A Novel Method for Interlaboratory Analysis of Total Package Oxygen

Abstract

The measurement of total package oxygen (TPO) is a standard procedure used in bottling to provide fundamental information for the prediction and improvement of beer flavor stability. The standard analytical procedure used today was developed more than 25 years ago. Sufficient repeatability and reproducibility are vital, and in the case of TPO, repeatability between laboratories is a problem and poses a serious threat to the significance of the analysis, especially in large breweries with multiplant operations. In this study, several approaches for the preparation of TPO samples were compared, among them modifications to containers and different matrices. Trials focused on bottles with pry-off crown corks. A selection of bottles tested according to volumetric capacity was necessary to minimize errors. The central issue was the method used to introduce well-defined amounts of oxygen into a container. Best results were obtained using crown corks custom fitted with a septum for gas chromatography, allowing injection of the desired amount of air using a gas-tight precision syringe. Matrix effects were evaluated by performing reproducibility and repeatability analyses with different liquids in sample containers. In the case of water, several modes of preparation were examined. None of the preparations yielded satisfactory reproducibility. Using beer as a matrix, statistical distribution was significantly better. Preparations assessed included fresh tunnel-pasteurized beer, pasteurized beer aged at 28 or 60°C, and pasteurized beer from a brewery pilot plant. Combining the most favorable conditions, an acceptable reproducibility and repeatability of 22% was achieved. Based on these findings, an interlaboratory analysis model was designed that consists of two levels of air injection into beer, which generates three concentrations of oxygen. This model has been put into practical use, and its reproducibility and repeatability is an acceptable 24%. This compares with an unacceptable 48% calculated from the data presented by the ASBC technical subcommittee.