Ive used them and they work well. But for GC injections I think they would be a bit cumbersome. I agree that the Chaney adapter would be well suited,but an auto-sampler is the best option. An internal standard might help too. That way,what your really measuring is the ratio of two peaks,so small variations in the amount injected is less critical. Variations in timing though can still cause problems. And large variations in the amount injected can similarly cause problems.
One thing that many people forget with syringes is that their accuracy and precision can be far less than you might think. Its typically specified as some percentage of full scale (or sometimes a percentage of some percentage full scale) For instance, Hamilton claims 1% of full scale for accuracy and 1% of 80% full scale (0.8%) for precision.
Practically speaking,suppose you have a particular 100uL syringe. It delivers its average volume at 101uL. Thats within spec. Now suppose furthermore,that syringe shows a variation of +/- 0.8uL. Thats also within spec. So,if you do your part of the job right,then your syringe will deliver between and 100.2uL and 101.8uL most of the time.
That seems pretty good,but it gets bad really fast. Suppose you use your syringe to deliver 10uL now. First,nothing says that just because your syringe was +1uL accuracy and +0.8uL precision at full scale,that its the same there. The only thing we can count on is,assuming the syringe is good,that it will deliver its volume within the manufacturers specification. Lets assume its at that LOWER edge of those specifications. So,it delivers a volume at 10uL that is -1uL on the low side and +/- 0.8uL. That comes out to 9uL +/- 0.8. if you do your job correctly it delivers volumes between 8.2uL and 9.8uL. Thats not so good. Thats why Hamilton recommends against using their syringes lower than 20% of full scale.
Even at 20%,if you pick up a random 100uL syringe,you can expect it to dispense between 18.2 and 21.8uL. This has some significant ramifications in the lab. Suppose someone spikes a vial with 20uL of analyte using that 100uL syringe. Now suppose I make a standard using a similar syringe and use that as a single calibration point to quantitate his value against. Numbers between 83.9% and 119.78% are perfectly normal. Where I work,we have a QC department that prepares blind QC samples. That would fail. Whats worse if you work out the chance of 4 or 5 points being off in the same direction,severely biasing a curve,its not as small as you might think. What that means is,over thousands of analytical runs and calibrations,you expect this to bite you quite a few times,unless you take steps to understand and eliminate or at least detect it.
