Name
Application of photoexcited muon spin spectroscopy to study excess charge carrier lifetimes in 4H-SiC epilayers
Description

The availability of ultra-thick (>100 μm) 4H-silicon carbide (SiC) epitaxial layers has already enabled high voltage (> 10 kV) devices, such as diodes [1], insulated-gate bipolar transistors (IGBTs) [2] and thyristors[3] for applications such as HVDC grid level power transfer. A requirement for good performance in these high voltage devices is a high carrier lifetime τeff (>1 µs) throughout the epitaxial thickness. This material system is typically grown with a τeff around 1 µs, which has been shown to be limited by the Z1/2 defect level. τeff enhancement is typically performed by thermal oxidation or carbon implantation. However, the enhancement throughout the drift thickness is diffusion limited due to the low diffusion coefficients of carbon and oxygen in SiC. Common carrier lifetime measurement techniques such as microwave-detected photoconductance decay (µPCD) only probe the near-surface region of the epilayer. Hence they provide only limited information on thick layers especially after enhancement. Here, we present a technique to assess the bulk quality of thick layers directly. Photoexcited muon spin spectroscopy (photo-µSR) has been suggested as a method to study the quality of crystalline silicon by assessing charge carrier lifetime [4]. It is based on the implantation of (anti)muons into the sample and analysis of the positrons emitted upon their decay. Muons can be implanted fully spin polarized and when deposited into semiconductors they interact with the electron gas to form the hydrogen isotope muonium. Upon interaction with the (random) electron spin, the initial muon spin is either maintained or lost, whenever muonium is formed or the electron is exchanged. This impacts the direction of positron emission, and our analysed quantity is the rate of depolarization after a muon implantation pulse. This depolarization rate scales with electron concentration, which can be influenced by illumination. Variation of illumination intensity or the delay between pump (laser) and probe (muons) allows analysis of carrier dynamics and inference of carrier lifetime. In this work we applied both approaches to the investigation of 100 µm thick epitaxial 4H-SiC layers after various thermal pretreatments. Epitaxial layers were grown at Warwick on 100 mm-diameter, 4° off-axis, SiC wafers using an LPE ACiS M8 chemical vapour deposition reactor. Low doped (< 2e1014cm-3) 4H-SiC homoepitaxial layers were grown at 1550 °C at a nominal thickness of 100 µm and using Trichlorosilane (HCl3Si, TCS) and Ethylene (C2H4) with a C/Si ratio of 0.796.We examined four distinct samples with photo-µSR: one in its untreated state (labelled as-grown), and three samples (labelled Ox-S1, Ox-S7, and Pass-S4) subjected to post deposition treatments. After RCA cleaning, thermal oxidation was conducted on all three samples at 1400 °C in a 1:4 O2:Ar atmosphere for 4 h in a high-temperature furnace, followed by a 10% HF SiO2 etching treatment for 5 min. The Ox-S7 sample underwent this process again followed by annealing at typical post-implantation activation temperatures (1750 °C for 4 h in Ar ambient). Sample Pass-S4 underwent phosphorus pentoxide (P2O5) deposition at 1,000°C after a dilute nitrous oxide (N2O) thermal growth [5]. Both samples underwent a 10 % HF etching treatment remove the SiO2 layers. τeff,PCD was assessed prior to and after muon exposure via µPCD in a custom Freiberg Instruments spotMDP setup. We condense the photo-µSR measurements performed on the samples to the total muon spin asymmetry decay difference between dark and illuminated measurements, as illustrated in Fig.1(b). This quantity encompasses the impact of carrier density Δn on the muon spin asymmetry. We compare this quantity to numerical simulation of the experiment to extract τeff,µSR. Even with simple assumptions (e.g. negligible interface recombination) we achieve good agreement between experimental data and simulation, as shown in Fig.2. We compare our results to µPCD measurements before and after muon exposure and get an overall good agreement as seen in Fig.3. Due to the surface-near signal detection, τeff,PCD is not very representative of deeper regions of the epitaxial layer. The photo-µSR method on the other hand is sensitive to what region or depth the muons are implanted in, which can be influenced through the use of degrader foils, as discussed e.g. by Murphy et al. [6]. This allows to assess τeff of epilayers with more practically relevant weighing compared to µPCD.

Speakers
Tim Niewelt - University of Warwick
Date
Friday, October 4, 2024
Time
2:20 PM - 2:40 PM
Location Name
Room 305
Track
Intrinsic Properties