In the view of developing high performance metal-oxide-diamond field effect transistor (diamond MOSFET), recent reports presents different approach in the choice of the gate material and, in particular, on the dielectric layer [1-2]. However, a nanometric analysis of the band levels is necessary to understand the electron dynamic across the MOS stack. Recently, STEM-EELS have been revealed as reliable technique to estimate the bandgap of SiO₂ dielectric materials. Nevertheless, its applicability has been limited to SiN_x materials.

To evaluate electron transitions near the bandgap energy in the EELS spectra, researchers have to overcome several experimental difficulties. Here, we evaluate the effect of Čerenkov radiation and volume plasmon-related peaks in low-loss range of EELS spectra. In this study, STEM-EELS techniques are used to analyse the O-terminated diamond/Al₂O₃ interface (similarly to previous studies presented in SiO₂ by other authors [3]). Indeed, volume plasmons (VP) and Čerenkov (Ch) radiation contribution are evidenced in Fig.1, which also shows diamond-related D1-2 peaks and Al₂O₃-related peak A1 however, in the Al layer, only plasmon-related peaks are revealed. Probe position of the previously presented EELS spectra are shown in Fig.2, as numbered dots. Figure 2 shows 001-BF TEM micrography of the diamond/Al₂O₃/Al layers. Inset of Fig.2 shows HREM micrography of the diamond/Al₂O₃ interface, revealing variations in the crystalline quality of the low temperature (100ºC) ALD-deposited Al₂O₃ layer.

In this work, we present a methodology to evaluate the influence of the Čerenkov losses and plasmon-related peaks in the low-loss EELS spectra. In some cases, such peaks are shown to mask the interband-related transition. Indeed, in such cases, Čerenkov-related and plasmon-related peaks have to be deconvoluted and removed, in order to accurately apply the linear-fit method [3], which allows calculating the diamond/oxide bandgaps.

The previously described methodology allows determining the bandgap variations in the oxygen-terminated diamond/oxide interfaces.

[1] S. Cheng, L. Sang, M. Liao, J. Liu, M. Imura, H. Li, and Y. Koide, Appl. Phys. Lett. 101, 232907 (2012).

[2] A. Maréchal, M. Aoukar, C. Vallée, C. Rivière, D. Eon, J. Pernot, and E. Gheeraert, Appl. Phys. Lett. 107 (14), 141601 (2015).

[3] Jucheol Park, Sung Heo, Jae-Gwan Chung, Heekoo Kim, HyungIk Lee, Kihong Kim, and Gyeong-Su Park,  ULTRAMICROSCOPY 109 (9), 1183 (2009).

Figures:

Figure 1: EELS spectra acquired in the different layers of the diamond/Al2O3/Al MOS stack (spots 1-3 in Figure 2). Some observed peaks (as Čerenkov -related and plasmon-related peaks) are not material-characteristic peaks and, thus, have to be removed for an accurate determination of the bandgap.

Figure 2: 001-BF TEM micrography of the MOS stack, diamond/Al2O3/Al layers are revealed. White arrow is used to highlight EELS probe position (spots 1-3). Inset shows HREM micrography of the diamond/oxide interface, revealing variations in the crystalline quality of the ALD-deposited Al2O3 layer.

Figure 3: Low loss EELS spectra acquired at diamond and Al2O3 layers. Black line is used to plot convoluted original EELS spectra; dashed line shows the fitting of the removed peaks (plasmon and Čerenkov -related), grey line is used to plot final spectra. Then, linear-fit method is applied to deduce diamond and Al2O3 bandgaps.

« Back to The 16th European Microscopy Congress 2016

EMC Abstracts - https://emc-proceedings.com/abstract/diamond-based-mosfets-bandgap-interface-profiling-by-stem-eels/