Fast phenomena in semiconductor lasers

Although diode lasers are almost ideal sources for ultrahigh-speed data communication systems, system performance remains critically dependent on the quality of the optical pulses that they generate. Uniquely among lasers, the output power can be modulated directly by modulating the diode current. However, this leads to relaxation oscillations and a roll-off at high frequencies that is superimposed on the frequency response of the drive circuit. This is limited by parasitics and maximum modulation frequencies range from 1 to 70 GHz, depending on the type of laser and its packaging. The higher values are obtained with short cavity lengths and tight optical confinement, the highest frequencies being achieved in vertical-cavity devices. Modulation bandwidth is usually limited by circuit parasitics, device heating and the maximum power-handling capability of the laser facets. The generation of picosecond and femtosecond pulses demands special techniques and three-gain switching, Q-switching and mode locking-are discussed in detail, with their relative advantages and disadvantages compared. Very short pulses inherently contain a significant spread of wavelengths and their generation requires the optical gain in the laser medium to extend across that wavelength range. While diode lasers satisfy this criterion better than many other types, the effect of gain non-linearities and carrier-transport effects prevent the Fourier-transform limit from being achieved in practice. As a result, external pulse-compression techniques, which exploit the detailed temporal and spectral properties of the laser pulses, such as frequency chirping, self-phase modulation and group velocity dispersion, are becoming more important, and diode lasers are increasingly challenged as primary sources by compact, efficient, diode-pumped solid-state lasers. The paper summarizes the levels of maximum pulse power and minimum duration that have been achieved using the various techniques.