A Statistical Evaluation of Methods of Chromatogram Interpretation-GPC

Abstract

Herein is reported a statistical evaluation of two common methods of determining the molecular weight calibration curve using broad molecular weight distribution standards. A Monte-Carlo simulation technique was employed to investigate and compare error propagation in these methods. It has been common practice to evaluate methods of chromatogram interpretation on an entirely theoretical basis without consideration of error propagation in the calculation procedures employed in the various methods. It is possible for a method which has a sounder theoretical basis to give less reliable results because of excessive error propagation. A statistical evaluation of error propagation also identifies the steps in calculation procedures where serious error propagation occurs leading perhaps to alternate and improved calculation paths.

The analysis of several sources of error is presented in detail including propagation errors due to errors in the molecular weight averages of the broad molecular weight distribution standards and chromatogram replication error.