Events and Announcements

Degradation of far-ultraviolet light emitting diodes on AlN substrate

3. Optical devices
Shashwat Rathkanthiwar¹ , Maki Kushimoto², Yudai Shimizu³, Kazutada Ikenaga³, Mayank Bulsara³, Keitaro Ikejiri³, Hiroshi Amano⁴, Leo J. Schowalter^1
1 Lit Thinking, Orlando, Florida 32826, USA
² Graduate School of Engineering, Nagoya University, Aichi 464-8603, Japan
³ Taiyo Nippon Sanso, Innovation Unit, Yokohama, Kanagawa 220-8561, Japan
⁴ Center for Integrated Research of Future Electronics, Institute of Materials Research and System for Sustainability, Nagoya University, Nagoya, 464-8601, Japan

Abstract:
The safety and efficacy of far-UVC radiation (<240 nm) for indoor pathogen inactivation has spurred research on far-UVC light sources.  While nitride LED technology promises to replicate the advantages of visible and near-UV LEDs like compact size, low operating power, and tunable wavelength, a major challenge is the sharp decrease in efficiency and lifetime (at practical current densities) at shorter wavelengths (high Al content). Perhaps this degradation is linked to the increased point defect incorporation during high Al-content AlGaN growth or to a greater susceptibility of point defects (or point defect complexes) to be activated by high current at high Al composition. The precise mechanism is unknown, and, in this study, we are investigating the impact of multi-quantum well (MQW), electron blocking layer (EBL), and graded-AlGaN hole injection layer (HIL) variations on 240-nm LED lifetimes. The LEDs were grown on AlN substrates using a low-pressure, resistive-heated, horizontal-flow metalorganic chemical vapor deposition (MOCVD) which achieved step-flow morphology and pseudomorphic growth. Cross-sectional transmission electron microscopy corroborated the high-quality epitaxial growth. The Si-doped Al_0.79Ga_0.21N n-contact layer exhibited a sheet resistance of 405 Ω/□. TLM contacts on the p+ GaN contact layer showed a linear I-V behavior confirming an Ohmic contact formation. LEDs exhibited a sharp electroluminescence with FWHM of ~11 nm. Accelerated constant-current degradation tests [1] were conducted in the range of 0.7 to 4 kA/cm² (corresponding to a forward voltage ranging from 7.9 to 11.3 V, respectively, at the start of the test). A 12-fold reduction (from 520 to 42 seconds) in L50 lifetime (50% drop from the initial output power) was observed for the 6-fold increase in current density. Notably, increasing the HIL start composition from 90% to 95% Al led to 2.5 and 5 times decrease in the L50 lifetime at 0.7 and 4 kA/cm², respectively, which supports the hypotheses that point defects at higher Al content are susceptible to higher degradation rates for far-UVC LEDs.

[1] Zhang et al., physica status solidi (a), 221 (2024) 2300946.

Advanced High-Flow-Velocity Horizontal MOCVD Technology for Nitride Semiconductor Growth
Keitaro Ikejiri¹ , Yudai Shimizu¹, Mizuki Yamanaka¹, Kenichi Eriguchi¹, Kazutada Ikenaga¹, Hiroki Tokunaga^1
1 TAIYO NIPPON SANSO Corporation, 10 Okubo, Tsukuba, Ibaraki 300-2611 Japan

Abstract:

The expansion of nitride semiconductor applications, such as LEDs, lasers, and high-power electronics, requires MOCVD systems that can precisely control crystal growth while maintaining high throughput. In this context, the MOCVD process has conflicting requirements. For high-quality GaN or high-indium-content InGaN deposition, relatively high pressures are necessary. However, increasing pressure leads to enhanced parasitic growth, resulting in deteriorated crystal quality and uniformity. Conversely, lower pressures improve the thickness and composition uniformity of AlGaN and control the carbon concentration in GaN. The optimal MOCVD system for flexible control of nitride semiconductors must handle pressures from low to atmospheric while maintaining high-flow velocities through narrow channels. The challenge is to meet these requirements effectively in both R&D and mass production. Taiyo Nippon Sanso’s MOCVD system achieves this with a high-flow-velocity system using triple gas injectors in a horizontal reactor with a flow channel height under 10 mm, and a gas control system operating from low to atmospheric pressure. In addition, zone-divided resistive heating technology allows the system to accurately regulate temperatures even for large diameter wafers as well as wafers that deform into concave or convex shapes during high-temperature processes.

The system’s effectiveness was demonstrated with AlGaN growth. Using the SR4000 MOCVD system for a single 4-inch substrate, we controlled the Al composition in AlGaN by adjusting the TMA to total MO supply ratio (TMG + TMA) linearly. This method maintained a favorable Al composition distribution (in-plane max-min ≤ 1.5%) and thickness distribution (in-plane max-min/average < 5.0%) for 40% to 80% Al composition. The same approach applies to the UR26K MOCVD system, a large-scale production system for 6 x 8-inch substrates. Optimizing growth conditions achieved an Al composition distribution (in-plane and inter-plane max-min ≤ 0.2%) and thickness distribution (in-plane and inter-plane max-min ≤ 1.0 nm, average: 18.1 nm) in the AlGaN barrier layer of AlGaN/GaN HEMT on Si. We will also discuss the flow channel effectiveness using numerical simulations and InGaN crystal growth characteristics.

ABSTRACT

Ultra-violet light emitting diodes (UV LEDs) and lasers based on the III-Nitride material system are very promising since they can enable compact, safe, and efficient solid-state sources of UV light for a range of applications. The primary challenges for UV LEDs are related to the poor conductivity of the p-AlGaN layers, hole injection (p-contact), and the low light extraction efficiency of the LED structures. In this abstract, we discuss the exploration of metal-semiconductor junctions that enhance both light extraction efficiency and hole injection.

Tunnel junction (TJ)-based UV LEDs provide a distinct and unique pathway to eliminate several challenges associated with UV LEDs [1-4]. Metal semiconductor tunnel junctions (MS TJs) have been demonstrated earlier with efficient injection into p-AlGaN [5]. The negative polarization charges at the GaN (InGaN)/AlGaN interface create the necessary band bending to facilitate carrier tunneling. This concept is similar to a traditional tunnel junction, but with metal replacing the top n-AlGaN layer. Eliminating the top n-AlGaN layer offers advantages for MOCVD growth, as surface activation is now possible. Efficient tunneling between the metal and p-AlGaN can be achieved using a polarization-engineered layer, such as GaN (InGaN), with an optimized thickness, and the use of UV-reflective metals like Aluminum are particularly promising since they offer a method to achieve highly reflective UV LEDs.

In this work, we discuss the design and demonstration of all-MOCVD grown metal-semiconductor tunnel junction UV LEDs. We then show how the design of the metal can greatly impact the hole injection and light extraction/reflectivity at the metal/semiconductor interface. Al-based metal/semiconductor junctions have excellent reflectivity, but show poor hole injection/p-contact. However, Ni-based metal/semiconductor junctions have poor reflectivity but have excellent hole injection. We show that hybrid ultra-thin Ni/Al layers can give excellent reflectivity while reducing the voltage drop significantly. Using this approach, we show a 43% improvement in the peak external quantum efficiency, and low on-resistance and operating voltage for ultrathin Ni/Al/GaN/AlGaN metal-semiconductor junctions.

The epitaxial structure was grown at TNSC with a Taiyo Nippon Sanso SR4000HT MOCVD reactor. It consists of an active region with three pairs of 1.7 nm Al0.42Ga0.58 N quantum wells (QWs) separated by 1.25 nm Al0.5Ga0.5N quantum barriers. The tunnel junction is composed of a 6 nm p++ Al0.5Ga0.5N layer (doped with Mg at which is capped with a 4 nm GaN layer. A control sample was grown with a similar structure, except for the tunnel junction which was replaced with a 50 nm thick p-GaN layer to represent a standard LED. The p-type layers were activated through rapid thermal annealing at for 28 minutes in N2. Three different metal contacts were deposited to the MS TJ sample to make top contact: Ni(20 nm)/Au, Ni(1 nm)/Al(100 nm)/Ni/Au and Al(100nm)/Ni/Au. It has been shown that 1 nm Ni contact improves the reflectivity from 30% to 60% at 290 nm[6]. The improved reflectivity in addition to the thinner GaN layer leads to less absorption and better light extraction.

The composition and thickness of each layer was extracted from the profile for both the control sample and MS tunnel junction sample. Current-voltage characteristics show a voltage drop of 8.3 V,11.8,15.3 V at 20 A/cm2 for the Ni, hybrid Ni/Al and Al contacts respectively. The increase in voltage drop is due to the different work function Wm of the metals. Previous work has shown that thin 1nm Ni has a lower Wm than thicker Ni [7]. Hence it has a slighter higher schottky barrier height than the Ni based contact. All of the samples showed a low on-resistance of 5×10-3 ohm.cm2 indicating efficient tunneling in the MS TJ structures. Replacing Ni with the reflective hybrid metal contacts results in a 43-57% increase in the peak EQE. The electroluminescence spectra of all samples show a peak wavelength emission shifting from 296 nm to 292 nm as the current density increases from 2 A/cm2 to 200 A/cm2.

In summary, we have successfully demonstrated metal semiconductor tunnel junctions for MOCVD-grown UV LEDs. By depositing hybrid Ni/Al contacts, low voltage drop, resistance and higher EQE is achieved. The results of this work provide an alternative method for reducing the absorbing layer thickness while improving the external quantum efficiency. Future works involve optimizing the polarization engineered interlayer and doping/thickness of the p++ AlGaN.

This work was funded by ARO DEVCOM Grant No. W911NF2220163 (UWBG RF Center, program manager Dr. Tom Oder)

References:

1. Zhang, Y., et al.,Applied Physics Letters, 2016. 109(12).

2. Pandey, A., et al., Photonics Research, 2020. 8(3).

3. Fan Arcara, V., et al., Journal of Applied Physics, 2019. 126(22): p. 224503.

4. Nagata, K., et al., Applied Physics Express, 2021. 14(8): p. 084001.

5. Zhang, Y., et al., Applied Physics Letters, 2017. 111(5).

6. Kneissl, M. and J. Rass, 2016, Springer.

7. Grodzicki, M., et al., Applied surface science, 2014. 304: p. 24-28.