The growing demand for precise atomic level control in device fabrication has led to the rising popularity of Atomic Layer Etching (ALE).1-3 As the name implies, the process facilitates layer-by-layer material removal in an atomic scale, making low surface damage one of the key features of this process. The low-power ion bombardment and the self-limited nature of different process steps help to reduce the damage induced by ALE. However, the damage is not easy to characterize. Atomic Force Microscopy (AFM) is one of the methods that is widely used for surface characterization. However, roughness measurements of the surface using AFM may be misleading as it primarily reflects the top surface, which is typically composed of native oxide rather than the actual Si surface.In this work, we discuss how we used X-Ray reflectometry (XRR) together with ellipsometry to measure the damage induced to the Si device layer of a Silicon on Insulator (SOI) wafer.The experiments were done on an SOI wafer with a 55 nm thick device layer and 145 nm box layer. In order to remove native oxide, all samples were treated with Buffered oxide etch (BOE) 10:1 right before the etch process. The cyclic process was performed in a commercial Inductively Coupled Plasma Reactive Ion Etcher (ICP-RIE) Takachi™ tool from Plasma-Therm LLC, USA. The process gases were Chlorine to activate the Si surface and Ar to remove the activated surface. There was no plasma during activation and purging, it was only ignited during the removal step. XRR, AFM, and ellipsometry characterizations were used to monitor the damage induced by the process. These methods were used to measure the surface roughness before and after the process. In the ellipsometry model, a damage layer is incorporated to evaluate the Si surface that has undergone plasma bombardment resulting in its transformation into amorphous Si. This layer is typically referred to as the surface damage layer. The XRR can measure the roughness on Si layer right beneath the native oxide which again demonstrates the surface damage. On the other hand, AFM measures the roughness at the surface of native oxide.Our results show that while in XRR and ellipsometry, the roughness change follows the same trend, the AFM measurement is not. Results have revealed that in the case of a non-ALE process, both XRR and ellipsometry techniques exhibit a substantial level of surface damage, whereas AFM indicates lower damage. This provides evidence that AFM alone may not be a reliable approach for assessing surface damage. In addition, both XRR and ellipsometry confirm a thick native oxide on top which is very much correlated to the damage. Typically, the thickness of the native oxide is in the ranges between 15-25 Å, and our results support that the thickness of the measured native oxide layer increases with the level of surface damage.
Read full abstract