Parametric study of STiGer etching process in order to reduce extended formation of scalloping defects on the sidewalls of silicon submicron trenches W. Kafrouni a, * , T. Tillocher a , J. Ladroue a, b , P. Lefaucheux a , M. Boufnichel b , P. Ranson a , R. Dussart a a GREMI, Universit e d'Orl eans/CNRS, 14 rue d'Issoudun, BP 6744, 45067 Orl eans Cedex 2, France b STMicroelectronics,16 rue Pierre et Marie Curie, BP 7155, 37071 Tours Cedex 2, France article info Article history: Received 27 May 2016 Received in revised form 26 August 2016 Accepted 27 August 2016 Available online 30 August 2016 Keywords: Silicon Deep etching STiGer Cryogenic SF 6 Scalloping Undercut abstract A first study was carried out to define the appropriate parameters to create a passivation layer by SiF 4 /O 2 plasma that resists lateral chemical etching by SF 6 plasma via the STiGer process at a substrate tem- perature of 85  C. The most efficient passivation layer was obtained for a SiF 4 /O 2 gas flow ratio of 0.65, a pressure close to 1 Pa, and high RF source power. Submicron trenches with a critical aperture of about 0.8 mm were etched by the STiGer cryoetching process, which consists of alternating etch (SF 6 or SF 6 /O 2 chemistry) and passivation (SiF 4 /O 2 chemistry) steps. The obtained trenches, observed by SEM, were vertical with a high aspect ratio equal to 46 with an average etches rate of about 1.8 mm/min. These features exhibit both an undercut and special defect, which called “extended scalloping”. This defect is composed of anisotropic cavities developed on the feature sidewalls, just below the mask and originates from scattered ions located at the feature entrance, which hit the top profile and locally remove the passivation layer. The formation of these defects, as well as trench profiles, strongly depends on the STiGer process parameters, especially chamber pressure and cycle times (passivation/etch steps), and with the optimization of these parameters, defect-free sidewall trenches are obtained with a high aspect ratio. © 2016 Elsevier Ltd. All rights reserved. 1. Introduction Dry and deep reactive-ion etching (DRIE) of silicon plays an important role in the development of MEMS and microelectronics systems [1]. The other main drivers of this growth of DRIE include advanced packaging [2], power electronics [3], passive capacitive components [4], complex microfluidics devices [5], and micro- optics [6]. All these applications require high-selectivity, vertical profiles free of defects and good uniformity, in addition to high etch rates for efficient production and commercialization [7]. Bosch [8e12] and cryoetching processes [13e16], used in silicon micromachining to etch high-aspect ratio structures, have both advantages and disadvantages. Both have been successful for etching micrometer and sub-micrometer features to great depths, and very high aspect ratio trenches have been reported in the literature [17e19]. The Bosch process is a time-multiplexed plasma etching pro- cess. It involves three different steps: depassivation, etching, and repassivation. Each of these steps may be performed concurrently. Some authors have been working on the relationship between the three steps in order to control excess bowing at the top of the trenches and narrowing at the bottom. When depassivation and etching were combined in a single step, better etch profiles was obtained. Both etching rate and selectivity decreased in this case because silicon etching takes place at a lower pressure than in the regular Bosch process [20]. Until now, one of the best processes for deep etching silicon has been the Bosch process, but extra cleaning steps, typically O 2 plasmas, are necessary to remove the polymer coating from the reactor walls. Cryogenic DRIE offers the benefit of producing vertical sidewalls with no observable roughness compared to the Bosch process. This advantage of the cryogenic process is of great significance, especially when etching sub-micrometer structures, resulting in an * Corresponding author. E-mail address: wassim.kafrouni@gmail.com (W. Kafrouni). Contents lists available at ScienceDirect Vacuum journal homepage: www.elsevier.com/locate/vacuum http://dx.doi.org/10.1016/j.vacuum.2016.08.019 0042-207X/© 2016 Elsevier Ltd. All rights reserved. Vacuum 133 (2016) 90e97