Have you tried increasing the mesh count around that region?
I've had issues many times with particles at high speed impacting a surface and the mesh is so coarse that the particle is now in a solid cell so it stops calculating it. I doubt that is the case here, but it is possible.
Otherwise have you tried adjusting the density of your particles to see if it is simply a matter of velocity and taper angle keeping particles suspended in that region? Flow does not take the particle volume into account (as far as affecting the flow), so you could have a billion particles accumulate in that region and the particle velocity won't change due to particle-to-particle interactions like it would in real life.