NAMBE 2025 Session NAMBE1-MoA: Photonic Devices

We present our latest results on the growth of ErAs/III-V semiconductor materials designed for photoconductive devices operating at 1.55 μm telecom-wavelength and for hybrid emitters. Terahertz (THz) technology holds significant promise across various fields, including biomedical imaging, communications, astronomy, and spectroscopy. Bridging the "Terahertz Gap" requires new materials engineered for high-performance and reliable technology. Photoconductive switches (PCS), key components for THz pulse generation and detection, face challenges under 1.55 μm excitation. ErAs:InGaAs and other nanocomposites have been explored for this purpose, but their high electron concentration, due to the Fermi level residing in the conduction band, limits dark resistivity. To address this, we propose ErAs:[(InGaBiAs)_x(InAlBiAs)_1-x)], a digital alloy with incorporated nanoparticles, as suitable PCS material for telecom-wavelength usage¹. ErAs nanoparticles ensure sub-picosecond carrier lifetimes ensuring high temporal resolution while simultaneously pinning the Fermi level within the bandgap. The freedom offered by the InGaAlBiAs short-period superlattice enables tunable bandgap engineering—aluminum raises the conduction band edge while bismuth, incorporated through low-temperature, stoichiometric growth, maintains a bandgap suitable for 1.55 μm excitation. This approach achieves both low carrier concentration and high dark resistivity, enhancing the performance of THz photoconductive devices at telecom wavelengths. We also demonstrate a Hybrid THz emitter that integrates a sputtered Ta/CoFeB/Pt spintronic emitter and an MBE grown ErAs:GaAs photoconductive antenna (PCA) into a single device². This device enables independent and tunable excitation of two THz emitters, facilitating control over emerging THz functionalities such as elliptical polarization and pulse shaping.

[1] W. Acuna, et al., Adv. Funct. Mater. 34, 2041853, (2024).

[2] W. Wu, et a.l, Adv. Opt. Mater. 2402374, (2024).

The photonic crystal surface-emitting laser (PCSEL) is promising device for scaling power and brightness of semiconductor lasers [1]. Embedding a 2D photonic crystal near the active region of semiconductor laser defines the cavity and enables surface-emission. Not unlike distributed feedback (DFB) lasers this requires an epitaxial regrowth step, though for the PCSEL the feature sizes are smaller than those of the DFB laser and the lithographic features are closer to the active region. To operate a PCSEL in single-mode at the Γ-point a high-index contrast and appropriate symmetry of the photonic crystal are necessary. This requires control of the regrowth profile of the 2D photonic crystal. Output powers of 50W continuous-wave have been reported in GaAs-based devices operating around a wavelength of 1µm [2] utilizing metal-organic chemical vapor deposition (MOCVD). While these devices can be extended to longer wavelengths the necessary regrowth becomes challenging [3], particularly in antimonide-based devices due to the higher adatom mobility and the larger lattice constant required for the extend-short-wave infrared (e-SWIR) wavelengths. To date a maximum of 30mW from a 250µm aperture device have been reported [4]. Performance gains are expected with improved regrowth and the ability to optimize device based on high-fill factor air-void photonic crystals. Here we report on initial regrowth samples and devices exploring the molecular beam epitaxy (MBE) parameter space for devices emitting nominally at a wavelength of 2µm. Initial samples show higher uniformity and regularity in the embedded 2D photonic crystal, based on scanning electron microscope (SEM) images, than what have previously been reported. Limitations to optimization of GaSb-based devices based on the material and growth decisions will also be discussed.

[1] S. Noda, et al., “High-power and high-beam-quality photonic-crystal surface-emitting lasers: a tutorial” Advances in Optics and Photonics, 15(4), 977-1032 (2023).

[2] M. Yoshida, et al., “High-brightness scalable continuous-wave single-mode photonic-crystal laser” Nature, 618(7966), 727-732 (2023).

[3] W. Lee, et al., “Comparison of Thermal and Atomic-Hydrogen-Assisted Oxide Desorption Methods for Regrowth of GaSb-Based Cascade Diode Lasers” J. Electron. Mater. 50, 5522–5528 (2021).

[4] L. Shterengas, et al., “Photonic Crystal Surface Emitting GaSb-based Type-I Quantum Well Diode Lasers” IEEE Journal of Selected Topics in Quantum Electronics, 31(2), 1-7 (2025).

The quaternary alloy, In_xGa_1-xAs_ySb_1-y, grown lattice-matched to GaSb substrates, is a promising alternative to extended InGaAs for Short-Wave Infrared (SWIR) detection due to its tunable wavelength range from 1.7 µm to 3 µm. This flexibility is essential for advanced imaging and sensing applications, positioning InGaAsSb as a key material for next-generation infrared devices. Yet, GaSb-based devices face challenges like high background doping and surface leakage, which hinder performance. Various growth techniques and architectures, including unipolar barrier designs, have been explored in the literature to mitigate these issues. Developing an optimal InGaAsSb-based structure requires a comprehensive understanding of the material’s properties and key device performance metrics ¹. This work presents the molecular beam epitaxial (MBE) growth and characterization of the nominal composition In_0.29Ga_0.71As_0.25Sb_0.75 on a GaSb substrate. The growth temperature was estimated to be in the range of 450°C–460°C. Reflection High-Energy Electron Diffraction (RHEED) patterns confirmed that the InGaAsSb layer grows with a (4×2) surface reconstruction. Following MBE growth, the surface and structural properties of the alloy were characterized using High-Resolution X-Ray Diffraction (HR-XRD), Atomic Force Microscopy (AFM), Nomarski imaging, and photoluminescence (PL). Two structures were fabricated: a bulk structure with a 500 nm thick InGaAsSb layer and a homojunction p-i-n structure with a 1 µm thick intrinsic layer. The bulk InGaAsSb exhibited a lattice mismatch of 0.164% and a surface roughness of 1.02 A, indicating good crystalline quality. The PL peak was observed at approximately 2.33 µm. However, HR-XRD scans of the p-i-n structure revealed evidence of spontaneous superlattice formation in the active region, which is currently under investigation. This behavior may be attributed to growth near the miscibility gap (x~0.3). PL measurements showed no detectable peak shift despite the presence of the superlattice. Ongoing studies focus on optoelectronic properties like minority carrier lifetime and background doping to assess the superlattice's impact on performance.

1. K. Mamić, L. A. Hanks, J. E. Fletcher, A. P. Craig and A. R. J. Marshall, Semiconductor Science and Technology 39 (11), 115002 (2024).
2. I. P. Ipatova, V. A. Shchukin, V. G. Malyshkin, A. Y. Maslov and E. Anastassakis, Solid State Communications 78 (1), 19-24 (1991).

Lasers based on self-assembled quantum dot (QD) gain media have attracted considerable attention due to their low sensitivity to operating temperature and low threshold current densities. Several laser architectures based on InAs QD have been demonstrated with excellent performance, e.g. edge emitting lasers (EELs), including distributed feedback(DFB) lasers, vertical cavity surface emitting lasers (VCSELs) or vertical external cavity surface emitting lasers (VECSELs). In this study, we demonstrate a photonic crystal surface emitting laser with a QD active region (QD-PCSEL). The PCSEL fabrication is based on epitaxial regrowth,which facilitatesthe photonic crystal (PC)to belocatedin the laser’s waveguide near the upper clad layers. The structure is grown using elemental source molecular beam epitaxy (MBE). In the first epitaxial step the bottom AlGaAs cladding layer and the GaAs waveguide including the QD active region are grown. Subsequently, the wafer is removed from the MBE reactor and the PC is fabricated into the GaAs waveguideusing electron beam lithography (EBL) patterning and etchingby inductively coupled plasma (ICP) dry etching.This is followed by a second epitaxial step, the regrowth, where the top AlGaAs cladding layer and a top contact layer are grown on the sample. One of themain challenges is thesurface preparation including the removal of the native oxide before the regrowth. This is accomplished by an acid etch prior to loading the sample into the vacuum chamber and a thermal surfacetreatment witharsenic supplyprior to growth. The thermal step at >600^oC can alter the QD gain medium, i.e. cause a blueshift and narrowing of the emission spectrum of the self-assembled QDs [1-3]. Alternatively, the regrowth can be performed by MOCVD at comparable high temperatures, where hydrogen is available for surface preparation.In our study the InAs QDsare embedded in a InGaAs quantum well (QW),a dot-in-a-well (DWELL) design. Our experiments indicate that the DWELL active region remains unchangedduring the regrowth process, i.e. there is no significantalteration to the emission wavelength. This is key to the realization of this laser. We present characterization dataof the QD-PCSEL and compare regrowth options.

1. Photonics2018, 5(3), 27

2. Journal of Lightwave Technology, vol. 35, no. 20, 4547-4552, 2017

3. Crystal Growth & Design 2021, 21, 6,3521-3527

The integration of lasers with silicon photonics is required to produce highly-efficient low-footprint photonic integrated circuits (PICs) for applications in data communication, LIDAR, and biosensing. Monolithic integration through the direct growth of III-V semiconductor materials on patterned silicon photonics wafers would enable large-scale production of PICs by utilizing 300 mm silicon wafers and eliminating costly III-V substrates from the process. However, growth on patterned silicon photonics wafers introduces new challenges in material quality, coupling efficiency, and growth uniformity. In this work, we report a two-step growth approach using metal-organic chemical vapor deposition (MOCVD) and molecular beam epitaxy (MBE) to create InAs quantum dot lasers and photodetectors coupled to silicon nitride waveguides on foundry-processed silicon photonics wafers.

By combining MOCVD’s high-quality selective III-V buffers on silicon with MBE’s precise control of growth parameters for quantum dots, we can mitigate many of the difficulties associated with III-V growth on patterned silicon photonics wafers. Photonic integrated circuits with silicon nitride couplers, silicon nitride waveguides, silicon ring resonators, and silicon nitride distributed Bragg reflectors all embedded in silicon dioxide were fabricated at AIM Photonics on 300 mm silicon-on-insulator wafers. Pockets were etched through the silicon dioxide leaving the silicon nitride couplers exposed on the silicon dioxide sidewalls. Anti-phase domain-free GaP, a GaAs buffer, and InGaAs strained layer superlattices were grown by MOCVD to reduce the defect density. Selective MOCVD growth was used in the exposed silicon pockets to prevent the deposition of polycrystalline III-V on the sidewalls containing the nitride couplers. A separate confinement heterostructure laser stack was then grown by MBE on diced 3.2 x 2.6 cm coupons with an InAs quantum dot active region aligned to the nitride couplers. This material was then fabricated into 4 mm long and 4 um wide ridge waveguide devices with contacts for electrical bias.

Using this method, we demonstrate waveguide-coupled lasers and photodetectors monolithically integrated with foundry-processed silicon photonics wafers. Lasers operate in the O-band for data communication applications and show mW-scale output powers measured in-fiber when coupled out of the chip. Photodetectors were characterized as a function of bias voltage when excited by 1.3 um input laser light coupled from off-chip. The photodetectors show a highly linear response with sub-nA dark current. Together with the integrated laser results, this represents a major step forward in creating scalable PICs with on-chip III-V devices.

This material is based on research sponsored by Defense Advanced Research Projects Agency (No. HR0011-20-C-0142) and Air Force Research Laboratory under AIM Photonics (agreement number FA8650-21-2-1000). The U.S. Government is authorized to reproduce and distribute reprints for Governmental purposes notwithstanding any copyright notation thereon. The views and conclusions contained herein are those of the authors and should not be interpreted as necessarily representing the official policies or endorsements, either expressed or implied, of the United States Air Force, the Air Force Research Laboratory, or the U.S. Government.

Epitaxial III-V/Si tandem solar cells offer a path to high efficiency at lower cost¹. 1.7 eV GaAs_yP_1-y (GaAsP) is an optimal top cell for Si-based tandems, leveraging a GaP nucleation and a GaAsP graded buffer. Despite advances in dislocation control in GaAsP subcells on Si with two-step growth, strained layer superlattices, tailored doping profiles, and compositional grading², high threading dislocation density (TDD) still limits minority carrier diffusion lengths. Thinner GaAsP absorbers can mitigate challenges with low diffusion lengths but incur optical losses due to incomplete absorption. A distributed Bragg reflector (DBR) resolves this limitation by reflecting transmitted photons back into the absorber, enhancing effective optical path length and carrier collection while enabling thinner GaAsP. Here, we demonstrate the first GaAsP single-junction (1J) cell on Si with a 1.46% efficiency boost from a 20-period Al_0.20Ga_0.80AsP/Al_0.80Ga_0.20AsP DBR, highlighting photon management for high-performance GaAsP/Si tandem solar cells.

We grew GaAsP solar cells via MBE on GaP/Si (001) templates with a 500 nm p-GaP buffer, a 1.8 µm p-GaAsP graded buffer and a 750 nm GaAsP absorber. Compared to our previous best GaAsP 1J devices³, we reduced the cell thickness by 2.2 µm, allowing >2 µm for the DBR without exceeding the thermal expansion cracking threshold of ~6 mm.

The 20-period AlGaAsP DBR calibration shows a 95.7% peak reflectance with a 55 nm bandwidth, closely matching calculations (97.2%, 60 nm). Electron channeling contrast imaging reveals similar TDD (~7-8×10⁶ cm^–2) in cells with or without a DBR, indicating negligible impact from DBR growth. External quantum efficiency (EQE) measurements show enhanced carrier collection at 660-730 nm due to DBR reflection, increasing EQE-calculated short-circuit current density (EQE-J_SC) by 1.42 mA/cm². EQE at 450-550 nm was lower than expected, likely due to elevated window/emitter interface recombination loss. Nevertheless, the DBR cell exhibits a 1.46% absolute efficiency gain due to increased J_SC and a 12 mV open-circuit voltage (V_OC) boost. Additionally, V_OC and fill factors closely match our previous best n+/i/p cells³.

This work demonstrates 15%-efficient GaAsP solar cells on Si with an AlGaAsP rear DBR, where DBR reflection improves near-bandedge EQE and J_SC. Our results show the promise of DBRs for high-efficiency, low-cost GaAsP/Si tandems, with potential for enhanced radiation hardness in space applications with a thin absorber.

[1] J. F. Geisz et. al., Semicond. Sci. Technol. 17 (2018). [2] J. T. Boyer et. al., Cryst. Growth Des., 20 (2020). [3] S. Fan et. al., Cell Rep. Phys. Sci., 1 (2020).

Solar cells, along with other optoelectronic devices such as photodiodes, light-emitting diodes (LEDs), and lasers, rely on p-n junctions to either collect photogenerated carriers in absorber regions or inject carriers into the active region. The use of p-n junctions in solar cells is advantageous because the electric field within the device yields the efficient extraction of photogenerated carriers for high power conversion. The doping levels in the p- and n-regions set the built-in voltage (V_bi) across the device, which in turn limits the maximum achievable open circuit voltage (V_oc). A higher V_bi is preferred as it creates a stronger electric field in the absorber region, reducing the transit time for photogenerated carriers, enhancing their collection at the contacts, and improving overall conversion efficiency.

Solar cells with a V_oc exceeding 1 V have been developed using an n-type CdTe/MgCdTe double-heterostructure (DH) absorber with an n-type indium tin oxide (ITO) transparent layer forming a hole-selective contact. No p-type doping is used in the devices. The ITO layer is directly deposited atop the MgCdTe barrier layer. Charge transfer from the ITO and the n-type absorber to the interface states between the ITO and the top MgCdTe barrier results in a Fermi level near the valence band edge of the CdTe layer. This charged interface functions effectively as a "p-region," achieving a V_bi of up to 1.01 V. A straightforward model is proposed to explain the relationship between Mg composition in the barrier layer and the corresponding V_bi. The modeling results are in good agreement with experimental results obtained from capacitance-voltage (C-V) measurements.X-ray photoelectron spectroscopy (XPS) measurements on samples having a MgCdTe top barrier layer with different Mg compositions confirm the correlation between the interface Fermi-level position and the observed V_bi. The devices, tested by the National Renewable Energy Laboratory (NREL), show a V_oc over 1 V (1.0164 ± 0.0026 V), consistent with the V_bi derived from C-V measurements, thereby confirming that V_bi limits V_oc in these devices. The integrated J_sc from EQE is 24.88 mA/cm², attributed to the absence of an absorptive hole-selective contact layer, such as a-Si:H. The efficiency of this device reaches 17.3%.

This innovative approach to addressing the challenge of low p-type doping in CdTe solar cells can potentially be applied to other interfaces involving semiconductors, transparent conductive oxides (TCOs), and certain metals, offering broad applications not only in photovoltaics but also in photodetectors, LEDs, and lasers.