I'd start with a 5% diphenyl dimethyl polysiloxane phase. A real workhorse. Maybe 0.32mm I.D. x 30M length and run the helium around 1.5 cc/min constant flow. Just a steady ramp from the appropriate solvent focusing temperature up to around 330°C over an hour.
Kurt, I think I have to disagree about the ability of low -vs- high res to make good identifications. Running in full-scan mode, a low-res can identify quite well assuming a non-coeluted peak. With coelutions, even a high res instrument won't be able to sort out the mix of masses from the coeluting compounds and the library fits will be poor. So the real limiting factor would be the complexity of the mixture... thus the ability of the column to resolve the close eluters.
You want your library spectra to be run at a resolution similar to the resolution that you're using.
Generally, you'll find the higher resolution instruments running in a selected ion monitoring mode where only a small subset of masses are being scanned thus increasing the mass dwell time and sensitivity, and using the filtering capability of higher resolution to distinguish masses of interest from very close masses coming from coeluting non-interest compounds.
The mass filtering capability of a high res instrument comes into play when doing such a SIM analysis since at 1000 resolution (low res), mass 162.26 can be distinguished from a mass about 0.16 amu different (162.26/1000), thus if you're trying to look at mass 162.26, you'll be seeing a mass range of 162.10 - 162.42, thus non-interest coeluters can interfere if the mass is close enough to mass 162.26. Raise the resolution to 10,000 (high res), and discrimination increases to 0.016 amu (162.26/10000), thus lowering the indistinguishable mass range to 162.244 - 162.276. Much higher resolutions can be used, but with higher resolution, there is a sacrifice in sensitivity due to the physical focusing of the ion beam.