Ion Trap question

[Ken]

Hi there. I recently encountered an interesting observation on the bruker ion trap instrument and would like to know if anyone can advise further.

Issue: Unable to obtain correct nominal mass of compound at 905 m/z at negative ionisation. The m/z obtained was 906 with a +1 mass error.

Instrument was well calibrated using the tuning mixture; even at high masses > 1000 m/z

When run with other working standards ie in the same mode - negative mode but lower m/z at 203, the mass is correctly identified.

Thus, can anyone advise why the mass at 905 is not correctly identified where else the instrument passed the the tuning mixture and also working well with lower masses?

Can the analyte of interest at m/z 906 be affected by the sample matrices thus mass is not correctly identified? Or...?

Please advise if can. Thanks a lot ~

[ods-at-pacific]

You do not give the elemental composition of the analyte. Is it possible that the error in the m/z assignment is due to the hydrogen mass defect That would be my guess.



Want to learn more about mass defect; consult Mass Spectrometry Desk Reference, 2nd Ed.

[Ken]

Dear David, thanks for the reply.

May I just ask if ion trap is more prone to have this mass defect issues? Especially with bigger molecules that have higher amount of hydrogen atoms?

I don't encountered this with the linear ion trap.

Please advise.

[ods-at-pacific]

All of the nuclides of the elements most often encountered in organic mass spectrometry (H, C, N, O, S, Si, P, and the halogens and Na and K when considering pos. ion ES) have mass defects with the exception of C-12. The mass defect of a nuclide is the difference between its integer (whole number) mass and its exact mass as determined relative to the exact mass of C-12 being defined as 12 u (unified atomic mass units). Hydrogen, deuterium, C-13, N-14, and N-15 all have positive mass defects (the exact mass is > the integer mass). All the nuclides with integer mass > 15 u have negative mass defects.

The whole number m/z values you stated in your original post probably represent the nominal mass of single charge ions.

The nominal mass of an element is the integer mass of the most abundant isotope of that element. The nominal mass of a molecule or ion is the integer mass calculated using the number of atoms of each element in the molecule’s or ion’s elemental composition times the nominal mass of the individual elements.

The monoisotopic mass of an element is the exact mass of the most abundant isotope express to a number of decimal places. The monoisotopic mass of a molecule or ion is calculated in the same way as the nominal mass but using the element’s’ monoisotopic masses.

Depending on how the mass spectrometer is calibrated and/or whether there is a mass defect correction applied to the acquired data before they are stored can result in a difference between an observed value and a nominal mass. Take the case of the molecular ion of C50H102. This ion has a nominal mass of 702 u; however, the observed m/z value for the single charged ion is at 703 using a quadrupole mass spectrometer that reports to the nearest integer value; i.e. 102 x 1.007825 = 102.7982, which when expressed as an integer is 103. The monoisotopic mass and the nominal mass of 50 atoms of C-12 is 600 u.

I know that the Xcalibur software has a 100 millimass units per 100 m/z-units correction as the default for the storage of acquired data. This value can be changed; however, most people never do because they do not understand mass defect. This may explain why you don’t have this miss-mass assignment problem with the linear ion trap which is using Xcalibur.

If an instrument is calibrated using oligomers of ethylene glycol (-CH2CH2O-) which have a monoisotopic mass of 44.0262 u and a nominal mass of 44 u, the calibration point at nominal 880 (20 oligomers) will have a exact mass of 880.524 u which could be rounded to the integer 881 u, depending on how the instrument rounds. This shows that the mass defect is important with respect to the calibration compound used and the algorithm used for rounding the observed values. In the Agilent ChemStation for GC/MS the rounding is done to the integer existing between – 0.351 and + 0.649. This not only helps with the positive hydrogen mass defect but also the negative mass defect due to multiple atoms of bromine. The 3D QIT does not respond differently to mass defect than does the LQT, the TOF, or any other type of mass spectrometer. It all has to do with calibration and what is used as the calibrant and correction applied before data storage.

PFTBA is used as the calibrant for most GC/MS instruments because the various CF ions have essentially no mass defect even at m/z 502 and m/z 614. This means that the monoisotopic mass of a PFTBA ion with nominal mass of 614 is almost exactly the same value (613.965 u).

Go back and calculate the mass of your ions based on monoisotopic masses and round this value to the nearest integer. If you are are going to use mass spectrometry, I recommend that you take some courses or least get a good book so that you can answer these questions on your own. Any time you have ions that have a mass >500 u, mass defect can be an issue. This iswhy you need to clearly understand the topic.

[lmh]

[rant mode]
I get really angry about this whole concept of mass defect. It's really very very simple. No element (apart from 12C) actually weighs an integer. We live in an era where any reputable lab almost certainly has chemical drawing packages that add up exact mass without rounding decimals. If you draw your structure and use such a package, the software will tell you what the molecule really weighs, not some silly approximation left over from the days when people couldn't add up decimals. Mass spec instruments measure the real mass of the ion. The two should be the same. No further complications necessary. No one should need to mention the words "mass defect" ever again.

Mass spec software is aware of the concept of mass defect, because some software knows that if a person types in 703 then they maybe want a mass window biased up a bit because by the time you reach a mass of 703, most typical organic molecules are a bit heavier. I do not believe, though, that any mass spec actually measures 703.5 and reports it as 703 as a result of some internal correction for mass defect. This would be utterly unjustified because the instrument has no right to assume what elements the user is studying. If I'm working on uranium salts, I will need very different mass defects to a person working on copper cluster ions, or a person working on things with a large alkane content. We all want to measure the actual mass of our analyte.
[/rant mode]

On the other hand, if, taking accurate mass into account, your measured mass in a well-calibrated ion trap is wrong, there are a number of possible good reasons. First reason: was your trap over-filled? If a trap contains too many ions, they begin to influence one another. Instead of experiencing only the electric field of the trap, ions also experience the electric field of nearby ions. This is called space-charging, and reduces resolution, while also distorting the peak-shape in a spectrum. The distortion can move the middle of the peak away from the correct mass. There are probably many other reasons why a trap might give bad measurements, but this one I know because I've done it too often for comfort.

[ods-at-pacific]

Imh,

I find what you say to be interesting. Maybe you can help me.

I drew the structure of bromobenzene in my drawing program and requested it to Calculate mass. It returned two values. The first was labeled the Molecular Mass and was 157.01 Da. The second was the Exact mass of 155.957461 u. I am using a Varian ion trap GC/MS that reports m/z values to the nearest integer. I want to determine if bromobenzen is in my sample. One way to do this is through the use of a mass chromatogram. What m/z-value do I use, 157, 156, or 155? The NIST Mass Spectral Database says the Mol. Wt of bromobenzene is 156.

Then I got interested in octobromobiphenyl ether. My drawing program listed its Molecular mass as 801.38 Da and its Exact mass as 793.357254 u. What m/z value do I use for a mass chromatogram in my Varian instrument (which reports to the nearest integer m/z value) to see if OBBE is present in my sample?

Using my Varian ion trap GC/MS, I observe a molecular ion peak at m/z 801. Based on the Nitrogen Rule, does that m/z value tell me that the analyte has an odd number of nitrogen atoms?

Maybe this concept of mass defect does have some significance after all.

[lmh]

Ah! I suspect you know a lot more about this than I do, but since I had the rant, I'm obliged to accept the consequences, and your quite appropriate challenge!

For brominated compounds, the major difference between molecular weight and exact mass isn't going to be anything to do with mass defect (in the sense of difference between integer mass of an isotope of an element and the true mass). The major difference is caused by the isotope peaks. A drawing package will return 155.9575 as exact mass because that is the exact mass of the most abundant isotopologue (note: I think isotopologues are structures with different heavy isotopes in them; isotopomers are structures with the same heavy isotopes in different places, but don't trust me, try IUPAC). The Molecular Mass of 157 (Chemdraw: 157.0079) is just the mean mass of all isotopologues present, weighted assuming natural abundance of all elements, so it's completely irrelevant in mass spec provided your resolution is sufficient to resolve isotopologues (i.e. you can see the isotope peaks, which you should do unless you're a protein person at high charge state). There may be no peak whatsoever at the molecular mass of a compound. Your measured mass is the exact mass plus or minus mass of any adduct and steps of about 1 for isotopologues.

Your Varian trap should be seeing pairs of almost equally intense peaks 2 units apart for bromobenzene parent ion. You will obviously choose which parent isotope peak seems most suitable (from the point of view of getting maximum S/N ratio at adequate specificity).

Octobromobiphenylether is going to have a disastrously huge number of isotope peaks. Any spectrum of it will look like a cross-section of a hedgehog having a bad spine day. As regards a molecular ion at 801, I rather feel that with big delocalised things, especially those heavily substituted with electron-withdrawing groups, it's not unthinkable that they could form a reasonably stable radical ion, so I wouldn't feel confident using the nitrogen rule (which breaks down when you can't be sure whether the ion is odd-electron or even-electron).

As regards Varian ion traps rounding their measurement to the nearest mass unit: I had no idea they did (I haven't used a Varian trap). I work mostly in LC-MS, where doubly charged ions are quite common. How do you round a mass that genuinely does fall exactly between two integers? Wherever you put the division between two masses, some things will fall just on the border. I really feel, strongly, that instruments should report the actual mass they measure, and leave it to us (or data-analysis software) to sort out what this means, using our knowledge of the resolving power, precision and accuracy of the instrument. Life's hard enough without having to guess how a measured value has been mangled and distorted before being given to me!

Sorry if I've caused upset, happy if I've merely sparked debate...

[ods-at-pacific]

Sorry for my delay in responding. I have had a lot going on the last few days with the start of two different Web courses on mass spectrometry.

You are correct, I do know a lot about this subject. Your post did not upset me; I think it is good to stimulate discussion. I don’t think you can dismiss mass defect as an unimportant consideration in mass spectrometry. You are correct that a mass spectrometer will measure whatever the actual m/z value is of an ion. Electron ionization (EI) databases are built on the basis of nominal mass and a value returned by the mass spectrometer for the mass of ion must be converted to nominal mass before the spectrum can be searched against such a database.

You were correct to point out the potential problem of space charge resulting from too many ions in a quadrupole ion trap. This is a very real problem. My experience is that there is a series of peaks, 1 m/z unit apart on either side of a peak with a maximum intensity which may be 1 to 2 m/z units greater than or less than the actual m/z of the ion; the so-called Christmas tree effect. Such an overload problem is obvious and cannot be mistaken for a miss-mass assignment due to mass defect.

I hope you find this discussion worthwhile. BTW the nominal mass of octobromobiphenyl ether is 794 u. Unless mass defect is considered, the integer mass peak of this nominal mass peak will be at m/z 793. The 801.36 Da value is the average molar mass of the compound based on the atomic masses (atomic weights) of the elements time the number of their atoms in the elemental composition.

The molecular ion peak at m/z 801 represents a hydrocarbon with a nominal mass of 800 u. Before the Nitrogen rule can be evoked, the m/z value has to be adjusted for mass defect.

I enjoyed this discussion.

[lmh]

I think your point is that mass defect is basically a set of rounding rules that you have to use to make modern measurements match up appropriately to unit-resolution libraries? I suppose I've been lucky to come into MS in an era when accurate mass was becoming accessible to more than just the specialist few with an FT instrument. Also my entry has been via LC-MS, where unit-resolution EI-libraries like NIST aren't greatly relevant; modern LC-MS libraries and databases tend to contain accurate mass information.

This hurts (NIST and its ilk have done so much good...) but I have to say it: it was never, never, right to save library data at unit resolution, and it still isn't right, even if we're forced to by history. It should, at least, be accepted as a historical fudge and not a fundamentally right thing to do.

It is absolutely basic in measuring anything that you need to measure (and record) with a little bit more precision than the minimum difference that matters to you. If you're checking the calibration of a pipette that needs to be correct to 1uL, you would never use a balance that is only correct to 1mg. Simplistically, the sum of acceptable errors on the pipette and balance can give you an answer that's wrong (pipette is out by 0.4uL upwards, balance out by 0.4mg downwards, balance thinks that the pipette is 0.8uL wrong and reports, to nearest uL, that the pipette is delivering 1uL too much, when it was actually in specification). If unit resolution was necessary for searching, EI libraries should have stored the first decimal place, and allowed for errors when matching experimental data to stored spectra, based on the precision and accuracy of both the instrument that collected the standard library entry, and the instrument that collected the sample spectrum to be searched for. But it's easy to say in hindsight...

Thanks for your thoughtful reply again.

[ods-at-pacific]

You are one of the growing number of people that use mass spectrometry without having or wanting to have an understanding of the science behind it. There is probably nothing wrong with that, but there is a lot that you may miss-interpret in pursuing such.

Mass accuracy and mass resolution are NOT the same. The mass accuracy is how close can the mass of an ion be measured compared to its exact mass as calculated from standard tables of isotopic masses determined relative to the most abundant isotope of carbon. Mass resolution is the differences in the m/z values of two ions that can be separated.

A transmission quadrupole may report the m/z value of an ion to the nearest 0.05 unit; however, the instrument has a maximum resolution of 0.3 m/z units. If two ions have the same nominal mass (defined in an earlier post in this thread) of 43, they could be a pair of acylium ions (exact mass 43.0185 u or a pair of propyl ions (exact mass of 43.0548 u); a difference of 36 mmu. It is also possible that there could be one of each which would have an assigned accurate mass of 43.0367 u, if there were an equal number of both. The mass spectrometer would have insufficient resolution to separate the propyl and acylium ions and by virtue of the fact that it takes ten measurements across a single m/z value, the mass accuracy for either monoisotopic ion would be such that both would be reported as having m/z 43.05.

If you look at the elements normally encountered in organic mass spectrometry (also stated in an earlier post in this thread) you will see that only the mass defect of hydrogen is significant for ions with a mass of < 500 Da. Much of what can be eluted from a gas chromatography has a mass of < 500 Da; therefore, building libraries that have integer mass peaks that were obtained from mass spectrometers that only reported integer mass due to their resolution and ability to assign masses was not only the RIGHT thing to do, it was the prudent thing to do. Electron ionization (EI) mass spectral libraries have been being built and used in the identification of unknowns since the 1960s, long before you started using the technique. Yes there were instruments that had the resolution and ability to assign accurate masses at that time (double-focusing magnetic and electric sector instruments), but there were many more unit resolution unit mass accuracy instruments in use and have been for many years. The TOF LCMS with a mass accuracy of 10 mmu was introduced in 1997, < 15 years ago. These early instruments did not produce very accurate isotope peak intensity data. That has only improved in the last 4 to 5 years and is still not universal among all instrument manufacturers.

For an ion that has a monoisotopic mass of 750.4 Da, there are > 600 elemental within 4 mmu. I believe this will push the accuracy of most commercial TOF instruments. Doing library searches against mass spectra where the m/z values are reported to nearest 0.01 m/z unit can be dangerous. If the peak from your instruments has an m/z value that is 0.05 units different than that in the library, should that peak in the library spectrum not be considered?

Before you criticize someone for what they have done, give it a little more thought to what they have accomplished. Do you have any idea how many CORRECT identifications of unknowns have been done using the various commercial EI mass spectral databases that have the masses recorded to the nearest integer?

Consideration of mass defect is important in the application of the Nitrogen Rule. Nominal mass is fundamental to the Nitrogen Rule. I use both everyday in the identification of ions obtained by EI, CI, ESI, etc.

[lmh]

I am so sorry to have caused offence, but please do re-read my post, and your reply (which I'll admit I find unfair and hurtful).

(1) I am aware of the difference between accuracy and resolution. It was sloppy of me to refer to NIST as a unit-resolution library, but I did so because it was created for instruments that were generally reckoned only to resolve things that were a unit apart. I correctly referred to modern databases and libraries as accurate mass libraries. In fact it makes no sense to use the term resolution for a library, as resolution is a property of an instrument. Accuracy and precision are terms that apply to libraries and instruments; NIST has a precision of 1 unit. My argument is that it should have had a precision of 0.1 units if the smallest mass difference that mattered to anyone was 1 unit.

(2) Also, although transmission quadrupoles can only resolve things a unit apart, the middle of the peak has been accurate to better than that for some while (from personal knowledge at least 15 years).

(3) I'm also aware of how many good identifications have been achieved using NIST and other similar libraries working to unit precision. I even apologised before criticising NIST. However, the fact that NIST-style libraries have served us brilliantly over a long period doesn't make them perfect. We have to look for weaknesses in things from the past (even small weaknesses in otherwise very good things) if we're to do things better in the future - and now that we have really good ToFs and orbitraps etc., there is a strong drive to do things better.

(4) I think you've missed the point of what I was getting at with library searches. To me it's a very simple statistical process. The ideal library (that I can envisage, at this point, and I don't mind in the faintest if this looks wrong in 20 years) contains masses that are precise, with known accuracy (i.e. 100.1234 measured on an instrument that has a precision of +/- 0.005). This is easy to search: if I have a measured mass of 100.1237 on an instrument with precision of +/- 0.002 it's a simple matter of statistics to work out the likelihood that my ion is significantly different to the original measurement.

If, on the other hand, the original measurement is rounded to 100, and my measurement is rounded to 100, I can't make any statistical comparison whatsoever, and I've wasted the fact that I had a precise, accurate, and highly-resolving instrument.

[ods-at-pacific]

Imh

I apologize if I have offended you. My reply was meant to inform and educate. I, and hope others, have enjoyed this back and forth. Unfortunately, we may have drifted off topic and the poor guy with the miss-mass assignment in his quadrupole ion trap may be no better off.

[lmh]

me too... sorry, I rather hijacked this thread... I've certainly learned quite a bit (one always should).

[Peter Apps]

I've certainly learned quite a bit.

Me to. One small point I'd like to add is that "identifications" based on integer resolution spectra (whether they be library searched or interpreted) ALWAYS have to be backed up by something else such as co-elution of an authentic reference if the substance is available in that form, and if the nature of the problem is such that this would provide a sensible answer to the question of whether or not the substance actually was present inthe sample. For example, if you run a selective sample prep looking for pesticides in foods, run the extract under specific conditions known to be appropriate, get a peak with a spectrum that library matches to Deadylate, spike the sample with pure Deadlyate and find an enhancement of the peak with no change in the spectrum, you are probably justified in reporting it as present. To definitively identify an unknown (as in bioprospecting or my own field of semiochemistry) would require an integer spectrum, an accurate, precise mass (to agree with the proposed empirical formula), NMR if you can accumulate enough of the unknown compound, and a specific synthesis (followed by a bioassay, but that's biology, not chemistry).

Peter

[Don_Hilton]

And some of us remember the "inovation" of reducing mass spectra to a small number of the most abundant ions (at unit mass values) to save library space and computing time. But that was in the days that the data system was an oversized desk calculator....