Homemade copper semiconductors
On the weekend I played around trying to make copper-oxide diodes. I succeeded in making something diode-like, but it behaved more like two opposite-facing diodes in parallel. In the rest of this post I'll detail how I made my 'un-odes' and their test results in the hope of encouraging other people to tinker with home-made semiconductors.
I've always wanted to make my own semiconductors. My dream is make N and P doped pastes that you can 'paint' or 'dab' onto bare PCBs to make diode circuits. In some ways this would truly be making my own chips, and it could be expanded to make more complex designs with printing or stencilling. Dreams :)
I have had little luck following many diode-making guides on the web. Often these are aimed toward making one-off diodes for crystal radios. Everything I have read and tried so far has involved hit-and-miss procedures where you hope to make or find a single tiny dot of semiconducting material on a large surface. I want something more reliable.
My motivation was resurrected after reading a page on making copper rectifiers by H.P Friedrichs. He details some of his own attempts at making them. Just like many other guides however he focuses on making a 'one off' that gets placed inside a complicated jig to keep contact on a single point of his creation.
Friedrichs uses borax and heat to create a layer of semiconducting cuprous oxide on the surface of a copper fitting. To clean off layers of unwanted cupric oxide he uses an acid. He uses a piece of (lead-tin) solder wire touching the cuprous oxide as the other half of his junction.
I didn't have have borax handy and being too lazy to walk down the street I decided to keep researching other methods of making cuprous oxide. A Russian video I found suggested I needed to keep a piece of copper accurately at 1000 degrees celsius for extended periods of time to build up a good layer. This discouraged me further.
In the end I decided to just have a play around with some copper, candles and various chemicals I had lying around. I tried dipping a small piece of copper in diluted hydrogen peroxide and boiling it off repeatedly. I tried slowly heating the copper and probing the resulting rainbow of oxides with pencils and solder. After a day's work I managed to make a single speck of semiconducting region on my copper that was quickly destroyed from me probing it with a pencil.
In the end I found a better way.
EDIT: Note that my final oxide is a black colour. I am uncertain which type of oxide (or combination of oxides) my final result produced.
To test my junctions I used an 'octopus circuit' hooked to to my oscilloscope. This let me see the 'I-V curves' of what I made.
'I-V curves' are just graphs of how much current flows through a part depending on the voltage applied to it. The IV curve of a resistor is just a straight line:
Here you can see that a positive voltage across a resistor makes a positive amount of current flow through it, and visa versa. The line follows ohm's law: V=IR
A diode has a non-linear IV curve:
(Almost) no current flows for negative voltages but lots of current can flow for positive ones. The point at which the curve suddenly sweeps up is often quoted as 0.7V for a normal silicon diode, however it can be much higher or lower depending on the current levels and particular diode you are using.
Also interesting is the curve of a zener diode:
The negative voltage that the curve suddenly sweeps down is referred to as the 'reverse breakdown' voltage. This is generally much larger than the 0.7-ish forward breakdown voltage and is sometimes used as a voltage reference in circuits.
An octopus circuit simply applies a voltage across the part you want to test and provides convenient points to measure voltage and current. I stole the circuit I'm using from a video made by w2aew
I made mine with an ordinary 1/4W 1K resistor, a breadboard and a wound AC power supply:
On the right you can see a large toroidal transformer I have had lying on my desk for a while. It outputs 12VAC on each of its secondary windings, which is a bit more than I wanted, so I wound several turns of black wire around it to get about 3VAC. This is an RMS measure so the actual voltage peaks were from about +4.3V to -4.3V.
Make sure you use an isolated power supply (eg a wound transformer)! In this circuit we arbitrarily hook up oscilloscope ground to the middle of the circuit. Please avoid setting fire to anything or harming anyone. Don't presume something is safe just because you read about it on the internet.
In the picture above I have a small piece of copper being held in vice-grips. One alligator leads connects to the vice-grips and another connects to the end of a pencil. You will want to choose a value of resistor that makes it safe to short these two leads as you will be constantly doing this when hunting for semiconductor junctions.
Make sure the grounded lead is hooked up to the vice-grips (and not the signal lead), otherwise touching the vice-grips with your hand will introduce noise into your signals. You will see a blurry graph on your oscilloscope when this happens.
Not shown in this picture is an analog oscilloscope in X-Y mode. I'm using a Goldstar OS-7040A that my university discarded. She's an old gal with some some quirks I need to fix, namely:
- her vertical amplification sometimes dies, temporarily revivable with percussive maintenance
- her lower metal housing ('skirt') keeps falling off during use
- her zero point drifts as she heats up
I thought I fixed the vert-amp problem a couple of years ago by cleaning some contacts, but it's resurfaced. The temp drift problem is something I had not noticed before -- during this experiment the centre of my X-Y graph would slowly move diagonally.
Producing my junctions
- Candle (& method of lighting)
- Small piece of flat copper
- Small pair of vice-grips
- Glass of water
- Kitchen scourer
- Misc. parts to make a stand for the vice-grips
Improvise everywhere: no one will have the same parts as me. Lots of things are made of copper. You don't need vice-grips if you have some wire.
(1) Clean your piece of copper using the scourer. I clean mine on top of a piece of copper PCB rather than directly on my desk so I don't scrape wood contaminants onto my copper.
(2) Cut a small handle into your piece of copper using your scissors. Make it small and thin so as little heat is lost through it as possible (vice-grips work very well as a heat-sink). Avoid touching the cleaned area of the copper with your fingers.
(3) Light your candle and let it burn for a few minutes so that the flame size stabilises. I had to close my window to get rid of a gentle breeze.
(4) Setup a way of heating the piece of copper above the candle flame. I did this:
A couple of rolls of tape were acting as a stand for the vice-grips. Atop the rolls of tape was a metal bowl to aid in some heat-sinking, as the vice-grips were getting decently warm in my experiments.
Make sure that the candle flame is not licking your piece of copper or it will start depositing carbon (and smoking).
(5) Keep the copper hot for a few minutes. A nice black layer of corrosion will appear on top:
(6) Drip a few drops of water onto the copper one-by-one, with a good time-gap (eg 15 seconds or more) inbetween for the copper to recover its temperature. If the water 'bounces' or 'runs' off the surface then try again. A successful drop will make some of the darkened surface corrosion 'rupture' or 'move', temporarily exposing more orange-looking flecks of copper.
(7) Keep the copper heated for a few more minutes. Eventually all the copper on the top surface will again look black, however you may notice some circular patterns from where you placed your drops of water.
(8) Blow out your candle and wait for everything to cool.
(9) Observe what you have made and compare it against what I found. Get angry if it's not the same and try changing a few of the variables I have been vague about.
You can see that the water drops created some circular patterns in the copper oxides. The surface of the un-treated copper is 'dotted' or 'noisy' whilst the drop-sites are made of concentric rings. I only cleaned the left and centre of my sample before starting: ignore the mess you see on the right half.
(10) Gently brush away the upper surface of copper oxide using a tooth-brush. Avoid brushing away the dark, circular regions which lie under the surface layer as much as possible.
This is what the sample from above looked like after I brushed it:
These round, black regions are what we are after.
Testing and Results
I had succes with both pencils and lead-free solder as a contact against the round oxide regions. Although the photos show me using a rather large 5B pencil, similiar success was obtained with an ordinary pencil labelled as 'dark' (I didn't have any HB at hand to try). Pressing a blunt pencil anywhere on the circular regions yielded curves that looked like these:
(vertical axis: 0.5V per division, horizontal axis: 1V per division)
I had to press very gently to get this result. When I pressed further the centre point on my oscilloscope (0V and 0A) screen would 'twist' away from being horizontal, suggesting I was putting a resistor in parallel with my junction. Pressing even further would cause the line to suddenly appear vertical (ie a short) until I released some pressure.
As mentioned in my opening paragraph: my junctions act like two diodes facing opposite directions in parallel ("antiparallel"). The reverse breakdown voltage is not that different to the forward breakdown voltage. This limits the usefulness of the diodes made by my exact method here to a few niche applications like distortion circuits and biasing. I really want to find a way of fixing this so that my diodes can be more practical.
I am not yet sure why there is an 'open-looped' region in my my I-V curves. My first thoughts were that I had created a memrister, however the scope probes I am using are not properly compensated and it might just be the effects of some capacitance. Although these tests were run at 50Hz (minimising the effects of the two afore-mentioned problems) I still want to test on another scope before I jump to any conclusions.
EDIT: Hmm, if it were capacitance then it should have affected the whole line, not just one half. Possibly some capacitance that gets shorted in the other half of the waveform?
The photos used in this article are from the second attempt at my method. I had hastily recorded the results of my first attempt on video just in case I was never able to repeat it again, but it seems this 'water-drop' method is somewhat reliable. I'll only properly know when I get some more free time to try again.
I was wondering if it would be possible to make this type of diode directly on a copper-clad PCB. It would be interesting to try, however I would want to do it in a temperature-controlled fashion (and outdoors) to avoid my base material outgassing nastily. You could then deposit a mixture of mostly graphite powder (and perhaps a little glue or clay) to bridge between a copper track and a copper-oxide region. More things to experiment with!
I'm open to comments: add one anonymously below or shoot me an email (and say whether or not you want your comment published). Make sure to say something if you try my method, especially if it does not work for you!