Tuesday, October 23, 2007

An Emailer looks at the chromatograms

An emailer sent the following. We've inserted the illustrations, captions, other links, and emphasis. The text was previously a comment in the Monday Roundup.

I am a technical lead in a radiochemistry laboratory. We are involved with a completely different type of spectroscopy and analysis. I am not expert in anything to do with organic analysis.

I am not by any means a chromatography expert, and small apparent discrepancies may be normal for the technique..... But, the graphs displayed in the majority decision look odd and inconsistent.


Fraction F1:

1. The 5AC peak is a relatively small peak in the GC/MS, but nearly equal to the peak attributed to the analyte in the GCCIRMS than the GC/MS.

Majority Award page 31; circle added; click for bigger.
The 5aAC is about 1/2 the height of the 11k.

Majority award Page 32, circles added, click for bigger. The 5aAC is about the same height as the 11k

In my unexpert eyes, either:

(1) These are not aliquots from the same sample, or,

(2) two instruments are so different that relative peak height is not a reliable indicator of pattern.

2. The GC shows two distinct doublets at about 20.4 and 21 minutes. The GCCIRMS shows 4 evenly spaced peaks. If the GCCIRMS was simply higher resolution, you should still have 2 sets of double peaks.

Majority Award, page 32 zoomed in; click for bigger. Two major peaks.

Majority Award, page 32, zoomed; click for bigger. Four peaks.

Either :

A). These are different samples, or

B) the characteristics of the two instruments are different such that the elution rate is not consistent. Peaks could even swap positions.

3. The A sample GC/MS shows peaks at 9.6 and 13.2 that are much stronger in relative intensity than the B sample.

Majority Award page 31; A sample peaks at 9.6 and 13.2 minutes; click for bigger

Majority award page 37, B sample peaks at 9.6 and 13.2 minutes. Click for bigger.


A) The separation chemistry was very inconsistent between samples, or

B) the A&B samples are from different people.

Fraction F2:

1. The small peak at 16.6 on the GC/MS shows a relatively small shoulder on the right side and both A and B are similar on the GC/MS. On the A sample GCIRMS, the doublet has changed such that the large and small peaks are the same size, however, on the B sample GCIRMS, the doublet looks similar to the GC/MS. ??????

Majority award, page 33, A sample F2 peak at 16.6 minutes has a right shoulder fractionally smaller than the main peak. Click for bigger.

Majority award page 34. In the A sample F2, the corresponding peaks in the F2 GCCIRMS are close to the same height, so peak heights don't seem to correspond. Click for bigger.

Majority award page 39, B sample F2 GCMS peaks at 16.6 are about the same as the A sample, with lower right shoulder. Click for bigger.

Majority award page 40, B sample F2 GCCIRMS peaks at 1400 have same relative height as GCMS, which is different than the A sample GCCIRMS, where they were nearly the same height. Click for bigger.

2. The peaks at 16.6 and 18.2 on the GC/MS are about ½ the height of the analyte peaks, but are relatively much smaller in the GCCIRMS.

Majority Award, page 33. A sample F2 GCMS peaks are about 1/2 size of the target peaks.

Majority award page 34, A sample F2 GCCIRMS peaks are much smaller than in the GCMS. Peaks heights don't seem to be relatively the same between instruments. Click for bigger.

3. The two analyte peaks on the GC/MS are very close together and spaced from the larger peak at 15.2. On the GCCIRMS, all 3 peaks appear to be evenly spaced.

Majority award, A sample F2 zoomed in; two peaks spaced from third major. Click for bigger.

Majority award page 34, zoomed in. Peaks are about evenly spaced.

Given F1 #2, above, could these peaks have swapped positions?

4. The peak at 15.2 looks to have a low side shoulder. Could the shoulder have moved under the Andro peak in the GCCIRMS? What if the GCCIRMS shoulder grew in intensity?

Majority award, page 34. Peak at 15.2 has left shoulder; Click for bigger.

5. Does the Eito peak have a high side shoulder, or the Andro peak have a low side shoulder?

Majority award, page 33, zoomed in. Potential right shoulder on left peak? Left shoulder on middle peak? Left shoulder on right peak?

Fraction F3:

1. On the GC/MS, the 5b-pregnandiol peak is broader than the 5a and 5b diol peaks.

Majority award page 35, A sample F3 GCMS. Pdiol peak looks wider than other analytes.

I think it may actually be 2 superimposed compounds?

2. It is extremely difficult to tell which peak on the GCCIRMS is the 5AC peak.

Majority award page 36, zoomed in. Which of these is the 5aAC peak?

3. The B sample peak at 12 min. is missing on the A sample.

Majority Award page 34, A sample F3 - very small peak at 12 seconds. Click for bigger.

Majority Award page 41, substantial peak at 12 seconds. Click for bigger.

4. The observed difference in peak heights in peak heights between the GC/MS and GCCIRMS (above) combined with the shift in peak position also described above, would make me concerned about the small baseline jitters around the 5a and 5b diol peaks.

Majority Award page 34, A sample F3 GCMS zoomed in. Baselines around the peaks of interest jittery?

Majority Award page 41, B sample GCMS zoomed in. Jitters in the baseline?

All said, I am not at all a GC expert. Maybe such details are normal for the field. I hope you guys can spend a bit of time in the lab and work it out.


By the way, in our lab, we look for exotic inorganic elements. If anybody were ½ as sloppy as the LNDD paperwork, I would expect to be fired, along with several others. If I saw that paperwork, I wouldn't give it a plug nickel.


Mike Solberg said...

It's taken me a while to understand this level of detail, but I think I get it now. It seems the two main points are that peak height in GCMS is clearly not always repeated in GC-IRMS, and that it looks to be possible that peaks could even switch order or position between the two machines (although this is not as clearly demonstrated in the graphs shown. But how would you know? If peaks have switched order, and you can't id them based on height, and you don't have reference materials run in IRMS to give you the retention time, you wouldn't have any way to tell what was what).

Those two facts (as revealed in LNDD's own work on Floyd's samples) would seem to show that matching of visual patterns between GCMS and GCIRMS is not a reliable way to identify the substances involved, and certainly not a reliable way to guarantee there is no "contamination" - more than one substance in a peak, especially one with shoulders, etc.

So, what was Brenna thinking?