Metrology by scatterometry
Transparent substrate imprinted by NIL
and characterized by scatterometry
The scatterometry is a metrology technique which allows analysing the light diffracted by a periodic pattern in order to determine the shape of this pattern. This kind of measurement is possible using an ellipsometer or a reflectometer. Electromagnetic codes based on the MMFE allow addressing the direct problem, i.e. computing the signature of a given pattern. The measurement consists in a inverse problem solving. Different algorithms are developed to address the inverse problem, but the method which consists in a research in a database previously built thanks to the direct problem remains the usual technique.
This technique is already almost routinely used in the industry for static measurements especially because of its non invasive and non destructive behaviours. Static and dynamic scatterometry is developed in LTM since several years. Static scatterometry is mainly improved for characterization of imprinted patterns, in connection with NIL activities. It is very interesting since it leads to a fast and non destructive measurement of printed structures. It is also very performant and accurate to characterize high resolution nanostructures. Scatterometry was developed during last years for 1D periodic lines, but modelisation is now improved for 2D structures. It was also performed mainly on Silicon wafers, but one important issue of development is now the adaptation of scatteromerty to transparent substrates such as glass or plastic films for optical applications.
One important achievement is the development of real-time scatterometry to characterize in situ the profiles of patterns during plasma etching processes. There is a huge interest from the microelectronic industry for a metrology tool able to follow dynamic processes in situ and in real-time. The real-time constraint induces a library approach. This technique consists in computing multiple signatures covering all expected shape parameters and comparing these signatures in real-time with the measurement in order to find which one is the most similar. The number of parameters governs the size of the library and a good accuracy leads to a large library size. Two main axes of improvement are developed. First, a parallelized computation to create a big library in a shorter time and second a smart method, using GPU (Graphics Processing Unit) allows speeding up drastically the signature research in the library. A software environment has been created, and an experimental set up has been built to study the evolution of nanometric patterns exposed to a plasma. The results exhibit a very good agreement between scatterometry and AFM. This technique is therefore very promising for real time characterization.
