Not getting linear curves with ECD

[MMJ88]

Hi there,

I'm looking at Chloroform, 1,1,1-Trichloroethane, and Carbon Tetrachloride on a GC/ECD setup. The sampler is a Gerstel TDSA, a thermal desorption autosampler that allows the desorption of our sampling tubes right onto the column.

I'm getting non-linear curves for both 1,1,1-Trichloroethane and Carbon Tetrachloride. (http://i.imgur.com/UkzcB.jpg). The fit is pretty nice, but I was under the impression I should be getting linear calibrations. What factors contribute to a response like this? Is there anything I can do?

Specifics: Agilent 7890A GC, Restex Rxi 624Sil MS column (30m x 0.25mm x 1.4um), Gerstel TDSA autosampler.

[Karen01]

Hi there,

I'm looking at Chloroform, 1,1,1-Trichloroethane, and Carbon Tetrachloride on a GC/ECD setup. The sampler is a Gerstel TDSA, a thermal desorption autosampler that allows the desorption of our sampling tubes right onto the column.

I'm getting non-linear curves for both 1,1,1-Trichloroethane and Carbon Tetrachloride. (http://i.imgur.com/UkzcB.jpg). The fit is pretty nice, but I was under the impression I should be getting linear calibrations. What factors contribute to a response like this? Is there anything I can do?

Specifics: Agilent 7890A GC, Restex Rxi 624Sil MS column (30m x 0.25mm x 1.4um), Gerstel TDSA autosampler.

You might want to read up on ECD detectors ... Google is your friend.

- Karen

[Ultimate Science 101]

If you do liquid injections using the chemicals & concentrations of interest, is the ECD response linear? This might tell you if the non-linearity is a problem with your detector.

There are obviously a lot of parameters involved with thermal desorption (tube desorption flow, CIS temperature, etc.) that could be contributing to the non-linearity. Can you post the analytical conditions? This would better help people to assist you.

[Karen01]

If you do liquid injections using the chemicals & concentrations of interest, is the ECD response linear? This might tell you if the non-linearity is a problem with your detector.

ECD is an inherently non linear detector.

- Karen

[AICMM]

mmj88,

Karen is dead on - ECD is non-linear. Can help yourself some by which make-up you use but I honestly cannot remember which.... sorry.

You are also looking at two compounds with vastly different responses, CT having fantastic response and CF,,, not so much.

You can look at changing your splits (altering your column/detector loadings) if you are taking multiple tubes. Or you can investigate doing real-time dilution at the detector. This is what a customer of mine did looking at CS2 and CT in air, also with vastly different response factors. When the CS2 eluted, they split 1:2 (for example) and when CT eluted they split 1:50.

Best regards,

AICMM

[MMJ88]

AICMM- certainly you are right, my chloroform and carbon tetrachloride have very different responses inherently on ECD. Chloroform is the "more important" one in this analysis, so I'm using a split ratio that's a lot lower for what I would if I was just looking at carbon tetrachloride. Any other info on this real-time dilution? Thank you for your suggestions.

Karen - Thank you for your responses. I will be reading up more on this. Do you have any good resources on ECD you could recommend to me?

Ultimate Science - parameters below

Conditions:
- Desorption - Solvent vent mode (vent for 1 min at 50C), then desorption ramp: 50C for 1 min, 60C/min to 275, hold for 5 min. Inlet maintained at 275C
- Oven - 50C for 2 min, 8C/min to 120C, then 80C/min to 200C, hold for 3.5 min
- Injection - split, 35:1
- Carrier gas - He, 1.3 mL/min
- Detector, uECD
- Makeup gas - Ar:CH4 (95:5), 40ml/min

Thanks for the input everyone.

[Ultimate Science 101]

Certainly the ECD has a narrow dynamic range compared to a lot of other detectors, but it can still be linear within a specific range of concentrations. I assume your ECD temperature is between 300-330C?

Are the samples collected onto the TD tubes coming from a vapor source? If so, I might suggest the TDS be at 25C when the tube enters. For these types of volatiles, a tube desorption flow rate of 10-20ccm should be sufficient. How cold is the CIS injector when the tube is desorbing? It seems the colder the better to pre-concentrate/trap these types of volatiles. Can you go to -60C or lower? Also, what type of packing are you using in the CIS? Tenax, carbopack or glass beads?

Gerstel has some great applications chemist who can help to ensure your method is optimal for trapping of these chlorinated components. However, as others have mentioned in this thread, the ECD isn't always the most linear detector. I've seen some that are more linear than others. Good luck!

[MMJ88]

The ECD is running at 240C. Is that abnormally low? I apologize for my inexperience with this, I have only been using MS detectors until now.

The sample tubes in the actual study are sampled from a vapor source, but these tubes for the calibration are simply spiked on the head of the tube with 1uL of methanol with the appropriate concentrations of analytes, then room temperature nitrogen is run through for 3 min at 50mL/min.

The liner is a glass bead one.

The CIS is another can of worms entirely. I'm not using it right now. I had (what looked to be) a pretty good CIS method. The peaks were completely separated, sharp, and symmetrical. However, the response would vary wildly. I could run the same spike multiple times and get very, very different responses. A spike could be 33000 counts one run and 8000 the next. I have been in correspondence with Gerstel about this. They are very nice and helpful, but honestly did not know what was going on (neither did I). So, right now I'm not using cryocooling. The inlet just sits at 275C and the desorption flow runs right onto the column. My chromatography is not as good, but it's more reliable.

Feel free to skip over this CIS issue I don't want to drag anyone else into this craziness, haha.

Thank you very much for your input.

EDIT: If anyone does want to tackle the CIS issues, I suppose the CIS method information would be relevant (this is a separate method than the one I'm currently using and already described)
Initial Temp: -120C
Equilibration Time: .5 min
Initial TIme: .2 min
Ramp:
Rate 12.0C/sec
End Temp: 275C
Hold Time: 3.00 min

[Ultimate Science 101]

It's probably best to run the ECD hotter but the upper temperature limit of your analytical column isn't compatible with temperatures much greater than 240C. If you did want to try running the detector hotter, you could cool the ECD and install a section of Siltek guard column into the detector and then connect your analytical column to this guard column via a glass pressfit connector. Then, you could put the ECD to 300 or 330C.

Are you dosing your calibration tubes all at once and leaving them in the autosampler for analysis? Or are you spiking them and immediately desorbing? As much as the Gerstel sampler is sealed, you might try spiking and then immediately running to see if there is a difference. Otherwise, are you able to run the calibration standards with the split closed just to see if the detector could potentially be linear?

It sounds like the -120C CIS method is ideally the way to go.

Are you able to interface your Gerstel sampler to a potentially more linear detector such as a mass spectrometer? Admittedly, I've seen other ECD data that look very similar to the calibration curve you have generated (via direct injection).

[MMJ88]

I just talked to Restek on the phone, they indicated that yes, that temperature will pretty much burn the coating off the column, but since it's a very short section just going into the detector, that shouldn't affect my chromatography or detection.

As far as the procedure for tubes, I've done it both ways, no change in results.

We do not have the opportunity to run it into another detector unfortunately. I will look into what you said about running it splitless. Would I just change the method around so that my desorbing flow is 1.3mL/min, since that's what I want my column flow to be, and there are no splits/vents?

Again, thank you for your input.

[Bigbear]

Where is it non linear? If it starts to flatten out at the top ( less response than expected) I would suspect you are overloading the detector. Please be aware that an ECD is a "subtractive" detector ( signal maximum is with no compounds in the detector). Samples going through it remove signal. that is why ECD's have such a narrow dynamic range.
For giggles and laughs try doubling your split rate and see if you get better llines.

[MMJ88]

Yeah, if that were the case (flattening out at the top), I would understand that, but it's not the case. See here: http://i.imgur.com/UkzcB.jpg.

As for your other suggestion, I am currently playing with the detector temperature and split rate to see if I can increase my split while still keeping a decent LOD to see if a higher split gives me a better line.

Where is it non linear? If it starts to flatten out at the top ( less response than expected) I would suspect you are overloading the detector. Please be aware that an ECD is a "subtractive" detector ( signal maximum is with no compounds in the detector). Samples going through it remove signal. that is why ECD's have such a narrow dynamic range.
For giggles and laughs try doubling your split rate and see if you get better llines.

[Ultimate Science 101]

I guess rather than "revamping" the method to be splitless, are you including an appropriate internal standard with your calibration standards you are dosing into the tubes? If not, try including and see how the results look.

Gerstel has literature they provide regarding setup of your ChemStation/Maestro for both splitless & split analyses. I'd suggest reviewing this material or emailing them for a copy if you are interested in trying the splitless route to see if results change. It's quite likely your results may be as linear as they're going to be.

[MMJ88]

Just to follow up on this (for anyone that may come across this in the search in the future), I was able to obtain linear results for this method by using an inlet liner packed with Tenax TA. I trapped at 5C in the CIS then desorbed at 275 onto the column. This works very well for me. It avoids the problems I was having at -150C (freezing, blockage), but gives the same benefit of cryofocusing. It also uses less liquid nitrogen.