Automated characterisation and analysis of large arrays of nanostructures fabricated at wafer scale

In this paper we are presenting a fast and automatic way of characterisation and data analysis using a general purpose SEM (Scanning Electron Microscope). This is a very interesting approach for us (International Iberian Nanotechnology Laboratory) as we are fabricating millions and millions of nanostructures everyday and we need to get feedback from how the structures look like in order to control and improve our fabrication tools and processes.

The paper has been published in Precision Engineering journal (Elsevier) and this is the link to the journal: 10.1016/j.precisioneng.2019.08.017

By this method we are able to check more than 300000000 (3 ∙ 10^8) pillars in one night or to study in detail 40000 (4 ∙ 10^4) pillars in different positions of the wafer (1600 SEM images in both cases). We developed the scripts for iFast (programming of the SEM acquisition), Labview (image processing) and Mathematica (data visualization). 

This is an example of the images we are processing:

This is the abstract of the work you can find in Precision Engineering (https://doi.org/10.1016/j.precisioneng.2019.08.017):
Large-area and dense arrays of nanometric scale structures are commonly fabricated for several applications. The characterisation of sub-micron structures (below 1 μm) at a large scale as well as the related data analysis are challenging tasks. Here, we present a method to address the image acquisition, the data extraction and the data analysis, applied to the evaluation of the uniformity of nanometric structures in a silicon master. Automated routines for both high and low magnification Scanning Electron Microscope (SEM) imaging have been successfully developed. SEM images have been automatically acquired by scripting routines, which collect large amounts of images of the nanometric structures covering multiple regions of the wafer in a few hours. Geometric parameters of the nanometric structures such as diameter, period and height have been extracted from the raw images using the developed image processing scripts. Finally, the data extracted from more than 4800 images acquired with three different characterisation scripts (1600 images for each study) have been analysed and plotted according to their position in the wafer.


And we can get visual results as:

To cite our work this is the reference:
Llobet J, Martins M, Calaza C, Antunes M, Martins S, Fonseca H, Pires B and Gaspar J
 Precis. Eng. 60, 320-325 (2019).

For any question or suggestion do not hesitate to contact me.