Automated Hydrogen Generator

Terrific Physics
40 followers
In this video, I'll show how I built my hydrogen generator which can make close to 20 liters per
hour of pure hydrogen. Using water electrolysis to make a mixture of hydrogen and oxygen,
or HHO, seems to be a popular project on the internet. It's easy to do,
and the balloons make great fireworks by detonating with a supersonic flame front.
But as fun as HHO is, I can't seem to find any practical use for it except maybe as a
cutting torch for really small-scale work or when you don't have access to an oxyacetylene flame.
Pure hydrogen, on the other hand, has innumerable uses. In my case,
I want to attempt to liquefy it in a Joule-Thompson cycle by using liquid nitrogen for pre-cooling.
Another use in cryogenics would be as a refrigerant for an oscillating flow cycle
like a Stirling or Pulse tube cryo-cooler. These typically use helium as a working fluid,
but it's a non-renewable resource and very difficult to get in some countries.
It can also replace helium in balloons or blimps, which has never gone wrong. Except for that one
incident, it's pretty great. In fact, it's even more efficient than helium as a lifting gas.
And in chemistry, there's a seemingly endless list of reactions hydrogen takes place in.
I originally tried doing this the dirty way, by dissolving aluminum and sodium hydroxide.
And it does work, but it's quite a bit more costly per unit volume of hydrogen,
and after getting blasted in the face with sodium hydroxide solution when my reactor
tank exploded, I decided that I'd look into a different method.
Industrially, the most common way of producing hydrogen is with superheated steam and methane
flowing over a platinum catalyst, which then breaks up into hydrogen and CO2.
Considering the price of platinum, I don't see myself doing that anytime soon.
That just leaves water electrolysis as the only reasonable approach.
Typically, the way this is done is with a stack of plates sandwiched together
with a whole bunch of bolts to hold them tightly in place,
and some sort of membrane material to keep the two gases from mixing.
If you read research papers on the subject, they're usually using some ridiculously expensive
fancy polymer membrane, but you really only need something like ripstop nylon,
which is porous enough to allow conduction,
but the holes are small enough that the little gas bubbles can't leak across it.
Problem is, every time I tried to do the typical sandwich method, the gas gets leaked,
and attempts I made with 3D printed cells also leaked, and let me tell you,
a solution with a pH of 13 is not a pleasant thing to have leaking all over the place.
Then I got another idea which should theoretically be leak-proof
and trap all the hydrogen that's produced until it's ready to be moved elsewhere.
After watching my dog drink from his upside-down water bowl thing,
I realized that would be a really useful geometry for a hydrogen generator cell.
The cathode would be placed inside of a glass bell,
and the anode would be placed on the outside so the electrodes would be concentric,
then the whole thing would be in a larger glass or plastic bowl full of the electrolyte.
The wall of the glass bell would prevent the gas bubbles coming off the electrodes from mixing,
since they're always traveling upwards due to the buoyancy.
Then a tube would be placed inside the bell that goes all the way up to the top,
which I'd suck all the air out of, allowing the bell to be completely
flooded while the outside water is still just open to the atmosphere without any seals.
Of course, this tube would need a check valve on it to ensure gases could only flow
out. Otherwise, air would backflow into the bell and push the water line inside of it back down.
As the hydrogen accumulated, the water line inside the bell would be pushed down,
which would allow me to see the exact volume of gas that had been produced.
Of course, as the water line of the bell moves down from accumulated hydrogen,
the water line of the bowl would move up.
Since the water is loaded with sodium hydroxide, it's very conductive,
so I could simply use a pair of smaller electrodes to detect when the water level
reaches a certain height and trigger a small pump to pull the gas out of the bell via the tube.
I think it's a solid idea in theory, but let's see if it actually works in reality.
To mount the cathode, I made this ring thing, which will support the glass bell in the middle
and has arms on the outside to mount the anode. Then the whole thing will live inside this
Tupperware bowl. I kept the lid, but it's not there to seal anything off. It's just there to
act as a baffle to keep the nasty salt spray from jumping out of the water, which tends to happen
with an open container undergoing electrolysis. The electrodes are 0.15 inch thick strips of 316
stainless steel. This material won't last forever like platinum, but it should last several hundred
hours. At the time of making this video, I've put close to 100 hours of runtime on my cell,
with only slight discoloration of the electrodes. The cathode sheet wraps around the fixture and
has a narrower, thicker strip of stainless steel to connect it to the outside world.
The strip has some heat shrink on it to prevent unwanted electrolysis from happening on its
surface. Here you can see how it wraps around the bottom of the glass. Then the rings for the
screws are also 316 stainless. In retrospect, I could have just welded these pieces of plastic
together with a soldering iron. Then I screw on the anode sheet, then install this little bracket
thing in the middle, which will serve as the base for mounting the little vacuum tube for extracting
gas from the cell. I used silicon caulk instead of super glue to attach everything because it
seems to hold up well in a high concentration of sodium hydroxide. Let's try pulling vacuum on the
bell. The sudden bubbling is from the water line on the outside dropping below the bottom of the
bell and letting a bunch of air in, so the cell needs more water in it. So here's the bell with
all the air removed from it. I didn't get very far into testing before all of my 3D printed parts
started to crumble, and it was at that point I learned that PLA will dissolve in sodium hydroxide.
So all the parts had to be reprinted with something else. A friend of mine reprinted
all the parts in ABS, which I didn't have at the time, and the cell was rebuilt after a couple of
days. Here's the leads for the new cell. As you can probably tell, I'm expecting to pull a lot
of current, which leads to an important question. How do you predict how much current you're going
to pull with electrolysis? After a little bit of googling, I found this nifty chart that gave
conductivity for a bunch of different solutions based on electrolyte concentration. Let's look
at sodium hydroxide. The chart says that at five percent concentration, we'll get a conductivity
of 206 millisieverts, or 0.206 sieverts per centimeter. A sievert is the inverse of an ohm,
so for example, 1 volt at 5 sieverts would give 5 amps, 1 volt at 10 sieverts would give 10 amps,
2 volts at 10 sieverts would give 20 amps, and so on and so forth. The per centimeter part is area
over distance. In other words, two 1 centimeter squared plates that are 1 centimeter apart in a
5% sodium hydroxide solution would give 0.206 sieverts, so in theory that's a little over 1
amp at 5 volts. However, when it comes to electrolysis, there's one more factor in that
equation which I'll talk about later, but first I did my own test to verify the numbers from the
chart. After all, you shouldn't just believe everything you read on the internet. To do this,
I made a small test fixture that had plates of a known area and distance and tested them with my
power supply and various concentrations of sodium hydroxide. The results I found were pretty
consistent with the chart, save for one very conspicuous difference. The solution didn't start
conducting for any concentration of electrolyte until about 2 volts, so the simple assumption of
I equals V over R isn't valid for electrolysis. Instead, the equation would be I equals V minus V
threshold over R. According to the Wikipedia article on water electrolysis, this threshold
voltage is theoretically 1.23 volts at room temperature, but in practice it seems to be
somewhere between 1.8 to 2.0 volts. Okay, so that's how to estimate current, but what about
gas production rate? A mole of water takes about 287 kilojoules of energy to separate at room
temperature. The hourly production rate of gas would be the threshold voltage times the current
times 3,600 seconds for an hour divided by 287,000. This gives moles per hour. At room temperature,
a mole of ideal gas occupies about 24 liters, so multiply that number by 24 and you get liters per
hour of gas output. This is the production rate of hydrogen, but the production rate of oxygen
would be exactly half of that. If you're building a cell, you may be tempted to just crank up the
voltage to increase the current, and this would certainly increase the gas output. Problem is,
it's also reducing the efficiency of the cell and dramatically increasing heating. If you had 12
volts and 10 amps going into a cell, 10 times threshold voltage, or about 20 watts, would be
going into electrolysis, and the remaining 100 watts would be going toward heat. Also, increasing
the temperature increases conductivity of the solution, which would cause even more current draw
and an eventual thermal runaway. This is why electrolysis is typically done with a constant
current power supply rather than a constant voltage supply. If it is done with a constant
voltage supply, designers typically try to keep the voltage as low as possible above the threshold
to maximize efficiency and minimize heat. In industrial devices, usually each cell is somewhere
between 2.2 to 2.5 volts. Based on my CAD design, I expected to have nearly 100 centimeters of
conduction area, so at 5% sodium hydroxide, that would result in 18.7 sieverts, or 53 milliohms.
However, with the glass bell blocking part of the path, that number would be a little bit lower.
Originally, I planned to use a 3.3 volt supply for the cell, but I wanted a little more output,
so I just got a 5 volt supply instead. I figured if the cell began to overheat,
I'd just dilute the electrolyte a little bit so it wasn't quite as conductive.
Based on these numbers, I should get 56 amps with a 5 volt supply. But again, the glass wall would
reduce conductivity somewhat, and voltage drop across the leads would reduce the current somewhat
too. I figured it would realistically be more like 20 to 30 amps. So I printed some extra widgets
and started to mount all my equipment on a wooden board. Here's the 5 volt power supply,
vacuum shutoff valve, vacuum pump, current meter, and a Gatorade bottle serving as a liquid trap in
case the pump pulled too much gas out of the bell and started sucking out the electrolyte.
I also got this 12 volt wall wart to run the pump and extra doodads which I repackaged into
this 3D printed box. Here's a look at the full system. I originally had a 1 milliohm shunt for
current measurement, but it was grossly inaccurate so I replaced it with this Hall Effect sensor
instead. I also added a little fridge compressor so the hydrogen could be stored in a tank under
pressure once it was removed from the cell. I did a couple of test runs but found that the hydrogen
was contaminated with oxygen when I did ignition tests. It's a little bit hard to tell from the
audio in this video, but those bubbles went off with a pretty sharp bang rather than the lazy
sort of whoosh sound you get if it was pure hydrogen. Turns out some of the gas bubbles from
the anode were migrating across the gap because part of the anode sat below the bottom of the
glass bell. The anode needed to be moved higher so that none of it hung below the bottom of the glass.
To do this, I just added some extensions onto the printed arms that hold the anode. Rather than
attempting to use glue that might react with the sodium hydroxide, I just used my soldering iron
to weld the ABS parts together, which actually worked pretty well. Fortunately, I had plenty
of headroom in the Tupperware bowl and my anode still fit inside with the lid on. This time, when
I did an ignition test on the gas, there was no bang whatsoever, just a lazy flame, so it looks
like the hydrogen is pretty pure now. Starting at room temperature, the current sensor reads 25 amps,
so right about what I expected. The output from the power supply is 5.21 volts, and the voltage
across the cell is 5 volts even, so there's a 0.21 volt drop across the leads, which is a little over
5 watts at 25 amps. That's not too surprising, since the leads get noticeably warm. Here's a
closer look at the salt spray I mentioned earlier. Even without any boiling or big bubbles or high
temperature, the microbubbles can launch themselves several inches out of the water when they hit the
surface, which is why the water has to be covered with some sort of baffle, and tubing extracting
the gas has to have a filter on it. The gas still seems to be pretty pure, so now it's time to add a
Arduino with a little breadboard shield I built a circuit on top of. The board connects to relays
that control the 5 volt power supply and compressor, and a MOSFET that controls the 12 volt
vacuum pump. A pair of stainless steel screws are mounted to the lid of the cell. When enough gas
accumulates in the glass bell, the water line inside the outer portion of the cell rises high
enough that it contacts the screws. This has the effect of closing a switch that causes a transistor
to pull down one of the digital pins on the Arduino. When the Arduino has read a low state
on that pin for one second, it kicks on the vacuum pump for three seconds, which transfers gas into a
beach ball or balloon. If the electrodes are triggered for more than 10 seconds, the program
assumes something has gone wrong and it goes into a so-called panic mode and shuts everything off.
The only way to clear the panic mode is to power cycle the entire device. There's also functionality
and wiring meant for a switch to detect when the beach ball or balloon is full and turn on the
compressor, but I never got around to actually setting up a physical fixture for that. Here's a
time lapse of the cell filling up a latex balloon. You may notice that over time the current is going
up. That's no error. As I mentioned earlier, higher temperature will result in higher conductivity of
the electrolyte. Here's a graph of the temperature versus time. After three hours of running, the
electrolyte reaches nearly 60 C. And here's a graph of amperage versus temperature. At room temperature,
we start off just a hair above 25 amps, but at 60 C, we're over 40 amps. Fortunately, I didn't get
any thermal runaway because the cell lost enough heat naturally and never really got above 60 C,
but it's not a great idea to run this cell at 60 C for a long time because the ABS parts are going
to be fairly soft at that temperature. Anyway, after one hour, my balloon started to float a
little bit, and after two hours, it had no problem holding up the fill adapter and tubing.
When I zero my scale with slack on the tether and then release the tether, I get minus 22 grams,
which is the lift it's producing. Alright, so far I've got a cheap and dangerous replacement
for helium to fill party balloons, but to fill something bigger, I'm going to need a way to store
a lot of hydrogen. I originally got a 10 gallon steel tank for this purpose, but then I read that
a phenomenon known as hydrogen embrittlement can cause the steel to crack and fail. This seems to
be the result of those pesky little hydrogen atoms being small enough to tunnel their way
through the lattice of iron atoms. I read that aluminum is far less affected by this issue,
but my aluminum tank was currently being used to hold the ethylene for my cryocooler project.
So I evacuated my steel tank, then set up a hose to equalize the pressure between the steel tank
and the aluminum tank. Once the tanks were equalized, I used my trusty compressor module
to force the remaining ethylene to the steel tank. Afterward, I removed any trace amounts of
leftover ethylene with a vacuum pump. Supposedly another thing that exacerbates the hydrogen
embrittlement problem is the presence of moisture. That's a problem because the gas coming out of the
cell is extremely humid because, you know, it's literally coming out of water. So I'm going to
add this bottle full of silica gel in line with my pump output to dry the hydrogen. Just to show
that it works, I have a little humidity meter inside this inverted Tupperware box. The humidity
starts off a little over 40% and when I blow into it, the container fogs up and goes close to 80%.
Now I'm going to blow through the drier bottle and we should see the humidity inside the container
drop. And there you go. I think the humidity actually went close to 0%, but the meter is
probably pegged at around 10%. So hopefully this will help mitigate the embrittlement problem.
So after spending a weekend running the cell, I had enough hydrogen to fill this 10-gallon
tank to 125 psi, which translates roughly to about 320 liters of hydrogen. The cell puts out
about 15 to 20 liters per hour, depending on the temperature of the electrolyte. To fill a much
larger balloon, I've made this adapter with a chunk of PVC pipe and some 3D printed bits. 320 liters
of hydrogen is enough to generate about 340 grams of lift, which is close to a pound. So let's strap
this GoPro to the balloon and try to catch the sunset with it. Not a bad view, but I definitely
need some stabilizer fins if I ever want to do aerial photography from a balloon. A drone might
be easier. My string was also way too thick, which limited my altitude. I think I made my point though.
When the flight was over, I removed the clip from the balloon nozzle and recovered the hydrogen
back into my tank for later use. So I think that's a pretty good demonstration of what's
possible with a little tabletop electrolysis cell. I think there's a lot of ways I could make this
device even more compact and energy efficient, but for now I'm pretty happy with it. I didn't
originally intend to do anything with balloons, but after flying one with hydrogen I made, I kind
of want to try launching one of those high altitude weather balloons or even have my own
RC Zeppelin like that guy who made an RC model of the USS Macon that actually floats. I used to
build and fly RC planes and drones, but something about lighter-than-air flight seems so much more
interesting, even if it's less practical. Anyway, there's one more thing I need to mention about
electrolysis. If you're using stainless steel electrodes, they'll last for a reasonably long
time, but not forever. Over time, they'll release various chromate salts into the water, which are
bad for your health. Some of these could even have hexavalent chromium, which is particularly nasty.
So if you have to get rid of your electrolyte after hundreds of hours of use, don't dump it
down the drain. Put it in a bucket, let the water evaporate, then collect the precipitate in a
plastic bag and drop it off to a toxic waste disposal site or stash it away in a government
warehouse. Anyway, thanks for watching.
