Wednesday, 24 February 2016

You want to do what?

This is another of "those" posts. The ones were I get excessively technical and start talking about ions and mass spectrometers and stuff. I build them (mass spectrometers, not ions) and they do end up taking up a lot of my brain space. Or the bits that are left over after I've had to decided what my favourite species of rorqual whale is...

So, let me introduce you to a few basic concepts of time-of-flight mass spectrometry, just to lay the groundwork for introducing you to the true depths of frustration I have at dealing with my customer.

Step1: make some ions. It doesn't really matter how, just make some and keep them in a tidy place.

Step 2: hit all of the ions with the same big cricket bat such that they all fly off down a long tube.

Step 3: count how long it takes each ion to get to the other end of the tube. The lightest ions will get there first, the heaviest ones last. By timing how long they take, you know how heavy they are.

Step 4: while those first ions are hurtling down the tube, be busy making some more in the tidy place.

Step 5: as soon as you're certain the heaviest ions have got to the other end, hit the second set of ions with the big cricket bat.

Keep doing the above, over and over and over again. Generally you can manage to do that 20,000 times a second. Quick isn't it? Each time you do it, you add the ions you've counted to a bar chart, and gradually you build up a record of which ions you're making and collecting. You might, for example, make a nitrogen ion every time, and so you'll collect 20,000 nitrogen ions every second, and get a really tall bar on your chart. Meanwhile you might only collect one sulphur dioxide ion every now and then and have maybe 200 of them every second.

There's a catch in all of this though (of course there is). Detecting a single ion is quite tricky. They're awfully small. Just as an example, there are roughly 1,000,000,000,000,000,000 of them in a grain of salt. Like I said, one is awfully small. So the equipment you use to detect one of the little buggers has to be very, very sensitive. And if something is very, very sensitive, it will have a tendency to notice other things happening. Like fleas farting. Actually, not quite that bad, but if you turn a computer on next to it, it will (falsely) believe it's just seen a load of ions. It picks up electrical noise like small children pick up rude words. So there's always low-level background noise being added to that bar chart. One or two false ions being recorded all the time, randomly scattered all over the chart. Which means that to be absolutely certain that you are actually seeing whatever you're looking for, it needs to have more real ions arriving than there are false ions.

There's some clever statistics you can use to work out whether you really believe you're seeing your interesting chemical, but roughly speaking, on one of our "normal" instruments, we get thousands of ions per second in things we're interested in. Thousands every second. Remember that please. We can, usually, manage to spot a bar on this chart and say it's definitely due to a real chemical if there are, say, at least 10-20 ions per second.

Here we go, this is what I'm talking about:

That scrubbly red line along the bottom is noise, and the spikes sticking up show the number of real ions that have been counted arriving at any given time. The things that take a long time (about 27 microseconds in this case) are the heavier ones, the things that take less time (about 14 microseconds) are the lighter ones.

Now, when you first get an instrument working, chances are it won't work very well. It will need tuning. Just like a piano - you'll get something out of it straight away, but it won't be very pretty. And to tune a mass spectrometer, the best thing to do is ask your software just to show you a running total of how many ions it's counting every second. Nice big number in the middle of the screen that you can see from the other side of the lab. And then you start knob-twiddling.

Up a bit, down a bit, left a bit, right a bit, more gas, less gas, stronger magnet, no magnet. Tweak everything while watching that total, and try and make the number as big as possible. Bingo! You've just tuned a time-of-flight mass spectrometer! (OK, to preserve some dignity and respect here, I should point out that it's not quite as random as that, and frequently takes quite a lot of thinking too, but once we send one of these things to a customer, it's all basically tuned, and they just need to tweak a few things to get everything just perfect. They might get as much as two or three times as many ions once they've fine-tuned things.)

And now we get to the ranty bit.....

I have spent the last few weeks attempting to help a customer via email. He didn't have very much money, so he refused to pay for us to install his instrument. He assured us he could assemble and commission it himself. Never, ever, ever believe a customer who says this. He has bleated about not having any signal, and I have tried every single test I can think of, most of which he has completely ignored because he thinks he knows better. Now, however, he's gone on holiday and his boss has taken over. On the plus side, his boss seems to be intelligent, competent and capable of following instructions and providing meaningful feedback. On the minus side, this has coincided with my boss getting involved because I was stomping round the office muttering about incompetent fuckwittery and he realised I needed some back-up. But now things are beginning to go smoothly, the customer (erroneously in my opinion) thinks it's because my boss is awesome and I'm useless. Sigh.

And things are also only beginning to go smoothly. We finally, in desperation, asked the customer how many ions he actually thought he was making in his tidy place.

3 ions per second.

Yes, that's right, 3. Not 3000, not 300, not even 30, but 3 piffling little ions per second.

And he's been complaining that he can't see them. He actually, honestly thinks it's unreasonable that an un-tuned mass spectrometer is failing to see this. When it was gently pointed out to him that he might have, perhaps, just maybe, not quite enough ions, he did concede we might have a point. So he ran an experiment overnight. And he finally saw some ions.

So now we have to try and tune his instrument with him.

With, say, 3000 ions per second, we can tune an instrument perfectly in a couple of hours. But then, we can watch a nice big number on screen changing in real time as we make adjustments. He has to run an experiment for several hours before he can tell if an adjustment improved things. That couple of hours it might take us will translate to hundreds, or possibly thousands of hours worth of experiments to get to the same point. Weeks or even months of work. To achieve something that should take an afternoon. I can understand why he might be a trifle disappointed.

To say that I'm irritated that he didn't mention this sooner would be an understatement.

To say I'm vexed with the colleague who sold him this instrument without discovering a fundamental problem like this would be putting it mildly.

To say that I'm annoyed that my boss has come up smelling of roses while I look like the incompetent numpty would be a pale shadow of the truth.

To say that I'm peeved that when my customers don't follow my instructions they conclude that I'm the idiot would be to wholly miss the point.

To say that I'm sick and tired of my customers blaming fundamental physical principles on me would be failing to grasp the magnitude of my ire.

On the other hand, the customer seems to be happy now, so I guess I'll just keep my head down...

No comments:

Post a Comment