You're seemingly tied to the static-headspace method so you have a few things to determine before you'll have a good method.
Triacetin is probably a good choice of a solvent because it's chemically similar to your sample.
If I have your problem and your equipment (static headspace sampling is an equilibrium technique), I would use the method of standard addition to calibrate. Make a benzene-in-triacetin standard, then make known additions to the same mass of PET (in different vials of course). The key is you'll have to "cook the benzene into the sample" after you add it to the vial to ensure you're at equilibrium. To do this, I make several vials with a known and the same amount of PET in them. I then add a known amount (the same) of the standard to each vial and seal them. Then I heat the vials to a temperature above the glass transition temperature of the solid sample for many different times. Analyze an unequilibrated vial as a the t=0 sample. Then analyze the samples at 0.5, 1, 1.5, 2 hours etc. and see how the benzene peak area decreases with time. Once it doesn't decrease anymore, you're at equilibrium. As the sample approaches equilibrium, the amount of benzene that's actually in the gas phase will be less than what was present at t=0.
At this point, you know what it takes to get your fortified samples to equilibrium. Now, you an just perform the method-of-standard addition to determine benzene in your samples. If you can verify that the calibration data is independent of the matrix, then you can probably buy some clean PET and use that as your universal calibration medium.
It may take less time to get your samples (no fortification with benzene) to equilibrium. You'll need to determine this too.
There's a great book out there by Ettre and Kolb called "Static Headspace Gas Chromatography" that discusses all of these techniques. I'd get a copy of that if you don't have it. The PerkinElmer app note employs the technique called Multiple Headspace Extraction (MHE) that reduces/eliminates the matrix dependence of the analysis. You may explore that as well. Perhaps that's why they heated the sample to 200 °C during their analysis.
https://resources.perkinelmer.com/lab-s ... 276907.pdf