Promoting Statistics of Distributions in Nanoscience: the Case of Improving Yield Strength Estimates from Ultrasound Scission

Sascha Vongehr,Shaochun Tang,Xiangkang Meng
DOI: https://doi.org/10.1021/jp305785w
2012-01-01
Abstract:Statistics often reveals unexpected errors. This is accepted and crucial in mature fields like medicine and experimental high energy particle physics, but usually underappreciated in still relatively novel areas. This work presents our strongest case in support for elevating the status of statistics in nanotechnology. Ultrasound fracturing can analyze tensile strengths of elongated nanostructures. It is rapidly growing in popularity. We calculate the location and shapes of the statistical distributions of fragment lengths and find that the usually employed analysis underestimates yield strengths. In unfortunate special cases, the true yield strength can be up to four times as high. Numerical simulations with generic assumptions still predict true values that are roughly twice as large as the result of the usually applied formulas. We suggest a different data analysis which removes the systematic error. It additionally provides an improved estimate of the statistical error, which combines the variation in yield strengths with the method's measurement accuracy. A detailed reanalysis of previously published data supports our interpretations. The general discussion of the involved systematic and statistical errors calls for caution. Further confirmation by, for example, micro manipulation, is needed before ultrasound fracturing analysis can be trusted.
What problem does this paper attempt to address?