T his thesis has described an experim ental and th e o re tica l in ve stig a tio n o f active m ode-locking o f sem iconductor lasers. The project was undertaken in c o lla b o ra tio n w ith N a tio n a l P hysical L a b o ra to ry who defined the objective o f the experim ental w ork as the generation o f lOps optical pulses. T h is objective has been m et and was accom plished u sin g an a c tiv e ly mode-locked, external g ra tin g sem iconductor laser in conjunction w ith a sem iconductor optical a m p lifie r. In addition, we have studied short pulse g e n e ra tio n u sin g com posite c a v ity InG aA sP -LiN bO g lasers. In these stru ctu re s, we have in ve stig a te d m ode-locking by c u rre n t m o du la tio n , A M m ode-locking, and fo r the fir s t tim e in th is type o f co n fig u ra tio n FM m ode-locking. We have com plem ented our e xperim ental in ve stig a tio n s w ith a th e o re tica l study o f active m ode-locking o f sem iconductor lasers usin g a frequency-dom ain fo rm u la tio n . We have perform ed num erous n u m e ric a l s im u la tio n s and dem onstrated m o de-locking by c u rre n t m odulation, AM , FM m ode-locking and FM -laser operation o f m o n o lith ic external ca vity semiconductor lasers. To the best o f our knowledge, th is is th e f ir s t u n ifie d d e s c rip tio n o f a ctive m o de -co u p lin g effe cts in sem iconductor lasers.
To review our experim ental w ork, we s ta rt w ith the fa b rica tio n o f lith iu m niobate m odulators. The waveguides were form ed by tita n iu m in d iffu s io n in to lith iu m niobate substrates. In order to prevent the o u td iffu s io n o f
L i2 0 we investigated both w et and d ry diffusions. The la tte r m ethod was
used fo r the waveguide fa b rica tio n o f o u r m odulators. We fa b rica te d a phase m odulator by depositing a tra ve llin g-w a ve electrode s tru c tu re on the waveguide substrate. For the in te n s ity m odulator, a Y -ju n c tio n was u tilis e d w hich acts as a M ach-Zehnder in te rfe ro m e te r w ith in a composite ca vity. The electrodes o f the in te n s ity m o dulator were fa b rica te d as a
tra ve llin g -w a ve push -p u ll configuration. B oth m odulators were designed as w aveguide devices u tilis in g o p tic a l TE-m ode p ro p a g a tio n . A n ti- re fle c tiv e coatings and m irro rs were deposited on a ll th e w aveguide samples and m odulators used in our experim ents.
W ith regard to our com posite-cavity InG aAsP-LiN bO g la se r experim ents, we have investigated several methods o f optical coupling between the laser and lith iu m niobate w aveguide. The m ost e ffic ie n t e x te rn a l c a v ity operation was achieved using a graded-index lens as a coupling elem ent. We also investigated both d ire ct coupling and the use o f spherical lenses. In these tw o cases we failed to produce proper external ca vity operation.
U s in g th re e co m p o site -ca vity In G a A sP -L iN b O g c o n fig u ra tio n s we dem onstrated active m ode-locking by c u rre n t m o du la tio n , A M and FM m ode-locking. Due to the low in tra c a v ity optical coupling efficiency and h ig h p a ra s itic reflections we obtained re la tiv e ly broad pulses. We could not com plete a more detailed experim ental in v e s tig a tio n o f active mode- locking o f InG aAsP-LiN bO g lasers because o f the low optical power levels. O ur experim ental in ve stig a tio n o f InG aAsP-LiN bO g lasers culm inated in the fo llo w in g results:
1) InG aAsP-LiN bO g laser, cu rre n t m odulation 35ps
2) InG aAsP -LiN bO g laser, A M m ode-locking 50ps
3) InG aAsP-LiN bO g laser, FM mode-locking 45ps
The re sults described above were obtained under consistent experim ental conditions and using tw o ve ry s im ila r B T R L b u rie d h e te ro s tru c tu re lasers. These m easurem ents were made u sin g a fa s t photodiode / sam pling oscilloscope com bination.
W ith in the same m easurem ent system we in ve stig a te d a b u lk e xte rn a l c a v ity laser and measured 27ps pulses. The b u lk e xte rn a l c a v ity in th is experim ent consisted o f a G R IN lens and a plane m irro r. The best mode- locked pulses w hich we measured a t U C L were 16ps. These 16ps pulses
were obtained from a m odified b u lk external ca vity con fig u ra tio n using an im proved m easurem ent system w hich incorporated a lensed fib re and an o p tic a l is o la to r. The m o dified c a v ity consisted o f a sp h e rica l lens, uncoated glass étalon and a plane m irro r. U n fo rtu n a te ly , we were only able to measure the spectrum o f broad mode-locked pulses (~40ps) due to th e lo w finesse o f the scanning F a b ry-P e ro t re so n a to r used in th is experim ent.The tim e-bandw idth product o f the 40ps pulses was estim ated to be 0.6.
The experim ental w ork undertaken a t N F L included the te stin g o f several e xte rn a l g ra tin g configurations. We experim ented w ith several g ratin g s and lenses in order to fin d the optim al com bination fo r m ode-locking. In a d d itio n we experim ented w ith two types o f electrical drive. S u rp ris in g ly , we found little difference between the perform ance o f the m ode-locked systems when d rive n e ith e r w ith a step-recovery diode or a sinusoidal drive. O ur fin a l configuration, w hich is now used fo r c a lib ra tio n purposes
a t N F L , in c o rp o ra te s a m ode-locked se m ico n d u cto r la s e r and
sem iconductor optical a m p lifie r. For th is system we have m easured 7.5ps pulses (fo r sech^ pulse shape) a fte r the optical a m p lifie r and obtained 15.4ps sam pling oscilloscope traces (Fig. 6.10).
W ith regard to the th e o re tica l w ork, we have developed a frequency- dom ain m odel o f active m ode-locking o f sem iconductor lasers. Several frequency-dom ain fo rm u la tio n s o f active m ode-locking o f sem iconductor lasers by cu rre n t m odulation have been reported in the lite ra tu re . To the best o f o u r knowledge, our a tte m p t is the fir s t to p rovide a u n ifie d d e s c rip tio n o f m ode-locking by c u rre n t m o d u la tio n , A M , F M mode- lo ckin g and FM -laser operation o f sem iconductor lasers. O u r fo rm u la tio n is based on a set o f m ulti-m ode fie ld equations, used in conjunction w ith th e ra te equation fo r the c a rrie r num ber. We have also u tilis e d the frequency-dom ain approach to show th a t long tra ve llin g-w a ve m odulators can s till provide s u ffic ie n t in tra c a v ity m odulation in A M and F M mode- locking when th e ir lum ped equivalents become in e ffic ie n t.
P a ra s itic in tra c a v ity re fle ctio n s are neglected in o u r m odel. These re fle ctio n s can p la y an im p o rta n t role in m ode-locking, and in practice th e y lim it the m inim um pulse w idths in b u lk external ca vity lasers. Since the fo rm u la tio n th a t we use neglects the sp a tia l v a ria tio n o f the c a rrie r concentration, the v a lid ity o f our model is probably re s tric te d to pulses longer th a n 5ps, as in the case o f o rd in a ry averaged ra te equations. The m a in advantage o f the frequency-dom ain approach is th a t i t provides u n ifie d description o f m ode-coupling effects caused by an active, fo rcing elem ent.
A com plete set o f program s dealing w ith the dynam ics o f sem iconductor lasers has been produced. These program s are w ritte n in F o rtra n 77 and th e y use NAG routines and G INO graphics. In the developm ent o f these program s, we s ta rte d w ith a single-m ode ra te d e s c rip tio n o f g a in - sw itch in g (both s o lita ry and external ca vity lasers) and Q -sw itching. The n e x t step was a m u lti-m o d e fo rm u la tio n o f these effects. O u r fir s t num erical model fo r m ode-locking by cu rre n t m odulation included a fla t gain and loss p ro file and assumed IG H z m odulation. We have compared our sim ulations w ith Reference 2 and have found excellent agreem ent fo r b o th ten-m ode and fifty-m o d e cases. We im p ro ve d upon th e m odel
presented in Reference 2 in tw o ways. F irs t, we m o dified the tim in g
b e tw een th e p h o to n and c a rrie r n u m b e r v a ria tio n s b y p ro p e r synchronisation o f the forcing te rm due to c u rre n t m o d u la tio n and the e q u iva le n t coupling coefficient. Second, we in trod u ce d a para b o lic loss w h ich re su lte d in the generation o f re a lis tic pulse shapes and spectra. R egarding the second im provem ent, i t was necessary to in clu d e an a d d itio n a l phase s h ift as an effect o f the lin e w id th enhancem ent factor.
A n in v e s tig a tio n o f active m ode-locking o f m o n o lith ic sem iconductor lasers a t 10 GHz was perform ed using tw e n ty five modes and a parabolic loss d is trib u tio n . The tim e e v o lu tio n o f m o de -lo ckin g by c u rre n t m o d u la tio n was studied using our standard in itia l co nditions, nam ely zero photon num bers and random phases. The effect o f d e tu n in g o f the m o du la tin g frequency and v a ria tio n o f the bias and RF c u rre n t have been
exam ined. F in a lly , the effect o f the lin e w id th enhancem ent fa cto r was investigated. We have also studied FM m ode-locking o f m o n o lith ic lasers. Two types o f behaviour were observed, depending on the in clu sio n o f the lin e w id th enhancem ent factor. For pç=0 we obtained double p u lsin g - a ch a ra cte ristic w hich we had obtained in our experim ent w ith composite c a v ity InG aA sP -LiN bO g lasers. A single sequence o f m ode-locked pulses was generated from our model fo r p^=5 . We have also dem onstrated FM - laser operation o f a m o no lith ic sem iconductor laser. As in the case o f FM m ode-locking, we observed tw o types o f behaviour. F o r P^=0 we obtained sym m etric FM spectra, whereas w ith pg=5 a d isto rte d F M spectra and enhanced p u ls a tio n were obtained. We also te ste d o u r th e o re tic a l fo rm u la tio n o f A M m ode-locking and dem onstrated m ode-locked pulse
sequences fo r both p^=5 and Pc=0.
We have achieved good consistency between or n u m e rica l sim u la tio ns o f m ode-locking by c u rre n t m o du la tio n , A M , FM m ode-locking and FM la se r operation. V a lid a tio n o f o u r n u m e rica l s im u la tio n s has proved d iffic u lt as very little has been reported on experim ental m easurem ents and th e o ry o f active m ode-locking o f m o n o lith ic sem iconductor lasers. M ost o f o u r com parisons are made w ith the p u b lish e d e xp e rim e n ta l investigations o f b u lk external cavity lasers.
O ur num erical results fo r active m ode-locking by c u rre n t m odulation are in good agreem ent w ith the experim ental traces produced by a m o no lith ic
A T T & B e ll l a s e r ^ 56 U n fo rtu n a te ly, there are no reported experim ental
in ve stig a tio n s o f frequency detuning or the effect o f v a ria tio n s o f the bias and RF cu rre n t to compare w ith our data.
D e tu n in g o f the m odulating frequency in our model re su lts in irre g u la r, b u t cyclic pulse sequences. The jit t e r and a m plitud e v a ria tio n s are n o t v e ry pronounced fo r sm a ll d e tu n in g . H ow ever, fo r la rg e d e tu n in g (>150M Hz), the pulses s ta rt to break-up. In the case o f p ositive detuning, the new pulse evolves from the back edge o f the old pulse. Conversely, fo r negative detuning, the new pulse develops from the tra ilin g edge o f the