In order to improve the process control during the steelmaking process, it is essential to obtain knowledge, as well as interpret information, on the characteristics of non-metallic inclusions in liquid steel samples. The purpose of the present study is to evaluate the effect of different sampling conditions on the inclusion characteristics. The changes in number, size and composition of primary inclusions due to the precipitation of secondary inclusions are studied. This is investigated in laboratory scale samples from an Fe-10 mass-%Ni alloy as well as industrial steel samples with varying contents of oxygen and sulphur. The results show that in most cases, the inclusion population can be separated, using a size condition, into primary (>0?6 mm) and secondary (,0?6 mm) inclusions. Overall, the Lollipop sample with a 6 mm thickness can be recommended, because most of the secondary inclusions in low sulphur steels can be removed from the total particle size distribution. However, during sampling of liquid steel containing high levels of sulphur, a significant amount of sulphides precipitate heterogeneously onto primary inclusions, making it impossible to use size or morphology to separate the population. Finally, the numbers of secondary inclusions in the steel samples were found to increase significantly with an increased sample cooling rate. However, the number of primary inclusions was found to be almost constant and independent of the sample cooling rate.

Many kinds of disposable samplers are available for sampling of liquid steel during steelmaking. Depending on slag protection, they can be divided into two main groups: metal‐cap‐protected and argon‐protected samplers. For high quality steels, a correct determination of inclusion characteristics in metal samples is important for an improved process control. This paper seeks to investigate the effect of slag protection system, and sample geometry, on deviation and total content of oxygen within samples from the ladle treatment of liquid duplex stainless steel. The homogeneity with respect to total oxygen content was determined for different zones in the body, pin and inlet part of Björneborg and Lollipop (6 and 12 mm thickness) samples. The results showed low, and stable, contents of oxygen in the argon‐protected samples. However, the total oxygen content, and deviation of oxygen, in metal‐cap‐protected samples was much higher, especially in the body part of the samples. Finally, it was found that the oxygen content heterogeneity in the metal‐cap‐protected samples increased with a decreased sample weight. Based on the obtained results, the 6 mm thick Lollipop sample with argon‐protection is recommended for determination of total oxygen content and oxide inclusion characteristics.