Type of presentation: Poster

MS-8-P-2815 Strain determination by CBED in Si-rib structures for photonic devices

Balboni R.1, Bolognini G.1, Corticelli F.1, Ferri M.1, Mancarella F.1, Marini D.1 2, Montanari B. G.3
1CNR, Istituto IMM, Bologna, Italy, 2Università degli Studi di Bologna, Bologna, Italy, 3Laboratorio MIST-ER, Bologna, Italy
balboni@bo.imm.cnr.it

Strained silicon opens interesting perspectives for photonic applications, since the modification in the crystal structure symmetry can give rise to non-linear effects. The deposition of a straining layer on top of a silicon waveguide can break the silicon lattice inversion thus enabling significant linear electro-optic effect [1] or second-harmonic generation [2]. It is therefore important to monitor the stress induced by the process steps used to define the structures and to check the validity of the stress model adopted in the computer simulations of the process itself. In this work the Convergent Beam Electron Diffraction (CBED) technique was used to estimate the lattice deformation induced by a silicon-nitride film deposited on micrometer-scale silicon rib structures.

The manufacturing process consisted in the deposition of low-temperature silicon oxide on Si wafers, followed by photolithography and selective removal through reactive-ion etching. An Si3N4 film was then deposited on the structures, inducing a significant strain inside the silicon ribs. In the analysed samples, shown in Fig. 1, the rib height and width were 450 nm and 2 μm respectively, while the deposited Si3N4 thickness was 375 nm. TEM samples thinned along the [110] orientation were analysed in the [230] projection in a Tecnai F20T transmission electron microscope operated at 200 kV and in the STEM mode; the lamella thickness was estimated about 350 nm. HOLZ lines patterns were recorded in a matrix of points using the procedure reported in [3] and analysed with the ASAC software [4]. The resulting stress configuration was also simulated using a finite-element method.

An overall compressive strain was measured, as induced by the deposited Si3N4. Fig. 2 shows a comparison of the measured and simulated strain component εZZ. Both results behave symmetrically with respect to the rib width with a reduced compressive strain at the centre. Fig. 3 shows the same results for the εXZ shear strain component: as expected, both curves show an antisymmetric behaviour with respect to the rib centre, which is directly reflected in the diffraction pattern features (see insets in Fig. 3). Although the simulation seems to slightly underestimate the crystal deformation, the overall agreement can be considered good.

In conclusion, the combination of process simulation and CBED strain measurement results proved to be effective in predicting the optical behaviour of strained crystal silicon structures.

[1] R. S. Jacobsen et al., Nature, vol. 441, pp. 199-202, May 2006.

[2] M. Cazzanelli et al., Nature Materials, vol. 11, pp. 148-154, Feb 2012.

[3] A. Armigliato, R. Balboni and S. Frabboni, App. Phys. Lett. 86 (2005), p.63508.

[4] http://stream.bo.cnr.it/docs/iTEM_Solution_ASAC.pdf


Fig. 1: ADF STEM image of Silicon rib section. Foil normal is [110] while vertical direction is [001]. The Si3N4 appears brighter in the image than the silicon substrate.

Fig. 2: Experimental (dots+bars) and simulated (squares) of the εZZ strain component across the Si rib (along [-110] direction) at a height of 65 nm with respect to the rib bottom floor.

Fig. 3: Experimental (dots+bars) and simulated (squares) of the εZZ strain component, same conditions of Fig. 2. In the insets a portion of the CBED patterns, registered at the points indicated by the arrows, show an antisymmetrical behaviour.