Charge movement in a GaN-based hetero-structure field effect transistor structure with carbon doped buffer under applied substrate bias

Abstract

Charge movement in a GaN-based hetero-structure field effect transistor structure with carbon doped buffer under applied substrate bias. Journal of Applied Physics, 118(21), [215701]. DOI: 10.1063/1.4936780Publisher's PDF, also known as Version of record University of Bristol -Explore Bristol Research General rightsThis document is made available in accordance with publisher policies. Please cite only the published version using the reference above. Full terms of use are available: http://www.bristol.ac.uk/pure/about/ebr-terms Charge movement in a GaN-based hetero-structure field effect transistor structure with carbon doped buffer under applied substrate bias Charge trapping and transport in the carbon doped GaN buffer of a GaN-based hetero-structure field effect transistor (HFET) has been investigated under both positive and negative substrate bias. Clear evidence of redistribution of charges in the carbon doped region by thermally generated holes is seen, with electron injection and capture observed during positive bias. Excellent agreement is found with simulations. It is shown that these effects are intrinsic to the carbon doped GaN and need to be controlled to provide reliable and efficient GaN-based power HFETs.

Impact of buffer charge on the reliability of carbon doped AlGaN/GaN-on-Si HEMTs

“…All epitaxial layer structures contain 800nm of intentionally carbon doped GaN buffer with a 250nm undoped channel region grown on various thicknesses of graded AlGaN:C SRL. The graded AlGaN layer, starting as pure AlN at the substrate inter-face transforming to pure GaN at the GaN:C interface, was chosen to not introduce any heterointerface between GaN:C and substrate that might lead to accumulation of free charges and/or hinder vertical charge flow similar to [20]. Two experiments were undertaken: a drain transient measurement to measure dynamic R ON and substrate bias ramp to characterize buffer charging and vertical leakage.…”

“…It is clear that CC in these devices mostly results from charge storage in deep levels in the buffer, with the difference in CC between Fe and C doping reported to be the result of their relative energy levels, respectively pinning the Fermi level in the upper and lower halves of the bandgap [17] - [22]. Monitoring the substrate bias dependence of the 2DEG current, and its dispersion as the ramp-rate and temperature are varied, allowed a model for the transport within each layer within the buffer to be constructed [18] - [20]. The structures investigated in this work generally consist of a AlN nucleation layer grown on a p-type Si substrate, followed by a superlattice or graded AlGaN layer to compensate the lattice mismatch between the substrate and the GaN.…”

Control of Buffer-Induced Current Collapse in AlGaN/GaN HEMTs Using SiNx Deposition

“…A full description of charge trapping in the HEMT will also depend on the horizontal transport. The C:GaN layer is known to be weakly p-type, [20,21] so to understand the charge storage in this layer we must consider the vertical structure of the device [15][16][17][18]. Figure 3 (Fig.…”

“…The technique is especially sensitive to changes in the resistivity of the UiD (Unintentionally Doped) channel and C:GaN [14,15]. Interpretation is discussed in more detail in [11,[16][17][18]. TLM structures with a contact gap of 35 µm were used and the substrate was swept to −600V and back to 0V at 9.2V/s while a source to drain voltage of 1V was applied.…”

Lateral Charge Transport in the Carbon-Doped Buffer in AlGaN/GaN-on-Si HEMTs

“…Substrate bias experiments provide an excellent tool to study charge trapping and transport in the buffer and effectively distinguish surface and bulk induced current collapse [14][15][16]. Monitoring the substrate bias dependence of the channel conductivity, and its dispersion as the ramp-rate and temperature are varied, allowed a model for the transport within each layer within the buffer to be constructed [13,[17][18][19]. Substrate bias ramps have been used to link buffer leakage in the upper part of the epitaxy to dynamic RON dispersion [20,21].…”