Page 1 of 1

Understanding the detector + "spillover"

PostPosted: Tue May 21, 2019 8:30 am
by jimbomahoney
Hi all,

I see a lot of reference to +/-1 spillover and +16 spillover.

My understanding is that the +16 is coming from oxidation of that isotope. I totally get that.

However, I want to understand the +/-1.

I had initially understood it to be because the metals that Fluidigm produce cannot be 100% pure and will always contain 1% of the +1 and -1 masses, which then leads to "spillover" when detecting.

On the other hand, a comment Mike had made about a sample being contaminated with La139 from somewhere else but still showing "overspill" in the +/-1 channels got me wondering:

1) What "resolution" does the detector have? (i.e. what is the smallest fraction of mass it can discriminate?).

2) How is this "overspill" actually happening? (i.e. is this a physical effect of the detector / something to do with saturation?)

Bear in mind my background is with 2D detectors like EMCCDs, so I'm thinking in terms of impact ionisation, saturation etc. I have no idea how the detector in a mass system works.

Re: Understanding the detector + "spillover"

PostPosted: Tue May 21, 2019 3:04 pm
by mleipold
Hi James,

The +1 and -1 are generally from isotopic impurities in the metal salts. As you might imagine, it's difficult to separate the M+1 from the M during purification.

Some of this depends on the isotopic distribution, though: if, say, you have the following natural abundance distribution:
1. M-1 = 70%
2. M = 20%
3. M+1 = 10%

Then M is probably going to have more M-1 impurity than M+1 impurity.

I should also note: this M +/-1 nomenclature is a bit misleading: not every element in nature has isotopes in complete sequence. For example, Indium only has 113 and 115 (no 114). Therefore, you'd be talking about +/-2. This is also a case of the above example: 115In is ~96%, 113In is ~4%. So, you can get isotopically purified 115In that's about 99-99.5% 115In, but the most pure I've seen 113In is about's not economical to try to get it more pure than that. This is also one of the reasons why 102Pd is so expensive: it's only about 1% of nat abund.

The above is separate from the mass resolution spillover, where a particular channel is so incredibly "bright"/abundant that it spills into the +/-1 regardless of whether Nature has an isotope there. Please note, this really only happens when you have signal above about 1e4. I'm not 100% sure exactly how it happens, mechanistically: for example, my *assumption* is that it's a lack of resolving power of the TOF reflectron when there are that many ions of a particular mass, but you'd have to ask a mass spec person for sure.

Basically, you largely keep the shape of the ion peak (h x w ratio). Normal intensity ion peaks usually have baseline resolution of at least 10 units (nanosecond, I think?) between adjacent peaks. However, if the height gets big enough, the peak winds up wide enough that the left leg spills into the M-1 channel and the right leg spills into the M+1 channel. I should also mention that the peak shapes are narrower at the lower mass/TOF end: 133Cs is narrower than 193Ir

I've seen this occur with 193Ir spilling into the 194(Pt) channel when the Ir staining was ridiculously bright. In that case, it was bad because we were using 194Pt as a a BC channel: we've since revamped the protocol so that if we're using 194Pt BC, we use the 103Rh intercalator instead.

I've seen this occur with 139La spilling into 138Ba and 140Ce.

I've attached examples of all these.

Re: Understanding the detector + "spillover"

PostPosted: Tue May 21, 2019 5:58 pm
by vitozan
If I remember all my analytical chemistry classes right, Mikes explanation is already excellent!

A perfect correspondence between mass-over-charge (m/z) and time-of-flight (TOF) assumes that all ions are EXACTLY accelerated to the same kinetic energy. However, the acceleration energies will in practice not be perfect, which results in a kinetic energy distribution. A reflectron will further narrow this energy distribution, but still cannot make it perfect. So the kinetic energy and thus also to the TOF of an isotope is a distribution, however a VERY narrow one. That means there is some chance for each ion to have an energy that is so different from the mean, that the TOF will fall in a bin that is assigned to a neighboring isotope mass. While this chance is very small, if you have a LOT of ions a small, but predictable, fraction of those ions will by chance end up in a neighboring +/- 1 mass TOF bin. Also if the mass calibration (TOF ~ m/z relationship) is not well done, this effect can be appreciably stronger.
I think in the mass cytometry field, this is usually called an 'abundance sensistivity' spillover (as opposed to +/- 1 spillover which is solely due to isotopic impurities).

In our empirical assessment of 'spillover' in mass cytometry, we actually found that +/- 1 spillover was virtually absent with the machine used (<0.05%): ... gure/fig2/
However, one reason why we initially started looking into spillover in mass cytomerty, was because we had a machine which must have not been setup correctly. This lead to a appreciable and consistent amount of +/- 1 spillover in a number of channels of a dataset I analysed.
Also, I can confirm seing it e.g. in Ir193→ Pt194 when having really high Ir193 levels.
Thus, while rare on a well calibrated machine, I would definitely recommend to check +/- 1 as part of spillover assessment, if only to confirm that the mass calibration was appropriate.

Btw: In the same paper we also tried to graphical capture the different sources of 'spillover' in mass cytometry in Figure 1a: ... gure/fig1/

Cheers, Vito

Re: Understanding the detector + "spillover"

PostPosted: Tue May 21, 2019 7:08 pm
by mleipold
Thanks Vito!

I should also say: the mass resolution thing (abundance sensitivity, left/right leg spillover) has changed over the models of the instrument. I remember at CYTO 2013, Scott Tanner gave a talk introducing the CyTOF2, and specifically mentioned some changes relative to CyTOF1.

Basically, when your ion cloud makes it through the quadrupole and is headed to be "deli-sliced" as almost 80kHz pushes into the TOF+Detector chamber, there's a distribution of energies within that slice. By this, I mean that in a perfect world, there would be an infinitely tight ion cloud, almost like a dot. That perfect slice would get uniformly pushed into the TOF, and uniformly reflected by the reflectron to be an infinitely thin line of ions slamming into the detector at precisely the same time.

However, in real life, will be a few ions slightly ahead of the exact center of the push-slice and a few ions slightly behind the middle (both part of the "thickness" of the push-slice), and then a few ions headed slightly more to the left of the middle and a few ions headed slightly more to the right of the middle (think width of slice). This creates a distribution even during injection, which as Vito said, can only be partly "fixed" by the reflectron.

Scott said that one of the ion optics upgrades to the CyTOF2 (and presumably, also in the Helios) is tighter control on the "left" and "right" width, narrowing that aspect of the ion distribution even before it enters the TOF+detector chamber.

In other words: if you look at CyTOF1 data, there's a good chance that +/- 1 things might be a bit different than data generated on a CyTOF2 or Helios.


Re: Understanding the detector + "spillover"

PostPosted: Wed May 22, 2019 7:26 am
by jimbomahoney
Brilliant, thanks both!