weber,

The ceramic backing will be partially melted away during welding. This is normal and shouldn't really be anything to be concerned about. I don't think contamination is an issue either. If the very use of ceramic backing would cause contamination of the root pass, it would defeat their purpose. If you're familiar with stud welding, take a look at the inside of ceramic ferrules after studs are shot. Part of the ceramic is melted and black. Ceramic ferrules for automatic stud welding are essentially doing the same thing as ceramic backing, with no adverse affect on the welds on the studs.
Ceramic backing is advertised that with its use, the joint doesn't have to be backgouged. In many cases, you can achieve these kinds of results, but not always. There are a few factors that you should be aware of. You need to use the proper shape and size of ceramic backing in order to obtain the type of results you want, and it must be properly placed on the joint. Fitup conditions are critical for the use of ceramic backing. You will find in many cases that even with the use of ceramic backing, variations in such fitup characteristics as root opening, root face, misalignment, etc., result in an unacceptable weld condition that will require backgouging and back welding, at least locally. Ceramic backing may also be difficult to use for some configurations, such as at the ends of a girder web splice, and backgouging and back welding will likely be necessary in areas such as these. Another consideration is that ceramic backing is porous, and is therefore subject to picking up moisture. If your application is sensitive to hydrogen, you will want to treat the ceramic backing much like the other weld consumables with regard to atmospheric exposure limits. Though not as sensitive as SMAW electrodes, you don't want to leave them lying around and exposed to rain, dirt, etc. They should be placed on the joint immediately prior to welding.

In the context of a tendon welding application, a ceramic backing was used to backup a SAW weld root. The root surface was subsequently finish ground smooth to the profile typical of tendon welds in high cycle fatigue service. In the effort to resolve the cause for UT and MT centerline indication at the ID surface after finish grinding, the root pass heat input was increased and coincidentally, the ceramic backing was more heavilly heat affected. This resulted in no new coincident discontinuity or defect features and was coincident with significant reduction of the centerline feature.

The increased heat effect upon the ceramic material was evident in the amount of "glazing", or glassification, of the ceraming surface interfacing with the weld metal. Whereas previously there was light or intermittent heat affect to the ceramic, now the effect was continuous and relatively deep. The glazed ceramic surface took a replica cast of the weld surface. As part of rationalizing the changed root pass parameters, and taken in the context of a slag metal system, such as SAW, it was predicted that the ceramic material, being essentially alumina and a singificant component of many SAW fluxes, was benign. Concern for contaminant minerals entrained in the raw material for the backing were not realized, nor identified from chemical analysis.

The suggestion to control the dryness or sundry contamination of the ceramic material should be taken as and absolute requirement. Interpreted as significant toward removing the ID indications, and other indications, was the implementation of "clean handling" and hydrogen contamination control practice for everything associated with the weld process.