The gemara asks whether a keli combines the two half עשרון that are in it even though they aren't touching. The gemara then questions whether one can take a kemitza from one half עשרון that is in a keli with the other half but isn't touching - the gemara understands that this question is fundamentally whether the concept of tziruf keli is d'oraysa or d'rabonon. The Rambam (Hil. Pesulei HaMukdashin 11:23) paskens that it is a safeik whether one can take the kemitzah on a half עשרון and have it count for the other half. The Steipler (kehilas ya'acov 16) asks, why does the Rambam consider this a safeik? The gemara makes it very clear that the only possibility of the kemitzah on one half working for the other is if tziruk keli would be d'oraysa, but the problem is that the Rambam himself paskens (sha'ar avos ha'tuma 12:7) that tziruf keli is only d'rabonon. Therefore, the Rambam should have paskened that a kemitza on one half definitely doesn't work for the other?
The steipler explains that the concept of tziruf keli in the world of tu'mah is very different from the concept of tziruk keli to allow kemitzah from one half on the other. To allow kemitzah the tziruf keli wouldn't have to consider the two halves to be one unit, because kemtiza is always done from flour that is not really one "gush" or unit. The concept of tziruf keli for kemitza would only require that we consider the two parts to be together, but the concept of tziruk keli for tu'mah would literally have to consider them as one unit. Therefore, the rambam can pasken in hilchos tu'mah that tziruk keli is only d'rabonon, yet have a safeik whether kemitzah on one half עשרון works for the other since for this purpose the tizruf keli may be d'oraysa.