The last weeks, i have been working on a method to detemrine mineral oil.
It was going fine when i went on my week vacation. In that week, the only thing that happened to the GC was a maintenance by Interscience, and they only did an alignment of the autosampler, so they did nothing on the column or injector or detector.
I come back, and suddenly ALL my values are +- 18% lower then before.
Is it possible that this "drop" in respons is due to the column being relatively new (we started working on it 4 weeks ago)?
By DaT on Friday, April 25, 2003 - 06:32 am:
Hmm, maybe I can add that none of my injection settings (injection speed, sample size, ...) or temperature program had changed either.
So the only possibility i see is that the column needed some time to stabilize for a reason i can't understand (yet?)
So if anyone knows what this could be, then I would appreciate it if you let me know what caused this :)
By Anonymous on Friday, April 25, 2003 - 07:47 am:
can you supply more details, column type, size, operating temperatures, etc.?
By Les on Friday, April 25, 2003 - 09:36 am:
Maybe the alignment included a new syringe. Check the syringe is properly installed. The vol of syringe is same and the injection volume the same.
By DaT on Monday, April 28, 2003 - 12:47 am:
We work with large volume injection. So first we have an uncoated column of 10metres (solvent = hexane) then this pre-column is connected to the analytical column through the SVE. Column type is a 95%dimethyl-5%difenylpolysiloxane cappilary column.
Length: 30metres, ID: 0,25mm and df = 0,25µm
Operating temperatures: i start at 60°C and i end at 300°C
The alignment did not include a new syringe, and even if they had done that, we changed syringes ourselfs in those 4 weeks and we never had problems then.
Well, problem is a big word, I remade my calibration curve, and am redoing the validation as we speak (the change in response happened just before i finished the validationso gotta redo it). But I would like to know why it happened because this is something they will ask when i have to bring forward my results.
I have already checked every setting of my temperature program, my injection, autosampler and detection. Nothing has changed. So I myself am thinking of 4 things:
- loss at injection. But the loss is the same for
"heavy" and volatile components, so not likely
in my opinion
- "problem" at detection. Maybe for some reason
not all the carrier gas (He) and components
are burned at the FID?
- the SVE: maybe the SVE valve is not functioning
properly, but since the loss is 20% for each
component this is not likely. If the SVE stays
open longer or shorter this would mainley
influence the volatiles.
- the column itself?
I am just saying my thoughts here. IT could be that some of my thoughts aren't very logic, but i don't have much experience working yet so some solutions i gave might very well be almost impossible.
By Les on Monday, April 28, 2003 - 04:53 am:
I've had FIDs that got partially blocked with material and caused response drops.
By DaT on Monday, April 28, 2003 - 06:33 am:
Wouldn't such a block build up over time, meaning your respons drop getting higher each time you measure?
By Anonymous on Monday, April 28, 2003 - 09:17 am:
Did you change the amount or the gas used for makeup gas in your FID? Did you bump your ratios for the gases, air, hydrogen, etc for the FID?
Did you change cylinders or the gas supply in some manner without knowing that you did?
Have you checked your valve for leaks or were you possibly backpressuring the loop previously without knowing it from a partially block sample vent line which then unblocked at some point?
By Anonymous on Monday, April 28, 2003 - 12:01 pm:
A company did maintanence on a GC and didn't even change the liner and septum? Sounds a little strange to me. Are you positive that nothing more was done than autosampler alignment?
By DaT on Monday, April 28, 2003 - 11:31 pm:
1) kame up gas is at same flow and pressure as before
2) bump my ratios for gases means? (sry, don't really get the word "bump")
3) changing the cilinders didn't happen, but even if we did, the GC itself keeps the flows steady itself (providing you give a minimum of pressure of course)
4) leakcheck, hmm, maybe i might do that :p I will do that today
The reason interscience went here for a maintenance is: It is a rather new GC, and the autosampler is not working properly. With that, I mean the AS sometimes takes the wrong vial (+- once every 20 measurements) and sometimes it just misses the injector port/vials septum/enters ial and turns on/ causing the needle to bend (once every 50 measurements +-). That is what they came to solve (and failed at it too btw). We do such an alignment ourselfs 2 times a week now to avoid the needle missing what it is aiming for, and we had np with bended needles for 8 days now. So that works (we think). But we think the AS got a software bug though.
Yet this has nothing do to (i think) with a respons loss. I checked the volume being sucked up in the needle during, ... and it performs that correctly.
Further more, I don't know what you mean with "liner" but we don't use a liner. We inject straight onto an uncoated column, whereas a liner (or so i have been thought) is not a piece of column.
I will try that leaktest (i'm wondering whether it would be this, if so, then stupid of me not to check it earlier)
By DaT on Tuesday, April 29, 2003 - 06:30 am:
Passed the leak check. So that wasn't it.
Maybe it is the detector after all, but i'm not checking that untill i got my validation fully done once.
If i check the detector afer that, and the respons goes to its original level, then i can always do the LOD and reproducability etc faster then a full validation.
I want to thank all who tried to help ;)