Directly modulated lasers at 1310 nm are useful as low cost short haul high speed transmitters. High speed requires large differential gain and high damping. Low cost dictates that the device can operate without a thermoelectric cooler (TEC) to temperatures of 85 C, preferably without introducing exotic designs or material features. A multiple quantum well (MQW) ridge waveguide (RWG) gain coupled (GC) distributed feedback (DFB) InGaAsP laser grown by metal-organic chemical vapor deposition (MOCVD) satisfies the above requirements.
Directly modulated lasers (DMLs) have two high performance applications: 1310 nm 10 Gb/s uncooled and 1550 2.5 Gb/s extended reach. Two key elements are gain coupled gratings and buried heterostructures. Gain coupled gratings simultaneously increase the DML's intrinsic relaxation oscillation frequency and damping, while the buried heterostructure reduces thermal chirp and parasitic capacitance. Large relaxation oscillation frequencies and reduced parasitic capacitance allow 85 °C operation; large damping and reduced thermal chirp enable extended reach.
Subcarrier multiplexed fiber-optic systems using direct modulation of semiconductor lasers have attracted much attention for analog and digital broadband services. Since analog modulation is adopted, the system performance can be degraded seriously due to inherent nonlinearities of semiconductor lasers. In this study, the injection-locking technique is applied to reduce the nonlinearities in directly modulated semiconductor lasers. In particular, the characteristics of the second harmonic distortion (SHD) and the third harmonic distortion (THD) are experimentally investigated. First of all, at a modulation frequency of 5 GHz, both SHD and THD in the injection-locked semiconductor laser are observed to reduce significantly and equally for a large range of modulation power, which are 15dB- and 23 dB-decreased, respectively. Second, at a fixed modulation power of 6 dBm, the reduction of SHD and THD is, however, found to vary with modulation frequency. The reduction of SHD is more substantial at around one-half of the relaxation resonance frequency of the free-running laser. The decrease in THD is more significant at around one-half and one-third of the relaxation resonance frequency of the free-running laser. Finally, how the reduction of harmonic distortions varies with the operational parameters of the laser system is also investigated. The results demonstrate the feasibility of the injection-locking method in reducing harmonic distortions for high-speed and high-power analog modulation applications.
A transmission-line laser model has been used for simulating distributed-feedback (DFB) lasers. Statistical distributions of laser parameters like threshold current, slope efficiency, front-to-back power ratio, or side-mode-suppression ratio (SMSR) are generated by varying randomly lasers’ facet phases. Model parameters were adjusted by comparing simulated and experimental distributions for a continuous wave (CW) index-coupled and a 2.5 Gbit/s gain-coupled directly-modulated (DM) DFB lasers.
For the index-coupled DFB laser, agreement with experimental data is excellent except for the front-to-back power ratio, which has a larger spread than measured experimentally. For the gain-coupled DFB laser, distributions are in excellent agreement with experimental data, but the SMSR is calculated to have a median about 6 dB larger than measurement.
Distributions of small-signal parameters and dispersion penalties after propagating in an optical fibre are also generated for various drive conditions and design parameters. It is shown that a grating with an index coupling larger than 4.0 and a gain coupling of around 0.05 gives the highest 2 dB dispersion penalty yields for a reach of 450 km. There is nevertheless a compromise between high dispersion penalty yields and CW single-mode yields when using large index coupling coefficients with only a small amount of gain coupling.
There is currently interest in using novel modulation formats for high bit-rate datacoms systems. 4-level modulation is an attractive method of halving the line-rate required for 40Gb/s systems. This 20GBaud line rate enables reduced bandwidth direct modulation of semiconductor lasers, thus reducing laser chirp, increasing transmission distances and also enabling simplified drive electronics to be used. In this experiment the 4-level signal is generated by electrically combining 2 de-correlated 20Gb/s data streams of differing amplitude from a pattern generator and then used to modulate a DFB laser. The directly modulated source is a DFB laser, emitting at 1310nm with a 3dB frequency response of 20GHz. This laser also has a very linear modulation response, with a spurious free dynamic range of over 100dBHz<sup>2/3</sup> at 25°C and over 90 dBHz<sup>2/3</sup> at 85°C. This highly linear behaviour is necessary to allow direct 4-level modulation source even at high temperature. The 40Gb/s 4-level signal is then transmitted along standard fibre and detected with an electrical receiver. In order to overcome the attenuation limited transmission distance of 20km a semiconductor optical amplifier, with a saturation power of 11dBm and fibre to fibre gain of 20dB, is used. The addition of an SOA enables transmission distances of 40km to be achieved with transmission penalties of as low as 2.6dB, even with the laser operating at 70°C. The robustness of the 4-level modulation is compared to NRZ and the impairments to both signals upon optical amplification are examined.
The characteristics of period-one oscillations in semiconductor
lasers subject to optical injection is experimentally and
quantitatively investigated. The changes in the frequency
separation and in the magnitude difference between the principal
oscillation and the sideband of the injected laser are studied
as a function of experimentally accessible parameters, the detuning frequency and the injection strength of the injection signal. The frequency separation decreases as the injection strength and the detuning frequency decrease. The magnitude of the principal
oscillation decreases with the decreasing injection strength and the
increasing detuning frequency, while that of the sideband grows
at the same time. At some operating conditions, these characteristics
leads to a situation that the magnitude of the sideband becomes larger than that of the original principal oscillation, resulting in a frequency shift of the principal oscillation from the injection frequency to the sideband.
Universal self-organisation on surfaces of semiconductors upon deposition of a few non-lattice-matched monolayers using MOCVD or MBE lead to the formation of quantum dots. Their electronic and optical properties are closer to those of atoms than of solids.
We have demonstrated for QD-lasers a record low transparency current density of 6A/cm<sup>2</sup> per dot layer at 1.16 μm, high-power of 12W, an internal quantum efficiency of 98%, and an internal loss below 1.5 cm-1. Relaxation oscillations indicate the potential for cut-off frequencies larger than 10 GHz.
GaAs-based QD-lasers emitting at 1.3 μm exhibit output power of 5 W and single transverse mode operation up to 300 mW. At 1.5 μm again an output power of 5 W has been obtained for first devices showing a transparency current of 700 A/cm<sup>2</sup>.
Single mode lasers at 1.16 and 1.3 μm show no beam filamentation, reduced M<sup>2</sup>, sensitivity to optical feedback by 30 db and α-parameter as compared to quantum well lasers.
Passive mode locking of 1.3 μm lasers up to 20 GHz is obtained.
Thus GaAs-lasers can now replace InP-based ones at least in the range up to 1.3 µm, probably up to 1.55 μm.
We report the analysis and application of uncooled, directly-modulated high-speed DFB lasers with emphasis on their analogue transmission performance. Fibre-optic links employing such lasers are shown to meet the most stringent requirements of analogue systems at both high carrier frequencies and high temperatures. Spurious-free dynamic ranges (SFDR) exceeding 100dB×Hz<sup>2/3</sup> and 90dB×Hz<sup>2/3</sup> and input third-order intercept points (IIP3) above 20dBm and 18dBm are reported for carrier frequencies up to 20GHz at 25°C and up to 10GHz at 85°C, respectively. The error-vector magnitude (EVM) for a 256-QAM modulated signal transmitted over 15km of SMF remains below 1.9% for carrier frequencies of both 2GHz and 5GHz for all measured temperatures. The link performance is assessed by using 3GPP W-CDMA, IEEE 802.11a and IEEE 802.11b signals. In all cases the EVM remains within the standard specification, for fibre-optic link lengths of up to 10km and laser operating temperatures of up to 70°C. Finally, an IEEE 802.11b WLAN demonstrator is presented, allowing antenna remoting over up to 1000m of 62.5/125μm MMF.
Directly modulated lasers (DMLs) have two high performance applications: 1310 nm 10 Gb/s uncooled and 1550 2.5 Gs/s extended reach. Two key elements are gain coupled gratings and buried heterostructures. Gain coupled gratings simultaneously increase the DML's intrinsic relaxation oscillation frequency and damping, while the buried heterostructure reduces thermal chirp and parasitic capacitance. Large relaxation oscillation frequencies and reduced parasitic capacitance allow 85°C operation; large damping and reduced thermal chirp enable extended reach.
We analyze the high-temperature continuous-wave performance of 1.3 micron AlGaInAs/InP laser diodes grown by digital alloy molecular beam epitaxy. Commercial laser software is utilized that self-consistently combines quantum well bandstructure and gain calculations with two-dimensional simulations of carrier transport, wave guiding, and heat flow. Excellent agreement between simulation and measurements is obtained by careful adjustment of material parameters in the model. Joule heating is shown to be the main heat source; quantum well recombination heat is almost compensated for by Thomson cooling. Auger recombination is the main carrier loss mechanism at lower injection current. Vertical electron escape into the p-doped InP cladding dominates at higher current and it causes the thermal power roll-off. Self-heating and optical gain reduction are the triggering mechanisms behind the leakage escalation.
The material physics of digitally grown InAlGaAs quaternary alloy systems are investigated using Molecular Beam Epitaxy (MBE) grown layers. With MBE, arbitrary epitaxial alloy compositions can be achieved, without changing the group III elemental constituents flux rates, by simple sequential shuttering of the relevant fluxes. Monolayer fluctuations create inhomogeneities that lead to a broadening of the photoluminescence (PL) spectra. Multiple PL peaks are also seen in select alloy compositions.
Experimental measurements of threshold current density as a function of temperature have been analyzed in terms of the characteristic temperature, T<SUB>0</SUB>, and temperature gradient (Delta) <SUB>T</SUB>J<SUB>th</SUB> equals (delta) J<SUB>th</SUB>/(delta) T, for a number of semiconductor laser device structures. These include AlInGaAs/InP, InGaAsP/InP, and AlGaAs/GaAs. A theoretical model is used to investigate the possible loss mechanisms in laser diodes that cause the superlinear increase of threshold current with temperature. The characteristic temperature T0 is found to vary with temperature and device length, thus making it somewhat misleading when quoted without qualification. A different approach based on plotting ln((Delta) <SUB>T</SUB>J<SUB>th</SUB>) vs. ln(J<SUB>th</SUB>) shows a linear relationship that is dependent on device structure only, allowing the use of a new figure of merit for the temperature performance of semiconductor lasers.
Issues relating to modeling the full spatiotemporal dynamics of wide aperture semiconductor lasers and amplifiers are discussed. Included in the discussion are the limitations of the usual beam propagation approach, characteristics of the many-body light- semiconductor material interaction, spurious nonphysical instabilities which mimic numerical grid oscillations and novel subpicosecond pulse reshaping and compression effects. An explicit simulation is presented for a flared amplifier structure and the results are compared with those using a linear gain model.