Plasma Surface Preparation of III-V Materials: Quasi In-Situ XPS Measurements and Integration Guidelines

Abstract

Obtaining a competitive III-V laser made with a CMOS-compatible process is a major issue in the framework of Silicon photonics. The choice of the material used to elaborate an ohmic contact for III-V substrate is quite limited and fabrication processes need to be developed [1, 2]. In this way, the surface preparation of the substrate before any deposition is the first step necessary towards success [3]. The native oxides need to be removed and the surface should not be damaged. A wet cleaning is used to remove the oxides and contaminants on the substrate surface, but the oxide regrowth occurs during the vacuum break undergone between the wet cleaning and the deposition reactor. Therefore, in-situ plasma cleaning is a mandatory complement in order to remove the substrate oxides before the deposition [3]. Helium and Argon plasma treatments have been used in this study on InGaAs and InP materials. Quasi in-situ analysis is ensured with the use of a vacuum wafer carrier [4]. Inbulk, InOx, InIn-In, Pbulk and POx concentrations have been extracted from In 3d5/2 and P 2p3/2 core level energy regions [5, 6]. Figure 1 shows the parallel angle-resolved XPS (pAR-XPS) measurements of the P 2p energy region of an InP substrate after an He plasma. Transport to the XPS have been completed with and without the use of the vacuum wafer carrier. Measurements have been performed at a collection angle of 55 ± 5° corresponding to the middle range (roughly 2-3 nm). Pbulk and Pox proportions correspond to the sum of the two doublets 3/2 and 1/2 of the phosphorus core level energy region respectively for the P-In and P-O bindings. Pox proportion is higher when an air vacuum break of 30 minutes occurs (Fig.1) whereas this concentration is almost zero using the vacuum wafer carrier. Thanks to the use of quasi in-situ XPS analysis, the removal of Pox (Fig.1) on InP is clearly demonstrated which was not the case in a previous study where an air vacuum break was realized between the surface treatment and the XPS analysis [3]. Thus, quasi in-situ characterization improves the accuracy of the results notably by limiting the oxides regrowth occurring during the wafer transfer (Fig. 1). Figure 2 and 3 show a summary of pAR-XPS measurements realized on InP samples after various plasma treatments. XPS spectra have been collected at angles of 25 ± 5° and 75 ± 5°, corresponding to photoelectrons escaping from the bulk surface (roughly 10 nm) (Fig.2) and the surface (roughly 1 nm) (Fig.3) . Two Argon plasma conditions were used. “Ar plasma soft” corresponds to the less powerfull recipe. InP substrate “as-received” was measured to determine oxide proportions of both elements, 21% for POx and 68.1% for InOx. Looking at the deep surface measurements (Fig.2), all plasma exhibit interesting results for the oxide removal. In-In bonds are created during the cleaning and may be assimilated to indium agglomerates at the InP surface [3]. Comparison with results in the near surface (Fig.3) shows that “Ar plasma” is the most efficient for the oxide removal with Pox and Inox proportions at 11.2% and 25.7%. “He plasma” is also effective but shows an higher Pox proportion close to the surface (23.7%). “Ar plasma soft” clearly shows the highest oxide proportions near the surface with Pox and Inox at 16.3% and 34.3%. From this study, it can be concluded that Ar and He plasmas are suitable candidates for the surface preparation of III-V materials prior to metallization. Moreover, the benefit to work with a quasi in-situ characterisation has been established. Measurements obtained give more accuracy to monitor the plasma surface preparation performance (Fig. 1). This work was supported by the French National Research Agency (ANR) under the ”Investissements d’avenir” programs: ANR 10-AIRT-0005 (IRT NANO-ELEC), ANR 10-EQPX-0030 (EQUIPEX FDSOI 11) and ANR-10-EQPX-0033 (EQUIPEX IMPACT). [1] L. Grenouillet, and al., 6th IEEE Int. Conf. on Gr. IV Photonics, 2009. [2] E. Ghegin, and al., EEE Trans. on Elect. Dev., vol. 64, n° 111, pp. 4408, 2017. [3] Ph. Rodriguez and al., ECS Trans., vol. 69, n° 18, pp. 251, 2015. [4] B. Pelissier and al., Microelec. eng., vol. 85, n° 11, pp. 151, 2008. [5] G. Hollinger and al., J. of Vac. Sci. & Tech. A: Vac., Surf., and Films, vol. 3, n° 16, pp. 2082, 1985. [6] G. Bruno and al., J. de Physique IV, vol. 5, n° C5, pp. 663, 1995. Figure 1