Search

The continuous development of advanced photonic devices based on 3-dimensional structuring of active materials calls for more efforts on characterization techniques. This statement applies, in particular, to processes, such as dry etching applied to III-V semiconductors. Dry (or plasma-based) etching is frequently used within nanofabrication platforms to realize semiconductor lasers, ridge waveguides, photonic integrated circuits, etc. Luminescence techniques (photo-luminescence or cathodo-luminescence) can analyze the material’s properties after dry etching, especially for direct band-gap semiconductors. The presence of non-radiative defects, changes in the local stress/strain, etc., can be probed in detail owing to the spatial resolution of this class of techniques, and their quantification can also be addressed. Our goal in this work is to highlight a robust methodology involving the design of specific test materials, including a series of different quantum wells (QWs) located at well-defined depths below the surface. The spectroscopic signature of these QWs provides valuable information, which can be used to assess the changes occurring to the III-V semiconductor material due to dry etching. The test structures have been designed on (100)-oriented InP substrates, with alternating InAsxP1-x QWs and InP barriers. Sequences of typically 8 QWs with graded As composition were grown, with a fixed thickness (7 to 8 nm), separated by 100 nm InP barriers. The shallowest QW is located 300 nm below the sample surface. By grading the As composition in the QWs (with x typically between 0.35 and 0.5), the luminescence signal for each QW could be unambiguously identified with sharp lines when measured at very low temperatures. These structures were grown by gas-source molecular beam epitaxy. Figure 1 illustrates such a structure. In a second step, a SiNx film was deposited by plasma-enhanced chemical vapor deposition and patterned using standard optical lithography. Elongated stripes were thus defined, whose width varies between 1 and 50 µm, and length is a few mm. Finally, plasma etching was used to fabricate stripes within the QW structures, using this SiNx film as a hard mask. Based on Cl2/CH4/Ar and H2/CH4/Ar, different gas mixtures were employed to perform this etching. We have characterized these etched stripes by micro-PL at 10 K in a specially designed optical cryostat, allowing a spatial resolution of approximately 1 µm and a step-size of 5 µm for mapping the luminescence signal. A 1064 nm laser source was chosen for the excitation of the PL signal in our samples. The choice of this wavelength allows selective excitation of the QWs, avoiding excitation of the InP barrier material. As shown in figure 2, the spectrum displays well-identified lines, which can be attributed clearly to the different QWs in the sample. The very sharp lines (full width at half maximum of the order of 4 meV) attest to the very high sample quality. A fitting procedure was implemented to determine the spectral characteristics of each transition, as illustrated by the red line (“Model”) in fig. 2-a. We have established ([1]) that the observed PL lines are associated with single optical transitions in each QW between the electron and the heavy-hole levels. The transition energies scale linearly with the As composition in each QW (fig. 2-b). By scanning across the etched stripes, we could determine the local changes of the spectral parameters for each QW line as a function of the laser beam position. The first observation is that the etching processes do not introduce any spectral broadening of the PL lines. In fact, for some of the etching processes evaluated, a sharpening of the lines is even observed. This is in contrast with the general assumption that dry etching produces some disorder-induced PL line broadening. The line positions are almost not affected by the etching processes. Finally, we observe some changes in the intensities of the QW lines across the stripes: these intensities decrease strongly near the edges. The different etching processes induce a different magnitude for this effect. These trends will be analyzed in terms of the modifications introduced within the etched material in the area close to the stripe edges. In some cases, the lateral extension of these modifications can reach 10 µm or more. [1] J. P. Landesman, N. Isik-Goktas, R. R. LaPierre, C. Levallois, S. Ghanad-Tavakoli, E. Pargon, C. Petit-Etienne and J. Jiménez, J. Phys. D: Appl. Phys. 54, 445106 (2021). Figure 1