On Heterotic Orbifolds from the M Theory Point of View: Brane Engineering the Eleventh Dimension and its Ends

Horava--Witten M theory -- heterotic string duality poses special problems for the twisted sectors of heterotic orbifolds. In [1] we explained how in M theory the twisted states couple to gauge fields apparently living on M9 branes at both ends of the eleventh dimension at the same time. The resolution involves 7D gauge fields which live on fixed planes of the (T^4/Z_N) x (S^1/Z_2) x R^{5,1} orbifold and lock onto the 10D gauge fields along the intersection planes. The physics of such intersection planes does not follow directly from the M theory but there are stringent kinematic constraints due to duality and local consistency, which allowed us to deduce the local fields and the boundary conditions at each intersection. In this paper we explain various phenomena at the intersection planes in terms of duality between HW and type I' superstring theories. The orbifold fixed planes are dual to stacks of D6 branes, the M9 planes are dual to O8 orientifold planes accompanied by D8 branes, and the intersections are dual to brane junctions. We engineer several junction types which lead to distinct patterns of 7D/10D gauge field locking, 7D symmetry breaking and/or local 6D fields. Another aspect of brane engineering is putting the junctions together; sometimes, the combined effect is rather spectacular from the HW point of view and the quantum numbers of some twisted states have to ''bounce' off both ends of the eleventh dimension before their heterotic identity becomes clear. Some models involve D6/O8 junctions where the string coupling diverges towards the orientifold plane. We use the heterotic-HW-I' duality to predict what should happen at such junc