 Hello, creatures. To be honest, I never thought that I would be introducing a talk on measuring radioactivity like ever in my life. But then again, considering the world's current state at large, it might be not such a bad idea to be prepared for these things. And gladly, our next speaker, Oliver Keller, is an expert in detecting radioactive stuff. Oliver is a physicist and works at one of the most prominent nerd-happy places. Ditzern, since 2013, is also doing a PhD project about novel instruments and experiments on natural radioactivity at the University of Geneva, and to even more, to add even more RC3 pizazz. Oliver is active in our open science community and passionate about everything open source. All that sounds really cool to me. So without further ado, let's give a warm virtual welcome to Oliver and let's hear what he has to say about measuring radioactivity with using low-cost silicon sensors. Oliver, the stream is yours. Thanks. That was a very nice introduction. I'm really happy to have this chance to present here. I'm a member since quite some years, and this is my first CCC talk. So I'm quite excited. Yeah, you can follow me on Twitter, or I'm also a master on not so active, and most of my stuff is on GitHub. Okay, so what will we talk about in this talk? I'll give you a short overview also about radioactivity because yeah, it's a topic with many different details, and then we will look at the detector more in detail and how that works in terms of the physics behind it and electronics. And then finally, we'll look at things that can be measured, how the measurement actually works, what are interesting objects to check, and how this relates to silicon detectors being used at CERN. So the project is on GitHub called DIY Particle Detector. It's an electronic design, which is open hardware. There's a wiki with lots of further details for building and for troubleshooting. There's a little web browser tool, I will show it later briefly, and there are scripts to record and nicely plot the measurements. Those scripts are BSD licensed and this is written in Python. There are two variants of this detector. One is called electron detector. The other one alpha spectrometer. They use the same circuit board, but one is using four dites. The other one, one photodite. There's a small difference between them, but in general, it's pretty similar, but the electron detector is much easier to build and much easier to get started using. Then you have complete part lists and even a complete kit can be bought on kidspace.org, which is an open hardware community repository. I really recommend you to check it out. It's a great community platform and everyone can register their own GitHub project quite easily. Now, this is a particle detector in a tin box. You can use the famous Altoids tin box or something for Swiss chocolate, for example. You can see it's rather small board about the size of a nine volt battery. Then you need, in addition, about 20 resistors, capacitors and these silicon diodes plus an operational amplifier, which is this little chip here, this little black chip here on the right side. You can see it's all old school, large components. This is on purpose, so it's easy to solder for complete electronic beginners. This picture is already one user of this project to post it their own build on Twitter. Okay, so natural radioactivity. I would say it's a story of many misconceptions. Let's imagine we are this little stick figure here on the ground. Below us we have uranium and thorium. We also have potassium-40 in the ground. Potassium-40 is pretty specific and peculiar. It actually makes all of us a little bit radioactive. Every human has about 4,000 to 5,000 radioactive decays every second because of the natural potassium. Natural potassium comes with a radioactive isotope, which is just everywhere. It's in bananas, but it's also in us because we need it for our body chemistry. It's really important. Even some of those decays are even producing antimatter. So how cool is that? Okay, so what would we be measuring on the ground? Well, there could be some gamma rays or electrons. Those are from beta decays. Or from the uranium, there is one radionuclide appearing in the decay chain, which is called radon. Radon is actually a gas. From the ground, the radon can diffuse upwards and travel with air and spread around. So it's a bit like a vehicle for radioactivity from the ground to spread to other places. And that radon would decay with alpha particles producing electrons in beta decays and also gamma radiation further down in the decay chain. So just to recapitulate, I've said it already twice. So alpha particles are actually helium nuclei. So it's just two protons and two neutrons. And the electrons are missing. And in a beta decay, basically one neutron is transformed into a proton and an electron. And there's also an electron anti-neutrino generated. But this is super hard to measure. So we're not measuring those. Mostly we'll be measuring electrons from beta decays. That's why you see all these little E's indicating beta decays. Okay, if we would go to the hospital here on the left side, we would probably find some X-rays from checking our bones or something like this. Or even gamma rays or alpha particles being used in treatments or very modern even proton beams are sometimes generated for medical applications. Now here on the right side, if you go close to a nuclear power plant, we probably measure nothing unless there's a problem. In this case, most likely we would find some gamma radiation, but only if there's a problem. Okay, and then actually this is not the whole story. This is terrestrial radiation. But we also have radiation coming from upwards, showering down on us every minute. And that's actually nothing we can do against it. So protons are accelerated from in the universe. Basically the biggest particle accelerator nature has. And once they hit our atmosphere, they break apart into a less energetic particles and it's many of them. So in the first stage there's lots of pions generated and also nutrients. But nutrients are really hard to measure. So I'll ignore them for most of the talk. Then those pions can decay into gamma rays and then trigger a whole chain of positron electron decays, which again create gamma rays and so forth. And this goes actually the whole way down to the earth. We will have a little bit of that on the sea level. And the other more known part of atmospheric radiation is actually muons. So some pions decay into muons, which is kind of a heavy electron. And also in neutrinos, but neutrinos are again very hard to measure. So I'll ignore them for most of this talk. And if you look here on the right side on this altitude scale, you'll see an airplane would be basically traveling where most of the atmospheric radiation is produced. And this is why if you go on such an airplane, you have actually several times more radiation in there than here on earth. And of course on the ground it also depends where you are. There are different amounts of uranium and thorium in the ground. And this is just naturally there. But it depends on the geology, of course. Okay, so I've talked quite a bit about radiation. And I'm saying I want to use silicon to detect it. So what radiation exactly maybe let's take a step back and think about what we know maybe from school. So we have this rainbow for visible light, right? This is in terms of wavelength. We have 800 to 400 nanometers spanning from the infrared red area to over green to blue and into the violet. And lower than those wavelengths or let's say bigger millimeter waves, meter waves and even kilometer that would be radio waves. Radio frequencies for our digital communication systems, Wi-Fi, mobile devices and so forth. But I want to look actually more towards the right because that's what we are measuring with these detectors. It's a shorter wavelength which actually means higher energy. So on the right side we would be having ultraviolet radiation, which is kind of at the border to what we can measure. And these 800 to 400 nanometers translate into 1.5 to 3 electron volts, which is a unit that particle physicists really prefer because it basically relates the energy of an electron after it has been accelerated by 1 volt and makes it much easier to work with nuclear or particle physics because everything, all the energy is always related to an electron. And this formula here is just a reminder that the wavelength can be always converted into energy and it's inversely proportional. So wavelength increases to the left and the energy to the right. And if you increase energy more from the visible range, so let's say thousands of electron volts, then we arrive here, millions, mega electron volts, even giga electron volts. And there is now a pretty important distinction between those two areas. And that is the right one is ionizing radiation and the left one is non-ionizing radiation. UV is a little bit in the middle of that, so some parts of the UV spectrum can be ionizing. It also depends a lot on the material that the radiation is interacting with. For these detectors I'm talking about today and alpha-beta gamma radiation, this is all ionizing. So some examples, lowest energy on the lower spectrum would be x-rays, then electrons, gammas from radioactive radionuclides that I already talked about in the previous slide, alpha particles. And then muons from the atmosphere would be more in the giga electron volts range and so forth. And for these higher energies, of course, you need something like the LHC to accelerate particles to really high energies. And then you can even access the Terra electron volts regime. Okay, silicon diodes. What kind of silicon diodes? I'm using, in this project, low-cost silicon pin diodes. One is called BPW34. It's manufactured from Wisch High or Osram. It costs about 50 cents. So that's what I mean with low cost. There's another one called BPX61 from Osram. It's quite a bit more expensive. This is the lower one here on the right. It has a metal case, which is the main reason why it's more expensive. But it's quite interesting because that one we can use for the alpha detector. If you look closely, there is a glass on top. But we can remove that. We have a sensitive area. So this chip is roughly 7 square millimeters large. And it has a thickness, a sensitive thickness of about a 50 micrometer, which is not a lot. So it's basically the half of the width of a human hair. And in total, it's a really small sensitive volume, but it's enough to measure something. And just as a reminder, how much of gammas or X-rays we would detect with this? Not a lot, because these high energetic photon radiation kind doesn't interact very well in any kind of matter. And because a sensitive area is so thin, it would basically permeate through it. And most of the times, not interact and doesn't make a signal. Okay, what's really important, since we don't want to measure light, we have to shield light away. We need to block all of the light. That means the easiest way to do that is to put it in a metal case. There it's electromagnetically shielded and completely protected from light as well. Electromagnetic radiation or radio waves can also influence these detectors because they are super sensitive. So it should be a complete Faraday cage, a complete metal structure around it. There's lots of hints and tips how to achieve that on the wiki on the on the GitHub of this project. Okay, let's think about one of those pin diets. Normally, there is one part in the silicon, which is n-doped, negatively doped, and there's another part, usually, which is positively doped. And then you arrive at a simple so-called PN junction, which is a regular semiconductor diet. Now pin diets add another layer, a so-called intrinsic layer here shown with the eye. And that actually is the main advantage why this kind of detector works quite well. It has a relatively large, sensitive sickness. So if you think about let's say a photon from an x-ray or a gamma decay or an electron hitting the sensor, so by the way, this is a cross section view from the side. But okay, that doesn't really matter. But let's say they come here from the top into the diet and we're looking at the side. Then we have actually ionization because this is ionizing radiation. So we get free charges in the form of electron hole pairs. So electrons would be here, the blue ball and the red circle would be the holes. And depending on the radiation kind, how this ionization takes place is quite different. But the result is if you get a signal, it means there was ionization. Now if just this would happen, we could not measure anything. Those charges would quickly recombine and on the outside of the diode there would be a little signal. But what we can do is we can apply actually a voltage from the outside. So here we just put a battery. So we have a positive voltage here, a couple of volts. And then what happens is that the electrons will be attracted by the positive voltage and the holes will travel to a negative potential. And we end up with a little net current or a small bunch of charges that can be measured across the diode as a tiny, tiny current peak. The sensitive volume is actually proportional to the voltage. So the more voltage we put, the bigger is our volume and the more we can actually measure with certain limits of course because the structure of the pin diode has a maximum thickness just according how it is manufactured. And these properties can be estimated with CV measurements. So here you see an example of a couple of diodes, a few of the same type, the two that I have mentioned. They are different versions. One has a transparent plastic case. One has a black plastic case. It doesn't really matter. You see basically in all the cases more or less the same curve. And as you increase the voltage, the capacitance goes down. This is great and basically shows us that those silicon chips are very similar, if not exactly the same chip. Those differences are easily explained by manufacturing variances. And then because this actually, if you think about it, it looks a bit like a parallel plate capacitor. And actually you can treat it as one. And if you know the capacitance and the size, the area, you can actually calculate the distance of these two plates or basically the width or the thickness of the diode. And then we arrive at about, yeah, 50 micrometer if you put something like 8 or 10 volts. Okay. Now we have a tiny charge current. Now we need to amplify it. So we have here a couple of diodes. I'm explaining now the electron detector because it's easier. We have four diodes at the input. Like, and this is the symbol for an operational amplifier. There are two of those in this circuit. The first stage is really the special one. So if you have a particle striking the diode, we get a little charge current hitting the amplifier. And then we have here this important feedback circuit. So the output is feedback into the input, which in this case makes a negative amplification. And the amplification is defined actually by this capacitance here. The resistor has a secondary role, the small capacitance. It is what makes the output voltage here larger, smaller the capacitance, the larger the output, and it's inverted. Then in the next amplifier step, we just increase the voltage again to a level that is useful for using it later. But all of the signal quality that has been achieved in the first stage will stay like that. So signal to noise is defined by the first stage. The second one is just to better adapt it to the input of the measurement device that's connected. So here this is a classical inverting amplifier where just these two resistors define the amplification factor. It's very simple. It's just a factor of 100 in this case. So if you think again about the charge pulse and this circuit here is sensitive, starting from about 1000 liberated charges in those diodes as a result from ionization, we get something like 320 microvolt at this first output. And this is a spike that quickly decreases. Basically these capacitors are charged and quickly discharged with this resistor. And this is what we see here. And then that is amplified again by a factor of 100. And then we arrive at something like at least 32 millivolts, which is conveniently a voltage that is compatible with most microphone or headset inputs of computers or mobile phones. So a regular headset here has these four connectors and the last ring actually connects the microphone. The other is ground and left right for the earbuds. Okay, how do we record those pulses? This is an example of 1000 pulses overlaid measured on an oscilloscope here. So it's a bit more accurate. You see the pulse is a bit better. This is kind of like the persistence mode of an oscilloscope. And the size of the pulse is proportional to energy that was absorbed. And the circuit is made in such a way that the width of the pulse is big enough such that regular sampling frequency of a sound card can actually catch it and measure it. Yeah, this is potassium salt. This is cut here. This is called low salt in the UK. There's also German variants. You can also just buy it in the pharmacy or in certain organic food stores as a replacement salt. On the right side is an example from the small Columbide stone, which has traces of uranium on it. And this is measured with the alpha spectrometer. You see those pulses are quite a bit bigger. Here we have 50 microseconds and here we have more like one millisecond of pulse width. Now there's a software on a browser. This is something I wrote using the web audio API and it works on most browsers. Best is Chrome. On iOS, of course, you have to use Safari. And that records, once you plug the detector, it records from the input at 48 or 44.1 kHz the pulses. Here's an example with the alpha spectrometer circuit. You get these nice large pulses. In case of the electron detector, the pulse is much shorter. And you see, you see the noise much more amplified. This red line is kind of the minimum level that the pulse needs to trigger. It's bigger than that's like the trigger level of an oscilloscope. You can set that with those buttons in the browser. You need to find a good value. Of course, if you change your input volume settings, for example, this will change. So you have to remember with which settings it works well. And this pulse, for example, is even oscillating here. So for an electron detector, it's basically nice to count particles for the alpha detector. It's really the case where the size of the pulse can be nicely evaluated. And we can actually do energy measurements. And these energy measurements can be also called spectrometry. So if you look closer at these many pulses that have been recorded, and we find that there is really like much more intensity, which means many more same pulses were detected. We can relate it to radium and radon. If we use a reference alpha source and I have done this, I have measured the whole circuit with the reference sources and provide the calibration on GitHub. And you can reuse the GitHub calibration if you use exactly the same sound settings that I have used for recording. And for example, these two very weak lines here are from two very distinctive polonium isotopes from the random decay chain. The top part here, which is really dark corresponds basically in the histogram view to this side on the left, which is electrons. Most of these electrons, they will actually enter the chip and leave it out without being completely absorbed by it. But alpha particles interact so strongly that they are completely absorbed within the 50 micrometers of sensitive volume of these diets. And okay, here's a bit difficult to see peaks, but far end of the high energy spectrum, you see two really clear peaks. And those can only stem from polonium, actually. I mean, we know it's uranium. And that can only be polonium, which is that isotope that produces the most energetic alpha particles in and which is natural. I said, if you use the same setting like me, you can use it. So the best is if you use actually the same sound card, because there, if you put it to 100% input sensitivity, you will have exactly the same result, like in my calibration case. And this sound card is pretty cheap, but also pretty good. It costs just $2 and has a pretty range and resolves quite well 16 bits. And I think, oh, you can do that with an Arduino as well. It's actually a bit hard to do a really well defined 16-bit measurement, even at 48 kilohertz. It's not so easy. And this keeps it cheap and kind of straightforward. And you can have just some Python scripts on the computer to read it out. And this is as a reminder in order to measure alpha particles, we have to remove the glass here on top of the diet. So I'm doing it just with cutting into the metal frame. And then the glass breaks away easily. That's not a problem. There's more on that on the Wiki. Now, we can kind of compare alpha and gamma spectrometry. And here's an example. This is an uranium glazed ceramics. The red part is uranium oxide that was used to create this nice red color in the 50s, 60s, 70s. And in the spectrum, we have two very distinctive peaks and nothing in the high energy regime, only this low energy range has a signal. And this corresponds actually to uranium 238 and 234 because they use actually purified uranium. So all of the high energy progeny or daughters of uranium, they're not present here because that was purified uranium. And this measurement doesn't even need vacuum. I put it just like this in a regular box. Of course, if you would have vacuum, you would improve these peaks by a lot. So this widening here to the left, basically that this peak is almost below the other one. That is due to the natural air at the regular air pressure, which already interacts a lot with the particles and absorbs a lot of energy before the particles hit the sensor. So in terms of pros and cons, I would say the small sensor is quite interesting here in alpha spectrometry because it's enough to have a small sensor. So it's cheap and you can measure very precisely on specific spots. And on the other hand, of course, the conditions of the object influence the measurement a lot. So for example, if there's some additional paint on top, the alpha particles might not make it through. But in most of these kind of samples, alpha radiation actually makes it through the top transparent paint layer. In terms of gamma spectrometry, you would usually have these huge and really expensive sensors. And then the advantage, of course, is that you can measure regardless of your object. You don't really need to prepare the object a lot. You might want some latch shielding around it. And that's again expensive. But okay, at least you can do it, you can improve the measurement like that. And it's basically costly because the sensor is quite expensive. While versus in this setup for 15 to 30 euro, you have everything you need. And here you're looking at several hundred to several thousands of euros. Okay, now I'm measuring, I have to be a bit quicker now, I noticed. So I talked already about the potassium salt. There's also fertilizer based on potassium baking powder. Uranium glass is quite nice. You can find that easily on flu markets. Often also old radium watches. Here's another example of a uranium glazed kitchen tile in this case was actually in the kitchen. So the chances are that you at home find actually some of those things in the cupboards of your parents or your grandparents. This is an example of toriated glass, which has this distinctive brownish color, which actually is from the radiation. And a nice little experiment that I can really recommend you to look up is called radioactive balloon experiment. Here you charge the balloon electrostatically and then it will catch polonium from the air. And that's really great. You basically get a radioactive balloon after it was just left for 15 minutes in a normal regular room. Okay, now as a last kind of context of all of this to end this presentation, I want to quickly remind how important the silicon detectors are for places like CERN. This is a cross-section of the atlas detector. And here you have basically the area where the collisions happen in the atlas detector. So this is just a fraction of a meter. And you have today 50 to 100 head-on collisions of two protons happening every 25 nanoseconds. Not right now, but soon again machines will be started again next year. And you also can, by the way, build a similar project, which has a slightly different name. It's called build your own particle detector. This is atlas made out of Lego. And on this website, you find a nice plan how to build or ideas how to build it from Lego to better visualize the size and interact more with particle physics. In case of the CMS detector, this is the second biggest detector at CERN. Here you see nicely that in the middle at the core of the collision, you have many, many pixel and microstrip detectors which are made of silicon. And these are actually 16 square meters of silicon pixel detectors and 200 square meters of microstrip detectors also made of silicon. So without basically that silicon technology, moran detectors wouldn't work because this fine segmentation is really required to distinguish all of these newly created particles as a result of the collision. So to summarize the website, some GitHub, there's really this big wiki which you should have a look at. And there's a gallery of pictures from users. There's some simulation software that I used as well. I didn't develop it, but I wrote how to use it because the spectrum can sometimes be difficult to interpret. And there's a new discussions forum that I would really appreciate if some of you had some discussions there on GitHub. And most of the things I showed today are actually written in detail in a scientific article, which is open access, of course. And I want to highlight two related citizen science projects. On the one hand is a safe cast, which is about a large, nice, sensitive Geiger-Müller-based detector that has a GPS and people upload their measurements there. This is quite nice. And also OpenGeiger is another website, mostly German content, but also some of it is English that also uses diode detectors, showed many nice places. He calls it Geiger caching, so places around the world where you can measure something, some old minds, things like this. And if you want updates, I would propose to follow me on Twitter. I'm right now writing up two other articles with more ideas for measurements and some of the things you have seen today. Thanks a lot. Well, thanks a lot, Oliver. I hope everyone can hear me now again. Yes, thanks for mentioning the citizen science projects as well. It's really cool, I think. We do have a few minutes for the Q&A and also a lot of questions coming up in our instance at the IRC. So the first question was, can you talk a bit more about the SNR of the system? Did you pick particular resistor values and OPMs to optimize for noise? Was it a problem? Yeah, so noise is a big issue here. Basically, the amplifier is one I found that there's around four euros. I'm trying to find the slide. You have to look it up on GitHub, the amplifier type. But this is the most important one, and then actually the resistors are here. The resistors in the first stage, sorry, the capacitors is the second important thing. They should be really small since I'm limited here with hand solderable capacitors. Basically, I choose the one that just still available, let's say, and this is basically, what is available is basically a 10 picofarad capacitor. You put two of them one after another, you have the capacitance, so you get five. And this, by the way, is also a 10 picofarad capacitor. So I kind of try to keep the same resistor values as much as possible. And here at the output, for example, this is to adjust the output signal for a microphone input. And the alpha spectrometer, I changed the values quite a bit to make a large pulse. But yeah, it's basically playing with the time constants of this network and this network. All right, I hope that answers the question for the person. Yeah, but people can get in contact with you right after the shock, maybe, as well. So there's another question. Have you considered using an I2S codec with a Raspberry Pi radiation marks radiation HAT should be almost now set up and completely repeatable. So last one's more of a comment. I don't know that component. But yeah, as I said, using a sound card is actually quite straightforward. But of course, there's many ways to get fancy. And this is really, this should actually attract teachers and high school students as well as projects. So this is one of the main reasons why certain technologies have been chosen rather simple than let's say fancy. Yeah, so it should work with a lot of people, I guess. And one another question was how consistent are the sound cards? Did you find the same calibration worked the same with several of them? So yeah, so if you want to use my calibration, you should really buy this $2 card from eBay CM 108. I haven't seen a big difference from card to card in this one. But of course, like from one computer to the mobile phone is a huge difference in input sensitivity and noise. And it's very difficult to reuse the calibration in this case. But you still can count particles. And the electron detector is is anyway, mostly it actually just makes sense for counting because the electrons are not completely absorbed. So you get an energy information, but it's not the complete energy of the electron. So yeah, you could use it for x rays, but then you need an x ray machine. So yeah, two doesn't need an x ray machine. So maybe one question I have because I'm not very familiar with the tech stuff, but what actually can be done with it right in the field. So you mentioned some working with teachers with with these detectors. What have you done with that in the wild, so to say? So what's quite nice is you can characterize stones with it, for example. So since you can connect it to a smartphone is completely mobile. And it goes quite well in combination with a Geiger counter in this case. So with a Geiger counter, you just look around where where's some hotspots. And then you can go closer with the alpha spectrometer and actually be sure that there is some traces of thorium or rhenium on the stone, for example. Or in this type of ceramic is old ceramics. You can go to the flu market and just look for these very bright red ceramics and measure them on the spot and then decide which one to buy. Okay, so that's what I'm going to do with right. But thanks for highlighting a bit the practical. So I think it's really cool to educate people about some scientific things as well. Another question from the IRC, didn't you have problems with common mode rejection while connecting your device to the sound card? Yes, have you tried to do a AD conversion digitization on the board itself already? Transfer wire SP diff? Yeah, so of course, I mean, this is the thing to do if you want to make like a super stable rock solid measurement device. But it is really expensive. I mean, that's we're looking here at 15 euros. And yeah, that's the reason to have this separate sound card just to enable with very low resources to do this. But I'm looking for these pulses here. So this common mode rejection is a problem. And also this kind of over-swinger, I'm missing the English term now. These kind of oscillations here. If you design a specific analog to a digital conversion, of course, you would take all of that into account and it wouldn't happen. But here this happens because the circuit can never be exactly optimal for a certain sound card input. It will always be some mismatch of impedances. All right. So maybe these special technical issues and details. This could be something you could discuss with Oliver on Twitter or baby Oliver, you want to join the IRC room for your talk as well. People were very engaged wearing your talk. So this is always a good sign in that sense. I'd say thank you for being part of this first remote case experience. Thanks again for your talk and for taking the time. And yeah, best for you and enjoy the rest of the conference, I'd say, of the Congress. And a warm round of virtual applause and big thank you to you Oliver. Thanks. I'll try in the chat room right now.