Influence of the carrier wafer during GaN etching in Cl2 plasma

Abstract

In this study, we have performed a thorough characterization of the GaN surface after etching up to 100 nm in Cl2 plasma under various bias voltages and according to the carrier wafer used (Si, SiO2, Si3N4, and photoresist). The objective of this article is to evaluate the etch damage and contamination of the GaN surface when materials with other chemical nature are present during etching. The effects of etching conditions on surface morphology and chemical compositions of etched GaN films are studied in detail using XPS and AFM measurements. To this aim, a universal methodology is proposed to estimate accurately by XPS the stoichiometry of the GaN surface exposed to reactive plasmas when only an Al Kα x-ray source is available. The results indicate that the GaN etching mechanisms are very sensitive to the chlorine radical density present in the plasma, the latter being strongly influenced by the carrier wafer. Substrates that are more chemically reactive with Cl2 plasma such as silicon or photoresist compared to SiO2 or Si3N4 will lead to a greater loading of atomic chlorine, which in turn will lead to lower GaN etch rates. Moreover, the GaN surface contamination will depend on the etch by-products ejected by the carrier wafer. The GaN surface exposed to Cl2 plasma shows a Ga-depleted surface because of the more important reactivity of Cl with Ga rather than N, except in the SiO2 carrier wafer case. In this latter case, the formation of Ga–O bond limits the Ga removal. Regarding the surface roughness, it seems that the contaminants play a little role in the roughness formation except for the oxygen released by the SiO2 carrier wafer. On the other hand, the surface roughness evolution is clearly driven by the chlorine radical flux reaching the GaN surface. At low bias voltage, a preferential crystalline orientation etching driven by the Cl radicals leads to the formation of hexagonal shaped defects that are associated to screw-type threading dislocations already present in the pristine GaN material. At higher bias, the enlargement of the defects is limited, leading to a very low surface roughness value but to amorphized surfaces.

Influence of the Cl2 etching on the Al2O3/GaN metal–oxide–semiconductor interface

“…Compared to the non-etched sample, the etched samples exhibit a flat band voltage shift toward negative values. This could be attributed to the amorphization of the GaN surface or the degradation of the surface stoichiometry after such plasma conditions as reported previously 7 .…”

“…However, the etching step modifies the GaN surface composition (e.g. N-depleted surface, etch residues) and the surface morphology 7,8 . The deterioration of the GaN surface necessarily leads to a poor dielectric/GaN interface quality.…”

“…In a previous study, we thoroughly investigate the GaN surface modifications after Cl2 etching under various bias voltage and according to the carrier wafer used (Si, SiO2, Si3N4, photoresist), by XPS and AFM analyses 7 . We showed that the GaN surface is differently damaged and contaminated, given that each carrier wafer produces etch byproducts of different chemical nature that can redeposit on the GaN surface.…”

Selective area doping of GaN toward high-power applications

“…Well-defined trenches with vertical sidewalls are routinely obtained when the dry-etching technique is optimized. Dry etching has been successfully implemented with plasma techniques: magnetron reactive ion etching (RIE) [7,8], ion beam etching [9][10][11], plasma etching [12][13][14], electron cyclotron resonance etching [15][16][17], RIE [18,19], inductively coupled plasma (ICP) [20,21], and ICP-RIE [22,23]. All of these may cause plasma-induced damage (PID) [24][25][26][27][28] to and contamination of the material under treatment [29].…”

Plasma induced damage on AlGaN/GaN heterostructure during gate opening for power devices

“…The angles of 23.75° and 76.25° correspond to photoelectrons escaping from the bulk (about 8 nm) and near-surface (about 2 nm), respectively. 15 As the AlGaN layer thickness is about 4 nm, we estimate from IMPF calculations that the angle of 53.25° is the most appropriate to probe the whole AlGaN layer with the minimal signal from underneath GaN. 15 In order to investigate the reactive layer composition formed after plasma exposure, and surface properties change, the angle of 76.25° is used and the values are compared to the reference ones at 53.25°.…”