Open Access
Add to BibSonomy
Abstract
Porous GaN has been proposed as a novel cladding material for visible light-emitting laser diodes (LDs). Fabrication of nanoporous-GaN bottom-cladding LDs was already realized by selective electrochemical etching (ECE) of the highly n-type doped GaN layer in the LD structure after epitaxy. In this work, we applied a reverse approach: in the first step, locally porous areas in GaN substrate were fabricated, and next, a LD structure was grown on top by plasma-assisted molecular beam epitaxy (PAMBE). We compare the electrical and optical properties of the devices with porous bottom cladding with the devices from the same wafer that was grown on top of a standard GaN layer. Continuous wave (CW) operation is achieved for porous LD at 435.4 nm and slope efficiency of 0.046 W/A. Standard LD was lased in CW mode at 442.6 nm and had a slope efficiency of 0.692 W/A. In porous LD, the internal losses were estimated using the Hakki-Paoli method to be 68 cm−1, while for standard LD, the losses were 25 cm−1. Near-field patterns recorded for the studied devices indicate light scattering on the porous layer to be the possible reason for the increased losses in porous LD.
Published by Optica Publishing Group under the terms of the Creative Commons Attribution 4.0 License. Further distribution of this work must maintain attribution to the author(s) and the published article's title, journal citation, and DOI.
1. Introduction
Nitride-based laser diodes (LDs) are of great importance for many critical applications: data recording and reading, displays, optical communication, gas detection, medical diagnostics, polymer curing and others. Although the fabrication and production technology of standard device designs is already mature, the scientific effort is still undertaken to improve the electrical and optical parameters of these devices [1,2]. Typically used AlGaN cladding layers offer relatively low refractive index contrast and cause issues with lattice mismatch to GaN. Therefore, other material solutions are explored, such as InAlN [3], highly doped GaN [4,5] or refractive index engineering using thick InGaN waveguides [6] to improve the light confinement. Nanoporous GaN is also considered as a potential material for this purpose due to wide range of available refractive indices, easily tunable with porosity degree [7–9]. A decrease in pumping power threshold for the optically pumped lasers with porous cladding layers [10,11] motivated the efforts to apply porous GaN also to electrically pumped edge-emitting LD structures. The high refractive index contrast to GaN offered by nanoporous layers [9] could bring straightforward improvements in light confinement when implemented as claddings in LDs, thus allowing for the reduction of threshold current density and increasing slope efficiency. Such a material is very attractive, especially for long-wavelength lasers: green and beyond, because the refractive index contrast of AlGaN to GaN decreases with increasing wavelength [12,13]. Since the first demonstration of optically pumped lasers with porous cladding layers there have been no reports on edge-emitting electrically pumped lasers for 8 years. Then, in the year 2022, three reports on electrically pumped LDs with porous GaN bottom cladding have been published [14–16]. The implementation of the porous layer into the LD structure was surprisingly disappointing because instead of improvement, the threshold current density and slope efficiency of porous-cladding LDs (por-LDs) were worse than their ‘classic’ counterparts. Despite the fact the epitaxial technique and substrate orientation were different, i.e., metalorganic chemical vapor deposition (MOCVD) on semipolar (20$\bar{2}$1) GaN or plasma-assisted molecular beam epitaxy (PAMBE) on (0001) GaN substrates, the fabrication strategy was essentially the same and involved porous layer fabrication after epitaxy. The authors considered different sources of increased losses and pointed to light scattering on porous layers [14–16]. Reported por-LDs emitted at 449 nm [14], 454 nm [15] and 512 nm [16], respectively; only in the second case continuous wave (CW) operation was obtained. Therefore, better understanding of the effects of porous GaN implemented as a cladding layer is needed to take advantage of the properties of this novel material, especially with respect to the impact of porosity and pore size distribution, emission wavelength, growth technology, or fabrication step sequence.
An approach previously applied to fabricate porous cladding LDs involved electrochemical etching (ECE) after the epitaxial process. This solution suffered from parasitic etching through the defects in oxide protective layer [14] or etching of other layers in the LD epitaxial structure of lower doping [15]. In this work we focused on the verification of the other approach, namely, etching porous layer before the LD structure growth. The proposed approach eliminates the problems of parasitic etching, e.g. of the active region, and provides a good control over the etching front position vs. the position of the laser ridge. We successfully fabricated blue LDs with and without porous bottom cladding on the same wafer using PAMBE, emitting at 435.4 and 442.6 nm, respectively. The CW laser operation is shown for both cases. Importantly, we confirmed that the pore structure is stable at 730°C, which is the temperature used during LD overgrowth. We observed inferior characteristics of the por-LDs compared to the standard LDs with respect to threshold current density and slope efficiency. Significantly higher losses measured by the Hakki-Paoli method and altered near-field patterns were observed for por-LDs.
2. Sample fabrication
The fabrication of por-LDs involved two main steps: (1) fabrication of locally porous GaN substrates and (2) epitaxy of LD structure. Commercially available HVPE GaN substrate with dislocation density of 5 × 106 cm−2 was used. Both, the bottom cladding structure with highly doped GaN:Si layer and LD structure we grown using PAMBE in VG Semicon MBE reactor. The growth conditions were basically the same as in our previous reports [14,17,18]. The growth temperature and presence of metallic layer during the epitaxy was controlled by laser reflectometry.
The fabrication steps of the por-LDs studied in this work are schematically shown in Fig. 1(a-d). The first step involved the epitaxy of the bottom cladding structure: 50 nm GaN:Si, 500 nm GaN:Si+ and 50 nm GaN:Si. The doping level in the GaN:Si+ cladding was 3·1019 cm−3, while in the other GaN:Si layers it was 2·1018 cm−3. The structure was capped with a 300 nm SiO2 protective layer to prevent parasitic etching from the top through dislocations [19] and 5 µm-wide grooves were made for conducting the ECE procedure. In this work we fabricated grooves of the same length as the laser ridge that provided homogeneous directionality of pores within the cladding along the ridge. In our previous demonstration, shorter grooves were applied [14]. Etching was carried out in 0.3 M oxalic acid at 3.1 V bias for 14 h. A platinum plate was used as a counter-electrode and a standard Ag/AgCl electrode as a reference. The etch front position after the process was assessed using an optical microscope to be 36 µm as measured from the groove that gives the average lateral etch rate of 44 nm/min.
Fig. 1. Schematics of the fabrication steps of porous-cladding LD: (a) epitaxy of highly doped GaN layer, (b) porous GaN etching, (c) LD structure regrowth on a porous substrate, (d) final LD structure after processing. (e) Details of the LD epitaxial structure.
After ECE, the oxide layer was removed and the sample was loaded back to the MBE reactor for LD structure regrowth. Details of the full epitaxial structure, including cladding and LD epitaxial parts are shown in Fig. 1(d). The LD structure consists of the following layers: 50 nm GaN:Si, 110 nm In0.04Ga0.96N waveguide, a single 10.4 nm In0.13Ga0.87N quantum well (QW), 110 nm In0.04Ga0.96N waveguide, 20 nm Al0.13Ga0.87N:Mg electron blocking layer (EBL), 100 nm GaN:Mg and 700 nm Al0.04Ga0.96N:Mg top cladding and InGaN:Mg contact layer. Lithography masks were designed in a way that devices with porous bottom cladding are located next to the devices with GaN:Si bottom cladding (on the same wafer), as presented in optical microscope image shown in Fig. 2(a). Some devices were moved away from the grooves so that there is no porous material below their ridges. Devices were processed as ridge-waveguide oxide isolated lasers. Afterwards, the substrate of the laser was thinned to 120 µm. Resonators were 2.2 µm wide and 700 and 1000 µm long. Mirrors were cleaved and left uncoated. No damage to the porous layer was observed after the full laser diode processing.
Fig. 2. (a) Optical microscope image of standard and porous LDs on the same wafer. Cross-sections are schematically marked. Contrast from porous cladding formed due to etching is marked with arrows. (b) schematics of the por-LD. SEM images taken at 45° of (c) a por-LD processed device cross-section of a showing complete etching of the cladding beneath the ridge (d) porous GaN cross-section perpendicular to etching direction, along the laser ridge (e) porous GaN viewed after cleaving along the etching direction, perpendicular to the laser ridge.
3. Results
3.1 Structural characterization of the por-LDs and standard LDs
Characterization of the porous bottom cladding pore morphology was performed by scanning electron microscopy (SEM). Two cross-sections were prepared as schematically marked in Fig. 2(a) and in the cartoon presented in Fig. 2(b): wafer with LD structure was cleaved perpendicular to the etching direction (along the groove used for ECE – cross section 1) and along the etching direction (perpendicular to the groove used for ECE – cross section 2). Porosity was assessed in the SEM image shown in Fig. 2(d) that is a cross-section (1), parallel to the laser ridge. The porosity and pore size of the porous GaN was estimated using image processing by open source ImageJ application. The porosity of the GaN:Si layer with doping level of 3·1019 cm−3 etched at 3.1 V is approximately 21% with an average pore size of 14 nm. It can be seen from the image presented in Fig. 2(d), the pore size distribution is rather uniform, but there are locally some pores of a diameter >30 nm. In our previous report presenting LD with porous bottom cladding [14] we used GaN:Si layer doped to level of 6·1019 cm−3 etched at 2.2 V. Then, a lower porosity, 15% and a pore size of ≈ 20 nm, elongated in the vertical direction, was achieved. Note that the porosity was assessed on a groove wall surface, while now we present a cross section of the porous layer, so the results cannot be directly compared. The pore morphology was observed on the groove wall also in the previous report [8]. Elongated, tubular pores are visible in the cross-section (2) presented in Fig. 2(e). In SEM images shown in Fig. 2(d-e), a line of pores visible 50 nm below the porous cladding, that is the position of the interface between the substrate and MBE-grown GaN:Si. However, it is not visible in Fig. 2(c). Fig. 2(c) presents a processed por-LD device cross-section that proves complete etching of the cladding beneath the ridge. Position of the etching front is a few micrometers away from the mesa.
It is important to emphasize, that in the pre-growth porosified cladding approach no parasitic etching channels through dislocations from the top surface were observed, see Fig. 2(a), as opposed to the after-growth etching process [14]. Additionally, note that SEM investigation is done on processed devices, that proves the pore structure is stable at 730°C during LD regrowth on locally porous GaN substrate by PAMBE.
3.2 Optical and electrical characterization of the por-LDs and standard LDs
Figure 3 shows the light-current-voltage (LIV) characteristics of LDs with porous and standard cladding layers. Measurements were taken under DC operation. The devices were c-mounted to ensure low thermal resistivity. The case temperature was stabilized by a thermoelectric cooler at 20°C. The threshold current densities were 7.95 and 9.94 kA/cm2 for por-LD and standard LD, respectively. Fitted slope efficiency for por-LD is 0.046 W/A, similarly to the previous reports. Standard LD slope efficiency is 0.692 W/A. High resolution spectra of the lasers above threshold were measured using 1000 Horiba Jobin Yvon 1-meter-long spectrometer and are presented in Fig. 4. Emission wavelength difference is observed between the studied LD. The blue shift of 7 nm is attributed to lower strain state in the QW in case of por-LD [20]. The value of the blue shift is similar to the one reported earlier in our work, [14] when the porosification of the cladding was carried out after growth of the whole epi structure of the laser diode. Therefore, we conclude that the shift is not caused by any structural differences between porous and nonporous regions, such as thickness of the quantum well or the indium content, but rather by the partial strain release.
Fig. 4. High resolution laser emission spectra above threshold collected for (a) LD with porous bottom cladding and (b) for standard LD from the same wafer. The mode spacing difference between both spectra results from different resonator lengths.
To estimate and compare the optical losses of the devices with the porous and standard bottom cladding layer, we performed the measurement of the optical modal gain using the Hakki-Paoli technique [5,21,22]. Fig. 5 presents the gain spectra for por-LD and standard LD for various currents. From the long wavelength part of the gain spectra we estimated the internal losses for the standard LD to be αi = 25 cm−1. For por-LD the value of internal losses αi cannot be estimated from the long-wavelength tail of the spectrum, because they exceed the experimental limit (35 cm−1) related to the noise level of the measurement of the internal losses for given length of the resonator [22]. The dependence of the maxima of the measured modal gain on current density is presented in Figure 5(c). The differential gain, dG/dj, was estimated to be 6.6 and 10.8 cm/kA for the standard and por-LD, respectively. Despite higher value of differential gain, the por-LD exhibits higher optical losses. The crossing point of the fitted differential gain of por-LD with the zero current point gives a rough estimation of the value of internal losses of 68 cm−1, which is of the same order of magnitude as suggested before for the LDs with porous cladding layers [14,15]. The origin of such high internal losses is believed to be caused by the scattering on the nanoporous GaN. The differential modal gain was fitted for each laser using the low current data points.
Fig. 5. Modal gain spectra of (a) por-LD and (b) standard LD collected for selected operating currents. (c) The dependence of the maximum modal gain of measured devices on the current density.
Both samples consist of the active region grown at the same time on the same wafer. Hence, the large discrepancy in the observed differential modal gain should not originate from the variation of the material gain of the quantum well. Therefore, we performed the 2D optical mode simulations using the COMSOL Multiphysics software to determine material gain [23]. The refractive index of porous GaN was calculated using the volume averaging theory [24] assuming 21% porosity. The optical mode distribution and material gain relation on current density are presented in Figure 6. The active region optical mode confinement factor (Γ) for por-LD is higher, which is expected since porous GaN has a much lower refractive index. The mode is pushed out of the bottom part of the device compared to the standard LD. Obtained values of the confinement factors allow to estimate the material gain g from the modal gain by the relation: G=Γg+αi. The material gain for the por-LD is just slightly higher than for the standard LD. This leads to the conclusion that the differences in modal gain arise mainly from the differences in the optical losses and change of the mode confinement due to the porosity.
Fig. 6. The 2D map of optical mode for (a) por-LD and (b) standard LD. The profile of refractive index and optical mode along vertical cross section for (c) por-LD and (d) standard LD. The shown confinement factor values are derived from the 2D simulations. (e) The material gain dependence on the current density for both lasers with the fitted differential material gain values.
In order to acquire information about possible scattering of the optical mode, the near-field images were taken using a Gaussian beam telescope setup, described in more detail in Ref. [25]. Figure 7(a-c) and (d-f) shows the two-dimensional (2D) intensity profiles of the near-field pattern of both LDs in the transversal direction for selected operating currents below threshold. All the maps are normalized and presented in the same x-y scale, 14 µm x 7 µm. Near-fields collected for por-LD clearly exhibit some additional features being a fingerprint of light scattering by the porous bottom cladding. Additionally, the standard LD exhibits a two mode emission, visible in particular for lower currents. The comparison of the intensity profiles for both studied devices is presented in Figure 7(g). Low light intensity in the near-field pattern presented in Figure 7(c) at the bottom side of the por-LD suggests a good light confinement. However, the additional few peaks visible at the top side of the device indicate that some light is lost due to scattering on the porous layer. This mechanism was not included in the simulation presented in Figure 6(a-b), which explains the differences in the mode distribution. Smaller pore size of higher density has been suggested to be a possible solution to minimize this issue [14,15].
Fig. 7. Near-field patterns for (a-c) por-LD and (d-f) standard LD collected for selected operating currents. In all the maps the intensity is normalized. (g) Intensity profiles of the near-field patterns in transversal direction for por-LD and standard LD extracted from maps presented in (c) and (f), respectively.
3.3 Discussion on the losses and injection efficiency of the por-LDs and the standard LDs
We will now try to estimate the injection efficiency and internal losses of both LDs. The slope efficiency measured from one side of uncoated LDs is given by:
In case of the standard LD using Eq. (1) we get an injection efficiency of 100%, assuming resonator length of 700 μm, internal losses of 25 cm−1 and slope efficiency of 0.692 W/A. The slope efficiency of the por-LD is relatively small compared to the standard LD. If one assumes injection efficiency equal to 100% as in case of a standard LD, then the internal losses of the por-LD would need to be on the order of 500 cm−1, which is extremely high. If one assumes more realistic internal losses of 68 cm−1, then the injection efficiency would be equal to 17%, a value which is not unreasonable. On the other hand, if we assume such a low injection efficiency, it is hard to explain why the material gain and threshold current densities of both LDs are similar. Therefore, it is evident that more work is needed to fully understand the internal parameters of LDs with porous GaN claddings.
4. Conclusions
We demonstrate CW operation of porous-bottom-cladding LD (por-LD) grown by PAMBE on locally porous GaN (0001) substrate. The proposed approach involving the use of porous substrate allowed for the elimination of the issues related to the parasitic etching. SEM investigation confirmed that the pore structure is stable at 730°C during LD regrowth on locally porous GaN substrate. The electrical and optical characteristics of por-LD are compared to those of its standard counterpart fabricated from the same wafer. Por-LD exhibited shorter emission wavelength 435.4 nm and much lower slope efficiency 0.047 W/A as compared to the standard device that emitted at 442.6 nm at 0.692 W/A. Significantly higher losses were measured for the device with porous bottom cladding and a fingerprint of light scattering in near-field was observed. Engineering the porosity in the bottom cladding could be a possible solution to improve the performance of the LDs with porous claddings. Nanoporous layers with smaller pore diameter with higher density should reduce the light scattering effect and value of optical losses while preserving the refractive index value of the porous cladding layer. Understanding of the impact of porous GaN implemented to device technology could be highly beneficial for the light confinement in long wavelength InGaN-based emitters by enabling the fluent refractive index engineering, both post- and pre- growth. However, more work has to be done to understand the present limits of the technology and origin of the increased optical losses.
Funding
HORIZON EUROPE Digital, Industry and Space (Vission ID:101070622); Narodowe Centrum Badań i Rozwoju (LIDER/35/0127/L-9/17/NCBR/2018, Norway Grants 2014-2021 NOR/SGS/BANANO/0164/2020); Narodowe Centrum Nauki (PRELUDIUM 2019/35/N/ST7/02968, SONATA no. 2019/35/D/ST5/02950); Fundacja na rzecz Nauki Polskiej (POWROTY/REINTEGRATION POIR.04.04.00-00-4463/17-00, TEAM-TECH POIR.04.04.00-00-210C/16-00).
Acknowledgments
This work received funding from the Foundation for Polish Science co-financed by the European Union under the European Regional Development Fund within the projects POWROTY/REINTEGRATION POIR.04.04.00-00-4463/17-00 and TEAM-TECH POIR.04.04.00-00-210C/16-00. This work was also financially supported by National Science Centre Poland within grants SONATA no. 2019/35/D/ST5/02950 and 2019/35/D/ST3/03008, PRELUDIUM 2019/35/N/ST7/02968 as well as the National Centre for Research and Development within grant no. LIDER/35/0127/L-9/17/NCBR/2018. The research leading to these results has also received funding from the Norway Grants 2014-2021 via the National Centre for Research and Development grant no. NOR/SGS/BANANO/0164/2020. This work received funding from the European Horizon 2020 project VISSION (Grant ID:101070622)
Authors would like to thank Krzesimir Nowakowski-Szkudlarek for skillful cleaving of the LDs.
Disclosures
The authors declare that there are no conflicts of interest related to this article.
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. G. Muziol, H. Turski, M. Siekacz, et al., “Laser diodes grown by molecular beam epitaxy,” in Nitride Semiconductor Technology (Wiley, 2020), p. 301–332.
2. Y. Kato, K. Miyoshi, T. Takeuchi, et al., “Low threshold current density in GaInN-based laser diodes with GaN tunnel junctions,” Appl. Phys. Express 14(6), 061002 (2021). [CrossRef]
3. A. Castiglia, E. Feltin, G. Cosendey, et al., “Al0.83In0.17N lattice-matched to GaN used as an optical blocking layer in GaN-based edge emitting lasers,” Appl. Phys. Lett. 94(19), 193506 (2009). [CrossRef]
4. P. Perlin, T. Czyszanowski, L. Marona, et al., “Highly doped GaN: a material for plasmonic claddings for blue/green InGaN laser diodes,” Proc. SPIE 8262, 826216 (2012). [CrossRef]
5. S. Stańczyk, T. Czyszanowski, A. Kafar, et al., “InGaN laser diodes with reduced AlGaN cladding thickness fabricated on GaN plasmonic substrate,” Appl. Phys. Lett. 102(15), 151102 (2013). [CrossRef]
6. G. Muziol, H. Turski, M. Siekacz, et al., “Elimination of leakage of optical modes to GaN substrate in nitride laser diodes using a thick InGaN waveguide,” Appl. Phys. Express 9(9), 092103 (2016). [CrossRef]
7. D. Chen, H. Xiao, and J. Han, “Nanopores in GaN by electrochemical anodization in hydrofluoric acid: formation and mechanism,” J. Appl. Phys. 112(6), 064303 (2012). [CrossRef]
8. M. Sawicka, N. Fiuczek, H. Turski, et al., “Revealing inhomogeneous Si incorporation into GaN at the nanometer scale by electrochemical etching,” Nanoscale 12(10), 6137–6143 (2020). [CrossRef]
9. P. H. Griffin and R. A. Oliver, “Porous nitride semiconductors reviewed,” J. Phys. D: Appl. Phys. 53(38), 383002 (2020). [CrossRef]
10. C. Zhang, G. Yuan, K. Xiong, et al., “(Invited) New directions in GaN photonics enabled by electrochemical processes,” ECS Trans. 72(5), 47–56 (2016). [CrossRef]
11. G. Yuan, K. Xiong, C. Zhang, et al., “Optical engineering of modal gain in a III-nitride laser with nanoporous GaN,” ACS Photonics 3(9), 1604–1610 (2016). [CrossRef]
12. M. Hajdel, G. Muzioł, K. Nowakowski-Szkudlarek, et al., “Influence of InGaN waveguide on injection efficiency in III-nitride laser diodes,” Opt. Appl. 50(2), 311–321 (2020). [CrossRef]
13. G. Muziol, H. Turski, M. Siekacz, et al., “Enhancement of optical confinement factor by InGaN waveguide in blue laser diodes grown by plasma-assisted molecular beam epitaxy,” Appl. Phys. Express 8(3), 032103 (2015). [CrossRef]
14. M. Sawicka, G. Muziol, N. Fiuczek, et al., “Electrically pumped blue laser diodes with nanoporous bottom cladding,” Opt. Express 30(7), 10709–10722 (2022). [CrossRef]
15. R. Anderson, D. Cohen, H. Zhang, et al., “Nano-porous GaN cladding and scattering loss in edge emitting laser diodes,” Opt. Express 30(2), 2759–2767 (2022). [CrossRef]
16. R. Anderson, H. Zhang, E. Trageser, et al., “Green edge emitting lasers with porous GaN cladding,” Opt. Express 30(15), 27674–27682 (2022). [CrossRef]
17. G. Muziol, M. Hajdel, H. Turski, et al., “Distributed-feedback blue laser diode utilizing a tunnel junction grown by plasma-assisted molecular beam epitaxy,” Opt. Express 28(23), 35321–35329 (2020). [CrossRef]
18. M. Siekacz, G. Muziol, H. Turski, et al., “Vertical integration of nitride laser diodes and light emitting diodes by tunnel junctions,” Electronics 9(9), 1481 (2020). [CrossRef]
19. F. C. P. Massabuau, P. H. Griffin, H. P. Springbett, et al., “Dislocations as channels for the fabrication of sub-surface porous GaN by electrochemical etching,” APL Mater. 8(3), 031115 (2020). [CrossRef]
20. R. Ley, L. Chan, P. Shapturenka, et al., “Strain relaxation of InGaN/GaN multi-quantum well light emitters via nanopatterning,” Opt. Express 27(21), 30081–30089 (2019). [CrossRef]
21. B. W. Hakki and T. L. Paoli, “Gain spectra in GaAs double−heterostructure injection lasers,” J. Appl. Phys. 46(3), 1299–1306 (1975). [CrossRef]
22. S. Stańczyk, Physical mechanisms determining the maximum optical power emitted from InGaN laser diodes, 2022, Institute of High Pressure Physics Polish Academy of Sciences,https://bip.unipress.waw.pl/images/stories/doktoraty/stanczykszymon.pdf.
23. COMSOL Multiphysics®, COMSOL ABStockholm, Sweden.
24. M. M. Braun and L. Pilon, “Effective optical properties of non-absorbing nanoporous thin films,” Thin Solid Films 496(2), 505–514 (2006). [CrossRef]
25. S. Rogowsky, H. Braun, U. T. Schwarz, et al., “Multidimensional near- and far-field measurements of broad ridge (Al,In)GaN laser diodes,” Phys. Status Solidi C 6(S2), S852–S855 (2009). [CrossRef]
