Scanning capacitance microscopy characterization of AIIIBV epitaxial layers

The GaAs samples were prepared by MOVPE epitaxy and consisted of four 250 nm thick n-type layers with different carrier concentration (A: n = 2.5 × 10¹⁸ cm⁻³, B: n = 2.0 × 10¹⁷ cm⁻³, C: n = 3.0 × 10¹⁶ cm⁻³, D: n = 9.5 × 10¹⁷ cm⁻³ and E: insulating substrate n ~ 1 × 10¹⁶ cm⁻³) deposited on an insulating substrate (Fig. 1). Such structures were fabricated to verify whether the SCM method could be used for quantitative characterization of doping level concentration in GaAs layers by creating calibration curves allowing to correlate the obtained SCM signal with carrier concentration magnitude.

Fig. 2a presents the profile of carrier concentration in the GaAs test sample measured by electrochemical CV method (E-CV) (the respective layers are marked on the top). Relevant SCM signal profiles obtained for V_AC = 2 V and two different DC sample bias voltages (V_DC = 0 V and 2 V) are presented in Fig. 2b. The resulting calibration curves of the SCM signal in dependence on carrier density are presented in Fig. 3. It could be noticed that for both polarization voltages V_DC, the calibration curves have monotonic character and allow to unambiguous determining carrier concentration based on SCM measurement. This is a clear advantage of SCM technique over SSRM method where similar calibration curves determined for the same sample have large variability which excludes its application for quantitative characterization of GaAs layers [12].

Fig. 4 presents a scheme of InGaAs tunnel junction structure intended for tandem solar cells. In the device designing process and layers growth controlling it is crucial to determine the exact position of p-n electrical junction. Both E-CV and SCM methods can be used for this purpose. However, as can be seen in Fig. 5, the electrical junction position can be determined more precisely using SCM measurements. Comparison of profiles presented in the Fig. 5 shows that the electrical junction position is more precisely localized based on SCM measurement for which a sharp transition from negative signal (p type doping) to positive (n type doping) can be easily seen. The SCM mode parameters (V_DC = 0.5 V and V_AC = 1 V) were selected to obtain higher difference of measured signal levels from p and n type doped layers. Because they differ from the parameters used during imaging of n-GaAs test sample, the calibration curves presented in Fig. 3 could not be directly applied for quantitative characterization of this structure.

Schematic view of a high electron mobility transistor based on AlGaN/GaN heterostructure grown on silicon substrate is presented in Fig. 6. Due to the difference in thermal expansion coefficient and lattice mismatch between epitaxial layer and substrate, the growth of such structure with a quality required for proper device operation is not a trivial task [13]. Basing on determined calibration curves (Fig. 3) SCM was used to verify the expected carrier concentrations levels (very high in transistor channel at AlGaN/GaN interface and very low in the rest of the structure) across the investigated structure and also its distribution uniformity for the respective layers. Fig. 7 presents the topography profile of the sample cross section and the SCM signal profile of AlGaN/GaN/Si structure with depicted position of each layer (to prevent the AFM tip from slipping from the semiconductor layer edge, a photoresist layer was deposited on the top of the structure, which was also included in the topography). The most interesting observation is the existence of high carrier concentration not only at the AlGaN/GaN interface but also in the region of LT (low temperature): AlN. This means that a parasitic channel for electrons exists in the structure which may result in significant deterioration of transistor parameters [14].