(Digital Presentation) Characterization of Passivation Dielectrics on Silicon Through Second Harmonic Generation: Effect of Fixed Charge

In applications such as image sensors or solar cells, the performance of the device can be improved thanks to the passivation of the silicon substrate using high-k dielectrics such as alumina, hafnium dioxide etc. To assess the quality of the dielectric/semiconductor interface, conventional electrical characterization techniques can be employed, but they require fabrication of dedicated test devices. Among the methods that can be directly used at wafer-level, COCOS [1] allows obtaining electrical information such as the interface states density, but the charging of the sample surface can be an issue. In this paper, we discuss an alternative sensitive and non-destructive optical technique to characterize the electrical properties of a passivation scheme: the second harmonic generation (SHG). In centrosymmetric materials, in the dipolar approximation, the bulk SHG response is zero [2], so the signal mainly contains interface-related information. In particular, when a static electric field (Edc) is present at the interface, the SHG is reinforced and a phenomenon called EFISH (electric field SHG) appears [3]. The electric field at the interface between a dielectric and a semiconductor is related to fixed charges (Qox) and/or interface states (Dit), so the SHG signal can provide these two parameters. Additionally the technique has a potential to probe small areas (within the micrometer size). In this work, the study is focused on the ability of the SHG to measure Qox. The samples used have a layer of Al2O3 of 13nm thickness on Si (100), deposited with different processes, followed or not by an annealing. The annealing step is known to activate negative charges in alumina layers, which promotes the field effect passivation and it can also modify the interface defects [4]. The samples were measured with the SHG tool from FemtoMetrix [5]. Complementary electrical characterizations were performed with the COCOS technique on the full wafer and with capacitance versus voltage (C-V) on dedicated MOS capacitors, in order to obtain Qox. The electric field at the interface can then be calculated through Gauss’ law. Figure 1a shows a typical experimental response of the SHG versus time, obtained on two different samples. The parameters Qtot and Dit were extracted through COCOS. The time-dependence of the SHG is due to the time-dependence of the electric field during the measurement because of charging/discharging effects by carriers generated under the incident laser of the SHG. The SHG signature is obviously stronger for a high Qtot, which indeed is associated with a higher electric field at the interface. Figure 1b shows (with symbols) the experimental results of SHG versus the incidence angle (AOI) for the two samples. The increase of SHG signal for the sample with higher Qtot is confirmed, regardless of the angle of incidence. A general calibration, only based on these types of measurements is not sufficient, since optical phenomena such as transmission & interferences depend on the thickness of the layers involved. In order to better anticipate the effect of these phenomena for a future calibration, we developed a home-made simulator, based on absorption in each layer, boundary conditions at each interface and which contains the “static” interface electric field between materials. Using the electric field calculated with the charge parameters extracted from the COCOS/C-V measurements, we traced the simulated SHG versus AOI (Figure 1b, lines). The tendency observed in experiments is reproduced, even though the simulations are not fully superimposed. This mismatch is probably due to the fact that some of the material parameters in the simulations (e.g. the non-linear susceptibility values) are not very well known. The full paper will show additional samples, corresponding C-V and SHG experimental and simulated data, with the aim of proposing a calibration method that would allow the direct extraction of Qox based only on the SHG measurement. Acknowledgements This work was supported by the French National Plan Nano2022, within the IPCEI Nanoelectronics for Europe program. [1] M. Wilson, J. Lagowski, L. Jastrzebski, A. Savtchouk, and V. Faifer, AIP Conference Proceedings, 2001, vol. 550, no. 1, pp. 220–225. [2] R. W. Boyd, Nonlinear optics. Academic press, 2020. [3] N. M. Terlinden, G. Dingemans, V. Vandalon, R. Bosch, and W. M. M. Kessels, J. Appl. Phys., vol. 115, no. 3, p. 033708, 2014. [4] D. K. Schroder. John Wiley & Sons, 2015. [5] http://femtometrix.com Figure 1