Optics and Quantum Electronics - RLE at MIT · 2019. 11. 20. · 24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24 RLE Progress Report 144 24-4

24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24 RLE Progress Report 144

24-1

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 Dr. Kazi S. Abedin, Dr. Stephane Bourquin, Dr. Mark Brezinski, Dr. Katherine Hall, Christian Koos, Dr. Chris Kroeger, Dr. Christina Manolatou, Dr. Chan H. Park, Dr. Lelia A. Paunescu, Dr. Poh-Boon Phua, Dr. Thomas R. Schibli, Karl Schneider, Dr. Ping Xue Graduate Students Aaron Aguirre, Juhi Chandalia, Ravi Ghanta, Juliet Gopinath, Felix Grawert, Matthew Grein, Paul Hertz, Pei-lin Hsiung, Leaf Jiang, Mohammed Jalal Khan, Tony Ko, Andrew Kowalevicz, O. Onur Kuzucu, J.P. Laine, Nirlep Patel, Milos Popovic, Rohit Prasankumar, Peter Rakich, Daniel Ripin, Bryan Robinson, Shelby Savage, Hanfei Shen, Jason Sickler, Laura Tiefenbruck, Aurea Tucay, Michael Watts, Samuel Wong Undergraduate Students Karen Robinson Technical and Support Staff Mary Aldridge, Donna Gale, Cindy Kopf Research Areas and Projects Ultrashort Pulse and Laser Generation Technology

Few-Cycle Pulse Generation and Dispersion Compensating Laser Optics • Double-Chirped Mirror Design • Record Pulse Generation of 5 fs Pulses with Octave Bandwidth • Few-Cycle Laser Pulses from Cr:YAG and Cr:Forsterite Lasers at 1.3 µm and 1.5 µm • Ultra-broadband Prismless Ti:sapphire lasers

Ultra-low-threshold, Low Cost, Femtosecond Laser Technology Continuous Wave and Q-switched Mode-locked Microchip Lasers Broadband Oxidized Saturable Bragg Reflector Ultrafast Cr:YAG Laser Non-epitaxially Grown Semiconductor-Doped Silica Films for Laser Modelocking Active Harmonically Modelocked Fiber Lasers High-Repetition-Rate Fiber Ring Laser Passively Modelocked With a Saturable Absorber Mirror Self-Stabilized Harmonic Passively Modelocked Stretched-Pulse Erbium Fiber Ring Laser Noise in Harmonically Modelocked Lasers Timing Jitter Reduction in Modelocked Semiconductor Lasers with Photon Seeding Quantum-Limited Noise Performance of a Semiconductor Modelocked Laser Experimental Demonstration of a Timing Jitter Eater Novel Low-Coherence Light Sources for Optical Imaging Applications

• Spectral Broadening in Tapered Fiber using a Femtosecond Nd:Glass Laser • Continuum Generation in the Visible Wavelength Region Using High Nonlinearity Air-

Silica Microstructure Optical Fibers • Broadband Fluorescence Sources Using Ti:Al2O3 Crystals


24-2

Optical Phase Control and Stabilization Techniques

Control and Stabilization of the Absolute Optical Phase Evolution Few-Cycle Nonlinear Optics and Absolute Optical Phase Effects in Carrier-Wave Rabi Flopping Nonlinear Fabry-Perots for Synchronization of Independent Laser Oscillators Control of the Absolute Optical Phase in Picosecond Lasers Control of Q-switching Instabilities in Mode-locked Lasers by Active Feedback

Ultrafast Phenomena and Quantum Electronics

Resonance Raman Studies on 0.4nm Single Wall Carbon Nanotubes Enhanced Light Extraction and Lasing in Two Dimensional Photonic Crystals Highly Nondegenerate Four-Wave Mixing in Tapered Microstructure Fiber Femtosecond Pump-Probe Spectroscopy Using a Two-Dimensional Smart Pixel Detector Array

Photonics and Devices

Resonant Channel Add/Drop Filters Fiber-Chip Coupling Polarization Mode Dispersion Suppression of Radiation from Quarter-Wave Shifted Bragg Resonators Micron-size Bending Radii in Silica-based Waveguides Micromachined Photonic Devices using Nonlinear Materials Processing

Publications Conference Papers


24-3

Ultrashort Pulse and Laser Generation Technology Few-Cycle Pulse Generation and Dispersion Compensating Laser Optics Sponsors U.S. Air Force – Office of Scientific Research Grant F49620-01-1-0084 MIT Lincoln Laboratory Grant ACC 334 MIT Presidential Fellowship Project Staff Felix Grawert, Onur Kuzucu, Dr. Uwe Morgner Dr. Thomas R. Schibli, Professor Franz X. Kärtner, Professor James G. Fujimoto, Professor Hermann A. Haus, Professor Erich P. Ippen, Richard Ell 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, 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. We have developed a theoretical model which provides a foundation for understanding and optimizing 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 Design Solid state lasers can have gain over extremely broad bandwidths of several hundred nanometers, enabling both the generation of few cycle pulse durations or longer pulse durations with broad tunability. Intracavity dispersion is the limiting factor in laser performance for sub-10 fs pulses due to their broad bandwidth. Intracavity prisms have been used for dispersion compensation. However, prisms have parasitic higher order dispersion, which limits pulse duration and also makes laser tuning difficult. Self-phase modulation (SPM) is a temporal nonlinear effect also originating in the Kerr nonlinearity at high intensities that generates new frequencies and spectrally broadens the pulse. When dispersion is carefully compensated, SPM can aid in generation of pulses with spectra that extend beyond the gain bandwidth. Double chirped mirrors (DCMs) have recently emerged as a powerful technology which permits intracavity dispersion management [1-6]. DCMs enable both broadband operation and intracavity dispersion compensation without prisms. Figure 1 shows the differences between standard Bragg mirrors, simple chirped mirrors, and double-chirped mirrors. In a simple chirped mirror the Bragg-wavelength of the grating increases with increasing penetration depth of light into the mirror. Light is reflected at the index discontinuity between the surrounding air and the first layer in the mirror and during chirping of the Bragg-wavelength. These reflections interfere with the strong reflection from the back of the mirror which leads to Gires-Tournois like interferences resulting in strong dispersion oscillations. In a double-chirped mirror we avoid these spurious reflections by a consistent impedance matching from the ambient air all the way to the classical turning point of the wave in the chirped mirror. This is achieved by matching from air to the first low index layer of the actual DCM by a broadband AR-coating. Additional spurious reflections inside the chirped mirror are avoided by slowly switching on the grating, which can be done by


24-4

increasing the thickness of the high-index layer adiabatically from an almost vanishing value to the corresponding quarter wave thickness. The proper design of DCMs enables the intracavity dispersion to be almost perfectly compensated, removing all higher order dispersion terms over the laser bandwidth which limit the generation of short pulses. We have recently demonstrated a 5 fs Ti:Al2O3 laser where we compensated in average dispersion up to sixth order over one octave of bandwidth [7].

Figure 1: Schematic drawing of layer sequence for various dielectric mirrors: a) standard quarter wave Bragg mirror, b) simple chirped mirror, c) double chirped mirror (DCM). We have recently developed an analytic theory for the design of dispersion compensating mirrors and mirror pairs [4, 5, 8, 9]. Dispersion compensating mirrors, and especially double-chirped mirrors developed in our group, provide high reflectivity and well controlled dispersion over 400 nm in the case of single mirrors or even over one octave using specially matched mirror pairs. These components are a prerequisite for miniaturized ultrabroadband femtosecond lasers necessary for biomedical optical imaging, ultrafast instrumentation, and other applications. In our previous work pulses shorter than 5.4 fs at a center wavelength of 800 nm, corresponding to a bandwidth greater than 350 nm, were generated directly by a Kerr-lens mode-locked Ti:sapphire laser at a repetition rate of 90 MHz and an average output power of 200 mW. In this laser, the pulse duration was limited by the bandwidth over which the DCMs can balance the dispersion inside the cavity. The gain bandwidth of Ti:Al2O3 extends from 600 nm to almost 1200 nm and would enable the generation of even shorter pulses, in the single-cycle regime. However, increasing the bandwidth of the DCMs results in stronger oscillations of the group delay, which limits the pulse duration. This effect is a consequence of fundamental properties of DCM operation. This problem can be solved by developing complementary sets of mirrors. The phase of the group delay oscillations can be controlled since it depends on the index and layer thickness of the dielectric layers. Two complementary sets of ultra-broadband DCMs can be designed with excess group delay oscillations that are exactly out of phase. Using these two mirror sets, the excess oscillations can be made to cancel. This novel approach enables dispersion compensation over a much greater spectral range than possible using a single mirror set and enables the generation of unprecedented bandwidths and pulse durations. Figure 2 shows the calculated and measured reflectivity and group delay dispersion of the mirror pairs. The mirrors are designed such that the dispersion of all intracavity components is exactly compensated up to sixth-order.

Bragg-Mirror: TiO / SiO2 2 - LayersλB 4

Chirped Mirror: Bragg-Wavelength chirped SiO - Substrate

2 λ1λ2 λ1λ2 >

Negative Dispersion:

Double-Chirped Mirror: Bragg-Wavelength and Coupling Chirped

Air

“Impedance” - Matching

d =h λB/4

SiO - Substrate

2 AR- Coating

AirSiO - Substrate

2

λB

a)

b)

c)


24-5

100

80

60

40

20

0

REF

LEC

TIVI

TY (%

)

140012001000800600WAVELENGTH, NM

-600

-400

-200

0

200

400

600

GD

D, FS

2

100

80

60

40

20

0

REF

LEC

TIVI

TY (%

)

140012001000800600WAVELENGTH, NM

-600

-400

-200

0

200

400

600

GD

D, FS

2

Figure 2: Calculated and measured reflectivity and dispersion of ultrabroad bandwidth DCM pairs specially designed to cancel parasitic dispersion oscillations. Record Pulse Generation of 5 fs with Octave Bandwidth Using this novel design, we recently demonstrated the generation of pulses with durations of only 5 fs and spectral bandwidths over one octave directly from a Ti:Al2O3 laser [7]. This is the shortest pulse ever generated directly from a laser. Figure 3 shows a schematic of our laser system. This laser has a standard z-cavity design with the addition of a second intracavity fold. A second focus is generated by M4 and M5 in which a 2.4 mm thick plate of BK7 is positioned. This provides enhanced self phase modulation and increases the laser bandwidth.

Figure 3: Schematic of Ti:sapphire laser that generates octave bandwidths and 5 fs pulse durations. A second focus was used to increase bandwidth via self phase modulation. The optical power spectrum at the laser output is displayed in Figure 4a) on a linear and logarithmic scale. On a log scale the spectrum extends from 600 to 1350 nm above the noise floor. The FWHM of the corresponding pulse assuming a flat phase would be 4.3 fs. The structure in the spectrum is correlated with oscillations in the measured intracavity GDD suggesting that improved DCM design would improve the laser performance. The two peaks at 700 nm and 1050 nm are caused by the increasing output coupling mirror transmission. Despite the large oscillations in the GDD caused by fabrication tolerances, the spectrum is relatively smooth, which can be explained by the enhanced SPM due to the second intracavity focus and the strong KLM action, which continuously cleans up the pulse shape. The interferometric autocorrelation measurement is displayed in Figure 4b). A phase retrieval algorithm [10] was used to reconstruct the actual pulse shape from the autocorrelation. The intensity envelope of the reconstructed pulse indicates a FWHM of 5 fs. The phase measurement suggests that reductions in duration to 4.5 fs should be possible.


24-6

1.0

0.8

0.6

0.4

0.2

0.0

SP

EC

TRU

M,

a.u.

12001000800600WAVELENGTH, NM

10-5

10-4

10-3

10-2

10-1

100

b)

8

6

4

2

0

IAC

3020100-10-20TIME DELAY, FSa)

Figure 4: a) Interferometric autocorrelation of the 5 fs pulse b) corresponding spectrum on a linear and logarithmic scale. These are the shortest pulses ever generated directly from a laser. Few-cycle laser pulses from Cr:YAG and Cr:Forsterite lasers at 1.3 µm and 1.5 µm Dispersion management techniques using DCMs can be applied to a wide range of solid state laser materials. The spectral range at 1.3 µm and 1.5 µm is of particular interest because it falls into the 2nd and 3rd telecommunication window. Our group recently demonstrated all-solid-state, Kerr-lens mode-locked Cr:forsterite and Cr:YAG lasers producing 14 fs pulses with 250 nm bandwidth at 1.3 µm and 19 fs pulses with 240 nm bandwidth at 1.5 µm, respectively [11,12]. In this wavelength range dispersion compensation is particularly challenging because higher order dispersion becomes the dominant factor near the zero dispersion wavelength. Higher order dispersion is hard to compensate over an extended wavelength range even with DCMs. The total spectral coverage on a logarithmic scale of the three few-cycle lasers, Ti:sapphire, Cr:forsterite and Cr:YAG, developed so far is shown in Figure 5. Figure 5: Total spectral coverage of the few-cycle laser systems Ti:sapphire, Cr:forsterite and Cr:YAG. The spectral coverage of these three novel laser sources will allow us to perform unique spectroscopic investigations and excitations of semiconductors and semiconductor devices in the next years. Some of them are further discussed below. Ultra-broadband prismless Ti:sapphire lasers To date, extremely stable, all mirror-dispersion controlled Ti:sapphire lasers emitting 8 fs pulses with spectra extending over 105 nm full width at half-maximum (FWHM) have been demonstrated [13]. As shown above, additional use of prism pairs for dispersion compensation results in an increased flexibility in the mirror design and, therefore, in pulses as short as 5 fs with octave spanning spectra directly from the laser. However, it has been found, that the fluctuations in the intracavity beam-pointing angle translates into undesired dispersion fluctuations in lasers with


24-7

prisms, which is detrimental in experiments concerned with carrier-envelope phase stabilization. Here, an all mirror-dispersion controlled Ti:sapphire laser, which generates sub-8fs pulses and most notably, emits over a spectral width as large as 250 nm (FWHM) on a linear scale and even generates significant spectrum over 400 nm on a logarithmic scale. This ultrabroadband laser is suitable for carrier-envelope phase stabilization based on interference of second and third harmonic light [14] and may have many other applications, such as in high-resolution optical coherence tomography [ 15]. Due to the reduced intracavity-losses, an optical to optical efficiency of more than 10% in the mode-locked state is achieved allowing for pump-powers below 3 W with a total output power of typically 300 mW.

Pump Laser

Telescope DCM - 10cm ROC

Ti:Al22OO

33

2.4% Output Coupler DCM DCM Figure 1. Schematic diagram of the astigmatically compensated, prism-less Ti:sapphire laser. All mirrors except the output coupler mirror are double-chirped mirrors (DCMs), which compensate the second and third order dispersion of the laser crystal and the air, which resides inside the laser cavity.

1.0000.9990.998

1.0

0.5

0.0

Ref

lect

ivity

-80

-60

-40

-20

0

GD

D[fs

2 ]

1000950900850800750700wavelength (nm)

b)a)

wavelength (nm)wavelength (nm) Fig. 2: a) Reflectivity of the DCMs. b) Desired (dotted line), designed (solid line) and measured (dashed line) group delay dispersion of the DCMs. Figure 1 shows a schematic diagram of the prism-less Ti:sapphire laser. All cavity mirrors except the output coupler are double-chirped mirrors (DCMs). The DCM-design is such that there is high transmission of the pump light at 532 nm and high reflectivity greater than 99.9% from 700-1000 nm. On average, the DCMs generate within six bounces per roundtrip the necessary dispersion to compensate the positive second and third order dispersion of the laser crystal and the 4.5 meters of air which are present in this 75 MHz laser cavity, see Fig. 2. The laser crystal has a path length of 2.3 mm and absorbs approximately 70% of the pump light emitted from a frequency doubled, diode pumped Nd:YVO4 laser. Figure 3 shows a typical spectrum and the corresponding interferometric autocorrelation of the pulses emitted by the laser. A phase-retrival algorithm [10] results in 7.8fs pulses. (The Fourier transform limit of the spectrum shown in Fig. 3 assuming constant spectral phase is 6.9 fs).


24-8

1.0

0.8

0.6

0.4

0.2

0.0Nor

m. S

pect

ral P

ower

14001 2001000800600Wavelength [nm]

-80

-60

-40

-20

0

Norm

. Spectral Power (log)

8

6

4

2

0

IAC

-40 -20 0 20 40Time [fs]

Measured Fitted (7.8fs)

Figure 3. Typical spectrum and interferometric autocorrelation generated by the laser shown in Figure 1. Since the resonator dispersion is independent of the alignment of the cavity, stable mode-locking is observed over a day without dropout. Also no fine-tuning of the dispersion is needed. The operation of such a laser is significantly easier than a laser with additional prism pairs for-dispersion compensation. 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 the in 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. Kärtner, 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. Kärtner and U. Keller, "Theory of Double-Chirped Mirrors," IEEE J. of 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. Kärtner, 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. N. Matuschek, F.X. Kärtner and U. Keller, "Analytic design of double-chirped mirrors with custom tailored dispersion characteristics," IEEE J. Quantum Electron. 35(2): 129-37 (1999)

9. .F.X. Kärtner, U. Morgner, T.R. Schibli, E.P. Ippen, J.G. Fujimoto, V. Scheuer, G. Angelow and T. Tschudi, "Ultrabroadband double-chirped mirror pairs covering for single cycle pulses," submitted to J. Opt. Soc. Am. B.

10. A. Baltuska, A. Pugzlys, M. Pshenichnickov, D. Wiersma, B. Hoenders and H. Ferwerda, “How to retrieve amplitude and phase from an autocorrelation and spectrum,“ Proceedings of Ultrafast Optics 1999, 221, Th10, Ascona, Switzerland, (1999).

11. C. Chudoba, J.G. Fujimoto, E.P. Ippen, H.A. Haus, U. Morgner, F.X. Kärtner, V. Scheuer, G. Angelow and T. Tschudi, "All-solid-state Cr:forsterite laser generating 14 fs pulses at 1.3 um," Opt. Lett. 26(5): 292-94 (2001).

12. D.J. Ripin, C. Chudoba, J.T. Gopinath, J.G. Fujimoto, E.P. Ippen, U. Morgner, F.X. Kärtner, V. Scheuer, G. Angelow and T. Tschudi, "Generation of 20 fs pulses by a prismless Cr4+:YAG laser," Opt. Lett. 27(1), 61-3 (2001).


24-9

13. A. Stingl, M. Lenzner, Ch. Spielmann, F. Krausz and R. Szipöcs, "Sub-10-fs mirror-dispersion-controlled Ti:sapphire laser," Opt. Lett. 20(3): 602-4 (1995).

14. U. Morgner, R. Ell, G. Metzler, T.R. Schibli, F.X. Kärtner, 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).

15. W. Drexler, U. Morgner, F.X. Kärtner, C. Pitris, Y. Chen, S.A. Boppart, X.D. Li, E.P. Ippen and J.G. Fujimoto, "In vivo ultrahigh resolution optical coherence tomography," Opt. Lett. 24(17): 1224 (1999).


24-10

Ultra-low-threshold, Low Cost, Femtosecond Laser Technology Sponsors National Science Foundation

Grant ECS-019452 Air Force Office of Scientific Research

Grant F49620-98-01-0084 Air Force Office of Scientific Research (MFEL)

Grant F49620-01-1-0186 Project Staff Andrew M. Kowalevicz, Rohit P. Prasankumar, Dr. Thomas Schibli, Dr. Ingmar Hartl, Dr. Uwe Morgner, Professor Franz X. Kärtner, Professor Erich P. Ippen, and Professor James G. Fujimoto Kerr lens modelocked (KLM) Ti:Al2O3 lasers can generate extremely short pulse durations with broad bandwidths and have widespread applications in ultrafast studies as well as in biomedical imaging [1]. A standard Kerr lens modelocked laser operating with a 5W pump can produce output powers of 500 mW with 5 nJ of pulse energy. 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.

Figure 1. Schematic of the ultra-low-threshold Ti:Al2O3 laser. Arm lengths are 163 cm and 130 cm for prismatic and OC arms respectively. Intracavity dispersion compensation and tuning is provided by double chirped mirrors (DCM) and a pair of fused silica prisms separated by 31 cm. Here we report the development of an ultra-low-threshold modelocked Ti:Al2O3 laser. Since the pulse energy of a laser is given by its average power divided by its repetition rate, increasing the cavity length produces an increase in pulse energy [2, 3]. This increased pulse energy enables high performance KLM operation at low powers while also decreasing the spot size of the laser mode to reduce pump thresholds. . Previous investigations have achieved low threshold operation in Ti:Al2O3 with a KLM threshold of 500 mW pump power and sustained modelocked operation below 400 mW, generating pulses of 18 fs with bandwidths of 66 nm [4]. We report what is, to our knowledge, the lowest threshold achieved to date for a Kerr lens modelocked Ti:Al2O3 laser. Modelocking can be started with <260 mW of incident pump power, and once initiated, maintained with as little as 170 mW pump. Pumping at 260 mW, pulse durations of 11 fs with bandwidths of ~125 nm are generated with 13 mW output power at 50 MHz repetition rate.


24-11

Figure 1 shows a schematic of the ultra-low-threshold KLM Ti:Al2O3 laser. The pump source is a frequency doubled, diode pumped Nd:Vanadate laser. The cavity is a standard folded X configuration with a 2 mm thick Ti:Al2O3 laser crystal having an α = 5.0 cm–1. The focusing mirrors are 7.5 cm radius of curvature and transmit >92% of the pump beam at 532 nm. The output coupler has a transmission of ~2% between 700 nm and 1000 nm and is low dispersion. All of the mirrors except the output coupler are doubled chirped mirrors (DCM). Intracavity dispersion compensation and tuning is provided by a pair of fused silica Brewster prisms separated by 31 cm.

Figure 2. (a) The output spectrum and (b) interferometric autocorrelation of the pulses. In order to achieve high performance KLM operation at low thresholds the laser cavity is extended and has a 50 MHz repetition rate. In addition to increasing the energy per pulse by a factor of 2 over the standard 100 MHz configuration, the increased arm lengths also reduces the laser mode size in the crystal to about 8 µm radius. Increasing the pulse energy by cavity length scaling enables stronger KLM and shorter pulses at low pump powers. The tight focusing of the laser mode also reduces the laser threshold. In order to make more efficient use of all available pump power, we use a double pass pump configuration where the unabsorbed pump transmitted through the laser crystal is collimated and retroreflected. Because our crystal absorbs only ~63% of the light on the first pass and the laser operates close to threshold, this second pass of the pump gives a factor of 2 increase in CW output power.

The use of both DCMs and prisms for intracavity dispersion compensation enables the dispersion operating point of the laser to be tuned while maintaining a low amount of third order dispersion. KLM can be started by inducing high intensity fluctuations in the laser by either translating the end mirror or one of the prisms. The KLM threshold is sensitive to the dispersion operating point. To start modelocking at low pump thresholds, the dispersion must first be set to a more negative value than for optimum pulse durations. After modelocking is initiated, the dispersion operating point can be tuned toward zero to increase the bandwidth and achieve minimum pulse duration. Typically we observe an output spectrum with 85 nm bandwidth at the dispersion operating point for starting KLM and bandwidths >125 nm when dispersion is tuned for optimum pulse duration. At higher pump powers, the lasers can be started at a dispersion operating point closer to the optimum pulse duration.

The pulse duration was measured with a collinear interferometric autocorrelator. The excess dispersion from the output coupler and external optical elements was compensated using two reflections from DCMs prior to the autocorrelator. Second harmonic was generated by focusing with a mirror into a thin KDP crystal. Fitting the interferometric autocorrelation with a sech2 pulse yields a pulse duration of 11 fs. The time-bandwidth product of 0.614 is above the transform limit of 0.315, indicating that there is some residual pulse chirp.


24-12

In conclusion, we have demonstrated ultra-low-threshold operation of a modelocked Ti:Al2O3 laser. Output bandwidths of 80 nm to greater than 125 nm FWHM can be generated with pulse durations as short as 11 fs with KLM thresholds of <260 mW. By reducing the pump power requirements a factor of 10x lower than conventional KLM lasers, ultra-low-threshold lasers should enable a significant reduction in the cost of femtosecond technology.

References 1. D. E. Spence, P. N. Kean, and W. Sibbett, “60-fsec pulse generation from a self-mode-

locked Ti:Sapphire laser,” Optics Lett., 16: 42-44, 1991. 2. S. H. Cho, B. E. Bouma, E. P. Ippen, and J. G. Fujimoto, “Low-repetition-rate high-peak

power Kerr-lens mode-locked Ti:Al2O3 laser using a multiple-pass cavity,” Optics Letters, 24: 417-419, 1999.

3. A. R. Libertun, R. Shelton, H. C. Kapteyn, and M. M. Murnane, “A 36 nJ-15.5 MHz extended-cavity Ti:sapphire oscillator,” presented at Conference on Lasers and Electro-Optics, Baltimore, MD, 1999.

4. K. Read, F. Blonigen, N. Riccelli, M. Murnane, and H. Kapteyn, “Low-threshold operation of an ultrashort-pulse mode-locked Ti:sapphire laser,” Optics Letters, 21: 489-491, 1996.


24-13

Continuous Wave and Q-switched Modelocked Microchip Lasers Sponsors MIT Lincoln Laboratory Grant ACC 333 National Science Foundation MIT Presidential Fellowship Project Staff Felix Grawert, Dr. Thomas R. Schibli and Professor Franz Kärtner University of Karlsruhe: Jochen Hetzler, Dr. Uwe Morgner, Professor Martin Wegener University of Stuttgart: R. Butendeich, J. Schwarz, Dr. H. Schweizer, Dr. Ferdinand Scholz Compact, reliable and cheap femtosecond laser sources with fundamental repetition rates of 10 GHz and more are needed for high-speed optical data transmission in the 1.5 µm wavelength region for optical analog to digital conversion and optical imaging techniques. A promising approach towards these goals are Cr4+:YAG microchip lasers (for microchip lasers, see e.g. [1]). Cr4+:YAG satisfies both the requirements for femtosecond pulse generation as well as for applicability in fiber-optic communication systems. First experimental results towards mode locking of Cr4+:YAG microchip lasers are reported. Continuous wave (cw) operation of the microchip laser with up to 300 mW of output power has been achieved. Kerr-lens mode-locked (KLM) operation of a 8.2 mm long Cr4+:YAG microchip laser using a Q-switched Nd:YVO4 laser for pumping as well as Q-switched mode-locked operation with saturable Bragg-reflectors (SBR) and cw pumping at a fundamental repetition rate of 10 GHz with 200 fs pulses has been demonstrated.

Telescope

Cr :YAG-Crystal4+

Pump beam

SaturableAbsorberMirror (SAM)

Outputcoupler

HeatsinkDichroic mirror

Laser Output

Nd:YVO /Laser diode

4

Fig. 1: Set-up of the modelocked femtosecond microchip laser.

-50

-40

-30

-20

-10

0

Spec

tral e

nerg

y de

nsity

[dB]

15001480146014401420Wavelength [nm]

1.0

0.5

0.0Inte

nsity

[a.u

.]

-0.4 0.0 0.4Time [ps]

Fig. 2: Mode-locked spectra emitted by the system sketched in Fig.1. The inset shows the intensity-autocorrelation trace of the pulses emitted by the micro-chip laser when operated in the Kerr-lens mode-locked regime.


24-14

The principle structure of the SBRs can be found in [2]. The experimental setup of the microchip laser is sketched in Fig. 1. A diode-pumped Nd:YVO4 laser which can be operated either in continuous-wave (cw) or in Q-switched mode is used for pumping the Cr4+:YAG material. The microchip laser consists of the active material sandwiched between two flat mirrors, one of them being used as the output-coupler mirror (OC). The other one is either a high-reflector (HR) for KLM operation or a saturable Bragg-reflector (SBR) as a mode-locker. Figure 2 shows a typical spectrum of the laser when Kerr-lens mode-locked using a Q-switched pump. The inset shows the intensity-autocorrelation of the pulse train of 200 fs pulses. A severe problem in building lasers with a high fundamental repetition-rate is the Q-switching instability, since the achievable intracavity pulse energy is too low to saturate the absorber strongly. In the past, we have investigated several passive schemes to overcome this instability, such as two-photon absorption [3] and soliton formation [4], respectively. To evaluate the effectiveness of these techniques in the micro-chip setup we performed numerical simulations of the temporal dynamics of the system. The results of such a simulation are shown in Fig. 3.

Fig. 3: Critical pulse energy Wcrit plotted as a function of the intracavity net-dispersion per roundtrip for different thicknesses of an intracavity indium-phosphide layer which is used as a two-photon absorber. The minimum pulse energy Wcrit needed for continuos-wave mode-locking of the system, shown in Fig. 1, is plotted as a function of the intracavity net-dispersion per round-trip for different thicknesses of an intracavity indium-phosphide (InP) layer which serves as a two-photon absorber. It can be clearly seen that appropriate dispersion compensation in conjunction with a two-photon absorber can lower Wcrit by one order of magnitude, which should allow cw-mode-locked operation of the microchip laser in the near future. References: 1. J. J. Zayhowski and A. Mooradian, ''Single-frequency microchip Nd lasers,'' Opt. Lett. 14(1):

24-6 (1989) 2. E. R. Thoen, E.M. Knootz, D.J. Jones, D. Barbier, F.X. Kärtner, E.P. Ippen and L.A.

Kolodziejski, ''Erbium-Ytterbium Waveguide Laser Mode-locked with a semiconductor saturable absorber mirror,'' IEEE Photon. Technol. Lett. 12(2): 149-151 (2000).

3. E.R. Thoen, E.M. Knootz, M. Joschko, P. Langlois, T.R. Schibli, F.X. Kärtner, E. P. Ippen and L. A. Kolodziejski, “Two-photon absorption in semiconductor saturable absorber mirrors,” Appl. Phys. Lett. 74(26): 3927-9 (1999)

4. S. Namiki, E.P. Ippen, H. A. Haus, and C. X. Yu, ''Energy rate equations for mode-locked lasers,'' J. Opt. Soc. Am. B. 14(8): 2099 (1997)


24-15

Broadband Oxidized Saturable Bragg Reflector

Sponsors U.S. Air Force – Office of Scientific Research Grant F49620-01-1-0084 MRSEC Program of the National Science Foundation Award DMR 98-08941 Project Staff Dr. Daniel Ripin, Juliet Gopinath, Hanfei Shen, Alexei Erchak, Dr. Gale Petrich, Professor Franz Kärtner, Professor Leslie Kolodziejski, Professor Erich P. Ippen

Kerr lens modelocked Cr4+:YAG lasers are used to generate femtosecond laser pulses in the wavelength range from 1300 nm to 1600 nm. Pulses as short as 20 fs with a spectral bandwidth of 190 nm fwhm have been produced from a laser using double-chirped mirrors (DCMs) [1] for group delay dispersion (GDD) compensation. In general, without precise alignment, Kerr lens modelocking (KLM) will not be self-starting. External perturbations are then used to initiate modelocking by creating transient power spikes. Saturable absorber mirrors based on semiconductor quantum wells have been used in a varietyof solid-state lasers to initiate modelocking. For Cr4+:YAG, saturable absorber mirrors have beendemonstrated consisting of InGaAs quantum wells grown upon a highly reflecting mirrors [2, 3]. In most cases the mirrors were GaAs/AlAs Bragg stacks. These mirrors typically have a bandwidth of ~100 nm and can therefore limit the minimum pulsewidth by spectral filtering. Overcoming this difficulty, Zhang et. al. generated 44 fs pulses from a laser started by a InGaAs/InAlAs quantum well saturable absorber bonded onto a broadband enhanced gold mirror [4]. It is likely that the pulsewidth in this laser was limited by higher-order dispersion. In this work, a novel high-index-contrast mirror-based saturable Bragg reflector (SBR) was used to generate 35 fs pulses with a fwhm bandwidth of 68 nm Kerr lens modelocked Cr4+:YAG laser. The laser DCMs to compensate the laser crystal GDD. The SBR consists of a broadband oxidized 7-period GaAs/AlxOy Bragg mirror substrate supporting an InGaAs/InP quantum well absorber.

The refractive index and square of the electric field standing wave pattern in the high-dielectric contrast SBR are shown as a function of position in the structure in Figure 1. The SBR consists of a 7-period GaAs/AlxOy Bragg stack a 10 nm InGaAs quantum well in a λ/2-thick InP layer. Each layer thickness was chosen for a center wavelength of λ = 1440 nm. GaAs and AlxOy layers have indices of refraction of 3.39 and 1.61 at 1.5 µm respectively, creating a high-index-contrast mirror that has a calculated reflectivity of 99.9% within the wavelength range of 1220 to 1740 nm using a dielectric stack of only 7-periods.

The SBRs are fabricated with III-V semiconductor-based material growth techniques. First, a GaAs/AlAs multilayer stack is grown by gas source molecular beam epitaxy (GSMBE). The AlAs layers are converted to AlxOy through a wet-oxidation process. It is estimated that the resulting AlxOy layers extended as far as 300 µm into the structure. Side-view scanning electron micrograph (SEM) images of an unoxidized and oxidized SBR structure are shown in Figure 2(a) and 2(b). Using pump-probe spectroscopy, the saturation fluence is on the order of ~ 10 µJ/cm2 and the maximum saturable loss is 0.3%.

The broadband SBR was introduced into a Cr4+:YAG laser cavity to start modelocking. The laser cavity was then optimized for Kerr-lens modelocking. A plot of the KLM pulse spectrum is shown with linear and logarithmic scales in Figure 3. The pulse spectrum is centered at 1490 nm, and has a full-width half maximum of 68 nm. Spectral components are detected from 1200 to > 1700 nm. An interferometric autocorrelation trace, used to determine the pulsewidth, is shown in


24-16

Figure 4. The autocorrelation fits well to a 32 fs sech-shaped pulse. A sech-shaped pulse fit may underestimate the true pulsewidth for non-sech-shaped pulses. Using the measured spectrum, a bandwidth limited pulsewidth of 35 fs is calculated.

Fig. 1. Index of refraction and electric field energy amplitude of the designed Bragg reflector (SBR) mirror consisting of a GaAs/AlxOy high-index contrast mirror and an InGaAs/InP quantum well. (a) (b) Fig. 2. Scanning electron micrograph (SEM) images of an (a) unoxidized and (b) oxidized SBR structure. After oxidation, the AlAs layers are converted to polycrystaline AlxOy, which appears to be granular.

0 1 2 3

1

2

3

4

Air

InP

InGaAsAl

xOy

GaA

s

Inde

x of

Ref

ract

ion

Position (µm)

0

2

4

|E(z)| 2


24-17

Fig. 3. Pulse spectrum from a self-started Cr4+:YAG laser plotted on a linear (black) and logarithmic (gray) scale.

Fig. 4. Interferometric autocorrelation of a saturable absorber Cr4+:YAG laser.

1200 1300 1400 1500 1600 1700

0.0

0.2

0.4

0.6

0.8

1.0

Inte

nsity

(Arb

. Uni

ts)

Wavelength (nm)

-70

-65

-60

-55

-50

-45

-40

-35

-30

-25

Inte

nsity

(dB)

-100 -50 0 50 100

0

2

4

6

8

Auto

corr

elat

ion

Time Delay (fs)


24-18

References:

[1] D. J. Ripin, C. Chudoba, J. T. Gopinath, J. G. Fujimoto, E. P. Ippen, U. Morgner, F. X. Kärtner, 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. Spälter, M. Böhm, M. Burk, B. Mikulla, R. Fluck, I. D. Jung, G. Zhang, U. Keller, A. Sizmann, and G. Leuchs, "Self-starting soliton-modelocked femtosecond Cr(4+):YAG laser using an antiresonant Fabry-Pérot saturable absorber," App. Phys. B 65, 335-338 (1997). [4] Zhigang Zhang, Tadashi Nakagawa, Kenji Torizuka, Takeyoshi Sugaya, and Katsuyuki Kobayashi, "Self-starting mode-locked Cr4+:YAG laser with a low-loss broadband semiconductor saturable-absorber mirror," Opt. Lett. 24,1768-1770 (1999).


24-19

Ultrafast Cr4+:YAG Laser

Sponsors U.S. Air Force – Office of Scientific Research Grant F49620-01-1-0084 MRSEC Program of the National Science Foundation Award DMR 98-08941 Project Staff Dr. Daniel Ripin, Juliet Gopinath, Hanfei Shen, Professor Franz Kärtner, Professor Erich P. Ippen Short pulsewidth optical sources are ideal for time-resolved studies of ultrafast phenomena and devices, as optical clocks with precise timing at the cavity repetition rate, and for ultra-high speed optical communications. Their broad coherent bandwidth can be exploited for spectroscopy, to generate synchronized multi-wavelength optical sources, or for metrological optical frequency standards. Specific laser crystals are chosen for applications based on their material properties. Cr4+:YAG is a promising laser crystal with broad emission from 1300 to 1600 nm. This gain spectrum make this laser crystal ideal for optical telecommunications applications. In general, intracavity dispersion due to the laser crystal, prisms, and mirrors limits the bandwidth and pulsewidth of the emitted light pulses. The shortest pulses previously reported in the Cr4+:YAG laser system are 43 fs pulses [1], which were generated in a laser cavity using two intracavity fused silica prisms to compensate the group delay dispersion (GDD) from the Cr4+:YAG crystal. It is thought that third-order dispersion (TOD) limits the pulsewidth in this system [2].

We have demonstrated the generation of 20 fs pulses from a Cr4+:YAG laser using double-chirped mirrors for dispersion compensation [3]. Double-chirped mirrors allow shorter pulse widths to be generated by reducing higher-order dispersion in the laser cavity. A schematic of the laser cavity is shown in Figure 1. A 2 cm Cr4+:YAG crystal was placed inside a Z-fold cavity designed to maximize Kerr-lens modelocking while simultaneously compensating for astigmatism. The laser is optically pumped by a Nd:YVO4 laser emitting cw light at 1064 nm. Six round-trip bounces off double-chirped mirrors (DCMs) are used to simultaneously compensate GDD and higher-order dispersion. A plot of the calculated Cr4+:YAG and net cavity dispersions are shown in Figure 2. By purging the laser cavity with N2 gas to eliminate water absorption, it was possible to generate optical pulses with a center wavelength of 1450 nm, and a bandwidth of 190 nm from 1310 to 1500 nm fwhm. Kerr-lens modelocking was used to generate ultrashort pulses. A minimum pulsewidth of 19.5 fs was determined by fitting an interferometric autocorrelation (shown in Figure 3) using a pulse phase retrieval algorithm. The average power of the 110 MHz pulse train was 200 to 400 mW, for 9 W of absorbed pump. An example of the modelocked pulse spectrum is shown on both linear and log scale in Figure 4. The spectrum has a full width at half maximum of 190 nm, extending from 1310 to 1500 nm.


24-20

Fig. 1. Schematic of the Cr4+:YAG laser Z-fold cavity. A 2 cm Cr4+:YAG crystal was surrounded by two 10 cm radius-of-curvature folding mirrors (M1 and M2). One arm contains an output coupler (OC), while the second arm contains two flat mirrors (M3 and M4). Pulses experience six reflections off double-chirped mirrors (M1-M3) each roundtrip through the cavity for dispersion compensation. The laser is optically pumped by a Nd:YVO4 laser emitting cw light at 1064 nm.

Fig. 2. Plot of Cr4+:YAG and net cavity dispersion. The net cavity dispersion includes the dispersion from Cr4+:YAG and reflections from 6 double-chirped mirrors (DCMs).

Nd:YVO4 Laser - 1064nm

M1

f = 10 cm

2 cm Cr4+:YAG

OCM2M3 M4

1300 1400 1500 1600-500

0

500

1000

1500

Cr4+:YAG and 6 DCM Bounces

Cr4+:YAG

GDD

(fs2 )

Wavelength (nm)


24-21

Fig. 3. Autocorrelation of ultrafast Cr4+:YAG laser pulses. The autocorrelation matches a fit corresponding with a 19.5 fs pulse.

Fig. 4. Spectrum from a Cr4+:YAG laser plotted on a linear and logarithmic scale. The pulse has a spectral bandwidth of 190 nm spanning from 1310 to 1500 nm fwhm. References:

[1] Y. P. Tong, P. M. W. French, J. R. Taylor, J. O. Fujimoto, "All-solid-state femtosecond sources in the near infrared"; Opt. Comm., 136, 235-238 (1997).

-75 -50 -25 0 25 50 75

0

2

4

6

8

IAC 19.5 fs Fit

Auto

corr

elat

ion

Time Delay (fs)

1100 1200 1300 1400 1500 1600 17000.0

0.2

0.4

0.6

0.8

1.0

1.2

Inte

nsity

(Arb

. Uni

ts)

Wavelength (nm)

-70

-60

-50

-40

-30

-20

-10

Inte

nsity

(dB)


24-22

[2] Tatsuya Tomaru and Hrvoje Petek, "Effect of third-order dispersion on the phases of solitonlike Cr4+:YAG-laser pulses characterized by the second-harmonic generation frequency-resolved optical gating method"; J. Opt. Soc. Am. B, 18, 388-393 (2001). [3] D. J. Ripin, C. Chudoba, J. T. Gopinath, J. G. Fujimoto, E. P. Ippen, U. Morgner, F. X. Kärtner, V. Scheuer, G. Angelow, and T. Tschudi, "Generation of 20-fs pulses by a prismless Cr4+:YAG laser," Opt. Lett. 27, 61-63 (2002).


24-23

Non-epitaxially grown semiconductor-doped silica films for laser modelocking 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 Project Staff Rohit P. Prasankumar, Aurea Tucay, Dr. Thomas Schibli, Dr. Christian Chudoba, Dr. Ingmar Hartl, Professor Michael Ruane and Paul Mak (Boston U), Dr. James N. Walpole and Leo J. Misaggia, (MIT Lincoln Laboratory), Professor Franz X. Kärtner, Professor Erich P. Ippen, and Professor James G. Fujimoto The development of more reliable, compact, and inexpensive modelocked lasers is a major thrust of current research. Semiconductor saturable absorbers are a common technology used to generate self-starting, stable femtosecond pulses in solid state lasers. These devices are typically fabricated by molecular beam epitaxy (MBE) and have been used for both saturable absorber modelocking and initiation of Kerr lens modelocking (KLM) in many solid state laser systems [1, 2]. 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. In previous work, we developed non-epitaxially grown saturable absorber devices and applied them to self-starting KLM in a Ti:Al2O3 laser [3]. The devices consist of InAs nanocrystallites doped into SiO2 films in a 10%InAs/90%SiO2 ratio and deposited on sapphire substrates using a non-magnetron radio frequency (RF) sputtering system. 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) from 500-750 °C was an effective method of controlling the absorption saturation dynamics of our saturable absorbers. The structural and optical properties were comprehensively characterized [4] and the devices were used to initiate KLM in a Ti:Al2O3 laser. 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 modelocking without KLM and also limits the minimum achievable pulsewidth.

3x10-3

2

1

0

-∆α/α

20151050time (ps)

Pump probe 10% InAs / 90% SiO2

925 nm

800 nm750 nm

Figure 1. Tunable pump probe measurements between 750 and 900 nm revealing a decrease in saturation fluence with wavelength. The magnitude of the signal is inversely proportional to the saturation fluence.


24-24

Recently, our focus in this work has been to lower the saturation fluence of these devices to obtain shorter pulses and enable saturable absorber modelocking without KLM. We developed a novel pump-probe system using a broadband 5.5 fs Ti:Al2O3 laser [5] to obtain 17 fs time resolution and independent pump and probe wavelength tunability over a range of 700 to 1000 nm. Using this system, we characterized the nonlinear optical properties of our non-epitaxially grown semiconductor-doped silica film saturable absorbers and discovered trends that aid in device optimization [6]. The devices used in this study were fabricated with a magnetron RF sputtering system, while varying growth parameters such as InAs/SiO2 ratio and substrate temperature. Degenerate pump probe experiments at wavelengths between 750 and 925 nm (Figure 1) and linear transmission measurements (Figure 2) indicate that operation closer to the band edge and fabrication of devices with larger quantum dots results in lower saturation fluences.

100

80

60

40

20

0

%T

30002500200015001000500wavelength (nm)

10% InAs / 90% SiO2

40% InAs / 60% SiO2

Figure 2. Linear transmission measurements on semiconductor-doped silica films with different InAs/SiO2 ratios, showing an increase in quantum dot size with increasing InAs/SiO2 ratio. Using these guidelines, InAs-doped silica film saturable absorbers with an InAs/SiO2 ratio of 40% / 60 % were fabricated and their saturation fluence at 1.25 µm was measured from pump probe experiments (Figure 9) to be 3.35 mJ/cm2, which is a significant improvement over the previous work in Ti:sapphire.

3x10-3

2.5

2

1.5

1

.5

0

-∆α/α

2.01.51.00.50.0-0.5time (ps)

40% InAs / 60% SiO2

1260 nm

800 nm

Figure 3. Comparison of pump-probe measurements at 1260 and 800 nm on annealed 40% InAs/60% SiO2 films. We investigated the application of these devices to self-starting modelocking in a Cr:forsterite laser operating at 1.25 µm. This laser produces 14 fs pulses in a standard 4 mirror configuration [7], which are currently the shortest pulses ever generated at this wavelength. The cavity was then modified to include an additional fold for focusing onto the saturable absorber in a transmissive geometry. Using films with a 40%/60% InAs/SiO2 ratio that were designed to have


24-25

1% absorption at 1.25 µm, self-starting KLM was obtained, with a bandwidth of 91 nm and pulsewidth of 30 fs measured by interferometric autocorrelation (Figure 4). 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 were also able to obtain saturable absorber modelocking without KLM, although a long background pulse coexisted with the shorter (~300 fs) modelocked pulse. We expect that optimization of the dispersion and the use of devices with lower saturation fluence will overcome this problem. Figure 4. Self-starting KLM in Cr:forsterite using 40%InAs/60% SiO2 semiconductor-doped silica films. References 1. U. Keller, K. J. Weingarten, F. X. Kaertner, D. Kopf, B. Braun, I. D. Jung, R. Fluck, C.

Honninger, N. Matuschek, and J. A. D. Au, “Semiconductor saturable absorber mirrors (SESAM's) for femtosecond to nanosecond pulse generation in solid-state lasers,” IEEE Journal of Selected Topics in Quantum Electronics, 2: 435-453, 1996.

2. S. Tsuda, W. H. Knox, S. T. Cundiff, W. Y. Jan, and J. E. Cuningham, “Mode-locked ultrafast solid-state lasers with saturable Bragg reflectors,” IEEE Journal of Selected Topics in Quantum Electronics, 2: 454-464, 1996.

3. I. P. Bilinsky, B. E. Bouma, and J. G. Fujimoto, “Self-starting KLM Ti:Al2O3 laser using semiconductor-doped glass structures,” Technical Digest. Summaries of Papers Presented at the Conference on Lasers and Electro Optics. Conference Edition, 6: 333-4, 1998.

4. I. P. Bilinsky, R. P. Prasankumar, and J. G. Fujimoto, “Self-starting mode locking and Kerr-lens mode locking of a Ti:Al2O3 laser by use of semiconductor-doped glass structures,” Journal of the Optical Society of America B Optical Physics, 16: 546-9, 1999.

5. U. Morgner, F. X. Kartner, 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,” Optics Letters, 24: 411-413, 1999.

6. 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,” presented at Quantum Electronics and Laser Science Conference, Baltimore, MD, 2001.

7. C. Chudoba, J. G. Fujimoto, E. P. Ippen, H. A. Haus, U. Morgner, F. X. Kärtner, V. Scheuer, G. Angelow, and T. Tschudi, “All-solid-state Cr:forsterite laser generating 14 fs pulses at 1.3 um,” Optics Letters, 26: 292 - 294, 2001.

1.1 1.2 1.3 1.4 1.5 1.60

1

FWHM 91 nm

Spec

tral I

nten

sity

(nor

m.)

Wavelength (µm)

IAC

Mea

sure

men

t (no

rm.)

0 50 100 0

1

2

3

4

5

6

7

8

Time Delay (fs) -100 -50


24-26

Active Harmonically Modelocked Fiber Lasers Sponsors U.S. Air Force – Office of Scientific Research Grant F49620-01-1-0084 DARPA - Defense Advanced Research Projects Agency Grant F49620-96-01266 Project Staff Matthew E. Grein, Leaf A. Jiang, 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 analog-to-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 signal-to-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 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 active modelocking of fiber lasers is achieved with two types of modelockers: amplitude (AM) and phase (PM) modulation, each yielding similar performance with respect to pulse shaping and stability. We have found, however, the modulation strongly affects the noise characteristics in qualitatively different ways[1-2]. The timing jitter is expressed as excitations (driven by noise) that are damped by characteristic time constants that describe the laser’s dynamic response to noise[3]. The time constants are related to the laser parameters, such as modulation depth, filtering, and group-velocity dispersion (GVD), and are qualitatively different for the cases of AM and PM. For the case of AM, the pulse timing is directly restored by the modulator, leading to an exponential-type of timing recovery. In contrast, for the case of PM, pulse timing is not directly restored. Mistimed pulses first experience a shift in frequency from the phase modulator. These frequency shifts then convert to a timing shift through group-velocity dispersion. Upon successive round trips, the accumulated frequency shifts are damped out by filtering. The characteristic timing recovery resembles that of a damped, harmonic oscillator that can be underdamped, overdamped, or critically damped depending on the relative strengths of the imposed frequency shift by the modulator and GVD, and the filtering strength. 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.


24-27

.

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 µm 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. The detected laser signal (10 GHz) is compared with that of the external microwave frequency reference using


24-28

Figure 3: Timing jitter measurement scheme. C is a 50/50 polarization-independent directional, PD 16 GHz photodiode, F microwave bandpass filter centered at 9.00 GHz, A microwave amplifier, M double-balanced microwave mixer, PS microwave phase shifter, LO microwave frequency reference.

homodyne phase detection, shown in Fig. 3. The laser pulses are first split in a directional coupler. The pulses impinge on a photodetector (16 GHz bandwidth), and the harmonics of the laser repetition rate are filtered from the photogenerated current, leaving only the photogenerated current at the laser repetition rate. This signal is amplified and compared to the microwave frequency reference using a double-balanced mixer. Because the microwave frequency reference and laser repetition rate are locked using a PLL, the output of the mixer is a voltage proportional to the phase difference between the microwave frequency reference and laser repetition rate. This phase-error noise voltage is displayed using an fast-Fourier transform (FFT) analyzer. By using two separate channels and electronically cross-correlating them using the FFT dual signal analyzer, we were able to reduce the measurement noise floor by an additional 15 dB[4]. The spectrum of the phase noise can be related to the spectrum of the timing jitter.

The resulting phase-noise spectrum L(f) is shown in Fig. 4 for the case where the

Figure 4: Timing jitter spectrum for the case of mostly AM. Upper solid curve, data; lower solid curve, measurement noise floor; dotted curve, theory.


24-29

modulation type was set for mostly AM. The phase noise of the LO (Poseidon Shoe-Box Oscillator) was measured in a separate measurement (-112 dBc/Hz @ 100 Hz, -141 dBc/Hz @ 1 kHz, -161 dBc/Hz @ 10 kHz, -172 dBc/Hz @ 100 kHz) and is lower than the laser noise. The many sharp spurs from 10 Hz to 10 kHz and 70 kHz and 90 kHz are due to environmental electrical interference. The peak near 20 kHz is the laser relaxation oscillation and contributes mostly amplitude noise[8]. The knee near 2.5 kHz is predicted from the theory (also shown in Fig. 4), after which the spectrum falls off by approximately 30 dB per decade. The supermodes at harmonics of 483 kHz have the same spectrum as the baseband mode. The noise floor in the data that is above the measurement noise floor near –150 dBc is due to the overlapping supermode tails, as shown in comparison with the theoretical curve that includes the overlapping supermode tails. Fig. 5 shows the case where the laser is modulated with an increased amount of PM. The laser retiming is underdamped, resulting in a characteristic oscillatory peak near 4 kHz and a rolloff of approximately 40 dB per decade for f >4 kHz, as expected from theory. The root-mean-square timing jitter is given by

∫= dffLfm

)(22

1π

σ

where fm is the 9.0 GHz pulse repetition rate frequency. The jitter integrated from 10 Hz to 241.5 kHz is 9.66 fs for Fig. 4 and 80.0 fs for Fig. 5. From the theoretical model, we have determined that for AM, the quantum-limited timing jitter is proportional to the square of GVD and inversely with the modulation slope. The minimum quantum-limited timing jitter is achieved using a vanishing GVD and strong optical filtering to reduce the Gordon-Haus contributions to the timing jitter. For PM, the minimum jitter requires an optimum dispersion because large dispersion increases the timing recovery from noise that comes from timing fluctuations, but dispersion also increases the noise that comes from frequency shifts via the Gordon-Haus effect. It turns out that the quantum-limited timing jitter grows linearly with GVD for large GVD and inversely proportional to GVD for small GVD. In both cases, the timing jitter is proportional to the internal ratio of signal photons to ASE photons.

Figure 5: Timing jitter spectrum for the case of mostly PM. Upper solid curve, data; lower solid curve, measurement noise floor; dotted curve, theory.


24-30

References

1. 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).

2. M. E. Grein, L. A. Jiang, Y. Chen, H A. Haus, and E. P. Ippen, “A study of the of the dynamics governing timing restoration in the actively modelocked soliton laser”, in Conference on Lasers and Electro-Optics, OSA Technical Digest (Optical Society of America, Washington DC, 1999), p. 100.

3. H. A. Haus and A. Mecozzi, “Noise of modelocked lasers”, IEEE J. Quantum Electron. 29(3): 983-996 (1993).

4. W. F. Walls, “Cross-correlation phase noise measurements,” In IEEE Frequency Control Symposium, (Institute of Electrical and Electronics Engineers, New York, 1992), pp. 257-260.

Publications

M. E. Grein, L. A. Jiang, H. A. Haus, E. P. Ippen, C. McNeilage, J. H. Searls, and R. S. Windeler, “Observation of quantum-limited timing jitter in an active, harmonically modelocked fiber laser”, submitted to Opt. Lett.

M. E. Grein, H. A. Haus, Y. Chen, and E. P. Ippen, “Timing jitter in actively modelocked fiber lasers”, to be submitted to IEEE J. Quantum. Electron.

Conference Papers

M. E. Grein, H. A. Haus, E. P. Ippen, and Y. Chen, “The quantum limit of timing jitter in actively mode-locked soliton fiber lasers”, in OSA Trends in Optics and Photonics (TOPS) Vol. 56, Conference on Lasers and Electro-Optics (CLEO 2001), pp. 243-244.

M. E. Grein, L. A. Jiang, H. A. Haus, and E. P. Ippen, “Timing jitter in modelocked lasers,” invited talk presented at the IEEE Lasers and Electro Optics Society Annual Meeting, November 12-14, San Diego CA, USA, paper MWP. J. J. Hargreaves, P. W. Juodawlkis, J. J. Plant, J. P. Donnelly, J. C. Twichell, F. Rana, M. E. Grein, R. J. Ram, E. P. Ippen, “Timing jitter in modelocked lasers,” invited talk presented at the IEEE Lasers and Electro Optics Society Annual Meeting, November 12-14, San Diego CA, USA, paper MWQ.

Reports H. A. Haus and M. E. Grein, “Quantum limit on timing jitter of actively mode-locked lasers,”

internal Optics and Quantum Electronics Memo No. 92 (1999).


24-31

High-Repetition-Rate Fiber Ring Laser Passively Modelocked with a Saturable Absorber Mirror Sponsors U.S. Air Force – Office of Scientific Research Grant F49620-01-1-0084 Japanese Science and Technology Agency Project Staff Dr. K. S. Abedin, J. T. Gopinath, M. E. Grein, Professor L. A. Kolodziejski, and Professor E. P. Ippen Recently, there has been significant interest in increasing the repetition rate of passively mode-locked erbium fiber lasers for applications such as high-speed optical communication and precision optical sampling. Mode-locked pulses at fundamental cavity repetition rates of 300 MHz have been generated from erbium/ytterbium fiber lasers with short cavity lengths [1]. In addition, harmonic operation of such lasers provided pulses with GHz repetition rates [1]. Recently, Er/Yb codoped phosphate glass waveguides, which exhibit higher pump absorption at ~980 nm and higher gain near 1.53 µm, have also been used in high repetition rate lasers [2]. Passively mode-locked fiber laser cavities contain, in addition to the gain media (Er-doped fiber), a section of undoped single mode fiber (SMF), with anomalous dispersion at 1.5 µm. This SMF, which is typically 2~4 times longer than the doped fiber, is intended to compensate the positive dispersion of the doped fiber and to initiate nonlinear pulse shaping. If the undoped fiber can be replaced with a doped anomalous dispersion fiber, it would be possible to increase the total gain and saturation energy of the laser. We have demonstrated a compact ring laser, consisting of erbium-doped fiber sections with both normal and anomalous dispersion, that is mode-locked with the aid of semiconductor saturable absorber mirror (SESAM). The self-starting laser produces a mode-locked pulse train with a fundamental cavity repetition rate of 140 MHz and average output power as high as 5.3 mW.

λ/4

λ/4 λ/4

Pump

SPLICE

λ/4

EDF1

(+0.075 ps2/m) 11 ~ 130 cm

Dichroic Mirror

PBS SESAM

Isolator

Output

EDF2

(-0.011 ps2/m) 0 ~ 96 cm

Spliceλ/2

Figure 1: Schematic diagram of the erbium fiber ring laser and the structure of the SESAM.


24-32

The experimental setup is shown in Fig. 1. The gain section consists of a section of high-gain erbium-doped fiber having normal dispersion (EDF1, β2 : +75± 10 ps2/km), and a section of erbium doped fiber with anomalous dispersion (EDF2, β2 : –10.8 ps2/km). The SESAM (Ref. 3) was incorporated in the ring cavity by using a polarization beam splitter and a quarter-wave plate.

0

0.02

0.04

0.06

0.08

-10

-9

-8

-7

-6

-5

0 0.02 0.04 0.06 0.08 0.1

Saturable Absorber

Gain

Dispersion

Chirp Parameter

Ch

irp P

arameter, C

Averaged Dispersion (ps*ps/km)

|σ |

, σ ,

σg

d

a

98.3 MHz

108.5 MHz

140.6 MHz

0

0.2

0.4

0.6

0.8

1

-10 -5 0 5 10

Inte

nsi

ty (

a.u

.)

Delay (ps)

-90

-60

-30

0 100 200 300

Inte

nsi

ty (

dB

)

RF Frequency (MHz)

2a

2b

Figure 2: (a) Autocorrelation trace showing the compressed output pulse. Inset shows the RF spectrum. (b) Plot showing the contributions of GVD, gain dispersion and saturable absorber action to pulse broadening /shortening at different average dispersion. σ g : shortening due to gain dispersion; σa : shortening due to saturable absorber; broadening due to cavity dispersion:σd .


24-33

The pulse width, spectral width, and time-bandwidth product were 5.9 ps, 4.6 nm, and 3.5, respectively. The chirped high-energy output pulses from the laser could be compressed to 450 fs by using a 400 m long large effective-area fiber (LEAF). The autocorrelation trace of the compressed pulse output is shown in Fig. 2(a), and in the inset RF spectrum is plotted. According to the theory that describes passively mode-locked lasers, the pulse envelope for such systems can be represented as a t( ) = Ao sech t / τ( ) 1+ jC( ) . For such pulses, the balance between the pulse broadening and shortening can be described by [4]

g

Ωg2τ o

2 2 − C2( )−3C ′ ′ β cav

2τ o2 = σ a (1)

Here, g is the saturated gain per pass, Ωg is the gain linewidth, C is the chirp parameter, βcav

" is the total GVD of the cavity, and σa is the modulation index due to saturable absorber. The first and second terms, defined as σg and σd , represent the pulse broadening due to gain dispersion and GVD of fiber, while the term on the right side represents the shortening due to the SESAM. Since it was difficult in our cavity to measure the contribution of the absorber, we estimated it by solving Eq. 1. Figure 2b plots the contribution of GVD, gain dispersion, absorber and the chirp parameter for different values of average cavity dispersion. Since the laser was operated with C >> 2 , the gain dispersion plays the role of pulse shortening rather than the broadening typically observed in a laser with small anomalous cavity dispersion. The combined effect of pulse shortening due to the saturable absorber and gain dispersion was counterbalanced by the broadening due to GVD. References 1 B.C. Collings, K. Bergman, S.T. Cundiff, S. Tsuda, J.N. Kutz, J.E. Cunningham, W.Y. Jan, M.

Koch, and W.H. Knox, “Short cavity erbium/Ytterbium fiber lasers mode-locked with a saturable Bragg reflector,” IEEE J. Sel. Topics Quantum Electron. 3: 1065-1075 (1997).

2 E. R. Thoen, E.M. Koontz, D.J. Jones, D. Barbier, F.X. Kartner, E.P. Ippen, L.A. Kolodziejski, “Erbium-Ytterbium waveguide laser mode-locked with a semiconductor saturable absorber mirror,” Photon. Technol. Lett. 12: 149-151 ( 2000).

3 E.R. Thoen, E.M. Koontz, M. Joschko, P. Langlois, T.R. Schibli, F.X. Kartner, E.P. Ippen and L.A. Kolodziejski, “Two-photon absorption in semiconductor saturable absorber mirrors,” Appl. Phys. Lett. 74, 3927-3929 (1999).

4 J.N. Kutz, B.C. Collings, K. Bergman, S. Tsuda, S.T. Cundiff, W.H. Knox, P. Holmes, and M. Weinstein, “Mode-locking pulse dynamics in a fiber laser with a saturable Bragg reflector,” Opt. Soc. Am. B. 14: 2681-2690 (1997).


24-34

Self-Stabilized Harmonic Passively Modelocked Stretched-Pulse Erbium Fiber Ring Laser Sponsors: U.S. Air Force – Office of Scientific Research Grant F49620-01-1-0084 Japanese Science and Technology Agency Project staff: Dr. K. S. Abedin, J. T. Gopinath, L. A. Jiang, M. E. Grein, Professor H. A. Haus, and Professor E. P. Ippen Multiple pulses generated in a soliton fiber laser cavity due to energy quantization effects are often randomly spaced. Grudinin et al. has reported [1] a means of self-stabilizing a harmonic passively mode-locked soliton fiber laser through the electrostrictive interaction between the successive pulses. In the anomalous dispersion regime, the electrostrictive interaction between the pulses in the cavity can lead to repulsive forces, thus producing a periodic harmonic pulse train. In this work, we demonstrate self-stabilized harmonic operation of a stretched pulse fiber laser [2]. Highly periodic harmonic pulses with a repetition rate as large as 220 MHz and with an average power of 80 mW have been achieved. These harmonic pulses organize themselves by repelling each other in a manner similar to that observed in soliton fiber lasers. The pulses have picosecond jitter, supermode noise suppression of > 75 dB, and could be compressed to ~125 fs by external chirp compensation.

1.8m Erbium-Doped Fiber 4.4 m SMF-28

Fiber

λ/4 λ/4 λ/4 λ/2

PBS

SMF-28 Fiber

MOPA

Isolator

Collimator

WDM Coupler

Birefringent Plate Nonlinear

Crystal

PMT

Chopper

Corner Cube

Corner Cube

Corner Cube

Lock-In Amplifier

Fig. 1. Experimental setup. The schematic diagram of the experimental setup is shown in Fig. 1. Highly doped Er fiber (1.77 m), is used as a gain medium. The large normal dispersion of this gain fiber is partially compensated with 4.35 m of standard single mode fiber, producing a total cavity dispersion of


24-35

0.036 ps2. The total length was 6.55 m and the fundamental cavity repetition rate was 31.52 MHz. By adjusting the waveplates, Gaussian-like pulses with widths varying between 1.7-1.9 ps and average output powers as high as ~96 mW were produced. For particular settings of the waveplates, it was possible to produce multiple uniformly-spaced pulses. As the waveplates were adjusted initially, the high energy pulse splits into a group of two to seven pulses with random spacing. Next, the pulses uniformly distribute themselves over the round-trip period in a time scale of 15 sec to about a minute. Stable operation, with lower order cavity harmonics suppressed by >70 dB (repetition rates of 157.6 MHz, N = 5) and >50 dB (repetition rate of 220.6 MHz, N = 7) can be achieved, as shown in Fig. 2. These states continued for many hours without further adjustment. For a pump power of 550 mW, output powers as much as 82 mW, as shown in Fig. 2(c), were obtained (7th harmonic operation) at a repetition rate of 221 MHz. Chirped pulses, with widths of ~1.7 ps and spectral width of 50 nm, were obtained from the laser. The pulses were compressed with 2.5-3 m of standard single mode fiber. The minimum pulse width obtained was 125 fs, and the time-bandwidth product 0.77.

-120

-100

-80

-60

-40

-20

Inte

nsi

ty (

dB

m)

N=5

-120

-100

-80

-60

-40

-20

0 50 100 150 200 250

Inte

nsi

ty (

dB

m)

RF Frequency (MHz)

N=7

(a)

(b)

0

20

40

60

80

100

0

1

2

3

4

5

6

7

0 100 200 300 400 500 600

Output Power (mW)

Harmonic Number (N)

Ou

tpu

t P

ow

er (

mW

)

Harm

on

ic Nu

mb

er (N)

Pump Power (mW)(c)

Fig. 2 Laser output. (a) RF spectra of for laser operating at the 5th and 7th harmonic of the fundamental cavity repetition rate. (b) output power and harmonic number as a function of launched pump power. To measure the noise of the harmonically mode-locked laser, we performed crosscorrelations between successive pulses in the train, delaying one arm by the pulse period (shown in Fig. 1). The envelope of the crosscorrelation trace was irregular, and had a typical width of about ~2 ps. From the scan speed of ~120 fs/s and an average separation of 50 fs between successive peaks, we conclude that the distance between successive pulses in harmonic operation varies on the order of 0.5 s. This slow change suggests that it takes a relatively "long" time for the laser to settle after perturbation, an indication that the repulsive force leading to self-stabilization is rather weak. This weak force can result either from dynamic saturation of the gain caused by the high energy pulses [3] or the electrostrictive interaction as seen in soliton lasers. The transfer function of electrostrictive response of fiber contains a number of peaks or resonants as a result of reflection from the cladding-coating boundary of the fiber. This transfer function also exhibits anti-resonants in which the real component changes its sign [4]. A periodic pulse train whose repetition rate approaches these anti-resonant points would experience electrostrictive phase modulation, with curvature opposite to what it assumes for resonant frequencies. In a stretched pulse fiber laser


24-36

with net normal dispersion, self-stabilization is expected to occur when the repetition rate becomes close to one of the anti-resonant frequencies. The frequency shift resulting from the electrostrictive effect and the group velocity dispersion will help to stabilize the pulse train. Investigations to further study this mechanism are now in progress. References 5 A.B. Grudinin, D.J. Richardson, D.N. Payne, "Passive harmonic modelocking in soliton fiber

lasers", J. Opt. Soc. Am. B 14: 144-153 (1997). 6 K. Tamura, E.P. Ippen, H.A. Haus and L.E. Nelson, "77-fs pulse generation from a stretched-

pulse mode-locked all-fiber ring laser", Opt. Lett. 18: 1080-1082 (1993). 7 J.N. Kutz, B.C. Collings, K. Bergman and W.H. Knox, "Stabilized pulse spacing in soliton

lasers due to gain depletion and recovery, IEEE J. Quantum Electron. 34: 1749-1757 (1998). 8 A. Fellegara, S. Wabnitz, "Electrostrictive cross-phase modulation of periodic pulse trains in

optical fibers", Opt. Lett. 23: 1357-1359 (1998).


24-37

Noise in Harmonically Modelocked Lasers Sponsor U.S. Air Force – Office of Scientific Research Grant F49620-01-1-0084 DARPA - Defense Advanced Research Projects Agency Grant F49620-96-01266 Project Staff Leaf A. Jiang, Mathew E. Grein, Farhan Rana, John M. Fini, Professor Rajeev J. Ram, Professor Erich P. Ippen, Professor Herman A. Haus The noise of modelocked lasers is often measured by converting the optical signal to a microwave signal and then comparing its phase noise to a quiet local oscillator. The interpretation of the resulting noise spectrum, which is a combination of both amplitude and timing noise, follows the work of von der Linde [1]. This model was developed for fundamentally modelocked lasers and does not directly apply to the complicated correlations that exist in harmonically modelocked lasers. By using von der Linde’s noise model, researchers have underestimated the noise of their harmonically modelocked lasers [2-4] and hence mistakenly report extremely low timing jitters. A model for amplitude and timing noise in harmonically modelocked lasers was developed for the case that the pulses in the cavity are uncorrelated. We show that the power spectral density of the first cavity axial mode is dominated by amplitude noise, which allows us to distinguish amplitude and timing fluctuations with a smaller electronic detection bandwidth. Our model predicts that increasing the cavity length while keeping the same gain and loss per round-trip changes the frequency content of the timing jitter but does not change the integrated value. In addition, we clarify how to interpret the spectrum from residual phase noise measurements and RF analyzer measurements. Models for the noise in gain-switched lasers [5,6] and fundamentally modelocked lasers [1,7] are well known, but there is currently no treatment for patterning effects, which are inherent in harmonically modelocked lasers. These two cases cover two extremes: in the case of gain-switched lasers, each pulse independently builds up from ASE and therefore the pulse-to-pulse timing jitter is mainly independent and uncorrelated. In the case of fundamentally modelocked lasers, the same pulse recirculates through the cavity and hence the pulse-to-pulse jitter is highly correlated. In a harmonically modelocked laser, there is a mix of these two effects. Just like a gain-switched laser, all M pulses in the cavity at any time instant are independent of each other since they all arise separately from spontaneous emission. On the other hand, similar to a fundamentally modelocked laser, the pulses recirculate in the laser cavity and hence the output is correlated, e.g. pulse 1 is correlated with pulse M+1, but pulse 1 and 2 are not strongly correlated. The assumption that all M pulses in the cavity are independent is a good approximation as long as the gain dynamics do not cause pulse-to-pulse correlations, i.e. the gain recovers between pulses (semiconductor laser) or is not significantly affected by a single pulse (Erbium laser). The case in which neighboring pulses are highly correlated has also been investigated [8]. References [1] D. von der Linde, “Characterization of noise in continuously operating mode-locked lasers,” Appl. Phys. B. 39: 201-217 (1986). [2] W. Ng, R. Stephens, D. Persechini and K. V. Reddy, “Ultra-low jitter modelocking of Er-fibre laser at 10 GHz and its application in photonic sampling for analogue-to-digital conversion,” Electron. Lett. 37(2): 113-115 (2001).


24-38

[3] T. R. Clark, T. F. Carruthers, P. J. Matthews, and I. N. Duling III, “Phase noise measurements of ultrastable 10 GHz harmonically modelocked fibre laser, “ Electron. Lett 35(9):720-721 (1999). [4] C. M. DePriest, P. J. Delfyett, J. H. Abeles, and A. Braun, “Low noise external-cavity semiconductor diode ring laser actively modelocked at 10 GHz,” Proceedings of the Conference on Ultrafast Electronics and Optoelectronics, Lake Tahoe, Nevada, Optical Society of America, January, 2001. [5] D. A. Leep and D. A. Holm, “Spectral measurement of timing jitter in gain-switched semiconductor lasers,” Appl. Phys. Lett. 60(20):2451-2453 (1992). [6] M. Jinno, “Correlated and uncorrelated timing jitter in gain-switched laser diodes,” IEEE Photon. Tech. Lett. 5(10):1140-1143 (1993). [7] U. Keller, K. D. Li, M. Rodwell, and D. M. Bloom, “Noise characterization of femtosecond fiber raman soliton lasers,” IEEE. J. Quantum Electron. 25(3): 280-288 (1989). [8] F. Rana, H. L. T. Lee, M. E. Grein, L. A. Jiang, R. J. Ram, E. P. Ippen, and H. A. Haus, “Characterization of the noise and correlations in harmonically mode-locked lasers,” submitted to J. Opt. Soc. Am. B. Publications L. A. Jiang, M. E. Grein, J. M. Fini, E. P. Ippen and H. A. Haus, “Noise of harmonically modelocked lasers,” presented at the Gordon Research Conference on Nonlinear Optics and Lasers, Colby-Sawyer College, New London, New Hampshire, July 29 - August 3, 2001. F. Rana, H. L. T. Lee, M. E. Grein, L. A. Jiang, R. J. Ram, E. P. Ippen, and H. A. Haus, “Characterization of the noise and correlations in harmonically mode-locked lasers,” submitted to J. Opt. Soc. Am. B.


24-39

Timing Jitter Reduction in Modelocked Semiconductor Lasers with Photon Seeding Sponsor DARPA - Defense Advanced Research Projects Agency Grant F49620-96-01266 Project Staff Leaf A. Jiang, Dr. Kazi S. Abedin, Mathew E. Grein, Professor Erich P. Ippen, Professor Herman A. Haus Reflecting a small fraction of the output light of a laser back into the cavity is called photon seeding and has been shown to reduce timing jitter in modelocked semiconductor lasers [1,2] and gain-switched semiconductor lasers [3]. Reduction of timing jitter is crucial for high-speed optical sampling systems [4] in which it is important to have both short pulses and low-timing jitter. These two requirements on optical sampling streams are difficult to achieve simultaneously [5]. In this work, we demonstrate improved noise performance though coherent photon seeding (CPS) without significant pulse broadening. Improved timing jitter performance from CPS has been attributed to the suppression of noise initially located just ahead of the pulse. This noise migrates into the pulse and causes severe perturbations of the pulse profile due to the random phase and amplitude of the spontaneous emission [6]. Numerical simulations have shown that the pulses suffer large scale perturbations originating from the weak spontaneous emission noise background and that the power of the seeding pulse should be large enough to swamp the noise background, but small enough so as to not affect the pulse forming dynamics of the laser [7].

Fig. 1. Experimental setup for photon seeding. The experimental setup for photon seeding with a modelocked semiconductor laser is shown in Fig. 1. The output pulses of the laser are 2.5 ps wide with a carrier wavelength of 1.5 µm. Less than one-thousandth of the photons are reflected back into the laser cavity by tapping 50% of the light with a beam-splitter, attenuating the signal and then reflecting the photons back into the laser cavity with gold mirror mounted on a micrometer stage. The number, delay, and polarization of the returning photons were optimized to obtain the best noise performance. To monitor the performance of the pulses as well as the quality of the laser pulses, the optical


24-40

intensity autocorrelation, optical spectrum, and RF spectrum of the optical pulses were measured simultaneously. Fig. 2 shows the RF spectrum of the detected laser output as a function of delay between the main and seed pulses. The RF noise skirts decrease as the optimal delay is approached. The inset plots show the corresponding intensity autocorrelations and optical spectra. These plots are almost invariant with respect to delay. This indicates that the pulse shape is not significantly changed with photon seeding and that the noise performance of the laser can be improved without degrading the output pulses.

-4 -2 0 2 40

2

4

6

SH

G S

igna

l (a.

u.)

Delay (ps)

-4.0M -2.0M 0.0 2.0M 4.0M-100

-90

-80

-70

-60

-50

-40

-30

RBW: 10 kHz

Pow

er (

dBm

)

Frequency Offset (Hz)

1540

1544

1548

1552

-70

-50

-30

-10

Pow

er (

dBm

)Wavelength (nm)

Noise Floor

Fig.2. The RF spectra, intensity autocorrelations (left inset), and optical spectra (right inset) of the modelocked laser. The RF spectra are centered at the second harmonic (18 GHz). The RF carrier power at the second harmonic was measured separately using a wider resolution bandwidth and was -15.50 to -17.00 dBm depending on the delay. The gray line is the noise floor of the RF measurement. The timing jitter of a hybridly modelocked semiconductor laser was reduced from 0.75 to 0.22 ps using CPS. There was no penalty to the pulse width of the laser indicating that CPS may be useful in obtaining short pulses and low noise simultaneously. References [1] H. Kawaguchi and K. S. Abedin, “Coherent photon seeding of actively mode locked laser diodes,” Appl. Phys. Lett. 62(18): 2164-2166 (1993). [2] P. Langlois, D. Gay, N. McCarthy, and M. Piché, “Noise reduction in a mode-locked semiconductor laser by coherent photon seeding,” Opt. Lett. 23(2): 114-116 (1998). [3] M. Schell, W. Utz, D. Huhse, J. Kässner, and D. Bimberg, “Low jitter single-mode pulse generation by a self-seeded, gain-switched Fabry-Perot semiconductor laser,“ Appl. Phys. Lett. 65(24): 3045-3047 (1994).


24-41

[4] P. W. Juodawlkis, J. C. Twichell, G. E. Betts, J. J. Hargreaves, R. D. Younger, J. L. Wasserman, F. J. O’Donnell, K. G. Ray, and R. C. Williamson, “Optically sampled analog-to-digital converters,” IEEE Trans. Microwave Theory Tech. 49(10): 1840-1853 (2001). [5] L. A. Jiang, M. E. Grein, E. P. Ippen, C. McNeilage, J. Searls, and H. Yokoyama, “Quantum-limited noise performance of a mode-locked laser diode,“ Opt. Lett 27(1):49-51 (2002). [6] P. Beaud, J. Q. Bi, W. Hodel, and H. P. Weber, “Experimental observation of the self-stabilization of a synchronously pumped dye laser,” Opt. Commun. 80(1): 31-36 (1990). [7] G. H. C. New, ”Self-stabilization of synchronously mode-locked lasers,” Opt. Lett. 15(22): 1306-1308 (1990). Publications L. A. Jiang, K. S. Abedin, M. E. Grein, and E. P. Ippen, “Timing jitter reduction in modelocked semiconductor lasers with photon seeding,” Appl. Phys. Lett., forthcoming.


24-42

Quantum-Limited Noise Performance of a Semiconductor Modelocked Laser Sponsor U.S. Air Force – Office of Scientific Research Grant F49620-01-1-0084 DARPA - Defense Advanced Research Projects Agency Grant F49620-96-01266 Project Staff Leaf A. Jiang, Mathew E. Grein, Professor Erich P. Ippen, Professor Herman A. Haus, Cameron McNeilage*, Jesse Searls*, Hiroyuki Yokoyama** *Poseidon Scientific Instruments, Unit 1, 95 Queen Victoria Street, Freemantle Western Australia 6160 **NEC Opto-Electronics Basic Research Lab, NEC Corporation 34 Miyukigaoka, Tsukuba Ibaraki 305-8501 Japan Quantum-limited noise performance of a semiconductor modelocked laser was demonstrated. The integrated timing jitter of the 6.7 ps output pulses was only 47 fs (10 Hz to 10 MHz) and 86 fs (10 Hz to 4.5 GHz), which is the best-reported figure of merit (pulsewidth4 x timing jitter) for semiconductor modelocked lasers. Record performance was achieved by using a narrow intracavity optical filter and an ultra-low noise sapphire crystal microwave oscillator at 9 GHz. High-speed optical sampling systems [1] and optical time-division multiplexed transmission (OTDM) systems [2,3] have stringent timing jitter requirements. Timing jitter less than 500 fs is required for OTDM transmission at 160 Gb/s (timing jitter should be less than 10% of 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 GS/s.

10 100 1k 10k 100k 1M 10M 100M 1G-160

-150

-140

-130

-120

-110

-100

-90

-80

Spectrum Analyzer Data

Vector Signal Analyzer Data

Spectrum Analyzer Noise Floor

Vector SignalAnalyzer

Noise Floor

L(f

) dB

c/H

z

Offset Frequency (Hz) Fig.1. Single-sideband phase noise of the hybridly modelocked laser diode and corresponding noise floor.


24-43

The timing jitter of the modelocked laser was measured with a residual phase noise technique [4] and the resulting single-sided phase noise is shown in Fig. 1. Spurs due to 60 Hz and harmonics thereof are apparent. Vibration spurs at harmonics of 20 Hz and 55 Hz due to AC fans and other ambient vibrations are also visible. At higher offsets, radio stations pickups appear in the 10-100 MHz decade. A spur due to wireless telephones is visible in the 100 MHz to 1 GHz decade. In addition, contributions of noise due to the fast gain dynamics in the semiconductor laser at the relaxation oscillation frequency are also evident in this decade [5]. In the 100 kHz to 1 MHz decade, the spurs in the laser noise are due to the switching power supply in the current source. 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 reduces to that produced by the fundamental amplified spontaneous emission (ASE) quantum noise. Since the noise less than 5 kHz contributes a small fraction of the integrated timing jitter, the predominant source of the total value is ASE. Fig. 4 shows the integrated timing jitter in each decade. 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 the microwave oscillator noise. References [1] P. W. Juodawlkis, J. C. Twichell, G. E. Betts, J. J. Hargreaves, R. D. Younger, J. L. Wasserman, F. J. O’Donnell, K. G. Ray, and R. C. Williamson, “Optically sampled analog-to-digital converters,” IEEE Trans. Microwave Theory Tech. 49(10): 1840-1853 (2001). [2] H. Yokoyama, “Highly stabilized mode-locked semiconductor diode lasers,'' Review of Laser Engineering 27: 750--755 (1999). [3] U. Feiste, R. Ludwig, C. Schubert, J. Berger, C. Schmidt, H.-G. Weber, B. Schmauss, A. Munk, B. Buchold, D. Briggmann, F. Kueppers, and F. Rumpf, “160 Gbit/s transmission over 116 km field-installed fibre using 160 Gbit/s OTDM and 40 Gbit/s ETDM,'' Electron. Lett. 37(7): 443--445 (2001). [4] D. J. Derickson, A. Mar, and J. E. Bowers, “Residual and absolute timing jitter in actively mode-locked semiconductor lasers,'' Electron. Lett. 26(24):2026-2028 (1990). [5] L. A. Jiang, M. E. Grein, H. A. Haus, and E. P. Ippen, “Noise of mode-locked semiconductor lasers," IEEE J. Select. Topics Quantum Electron. 7(2): 159-167 (2001). Publication L. A. Jiang, M. E. Grein, E. P. Ippen, C. McNeilage, J. Searls, and H. Yokoyama, “Quantum-limited noise performance of a mode-locked laser diode,“ Opt. Lett 27(1):49-51 (2002).


24-44

Experimental Demonstration of a Timing Jitter Eater Sponsor DARPA - Defense Advanced Research Projects Agency Grant F49620-96-01266 Project Staff Leaf A. Jiang, Mathew E. Grein, Professor Erich P. Ippen, Professor Herman A. Haus The timing jitter of a pulse train can be reduced with dispersive fiber and a phase modulator. We experimentally demonstrate the reduction of the timing jitter of a 10 GHz modelocked semiconductor laser by 60%. Timing jitter can severely limit system performance in fiber-optic communications [1] and high-speed optical sampling [2]. Recently, we proposed that the timing jitter could be reduced at the expense of frequency noise using group velocity dispersion and phase modulation [3]. Here, we report on experimental results.

Fig. 1. Experimental setup of the timing jitter eater. The setup used to reduce the timing jitter of a modelocked laser is shown in Fig. 1. A train of pulses with a given amount timing jitter <∆t2> and frequency noise <∆f2> is shown at the top of the figure. The pulses are then sent through pre-chirp fiber and then through a phase modulator. The offset phase of the phase modulator is chosen so that correctly timed pulses do not get


24-45

chirped. Pulses that are mis-timed so that they arrive too early in the modulator are positively (or negatively; depends on the cycle of the phase modulator). Pulses that are too late are negatively (or positively) chirped. If the sign of the dispersion is chosen so that the positively chirped pulses are delayed and the negatively chirped pulses are advanced, then at the end of the dispersive fiber, the timing jitter is reduced. The frequency noise, on the other hand, increases since each mistimed pulse has a slightly different carrier frequency due to the modulator.

10.0030G 10.0035G 10.0040G 10.0045G 10.0050G-90

-80

-70

-60

-50

-40

-30

-20

-10

Phase Mod (135 deg)

Noise Floor

Phase Mod (0 deg)

No Mod

39.50 dB

43.50 dB

RBW: 30 kHz

Pow

er (

dBm

)

Frequency Offset (Hz) Fig.2. RF spectra of the output pulses show 4 dB reduction. In our particular experiment, we started with 6.7 ps pulses from a semiconductor modelocked laser at a repetition rate of 10 GHz [4]. The gain current was 65 mA, the saturable absorber bias was -1.5 V, and the RF drive to the saturable absorber was 11 dBm at 10 GHz. To simulate the case of a laser with quantum-limited timing jitter with a relatively noisy RF oscillator (our HP83732B synthesizer has an integrated jitter of 300 fs from 10 Hz to 10 MHz), it was necessary to decrease the rf power to the saturable absorber to a low value of 11 dBm (usually operated at 25 dBm). When the modelocker is weak, spontaneous emission is the dominant contributor to timing jitter. In a hero experiment, one would use a low noise sapphire oscillator and one would not need to decrease the performance of the laser. Our experimental setup had no pre-chirp fiber, a phase modulator capable of phase shifting by almost one optical cycle at maximum applied peak-to-peak voltage, and 21 km Corning SMF-28 post-chirp fiber. The linewidth of the actively modelocked laser was 4 MHz. The linewidth is narrow enough in this case that frequency noise to timing jitter conversion is relatively small. An EDFA was used directly after the semiconductor laser to increase the output power from 0.5 mW to 10 mW. The RF spectra of the output pulses are shown in Fig. 2. The output noise is shown to decrease by 4 dB or 60% when the phase modulator has the correct phase with respect to the pulses. When the phase of the phase modulator is detuned by 135 degrees, there is no improvement in the rf noise spectrum as expected. In summary, we have demonstrated 4 dB of timing jitter reduction of a semiconductor modelocked laser. We are currently looking into optimal designs for telecommunication applications and optical sampling applications as well as performance improvements with solitons.


24-46

References [1] R. A. Barry, V. W. S. Chan, K. L. Hall, E. S. Kintzer, J. D. Moores, K. A. Rauschenbach, E. A. Swanson, L. E. Adams, C. R. Doerr, S. G. Finn, H. A. Haus, E. P. Ippen, W. S. Wong, and M. Haner, “All-optical network consortium- ultrafast TDM networks,” IEEE Journal on Selected Areas in Communications 14(5): 999-1013 (1996). [2] P. W. Juodawlkis, J. C. Twichell, G. E. Betts, J. J. Hargreaves, R. D. Younger, J. L. Wasserman, F. J. O’Donnell, K. G. Ray, and R. C. Williamson, “Optically sampled analog-to-digital converters,” IEEE Trans. Microwave Theory Tech. 49(10): 1840-1853 (2001). [3] M. E. Grein, H. A. Haus, L. A. Jiang, and E. P. Ippen, “Action on pulse position and momentum using dispersion and phase modulation,” Opt. Express 8(12): 664-669 (2001). [4] L. A. Jiang, M. E. Grein, E. P. Ippen, C. McNeilage, J. Searls, and H. Yokoyama, “Quantum-limited noise performance of a mode-locked laser diode,“ Opt. Lett 27(1):49-51 (2002). Publications M. E. Grein, H. A. Haus, L. A. Jiang, and E. P. Ippen, “Action on pulse position and momentum using dispersion and phase modulation,” Opt. Express 8(12): 664-669 (2001). L. A. Jiang, M. E. Grein, B. S. Robinson, E. P. Ippen, and H. A. Haus, “Experimental demonstration of a timing jitter eater,” submitted to the Conference on Lasers and Electro-Optics 2002.


24-47

Novel Low-Coherence Light Sources for Optical Imaging Applications 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 Paul Herz, Pei-Lin Hsiung, Tony Ko, Andrew M. Kowalevicz, Rohit P. Prasankumar, Aurea Tucay, Dr. Thomas Schibli, Dr. Christian Chudoba, Dr. Wolfgang Drexler, Dr. Ingmar Hartl, Dr. Xingde Li, Dr. Uwe Morgner, Dr. Ping Xue, Dr. Timothy Birks and Dr. William Wadsworth (U. Bath), Dr. Daniel Kopf and Dr. Udo Bunting (High Q Laser Productions), Prof. Rene Salathe and Dr. Markus Pollnau (ETH Lausane), Dr. Robert Windeler (Lucent Technologies), Professor Franz X. Kartner, and Professor James G. Fujimoto 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. These sources are relatively inexpensive 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 developed several novel low-coherence light sources. Spectral broadening in tapered fiber using a femtosecond Nd:Glass Laser High nonlinearity, air-silica microstructure fibers [6] or tapered fibers [7] 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 130 mW average power at 75 MHz repetition rate and 1.06 µm wavelength. The Nd:Glass is pumped by two 1 W diode laser diodes. The Nd:Glass laser is soliton modelocked [8] using a SESAM [9, 10] 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 [7]. 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.


24-48

Frequency Offset to First Harmonic [Hz]

Spe

ctra

l Pow

er [d

Bc

/ Hz]

Detection System Noise Level

Nd:Glass LaserContinuum (Tapered Fiber)

102

103

104

105

106

(b)

-150

-100

-50

Inte

nsity

[a.u

.]

132 nm

1.00 1.20 1.40

(a)

Wavelength [µm]

0.0

0.2

0.4

0.6

0.8

1.0

8 nm

Figure 1. (a) The optical spectrum of the femtosecond Nd:Glass laser and the optical spectrum of the continuum generated in a tapered fiber, and (b) noise spectrum of the continuum generated in a tapered fiber compared to the pump laser. Figure 1 (a) shows a typical laser output spectrum and the continuum generated by the tapered fiber with an average power of 60 mW. 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. In order to measure the noise in the continuum, a 110 nm wide portion of the spectrum centered at 1.3 µm was selected and monitored with a fast InGaAs photodiode. Figure 1 (b) shows the RF spectrum of the photodiode signal compared to the laser. These measurements show that the continuum generation produces a negligible increase in noise above the laser noise itself. Continuum generation in the visible wavelength region using high nonlinearity air-silica microstructure optical fibers High nonlinearity, photonic crystal fibers or air-silica microstructure fibers and tapered fibers can generate an extremely broadband continuum extending from 390 nm to 1600 nm using low energy femtosecond pulses [7, 11]. This large bandwidth enables ultrahigh resolution OCT because of the λ2/∆λ dependence of the axial OCT resolution. We demonstrate that extremely broad bandwidth of the continuum centered at 525 nm in the spectral region 450 nm to 700 nm can be generated by femtosecond pulses from a commercial Ti:Sapphire laser coupled into an air-silica microstructure fiber, enabling OCT resolutions below 1 um. The spectrum is limited at shorter wavelengths by the transmission properties of the microscope objective. Longitudinal OCT resolutions below 0.9 µm are feasible using the visible part of the continuum. Because of the high dispersion of typical optical materials in this wavelength range and the lack of broadband fiberoptic couplers, a free space optical setup is used to demonstrate ultrahigh resolution OCT in the visible spectral range. Figure 2 shows a schematic of the experimental setup.


24-49

Figure 2. Ultrahigh resolution OCT system using continuum generation in an air-silica microstructure fiber as the light source. MO: Microscopic Objective Lens, BS:Beam Splitter, D1,D2: Detector, λ/2:Half-wave Plate, A: Aperture. An isolated reflection from a silver mirror demonstrates that transverse resolutions below 0.9 µm in air can be achieved using the supercontinuum light source. The spectrum of the collimated continuum (Figure 5 (a)) was measured using a calibrated optical spectrum analyzer to have a FWHM bandwidth of 100 nm centered at 525 nm. The interferometric signal is shown together with its envelope in Figure 5 (b). A free-space axial OCT resolution of 0.86 µm was determined by measuring the full width at half maximum of the envelope of the interferometric signal. This is to our knowledge the highest longitudinal OCT resolution demonstrated to date in this wavelength range.

Z-scan

Reference

MO Filter

Amplifier Demod

Computer

D1

D2

Photonic Crystal Fiber

Diode Pumped SHG Nd:YVO4

KLM Ti:Sappire

Laser Faraday Isolator

Sample X-scan

BS

BS

MO

MO

MO

MO

Spectral Filtering λ/2


24-50

0.40 0.60 0.80Wavelength [µm]

0.0

0.2

0.4

0.6

0.8

1.0In

tens

ity [a

.u.]

100 nm


0.0

0.2

0.4

0.6

0.8

Inte

nsity

[a.u

.]

190 nm

-10 -5 0 5 10

Delay [µm]

0.0

0.5

1.0

Inte

nsity

[a.u

.] 0.86 µm


-400

-200

0

200

400

Gro

up D

elay

Dis

pers

ion

[fs2 ]

(a) (b)

(d)(c)

Figure 3. (a) Typical output spectrum of microstructure fiber before spectral filtering. (b) Interference fringes recorded by using an isolated reflection. The full width at half maximum was measured to be 0.86µm. Detected optical spectrum (c) and group delay dispersion mismatch of the interferometer arms (d) obtained from the Fourier transform of the interferometric signal. The vertical dashed lines mark pronounced spectral features.

The detected optical spectrum was calculated by Fourier transforming the interferometric signal. The detected spectral bandwidth is 190 nm. This broadening of the bandwidth occurs due to an attenuation of the short wavelength part of the light source and also the spectral responsibility of the detector. Pronounced spectral features of the light source are still visible after spectral shaping and are marked with dashed vertical lines between Figure 3 (a) and Figure 3 (c). The group delay dispersion mismatch between the two interferometer arms is obtained from the phase of the Fourier transformed interferometric signal and is shown in Figure 3 (d) and is relatively low due to the highly symmetrical setup. Pinholes of 15 microns in diameter are placed before the two detectors respectively to eliminate the background due to multiscattering of tissue samples. To reduce the noise level, dual balance detection is further utilized. The light intensity on the photodiodes is balanced by adjustment the diameter of aperture A before the detector D2 and MO. The measured sensitivity is 97db with incident light power 0.5mw at sample. Imaging studies using this system and further study of fiber system are in progress. Broadband fluorescence sources using Ti:Al2O3 crystals Fluorescence from Ti:Al2O3 and laser dyes have been demonstrated as sources for ultrahigh resolution low coherence reflectometry [12, 13]. Low coherence reflectometry with <2 µm axial resolution was first demonstrated in Ti:Al2O3 using 4.8 µW fluorescence produced from a Ti:Al2O3 crystal pumped by a 20 W Argon laser. While this result demonstrated the highest resolution low coherence reflectometry achieved at that time, the output powers were not sufficient to permit imaging.


24-51

Output powers can be improved significantly using a high doping density crystal and more efficient fluorescent coupling. By matching the numerical aperture of the single-mode fiber as well as retroreflecting the transmitted pump and fluorescent light, we demonstrate a more than one order of magnitude improvement in light source efficiency. Using 5 W cw laser pumping of a Ti:Al2O3 crystal, 40.3 µW of fluorescence with bandwidths of 138 nm can be coupled into a single-mode optical fiber.

Figure 4. (a) Coupled power vs. pump power for different temperatures and comparison between single pass power and retroreflecting. (b) Coupled spectrum with 138 nm FWHM. Figure 4 (a) shows the fluorescent power coupled into the single mode fiber versus incident pump power. Figure 6 (b) shows the spectrum of the fiber-coupled fluorescence. The FWHM bandwidth was 138 nm and follows the expected fluorescence spectrum of Ti:Al2O3. Both the power and spectrum were stable for periods of several hours without realignment. A maximum fluorescence power of 40.3 µW was achieved at 5 W of incident pump, cooling the crystal to –15 °C. Operating at room temperature without cooling, 34.1 µW could be generated at 5 W pump. A family of curves was obtained with the crystal cooled to different temperatures. The output power increased sublinearly at higher pump powers and this reduced efficiency was probably the result of thermal effects. These results suggest that higher power should be possible with better thermal control. Using this source enables ultrahigh resolution OCT imaging with 2.2 µm axial resolution in air (1.7 µm in tissue) and >86 dB sensitivity. This demonstrates a simple, robust light source for spectroscopy and ultrahigh resolution OCT imaging without the need for complex femtosecond Ti:Al2O3 lasers. References 1. D. Huang, E. A. Swanson, C. P. Lin, J. S. Schuman, W. G. Stinson, W. Chang, M. R.

Hee, T. Flotte, K. Gregory, C. A. Puliafito, and J. G. Fujimoto, “Optical coherence tomography,” Science, 254: 1178-1181, 1991.

2. G. J. Tearney, M. E. Brezinski, B. E. Bouma, S. A. Boppart, C. Pitris, J. F. Southern, and J. G. Fujimoto, “In vivo endoscopic optical biopsy with optical coherence tomography,” Science, 276: 2037-9, 1997.

3. S. A. Boppart, B. E. Bouma, C. Pitris, J. F. Southern, M. E. Brezinski, and J. G. Fujimoto, “In vivo cellular optical coherence tomography imaging,” Nature Medicine, 4: 861-5, 1998.


24-52

4. W. Drexler, U. Morgner, F. X. Kärtner, C. Pitris, S. A. Boppart, X. D. Li, E. P. Ippen, and J. G. Fujimoto, “In vivo ultrahigh resolution optical coherence tomography,” Optics Letters, 24: 1221-1223, 1999.

5. W. Drexler, U. Morgner, R. K. Ghanta, F. X. Kaertner, J. S. Schuman, and J. G. Fujimoto, “Ultrahigh resolution ophthalmic optical coherence tomography,” Nature Medicine, in press, 2000.

6. J. K. Ranka, R. S. Windeler, and A. J. Stentz, “Visible continuum generation in air-silica microstructure optical fibers with anomalous dispersion at 800 nm,” Optics Letters, 25: 25-7, 2000.

7. T. A. Birks, W. J. Wadsworth, and P. S. J. Russell, “Supercontinuum generation in tapered fibers,” Optics Letters, 25: 1415-1417, 2000.

8. 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), 2: 540-556, 1996.

9. D. Kopf, F. X. Kärtner, K. J. Weingarten, and U. Keller, “Diode-pumped modelocked Nd:glass lasers using an A-FPSA,” Optics Lett., 20: 1169-1171, 1995.

10. U. Keller, K. J. Weingarten, F. X. Kaertner, D. Kopf, B. Braun, I. D. Jung, R. Fluck, C. Honninger, N. Matuschek, and J. A. D. Au, “Semiconductor saturable absorber mirrors (SESAM's) for femtosecond to nanosecond pulse generation in solid-state lasers,” IEEE Journal of Selected Topics in Quantum Electronics, 2: 435-453, 1996.

11. J. K. Ranka, R. S. Windeler, and A. J. Stentz, “Visible continuum generation in air-silica microstructure optical fibers with anomalous dispersion at 800nm,” Optics Letters, 25: 25-27, 2000.

12. X. Clivaz, F. Marquis-Weible, and R. P. Salathe, “Optical low coherence reflectometry with 1.9 mm spatial resolution,” Electronics Letters, 28: 1553-1554, 1992.

13. H. H. Liu, P. H. Cheng, and J. Wang, “Spatially coherent white light interferometer based on a point flourescent source,” Optics Letters, 18: 678-680, 1993.


24-53

Optical Phase Control and Stabilization Techniques Control and Stabilization of the Absolute Optical Phase Evolution Sponsors U.S. Air Force – Office of Scientific Research Grant F49620-01-1-0084 MIT Lincoln Laboratory Grant ACC 334 National Science Foundation Project Staff Onur Kuzucu, Dr. Thomas Schibli, Richard Ell, G. Metzler, and Dr. Uwe Morgner Professor James G. Fujimoto, Prof. Hermann A. Haus, Professor Erich P. Ippen and Professor Franz X. Kärtner. When pulses become only a few cycles in length the precise shape of the electric field underneath its intensity envelope becomes significant. Then the shape of the field sensitively depends on the phase-difference between pulse-envelope and optical carrier wave [1,2]. The 5 fs pulses with octave-spanning spectra from a Ti:sapphire laser oscillator, demonstrated above, have been directly used to observe this phase evolution in a pulse train emitted from a mode-locked laser. Due to the difference between phase and group velocity in a laser and additional nonlinear effects [3], the carrier-envelope phase φ slips by a given amount per roundtrip [4], leading to an offset of the frequency comb of the modelocked laser from the origin by fφ=φ/(2πTR) [5]. This offset is of great importance if the comb shall be used as a standard for frequency metrology. Phase controlled lasers are considered to be the clock work of improved optical clocks for frequency standards. Mode-locked lasers have already greatly advanced frequency metrology over the last two years [6], [7]. The frequency shift of the comb due to the carrier-envelope phase-slip can be detected by nonlinear optical techniques. When the spectrum of the pulse is broad enough, i.e. one octave, one can beat the second harmonic of the long wavelength part of the pulse spectrum with the short wavelength part of the spectrum, see Fig. 1.

Fundamental

χ(2)

ω00

e2iφ

χ(3)eiφ e3iφ

eiφ

2ω0 3ω0 Figure 1: Schematic spectra of a pulse covering one octave on an optical frequency scale together with the spectra generated by instantaneous χ(2)- and χ(3)-processes, given by successive convolution of the fundamental spectrum. Figure 2 shows our experiment. 5-fs pulses from our Ti:sapphire laser oscillator are focused into a BBO crystal for second-harmonic generation. At the output the interference signal between second-harmonic and fundamental is enhanced before detection by spectral filtering of the overlap regime near 600 nm. As Figure 1 indicates, greater, phase-dependent, spectral overlap can be obtained between second and third harmonics as has been also demonstrated [4]. Using


24-54

this phase-dependent signal we could stabilize the carrier-envelope phase evolution to a microwave oscillator, as shown in Figure 3.

-50

-40

-30

-20

-10

0

RF s

pect

ral p

ower

den

sity

[dB]

706050403020100RF Frequency [MHz]

Figure 2: Interference between fundamental and second harmonic in a compact detector. Top: Set-up; Second harmonic light generated in a BBO-crystal interferes with the other end of the fundamental spectrum in a polarizer. The beat signal is detected with a photomultiplier tube behind a wavelength filter. Bottom: RF-spectrum of the signal at the photomultiplier tube. The right peak at 65 MHz is from the repetition rate fR. The two inner peaks represent the carrier-envelope phase evolution frequency fφ and the mixing product fR -fφ . The background is due to shot noise.

0 1 2 3 4010203040506070

PLL onPLL off

fR

fφ

fR - fΦ

RF

frequ

ency

[MH

z]

time [minutes]

Figure 3: Stability of the carrier-envelope phase evolution frequency fφ with and without external stabilization to an RF-oscillator by a phase-locked loop (PLL). So far we have demonstrated how to control the phase evolution of optical pulses and we could stabilize it. However, we still do not know what the phase of the optical pulse is. Therefore, there is an ongoing quest for nonlinear optical effects that reveal the value of the absolute optical phase of a pulse. References: 1. T. Brabec and F. Krausz, Rev. Mod. Phys. 72(2): 545-91 (2000). 2. 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).

3. H.A. Haus and E.P. Ippen, “Group velocity of solitons,“ Opt. Lett. 26(21): 1654-56 (2001).


24-55

4. L. Xu, C. Spielmann, F. Krausz, and R. Szipöcs, “Ultrabroadband ring oscillator for sub-10-fs pulse generation,” Opt. Lett. 21(16): 1259-61 (1996).

5. J. Reichert, R. Holzwarth, T. Udem, and T. Hänsch, “Measuring the frequency of light with mode-locked lasers,” Opt. Comm. 172(1): 59-68 (1999).

6. S.A. Diddams, D.J. Jones, J. Ye, S.T. Cundiff, J.L. Hall, J.K. Ranka, R.S. Windeler, R. Holzwarth, T. Udem and T.W. Hänsch, "A direct link between microwave and otpical frequencies with a 300 THz femtosecond laser comb," Phys. Rev. Lett., 84(22): 5102-5 (2000).

7. S.A. Diddams, L. Hollberg, L.S. Ma and L. Robertsson, "Femtosecond-laser-based optical clockwork with instability ≤ 6.3x10-16 in 1s," Opt. Lett. 27(1): 58-60 (2002).

8. U. Morgner, R. Ell, G. Metzler, T. R. Schibli, F. X. Kärtner, 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).


24-56

Few-Cycle Nonlinear Optics and Absolute Optical Phase Effects in Carrier-Wave Rabi Flopping Sponsors MIT Lincoln Laboratory Grant ACC 334 National Science Foundation Project Staff Professor Franz X. Kärtner University of Karlsruhe: Dr. Uwe Morgner, Oliver D. Mücke, Thorsten Tritschler, and Professor Martin Wegener As already discussed above it is important to have control over the absolute optical phase. In frequency metrology only control over the phase-slip rate is of importance. However, there are other areas of research, e.g. high-harmonic generation, where knowledge and stabilization of the absolute optical phase is absolutely necessary [1] for further progress in this area. In these experiments the outcome depends on the optical phase of the pulses in use. Therefore, there is an ongoing quest to find nonlinear optical effects that depend on the value of the optical phase of a pulse. Such effects have been predicted to occur in extreme nonlinear optics of atoms [2] and have been observed recently [3].

In cooperation with researchers from the University of Karlsruhe, our sub-two-cycle laser pulses have been used to study carrier wave Rabi flopping in GaAs [4], which was predicted to occur by Hughes [5], a few years earlier. In short, carrier wave Rabi-flopping means, that the laser field becomes so intense, that the Rabi frequency is on the order of the carrier frequency of the optical pulse. For example in GaAs, electric fields as large as 1 GV/m, equivalent to intensities of 0.1 TW/cm2, are necessary to achieve such large Rabi frequencies. In earlier work on optical resonance phenomena, it was thought, that such a regime is not accessible [6]. Such high fields are readily available by focusing few-cycle laser pulses directly from an oscillator to a diffraction limited spot. Matter can sustain such strong fields for the short period of time comprising only a few optical cycles of light. In the regime of carrier wave Rabi flopping, the rotating wave approximation in the Bloch-Equations is not any longer valid and new phenomena, such as breakdown of the area-theorem are expected [5]. Beyond the Rotating Wave Approximation the two-level system also emits radiation at all odd harmonics as an inversion symmetric medium does and which has been observed in GaAs [4]. Very recently, we found by computer simulations, that the dipoles undergoing carrier wave Rabi-flopping also show a phase sensitive emission [7]. Figure 1 (a-c) shows the spectral intensity of a sub-two-cycle pulse with a rectangular pulse spectrum at the input, after propagation through a 20 nm thick layer of GaAs transferred to a sapphire substrate and front coated with an antireflection coating. The assumption of a square shaped pulse spectrum is close to the experimental conditions [8]. The driven dipoles radiate at the fundamental and odd higher order harmonics. The oscillation of the inversion with the Rabi frequency modulates the optical polarization, which generates sidebands in the emission. When the Rabi-frequency becomes twice as large as the carrier-frequency, the modulation sidebands of the fundamental and the third harmonic overlap at twice the fundamental frequency. Very similar to the above discussed phase dependent nonlinear effects, where higher order harmonics of different spectral components of the pulse overlap, to give rise to a phase dependent interference of the sidebands. Figure 1 shows the intensity of the output light around twice the carrier frequency for different phase values of the input pulse. As the simulation shows, the signal is 12 orders of magnitude less intense then the incoming laser light. Thus the signal is in the sub-pW range. Surprisingly, we also found, that the minimum and maximum of the phase sensitive emission does not exactly occur at zero or π phase shift and the location of the extremes depend on the Rabi frequency, see Figure 2. It remains to be investigated, whether the


24-57

signal is strong enough to lead to a practical phase detector. The availability of a compact phase detector for few-cycle laser pulses directly from the laser may lead to much more compact and improved frequency standards for frequency metrology and will help to synthesize arbitrary electrical field waveforms in the optical domain [9].

Figure 1: (a) Signal intensity (linear scale, normalized to the maximum intensity, Imax, of the incident laser spectrum) emitted into the forward direction versus spectrometer frequency ω in units of the laser center frequency ω0 for different values of the absolute optical phase φ . The GaAs film with L=20nm thickness on a sapphire substrate has a frontside antireflection coating. E0 is measured in units of GV/m. The inset illustrates the interference of peaks from the different Rabi doublets as the Rabi frequency ΩR= ħdE0. (b) as (a), but signal intensity (normalized) on a logarithmic scale for fixed φ = 0 and for different incident electric field amplitudes as indicated. (c) shows the signal intensity as a function of the electric field strength and frequency [7].


24-58

Figure 2: Grey scale image of the emitted intensity as a function of ω and φ for a thin GaAs film with thickness L on a sapphire substrate. (a) L = 100 nm and E0 = 3.5 GV/m, (b) as in (a) but L = 20 nm, (c) as (b) but with an additional front-side antireflection coating (as in Fig. 9), (d) as (c) but for an electric field amplitude of E0 = 4 GV/m [7]. References: 1. A. Durfee, A. Rundquist, S. Backus, C. Herne, M. Murnane, and H. Kapteyn, “Phase

matching of high-order harmonics in hollow waveguides,” Phys. Rev. Lett. 83(11): 2187-90 (1999).

2. P. Dietrich, F. Krausz and P.B. Corkum, "Determining the absolute carrier phase of a few-cycle laser pulse," Opt. Lett. 25(1): 16-8 (2000).

3. G.G. Paulus, F. Grasbon, H. Walther, P. Villoresi, M. Nisoli, S. Stagira, E. Priori and S. d. Silvestri, "Absolute-phase phenomena in photoionization with few-cycle laser pulses," Nature 414: 182-4 (2001).

4. O.D. Mücke, T. Tritschler, M. Wegener, U. Morgner and F.X. Kärtner, "Signatures of Carrier-Wave Rabi-Flopping in GaAs," Phys. Rev. Lett. 87(057401): 1-4 (2001)

5. S. Hughes, "Breakdown of the Area Theorem: Carrier-Wave Rabi Flopping of Femtosecond Optical Pulses," Phys. Rev. Lett. 81(16): 3363-6 (1998).

6. L. Allen, J. H. Eberly, Optical Resonance and Two-Level Atoms. (J. Wiley & Sons, New York, 1975).

7. O.D. Mücke, T. Tritschler, M. Wegener, U. Morgner and F.X. Kärtner, "Role of the absolute optical phase of few-cycle pulses in non-perturbative resonant nonlinear optics," submitted to Phys. Rev. Lett.

8. U. Morgner, F.X. Kärtner, S.T. 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. 34(6): 411-3 (1999)

9. A. Poppe, R. Holzwarth, A. Apolonski, G. Tempea, C. Spielmann, T. W. Hänsch and F. Krausz, "Few-cycle optical waveform synthesis," Appl. Phys. B 72(3): 373-6 (2001).


24-59

Nonlinear Fabry-Perots for Synchronization of Independent Laser Oscillators Sponsors MIT Lincoln Laboratory Grant ACC 334 Project Staff Dr. Thomas Schibli, Franz X. Kärtner University of Karlsruhe: Wolfgang Seitz and Dr. Uwe Morgner Synchronized ultrashort laser pulses at different independently tunable wavelengths increase the flexibility of time-resolved optical spectroscopy, allowing for sophisticated excitation and probe techniques [1,2]. Self-synchronized two-color single crystal Ti:sapphire lasers have been reported [3]. Other methods to synchronize two independent Ti:sapphire lasers are based on active stabilization using electronic control-loops [4]. Here, we report a novel method of passive synchronization of two independently mode-locked lasers at discretely separate wavelengths. In our case a passively ps-mode-locked Nd:YVO4 laser oscillator at λ1 = 1064 nm is synchronized to a fs-mode-locked Ti:sapphire laser at λ2 = 850 nm. Synchronization is achieved by optical modulation of the intracavity losses of the slave oscillator (Nd:YVO4) by the master oscillator (Ti:sapphire). The loss modulator is realized by a nonlinear semiconductor Fabry-Perot mirror (FPM). Synchronization of the two pulse trains is observed and verified via cross correlation measurements over a cavity length detuning range exceeding 20 µm between the two laser oscillators. The experimental set-up for synchronization is schematically shown in Fig.1. A diode-pumped Nd:YVO4 laser (dashed box) is passively mode-locked using a semiconductor saturable absorber mirror, which is mounted on a translation stage to adjust the cavity length. The free running Nd:YVO4 laser (without FPM) produces optical pulses with a duration of 8 ps at 1064 nm wavelength. For the master laser a commercially available Ti:sapphire laser was used, emitting transform-limited optical pulses of 180 fs duration. For synchronization, one mirror in the Nd:YVO4 laser cavity is replaced by the FPM.

Ti:sapphirelaser

BBO (1mm)

PM

filter(472 nm) 4

OC

DBS

BS

BS

Fig. 1: Experimental setup of the two synchronized laser oscillators. PM, Photomultiplier; SESAM, Semiconductor Saturable Absorber Mirror; FPM, Fabry-Perot-Modulator; OC, Output Coupler; DBS, Dichroic Beam Splitter; BS, Beam Splitter. The nonlinear Fabry-Perot mirror used consists of a 5-µm-thick InxGa1-xAs layer (x = 0.09) on top of a Bragg reflector grown by solid source molecular beam epitaxy at 580 °C. The Bragg reflector consists of 15 pairs GaAs/AlAs quarter-wave layers at 1064 nm grown on a GaAs substrate at 615°C. For external optical modulation, a fraction of the Ti:sapphire laser output power


24-60

(approximately 100 mW) is focused on the FPM in a spot overlapping with the Nd:YVO4 laser spot. By controlling the temperature of the FPM, the resonance frequency of the FPM can be shifted in wavelength, which allows choosing the operating point. For maximum modulation the resonance frequency is tuned just below 1064 nm resulting in a slightly reduced reflectivity of the linear FPM. With incident Ti:sapphire radiation, free carriers are generated inside the InGaAs layer causing a shift of the resonance frequency to shorter wavelengths. The shift results in an increased reflectivity of the FPM and in reduced losses for the Nd:YVO4 in sychronism with the repetition rate of the Ti:sapphire master oscillator. To determine the tolerance in cavity length mismatch, the round-trip frequency of the Nd:YVO4 laser versus cavity length variation is detected. At 150 mW of control power the tolerance in cavity length mismatch exceeds 20 µm (see Fig. 2b).

a) b)1000

900

800

700

600

500

f Nd:

YVO

4 (H

z) -

82.6

MH

z

-15 -10 -5 0 5cavity length variation (µm)

24

20

16

12

8

lock

ing

rang

e (µ

m)

17515012510075control power (mW)

Fig.2: (a) Measured repetition rate of the Nd:YVO4 laser versus cavity length variation. The synchronization regime (locking range) is emphasized by the grey bar. For clarity, the value 82.6 MHz is subtracted from the exact repetition rate. (b) Locking range plotted as a function of the Ti:sapphire laser power driving the FPM. The dashed line is a linear fit to the measured points (dots).

Fig.3: (a) Measured cross correlation between the synchronized Ti:sapphire and Nd:YVO4 laser which directly reveals the pulse shape of the Nd:YVO4 laser pulses. The FWHM is approximately 13 ps. (b) Measured intensity autocorrelation of the synchronized Nd:YVO4 laser (solid curve) and calculated autocorrelation of the measured cross correlation function from (a) (dotted curve). Fig. 3 (a) shows the cross correlation between the two laser pulse trains by sum-frequency generation in a 1-mm-thick BBO crystal. The asymmetry of the cross correlation trace is due to the asymmetric temporal response of the FPM as described above. The simultaneously measured intensity autocorrelation trace of the synchronized picosecond Nd:YVO4 pulses has a FWHM of 22.3 ps and is displayed in Fig. 3 (black line). The computed autocorrelation function of the measured cross correlation signal, which ideally equals the autocorrelation, is also shown in

1.0

0.8

0.6

0.4

0.2

0.0

inte

nsity

(arb

.)

-80 -40 0 40 80time delay (ps)

1.0

0.8

0.6

0.4

0.2

0.0cros

s co

rrela

tion

(arb

.)

1007550250-25

time delay (ps)

b)a)


24-61

Fig. 3 (dotted curve). From the cross correlation signal, a width of about 13 ps is derived for the pulses from the Nd:YVO4 laser. The two traces in Fig. 3 agree within the measurement accuracy. A conservative estimation of the timing jitter between both lasers was performed by analyzing the intensity fluctuations in the rising / falling edge of the cross correlation [4]. A time record in 100-kHz bandwidth over 1 s revealed a timing jitter between the two lasers below 1 ps, which is much smaller than the duration of the Nd:YVO4 pulses. In fact, subsequent synchronization experiments as demonstrated in the following chapter indicate a timing jitter much shorter than 3 fs. The great advantage of the passive scheme demonstrated here, in comparison to an active scheme is, that it removes also the fast timing-jitter components in the slave-oscillator and tightly links the two pulses. References: 1. J. Feldmann, S. T. Cundiff, M. Arzberger, G. Böhm and G. Abstreiter, “Carrier capture into

InAs/GaAs quantum dots via multiple optical phonon emission,” J. Appl. Phys. 89(2): 1180-4 (2001).

2. M.R.X. de Barros and P.C. Becker, “Two-color synchronously mode-locked femtosecond Ti:sapphire laser,” Opt. Lett. 18(8): 631-3 (1993).

3. A. Leitenstorfer, C. Fürst 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).

4. R. K. Shelton, L.-S. Ma, H. C. Kapteyn, M. M. Murnane, J. L. Hall and J. Ye, ”Phase-Coherent Optical Pulse Synthesis from Seperate Femtosecond Lasers,” Science 293(5533): 1286-9 (2001).


24-62

Control of the Absolute Optical Phase in Picosecond Lasers Sponsors MIT Lincoln Laboratory Grant ACC 334 National Science Foundation Project Staff Dr. Thomas R. Schibli and Professor Franz X. Kärtner University of Karlsruhe: Richard W. Ell, Wolfgang Seitz and Dr. Uwe Morgner Recently, coherent superposition of two independently mode-locked, but spectrally overlapping Ti:sapphire laser oscillators has been demonstrated by means of external active control [1]. This concept opens up new perspectives in metrology with femtosecond mode combs as discussed before, and might lead to phase coherent emission over much broader bandwidth as is possible from a single mode-locked laser (e.g. [4]). The corresponding pulses may reach the single-cycle regime.

Here, we demonstrate a possibility to measure the field repetition rate of a picosecond laser with only 0.1 nm bandwidth by passive synchronization of the picosecond laser to an octave spanning Ti:sapphire oscillator. Phase controlled picosecond lasers might become important for particle accelerators and in metrology. The optical spectrum of a mode-locked laser is characterized by two frequencies, the pulse repetition rate fR, which is the repetition rate of the pulse envelope and the field repetition rate fφ that describes the carrier-envelope phase shift rate (see Fig. 1). Recently, the measurement of fφ in pulse trains with octave spanning spectra has been demonstrated [5-7].

Here, we demonstrate a possibility to measure the carrier-envelope phase-shift frequency of a picosecond laser with only 0.1 nm bandwidth by synchronizing the picosecond laser to a broadband Ti:sapphire oscillator using the Fabry-Perot mirror discussed before.

The master laser – a few-cycle-Ti:sapphire ring laser emitting around 800 nm – modulates the losses in a Nd:YVO4 oscillator, emitting 10-ps-pulses at 1064 nm, via the intracavity nonlinear Fabry-Perot semiconductor mirror.

The set-up is shown in Fig.2: A dichroic mirror splits up the output light of the master laser, a Ti:sapphire ring laser emitting octave spanning spectra from 600 to 1200 nm. The short wavelength part

f

fR

fΦ1

f

fR

fΦ2

Ti:Sa

Nd:YVO 4

Fig.1 Mode combs of two lasers with equal pulserepetition rate fR and independent field repetitionrate fφ1 and fφ2.

Ti:sa - ring(600-1200nm)

PD

SMF

Nd :YVO 4(1064nm)

600 - 1000nm

1000-1200nm GDM

1064nm

FPM

Fig.2 Experimental set-up for the measurement ofthe difference field repetition rate |fφ1 - fφ2|. DM:dichroic mirror, FPM: Fabry-Perot-Modulator,SMF: single mode fiber, G: grating, PD:photodiode


24-63

below 1 µm is used to drive the intracavity Fabry-Perot-Modulator (FPM) in the Nd:YVO4-slave-oscillator to provide synchronism. Within a few micron of cavity length detuning the lasers have the same pulse repetition rate fR . An upper boundary for the timing jitter between the two pulse trains has been estimated by cross-correlation measurements to be well below 1 ps. The long wavelength part of the Ti:sapphire spectrum is spatially superimposed with the output of the

Nd:YVO4 -laser in a single-mode fiber and temporally by means of an optical delay line. A PIN photodiode detects the beat signal behind a narrow band spectral filter. Fig.3 shows the measured RF-spectrum which indicates the same pulse repetition rate of both lasers fR and the position of the difference field repetition rate |fφ1 - fφ2| between the two lasers. This peak can be shifted by changing the intracavity dispersion in one of the two lasers. The measurement has been done with a resolution bandwidth of 100 kHz which approximately corresponds to the width of the beat signal. The finite width of this line indicates residual phase fluctuations between the two lasers. On one hand, the observation time interval, 10 µs, is longer than the timing constant for timing restoration between the two lasers due to the passive synchronization. On the other

hand it is shorter than the bandwidth of typical fluctuations in the carrier-envelope phase of both lasers [7]. Therefore, the observation of the beat signal proofs that the timing jitter of both lasers within the observation time is much smaller than one optical cycle equivalent to about 3 fs.

An independent measurement of the carrier-envelope phase shift frequency fφ1 of the Ti:sapphire oscillator with the scheme shown before allows for the control of the carrier-envelope phase of the picosecond laser.

References: 1. R.K. Shelton, L.-S. Ma, H.C. Kapteyn, M.M. Murnane, J.L. Hall, and J.Ye, “Phase-Coherent

Optical Pulse Synthesis from Seperate Femtosecond Lasers,” Science 293(5533): 1286-9 (2001).

2. J. Reichert, M. Niering, R. Holzwarth, M. Weitz, T. Udem, and T.W. Hänsch, “Phase Coherent Vacuum-Ultraviolet to Radio Frequency Comparison with a Mode-Locked Laser,” Phys. Rev. Lett. 84(15): 3232-5 (2000).

3. J. Stenger, C. Tamm, N. Haverkamp, S. Weyers, and H.R. Telle, “Absolute frequency measurement of the 435.5-nm 171Yb+-clock transition with a Kerr-lens mode-locked femtosecond laser,” Opt. Lett. 26(20): 1589-91 (2001).

4. 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, “Generation of 5 fs pulses and octave-spanning spectra directly from a Ti:sapphire laser,” Opt. Lett. 26(6): 373 (2001).

5. 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 Modelocked Lasers and Direct Optical Frequency Synthesis,” Science 288: 635-9 (2000).

-70

-60

-50

-40

-30

-20

-10

0

RF

spec

tral p

ower

den

sity

(dB)

80706050403020100RF frequency (MHz)

| fφ1 − fφ2 |

fR

fR - | fφ1 − fφ2 |

Fig.3 RF spectrum of the beating signal betweenthe two different lasers indicating the differencecarrier-envelope phase-slip rate |fφ1 - fφ2|. Theresolution bandwidth was 100 kHz.


24-64

6. 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).

7. U. Morgner, R. Ell, G. Metzler, T. R. Schibli, F. X. Kärtner, 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).


24-65

Control of Q-Switching Instabilities in Mode-locked Lasers by Active Feedback Sponsors MIT Lincoln Laboratory Grant ACC 333 National Science Foundation Project Staff Keren Robinson, Dr. Thomas R. Schibli and Professor Franz X. Kärtner University of Karlsruhe: Dr. Uwe Morgner Since the early days of passive mode locking additional undesired Q-switching is a major concern [1]. Especially in the case of fiber and solid state lasers with long upper-state lifetimes mode-locked by semiconductor saturable absorbers, the suppression of the Q-switching instability is a challenging task. Over the last year several passive schemes for the suppression of Q-switching instabilities in passively mode-locked lasers using a variety of absorber characteristics have been investigated [2,3,4]. Even if those methods are used, there still exists a rather limited parameter range (i.e. pump power, absorber saturation, modulation depth of the absorber, pump volume, and gain cross section, etc.) over which the laser is continuous wave (cw) mode-locked. In the last year we demonstrated, that the Q-switching instability in mode-locked lasers can be controlled by intracavity loss- or gain-modulation driven by the average output power of the laser via a feedback loop [5]. Stabilization of an unstable system by use of a feedback loop is a well known concept [6] but has not been applied to suppress Q-switching of mode-locked lasers to our knowledge, even though it has been used to reduce undesired output power fluctuations. This technique also allows reducing the energy fluence on the absorber, since the stabilization against Q-switching is done actively and not by strong saturation of the absorber as is needed especially in picosecond laser systems. For the case of directly diode-pumped lasers, gain modulation can be achieved without any additional intracavity elements by direct modulation of the diode current. The Q-switching instability can be enhanced by positive- or suppressed by negative feedback. We have derived stability conditions, that help to design the feedback loop for the case of intracavity gain modulation via pump current modulation in various forms [5]. To demonstrate this concept, we used a commercially available diode-pumped Nd:YVO4-laser (High-Q-Laser: Model-SC-400) mode-locked with a semiconductor saturable absorber mirror. For the experiment the laser is operated at low pump power levels, where mode-locked Q-switching does occur. The feedback loop can be opened or closed by a switch. Fig. 3 shows the output power versus pump power characteristic of the laser system extracted from a simulation and in the experiment with and without the feedback loop closed. The feedback technique is able to stabilize the laser in cw-mode-locked operation over the full range of pump power available. This method of controlling the Q-switching instability can be applied to many other laser systems. It can be used to scale up the output power, repetition rate and the active volume of this and other laser systems, which is of great importance for diode-pumped laser systems because it relaxes the requirement on the brightness of the pump diodes. Furthermore, it enables lower power-densities on the saturable absorber mirror which is beneficial in respect to longterm stability of the laser system. Fig. 2 shows the microwave spectrum of a higher power, high-repetition rate version of the laser studied above. Without a feedback stabilization, the laser only operates in the Q-switch modelocking regime at any pump power level. With the feedback loop on, the Q-switching is highly suppressed, see Fig. 2. The remaining sidebands at -76dBc are due to a parasitic resonance in the described current source and can be further suppressed by appropriate shielding. The noise floor at -82dB is due to the available RF-spectrum analyzer.


24-66

cw cwML

QSML

cw cwML

c) d)

a) b)

Fig. 1: Output power of a diode-pumped, mode-locked Nd:YVO4 laser as a function of pump power. Results of a numerical simulation with (a) and without (b) the feedback loop. The feedback loop stablized the laser system in the cw-modelocked opteration for all pump conditions. (c) and (d) show the corresponding experimental results, which agree well with the simulations.

(a) (b) Fig. 2: Microwave spectra of a feedback stabilized high-power, high-repetition rate Nd:YVO4-laser: a) feedback loop open, strong Q-switched mode locking occurs; b) feedback loop closed, Intensity fluctuations are suppressed by 80 dB.


24-67

Ultrafast Phenomena and Quantum Electronics Resonance Raman Studies on 0.4nm Single Wall Carbon Nanotubes Sponsors U.S. Air Force – Office of Scientific Research Grant F49620-01-1-0084 MRSEC Program of the National Science Foundation Award DMR 98-08941 Project Staff Peter Rakich, Dr. Ado Jorio, Professor Gene Dresselhaus, Professor Mildred Dresselhaus, Professor Erich Ippen, Professor Katrin Kneipp, Professor Z. M. Li, Professor Z. K. Tang Due to their remarkable physical properties, carbon nanotubes have been studied extensively during the past several years [1]. Single wall carbon nanotubes (SWNTs) exhibit a wide range of electronic structures as a result of their reduced dimensionality. Depending on the specific diameter and chirality, a nanotube may be either semiconducting or metallic. As a result, they are promising for many future applications. Recently, fabrication of very small diameter SWNT's has been made possible by growth inside channels of AlPO4-5 (AFI) zeolite single crystals [2]. Through high-resolution transmission electron microscopy studies, fabricated nanotubes have been found to have a diameter of 4.2 ± 0.2 Å, which is consistent with the diameter of the pores inside of the AFI crystal (AFI having an inner diameter of 7.3 Å). Nanotubes of this size may prove to be among the smallest theoretically possible. It is note worthy that the smallest possible fullerene C20 are of the same diameter, and recent reports show that they exhibit superconductivity below 15K. Only three different SWNT structures could be consistent with the measured diameter: metallic (5,0) and (3,3) and semiconducting (4,2) structures. (Here the indices (m,n) specify chiral vector and the unit cell of the SWNT.[1]) To experimentally determine the structure of these nanotubes, both polarized absorption studies [2] and polarized micro-Raman measurements [3] have been performed, however neither have yielded conclusive evidence of the exact nanotube structures. Shown in Figure 1 are the polarized absorption spectra of a zeolite crystal with imbedded nanotubes. The several absorption bands labeled as S, A, B and C in these spectra are consistent with dipole transitions between Van Hove singularities of the nanotube structures. Thus, one can only conclude that all of the structures could be present [2].

Alternatively, resonant Raman spectra can yield detailed information about the structure of nanotubes. A unique chirality can be assigned if the radial-breathing mode is identified through resonance Raman measurement. Resonance Raman measurements yield two pieces of information that are necessary to classify the structure of any given nanotube: the precise measurement of an electronic resonance as well as the frequency of the radial breathing mode. The absorption peak identified as peak A (at 1.37 eV) in Figure 1 would be consistent with an absorption resonance of either (5,0) or the (4,2) structure. In order to determine which of these structures has formed in the zeolite crystal, we have developed a confocal Raman microscope that is capable of tuning from 1.3 eV to 1.77 eV. Through resonant Raman measurements over these energies we will identify the structure of these nanotubes, which will allow further study of these new molecules.


24-68

References [1] R. Saito, G. Dresselhaus, and M. S. Dresselhaus, in Physical Properties of Carbon Nanotubes (Imperial College Press, London, 1998) [2] Z. M. Li, Z. K. Tang, H. J. Lui, N. Wang, C. T. Chan, R. Saito, S. Okada, G. D. Li, and J. S. Chen, "Polarized Absorption Spectra of Single-Walled 4 Å Carbon Nanotubes Aligned in Channels of an ALPO4-5 Single Crystal"; Phys Rev. Lett. 87, 127401(1-4) (2001) [3] A. Jorio, A. G. Souza Filho, G. Dresselhaus, M. S. Dresselhaus, A. Righi, F. M. Matinaga, M. S. S. Dantas, M. A. Pimenta, J. Mendes Filho, Z. M. Li, Z. K. Tang and R. Saito, "Raman studies on 0.4nm diameter single wall carbon nanotubes", Chem. Phys. Lett. 351, 27-34 (2002)

Fig. 1 The polarized optical absorption spectra of the SWNT containing AFI crystal. The Curves Labeled by θ = 0° and θ = 90° are for the electric field 7 AFI channels and the electric field ⊥ AFI channels respectively. The dotted Curve is the spectrum of the AFI crystal [2].


24-69

Enhanced Light Extraction and Lasing in Two Dimensional Photonic Crystals Sponsors U.S. Air Force – Office of Scientific Research Grant F49620-01-1-0084 MRSEC Program of the National Science Foundation Award DMR 98-08941 Project Staff Peter Rakich, Alexei Erchak, Professor Shanhui Fan, Dr. Daniel Ripin, Dr. Gale Petrich, Professor Erich P. Ippen, Professor John D. Joannopoulos, Professor Leslie A. Kolodziejski Semiconductor light-emitting diodes (LEDs) are used in a variety of displays, indicators, lighting applications, and even short distance communication systems. They are popular because of the high brightness, low power consumption, and long lifetime. Depending on the geometry of a particular structure, semiconductor LEDs can be plagued by low efficiencies due to poor light extraction. In a typical structure, only 2.5% of the light generated will be extracted out of the device's top surface, while the rest is trapped by guided modes in the high-index semiconductor material. Photonic crystals (PC), materials with a spatially periodic refractive index, can be designed to efficiently couple light from the dielectric guided modes into free space. [1] In particular, two-dimensional photonic crystals with the proper dimensions can coherently couple light from waveguide modes into highly directional radiation modes through diffraction. To demonstrate this effect we have fabricated a InGaP/InGaAs quantum well LED, with emission centered around 980 nm. A two-dimensional photonic crystal microstructure is then fabricated within the LED by electron beam lithography, and is designed to inhibit waveguide modes at the emission frequency. Surface normal photoluminescence (PL) is observed to increase by as much as a factor of 8 in the photonic crystal region compared to regions on the same LED structure without photonic crystals. The two-dimensional photonic crystal LED under study consists of an InGaP/InGaAs active quantum well region, a low refractive index AlxOy spacer layer, and an AlxOy/GaAs distributed Bragg reflector (DBR) with a calculated stop band from 800nm to 1400nm. The structure is fabricated using gas source molecular beam epitaxy, direct-write electron beam lithography, reactive-ion etching, and oxidation processes [2]. The two-dimensional photonic crystal consists of a triangular lattice of holes etched within the upper InGaP cladding layer, with a nominal hole-to-hole spacing of 382nm and a nominal hole diameter of 193nm. The holes do not penetrate into the InGaAs quantum well layer, to avoid creating additional non-radiative surface recombination. Each two-dimensional photonic crystal LED is a 30µm by 30µm region within a 50µm by 50µm LED mesa. Structure schematics and scanning electron micrographs of a structure are shown in Figure 1. Room temperature PL measurements have been performed with use of a tunable Ti:Al2O3 laser and a cooled CCD spectrometer. 785nm laser light was focused with a microscope objective onto the photonic crystal and the resulting photoluminescence was coupled through the same objective into the spectrometer. Total PL enhancement factors as high as 100 have been achieved at individual wavelengths, and the total light extraction efficiency has been improved by a factor of 8 [1]. As pump powers were increased a sharp spike in the PL spectrum was observed at 1005nm indicating laser action (see Figure 2a). In order to understand the nature of the lasing, careful band structure calculations and simulations were necessary. In addition, several new measurements were needed to connect theory to experiment.


24-70

Understanding the Band structure of a photonic crystal is crucial to understanding how it interacts with and manipulates light. Several spectroscopic measurements can yield evidence of the photonic crystal band structure. In our study we have performed both photoluminescence measurements and white light reflection measurements. Photoluminescence measurements give a good indication of the integrated density of modes in the vicinity of the Gamma point. One expects that where the density of modes (above the light line) is highest, the PL will be enhanced the most. White-light reflection measurements may also yield similar information through phase matching of incident light into guided modes. Coupling of the radiation into these modes will result in dips in the reflection spectrum, provided that the appropriate phase matching conditions are met. These dips tell us where the photonic crystal density of modes is large. In many ways these two measurements yield the same information, however photoluminescence measurements are limited to the wavelengths of emission whereas white-light is not. In order to understand the origin of the lasing, detailed band structure calculations based solely on the SEM measurements of the device parameters were performed yielding the band diagram shown in Figure 2b. The horizontal dotted line indicates the lasing frequency. As can be seen, this line matches well with bending of the bands at the M-point, which occurs at 1007nm. Flattening of the bands indicates that the group velocity goes to zero, producing a high density of modes, and likewise a lower threshold for lasing. Therefore, this band diagram predicts the lasing frequency with good accuracy. To further test the validity of this band diagram, narrow collection angle (5 degree) PL and white-light reflection measurements were performed, which precisely determine the positions of the bands at the Γ point. As can be seen in Figure 3a, plots of both PL and white-light reflection have sharp spectral features labeled (A) and (B) indicating a large density of modes in the vicinity of the Γ point. Due to the high symmetry of the photonic crystal at the Γ point, phase matching of the guided modes to radiation can only occur where bands are degenerate. Upon inspection of Figure 3b, one finds that there are only two degenerate bands that one would expect to phase match at the Γ point, these include bands 2 and 3 as well as bands 5 and 6. The frequencies of these degenerate points are in close agreement with features (A) and (B) identified in the PL and white-light reflection spectrum seen in Figure 3a. In conclusion, we have observed PL enhancement factors as high as 100 at individual wavelengths. As pump powers are increased, two-dimensional distributed feedback results in laser action at 1005nm. Through band structure calculations based entirely on device parameters, the lasing was found to be consistent with bending of the bands at the M point. To our knowledge this is the first observation of lasing at the M point in a photonic crystal. References

[1] Alexei A. Erchak, Daniel J. Ripin, Shanhui Fan, Peter Rakich, John D. Joannopoulos, Erich P. Ippen, Gale S. Petrich and Leslie A. Kolodziejski, "Enhanced coupling to vertical radiation using a two-dimensional photonic crystal in a semiconductor light-emitting diode"; APL., 78, 563-565 (2001). [2] See RLE Progress Report 2002, L. A. Kolodziejski.


24-71

Fig. 1. a) The two-dimensional PC Structure b) Scanning electron micrograph of fabricated PC structure.

Fig. 2. a) Plot of measured PL at higher pump powers vs. wavelength and frequency. Pronounced peak visible at 1005 nm. b) Calculated band diagram showing frequency vs. wave number. The horizontal dotted line indicates the frequency of observed lasing

a b

a

Freq

uenc

y (2

πc/a

)

b


24-72

Fig. 3 a) Top (bottom) curve is reflectivity (PL enhancement) versus wavelength for 5 degree Collection angle. b) Calculated band diagram in the vicinity of the Γ point showing frequency vs. wave number. The line crossing bands indicates cutoff for a collection angle of 5 degrees.

0

50

100

150

200

250

300

350

400

0

0.2

0.4

0.6

0.8

1

1.2

850 900 950 1000

PL Enhancement

Photonic Crystal Reflectivity

Wavelength

A

B

0.449 0.402 0.424 0.382 Frequency (2πc/a)

(nm) a b


24-73

Highly Nondegenerate Four-Wave Mixing in Tapered Microstructure Fiber Sponsors U.S. Air Force – Office of Scientific Research Grant F49620-01-1-0084 Japanese Science and Technology Agency Project Staff Dr. K. S. Abedin, J. T. Gopinath, J. Chandalia, Professor E. P. Ippen, and Professor H. A. Haus Four-wave mixing (FWM) in fiber and semiconductor amplifiers/lasers is a promising technique for wavelength conversion in telecommunications. Fiber is attractive because it is a passive device with a large bandwidth. Non-degenerate FWM has previously been observed in 50 m of SMF for a frequency difference of 96 THz (wavelength separation of 600 nm) [1], and in 6.1 m of microstructure fiber for a frequency difference of 32 THz (wavelength separation of 20 nm) [2]. We have achieved highly non-degenerate FWM, involving wavelengths of 1540 nm, 800 nm, and 537-575 nm, with conversion efficiencies up to 10%, in an 18 cm tapered air silica microstructure fiber (T-ASMF) [3]. In the tapered region, the core diameter is reduced to ~3 µm resulting in an enhancement of the nonlinear effect by an order of magnitude. The large core-clad index contrast significantly helps the phase matching for FWM.

Figure 1. Experimental setup for highly non-degenerate FWM. The experimental setup is shown in Figure 1. Light from an femtosecond optical parametric oscillator (OPO, 1540 nm) with repetition rate of 80 MHz, and a Ti:Sapphire laser (810 nm), was focused into an 18 cm-long T-ASMF, using an aspheric lens with a focal length of 18 mm. Coupling efficiencies for the OPO and Ti:Sapphire ranged from 30-40 %. Because of the large wavelength separation of the Ti:Sapphire and the OPO, the pulses walk-off as they propagate through the microstructure fiber, due to the difference in the group index. We performed


24-74

numerical simulations to determine the group index and the GVD of the T-ASMF. We obtain a group index difference ng(1540 nm)-ng(810 nm) of 0.0125 for the lowest order LP mode in the T-ASMF. This gives an interaction length of 15 mm in the T-ASMF, the distance over which the pulses overlap. The zero dispersion wavelength was estimated to be ~884 nm. When the Ti:Sapphire pulse is launched alone, self phase modulation caused the 10 nm spectrum to broaden to 55 nm. When the OPO is added, we observed generation of strong green light from the T-ASMF with wavelengths ranging from 537 nm to 575 nm. With a maximum incident Ti:Sapphire power of 180 mW and OPO power of 80 mW, we were able to generate about 4 mW of green light (idler) from the fiber. Assuming coupling efficiencies of 32.5% and 43.5%, respectively, for the pump and signal beams, an efficiency Pidler/Psignal of about 10 % was obtained. The parametric gain of four-wave mixing process can be expressed as:

( ) ( )[ ]22 2/κγ −= pPg where ( )( )effp An //2 2λπγ = is the nonlinearity coefficient of the T-ASMF, λ p is the pump wavelength, n2 is the nonlinear refractive index, Aeff is the effective mode-field area, and Pp is the pump power. The parameter κ is the phase mismatch and is defined as: κ = ∆k + 2γPp Here, the term 2γPp is the induced pump nonlinearity, and ∆k = ks + ki − 2kp is the wave vector mismatch between the signal, idler and the pump beam involved in the FWM process. In an optical fiber, the effective refractive index of the waves (hence, also ∆k ), is determined by the dispersion of the bulk medium (core) as well as the dispersive properties of the waveguide. For the T-ASMF fiber used in the experiment, the large index difference and the small core size can result in a waveguide contribution which is comparable to the material contribution and significantly affects the value of ∆k . In Fig. 2, we plot the wave vector mismatch for T-ASMF with core diameter of 2 µm and 3 µm as a function of signal wavelength (pump wavelength kept constant at 810 nm). The material contribution of ∆km is also shown for comparison. The figure clearly indicates that the T-ASMF with large index difference and small core size can considerably reduce the phase mismatch, and through proper design it may even be possible to make ∆k small enough to be compensated for by the induced pump nonlinearity to ensure κ = 0 . This work is being carried out in collaboration with C. E. Kerbage and Dr. B. J. Eggleton at Lucent Technologies.


24-75

-400

-200

0

200

400

600

0.8 1 1.2 1.4 1.6 1.8

∆ k

m

∆ k

m+w

,

(cm

-1 )

Signal Wavelength (µm)

∆ k m

∆ k m+w

∆ k m+w

φ = 2 µm

φ = 3 µm

Fig. 2. Calculated wave-vector mismatch (waveguide and material contribution combined) plotted a function of signal wavelength (pump wavelength kept constant at 810 nm) for T-ASMF with core diameters of 2 µm and 3 µm. The dotted curve shows the contribution of material dispersion to wave-vector mismatch. References

1. C. Lin, W. A. Reen, A. D. Pearson, and H.-T. Shang, "Phase matching in the minimum chromatic-dispersion region of single-mode fibers for stimulated four-photon mixing," Opt. Lett. 6: 493-495 (1981).

2. J. E. Sharping, M. Fiorentino, A. Coker, P. Kumar, and R. S. Windeler, "Four-wave mixing in microstructure fiber," Opt. Lett. 26: 1048-1051 (2001).

3. J. K. Chandalia, B. J. Eggleton, R. S. Windeler, S. G. Kosinski, X. Liu, C. Xu, "Adiabatic coupling in tapered air-silica microstructured optical fiber," IEEE Phot. Tech. Lett. 13: 52-54 (2001).


24-76

Femtosecond Pump-Probe Spectroscopy Using a Two-Dimensional Smart Pixel Detector Array Sponsors Air Force Office of Scientific Research (MFEL)

Grant F49620-01-1-0186 Air Force Office of Scientific Research

Grant F49620-98-01-0084 Project Staff Dr. Stephane Bourquin, Rohit P. Prasankumar, Dr. Rene Salathe (ETH Lausane) and Professor James G. Fujimoto Femtosecond pump-probe spectroscopy is a well-known method used to investigate ultrafast excited state dynamics in condensed matter, chemical, and biological systems. The standard technique uses a narrowband pump and continuum probe to excite the sample, and employs a spectrometer to acquire the wavelength dependent absorption saturation dynamics. This technique often requires the use of a femtosecond amplifier to achieve sufficient intensities for continuum generation and sufficient pump-probe signal magnitudes. With the development of 5 fs lasers, which can generate spectra spanning one octave, spectrally resolved pump-probe measurements can be performed without the need for amplifiers [1-3]. Although the pulse repetition rate from a laser oscillator is extremely high, pulse energies are low and signal levels are small. Standard CCD detectors cannot detect modulated signals and thus it is not possible to take advantage of high signal to noise measurements that are possible with high repetition rate laser sources. Here, we present a new technique that enables the parallel acquisition of pump-probe measurements for multiple wavelengths. This is made possible using a novel, two-dimensional smart pixel detector array, which was originally developed for high speed optical coherence tomography (OCT) [4, 5]. Each pixel performs amplitude demodulation, and in combination with a diffraction grating, probe transmission signals can be acquired in parallel for multiple wavelengths. The smart pixel array can achieve sensitivities comparable to lock-in amplification but can perform demodulation of probe transmission signals at multiple wavelengths simultaneously, enabling time and wavelength resolved femtosecond pump-probe spectroscopy. In this work, we demonstrate spectrally resolved measurement of carrier dynamics in a thin sample of GaAs using this smart pixel detector array in combination with a broadband Ti:sapphire laser and femtosecond pump-probe scheme. This system uses a 100 MHz, 5 fs, 275 nm bandwidth Ti:sapphire laser, the output of which is separated into a 71 mW pump beam and 0.3 mW probe beam. A schematic of the beam configuration is shown in Figure 1. Reflective optics are used to preserve pulse duration. A 3.8 kHz modulation of the pump beam is performed by use of a chopper CH. A diffraction grating DG spectrally spreads the probe beam after the sample and a lens L2 focuses the beam onto one row of the detector array DA. A row of this detector is read out multiple times for each pump-probe 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.


24-77

Probe

Pump

CHS

DA

DG

L2

L1DL

M

λ1 … λn Figure 1. Schematic diagram of the experiment: M, mirror; DL, delay line; CH, chopper; S, sample; L1,L2, achromat lenses; DG, diffraction grating; DA, detector array. Figure 1 shows a photograph of a section of the detector array and a schematic diagram. 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 filter cut-off frequencies are adjusted by off-chip reference voltages. The generated amplitude modulation signals are selected sequentially by a row and a column address decoder, via the pixel and the column buffers. The analog output signal is read out of the chip, digitized by a 12-bit data acquisition card and transferred to a computer.

Referencevoltages

Photodiode

Electroniccircuitry

Columnaddress

Pixel buffer

Rowaddress

Chipoutput

Columnbuffer

Column address decoder

Row

add

ress

dec

oder

Pixel cell

Figure 2. (left) Photograph of a section of the smart pixel detector array. The address decoders are visible on the left and bottom. (right) Detector array architecture corresponding to the photograph. Each pixel performs amplitude demodulation of the modulated optical signal and its output is addressed sequentially by a row and a column address decoder. Results of initial experiments are shown in Figure 3. It can be seen that the dynamics are a strong function of wavelength. Measurements performed far above the band edge of 875 nm have a rapid relaxation 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.


24-78

-1 0 1 2 3 4

-1 .5

-1

-0 .5

0

0 .5

1

1 .5

2

2 .5

de lay (ps )

735 nm

794 nm

828 nm

∆T/T(x10-2)

-1 0 1 2 3 4

-1 .5

-1

-0 .5

0

0 .5

1

1 .5

2

d e la y (p s )

8 5 4 n m

8 8 0 n m

8 9 7 n m

∆T/T(x10-2)

Figure 3. Femtosecond pump-probe measurements at selected probe wavelengths as a function of time, extracted from the 3-dimensional data acquired from the smart pixel detector array. The traces are separated for clarity. This measurement demonstrates the feasibility of performing parallel spectrally resolved pump-probe measurements over 58 simultaneous wavelengths with high modulation sensitivities when averaging each pixel 1000 times, corresponding to 52 seconds of measurement time for a 5.67 ps scan. Even higher sensitivities are possible with increased averaging. This technology can be applied to a wide range of pump-probe measurements in condensed matter, chemistry, and biology. Using the two dimensional imaging capabilities of the smart pixel array, an additional dimension of detection is also possible, enabling for example, one-dimensional spatial imaging combined with spectrally and time resolved measurement. References 1. 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,” Optics Letters, 24: 411 -- 413, 1999.

2. 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,” Optics Letters, 24: 631-3, 1999.

3. R. Ell, U. Morgner, F. X. Kärtner, 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,” Optics Letters, 26: 373-5, 2001.

4. S. Bourquin, P. Seitz, and R. P. Salathe', “Two-dimensional smart detector array for interferometric applications,” Electronics Letters, 37: 975-976, 2001.

5. S. Bourquin, P. Seitz, and R. P. Salathe', “Optical coherence topography based on a two-dimensional smart detector array,” Optics Letters, 26: 512-514, 2001.


24-79

Photonics and Devices Resonant Channel Add/Drop Filters Sponsor The MIT-Pirelli Lab S.p.A. Research Agreement Project Staff Professor H. A. Haus, Dr. C. Manolatou, M. Popović Channel add/drop filters are basic components in WDM systems. Their function is to add/extract a channel to/from a WDM stream without affecting the other channels. The filters that we are studying are based on resonators evanescently coupled to two waveguides: the bus that carries all the WDM channels and the receiver that carries the selected channel. If the resonant wavelength coincides with the wavelength of the selected channel and with appropriate waveguide-resonator couplings, all the channel power is transferred between the bus and the receiver. To satisfy the WDM system specifications, (e.g. channel bandwidths 10GHz, channel spacings 50-100GHz, crosstalk from adjacent channels <-25dB), the shape of the filter’s frequency response must be almost square. Using a number of identical resonators with appropriate quality factors and couplings the filter response can be engineered to satisfy these constraints in analogy with filter design using lumped circuit elements. The design of these filters is tested by Finite Difference Time Domain (FDTD) simulations. An analytical method that approximates well the filter response around the resonance for given the resonant frequency and quality factors is the Coupled-Mode Theory (CMT). Fitting the numerically obtained response with the curves given by the CMT expressions we can extract basic filter parameters, (i.e. radiation and external quality factors) and understand how we can improve the filter performance. Our design starts from the first order filter. A basic implementation of first order filter consists of a ring symmetrically placed between the bus and the receiver. The bandwidth and the peak of the filter response depend on the ring-waveguide coupling which in turn depends on their mutual separation and the interaction length. Because a circular ring has a very short interaction length, its separation from the waveguide is too small to be successfully fabricated. For this reason we have preferred to use a racetrack instead which, due to its straight waveguide segments, has a longer interaction length and can thus be placed at a practical distance from the waveguide. A potential problem with the racetrack is a mode mismatch at the transition from the straight to the curved section which can result in reflection into the counter-propagating mode as well as radiation in addition to that produced by the bend itself, thus degrading the filter performance. Specifically if the racetrack has circular bends then at the transition point the radius of curvature changes discontinuously from infinite to that of the circular bend. Mode-matching can be improved and the reflection and radiation reduced by designing the bend to be non-circular with a continuous curvature change. Our designs will be implemented with silicon nitride waveguides (refr. index 2.2) buried in SiO2 (refr. index 1.46). The core cross-section is chosen 600nmx600nm to ensure single mode operation and polarization independence. In our FDTD simulations we use a 2D model of this waveguide that has the same effective index and mode profile in the horizontal direction. An example of such a 1st -order filter simulated by FDTD is shown in Fig.1(a) and its numerically obtained response fitted by CMT is shown in Fig.1(b). The racetrack fits 40 optical wavelengths and its separation from the waveguides is 320nm. A 3rd-order filter made up of 3 such racetracks placed between the bus and receiver is shown in Fig. 3(a) with the electric field found by FDTD. The numerical response curves at the drop and through ports are shown in Fig. 3(b) and (c). CMT provides the relation between the resonator-waveguide coupling and the resonator-resonator couplings for maximum received power and Butterworth response; the corresponding theoretical


24-80

curves are also shown in Fig. 3(b) and (c) (dotted line). Comparing the numerical and the theoretical curve it is clear that the coupling between the racetracks is stronger than desired; in fact their separation is about 40-60nm too small. Our FDTD simulations have also revealed certain interesting effects, as we go from 1st to 3rd order filter, that we can quantify by fitting the 3rd-order response obtained by FDTD with the expressions of the CMT. For this particular example:

a) There is a 21GHz downward shift of the spectrum center. b) The radiation loss of the resonant modes of 3 coupled racetracks drops almost to one

third of the single racetrack. c) The spectrum appears skewed which is an indication that each of the 3 modes has a

different radiation loss. Our explanation for a) is that the resonator-waveguide and the resonator-resonator couplings bring about different shifts of the propagation constants in the racetracks and thus shift the resonant frequencies. We attribute b) to radiation cancellations occurring when 3 resonators are coupled together and c) to the fact that these are different for each of the excited modes of the 3 racetrack system. Our goal is to develop a theory that can predict these effects quantitatively.

Figure 1: a) Electric field amplitude in a 1st-order resonant channel dropping filter based on a racetrack with non-circular bends, b) Frequency response obtained by FDTD and fitted by CMT

λo =1550.92nmQ d = 7600

(b)

(a)


24-81

Fiber-Chip Coupling The high index-contrast (HIC) integrated waveguides used in our structures have sub-micron cross-sections while standard optical fibers have diameters of 8-10 microns. The large mode mismatch makes it impossible to couple light efficiently from the fiber directly to the optical chip (more than 80% of the power is lost to radiation).To improve the coupling efficiency a spot-size converter must be placed between the fiber and the integrated waveguide with certain desirable

(a)

(b) (c)

Figure 2: a) Electric field amplitude found in a 3rd-order channel add/drop filter based on 3 racetracks identical to that of Fig.1 and, b), c) drop and throughput spectra, respectively, found by FDTD and compared with the desired 3rd-order response found by CMT.


24-82

characteristics such as: compactness, broadband operation and low polarization dependence. In order to ensure compactness and avoid misalignment problems it is preferable to have the spot-size converter integrated on the chip. In the literature such devices are hundreds of microns long, e.g. adiabatic tapers, and the coupling losses are 1-3dB. We propose an integrated 3D mode-size converter based on two lensing mechanisms, as shown in the schematic of Fig. 3(a): The refractive index in the vertical direction is graded quadratically as n(y) = ni(1-y2/2h2), where ni is the index at the center, by deposition of thin layers of varying index. A material system suitable for this is structure is silicon oxynitride (SiON) which allows an index variation from 1.45 (SiO2 index) up to 2.2 (silicon nitride index). It is well known that a Gaussian beam propagating along the axis of this index distribution contracts and expands periodically, the period and the minimum beam width depending on the parameter h. If the length of this layered structure is half a period long it acts as a lens. In the lateral direction the curved interface also acts a lens with focal length determined by the curvature and the index contrast of the interface. Both lensing structures are preceded by a quarter-wave thick antireflecting layer. These two lensing mechanisms combined as in the schematic of Fig. 3(b) will result in a 3D mode-size conversion. The size of the structure is prohibitive for a 3D simulation so for the analysis we decompose the problem beam optics which gives a very good prediction for the evolution of the beam. The beam-width variation of into two 2D problems linked by effective index. The initial design is carried out analytically using Gaussian a Gaussian beam is associated with an effective index variation along the propagation direction. From he analysis of the vertical mode-conversion (yz-plane) we obtain the index to use in our analysis and design of the lateral mode-conversion (xz-plane). The FDTD

vertical conversion

graded index lateral conversion

HIC lens

fiber mode

waveguide mode

AR coating

y x

z

Figure 3: (a) Schematic of 3D mode-size conversion using a graded index distribution in the vertical direction and high index-contrast (HIC) curved interface in the lateral direction (b) Combination of the two mechanisms to form a single, compact 3D mode-size converter.

(a)

(b)


24-83

simulations for each plane and the 2D coupling efficiencies are shown in Fig. 4(a). At the input we launch a 10micron wide Gaussian beam that represents the fiber mode and at the output we calculate the power transmitted into the fundamental mode of a 0.3 micron wide silicon nitride waveguide. Due to reciprocity the coupling efficiency is the same as in the opposite direction (i.e. from the fundamental mode of the waveguide to the gaussian - shaped mode of the fiber). The estimated 3D coupling loss is found as the sum of the 2D coupling losses from the vertical and the lateral mode conversion. In this example the coupling loss is less than 1dB achieved within only within 11 microns. As shown in the wavelength response curves in Fig.4(b) this result holds over a broad bandwidth and for both polarizations.

Figure 4: (a) FDTD simulation of vertical and lateral mode-size conversion and the respective 2D coupling efficiencies and b) 3D estimate for the coupling efficiencies showing broadband operation and small polarization dependence.

-0.59dB (87%)

-0.71dB (85%)

3D estimate

Tra

nsm

issi

on (d

B)

(b)

x

zy

11 µm

TE: -0.4dB (91%) TM: -0.3dB (93.3%)

(a) 11 µm

y

zx

12 µm

TE: -0.31dB (93%) TM: -0.29dB (93.5%)


24-84

Polarization Mode Dispersion Sponsor MRSEC Program of the National Science Foundation Grant DMR 98-08941 Project Staff: Professor Hermann A. Haus, Professor Erich P. Ippen, John M. Fini, P. B. Phua Polarization mode dispersion (PMD) is a limiting factor in long distance ultrahigh speed optical telecommunication systems. It broadens and distorts the signal propagating through the fiber. This leads to inter-symbol interference, which causes detection errors. By performing appropriate compensation to this pulse broadening effect, one can reduce the bit error rate of the digital transmission.

Our group’s previous work on real-time estimation of 1st order PMD was based on scrambling the input state of polarization (SOP) so that the output time-averaged state of polarization (SOP) is distributed on an ellipsoid in the Stokes space representation. The ellipsoid is in contact with the unit Poincaré sphere at the two points corresponding to the principal states of polarization. Recently, we investigated a new PMD estimation technique which provides first and second order PMD information by exploiting the bandwidth dependence of the averaged SOP measured by a polarimeter [1]. By optically filtering the tapped signal at the receiver end, we measure the averaged SOP for different spectral bandwidths and deduce the PMD parameters. The technique allows real-time PMD monitoring in a feed-forward PMD compensation. In addition, this technique is easily implemented since the PMD diagnosis is done at the receiver end and it utilizes components that are installed in existing telecommunication systems.

In a feedforward PMD compensation scheme, once the PMD parameters are characterized, a compensator needs to be set so as to compensate for the observed PMD. For full PMD compensation up to 2nd order, we need at least three concatenated first order PMD segments, since a two-segment PMD compensator cannot compensate the second order PMD completely because of its frequency-independent differential group delay (DGD). In [2], we studied a particular compensator consisting of three first order PMD segments, one of which is adjustable, concatenated via polarization rotators. Since this is a deterministic approach, we have solved analytically the required individual rotation matrices of the polarization rotators and the required DGD for the variable delay line in the compensator in order to compensate completely any 1st and 2nd order PMD of the transmission cable.

For PMD emulation in the laboratory, emulators have been built by physically concatenating as many as 15 segments of polarization maintaining fiber with polarization scramblers at every PMF junction. The reason for using so many segments is to generate statistical distributions of PMD parameters that resemble those present in the real optical communication fiber. From Figure 1, it is obvious that a small number of concatenated PMF segments are not adequate to produce the tail in the Maxwellian distribution of the differential group delay, if only the polarization is randomly scrambled between these segments. In [3], we propose a novel design of a first and second order PMD emulator based on a configuration of 4 segments: one variable DGD segment concatenated with 3 fixed DGD segments. There are three polarization rotators, one at each junction of the segments. Instead of polarization scrambling at each junction, we set these polarization rotators according to a statistical schedule so as to produce the probability density functions (pdf) of first and second order PMD vectors that resemble those present in a long haul transmission cable (see Figure 2). The essence of this approach is to randomly select a pair of isotropically distributed 1st and 2nd order PMD vectors in Stokes space chosen to follow the pdf of real transmission cables. The 4-segment emulator is set to produce these PMD vectors. Thus, over a large number of independent samplings, the 1st and 2nd order PMD vectors generated by the emulator, will follow the pdf of the real transmission cable.


24-85

Equipment for this project has been provided by The 3M Company.

References [1] P.B. Phua, J.M. Fini, H.A. Haus, E.P. Ippen, “New 1st and 2nd order PMD

characterization using time-averaged state-of-polarization variation with signal’s bandwidth”, OSA Annual Meeting 2001, Long Beach, TuY4

[2] P.B. Phua and H. A. Haus, “Deterministic Approach to Pure 2nd Order Polarization Mode Dispersion Compensation”, to be presented at Optical Fiber Communication Conference 2002, Anaheim.

[3] P.B. Phua and H. A. Haus, “A Deterministically Controlled Four-Segment PMD Emulator”, to be submitted for publication.

Figure 1: Monte Carlo simulation of 1st and 2nd order PMD generated by a concatenation of 4-DGD segments with polarization scramblers between them

Figure 2: Monte Carlo simulation of 1st and 2nd order PMD generated by the same concatenated 4-DGD segments. However, instead of polarization scrambling, the polarization rotators and adjustable DGD segment, are controlled deterministically to produce the realistic pdfs of transmission fiber.

100 200

0.015

τr (ps)

Prob

. Den

sity

(ps-1

)

5000 10000

0.0002

Prob

. D

ensi

ty (p

s-2)

ωτr

(ps2)

100 200

0.015

τr (ps)

Prob

. Den

sity

(ps-1

)

100 200

0.015

τr (ps)τr (ps)

Prob

. Den

sity

(ps-1

)

5000 10000

0.0002

Prob

. D

ensi

ty (p

s-2)

ωτr

(ps2)

5000 10000

0.0002

Prob

. D

ensi

ty (p

s-2)

ωτr

(ps2)ωτr

(ps2)

50 100 150 200 250

0.015

10 20 30 40

0.05

0.1

Prob

. Den

sity

(ps-1

)

τr (ps) ωτr (ps2)50 100 150 200 250

0.015

10 20 30 40

0.05

0.1

Prob

. Den

sity

(ps-1

)

τr (ps)τr (ps) ωτr (ps2)ωτr (ps2)

Prob

. D

ensi

ty (p

s-2)


24-86

Suppression of Radiation from Quarter-Wave Shifted Bragg Resonators Sponsor MRSEC Program of the National Science Foundation Grant DMR 98-08941 Project Staff Professor H. A. Haus, Professor J. D. Joannopoulos, Dr. S. G. Johnson, M. R. Watts. Resonators formed of an optical waveguide Bragg grating with a “defect", a quarter wave shift, have been proposed as channel dropping filters [1,2]. The free spectral range of such filters depends on the strength or index contrast of the Bragg grating which determines the bandgap. As the bandgap is increased, the effective volume of the resonator decreases. A decrease of the effective volume is generally accompanied by an increase of radiation into the surrounding media. We propose designs that, in principle, suppress all radiation, and promise in practice to operate with greatly reduced radiation loss. Figure 1: Cross-section of a junction of two step index waveguides. Our approach is to suppress all radiation at the junction of two waveguides and then form a Bragg resonator from a series of such junctions. We show that under certain conditions radiation can be eliminated at the junction of a pair of step index waveguides (depicted in Figure 1) : one with a core of dielectric constant 1ε and a cladding of 2ε , and the other with a core of 1

~ε and a cladding of 2

~ε . To demonstrate this result, we begin by using the wave equation to develop necessary conditions on both the materials and the modes for the junction to be radiation-free. We then show that a superposition of forward and backward propagating guided modes satisfy the boundary conditions entirely, precluding any coupling to radiation modes. The propagation directions are defined along z and the guided modes e and e~ with z dependences )zjexp( β− and )z~jexp( β− obey the respective wave equations ( ) 022

02T =β−εωµ+∇ e (1)

and ( ) 0~~~ 22

02T =β−ωεµ+∇ e (2)

Since the transverse field profiles must be continuous at the junction of the two waveguides, the transverse mode profiles must have the same functional dependence (i.e. ΤΤ ee ~∝ ) if the field solutions are to be composed solely of guided modes. This requirement in conjunction with (1) and (2) leads directly to the condition that the dielectric contrast must be conserved across the junction

1ε 1~ε

2ε 2~ε

2ε 2~ε

z

n


24-87

2121

~~ ε−ε=ε−ε (3) It is important to note that this condition and the boundary condition at the waveguide walls ( ) 0εε 2211 =−⋅ EEn (4) cannot both be satisfied except in the trivial case for which 11

~ε=ε and 22~ε=ε or when the

normal component of the electric field is zero. Therefore, we restrict our attention to transverse electric (TE) modes. Matching of the transverse E-field profile of a TE wave automatically matches the transverse H-field profile. Indeed, from Faraday’s law

TT HEz 0jz

ωµ−=∂∂

× (5)

we conclude that the matching is automatic without any further constraints. Mode-matching assures that at any junction of two waveguides no radiation modes are excited, only reflection and transmission of the propagating mode occurs. A series of mode-matched junctions can then be spaced at quarter wave separations and a central quarter wave defect introduced to generate a high-Q cavity. Since the subsections are mode-matched, the field is reflected and transmitted at each junction, but no radiation occurs since the mode-match is complete with the guided-mode solutions alone. With no radiation at any interface within the structure, the resonator is radiation-free and can be designed to possess arbitrarily high-Q. A schematic of an ideal two-dimensional resonator and finite-difference time domain (FDTD) simulation of the field are superimposed in Figure 2.

a T

defect-layer

2n~

n2

n1

1n~

Figure 2: Field profile and schematic of an ideal slab resonator As discussed above, in three dimensions the TE01 modes of dissimilar cylindrical waveguides can be matched as well, and an analogous resonant structure may be formed. We do so and calculate the Q's of both structures as a function of the cladding thicknesses (Figure 3). As the cladding thickness is increased, the Q's increase until they are limited by the external Q imposed by the limited number of layer pairs in the Bragg mirrors.


24-88

1 2 3 4 5 6103

104

105

106

107

cavi

ty Q

cladding thickness T (a)

N=10 N=14

2 3 4 5 6101

102

103

104

105

106

cavi

ty Q

cladding diameter T (a)

N=5 N=10

Figure 3: Cavity Q plotted as a function of cladding width for (a) slab and (b) cylindrical cases with N layer pairs on both sides of the “defect". In both cases the indices are 3n~,1n,2n 121 === and

6n~2 = . Note that the effective index contrast is higher in the cylindrical case leading to higher Q for a given number of layer pairs. Practical integrated optic structures typically require rectangular waveguide geometries. Although it does not appear to be possible to achieve a perfect mode-match with a rectangular geometry, strong mode matches can be obtained and radiation substantially suppressed by manipulating the aspect ratio and index contrast of the guide sections. References [1] H. A. Haus and Y. Lai, ``Narrow-band optical channel dropping filter," J. Lightwave Technology, 10, 57-62 (1992) [2] J. N. Damask and H. A. Haus, ``Wavelength division multiplexing using channel-dropping filters," J. Lightwave Technology, 11, 424-428 (1993) [3] H. A. Haus,``Waves and Fields in Optoelectronics," Prentice Hall, 1984


24-89

Micron-size bending radii in silica-based waveguides Sponsor MRSEC Program of the National Science Foundation Grant DMR 98-08941 Project Staff Prof. H. A. Haus, Dr. K. Wada, M. Popović

Low index contrast silica bench technology is widely employed in the fabrication of passive integrated optical components, by virtue of its use of well-tested IC industry manufacturing processes and technology. Large waveguide cross-sections offer low fiber-to-chip coupling and propagation losses. A major drawback is the relatively large component size and thus low density of integration, where a limiting factor is the minimum waveguide bend radius. Bend radii are of millimeter size in the low index contrasts (∆ = 0.25-1.5%) found in silica.1 On the other hand, high index contrast such as in silicon-on-insulator (SOI), while offering dense integration, poses challenges of fiber-to-chip coupling due to mode mismatch and misalignment, and sensitivity to fabrication defects.

A technology that allows a drastic reduction in the bending radius would overcome one of silica’s major obstacles to attaining large-scale optical integration. We propose a scheme using tapered air trenches around bends to provide locally enhanced lateral mode confinement. We use adiabatic tapering to avoid mode mismatch and Fresnel reflection losses at abrupt junctions in order to miniaturize waveguide bends while maintaining broadband low-loss performance.2

Air trenches have been proposed for suppressing bend radiation in several ways.3, 4 When they replace the cladding to enhance lateral mode confinement, mode mismatch-induced junction loss is incurred at points of abrupt change in refractive index and limits the success of the approach in low index contrast. To our knowledge, no attempt has been made to use air trenches to design small, low-loss bends in low index contrast by properly addressing the mode mismatch issue introduced by the present air trench.

We introduce a pair of “inverted” or “cladding” tapers as an integral part of the air trench at the bend (Fig. 1), in order to provide fast mode transition to and from the high index contrast trench region with low radiation loss and low reflection.

Polarization insensitivity was not a design consideration. It is inevitably poor in the high aspect ratio trench in Fig. 1d. Polarization insensitivity of optical processing is to be achieved by splitting the polarizations and rotating one of them, the signal being processed by identical circuits. Following, the polarizations are recombined after another rotation. We use the vertical (out of the wafer plane, Fig. 1) electric field polarization for our designs.

In a junction of two straight waveguides (with a junction similar to the one in Fig. 2a) the trench waveguide width can be optimized for surprisingly low junction loss (<0.1dB even for fiber-like ∆ ~ 0.25%). At optimum, it is wider than the low index input guide. Unfortunately, with a bent trench waveguide as in Fig. 2a this requirement for low junction loss is in direct conflict with the objective of minimizing the bend radius. At small radii, the modal width is determined by the bend radius, not by the waveguide width (whispering gallery regime, e.g. see Ref. 5). Bending loss may still be very small, but a good mode match to the straight waveguide is no longer possible. A straight waveguide has a minimum mode width determined by the index contrast, and thus a priori cannot match well to much narrower modes in the air trench bend. Thus, junction loss sets a lower bound on the bend radius of a trench bend with abrupt junctions (e.g. 15mm for ∆ = 0.1%). To make use of small radii, a low loss mechanism to compress the input mode is needed. We propose using an adiabatic “cladding taper” (Figs. 1, 2c). The taper allows reduction of the radius in the trench region to the point where it is limited by bending loss rather than the modal distribution width required for acceptable junction loss.


24-90

A progression from an abrupt junction to a cladding taper is illustrated in Fig. 2. An abrupt junction (Fig. 2a) requires a large bending radius. A taper after the abrupt junction can be used to compress the mode so that a tighter bend radius can be used (Fig. 2b), but a more natural way to build the transition is by introducing the taper as a slow perturbation of the waveguide’s index profile, as shown in Fig. 2c. This eliminates the Fresnel reflection at the junction of Fig. 2b, and allows an adiabatic evolution of the mode to a shape that is better matched to the leaky bend mode.

To demonstrate a reduction in overall size, we compare conventional waveguide bends and air trench bends designed for the same total loss (98% or 0.1dB/90°). For the chosen transmission, optimization of the present structure - consisting of a bend and two tapers - for minimum total size is a complex problem in many parameters. For simplicity, we choose a bend first, and optimize the taper parameters with respect to the chosen bend design. In lower index examples, the bend contributes little to overall structure size, so we make the radius larger than needed for low bend loss in order to also lower the taper-bend junction loss (Fig. 1a).

In a conventional curved waveguide, bending loss is minimized by coupling to the lowest order leaky mode.5–10 The total loss in a 90° waveguide bend comes from the bending loss of the chosen leaky mode in the bend and from mode mismatch (“junction loss”) at the interfaces between the straight and bent waveguides. In the air trench bend, additional loss is incurred by propagation through each cladding taper, and at the junction where each taper meets the low index contrast waveguide (Fig. 1). Radiation loss due to curvature in regular and air trench bends is evaluated numerically according to the approaches in Refs. 7–9, by the WKB method or by Airy functions. Conventional bend radii for 98% transmission and ATB radii in the bending region are shown in Table 1. In low index contrast conventional waveguide bends, bending loss is dominant and the junction loss can be ignored.

The “cladding tapers” adiabatically shape the fundamental mode of the low index contrast input waveguide (at the input of section I, Fig. 1a) to the shape of the fundamental mode of the high index contrast output (at the output of section II, Fig. 1a). In low index contrast, this output mode can be much smaller than the minimum possible width of the input waveguide mode. The output waveguide width is fixed by the optimal bend design. The cladding tapers we consider are piecewise linear.

The tapers are designed using some simple ray optics as a starting point and characterized by 2D FDTD simulations. Individual taper efficiencies are listed in Table 2, with equal throughput in either direction dictated by reciprocity (assuming lossless dielectrics). We present 2D simulation results for several example air trench bends, designed in the manner described above and chosen to demonstrate the proposed idea. Examples A-C range in index contrast from 0.25% to 7% (Table 1), where the last example (Fig. 3a) is above the range of typical index contrasts found in silica waveguides (∆< 1.5%),1 but was our first structure, used because of its small size in terms of wavelengths and thus short simulation time. The refractive indices that are chosen for all examples are ones that could be produced in SiOxNy core/SiO2 cladding waveguides. Waveguide cross-sections are chosen to be square (in the low index contrast region), and at the cutoff of the second guided mode.

Finite Difference-Time Domain (FDTD) results for transmission loss are obtained by launching a short pulse (on the order of 50fs) into the input waveguide with enough spectral width to span the wavelength spectrum of interest. A discretization of ~20 points per wavelength is used, with the perfectly matched layer (PML) absorbing boundary condition imposed on the edges of the 2D computational domain.

Figures 3a-b are plan view plots of the electric field amplitude superimposed on the outline of the structure. The absence of clear radiation patterns gives qualitative evidence that these bends exhibit low loss. The new bend radii and total edge lengths including tapers (i.e. effective radii) are listed in Table 1 alongside regular bend data, and show drastic size reduction for the same performance. As a metric for the size of bends, the “total” size, we use a box placed around the


24-91

bend which encompasses the bend structure as well as >99.9% of the input and output waveguide modes (Fig. 3a-b).

Figure 4 shows the transmission and back-scattering over the C-band communications window of 1530-1570nm, exhibiting little wavelength dependence as expected, because the structure makes no use of either resonance or multi-mode/path interference. The exception is backscattering in the 180° turn (Example A), where weak Fabry-Perot resonances show up due to the proximity of the input and output waveguides. Markers placed in Fig. 4 at the central wavelength (1550nm) show the values of transmission and reflection into the fundamental waveguide mode (obtained using overlap integrals). These represent the bend transmission loss values of interest, where 98% was targeted in order to compare the bend sizes. Reflection is well below -30dB in all cases, and thus does not present a problem, at least in theoretical design.

For our 2D simulations, effective indices were obtained assuming a guided mode, hence a waveguide cross-section in which the air trench and cladding regions extend infinitely above and below the core (Fig. 5b). The air trench and cladding are in reality of finite vertical extent (Fig. 5a) and there will be substrate loss if the modal index is lower than the index of the bulk cladding below the trench. This effect is enhanced in lower index contrast structures, where the trench is of high aspect ratio, and the mode has a proportionately larger vertical extent (Fig. 5c). In all of these cases, an ideal air trench of infinite extent would be guiding, and it is its truncation at a finite depth that results in leakage loss.

For low loss values, we calculate, using a perturbative method, the required depth of the air trench in Fig. 5a given an acceptable substrate loss (chosen to be much less than total ATB loss). We define an equivalent current sheet at the trench-bulk cladding interface to replace the source mode (of the infinite trench in Fig. 5a), that exactly reconstructs the field in the region of interest below the interface. We then integrate a cylindrical vector Green’s function (e.g., Ref. 11) over the source for the perturbed substrate region (changed from Fig. 5a to Fig. 5b) to evaluate the far field radiation, and thus loss per unit length. In this calculation, we assume that silica cladding continues below the trench - a Si substrate (n = 3.2) below the trench would provide lower loss due to enhanced reflection. We also assume the mode shape remains unperturbed above the trench-bulk cladding interface with the appearance of the leaky substrate region.

We show substrate loss results obtained for ATB examples A, B and C in Fig. 6, where 2D cross-section modal solutions from a vectorial mode solver (e.g. see Fig. 5c) were used in the calculation. From these plots we choose a trench depth for which the substrate loss is much lower than the ATB loss of 0.1dB.


24-92

Figure 1. Air trench bend schematic: (a) plan view; and cross-sectional views in the (b) low index and (c) air trench regions. Contour plots representative of the mode pattern are superimposed on cross-sections.

Figure 2. Air trenches in waveguide bends: a) an abrupt junction, b) an abrupt junction followed by a taper, and c) the proposed adiabatic cladding taper. The bend radius is junction loss limited in (a), but curvature loss limited in (b),(c). Junction loss is present in (a) and (b), but is virtually eliminated in (c).


24-93

Figure 3. Electric field plot from FDTD simulation: (a) Example C (Table 1), (b) Example A (Table 1). The total size, as referred to in the text, is outlined by a box (dash-dot). The inset in (b) shows the 180°bend simulation used for the throughput efficiency of the low index contrast Example A. The key in the lower right corner refers to transmission data in Fig. 4. Table 1: Regular and Air Trench Bend radii and total sizes for 98% transmission.

aDefined as the edge length of a box enclosing the entire bend structure and accommodating >99.9% of the input and output mode power. The minimum possible bend size is a square with an edge equal to this 99.9% mode width. Table 2: Air Trench Bend loss budget.

aTotal loss is due to 2 tapers, 2 junctions and 90°of bending. Estimated individual losses do not exactly add to equal the total loss, because the latter is the result of a simulation of the entire structure and as such is more accurate. Junction loss at the interface between the low index waveguide and the cladding taper is deemed negligible and ignored. bLoss from 180°ATB (note that bending loss is only 0.002 dB/90°).


24-94

Figure 4. (a) Transmitted and (b) backscattered power ratio spectra for examples A (dash-dot), B (dash) and C (solid). True insertion loss and return loss (into the fundamental mode) are shown at the central wavelength (A - triangle, B -circle, C - square).

Figure 5. Air trench cross-section: (a) actual air trench of finite depth; (b) idealized trench, free of substrate loss; and (c) the dominant E-field of the fundamental quasi-TE and -TM modes of the ideal trench for example C.


24-95

Figure 6. Bulk cladding (substrate) loss: loss per unit length for examples A, B and C (from Fig. 3), as a function of displacement of the air trench-bulk cladding interface from the core axis (measured in core heights). References 1. T. Miya, “Silica-based planar lightwave circuits: passive and thermally active devices,” IEEE

J. of Sel. Topics in Quan. Electron. 6, 38–45, Jan 2000. 2. K. Wada, M. Popovic, S. Akiyama, H. A. Haus, and J. Michel, “Micron-size bending radii in

silica-based waveguides,” in Proc. IEEE/LEOS Summer Topical Meeting on WDM Components, 13–14, (Copper Mountain, Colorado), Aug 2001.

3. L. H. Spiekman, Y. S. Oei, E. G. Metaal, F. H. Groen, P. Demeester, and M. K. Smit, “Ultrasmall waveguide bends: the corner mirrors of the future?,” IEE Proc.-Optoelectron. 142,. 61–65, Feb 1995.

4. J. Yamauchi, M. Ikegaya, and H. Nakano, “Bend loss of step-index slab waveguides with a trench section,” Microwave and Optical Technol. Lett. 5, 251–254, Jun 1992.

5. E. C. M. Pennings, Bends in Optical Ridge Waveguides: Modeling and Experiments. PhD thesis, T. U. Delft, The Hague, Netherlands, 1990.

6. E. A. J. Marcatili, “Bends in optical dielectric waveguides,” Bell Sys. Tech. J. 48, 2103–2132, Sep 1969.

1. M. Heiblum and J. H. Harris, “Analysis of curved optical waveguides by conformal transformation,” IEEE J. of Quan. Electron. QE-11, 75–83, Feb 1975.

2. D. Rowland, “Nonperturbative calculation of bend loss for a pulse in a bent planar waveguide,” IEE Proc.-Optoelectron. 144, 91–96, Apr 1997.

3. I.C. Goyal, R.L. Gallawa, and A.K. Ghatak, “Bent planar waveguides and whispering gallery modes:a new method of analaysis,” J. Lightwave Technol. 8, 768-774, May 1990.

4. M. K. Smit, E. C. M. Pennings, and H. Blok, “A normalized approach to the design of low-loss optical waveguide bends,” J. Lightwave Technol. 11, 1737–1742, Nov 1993.

5. J. A. Kong, Electromagnetic Wave Theory, EMW Publishing, Cambridge, MA, USA, 1999.


24-96

Micromachined Photonic Devices using Nonlinear Materials Processing Sponsors Air Force Office of Scientific Research (MFEL) Grant F49620-01-1-0186 Air Force Office of Scientific Research Grant F49620-98-01-0084 Project Staff Andrew M. Kowalevicz, Dr. Ingmar Hartl, Dr. Kaoru Minoshima, Professor Erich P. Ippen, and Professor James G. Fujimoto Photonic device fabrication using nonlinear material processing with near-IR femtosecond pulses can generate localized, clean, three-dimensional (3D) structures in many materials [1-3]. Recently, fabrication of glass waveguides using pulses directly from femtosecond lasers oscillators, without the need for amplification, has been reported [4-6]. Unamplified laser oscillators have several advantages over amplified systems for device fabrication including lower cost and complexity as well as greater control of exposure parameters. Because of their high repetition rates, multiple low intensity shots interact within the focal volume, leading to well controlled cumulative effects. Several photonic devices have been demonstrated including single-mode waveguides [5], X-couplers [6], and directional couplers [4]. Our group has developed a high-power femtosecond Ti:Sapphire laser oscillator using a novel long cavity design [7]. Because of its high pulse energies and high repetition rate, this laser enables the versatile fabrication of a wide range of photonic devices.

We demonstrate the fabrication and characterization of directional couplers and what is, to our knowledge, the first Mach-Zehnder interferometer. Interferometers are of interest because of their use in practical devices such as sensors and switches. These devices are characterized using a variety of techniques including near field and far field mode measurements, interaction length dependence analysis, and wavelength dependent device properties. Laser processing enables rapid prototyping and fabrication of three dimensional device geometries. The laser used to perform the nonlinear material processing is a high pulse energy Ti:Sapphire laser oscillator. The laser cavity has been extended using a Herriott type multipass cell to reduce the repetition rate to 4MHz and increase the pulse energy. Operating in net negative dispersion with a saturable bragg reflector (SBR) to stabilize operation, 80 fs pulses with up to 100 nJ energies can be generated [7]. The laser is focused into a 1 mm thick glass plate using a 100X oil-immersion microscope objective with an effective 0.6 NA. Device structures were fabricated by 2D translation of the sample perpendicular to the incident light using precision computer controlled stages. Figure 1. (a) Phase contrast image of directional waveguide couple with interaction length shown and (b) coupling ratio vs. interaction length for directional waveguide couplers that demonstrates coupled-move behavior.

Interaction length (mm)

Cou

plin

g ra

tio 0.6

0.4

0.2

0.01086420

(C)


24-97

Couplers such as X couplers as well as coupled mode devices have been demonstrated, however detailed characterization requires more than only output spot observation. Here we analyzed mode-coupling behavior. Figure 1 (a) shows a two waveguide directional coupler fabricated inside a glass plate. Waveguides were fabricated by scanning the plate in the X-Y plane at 10 mm/s with an incident pulse energy of ~10 nJ. Evaluation of the mode profile at the output of the coupler shows that the waveguides have diameters of 4-5 µm and are single mode in the near-IR. We fabricated a series of couplers with several different interaction lengths (indicated with an arrow in Figure 1 (a). Figure 1 (c) shows the coupling ratio as a function of the interaction length with 8 µm separation. The data shows that the coupling ratio oscillates as a function of interaction length. The period of oscillation changes as a function of the separation. The oscillatory behavior is the first clear evidence of mode coupling between two waveguides fabricated by the nonlinear laser processing. As an example of a more complex device, we demonstrate, what we believe is for the first time, the fabrication and characterization of a Mach-Zehnder interferometer by the nonlinear laser processing. The interferometer, consisting of two X-couplers placed back-to-back, with crossing angles of 2 degrees. The path length difference between the two arms is approximately 10 µm. In order to characterize the operation of the interferometer, we used the output of a KLM Ti:Sapphire laser which can generate ~5 fs pulses and bandwidths of ~300 nm [8]. This laser source provides a high brightness, single mode beam, which can be easily coupled into the waveguides by using a microscope objective. The edges of the glass substrate were polished to optical quality in order to enable efficient coupling. The output spectrum from the interferometer waveguide was collimated using a second microscope objective. The spectral data was measured with an optical multichannel analyzer (OMA). If a given frequency of light is coupled into the interferometer, it will be split into the two arms of the interferometer at the first X-coupler, travel different path lengths, and will either constructively or destructively interfere after being recombined at the second X-coupler. Thus, the unbalanced path length Mach-Zehnder functions as a filter. The frequency or wavelength dependence of the interferometer can be measured using a broadband light source. Figure 2 (a) shows the input spectrum with a FWHM of 130 nm. Figure 2 (b) shows the wavelength transfer function in the crossed interferometer arm. The transfer function is constructed by normalizing the output spectrum by the input spectrum. With an arm length difference of 10 µm and a spectrum centered at 800 nm, oscillations in the output should occur with periodicity of ~65 nm. We observe a full modulation of the transfer function with a periodicity of approximately 50 nm, in good agreement with theory.

Figure 2. Broad input spectrum, and normalized output spectrum demonstrating the interferometer is acting as a filter with periodic modulation of the spectrum.


24-98

We demonstrate photonic device fabrication using nonlinear materials processing in glass with a high power femtosecond laser oscillator. The building blocks of photonic devices such as X couplers, coupled mode devices, as well as interferometers can be fabricated. Interaction length dependence and wavelength dependent characteristics, using femtosecond broadband lasers as light sources, can be measured. Femtosecond materials processing will enable rapid and versatile device prototyping as well as the fabrication of three-dimensional structures not possible with planer techniques. References 1. K. M. Davis, K. Miura, N. Sugimoto, and K. Hirao, “Writing waveguides in glass with a

femtosecond laser,” Optics Letters, 21: 1729-1731, 1996. 2. D. Homoelle, S. Wielandy, A. L. Gaeta, N. F. Borrelli, and C. Smith, “Infrared

photosensitivity in silica glasses exposed to femtosecond laser pulses,” Optics Letters, 24: 1311-1313, 1999.

3. Y. Sikorski, A. A. Said, P. Bado, R. Maynard, C. Florea, and K. A. Winick, “Optical waveguide amplifier in Nd=doped glass written with near-IR femtosecond laser pulses,” Electronic Letters, 36, 2000.

4. A. M. Streltsov and N. F. Borrelli, “Fabrication and analysis of a directional coupler written in glass by nanojoule femtosecond laser pulses,” Optics Letters, 26: 42, 2001.

5. C. B. Schaffer, A. Brodeur, J. F. Garcia, and E. Mazur, “Micromachining bulk glass by use of femtosecond laser pulses with nanojoule energy,” Optics Letters, 26: 93, 2001.

6. K. Minoshima, A. M. Kowalevicz, I. Hartl, E. P. Ippen, and J. G. Fujimoto, “Photonic device fabrication in glass by use of nonlinear materials processing with a femtosecond laser oscillator,” Optics Letters, 26: 1516-1518, 2001.

7. S. H. Cho, F. X. Kärtner, U. Morgner, E. P. Ippen, J. G. Fujimoto, J. E. Cunningham, and W. H. Knox, “High energy pulse generation using a 4 MHz repetition rate KLM Ti:Al2O3 laser operating with positive and negative dispersion,” Optics Letters, 26: 560-562, 2001.

8. 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,” Optics Letters, 24: 411 -- 413, 1999.


24-99

Publications 1. L.A. Jiang, M.E. Grein, E.P. Ippen, C. McNeilage, J. Searls, and H. Yokoyama, “Quantum-

limited noise performance of a mode-locked laser diode,“ Opt. Lett 27(1):49-51 (2002). 2. M.E. Grein, H. A. Haus, L. A. Jiang, and E. P. Ippen, “Action on pulse position and

momentum using dispersion and phase modulation,” Opt. Express 8(12): 664-669 (2001). 3. C. Chudoba, J.G. Fujimoto, E.P. Ippen, H.A. Haus, U. Morgner, F.X. Kärtner, V. Scheuer, G.

Angelow, and T. Tschudi, “All solid-state Cr:forsterite laser generating 14 fs pulses at 1.3 µm,” Opt. Lett. 26(5): 292-4 (2001).

4. 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, “Generation of octave filling spectra, 5-fs-pulses directly from a Ti:Sapphire oscillator with enhanced dispersion-managed-soliton formation,” Opt. Lett. 26(6): 373-5 (2001).

5. F.X. Kärtner, 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. of the Opt. Soc. of Am. B 18(6), 882-5, (2001).

6. U. Morgner, R. Ell, G. Metzler, T.R. Schibli, F.X. Kärtner, 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): 5462-65 (2001).

7. O.D. Mücke, T. Tritschler, M. Wegener, U. Morgner and F.X. Kärtner, ''Signatures of Carrier-

Wave Rabi-Flopping in GaAs,'' Phys. Rev. Lett. 87 (057401): 1-4 (2001). 8. T.R. Schibli, T. Kremp, U. Morgner, F.X. Kärtner, R. Butendeich, J. Schwarz, H. Schweizer,

F. Scholz, J. Hetzler and M. Wegener ''CW-operation and Q-switched mode locking of Cr4+:YAG-microchip lasers,'' Opt. Lett. 26(12): 941-3 (2001).

9. D. J. Ripin, C. Chudoba, J. T. Gopinath, J. G. Fujimoto, E. P. Ippen, U. Morgner, F. X.

Kärtner, V. Scheuer, G. Angelow and T. Tschudi, ''Generation of 20 fs pulses by a prismless Cr4+:YAG laser,'' Opt. Lett. 27(1): 61-3 (2002).

10. W. Drexler, U. Morgner, R. K. Ghanta, F. X. Kärtner, J. S. Schuman and J. G. Fujimoto,

"Ultrahigh resolution ophthalmic optical coherence tomography," Nature for Medicine 7(4): 502-7 (2001)

11. S. H. Cho, F.X. Kärtner, U. Morgner, E.P. Ippen, J.G. Fujimoto, J.E. Cunningham and W.H.

Knox, “90 nJ pulse generation using a 4 MHz repetition rate KLM Ti:sapphire laser operating with net positive and negative intracavity dispersion,“ Opt. Lett. 26(8): 560-562, (2001).

12. W. Drexler, U. Morgner, R. K. Ghanta, F. X. Kärtner, J. S. Schuman, J.G. Fujimoto, “Ultrahigh

resolution ophthalmic optical coherence tomography,” Nature Medicine 7, 502-507, (2001). 13. C. Chudoba, J. G. Fujimoto, E. P. Ippen, H. A. Haus, U. Morgner, F. X. Kartner, V. Scheuer,

G. Angelow, and T. Tschudi, “All-solid-state Cr:forsterite laser generating 14 fs pulses at 1.3 µm,” Opt. Lett. 26, 292-294, (2001).

14. 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, “Generation of 5-fs pulses and octave-spanning spectra directly from a Ti:Sapphire laser,” Opt. Lett. 26, 373-375, (2001).


24-100

15. I. Hartl, X.D. Li, C. Chudoba, R. Ghanta, T. Ko, J.G. Fujimoto, J. K. Ranka, R. S. Windeler,

and A. J. Stentz, “Ultrahigh resolution optical coherence tomography using continuum generation in an air-silica microstructure optical fiber,” Opt. Lett. 26, 608-610, (2001).

16. F. X. Kärtner, U. Morgner, R. Ell, T. Schibli, J. G. Fujimoto, E. P. Ippen, V. Scheuer, G.

Angelow, and T. Tschudi, “Ultrabroadband double-chirped mirror pairs for generation of octave spectra,” J. Opt. Soc. Am.B, 18, 882-885, (2001).

17. U. Morgner, R. Ell, G. Metzler, T. R. Schibli, J. G. Fujimoto, E. P. Ippen, and F. X. Kärtner, “Nonlinear optics with phase-controlled pulses in the sub-two-cycle regime,” Phys. Rev. Lett., 86, 5462-5465, (2001).

18. K. Minoshima, A. M. Kowalevicz, I. Hartl, E. P. Ippen, and J. G. Fujimoto, “Photonic device fabrication in glass by use of nonlinear materials processing with a femtosecond laser oscillator,” Opt. Lett. 26, 1516-1518, (2001).

Conference Papers 1. L. A. Jiang, M. E. Grein, J. M. Fini, E. P. Ippen and H. A. Haus, “Noise of harmonically

modelocked lasers,” Gordon Research Conference on Nonlinear Optics and Lasers, Colby-Sawyer College, New London, New Hampshire, July 29 - August 3, 2001.

2. L. A. Jiang, M. E. Grein, B. S. Robinson, E. P. Ippen, and H. A. Haus, “Experimental

demonstration of a timing jitter eater,” submitted to the Conference on Lasers and Electro-Optics 2002

3. M. E. Grein, H. A. Haus, E. P. Ippen, and Y. Chen, “The quantum limit of timing jitter in

actively mode-locked soliton fiber lasers”, in OSA Trends in Optics and Photonics (TOPS) Vol. 56, Conference on Lasers and Electro-Optics (CLEO 2001), pp. 243-244.

4. M. E. Grein, L. A. Jiang, H. A. Haus, and E. P. Ippen, “Timing jitter in modelocked lasers,”

IEEE Lasers and Electro Optics Society Annual Meeting, November 12-14, San Diego CA, USA, paper MWP. Invited Talk

5. J. J. Hargreaves, P. W. Juodawlkis, J. J. Plant, J. P. Donnelly, J. C. Twichell, F. Rana, M. E.

Grein, R. J. Ram, E. P. Ippen, “Timing jitter in modelocked lasers,” IEEE Lasers and Electro Optics Society Annual Meeting, November 12-14, San Diego CA, USA, paper MWQ. Invited Talk

6. C. Chudoba, J.G. Fujimoto, E.P. Ippen, H.A. Haus, U. Morgner, F.X. Kärtner, V. Scheuer, G.

Angelow and T. Tschudi, ''All-solid-state Cr:forsterite laser generating 14-fs pulses with 250 nm bandwidth at 1.3 µm,'' Photonics West, San Jose, California, January 20-26, 2001.

7. O. D. Mücke, T. Tritschler, M. Wegener, U. Morgner and F.X. Kärtner, ''Carrier-Wave Rabi-

Flopping in GaAs,'' Quantum Electronics and Laser Science Conference (QELS 2001), Baltimore, USA, May 6-11, 2001.

8. U. Morgner, R. Ell, G. Metzler, T.R. Schibli, F.X. Kärtner, J.G. Fujimoto and E.P. Ippen,

''Nonlinear Optics with Phase-Conrolled Pulses in the Sub-Two-Cycle regime,'' Quantum Electronics and Laser Science Conference, Baltimore, Maryland, May 6-11, 2001.

9. C. Jirauschek, U. Morgner and F.X. Kärtner, ''Spatio-temporal Gaussian Pulse Dynamics in

Sub-10-fs Lasers,'' paper presented at Quantum Electronics and Laser Science Conference, Baltimore, Maryland, May 6-11, 2001.


24-101

10. O. D. Mücke, T. Tritschler, M. Wegener, U. Morgner and F.X. Kärtner, ''Signatures of carrier-

Wave Rabi-Flops in GaAs,'' DFG-Spring-Meeting, Hamburg, Germany, March 26-30, 2001. 11. T.R. Schibli, T. Kremp, U. Morgner, F.X. Kärtner, R. Butendeich, J. Schwarz, H. Schweizer,

F. Scholz, J. Hetzler and M. Wegener, ''CW and Q-switched mode-locked Cr:YAG micro-chip lasers,'' Advanced Solid-State Lasers 2001, Seattle, Washington, January 28-31, 2001.

12. U. Morgner, R. Ell, F.X. Kärtner, J.G. Fujimoto, E.P. Ippen, V. Scheuer, G. Angelow, T.

Tschudi, M.J. Lederer, A. Boiko and B. Luther-Davies, ''Octave-spanning Spectra Directly from a two-foci Ti:sapphire laser with enhanced self-phase modulation,'' Conference on Lasers and Electro-Optics (CLEO 2001), Baltimore, Maryland, May 6-11, 2001.

13. F.X. Kärtner, R. Butendeich, J. Schwarz, H. Schweizer, F. Scholz, J. Hetzler and M.

Wegener, ''Mode-locking of Cr:YAG Microchip Lasers,'' at Conference on Lasers and Electro-Optics, Baltimore, Maryland, May 6-11, 2001.

14. F.X. Kärtner, U. Morgner, E.P. Ippen, J.G. Fujimoto, V. Scheuer, G. Angelow and T. Tschudi,

''Few-Cycle-Pulse Generation and its Applications'', International Conference on Lasers and Electro-Optics (CLEO Pacific 2001), Tokyo, Japan, July 15-19, 2001.

15. W. Drexler, U. Morgner, F.X. Kärtner, C. Spielmann, A. Stingl, F. Krausz, E.P. Ippen, J.G.

Fujimoto and A.F. Fercher, ''Ultrahigh resolution, functional optical coherence tomography using state of the art femtosecond laser technology,'' International Conference on Lasers and Electro-Optics (CLEO Pacific 2001), Tokyo, Japan, July 15 - 19, 2001.

16. U. Morgner, F.X. Kärtner, J.G. Fujimoto and E. P.; Ippen, ''Sub-Two-Cycle Pulse Generation

and Phase-sensitive Nonlinear Optics,'' Gordon Research Conference, New London, New Hampshire, USA, July 30-- August 2, 2001. Invited Talk

17. F. X. Kärtner, ''Noise in optical systems,'' paper presented at Workshop on Modeling,

Simulation and Optimization of Integrated Circuits, Oberwolfach, Germany, 2000. 18. Hartl, X. Li, C. Chudoba, R. K. Ghanta, T. H. Ko, J. G. Fujimoto, J. K. Ranka, R. S. Windeler,

and A. J. Stentz, “Ultrahigh resolution OCT using continuum generation in an air-silica microstructure optical fiber,” International Biomedical Optics Symposium, BIOS’2001, San Jose, CA, January 20-26, 2001, paper 4251-07.

19. X. Li, C. Chudoba, T. H. Ko, C. Pitris, R. K. Ghanta, and J. G. Fujimoto, “Imaging solid

tissues with an OCT imaging needle,” International Biomedical Optics Symposium, BIOS’2001, San Jose, CA, January 20-26, 2001, paper 4251-09.

20. P. Hsiung, X. Li, I. Hartl, C. Chudoba, T. H. Ko, R. K. Ghanta, and J. G. Fujimoto, “High-

speed path length scanning for optical coherence tomography,” International Biomedical Optics Symposium, BIOS’2001, San Jose, CA, January 20-26, 2001, paper 4251-15.

21. R. Ell, U. Morgner, F. X. Kärtner, J. G. Fujimoto, E. P. Ippen, V. Scheuer, G. Angelow, T.

Schudi, M. J. Lederer, A. Boiko, and B. Kuther-Davies, “Octave-spanning spectra directly from a two-foci Ti:sapphire laser with enhanced self-phase modulation,” Conference on Laser and Electro-Optics, CLEO’01, Baltimore, MD, May 6-11, 2001, invited paper CMF1.

22. 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 films saturable absorbers,” Quantum Electronics and Laser Science Conference, QELS’01, Baltimore, MD, May 6-11, 2001, paper QtuF1.


24-102

23. K. Minoshima, I. Hartl, E. P. Ippen, and J. G. Fujimoto, “Versatile photonic device fabrication using nonlinear processing in glass with a femtosecond laser oscillator,” Quantum Electronics and Laser Science Conference, QELS’01, Baltimore, MD, May 6-11, 2001, paper QTuH2. Invited Talk

24. Hartl, C. Chudoba, T. Ko, X. D. Li, J. G. Fujimoto, M. Brezinski, and R. S. Windeler, “Imaging

water absorption with spectroscopic optical coherence tomography,” Conference on Laser and Electro-Optics, CLEO’01, Baltimore, MD, May 6-11, 2001, paper CWE2.

25. U. Morgner, R. Ell, G. Metzler, T. R. Schibli, F. X. Kärtner, J. G. Fujimoto, and E. P. Ippen,

“Nonlinear optical with phase-controlled pulses in the sub-two-cycle regime,” Quantum Electronics and Laser Science Conference, QELS’01, Baltimore, MD, May 6-11, 2001, paper QFC2.

26. P. Hsiung, X. D. Li, I. Hartl, T. Ko, C. Chudoba, and J. G. Fujimoto, “High-speed path length

scanning using a Herriott cell delay line,” Conference on Laser and Electro-Optics, CLEO’01, Baltimore, MD, May 6-11, 2001, poster CWA62.

27. Hartl, C. Chudoba, T. Ko, X. D. Li and J. G. Fujimoto, R. S. Windeler, “Optical coherence

tomography using continuum generation in an air-silica microstructure optical fiber,” European Conference on Biomedical Optics, ECBO, Munich, June 17-22, 2001, paper MH2.

28. X. Li, I. Hartl, C. Chudoba, W. Drexler, T. Ko, P. Hsiung, C. Pitris, R. Ghanta, F. X. Kärtner,

U. Morgner, M. E. Brezinski, and J. G. Fujimoto, “Ultrahigh resolution and spectroscopic optical coherence tomography,” OSA Annual Meeting, ILS-XVII: 17th Interdisciplinary Laser Science Conference, Long Beach, CA, October 14-18, 2001, paper WV1, Invited Talk.

29. J. G. Fujimoto, I. Hartl, P. Xue, T. H. Ko, C. Chudoba, W. J. Wadsworth, T. A. Birks, and R.

S. Windeler, “Supercontinuum generation in photonic crystal fibers as light sources for OCT,” Photonics West, SPIE, San Jose, CA, January 19-25, 2002, paper 4633-26, Invited Talk.

30. A. M. Kowalevicz, T. R. Schibli, R. P. Prasankumar, F. X. Kärtner, and J. G. Fujimoto, “Ultra-

low-threshold, low cost, Kerr lens modelocked Ti:Al2O3 laser,” Conference on Lasers and Electro-Optics (CLEO), Long Beach, CA, May 19-24, 2002, to be presented.

31. Hartl, P. Hsiung, T. H. Ko, J. G. Fujimoto, T. A. Birks, W. J. Wadsworth, U. Bünting, and D.

Kopf, “High resolution OCT imaging using a spectrally broadened femtosecond Nd:Glass laser,” Conference on Lasers and Electro-Optics (CLEO), Long Beach, CA, May 19-24, 2002, to be presented.

Documents

Optics and Quantum Electronics - RLE at MIT · 2019. 11. 20. · 24 - Photonic Materials, Devices and Systems – Optics and Quantum Electronics – 24 RLE Progress Report 144 24-4