We use the Evolution for 524 and have dealt with similar problems in the past, even with the first gen Encon and Tekmar 2000.
The cause is related to the total analyte concentration. The higher the total concentration the better the recovery especially with the later eluting compunds. I had a three or four page thread going on that here in the past. I did a study where I calculated the total mass of analytes and compared that to recoveries of the 1,4-Dichlorobenzene. As the total concentration increases the area of the internal standard also increases. One solution was to increase the concentration of the internal/surrogate spike. I currently use 25ppb in the sample for all the internal/surrogate compounds and it works well, if you increase to 50ppb or 75ppb it works even better.
With a calibration from 0.5ppb to 200ppb the 25ppb internal works well. I discovered the relationship when I was trying to do calibrations from 0.05ppb to 40ppb using 1ppb or 5ppb internal standard/surrogate concentrations, the effect was greatly exaggerated. I then tried the same calibration range but had the internal/surrogates at 50ppb and the areas over the curve had very low %Diff sometimes as low as 5% or less. If you add up the total analyte concentration plus internal standard/surrogate in each standard you get much less change in total concentration with the higher internal/surrogate and therefore have better results on the calibration and less change between a blank or clean sample and spiked calibration checks.
There was much speculation in that old thread about the possible causes for the effect. I did find I have better results if I only bake the MORT at 180C though. I wonder if possibly a higher bake temperature causes char formation, just a very thin film, on the MORT that adsorbs small amounts of analytes but becomes saturated with large amounts which allows more analyte to make it through to the trap in the more concentrated standards. MORT bake temperature and turning off the gas saver and allowing the full split flow during analysis are the only changes I have really made to the system, the latter to help flush out any moisture in the system between samples. So far I can go much longer before I see any problems with high %rec of the internal/surrogates and increasing area counts over the calibration range for both my 524 and 8260 setups.