If you have ruled out leaks and the FID jet is clean, maybe it could be inlet discrimination.

http://www.chromacademy.com/gc-troubleshooting-guide-peak-area-reproducibility.html

1. Use splitless (by reducing of course your compound 's on-column amount!) - avoids variations by split which can be less reproducible.

2. Avoid Liners with glass-wool. Use deactivated empty liners.

3. Increase injection volume up to say 2 uL and reduce the amount of samples by a factor x- as you do not want to go beyond the dynamic range.

4. Use lower injection port like 220 C, as you wish to avoid thermal breakdown at injection port.

5. After derivatization reactions with TMSI/TMCS(2:1), make sure to centrifuge well and avoid white particles/ crystals that these compounds invariably form, and take very little.

6. Make larger volume reactions to avoid handling variations, clean syringes with 2-3 washes or more to avoid carry over, do not store derivatized samples more than few hours. Change liner and septa every 24 hours if you keep running it all the time. If few samples, then not an issue.

7. Change a new column syringe, liner, septa, and see if the problem persists.

" if I provide repeated injections from the same vial peak areas are getting higher and higher, " is also seemilngy a "carry over" issue from syringe, septa, liner, columns! So wash and clean well, and run blanks in between every samples! : )

Also, what magnitude (x times) of variation you see? Calculated CV/ RSD as well?

I guess problems would be in points 1-4 and if you can address them one by one, I see the problem going away!

Thanks,

Biswa

Lactose is a carbohydrate.  Are you sure it can be derivatized with TMSI/TMCS?

Do you have no excess of pyridine in your derivative solution. Normally the carbohydrate derivative agens work with excess of pyridine to scavenge the released protons. Perhaps it will help if you add 90% pyridine to your derivative solution.