Open Access
Add to BibSonomyAbstract
In this paper we report a method to overcome the limitations of gain-saturation and two-photon absorption faced by developers of high power single mode InP-based lasers and semiconductor optical amplifiers (SOA) including those based on wide-waveguide or slab-coupled optical waveguide laser (SCOWL) technology. The method is based on Y-coupling design of the laser cavity. The reduction in gain-saturation and two-photon absorption in the merged beam laser structures (MBL) are obtained by reducing the intensity of electromagnetic field in the laser cavity. Standard ridge-waveguide lasers and MBLs were fabricated, tested and compared. Despite a slightly higher threshold current, the reduced gain-saturation in MBLs results in higher output power. The MBLs also produced a single spatial mode, as well as a strongly dominating single spectral mode which is the inherent feature of MBL-type cavity.
©2013 Optical Society of America
1. Introduction
A lot of effort has been devoted to increasing the power output of semiconductor lasers. Some of the device structures used are evanescently coupled arrays [1], leaky-wave coupled arrays [2], wide aperture [3], waveguide tapering [4], wide-waveguide structure [5], and slab-coupled optical waveguide [6].
Only wide-waveguide lasers and slab-coupled optical waveguide lasers produce output beam quality required for simple, reliable and low loss coupling to a single mode optical fiber. Wide-waveguide lasers also produce significantly higher optical power output [5]. By increasing the thickness of the waveguiding layers (transverse direction) the optical mode penetration into the adjacent highly doped regions is significantly reduced. Free carrier absorption in highly doped layers is one of the main contributors to optical losses in semiconductor lasers [5,7]. Hence, reducing the interaction of the optical field with doped regions leads to reduced optical loss in the cavity and as a result higher net gain. In this case the limiting factor for the thickness of the waveguiding layer is the onset of higher order modes.
The best results in output power have been obtained by slab-coupled optical waveguide lasers (SCOWLs) [6]. The SCOWL design is based on Marcatili’s coupled-mode analysis [8]. It shows that higher order modes in a large multimode waveguide can be coupled into the slab leaving only the fundamental mode in the waveguide. This approach allows the fabrication of even larger cavities than those of wide-waveguide lasers and yet maintains fundamental mode operation. Extension of the laser cavity in the transverse direction results in significantly reduced optical losses due to reduced mode interaction with the highly doped regions of the device. At the same time, increased laser cavity width results in higher modal gain as a result of the increased effective volume of the optical mode.
Nevertheless these types of lasers also have their limitations. The two main mechanisms limiting the output power of these InP-based lasers are gain-saturation and two-photon absorption. Both of these processes have the same origin and are proportional to the photon concentration in the laser cavity. Therefore the approach for gain-saturation reduction presented in this paper will also be effective in reducing two-photon absorption.
Two-photon absorption is a nonlinear effect, which depends on square of the optical field intensity. It can be expressed as:
where β is the two-photon absorption coefficient, z is the optical field propagation direction, P is the optical field intensity, P0 is the initial intensity (at z = 0). Two-photon absorption becomes significant only at very high optical field intensity. On the other hand gain-saturation affects all gain media regardless of optical field intensity.The gain of a laser medium taking into account the gain-saturation process, g, can be expressed as:
where g is the gain coefficient, go is the gain coefficient in absence of an optical field, P is the optical field intensity, Psat is the saturation intensity (saturation intensity is the optical field intensity at which gain is reduced by a factor of two). Since the coefficients go and Psat are constants for a given device structure, hence the only option to reduce the gain-saturation for a given laser structure is to reduce the optical field intensity P in the active region. Our MBL design accomplishes this by doubling the cross-sectional area of the optical mode while maintaining single mode operation, unlike broad aperture lasers. Figure 1 shows simulated gain curves as a function of optical intensity (Eq. (2)) for a standard ridge waveguide laser and a laser with twice the cross-sectional area of the mode, such as the MBL. In both cases the gain drops monotonically. However, gain-saturation is noticeably reduced in case of the laser with twice the mode area.
Fig. 1 Solid line is an example of gain-saturation (g/g0) as a function of electromagnetic field intensity (P) of a standard ridge waveguide laser. Dashed line is an example of gain-saturation of a laser with doubled cross-section area of the mode.
In this study we design and fabricate MBLs and compared their performance with standard ridge waveguide lasers. MBLs produced more than 30% higher maximum output power as compared to the ridge waveguide lasers while maintaining the single spatial mode regime.
2. Structure and fabrication
Both MBLs and conventional ridge waveguide lasers were fabricated on a single wide-waveguide laser structure. The sample was MOCVD grown on (001) InP n+ substrate. An n+ (Si: 2x1018 cm−3) InP buffer was first grown, followed by 1 µm of undoped In0.86Ga0.14As0.30P0.70 waveguiding region, which also contained the active region in the middle. The active region consisted of five In0.60Ga0.40As compressively strained quantum wells separated by In0.79Ga0.21As0.45P0.55 barriers. After the waveguiding layer a 2 µm InP p-cladding (Zn: 2x1018 cm−3) layer was grown. The structure was terminated with a thin 50 nm In0.53Ga0.47As (Zn: 1x1019 cm−3) p+ contact layer.
Standard contact optical lithography was used to define the laser cavity. Ridges were formed by RIE etching with CH4-H2 chemistry using a previously established polymer free single step etching recipe [9] suitable for “soft mask” etching, followed by PECVD deposition of 200 nm of SiNx at 100 °C. After lift-off to expose the top of the ridge and solvent cleaning to remove photoresist residue, the top contact (Ti/Pt/Au) was deposited using an e-beam evaporator. After that the substrate was polished down to thickness of about 200 µm and cleaned with solvents, the bottom contact (Ge/Ni/Au) was deposited by e-beam evaporator. To form good ohmic contacts the samples were annealed in RTA for 1 min at 450 °C under Ar ambient. Subsequently the samples were cleaved into 3 mm long devices and tested, with facets “as cleaved” (without any facet coating). The testing was done at a constant temperature of 10 °C. To avoid device heating the lasers were tested under pulsed condition (duty cycle 1%) and pulse duration of 100 ns. The light output was collected from a single facet and measured with integrating power meter.
3. Design and numerical modelling
For numerical modelling we used commercially available software package BeamPROP from RSOFT. BeamPROP is a general numerical modelling package for computing of electromagnetic field propagation in arbitrary waveguides. This package is based on finite difference beam propagation method (BPM), which is the most widely used technique for modeling field propagation in integrated and fiber-optic photonic devices [10]. We used BeamPROP to simulate one directional propagation of the optical field through the MBL waveguide structure, since one of the limitations of BPM is its inability to deal with reflections, which makes it impossible to simulate laser resonators.
Figure 2 illustrates the numerical modelling results of optical field propagation in the MBL cavity. Optical mode launched at point (0,0) splits evenly at the first Y-junction, propagates through the branches and then merges together at the other Y-junction, resulting in mode intensity at the end point (0,3000) of near unity. Dark blue ripples outside the waveguide represent the electromagnetic field losses. From numerical modelling it is obvious that the majority of the losses occur at both Y-junctions.
Fig. 2 Color-coded result of BeamPROP (RSoft) numerical modelling of the electromagnetic field propagation in MBL. The mode launched at the beginning (0,0) of the waveguide evenly splits at the first Y-junction and then merges at another Y-junction. At the end (0,3000) of the waveguide the mode nearly reaches unity.
Obviously it is very important in this type of device to minimize the optical loss at the Y-junction. A priori limit for additional optical losses in the structure was set to less than 5%, which for our cavity translates into a loss equivalent to about 0.1 cm−1. Several parameters of the MBL structure are required to be optimized to achieve this value, such as the radius of the Y-junction (or equivalently the distance between two waveguides), the etch depth of the ridge and the refractive index of the SiNx insulating layer. To determine the optimum values of these parameters, three sets of interdependent numerical modelling were performed. The aim of the numerical modelling was the determination of the optical field transmission from one facet of the MBL to the other facet. The results are illustrated in Fig. 3 -5, respectively.
Fig. 3 Dependence of electromagnetic field transmission through MBL-type waveguide as a function of the distance between the branches (BeamPROP numerical modelling).
From Fig. 3 transmission of the electromagnetic field through the MBL waveguide gradually decreases with distance between the branches up to about 20 µm. Beyond this value there is a much faster drop. For the distance between the branches of 50 µm, only ~77% of the light is transmitted. The oscillating behavior of the curve observed between values 30 µm – 50 µm are due to reduced confinement and more pronounced interference from the leaking field. Hence as long as the distance between the waveguides is maintained below 20 µm (Y-junction radius longer than 25 mm) the optical propagation losses will be less than 5%.
Similarly numerical modelling was done for the dependence of field transmission through the structure as function of the insulating material refractive index. For the numerical modelling of the effect of insulating layer refractive index, we used the refractive index values of SiNx ranging from 1.6 to 2.1, values that are regularly obtained in our PECVD system by changing the deposition conditions. The results in Fig. 4 show that as long as the refractive index of the SiNx layer is above 1.7 transmission of 95% or more can be obtained. Below this value, there is a sudden drop in the transmission across the MBL waveguide.
Fig. 4 Dependence of electromagnetic field transmission through the MBL as a function of the SiNx insulating layer refractive index. (BeamPROP numerical modelling).
The precise control of etch depth is even more important as illustrated in Fig. 5 . There is a small widow around the etch depth of 1.9 µm in which 95% optical transmission is achievable. Too shallow etch (less than 1.86 µm) results in reduced lateral confinement of the mode and increased mode leakage from the waveguide at the Y-junctions, and as a result increased transmission losses. On another hand, etching too deep into the waveguide layer (more than 2.0 µm) results in additional losses from extensive mode interaction with the etched surface.
Fig. 5 Dependence of electromagnetic field transmission through the MBL as a function of ridge etching depth. (BeamPROP numerical modelling).Vertical bar represents the cladding and the waveguide layer interface.
Obviously high transmission losses at the Y-junctions would negate the benefits of reduced gain-saturation and should therefore be avoided.
From numerical modelling the following optimized parameters were determined: laser length 3 mm, laser ridge width 3 µm, the distance between the waveguides 15 µm (Y-junction radius of 67 mm), refractive index of 1.88 for the SiNx insulating layer and the etch depth of 1.9 µm (0.1 µm above the waveguide layer).
Using these optimized parameters, very low additional transmission losses in the MBLs can be obtained, as shown in Fig. 2. The optical field introduced into the waveguide at point (0,0), splits evenly, passes through the waveguides, recombines and emerges at the exit point (0,3000) with losses of less than 5%.
4. Results and discussion
The light versus current (L-I) curves of both type of devices are shown in Fig. 6 . Both curves do not have any “kinks”, implying that they operate in a single mode up to injection current of 3000 mA.
Fig. 6 Light – current (L-I) curves of MBL (solid line) and standard ridge waveguide laser (dash-dot line). The output light was collected from a single facet. The devices’ facets are “as-cleaved”.
The L-I curve of the MBL is nearly perfectly straight line in contrast to the standard ridge waveguide laser which has a noticeable rollover. Since the testing of both devices was carried out under the same conditions, on a temperature controlled stage and at very short current pulses of 100 ns with 1% duty cycle, the effect of thermal heating is negligible. Hence theL-I curve rollover can only be the result of the stronger material gain-saturation.
Additional losses introduced by the Y-junctions lead to a slight increase of the threshold current value in the MBL. However the reduction of gain-saturation in MBL results in a maximum power output of 228.5 mW at 3000 mA (limited by the current source), which is 30% higher than the maximum power output of 172.6 mW from conventional ridge waveguide laser.
As was shown in Fig. 1 doubling of the cross-sectional area of the mode results in significant reduction of gain-saturation. This reduction of gain-saturation in MBL can be clearly seen in the slope efficiency as a function of output power as shown in Fig. 7 . The slope efficiency of a laser is the slope of L-I curve and represents the rate of output power increase with injected current. Hence higher gain results in higher slope efficiency. In Fig. 7, the slope efficiency of the standard ridge waveguide laser is initially higher, however it drops quickly with the increase of the output power, while that of the MBL is almost constant over the entire range of measurement. The sharp drop of both curves at high output power is due to instabilities of the current source at the end of the range (consistent for all tested devices).
Fig. 7 Slope efficiency vs. output power for the MBL (solid line) and standard ridge waveguide laser (dashed line). The sharp drop at the end of the curves is due to instabilities of the current source and happened for all devices.
Since the objective of this study is to produce an increased power output from a single mode laser, a near-field and a far-field study of the MBL laser were carried out to determine the mode profile. For a near-field study a confocal microscopy was employed to increase the optical resolution for this wavelength. Only a single lobe is visible in the near-field image of the MBL shown in Fig. 8 .
For the far-field study a scanning slit method was used. Beam divergence for the slow axes of 16.9° was derived for MBL, whereas a value of 16.7° was obtained for the standard single mode ridge waveguide laser. Single lobe near-field, together with beam divergence value close to the value for the standard single mode ridge waveguide laser and a “kink”-free L-I curve are obvious indicators of a single spatial mode operation of the MBL.
As for the spectral characteristics, laser resonator formed by cleaving is a form of Fabry-Perot interferometer, with a set of Fabry-Perot modes supported by the geometry of the resonator. Overlap of Fabry-Perot modes with material gain spectrum determines the set of spectral modes supported by a given laser. Usually semiconductor lasers produce multimode spectrum, unless special measures are undertaken like distributed feedback (DFB), distributed Bragg reflector (DBR) or coupled cavity resonators.
A strong domination of one spectral mode is the inherited characteristic of MBL-type cavity due to coupling of two waveguide branches inside a Fabry-Perot optical resonator. In an MBL device the spectral modes are the result of the convolution of two sets of Fabry-Perot modes from two waveguides that form the MBL cavity. Due to a small difference in the path length of each waveguide (due to fabrication imperfection), only a few overlapping modes exist. This results in a dominating mode in the vicinity of gain curve maximum. Shown in Fig. 9 are the spectral modes of the MBL at 1000 mA injection current (CW operation). It is worth mentioning, that no optimization has been performed to increase spectral mode selectivity of these MBLs and both branches were designed to be of the same geometry, while for the single spectral mode operation with sufficient side mode suppression the branches should be of significantly different optical lengths. The mode spacing of the MBL is 0.12 nm. It has been shown previously that by injecting current separately into different parts of this type of device spectral mode control can be achieved [11].
5. Conclusion
We have presented merged beam laser (MBL) and compared it with the standard ridge waveguide laser. This type of laser cavity is capable of increasing the laser output power by reducing the optical field intensity in the cavity, hence reducing the gain saturation of the active medium. The comparative study of an MBL and a standard ridge waveguide laser resulted in 30% higher maximum output power produced by the MBL. Since the mechanisms of gain-saturation and two-photon absorption are similar and strongly depend on optical field intensity in the laser cavity, this type of laser cavity can also be used to reduce two-photon absorption.
Our study clearly shows that MBL is operating in a single spatial mode regime. Additionally due to its unique cavity a single spectral mode is clearly dominating in the spectrum of the MBL.
It is very important in this laser cavity design to reduce the additional losses introduced by the Y-junctions. By carefully designing and optimizing parameters like radius of the Y-junctions, insulating material refractive index and etch depth the additional losses introduced by the Y-junctions were limited to less than 5% or equivalently to less than 0.1 cm−1 for our MBL cavity.
This design can be extended to lasers with higher number of branches. Our numerical modelling of light propagation through a four branch MBL indicate that transmission losses of less than 5% can be achieved (not shown here). This opens up the way for a further increase of output power.
Acknowledgments
The author would like to acknowledge the help of Dr. Aliaksandr E. Minovich with confocal microscopy measurement of devices’ near-field. The Australian Research Council is acknowledged for its financial support. This work has been made possible through the access to the ACT Node of the Australian National Fabrication Facility.
References and links
1. D. F. Welch, B. Chan, W. Streifer, and D. R. Scifres, “High-power, 8 W cw, single-quantum-well laser diode array,” Electron. Lett. 24(2), 113–115 (1988). [CrossRef]
2. D. Botez, L. J. Mawst, and G. Peterson, “Resonant leaky-wave coupling in linear arrays of antiguides,” Electron. Lett. 24(21), 1328–1330 (1988). [CrossRef]
3. A. Al-Muhanna, L. J. Mawst, D. Botez, D. Z. Garbuzov, R. U. Martinelli, and J. C. Connolly, “High-power (>10 W) continuous-wave operation from 100-μm-aperture 0.97-μm-emitting Al-free diode lasers,” Appl. Phys. Lett. 73(9), 1182–1184 (1998). [CrossRef]
4. B. Sumpf, H. Wenzel, and G. Erbert, “High-power, high-brightness semiconductor tapered diode lasers for the red and near infrared spectral range,” Proc. of SPIE 7616, 76161L (2010).
5. D. Garbuzov, L. Xu, S. R. Forrest, R. Menna, R. Martinelli, and J. C. Connolly, “1.5 µm wavelength, SCH-MQW InGaAsP/InP broadened-waveguide laser diodes with low internal loss and high output power,” Electron. Lett. 32(18), 1717–1719 (1996). [CrossRef]
6. J. J. Plant, P. W. Juodawlkis, R. K. Huang, J. P. Donnelly, L. J. Missaggia, and K. G. Ray, “1.5-µm InGaAsP-InP slab-coupled optical waveguide lasers,” IEEE Photon. Technol. Lett. 17(4), 735–737 (2005). [CrossRef]
7. S. Adachi, “Free-carrier effects on optical properties,” in Physical Properties of III–V Semiconductor Compounds: InP, InAs, GaAs, GaP, InGaAs, and InGaAsP, (John Wiley & Sons, USA, 1992).
8. E. A. J. Marcatili, “Slab-coupled waveguides,” http://www.alcatel-lucent.com/bstj/vol53-1974/articles/bstj53-4-645.pdf
9. M. Lysevych, H. H. Tan, F. Karouta, and C. Jagadish, “Single-step RIE fabrication process of low loss InP waveguide using CH4 / H2 chemistry,” J. Electrochem. Soc. 158(3), H281–H284 (2011). [CrossRef]
10. “BeamPROP user guide,” RSOFT Design Group, Inc.
11. M. Schilling, W. Idler, D. Baums, G. Laube, K. Wunstel, and O. Hildebrand, “Multifunctional photonic switching operation of 1500 nm Y-coupled cavity laser (YCCL) with 28 nm tuning capability,” IEEE Photon. Technol. Lett. 3(12), 1054–1057 (1991). [CrossRef]
