I was involved in a project south of Fresno, CA in 1991 with a slab on grade first floor, very sandy, arid soil, and the contractor wanted to delete the vapor retarder so we executed a deduct change order. In hindsight, this required an adjudication since it is a code requirement. If I knew then....
The floor finish was a vinyl backed carpet tile. About 9 months later, a portion of the vacant floor was being built out and when the tile was lifted, mold was growing under the vinyl. We normally did a full spread glue but on this project, we again executed a deduct change order to use the grid method of glue, as recommended by Milliken.
We did a mositure test of the salb, inconclusive; monitored temperature and humidity for a couple of weeks, inconclusive; reviewed the concrete test reports to ensure the maximum slump was not exceeded; inconclusive.
We cleaned up the mold, replaced the carpet, and laid it down. We found the main lobby also had a lot of moldy areas as well. They were cleaned up and the carpet replaced.
The only thing we could think of - since the floor one slab was placed well after floor two - since it had an underfloor cellular system like that used on floor 2, hydration may not have been complete and the concrete was moist enough foster mold growth under the vinyl.
The next project in Portland, OR in 1992, we did not delete the vapor retarder, we had the concrete tested to ensure it was "dry" enough (by that time, glue was becoming water based and sensitive to moisute in the concrete) and we went back to the full spread glue method. We monitored it and did not find a recurrence.
Don Phillips