Butt-coupled laser diode: wavelength tuning and optical noise.

Transcription

1 Butt-coupled laser diode: wavelength tuning and optical noise. Yakov Sidorin and Dennis Howe Optical Data Storage Center, Optical Sciences Center University ofarizona, Tucson, AZ ABSTRACT We consider a high power laser diode with significant external feedback from the entrance facet of a bun-coupled fiber. The efffective reflectance offabry-perot etalon, formed by the laser and fiber facets, undergoes deep modulation due to the change in the laser-to-fiber separation. This results in the variation of nonlinear lasmg characteristics and the optical properties of the oscillalion cavity, which leads to a significant wavelength tuning (up to 15 mu). The developed model succesfully predicts expaimental results. The longitudinal mode specirum, tolerances and relative intensity noise are characterized as well. Keywords: wavelength tuning, laser diodes, extremely short external cavity, optical feedback, buttcoupling, relative intensity noise. 1. NTRODUCTON n recent publication1 we presented a theoretical model that accurately predicts the observed wavelength tuning that results from changes in the butt-coupling spacmgz that separates the output facet of a high output power Fabiy-Perot laser diode (SDL 5400 series) and theuncoated entrance facet of a singlemode fiber which is butt-coupled to the diode. We consider the etalon of length z formed by these two facets to be an exiremely short external cavity which can be tuned (by varying the spacingz) to produce a very broad range of feedback into the active laser diode cavity. The model takes into consideration the variation in active medium carrier density versus zand the resulting influence of this parameter on the aclive medium's gain properties. n this paper, we give a brief description of the model and also discuss the stability and repeatability of wavelength tuning realized by control of the butt-coupling spacing z. Discussion ofhow optical noise (characterized by relative intensity noise, or RN) depends on zis presented as well. 2. EXPERMENTAL SETUP We studied butt-coupling of a A1GaAS Fabry-Perot high power laser diode (HPLD) that has low output facet reflectance of -4% (SDL 5400C/5410C; absorption rate '2 cni1, 750 pmcavity length, nominal wavelength 845 nm, rear facet reflectance.-95%). Optical feedback was provided by a '4 m long single-mode silica fiber with an uncoated cleaved front facet that was precisely positioned with respect to the HPLD ( see Fig. 1). Reflections from the back end of the fiber were eliminated by angle-cleaving. The HPLD output and fiber entrance facets comprise a short external etalon (several pmin length)2. Angular misalignment was allowed to within 10 mrad. The diode drive current and temperature were kept constant during the experiment. The end facet of the fiber was imaged onto the receiver of an optical spectrum analyzer (HP70950A) to observe SPE Vol X/97/$

2 LASER FBER LENS BEAM OPTCAL SPECTRUM DODE SPLTTER ANALYZER z:; POSTONER zz:o:r::t ;..._ * VOLTMETER DETECTOR RF SPECTRUM ANALYZER Fig.1. Experimental setup for measurement ofspectral properties and thtsity noise of a HPLD in a bun-coupling configuration. The fiber positioncan be varied in 20 mn increments. the variation of BPLD spectrum with a resolution of 0.08 am. The variations of the output power were monitored with a photodiode. RN measurements were earned out via an RF-spectrum analyzer (HP8592B). 3. THE MODEL An extremely short external cavity diode laser and its equivalent are schemalically shown in Fig. 2: here rk is the field reflectance of a kth boundary, d is the solitary HPLD cavity length and z is the external cavity length. The effective field reflectance of the externaletalon formed by the surfaces with refiectances r andr characterizes the output coupling ofthe extremely short ECLD and can be written as2 1i r2co p. Teff (v,z) = r2 + 2 C (z,0,5)[rr exp( ik2z)] (1) where z is the length of external cavity, k =2icv c is the propagation constant for light in the external cavity and CLL (z,, 5) is the field overlap integral between the laser mode that appears (internal to the laser active medium) at the laser output facet and the one that has made p-round trips inside the external etalon. The effeclive power reflectance Reff(Z,) = rcff(z,2 varies periodically versus both k and z within4 to 10% (see Fig. 3), although it can get as high as 16% when z <1 jim. The solitaiy HPLD has a 25 ma cw threshold and usually operates in a multi-longitudinal mode regime. Axial mode selection inour system with feedback is determined by the cavity loss and medium gain modulation that is a function of the external etalon length z.onecan find the required gain condition for the ECLD realized via butt-coupling to be g(z,)=a 1n[rreff(z,)]/d (2) 134

3 laser external mirror Fig.2: a) External cavity laser diode, realized via butt-coupling to a fiber having input facet amplitude reflectance r3. a and 0 are transverse and angular misalignments between fiber and laser diode optical axes. ri compound laser d re(a,z) b) Equivalent scheme; r is the effective reflectance ofthe external etalon oflength z. -.. Q,.'' ' : S,.. () ) S SS.S.. S S :. i.. Fig. 3 : Spectral disiribulion ofthe external etalon effective reflectance versus external etalon length z. where am the HPLD cavity internal loss. We conirol the center frequencies of the etalon output coupler's reflection bands according to Eq.(l) by varying the length zofthe external etalon. Minima of radiative loss (defined as the second term in Eq.(2)) correspond to the external cavity reflection peaks; these shift in frequency as zis varied (see Fig.4). Prospective laser oscillation occurs at the optical frequencies where the gain is equal to its threshold value g( ) which is given by Eq.(2). f, due 135

4 Wavelength (nni) Fig.4: Change in spectral distribution ofthe radiative loss ofcompound laser diode due to change in external cavity length z. 40 E U U 0 U Separation, z (pm) Fig.5: Threshold current modulation of a butt-coupled HPLD. to some mechanism, the HPLD threshold cunent we changed, a new threshold gain level. together with a new oscillation frequency, would result through the influence of the canier density corresponding to the new threshold current value. The dependence ofthe threshold cunent on external feedback is known to be3 thr Cfr [ m(1 + hi[(rlreffi"2 1] (3) where the constant C depends on the structure and composition of the gain medium. Obviously, the threshold ciment (and therefore the carner density in the HPLD active region and threshold gain) varies as 136

5 we change z through its dependence on R(z). The calculated modulation of threshold current versus zin our experimental system is presented in Fig. 5. The gain spectral distribution for the SDL 5400C/5410C HPLD active medium was estimated using certain assumptions about the laser structure which are based on guidance provided by the manufacturer. The traditional functional form ofthe gain specirum was used4 where E21 g(e21 ) = umax (E21 )(f2 f1) (4) is the energy of the iransition at frequency V21, g,(e1) is the maximum gain and(f2 fi ) the Fermi factor that depaids on the injection level We assumed a bulk A1Ga1As active layer (thickness O.1pm, x 0.0 1), parabolic shaped transition bands and used tabulated optical characteristics for GaAs and A1GaAs5'6. The presence of light-holes, lineshape broadening and polarization dependence was neglected. Carriers were ireated as an idealfermi gas in 3D4'7. Calculation shows that the maximum of the threshold gain spectral distribution shifts in response to the change in threshold current that occurred our experiment by as much as lsnm. This shift, combined with the periodic variation of the compound laser radiative loss, leads to the observed mode selection and wavelength tuning. 4.EXPERMENTAL RESULTS AND DSCUSSON 4.1 Wavelength tuning. Fig. 6 (a,b) presents typical experimentally observed and calculated wavelength tuning curves for two different starting conditions (initial separations of z045 pm and 5.tm respectively; external cavity length increment of 60 nm). The spectral structure ofthe light output that occurs at different points of the tuning curves is also shown for several values ofz z +öz, where öz is the scan increment on the abscissa offig. 6. For relatively large separations z > 15 im the position ofthe external etalon's reflectance maxima and shape of the lasing medium's gain specirum collaborate to cause single-mode operation, provided that the local maximum of the external etalon's reflectance versus frequency is sufficiently narrow. At shorter separations the feedback conditions support several laser cavity modes simultaneously and multi-mode operation occurs (compare Figs. 6a and 6b). The number oflongitudinal modes lased simultaneously in this case grows as z decreases. t can be seen from Fig. 6a that the shallow slopes of the wavelength tuning curve are formed by the tuning of a single mode. The steep vertical"transinon regions", or "dips", that occur in both the multi-mode and single-mode tuning regimes correspond to simultaneous multi-mode lasing ofmodes at both ends of the tuning range spectrum (Fig. 6a -pointe; and Fig. 6b - pointj). n Fig. 6(a,b) the spectra at points a andfare analogous to those at points d andj, respectively. The error bars shown in Fig.6 correspond to the standard deviation of measured wavelengths obtained over the course of multiple experiments. t should be pointed out that in the regions of z that correspond to multi-mode operation (z < 10.tm) this tuned wavelength uncertainty is higher than for single mode operation. The explanation stems from the fact that the change in external cavity length & necessary to switch the dominantly lasing mode of the laser cavity to a neighboring one dependsnonlinearly on the absolute length zofthe external cavity (see below). 137

6 4.2 Tolerances. Timing arises in the form of controlled shifts, or well-behaved (unidirectional) jumps, between neighboring 1as cavity modes. When working in a well-defined single-mode tuning reme (z> 35 jim), we resolve almost every resonant mode ofthe laser cavity with satisfactory stability and repeatability. The typically observed "stair-case" tuning behavior is shown in Fig. 7, which contains tuning data obtained from three consecutive scans performed at an initial separation ofz0 40 p.m using a scan step increment of 60 nm (wavelengths measured during the first of the three scans in Fig. 7, which extends from pointa to point b are plotted as "-" symbols; wavelengths measured during the second scan, which extends from point b to point C, are denoted by the "0" symbols) a) 845 U Change n External CaVity Length,.z (jim) b) Change n External Cavity Length, iz (jim) Wavelength,. (nm) Fig.6: Wavelength tuning curves(experiment - solid, modeling - dashed),andcorrespoding spectra, scan increment 6O am., evely 3rd point shown. nitial separation between the HPLD and the fiber: a)---lspin; b)-5 pin. 138

7 'a'to'b' 'b' to 'c' 846 'c'to'b' : \ : eisa * i.: U.. --la Change n External Cavity Length, Az (j.txn) b 0.6 Fig.7: ; -rio 000 z -120 Two consecutive experimental scans of a single period ofthe tuning curve, performed at 4O1.tm separation with a 60 mu scan increment :i Frequency, MHz a '*! --- A: '?- H- -;AJZ (m+1)j2 ' Separation, z (a.u.) E z C) Z0.7 std Frequency (MHz) Frequency (Mhz) 10 Fig.8: Relative intensity noise (a) solitary HPLD; (b,c,d) butt-coupled HPLD averaged over 10 scans. 139

8 We previously determined that a simple coupled-cavity model of a butt-coupled, high output power diode laser configuration which does not include the variation in the active medium carrier density versus z and the resulting influence of this parameter on the active medium's gain properties does not adequately describe the extremely short external cavity laser. f only optical (coupled-cavity) effects are considered, a cursory estimate will show that, for our butt-coupling configuration, an external cavity length change of öz _ 1.5 nm is necessary to produce a jump between neighboring laser diode modes. However, a more rigorous calculation - and this is the strength of our model - reveals that the magnitude of öz required to cause switching between oscillating modes on the "shallow slopes" ofthe tuning curve depends nonlinearly on the value ofz0.atinitial separations ofthe order ofz0 5 microns such a switch occurs foröz -3 nm, it increases to & - 15 mu at z 25 microns and atz0 100 microns öz is as high as 45 nm (we would like to emphasize that the scan resolution in our experiments is 20 mn); Other observations include: - exireme sensitivity of the tuned wavelength to öz is exhibited in the vicinity of the "vertical iransition regions" ofthe tuning curve for any value of z0; - thepeak-to-peak range oftuned wavelengths decreases non-linearly versus z0. Since thelasing behavior with an exiremely short external cavity is multi-mode (cf., Fig 6b), we might be interested in working at a somewhat larger external cavity separation where the tuned laser spectrum is purely single-mode (cf., Fig. 6a). t is quite possible that the combination of a sufficiently large, single longitudinal mode tuning range and reasonably low sensitivity to external cavity length perturbations öz may exist to make this method ofdiode laser wavelength tuning useful for certain applications. 4.2 Relative intensity noise. Finally, the optical noise (RN) characteristics for the solitary and bun-coupled HPLDs are presented in Fig. 8. As shown in Fig. 8b (for noise measurements at 1 MHz) significant increases in the value of Rll'T can be observed at external cavity separations zthat correspond to the "vertical iransition regions" of the tuning curves (e.g., at the points e andj in Figs. 6a and 6b, respectively). On average, the single-mode regime is less noisy than the multi-mode case, except at tuning curve "transition regions" where two welldefined modes are present simultaneousiy at opposite ends of the tuning range. Additional insight can be obtained from the measured statistical characteristics ofrjn that are presented in Figs. 8(c,d). 5. CONCLUSON n our butt-coupling experiments we demonstrated that by working with external cavity lengths on the order ofthe Rayleigh range ofthe laserbeam it is possible to tune the working wavelength ofthe HPLD by about 15 nm using only the variation of the external cavity length, while maintaining constant driving current and temperature. We consider the two reflectors that form the extremely short external cavity (i.e., the BPLD output facet and coupled fiber entrance facet) to comprise anetalon that acts as a tunable output coupler ofthe bufl-coupled HPLD. The fact that the two mirrors that form the extemaletalon have similar reflectances and that they are closely spaced causes the net reflectance oftheetalon to be highly modulated as the spacing varies. This large depth of modulation ofetalon net reflectance is the key to the achievement of a large tuning range. An analytical model that includes the effect of multiple reflections in the external etalon, each attenuated by a modal overlap integral, and which is based on phenomenological principles, shows good agreement with the experimental results. Our experiments show that in there exists a trade-off among the tuning range that can be achieved, coupling efficiency and the spectral purity of light. That tradeoff may be an important issue for integrated optics application. 140

9 ACKNOWLEDGMENTS This work was supported by Optical Data Storage Center, Optical Sciences Center, University of Arizona. REFERENCES 1. Y. Sidorin and D. Howe, Laser diode wavelength tuning due to butt-coupling into an opticalfiber, accepted for publicaton, Opt. Lett., Y. Sidorin and D. Howe, Diode laser- to-fiber butt-coupling: extremely short external cavity, accepted for publication, Appl. Opt., G.P. Agrawal and N.K. Dulta, Semiconductor Lasers, Chapter 2,2nd Ed. (VanNostrand Reinhold), New York, C. Kittel, ntroduction to Solid-State Physics, (John Wiley & Sons), New York, Handbook of Optical Constants of Solids, (Academic Press), Handbook of Optical Constants of Solids, (Academic Press), H. Hang and S.W. Koch, Quantum theoiy of the optical and electronic properties of semiconductors, Chapter 6 (World Scientific),