How do loop-suppressed flavor processes constrain diquarks more stringently than tree-level collider signatures?

This is an excellent question that gets at a subtle but important hierarchy in how particle physics constrains exotic new physics.

Why loop-suppressed flavor processes bite harder

Tree-level collider signatures are blunt instruments for diquarks. At the LHC, diquarks that couple only to right-handed down-type quarks are nearly invisible: they produce jets that drown in QCD backgrounds, and their effects on Higgs and electroweak processes are tiny [AG-2024.10-1055]. A diquark mediator sitting at the LHC just looks like noise.

Loop-suppressed flavor processes are precision scalpels. When a diquark exchanges in a flavor-changing process—say, a rare decay like *b* → *s γ*—it contributes at loop level, meaning the amplitude is suppressed by factors of the weak coupling constant and loop factors. But here's the key: these processes have been measured to exquisite precision, sometimes to parts per thousand, by experiments like LHCb and Belle. A new particle's contribution, even if small, gets caught by the experimental accuracy.

The practical upshot: flavor constraints can exclude down-type diquarks in the hundreds of TeV mass range [AG-2024.10-1055]—far beyond what direct LHC searches can probe. This matters because it closes off the hiding places where a diquark could lurk undetected in collider data.

The catch: flavor structure can be a shield

The story doesn't end there. If you impose a specific flavor structure on how the diquark couples to different quark generations, some of those precision constraints weaken or disappear [AG-2024.10-1055]. This is why both methods matter: in combination, flavor *plus* collider constraints narrow the remaining viable parameter space down to multi-TeV scales, eliminating most of the window where diquarks could hide.