Mode locked lasers are a compact, inexpensive source of optical pulses. Currently, some problems with semiconductor sources are low output power, self phase modulation, secondary pulse formation, and timing jitter. These problems are addressed here

Experimental results from directly modulated semiconductor lasers at 30 Gbit/s are presented. Critical problems for 30 Gbit/s electronic TDM systems are discussed as well as advances in component technology for 30 Gbit/s transmission systems.

A comprehensive timing jitter comparison is made for mode-locked semiconductor lasers using active, passive, and hybrid mode-locking techniques in both external and monolithic cavity configurations. Active mode locking gives the lowest residual rms timing jitter of 65 fs (150 Hz-50 MHz), followed by the hybrid and passive mode-locking techniques. It is found that monolithic cavity devices with all active waveguides have higher timing jitter levels than the comparable external cavity case.

The issue of optimum facet reflectivity for large bandwidth semiconductor lasers is addressed. The laser facet reflectivity, is chosen for maximising the relaxation oscillation frequency, by optimisation of the photon lifetime and maximum obtainable intracavity photon density. The optimum mirror reflectivity also depends on the relative importance of photon density-induced facet damage, parasitic ...

A new, computationally efficient semiconductor-laser model has been developed for subpicosecond mode-locked laser analysis. It consists of a set of infinitesimal laser sections resulting from longitudinal integration of the laser rate equations. The model is used to predict the response of an actively mode-locked semiconductor laser for various pulsed-current drives.

Pulses shorter than 4 ps (deconvolved) have been obtained by optically gain switching a GaAs multiple quantum well vertical-cavity surface-emitting laser with a picosecond dye laser. Pulse width and relative peak delay were measured as a function of pump power. A theoretical model of the large signal response agrees well with the measured data. The model predicts the minimum achievable pulse width and pulse delay for this device structure. Experimental results and calculated values indicate that very high modulation rates are possible with vertical-cavity surface-emitting lasers.

Important design considerations for high-speed GaInAsP 1.3-μm surface-emitting lasers are described. Modified rate equations for surface-emitting lasers and a small-signal analysis are used to calculate the frequency response versus the mirror reflectivity. For the structure analyzed, the results predict optimum reflectivities of 98% for high-frequency operation and maximum quantum efficiency. Strong gain-saturation effects are predicted owing to the high photon densities that occur in these devices. As the mirror thickness increases, the frequency response of the laser reduces greatly. The effect of heating is found to be important and is included in the analysis.

We describe the first results of hybrid mode locking combining both active and passive mode locking of a semiconductor laser. These functions are integrated into a monolithic device with a 1.3 μm GaInAsP gain region, an active waveguide, and a saturable absorber. The devices have low threshold currents, and exhibit hysteresis in their light/current characteristics. The long integrated waveguides allow mode locking at a repetition rate of 15 GHz without the need for an external cavity. Pulse widths as short as 1.4 ps have been demonstrated using the combined effects of active and passive mode locking.

Some published results on mode locked monolithic lasers are reviewed. Optical spectra of a typical hybrid mode-locked monolithic device for different levels of RF power and frequency are shown and discussed, as are effects of detuning the modulation frequency. Design considerations are examined. These include the length of the device, the use of hybrid mode locking, and the choice of active or passive waveguides.<>

A new mode-locking technique, self-mode-locking, is described which uses the detected optical pulses from the mode-locked laser as the active driving source. This technique forms narrow-width mode-locked optical pulses at low repetition rates without the use of a microwave synthesizer.

Measurements of actively mode-locked semiconductor lasers are described and compared to calculations of the mode-locking process using three coupled traveling-wave rate equations for the electron and photon densities. The dependence of pulse width on the modulation current and frequency are described. A limitation to minimum achievable pulse widths in mode-locked semiconductor lasers is shown to b...

1.3 mu m GaInAsP lasers were integrated with active waveguides and satirable absorbers to form arrays of monolithic devices. These devices have low threshold currents, and long integrated waveguides have led to picosecond pulse widths at a 15-GHz repetition rate. The integrated active waveguides decrease the effective coupling losses from laser to integrated cavity, and the saturable absorber prov...

Harmonic and intermodulation distortion are two crucial performance parameters of laser diodes in determining their utility in wide-bandwidth analog links. The multimode rate equations are solved numerically to give information on the bandwidth phase, harmonic, and intermodulation distortion of GaInAsP laser diodes at both small- and large-signal levels. The results of the theoretical analysis agr...

The authors describe a new limit on the achievable pulsewidth from actively mode-locked semiconductor lasers which is due to dynamic detuning. Dynamic detuning sets a higher limit on pulsewidth than the effects of finite gain bandwidth and dispersion, agreeing with experimental results which show pulsewidths much longer than expected if dynamic detuning is neglected. The dynamic detuning mechanism...

The transient behaviour of the output of a twin stripe laser is found to be dependent upon the rise time and levels of the injection currents. Under certain biasing conditions the transient output has a high initial value which relaxes to a much lower value if the pulse rise time is fast but not if it is slower. This effect is found to be associated with beam steering.