varian 3800 reproducibility problems

[sheepy98]

Hi there, Am using a Varian 3800 GC with 1177 injector and FID and have no reproducibility with various capillary columns and methods. RSDs 8% or more with generally 0.2 microlitre inj and 50:1 split . New syringes, septa , injector liners all tried to no avail . Recent gas supply filters fitted but still no joy . Found recently with new Electronic Flow Control unit fitted got seven consecutive injections with RSD of ~1% but erratic reproducibility returned and still bad . Any help would be gratefully received .

[BG]

Hi,

We need to know the type of solvant you have , the type of injection (manually or autosampler), speed , technique of injection and the purge septum flow.
Eventually the type of sample (volatil or not)
Because your probleme must be a pb of bad transfer in the column due to the vaporisation and all these informations are necessary

[sheepy98]

Hi BG, Thanks for reply . Injection is manual with 0.2 microlitre of sample drawn in syringe , ~0.2 microlitre of air drawn back , place syringe needle in injector port for 1 second , fast injection , leave in port for 1 second then withdraw . I think my technique is relatively sound but know an autosampler would be better . I assume purge septum flow is correct as have had a GC engineer recently checking all flows even though I dont know its exact figure and know we should really purchase an accurate flow meter . The two main methods we are trying are Triallate samples made up in Acetone and Dichlobenil samples made up in Ethyl Acetate . Both methods are established industry methods . Triallate performed on DB5 column and Dichlobenil on CP-Sil 19 as per method but both analyses give similar reproducibility problems .

[CE Instruments]

You have not stated the syringe size ? If 10ul it is unlikely that you will get good reproducability on a 0.2ul injection. If you are using a 1ul syringe, plunger in needle you need to be sure that your sample does not contain high boiling point components that could precipitate out in the needle. Have you tried an internal standard to see if it is the sample transfer from the syringe that is causing the problem or something else ? Try using a 1ul injection volume with increased split to see if the small injection volume is a problem.
If you want better than 1% RSDs on a daily basis buy an autosampler

[BG]

hello sheepy 98,

Ok for the answer but effectively, as sayed CE instruments, the seringe and sample volumes are important.
In an other hand, when you say purge septum is OK and fast injection.........what are the values?

What is the mode of injection you choose on your autosampler (and the model of autosampler) because sometimes, if you use preprogrammed injection mode, injection speed could not corresponding to your sample and generate pb of RSD ...!!

BUT your pb had always been or did it progressively appear ?

[sheepy98]

Hi , Thanks for both replies . Syringe is a 1.0 microlitre . Both methods I have tried have internal standards but no reproducibility . I shall post some typical figures tomorrow from previous analyses . In the past have tried varying sample injection sizes and differing split ratios but the RSDs of 8% or more still occur . The machine is over 3 years old but from day 1 this problem has been evident , have never had decent analyses from this GC . I was not at company when GC first bought and Varian seemed sure it was method or user problems rather than machine . Now my boss is reluctant to spend maybe £5000 on an autosampler until he is sure GC is functioning properly . [/u][/url]

[xxx]

Is the 8% RSD relative to your standard area or relative to the ratio area standard/area ISTD ?

Bye

[BG]

Hi sheepy98

SORRY i made a mistake i understood you had an autosampler.........but in an other hand you should buy one !!!!

[sheepy98]

Thanks for all the feedback . The following is a series of typical analyses from this GC .

Triallate = 0.0952 g } in 50 ml
Di Ethyl Phthalate = 0.1019 g } Acetone

Counts Counts RATIO

1. 14903 7232 = 1.92521
2. 13419 6545 = 1.91546
3. 17014 7584 = 2.09590
4. 12674 6037 = 1.96135
5. 15627 7593 = 1.92276
6. 16153 8282 = 1.82214
7. 16163 8603 = 1.75523
8. 17843 9687 = 1.72084
9. 14364 6581 = 2.03914

As mentioned in first note these figures are typical , yet when an engineer replaced the Electronic Flow Control unit the following was achieved :

1. 11431 5653 = 1.88916
2. 14550 7237 = 1.87831
3. 18105 8906 = 1.89934
4. 16036 8223 = 1.82192
5. 11395 5792 = 1.83801
6. 18387 9151 = 1.87718
7. 17961 8835 = 1.89927

As good figures as we have ever had but ultimately the RSD reverted to the type of level as first figures . I believe something key has been altered whilst changing EFC unit but what I dont know , any suggestions ?

Thanks.

[Rick Jagielski]

One thought,

An easy way to eliminate/reduce the variability from manual injection technique is to use a needle only injection. If you flush the syringe with the sample until you are sure that it is representitive, then perform the injection with no volume, just the needle filled with sample, you can leave it stuck through the septa for a set time.(A few seconds for very volitile substances, or longer for high bp materials.) The sample that vaporizes out of the needle should be very reproducible.

Hope this helps

Rick

[JI2002]

One more suggestion: Increase injection volume by 5x (1.0 ul), use splitless instead of split, but need to decrease the sample concentration.