Soluble periplasmic materials will be in the supernatant. Membrane-bound materials will be in the pellet if the centrifugation is hard enough. If it was a short, low-speed spin, most of the membranes will still be in the supernatant, however. To thoroughly pellet the membranes, spin for an hour at 100,000 x g in an ultracentrifuge.
as Adam told, periplamic matherial will be in the surnatant after sonication but it will be mixed with soluble citoplamic fraction because sonication is strong extraction method and it is not able to selectivelly extract the periplamic fraction. if you would like to selectivelly recover this fraction you need to use a more gentle lysys approach as osmotic shock using sucrose or treatment with polimixun with destabilize the periplasmic membrane.