Arcing is always accompanied with a high current. To prevent arcing, you must prevent the appearance of the current. In our lab we are working with RF-plasma discharges and therfore an electrical insulation of the substrates with a capacitor is working fine.

Unfortunatley we can't isolate the substrate becasue of the HDPCVD process itslef, that use a bias voltage, and the wafer structure, that has uncovered Al lines.

While the first issue is fixed, for the second one we are working on pre deposition steps in order to cover the lines but it seems to be not enough.

I am dealing with the same material but with another geometry. I want to coat Al-coils with an electrical isolation by a ion assited PE-CVD process. The first try to prevent the extreme (!) arcing was to make a precoating on the Al-colis by an anodization process.  It works, but anodization is a topic on its own.

I think your substrate is placed direct on the powered electrode. If the applied bias is the self bias voltage than a thin plate (about 0.5 mm) of an isolator (glas, ceramics) on the electrode did not change the bias you need for the SiOx deposition process.