Optical bistability in diode-laser amplifiers and injection-locked laser diodes

Optical bistability (OB) in semiconductor lasers andin resonant-type laser amplifiers has attracted muchattention recently because of its potential applicationin optical computing and optical communicationnetworks. When a semiconductor laser, biased justbelow threshold, is operated as an optical amplifier,dispersive OB has been demonstrated.'1-4 On theother hand, OB has also been found recently in opticallyinjection-locked semiconductor lasers.5 ' Thesephenomena have been theoretically explained eitherwhen a laser amplifier operates below threshold 3'4or when an injection-locked laser works abovethreshold.5',7 Quite different physical images havehitherto been considered for these two cases. Sincethere is no discontinuity at the threshold of asemiconductor laser, the OB should be continuousfrom below to above threshold. However, a unifiedtreatment has not yet been found. In this Letter wepresent our experimental measurement on the OBproperties of a distributed-feedback semiconductorlaser biased from below to above threshold.The experimental setup is described as follows.Two identical distributed-feedback laser diodes withan emission wavelength of 1554 nm are used. Thefirst laser (master), biased at three times its threshold,is used to generate the signal light, and its outputis injected into the second laser (slave) that worksas an OB element; both lasers have one facet antireflectioncoated and the other facet cleaved. Eachlaser used is isolated from external reflection with adouble-section Faraday optical isolator that providesmore than 70 dB of isolation. A monochromator anda scanning Fabry-Perot interferometer are used forrough and fine measurement of the optical spectrum.Frequency matching and detuning between the twolasers is accomplished by adjusting the heat-sinktemperature of the master laser. The injected opticalpower into the slave laser can be evaluated asfollows: first, measure the linewidth-enhancementfactor a of the slave laser by using the injectionlockingmethod8 ; with this ae value known, the injectedoptical power can then be obtained throughthe measured injection-locking bandwidth when theslave laser is biased above threshold.9 At each biaslevel of the slave laser and with a definite opticalpower injected from the master, a bistable outputfrom the slave laser can be obtained versus the signalfrequency detuning. As an example, the squares inFig. 1 give the measured bistable loop width versusthe relative bias level of the slave laser from belowto above threshold while the injected optical power iskept at approximately -23 dBm. Below threshold,the OB loop increases its width with the increaseof the bias level, while at above threshold, the loopwidth decreases with the bias level. The maximumOB loop width is obtained, in our case, at -1.03 timesthe free-running laser threshold, and this value is, ingeneral, dependent on the amount of injected opticalpower.In order to simulate the OB operation, a unifiedtreatment is obviously necessary, which allows oneto consider the laser biased from below to abovethreshold. Our theoretical analysis is based on therate-equation model of Lang

Fig. 1. Measured bistable loop width versus the normalizedinjection current with the optical injection levelat approximately -23 dBm (squares) together with thecalculated results with an injected optical power Pi. of-25 dBm (short-dashed curve), -23 dBm (solid curve),and -21 dBm (long-dashed curve). Ith is the free-runinglaser threshold.

Fig. 1. Measured bistable loop width versus the normalizedinjection current with the optical injection levelat approximately -23 dBm (squares) together with thecalculated results with an injected optical power Pi. of-25 dBm (short-dashed curve), -23 dBm (solid curve),and -21 dBm (long-dashed curve). Ith is the free-runinglaser threshold.