The
dd(1) program can use a device node as the input file, to copy an "image" of a partition, sector-by-sector. It would need to be restored in the same fashion. Splitting the output for backup onto smaller media is possible, but the splits would need to be concatenated and dd used during restore. Was that what you meant? If so:
# dd if=/dev/r<partition> | split -b 100m -
Or were you thinking of "slicing" for performance, perhaps in RAID 0? If so, you would create a RAID device which has the underlying slices, then copy your filesystem to it. You could use
ccd(4) to create a logical device, and perhaps use dd, though I would recommend
dump(8)/
restore(8),
pax(1), or
tar(1). You could also use a
zfs(1M) filesystem, where dd would not be usable at all.
If you explain what you actually want to accomplish, that might be helpful.