Life-time of a GC-TCD

[Narsil]

Hi everyone!

I am working on a GC with a TCD for several months and I am wondering what the average life-time of a TCD is. Usually the major concern is a not appropriate baseline noise but because I am not the first person on this GC I do not know if the noise got worse from time to time. Is there any guideline?

Thanks for any information!

edit:
"Auto-zero" is far away from zero, is that normal?

[Ron]

The lifetime is very dependent on proper use. Turning off carrier and makeup flow while the filaments are hot will kill a TCD quickly. You want to run the minimum current necessary for the sensitivity you need. Higher current than needed will decrease the life of the detector and increase the noise level.

[Narsil]

Yes, i think so, that the combination of temperature and current is really critical.
...but isn't it, the higher the current, the higher the signal and the better the detection limit or does the S/N ratio increase as well and therefore the detection limit stays the same or even worse?

What do you think is better....lowering the current and rising the detector temperature for better detection limits or the other way around...or both as low as possible?

Thanks for help.

[HW Mueller]

Narsil, you apparently wanted to ask whether noise increases rather than the S/N ratio??

[Ron]

I would run the detector as cool as possible and a low current. The sensitivity is better at low temperatures. The noise increase is usually roughly proportional to the signal increase, so even though the signal is higher the signal to noise may not be higher. The best way to find out is to experiment a little with your sample, change one parameter at a time. and try to maximize your signal to noise.

[Narsil]

@ron, thank you!

@hw mueller, yes, I meant noise increase and S/N ratio decrease

[AICMM]

Narsil,

Sensitivity is proportional to current. Filament lifetime is inversely proportional to filament current. Filament current is also dictated by the gas being (for most vendors.) Cooler detector is more sensitive but you must keep it above the final column temperature to avoid condensation. One thing that has not been noted here so far, TCD is a concentration dependent detector so flow rate through the detector is a critical factor. Lower flow means more sensitivity if you do not sacrifice peak width too significantly. For example, don't use make-up unless you have to.

REgarding your zero, it sounds like you may have poor balance of flows between reference and carrier but without further information it is difficult to say.

Best regards.

[AICMM]

Narsil,

Sensitivity is proportional to current. Filament lifetime is inversely proportional to filament current. Filament current is also dictated by the gas being (for most vendors.) Cooler detector is more sensitive but you must keep it above the final column temperature to avoid condensation. One thing that has not been noted here so far, TCD is a concentration dependent detector so flow rate through the detector is a critical factor. Lower flow means more sensitivity if you do not sacrifice peak width too significantly. For example, don't use make-up unless you have to.

REgarding your zero, it sounds like you may have poor balance of flows between reference and carrier but without further information it is difficult to say.

Best regards.

[Narsil]

Thanks for the helpful comments!

I always tried to increase the signal but I forgot to have a look at the noise - which in fact also increased.

Concerning the TCD lifetime: I think a major problem is/was that we are using nitrogen as a carrier gas which has to have a lower level of the current at a respective temperature compared to helium and that several people worked at that GC. Therefore I cannot guarantee that the TCD was always flushed at least 10 min prior switching it on to get rid of remnants of oxygen....Might that influence a zero shift or a lower S/N ratio?

[AICMM]

Narsil,

Depending on what you are looking for, switching to helium or hydrogen could help your sensitivity substantially (without much of an increase in noise, if any.) Which leads to, what is it that you are looking for? Fixed gases or something else? If something else, is it a solvent injection because you might want to use a run time event to lower the filament current for solvent then raise it for analytes of interest.

It is entirely possible that what others have run may have damaged your filaments...

Best regards.

[Narsil]

I am analyzing hydrogen, methan and carbon dioxide in one single run. Therefore I have to take nitrogen as a carrier gas because helium has a similar thermal conductivity than hydrogen.
It is the only way I suppose, however, I have to life with the worse sensitivity for carbon dioxide because nitrogen is similar to that.
As you can see, it is not that easy.

The TCD is about 4 years old/like the GC - I hope it might run a few years more. I am not sure if a replacement will help.

Is there a or does someone probably know a respective value of a common noise in µV/mV at the respective carrier gas nitrogen 5.0, at a given current (30 mA) and temperature of the TCD (150°C) or is that not possible to estimate?!

[AICMM]

I am starting to sound like a broken record here but why does everyone try to do this application by TCD? It is so easy to do this by HID that it is not even funny. Great sensitivity to all components, really easy to set up, extremely robust method.... Yeah, helium may be a little more expensive, but how important is it to see (and integrate) less than 5% hydrogen?...

Sorry, I will get off my soap box now....

By the way, absolutely nothing personal Narsil, just that this problem rears it's head all over the forum.

[CE Instruments]

I am starting to sound like a broken record here but why does everyone try to do this application by TCD?

Because TCD GCs are far more common and can be picked up for next to nothing. An HID costs $$$$$ from new and there are few available second hand. Add to that the massive sensitivity means that leaks and contamination require lots more attention

To respond re TCD life, they go on forever as long as you do not turn the filaments on with no carrier gas I have seen plenty of 30 year old GC/TCD still running

[AICMM]

CE Instruments,

An new HID for $2500, if you have an old FID electrometer hanging around does not seem like too much money if you compare it to the headache of trying to measure hydrogen at the same time as other fixed gases. It is really cost effective if you are considering two TCD's, one for hydrogen and one for everything else. Then it will beat that price. Especially after you add in a second valve and second column for the second gas channel.

Yes you have to be leak free but you don't have to be pristine by any shake of the imagination. I use copper tubing, and standard flow fixtures from a 5890 or 6890 and I don't have any problems on mine. You also don't need more than five 9's helium and I don't use a getter either. Besides, a system that is leak proof enough for a TCD could very easily be not leak proof enough considering the poor sensitivity of the TCD in many applications.

If you are looking at a couple of hundred ppm to percent levels, which a lot of applications do, then the extra noise of an HID doesn't matter. Not that it is THAT much noisier (about an order of magnitude) if you are leak free. I have looked at 0.5 to 1% of hydrogen on a TCD and you can barely see it. That's a great peak on my HID. I have also looked at 20 ppm with a 30 uL loop if you need to do low levels.

The only significant issue with an HID in this application is working near 100%. Then your peak will be distorted and non-linear compared to a TCD but a lot of people don't care about the near 100% peak.

Back off my soap box...

Don't get the wrong impression. I still believe in the TCD as a detector for certain applications, especially where you cannot use helium. However, the use of TCD for measuring hydrogen and.... is not one of them in my opinion.

Best regards.

P.S., I have run my HID out of helium a couple of times (typically over a weekend) and once I restore the helium flow to it on Monday, it starts right back up. Yeah you have to bake it, mostly to get rid of water (water, water everywhere....) but the detector itself will be fine.