If air atmosphere is used, some degree of oxidation ― and possibly also hydroxylation ― can be typically expected. Bubbles can possibly be generated by redox galvanic processes involving the melt (fused electrolyte) and the two possibly distinct contacting materials (crucible and riser) which may act as electrodes if somehow bridged by electron conductor materials (e.g. graphite or metallic alloys), particularly under unsatisfactory homogenization of the melt. Melting under blanketing inert gas or under reducing atmosphere can help to avoid bubbles.
I melt in electric arc furnace in argon atmosphere it works good. Oxide layers only appeared with me in case of leakage. If you have facilities to melt in inert gas atmosphere I think you can overcome this problem.