Open Access
Add to BibSonomyAbstract
We investigate the dynamical response of a quantum dot photonic integrated circuit formed with a combination of eleven passive and active gain cells operating when these cells are appropriately biased as a multi-section quantum dot passively mode-locked laser. When the absorber section is judiciously positioned in the laser cavity then fundamental frequency and harmonic mode-locking at repetition rates from 7.2GHz to 51GHz are recorded. These carefully engineered multi-section configurations that include a passive wave-guide section significantly lower the pulse width up to 34% from 9.7 to 6.4 picoseconds, as well increase by 49% the peak pulsed power from 150 to 224 mW, in comparison to conventional two-section configurations that are formed on the identical device under the same average power. In addition an ultra broad operation range with pulse width below ten picoseconds is obtained with the 3rd-harmonic mode-locking configuration. A record peak power of 234 mW for quantum dot mode-locked lasers operating over 40 GHz is reported for the first time.
©2007 Optical Society of America
1. Introduction
The compact size, low power consumption, and direct electrical pumping of monolithic mode-locked lasers (MLLs) make them promising candidates for inter-chip/intra-chip clock distribution [1–3] as well as high bit-rate optical time division multiplexing, electro-optic sampling, and impulse response measurement of optical components [4]. Typically these applications require high repetition rate of the order of tens of GHz, a peak power larger than 1 W and a pulse width shorter than 10 picoseconds, all of which can be met by quantum dot mode-locked lasers.
Some unique advantages of quantum dot (QD) materials, such as ultra broad bandwidth, ultra fast gain dynamics, and easily saturated gain and absorption, make them an ideal choice for semiconductor monolithic MLLs [5–9]. In this letter, we present novel reconfigurable multi-section monolithic passive MLLs that emit near 1.3-μm using quantum dot active regions on a GaAs substrate. In the same device, we show experimentally that these optimized multi-section configurations containing gain, absorber, and passive sections significantly lower the pulse width (9.7 ps to 6.4 ps, up to 34%) and increase the peak pulsed power (150 mW to 224 mW, up to 49%) of the QD MLL compared with conventional 2-section configurations consisting of only gain and absorber regions. The multi-section geometry also allows modification of the absorber position within the same optical cavity to achieve complete harmonic mode-locking up to the 7th harmonic and incomplete locking as high as the 15th harmonic. With an optimized 3rd harmonic configuration, an extra-broad operation range with pulse width below 10 picoseconds was obtained.
This work is organized in the following manner. In the device design section the basic monolithic laser operating principles are outlined. In the materals section the details of the growth and composition of the device are presented. In the section on results and discussion, all the recorded data are presented with detailed stability maps as well examples of the pulse width versus power. Finally in the concluding section we summarize our findings, draw conclusions and present a series of open problems and future applications for these kinds of photonic integrated circuits.
2. Device design
A large body of both theoretical and experimental research has been performed on the issues of the semiconductor quantum well and quantum dot passive mode-locked lasers [10–17]. A common condition for passive mode-locking in lasers that elucidates why quantum dot active regions are desirable is the requirement on the stability parameter, s, defined as:
Where Esat,abs is the saturation energy of the absorber, Esat,g is the saturation energy for the gain section, h is Planck’s constant, ν is the optical frequency, Gg is the differential gain in the gain section, Ga is the differential loss in the absorber and A is the optical mode cross-sectional area, which is equal in the absorber and gain sections of the monolithic semiconductor laser [18]. The parameter s can also be expressed as:
where ntr 2D is the transparency density of 2D system such as a quantum well (QW), and Γ is the optical confinement factor [19,20]. In Eq. (2), we have substituted the transparency condition for the quantum dot case, which highlights the inverse relationship between s and the dot density, nQD. Unlike QW devices, QD lasers have the advantage that nQD can be directly controlled during crystal growth.
According to the work of Lau and Paslaski [15, 16], optical pulses will be generated in a passive mode-locked laser without being contaminated with self-pulsations if the following inequality is fulfilled
In this inequality g0 is the steady-state gain, and gth is the cavity loss of the resonator. One of the key assumptions determining the validity of Eq. (3) is:
where S0 is the photon density in the laser. Equation (4) states that the stimulated lifetime must be much less than the carrier lifetime in the respective sections of the laser. This requirement is easily met in the absorber, but in the gain section near threshold in a quantum well device, this condition is difficult to achieve. In the quantum dot active region, however, GgS0>>1/τg is easier to achieve because the carrier lifetime does not vary strongly with carrier density as it does in quantum wells [21]. This fact and ultra-fast gain recovery may explain why self-pulsation is rarely observed in quantum dot mode-locked lasers [13].
Equation (3) suggests that the design strategy of a passive MLL is to minimize the ratio of gth/g0 and Gg/Ga. Although this equation is not directly correlated to pulse width and peak power, it is reasonable to expect that improving these ratios will benefit these operating parameters. Since the threshold condition requires the average gain minus the average loss, a0, to equal the threshold gain, g0 - a0 = gth, it is difficult to decouple gth and g0 while optimizing gth/g0. The preferred approach as described by Lau and Paslaski [16] is to decrease gth by decreasing the internal loss, applying a high reflection coating to the mirrors or increasing the total cavity length. To maximize s it is noted that the gain of QDs is easily saturated, and, consequently, the differential gain Gg is very small at high current density [22]. Also, applying a high reverse voltage on the absorber will increase the differential loss Ga. However, a standard two-section passive mode-locked laser can operate with a low Gg and high g0 only by simultaneously increasing gth, which is undesirable. The multi-section laser including a passive section avoids this tradeoff.
Fig. 1. The layouts and probing scheme of the reconfigurable MLLs designed all within the same multi-section device. The probes biased with reverse voltage, gain currents and transparency current are colored in blue, red and cyan respectively. (a) A0.5G5.0, (b) A0.5P0.5G4.5. (c) A0.5P1.0G4.0. (d) A0.5G4.0P1.0. (e) A0.5G3.0G2.0. The light red indicts a lower current density. (f) G2.0A0.5G3.0, the asymmetrical configuration for higher order harmonic excitation (7th harmonic in this example).
A multi-section MLL device consists of an additional passive waveguide section besides the absorber and gain sections. The passive waveguide section is used to decrease the gain section length, which forces the desirable result: g0 is higher, Gg is lower and at the same time, gth, remains unchanged; hence, the mode-locked condition will be more easily satisfied. This design increases the saturation photon density by running in strong population inversion, decreases nonlinear gain suppression and will result in higher peak powers also. A reconfigurable multi-section device consists of a single-mode ridge waveguide divided into electrically isolated sections that can be biased independently. By altering the bias on each section, the reconfigurable multi-section device operates either as a conventional two-section or multi-section MLLs with different section configurations as shown in Fig. 1. Therefore, the characteristics of different geometrical designs can be compared within the same waveguide device eliminating the impact of variation between different devices [23]. Considering that it is desirable to locate the absorber near a high-reflectivity (HR) mirror facet to induce the quasi-colliding pulse effect [24], there are two realistic layouts for mode-locking the three-section MLL at the round-trip frequency of the laser cavity. In the absorber-gain-passive (AGP) design (seen in Fig. 1), the gain section is positioned between the absorber and the passive section, and in the absorber-passive-gain (APG) layout, the passive section is between the absorber and the gain areas. During operation, the passive section is biased with an appropriate level of current to achieve optical transparency since we are not able to fabricate a truly passive section in these particular designs. A third approach for fundamental harmonic mode-locking is the dual-gain or AGG geometry in which the amplifiers are pumped at different current densities. This AGG design is intended to widen the gain spectrum in the MLL device and generate a narrower pulse width due to an increase in the number of locked modes. The final triple-section configuration of interest locates the absorber between gain sections (GAG) and can be implemented to excite higher order harmonics of the fundamental repetition rate using the asymmetrical colliding pulse technique [25, 26].
3. Materials and fabrication
The laser epitaxial structure of this device is a multi-stack “Dots-in-a-WELL” (DWELL) structure that is an optimized six-stack QD active region grown by solid source molecular beam epitaxy on a (001) GaAs substrate [27, 28]. The 3.5-μm wide ridge-waveguide devices are fabricated following standard multi-section device processing [22]. One cleaved facet was HR-coated (R ≈ 95%) and the other facet was low reflection (LR)-coated (R ≈ 15%). The internal loss is 2.0 cm-1 and the threshold gain gth is 3.8 cm-1. The total device length of 5.5-mm is segmented into 11 implant-isolated 0.5-mm sections that are pumped through an eleven-pin probe card by either a multi-channel current source to the various individual gain portions or a voltage source supplying reverse bias to the absorber. The eleven-section device is systematically configured in the APG, AGP, AGG and GAG layouts. For convenience, the different configurations are labeled with subscripts denoting the length in mm’s. For example, an AGP structure with a 0.5-mm absorber, a 4.5-mm gain section and a 0.5-mm passive waveguide section is labeled as A0.5G4.5P0.5. The measurements are performed at a controlled substrate temperature of 20°C.
4. Results and discussion
For pulse width and peak power improvement at the fundamental repetition rate of 7.2 GHz, six distinct geometrical configurations were tested: A0.5G5.0, A0.5P0.5G4.5, A0.5P1.0G4.0, A0.5G4.0P1.0, A0.5G4.5P0.5 and A0.5G3.0G2.0. The absorber was biased with a voltage of -3V during testing. From the results of the net modal gain and absorption spectra shown in the Fig. 2, which were obtained with the segmented contact method [22, 29], the passive section was biased with a current density of 1.5 mA/section (87.5 A/cm2) to achieve optical transparency.
As generally shown in Figs. 3(a), 3(b), and 3(c), the pulse widths and the corresponding peak powers as a function of the average output power for each of the 3-section MLL configurations are narrower and higher compared to the traditional 2-section configuration. In Fig. 3(a), the trends as a function of the passive section length are examined. Compared to the A0.5G5.0 configuration, the A0.5P0.5G4.5 device has at best an 11% narrower pulse width at an average power of 20 mW. By increasing the passive waveguide section to 1 mm, the A0.5P1.0G4.0 configuration decreases the pulse to 15.3 ps at the same average power of 20 mW and achieves a 22.2% improvement compared to the A0.5G5.0 configuration. The Fig. 3(a) data of the A0.5P1.0G4.0 device showing a significantly smaller increase in pulse width as function of average power clearly exhibits that lengthening the passive section increases the saturation power in the gain section by running in stronger inversion. However, the data also shows that the A0.5P1.0G4.0 is not as stable near threshold probably because the peak photon density has been shifted away from the absorber making the device more susceptible to spontaneous emission noise.
Fig. 2. (a). The measured net modal gain under CW pump and absorption spectra of the quantum dot active region. (b) The net modal gain at the wavelength of 1216nm from (a). The transparency current density is 1.5mA/section as shown with the arrow. The steady-state gain g0 of the configuration A0.5G5.0 is 4.2 cm-1 and the corresponding dg/dJ is 5.1×10-3 cm/A (point A). The g0 of the configuration A0.5P0.5G4.5 is 5.2cm-1 and the corresponding dg/dJ is 2.7×10-3 cm/A (point B). With a 1-mm passive waveguide section, the A0.5P1.0G4.0 configuration has a g0 of 5.8 cm-1 (point C) and a small dg/dJ of 0.75×10-3 cm/A. The A0.5G4.0P1.0 configuration has a near saturated g0 of 5.85 cm-1 (point D) and the smallest dg/dJ of 0.6×10-3 cm/A.
Figure 3(b) demonstrates that the AGP configuration is superior to the APG configuration. Above 12 mW average power, the AGP devices have 1–3 ps narrower pulse widths than the APG’s. In comparison to the conventional A0.5G5.0 configuration, the A0.5G4.0P1.0 configuration significantly lowers the pulse width up to 34% from 9.7 ps to 6.4 ps at an average power of 10.5mW and increases the peak pulsed power by 49% from 150 mW to 224 mW. These results are the best improvements attained.
Table 1. Pulse width and peak power improvements in the APG and AGP configurations.
Fig. 3. The pulse width (left) and the peak power (right) of the different multi-section QD MLL configurations compared with the A0.5G5.0 configuration. (a) A0.5P0.5G4.5 and A0.5P1.0G4.0 configurations. (b) A0.5G4.5P0.5 and A0.5G4.0P1.0 configurations. (c) A0.5G3.0G2.0 configuration.
The reality that the MLLs with a longer passive section have superior performance in terms of pulse width and peak power confirms the hypothesis that running the gain section in strong population inversion improves the MLL performance [30]. With the measured gain spectra shown in the Fig. 2(a), we can obtain the steady-state gain, g0, and the differential gain, Gg, for the different configurations, which are shown in the Table 1, and understand what degree of change in these parameters is necessary to produce the observed pulse width and peak power improvements. For the conventional 2-section MLL configuration A0.5G5.0, the gain section operates at g0 of 4.2 cm-1, which is shown as point A in the Fig. 2(b). It is assumed that the differential gain with respect to the current density, dg/dJ, is proportional to the differential gain Gg. At point “A”, dg/dJ is 5.1×10-3 cm/A. For the A0.5P0.5G4.5 configuration, the device works under a higher steady-state gain (g0 = 5.2 cm-1) and a smaller differential gain (dg/dJ = 2.7 ×10-3 cm/A) than the A0.5G5.0 configuration. Increasing the passive waveguide section to 1 mm (A0.5P1.0G4.0) forces the gain section to function under even higher current density where g0 is almost saturated at 5.8 cm-1 and the dg/dJ is about 0.75×10-3 cm/A. Thus to decrease the pulse width by as much as 27% and increase the peak power by 36%, g0 was increased by 38% and dg/dJ was decreased by a factor of 6.8. Evidently, there are other physical effects that limit pulse width and peak power performance such as chirp in the pulse.
Point C in the Fig. 3(b) corresponds to the operating point of the alternative A0.5G4.0P1.0 configuration. It has a slightly higher g0 (5.85 cm-1) and lower dg/dJ of 0.6×10-3 cm/A than the related A0.5P1.0G4.0, but noticeably higher peak power on average—about 12%. Besides the higher steady-state gain and lower differential gain, the improvement between the A0.5P1.0G4.0 configuration and A0.5G4.0P1.0 configuration is probably caused by reduced non-linear gain effects from a shorter continuous gain path in the AGP. Finally, neither the AGP nor the APG lase when the length of the passive waveguide section increases to 1.5 mm because g0 exceeds the maximum ground state gain in the quantum dot.
There are many approaches to the AGG configuration since the saturation in the gain section is less of an issue than in the APG and AGP layouts. In the test, the A0.5G3.0G2.0 configuration was found optimum to achieve a flat gain spectrum with the widest bandwidth. At an average emission power of 10.9 mW, the best peak power of the A0.5G3.0G2.0 configuration is 201 mW, which is a 30% improvement compared to A0.5G5.0, at a current density ratio between the short and long gain sections of 3. The AGG geometry has no significant advantage over the APG or AGP layouts probably because the gain spectrum as shown in Fig. 2(a) is already rather inhomogeneous under high injection for this particular QD laser wafer. It is noticed that the minimum time-bandwidth product (TBP) occurs at the minimum pulse width condition and increases in the configurations with a passive waveguide section (3.4–6.9) compared to the 2-section configuration (1.9). This difference is believed to be due to the influence of a larger linewidth enhancement factor in the higher pumped gain section.
To achieve harmonic mode-locking at multiples of the 7.2 GHz fundamental frequency, the 0.5-mm absorber was systematically placed in one of the possible cavity sections that are numbered sequentially from 1 to 11, where 1 represents the position next to the LR-coated mirror. Asymmetric mode-locking at the n th harmonic is expected by placing the absorber at the location L/n, where L is the length of the cavity [25, 26]. In an asymmetrical colliding pulse mode-locked laser with discrete section-lengths like the multi-section GAG MLL described here, the number n representing the harmonic supported in the MLL is a function of the total number N of equal-length segments in the laser cavity and the segment position number m of the absorber:
Since positions m =1 and m = 11 are contiguous to a mirror and result in fundamental harmonic mode-locking, Eq. (5) is valid only for 1<m<N. In practice, Eq. (5) is used to see how close the right-hand side of the equation is to a positive integer given the experimental constraints. Since there are 11 equal segments in the cavity, (N=11), the position m = 6 yields n = 2 (exactly) and mode-locking at the 2nd harmonic, typically called colliding pulse mode-locking, is anticipated at 14.4 GHz. All other positions m yield n close to some integer. In other words, the cavity is not set up perfectly for harmonic mode-locking for n ≥ 3, although for m = 4 or 8, n = 3.14, which is reasonably close to 3. Thus, the third harmonic works acceptably in this absorber arrangement as our experiment will show. m = 3 or 9 gives n = 4.40, which is ambiguous as to which harmonic should lock. m = 2 or 10 gives n = 7.33, also rather indeterminate.
Fig. 4. Stabilty maps of the optical pulse width (left) and peak power (right) vs. gain current and absorber voltage for the GAG configurations. (a) The fundamental harmonic mode-locking. Section 11 was reverse biased as the absorber. (b) The 2nd harmonic mode-locking. The section 6 was reverse biased as the absorber. (c) The 3nd harmonic mode-locking. The section 8 was reverse biased as the absorber. An extra-broad operation range with pulse width below 10 ps was obtained due to symmetric pulse collision.
Inserting the absorber at position 9, 8, 6, 5 or 4 experimentally results in background-free mode-locking (Fig. 5), and the corresponding repetition rates for the different absorber locations are listed in Table 2. The pulse shape traces shown in the Fig. 5 and Fig. 6 are obtained from a background free Femtochrome autocorrelator.
Fig. 5. (778 KB) Movie showing the absorber location and the corresponding mode-locked pulse shape [Media 1]
With the absorber positioned at sites 1, 2, 3, and 10, no mode locking was observed. For the first 3 cases, the result is attributed to the dissimilar intensity of the colliding pulses caused by the low-reflection coating on the nearby mirror. Location 10 is close to ideal for locking on the 7th harmonic, but this is not seen. Locations 4 and 8 are geometrically equivalent and give the same result of 21.6 GHz, which is the 3rd harmonic. The mode-locking pulse and peak power of layouts with the absorber at location 11, 6 and 8 are mapped as a function of gain current and absorber voltage in Fig. 4. For the colliding pulse mode-locking (site 6), the absorber is more efficiently bleached with a relatively symmetric pulse collision, and the operation range expands compared to mode-locking on the fundamental harmonic (site 11). Next by inserting the absorber at location 8, a further increase in the sub-10 ps working range is obtained due to an even more symmetric pulse collision. The gain current-absorber voltage operating area for m = 8 is almost 3 times larger than the area of the arrangement functioning at the fundamental repetition rate. In contrast, the stable mode-locking range of the m = 4 MLL (also 3rd harmonic) covers only a very narrow area because of a more asymmetrical pulse collision, even though the intensity sum of the colliding pulses is higher compared to the m = 8 device. This comparison of the m = 4 and m = 8 devices emphasizes that symmetric pulse collision is more important than the total intensity for ensuring stable mode-locking.
Fig. 6. (a). The pulse shape of the G3.0A0.5G2.0 configuration with a gain current density of 857 A/cm2 and a reverse bias of 1V. The absorber is positioned at site 5, corresponding to 7th harmonic mode locking. The pulse width is 6 ps and the peak pulsed power is 70 mW. (b) The pulse shape of the incomplete mode-locking at 115 GHz with the absorber at section 9 and biased at 0 V. The gain current is 120 mA.
Positions 5 and 7 are sites of very low symmetry (predicted to mode-lock between the 2nd and 3rd harmonic). However, robust mode-locking was achieved at the 7th harmonic of 50.7 GHz with the absorber at site 5. Although this observation is not well understood at the time, the result may arise from either non-linear dynamic interaction between the gain and absorber or a slight offset in the cavity segmentation caused by mirror cleaving uncertainty. Using the 50.7 GHz configuration, a pump current density of 857 A/cm2 and a reverse bias of 1V, yielded pulse widths of 6 ps and a peak power of 70 mW. The pulse shape is shown in Fig. 6(a) demonstrating complete mode-locking. Location 9 gave the result of 42.4 GHz. Under conditions of a reverse bias of 8 V and a gain current of 340 mA, a record peak power of 243 mW for QD mode-locked lasers operating over 40 GHz. was achieved at a pulse width of 6.6 ps.
Table 2. Harmonic mode-locking repetition rates and TBP achieved with the absorber placed at different positions
With the absorber placed at section 9 and the bias voltage set at zero V, we obtained incomplete mode-locking at 115 GHz with pump currents ranging from 100 mA to 130 mA. The pulse shape is shown in the Fig. 6(b). It is expected that we can obtain even higher frequencies by cleaving the end sections to different lengths without decreasing the absorber’s length. This technique offsets the center position of the absorber to alternate locations that can more efficiently excite higher order harmonics.
5. Conclusion
Novel multi-section quantum dot passively mode-locked lasers have been discussed. Reconfigurable multi-section MLL devices were fabricated that permit the comparison of the characteristics of different MLL structures within the same device. The multi-section MLL configurations significantly increase the peak pulsed power up to 49% and reduce the pulse width up to 34% in the QD mode-locked lasers because the passive waveguide section increases the saturation photon density, decreases nonlinear gain suppression in the device and makes the passive MLL devices mode-lock more easily. A minimum pulse width of about 6.5 ps has been realized for a 5.5-mm cavity length, 7.2 GHz repetition rate device. By positioning the absorber at different locations along the laser cavity, we obtained high-order harmonic MLLs with repetition rate up to 50.7 GHz and incomplete mode locking device with repetition rate as high as 115 GHz. A record peak power of 234 mW for QD mode-locked lasers operating over 40 GHz is recorded. Future photonic integrated circuits such the simple ones described in this work may be key building blocks [23, 31] on future “Lidar on a Chip” systems that significantly reduce the LIDAR system size, cost, and weight, as well as allow for dramatic frequency and waveform agility [31, 32]. This type of technology solution using QD photonic interacted circuits will enable the development of small-scale seeker systems such as small UAV’s for urban operations and sensing.
Acknowledgments
The authors gratefully acknowledge sponsorship of this work from AFRL grant FA8750-06-1-0085, AFOSR grant FA9550-06-1-0411 and the STEPP/NRC program. They also are grateful to the many individuals for discussions and perspective on the LIDAR and AWG issues including John Malowicki, Dr. John Albrecht, Rick Fedors and Dr. Paul F. McManamon.
References and links
1. D. A. B. Miller, “Rationale and challenges for optical interconnects to electronic chips,” Proceedings of the IEEE88, 728-749 (2000).
2. G. A. Keeler, B. E. Nelson, D. Agarwal, C. Debaes, N. C. Helman, A. Bhatnagar, and D. A. B. Miller, “The benefits of ultrashort optical pulses in optically interconnected systems,” IEEE J. Sel. Top. Quantum Electron. 9, 477–485 (2003). [CrossRef]
3. K. A. Williams, M. G. Thompson, and I. H. White, “Long-wavelength monolithic mode-locked diode lasers,” New J. Phys. 6, 179 (2004). [CrossRef]
4. L. Zhang, L. Cheng, A. L. Gray, S. Luong, J. Nagyvary, F. Nabulsi, L. Olona, K. Su, T. Tumolillo, R Wang, C. Wiggins, J. Ziko, Z. Zau, P. M. Varangis, H. Su, and L. F. Lester, “5 GHz Optical Pulses From a Monolithic Two-Section Passively Mode-locked 1250/1310 nm Quantum Dot Laser for High Speed Optical Interconnects,” Optical Fiber Communication Conference. Technical Digest. OFC/NFOEC 3, (2005). [CrossRef]
5. X. D. Huang, A. Stintz, H. Li, L. F. Lester, J. Cheng, and K. J. Malloy, “Passive mode-locking in 1.3 μm two-section InAs quantum dot lasers,” Appl. Phys. Lett. 78, 2825–2827 (2001). [CrossRef]
6. X. D. Huang, A. Stintz, H. Li, A. Rice, G. T. Liu, L. F. Lester, J. Cheng, and K. J. Malloy, “Bistable operation of a two-section 1.3-μm InAs quantum dot laser - Absorption saturation and the quantum confined Stark effect,” IEEE J. Quantum Electron. 37, 414–417 (2001). [CrossRef]
7. R. L. Sellin, C. Ribbat, M. Grundmann, N. N. Ledentsov, and D. Bimberg, “Close-to-ideal device characteristics of high-power InGaAs/GaAs quantum dot lasers,” Appl. Phys. Lett. 78, 1207–1209 (2001). [CrossRef]
8. M. Kuntz, G. Fiol, M. Lammlin, D. Bimberg, M. G. Thompson, K. T. Tan, C. Marinelli, R. V. Penty, I. H. White, V. M. Ustinov, A. E. Zhukov, Y. M. Shernyakov, and A. R. Kovsh, “35 GHz mode-locking of 1.3μm quantum dot lasers,” Appl. Phys. Lett. 85, 843–845 (2004). [CrossRef]
9. A. R. Rae, M. G. Thompson, R. V. Penty, I. H. White, A. R. Kovsh, S. S. Mikhrin, D. A. Livshits, and I. L. Krestnikov, “Harmonic mode-locking of a Quantum-Dot Laser Diode,” 874–875. 2006IEEE LEOS Annual Meeting . [CrossRef]
10. H. A. Haus, “Mode-locking of lasers,” IEEE J. Sel. Top. Quantum Electron. 6, 1173–1185 (2000). [CrossRef]
11. H. A. Haus and Y. Silberberg, “Theory of mode-locking of a Laser Diode with a multiple-quantum-well structure,” J. Opt. Soc. Am. B 2, 1237–1243 (1985). [CrossRef]
12. G. New, “Pulse evolution in mode-locked quasi-continuous lasers,” IEEE J. Quantum Electron. 10, 115–124 (1974). [CrossRef]
13. E. A. Viktorov, P. Mandel, A. G. Vladimirov, and U. Bandelow, “Model for mode locking in quantum dot lasers,” Appl. Phys. Lett. 88, 201102 (2006). [CrossRef]
14. F. Lelarge, B. Dagens, J. Renaudier, R. Brenot, A. Accard, F. van Dijk, D. Make, O. Le Gouezigou, J. G. Provost, F. Poingt, J. Landreau, O. Drisse, E. Derouin, B. Rousseau, F. Pommereau, and G. H. Duan, “Recent advances on InAs/InP quantum dash based, semiconductor lasers and optical amplifiers operating at 1.55 μm,” IEEE J. Sel. Top. Quantum Electron. 13, 111–124 (2007). [CrossRef]
15. K. Y. Lau and J. Paslaski, “Condition for short pulse generation in ultrahigh frequency mode-locking of Semiconductor-Lasers,” IEEE Photon. Technol. Lett. 3, 974–976 (1991). [CrossRef]
16. J. Palaski and K. Y. Lau, “Parameter ranges for ultrahigh frequency mode-locking of semiconductor-lasers,” Appl. Phys. Lett. 59, 7–9 (1991). [CrossRef]
17. H. Haus, “Parameter ranges for CW passive mode locking,” IEEE J. Quantum Electron. 12, 169–176 (1976). [CrossRef]
18. P. Vasilev, Ultrafast diode lasers Fundamentals and applications, (Artech House, Boston, 1995), Chap. 4.
19. R. G. M. P. Koumans and R. vanRoijen, “Theory for passive mode-locking in semiconductor laser structures including the effects of self-phase modulation, dispersion, and pulse collisions,” IEEE J. Quantum Electron. 32, 478–492 (1996). [CrossRef]
20. K. Yvind, D. Larsson, L. J. Christiansen, Leif K. Oxenlowe, Jesper Mork, Jorn M. Hvam, and Jesper Hanberg, “Design and evaluation of mode-locked semiconductor lasers for low noise and high stability,” Proc. SPIE 5825, 37–48 (2005). [CrossRef]
21. P. Blood, H. Pask, I. Sandall, and H. Summers, “Recombination in quantum dot ensembles,” Proc. SPIE 64850J, Novel in-Plane Semiconductor Lasers VI 6485, (2007).
22. Y. C. Xin, Y. Li, A. Martinez, T. J. Rotter, H. Su, L. Zhang, A. L. Gray, S. Luong, K. Sun, Z. Zou, J. Zilko, P. M. Varangis, and L. F. Lester, “Optical gain and absorption of quantum dots measured using an alternative segmented contact method,” IEEE J. Quantum Electron. 42, 725–732 (2006). [CrossRef]
23. M. Yousefi, Y. Barbarin, S. Beri, E. A. J. M. Bente, M. K. Smit, R. Notzel, and D. Lenstra, “New role for nonlinear dynamics and chaos in integrated semiconductor laser technology,” Phys. Rev. Lett. 98, 044101 (2007). [CrossRef] [PubMed]
24. D. J. Derickson, R. J. Helkey, A. Mar, J. R. Karin, J. G. Wasserbauer, and J. E. Bowers, “Short pulse generation using Multisegment Mode-Locked Semiconductor-Lasers,” IEEE J. Quantum Electron. 28, 2186–2202 (1992). [CrossRef]
25. T. Shimizu, X. L. Wang, and H. Yokoyama, “Asymmetric colliding-pulse mode-locking in InGaAsP semiconductor lasers,” Opt. Rev. 2, 401–403 (1995). [CrossRef]
26. T. Shimizu, I. Ogura, and H. Yokoyama, “860 GHz rate asymmetric colliding pulse modelocked diode lasers,” Electron. Lett. 33, 1868–1869 (1997). [CrossRef]
27. G. T. Liu, A. Stintz, H. Li, K. J. Malloy, and L. F. Lester, “Extremely low room-temperature threshold current density diode lasers using InAs dots in In0.15Ga0.85As quantum well,” Electron. Lett. 35, 1163–1165 (1999). [CrossRef]
28. A. Stintz, G. T. Liu, A. L. Gray, R. Spillers, S. M. Delgado, and K. J. Malloy, “Characterization of InAs quantum dots in strained InxGa1-xAs quantum wells,” J. Vac. Sci. Technol. B 18, 1496–1501 (2000). [CrossRef]
29. S. W. Osborne, P. Blood, P. M. Smowton, Y. C. Xin, A. Stintz, D. Huffaker, and L. F. Lester, “Optical absorption cross section of quantum dots,” J. Phys. Condens Matter 16, S3749–S3756 (2004). [CrossRef]
30. K. Y. Lau, “Short-Pulse and High-Frequency Signal Generation in Semiconductor-Lasers,” J. Lightwave Technol. 7, 400–419 (1989). [CrossRef]
31. Y. Liu, Z. Wang, M. Han, S. Fan, and R. Dutton, “Mode-locking of monolithic laser diodes incorporating coupled-resonator optical waveguides,” Opt. Express 13, 4539–4553 (2005). [CrossRef] [PubMed]
32. F. Y. Lin and H. M. Liu, “Diverse waveform generation using semiconductor lasers for radar and microwave applications,” IEEE J. Quantum Electron. 40, 682–689 (2004). [CrossRef]
