I have encountered a GC-headspace sampler problem and could use some help. We have been using an HP 6890 GC with an HP 7694 headspace sampler and an FID detector. Our method requires that the linearity correlation coefficient of three standards injected before and after the samples be greater than 0.995. Normally this requirement is easily met with R values of 0.999 and 1.000 for most runs. Occasionally however, the system fails. The response of the second set of standard injections is sometimes as low as 80% of the first set, and the R value for the set of six injections is consequently less than 0.995. The drop off in peak response is a puzzle. I am a little new to GC (especially headspace analysis) and a little uncertain as to what the most likely cause of the problem might be. I've been advised that the FID is not likely the source of the response changes. That leaves the headspace sampler. I am aware that changes in headspace carrier gas flow rates have large effects on the amount of sample introduced into the GC, but I don't know whether it is likely that those flow rates might change during the course of a run or what might cause such changes. All other aspects of the chromatograms are fine. Retention times and peak shapes never vary. Any thoughts? Thanks.
By Anonymous on Thursday, September 19, 2002 - 02:11 am:
Maybe you can check the connection between sampler and GC. Sometimes it is leaking.
By Rod on Thursday, September 19, 2002 - 05:25 am:
The most common cause of this error is a mistake in making the stds.
The second most common cause is improperly sealed vials.
The third most common cause is a valve rotor or other leak.
Another cause is contaminated lines and/or column which will adsorb a certain amount of analytes each injection which will curve your regression line.
The fact that retention times are consistent indicate that the analyzer is injecting properly and the GC is OK.
Check the compliance with making the stds. If that is OK and it probably is, then the most likely problem is the crimper of your vials needs to be re-calibrated for a tight non-leaking seal.
Your analyzer I believe uses a fixed loop injection from a pressurized vial. Make certain these components are working correctly.
Senior Research and Development Chemist
Gas Separations Research
595 North Harrison Road
Bellefonte, PA 16823
By Harvey G. on Monday, September 23, 2002 - 08:45 pm:
I have used HS/GC extensively. The major problem I have run into regarding inconsistent area response counts, has been exclusively traceable to the caps, vials and septa leaking. This does not mean that the other sources of errors mentioned are not valid. They are. Every one of them. However, it has been my experience that the caps, vials and septa are the main culprits with head space analysis. When you clamp the cap you should not be able to turn it by hand. If you can, it will leak. When it comes from the instrument, you should still not be able to turn it by hand. If you can, it MAY have leaked. If you notice any curving of the septa, up or down, it will tear upon injection and you will lose pressure from a leak around the needle. The curve is caused by over tightening the cap. After the vials have been run through check the septa. If there is a small round hole you are probably OK. If there is any indication that the point of injection is a small straight line instead of a round dot, then the septa tore upon injection. You need to match caps with septa and vials such that they will give a tight seal yet not distort the septa. Also examine the needle. A build up of hard material on the outside of the needle can cut the septa after the point has penetrated. If possible use an internal standard technique. Otherwise a surogate standard can be used to correct peak areas for leaks. Unfortunately, the most important headspace analysis I perform reqires standard addition. I have to prepare a minimum of 6 vials for injection to obtain one result, with confidence for each sample. Two leaks and I am preparing 6 more vials for that sample. I am very fussy about purchasing vials, caps and septa. The last few times, I drove from Toronto to Life Sciences in Peterboro with my capping device in hand and tried each combination of caps, septa and vial they had before placing my order. In the last 3 years, I have not had to prepare a second 6 vials for any sample.
If you want more information, contact me directly through e-mail by clicking on my name at the top of this memo.
By Anonymous on Tuesday, September 24, 2002 - 07:31 am:
To add to what Harvey said above, all crimpers are not created equal. Some crimpers that work fine with one vial, septum, and seal combination will give nonreproducible results.
By Greg Burtenshaw on Wednesday, September 25, 2002 - 10:16 pm:
I too have had a very similar problem recently. We run a HP6890GC with ChemStation and have been seeing a response about 10 times less than we expect. This application uses a 600ml sample jar and 45mm septa with a manual injection onto a capillary column.
Just to make matters worse every now and then we check the same sample four or fives times and get the expected result once or twice. It does not seem consistent.
We have changed liners, o-rings, gold seal, column, FID injector, checked flows, leak tested and tried 3 different syringes, re-made standards... all to no avail.
If our problem was caused by sample introduction/handling then we would see varying results not consistently 10 times less then the odd good result.
Any help would be greatly appreciated.
By Anonymous on Thursday, September 26, 2002 - 03:23 pm:
Use automated Headspace analysis.
Anything else may be VERY irreproducible. The absorption in the needle metal, glass, and plunger components causes great variation. Also the temperature of your needle may vary which affects the transfer of the analytes of interest.
You would not use a hand pump to pump your mobile phase for HPLC would you?
The factors needed to get good answers with manual HS analysis are too varied for the casual user to master. (IMHO) Good luck.
By Nazri on Thursday, December 26, 2002 - 07:08 am:
I am having the same thing. But could it be because, I have the same temperature for all levels of calibration?
I am using a 16 solvent calibration, from alcohol to aromatic hydrocarbon. The alcohol shows a beautiful linear line but the aromatics, at certain calibration points, tends towards a polynomial curve insteaad.
Is this normal ?
I am very inexperienced in HS-GC
Thanks a zillion
By Rodney on Thursday, December 26, 2002 - 07:42 am:
You will minimize this problem by using stds freshly made before use. Use the same temperature for all your std samples. Make certain you do not have leaking seals, check your crimper for proper adjustment.
I have made stds containing solvents from 1ng per vial to 25mg per vial with no problems if you follow the above.
Good luck and best wishes in the New Year
By Nazri on Friday, December 27, 2002 - 06:33 am:
Thanks for your kind reply.
I did as you had described above whenever I attempt calibration. I make 10 different amounts, from 10ul to 100ul of the standard.
The thermostating temperature and time is the same throughout. The crimper is adjusted the best I could, with the above reminders in view.
Hence, with all these parameters in place, I still get a quadratic polynomial curve for acetates and beyond (esters, aromatic hydrocarbons) while the 'light' ones such as methanol and propanols give me a linear curve when all the ten points are plotted.
So again, am I on the right track?
Thank you and warmest regards for the comin New Year
By Rod on Saturday, December 28, 2002 - 05:18 pm:
As always it seems to be on this forum, questions are asked without enough information given to return solid answers.
How do you make up your stds? How long after being made do you use them? How much heating, temp and time, do you use? How are the vials sampled and injected?
Are you placing the same total amount of liquid in each vial?
By Nazri on Tuesday, December 31, 2002 - 09:17 am:
I have a vial containing 16 different solvents which is supplied by my headquarters in Rickenbach.
Normally, the calibration standards are like I'd say, from 10ul to 100ul. So using a syringe, I withdrew them from this vial and pump in the required amount into separate vials. Immediately after this, I test the 10 samples containing different amounts of the 16 calibration solvents.
Heating will be at 80 degrees, time will be 60 mins and this is for ALL vials.
Please help me out here buddy.
Feel free to ask me anything else that you might wanna know
By Anonymous on Tuesday, December 31, 2002 - 01:10 pm:
Are you overloading the column or detector at the high concentrations? This will cause nonlinearity in a calibration. You have a linear plot for the most soluble component, alcohols, which do not partition efficiently into the headspace, and nonlinear response for the less soluble compounds which will partition well. This makes me suspect overloading.
By Nazri on Thursday, January 9, 2003 - 06:18 am:
I mix the 12 calibration solvents (Ethanol, n-propanol, iso-propanol, ethyl acetate, isopropyl acetate, n-propyl acetate, iso-propyl acetate, Ethoxypropanol, toluene, ethyl benzene, p-xylene, o-xylene) at 0.3mg each into a 100ml flask. Then I top up with Triacetin till the whole mixture reaches the 100ml mark. Basically, this is my calibration solvent.
By Rodney on Thursday, January 9, 2003 - 12:03 pm:
You place 10 to 100無 into ??mL vials? 20, 12, 6mL?
How much (dissolution?) solvent is then added to each vial? What is this solvent?
Your calibration solvent is 3痢/mL of each of 12 solvents listed above. How much do you add to each vial? What else is in each vial?
60 minutes is excessive unless you have a lot of sample in each headspace vial.
Are you placing the same total amount of liquid in each vial? This can affect the linearity of some solvents more than others.
Feel free to email be directly if you do not wish to discuss details publicly on this forum.
Senior Research Chemist
Bellefonte, PA USA
rgeorge @ sial.com
By Nazri on Wednesday, January 15, 2003 - 05:14 am:
Thanx for the help.
Ok, let me describe what i'd been doing so far. Before i inject the calibration solvent into the separate 20ml vials, ranging from 10ul to 100ul each (10 vials), I first make the cal solvent as a mixture of 16 individual solvent altogether.
This 16 individuals, @ 0.3mg each are mixed in a 100ml flask. I top up the mixture to the 100ml mark with Triacetin.
Basically, all the solvents are equal in terms of the quantity.
I used 60 minutes because, the main analysis I'd be doing will be on printed materials, in which the thermostating time are set at 60 minutes so the calibration have to be using the exact parameters.
Thanks for the kind e-mail
By Rodney on Wednesday, January 15, 2003 - 06:43 am:
You will achieve better linearity if you add the same volume of your carrier solvent (triacetin) to each vial. Add 90無 of triacetin and 10無 of std to one vial, add 80無 of triacetin and 20無 of std to another vial, etc and finally add 100無 of std alone to last vial.
Your linearities should improve.
By Nazri on Friday, January 17, 2003 - 05:11 am:
I think when I mixed the solvents up and topped it in the 100ml flask thoroughly, I am already having the same volume throughout didn't I?
If I should be doing this manually by adding 10ul and having to top up again with triacetin in the sample vials again later, this is going to be less accurate wouldnt it? cos ethanol is one of them solvents.
Also, like i said, i need 10ul in one vial up to 100ul in the last vial which makes me having 10 separate vials containing amounts in steps of 10ul each, and not 100ul for every vial.
Did I explain clearly? Sorry if my explanation had been dreadful.
By Rodney on Friday, January 17, 2003 - 05:45 am:
You are varying the amount of liquid in the std vials are you not? Varying the total amount of liquid in the vials from 10 to 100 無?
If you are doing this, then this is the primary source of your non-linearity.
Each vial must be the same except for the amount of analyte (I do not mean the carrier solvent - triacetin).
Vial 1 should contain 10痢 of ethanol in 100無 of Triacetin.
Vial 2 should contain 20痢 of ethanol in 100無 of Triacetin.
Vial 3 should contain 30痢 of ethanol in 100無 of Triacetin, and so on until you have 100痢 of ethanol in Triacetin in the last vial.
You cannot have differing amounts of carrier solvent in each vial and expect to show linearity of response.
You cannot have differing amounts of carrier solvent in each vial and expect to show linearity of response.
I have tried to be clear. I apologize if I am not.
Headspace analysis requires the equilibration of liquid and gas phases to be linear. You cannot form the same equilibration between liquid and gas if you vary the amount of liquid in the vial or change the size of the vial in a series of stds.
if you varied the sizes of the vials in your series of stds would you expect the response of the various stds to be linear? Of course, not.
One more time: each vial must contain the same amount of triacetin. Only the amount of analyte dissolved in the triacetin should vary.
See my paper of in the Journal of Chromatography
Vol 69 No.11 pages 2221-2223 for details concerning my preparation of stds from 0.5痢/mL to 1000痢/mL and added 1無 of each std solution to 25無 of water in each vial.
By Nazri on Monday, January 20, 2003 - 06:01 am:
A zillion thanks for being patient in guiding me along.
I shall try it as soon as I can and revert back to your kindself.
Once again, thank you very much
By Vem on Saturday, January 25, 2003 - 06:10 am:
Guys, wouldn't this make life easier if you run internal stds and that way who cares about leakage.
By diana beck on Thursday, June 3, 2004 - 09:24 am:
Hi ! My opinion : when the needle is penetrating the septa is possible that a micro particle of the silicone or the rubber goes into the needle and create not "perfect" assumption. Diana.