organics analysis on 5973N GC-MS

[Archimonde1308]

I am trying to separate a mixture of benzene, toluene, naphthalene and other hydrocarbons(which i can't identify) dissolved in isopropanol (i am not sure about the purity of the isopropanol. it says 99.5% pure but i doubt it tbh). I have an Agilent 5973N GC-MS system. I am currently using a restek rtx-5 30m x .25 mm column with a temperature program as follows: 40C for 2 min, ramp to 150C at 10C/min, dwell time 5 min. I am getting secondary peaks next to those corresponding to isopropanol and benzene. The MS cannot identify these peaks accurately enough (qual ~ 43). Also, the peak next to benzene does not occur when I do a blank isopropanol or a benzene-isopropanol mix sample. Is it because I the column that I am using is unsuitable for this analysis? If yes, which column would be a better option? If the column is appropriate, I would be very happy if i get a few suggestions on the right way to perform organics analysis.
Thanks in advance

[Peter Apps]

The simplest explanation is that these peaks are due to some of the other hydrocarbons in the mixture. Do you have any reason to think that they are something else ?

Peter

[AICMM]

You might consider a different solvent like methanol. You should also post the inlet conditions (temp., splitless or split, inj volumne, etc...) A -5 is not a great column for this separation since it does not hold up benzene all that much, might be better to use a -624 or even a Wax.

Best regards,

AICMM

[Archimonde1308]

Apart from benzene, i would expect to see toluene and ethylbenzene in the MS but they don't show up. Also, there is a peak immediately next to the benzene peak. Is it possible that it is a benzene split peak? The MS suggests 10 different possible compounds for this second peak but all have quals less than 50 i.e. bad fits. (toluene and ethylbenzene are not among the suggested compounds)

AICMM, I don't want to use methanol since its pretty volatile. Isopropanol is my preferred solvent since this MS is downstream a reactor which is operated for 5 hrs and then the liquids are analysed. I know rtx-5 is a pretty bad column. I will try to run some samples through a colleague's GC which I believe has the DB-5HT capillary polar column (30 m* 0.25 mm). Any thoughts on this column?

[Peter Apps]

Apart from benzene, i would expect to see toluene and ethylbenzene in the MS but they don't show up. Also, there is a peak immediately next to the benzene peak. Is it possible that it is a benzene split peak? Presumably the MS identifies the benzene as benzene ?, and so its failure to identify the extra peak as benzene is probably because it is something else

The MS suggests 10 different possible compounds for this second peak but all have quals less than 50 i.e. bad fits. (toluene and ethylbenzene are not among the suggested compounds). Bad fits are often due to co-elutions.

AICMM, I don't want to use methanol since its pretty volatile. Isopropanol is my preferred solvent since this MS is downstream a reactor which is operated for 5 hrs and then the liquids are analysed. I know rtx-5 is a pretty bad column. I will try to run some samples through a colleague's GC which I believe has the DB-5HT capillary polar column (30 m* 0.25 mm). Any thoughts on this column? As far as I know a DB5 and an Rtx-5 are very similar non-polar columns

Are you running samples or standards ? If you are running standards that you know contain the target compounds then there is a problem with the analysis. If you are running samples the simplest explanation for the lack of toluene and ethylbenzene peaks is that these to compounds are not in the samples.

Peter

[Archimonde1308]

Hey thanks for the reply, Peter. I don't mind if toluene and ethylbenzene don't show up because it does indicate that they aren't formed in the first place. I am not running standards, but the products of an experiment. The MS identifies benzene as benzene and it cannot identify the peak that partially overlaps with it. The MS also identifies isopropanol as isopropanol but again it cannot identify the peak that partially overlaps with it. One instance maybe a coincidence. When 2 compounds have partially overlapping unidentifiable peaks, it generally means something is wrong.

[Steve Reimer]

isopropanol boils at 83 degrees, benzene at 80. That isn't good for your analysis.
For your peak before benzene, does its spectrum have one or two ions much larger than the rest or a series of clusters at say 43, 57 and 71 suggesting a hydrocarbon? On my Vocol column n-heptane elutes just before benzene.

[Archimonde1308]

No. I have the unknown peak after the benzene peak. Also, it's spectrum has clusters around 27, 43, plus a single ion at 59. The benzene and isopropanol peaks are separated by a good distance so the column isn't too bad.

[Peter Apps]

Hey thanks for the reply, Peter. I don't mind if toluene and ethylbenzene don't show up because it does indicate that they aren't formed in the first place This conclusion is valid if and only if you can show that your method can detect these two compounds if they are present - to do this you have to run standards


. I am not running standards, but the products of an experiment. The MS identifies benzene as benzene and it cannot identify the peak that partially overlaps with it. if the peaks are overlapping the spectrum is a mixture, which is why it is not characteristic of any pure compound. Try doing a manual baseline subtraction to get a cleaner spectrum The MS also identifies isopropanol as isopropanol but again it cannot identify the peak that partially overlaps with it.ditto One instance maybe a coincidence. When 2 compounds have partially overlapping unidentifiable peaks, it generally means something is wrong. What is wrong is that the peaks are overlapping - MS spectral identification only works with pure compounds - i.e. clean peaks

Peter

[Bigbear]

Try slowing your ramp to 5cc/min to see if the peaks resolve better.

[Don_Hilton]

Also, keep you linear velocity in the column in a range to obtain the best chromatogram possible - I tend to like something in the range of 23 to 35 cm/sec. Use a sampling rate that gives a sufficient number of data points across the peak without over sampling - I like a few more than 8 points across a peak. And if your mixture is highly complex, get a copy of AMDIS and use it to deconvolute the spectra. AMDIS is comes with the NIST library, if you have a copy of that, or may be found on the NIST web site. If you use AMDIS, supposedly you can get by with even fewer than 8 points across the peak because of some of the assumptions that are made in AMDIS. And, keep the acquisition threshold level down to where you do see some background noise. (I am a strong proponent of a 0 threshold and allowing AMDIS to find things that might not be otherwise found.)