Knowledge Management System Log in
EN
Home PROGRAMMES Technical Committee Initiatives Influence of data analysis on the results of x-ray reflectiv...

Influence of data analysis on the results of x-ray reflectively measurements for ultra-thin film thickness of complex multilayers

Project overseer
Lingling REN
Email
renll@nim.ac.cn
Project timeline
2018/5/1-2018/12/31
Project status
Completed

Project description

Multi-layer ultra-thin films of heterogeneous materials are critical to many classes of precision semiconductor devices e.g. dynamic random access memory and giant magneto resistance. The interfacial structure, uniformity and thickness of such films are critical parameters for control. Non-contact surface analysis techniques are a critical component for the characterisation of such ultra-thin film structures, and the quality control of manufacturing processes. X-ray reflectivity (XRR) is the most promising of these techniques for the accurate determination of ultra-thin multilayer films. A recent pilot comparison held within APMPs TCMM on XRR thin-film thickness determination of complex multilayers, indicated that some variability in result reporting was observed, arising from the data analysis method chosen by the participant laboratories. In preparation for a supplementary comparison on ultra-thin film thickness by XRR, it is desirable to de-convolute the influence of data analysis methods from variation in measurement results. In order to do this, TCMM proposes to hold a workshop for participants which will cover XRR analysis methods and approaches, whilst also providing time for participants to conduct a round-robin on analysis of the same set of measurement data, in order to determine the influence of data analysis on measurement results.

 

TCMM ran a pilot study on ultra-thin film thickness measurement by X-ray reflectivity (XRR), in which for complicated multi-layer samples a high degree of variability in the reported results was observed. An analysis of the measured data indicated that the observed variability was arising as a result of variations in the fitting/data analysis process, rather than in the collection of raw data. This TCI project was proposed to address this variability, through workshop training and a data-analysis round robin. The resulting TCI workshop was held in October 2018.  

(Expected impacts (for on-going projects) and/or realised impacts (for completed project))

a significant improvement in the data-fits was observed:

  • the standard deviation of the fitting results for each layer, using the same original data set, was smaller than 0.04 nm among the 4 participating labs;

  • the standard deviation of the fitting results for each layer, using the same original data set, was smaller than 0.06 nm among the 4 participating labs ;

  • all 4 labs were able to validate their measurement capability and develop/refine/validate their data analysis method for XRR;

  • the standard deviation of the fitting results for every layer using the measured sample 4 data from D institute was smaller than 0.02 nm among 3 Labs;
    - the initial input data used for the oxide and contamination layer are very important for generating high-quality fitting results;

  • the buffer layer in the fitting model has little effect on the final results.

Other factors in the analysis process could be more deeply investigated in future workshops. Suggestion: a guideline for the fitting process is necessary for reliable results, for ordinary GIXRR users.


The TCI project was a wonderful way to overcome some of the challenges uncovered during the pilot study.

Log in

User *
Password *