Premium
High‐Precision Density Measurements of Energetic Materials for Quality Assessment
Author(s) -
Herrmannsdörfer Dirk,
Klapötke Thomas M.
Publication year - 2021
Publication title -
propellants, explosives, pyrotechnics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.56
H-Index - 65
eISSN - 1521-4087
pISSN - 0721-3115
DOI - 10.1002/prep.202000272
Subject(s) - repeatability , statistics , gas pycnometer , standard deviation , materials science , amorphous solid , crystal (programming language) , helium , mathematics , analytical chemistry (journal) , chemistry , computer science , chromatography , crystallography , organic chemistry , programming language
Even though helium pycnometry is considered a very precise method of density determination, its use to analyse factors influencing crystal quality (e. g. solvent inclusions, amorphous states, crystal defects…) is limited due to the unknown inherent accuracy of the method. Previous studies have attempted to identify the best possible accuracy of helium pycnometry. However, no robust method has been proposed to validate even the precision obtained. This means that the often occurring, seemingly random density deviations have never been satisfactorily explained. In this study the influencing factors on density measurements of the CL‐20/HMX cocrystal and phase stabilised ammonium nitrate carried out with an AccuPyc 1340 TEC were quantified and a robust method of data analysis was developed to judge with certainty the quality and significance of the data obtained. This study shows that statistical tests such as the t‐test and ANOVA cannot be utilised to differentiate between samples or repeat measurements, even though the density data points exhibit a normal distribution. By utilising confidence intervals and quantifying the three major sources of error (sampling error, change in barometric pressure during the measurement, and weight in error), a repeatability of 0.015 % was achieved and a method was developed that enables differentiation between samples with a certainty as good as 0.05 %.