs.
Applied Physics Letters 70, 1998-2000 (1997).
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
83
[16] S. Gupta, J. F. Whitaker, and G. A. Mourou. Ultrafast Carrier Dynamics in III-V Semiconductors Grown by Molecular-Beam Epitaxy at Very Low Substrate Temperatures. IEEE Journal of Quantum Electronics 28, 2464-2472 (1992).
[17] M. Haiml, U. Siegner, F. Morier-Genoud, U. Keller, M. Luysberg, P. Specht, and E. R.
Weber. Femtosecond Response Times and High Optical Nonlinearity in Beryllium-doped
Low-temperature Grown GaAs. Applied Physics Letters 74,1269-1271 (1999).
[18] M. Hofer, M. H. Ober, F. Haberl, and ME. Fermann. Characterization of Ultrashort Pulse
Formation in Passively Mode-Locked Fiber Lasers. IEEE Journal of Quantum Electronics 28,
720-722 (1992).
[19] H. A. Haus. Parameter Ranges for CW Passive Mode Locking. IEEE Journal of Quantum
Electronics QE-12, 169-176 (1976).
[20] H. A. Haus, J. G. Fujimoto, and E. P. Ippen. Structures for Additive Pulse Mode Locking 8,
2068-2076 (1991).
[21] H. A. Haus. Theory of Mode Locking With a Fast Saturable Absorber. Journal of Applied
Physics 46, 3049-3058 (1975).
[22] E. P. Ippen, D. J. Eichenberger, and R. W. Dixon. Picosecond Pulse Generation by Passive
Modelocking of Diode Lasers. Applied Physics Letters 37, 267-269 (1980).
[23] E. P. Ippen. Principles of Passive Mode Locking. Applied Physics B 58, 159-170, (1994).
[24] E. P. Ippen. Femtosecond Pulse Generation: Principles and Fiber Applications. Erice Summer School Plenum 1999.
[25] M. N. Islam, E. R. Sunderman, C. E. Soccolich, I Bar-Joseph, N. Sauer, T. Y Chang, and B.
I. Miller. Color Center Lasers Passively Modelocked by Quantum Wells. IEEE Journal of
Quantum Electronics 25, 2454-2463 (1989).
[26] M. Joschko, P. Langlois, E. R. Thoen, E. M. Koontz, E. P. Ippen, and L. A. Kolodziejski.
Ultrafast Hot-Carrier Dynamics in Semiconductor Saturable Absorbers. Applied Physics Letters 76, 1383-1385 (2000).
[27] P. W. Juodawlkis, D. T. McInturff, and S. E. Ralph. Ultrafast Carrier Dynamics and Optical
Nonlinearities of Low-temperature Grown InGaAs/InAlAs Multiple Quantum Wells. Applied
Physics Letters 69, 4062-4065 (1996).
[28] I I. D. Jung, F. X. Kartner, N. Matuschek, D. H. Sutter, F. Morier-Genoud, G. Zhang, U
Keller, V. Scheuer, M. Tilsch, and T. Tschudi. Self-Starting 6.5 fs Pulses from a Ti:Sapphire
Laser. Optics Letters 22, 1009-11 (1997).
[29] I. D. Jung, F. X. Kartner, N. Matuschek, D. H. Sutter, F. Morier-Genoud, Z. Shi, V. Scheuer,
M. Tilsch, T. Tschudi, and U. Keller. Semiconductor Saturable Absorber Mirrors Supporting
sub-10-fs Pulses. Applied Physics B 65, 137-150 (1997).
[30] F. X. Kartner, L. R. Brovelli, D. Kopf, M. Kamp, I. Calasso, and U. Keller. Control of
Solid- State Laser Dynamics by Semiconductor Devices. Optical Engineering 34, 2024-2036
(1995).
[31] F. X. Kartner and U. Keller. Stabilization of Solitonlike Pulses with a Slow Saturable
Absorber. Optics Letters 20; 16-18 (1995).
[32] U. Keller, K. J. Weingarten, F. X. Kartner, D. Kopf, B. Braun, I. D. Jung, R. Fluck, C. Honninger, N. Matuschek, and J. Aus der Au. Semiconductor Saturable Absorber Mirrors
(SESAM's) for Femtosecond to Nanosecond Generation in Solid State Lasers. IEEE Journal
of Selected Topics in Quantum Electronics 2, 435-453 (1996).
[33] U. Keller. Semiconductors and Semimetals. Academic Press: Boston, 1998.
[34] E. M. Koontz. Personal Communication. August 2000.
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
84
[35] E. M. Koontz. The Development of Components for Ultrafast All-Optical Communication
Networks, MIT Doctoral Thesis. June 2000.
[36] J. N. Kutz, B. C. Collings, K. Bergman, S. Tsuda, S. T. Cundiff, W. H. Knox, P. Holmes, and
M. Weinstein. Mode-Locking Pulse Dynamics in a Fiber Laser with a Saturable Bragg Reflector. Journal of the Optical Society of America B 14, 2681-2690 (1997).
[37] I. Lahiri, D. D. Nolte, E. S. Harmon, M. R. Melloch, and J. M. Woodall. Ultrafast-Lifetime
Quantum Wells with Sharp Exciton Spectra. Applied Physics Letters 66, 2519-2521 (1995).
[38] M. Lambsdorff, J. Kuhl, J. Rosenzweig, A. Axmann, and J. Schneider. Subpicosecond Carrier Lifetimes in Radiation-Damaged GaAs. Applied Physics Letters 58, 1881-1883 (1991).
[39] K. F. Lamprecht, S. Juen, L. Palmetshofer, and R. A. Hopfel. Ultrashort Carrier Lifetimes
in H+ Bombarded InP. Applied Physics Letters 59, 926-929 (1991).
[40] P. Langlois, M. Joschko, E. R. Thoen, E. M. Kootnz, F. X. Kartner, E. P. Ippen, and L. A.
Kolodziejski. High Fluence Ultrafast Dynamics of Semiconductor Saturable Absorber Mirrors. Applied Physics Letters 75, 3841-3843 (1999).
[41] M. J. Lederer, B. Luther-Davies, H. H. Tan and C. Jagadish. An Antiresonant Fabry-Perot
Saturable Absorber for Passive Mode-Locking Fabricated by Metal-Organic Vapor Phase Epitaxy and Ion Implantation Design, Characterization, and Mode-Locking. IEEE Journal of
Quantum Electronics 34, 2150-2161 (1998).
[42] M. J. Lederer, B. Luther-Davies, H. H. Tan, C. Jagadish, M. Haiml, U. Siegner, and U.
Keller. Nonlinear Optical Absorption and Temporal Response of Arsenic- and OxygenImplanted GaAs. Applied Physics Letters 74, 1993-1995 (1999).
[43] W. H. Loh, D. Atkinson, P. R. Morkel, M. Hopkinson, A. Rivers, A. J. Seeds, and D. H.
Payne. All-Solid-State Subpicosecond Passively Mode Locked Erbium-Doped Fiber Laser.
Applied Physics Letters, 63, 4-6 (1993).
[44] W. H. Loh, D. Atkinson, P. R. Morkel, M. Hopkinson, A. Rivers, A. J. Seeds, and D. H.
Payne. Passively Mode-Locked Er3+ Fiber Laser Using a Semiconductor Nonlinear Mirror.
IEEE Photonics Technology Letters, Vol. 5, pp. 35-37.
[45] H. S. Loka, S. D. Benjamin, and P. W. E. Smith. Influence of Material Growth and
Annealing Conditions on Recombination Processes in Low-temperature Grown GaAs. Optics
Communications 161, 232-235 (1999).
[46] M. Lynch, J. Hegarty, A. Ginty, W. M. Kelly, and W. T. Tsang. Optical Measurement of Ion
Implantation Damage Depth in Multiple-quantum Well Mesa Structures. Applied Physics Letters 59, 1013-1015 (1991).
[47] H. W. Mocker and R. J. Collins. Mode Competition and Self-Locking Effects in a QSwitched Ruby Laser. Applied Physics Letters 7, 270-273 (1965).
[48] U. Morgner, F. X. Kartner, S. H. Cho, Y Chen, H. A. Haus, J. G. Fujimoto, and E. P. Ippen.
Sub-Two-Cycle Pulses from a Kerr-Lens Mode-Locked Ti:sapphire Laser. Optics Letters 24,
411-413 (1999).
[49] J. Shah. UltrafastSpectroscopy of Semiconductors and Semiconductor Structures.
Springer: Berlin, 1999.
[50] U. Siegner, M. Haiml, F. Morier-Genoud, R. C. Lutz, P. Specht, E. R. Weber and U. Keller.
Femtosecond Nonlinear Optics of Low-Temperature Grown Semiconductors. Physica B 273274, 733-736 (1999).
[51] Y Silberberg, P. W. Smith, D. A. B. Miller, B. Tell, A. C. Gossard, and W. Wiegmann. Fast
Nonlinear Optical Response from Proton-Bombarded Multiple Quantum Well Structures.
Applied Physics Letters 46, 701-703 (1985).
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
85
[52] Y Silberberg, P. W. Smith, D. J. Eilenberger, D. A. B. Miller, A. C. Goosard, and W. Wiegmann. Passive Modelocking of a Semiconductor Diode Laser. Optics Letters 9, 507-509
(1984).
[53] P. W. Smith, Y Silberberg, and D. A. B. Miller. Mode Locking of Semiconductor Diode
Lasers Using Saturable Excitonic Nonlinearities. Journal of the Optical Society of America B
2, 1228-1235 (1985).
[54] P. W. E. Smith, S. D. Benjamin, and H. S. Loka. Tailoring of Trap-related Carrier Dynamics
in Low-Temperature Grown GaAs. Applied Physics Letters 71, 1156-1158 (1997).
[55] D. H. Sutter, I. D. Jung, F. Morier-Genoud, F. X. Kartner, and U. Keller. Self-starting 6.5 fs
Pulses from a Ti:Sapphire Laser Using a Semiconductor Saturable Absorber and DoubleChirped Mirrors. IEEE Journal Selected Topics on Quantum Electronics 3, 197-208 (1998).
[56] D.H. Sutter, G. Steinmeyer, L. Gallmann, N. Matuschek, F. Morier-Genoud, U. Keller, V.
Scheuer, G. Angelow, and T. Tschudi. Semiconductor Saturable-Absorber Mirror-Assisted
Kerr-lens Modelocked Ti:sapphire Laser Producing Pulses in the Two-cycle Regime. Optics
Letters 24, 631-633 (1999).
[57] R. Takahashi, Y Kawamura and H. Iwamura. Ultrafast 1.55 gm All-optical Switching
Using Low-Temperature Growth Multiple Quantum Wells. Applied Physics Letters 68, 153156, (1996).
[58] E. R. Thoen. Development of Ultrashort Pulse Fiber Lasers for Optical Communication
Utilizing Semiconductor Devices. MIT Doctoral Thesis. June 2000.
[59] E. R. Thoen, E. M. Kootnz, D. J. Jones, D. Barbier, F. X. Kartner, E. P. Ippen, and L. A.
Kolodziejski. Erbium-Ytterbium Waveguide Laser Mode-Locked with a Semiconductor Saturable Absorber Mirror. IEEE Photonics Technology Letters 12, 149-151 (2000).
[60] E. R. Thoen, M. E. Grein, E. M. Koontz, E. P. Ippen, H. A. Haus, and L. A. Kolodziejski.
Stabilization of an Active Harmonically Mode-locked Fiber Laser Using Two-Photon Absorption. Optics Letters 25, 948-950 (2000).
[61] E. R. Thoen, E. M. Koontz, M. Joschko, P. Langlois, T. R. Schibli, F. X. Kartner, E. P. Ippen,
and L. A. Kolodziejski. Two-Photon Absorption in Semiconductor Saturable Absorber Mirrors. Applied Physics Letters 74, 3927-3929 (1999).
[62] J. P. van der Ziel, W. T. Tsang, R. A. Logan, R. M. Mikulyak, and W. M. Augustyniak. Subpicosecond Pulses from Passively Modelocked GaAs Buried Optical Guide Semiconductor
Lasers. Applied Physics Letters 39, 525-527 (1981).
[63] S. Weiss, J. M. Weisenfield, D. S. Chenmla, G. Raybon, G. Sucha, M. Wegener, G. Eisenstein, C. A. Burrus, A. G. Dentai, U. Koren, B. I. Miller, H. Temkin, R. A. Logan, and T. Tanbun-Ek. Carrier Capture Times in 1.5 gm Multiple Quantum Well Optical Amplifiers.
Applied Physics Letters 71, 9-11 (1992).
[64] M. Zirngibl, L. W. Stulz, J. Stone, J. Hugi, D. DiGiovanni, and P. B. Hansen. 1.2 ps Pulses
from Passively Mode-Locked Diode Pumped Er-doped Fiber Ring Laser. Electronic Letters
27, 1735-1737 (1991).
The Effect of Proton Bombardment on Semiconductor Saturable Absorber Structures
86