Proton Proton Collisions Cross Section Plot by Stirling

This is the inclusive cross section for b-quark pair-production which is something experiments can not really measure very precisely due to limited acceptance. The main way a b-quark pair is produced at the LHC is two initial state gluons fuse to one intermediate one that then splits into a final state b-quark pair. This is due to the fact that the gluon contribution dominates amongst the proton's pdf at very low Bjorken x. However, at very low x, the pdfs have huge uncertainties, e.g. for the pdf set Stirling used in the calculation (it can be seen here: https://mstwpdf.hepforge.org/). If you e.g. consider the case of having a momentum transfer just right to produce two b-Quarks at rest (~9 GeV) - at the Tevatron that corresponded to a larger x value than it does at the LHC. Unfortunately for a given set of PDFs with associated uncertainties one can only use the DGLAP equations to go from one Q^2 to another but not to move from one x to another. Therefore, one would need to include more data at low x in order to constrain these uncertainties, but they only used data mainly collected at HERA and the Tevatron. However, the set is as well outdated and newer PDF sets including LHC data exist (see e.g. https://arxiv.org/abs/1706.00428). This is the reason why the line starts to split at really high sqrt(s) - schematically x ~ M_x/sqrt(s). This is also the reason why you don't see it e.g. for top-quark pair-production because to produce them at rest a much higher x is needed, which in turn has lower uncertainties. In addition to the pdf-related uncertainties, there are as well large contribution coming from the Matrix element itself, however, I do not elaborate much more on that...