Low-threshold lasing of optically pumped micropillar lasers with Al0.2Ga0.8As/Al0.9Ga0.1As distributed Bragg reflectors

Semiconductor quantum dots (QDs) embedded in an optical microcavity have become an attractive and powerful platform not only for the study of cavity quantum electrodynamics (cQED) in the solid state but also to realize bright photonic sources to enable advanced technologies, including quantum communication,1,21. D. A. Vajner, L. Rickert, T. Gao, K. Kaymazlar, and T. Heindel, Adv. Quantum Technol. 5, 2100116 (2022). https://doi.org/10.1002/qute.2021001162. S. Rodt and S. Reitzenstein, APL Photonics 6, 010901 (2021). https://doi.org/10.1063/5.0031628 optical sensing,33. L. He, Ş. K. Özdemir, and L. Yang, Laser Photonics Rev. 7, 60 (2013). https://doi.org/10.1002/lpor.201100032 and photonic neuromorphic computation.4,54. A. Skalli, J. Robertson, D. Owen-Newns, M. Hejda, X. Porte, S. Reitzenstein, A. Hurtado, and D. Brunner, Opt. Mater. Express 12, 2395 (2022). https://doi.org/10.1364/OME.4509265. T. Heuser, J. Große, A. Kaganskiy, D. Brunner, and S. Reitzenstein, APL Photonics 3, 116103 (2018). https://doi.org/10.1063/1.5050669 An interesting experimental platform is the quantum dot micropillar system, where the quantum dots are located in a central cavity layer, sandwiched between epitaxially grown dielectric Bragg mirrors (DBRs) to offer a high Q-factor, a small mode volume, and, therefore, an enhanced Purcell factor.6,76. A. Kaganskiy, S. Kreinberg, X. Porte, and S. Reitzenstein, Optica 6, 404 (2019). https://doi.org/10.1364/OPTICA.6.0004047. S. Reitzenstein, A. Bazhenov, A. Gorbunov, C. Hofmann, S. Münch, A. Löffler, M. Kamp, J. P. Reithmaier, V. D. Kulakovskii, and A. Forchel, Appl. Phys. Lett. 89, 051107 (2006). https://doi.org/10.1063/1.2266231 Particularly, partial injection locking has recently been reported with micropillar lasers,88. E. Schlottmann, S. Holzinger, B. Lingnau, K. Lüdge, C. Schneider, M. Kamp, S. Höfling, J. Wolters, and S. Reitzenstein, Phys. Rev. Appl. 6, 044023 (2016). https://doi.org/10.1103/PhysRevApplied.6.044023 making them a promising alternative to conventional vertical cavity surface emitting lasers (VCSELs) as photonic neurons with cQED-enhanced properties and a so-far unparalleled integration density. In particular, for neuromorphic computing applications, spectral homogeneity across the array is the key and has recently been addressed by means of tuning the diameter of each micropillar individually, leading to notable progress in reducing the inhomogeneous broadening of fabricated micropillar arrays.5,95. T. Heuser, J. Große, A. Kaganskiy, D. Brunner, and S. Reitzenstein, APL Photonics 3, 116103 (2018). https://doi.org/10.1063/1.50506699. J. Bueno, D. Brunner, M. C. Soriano, and I. Fischer, Opt. Express 25, 2401 (2017). https://doi.org/10.1364/OE.25.002401Furthermore, the number of neurons, i.e., microlasers, in this case is of fundamental importance. In order to increase the number of photonic neurons in an energy-efficient manner, it is crucial to maximize their power efficiency and reduce their lasing threshold. Although electrically injected micropillar lasers have a low-lasing threshold,1010. S. Reitzenstein, T. Heindel, C. Kistner, A. Rahimi-Iman, C. Schneider, S. Höfling, and A. Forchel, Appl. Phys. Lett. 93, 061104 (2008). https://doi.org/10.1063/1.2969397 upscaling dense arrays of such devices requires a complex electrical contact layout and even poses limitations on lasers' rf-modulation bandwidth. Optically pumped micropillar lasers based on GaAs material systems, on the other hand, are often reported to suffer from a low power conversion efficiency (PCE) of around 3% due to the strong optical pump power absorption in the top Al(Ga)As/GaAs DBRs before it reaches the GaAs cavity where the quantum dots are located.1111. L. Andreoli, X. Porte, T. Heuser, J. Große, B. Moeglen-Paget, L. Furfaro, S. Reitzenstein, and D. Brunner, Opt. Express 29, 9084 (2021). https://doi.org/10.1364/OE.417063 As a result, despite cavity-enhanced operation, the high lasing threshold of optically pumped micropillar lasers hinders the upscaling of photonic neural networks leveraging such devices as neurons.A simple and yet long-overlooked solution to the low-PCE problem has been proposed recently, namely, to replace the conventional Al(Ga)As/GaAs DBRs with Al(Ga)As/AlGaAs DBRs to displace the absorption edge of the Al(Ga)As DBRs relative to the GaAs cavity, opening an ideal optical pumping window over a wide wavelength range.11,1211. L. Andreoli, X. Porte, T. Heuser, J. Große, B. Moeglen-Paget, L. Furfaro, S. Reitzenstein, and D. Brunner, Opt. Express 29, 9084 (2021). https://doi.org/10.1364/OE.41706312. V. A. Pammi, K. Alfaro-Bittner, M. G. Clerc, and S. Barbay, IEEE J. Sel. Top. Quantum Electron. 26, 1500307 (2020). https://doi.org/10.1109/JSTQE.2019.2929187 Figure 1(a) illustrates the propagation of pump light in a micropillar, showing for the conventional design the exponential damping of pump light due to absorption in the top DBR before reaching the optical cavity (black data), in contrast to the low-absorbing design where pump light reaches the optical cavity without such absorption (red data). The simple modification is fully compatible with the conventional fabrication environment, but to date is not commonly adopted by the micropillar community. In this work, we report on the realization and systematical study of such modified micropillar lasers with the Al0.2Ga0.8As/Al0.9Ga0.1As DBRs and stress its role to the reduction in lasing threshold.The epitaxial layers of our planar micropillar samples are grown by means of metal–organic chemical vapor deposition (MOCVD). The resonator consists of 33.5 pairs of λ/4n-thick Al0.9Ga0.1As/Al0.2Ga0.8As layers in the bottom DBR, followed by a single layer of self-assembled InGaAs QDs embedded in a λ/n-thick GaAs cavity, and 29.5 pairs of λ/4n-thick Al0.9Ga0.1As/Al0.2Ga0.8As forming the top DBR, which is capped with a λ/4n-thick GaAs layer to prevent oxidation. The InGaAs QDs are optimized to have a high density of ∼1.4×1010 cm−2 providing sufficient optical gain for lasing. Following the growth, micropillar lasers are processed by means of electron beam lithography (EBL), the details of which have been reported in our previous work.55. T. Heuser, J. Große, A. Kaganskiy, D. Brunner, and S. Reitzenstein, APL Photonics 3, 116103 (2018). https://doi.org/10.1063/1.5050669The power conversion efficiency is defined as the percentage of the pump power that is absorbed within the GaAs cavity, from where the photons are converted into excitons in the QDs which contribute to lasing. Since the PCE linearly scales with the applied pump power that is not absorbed or reflected by the DBRs, its value strongly influences the transparency threshold and the lasing threshold. As a first step, this quantity is determined via the transfer matrix method (TMM). In the calculation, absorption from the wetting layer and the QDs are omitted since the active layer thicknesses are too thin to effectively absorb the pump laser. The optical constants of the (Al)GaAs at cryogenic temperature are estimated by extrapolation and interpolation from reported values.13–1513. D. W. Jenkins, J. Appl. Phys. 68, 1848 (1990). https://doi.org/10.1063/1.34662114. D. T. F. Marple, J. Appl. Phys. 35, 1241 (1964). https://doi.org/10.1063/1.171360115. B. Monemar, K. K. Shih, and G. D. Pettit, J. Appl. Phys. 47, 2604 (1976). https://doi.org/10.1063/1.322979 Figure 1(b) shows the PCE of our modified design compared to the conventional design with 24/27 pairs Al0.9Ga0.1As/GaAs DBRs.1111. L. Andreoli, X. Porte, T. Heuser, J. Große, B. Moeglen-Paget, L. Furfaro, S. Reitzenstein, and D. Brunner, Opt. Express 29, 9084 (2021). https://doi.org/10.1364/OE.417063 Note that the conventional DBR design has six mirror pairs less than the modified design in order to provide a similar Q-factor and Purcell factor which is discussed in the supplementary material.

Separated by the optical absorption edge of GaAs (λαGaAs) around 820 nm and the optical absorption edge of Al0.2Ga0.9As (λαAl0.2Ga0.8As) around 700 nm, three optical pumping schemes can be readily identified: (i) when the pump wavelength is longer than λαGaAs, there is only negligible absorption occurring in the InGaAs wetting layer and QDs, leading to low PCE and (ii) when the pump wavelength is shorter than λαGaAs but longer than λαAl0.2Ga0.8As, the modified DBR design becomes superior to the conventional DBRs, because the modified top DBR is transparent at the laser wavelength such that the pump light can efficiently reach the GaAs cavity layer where it is then absorbed, creating excitons. One can also notice that the PCE of the modified DBR design rises with shortening pump wavelength thanks to the increasing optical absorption in the GaAs cavity, while the PCE of the conventional design decreases almost monotonously due to the strong depletion of the pump power. (iii) When the pump wavelength is shorter than λαAl0.2Ga0.8As, even the Al0.2Ga0.8As layers in the modified DBRs start absorbing, which drastically decreases the PCE. Note that absorption is not the only factor that could affect the PCE, reflectance fringes from the DBRs sideband but also plays an important role and leads to the oscillatory behavior of PCE.

Scheme (ii) clearly provides the better pumping condition, and the PCE of the modified design can potentially reach 29%, which is nearly six times higher than the 5% of the conventional design. The field propagation in both designs at commonly used pump wavelengths, such as 532, 671, and 781 nm, is visualized in Fig. 2. In the conventional design presented in Fig. 2(a), the light field is strongly absorbed in the DBR layers and decays exponentially regardless of the pump laser wavelength. In the modified design shown in Fig. 2(b), the DBRs become transparent to the 781 nm pump laser and the optical power is only absorbed within the GaAs cavity. Note that despite the hugely improved PCE, more than 70% of the pump power is either reflected or still transmitted to the bottom GaAs wafer due to the finite thickness of the GaAs λ-cavity; the details of which are shown in the supplementary material.In order to experimentally support the numerical results, pump-wavelength-dependent I/O measurements are carried out on our fabricated micropillar lasers in a typical micro-photoluminescence setup with a He-flow cryostat operated at 77 K. For the optical pumping, a 532, a 671, and a 781 nm diode pump solid state lasers (DPSSLs) along with a Ti:sapphire tunable continuous-wave laser (Ti–Sa) are integrated into the setup to investigate the three above-mentioned pumping schemes (i, ii, and iii). A variable magnification beam expander is installed in front of the lasers to carefully adjust the laser beam width, leading to a focused spot size around 1–2 μm with an optical objective of NA = 0.4 in front of the sample holder. In the detection path, a linear polarizer followed by a half waveplate are installed in front of the input slit of a high-resolution spectrometer to distinguish between the two linearly polarized modes of fundamental mode, which arises from unintentional micropillar ellipticity in fabrication.16,1716. C. Gies and S. Reitzenstein, Semicond. Sci. Technol. 34, 073001 (2019). https://doi.org/10.1088/1361-6641/ab155117. N. Heermeier, T. Heuser, J. Große, N. Jung, A. Kaganskiy, M. Lindemann, N. C. Gerhardt, M. R. Hofmann, and S. Reitzenstein, Laser Photonics Rev. 16, 2100585 (2022). https://doi.org/10.1002/lpor.202100585 In the following, they are labeled as modes 1 and 2, respectively.Figure 3 shows the optical characteristics of a micropillar laser with a diameter of 3.6 μm pumped by a 671 and a 781 nm lasers, corresponding to the absorbing scheme (iii) and the low-absorbing scheme (ii), respectively. Figures 3(a) and 3(d) depict the normalized emission spectra of mode 2 at different pump powers. The investigated structures under both pumping schemes exhibit a clear S-curve in the double log I/O plot along with drastically reduced linewidth reaching the resolution limit of around 60 μeV of our monochromator with 500 mm focal length, clearly signaturing the laser transition. By pseudo-Voigt line shape fitting,1818. A. Koulas-Simos, J. Buchgeister, M. L. Drechsler, T. Zhang, K. Laiho, G. Sinatkas, J. Xu, F. Lohof, Q. Kan, R. K. Zhang, F. Jahnke, C. Gies, W. W. Chow, C.-Z. Ning, and S. Reitzenstein, Laser Photonics Rev. 16, 2200086 (2022). https://doi.org/10.1002/lpor.202200086 we determine an experimental Q-factor of 12 200 with the dominant mode at the lasing threshold. As shown in Figs. 3(b) and 3(e), far above the lasing threshold, we observe a mode crossing behavior originating from the complex gain-competition dynamics where the dominant mode suddenly switches from mode 2 to mode 1 with further increasing pump power.19,2019. S. Holzinger, C. Redlich, B. Lingnau, M. Schmidt, M. von Helversen, J. Beyer, C. Schneider, M. Kamp, S. Höfling, K. Lüdge, X. Porte, and S. Reitzenstein, Opt. Express 26, 22457 (2018). https://doi.org/10.1364/OE.26.02245720. M. Schmidt, I. H. Grothe, S. Neumeier, L. Bremer, M. von Helversen, W. Zent, B. Melcher, J. Beyer, C. Schneider, S. Höfling, J. Wiersig, and S. Reitzenstein, Phys. Rev. Res. 3, 013263 (2020). https://doi.org/10.1103/PhysRevResearch.3.013263Our micropillar lasers operate in the weak coupling regime as conventional photon lasers based on excitonic gain provided by the integrated QDs.1616. C. Gies and S. Reitzenstein, Semicond. Sci. Technol. 34, 073001 (2019). https://doi.org/10.1088/1361-6641/ab1551 To qualitatively characterize the micropillar lasers, we fit the I/O curve by solving the laser rate equations and expressing the pump power Ppump as a function of the output power Pout,11,2111. L. Andreoli, X. Porte, T. Heuser, J. Große, B. Moeglen-Paget, L. Furfaro, S. Reitzenstein, and D. Brunner, Opt. Express 29, 9084 (2021). https://doi.org/10.1364/OE.41706321. G. Björk and Y. Yamamoto, IEEE J. Quantum Electron. 27, 2386 (1991). https://doi.org/10.1109/3.100877 PpumpA,B,β,ξ=AγβBPout1+BPout1+ξ1+BPout−βξBPout.The input scaling factor A∝(ηPCEδ)−1 links the pump rate to the measured pump power, where ηPCE is the power conversion efficiency and δ is efficiency that converts the absorbed pump power into QDs excitons which provide optical gain. The output scaling factor B factor links the intracavity photon number to the output power. ξ=n0β/γτsp is a dimensionless factor, including the exciton number n0 at transparency threshold, the spontaneous emission factor β, the cavity decay γ, and the spontaneous emission lifetime τsp. In order to unambiguously fit the equation, we take τsp= 1 ns and n0= 2.9 × 103 as have been reported for structures with nominally similar QD gain medium.1111. L. Andreoli, X. Porte, T. Heuser, J. Große, B. Moeglen-Paget, L. Furfaro, S. Reitzenstein, and D. Brunner, Opt. Express 29, 9084 (2021). https://doi.org/10.1364/OE.417063The lasing threshold, defined at the pump power when the mean photon number in the mode is unity, can be expressed as Pth=Aξ1−β+1+βγ/2β.2121. G. Björk and Y. Yamamoto, IEEE J. Quantum Electron. 27, 2386 (1991). https://doi.org/10.1109/3.100877 Comparing Figs. 3(b) and 3(e), we observe a remarkable lasing threshold Pth reduction of nine times from (263.3 ± 14.0) to (28.8 ± 1.6) μW by simply changing the pump laser wavelength from 671 to 781 nm and, therefore, entering the low-absorbing pumping scheme (ii) of the microcavity. The observed lasing threshold reduction factor is nearly identical to the ten times PCE change as estimated in Fig. 1(b), demonstrating that the reduction in pump laser absorption in the top DBRs is a leading factor to improving the pump efficiency.It is interesting to note that optical absorption in the top DBRs of conventional microcavities also leads to unwanted heating of the devices, which can be monitored by observing the redshift of the mode emission energy.2222. P. Jaffrennou, J. Claudon, M. Bazin, N. S. Malik, S. Reitzenstein, L. Worschech, M. Kamp, A. Forchel, and J.-M. Gérard, Appl. Phys. Lett. 96, 071103 (2010). https://doi.org/10.1063/1.3315869 Undesired heating not only obstructs the realization of a spectrally homogeneous micropillar laser array for neuromorphic computing, which requires a inhomogeneous broadening less than 200 μeV5,95. T. Heuser, J. Große, A. Kaganskiy, D. Brunner, and S. Reitzenstein, APL Photonics 3, 116103 (2018). https://doi.org/10.1063/1.50506699. J. Bueno, D. Brunner, M. C. Soriano, and I. Fischer, Opt. Express 25, 2401 (2017). https://doi.org/10.1364/OE.25.002401 but also hinders the study of physical effects such as high-β lasing in such microstructures. In the absorbing pumping scheme shown in Fig. 3(c), both modes redshift monotonously due to a strong thermal effect; while in the absorption-free scheme shown in Fig. 3(f), we could observe obvious effects from free-carrier dispersion, namely, carrier-induced change in the refractive index, before the thermal effects dominate. Under low pump power, the modes first exhibit a minor redshift of around 4 μeV resulting from bandgap shrinkage, and then a major blueshift around 70 μeV that can be attributed to band filling and the plasma effect.22–2422. P. Jaffrennou, J. Claudon, M. Bazin, N. S. Malik, S. Reitzenstein, L. Worschech, M. Kamp, A. Forchel, and J.-M. Gérard, Appl. Phys. Lett. 96, 071103 (2010). https://doi.org/10.1063/1.331586923. B. R. Bennett, R. A. Soref, and J. A. Del Alamo, IEEE J. Quantum Electron. 26, 113 (1990). https://doi.org/10.1109/3.4492424. H. C. Huang, S. Yee, and M. Soma, J. Appl. Phys. 67, 1497 (1990). https://doi.org/10.1063/1.345658 Notice that the underlying cause of the emission blueshift is not due to repulsive polariton–polariton interactions, which occurs only in optically triggered2525. S. Christopoulos, G. B. von Högersthal, A. J. Grundy, P. G. Lagoudakis, A. V. Kavokin, J. J. Baumberg, G. Christmann, R. Butté, E. Feltin, J.-F. Carlin, and N. Grandjean, Phys. Rev. Lett. 98, 126405 (2007). https://doi.org/10.1103/PhysRevLett.98.126405 and electrically injected polariton lasers.26,2726. C. Schneider, A. Rahimi-Iman, N. Y. Kim, J. Fischer, I. G. Savenko, M. Amthor, M. Lermer, A. Wolf, L. Worschech, V. D. Kulakovskii, I. A. Shelykh, M. Kamp, S. Reitzenstein, A. Forchel, Y. Yamamoto, and S. Höfling, Nature 497, 348 (2013). https://doi.org/10.1038/nature1203627. A. Bhattacharya, M. Z. Baten, I. Iorsh, T. Frost, A. Kavokin, and P. Bhattacharya, Phys. Rev. Lett. 119, 067701 (2017). https://doi.org/10.1103/PhysRevLett.119.067701 Under high pump power, despite the blueshift coming from increasing free carrier density, thermal effects dominate and result in a strong redshift. The high-absorbing case exhibits a redshift from 0.1 Pth to 10 Pth as high as (0.83 ± 0.13) meV, while the low-absorbing case exhibits a small blueshift of only (71.3 ± 1.8) μeV. Note that all effects do coexist in both pumping schemes, while in the low-absorbing pumping scheme the plasma effect is more dominant thanks to the weakened thermal effects. With the related free-carrier dispersion data from Huang et al.,2424. H. C. Huang, S. Yee, and M. Soma, J. Appl. Phys. 67, 1497 (1990). https://doi.org/10.1063/1.345658 we present a simple model as well to show the contributions as a function of optical pump power taking into accounts the above-mentioned effects (see the supplementary material).Additionally, in the absorbing scheme, the emission modes above lasing threshold become unstable and fluctuate due to thermal effects, leading to an inhomogeneous broadening of the laser linewidth as we can see in Fig. 3(b). In the absorption-free scheme, the modes, however, remain stable over the whole pump power range. This even allowed us to observe the linewidth broadening of the weak mode when the other stronger mode dominates and takes over lasing, highlighting an often-overlooked characteristics that one can only see in the absence of severe heating.Further pump-wavelength dependent I/O measurements are performed on the fabricated-micropillar lasers with diameters of 2.9, 3.6, and 5.4 μm for a broader pump laser wavelength range, covering all three pumping schemes to obtain comprehensive insight into the lasing characteristics. Again, the lasing thresholds are extracted via laser rate equations and are shown in Fig. 4(a). The Q-factors extracted at lasing threshold are 12 000, 12 200, and 15 000, respectively. As expected, under the wetting layer pumping scheme (iii) where the laser wavelength is longer than 820 nm, the micropillar lasers exhibit a high lasing threshold above several hundreds of microwatts at 833 nm and increases even up to milliwatts scale at 899 nm due to a weaker pump absorption by the wetting layer. Under the low-absorbing pumping scheme (ii) between 700 to 820 nm, the lasing threshold drastically reduces to tens of microwatts and even reaching a record low value of (12.8 ± 0.3) μW pumped at 708 nm for the 2.9 μm micropillar laser (see the supplementary material). Finally, when the laser wavelength is shorter than 700 nm in absorbing scheme (iii), the lasing threshold increases rapidly again up to several hundreds of microwatts under 671 nm pump laser. Additionally, although not included in the plot, the micropillars could not complete the laser transition when pumped by a 532 nm laser due to severe heating of the devices. The lasing thresholds depicted in Fig. 4(a) clearly signature an inverse bell-shape as predicted by the simulation of PCE.We also obverse that the 2.9 μm pillar, in general, shows a lower lasing threshold than the 5.4 μm pillar. This can be attributed to smaller mode volume and, therefore, higher Purcell- and β-factor of the smaller micropillar laser. The averaged extracted β-factors are (3.9 ± 1.2)%, (2.3 ± 0.3)%, and (1.0 ± 0.3)% for the pillars with 2.9, 3.6, and 5.4 μm diameter, respectively. Meanwhile, from the inverse of the input scaling factor A in the rate equations fit, we obtain the pump efficiency plotted in Fig. 4(b), which is a product of PCE and the exciton conversion efficiency δ. We could clearly observe an increase in pump efficiency in pumping scheme (ii). Note that due to the complex mode competition and crossing behavior above the lasing threshold, the extraction of the β-factor and pump efficiency becomes less accurate. Interestingly, despite a lower lasing threshold from the smaller pillar, its pump efficiency is lower than for the large diameter pillar. This behavior can be attributed to the finite pump laser spot size, which is bigger than the fundamental mode area of the small pillars, leading to a reduced fundamental mode pump efficiency (see the supplementary material).2828. G. Ctistis, A. Hartsuiker, E. van der Pol, J. Claudon, W. L. Vos, and J.-M. Gérard, Phys. Rev. B 82, 195330 (2010). https://doi.org/10.1103/PhysRevB.82.195330

In conclusion, we demonstrated that a modification of the epitaxial layer design of planar QD-microcavity structures, namely, the change in binary GaAs in the DBRs by ternary Al0.2Ga0.8As, leads to a low-absorbing pumping scheme. The reduction in threshold pump power of micropillar lasers can easily reach more than an order of magnitude, from (363.0 ± 18.5) μW pumped at 671 nm to values as low as (12.8 ± 0.3) μW pumped at 708 nm. Simulations predict that the upper limit of the associated PCE can be increased by nearly six times under ideal pumping, comparing the conventional design to our modified design. The reduced lasing threshold strongly enables scaling up of the number of micropillar lasers in applications requiring high energy efficiency. It also reduces undesired thermal heating effects, leading to more stable emission characteristics. The developed design is, therefore, highly interesting for both the fundamental study and applications of micropillar devices.

Note that the design of the devices and the measurements in this study are all conducted under 77 K to benefit from sufficiently high optical gain with a single layer of QDs. To provide sufficient optics for application-relevant room temperature operation, one can increase the carrier confinement by including AlGaAs barriers near the QD layer or by introducing stacked layers of QDs2929. H. Saito, K. Nishi, I. Ogura, S. Sugou, and Y. Sugimoto, Appl. Phys. Lett. 69, 3140 (1996). https://doi.org/10.1063/1.116808 as well as using optimized bottom DBR designs to further increase the PCE. For instance, one could additionally modify the bottom DBR design so that it would also be reflecting at the pump laser wavelength.3030. T. Elsass, K. Gauthron, G. Beaudoin, I. Sagnes, R. Kuszelewicz, and S. Barbay, Eur. Phys. J. D 59, 91 (2010). https://doi.org/10.1140/epjd/e2010-00079-6 On the other hand, to extend the low-absorbing pump window, one could further increase the Al composition in the AlGaAs alloy at the cost of a reduced refractive index contrast and, therefore, more required pairs of DBRs. Additionally, as an alternative to an epitaxially grown upper DBR, one could combine large bandgap dielectric DBRs deposited by means of plasma electric chemical vapor deposition (PECVD).3131. P. Qiu, B. Wu, P. Fu, M. Li, Y. Xie, and Q. Kan, IEEE Photonics J. 13, 1500106 (2021). https://doi.org/10.1109/JPHOT.2021.3089710 Our work provides a crucial insight into the fundamental design of micropillar lasers and paves the way to scaling up the number of photonic neurons in neuromorphic schemes based on the coupled laser array.See the supplementary material for additional information on I/O plots, optical numerical simulations, TMM simulations of reflectance and transmittance, modeling of the mode energy shift resulting from thermal effects and free-carrier dispersion, as well as diameter-dependent fundamental mode pump efficiency.

This work was funded by German Research Foundation (Nos. Re2974/20-1, Re2974/21-1, Re2974/26-1, and INST 131/795-1 FUGG) and Volkswagen Foundation (NeuroQNet II). The authors would like to thank K. Schatke, R. Schmidt, and R. Linke for the technical support on the epitaxy and processing. They would like to thank J. Große, T. Heuser, and S. Richter for assistance and discussion.

Conflict of Interest

The authors have no conflicts to disclose.

Author Contributions

Ching-Wen Shih: Conceptualization (lead); Data curation (lead); Formal analysis (lead); Investigation (lead); Methodology (equal); Software (lead); Visualization (lead); Writing – original draft (lead); Writing – review & editing (equal). Imad Limame: Investigation (supporting); Methodology (equal); Resources (lead). Sebastian Krüger: Formal analysis (supporting); Investigation (supporting). Chirag Palekar: Investigation (supporting); Methodology (supporting). Aris Koulas-Simos: Investigation (supporting); Methodology (supporting). Daniel Brunner: Funding acquisition (equal); Validation (equal). Stephan Reitzenstein: Funding acquisition (equal); Supervision (lead); Validation (equal); Writing – original draft (equal); Writing – review & editing (equal).

The data that support the findings of this study are available from the corresponding author upon reasonable request.

REFERENCES

1. D. A. Vajner, L. Rickert, T. Gao, K. Kaymazlar, and T. Heindel, Adv. Quantum Technol. 5, 2100116 (2022). https://doi.org/10.1002/qute.202100116, Google ScholarCrossref2. S. Rodt and S. Reitzenstein, APL Photonics 6, 010901 (2021). https://doi.org/10.1063/5.0031628, Google ScholarScitation, ISI3. L. He, Ş. K. Özdemir, and L. Yang, Laser Photonics Rev. 7, 60 (2013). https://doi.org/10.1002/lpor.201100032, Google ScholarCrossref4. A. Skalli, J. Robertson, D. Owen-Newns, M. Hejda, X. Porte, S. Reitzenstein, A. Hurtado, and D. Brunner, Opt. Mater. Express 12, 2395 (2022). https://doi.org/10.1364/OME.450926, Google ScholarCrossref5. T. Heuser, J. Große, A. Kaganskiy, D. Brunner, and S. Reitzenstein, APL Photonics 3, 116103 (2018). https://doi.org/10.1063/1.5050669, Google ScholarScitation, ISI6. A. Kaganskiy, S. Kreinberg, X. Porte, and S. Reitzenstein, Optica 6, 404 (2019). https://doi.org/10.1364/OPTICA.6.000404, Google ScholarCrossref7. S. Reitzenstein, A. Bazhenov, A. Gorbunov, C. Hofmann, S. Münch, A. Löffler, M. Kamp, J. P. Reithmaier, V. D. Kulakovskii, and A. Forchel, Appl. Phys. Lett. 89, 051107 (2006). https://doi.org/10.1063/1.2266231, Google ScholarScitation, ISI8. E. Schlottmann, S. Holzinger, B. Lingnau, K. Lüdge, C. Schneider, M. Kamp, S. Höfling, J. Wolters, and S. Reitzenstein, Phys. Rev. Appl. 6, 044023 (2016). https://doi.org/10.1103/PhysRevApplied.6.044023, Google ScholarCrossref9. J. Bueno, D. Brunner, M. C. Soriano, and I. Fischer, Opt. Express 25, 2401 (2017). https://doi.org/10.1364/OE.25.002401, Google ScholarCrossref10. S. Reitzenstein, T. Heindel, C. Kistner, A. Rahimi-Iman, C. Schneider, S. Höfling, and A. Forchel, Appl. Phys. Lett. 93, 061104 (2008). https://doi.org/10.1063/1.2969397, Google ScholarScitation, ISI11. L. Andreoli, X. Porte, T. Heuser, J. Große, B. Moeglen-Paget, L. Furfaro, S. Reitzenstein, and D. Brunner, Opt. Express 29, 9084 (2021). https://doi.org/10.1364/OE.417063, Google ScholarCrossref12. V. A. Pammi, K. Alfaro-Bittner, M. G. Clerc, and S. Barbay, IEEE J. Sel. Top. Quantum Electron. 26, 1500307 (2020). https://doi.org/10.1109/JSTQE.2019.2929187, Google ScholarCrossref13. D. W. Jenkins, J. Appl. Phys. 68, 1848 (1990). https://doi.org/10.1063/1.346621, Google ScholarScitation, ISI14. D. T. F. Marple, J. Appl. Phys. 35, 1241 (1964). https://doi.org/10.1063/1.1713601, Google ScholarScitation, ISI15. B. Monemar, K. K. Shih, and G. D. Pettit, J. Appl. Phys. 47, 2604 (1976).

留言 (0)

沒有登入
gif