How to make a Tritium Battery

tritium battery

Warning: Tritium battery is radioactive but safe to handle as long as it is contained within its vials. Work carefully to avoid breaking the vials. If they do break, leave the area and ventilate it for a few hours to disperse the gas. Greetings fellow nerds. In this video we’re going to make a very simple nuclear battery. But as usual I need to crush your expectations. Due to the limited amount of radioactivity that can be safely handled in a home lab, our nuclear devices must necessarily be weaker than the professional devices. Do not expect free energy or vast amounts of power.

Our Objective Regard Tritium battery:

Our objective here is to explore the science. As you may know nuclear batteries, or radioisotope thermo-electric generators, are famously used on space probes due to their high power density, reliability and longevity. They are also occasionally used in remote locations like unmanned lighthouses in the former Soviet Union. The military also uses such batteries for remote radar stations. Going to change the batteries on a regular basis was extremely costly so a nuclear battery was thought to be a better option. Unfortunately these types of nuclear batteries are strictly regulated and totally impossible for the average person to obtain. Now there exists smaller nuclear batteries you can buy commercially.

This one from City Labs is well known and can be purchased without any special nuclear device permits. It works by converting the beta radiation from radioactive tritium into electricity using what is essentially a solar cell. But instead of photons it absorbs beta particles. It is thus called a betavoltaic cell rather than a photovoltaic cell. My inquiries on the exact pricing were ignored but you can find them online from resellers for three thousand dollars. In terms of power their website lists their top unit has a voltage of two.four volts with a short circuit current of three fifty Nano Amps. This is quite small but it’s meant more for lower power devices like the memory or clock circuits in highly critical machines where chemical batteries are not as reliable. Other uses include medical implants like pacemakers, or remote sensors where changing the battery is not physically allowed or even possible.

In hardware security modules changing the battery isn’t allowed for security reasons. So the battery must last on its own for the lifetime of the device. Let’s try and make our own nuclear battery and see how it compares. It’ll likely be more expensive and less powerful than the device from City Labs but we’re here to explore the science. Now the City Labs’ device is categorized as a direct conversion radio isotopic battery in that the radiation directly interacts with betavoltaic cell to produce power. There isn’t a commercial source of betavoltaic cells so we’re going to have to use a photovoltaic cell. But photovoltaic cells only convert light into electricity, not beta radiation. We need to convert the beta radiation into light first. This technique is already well known and is categorized as indirect conversion. The simplest way of doing this is to put a phosphor in front of the radiation source which converts it to light. Fortunately for us, there is an easy to get commercial supply of safe radiation sources that have light conversion already built in. It’s these tritium key ring lights.

You can also find them as fishing lures, gun sights, emergency signs, military lights and so on. They work by having a sealed glass vial of tritium gas and an internal phosphor coating. The radioactive tritium gas emits beta radiation that strikes the phosphor and emits light. So the hard part of generating light from a radioisotope has been solved for us. This lets us avoid the safety issue of handling a radioactive gas. Now I bought mine online and they come encased in plastic. The actual tritium battery vial is much smaller. I could have bought the vials directly but the seller I got these lights from must have had a pricing error because they were selling for less than the price of the bare vials. So I took advantage and I bought fourteen before they pulled their listing, presumably they realized their mistake. Score one for me. But I still have to remove the plastic casing to get at the vials. Now the casings dissolve in dichloromethane solvent but only sparingly.

Use Soxhlet Extractor:

Soxhlet extractor

To help along the process I use this Soxhlet extractor I showed in a previous video. Briefly, the extractor constantly exposes the tritium battery lights to freshly distil dichloromethane and then drains it away where it is reboiled and distilled again. Thus I can use very little dichloromethane but thoroughly dissolve the plastic from the lights. Another reason why I used the extractor rather than just stirring with lots of dichloromethane is that the extractor is very gentle. There isn’t a fast spinning stir bar to smash the delicate glass vials of tritium battery once they are exposed. While the radioactivity is small it’s not a good idea to take any chances. Now i can do tritium vial decapsulation cheaply because i already have a soxhlet extractor. If you’re making your own nuclear battery then just buy the vials directly. Unless you take advantage of pricing errors like I did the naked vials are usually cheaper than the encased lights. Anyway, there we go, tritium vials. Before you get too excited, they’re not actually as bright as my camera suggests, I’m just maxing out the exposure settings. Anyway, I have more to go so I’m going put a fresh batch in and keep going. Now I noticed my vials had some insoluble rubber at the ends.

Presumably it’s silicone rubber and likely used to protect the vials from damage by cushioning from hitting the ends. If you want you can remove them by placing the vials in concentrated sulphuric acid. Silicone rubber is easily destroyed in sulphuric acid and after an hour they can be removed and washed with water. Nonetheless this is strictly optional and not actually necessary. Now we have cleaned tritium vials. And here we have it, fourteen vials of tritium battery. Turn off the lights and make sure they still work and haven’t leaked. Now to generate our power we put them on a solar cell. I’m going to use this amorphous silicon type solar panel you might find in calculators and other devices meant for indoors. Don’t worry i’ll test the more efficient Monocrystalline solar panels later. Now I intend to build a box but for now a quick and easy way of mounting the tritium is to first align them so they match the shape of the panel. And then taking some tape and picking them up.

Make sure they match. Now just place the panel on them and tape over. And now we have our simple nuclear battery. This is also called a radioisotope photovoltaic generator. It’s not a thermoelectric generator because it doesn’t use heat but just light. And we’re using a photovoltaic cell and not a thermoelectric one, so it’s photovoltaic. Now I’m going to put a second panel on top in the final version but for now this is good enough for performance testing. To ensure we get accurate results we encase the unit in aluminium foil to block outside light. Make sure the lead wires stick out. Now outside light can only help in terms of power generation. But we want to focus on just the nuclear energy part for our analysis. So let’s get some performance numbers. We can measure the voltage and the panel seems be generating around one.nine volts. So we’re off to a good start. Let me switch over to measure current. Let’s see what we get. And it looks like we have extremely low current. Barely one micro amp. Now this reading is likely completely inaccurate. For all we know our device is generating half that current and my meter is rounding up.

My meter could even have measurement errors and the true value could even be lower. Before you get too disappointed, keep in mind the professional device built by City Labs only produces three Nano amps, or about zero.thirty five micro amps, and it’s about half the size of our device. We shouldn’t expect to be able to beat actual nuclear scientists and engineers. Now I still want accurate performance numbers to compare to the professional device. We’re going to need a way of measuring Nano amps of current at various voltages. My multimeter can’t go that low. I could buy a better ammeter but I’ll save some cash and build a circuit. And this is my circuit. Here is my circuit diagram showing how it works. It’s more complicated than a simple ammeter circuit because this circuit will also let us examine other performance characteristics like the current versus voltage curve and find the maximum power point. First, this potentiometer here, these resistors over here and this battery form a simple voltage source and divider that will allow us to impose a voltage on the nuclear battery.

We’ll be able to control the magnitude and even the direction of that voltage with this potentiometer. The voltage will be measured by this voltmeter I’ve inserted here. Now to measure the current we’ll be using this voltmeter and this one megohm shunt resistor. By using such a large resistor the voltage difference across it caused by Nano amps of current reaches millivolt levels. A voltmeter can easily measure millivolt voltages and thus we can calculate the current by simply dividing the observed shunt voltage by the shunt resistance. Now a special point I need to make. The shunt resistance is not just this one megohm resistor. There is actually a ten megohm equivalent resistance across the voltmeter. Ideally a voltmeter will have infinite resistance but practically a real voltmeter has some very high but finite resistance. So the actual shunt resistance is actually a combination of the voltmeter internal resistance and the external shunt resistance.

Now we don’t know exactly what the internal resistance is but we don’t have to. We can simply measure the combined resistance directly by connecting the other voltmeter to it switching to ohmmeter mode and measuring it on the megohm scale. That value there is the combined shunt resistance of the voltmeter and the external shunt and we can use that in our calculations. It’s not often that the non-idealities of your multimeter will affect your circuits, but you need to be aware when they do. Now at this point you might ask, if the voltmeter has significant error at the Nano amp current levels. Wouldn’t my measurement with this ohmmeter also have errors in it as well? You’re absolutely right but in that case the makers of the multimeter know this and have calibrated the internal circuitry to compensate for that. So measurements, even in the megaohm range, are still accurate to within one% or two%. Okay now that we know the total shunt resistance and have recorded it in our lab notebook let me reassemble my circuit. Now if you look at my circuit again knowing how big the shunt resistance is you’ll notice something. I’ve put the voltage measuring voltmeter across both our current measuring shunt and our solar panel. If I want to measure just the solar panel, why am I putting it across both? The voltage difference across them will be the shunt and the solar panel combined. Since the shunt voltage will be on the millivolt range that would directly and significantly affect my results. So why am I doing that? Well you can put the voltmeter directly across the solar panel. But this actually makes the measurement harder. Remember when I said the voltmeter itself has a large but finite resistance? Well at Nano amp current levels this affects the results of the voltmeter and the ammeter circuit.

I can prove it here. I’ve connected the voltmeter directly to the solar cell. As you can see, when I connect or disconnect the voltmeter the ammeter shows a change. The current that is being drawn off by the voltmeter and its internal circuitry is significantly affecting results. While it is possible to measure the internal resistance of the voltmeter and mathematically correct for it, it is mathematically simpler to measure the shunt voltage and the solar cell voltage together and just subtract the shunt voltage which we can read off directly on the ammeter circuit. So I put the voltmeter on these points. In my case though since my current measuring voltmeter is connected more or less backwards. I just add the voltages rather than subtract but the mathematical principle is the same. I put it backwards so I can read current on the positive axis. Now before we start taking measurements be absolutely certain no extra light is getting in. Just shade the aluminium envelope containing the nuclear battery to check. If the readings change then there is some light leaking in and you need to encase it better. This was a problem in my earlier attempts before I finally wrapped it properly. Anyway, we can start taking measurements. It’s just a matter of reading off the voltages and rotating the potentiometer to read the next voltage.

We can sweep across the voltage range and get current readings each voltage. Granted you have to sit down and take a bunch of data points but it works. Now at this point, I’m pretty sure you actual electrical engineers are reacting to my circuit like this.

Bender Laugh:

Bender: Ahahahah! Oh wait, you’re serious.

Bender: Let me laugh even harder. AHAHAHAHHAH!

Yes I know, a completely automated circuit would have some sort of digital voltage regulator and a couple of ADCs and could sweep all the voltages in a few seconds and record them into a computer. I’m sure there are arduino or raspberry pi devices and shields that handle everything in one go. But then again that’s why I’m the chemist and you guys are the engineers. You may continue laughing.


Okay once we collect all our data we can now start processing it. We add together the shunt voltage and total voltage to find solar cell voltage. I must emphasize again that I can add them because I wired the currenting measuring voltmeter in the opposite direction of the total voltage measuring voltmeter. If it were the same direction you’d subtract the voltages. Now to find the current we divide the shunt voltage by the total shunt resistance that we found to be around zero.eighty nine megohms for the combined external and internal resistors.

Finally we find the power by multiplying the solar panel voltage by the current. Now we plot the current and the power versus voltage. And that is beautiful. There is nothing quite like plotting scattered data points and watching the underlying science emerge through the curves. One of the tiny little joys of research. So let me explain what we’re seeing. This blue line here is the current versus voltage. The current at zero volts represents the short circuit current is the maximum current the solar cell can produce. Now to the right where the current is zero this is the open circuit voltage. This is the voltage if the solar cell is completely unloaded. As you can see we can draw more and more current very easily until we reach a kink or knee where the solar cell can no longer generate much more current to meet demand. At that point the voltage decreases rapidly with only modest increases in current draw. This is the maximum power point and can be found easily by simply plotting the power output versus voltage. We can see it as this orange plot. The very top of the curve is the maximum power point and as the name suggests you get the most power from your solar cell if you can design your circuitry to load your solar cell to this voltage.

In this case around one.sixv. Now you might be wondering how I can have numbers into the negative region of my chart. My measuring circuit can apply voltages beyond the limits of the solar cell. There is nothing particularly special about those limits so the trends in solar cell behaviour continue. We’re just now draining power rather than generating it. At the negative voltage region we have a very sensitive light detector and this is the basis of some photodiode circuits. Beyond the voltage maximum is where light emitting diodes work and instead generate light rather than absorb it. A solar cell however won’t actually emit light since it’s not designed that way and will just heat up. Anyway on the whole I’m very satisfied with these results. We were able to measure the performance characteristics like short circuit currents and maximum power. We’re getting almost exactly microwatt of power. Let’s see if we can do better. Now amorphous solar cells are not very efficient, about several percent and that’s it. If you’ve looked into solar cells you know there are monocrystalline solar cells that you can buy at around twenty percent efficiency or higher. So let’s try that and see if we can do better. Now I know what you’re thinking, this monocrystalline solar panel I’m using is much smaller than my amorphous cell. It won’t be a fair comparison because it will generate less power than a solar cell of equivalent size.

You’re absolutely right but we can scale our results by measuring for area. I measured the monocrystalline panel to be four hundread fifty square millimetres in size and I measure the amorphous panel to be one thousand ninety two square millimetres in size. So the scaling factor is around two.four thousand two hundred sixy seven. I just need to multiply the current by that number and they’ll be comparable. Voltage is constant regardless of surface area so it doesn’t need to be scaled. So I’ve assembled the monocrystalline solar panel into our measuring circuit. Let me gather the data and start plotting. And here are our results. As said before I’m multiplying the current by the scale factor so we can compare them to our amorphous silicon panel. I had to put the power and current in separate charts since the numbers were so different. Now the monocrystalline cell produced a lot more current compared to the amorphous cell but the voltage max was much less. We can attribute some of this to the monocrystalline panel having fewer individual solar cells inside of it. But the most interesting difference is this fairly constant slope for the current trace and maximum power point is much lower. At best we’re only extracting nanowatts. And that’s already been scaled to match the size of the amorphous panel.

This monocrystalline panel is only giving us seven percent the performance. At first I thought I was doing something terribly wrong but I looked up the research by other groups and found this is a fairly well known observation. High efficiency monocrystalline solar cells are only efficient in very bright light. In very dim light like the weak glow of these tritium vials they actually perform very badly. Amorphous silicon solar cells while being very inefficient in bright light give much better performance in low light conditions. In thinking about this, I’ve actually seen this before but never really noticed it. Devices meant to work indoors with low light like this solar keyboard or calculators almost exclusively use amorphous solar cells. I’ve never seen one use monocrystalline solar cells. I even found this advertisement page extolling the virtues of using amorphous solar cells for low power. So apparently everyone knew that amorphous solar cells are better for indoor light except me.

Bender: AHAHAHAHAHAHA! Now to understand why this is we can look at this solar cell model circuit.

A solar cell behaves like a current source in parallel with a diode and shunt resistor. There is also an equivalent series resistance. But for this analysis we can ignore this. Now the shunt resistance isn’t an actual device you can find, it’s just a representation of the combined properties of the solar cell like crystal structure, defects, purity of silicon and so on. In a monocrystalline cell the current source is very powerful and produces a lot of current but the shunt resistance is rather low so at small power levels like low light it dominates the circuit and shorts out what meager power is available. In an amorphous solar cell the current generator is weak but the shunt resistance is huge so it has very little effect at low light. So for low light conditions the amorphous solar cell produces more usable power than the monocrystalline one.

For bright sunlight though the massively powerful current source dominates the circuit in a monocrystalline cell and thus they are used in bright sunlight where they give the most benefit. Anyway, I was going to try other solar panel types like polycrystalline but it seems the consensus among all the other researchers is that amorphous solar panels are the best for low light nuclear batteries. So I won’t bother. Okay so since we’re going to use amorphous solar cells exclusively let me rebuild it. You might be wondering if different colours or shapes would be better. I was only able to get discounted tritium vials for the colour green, so that was all I was able to test. The articles I read from other researchers mostly used green and some tested blue but I didn’t find any broad rigorous testing to determine the relative qualities of all the available colours. So that’s something you could try looking into if you want. As for shapes, I have seen rectangular tritium lights but those seemed to be more expensive for the same given area. Now we should be able to almost double our output doing the obvious, sandwiching a second solar panel on the device. And there we go, a double sided nuclear battery.

Once again I’m going to put the unit in foil to protect it from external light so we get accurate measurements. I’ve inserted the device back into my measuring circuit. Now I’ve connected the solar cells in parallel to get increased current. You can connect them in series for increased voltage if you want, the total power will still be the same. Running the voltage sweep again we then plot the data and see how we did. Interestingly enough rather than getting twice the maximum power we’re only getting about twentry three percent more. The maximum power point is one.twenty three microwatts rather than the microwatt of the single panel device. At first I thought I did something horribly wrong and rechecked my connections and work. Turns out this data is correct. The reason why the single panel nuclear battery attempt had such a good power rating was because the aluminium foil I used to block the light actually reflected the other side of the tritium vials and returned the light back to solar panel. Now it had to go through the tritium tubes again so it lost some power and thus didn’t give the same power as a double panel device. But it still gave more than half a double panel device, a good sixty percent more power. In my double panel device there is no reflection so we’re not getting that boost and thus not getting double power.

Still, a % increase is not negligible and worth keeping the extra panel since that’s still cheaper than % more tritium vials. Now that we have our relatively optimized device, let’s compare it to the professional one. Now betavoltaic technology is different from photovoltaic technology but it should have a maximum power point as well. Unfortunately City Labs don’t give us a current voltage graph like mine so we don’t know what that maximum power point is. But let’s give them the benefit of the doubt and assume they have perfect technology and their device gives perfect power. According to their specifications they can provide three hundred fifty Nano amps at two.four volts or zero.eightyfour microwatts. Our device at one.twenty three microwatts is actually better than the professional device. And we haven’t even factored in cost yet. While I got my tritium vials for a steep discount, if you bought the components and the tritium tubes at market price then cost comes out to about three hundred dollar Canadian. Meanwhile the reseller was selling the City Labs device for three thousand dollar Canadian. We could build ten of our devices.

That being said the City Labs device does beat mine in terms of size and weight. It is both smaller and lighter and that could be an important consideration in devices where that is a premium like in spacecraft or in medical devices like pacemakers. Nonetheless I’m totally blown away. I’m not even an electrical engineer let alone a nuclear engineer and I’ve got something that is both cheaper and more powerful than the professional version. Now why is their stuff more expensive? They got two things working against them. First, they need nuclear regulatory approval to build and sell their devices. I’d imagine getting the legal paperwork done and meeting nuclear safety standards is not exactly cheap. Second, at just zero.eighty four microwatts the number of applications isn’t that high and it’s unlikely they’re selling in massive volumes. So they need to charge more per device to stay in business. We kind of cheat in that tritium vials are mass produced in huge quantities for all sorts of uses so economies of scale and the free market have made them affordable and well within the reach of the average person.

We also don’t need nuclear regulatory approval. Anyway, so at this point you’re probably wondering, how long does this last? Well tritium has a half-life of. years. So ideally this should gradually halve its power output every twevel point three years. But there is another decay mode. The phosphor in the tubes themselves are constantly being bombarded with beta radiation and they will decay as well. So the total power output will decay faster than what the tritium half-life suggests. I don’t know what that rate is but it’s something to keep in mind if you want to use this device for the very long term. I also recommend using a better case than plastic tape. A custom metal box is best for the long term.

So what can we power this? As I said before, this is meant for extremely low power devices where longevity and reliability are absolutely paramount. For comparison, if we wanted half a watt, which is enough to run a low end cell phone, we’d need four one five zero of these, or about one hundred twently million Canadian dollars’ worth. If your expectations weren’t crushed already they should be now. Now I did want to use this for something but I’m not an electrical engineer so I can’t build the Nano watt level circuits that would take advantage of this battery. But I’m sure guys have better ideas and we should discuss them in the comments. So, that is how you make a very simple nuclear battery or radioisotope photovoltaic generator, for a fraction of the cost of the commercial units. Thanks for watching.

Special thank you to all of my supporters on patreon for making these science videos possible with their donations and their direction. If you are not currently a patron, but like to support the continued production of science videos like this one, then check out my patreon page here or in the video description. I really appreciate any and all support.

Be the first to comment

Leave a Reply

Your email address will not be published.