Is actinometry reliable for monitoring Si and silicone halides produced in silicon etching plasmas? A comparison with their absolute densities measured by UV broad band absorption

¹ Laboratoire de Spectrométrie Physique, Université Joseph Fourier-Grenoble, BP 87, 38402 St Martin d'Hères, France
² Laboratoire des Technologies de la Microélectronique, CNRS, 17 rue des Martyrs, CEA-LETI, 38054 Grenoble Cedex 9, France

Corresponding author: martin.kogelschatz@cea.fr

Received: 8 March 2005
Revised: 7 July 2005
Accepted: 14 November 2005
Published online: 22 February 2006

Abstract

SiCl_x radicals, the silicon etching by-products, are playing a major role in silicon gate etching processes because their redeposition on the wafer leads to the formation of a SiOCl_x passivation layer on the feature sidewalls, which controls the final shape of the etching profile. These radicals are also the precursors to the formation of a similar layer on the reactor walls, leading to process drifts. As a result, the understanding and modelling of these processes rely on the knowledge of their densities in the plasma. Actinometry technique, based on optical emission, is often used to measure relative variations of the density of the above mentioned radicals, even if it is well known that the results obtained with this technique might not always be reliable. To determine the validity domain of actinometry in industrial silicon-etching high density plasmas, we measure the RF source power and pressure dependences of the absolute densities of SiCl_x (x = 0−2), SiF and SiBr radicals, deduced from UV broad band absorption spectroscopy. These results are compared to the evolution of the corresponding actinometry signals from these radicals. It is shown that actinometry predicts the global trends of the species density variations when the RF power is changed at constant pressure (that is to say when only the electron density changes) but it completely fails if the gas pressure, hence the electron temperature, changes.