Analysing chlorine in pressed pellets by using of wet binding material often gives erroneously high results, because soluble in water chlorine concentrated on the surface of pellet during drying of the pellet. Enhancement of the signal in repeated exposure to x-rays may have similar reason.

For the pellets, is the loss of chlorine due to exposure to the x-rays, or due to exposure to moisture in the air? I would suggest that you may want to try using a fairly high concentration of a liquid polymer binder to "seal" the powder particles from the environment. It may also help to contain the chlorine in the sample if the loss is due to exposure to x-rays.

Your results are perfectly reasonable given the calculated LLD's you reported for your set up( instrument, application set up , sample prep etc.) Typically reproducible results can be obtained at 3-5 times higher concentration than the LLD, which is about what you are reporting. Remember that the LLD is not the LOQ, but where the analyte peak disappears into the background. Its all about counting statistics of the counts above background. That is not to say that your application could not be improved to obtain better the levels of quantification, but all else equal you are obtaining the reasonable results based on the your LLD's.