Download Optics and Quantum Electronics—J. G. Fujimoto, H. A. Haus, E. P. Ippen, F. X. Kärtner

24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
Optics and Quantum Electronics
Academic Staff
Prof. James G. Fujimoto, Prof. Hermann A. Haus, Prof. Erich P. Ippen,
Professor Franz X. Kärtner
Research Staff, Visiting Scientists and Affiliates
Samuel Adams, Dr. Stephane Bourquin, Dr. Mark Brezinski, Dr. Linze Duan, Dr. Jan-Molte
Fischer, Dr. Matthew Grein, Dr. Katherine Hall, Christian Jirauschek, Dr. Alexander Killi, Christian
Koos, Dr. Chris Kroeger, Dr. Xinbing Liu, Dr. Francisco Lopez-Royo, Dr. Shun-ichi Matsushita,
Dr. Christina Manolatou, Dr. Chan H. Park, Dr. Lelia A. Paunescu, Dr. Mark Roberts, Dr. Thomas
R. Schibli, Karl Schneider, Dr. Wolfgang Seitz, Prof. Alphan Sennaroglu, Dr. Luciano Socci, Dr.
G. Hugh Song, Dr. Hideyuki Sotobayoshi, Dr. Debra Stamper, Dr. Yuichi Takushima, Dr. Ping
Xue, Dr. Rebecca Younkin
Graduate Students
Desmond Adler, Aaron Aguirre, Juhi Chandalia, Marcus Dahlem, Fuwan Gan, Ravi Ghanta, Juliet
Gopinath, Felix Grawert, Paul Herz, Pei-lin Hsiung, Leaf Jiang, Aristidis Karalis, Mohammed Jalal
Khan, Jung-Won Kim, Tony Ko, Andrew Kowalevicz, O. Onur Kuzucu, J.P. Laine, Ryan Lang, Lia
Matos, Nirlep Patel, Milos Popovic, Poh-Boon Phua, Rohit Prasankumar, Peter Rakich, Daniel
Ripin, Bryan Robinson, Karen Robinson, Vikas Sharma, Shelby Savage, Hanfei Shen, Jason
Sickler, Michael Watts, Jade Wang, Aurea Tucay Zare
Technical and Support Staff
Mary Aldridge, Donna Gale, Cindy Kopf
Research Areas and Projects
Ultrashort Pulse Generation and Laser Technology
Octave Spanning Lasers and Dispersion Compensating Laser Optics
Compact Low-Threshold Ti:Al2O3 Laser
Generation of 150nJ Pulses from a Multiple-Pass Cavity KLM Ti:Al2O3 Oscillator
MPC Laser Development
Ultrafast Cr4+: YAG Laser
Cr:LiSAF Laser System
10 fs Diode Pumped Cr:LiCAF Laser
Spectral Broadening in a Tapered Fiber and High Numerical Aperture Fiber using Femtosecond
Nd:Glass Laser
1Pm Stretched-Pulse Laser with Microstructured Fiber for Dispersion Compensation
Timing Jitter Studies in a Passively Modelocked Regeneratively Synchronized Fiber Laser
Timing Jitter and Correlations in Harmonically Modelocked Fiber Lasers
Timing Jitter Reduction Using a Timing-Jitter Eater
Timing Jitter Studies in Hybridly Modelocked Semiconductor Lasers
Variational Analysis of Spatio-temporal Pulse Dynamics in Dispersive Kerr Media
Ultrafast Phenomena and Quantum Electronics
Ultrafast Pump-Probe Studies of Silicon- and III/V-based Devices
Materials for Modelocking
High-Speed Femtosecond Pump Probe Spectroscopy Using a Smart Pixel Detector Array
Photonics and Devices
Micromachined Photonic Devices
24-1
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
Development of First, Second, and Third Order Ring Resonators for Channel Dropping Filter
Applications
Tuning and Switching of Ring Resonator Through Perturbation of Effective Index
Guiding and Band Edge Measurements of 2-Dimensional Photonic Crystal Slab Formed by Posts
Integrated Tunable/Switchable Optical Add-Drop Multiplexer
Grating Filters and Reduced Radiation
Polarization Mode Dispersion
Optical Phase Control and Stabilization Techniques
Attosecond Synchronization of Modelocked Lasers
Few-Cycle Nonlinear Optics and Carrier-Envelope Phase Effects
Active Modelocking Using a Nonlinear Fabry-Perot Modulator
24-2
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
Ultrashort Pulse Generation and Laser Technology
Octave Spanning Lasers and Dispersion Compensating Laser Optics
Sponsors
National Science Foundation ECS-0119452
MIT Presidential Fellowship
Office of Naval Research N-00014-02-1-0717
Project Staff
Christian Koos, Onur Kuzucu, Lia Matos, Dr. Thomas R. Schibli, Dr. Lingze Duan,
Professor Franz X. Kaertner
The generation of ultrashort laser pulses continues to be a very active field of research. This
technology has found applications in the areas of biomedical optics, high speed communications,
frequency metrology and the investigation of ultrafast nonlinear processes in semiconductor
materials and devices. Generally, these laser sources aim to be cost effective, robust, and
technologically simple. Kerr-lens modelocking (KLM), which utilizes the electronic Kerr effect to
create an artificial fast saturable absorber, has been the most successful technique for the
generation of ultrashort pulses. Working in collaboration with Professors Erich P. Ippen,
Hermann A. Haus, and James G. Fujimoto, we have developed a theoretical model which
provides a foundation for understanding and optimization of short-pulse KLM lasers. Our program
investigates several areas of ultrafast laser technology, with the objective of developing new
technologies that can be applied across a range of laser materials and systems.
Double Chirped Mirror Pairs for Prismless Octave Spanning Lasers
Solid state lasers can have gain over extremely broad bandwidths of several hundred
nanometers, enabling the generation of few cycle pulse durations or longer pulse durations with
broad tunability. In addition, self-phase modulation (SPM) is a temporal nonlinear effect
originating in the Kerr nonlinearity at high intensities that generates new frequencies and
spectrally broadens the pulse. The broadband gain media, together with SPM, allow for
emission over one octave of bandwidth. The development of compact and robust octave
spanning lasers is of prime importance for optical frequency metrology and investigation of phase
sensitive nonlinear optical processes. To achieve this broadband emission directly from the laser
precise dispersion compensation is indispensable. Double-chirped mirrors (DCMs) have recently
emerged as a powerful technology that permits intracavity dispersion management [1-6]. Using a
combination of prism pairs and pairs of matched DCMs, octave spanning spectra have been
obtained directly from the laser [7]. However, the prism sequence prevents a compact layout of
the laser, which is also susceptible to long term drifts, because of beam variations in the prism
pair. We have designed and fabricated novel DCM-pairs[8] that cover an octave of bandwidth and
compensate dispersion using mirrors and thin BaF2-wedges only. BaF2 is chosen, because it has
the lowest third order dispersion in comparison with other fluorides and glasses transparent in the
visible to near infrared region. Figure 1 shows the calculated reflectivity of one mirror of the DCMpair, which is also transmissive for the pump light at 532nm.
The dispersion compensating mirror pairs provide on average more than 99.9% reflectivity with
only a small dip in the range of 800-900 nm. The angle of incidence on one mirror type of the pair
is increased by 4o in comparison with the design for optimum cancellation of the dispersion
oscillations without noticeable change in the average dispersion. The dispersion oscillations
cancel very well, considering the high sensitivity of the overall design on fabrication tolerances.
24-3
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
1.0
100
0.8
80
0.6
60
0.4
40
0.2
20
0.0
0
600
800
1000
Wavelength, nm
Group Delay (fs)
Reflectivity
RLE Progress Report 145
1200
Figure 1: The calculated reflectivity (red solid line) of one mirror of the DCM-pair, which is
also transmissive for the pump light at 532nm. The dispersion measured from 650 nm to
1100 nm (blue solid line) follows closely the design goal (green dashed line). The dispersion
measurement was limited to 1100nm because of the Si-detector used.
Prismless Octave Spanning Ti:Sapphire Laser
Using these novel dispersion compensating laser mirrors a compact Ti:sapphire laser design as
shown in Figure 2 is possible. The laser has a standard z-cavity design with an additional BaF2
plate in one arm of the resonator and BaF2-wedges in the other arm for fine adjustment of the
overall dispersion. The footprint of the laser is only 30x25 cm even at a repetition rate of only 82
MHz. In contrast to our earlier work [7], in the current setup second as well as third order
dispersion is almost symmetrically balanced in both laser arms giving rise to ideal conditions for
dispersion managed soliton formation and the generation of ultrabroadband spectra when the
wedges are adjusted to practically zero average intracavity dispersion.
Figure 2: Setup of the prismless, octave spanning Ti:sapphire laser. The possible overall
footprint of the laser is only 30x25cm.
24-4
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
1.0
0
0.8
-10
-20
0.6
-30
0.4
-40
0.2
-50
0.0
-60
800
1000
Wavelength [nm]
Power Spectral Density [dB]
Power Spectral Density [a.u.]
RLE Progress Report 145
1200
Figure 3: Typical output spectrum of the prismless Ti:sapphire laser. It covers one octave
from 600-1200 nm at about –30 dB below the average power level.
Figure 3 shows first spectra generated directly from this laser. It shows a pronounced peak at 670
nm due to the early roll-off of the output coupler. On a logarithmic scale the octave is reached at
about –30dB below the average power level. This is already good enough to use this laser for
direct optical frequency referencing using a 1f-2f technique [9]. A detailed pulse characterization
is in progress. These preliminary results show that the current limitations are not the mirror
bandwidth and the dispersion compensation but rather the early roll-off of the output coupler at
the short wavelength side of the spectrum. More broadband output couplers may lead to
significantly more output in the short and long wavelength range.
References
1. R. Szipöcs, K. Ferencz, C. Spielmann and F. Krausz, "Chirped multilayer coatings for
broadband dispersion control in femtosecond lasers," Opt. Lett. 19(3): 201-3 (1994).
2. R. Szipöcs, A. Stingl, C. Spielmann and F. Krausz, "Chirped dielectric mirrors for dispersion
control in femtosecond laser systems," paper presented at Generation, Amplification, and
Measurement of Ultrashort Laser Pulses II, Proc. SPIE, San Jose, California. Feb. 6-7, 1995.
3. R. Szipöcs and A. Kohazi-Kis, "Theory and design of chirped dielectric laser mirrors," Appl.
Phys. B 65(2): 115-136 (1997).
4. F.X. Kaertner, N. Matuschek, T. Schibli, U. Keller, H.A. Haus, C. Heine, R. Morf, V. Scheuer,
M. Tilsch and T. Tschudi, "Design and fabrication of double-chirped mirrors," Opt. Lett.
22(11): 831-33 (1997).
5. N. Matuschek, F.X. Kaertner and U. Keller, "Theory of Double-Chirped Mirrors," IEEE J.
Select. Topics Quantum Electron. 4(2): 197 (1998)
6. U. Morgner, F.X. Kaertner, S.H. Cho, Y. Chen, H.A. Haus, J.G. Fujimoto, E.P. Ippen, V.
Scheuer, G. Angelow and T. Tschudi, "Sub-two-cycle pulses from a Kerr-lens mode-locked
Ti:sapphire laser," Opt. Lett. 24(6): 411-13, (1999).
7. R. Ell, U. Morgner, F.X. Kaertner, J.G. Fujimoto, E.P. Ippen, V. Scheuer, G. Angelow and T.
Tschudi, "Generation of 5 fs pulses and octave-spanning spectra directly from a Ti:sapphire
laser," Opt. Lett. 26(6): 373-5 (2001).
8. F.X. Kaertner, U. Morgner, T.R. Schibli, E.P. Ippen, J.G. Fujimoto, V. Scheuer, G. Angelow
and T. Tschudi, “Ultrabroadband double-chirped mirror pairs for generation of octave
spectra,'' J. Opt. Soc. of Am. B 18(6): 882-5, (2001).
9. D.J. Jones, S.A. Diddams, J.K. Ranka, A. Stentz, R.S. Windeler, J.L. Hall, S.T. Cundiff,
“Carrier-Envelope Phase Control of Femtosecond Modelocked Lasers and Direct Optical
Frequency Synthesis,” Science 288(5466), 635-9 (2000).
24-5
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
Novel Low-Coherence Light Sources
The need for simple, robust sources of broadband light exists in fields such as spectroscopy as
well as biomedical imaging applications such as Optical Coherence Tomography (OCT) [1-3].
The achievable longitudinal resolution is inversely related to the bandwidth of the source.
Standard, ~10 µm axial resolution OCT imaging can be performed using superluminescent
diodes (SLD) that have ~20-30 nm FWHM bandwidths centered near 800 nm for ophthalmic
imaging, while SLD at longer wavelengths ~1300 nm are used for imaging of tissue. These
sources are relatively inexpensive, portable and have turn-key operation suitable for clinical use,
but provide limited resolutions due to their narrow bandwidths. Recently, OCT imaging with axial
resolutions of ~1 µm has been achieved using a Ti:Al2O3 laser with a bandwidth of ~300 nm [4,
5]. Unfortunately, these systems are expensive and complex, limiting their widespread use.
Our group has conducted ongoing work to develop novel low-coherence light sources. We have
introduced several alternative sources for ultrahigh resolution imaging. These sources focus on
reducing the cost, increasing the reliability and portability of systems, while maintaining high
performance capability.
Clinical compact low-threshold Ti:Sapphire laser
Sponsors
Air Force Office of Scientific Research (MFEL)
Grant F49620-01-1-0186
Air Force Office of Scientific Research
Grant F49620-98-01-0084
National Science Foundation
Grant ECS-019452
National Institute of Health
Grant NIH-5-R01-CA75289-04
National Institute of Health
Grant NIH-2-R01 EY11289-15
Project Staff
Stephane Bourquin, Aaron D. Aguirre, Ingmar Hartl, Pei-Lin Hsiung, Paul R. Herz, Tony H. Ko,
Tim A. Birks, William J. Wadsworth, Udo Bünting, Daniel Kopf, and James G. Fujimoto
Kerr lens modelocked (KLM) Ti:Al2O3 lasers can generate extremely short pulse durations with
broad bandwidths that are particularly useful in biomedical imaging [6]. A standard Kerr lens
modelocked laser operating with a 5W pump can produce output powers of 500 mW and
bandwidths in excess of 150 nm [7]. Unfortunately, the high cost of today’s femtosecond lasers
severely limits their widespread use. The cost of femtosecond Ti:Al2O3 lasers is strongly
dependent on the pump power requirements. Diode pumped solid-state lasers capable of
generating 5 W can be prohibitively expensive, while lasers generating several hundred mW are
considerably more affordable.
24-6
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
f = -500 mm
1.3 W
Max
Quartz
O/2
Crystal 2.05 mm
Optical Path D=5 cm-1@ 514nm
f = 50 mm CM1
CM2
HR
L = 40 cm
L = 54.5 cm
M2
L = 80 cm
L = 94.5 cm
M1
OC
Figure 1. Schematic of the low-threshold prismless Ti:Al2O3 laser. Arm lengths are
120 cm and 149 cm for the HR and OC arms respectively. Intracavity dispersion
compensation and tuning is provided by solely by double chirped mirrors (DCM).
In previous work we were able to develop an ultra-low-threshold modelocked Ti:Sapphire laser
which reduced the modelocking threshold to under 200 mW, significantly reducing the cost of
femtosecond sources [8]. Unfortunately, because standard mirrors were used, output bandwidth
of only 100 nm was achievable. While this effort significantly reduced the cost of a broadband
source, its spectral width, since longitudinal resolution is inversely proportional to the bandwidth
of the source, was not sufficient to achieve the ultra-high resolution OCT that other lasers with
double-chirped mirrors (DCMs) were capable of.
By following similar design criteria as our previous work, but by using 3rd generation DCM
technology to compensate higher order dispersion, we are able to develop a low-threshold
modelocked Ti:Al2O3 laser which is suitable for clinical ultra-high resolution OCT imaging. The
laser cavity is shown in figure 1. The entire laser cavity, as well as the pump source, has been
placed on a single lightweight breadboard measuring 19” by 45”, making it compact and portable.
We are using a compact pump source operating at 1.0 W output power. The modelocking is
initiated by a rapid translation of the end mirror high reflector. Once modelocked, the output
power is 50 mW with the broadband and smooth output spectrum with 124 nm FWHM, shown in
figure 2a. Even though the output power is modest compared with a standard laser, ophthalmic
imaging has exposure limitation of ~750uW, making this amount of output more than sufficient.
In figure 2b we see the measured resolution from the interference fringes. The 3.9 um resolution
in air corresponds to 3 um resolution in tissue, making ultrahigh resolution imaging possible.
Figure 2. (a) The output spectrum showing smooth 124 nm spectrum and (b) interferometric
fringes indicating a longitudinal resolution of 3.9 um in air corresponding to 3.0 um in tissue.
24-7
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
In conclusion, we have developed a low-threshold, low-cost, compact laser system. By replacing
standard mirrors and prisms with 3rd generation DCMs, we are able to develop a broadband
portable source to be used in the clinical setting for ultrahigh resolution OCT imaging with 3 um
longitudinal resolution.
References
1. Huang, D., et al., Optical coherence tomography. Science, 1991. 254(5035): p. 1178-1181.
2. Tearney, G.J., et al., In vivo endoscopic optical biopsy with optical coherence tomography.
Science, 1997. 276(5321): p. 2037-9.
3. Boppart, S.A., et al., In vivo cellular optical coherence tomography imaging. Nature Medicine,
1998. 4(7): p. 861-5.
4. Drexler, W., et al., In vivo ultrahigh resolution optical coherence tomography. Optics Letters,
1999. 24: p. 1221-1223.
5. Drexler, W., et al., Ultrahigh resolution ophthalmic optical coherence tomography. Nature
Medicine, 2000. in press.
6. Spence, D.E., P.N. Kean, and W. Sibbett, 60-fsec pulse generation from a self-mode-locked
Ti:Sapphire laser. Optics Lett., 1991. 16: p. 42-44.
7. Zhou, J., et al., Pulse evolution in a broad-bandwidth Ti:sapphire laser. Optics Lett., 1994. 19:
p. 1149-51.
8. Kowalevicz, A.M., et al., Ultralow-threshold Kerr-lens mode-locked TiAl 2 O 3 laser. Optics
Letters, 2002. 27(22): p. 2037-2039.
24-8
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
Generation of 150 nJ pulses from a multiple-pass cavity KLM Ti:Al2O3
Oscillator
Sponsors
National Science Foundation - ECS-019452
Air Force Office of Scientific Research - F49620-98-01-0084
Air Force Office of Scientific Research (MFEL) - F49620-01-1-0186
Project Staff
Andrew M. Kowalevicz, Aurea Tucay, Rohit P. Prasankumar, Professor Alphan Sennaroglu,
Professor James G. Fujimoto
This year, emphasis was placed on developing solid-state femtosecond lasers that are cheaper
and have improved performance. In particular, laser cavity configurations that generate high
pulse energies have been examined. Reducing the pulse repetition rate by increasing the cavity
length allows us to achieve higher pulse energies. However, a standard 100 MHz repetition rate
laser requires a 3 meter round-trip cavity length, so reducing the repetition rate to below 10 MHz
requires an unrealistically large laser.
A laser that contains a multi-pass cavity (MPC) allows us to obtain long cavity lengths without
compromising the need to place the laser on a typical optical table. A multi-pass cavity is
essentially a resonator that is comprised of two curved mirrors, into which an off-axis laser beam
tilted in either or both transverse directions is introduced (see Fig. 1). In this configuration the
beam bounces between the two mirrors, walking around either mirror in an elliptical spot pattern
(see Fig. 2). The ellipticity of the mirror spot pattern is determined by the tilt of the input beam.
The angle between the spots on successive bounces on one mirror is governed by the radii of
curvature of the mirrors and the distance between them. The choice of this angle and placement
of notches on the mirrors (or additional small mirrors) to inject and extract the beam allows us to
control the number of round trips the beam makes in the MPC before being extracted. Hence the
careful consideration of these parameters allows us to design an MPC that, when used inside a
laser, increases the laser cavity length and decreases the pulse repetition rate in a controlled
fashion. With the help of the ABCD matrix formalism for periodic optical systems, it is possible to
design the MPC to be a “unity q” transformation – that is, upon exiting the MPC, the beam’s
focused spot size and focus position (i.e., the Gaussian beam q-parameter) are the same as
those it had when it entered the MPC. When such an MPC is inserted into an existing laser, the
MPC increases the cavity length while leaving the focusing conditions in the laser crystal
unchanged.
Figure 1. A multi-pass cavity. A beam can be injected in the cavity through notches or small
mirrors.
24-9
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
Figure 2. The spot pattern on one of the mirrors in the multi-pass cavity shown above.
In addition to ultrafast studies, femtosecond lasers are utilized in nonlinear material processing.
Even though pulse durations as short as around 5 fs [1,2] have been reached from conventional
Ti:Sapphire lasers, the pulse energy is limited to a few nanojoules, because of its high repetition
rate. For material processing high energy pulses at moderate repetition rates are desirable.
We utilize a multiple pass cavity (MPC) based on the Herriott cell to produce a unity-q
transformation [3] that facilitates the lengthening of the cavity while maintaining the operating
point of a standard laser. The long cavity lengths introduced significant dispersion from air and
prismatic compensators produced higher order dispersion mismatch. Our current work makes
use of specially designed double-chirped mirrors (DCMs) that compensate dispersion without the
need for other intracavity dispersion compensating elements.
M2
M1
L1
Pump
Retroreflector
Pump
M3
M6
M4
OC
M5
M7
M8
M10
M9
M11
Multiple Pass Cavity (MPC)
Figure 3. Schematic layout of the high pulse energy laser cavity. All shaded mirrors are DCMs.
The pump source is a frequency doubled Nd:Vanadate capable of 10W of light at 532 nm. The
crystal is 3 mm thick and absorbs ~56% of the pump light on a single pass.
Figure 4 shows the schematic of the high pulse energy Ti:Al2O3 laser. The cavity length has been
increased to 5.85 MHz repetition rate. Since our cavity length is approximately 20 times longer
than a standard laser, we expect a similar scaling of the pulse energy for a given average output
power. This substantially higher pulse energy leads to enhanced self-phase modulation (SPM).
The additional frequency components from SPM would typically lead to progressively shorter
pulses and considerably higher intensity in the gain medium, which, if left unbalanced, would
overdrive the nonlinearities that lead to stable pulse generation. In order to balance the SPM, we
increase the net negative dispersion. Our MPC adds 48 DCM bounces with approximately -46 fs2
24-10
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
/ bounce, leading to a net negative dispersion of –1250 fs2 after accounting for the additional air
path of 48 m.
In order to produce high pulse energies, we focus 9.4 W of pump light into our laser crystal. Since
the crystal absorbs approximately 56% of the incident light, there are several watts of unabsorbed
pump light that gets transmitted. We retroreflect this transmitted light with a 20 cm ROC mirror
back into the crystal, which increases the average output power of the laser while also allowing
for the use of a 25% OC. This high percentage of output coupling, along with the large net
negative dispersion, prevents the intracavity intensity from becoming too large. KLM is initiated
by translating the end mirror (M11), which results in single pulsed modelocked operation with
output powers as high as 877 mW.
In order to verify that the laser was, in fact, producing single pulses, the output was measured
with a fast photodiode, an Optical Multichannel Analyzer (OMA), as well as an intensity
autocorrelator. The oscilloscope trace, shown in Figure 4a, shows spikes with 171 ns separation
corresponding to the cavity roundtrip time for 5.85 MHz. At the same time, the OMA monitored
the laser spectrum, shown in Figure 4b. The modelocked spectrum of the laser has 16.5 nm
FWHM centered at 788nm with dual symmetric sidebands, which are due to operation at large
negative dispersion. In order to establish the duration of our pulses, we performed an intensity
autocorrelation with a thin 300 um KDP crystal. The measurement yielded a FWHM of 67 fs
resulting in a pulse width of 43 fs (Figure 4c), which is close to the transform limit of 39 fs,
assuming a sech2 pulse shape.
FWHM 779.0nm - 795.5nm = 16.5nm 39 fs Limit
FWHM 67 fs -> 43 fs Assuming Sech Pulse
1.0
0.8
0.8
0.8
0.6
0.4
0.2
Intensity (a.u.)
1.0
Amplitude (a.u.)
Amplitude (a.u.)
171ns Separation -> 5.85 MHz
1.0
0.6
0.4
0.2
0.6
0.4
0.2
0.0
0.0
-0.2
-225
-150
-75
0
Time (ns)
75
150
225
725
0.0
750
775
800
Wavelength (nm)
825
850
-150
-100
-50
0
Delay (fs)
50
100
150
Figure 4. a) Pulse spikes from a fast photodiode at the repetition rate of the laser, b) the
modelocked spectrum of the laser with 16.5 nm FWHM, and c) the measured pulse duration of 43
fs which is close to the transform limit of 39 fs.
In conclusion, we have demonstrated a prismless, KLM Ti:Al2O3 laser operating at 5.85 MHz
based on a Herriott-style MPC. Because of its unity transformation of the guassian beam in the
MPC, we have achieved long cavity laser performance with standard cavity laser stability. We
have demonstrated 150 nJ pulses with 43 fs duration corresponding to 3.5 MW peak power. We
expect this never-before-achieved performance to open new avenues for the micromachining of
materials previously only possible with amplified laser systems. It will also be a useful tool to
eliminate thermal parasitics in pump probe and nonlinear optics experiments.
References
1. Morgner, U., et al., Sub-two cycle pulses from a Kerr-Lens modelocked Ti:sapphire laser.
Optics Letters, 1999. 24: p. 411 -- 413.
2. Sutter, D.H., et al., Semiconductor saturable-absorber mirror-assisted Kerr-lens mode-locked
Ti:sapphire laser producing pulses in the two-cycle regime. Optics Letters, 1999. 24(9): p.
631-3.
3. Herriott, D., H. Kogelnik, and R. Kompfner, Off-axis paths in spherical mirror interferometers.
Applied Optics, 1964. 3(4): p. 523-526.
24-11
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
MPC Laser Development
Sponsors
National Science Foundation - ECS-019452
Air Force Office of Scientific Research - F49620-98-01-0084
Air Force Office of Scientific Research (MFEL)- F49620-01-1-0186
Project Staff
Andrew M. Kowalevicz, Aurea Tucay, Rohit P. Prasankumar, Professor Alphan Sennaroglu,
Professor James G. Fujimoto
High Performance, Compact, Prismless, Low-Threshold 30 MHz Ti:Al2O3 Laser
Practical femtosecond laser sources need to meet several important requirements so that they
can be readily integrated in systems and used in a wide range of scientific and technological
applications such as pump-probe spectroscopy, medical imaging, and communications. These
include low-cost system design, compactness, and efficient all-solid-state operation with
reasonably high pulse energies. One possible method to lower the overall laser cost involves the
development of resonator designs that enable low-threshold laser operation [1,2]. Since the
pump laser is one of the major components of a Ti:Al2O3 system, this results in a dramatic cost
reduction. The resulting decrease in the average output power, however, leads to a decrease in
the pulse energy and peak intensity, limiting their use in nonlinear optics experiments. Previous
studies have shown that laser output pulse energy can be scaled up by reducing the pulse
repetition rate. In particular, multi-pass cavity configurations have been introduced which
increase the effective cavity length, while preserving the characteristics of the laser beam inside
the gain medium, to generate high-energy pulses from femtosecond oscillators with low-tomoderate average output powers [3-5].
L1
M1
pump
M3
xtal
M6
M4
M5
M2
M7
M8
OC
zR1
Fig. 1: Schematic of the compact prismless, low-threshold 30 MHz Ti:Al2O3 laser.
In this project, we constructed a novel femtosecond Ti:Al2O3 laser which combines several
favorable features to meet the above system requirements. A schematic of the laser is shown in
Fig. 1. The resonator contains the highly reflecting mirrors M1-M8 as well as a 11% output
coupler (OC). Light amplification occurs inside a 2-mm-long Ti:Al2O3 crystal (xtal) which is end
pumped by a diode-pumped frequency-doubled Nd:YVO4 laser operated at 532 nm. The input
lens L1 focuses the pump beam to an estimated 7-Pm radius inside the crystal. The absorption
of the crystal is 74%. The laser has prismless dispersion compensation with double-chirped
mirrors (M1-M6) [6]. The effective resonator length is extended by using a multi-pass cavity
consisting of the highly reflecting mirrors M7 and M8, separated by 23.4 cm. Although the cavity
24-12
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
length is approximately 5 meters long, an extremely compact design measuring only 30 x 55 cm
has been achieved by using the multi-pass cavity design.
Autocorrelation Intensity (a.u.)
1.25
1
0.75
0.5
0.25
0
-0.25
-100
-75
-50
-25
0
25
50
75
100
Delay (fsec)
Fig. 2. Intensity autocorrelation of the femtosecond pulses generated
compact prismless, low-threshold 30 MHz Ti:Al2O3 laser .
with the
Tight focusing geometry enabled efficient low-threshold operation of the compact Ti:Al2O3 laser.
With only 1.5 W of pump power, the laser generates 19-fs pulses with an average output power of
115 mW, corresponding to a pulse energy of 3.7 nJ at a repetition rate of 31 MHz. The measured
autocorrelation and the spectrum of the pulses are displayed in Figs. 2 and 3, respectively. The
pulses are centered at the wavelength of 780 nm with a spectral bandwidth of 42 nm, indicating
that they were nearly transform-limited. The output energy of the laser is comparable to that of a
conventional 100-MHz femtosecond Ti:Al2O3 laser with an average output power of 365 mW.
The reduced average output power of the present design should also significantly decrease the
role of unwanted thermal effects in pump probe measurements.
Spectral Intensity (a.u.)
1.25
1
0.75
0.5
0.25
0
700
750
800
850
900
Fig. 3. Spectrum of the femtosecond pulses obtained from the compact Ti:Al2O3 laser.
The pulse wavelength is centered around 780 nm with a spectral bandwidth of 42 nm.
24-13
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
References
1.
2.
3.
4.
5.
6.
K. Read, F. Blonigen, N. Riccelli, M.M. Murnane, H.C. Kapteyn, “Low-threshold operation
of an ultrashort-pulse mode-locked Ti:sapphire laser,” Optics Letters, 21, 489-491, 1996.
A. M. Kowalevicz, T. R. Schibli, F. X. Kartner, and J. G. Fujimoto, “Ultra-low-threshold
Kerr lens modelocked Ti:Al2O3 lasers,” Optics Letters 27, 2037, 2002.
A.R. Libertun, R. Shelton, H.C. Kapteyn. M.M. Murnane, “A 36 nJ-15.5 MHz extendedcavity Ti:sapphire oscillator” CLEO ’99 Technical Digest. p.469-70.
S.H. Cho, B.E. Bouma, E.P. Ippen, and J.G. Fujimoto, “A low repetition rate high peak
power KLM Ti:Al2O3 laser using a multiple pass cavity,” Optics Letters, 24, 417-419,
1999.
S. H. Cho, F. X. Kärtner, U. Morgner, E. P. Ippen, J. G. Fujimoto, J. E. Cunningham, W.
H. Knox, “Generation of 90-nJ pulses with a 4 MHz repetition-rate Kerr-lens mode-locked
Ti:Al2O3 laser operating with net positive and negative intracavity dispersion,” Optics
Letters 26, 560-562, 2001.
F. X. Kärtner, N.Matuschek, T. Schibli, U. Keller, H. A. Haus, C. Heine, R. Morf, V.
Scheuer, M. Tilsch, T. Tschudi, "Design and fabrication of double-chirped mirrors," Opt.
Lett., 22, 831, 1997.
24-14
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
Ultrafast Cr4+:YAG Laser
Sponsors
U.S. Air Force – Office of Scientific Research - F49620-01-1-0084
MRSEC Program of the National Science Foundation - DMR 98-08941
U.S. Navy – Office of Naval Research
Project Staff
Hanfei Shen, Juliet T. Gopinath, Sheila Tandon, Dr. Gale Petrich, Dr. Daniel Ripin, Dr. Alexei
Erchak, Professor Franz X. Kaertner, Professor Leslie A. Kolodziejski, Professor Erich P. Ippen
Several transition-metal-doped solid-state materials are useful as ultrafast laser media because
of their chemical and mechanical robustness and stability, broad gain bandwidths, and high
nonlinear coefficients, which allow for the design of efficient Kerr-lens modelocking. Optical
sources obtained with such media can be exploited for both their short temporal pulse duration
and their large spectral bandwidth. The former property makes such lasers ideal for timeresolved studies of ultrafast phenomena and devices, such as optical clocks with precise timing at
the cavity repetition rate and ultra-high speed optical communications; while the latter can be
used for spectroscopy, for example, to generate synchronized multi-wavelength optical sources,
or for optical frequency standards in metrology.
Cr4+:YAG is one such material, with broad emission from 1300 to 1600 nm. This gain spectrum
makes Cr4+:YAG ideal for studying applications associated with optical telecommunications. We
have previously demonstrated the generation of 20-fs pulses directly from a prismless Cr4+:YAG
laser using double-chirped mirrors for dispersion compensation [1]. The modelocked spectrum
peaked at 1490 nm and had a full-width at half-maximum of 190 nm, extending from 1310 to 1500
nm. The laser cavity was a standard Z-fold configuration, designed to maximize Kerr-lens
modelocking (KLM). In general, however, KLM is not self-starting without precise alignment of
the laser cavity. Instead, external perturbations are required to initiate modelocking by creating
transient power spikes.
Saturable absorber mirrors based on semiconductor quantum wells, capable of initiating
modelocking without sensitive alignment, have been used to overcome this difficulty in a variety
of solid-state lasers. In Cr4+:YAG lasers, saturable absorber mirrors, consisting of InGaAs
quantum wells grown upon highly-reflecting mirrors, have been demonstrated [2,3]. In most
cases, these mirrors were GaAs/AlAs Bragg stacks, whose bandwidth of ~100 nm typically
limited the minimum pulse width due to spectral filtering.
Recently, we have demonstrated a novel high-index-contrast mirror-based saturable Bragg
reflector (SBR), which was used to generate self-starting pulses with duration of 35 fs directly
from a Cr4+:YAG laser [4]. The SBR consisted of a broadband 7-period GaAs/AlxOy Bragg mirror
substrate supporting a 10-nm InGaAs quantum well absorber in a O/2-thick InP layer [5]. The
GaAs and AlxOy layers have refractive indices of 3.39 and 1.61 at 1.5 Pm, respectively, creating a
high-index-contrast mirror that has a calculated reflectivity of 99.9% over the wavelength range
1220 to 1740 nm. This represents a substantial improvement in bandwidth over previous Bragg
mirrors that used GaAs/AlAs stacks. Mirror reflectivity was measured using Fourier transform
infrared spectroscopy (FTIR), and is shown in Figure 1. The SBR has a stopband from 1300 to
1800 nm, and its nonsaturable loss is estimated to be <0.8% using Findlay-Clay analysis.
Furthermore, an absolute reflectivity greater than 99% is inferred by the successful use of the
mirror in the low gain Cr4+:YAG laser itself.
24-15
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
Figure 1. Measured reflectivity of the high-index-contrast mirror-based saturable Bragg reflector.
The broadband SBR was introduced into the Cr4+:YAG laser cavity to initiate modelocking. The
laser was then optimized for KLM. Plots of the pulse spectrum and autocorrelation from the selfstarting Cr4+:YAG laser are shown in Figure 2. The pulse spectrum is centered at 1490 nm, and
has a full-width at half-maximum of 68 nm. The bandwidth-limited pulse width is determined to be
35 fs. We believe that these parameters can be improved further, to match the performance of
the previously-described 20-fs Cr4+:YAG laser. One possible hindrance in the current experiment
is two-photon absorption (TPA) in the saturable absorber, which limits the minimum pulse
duration. This difficulty could be overcome by focusing the laser light onto a larger spot size on
the SBR, which would result in a lower beam intensity that counteracts TPA. The growth
technique for the SBR used in this experiment, however, cannot provide a large enough usable
mirror surface to do this. It is estimated that the usable mirror surface extends only as far as 200
Pm a side into the structure.
0.8
-35
-40
0.6
-45
-50
0.4
-55
0.2
-60
0.0
-65
1200
1300
1400
1500
1600
8
Autocorrelation
-30
Intensity (dB)
Intensity (Arb. Units)
-25
1.0
6
W ~ 32 fs
with sech fit
4
2
0
-70
1700
-100
-50
0
50
Time Delay (fs)
Wavelength (nm)
Figure 2. Spectrum and interferometric autocorrelation of self-starting modelocked Cr4+:YAG
laser pulses.
24-16
100
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
A promising alternative growth is now being investigated that uses AlGaAs in place of GaAs in
the SBR mirror substrate. Previously, the broadband GaAs/AlxOy Bragg mirrors were created by
steam oxidation of GaAs/AlAs. However, lattice contraction during oxidation resulted in
delamination between the GaAs and AlxOy layers, creating damaging areas of low reflectivity on
the SBR. Use of AlGaAs in place of GaAs greatly extends the oxidation dimensions for stable
mirror layers. A comparison of the usable mirror surface area in (a) the GaAs/AlxOy and (b) the
AlGaAs/AlxOy -based SBRs is shown in Figure 3. The allowable spot size for laser light has been
increased from 200 to 500 Pm. In addition, the new structures are mesas that are placed
throughout the surface of the structure, as opposed to from the side before. The new larger area
SBRs would allow us to focus the laser beam onto larger spot sizes and avoid significant TPA. A
similar AlGaAs/AlxOy -based SBR has already been used to demonstrate self-starting in a
Cr:forsterite laser [6] and will next be tested in the Cr4+:YAG.
Oxidized Region
Unoxidized
Region
Oxidized
Region
500 Pm
200 P m
(a)
(b)
Figure 3. Top-down view comparison of the usable mirror surface in (a) the GaAs/AlxOy and (b)
the AlGaAs/AlxOy -based SBRs. The maximum spot size for laser light has been increased from
200 to 500 Pm.
References
1. D. J. Ripin, C. Chudoba, J. Gopinath, J. G. Fujimoto, E. P. Ippen, U. Morgner, F. X. Kaertner,
V. Scheuer, G. Angelow, and T. Tschudi, "Generation of 20-fs pulses by a prismless Cr4+:YAG
laser," Opt. Lett. 27, 61-63 (2002).
2. B. C. Collings, J. B. Stark, S. Tsuda, W. H. Knox, J. E. Cunningham, W. Y. Jan, R. Pathak and
K. Bergman, "Saturable Bragg reflector self-starting passive mode locking of a Cr4+:YAG laser
pumped with a diode-pumped Nd:YVO4 laser," Opt. Lett. 21, 1171-1173 (1996).
3. S. Spalter, M. Bohm, M. Burk, B. Mikulla, R. Fluck, I. Jung, G. Zhang, U. Keller, A. Sizmann
and G. Leuchs, "Self-starting soliton-modelocked femtosecond Cr4+:YAG laser using an
antiresonant Fabry-Perot saturable absorber," App. Phys. B 65, 335-338 (1997).
4. D. J. Ripin, J. T. Gopinath, H. M. Shen, G. S. Petrich, L. A. Kolodziejski, F. X. Kaertner and E.
P. Ippen, "Oxidized GaAs/AlAs mirror with a quantum-well saturable absorber for ultrashort-pulse
Cr4+:YAG laser," Opt. Comm. 214, 285-289 (2002).
24-17
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
5. A. A. Erchak, D. J. Ripin, J. T. Gopinath, H. M. Shen, F. X. Kaertner, G. S. Petrich, L. A.
Kolodziejski and E. P. Ippen, "Large scale oxidation of AlAs layers for broadband saturable Bragg
reflector," CLEO 73 (2002), CTuK43 225.
6. T. R. Schibli, J. W. Kim, L. Matos, A. W. Killi, J. T. Gopinath, S. N. Tandon, G. S. Petrich,J. G.
Fujimoto, E. P. Ippen, F. X. Kaertner and L. A. Kolodziejski, "300 attosecond
activesynchronization of passively mode-locked lasers using balanced cross-correlation,
submitted to CLEO 2003.
24-18
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
Cr:LiSaF Laser System
Sponsors
National Science Foundation - ECS-019452
Air Force Office of Scientific Research - F49620-98-01-0084
Air Force Office of Scientific Research (MFEL) - F49620-01-1-0186
Project Staff
Rohit P. Prasankumar, Dr. Yasuyuki Hirakawa, Andrew M. Kowalewicz, Jr., Professor Franz X.
Kaertner, Professor James G. Fujimoto.
An 8.6 MHz extended cavity femtosecond Cr:LiSAF laser pumped by low cost diode lasers
Femtosecond lasers are an essential technology for many applications including ultrafast
spectroscopy, high speed measurement, laser micromachining and biomedical imaging. In order
to make femtosecond technology more accessible to users outside of the laboratory, methods of
reducing cost while maintaining high performance must be developed. Laser diode pumped solid
state lasers are an attractive alternative to conventional pumping with expensive gas or solid
state lasers. Cr:LiSAF is a well established laser material operating around 850 nm that can be
pumped with red laser diodes to obtain mode-locked pulse durations as short as 10 fs (Uemura
and Torizuka 2002; Wagenblast, Morgner et al. 2002). Previous efforts in diode pumping
Cr:LiSAF used broad-stripe diodes with powers of hundreds of mW (Dymott and Ferguson 1995;
Uemura and Torizuka 2002). These pump diodes are still relatively expensive and have poor
mode quality, making efficient mode matching and Kerr lens mode-locking difficult. An attractive
alternative is to pump with single spatial mode diodes, which significantly improves mode
matching and laser efficiency. Single mode diodes with powers of 50-60 mW at wavelengths
ranging from 660-690 nm are available for only $20 each, making this pump source extremely
inexpensive. Previous work demonstrated compact mode-locked Cr:LiSAF lasers in several
compact configurations pumped by single spatial mode diodes (Agate, Stormont et al. 2002;
Hopkins, Valentine et al. 2002). These lasers typically generated 20 mW output power and 120
fs pulses at 430 MHz, corresponding to a pulse energy of 0.05 nJ. The maximum pulse energy
achieved was 0.14 nJ (Agate, Stormont et al. 2002), which may be too low for some applications.
Multi-pass cavities (MPC) providing a unity q parameter transformation have been used to reduce
repetition rates from laser oscillators and thereby increase pulse energies without requiring
external amplification (Cho, Bouma et al. 1999). By using an MPC in a single mode diodepumped Cr:LiSAF laser, an inexpensive source of femtosecond pulses with energies comparable
to those generated by standard Ti:sapphire lasers can be achieved. The multi-pass cavity used
in this work consists of one large plane mirror and one large curved mirror separated by a
distance. For a desired repetition rate, the radius of the curved mirror, number of bounces on
each mirror, and distance between the two mirrors can be optimized to give a unity q parameter
transformation using ABCD matrix analysis. Two smaller mirrors, one plane and one curved, are
used to introduce and extract the laser beam from the MPC.
A schematic of the experimental setup is shown in Figure 1. The diode pump source consisted of
two diodes at 663 nm (Hitachi HL6503MG) and one diode at 685 nm (Mitsubishi ML1013R). The
diodes were microlensed by Blue Sky Research to provide a circular output beam. The diodes
were collimated and one diode at 663 nm (D1) was combined with the 685 nm diode (D2) using a
dichroic mirror (DM). The other 663 nm diode (D3) was polarization rotated using a half-wave
plate (WP) and the 3 beams were multiplexed with a polarizing beam splitter (PBS). This yielded
a collimated beam with a total power of 137 mW when each diode was driven by a current of 117
mA. The pump spot was focused to a minimum radius of 15 x 18 µm using a combination of a
R=-100 mm diverging lens (P1) and a R=76.3 mm antireflection coated achromatic lens (P2).
24-19
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
Two curved mirrors (M1 and M2, R=10 cm) were used to tightly focus the laser mode within the 5
mm long Brewster cut, 1.5 % doped Cr:LiSAF crystal (CR). The output coupler (OC) had a
transmission of 1% at 860 nm. Mirrors M3 and M4 were plane mirrors used to increase the arm
length. Initially, we optimized the laser in a standard x cavity without an MPC, SBR, or prisms,
obtaining a cw power of 28.5 mW. We then introduced the MPC, consisting of one plane 1.5”
diameter mirror (M6) and one 1.5” diameter R=4 m mirror (M7), separated by 2 meters. These
mirrors were DCMs that provided –42 fs2 group delay dispersion (GDD) per bounce around 860
nm. The beam passes through the MPC 16 times, resulting in a total added cavity length of 32m.
This MPC is designed to introduce a negligible amount of GDD, since each bounce on an MPC
mirror compensates the dispersion from the 2 meters of air between the mirrors. A flat mirror
(M5) and a curved R=4 m DCM (M8) are used to introduce and extract the beams from the MPC.
An extra fold for focusing onto the SBR was added, consisting of mirrors M9, M10, and the SBR;
this extra fold was set for a unity transformation of the q parameter. The SBR was similar to that
described in ref (Tsuda, Knox et al. 1996), with a reflectivity of 99.5% from 825-900 nm.
D3
M1
CR
D2
M2 P 2 P 1
D1
M3
PBS
M6 M
5
M7
M4
PR1
DM
OC
PR2
M10
M9
M8
OC
SBR
Figure 1. Experimental setup of the extended cavity Cr:LiSAF laser. Details are given in the
text.
Initial experiments were performed using prisms for dispersion compensation. Two fused silica
prisms (PR1 and PR2) separated by 50 cm were used for compensating the dispersion of the
crystal and tuning the dispersion operating point. The prism arm was 70 cm and the arm
including the MPC had an effective length of 90 cm. A standard high reflecting R=20 cm mirror
(M10) was used to tightly focus the laser mode onto the SBR. We obtained 43 fs pulses with
18.5 nm bandwidth at an 8.4 MHz repetition rate (figure 2). The average power was 5.5 mW in
this configuration, corresponding to a pulse energy of 0.66 nJ.
24-20
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
Intensity (arb.units)
1.0
0.8
0.6
0.4
0.2
0.0
FWHM 43 fs
-200
0
delay (fs)
200
(a)
Intensity (arb.units)
1.0
0.8
0.6
FWHM 18.5 nm
0.4
0.2
0.0
820
840
860
delay (fs)
880
900
(b)
Figure 2. Intensity autocorrelation (a) and spectrum (b) of the extended cavity Cr:LiSAF laser
with prisms. The repetition rate was 8.4 MHz and the pulse energy was 0.66 nJ.
Subsequently, we configured the cavity to operate without prisms. A total of ten bounces on
DCMs outside the MPC, providing -420 fs2 GDD, were used to compensate the dispersion of the
crystal and excess dispersion from the MPC and air in the cavity. M1 was replaced by a R=10
cm DCM. M3, M4, and M9 were replaced by flat DCMs. M10 was replaced by a R=30 cm
standard high reflecting mirror. PR1 and PR2 were removed. The arm lengths were 40 and 80
cm. All other components remained the same.
With this prismless setup, we obtained 39 fs pulses with 20 nm bandwidth (figure 3). The
average power was 6.5 mW at an 8.6 MHz repetition rate, corresponding to 0.75 nJ pulse energy.
We believe that the improved pulse energy in this configuration is primarily due to the lower
intracavity loss of the DCMs as compared to the prisms. The pulsewidth is most likely limited by
the bandwidth of the SBR in both configurations; we mode-locked the short cavity with no DCMs
and only prisms for dispersion compensation and obtained similar output spectra.
24-21
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
Intensity (arb.units)
RLE Progress Report 145
Intensity (arb.units)
1.0
0.8
0.6
0.4
0.2
0.0
FWHM 39 fs
1.0
0.8
0.6
0.4
0.2
0.0
FWHM 20 nm
820
-200
-100
0
delay (fs)
100
200
(a)
840
860
880
900
wavelength (nm)
920
(b)
Figure 3. Intensity autocorrelation (a) and spectrum (b) of pulses from the prismless extended
cavity Cr:LiSAF laser. The repetition rate was 8.6 MHz and the pulse energy was 0.75 nJ.
We measured the threshold and slope efficiencies for cw and mode-locked operation in the
prismless configuration as a function of different combinations of the three diodes. Thresholds for
both cw and mode-locked operation were typically between 69 and 81 mW, depending on the
particular diode combination tested. Mode-locked slope efficiencies were between 8 and 13%,
again depending on the combination of diodes. Diode D1 pumped the Cr:LiSAF crystal most
efficiently, as expected since its polarization and wavelength are most strongly absorbed. Diode
D3 was the least efficient, also expected since the absorption coefficient was a factor of 1.75
lower for this polarization.
In conclusion, a Cr:LiSAF laser incorporating an MPC and pumped by three single spatial mode
diodes has been demonstrated. Pulses as short as 39 fs with 20 nm bandwidth and 0.75 nJ
energy per pulse were generated at an 8.6 MHz repetition rate using only DCMs for dispersion
compensation. With prisms used for dispersion compensation, 43 fs pulses with 18.5 nm
bandwidth and 0.66 nJ pulse energy were generated at an 8.4 MHz repetition rate. This laser
has the potential to be significantly less expensive than conventional Ti:sapphire lasers due to the
low cost of its pump source. It could be useful for many applications requiring moderate pulse
energies and short pulse durations. Future work will include development of higher reflectivity
MPC mirrors and development of an SBR with higher reflectivity and a wider bandwidth.
References
1. Agate, B., B. Stormont, et al. (2002). "Simplified cavity designs for efficient and compact
femtosecond Cr:LiSAF lasers." Opt. Comm. 205: 207-213.
2. Cho, S. H., B. E. Bouma, et al. (1999). "Low-repetition-rate high-peak power Kerr-lens modelocked Ti:Al2O3 laser with a multiple-pass cavity." Opt. Lett. 24: 417-419.
3. Dymott, M. J. P. and A. I. Ferguson (1995). "18-fs-pulse generation from a diode-pumped selfmode-locked Cr:LiSAF laser". Conference on Lasers and Electro-Optics, CLEO, Baltimore.
4. Hopkins, J.-M., G. J. Valentine, et al. (2002). "Highly compact and efficient femtosecond
Cr:LiSAF lasers." IEEE J. Quant. Elect. 38(4): 360-368.
5. Tsuda, S., W. H. Knox, et al. (1996). "Mode-locking ultrafast solid-state lasers with saturable
Bragg reflectors." IEEE J. Sel. Top. Quant. Elect. 2: 454-464.
6. Uemura, S. and K. Torizuka (2002). Characteristics of 10-fs diode-pumped Kerr-lens modelocked Cr:LiSAF and Cr:LiSGAF lasers. Conference on Lasers and Electro-Optics.
7. Wagenblast, P. C., U. Morgner, et al. (2002). "Generation of sub-10-fs-pulses from a KerrLens mode locked Cr3+:LiCAF laser oscillator using third order dispersion compensating doublechirped mirrors." Opt. Lett. 27(19): 1726.
24-22
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
10 fs Diode Pumped Cr:LiCAF Laser
Sponsors
National Science Foundation - ECS-0119452
Project Staff
Felix Grawert, Phillip Wagenblast, Dr. Uwe Morgner, Professor Franz X. Kaertner
Cr3+-doped Colquiriite (Cr3+-:LiSAF, Cr3+-:LiSGaF, Cr3+-:LICAF) crystals are promising
materials for compact femtosecond laser sources. They show high quantum efficiency, broad
absorption bands in a wavelength range where high-brightness laser diodes are available, and
broad-band emission from 700 nm to 1000 nm, which supports pulses substantially shorter than
10 fs. Among them, Cr3+-:LiCAF has the highest quantum efficiency and the most favorable
thermal properties. Presently, Ti:sapphire lasers pumped by frequency-doubled solid-state lasers
are the only systems that deliver sub-10 fs pulses directly from the oscillator. Diode pumped sub10 fs laser systems are highly desirable as an inexpensive alternative for Ti:sapphire lasers in
spectroscopy, metrology, optical coherence tomography, and THz-generation. Initially. we used
the diffraction limited beam of a Ti:sapphire laser to generate 9 fs pulses with 220 mW average
power and 97 MHz repetition rate from Kerr-lens mode-locked Cr3+:LiCAF laser using broadband double-chirped mirrors for second- and third-order dispersion compensation [1]. Figure 1
shows the new setup using diode pumping. It is a standard z-folded resonator at 110 MHz
including a fused-silica prism sequence. The pump sources are two 500 mW laser diodes
(COHERENT S670C-500C) with an emitting area of 1x100 µm, where the large divergence of the
fast axis is collimated by a fiber lens. The two diodes are polarization multiplexed using a
polarizing beam splitter.
DCM
quartz prisms
O
2
DCM
R=75
PBS
f=100
f=50
OC
Pump mirror
R=75
Figure1: Set-up of the laser resonator and pumping scheme.
The total absorbed pump power amounts to 800 mW at maximum, and in cw mode of operation,
the laser emits up to 150 mW at a wavelength of 780 nm, and operates eight times above
threshold. Due to the low beam quality of the transverse multimode pump diodes Kerr-Lens
mode locking in diode pumped systems is usually achieved by hard apertures in the cavity [2,3].
Here, we found that by exploiting the gain guiding effect [4] we were able to demonstrate for the
first time a diode-pumped, soft-aperture KLM laser without internal aperture. In the mode-locked
state of operation, the laser emits 10 fs pulses with 40 mW of average power in a near-diffraction
limited beam. The mode-locked spectrum is shown in Fig. 2b. It extends from 750 to 900 nm and
shows an additional peak at 980nm, which is beyond the cavity bandwidth. The modulation in the
main part of the spectrum is due to the group delay oscillations of the mirrors. Assuming a flat
phase the transform-limited pulse duration would be 8.4 fs. The pulse train has been
characterized by spectral shearing interferometry [2] (SPIDER), a method which directly provides
the spectral phase of the pulses, and by Fourier transform the exact pulse shape.
24-23
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
SPIDER interferogram / arb.
RLE Progress Report 145
1
(a)
0.1
0.01
400
420
440
460
wavelength / nm
480
1.0
(b)
2
0.8
0
0.6
0.4
-2
0.2
-4
0.0
700
-6
750
800
850
900
950
wavelength / nm
1000
1.0
1050
10
(c)
'tFWHM = 9.3 fs
8
6
0.5
4
2
0
0.0
phase / radian
power / arb.
spectral phase / radian
spectral intensity / arb.
0.001
-2
-40
-20
0
time / fs
20
40
Figure 2: Results of the SPIDER characterization of the mode-locked pulses. a): sheared
interferogram, b): mode-locked spectrum and spectral phase, c): Reconstructed pulse and
temporal phase.
The interferogram of the upconverted, spectrally sheared pulses is shown in Fig. 2a on a
logarithmic scale. Over the whole spectral range including the peak at 980 nm, the pulses show
interference. The duration of the reconstructed pulse, shown in Figure 2.c is 9.3 fs (FWHM), and
prepulses are suppressed by one order of magnitude. This result represents a ten fold
improvement in output power compared to previous results [2].
References
1. P. Wagenblast, U. Morgner, F. Grawert, V. Scheuer, G. Angelow, M. J. Lederer, and F. X.
Kaertner, “Generation of sub-10-fs pulses from a Kerr-lens modelocked Cr3+:LiCAF laser
oscillator using third order dispersion compensating double chirped mirrors,” Opt. Lett. 27(19),
1726-9, 2002.
2. S. Uemura and K. Torizuka, “Development of a Diode-Pumped Kerr-Lens Mode-Locked
Cr:LiSAF Laser”, IEEE JQE 39(1), 68 (2003).
3. K. M. Gäbel, P. Rußbüldt, R. Lebert, and A. Valster, “Diode pumped Cr3+:LiCAF fs-Laser,”
Opt. Comm. 157, 327, (1998).
24-24
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
Spectral broadening in tapered fiber and a high numerical aperture fiber
using a femtosecond Nd:Glass Laser
Sponsors
Air Force Office of Scientific Research (MFEL) - F49620-01-1-0186
Air Force Office of Scientific Research - F49620-98-01-0084
National Science Foundation - ECS-019452
National Institute of Health - NIH-5-R01-CA75289-04
National Institute of Health - NIH-2-R01 EY11289-15
Project Staff
Andrew M. Kowalevicz, Rohit Prasankumar, Tony H. Ko, Alphan Sennaroglu, Thomas Schibli,
Professor Franz X. Kaertner, Professor Erich P. Ippen, Professor James G. Fujimoto
High nonlinearity, air-silica microstructure fibers [1] or tapered fibers [2] can generate an
extremely broadband continuum using low energy femtosecond pulses. The anomalous
dispersion characteristics of the fibers, which shift the zero dispersion to shorter wavelengths,
and the small core diameters, which provide tight mode confinement, help exploit the high
nonlinearities of the fiber.
We have demonstrated a new low coherence light source using a compact Nd:Glass
femtosecond laser spectrally broadened in a tapered single mode fiber. Our setup uses a
compact diode pumped femtosecond Nd:Glass laser (High Q Laser Production GmbH) which
generates pulses with 110-150 fs duration and 150 mW average power at 75 MHz repetition rate
and 1.06 µm wavelength with a 12 nm bandwidth. The Nd:Glass is pumped by two 1 W diode
laser diodes. The Nd:Glass laser is soliton modelocked [3] using a SESAM [4,5] for self starting
and intracavity prisms for dispersion compensation. The laser pulses are coupled into a single
mode fiber (Corning SMF-28). The fiber was tapered by stretching in a flame so that after a short
length (~20 mm) of normal 125 µm diameter fiber, the fiber tapers down to a uniform waist with a
diameter of 2 µm and a length of 90 mm, before tapering up again to normal fiber. This thin
uniform waist enables the efficient generation of continuum [2]. The output of the tapered fiber is
fusion spliced to a 10 m length of dispersion shifted fiber (zero dispersion at longer wavelengths)
to reduce parasitic contributions from four wave mixing.
Figure 1 (a) shows a typical continuum generated by the tapered fiber with an average power of
50 mW. The spectrum is centered at 1.3 µm with a bandwidth of 132 nm. The continuum is
asymmetrically shifted toward longer wavelengths. The shift in the spectrum may be the result of
Raman effects and the soliton self frequency shift as well as other mechanisms.
24-25
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
1.0
(a)
(b)
0.8
0.8
Intensity [a.u.]
Intensity [a.u.]
1.0
0.6
132 nm
0.4
0.2
0.6
123 nm
0.4
0.2
0.0
0.0
1000
1100
1200
1300
1400
1500
Wavelength [nm]
800
900
1000
1100
1200
1300
Wavelength [nm]
Figure 1. (a) The optical spectrum of the continuum generated in a tapered fiber, and (b)
optical spectrum of the continuum generated in a high numerical aperture fiber.
Based on the same principle than the tapered fiber, we have demonstrated supercontinuum
generation in a high numerical aperture fiber. The light from the Nd:glass laser is coupled into a 2
meter length of commercially available ultrahigh numerical aperture (NA) single mode fiber. The
germanium doped, 2.5 Pm core provides enhanced nonlinear effects and efficient continuum
generation.
A typical optical spectrum of the continuum at the output of the ultrahigh NA fiber is presented in
Figure 1b. The broadening is mainly due to the self phase modulation effect. A slight shift of the
central wavelength from 1064 nm to 1080 nm is observed, which may results from Raman effect.
With 90 mW of average output power, a bandwidth of 123 nm could be generated
These compact and portable light sources are well suited for ultrahigh resolution optical
coherence tomography. Axial resolution in air of 5.6 µm and 4.2 µm are theoretically achievable
with the tapered fiber and the high numerical aperture fiber, respectively.
References
1. J. K. Ranka, R. S. Windeler, and A. J. Stentz, “Visible continuum generation in air silica
microstructure optical f ibers with anomalous dispersion at 800nm,” Optics Letters, vol. 25,
pp. 25-27, 2000.
2. T. A. Birks, W. J. Wadsworth, and P. St. J. Russell, “Supercontinuum generation in tapered
fibers,” Optics Letters, vol. 25, pp. 1415-1417, 2000.
3. F. X. Kärtner, I. D. Jung, and U. Keller, “Soliton Modelocking with Saturable Absorbers,”
Special Issue on Ultrafast Electronics, Photonics and Optoelectronics, IEEE J. Selected
Topics in Quantum Electronics (JSTQE), vol. 2, pp. 540-556, 1996.
4. D. Kopf, F. X. Kärtner, K. J. Weingarten, and U. Keller, “Diode-pumped modelocked Nd:glass
lasers using an A-FPSA,” Optics Lett., vol. 20, pp. 1169-1171, 1995.
5. U. Keller, K. J. Weingarten, F. X. Kärtner, D. Kopf, B. Braun, I. D. Jung, R. Fluck, C.
Hönninger, N. Matuschek, and J. Aus der Au, “Semiconductor saturable absorber mirros
(SESAMs) for femtosecond to nanosecond pulse generation in solid-state lasers,” Special
issue on Ultrafast Electronics, Photonics and Optoelectronics, IEEE J. Selected Topics in
Quantum Electronics (JSTQE), vol. 2, pp. 435-453, 1996.
24-26
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
1Pm Stretched-Pulse Laser with Microstructured Fiber for Dispersion
Compensation
Sponsors
U.S. Air Force – Office of Scientific Research - F49620-01-1-0084
Project Staff
J. T. Gopinath, Dr. K. S. Abedin, Dr. M. E. Grein, and Professor E. P. Ippen
There is considerable interest in femtosecond pulse generation at 1 Pm for medical imaging and
procedures, spectroscopy and microscopy. At 1 Pm, Yb-doped silica fiber has excellent
conversion efficiency, broad gain-bandwidth, can be pumped by telecomm laser diodes at 975
nm, and should produce short high-energy pulses. However, in order to produce ultrashort
pulses, one must compensate the normal dispersion of the Yb fiber. Unfortunately, other
conventional fiber, such as single mode fiber at 1550 nm, also have normal GVD at this
wavelength. Previous femtosecond Yb fiber lasers have used intracavity prisms or grating pairs
for dispersion compensation. Because these elements are lossy, the output pulsewidth and
power of the laser are limited. Thus, it is desirable to replace these elements with low loss fiber.
Photonic crystal and microstructure fiber can provide anomalous dispersion at 1 Pm.
Microstructure fiber, fiber with a pattern of air holes, that owes its guiding properties purely due to
index contrast, took the world by surprise a few years ago, with its high profile application to
frequency metrology.
This fiber can be used for many nonlinear processes including
supercontinuum generation (applications: frequency metrology, medical imaging etc.), four-wave
mixing [1], and high harmonic generation. However, the loss of these fibers, which can be orders
of magnitude higher than the 0.2 dB/km loss of conventional fiber, is a major drawback and
makes them generally unsuitable for use in lasers. We are collaborating with Dr. Benjamin
Eggleton, Dr. Robert Windeler, and Charles Kerbage at OFS-FITEL, who make tapered air-silica
microstructure fiber (TASMF), with losses as low as 0.3 dB/taper (18 cm) [2]. The fiber can be
spliced reliably with relatively low losses of 0.1 dB/splice . Untapered, this fiber behaves similarly
to SMF; tapered, it has a nonlinearity an order of magnitude higher and a unique dispersion
profile (see Figure 1).
a)
b)
Figure 1: a) Cross-section of microstructure fiber. b) Calculated group index and GVD vs.
wavelength for the tapered air-silica microstructure fiber versus the taper diameter. We are using
tapers of ~1.44 Pm diameter in the laser.
24-27
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
We plan to use this fiber to generatefemtosecond pulses with the Yb fiber laser. One group has
already demonstrated 100 fs pulses with photonic crystal fiber in this system [3], Our goal is to
improve on this result by extending the laser operation into the stretched-pulse regime. Figure 2
shows the schematic of the laser we are building.
Figure 2: Schematic of laser
Initial results, in a ring configuration, incorporating a fiber WDM, have shown that very broadband
output pulses can be generated. The free space configuration shown above should provide more
precise dispersion balancing and lead to shorter pulses. Two tapers will be used in the laser, in
conjunction with a saturable absorber for self-starting modelocked operation.
References
1.
K. S. Abedin, J. T. Gopinath, E. P. Ippen, C. E. Kerbage, R. S. Windeler and B. J. Eggleton,
"Highly nondegenerate femtosecond four-wave mixing in tapered microstructure fiber,"
Applied Physics Letters, 81(8): 1384-1387 (2002).
2.
J. K. Chandalia, B. J. Eggleton, R. S. Windeler, S. G. Kosinski, X. Liu and C. Xu, "Adiabatic
coupling in tapered air-silica microstructured optical fiber," IEEE Photonics Technology
Letters, 13(52-54 (2001).
3.
H. Lim, F. O. Ilday and F. W. Wise, "Femtosecond ytterbium fiber laser with photonic crystal
fiber for dispersion control," Optics Express, 10(25): 1497-1502 (2002).
24-28
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
Timing Jitter Studies in a Passively Modelocked Regeneratively
Synchronized Fiber Laser
Sponsors
U.S. Air Force – Office of Scientific Research - F49620-01-1-0084
DARPA – Defense Advanced Research Projects Agency - F49620-96-01266
U. S. Navy - Office of Naval Research
Project Staff
Jason W. Sickler, Matthew E. Grein, Leaf A. Jiang, Professor Erich P. Ippen, Professor Hermann
A. Haus
Currently, a strong demand exists for modelocked lasers that produce sub-picosecond pulses at
high repetition rates. Such lasers would be extremely useful as clocks for high frequency, high
resolution analog-to-digital optical sampling, and as sources for high-speed time-division
multiplexed optical communications systems. The timing noise, or timing jitter, of these lasers
often limits the performance of systems in which these lasers would be used [1], thus
understanding and reducing the timing jitter in these lasers is important.
A primary candidate for satisfying the desire for such lasers is harmonically modelocked fiber
lasers. Previous work on modelocked fiber lasers sought to generate short pulses at high
repetition rates.
This includes work using polarization additive-pulse modelocking (P-APM)
schemes for short pulse generation, combined with regenerative feedback as a means to
harmonically modelock and thus increase the repetition rate [2, 3, 4]. In this work, we seek to
reduce the timing jitter of an harmonically modelocked regeneratively synchronized fiber laser
using an intracavity fiber loop.
The nature of timing jitter in
harmonically
modelocked
lasers can be described in
both the time and frequency
domain. Both domains, and
the relationship between
them, are illustrative. In the
time
domain,
a
fundamentally modelocked
laser produces a pulse train
where
all the
pulses
originate from the same
intracavity pulse. Because
the noise of the intracavity
pulse at one time is partially
correlated to the noise of
that same intracavity pulse
at another time, the noise of
the output pulses tends to
be correlated.
Because
Fig. 1. Theoretical plot of the spectral noise power density
noise
is
added
(via
of a fundamentally modelocked laser. [5]
spontaneous emission, and
other classical sources) and
removed (via filtering, and other loss mechanisms) with each round trip, the degree of correlation
between the intracavity pulse at two different times generally decreases when the times
considered are further apart. One expects, then, that timing jitter in a fundamentally modelocked
laser appears primarily at low frequencies, as the theory shown in Fig. 1 predicts.
24-29
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
The timing jitter in an
harmonically modelocked
laser differs from that of a
fundamentally modelocked
laser. The timing jitter of
any two pulses in the output
pulse train may or may not
be correlated. In general,
as long as effects that
would
lead
to
pulse
interaction can be ignored,
such as those resulting
from gain recovery, the
timing jitter of output pulses
that originate from different
intracavity pulses will not be
correlated. The timing jitter
of those output pulses
originating from the same
intracavity pulse will be tend
Fig. 2. Theoretical plot of the spectral noise power density
to be correlated, much like
of a harmonically modelocked laser for phase modulation
output
pulses
of
a
and filtering [5].
fundamentally modelocked
laser. Thus, for a laser
harmonically modelocked with harmonic number, N, one can think of N interleaved fundamentally
modelocked laser output pulse trains. The pulses within each interleaved pulse train tend to jitter
together, but the pulses contained in different interleaved trains are independent. The possibility
of pulse-patterning, using this heuristic, becomes clear. The spectral noise power density will
show timing jitter at low frequencies, as well as at harmonics of the fundamental frequency of the
laser. The theoretical plot in Fig. 2 shows the low frequency noise pedestal, as well as N-1 noise
pedestals occurring at harmonics of the fundamental frequency.
We are attempting to use an
intracavity fiber loop to
correlate the noise of the
intracavity pulses, in order to
effect and hopefully reduce
the timing jitter of the output
pulses.
The fiber loop
functions as a Gire-Tournois
interferometer.
When the
fundamental frequency of the
fiber loop is at a harmonic of
the cavity repetition rate,
pulses exiting the fiber loop
will overlap pulses passing by
the loop. In the shuffling of
photons among the intracavity
pulses we hope to correlate
their noise. The fiber loop will
be placed in a regeneratively
Fig. 3. Regeneratively synchronized harmonically
synchronized
P-APM
modelocked fiber laser.
harmonically
modelocked
fiber laser, shown in Fig. 3,
harmonically modelocked at 1GHz (N~70). [2, 4] Comparisons between the spectral noise power
24-30
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
densities, using a frequency discrimination technique [6, 7] shown in Fig. 4, as well as optical
correlations, will be made.
Interesting to note is that
the use of the frequency
discriminator technique to
measure timing jitter is not
typically done.
Rather,
timing jitter measurements
are more commonly made
in reference to a stable local
oscillator
[8].
The
advantage of the frequency
discriminator technique is
that a low-noise external
oscillator is no longer
Fig. 4. Diagram of a frequency discriminator setup.
required. When measuring
the timing jitter of extremely
low noise systems, this is a valuable advantage.
Results relevant to this work appear in the literature. These include experiments using intracavity
Fabry-Perot etalons [9, 10, 11] and fiber loops [12]. To our knowledge, however, the nature of
the fiber loop’s effect on noise has not been fully explored.
We predict two general possible results from this work. The first is that the correlation of pulses
will shift timing jitter power to lower frequencies (i.e. suppress the supermode). The pulses of the
pulse train will be “tied together” so that they jitter together, but the total timing jitter, the integral
of the noise power spectrum out to the Nyquist frequency, will not be reduced. The second
possibility is that the correlation of pulses will shift timing jitter power to lower frequencies, as well
as reduce the total timing jitter. The pulses of the pulse train will be “tied together” so that they
jitter together, and the “inertia” of the pulse train will reduce the total timing jitter. In either case,
successful correlation of the pulses in the pulse train should result in a noise spectrum that
qualitatively approaches that of a fundamentally modelocked laser.
References
1. P. W. Juodawlkis, et. al, “Optically Sampled Analog-to-Digital Converters,” IEEE Transactions
On Microwave Theory and Techniques 49(10): 1840-1853 (2001).
2. M. Margalit, et al, “Harmonic Mode-Locking Using Regenerative Phase Modulation,” IEEE
Photonics Technology Letters 10(3): 337-339 (1998).
3. C. X. Yu, et. al, “Noise of a Regeneratively Synchronized GHz Passively Modelocked Fiber
Laser,” in Conference of Laser and Electro-Optics - Europe, IEEE, (Nice, France), pp. 1,
IEEE, 2000.
4. C. X. Yu, et. al, “A GHz Regeneratively Synchronized Passively Mode-locked Fiber Laser for
Spectrum Generation In The 1.5Pm Region,” in Conference of Laser and Electro-Optics, OSA
Technical Digest – Postconference Edition, (San Francisco, CA), pp. 80, Optical Society of
America, 2000.
5. M. E. Grein, Noise and Stability of Actively Modelocked Fiber Lasers, Ph.D. Thesis,
Department of Electrical Engineering and Computer Science, MIT, 2002.
24-31
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
6. D. Scherer, “The Art of Phase Noise Measurement,” RF & Microwave Measurement
Symposium and Exhibition (August 1985).
7. T. Decker and B. Temple, “Choosing A Phase Noise Measurement Technique: Concepts
and Implementation,” RF & Microwave Measurement Symposium and Exhibition (?).
8. R. P. Scott, et. al, “High-Dynamic-Range Laser Amplitude and Phase Noise Measurement
Techniques” IEEE Journal of Selected Topics in Quantum Electronics 7(4): 641-655 (2001).
9. C. M. DePriest, et. al, “Ultralow noise and supermode suppression in an actively mode-locked
external-cavity semiconductor diode ring laser,” Optics Letters 27(9): 719-721 (2002).
10. J. S. Wey, et. al, “Performance Characterizaton of a Harmonically Mode-Locked Erbium Fiber
Ring Laser” IEEE Photonics Technology Letters 7(2): 152-154 (1995).
11. J. S. Wey, et. al, “Active Harmonic Modelocking of an Erbium Fiber Laser with Intracavity
Fabry-Perot Filters” IEEE Journal of Lightwave Technology 15(7): 1171-1180 (1997).
12. E. Yoshida, et. al, “Laser diode-pumped femtosecond erbium-doped fiber laser with a subring cavity for repetition rate control,” Applied Physics Letters 60(8): 932-934 (1992).
13. [13] M. E. Grein, et. al, “Experimental Observation of Quantum-Limited Timing Jitter in an
Active, Harmonically Modelocked Fiber Laser”, in Conference of Laser and Electro-Optics,
OSA Technical Digest, (Washington, D.C.), pp. 561-562, Optical Society of America, 2002.
14. M. E. Grein, et. al, Active Harmonically Modelocked Fiber Lasers, RLE 2001(Cambridge:
MIT Research Laboratory of Electronics, 2001).
15. H. A. Haus, et. al, “Noise of Mode-Locked Lasers,” IEEE Journal of Quantum Electronics.
29(3): 983-996 (1993).
16. L. A. Jiang, Ultralow-Noise Modelocked Lasers, Ph.D. Thesis, Department of Electrical
Engineering and Computer Science, MIT, 2001.
17. L. A. Jiang, et. al, Noise in Harmonically Modelocked Lasers, RLE 2001(Cambridge: MIT
Research Laboratory of Electronics, 2001).
18. F. Rana, et. al, “Characterization of the noise and correlations in harmonically mode-locked
lasers,” J. Opt. Soc. Am. B 19(11): 2609-2621 (2002).
19. D. von der Linde, “Characterization of the Noise in Continuously Operating Mode-Locked
Lasers,” Applied Physics B. 39: 201-217 (1986).
20. C. X. Yu, Soliton Squeezing in Optical Fibers, Ph.D. Thesis, Department of Electrical
Engineering and Computer Science, MIT, 2000.
24-32
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
Timing Jitter and Correlations in Harmonically Modelocked Fiber Lasers
Sponsors
U.S. Air Force – Office of Scientific Research - F49620-01-1-0084
DARPA – Defense Advanced Research Projects Agency - F49620-96-01266
U. S. Navy - Office of Naval Research
Project Staff
Matthew E. Grein, Leaf A. Jiang, Jason Sickler, Professor Erich Ippen, Professor Hermann A.
Haus
Actively modelocked fiber lasers can generate streams of transform-limited picosecond pulses
locked to an external frequency reference at GHz repetition rates with low amplitude and timing
jitter. Such a source can potentially be used for optical sampling in precision, high-speed analogto-digital converters and as optical transmitters in a high-speed time-division-multiplexed
transmission system. Much of the low-noise performance of fiber lasers—compared with
semiconductor lasers—arises due to the much larger intracavity pulse energy and larger signalto-noise ratio. The goal of this work has been to study the timing jitter in actively modelocked
fiber lasers. Pursuant to that goal, we have developed a theory for the quantum-limited timing
jitter, identified the characteristic retiming constants that govern the timing jitter for the case of
amplitude (AM) and phase (PM) modulation, developed a timing-jitter measurement scheme
using a balanced microwave homodyne detection scheme with high dynamic range, and built an
actively modelocked fiber laser that produces picosecond pulses at 10 GHz whose timing jitter is
quantum limited. The particular goal of the present work is to understand the noise and
correlations particular to harmonically modelocked fiber lasers (and harmonically modelocked
lasers generally).
Due to the low gain per unit length of rare-earth doped fiber (e.g., erbium and erbium-ytterbium),
fiber lasers are generally very long. High repetition rates are achieved by modelocking the cavity
at some harmonic, N, of the laser cavity frequency, resulting in N pulses per round trip. This
results in a timing jitter spectrum that is more complicated than that for a fundamentally
modelocked laser and depends strongly on the pulse-to-pulse correlations. To date, much of the
published work on measuring and interpreting the timing jitter has not properly taken these
correlations into account. In this work we have shown how the pulse-to-pulse correlations are
related to the total timing jitter, developed a theoretical model, and demonstrated experimental
confirmation of the model.
The fiber laser setup—shown in Fig. 1--is arranged in a sigma-type configuration in which the
linear portion is composed of non-polarization-maintaining elements. The amplifying medium is
an Er:Yb double-clad fiber side-pumped with a multimode 980 nm laser diode. The sigma laser
works as follows: a pulse exiting the polarizing beam splitter (PBS) from the ring depolarizes due
to environmentally-induced birefringence in the linear segment. A faraday rotator at the end of
the linear segment ensures that the backward-propagating pulse travels along the orthogonal
polarization axis with respect to the forward-traveling pulse. In this way, the polarization effects in
the forward and backward propagating directions are averaged out so that the pulse arrives at the
polarization beam splitter again with a linear polarization.
24-33
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
.
Figure 1. Sigma laser configuration. OSC is an external microwave frequency reference, G
microwave amplifier, F optical bandpass filter, HWP and QWP half- and quarter-wave plates,
PBS polarizing beam splitter, DSF and DCF dispersion-shifted and dispersion-compensating
fiber, EYDFA erbium-ytterbium co-doped fiber amplifier, FR faraday rotator, AS aspheric lens,
MR dielectric mirror.
The laser produces transform-limited, hyperbolic-secant pulses at 1.5 Pm with repetition rates
upwards of 10 GHz with pulsewidths from 900 fs to 2 ps, depending on the optical filtering and
pump power. The suppression of supermodes in the RF spectrum is typically greater than 70
dB, indicative of excellent laser stability. A typical autocorrelation trace and microwave RF plot
are shown in Fig. 2. The laser is locked to the external microwave frequency reference by
Figure 2. Background-free autocorrelation trace showing a fit to an hyperbolic secant with a
pulsewidth of 1.55 ps, and RF spectrum of the directly-detected photocurrent, showing greater
than 70 dB of supermode suppression.
stabilizing the cavity length using a phase-locked loop (PLL) consisting of a microwave phase
detector, control electronics, and a fiber-wound piezoelectric transducer.
The measurement of the laser timing jitter is achieved using a residual phase-noise technique
that is typically used to compare the relative phase noise between two microwave frequency
sources, described previously in Refs 1 and 2.
24-34
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
Figure 3. Timing jitter spectrum for the case of mostly AM. Upper solid curve, data; lower solid
curve, measurement noise floor; dotted curve, theory.
A typical phase-noise spectrum L(f) is shown in Fig. 3 for the case where the modulation is set to
mostly AM. The characteristic spectrum of the jitter at low frequencies (f < 250 kHz) is well
understood [3]. However, the significance and a complete explanation for the features appearing
at harmonics of the fundamental round-trip frequency—referred to as supermodes--have not
been explained in the literature. We have shown that the supermodes are, in fact, aliased
versions of the baseband mode—as shown in Fig. 4—and contribute to the overall timing jitter by
a factor of ¥N, where N is the harmonic number. This is consistent with the explanation that
pulses within the laser cavity are uncorrelated with each other. We confirmed this by measuring
the timing jitter by comparing the accumulated timing jitter from one pulse to the next using optical
cross-correlation—as shown in Fig. 5—and comparing it with the integrated timing jitter. The
width of the optical cross correlations did not change as the pulse was delayed from one to the
next, and the accumulated timing jitter agreed closely with that computed by integrating the timing
jitter spectrum of Fig. 3. We are currently working on a scheme to correlate each of the pulses in
the laser cavity (positively correlated) using an etalon or GTI to reduce the overall timing jitter,
which we believe should result in timing jitter reduction by a factor of ¥N.
Figure 4. Overlap of the first, tenth, and twentieth harmonics of the timing jitter spectrum.
24-35
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
Figure 5. Background-free autocorrelation and crosscorrelation of adjacent pulses traces
References
1. M. E. Grein, L. A. Jiang, H. A. Haus, and E. P. Ippen, C. McNeilage, J. H. Searls, and R.
S. Windeler, “Observation of quantum-limited timing jitter in an active, harmonically
mode-locked fiber laser,” Opt. Let. 27, 957-959 (2002).
2. W. F. Walls, “Cross-correlation phase noise measurements,” In IEEE Frequency Control
Symposium, (Institute of Electrical and Electronics Engineers, New York, 1992), pp. 257260.
3. M. E. Grein, L. A. Jiang, Y. Chen, H A. Haus, and E. P. Ippen, “Timing restoration
dynamics in an actively mode-locked fiber ring laser”, Opt. Lett. 24(23): 1687-1689
(1999).
Publications
Journal Articles, Published
M. E. Grein, L. A. Jiang, H. A. Haus, and E. P. Ippen, C. McNeilage, J. H. Searls, and R.
S. Windeler, “Observation of quantum-limited timing jitter in an active, harmonically
mode-locked fiber laser,” Opt. Let. 27, 957-959 (2002).
L. A. Jiang, M. E. Grein, J. K. Chandalia, E. P. Ippen, and H. Yokoyama, “Retiming
dynamics of modelocked semiconductor lasers”, Electron. Lett. 38, 1446-1447 (2002).
F. Rana, H. L. T. Lee, M. E. Grein, L. A. Jiang, and R. J. Ram, “Characterization of the
noise and correlations in harmonically mode-locked lasers,” J. Opt. Soc. Am. B 19, 26092621 (2002).
Journal Articles, Submittted for Publication
M. E. Grein, H. A. Haus, Y. Chen, and E. P. Ippen, “Quantum-limited timing jitter in
actively mode-locked lasers,” submitted to IEEE J. Quantum Electronics.
24-36
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
Meeting Papers
Published
M. E. Grein, L. A. Jiang, H. A. Haus, E. P. Ippen, C. McNeilage, and J. H. Searls,
“Observation of quantum-limited timing jitter in an actively modelocked laser,” in OSA
Trends in Optics and Photonics (TOPS) Vol. 73, Conference on Lasers and ElectroOptics, OSA Technical Digest, Postconference Edition (Optical Society of America,
Washington DC, 2002).
L. A. Jiang, M. E. Grein, H. A. Haus, E. P. Ippen, “Experimental demonstration of a
timing-jitter eater,” in OSA Trends in Optics and Photonics (TOPS) Vol. 73, Conference
on Lasers and Electro-Optics, OSA Technical Digest, Postconference Edition (Optical
Society of America, Washington DC, 2002).
24-37
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
Timing Jitter Reduction Using a Timing-Jitter Eater
Sponsors
U. S. Air Force – Office of Scientific Research - F49620-01-1-0084
U. S. Navy – Office of Naval Research
Project Staff
Leaf A. Jiang, Matthew E. Grein, Professor Erich Ippen, Professor Hermann A. Haus
Removing the timing jitter from a train of pulses can be useful for optical communications and
optical sampling. In long-haul optical soliton communication systems, where the transmitted
pulses propagate through many amplifiers, the timing jitter of the pulses can become significant.
Retiming the data pulses before detection can lead to improved bit-error rates. For optical
sampling applications, the timing jitter of the pulse train puts an ultimate limit on the speed and
resolution of the system.
There have been many approaches to retiming pulses using active modulation [1-3], including the
use of phase modulation and dispersion [4-6]. The novelty of the present work is the effective
application of active retiming to nonsolitonic pulses with a single phase modulator, which is
difficult in two respects: (1) short nonsolitonic pulses disperse easily in fiber, thereby putting a
severe limitation on retiming, whereas solitons are immune to dispersive broadening; and (2)
frequency-to-timing noise conversion puts a limit on the amount of timing jitter reduction. We
show how to overcome these problems by introducing prechirp fiber.
The experimental setup of the timing jitter eater is shown in Figure 1. A train of mistimed pulses
enters the eater from the top left-hand corner. The first pulse is slightly delayed, and the third
pulse is slightly advanced. The jittery pulse train propagates through prechirp fiber.
Figure 1. Timing-jitter eater experimental setup.
Next, the pulses enter a phase modulator that is sinusoidally modulated at the pulse repetition
rate. The phase of the modulation signal is set so that the desired pulse positions are aligned to
the peak of the sinusoidal phase modulation. A pulse that arrives early is redshifted, and a pulse
that arrives late is blueshifted. The pulses from the phase modulator then propagate through the
anomalous dispersion postchirp fiber in which blueshifted light travels faster than redshifted light.
If the length of the postchirp fiber is chosen correctly, the output yields jitter-reduced pulses. Note
that the eater will also work if the phase modulation is delayed by 180° and the sign of the
postchirp fiber is flipped. At the output of the timing jitter eater, each mistimed pulse has a slightly
24-38
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
different color, and hence the timing jitter reduction comes at the cost of increased wavelength
jitter.
Fig. 2 shows the jitter reduction revealed by the directly detected spectrum of the photocurrent
displayed on an rf spectrum analyzer. The reduction of the pedestals surrounding the 10.0 GHz
carrier is indicative of timing jitter reduction.
Figure 2. Spectrum of the first harmonic before and after the timing-jitter eater
There are two additional complications: (1) the output pulse width may be significantly larger than
the input pulse width due to dispersive broadening, and (2) the input wavelength jitter can turn
into timing jitter at the output of the system through dispersion. The prechirp fiber ameliorates
these two problems since it has the opposite dispersion of the postchirp fiber, so that the total
dispersion of the timing jitter eater is close to zero.
Figure 3 shows theoretical calculations of the best timing jitter reduction for a given input and
output pulse width made using a single phase modulator. We computed the amount of timing
jitter reduction by propagating a mistimed Gaussian pulse through the timing jitter eater and then
comparing the initial and final mistimed positions. The timing jitter reduction (defined as the rms
output timing jitter divided by the input timing jitter) is plotted in Fig. 3 on a decibel scale and
where the input pulse is assumed to be transform limited. The solid curves show the best that
can be done with both prechirp and postchirp fibers, and the dashed curves show the best
reduction possible when only postchirp fiber is used. Figure 3 shows that (1) it is difficult to retime
short pulses since their large spectra lead to significant dispersive broadening; (2) the prechirp
fiber improves performance when the output pulse width is close to the input pulse width; (3) the
output pulses can be shorter than the input pulses, as in the case of 10-ps input pulse with ±3S
phase modulation, since the phase modulator adds spectrum to the input pulses; and (4) the
dashed and solid curves come together at large output pulse widths, since the final pulse width
and noise reduction is insensitive to the amount of prechirp dispersion when the input and output
timing jitter are the same.
24-39
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
Figure 3. Theoretical calculations of the best possible jitter reduction for a given input and output
pulse width made using one phase modulator with M = 0.6S and M =3S. The input pulse is
transform limited, but the output pulse is not necessarily transform limited. Solid curves, best
noise reduction with prechirp and postchirp fiber; dashed lines, best noise reduction with only
postchirp fiber. The star corresponds to the experimental results in Fig. 2.
References
1. M. Nakazawa, E. Yamada, H. Kubota, and K. Suzuki, Electron. Lett. 27, 1270 (1991).
2. N. J. Smith, K. J. Blow, W. J. Firth, and K. Smith, Opt. Commun. 102, 324 (1993).
3. J. P. King, I. Hardcastle, and H. J. Harvey, “Method and apparatus for conditioning optical
solitons,” U.S. patent 6,130,767 (October 10, 2000).
4. M. E. Grein, L. A. Jiang, E. P. Ippen, and H. A.Haus, Opt. Express 8, 664 (2001),
http://www.opticsexpress.org.
5. L. Mollenauer and C. Xu, in Conference on Lasers and Electro-Optics, Vol. 73 of OSA
Trends in Optics and Photonics Series (Optical Society of America, Washington,D.C.,
2002), postedeadline paper CPDB1-1.
6. L. A. Jiang, M. E. Grein, E. P. Ippen, and H. A. Haus, in Conference on Lasers and
Electro-Optics, Vol. 73 of OSA Trends in Optics and Photonics Series (Optical Society of
America, Washington, D.C., 2002), postconference edition, pp. 164–165.
Publications
Journal Articles, Published
L. A. Jiang ,M. E. Grein, H. A. Haus, and E. P. Ippen, “Timing jitter eater for optical pulse trains”,
Opt. Lett. 28, 78-80 (2003).
Meeting Papers, Published
L. A. Jiang, M. E. Grein, H. A. Haus, E. P. Ippen, “Experimental demonstration of a timing-jitter
eater,” in OSA Trends in Optics and Photonics (TOPS) Vol. 73, Conference on Lasers and
Electro-Optics, OSA Technical Digest, Postconference Edition (Optical Society of America,
Washington DC, 2002).
24-40
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
Timing Jitter Studies in Hybridly Modelocked Semiconductor Lasers
Sponsors
U. S. Air Force – Office of Scientific Research
Grant F49620-01-1-0084
U. S. Navy – Office of Naval Research
Project Staff
Leaf A. Jiang, Matthew E. Grein, Professor Erich Ippen, Professor Hermann A. Haus
High-speed optical sampling systems and optical time-division multiplexed transmission systems
have stringent timing-jitter requirements. Timing jitter less than 500 fs is required for optical timedivision multiplexed transmission at 160 Gbits/s (timing jitter should be less than 10% of the bit
period). An optical source with less than 50 fs of timing jitter (10 Hz to 5 GHz) would be needed
for a sampler with 8-bit quantization at a sampling rate of 10 Gsamples/s. External cavity
modelocked laser diodes (EC-MLLD) can produce high-quality picosecond pulses at GHz
repetition rates, and thus could potentially be used as a pulse source for the aforementioned
applications. Beyond the state-of-the-art performance aspects, in this work we investigated the
quantum-limited timing jitter of an EC-MLLD and the effects of device optimization
The EC-MLLD used in our noise measurements is shown in Fig. 1. The semiconductor chip
consisted of two sections [1] a 50-Pm saturable-absorber section and a 500-Pm gain section. The
gain section was biased at 65 mA, and the saturable-absorber section was reverse biased at 21.4
V. The saturable absorber was modulated externally with a low-noise Poseidon Scientific Shoe
Box Oscillator (SBO) at 9 GHz, amplified to give 24 dBm of rf power. The cavity length was set so
that the round-trip frequency matched the rf drive frequency (the laser was fundamentally mode
locked).
The residual phase-noise measurement setup, shown in Fig. 2, consists of the SBO microwave
oscillator, the device under test, and a delay arm for setting the microwave phase into quadrature
so that the output is proportional to the phase difference between arms. The device under test
consists of the MLLD and the photodetector for the noise measurement or an attenuator with
equivalent loss for the noise-floor measurement. The mixer output is then observed on a vector
signal analyzer for low-frequency offsets (0–10 MHz) and on a rf spectrum analyzer for high
offsets (10 MHz–4.5 GHz). The delays through both arms were closely matched so that the
oscillator noise was suppressed. The rf spectrum analyzer measurement was calibrated by
measurement of the noise down to 1 MHz, where the noise overlapped the frequency range of
the vector signal analyzer measurement.
The single-sided phase noise is shown in Fig. 3a. Spurs that are due to 60-Hz wall current and
harmonics thereof are apparent. Vibration spurs at harmonics of 20 and 55 Hz that are due to ac
fans and other ambient vibrations are also visible. At higher offsets, radio station signals appear
in the 10–100-MHz decade. A spur that is due to wireless telephones is visible in the 100-MHz–1GHz decade. In addition, contributions of noise that are due to the fast gain dynamics in the
semiconductor laser at the relaxation oscillation frequency are also evident in this decade [2]. In
the 100-kHz–1-MHz decade, the spurs in the laser noise are due to the switching power supply in
the ILX-3207B current source (ILX Lightwave). For offsets less than 1 kHz, the laser noise
increases by approximately 13 dB/decade. A direct measurement of the noise of the voltage
supply to the saturable absorber and current source reveals an increase by 13 dB/decade for
offsets less than 1 kHz, which indicates that the low-frequency noise is due to the current-source
driver and voltage supply. For offsets greater than 5 kHz, the noise is reduced to that produced
by the fundamental amplified spontaneous emission quantum noise. Since noise less than 5 kHz
contributes a small fraction of the integrated timing jitter, the predominant source of the total value
is amplified spontaneous emission. Figure 3b shows the integrated timing jitter in each decade.
24-41
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
The timing jitter from 10 Hz to 4.5 GHz is 86 fs, or 154 fs if all spurs are included. With an
integrated timing jitter of 86 fs, it is important to have a quiet oscillator drive these lasers, since
the oscillator can easily be the dominant noise source. The SBO used in these measurements
had 5.6 fs of timing jitter (10 Hz to 10 MHz), which allows quantum-limited noise performance of
the MLLD. Quantum-limited noise performance means that the noise is dominated by
spontaneous-emission noise and not by microwave oscillator noise. The noise of a laser diode
depends on many laser parameters [3]. We have found some rules of thumb for achieving low
noise: (1) reduce cavity loss, (2) use tighter optical filtering, and (3) use a quieter oscillator.
Reducing cavity loss increases the ratio of signal to noise photons, while tighter optical filtering
limits Gordon–Haus jitter [4].
Fig. 1. Hybridly mode-locked semiconductor laser used in our experiments.
GRIN, graded-index; BPF, bandpass filter
Fig. 2. Residual phase-noise experimental setup. The device under test (DUT)
is the hybridly modelocked laser diode and photodetector. OSC, oscillator; AMP,
amplifier
24-42
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
1
Fig. 3. a) Single-sideband phase noise of the EC-MLLD and corresponding noise floor, pieced
together from the vector signal analyzer (10 Hz-10 MHz) and rf spectrum analyzer (10 MHz-5
GHz). The bandwidth of the mixer’s IF port was limited to 2 GHz. b) Integrated timing jitter in
each decade of the phase noise shown in Fig. 3a. In the last decade from 1 to 4.5 GHz, the white
bar corresponds to the timing jitter, assuming that the noise is equal to the noise floor; the black
bar corresponds to a theoretically expected –20 dB/decade roll off.
References
1. H. Yokoyama, “Highly stabilized mode-locked semiconductor diode lasers,” Rev. Laser Eng. 27,
750–755 (1999).
2. D. J. Derickson, A. Mar, and J. E. Bowers, “Residual and absolute timing jitter in actively modelocked semiconductor lasers,” Electron. Lett. 26, 2026–2028 (1990).
3. L. A. Jiang, M. E. Grein, H. A. Haus, and E. P. Ippen, “Noise of mode-locked semiconductor
lasers,” IEEE J. Sel. Top. Quantum Electron. 7, 159–167 (2001).
4. J. P. Gordon and H. A. Haus, “Random walk of coherently amplified solitons in optical fiber
transmission,” Opt. Lett. 11, 665–667 (1986).
Publications
Journal Articles
L. A. Jiang, M. E. Grein, J. K. Chandalia, E. P. Ippen, and H. Yokoyama, “Retiming dynamics of
modelocked semiconductor lasers”, Electron. Lett. 38, 1446-1447 (2002).
L. A. Jiang, M. E. Grein, H. A. Haus, and E. P. Ippen, “Quantum-limited noise performance of a
modelocked laser diode,” Opt. Lett. 27, 49-51 (2002).
F. Rana, H. L. T. Lee, M. E. Grein, L. A. Jiang, and R. J. Ram, “Characterization of the noise and
correlations in harmonically mode-locked lasers,” J. Opt. Soc. Am. B 19, 2609-2621 (2002).
Meeting Papers
L. A. Jiang, M. E. Grein, H. A. Haus, E. P. Ippen, “Experimental demonstration of a timing-jitter
eater,” in OSA Trends in Optics and Photonics (TOPS) Vol. 73, Conference on Lasers and
Electro-Optics, OSA Technical Digest, Postconference Edition (Optical Society of America,
Washington DC, 2002).
24-43
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
Variational Analysis of Spatio-temporal Pulse Dynamics in Dispersive Kerr
Media
Sponsors
National Science Foundation ECS-0119452
Project Staff
Christian Jirauschek, Dr. Uwe Morgner, Professor Franz X. Kaertner
The full spatio-temporal dynamics of Kerr-lens mode-locked (KLM) lasers can be studied in
numerical simulations, either by complete 3-D space modeling [1] or by various approximation
schemes [2, 3]. Alternatively, simplified models, allowing for quick numerical algorithms or even
handy formulas, are available for studying either the spatial or the temporal dynamics. Spatial
models for understanding and optimizing the self-focusing dynamics in a KLM laser are based on
an ABCD- or q-parameter analysis, neglecting the temporal breathing of the pulse in the Kerr
medium. In these approaches, the Kerr lens is modeled by an intensity-dependent lens, and an
iterative solution scheme is used [4, 5, 6]. As for the temporal dynamics, it has been shown [7, 8]
that dispersion-managed soliton formation is the most important pulse shaping process in KLM
lasers. The KLM action stabilizing the pulse against gain filtering can be considered a
perturbation to this dynamics. In the model introduced in [7], the temporal dynamics is studied
based on the variational principle [9], taking the nonlinearity in the Kerr medium and the actual
dispersion management into account.
We extended the variational principle to include the spatial effects due to self-focusing and
diffraction as well as the temporal effects due to self-phase modulation and second-order
dispersion. As a result, the equations of motion in the Kerr medium for the pulse parameters,
which are pulse width, pulse cross-sections in x and y directions, and temporal as well as spatial
chirp, are obtained [10]. Considering only the energy-preserving effects, the spatio-temporal
dynamics in the laser can then be described in Gaussian approximation, and the Gaussian
steady-state solutions can be extracted from the model [11].
To verify the validity of the Gaussian approximation, the Gaussian steady-state solutions of the
laser system are compared to the results of spatio-temporal numerical simulations [11]. By
including gain and loss in the numerical simulations, the influence of non-energy-preserving
effects on the pulse shaping process can be assessed. For obtaining the numerical solutions, an
initial laser pulse is propagated through the laser resonator over many roundtrips, until steady
state is reached. A transversally and spectrally dependent gain is allowed in the Kerr medium,
and loss is introduced by renormalizing the pulse after each roundtrip.
The Gaussian steady-state solution is computed for a Ti:sapphire laser setup with a 2.5 mm thick
Kerr medium and resonator arm lengths of 80 cm (left arm) and 110cm (right arm). The Gaussian
approximation is compared to numerically obtained results. In order to reduce the computation
time for the numerical simulation, axial symmetry has to be assumed. For the simulation, an
increase in intracavity pulse energy of about 2% during one roundtrip is assumed, corresponding
to a typical gain in today’s ultrashort-pulse KLM lasers. The spatial extension of the gain is
chosen to be somewhat narrower than the transversal pulse profile to enable soft-aperture Kerrlens mode-locking. In Fig. 1, the pulse duration, the spectral width and the beam width are shown
for the Gaussian steady-state solution subject to the position z in the Kerr medium. In Figs. 2 and
3, the Gaussian steady-state solution (dashed curve) is compared to the numerically obtained
result (solid curve). Displayed are the temporal pulse shape, characterized by the instantaneous
power P(t), the power spectrum S(f) and the transversal profile, described by the fluence )(r), at
the right end of the Kerr medium and at the right end mirror.
24-44
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
As can be seen from Figs. 2 and 3, the Gaussian steady-state solution is in good agreement with
the numerically obtained result for pulse durations down to a few optical cycles.
The Gaussian model provides insight into the elementary spatio-temporal dynamics of a KLM
laser system. An application of this model is to roughly determine the range of laser parameters
for which a steady-state solution exists, and to get a first approximation of the solution, avoiding
lengthy numerical simulations. Due to the low computational effort, even a scan over a wide
range of laser parameters becomes possible, allowing for a further optimization of KLM lasers.
Figure 1: Steady-state pulse dynamics in the Kerr medium: (a) Pulse duration. (b) Spectral width. (c)
Beam width.
Figure 2: Comparison between numerical solution (solid
curves) and Gaussian approximation (dashed curves) at the
right end of the Kerr medium. (a) Instantaneous power. (b)
Power spectrum.
24-45
Figure 3: Comparison between numerical solution
(solid curves) and Gaussian approximation (dashed
curves) at the right end mirror. (a) Instantaneous
power. (b) Power spectrum. (c) Fluence.
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
References
1. I.P. Christov and V.D. Stoev, “Kerr-lens mode-locked laser model: role of space-time
effects,” J. Opt. Soc. Am. B 15(7): 1960-66 (1998).
2. O.E. Martinez and J.L.A. Chilla, “Self-mode-locking of Ti:sapphire lasers: a matrix
formalism,” Opt. Lett. 17(17): 1210-12 (1992).
3. V.P. Kalosha, M. Müller, J. Herrmann and S. Gatz, “Spatiotemporal model of
femtosecond pulse generation in Kerr-lens mode-locked solid-state lasers,” J. Opt. Soc.
Am. B 15(2): 535-50 (1998).
4. H.A. Haus, J.G. Fujimoto, and E.P. Ippen, “Analytic theory of additive pulse and Kerr lens
mode locking,” IEEE J. Quantum Electron. 28(10): 2086-96 (1992).
5. G. Cerullo, S. De Silvestri, and V. Magni, “Self-starting Kerr-lens mode locking of a
Ti:sapphire laser,” Opt. Lett. 19(14): 1040-42 (1994).
6. A. Penzkofer, M. Wittmann, M. Lorenz, E. Siegert, and S. Macnamara, “Kerr lens effects
in a folded-cavity four-mirror linear resonator,” Opt. Quant. Electron. 28(4): 423-42
(1996).
7. Y. Chen and H.A. Haus, “Dispersion-managed solitons in the net positive dispersion
regime,” J. Opt. Soc. Am. B 16(1): 24-30 (1999).
8. Y. Chen, F.X. Kaertner, U. Morgner, S.H. Cho, H.A. Haus, E.P. Ippen, and J.G. Fujimoto,
“Dispersion-managed mode locking,” J. Opt. Soc. Am. B 16(11): 1999-2004 (1999).
9. D. Anderson, “Variational approach to nonlinear pulse propagation in optical fibers,”
Phys. Rev. A 27(6): 3135-45 (1983).
10. Ch. Jirauschek, U. Morgner, and F.X. Kaertner, “Variational analysis of spatio-temporal
pulse dynamics in dispersive Kerr media,” J. Opt. Soc. Am. B 19(7): 1716 - 21 (2002).
11. Ch. Jirauschek, F.X. Kaertner and U. Morgner, “Spatio-temporal Gaussian pulse
dynamics in Kerr-lens mode-locked lasers,” J. Opt. Soc. Am. B, forthcoming.
24-46
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
Ultrafast Phenomena and Quantum Electronics
Ultrafast Pump-Probe Studies of Silicon- and III/V-based Devices
Sponsors
U.S. Air Force – Office of Scientific Research - F49620-01-1-0084
MRSEC Program of the National Science Foundation - DMR 98-08941
Project Staff
J. T. Gopinath, J. M. Fischer, D. Cannon, Dr. G. S. Petrich, Professor F. Kaertner, Professor L.
Kimerling, Professor L. A. Kolodziejski, Professor E. P. Ippen
In order to improve designs of devices for applications in lasers, opto-electronics and
telecommunications, it is necessary to understand the fundamental properties of materials and
device structures. In this report, we describe the use of a femtosecond optical parametric
oscillator (OPO), producing 150 fs tunable between 1400 and 1600 nm, for characterization of
several important photonic materials. The pump-probe technique, which yields information about
absorption and index changes, as well as recovery times, is used to study the samples. In pumpprobe, a powerful pump pulse excites a sample at time t = 0. As the sample relaxes back to
equilibrium, its properties are sampled with weak, non-perturbing probe pulses at varying time
delays. The pump probe setup used has a measurement sensitivity of ~ 10-4 for transmission
and reflection changes.
Semiconductor structures designed for several different applications were characterized with a
degenerate pump-probe technique, in which the OPO signal wavelength is used for both the
pump and the probe. The pump and probe are in a ratio of 10:1 in fluence, and are crosspolarized. A schematic of the setup is shown below.
Figure 1: Schematic of pump probe setup.
A 543-nm-thick film of crystalline germanium on an undoped silicon wafer was characterized to
determine its suitability for use as a broadband ultrafast saturable absorber. The maximum
absorption of the film at 1540 nm is ~ 8.5%. Pump-probe traces as a function of incident fluence
are shown below. The fast component, due to spectral hole burning and thermalization of
carriers recovers quickly, and is a large percentage of the overall signal. However, there is still a
long-lived component of the signal, due to recombination and thermal effects. Future studies of
24-47
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
these effects as a function of fabrication conditions will help determine the optimum design for a
broadband saturable absorber.
Figure 2: Pump-probe trace of 543 nm Germanium film at 1540 nm.
In other experiments, a high-modulation-depth InP-based saturable Bragg reflector (SBR),
designed for use in a high repetition rate laser system, was studied.. The structure, fabricated by
Professor Kolodziejski’s group, consists of 12 InGaAs quantum wells in a O layer of InP. This is
grown on top of an MOCVD mirror consisting of a 22-pair GaAs/AlAs Bragg stack, centered at
1550 nm with a 100 nm bandwidth. Additional InP wasovergrown on the structure to enhance
two-photon absorption, which can help stabilize lasers against Q-switching instabilities [1].
Structures with either antireflection or resonant coatings to enhance the modulation depth have
been investigated Below is a pump-probe trace of the resonantly coated structure, in which a
maximum modulation depth of 20% has been achieved. The recovery time of this device is ~43
ps, due to the high lattice mismatch between the InP and the GaAs.
In addition to degenerate pump-probe, two-color pump-probe has been performed, using the
leftover 800 nm pump of the OPO and the OPO signal beam. Because the OPO is
synchronously pumped, these two colors have less than 20 fs of timing jitter [2] between them. In
addition, it is possible to perform two-color pump probe with the OPO signal and idler.
Microstructured fiber [3] and a Fabry Perot device have been studied using this technique. The
Fabry Perot device consisted of 1 Pm of polysilicon on a 1 mm quartz substrate, forming a Fabry
Perot. The Ti:Sapphire laser was used as the pump, and the OPO as the probe. The
Ti:Sapphire generated free carriers in the sample, causing the index to change slightly. This in
turn, leads to a shift of the resonances of the Fabry Perot, causing a change in transmission as a
function of wavelength. From this pump-probe experiment, a carrier density of 3 x 1018 /cm3
produced an index change of ~0.15%, a number that agrees well with values in the literature.
24-48
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
Figure 3: Pump probe of high modulation depth SBR at 1575 nm.
References
1. T. R. Schibli, E. R. Thoen, F. X. Kaertner and E. P. Ippen, "Suppression of Q-switched mode
locking and break-up into multiple pulses by inverse saturable absorption," Applied Physics B
Lasers and Optics, B70: S41-S49 (2000).
2. J. D. Kafka, M. L. Watts and J. W. Pieterse, "Synchronously pumped optical parametric
oscillators with LiB3O5," Journal of Optical Society of America B, 12(11): 2147-2157 (1995).
3. K. S. Abedin, J. T. Gopinath, E. P. Ippen, C. E. Kerbage, R. S. Windeler and B. J. Eggleton,
"Highly nondegenerate femtosecond four-wave mixing in tapered microstructure fiber,"
Applied Physics Letters, 81(8): 1384-1387 (2002).
24-49
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
Materials for Modelocking
Sponsors
Air Force Office of Scientific Research (MFEL) - F49620-01-1-0186
Air Force Office of Scientific Research - F49620-98-01-0084
National Science Foundation - ECS-019452
Project Staff
Rohit P. Prasankumar, Paul Mak, Professor Michael Ruane, Professor James G. Fujimoto
Non-epitaxially grown semiconductor-doped silica films for laser modelocking
Ultrafast laser technology has matured in recent years, resulting in the widespread availability of
ultrafast laser systems. However, the suitability of ultrashort pulse lasers for some applications is
still limited by their reliability and high cost, necessitating novel approaches to developing stable,
inexpensive ultrashort pulse laser sources. Semiconductor saturable absorbers are a well
established technology for generating stable, self-starting pulses in solid-state lasers. These
devices, known as semiconductor saturable absorber mirrors (SESAM) or saturable Bragg
reflectors (SBR), typically consist of semiconductor quantum wells grown in a semiconductor
mirror structure by molecular beam epitaxy (MBE) [1-3]. SESAMs and SBRs have been very
successful in mode-locking solid-state lasers, helping start and support pulses as short as 5.5 fs
in a Ti:sapphire laser [4]. However, they suffer from some disadvantages, such as lattice
matching constraints that limit the choice of semiconductor materials as well as reliance on a
complicated, expensive fabrication system.
The goal of this project is to develop a more versatile, lower cost alternative to epitaxially grown
semiconductor saturable absorbers. In previous work, we developed non-epitaxially grown
saturable absorber devices and applied them to self-starting Kerr lens mode-locking (KLM) in
Ti:sapphire and Cr:forsterite lasers [5, 6]. The devices consist of InAs nanocrystallites doped into
SiO2 films and deposited on sapphire substrates using magnetron and non-magnetron radio
frequency (RF) sputtering systems. RF sputtering is an inexpensive, simple device fabrication
technique that offers flexibility in the choice of semiconductor dopant and substrate materials.
We found that rapid thermal annealing (RTA) in nitrogen from 500-750qC was an effective
method of controlling the absorption saturation dynamics of our saturable absorbers. In
Ti:sapphire, self-starting 25 fs pulses were obtained with a bandwidth of 53 nm and tuning range
of 80 nm. The saturation fluence of these devices was measured to be 25 mJ/cm2, which is too
high to enable saturable absorber mode-locking without KLM and also limits the minimum
achievable pulsewidth.
Recently, we have developed guidelines for designing semiconductor-doped silica film saturable
absorbers with an optimized saturation fluence for a given solid state laser system. These
guidelines are based on extensive device characterization using linear and nonlinear optical
techniques while varying fabrication parameters. Growth parameters including choice of
semiconductor and target materials, annealing time and temperature, and ratio of semiconductor
to glass were varied to determine the most important factors influencing device performance.
Nonlinear optical measurements were performed using a previously developed pump-probe
system based on a broadband 5.5 fs Ti:Al2O3 laser [7] to obtain 17 fs time resolution and
independent pump and probe wavelength tunability over a range of 700 to 1000 nm [8]. The
magnitude of the measured pump probe signal can be shown to be inversely proportional to the
saturation fluence.
Linear optical measurements were performed using a Cary
spectrophotometer.
Initially, we examined the saturation fluence near the absorption edge of the semiconductor
nanocrystallites in a degenerate pump probe measurement between 750 and 925 nm (Fig. 1).
24-50
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
The measurement was performed on a 10%InAs/90% SiO2 film deposited on a sapphire
substrate. The film was annealed at 600qC for 60 seconds in nitrogen. The absorption edge of
this device was approximately 1100 nm. The saturation fluence of this film decreased with
increasing wavelength, demonstrating that operation close to the absorption edge is desirable to
minimize the saturation fluence.
10% InAs / 90% SiO Pump probe
-3
3x10
925 nm
-'DD
2
2
800 nm
750 nm
1
0
0
5
10
time (ps)
15
20
Figure 1. Tunable pump probe measurements between 750 and 900 nm revealing a decrease in
saturation fluence with wavelength.
The size of the InAs nanocrystallites was controlled by varying the ratio of InAs to SiO2 on the
sputtering target. As expected, the absorption edge shifted to longer wavelengths for higher
InAs/SiO2 ratios in linear transmission measurements. Pump probe measurements were
performed at 925 nm on 10%InAs/90% SiO2 films using the Ti:sapphire based pump probe
system, and on 40%InAs/60% SiO2 films at 1260 nm using a Cr:forsterite based pump probe
system. From the two graphs shown below (Fig. 2), it is clear that the magnitude of the signal is
significantly larger for the 40%InAs/60% SiO2 films. This data shows that the saturation fluence
strongly decreases with an increase in nanocrystallite size at wavelengths approximately the
same distance from the absorption edge.
-3
40%InAs/60%SiO2,
annealed at 600 C
Pump/probe wavelength 1260 nm
-'D D
30
20
3x10
-'D D
40x10
10
0
10%InAs/90%SiO2,
annealed at 600 C
Pump/probe wavelength 925 nm
-3
2
1
0
0
10
20
30
time (fs)
40 50x10
3
0
(a)
5
10
time (fs)
15x10
3
(b)
Figure 2. Pump probe measurements on (a) 40%InAs/60% SiO2 films at 1260 nm and (b)
10%InAs/90% SiO2 films at 925 nm. The saturation fluence is significantly lower for the
40%InAs/60% SiO2 films.
We also varied the annealing time and temperature of 10%InAs/90% SiO2 films to determine the
effect of RTA on the nonlinear absorption saturation dynamics. We observed discrete changes in
the dynamics as the annealing temperature was varied. Fig. 3 depicts the changes in the
24-51
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
measured pump-probe signal at 800 nm as a function of annealing temperature for 10%InAs/90%
SiO2 films annealed for 60 seconds in nitrogen. For temperatures of 350qC and below, a
negative pump-probe signal was measured, indicating no absorption saturation; this was identical
to measurements done on unannealed samples. In the range of 400-500qC, the absorption
saturation was positive but small. At temperatures of 550qC and above, the absorption saturation
was relatively large; samples annealed at these higher temperatures self-started KLM in
Ti:sapphire and Cr:forsterite lasers. Samples annealed at 700qC had a faster relaxation time
than those annealed at 550-600qC, although the magnitude of the signal and therefore the
saturation fluence was nearly the same as that of the sample annealed at 550qC.
We believe that these discrete changes in the pump-probe signal as a function of annealing
temperature are related to the glass melting and transition temperatures for SiO2. We tested this
by fabricating films with InAs nanocrystallites doped into a borosilicate glass matrix that had
different glass melting and transition temperatures. Pump-probe measurements demonstrated
that the absorption saturation did not change discretely with annealing temperature, instead
increasing continuously as the RTA temperature increased. However, at high temperatures the
dynamics were very similar to those displayed in Fig. 3 with a SiO2 matrix and there was no
significant improvement in the saturation fluence.
Temperature dependence of RTA
10%InAs/90%SiO2
-3
3x10
' T/T
550 C
2
700 C
1
500 C
350 C
0
0
no RTA
1000
2000
time (fs)
3000
Figure 3. Pump probe measurements on 10%InAs/90% SiO2 films at 800 nm as a
function of rapid thermal annealing temperature. The samples were all annealed for
60 seconds in nitrogen.
From these experiments, guidelines were formulated to optimize the saturation fluence of
semiconductor-doped silica film saturable absorbers for a given laser system. As shown in fig. 2,
films with larger nanocrystallites have a significantly lower saturation fluence than those with
smaller nanocrystallites. It is also clear that the films should have an absorption edge close to the
laser wavelength, since this also lowers the saturation fluence (Fig. 1). Therefore, the
semiconductor material and semiconductor/glass ratio should be chosen to satisfy these two
criteria. Finally, the annealing temperature should be above 550qC for strong absorption
saturation based on the results displayed in Fig. 3; however, the optimum temperature may vary
for different semiconductor and glass materials.
We applied these guidelines to design semiconductor-doped glass film saturable absorbers for
self-starting modelocking in a Cr:forsterite laser operating at 1.26 Pm [5, 6]. We deposited thin
40%InAs/60% SiO2 films on sapphire substrates and annealed them at 600qC. Their saturation
24-52
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
fluence at 1.26 Pm was measured from pump probe experiments to be 3.35 mJ/cm2, which is
nearly an order of magnitude improvement over the previous work in Ti:sapphire. Self-starting
KLM was obtained, with a bandwidth of 91 nm and pulsewidth of 25 fs measured by
interferometric autocorrelation. We measured the modelocking buildup time in this system and
found it to be approximately 2.5 ms, about 20 times faster than in Ti:sapphire; this can also be
linked to the lower saturation fluence at this wavelength.
We also investigated pure saturable absorber mode-locking without KLM as a function of the
intracavity dispersion. Self-starting saturable absorber mode-locking was observed; however,
intensity autocorrelations showed that the output consisted of a short pulse with a long
background pulse (Fig. 4). This is consistent with theories of fast saturable absorber modelocking with KLM and soliton-like pulse shaping. Adjustment of the intracavity dispersion towards
zero by varying the prism insertion results in a shorter pulse duration. However, the low self
amplitude modulation of the saturable absorber cannot support this short pulse, therefore it sheds
energy to a longer background pulse with a duration determined by the recovery time of the
saturable absorber. Autocorrelation measurements of the pulses for different intracavity
dispersion operating points agree with theoretical expectations. For small positive dispersions, a
long pulse with 5.3 ps duration is generated. The corresponding spectrum is 11 nm and the pulse
is strongly chirped. As the intracavity dispersion is made increasingly negative, a short solitonlike pulse of ~150 fs duration is generated with a longer background pulse. For more negative
values of dispersion, the intensity of the longer background pulse decreases. Pulse shaping
would be improved if the saturable absorber saturation fluence was reduced or if the relaxation
dynamics were made faster.
1.0
SHG signal (arb. units)
positive
2
dispersion ~+150 fs
0.8
slightly positive dispersion
0.6
slightly negative dispersion
0.4
negative
2
dispersion ~-340 fs
0.2
0.0
-10
-5
0
time (ps)
5
Figure 4. Intensity autocorrelations at different dispersion operating points for
saturable absorber modelocking without KLM. The peak intensity of the longer
background pulse decreases as the negative dispersion increases.
Future work will include applying non-epitaxially grown semiconductor saturable absorbers to
other laser systems, designing devices in different geometries, and testing other semiconductor
and glass materials to further reduce the saturation fluence.
24-53
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
References
1. U. Keller, K. J. Weingarten, F. X. Kärtner, D. Kopf, B. Braun, I. D. Jung, R. Fluck, C.
Hönninger, N. Matuschek and J. Aus der Au, "Semiconductor saturable absorber mirrors
(SESAMs) for femtosecond to nanosecond pulse generation in solid-state lasers," IEEE J.
Sel. Top. Quant. Elect. (JSTQE), 2, pp.435-453, 1996.
2. S. Tsuda, W. H. Knox, S. T. Cundiff, W. Y. Jan and J. E. Cunningham, "Mode-locking
ultrafast solid-state lasers with saturable Bragg reflectors," IEEE J. Sel. Top. Quant. Elect., 2,
pp.454-464, 1996.
3. D. Jung, F. X. Kärtner, N. Matuschek, D. H. Sutter, F. Morier-Genoud, Z. Shi, V. Scheuer, M.
Tilsch, T. Tschudi and U. Keller, "Semiconductor saturable absorber mirrors supporting sub10 fs pulses," Appl. Phys. B, 65, pp.137-150, 1997.
4. 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 mode-locked Ti:sapphire laser producing pulses in the two-cycle regime," Opt. Lett.,
24, pp.631-633, 1999.
5. P. Bilinsky, J. G. Fujimoto, J. N. Walpole and L. J. Misaggia, "Semiconductor-doped silica
saturable absorber films for solid state laser mode locking," Opt. Lett., 23, pp.1766-1768,
1998.
6. R. P. Prasankumar, C. Chudoba, J. G. Fujimoto, P. Mak and M. F. Ruane, "Self-starting
mode locking in a Cr:forsterite laser by use of non-epitaxially-grown semiconductor-doped
silica films," Optics Letters, 27, pp.1564-66, 2002.
7. U. Morgner, F. X. Kärtner, S. H. Cho, H. A. Haus, J. G. Fujimoto, E. P. Ippen, V. Scheuer, G.
Angelow and T. Tschudi, "Sub-two cycle pulses from a Kerr-Lens modelocked Ti:sapphire
laser," Opt. Lett., 24, pp.411 -- 413, 1999.
8. R. P. Prasankumar, I. Hartl, J. T. Gopinath, E. P. Ippen, J. G. Fujimoto, P. Mak and M. F.
Ruane, "Ultrafast dynamics of non-epitaxially grown semiconductor-doped silica film
saturable absorbers," Quantum Electronics and Laser Science Conference, 2001.
24-54
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
High-Speed Femtosecond Pump-Probe Spectroscopy Using a Smart Pixel
Detector Array
Sponsors
Air Force Office of Scientific Research (MFEL) - F49620-01-1-0186
Air Force Office of Scientific Research - F49620-98-01-0084
Project Staff
Stephane Bourquin, Rohit P. Prasankumar, Professor Franz X. Kaertner T. Lasser, R.-P. Salathe,
Professor James G. Fujimoto
The standard pump-probe technique uses an ultrashort pulse, narrow-band pump to excite the
sample and a continuum probe and spectrometer to acquire the wavelength-dependent
dynamics. This technique usually requires a femtosecond amplifier to achieve intensities
necessary for continuum generation and sufficient pump-probe signal magnitudes. With the
development of few-cycle, ultra broad bandwidth lasers which can generate spectra that span
nearly one octave, spectrally resolved pump-probe measurements can be performed without the
need for amplifiers [1, 2, 3]. Although the pulse repetition rate from a laser oscillator is extremely
high, the pulse energies are low and therefore the signal levels are small. Standard CCD
detectors cannot detect modulated signals, thus it is not possible to take advantage of high
signal-to-noise measurements that are theoretically possible with high repetition rate laser
sources [4].
We have developed a new technique that enables the parallel acquisition of pump-probe
measurements at multiple wavelengths simultaneously, using a novel, two-dimensional, smart
pixel detector array, which was originally developed for high-speed optical coherence tomography
(OCT) [5, 6]. Each pixel performs amplitude demodulation, and, in combination with a
spectrometer, probe transmission signals can be acquired in parallel for multiple wavelengths.
The smart pixel array can achieve sensitivities comparable to lock-in amplification while
simultaneously demodulating probe transmission signals at multiple wavelengths, thus enabling
time- and wavelength-resolved femtosecond pump-probe spectroscopy.
Figure 1 left shows a schematic of the experiment. A 100 MHz repetition rate, 5 fs, 250 nm
bandwidth Ti:sapphire mode-locked laser [1] is used with a femtosecond pump-probe system. A
beamsplitter (BS) separates the output of the laser into a 90 mW pump beam and 0.4 mW probe
beam. A pair of prisms is used in both pump and probe beams to permit independent adjustment
of pulse duration and chirp. A conventional delay line (DL) is used to scan the pump beam. The
pump and the probe beams are focused onto the sample by a parabolic mirror (PM) which is
used to avoid dispersion and to preserve pulse duration. An aperture stop (AS) is inserted into the
probe beam after the sample to prevent pump scattering into the detector. The probe beam is
then collimated by a lens (L1) and spectrally spread by a diffraction grating (DG). A lens (L2)
focuses the spread beam onto one row of the two-dimensional smart pixel detector array (DA).
The chopper (CH) modulates the pump beam at a frequency of 3.8 kHz. Each pixel in the row
demodulates the transmitted probe signal at a specific wavelength. The row is read multiple times
for each pump time delay, and averaging is performed to increase the signal-to-noise ratio. The
probe beam is subsequently chopped and the same row is read to acquire the linear transmission
of the sample for data normalization.
24-55
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
5 fs
Ti:Sapphire
laser
BS
Probe
Prism
compressor
PD
PM
Pump
M
Prism
compressor
M
EC
Sample
CH
L1
DL
AS
O1 … On
DA
DG
L2
Computer
RD
CD
100 Pm
Figure 1. (left) Schematic diagram of the experiment: BS, beamsplitter; M, mirror; DL, delay
line; CH, chopper; PM, parabolic mirror; L1,L2, lenses; DG, diffraction grating; DA, detector
array. (right) Photograph of a section of the smart pixel detector array. PD; photodiode; EC,
electrical circuit; RD, row address decoder; CD, column address decoder.
A photograph of a section of the detector array is shown in Figure 1 right. The silicon detector
chip is realized with a 2 µm complementary metal-oxide semiconductor process with a bipolar
transistor option. The die size is 7.2 mm x 7.2 mm, which allows a 58 x 58 pixel array. Each pixel
is 110 µm x 110 µm and contains a 35 µm x 35 µm photodiode and electronic circuitry for
performing amplification, band-pass filtering centered at the modulation frequency, rectification
and low-pass filtering. The generated amplitude modulation signals are selected sequentially by a
row and a column address decoder. The analog output signal is read out of the chip, digitized by
a 12-bit data acquisition card and transferred to a computer.
Measurements were performed in a thin sample of bulk GaAs. A 3D data set acquired by the
smart pixel detector array is shown in Figure 2 left. The normalized differential probe transmission
is plotted versus the time delay and the probe wavelength. The pump beam spectrum was set to
a bandwidth of 700 nm to 770 nm to excite conduction band states above the band gap of ~870
nm. The detected probe wavelength ranges from 700 nm to 900 nm and the pump-probe signals
are recorded over a 5 ps time delay. For each delay, the pixels are read 5000 times and
averaged, yielding a sensitivity to normalized differential probe transmission signal changes as
small as 2 x 10-4. The pixels are read at 625 kHz, corresponding to 188 seconds of total
measurement time. This time is a factor of 58 times faster than possible using a conventional
single detector pump-probe spectroscopy system.
Figure 2 right presents pump-probe measurements at selected wavelengths selected from the 3D
data set. The dynamics are a strong function of wavelength. Measurements performed far above
the band edge of 870 nm show a rapid relaxation of absorption saturation due to carrier-carrier
and carrier-phonon scattering, which remove carriers from their initial optically excited states.
Measurements closer to the band edge show increased absorption saturation as a function of
time from carrier relaxation and state filling.
24-56
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
3.0
879 nm
834 nm
2.5
807 nm
2.0
-2
'T/T (10 )
'T/T (10-2)
1.0
0.5
0.0
av 850
ele 800
ng
th 750
[n
m 700
]
1.5
772 nm
1.0
900
W
779 nm
769 nm
755 nm
0.5
0
1
Tim
2
3
4
728 nm
0.0
s]
ay [p
e del
703 nm
0
1
2
3
4
5
Time delay [ps]
Figure 2. (left) Spectrally resolved femtosecond pump-probe measurements of a thin sample of
bulk GaAs as a function of probe wavelengths and time delay. The data are acquired
simultaneously by the smart pixel detector array. (right) Pump-probe measurements at selected
probe wavelengths extracted from the three-dimensional data set. The traces are separated for
clarity.
These results demonstrate the feasibility to perform parallel spectrally resolved pump-probe
measurements across 58 simultaneous wavelengths with a detection sensitivity to normalized
differential probe transmissions as small as ~2 x 10-4 with a measurement time of only ~3 minutes
for a 5 ps scan. The current limitations of the technique are set by the low pixel readout rate and
low sensitivity due to the first-order filters implemented in the detector array. A redesign of the
detector array with second-order bandpass filters and with a higher readout rate to increase the
number of averages per time delay point would improve the signal-to-noise ratio of the system
and sensitivities to differential probe signal changes in the 10-5 to 10-6 range should be
achievable.
References
1. U. Morgner, F.X. Kärtner, S.H. Cho, Y. Chen, H.A. Haus, J.G. Fujimoto, E.P. Ippen, V.
Scheurer, G. Angelow, and T. Tschudi, Opt. Lett. 24, 411 (1999).
2. D.H. Sutter, G. Steinmeyer, L. Gallmann, N. Matuschek, F. Morier-Genoud, U. Keller, V.
Scheuer, G. Angelow, T. Tschudi, Opt. Lett. 24, 631 (1999).
3. R. Ell, U. Morgner, F. X. Kärtner, J. G. Fujimoto, E. P. Ippen, V. Scheuer, G. Angelow, T.
Tschudi, M. J. Lederer, A. Boiko, and B. Luther-Davies, Opt. Lett. 26, 373 (2001).
4. G. P. Wakeham and K. A. Nelson, Opt. Lett. 25, 505 (2000).
5. S. Bourquin, P. Seitz, and R. P. Salathé, Opt. Lett. 26, 512 (2001).
6. S. Bourquin, P. Seitz, and R. P. Salathé, Electron. Lett. 37, 975 (2001).
24-57
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
Photonics and Devices
Micromachined Photonic Devices using Nonlinear Materials Processing
Sponsors
Air Force Office of Scientific Research (MFEL)- F49620-01-1-0186
Air Force Office of Scientific Research - F49620-98-01-0084
Project Staff
Andrew M. Kowalevicz, Dr. Ingmar Hartl, Dr. Kaoru Minoshima, Professor Erich P. Ippen, and
Professor James G. Fujimoto
Nonlinear materials processing for photonic device fabrication using near-IR femtosecond pulses
has emerged as an active area of research because it is possible to fabricate localized, clean,
three-dimensional structures in a wide range of materials without the need for linear absorption.
A variety of devices in glasses such as waveguides [1, 2] [3], couplers [4] [5, 6], gratings [7], 3D
structures [8] [6], active waveguides [9], and void structures [10-12] have been successfully
fabricated. While laser amplifiers have often been used for fabrication, laser oscillators have
many advantages over amplifier systems, and moreover, the higher pulse repetition rate enables
faster and more efficient waveguide fabrication. Here we demonstrate the fabrication of coupled
mode devices. The behavior of device function is investigated by varying structural parameters
such as interaction length and separation. Mach-Zehnder interferometers were also fabricated
and spectral filtering was demonstrated using an unbalanced path length interferometer. These
devices constitute the basic building blocks of photonic devices.
Studies were performed using a novel, extended cavity modelocked titanium sapphire laser [13].
Since the total output power of a laser is limited, pulse energies can be increased by increasing
the laser cavity length and reducing the pulse repetition rate. Using 4 MHz repetition rate titanium
sapphire laser, pulses of up to 100 nJ can be generated with 80 fs pulse duration. Waveguides
were fabricated inside glass by focusing the femtosecond laser beam and translating the glass
perpendicular to the incident beam to write the waveguides. The refractive index differentials
were measured using OCT to be 10-3 to 10-2 depending upon exposure parameters. The edges
of the glass substrate were polished to optical quality in order to enable efficient coupling.
(a)
Input
Coupling
He-Ne laser
(b)
Separation
Output
Input
Interaction
Output
25
Fig. 1. (a) Phase contrast microscopic image of one of the directional couplers. (b) Schematic of
the coupler. Separation, d and interaction length, L are varied with the fixed total length, 25 mm.
Directional couplers were fabricated in which two single mode waveguides have an interaction
region without intersecting. Figs. 1(a) and (b) show the phase contrast microscopic top view and
24-58
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
the schematic of a directional coupler. The coupler has an interaction region of length L where
the two waveguides are parallel and closely spaced with a separation d. Each waveguide has
two bends of 1o. When a He-Ne laser beam (633 nm) was coupled into one input port, power
was transferred to the two output ports, suggesting coupling between the two waveguides.
Coupling ratio, R
Directional couplers operate using coupled mode effects. When two waveguides are brought into
close proximity with a small separation between them, the mode that is guided in one of the
waveguides can couple into the other waveguide by evanescent field interaction. The operation
of these devices can be described by coupled mode theory [14]. This oscillation of power
between the two waveguides is characteristic of coupled mode behavior. When the separation d
between the waveguides becomes larger, the overlap between the eigenmodes is reduced and
the coupling coefficient is smaller, so the oscillation of power between the two waveguides
becomes slower.
0.6
0.4
0.2
(a)
0.0
0
2
4
6
8
10
0.3
Coupling ratio, R
Coupling ratio, R
Interaction length, L (mm)
(b)
0.2
0.1
0.0
0
2
4
6
8
0.08
(c)
0.06
0.04
0.02
0.00
10
0
Interaction length, L (mm)
2
4
6
8
10
Interaction length, L (mm)
Fig. 2. (a) Interaction length dependence of the coupling ratio for d = 8 um, (b) d = 10 um, and
(c) d = 12 um. Experimental results (dots) and their best fit results to sinusoidal curves (lines) are
shown. The period of the oscillation increases from (a) 5, to (b) 9, to (c) 14 mm, which is
consistent with the coupled mode theory.
Figure 2 shows the measured results from our directional couplers. When the separation, d,
between the waveguides increases from (a) 8 um, to (b) 10 um to (c) 12 um, the oscillation period
of the coupling ratio, R, increases from (a) 5 mm, to (b) 9 mm, to (c) 14 mm. This behavior is
consistent with coupled mode theory which predicts that decreasing in coupling coefficient, yields
an increase in the oscillation of the coupling ratio with interaction length. The oscillatory behavior
of the coupling ratio as a function of interaction length, L, that is predicted by coupled mode
theory is clearly evident. The lines in the figures show best-fit sinusoidal functions. To the best of
our knowledge, this is the first demonstration of the oscillatory behavior of the coupling coefficient
and confirms coupled mode operation in devices fabricated by femtosecond nonlinear material
processing.
24-59
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
The fabrication and characterization of a Mach-Zehnder interferometer filter is demonstrated as
another example of a more complex photonic device. The interferometer consists of two Xcouplers placed back-to-back, with crossing angles of 2o as shown in Fig. 3(a). The path length
difference between the two arms is approximately 10 um. Light coupled into the interferometer is
split into the two arms of the interferometer at the first X-coupler, travels different path lengths,
and will either constructively or destructively interfere at the second X-coupler. The unbalanced
path length Mach-Zehnder interferometer functions as a wavelength dependent filter. The
frequency or wavelength dependence of the interferometer can be measured using a broadband
light source[15]. Fig. 3(b) shows the input spectrum with a FWHM of 130 nm, together with the
output spectrum. Fig. 3(c) shows the wavelength transfer function in the crossed interferometer
arm. The transfer function is constructed by normalizing the output spectrum by the input
spectrum. The red line shows the theoretically predicted wavelength dependence of the transfer
function for a path length difference of 14.1 um. The experimental measurements are in close
agreement with the theory. The difference between design arm length and actual arm length of
~4.1 um over 16.5 mm is quite small and is probably due to inaccurate translation (overshoot) of
the motorized stages used for fabrication and can be corrected with more precise mechanical
control.
(a)
Modelocked
Ti:Al2O3 Laser
1.0
Intensity (a. u.)
Intensity (a. u.)
1.0
0.8
0.6
0.4
0.2
0.0
700
(b)
2O Crossing
16.5
750
800
850
Wavelength (nm)
900
0.8
0.6
0.4
0.2
0.0
700
(c)
750
800
850
900
Wavelength (nm)
Fig. 3. (a) Schematic representation of Mach-Zehnder Interferometer with phase-contrast
microscope images of waveguides; (b) Input (red line) and output (black line) spectra from
broadband Ti:Al2O3 and, (c) Normalized output spectrum (black line) demonstrating the filtering
effect of the interferometer compared to theoretical model (red line).
In conclusion, we have demonstrated the fabrication of coupled mode devices and
interferometers using nonlinear femtosecond materials processing in glass. Oscillation of the
power between the two waveguides has been observed as a function of the waveguide
interaction length and coupling coefficient as well as a function of wavelength. These results are
24-60
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
consistent with coupled mode theory and demonstrate that the directional couplers operate by
mode coupling. The fabrication of an unbalanced Mach-Zehnder interferometer is also
demonstrated as an example of a more complex device. The operation of the unbalanced
interferometer as a wavelength filter is demonstrated and is in agreement with theory. These
results demonstrate the practical photonic device fabrication is possible using femtosecond
nonlinear materials processing.
References
1. Davis, K.M., et al., Writing waveguides in glass with a femtosecond laser. Optics Letters,
1996. 21: p. 1729-1731.
2. Miura, K., et al., Photowritten optical waveguides in various glasses with ultrashort pulse
laser. Applied Physics Letters, 1997. 71: p. 3329-3331.
3. Streltsov, A.M. and N.F. Borrelli, Fabrication and analysis of a directional coupler written
in glass by nanojoule femtosecond laser pulses. Optics Letters, 2001. 26: p. 42-43.
4. Homoelle, D., et al., Infrared photosensitivity in silica glasses exposed to femtosecond
laser pulses. Optics Letters, 1999. 24: p. 1311-1313.
5. Schaffer, C.B., et al., Micromachining bulk glass by use of femtosecond laser pulses with
nanojoule energy. Optics Letters, 2001. 26: p. 93.
6. Minoshima, K., et al., Photonic device fabrication in glass by use of nonlinear materials
processing with a femtosecond laser oscillator. Optics Letters, 2001. 26: p. 1516-1518.
7. Kondo, Y., et al., Fabrication of long-period fiber gratings by focused irradiation of
infrared femtosecond laser pulses. Optics Letters, 1999. 24: p. 646-648.
8. Glezer, E.N., et al., Three-dimensional optical storage inside transparent materials.
Optics Letters, 1996. 21: p. 2023-2025.
9. Sikorski, Y., et al., Optical waveguide amplifier in Nd doped glass written with near-IR
femtosecond laser pulses. Electronic Letters, 2000. 36(226).
10. Varel, H., et al., Micromachining of quartz with ultrashort laser pulses. Applied Physics A,
1997. 65: p. 367-373.
11. Glezer, E.N. and E. Mazur, Ultrafast-laser driven micro-explosions in transparent
materials. Applied Physics Letters, 1997. 71: p. 882-884.
12. Watanabe, W., et al., Optical seizing and merging of voids in silica glass with infrared
femtosecond laser pulses. Optics Letters, 2000. 25: p. 1669-1671.
13. Cho, S.H., et al., High energy pulse generation using a 4 MHz repetition rate KLM
Ti:Al2O3 laser operating with positive and negative dispersion. Optics Letters, 2001. 26:
p. 560-562.
14. Saleh, B.E.A. and M.C. Teich, Fundamentals of Photonics. 1991: John Wiley & Sons, Inc.
15. Morgner, U., et al., Sub-two cycle pulses from a Kerr-Lens modelocked Ti:sapphire laser.
Optics Letters, 1999. 24: p. 411 -- 413.
24-61
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
Development of First, Second, and Third Order Ring Resonators for
Channel Dropping Filter Applications
Sponsor
Pirelli
Project Staff
Peter T. Rakich, Tymon Barwicz, Milos Popovic, Christina Manolatou, Michael R. Watts, Giacomo
Gorni, Professor Henry I. Smith, Professor Herman A. Haus, Professor Erich P. Ippen
Ring resonators are particularly interesting for many applications because they can be designed
to function as efficient, high-Q filters which are extremely small in size. Ring resonators have
been used in microwave electronics for years. However, only recently have materials processing
and micro-fabrication become precise enough to demonstrate such devices at optical
wavelengths in dielectric waveguides [1-3]. As the demand for high data transmission in
telecommunications increases, the need for a fully integrated means of filtering and switching
telecommunications channels is ever increasing. Ring resonators appear to be a promising
solution for the integration problem.
The goal of our current research is to construct second and third order ring resonator filters which
would be suitable for telecommunications applications. A SiN2/SiO2 material system has been
chosen for the rings fabricated in this study. First, second and third order ring resonators have
been fabricated through e-beam lithography. A small ring radius of 20Pm produces a free
spectral range which extends over the C-band of telecommunications channels.
This is
important to ensure that only one resonance from a given filter can overlap with the C-band at a
given time. A schematic of the device layout can be seen in Figure 1. The SiN2 waveguides are
supported by a SiO2 undercladding, and a bus waveguide couples horizontally to the ring
resonator through evanescent coupling. The resonant wavelengths preferentially couple to the
drop port while all other wavelengths pass, unaffected, through the bus waveguide. In second
and third order ring resonators, the first ring transfers its power to one or two more rings before it
transfers power to the drop waveguide; however, the same concepts of the first order ring
resonators still apply. Additionally second and third order ring resonators offer added flexibility of
the transfer function. These devices were designed to more closely approximate top hat like
transfer functions which can serve as telecommunications filters.
The filter responses of the first, second and third order ring resonators can be seen in figures 1,
2, and 3 respectively. The spectrum in Figure 1 exhibits a Lorenzian-type transfer function which
is characteristic of first-order ring resonators. In comparing the through and drop ports, a 3db
transfer of power occurs on resonance. The second order ring resonator exhibits a symmetric
double-peak structure. This can be understood as a symmetric breaking of degeneracy through
the coupling of the two rings to one-another. It is important to notice that the roll-off of the second
order and third order rings is significantly faster for off-resonance wavelengths. This is an
important aspect of the higher order coupled ring structures in order to reduce cross-talk between
telecommunications channels. The third order ring resonator exhibits a somewhat more
complicated spectrum. In particular, the Through-port signal exhibits a sharp dip at longer
wavelengths. This can be attributed to the fact that the middle ring is slightly detuned from the
neighboring rings due to its environment. Despite this fact, the third order device exhibits a
120GHz spectral width and a 3db transfer of power from the through waveguide.
In conclusion, we have demonstrated respectable performance of first, second and third order
ring resonator devices in a SiN2 material system. Currently we work to calibrate our device
simulations with experiments and fabrication to improve the device performance and obtain filters
which could conceivably be useful as telecommunications channel dropping filters for large scale
integration in telecommunications networks.
24-62
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
First Order Ring Resonator: Comparison of Drop and
Through Ports For TE Polarization (Device E1)
Power (a.u.)
0.1
0.01
0.001
0.0001
1564
1566
1568
1570
1572
1574
1576
Wavelength (nm)
Figure 1. Schematic and transmission spectrum of first order ring resonator. The through-port
and drop-port signals are displayed in red and blue respectively.
Second Order Ring Resonantor (F6) Comparitive
Measurement TE Launch and Analyzer
0.1
Power (a.u.)
0.01
0.001
0.0001
10-5
1553
1554
1555
1556
1557
1558
1559
Wavelength (nm)
Figure 2. Schematic and transmission spectrum of second order ring resonator. The throughport and drop-port signals are displayed in red and blue respectively.
Third Order Ring Resonator Comparison of Through
and Drop Port Transmission for TE Polarization (Device G2)
0.1
Power (a.u.)
0.01
0.001
0.0001
10-5
1565
1566
1567
1568
1569
1570
1571
1572
1573
Wavelength (nm)
Figure 3. Schematic and transmission spectrum of third order ring resonator. The through-port
and drop-port signals are displayed in red and blue respectively.
24-63
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
References
1. B. E. Little, T. S. Chu, W. Pan, D. Ripin, T. Kaneko, Y kokubun, and E. Ippen “ Vertically
coupled Glass Microring Resonator Channel Dropping Filters” IEEE Photonics Tech. Lett.
11(2): 215-217 (1999)
2. B. E. Little, J. S. Foresi, G. Steinmeyer, E. R. Thoen, S. T. Chu, H. A. Haus, E. P. Ippen, L. C.
Kimmerling, W. Greene “Ultra-compact Si-SiO2 Microring Resonator Optical Channel
Dropping Filters,” IEEE Photonics Tech. Lett. 10(4): 549-551 (1998)
3.
J. V. Hryniewicz, P. P. Absil, B. E. Little, R. A. Wilson and P. T. Ho, “Higher Order Filter
Response in Coupled Microring Resonators” IEEE Photonic Tech. Lett. 12(3): 320-322
(2000)
24-64
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
Tuning and Switching of Ring Resonator Through Perturbation of Effective
Index
Sponsors
Pirelli
National Science Foundation - ECS 0119452
MRSEC Program of the National Science Foundation - DMR 98-08941
US Air Force – Office of Scientific Research - F49620-01-1-0084
Project Staff
Peter T. Rakich, Danielle Faccio, Professor Herman A. Haus, Professor Erich P. Ippen
Since their conception, ring resonators have demonstrated their promise as narrow band channel
dropping filters, and as laser cavities for integrated quantum well lasers. However, trimming and
tuning of such filters, for use in precisely defined telecommunications channels, proves to be a
challenging task. Several methods of trimming and tuning have been developed [1-2]; however,
none of them can promise tuning and switching on the microsecond timescale over large
wavelength ranges.
We demonstrate a new means of tuning and switching which is based on the external
perturbation of a ring resonator with a slab of high index material. The mode of an unclad ring
resonator (index guided) has an evanescent field which extends outside of the guide in the
direction normal to the plane of the ring. A slab of high index material, whose surface is parallel
to the plane of the ring, can be placed in the mode such that the guided mode of the ring
uniformly penetrates it (see Figure 1). This produces in increase in effective index of the mode,
and results in tuning of the resonance frequency of the ring.
Figure 1. Diagram depicting tuning concept. Glass perturbing body penetrates the
evanescent field which extends outside of the guide, changing its effective index
and resulting in a tuning of the resonance frequency of the ring.
For the purposes of this demonstration, a vertically coupled unclad ring resonator composed of
Ta2O5 used with a SiO2 lower cladding material. For device specifics see ref [3]. Tuning-fork
based shear-force feedback [4] was implemented to monitor and control the distance between a
cleaved piece of fiber-optic, which produces the perturbation, and the ring resonator. The
apparatus can be seen in Figure 2 (a). A small piece of fiber optic is attached to the end of a
quartz tuning fork. The fiber’s interaction with the surface is sensed through mechanical coupling
to the tuning-fork feedback system. The strength of the fiber-surface interaction is used as a set
point for the feedback system.
Various transmission spectra were acquired from the through-port of the ring resonator as a
function of the fiber-ring distance as seen in figure 2 (b). The shortest wavelength resonance
corresponds to the unperturbed ring. The series of red-shifted spectra correspond to increasing
24-65
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
C
R
Figure 2. (a) Simple schematic of tuning apparatus. Tuning fork signal is monitored to
maintain the distance of probe from surface of ring. (b) Transmission data from the
through port of ring resonator for various positions of the probe. The left-most
resonances are those with no perturbation, while the progression of red-shifted
resonances demonstrates increasing perturbation of the ring resonator.
perturbations of the ring, where greater shifts correspond to greater shifts correspond to greater
perturbations. Notice that the resonance which corresponds to contact of the fiber to the ring is
sharper than the other intermediate spectra. This is due to the fact that the feedback mechanism
which maintains the fiber-ring distance is too slow to compensate for the high frequency (kHz)
ambient vibrations on the optical table. This tends to artificially broaden the transmission spectra.
However, when the fiber makes contact with the ring the fluctuations in distance are removed and
a significantly sharper spectrum is observed. Additionally, when the fiber is raised again the
unperturbed resonance is fully recovered.
In conclusion, we have demonstrated a continuous and reversible tuning mechanism based on
external perturbation of the ring resonator mode which can be implemented on the microsecond
timescale. A total tuning range of 1.8nm has been achieved with a relatively low index probe.
Future research will seek to improve tuning ranges to as much as 20nm using higher index
probes with a multilayer design.
References
1. Yasuo Kokubun, Hirofumi Haeiwa, Hiroaki Tanaka, “Precise Center Wavelength Trimming of
Vertically Coupled Microring Resonator Filter by Direct UV Irradiation to Ring Core”
Proceedings LEOS Annual, Edinburgh, 2002
2. Heimala,-P.; Katila,-P.; Aarnio,-J.; Heinamaki,-A. “Thermally tunable integrated optical ring
resonator with poly-Si thermistor,”J.Lightwave Technol. 14(10): 2260-7 (1996)
3. B. E. Little, T. S. Chu, W. Pan, D. Ripin, T. Kaneko, Y Kokubun, and E. Ippen “ Vertically
coupled Glass Microring Resonator Channel Dropping Filters” IEEE Photonics Tech. Lett.
11(2): 215-217 (1999)
4. Khaled Karai, Robert D. Grober “ Piezoelectric tip-sample distance control for near field
optical microscopes” Appl. Phys. Lett. 66(14): 1842-44 (1995)
24-66
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
Guiding and Band Edge Measurements of 2-Dimensional Photonic Crystal
Slab Formed by Posts
Sponsors
National Science Foundation ECS 0119452
MRSEC Program of the National Science Foundation - DMR 98-08941
US Air Force Office of Scientific Research - F49620-01-1-0084
Project Staff
Peter T. Rakich, Solomon Assefa, Dr. P. Bienstman, Dr. Steven G. Johnson, Juliet Gopinath, Dr.
Hideyuki Sotobayashi, Dr. Gale S. Petrich, Professor John D. Joannopoulos, Professor Leslie A.
Kolodziejski, Professor Erich P. Ippen, Professor Henry I Smith
Photonic crystals provide a unique way of developing materials with novel optical properties for
the manipulation of light. For this reason, photonic crystals have generated a great deal of
interest as a scientific pursuit and for numerous applications. During the past several years,
many studies have examined the guidance of light in photonic crystals formed by a periodic array
of holes in slab waveguides [1-2]; however, little work has been done on photonic crystals formed
by a periodic array of posts. We demonstrate guidance and present evidence of a photonic band
gap in a two-dimensional photonic crystal, at optical wavelengths, formed by posts [3].
The two types of photonic crystals vary in several respects. However, the most important feature
to note is that a complete photonic band gap can only occur for TE light in structures made with
holes, while this is true only for TM light in photonic crystals formed by posts. This is because the
energy density inside of a post is highly confined when the electric field is aligned with the post,
resulting in a much larger splitting of electromagnetic states. This allows the formation of a
complete photonic band gap with TM polarized light.
The device under study is an asymmetric two-dimensional waveguide formed from posts. The
high index material which forms the guide is composed of GaAs which is grown through
molecular beam epitaxy, while the low index undercladding material AlxOy is formed through an
oxidation of epitaxially formed AlAs. The material surrounding the posts is purposefully over
etched to make the structure as symmetric as is possible. Fabrication specifics can be found in
ref [4]. A schematic of the device can be seen in Figure 1. The device dimensions were chosen
to produce a complete photonic band gap from 1540nm -1740nm for TM polarized light, while
transmitting for TE polarization at these same wavelengths.
Figure 1. Diagram of photonic crystal post devices used for experiment. Green
material is GaAs (n = 3.4) and light blue is AlxOy (n = 1.6).
24-67
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
Measurements were performed with the use of tunable lasers spanning wavelengths 1430 1610nm. A fiber lens assembly and piezo controlled stages were used for coupling at the input.
Collection was performed with nondispersive imaging optics, video cameras and a Ge detector.
The photocurrent signal was processed with a lock-in amplifier and acquired point by point as the
tunable laser was scanned. During all measurements, the guided mode at the output end facet
was simultaneously imaged and measured. In all cases, the polarization of the input-coupled
light was calibrated with a free space polarizer, and the collected light was analyzed with a
polarizer before detection.
Figure 2 is a plot showing the transmittance of the guided light through the photonic crystal for
both TE and TM guided light. Notice that the TM polarization is transmitting at shorter
wavelengths and the transmission rolls off at 1540nm while the TE spectrum is transmitting over
all wavelengths. These initial observations appear to be consistent with the formation of a band
gap for TM light. Additionally, oscillations in the transmitted spectrum appear to be consistent
with the Fabry-Perot cavity formed by the interfaces of the photonic crystal. This is further
evidence of guidance in the photonic crystal. Nearly identical results were observed in three
identically fabricated devices, demonstrating that these measurements and devices are
reproducible.
Figure 2. Transmission spectrum for both TM and TE polarization through photonic crystal
device shown in Figure 1.
In conclusion, preliminary measurements show strong evidence of guidance and the formation of
a photonic band gap in a photonic crystal formed by an array of high index posts. A 20db
suppression of the transmission is observed for TM polarization near the theoretical wavelengths
of the band gap. However, for the purposes of completeness we work to obtain a measurement
of both bandages. Currently under development is a measurement system which is based on a
supercontinuum white light source extending the measurement range from 1200nm-1800nm. A
characteristic supercontinuum spectrum can be seen in figure 3 which was produced by a novel
nonlinear fiber, the properties of which are currently under study. This will provide a more
complete measurement of the photonic band structure, and thus a more thorough analysis of
these devices.
Additionally, we have developed a near-field optical microscope which will enable the detection of
the evanescent field of the waveguide mode which extends outside of the photonic crystal
waveguide. As a near-field fiber probe is brought in close proximity to the waveguide a small
amount of light can tunnel through the probe into a fiber optic. As the probe is scanned over the
waveguide surface, phase and amplitude information can be obtained. This technique promises
to allow many new experimental studies of Bloch waves, lensing and negative refraction.
24-68
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
Figure 3. Supercontimuum generated by novel nonlinear optical fiber. The spectrum was
generated with 150fs pulses at 1550nm.
References
1. S. Rowson, A. Chelnokov, J.-M. Lourtioz, “Two-Dimensional Photonic Crystals in
Macroporous Silicon: From Mid-Infrared to Telecommunications Wavelengths,” J. Lightwave
Tech. 17(11):1989-95 (1999)
2. T. F. Krauss, R. M. De La Rue, and S. Brand, “Two-Dimensional photonic-bandgap
structures operating at near-infrared wavelengths,” Nature 383(6602): 699-702 (1996)
3. S. Lin, E. Chow, V. Hietala, P. Villeneuve, J. D. Joannopoulos, “Experimental Demonstration
of Guiding and Bending of Electromagnetic Waves in a Photonic Crystal” Science, 282(5387):
274-6 (1998)
Reports
Solomon Assefa, “Coupling into Photonic Crystal Waveguides” RLE TR-566 (Cambridge: MIT
Research Laboratory of Electronics, 1998)
24-69
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
Integrated Tunable/Switchable Optical Add-Drop Multiplexer
Sponsors
Pirelli
Project Staff
Dr. Matteo Cherchi, Dr. Daniele Faccio, Dr. Giacomo Gorni, Professor Hermann A. Haus, Dr.
Christina Manolatou, Miloš Popoviü, Dr. Maurizio Tormen, Michael R. Watts
This research covers the functional and electromagnetic design of tunable, switchable integrated
optical add/drop filters. It is part of a project to design a fully integrated optical add/drop module
(OADM) within the wavelength division multiplexing (WDM) framework. The OADM is intended to
add/drop a selected number of channels on a C-band WDM bus within prescribed drop and thru
channel response specifications such as insertion loss, cross-talk, dispersion, polarization
dependent loss (PDL), etc. Finally, the design is to be integrated on a chip, using high index
contrast design for miniaturization.
Polarization-independent performance is extremely difficult to obtain in a high-index contrast
device. The TE-like and TM-like modes in these structures exhibit marked differences in
coupling, propagation in bends, and sensitivity to fabrication errors. In order to obtain polarization
independence, we developed a polarization independent topology (Fig. 1) in which we split the
polarization states at the input, rotate one in order to obtain identical on chip polarizations, and
operate on the signals in parallel with identical structures. The signals dropped from the chip are
recombined electronically, and those continuing in the optical domain are recombined at the
output using the reverse process.
Figure 1. Polarization-independent device topology for a polarization-dependent optical
circuit using a symmetric configuration of integrated polarization splitters and rotators.
The required elements of an integrated polarization splitter and polarization rotator have been
concept proven and are entering the fabrication stage. Diagrams of the basic structures are
presented in Fig. 2. A three layered core was adopted in order to rotate and split the
polarizations using adiabatic following. Adiabatic following makes the structures to be both
fabrication tolerant and wavelength insensitive. For rotator and splitter device lengths of 400Pm
and 150Pm, respectively, a core index of 2.2 and a cladding index of 1.445, a = 0.25Pm and b =
0.75Pm, greater than 99% of the power is transmitted over the entire 1.5-1.6Pm band.
24-70
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
Figure 2. (a) Polarization rotator, (b) polarization splitter,
and (c) integrated polarization splitter and rotator.
Microring resonator channel add/drop filters are being designed to be tolerant to fabrication error.
A third-order drop-port response in Fig. 3 shows 100 overlaid filters with a uniform statistical error
distribution in the coupling coefficients between +6% and -6%, always meeting the prescribed
specification. A dielectric slab MEMS is placed at a controlled distance above the filter rings to
alter the propagation constant within the rings and tune the filter across the frequency spectrum.
Numerical simulation tools have been developed in order to accurately analyze and predict the
performance of the OADM device. These include fully numerical 3D FDTD (temporal) and leaky
waveguide modesolver (spectral) methods, as well as semi-analytic methods based on coupledmode theory.
The most reliable numerical tool for electromagnetic modeling of our dielectric structures is the
Finite Difference Time Domain (FDTD) method. In 3D, the full-wave Maxwell's equations are
propagated by discretization in space and time. No approximations other than the discretization
lead to very accurate results for arbitrary structures over wide frequency spectra. However, large
time and memory requirements of 3D FDTD make it impractical to simulate devices larger than a
few microns in each direction. Hence, the structure under consideration is divided into critical
segments which are simulated individually, and the results are assembled for a full simulated
response.
Figure 3. Fourth-order inline microring filter drop channel response
function with a +/-6% uniform error distribution applied to the ring-ring
coupling coefficients. The filter meets specifications.
24-71
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
In the case of complex resonant structures such as the add/drop filters with vertically coupled ring
resonators and with MEMS tuning slabs, full 3D simulations are required. While final tests of an
optimized structure are best verified by 3D FDTD, the optimization process is done in parts
because of the large computational cost of FDTD. Instead, the parameters of analytic models of
the device under considerations are extracted from smaller FDTD simulations during the design
and optimization process. Fig. 4 shows a logarithmically-scaled electric field map of a ring-ring
coupler simulated in FDTD.
Figure 4. Ring-ring coupling simulation detail (3D FDTD). Logarithmicallyscaled intensity map is shown.
In the case of large non-resonant, nearly reflectionless structures such as the polarization
splitter/rotator, 3D simulations are possible with alternatives such as a "sliding window" FDTD
method.
Electromagnetic design of the ring resonators themselves (radiation Q contrast of leaky modes),
of the MEMS tuning mechanism, and of some couplers is most efficiently done using spectral
methods. To this end a full-vector field modesolver was developed to find the guided and leaky
modes of arbitrarily shaped waveguides with a straight or cylindrical propagation axis. Fig. 5
shows the bend mode of a high loss ring for illustration.
The combination of semi-analytic methods to guide design, and numerical FDTD and
modesolvers has allowed our OADM designs to develop quickly, and the first fabricated filters are
forthcoming.
24-72
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
Figure 5. Leaky mode of primarily horizontal polarization in a 5.8Pm circular ring
resonator showing radiation loss (complex cross-section modesolver computation).
The contours of constant |EU| are non-linearly spaced to better show the small
radiation loss.
24-73
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
Grating Filters with Reduced Radiation
Sponsors
National Science Foundation MRSEC Program - DMR 98-08941
Project Staff
Aristidis Karalis, Professor Hermann A. Haus
The goal of this research project is the design of a radiation-free integrated broad-band FabryPerot filter using deeply etched high index contrast gratings separated by a defect. The radiation
from the defect is to be investigated and means for its reduction or elimination are sought.
A defect between two semi-infinite gratings causes radiation, in general, due to the mismatch of
the mode patterns on either side of the defect. An exception was found for some ideal structures,
when perfect mode-matching between segments is enforced[1]. Ideal structures cannot be
fabricated. We are searching for structures that approach the ideal and can be fabricated.
An infinite periodic guiding structure does not radiate. In the cut-off regime, the electric and
magnetic field are in quadrature throughout the entire cross-section, if the structure supports one
single bounded mode. The aim is to find a defect configuration that matches the mode patterns of
the electric and magnetic fields of the two grating structures to either side of the defect.
Reference
1. M. R. Watts, S. G. Johnson, H. A. Haus, and J. Joannopoulos,” Electromagnetic cavity with
arbitrary Q and small modal volume without a complete photonic band-gap,” Opt. Lett., 27,
pp. 1785-1787, (2002)
24-74
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
Polarization Mode Dispersion
Sponsor
3M Company, DSO National Laboratories
Project Staff
Professor Hermann A. Haus, Professor Erich P. Ippen, Poh Boon Phua
Polarization mode dispersion (PMD) limits long distance ultrahigh-speed optical
telecommunication systems. Due to the statistical nature of PMD, PMD compensators and
emulators have to be adjustable. A variable Differential Group Delay (DGD) module is an
important component in these applications. The common approach to generate variable DGD is
to separate the two orthogonal polarization components using a polarization beam splitter and to
introduce a path difference between them. The two polarization components are then recombined
using a polarization beam combiner. This approach requires mechanical movements, and tends
to suffer from slow speed (sub-second), large output polarization fluctuation and poor control
stability. Alternatively, one can generate variable DGD by concatenating two fixed DGD segments
via a polarization controller. However, this results in a second order PMD vector perpendicular to
the resultant 1st order PMD vector, which causes rotation of the principal state of polarization as
one moves away from the center wavelength.
Recently, we proposed a symmetrical way of concatenating 4 identical fixed DGD segments [1]
so that the resultant DGD is variable while no second order PMD is produced. In addition, the
third order PMD produced is only half the value of the one produced in the concatenation of two
fixed segments with the same DGD tuning range. The schematic of the module is shown in
Figure 1. Two identical blocks are concatenated via a tunable phase-plate, C1 , whose rotation
axis is the x-direction in Stokes space (equivalent to a horizontal linearly polarized birefringence
axis). Each block itself consists of two identical fixed DGD segments of negligible second order
PMD. They are concatenated via C 0 , which is a tunable phase-plate whose rotation axis is the ydirection (equivalent to a 45 degree linearly polarized birefringence axis). In Stokes space
representation, each fixed DGD segment has first order PMD,
U
W
matrix, R , whose rotation axis is the x-direction and rotation angle is
U
{W ,0,0} , and a rotation
TR .
Since the module is constructed with tunable phase-plates, which can either be electro-optic or
magneto-optic, high speed tuning of DGD is possible. Moreover, instead of using polarizationmaintaining fibers as fixed DGD segments, one could use birefringent crystals, which provide
compactness and stability. Alternatively, one could integrate the whole variable DGD module on a
wafer based on MEMS technology, since the fixed DGD segments are simply fixed delay lines in
free space while the tunable phase-plates are finely-adjustable delay lines in free space.
Based on similar concept of coherent 4-segment, we have also proposed another module [2] (see
Figure 2) that can be deterministically controlled to produce variable magnitude of second order
PMD without generating any first order PMD. When it is used with an additional polarization
controller, it can generate an arbitrary second order PMD vector within its designed operation
range. By applying both modules in Figure 1 and 2 together, we can achieve deterministically
controlled PMD compensation and emulation up to second order.
24-75
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
U
WR
U
(W ZR
C1
Block
Block
UU
U
U
W
C0
( RB
Tunable
PhasePlate
W ' W Z'
W
RC0 R)
UU
W ' W Z'
U
C1
W
R
R
0)
U
C0
W
R
Tunable
PhasePlate
R
Tunable
PhasePlate
Figure 1
C
Block 1
U
W
U
W
U
W R C0 R
C0
R
Block 2
R
R
Tunable
PhasePlate
U
W
C0
R
Tunable
PhasePlate
Fixed
PhasePlate 1
Fixed
PhasePlate 2
Tunable
PhasePlate
Figure 2
References
[1] P.B. Phua and Hermann. A. Haus, “Variable Differential Group Delay Module Without Second
Order PMD”, J. Lightwave Technol., 20, pp.1788-1794 (2002)
[2} P.B. Phua and Hermann. A. Haus, “Variable Second Order PMD Module Without First Order
PMD”, J. Lightwave Technol., 20, pp.1951-1956 (2002)
24-76
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
Optical Phase Control and Stabilization Techniques
Attosecond Synchronization of Modelocked Lasers
Sponsors
MIT Lincoln Laboratory - ACC 334
National Science Foundation - ECS-0119452
Office of Naval Research - N-00014-02-1-0717
Project Staff
Dr. Thomas R. Schibli, Jung-Won Kim, Alex Killi, Onur Kuzucu, Juliet T. Gopinath, Sheila
N.Tandon, Dr. Gale S. Petrich, Professor Leslie A. Kolodziejski, Professor James G. Fujimoto,
Professor Erich P. Ippen, Professor Franz X. Kaertner
The synchronization of pulse trains from independent modelocked lasers with sub-cycle timing
fluctuations is the most important step for the coherent synthesis of optical single-cycle pulses.
Ideally, the relative timing jitter should be less than a tenth of an optical cycle for a high quality
synthesized pulse stream. At a wavelength of 1 Pm, this requires a timing jitter of 330
attoseconds or less, measured over the full Nyquist bandwidth, i.e. half the laser repetition rate.
Several groups have investigated the possibility of active [1] and/or passive [2,3] synchronization
of multiple lasers. However, a sub-femtosecond timing jitter over the Nyquist bandwidth has never
been achieved.
We demonstrated a new method of synchronization for modelocked lasers, in which the timing
jitter between the two lasers is detected by a balanced cross-correlator, see Fig. 1, the optical
equivalent of a balanced microwave phase-detector. The signal is then fed back via an electronic
control loop in order to keep the two lasers synchronized. This method enables a drift-free and
temperature-independent synchronization between two individual lasers, a task that is difficult to
achieve with all-electronic schemes. To ensure the long-term stability of the system, the two laser
beams are combined inside the control loop.
Figure 1: Experimental setup of the synchronized lasers. Cr:fo: passively
mode-locked Cr:forsterite laser, Ti:sa: passively mode-locked Ti:sapphire laser;
SFG: sum-frequency generation; all bandpass filters transmit only the sumfrequency (1/496nm = 1/833nm+1/1225nm). The two beam splitters consist of
a thin fused silica glass substrate coated with a semi-transparent metal film.
The third correlator is used to generate the graphs shown in Fig. 3a.
24-77
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
The coherent synthesis of a single cycle pulse demands an optical spectrum as wide as 1.5
octaves. This requires both lasers to generate an ultra-wide and very stable spectrum. For long
term stable operation, we use the octave spanning [4], prism-less Ti:sapphire laser [5,6]
described above. The repetition rate of the laser is 82 MHz, and it emits about 100 mW of modelocked power through a 1% output coupler when pumped with 3.8 W of 532 nm light. To obtain
stable modelocking in the Cr:forsterite laser, we used a novel broadband InGaAs saturable
absorber on a large area, high-index contrast AlGaAs/AlxOy mirror [7], which enables the
generation of sub-30 fs pulses at 1230 nm wavelength in a self-starting configuration. This laser
emits about 60 mW of modelocked power through a 3% output coupler when pumped with
approximately 2 W of 1064 nm light.
Fig. 2 shows the spectrum of the two lasers
(solid line) at the output. The dashed lines
indicate the extent of the individual laser
spectra in the vicinity of the overlap region.
The shaded region indicates the spectral
region filtered out to record the difference in
carrier envelope offset frequency between
the two lasers (see. Fig. 4). For phase
coherent superposition of the two lasers, the
pulse envelopes of the two lasers must be
synchronized; and, in addition, the
Figure 2. Optical spectra of the mode-locked
difference in the carrier envelope offset
Ti:sapphire and Cr:forsterite laser. The dashed
frequency between the two lasers must be
lines indicate the spectra of the individual
stabilized. Synchronizing the pulse trains
lasers in the vicinity of the spectral overlap,
with sub-cycle precision is the most
and the shaded region indicates the spectral
challenging part of this synthesis process.
region used to detect the difference in carrier
To overcome the typical problems posed by
envelope offset frequency between the two
balanced microwave mixers previously used
lasers shown in Figure 4.
for this task [1], we employ the optical
equivalent of such a device: a balanced
cross-correlator. As shown in Fig. 1, the outputs of the two lasers are combined on a broadband
metallic beam splitter grown by P. O’Brien at MIT Lincoln Laboratory. One part of the combined
beam is directed to two nearly identical cross-correlators using 1 mm-thick LBO-crystals phase
matched for SFG of 833nm and 1225nm light. The only difference between the two correlators is
a 3 mm-thick fused silica window in the optical path of one of them. This glass inserts a group
delay between 833 nm and 1225 nm to offset the pulses emitted by the Cr:forsterite and
Ti:sapphire lasers by about 45 fs with respect to each other. The balanced detector output is then
proportional to the time difference between both laser pulses, and in the vicinity of zero timing
offset this detector acts like a balanced phase detector operating in the range of tens of THz. The
output of this balanced cross-correlator as a function of time difference between the Cr:forsterite
and the Ti:sapphire pulses is shown in Fig. 3b).
We used the signal from this balanced correlator to lock the repetition rates of the two lasers by
controlling the cavity length of the Ti:sapphire laser with cavity mirrors mounted on piezo-electric
transducers in a manner similar to that of ref. [1]. The first beam splitter used to combine the two
output beams from the lasers is inside this control loop. Since the output beam shown in Fig. 1
originates from this beam splitter, temperature drifts, acoustic noise or beam-fluctuations always
affect both laser beams in the same way as they both travel along identical paths. Therefore,
external noise cannot corrupt the relative jitter, and the output behaves as if it originated from the
same source. Fig. 3a) shows the resulting timing jitter measurement made with the out-of-loop
cross-correlator shown in Fig. 2. The residual timing jitter over the detector’s bandwidth of
2.3 MHz is 299 as ±104 as. The stated error is determined from the amplitude noise measured at
the peak of the cross-correlation. As in most passively modelocked laser systems, the main
contribution to the timing jitter has frequency components up to a few times the relaxation
oscillation frequency of the laser [8]. In the current system, the relaxation oscillation frequencies
24-78
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
are roughly 70 kHz for the Ti:sapphire laser and 140 kHz for the Cr:forsterite laser. Therefore,
noise above 2.3 MHz can usually be neglected.
Figure 3. a) Timing jitter determined from the
amplitude noise of the SFG of the out of loop
cross-correlator (see Fig.1). The rms-jitter
measured in a 2.3 MHz BW results in 299 as
±104 as. b) The output of the balanced crosscorrelator as a function of time difference
between the two laser pulses.
Figure 4. Heterodyne beat between the
Cr:forsterite and the Ti:sapphire lasers obtained
from the spectral region shown in Fig. 2. The two
beat signals below the repetition rate of 82 MHz
represent the difference in the carrier envelope
offset frequency of both lasers. The RF-analyzer
filter-bandwidth is 30kHz. The noise-floor is
caused by the trans-impedance amplifier and
poses only a technical limitation.
As soon as the repetition rates of the two lasers are locked together, we observe a strong beat
signal in the overlap region of the optical spectrum (see Fig. 4). To avoid saturation of the
detector only a small part of the optical spectrum was directed to the diode (shaded region in Fig.
2). As described in ref. [9], the beat signal represents the difference in the carrier-envelope offset
frequency 'fCEO between the two lasers. In contrast to previous results, it is now possible to
obtain this beat without the use of spectral broadening of the mode combs, which helps to provide
an exceptionally large signal-to-noise ratio of about 50 dB in a 30 kHz bandwidth. This signal can
also be used to lock the optical frequencies in the overlap region together, and consequently, we
can generate a fully coherent mode comb consisting of the sum of the two laser spectra that
currently span more than an octave in bandwidth.
References
1.
2.
3.
4.
5.
6.
R. K. Shelton, S. M. Foreman, L. Ma, J. L. Hall, H. C. Kapteyn, M. M. Murnane, M. Notcutt,
and J. Ye, “Subfemtosecond timing jitter between two independent, actively synchronized,
mode-locked lasers,” Opt. Lett. 27(5): 312-4 (2002).
A. Leitensdorfer, C. Fuerst, and A. Laubereau, “Widely tunable two-color mode-locked
Ti:sapphire laser with pulse jitter of less than 2 fs,” Opt. Lett. 20(8): 916-8 (1995).
Z. Wei, Z. Kobayashi, Z. Zhang, and K. Torizuka, "Generation of two-color femtosecond
pulses by self-synchronizing Ti:sappire and Cr:forsterite lasers," Opt. Lett. 26(22): 1806-8
(2001).
R. Ell, U. Morgner, F. X. Kaertner, J. G. Fujimoto, E. P. Ippen, V. Scheuer, G. Angelow, T.
Tschudi, M. J. Lederer, A.Boiko, B. Luther-Davies, "Generation of 5-fs pulses and octavespanning spectra directly from a Ti:sapphire laser," Opt. Lett. 26(6): 373-5 (2001).
T. R. Schibli, L. M. Matos, F. J. Grawert, and F. X. Kaertner, "Continuum generation in a
prism-less Ti:sapphire laser," Proceedings of the Ultrafast Phenomena, Vancouver, Canada,
May 12-17, 2002.
F. X. Kaertner, U. Morgner, T. R. Schibli, E. P. Ippen, J. G. Fujimoto, V. Scheuer, G.
Angelow, and T. Tschudi, "Ultrabroadband double-chirped mirror pairs for generation of
octave spectra," JOSA B 18(6): 882-5 (2001).
24-79
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
7.
8.
9.
D. J. Ripin, J. G. Gopinath, H. M. Shen, A. A. Erchak, G. S. Petrich, L. A. Kolodziejski, F. X.
Kaertner, and E. P. Ippen, "Oxidized GaAs/AlAs mirror with a quantum-well saturable
absorber for ultrashort-pulse Cr4+:YAG laser," Opt. Comm. 214: 285-9 (2002).
J. Son, J. V. Rudd, and J. F. Whitaker, “Noise characterization of a self-mode-locked
Ti:sapphire laser,” Opt. Lett. 17(10): 733-5 (1992).
Z. Wei, Y. Kobayashi, and K. Torizuka, “Relative carrier-envelope phase dynamics between
passively synchronized Ti:sapphire and Cr:forsterite lasers,” Opt. Lett. 27(23): 2121-3
(2002).
24-80
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
Few-Cycle Nonlinear Optics and Carrier-Envelope Phase Effects
Sponsor
National Science Foundation – ECS 0217358
Project Staff
Christian Jirauschek, Juhi Chandalia, Dr. Uwe Morgner, Oliver D. Mücke, Thorsten Tritschler,
Professor Martin Wegener, Professor Franz X. Kaertner
The carrier-envelope offset (CEO) phase, ĭ, is the relative phase between the rapidly oscillating
carrier wave of a laser pulse and its electric field envelope. The time derivative of ĭ is the CEO
frequency, fCEO. As mentioned above, control over fCEO is of great importance for frequency
metrology. Several approaches have been reported to measure the CEO frequency for pulses
directly emitted from mode-locked oscillators [1-5]. The underlying idea is to consider the
interference resulting from the superposition of the fundamental pulse spectrum, which by
definition has a phase 1ĭ, and its second (third) harmonic spectrum, which has a phase 2ĭ (3ĭ).
The resulting interference on a photo detector shows the difference phase 1ĭ (2ĭ), which can be
used to determine the CEO frequency fCEO (2fCEO). For extremely short pulses whose spectrum
covers one octave, such interference can be immediately realized by generating the second
harmonic in a suitable crystal [4]. However, if the laser does not yet reach an octave, the
fundamental needs to be spectrally broadened to generate sufficient spectral overlap. This has
been realized by using self-phase modulation (SPM) in optical fibers [1, 2]. However, such an
approach has the inherent drawback that the CEO phase changes within the setup due to the
fiber dispersion. Ideally, one would like to generate the spectral components, for example due to
SPM and SHG, in a medium (see Figure 1), which is so thin that ĭ does not change within the
medium. Ultimately, this would allow us not only to measure the CEO frequency but also to
determine the CEO phase directly.
eiI
Fundamental
F(2)
e2iI
SPM
eiI
0
Z0
2Z 0
3Z0
Figure 1: Schematic spectra of a pulse covering slightly less than one octave on an
optical frequency scale (top), together with the spectra generated by instantaneous F(2)processes (middle) and self-phase modulation (bottom). The red shaded area is where
the interference occurs.
In collaboration with researchers from the University of Karlsruhe, Germany, a simple technique
of determining the CEO frequency [6] has been developed. In the experiment, 5-fs optical pulses
from one of our broadband Ti:sapphire lasers with an average power of several tens of milliwatts
were tightly focused onto a ZnO sample, which has no inversion symmetry. The light emitted into
the forward direction was filtered to remove the prominent fundamental beam and spectrally
analyzed. Two sample thicknesses, 100 µm and 350 nm, were tested in the experiment. The
optical and electrical spectra of the output are shown in Figure 2 and Figure 3, respectively.
These results reveal the following two aspects: 1) Using about 100 µm thick ZnO single crystals,
24-81
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
we find a large peak (> 30 dB above noise level at 10 kHz bandwidth and 64 mW average power)
at the CEO frequency due to the overlap of the broadened spectrum via SPM and the generated
SHG. The simplicity and robustness of the approach make it attractive for applications in
frequency metrology or for stabilization of the CEO frequency. 2) A peak at the CEO frequency is
still observed using a 350 nm thin ZnO film, within which the CEO phase has negligible change.
This might pave the way for measuring the CEO phase itself and ultimately completing the
characterization of laser pulses. A further interesting result is the observation of a beat signal at
twice the carrier-envelope frequency. The origin of this signal could be related to the recently
discovered phenomenon of Carrier-Wave-Rabi-Flopping [5]. However, several competing
processes might be at its origin and further investigations are necessary.
Figure 3: RF spectra, 10 kHz resolution and
video bandwidths. (a) 100 µm ZnO single
crystal corresponding to Fig. 2, 455-480 nm
optical filter, (b) 350 nm ZnO epitaxial layer,
455-500 nm optical filter. The peaks at the
repetition frequency fr, the CEO frequency fĭ,
its second harmonic 2 fĭ and the mixing
Figure 2: The output spectra from the 100-µm
ZnO vs. time delay of the Michelson
interferometer. (a) I = 0.15 I0, (b) I = 2.04 I0,
and (c) I = 2.04 I0 with different saturation of
the grey scale. The curve labeled IAC is the
independently
measured
interferometric
References
1. A. Apolonski, A. Poppe, G. Tempea, C. Spielmann, T. Udem, R. Holzwarth, T. W. Hänsch,
and F. Krausz, "Controlling the Phase Evolution of Few-Cycle Light Pulses," Phys. Rev. Lett.,
85(4): 740-43 (2000).
2. D. J. Jones, S. A. Diddams, J. K. Ranka, A. Stentz, R. S. Windeler, J. L. Hall, and S. T.
Cundiff, ”Carrier-envelope phase control of femtosecond mode-locked laser and direct optical
frequency synthesis,” Science 288(5466): 635 (2000).
3. H. R. Telle, Steinmeyer, A. E. Dunlop, J. Stenger, D. H. Sutter, and U. Keller, “Carrierenvelope offset phase control: a novel concept for absolute optical frequency measurement
and ultrashort pulse generation,” Appl. Phys. B 69(4): 327-332 (1999).
24-82
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
4. U. Morgner, R. Ell, G. Metzler, T. R. Schibli, F. X. Kaertner, J. G. Fujimoto, H. A. Haus and E.
P. Ippen, "Nonlinear optics with phase-controlled pulses in the sub-two-cycle regime," Phys.
Rev. Lett., 86(24): 5462-5 (2001).
5. O.D. Mücke, T. Tritschler, M. Wegener, U. Morgner and F.X. Kaertner, "Role of the absolute
optical phase of few-cycle pulses in non-perturbative resonant nonlinear optics," Phys. Rev.
Lett. 89(16), 127401-04, (2002).
6. O. D. Mücke, Th. Tritschler, M. Wegener, U. Morgner, and F. X. Kaertner, “Determining the
Carrier-Envelope Offset Frequency of 5fs Pulses Using Extreme Nonlinear Optics in ZnO,”
Opt. Lett. 27(23): 2127-2129 (2002).
24-83
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
Active Modelocking Using a Nonlinear Fabry-Perot Modulator
Sponsor
National Science Foundation ECS-0217358
Project Staff
Richard Ell, Wolfgang Seitz, Dr. Uwe Morgner, Dr. Thomas Schibli, Professor Franz X. Kaertner
The dynamics of laser oscillators can be directly controlled by modulating the intracavity losses.
Over the last two years we developed a new approach of optically driven loss modulation by
means of a nonlinear semiconductor mirror based on a Fabry-Perot structure (Fabry-Perot
modulator, FPM). The structure of the device is shown in the inset of Figure 1.
100
reflectivity (%)
80
60
(1)
40
(2)
20
(3)
(4)
0
0.98 1.00 1.02 1.04 1.06 1.08 1.10 1.12 1.14
wavelength (µm)
Figure 1. Measured reflectivity of the type-I-FPM (solid curve) and calculated reflectivity
(dashed curve). The structure of the FPM is shown below the measured curve: (1) Fresnel
reflection at the air-InGaAs interface, (2) 5 µm InGaAs layer, (3) 15 pair GaAs/AlAs quarterwave Bragg stack centered at 1.064 µm, (4) GaAs(100)-substrate.
The modulation depth can be several percent and the response time is dominated by the
recombination time of the generated free carriers inside the semiconductor, which can be
reduced by ion-implantation. The optical characteristics have been studied via spectrally resolved
two-color pump-probe spectroscopy. Figure 2 shows the differential reflectivity change of the
Fabry-Perot. Applications of the FPM are the synchronization of the pulse trains of independently
mode-locked laser oscillators, i.e. a ps-Nd:YVO4 laser locked to a fs-Ti:sapphire laser in a
master-slave configuration[1]. However, the device can have a large variety of applications. We
demonstrated active mode locking of a solid state laser by an optically driven FPM [2] as shown
in Figure 3. The resulting pulse widths of the actively mode-locked Nd:YVO4 laser are as short as
6 ps, which is comparable to passively generated pulses.
So far, these experiments have been performed using a Ti:sapphire laser as the optical drive for
the modulation. However, in many cases this expensive source can be replaced by a cheap,
actively modulated diode laser. Then the FPM together with the drive might become an
interesting alternative to other mode locking schemes.
24-84
24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24
RLE Progress Report 145
delta R / R (%)
2%
timedelay:
4.25 ps
3.5 ps
2.75 ps
2 ps
1.25 ps
0.5 ps
-0.25 ps
-1 ps
-1.75 ps
-2.5 ps
-3.25 ps
-4 ps
1016.0
1018.0
1020.0
1022.0
wavelength (nm)
Figure 2. Temporally and spectrally resolved measurement of reflectivity change of the
Fabry-Perot modulator during the carrier generation process. The time delay between
adjacent curves is 750 fs. The dots indicate the shift of the resonance. (For clarity the curves
are vertically displaced.)
1
(b)
WFWHM (ps)
0.1
0.01
-25
0
25
pulse width
intensity (arb.)
(a)
12.5
(c )
10.0
7.5
5.0
5
time delay (ps)
10
Wr1/2
15
20
1/2
)
( ps
Figure 3. (a) Measured intensity autocorrelation traces of an actively mode-locked Nd:YVO4
laser on a logarithmic scale for FPMs with different carrier life times. (b) Measurement of the
pulse width dependence on the modulating power on the FPM1 on a double logarithmic
scale. The dashed line is a linear fit of the measured data points (filled circles) representing
the dependence WFWHM v Pm-1/4 in agreement with active modelocking theory. (c) Shortest
pulse widths WFWHM achieved with the different samples plotted as a function of the square root
of the carrier lifetime Wr of the samples. The dashed line is a linear fit of the measured data
points (filled circles).
References
1. W. Seitz, T. R. Schibli, U. Morgner, F. X. Kaertner, C. Lange, and W. Richter, “Passive
synchronization of two independent laser oscillators with a Fabry-Perot modulator,” Opt. Lett.
27(6), 454-456, 2002.
2. W. Seitz, R. Ell, U. Morgner, T. R. Schibli and F. X. Kaertner, M. J. Lederer and B. Braun, “Alloptical Active Mode-locking with a Nonlinear Semiconductor Modulator,” Opt. Lett. 27(24), 220911 (2002).
24-85