Use options slice if you are using slices, as mentioned !
Or use entire device (s2) without options=slice, then partition s0 inside LDOM (or outside) and give it to ASM.

Never give s2 to anything except zpools on VTOC labels.
ZFS when s2 (full scsi lun) or without slice, will repartition disk(s) properly and create zpool (EFI)
When using s2 or no slice at all, ZFS will be endianess aware, you will be able to import the zpool on both solaris architectures (SPARC or x86).

Also, clusters that do SCSI reservation and fencing (like Solaris cluster) will require s2 inside LDOM, or else it will not work.

As for recovering, well, i have no good news about that.
Perhaps some hackery could be done if you had original intact disks without creating zpools out of ASM disk(s)
Why did you do that ... it's like having bad UFS and creating ZFS out of it to recover, it just doesn't make sense

Hackery in a way that you could (from primary domain) :
Present a bit bigger lun, label them, create slice 0 matching the size of entire old lun exactly
Using dd copy entire s2 disk (broken) to s0 slice (new presented and labeled above)
At the end, you would have a labeled disk with s0 slice containing ASM label.
See if ASM now recognizes those luns as proper ones present to LDOM with options=slice
I haven't tried this, get back with results

In my opinion, it is just easier and less prone to error to create from scratch and restore backup and continue with standby DB.
I presume you are replicating with database tools, not storage backend, since if you used storage, you would get properly labeled disks on DR site.

Regards
Peasant.