Reproducibility GCMS (GCMS-QP2010)

[bobjakobs]

Hi,

I am not a regular GC user, but lately I am trying to measure the activity of my catalyst, using GCMS. I want to follow conversion vs time for several substrates. I was hoping that I could skip making a baseline for all products (which I then have to synthesize and purify) by the following method:
-Inject reaction mixture at t=0 containing my reactant and an internal standard (hexadecane).
-Follow decrease of reactant peak relative to internal standard as time goes by and reactant is consumed. I don't do anything with the upcoming product peak in my GC spectrum.

I would think this method gives pretty reliable results, as long as concentration are in the lineair regime of the MS detector.

My conversions vary between rougly 4 and 30 %, which is the actual problem.
To test the reproducibility of our system using an internal standard, I injected a solution containing tetradecane and hexadecane 10 times. The ratio of the area of the two is .79 (which is fine), but the standard deviation is 2.1% of this ratio.
Of course this deviation is unacceptable if I want to substract the reactant peaks at 4% conversion. I always thought the reproducibility is way higher than this.
I detect using single ion mode, for the three highest peaks in the mass spectrum of C14 and C16.

What could be the problem of this rather low reproducibility, or is this about the best I could get and should I look for other methods.

Thanks in advance,

Bob

[bobjakobs]

By the way, the GCMS I am using is quite state of the art (I think), it is a Shimadzu GCMS-QP2010, so that should not be the problem.

Thanks again,

Bob

[Peter Apps]

My money is on injection and inlet problems, but that is just guesswork in the absence of any information on set-up and running conditions.

Peter

[bobjakobs]

According to our analytical staff, the used method is ok. But if you tell me what kind of information is important, I could post that.

[Peter Apps]

Inlet and column temperatures, type of inlet liner, manual or automated injections, solvent for the standards and concentration of target compounds, volume injected, split or splitless injections, if splitless the splitless time, column phase and dimensions, transfer line temperature.

Do you have an FID as well as an MS ?

Is your target analyte (the reactant) chemically simialr to a stright chain alkane ? and is its boiling point simialr to C16's, if not you need to select another internal standard.

Peter

[bobjakobs]

[quote="Peter Apps"]Inlet and column temperatures, type of inlet liner, manual or automated injections, solvent for the standards and concentration of target compounds, volume injected, split or splitless injections, if splitless the splitless time, column phase and dimensions, transfer line temperature.

Do you have an FID as well as an MS ?

Is your target analyte (the reactant) chemically simialr to a stright chain alkane ? and is its boiling point simialr to C16's, if not you need to select another internal standard.

Peter[/quote]

Inlet T: 250C, column T:170C. Type of inlet liner: will check, don't really know. Automated injections. Solvent is hexane, concentrations are around 40 mg/L. 1uL injected using a 10uL syringe (autosampler). Splitted injection: ratio 100. Transfer line T if that is the same as interface T, 300C
Column phase and dimensions, I will check, but I think it is Zebron-5, 15 m, 0.1 um, 0.25mm column
We do have a FID detector as well, but on a different setup, could check that.
My reactant in this case would be diethyldiallylmallonate (DEDAM), which is chemically not really similar to C16, but its retention time in GC is pretty similar, so that's the reason I took it.
But the deviation I was referring to was using only C14 and C16, which are rather similar of course. I would say that ratios of the surface of these two would be really reproducible.

By the way, surface area's of the individual compounds (C14 and C16) vary up to 15% while injecting the same vial.

Thanks, and I will check indicated things asap (probably tomorrow)

[Ron]

There may be more than one thing going on here. What is the event time in you mass spec parameter table? Are you peaks smooth, or can you see a change in slope at each data point?

The interface temperature shoudn't need to be over 260C for this analysis, although that shouldn't cause problems with the reproducibility.

What is the flow rate and head pressure of your method?

What autosampler is being used?

[Peter Apps]

If my calculations are correct you are putting 0.4 ng of each alkane onto the column, which is getting towards the low side. Do you need to have the sample solution so dilute ?, and can you decrese the split ratio to say 20:1 ?.

Check the scan rate on the MS - if it is too low you get too few points across the peak to give a repeatable area count - the peaks will look like a series of straight lines rather than a smooth curve if you blow them up on the screen.

I still think that the inlet liner is a potential problem.

Peter

[bobjakobs]

Ok here is the extra info I found so far

Autosampler: Shimadzu AOC-20i
Event time: 0.05s (peaks look really smooth)

The actual column is (put it in wrong yesterday): Phenomenex Zebron ZB-5ms, 30m, 0.25 mm and 0.25 um
Flow rate: total flow: 142.7 ml/min column flow: 1.38 mL/min velocity: 45.0 cm/s purge flow: 3.0 mL/min
Pressure: 135.4 kPa

Ion source T: 200C
Detector voltage: relative to tuning result is selected. not absolute (not sure what this means)
Threshold (of MS): 500

C14 comes out at 2.2 min, C16 at 3.8 min, and this time is really reproducible.
Overall, I think the peak area of the first measurement is almost allways the highest, then it decreases pretty rapidly during the ones that follow, and then stabilize a bit. Sometimes however, I get this sudden (usually) decrease in peak area, which can be as high as 20% The area never really seems to recover from this, and fluctuates around this new level.

Furthermore, it looks like sensitivity is decreasing over the long range, a period of a few weeks, can result in a decrease of a factor of 5.

What do you think the standard deviation of this ratio I am talking about should be? If the optimal is not below, let's say 0.2% than I don't think it is usefull for my measurement and I should think of something else.

Should I still ask for the type of inlet liner?

Thanks

[bobjakobs]

I could easily go higher in concentration, or decrease the split ratio. However, peak area is now around half a million, which seems pretty nice to me.

[Peter Apps]

If the system is set up and operated properly you should be able to get an order of magnitude more precision than you need.

The raw peak areas are in arbitrary units, what is important is the signal to noise ratio, and you can most easily increase that by injecting a more concentrated solution and/or decreasing the split ratio.

YES, the inlet liner is important (that's why I asked). The symptoms you describe point to non-reproducible discrimination between the two alkanes. This can be fixed by appropriate choice of liner and liner packing.

Are you doing the repeat injections from a single vial ? - if you are a vacuum might develop in the vial, which causes variable injection volumes.

Peter

[Ron]

I'll bet if you look at the baseline of one of the extracted ions there is no noise showing at all. Set the threshold to 0 and see what the baseline looks like. At that high a threshold in SIM mode I would be very surprised if you aren't cutting off the bottom of the peaks. This is not good for precise measurements.

[bobjakobs]

So here is what I tried:

Decreased the interface temperature to 260C, put the threshold to 0, put the split ratio to 20 and opened the screw caps of the vials a little bit, since I inject from the same vial everytime.

The result is that the baseline of the individual spectra still looks rather good, I have some small noise up to a few thousand, while my peaks are in the few million regime. The peaks look a bit tailling, bit that's what I always see.
The peak area now slowly increase with time (so opposite of what I saw earlier), leading to ultimately a 25% higher peak area of sample 10 compared to sample 1. Since I opened the vials a bit, it might be that some solvent evaporated, but by the level of the solvent after all measuments are completed, this is really limited, and definitely not in the 25% order.
Still the standard deviation of the ratio between the two peaks is 2.2%.

The question about the inlet liner type is still pending...

Thanks for your help so far, I hope I can fix this problem ultimately.

[Peter Apps]

Maybe the quickest way to find out what type of inlet liner is in there is to open it up and look !! And if it takes three days for the instrument caregivers to asnwer a simple question I wonder how up to date they are with inlet maintenance like liner replacement and septum changes

Peter

[Peter Apps]

Another thought - does it make any difference if you work with the total ion scan rather than SIM ?

Peter