In-Situ Spectroscopic Ellipsometry and Optical Emission Studies of CF 4 /O 2 Plasma Etching of Silicon Nitride
- PDF / 477,462 Bytes
- 6 Pages / 420.48 x 639 pts Page_size
- 109 Downloads / 213 Views
EXPERIMENTAL SixNy films were deposited on GaAs (100) substrates by plasma enhanced chemical vapor deposition (PECVD) in an Alcatel ADF300A reactor operating at 13.56 MI-z. The SiXNy films were deposited at 300W RF power, 850 mTorr working pressure, 300'C substrate temperature, with 196.5 sccm N 2, 0.7 sccm Sill 4 , and 2.8 sccm NH 3 gas flow. The plasma etch experiments were carried out in an electron cyclotron resonance (ECR) etch chamber. The ECR chamber is equipped with an 88-wavelength in-situ spectroscopic ellipsometer (J. A. Woollam Co.) as well as a Princeton Instruments OE system which includes a 0.3 m focal length Acton Research spectrograph with a 1200 groove/mm grating and an EEV 256x1024 CCD detector. ECR etch experiments were carried out with microwave power settings between 100 and 500W, chamber pressure (controlled by means of a throttle valve) between 3 and 17 mTorr, and the CFS5%0 2 inflow rate between 5 and 10 sccm. The ECR chamber is pumped by a turbomolecular pump
8 backed by a mechanical pump and typically has a base pressure of 1x 10 Torr. Prior to etching, each sample was characterized by SE. A model consisting of a SixNy layer and a GaAs substrate layer was employed to extract the thickness and refractive index of the SixNy film from the SE data. Though the ellipsometer spans the spectral range of 280 nm to 760 nm, we only employed data acquired between 500 nm and 760 nm for model fitting since in this spectral range we can reasonably represent the SixNy film as transparent, thus enabling us to reduce the number of fit parameters in the model by setting adsorption to zero for the SixNy layer. The dependence of the refractive index of the SixNy layer on wavelength is represented with a dispersion relation defined as n = A + B/X2 + C/X4 (X is wavelength and A, B, C are fit parameters). In this work the GaAs substrate is represented as an effective substrate characterized by the optical constants determined by SE measurements of GaAs substrates and is therefore slightly different (n, k modified by < 0.03, 0.2 respectively) from the bulk optical constants over the wavelength regime of 500 nm to 760 nm used. A more physical model would be one that introduces an interfacial layer of unknown thickness and wavelength dependent optical constants in between bulk crystalline GaAs as the substrate and the SixNy film. Attempts to model the SixNy / GaAs system as such a 3 layer system show that the effect of an interfacial layer can be represented as an effective substrate as well. For the purpose at hand, i.e. formulation and testing of an etch control methodology we have chosen the simpler model for implementation of control. With independent information on the nature of the interfacial region, and indeed on the SiXNy surface roughness as well, improved physical models of the system can be readily incorporated in the real-time feedback control scheme for etching. During the ECR plasma etch, the system conditions (i.e. pressure, microwave power, etc.) are regularly measured and recorded by means of an
Data Loading...
