Open Access
Add to BibSonomy
Abstract
We present a study of the propagation of dark line defects (DLDs) in catastrophically damaged 808 nm laser diodes, based on cathodoluminescence (CL) measurements and laser mode propagation simulations. Room temperature CL images show blurred DLDs running parallel to the laser cavity. Remarkably, low temperature images reveal their true morphology: the blurred lines are resolved as parallel narrow discontinuous DLDs. This morphology does not match the usually reported molten front scenario of DLD propagation. Low temperature images show that DLDs consist of a sequence of catastrophic optical damage (COD) events separated a few micrometers from each other. Consequently, a different propagation scheme is proposed. The points where the CODs occur suffer a temperature increase and these hot spots play a capital role in the propagation of the DLDs. Their influence on the beam distribution is modelled using finite element methods. The calculations evidence changes on the intensity distribution of the laser that qualitatively reproduce the DLD shapes. Additionally, the COD events result in the generation of defects in the region that surrounds them. The successive CODs in the discontinuous DLDs are rationalized in terms of the enhanced laser absorption in these sensitized regions where the laser beam is concentrated by thermal lensing.
© 2022 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
1. Introduction
Reliability is a critical issue of laser diodes, especially as increasing powers are demanded for many applications. Unfortunately, high-power laser diodes suffer from catastrophic optical damage (COD) [1–3]. This failure mode occurs suddenly after many hours of regular operation. The causes and the mechanisms driving this phenomenon are still far from being understood [1–3]. Typically, the COD is described as a thermal runaway process, in which the laser light energy is transferred by optical absorption to a hot spot in the QW, which is the active absorption medium. As a consequence of the local temperature increase, the bandgap locally shrinks, with the concomitant increase of laser light absorption localized in that hot region, further increasing its temperature in a positive feedback loop, leading to melting, as classically described in the seminal paper by Henry et al [4].
Usually, failure analysis is based on the study of the defects resulting from the degradation. Different strategies aiming to provoke the COD in a controlled manner have been considered, e.g., step tests, in which high-intensity current pulses of different durations are applied to induce the COD [5]. These tests provide information about the genesis of the degradation, but they hide some unknown aspects of the process. The failure is caused under extreme operation conditions, producing current crowding [6–8], or laser-induced breakdown [9,10]. Usually, these tests run with fresh defect-free lasers, while the COD under normal operating conditions occurs in aged devices for which the defect population is progressively enhanced with the aging time. This leads to the reduction of the power threshold to COD in aged lasers, well documented in the literature [11,12]. Therefore, we will refer to this degradation as long-term COD to distinguish it from the COD forced under extreme power conditions.
The root causes of both degradations, forced and long term, cannot be a priori assimilated, even if the final results might be comparable in terms of the defects generated, and its propagation (as in both cases, the degradation is optically driven), one cannot affirm that the initial phase of the degradation is the same in the two cases. The fingerprint of the degradation consists of dark line defects (DLDs), which are elongated defects aligned along the laser cavity following the optical field instead of specific crystallographic directions [1–3]. The DLDs present dark contrast in cathodoluminescence (CL), and/or electroluminescence (EL) images. One should distinguish the DLDs associated with COD from the DLDs related to the propagation of dislocations, which spread at a much smaller velocity by either climb or glide mechanisms [13].
The COD commonly starts in a weak point of the front facet of the device. Still, it can also begin inside the cavity when the facet mirror technology is improved to make robust mirrors, or if local current crowding triggers the degradation. Subsequently, it propagates by gaining the energy necessary for its motion from laser self-absorption. At this point, we have to divide the problem into two different stages. On one side, the mechanism launching the process, i.e., the root cause of the degradation. On the other side, the mechanism responsible for the propagation of the DLDs, causing the final failure of the device.
As stated above, the COD is initially triggered at a weak point of the active zone, which produces a tiny hot region, susceptible of absorbing the laser radiation. Most frequently, it begins at the front facet of AlGaAs/GaAs lasers [14,15]. The mirror heating is due to non-radiative recombination at surface states formed at the interface between the mirror coating layers and the (110) facet of the semiconductor laser structure. Previous literature about thermal runaway shows that both temperature and stress play a paramount role in the COD process. In fact, the most probable scenario of degradation can be drawn as a local heating of the active zone and the subsequent breaking of atom bonds due to significant local thermal stresses. This will result in the formation of dislocations, which behave as optical absorbers that generate e-h pairs, which subsequently recombine non radiatively increasing the local temperature. However, it has been claimed that the temperature increase is insufficient if just the classical runaway feedback loop is considered [8,16]. In order to reach the high temperatures involved in the COD process, a drastic decrease of the local thermal conductivity must occur. This could be achieved by the sudden formation of a dense array of dislocations, with the density scaling above 107 cm-2 [17]. Alternatively, Adams et al [8] proposed current crowding at the hot spots as responsible for the sudden additional temperature increase.
The second stage of the problem concerns the propagation of the DLDs along the laser cavity. We present herein a study of the DLD propagation in AlGaAs/GaAs lasers centered on the analysis of the morphology of DLDs by temperature-dependent CL measurements. A model for the propagation of DLDs, based on the thermal lensing of the laser light by the local hot spots generated inside the laser cavity, is also proposed to further support the results.
2. Experimental and computational model
The devices referenced here are graded-index separate confinement heterostructure quantum well (GRINSCH QW) AlGaAs/GaAs laser bars emitting at 808 nm with output powers of a few tens of watts. Typically, these bars consist of 25 emitters (200 µm wide) separated from each other by optically and electrically isolated channels, with a period of 400 µm between emitters. Every emitter is divided into 20 injection channels separated by current blocking dielectric stripes, 10 µm pitch, Fig. 1. These bars are soldered p-side down, typically using a CuW heat sink with AuSn solder. The QW is AlGaAs (10% Al), with graded AlGaAs (26-65% Al) waveguide layers, AlGaAs (65% Al) cladding layers, and GaAs contact layer (Fig. 2). The lasers studied here have undergone a burn-in aging test at 30 °C and continuous optical output power of 30 W. The current vs. aging time for a given emitter under these conditions exhibits a sharp current increase characteristic of the COD after 100 hours of operation (Fig. 3). After the aging step, the individual emitters were inspected, and some of them were found to be catastrophically damaged. One of those catastrophically degraded emitters is the object of this study.
Fig. 1. Top view SEM image of the laser diode after removing the metal layer, where one can see the injection and blocking channels.
Fig. 2. Laser structure: (a) compositions and thicknesses of the layers; (b) simplified scheme depicting the reference system used for the modelling.
CL measurements were carried out with a Mono CL2 (Gatan) system attached to a field-emission scanning electron microscope (LEO1530 Zeiss). A photomultiplier (PMT) was utilized for the panchromatic CL (pan-CL) images, while a CCD camera detected the spectrally resolved CL and hyperspectral CL images. The CL images were recorded for temperatures varying between 80 K and room temperature. We will show herein the importance of the low-temperature panchromatic CL images for revealing the true morphology of the DLDs in the studied laser. CL measurements were done on the front facet and in top view, allowing the entire laser cavity to be explored. The lasers were etched in a diluted HF solution (5%), which under-etched the insulating dielectric layers between the metal and the semiconductor, permitting the lift-off of the top metal layer. This treatment opens the laser cavity to the e-beam thus allowing CL observation of the full cavity, revealing the DLDs path inside the cavity.
Finite element methods (FEM) were used for simulating the electromagnetic field inside the laser cavity in the presence of hot spots, which are critical for the DLD propagation. 2D FEM simulations were carried out on COMSOL Multiphysics using the Electromagnetic Waves in Frequency Domain (EWFD) module. This module solves Maxwell equations for monochromatic electromagnetic (EM) radiation traveling through a certain medium. The present model considers an 808 nm monochromatic plane wave traveling through an Al0.1Ga0.9As QW. Local heat sources in the QW simulate ongoing COD events. A Gaussian temperature distribution is adopted for the hot spots,
The local hot spot was shown to induce laser mode filamentation and thermal lensing [18]. This effect was modelled by considering the thermally induced changes in the refractive index around the hot spot, which modify the optical field distribution inside the cavity. The optical absorption threshold redshifts because of the bandgap shrinkage as the temperature is increased [19]. The temperature dependence of the bandgap is calculated through Varshni’s equation:[20]
In the case of Al0.1Ga0.9As, Eg(0)= 1.638 eV is the bandgap at 0 K, α = 5.405·10−4 eV/K and β = 204 K [21,22].
The modeled system consists of an 8 µm x 42 µm portion of Al0.1Ga0.9As resembling the active region (QW) of one of the injection channels of a single emitter surrounded by the corresponding blocking channels. This is a sufficiently large area to observe the effect of the hot spots on the optical field distribution and small enough to keep a reasonable computational cost. Further details on the simulations can be found elsewhere [18].
3. Results
3.1 Front facet CL
CL images were recorded on the front facet of the degraded emitter. Figure 4(a) shows a pan-CL image of the mirror facet. The confinement of carriers permits observing the CL emission from the QW, which appears as a bright line with a series of dark zones corresponding to degraded regions where the emission is totally or partially quenched. The most damaged points are those associated with the so-called V defects, which are typically observed in degraded centimeter laser bars [23]. The V defects are formed at the QW, and the two branches of the V propagate across the n-barrier, the cladding layer, and the substrate, forming an angle of 57° with the epitaxial plane, which corresponds to the intersection of the dislocation glide (111) planes and the (110) front facet plane. The V defects are formed by the thermal stresses generated by the sharp temperature gradients produced by the local heating at the front facet [24]. The V defect apex is located at the QW, evidencing that the hot spot lies within it. This is due to the low thermal conductivity of the QW, which concentrates most of the absorbed energy inside itself [24,25]. Typically, the COD of this type of laser is triggered at the front facet, where high temperatures can be reached due to non-radiative recombination at surface states.
Fig. 4. (a) Panchromatic CL image of the front facet showing the damage in the QW, and the V defects. (b) SEM image of the front facet
The SEM image of the front facet does not show external damage (Fig. 4(b)). The presence of blisters at the front facet of degraded lasers, which are sometimes observed, is a proof of melting of the active layers [26,27]. The blisters are formed by the extrusion of molten material close to the front facet. Its absence indicates that melting was not reached, as we will show later on.
The V-branches have been related to the relaxation of the stress induced by bar soldering [23,28]. However, they are only observed in degraded lasers, suggesting that, in order to arise, thermal stress induced by local heating of the QW needs to be added to the packaging stress. The V-branches can penetrate deeply into the substrate, pointing to a huge stress field associated with the sharp temperature gradients generated at the front facet. One can argue that the root cause of the degradation are the defects formed at the front facet, which behave as non-radiative recombination centers with the concomitant local temperature increase. The thermal energy storage in tiny regions of the QW in the front facet, consequence of the low thermal conductivity of QWs [29] and the thermal boundary resistance at the interfaces, would be responsible for the massive thermal stress gradients [16,30–32]. The role of local hot spots in the COD has been modeled in previous articles [16,32,33]. Here, we are mainly concerned with the mechanism of defect propagation responsible for the formation of DLDs. Thus, one has to observe what happened inside the laser cavity during the degradation.
3.2 Top-view CL
The CL images were recorded after removing the top metal layer. In Fig. 5, the room temperature pan-CL image of the cavity shows two long DLDs and other shorter DLDs. All these DLDs start in the front mirror zones where the most severe damage was observed in the CL images of Fig. 4, i.e., those corresponding to the points of the QW seeding the V defects. However, other front facet QW defects without associated V-branches did not develop DLDs along the cavity. The temperature reached in those regions is probably not high enough to induce stresses that could generate V defects and DLDs propagating along the laser cavity.
Fig. 5. Panchromatic CL image (top view) revealing the DLDs along the laser cavity (300 K). The scratch was produced in the metal lift-off process
The long DLDs run along the cavity and nearly reach the rear facet. These DLDs appear as two parallel continuous dark lines extending along two neighbor gain channels. This supports the role of the laser light as the driving energy for the defect propagation.
CL images were recorded at low temperatures to picture the DLDs with a higher spatial resolution. The spatial resolution of the CL images depends on the interaction volume of the e-beam with the solid, which is related to the e-beam energy and the density of the target. This is the so-called generation volume where the e-h pairs are generated. Besides, one must consider carrier diffusion, as the recombination can occur outside the generation volume. This contribution would also be collected because of the large numerical aperture of the parabolic mirror of the CL setup [34]. The diffusion of carriers can effectively enlarge the recombination volume from which the CL signal arises [35]. Therefore, the spatial resolution of the CL images is determined not only by the generation volume but also by either the minority carrier, or the ambipolar, diffusion length (Ld), which follows the relation:
where µ is the carrier mobility and τ the mean carrier lifetime.Interestingly, one observes differences in the DLD morphology between the CL images recorded at different temperatures. At room temperature (Figs. 5 and 6(c)), two fuzzy continuous dark lines are observed along the laser cavity. These lines have a lateral dimension of approximately 7 µm; therefore, they exceed the injection channel width. However, at 80 K finer features in the defect structure are revealed, Fig. 6.a. The DLDs appear as a series of much thinner dash-like lines consisting of dark contrasted irregular traces aligned along the laser cavity. As the temperature of the CL images increases (Fig. 6.b), the dark traces are progressively blurred with respect to the lowest temperature image. The merging of the lines would be related to the rise of the carrier diffusion length with temperature. This effect depends on the zone where the DLDs are observed: close to the front facet (right side of the image), the dark lines are thoroughly blurred, while they are still seen as discontinuous dark lines closer to the rear facet (left side).
Fig. 6. Panchromatic CL images of the DLDs at 80 K (a), 140 K (b) and 300 K (c), showing the change from a discontinuous dashed DLD at low temperature to a continuous DLD at high temperature.
The progressive increase of the recombination volume can be more clearly grasped from the series of images displayed in Fig. 7 for a larger set of temperatures. Note that the fact that Ld increases with T suggests that τ also increases with temperature. This would be due to the presence of trap levels, which reduce the diffusion length at low temperature; however, as the temperature increases, the carriers would be thermally liberated and recombine non radiatively at the extended defects produced by the COD, thereby enlarging the recombination atmosphere around them. The presence of trapping levels renders the measurement of the minority carrier diffusion length difficult; however, this effect is clearly observed in the expansion of the recombination atmosphere around the DLDs (Figs. 6 and 7) which modifies the apparent morphology of the DLDs when varying the temperature of the CL images.
Fig. 7. Pan-CL images of COD defects revealing the evolution of the recombination atmosphere when increasing the temperature of the CL measurements
The parallel discontinuous DLDs can be rationalized in terms of the lateral laser mode distribution due to the presence of hot spots, which rearrange the internal optical modes and produce the filamentation of the laser beam [13,36]. The dark lines run parallel to the cavity, and, as observed in Fig. 6.a, some of the DLDs disappear when approaching the rear facet. At low temperatures, only one line of the tracked set remains close to the end of the resonant cavity. The fading of the DLDs when approaching the rear facet results from the progressive optical power loss as they extend along the cavity [37].
Interestingly, the dark lines are also observed along the non-gain stripes (channels coated with the dielectric blocking layer). This fact supports the role of the optical field in the propagation of the DLDs, as there is no charge injection below the dielectric stripes. The filamentation induced by the hot spots can also locally enhance the optical power density inside the blocking channels, where those DLDs are generated. Note the absence of branching in the DLDs, which is usually observed in degraded broad-area emitters [38,39]. In fact, the laser structure, with 5 µm wide injection channels, does not support internally circulating modes [13]. This would explain the different DLD morphology in these lasers with respect to non-channeled broad area emitters, which present a dissimilar mode branching induced by circulating modes [13,39]. The morphology of the DLDs provides valuable information about the mechanism that drives their propagation. The discontinuous lines rule out the hypothesis of a melting front spreading along the laser cavity, as reported for InGaAsP/InP lasers, and which has been generally assumed to be the fingerprint of the thermal runaway [40,41]. This is a noteworthy result, as it points to an alternative scenario.
Hyperspectral CL images were recorded in selected zones of the laser cavity. Figures 8 and 9 display monochromatic images of the emission from the QW, waveguide, and contact layers, together with CL spectra characteristic of zones with different levels of damage.
Fig. 8. CL spectra and monochromatic images (T = 80 K) extracted from the hyperspectral CL image of a region close to the front facet, where the COD damage is high. Left panel: spectra of points with different degree of damage. Right panel, top to bottom: monochromatic images of the barrier layer, QW and contact layer. The points marked P1, P2 and P3 correspond to the spectra plotted in the left panel.
Fig. 9. CL spectra and monochromatic images (T = 80 K) extracted from the hyperspectral CL image of a region close to the rear facet, where the COD damage is lower than that in Fig. 8. Left panel: spectra of points with different degree of damage. Right panel, top to bottom: monochromatic images of the barrier layer, QW and contact layer. The points marked P1, P2 and P3 correspond to the spectra plotted in the left panel. Contrarily to Fig. 8, the damage is restricted to the injection channels.
As observed in the lower image in Fig. 8, the GaAs contact layer is free of damage close to the front facet. Despite having been etched from the device for the measurements, the dielectric has reduced the surface recombination velocity (SRV), giving the bright CL contrast in the blocking stripes with respect to the injection stripes, which appear darker. As mentioned above, the QW emission reveals degraded zones, which also extend beneath the blocking channels. One appreciates the discontinuity of the degradation pattern. The waveguide shows some damage along the injection channels, although the most severe degradation mainly concerns the QW. This is the expected behavior because the QW is the absorbing medium of the laser radiation when the local temperature increases. The GaAs contact layer is not affected by the degradation anywhere along the cavity.
Figure 9 evidences that, close to the back mirror, the damage is limited to the optically active region as well. The QW is degraded in the rear part of the cavity, but to a substantially lower degree than that observed close to the front facet. Furthermore, as the DLDs propagate, the lines observed beneath the blocking stripes start to disappear, and, as the rear facet is approached, the DLDs are strictly observed along the gain stripes. In fact, the degradation progressively reduces the optical power density because of the decrease of the gain volume; consequently, the damage is reduced and appears concentrated along the injection channels, where the laser power density is higher. As seen in the upper images of Figs. 8 and 9, even if the emission from the waveguide layers is only barely observed because of their composition and structure, the barriers are significantly less damaged than the QW. This means that even if the gain is substantially reduced when a large part of the QW is damaged, the cavity operates until the end of lasing.
The CL spectra acquired in points with different degrees of degradation show that the main differences among them are related to the emission intensity of the QW, without significant changes in the peak energy and shape. This rules out phenomena as melting or intermixing, which should result in compositional changes. According to our results, melting might occur, but it would be a very local event and close to the front facet. This suggests that melting is not necessarily a by-product of COD, even if previous reports have reported evident signs of melting in COD degraded lasers. [27,38,42]
3.3 Modelling of the degradation pattern
The laser mode distribution when hot spots are formed in the laser cavity has been modeled and computed. First, one or several tiny hot spots (≈1µm diameter) were placed in the simulated region by specifying their corresponding gaussian temperature profiles. The maximum temperature of those profiles ranged from 300 K to 1500 K, right below the melting temperature of the QW. Then, the impact of these hot regions on the laser mode distribution was calculated; for more details see, [18] where the thermal lensing effect was studied for a single hot spot. Here, and according to the defects seen in the CL images, we analyze the effect of multiple hot spots arranged perpendicularly to the direction of light propagation to account for the experimentally observed lateral mode filamentation.
Figures 10(a) and 10(b) correspond to simulations of two hot spots in the GaAs QW, inside the injection channel and with a lateral distance between them of 4 µm. The filamentation pattern is determined by the relative position of the two hot spots, depending mainly on the lateral separation between them (Δx), and only slightly on the distance shift of the two spots along the laser cavity (Δy), see Fig. 10(b). The thermal lensing associated with the temperature distribution of the hot spots focuses the laser light a few micrometers ahead of each defect and gives rise to a discontinuous filamentary-like laser intensity distribution along the resonant cavity. Furthermore, one observes that the laser filaments invade the blocking channels (outside the regions delimited by the black vertical lines in Fig. 10). Hence, DLDs can be generated in the absence of current injection, as observed in the CL images of Figs. 6 and 8. The successive CODs in Fig. 6 can be associated with the simulated patterns. First, defects (point defects, small loops…) would be formed around the hot spot because of the local high temperatures, and thermal stresses. The subsequent thermal focusing on this previously sensitized region would lead to enhanced light absorption, giving rise to favorable sites for a new COD event. Following this propagation sequence, the discontinuous DLDs would be formed.
Fig. 10. a) Simulation of two parallel hot spots with a lateral distance of 4 µm. b) Idem but adding a distance shift of 10 µm along the laser cavity. c) Simulation of 4 parallel hot spots with a lateral distance of 2µm. The blue arrows indicate the position of the hot spots, for which the peak temperature was 1400 K, about 100 K below the melting temperature. The injection channel is defined by the black vertical lines. Animations of the lateral mode distribution in the presence of hot spots for various relative positions are provided in the supplementary information (see Visualization 1 and Visualization 2).
The result obtained for four hot spots laterally spaced 2 µm is shown in Fig. 10(c), where one observes how the filamentation and thermal lensing result in a series of parallel alignments of high intensity power traces (where potential COD events would take place) that reproduce qualitatively the results observed experimentally.
4. Discussion
The thermal runaway process is usually assumed to be due to the melting of the active zone of the laser and the propagation of a melting front towards the incident laser light [2,4,37,43]. The discontinuous damage observed on CL images at low temperature clearly shows that this is not the propagation mechanism in the studied laser. On the other hand, the spectral analysis of the damage does not support the melting of the active zone, Figs. 8, and 9. Only close to the highly damaged front facet, local melting might have been reached. However, the discrete morphology of the DLDs rules out the propagation of a molten front.
The COD has been described as a microexplosion [5,44]. In fact, the COD results from the local accumulation of energy supplied by the laser at a tiny absorbing volume of the QW. In step tests, the COD has been associated with the occurrence of a flash of Planck’s radiation, which was monitored with a thermo-camera [38]. Successive current pulses produced new ignitions, each spatially separated a few micrometers from the previous one. This propagation mode would create a discontinuous damage pattern. The ignition mechanism is related to local heating produced by the absorption of the laser radiation generated in the cold parts of the device. The hot local region of the QW acts as an absorbing medium because of the local shrinkage of the bandgap, generating e-h pairs that recombine non radiatively. The local energy accumulated in the QW would be the balance between the amount of energy absorbed and dissipated. Because of the low thermal conductivity of the QW, which is limited by its narrow thickness [29] and the thermal boundary resistance (TBR) at the interfaces, [30,31] the absorbed energy is mainly accumulated in the QW, producing a sharp temperature drop at the QW/waveguide interface with the concomitant thermal stresses [45]. Once the critical stress is reached, a tangle of dislocations is produced, further reducing the thermal conductivity, [46] and enhancing the optical absorption. This results in a sudden increase of the temperature, i.e. the flash of Planck’s radiation reported by Tomm et al., [8,48] or the micro-explosion mentioned by Eliseev [44]. Note that this mechanism of dislocation generation is not assimilable to the generation of misfit dislocations. Rather than that, it consists of an abrupt plastic deformation, which generates a dense array of dislocations, necessary to account for a drastic reduction of the local thermal conductivity. The time constants involved in this process are much shorter than those governing typical climb or glide dislocation motion [47,48].
In the present case of long-term COD, the morphology of the DLDs revealed by the low-temperature CL images shows that their propagation is not continuous, as usually described in the thermal runaway model. Rather, it involves a sequence of COD events spatially separated along the cavity. In the studied laser, the COD starts at the front facet and is followed by a sequence of bulk CODs (COBD), [49,50] which are COD events produced in the inner cavity, so as to be distinguished from the CODs in the mirror face (COMD). A COD event has two relevant outcomes. The thermal flash produced in a COD event would give rise to a thermal wave, which will generate defects (e.g., point defects, small dislocation loops…,) extending to a certain distance around the hot spots. These defects behave as non-radiative recombination centers and sensitize that region to degradation. On the other hand, the local temperature distribution produced by the COD ignition will cause the thermal lensing of the laser light, which is focused on a nearby defect rich region of the cavity. The laser focusing on these sensitized regions provides the ideal scenario for triggering a new COBD event [18]. The DLD would then propagate by a sequence of COBDs, which will end when the progressive quenching of the laser power reduces the local laser intensity below the threshold necessary for achieving another COBD event.
5. Conclusion
Catastrophically degraded 808 nm laser diodes were studied in a wide range of temperatures (80 – 300 K). The morphology of the DLDs as revealed by CL images drastically changes as the temperature is progressively decreased. At low temperatures, the DLDs are observed as a series of parallel discontinuous lines evidencing the lateral mode filamentation of the laser field and the thermal self-focusing. As the temperature is raised, the recombination atmosphere around the defects is broadened due to the longer carrier diffusion length. As a result, DLDs appear as continuous and blurred shaped in the room temperature CL images.
An in-depth spectral analysis of the damaged regions, both close to the front and rear mirrors, suggests that local melting did not take place. Rather than that, the degradation was probably due to the formation of dense arrays of extended defects, mainly localized in the QW.
The mode filamentation was produced by the existence of hot spots in the QW inside the injection channels. The interaction between the hot spots and the laser radiation produces the filamentation and the local increase of the laser power density by self-focusing due thermal lensing, which reignites the degradation in a point separated a few µm behind the previous hot spot.
The evolution of the DLDs along the laser cavity clearly shows that the damage is progressively reduced when approaching the back facet. Though the damage extends over the injection and, to a lesser extent, blocking channels close to the front facet, the number of DLDs propagating towards the end of the channel gradually diminishes. Close to the rear facet, the DLDs only appear beneath the injection channel.
The laser field inside the cavity was modeled in the simultaneous presence of several hot spots. The results, irrespective of the specific positions of those hot spots, yield a discontinuous distribution of the laser intensity along filamentary paths aligned parallel to the cavity. The points of higher laser intensity would be responsible for the initiation of the successive reignitions resulting in a sequence of COBDs, which would reproduce the morphology revealed by the CL images.
Funding
Agencia Estatal de Investigación (PID2020-113533RB-C33, PID2021-12604OB-C21); Consejería de Educación, Junta de Castilla y León (VA283P18, UVA09-IR2021).
Disclosures
The authors declare no conflicts of interest
Data Availability
Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon reasonable request
References
1. R. G. Waters, “Diode laser degradation mechanisms: A review,” Prog. Quantum Electron. 15(3), 153–174 (1991). [CrossRef]
2. O. Ueda, Reliability and Degradation of III-V Optical Devices (Artech House, 1996).
3. J. Jiménez, “Laser diode reliability: crystal defects and degradation modes,” Comptes Rendus Phys. 4(6), 663–673 (2003). [CrossRef]
4. C. H. Henry, P. M. Petroff, R. A. Logan, and F. R. Merritt, “Catastrophic damage of AlxGa1− xAs double-heterostructure laser material,” J. Appl. Phys. 50(5), 3721–3732 (1979). [CrossRef]
5. J. W. Tomm, M. Ziegler, M. Hempel, and T. Elsaesser, “Mechanisms and fast kinetics of the catastrophic optical damage (COD) in GaAs-based diode lasers,” Laser Photonics Rev. 5(3), 422–441 (2011). [CrossRef]
6. P. G. Eliseev, A. G. Glebov, and M. Osinski, “Current self-distribution effect in diode lasers: analytic criterion and numerical study,” IEEE J. Sel. Top. Quantum Electron. 3(2), 499–506 (1997). [CrossRef]
7. W. C. Tang, H. J. Rosen, P. Vettiger, and D. J. Webb, “Raman microprobe study of the time development of AlGaAs single quantum well laser facet temperature on route to catastrophic breakdown,” Appl. Phys. Lett. 59(9), 1005–1007 (1991). [CrossRef]
8. M. Adams, S. Rauch, C. Holly, M. Traub, and H. D. Hoffmann, “Modeling of the impact of current crowding on catastrophic optical damage in 9xx-nm high power laser diodes,” Proc. SPIE 11688, 116680C (2021).
9. E. Yablonovitch, “Optical Dielectric Strength of Alkali- Halide Crystals Obtained by Laser- Induced Breakdown,” Appl. Phys. Lett. 19(11), 495–497 (1971). [CrossRef]
10. N. Bloembergen, “Laser-induced electric breakdown in solids,” IEEE J. Quantum Electron. 10(3), 375–386 (1974). [CrossRef]
11. A. Chavan, R. Radionova, G. W. Charache, R. J. Menna, H. J. Schluter, and J. L. Hostetler, “Comparison of facet temperature and degradation of unpumped and passivated facets of Al-Free 940-nm lasers using photoluminescence,” IEEE J. Quantum Electron. 41(5), 630–635 (2005). [CrossRef]
12. P. G. Eliseev, “Optical strength of semiconductor laser material,” Prog. Quantum Electron. 20(1), 1–82 (1996). [CrossRef]
13. A. K. Chin and R. K. Bertaska, “Catastrophic Optical Damage in High-Power, Broad-Area Laser Diodes,” in Materials and reliability handbook for semiconductor optical and electron devices, S. Pearton and O. Ueda, eds. (Springer, 2013).
14. G. Chen and C. L. Tien, “Facet heating of quantum well lasers,” J. Appl. Phys. 74(4), 2167–2174 (1993). [CrossRef]
15. A. Moser, E. E. Latta, and D. J. Webb, “Thermodynamics approach to catastrophic optical mirror damage of AlGaAs single quantum well lasers,” Appl. Phys. Lett. 55(12), 1152–1154 (1989). [CrossRef]
16. J. Souto, J. L. Pura, and J. Jiménez, “Thermomechanical issues of high power laser diode catastrophic optical damage,” J. Phys. D: Appl. Phys. 52(34), 343002 (2019). [CrossRef]
17. Z. Su, L. Huang, F. Liu, J. P. Freedman, L. M. Porter, R. F. Davis, and J. A. Malen, “Comparison of facet temperature and degradation of unpumped and passivated facets of Al-Free 940-nm lasers using photoluminescence,” Appl. Phys. Lett. 100(20), 201106 (2012). [CrossRef]
18. J. L. Pura, J. Souto, and J. Jiménez, “Effect of thermal lensing and the micrometric degraded regions on the catastrophic optical damage process of high-power laser diodes,” Opt. Lett. 45(7), 1667–1670 (2020). [CrossRef]
19. H. Yao, P. G. Snyder, and J. A. Woollam, “Temperature dependence of optical properties of GaAs,” J. Appl. Phys. 70(6), 3261–3267 (1991). [CrossRef]
20. Y. P. Varshni, “Temperature dependence of the energy gap in semiconductors,” Physica 34(1), 149–154 (1967). [CrossRef]
21. J. S. Blakemore, “Semiconducting and other major properties of gallium arsenide,” J. Appl. Phys. 53(10), R123–R181 (1982). [CrossRef]
22. D. E. Aspnes, “GaAs lower conduction-band minima: Ordering and properties,” Phys. Rev. B 14(12), 5331–5343 (1976). [CrossRef]
23. A. V. Andrianov, S. R. A. Dods, J. Morgan, J. W. Orton, T. M. Benson, I. Harrison, E. C. Larkins, F. X. Daiminger, E. Vassilakis, and J.P. Hirtz, “Optical and photoelectric study of mirror facets in degraded high power AlGaAs 808 nm laser diodes,” J. Appl. Phys. 87(7), 3227–3233 (2000). [CrossRef]
24. J. Souto, J. L. Pura, and J. Jiménez, “About the physical meaning of the critical temperature for catastrophic optical damage in high power quantum well laser diodes,” Laser Phys. Lett. 13(2), 025005 (2016). [CrossRef]
25. J. Souto, J. L. Pura, and J. Jiménez, “Nanoscale effects on the thermal and mechanical properties of AlGaAs/GaAs quantum well laser diodes: influence on the catastrophic optical damage (COD),” J. Phys. D: Appl. Phys. 50(23), 235101 (2017). [CrossRef]
26. M. Ziegler, J. W. Tomm, D. Reeber, T. Elsaesser, U. Zeimer, H.E. Larsen, P. M. Petersen, and P. E. Andersen, “Catastrophic optical mirror damage in diode lasers monitored during single-pulse operation,” Appl. Phys. Lett. 94(19), 191101 (2009). [CrossRef]
27. R. E. Mallard, R. Clayton, D. Mayer, and L. Hobbs, “Failure analysis of high power GaAs-based lasers using electron beam induced current analysis and transmission electron microscopy,” J. Vac. Sci. Technol., A 16(2), 825–829 (1998). [CrossRef]
28. J. W. Tomm, A. Gerhardt, R. Müller, V. Malyarchuk, Y. Sainte-Marie, P. Galtier, J. Nagle, and J. P. Landesman, “Spatially resolved spectroscopic strain measurements on high-power laser diode bars,” J. Appl. Phys. 93(3), 1354–1362 (2003). [CrossRef]
29. E. Gȩsikowska and W. Nakwaski, “An impact of multi-layered structures of modern optoelectronic devices on their thermal properties,” Opt. Quantum Electron. 40(2-4), 205–216 (2008). [CrossRef]
30. G. Chen and C. L. Tien, “Thermal conductivities of quantum well structures,” J. Thermophys. Heat Transf. 7(2), 311–318 (1993). [CrossRef]
31. E.S. Landry and A. J. H. McGaughey, “Thermal boundary resistance predictions from molecular dynamics simulations and theoretical calculations,” Phys. Rev. B 80(16), 165304 (2009). [CrossRef]
32. A. Martín-Martín, M. P. Iñiguez, J. Jiménez, M. Oudart, and J. Nagle, “Role of the thermal boundary resistance of the quantum well interfaces on the degradation of high power laser diodes,” J. Appl. Phys. 110(3), 033113 (2011). [CrossRef]
33. A. Martín-Martín, M. Avella, M. P. Iñiguez, J. Jiménez, M. Oudart, and J. Nagle, “Thermomechanical model for the plastic deformation in high power laser diodes during operation,” J. Appl. Phys. 106(7), 073105 (2009). [CrossRef]
34. J. Jimenez and J. W. Tomm, Spectroscopic Analysis of Optoelectronic Semiconductors202, (Springer, 2016).
35. S. Dadgostar, J. Souto, and J. Jiménez, “CL as a tool for device characterisation: the case of laser diode degradation,” Nano Ex. 2(1), 014001 (2021). [CrossRef]
36. M. Hempel, J. W. Tomm, T. Elsaesser, and M. Bettiati, “High single-spatial-mode pulsed power from 980 nm emitting diode lasers,” Appl. Phys. Lett. 101(19), 191105 (2012). [CrossRef]
37. M. B. Sanayeh, A. Jaeger, W. Schmid, S. Tautz, P. Brick, K. Streubel, and G. Bacher, “Investigation of dark line defects induced by catastrophic optical damage in broad-area AlGaInP laser diodes,” Appl. Phys. Lett. 89(10), 101111 (2006). [CrossRef]
38. M. Hempel, F. La Mattina, J. W. Tomm, U. Zeimer, R. Broennimann, and T. Elsaesser, “Defect evolution during catastrophic optical damage of diode lasers,” Semicond. Sci. Technol. 26(7), 075020 (2011). [CrossRef]
39. M. Hempel, J. W. Tomm, V. Hortelano, N. Michel, J. Jiménez, J.M. Krakowski, and T. Elsaesser, “Time-resolved reconstruction of defect creation sequences in diode lasers,” Laser Photonics Rev. 6(6), L15–L19 (2012). [CrossRef]
40. S. Mahajan, H. Temkin, and R. A. Logan, “Formation of optically induced catastrophic degradation lines in InGaAsP epilayers,” Appl. Phys. Lett. 44(1), 119–121 (1984). [CrossRef]
41. C. W. Snyder, J. W. Lee, R. Hull, and R. A. Logan, “Catastrophic degradation lines at the facet of InGaAsP/InP lasers investigated by transmission electron microscopy,” Appl. Phys. Lett. 67(4), 488–490 (1995). [CrossRef]
42. M. Hempel, J. W. Tomm, F. La Mattina, I. Ratschinski, M. Schade, I. Shorubalko, M. Stiefel, H. S. S. Leipner, F. M. M. Kießling, and T. Elsaesser, “Microscopic origins of catastrophic optical damage in diode lasers,” IEEE J. Sel. Top. Quantum Electron. 19(4), 1500508 (2013). [CrossRef]
43. O. Ueda, K. Wakao, S. Komiya, A. Yamaguchi, S. Isozumi, and I. Umebu, “Catastrophic degradation of InGaAsP/InGaP double-heterostructure lasers grown on (001) GaAs substrates by liquid-phase epitaxy,” J. Appl. Phys. 58(11), 3996–4002 (1985). [CrossRef]
44. P. G. Eliseev, “Degradation of injection lasers,” J. Lumin. 7, 338–356 (1973). [CrossRef]
45. G. Chen, “Thermal conductivity and ballistic-phonon transport in the cross-plane direction of superlattices,” Phys. Rev. B 57(23), 14958–14973 (1998). [CrossRef]
46. P. E. Hopkins, J. C. Duda, S. P. Clark, C. P. Hains, T. J. Rotter, L. M. Phinney, and G. Balakrishnan, “Effect of dislocation density on thermal boundary conductance across GaSb/GaAs interfaces,” Appl. Phys. Lett. 98(16), 161913 (2011). [CrossRef]
47. B. D. Schwartz, “Dark line defect growth in optically pumped AlxGa1− xAs laser material,” J. Appl. Phys. 58(2), 677–682 (1985). [CrossRef]
48. A. K. Chin, R. K. Bertaska, M. A. Jaspan, A. M. Flusberg, S. D. Swartz, M. T. Knapczyk, I. Smilanski, and J. H. Jacob, “Catastrophic Optical-Damage in High-Power, Broad-Area Laser-Diodes,” MRS Proc. 1195, 40–51 (2010). [CrossRef]
49. Y. Sin, Z. Lingley, M. Brodie, N. Presser, and S. C. Moss, “Catastrophic optical bulk degradation (COBD) in high-power single- and multi-mode InGaAs-AlGaAs strained quantum well lasers,” Proc. SPIE 10086, 100860S (2017). [CrossRef]
50. Y. Sin, Z. Lingley, M. Brodie, N. Presser, N. Ives, and S. C. Moss, “Catastrophic Optical Bulk Damage in High-Power InGaAs-AlGaAs Strained Quantum Well Lasers,” IEEE J. Sel. Top. Quantum Electron. 23, 1500813 (2017). [CrossRef]
