HELP: Loss of phenols, erratic behavior

[cjm]

My lab is suddenly struggling with a few of our mass specs. The issue is the loss of pentachlorophenol in the tune standard and all phenols/aniline compounds in the calibration standard (we are running method 8270). The issue is new, we've been running this method for years on multiple instruments with little trouble. At first signs of loss we would do routine maintenance, clip column, change liner, etc. I just did maintenance on one instrument yesterday and today pcp in the tune standard is almost non-existent. So my biggest question is where do I focus my attention? The inlet? The column?

Oh and one of our analysts mentioned that he would struggle with a new column for weeks then all of a sudden the response for these compounds was good, and calibration would hold well for a couple weeks, then all of a sudden it would fall out. He even ran back to back injections of a tune standard, pcp response was good in first and poor in second injection. Instrument conditions will follow shortly.....

We are running a couple different setups, one pulsed splitless, for trace analysis of water samples, and the other is split injection (10:1). We are having fewer issues with the instrument running pulsed splitless.

Column: Rtx-5 Sil MS w/10m integra guard. flow:1.4mL/min

EDIT: Update: I just ran tune standard by pulsed splitless and response of pcp relative to dftpp was MUCH better than with a split.
If only I could come up with a reliable split injection technique that wouldn't result in loss of these compounds. We're going to be running dirty (not trace) samples and high MW compound contamination may be a great concern in the near future. As for now the pulsed splitless is the only thing working. Calibration curve running now....

[cjm]

So I am going through the curve right now and there's still an issue with the difficult compounds. 2,4-dinitrophenol response factors for 2.5, 5, 10, 20, 50 ppm DOUBLE each level so calibration is impossible. There's definitely still something wrong with the system, and I hesitate to use this splitless method for ugly soil sample extracts, it may ruin the column quickly.