 Hello, Nanak Martens again from Gent University and PIB. We're now at the fourth lecture in our series on mass spectrometry basics. We talked about amino acids and proteins and their properties, how they influence the behavior of the anilines in the mass spectrometers, which gives us the ability to measure them. We talked about the general concept of a mass spectrometer, how it was built, that was the second lecture. We talked about ion sources, that was still the subject of the second lecture. We had electrospray and matrix-assisted laser desorption ionization. Then in the third lecture, we talked about the analyzers, we talked about the time of flight, we talked about the ion trap, and we talked about quadruple. And then we had a little digression about resolution and why it matters. And it matters because it gives us the charge state of molecules, because the mass spectrometry analyzers can only give us mass over charge and they can never give us mass by itself. So we need to determine the charge in a separate way. And that was important because electrospray sources, which are very popular, do not give us charge states. So we're right at the end of the mass spectrometry, we've kind of gone through the instrument the way the sample sees the instrument. And we're now at the detector, which is the thing that will finally record the actual presence of something. And then that presence is then through timing or voltages or other things that we know about, the system about the analyzer, we translate that into a mass of charge. So let's see how a detector works. This is not how detectors are built anymore, but it helps to illustrate exactly how it works. Basically, it's a very simple thing. It's an amplifier. Think about it this way, we're pushing ions through a system, a tube or a box or what have you, and these ions, they generate electrical currents. So if we were to have very sensitive measurement devices, we could measure the current generated by these ions passing through our detector, and this current actually gives us our signal intensity. And that's what we've been talking about. We talked about how essentially over time you see current coming through, and that's how a detector works. The problem is our detectors by themselves are not sensitive enough. We need amplification. We need to take this very weak signal coming from these very few ions in our mass spectrometer, and very few means millions of them, or hundreds of billions of them, so very many, but still it's not a lot when you look at it on a scale of everyday usage. We need to boost this signal. And the way you do that is with an amplifier. So let's have a look how these amplifiers work. Essentially you use something known as a Faraday cup that are organized in dynodes, two electrodes. I'm being a little bit facetious that I'm assuming a single ion coming in. In real life, a single ion would never be enough to generate a meaningful current at the end, but for the sake of argument, let's stick with one ion. So an ion would come in and we put a very small voltage on one of these dynodes, on the first one in the chain. Now, the ion sees this dynode and is charged and so it rushes towards it. We've seen this many times. By putting an electrical field, we can motivate ions to go in a certain direction to attain a certain kinetic energy, at a certain velocity. Now, unfortunately, this is also where the story of our analyte ends. Okay? So because the analyte is going to literally bump into this dynode, it's going to make quite a bit of a smash when it hits the surface of the dynode. In fact, there is sufficient energy, and this has to do with the tweaking of this voltage that is in the right range, to knock free a few electrodes. Now, since we have another node, it's a dynode, with a higher voltage somewhere, these electrons or positrons are now attracted to the second dynode and they start accelerating towards that electrode. Day two are going to hit that electrode and every time one of these two, three, four electrons hits the plate, they're going to knock free a further few electrons, which are then attracted to the next plate and the next plate and the next plate. And what you can see is a cascade. You get an exponential amplification. One ion yields two electrons, yields four, yields eight, etc. So you get very many more. It's just like a runaway reaction in a nuclear facility. You get ultimately a very large amount of electrons out and the amount of electrons you get out depends on how many get kicked loose at every hit and of course the number of steps you take. So the idea is that you take a small signal, very, very weak undetectable, but you actually make it into a detectable signal that you could see on an oscilloscope, which is essentially what our detectors are. They are oscilloscopes and then remember there's a digitizer that scans this oscilloscope over time and gives you a fairly good idea of what happens when and the faster it does that, the higher the resolution. Obviously, this is not how they are built these days. They are built in manual materials where you have silicon chips and the silicon chips have little bubbles in them and these bubbles actually act as dinodes and it's all very complicated. So while this is not a realistic representation, it does show you exactly what is going on. Now, having said that, there are some limitations and this is what I want to spend the majority of this lecture about. I want to tell you what the limitations are of any detector, not just the ones we use in mass spectrometry, but in anything, even our eyes or our ears or our skin, which are all detectors for light, sound and touch. Before we get into these limitations, let me first tell you why this is important. Why do we want to talk about limitations? We want to talk about limitations because we actually use the intensity information in our mass spectra for good purpose. We claim that if we see a signal like this, which is an isotopic envelope for one of our peptides, and another signal like this, which is say the same peptide, but with a spike in of a heavy compound, we can actually do that by the way, we can add neutrons to some of the atoms that we use in the synthetic versions of a peptide, which makes them heavier. So they move on the M over Z axis. In fact, it moved here by one, two, three, four daltons. So we made this one four daltons heavier. We did that in the lab. This is not in real life, but it gives us a reference. And suppose we know just how much of this compound we spiked into the sample. We can now say this one is roughly half of that in intensity. So since they're the same molecule, the only difference is that this one contains a few extra neutrons, which should not change the physical chemical properties apart from the weight. We can say if this is say 10 nanograms per milliliter, then this one will be five nanograms per milliliter, you see, because it's half the signal. If the signal would be roughly equal, and do notice it's roughly equal, it's not exactly equal. Obviously, there's going to be a little bit of noise. There's going to be a little bit of variation. And so this is as probably as good as it gets to equal, just like this isn't exactly half either. There's always this measurement in. These are roughly equal. So we could say this is present at the same concentration as this. And since we know this at 10 nanograms per milliliter, this will also be 10 nanograms per milliliter. And here, obviously, we have the opposite situation. 10 nanograms per milliliter for spike-in, artificial, heavy neutron analyte. And this is the real biological analyte that we are interested in. And we can see that there's two times as much. So this will be 20 nanograms per milliliter. This is this theory. This is what we want to achieve. We want to be able to quantify the amount of analyte in our sample by relating the quantity to the signal that it generates on the master chrono. Now, that works, but it doesn't work all the time or everywhere. So let's look at that. Any detector, including our eyes, our ears, and our skin, and anything else we've ever built, or nature has ever built, follows a certain type of response curve. Think of it this way. If you're in a very, very dark room where there's not a single photon of light, it will look dark and completely black to your eyes. Obviously, there are no photons, so your eyes cannot see anything. Now, if I were to bounce, say, five photons around in that room, it would probably still look dark, because five photons simply isn't enough for your eyes to perceive anything. You may need 100 photons before you start seeing something. So if this is the axis of the number of photons, and this is the axis of the response of your neurons in your brain, you can say that if the number of photons is too low, there will be no response. And it will remain so until you hit a certain threshold from which point onwards your neurons will start to fire. And so it will be flat, and then it will start to go up. And for a little while, it will be such that if I put twice the number of photons in, your eyes are going to say, oh, this is twice as bright. And so you get what we call a linear response for a while. And then suddenly, your eyes will stop being able to see any differences. So for instance, I'm now looking at a projector, and this projector is extremely bright. So to me, that looks completely washed out white. If I were to look into the sun, which would be a really bad idea because it can damage your eyes very quickly, it would also look completely washed out white. And in fact, to my eyes, it wouldn't look brighter than the projector. But to my actual physical structure of my eye, it would be much more destructive because the intensity of sunlight is much higher than the intensity of the beamer light. But my eyes are unable to perceive that difference. So what we get is first we see nothing for a while, then we start to see stuff, and we are kind of capable of saying it's twice as bright. And then suddenly, we get to really bright things, and then it levels off again. And whether the beamer is brighter than the sun, I cannot tell. I will be able to tell after a while because I go blind from one end of the other, but at least I cannot tell from a glance. My eyes are saturated. It is exactly the same with sound. With a very small amount of sound, you will hear nothing until you suddenly will hear something, and then you will be capable of telling the difference between something that is loud, like my voice, or say twice the volume of that sound. And you will be able to hear that. But if a fighter jet passes by, or somebody detonates an explosive, the explosive is probably going to make more sound, but you will not be able to tell the difference. It's just going to hurt and it's going to wash out the ability of you to hear anything. So again, we have a detection curve that follows this S-shaped curve. And that is what we call a sigmoid curve because it looks like an S, like a sigma in Greek. And this is very typical of any detector. It is therefore also typical of a mass spectrometer detector. So this is what you see here. We are actually looking at a log 2 scale of a measured ratio of two components, like we saw on this slide. We're actually measuring the ratio between two components at a log 2 range of what we measure versus what we know. This is the expected ratio. We've spiked this and we know exactly what kind of concentrations there are in the sample. And when it's zero, it means one over one because the log 2 of one is zero. And you can see we measure reasonably close to zero within tolerances. We measure a lot of compounds so we can do error bars. But as soon as we go away from zero and we go to here negative log ratio, so here there is less of the measured ratio, then what we can see is that it levels off. Can you see that? So it does not follow a linear curve. It follows a sigmoid curve and the same at the top. So it levels off. And this is very important to realize because our detectors really do work like that. The second thing you may notice is that not only can we not tell the difference between a large or a not so large difference anymore, we can also see that the error bars increase. So our measurements become less precise and they also become less accurate. We no longer see two times more as two times more here and each measurement has a larger margin of error. So essentially we want to measure somewhere here where the mass spectrometer is really good at measuring things. You do not want to measure things here because the mass spectrometer is much worse at measuring things and it's an inherent limitation of any kind of detector. Let's look at this again. We're actually scanning here from minus three point something to three point something, which is essentially one over 10 to 10 over one. So it's a 10 fold difference. So between here and here we got a 100 fold difference. That's two orders of magnitude. This is a reasonably old experiment that we actually did in 2005 and published in 2007. Let's look at a more modern version. This is a curve that was made by Marc Faudel and it has a one over 100 to 100 over one ratio. So the previous chart showed us this range and now we're looking at this range. This is on the log 10 scale. What you can see here is that this is the error on the measurement. It stays more or less within boundaries close to one but it gets worse and worse as you go away from one. We've seen that the previous graph went to here and then you can see that it really balloons once you get outside of this range. So the error becomes huge. So the reliability of measuring something at less than one over 10 is going to go down really quickly and the same thing goes for measuring something that is over 10 over one. You can go one step beyond that. This is work done in Ghent by Anstaz and Evie Timmermann and they actually went from one over 100 all the way to a thousand over one. So this is new compared to the previous and what you can see again is this beautiful sigmoid curve which is reasonably flat because we have such an enormous range here and you can see on these box plots the errors are smallish here but they get bigger very quickly with many, many outliers here and here and very few of them here. So what this teaches you is that if you want to measure an absolute quantity your spike in should ideally be close to one over one which is kind of counterintuitive and counterproductive in a way because we do need to know roughly at what concentration a certain peptide or protein should be in a sample before we can match the spike in. So of course you cannot do that. You're interested in the amount in the sample. You don't have a clue about what it could be so what people do is they try a range of options. They spike in our synthetic peptide which we use to measure the original one at concentration x and then 10 times that, 15 times that, 30 times that and then also one over 10, one over 15, etc. So you just make a range of possible spike ins. You measure each of them and you go and make a series of measurements. So you start off with the sample analyte being much more intense than our spike in and then over time this will grow and then it will start to look like this and it gives you a complete curve. You can do a regression on that. We can get a much more accurate measurement but that is really crucial if you want to do it right because our detectors are imperfect and if you happen to spike in somewhere here or somewhere here your measurement is essentially useful. So keep that in mind these detectors are not perfect and there's nothing you can do to change that. We've looked at more than 12 years' worth of detector evolution over these past three charts all the way from 2005 to 2014 and what we see is the detectors are not getting better. They have reached their limitations. I'll tell you one more thing about a detector which is funny. Remember that you had this electrode and then the ion hits it and then things fly off and they result into more things flying off and so forth. You could wonder can this plate continue to have things fly off of it because at some point the plate must become depleted. There should be nothing left to fly off the plate and that's exactly what happens. So when you use a detector over time the sensitivity of that detector which is the amount of signal it gives out depending on the amount of signal coming in will start to go down. If I put in exactly the same amount of ions on the top of the detector over time the signal the detector will give will start to go down. So the detector is losing its capability of ejecting enough stuff every time something hits the plate. So what people do to compensate for that is just like what you would do in the stereo is you just turn up the volume. They essentially continuously add amplification so they put higher voltages on the plates and that actually gives a bigger signal. Now the problem is just like your car stereo if you turn it up too far it will start to crackle and it will start to sound really bad because you've exceeded the limitations of the device. This happens with the detector too. When it becomes sufficiently depleted it will no longer be possible to continue to turn up the volume if you like the amplification. So at some point you have to replace the detector and that's literally what happens. They take the old detector out take it away hopefully for recycling and they put in a new one. So if you study say a standard sample that is being done for quality control over time in a mass spectrometer and you look at the signal what you will see is you will see a deterioration that is compensated and then at some point you will see a spike and that spike is usually the new detector but you will see this fluctuation and something to keep in mind detectors are not inherently stable either. Okay so these two things I think are very important if you want to do quantification because they affect what your result will do. With that we come to the end of this fourth lecture and in the next lecture we'll talk about Fourier transform, ion cyclotron resonance and orbitrap instruments because they're a kind of special beast they actually combine the analyzer and the detector in one. So that will be for the next lecture. Thank you very much and see you then.