This question is for more than a theory.

I would like to disprove that what I'm working on is not a Group II, Group IV transistor.

I know they CAN exist. I don't know why they don't exist or what their properties would be.

So, q1. What would their properties be as an npn and as a pnp?

q2 How could a circuit be formed to prove or disprove that the device is this type of transistor?

It definitely has Group II and Group IV elements with silicon and heavily doped elements and I think most likely in the PNP configuration. But it doesn't act like a normal Group III-V transistor.

Here's what I have so far. I have explored this in EE.SE as a battery cell, but with such random results that I wanted to pursue down this rabbit hole at the same time.

I began experimenting with a technology from the 1890's which generated power for the early telegraph system (see https://en.wikipedia.org/wiki/Earth_battery) thinking that with the availability of newer materials such as graphite and with todays tech I might be able to find a way to make this more viable. I've succeeded in making this viable but it is extremely difficult to characterize an 'Earth Battery' cell as a battery cell - the numbers simply are non-linear.

However, consider a few things and tell me if this could actually be a transistor.

Let's take a bin of soil that is mostly made up old compost - very heavy in clay, which after doing some research I came to realize contains not only a mix of heavy metals but a lot of silicone.

Into this bin we place an 8"x8"x.090" plate of Magnesium (AZ31B - 96% Mg, 3% Zn, 1% Other) and at the other end of the bin a 4"x4"x.090" plate of pure Graphite.

If this were a battery cell, the Carbon has a potential difference of 2.3 V compared to the Mg when in an electrolyte. Because of pH differences, moisture difference, etc, we only get 1.3V ~ 1.8V and we look for a source resistance by shorting the cathode with the anode and looking for the instant current and we also take various measurements with resistors and we've done this on Electronics Engineering Stack Exchange with results that don't make sense.

Lots of reasons are given why these numbers don't make sense. However, I modeled a pnp transistor on Ltspice because of a suspicion I had and was able to get some of the results.

Here's the theory: enter image description here

We have Carbon, which is a Group IV element, so in silicon that would make it a p-type material, heavily doped because it fouls in soil. Then we have the Mg which is a Group II element in silicon which is also a p-type element and is doped by the oxidation process (the other heavy metals in the surrounding clay embed themselves into the Mg as it corrodes, I've witnessed this.

Now we have a PNP transistor with the Emitter being the Carbon, the Base being the Soil and the Collector being the Mg. Also, because the soil is moist, creating an electrolyte, it allows electrons to flow.

Second, when I do resistor tests over time (over 24 hour periods) I see rises and falls in the current around sunset and sunrise - proving the telluric current concept spoken of in Wikipedia. That would be something more familiar to the base of a transistor than to a battery, wouldn't it?

Also, Here are some of the tests I performed when trying to characterize it as a battery:

Shorting the device and reading the current. Using the ammeter setting on my multimeter which has separate settings for μA, mA, and A. Set to μA. Positive lead attached to Cathode, Negative Lead to Anode. Result: 0.00 (possibly not sensitive enough?)

Finding the Open Voltage. Using the Voltage Setting with the multimeter still attached as in step 1 but leads are separated so as not to touch. Result: 1.327 and slowly rising [over an hour this will rise to 1.68]

  1. Attach various resistances and measure voltage.

Voltage setting with multimeter still attached as above. The resistor is measured first separately, then placed in between the multimeter leads. (Voltage shorted to 1.327V prior to each test) Resistors are Carbon film except in the one instance mentioned. The voltage source has an unstable frequency measured on an oscilloscope between 2Hz and 10Hz


  • 2.3 ohms, 1.4mV
  • 5.1 ohms, 6.7mV
  • 21.9 ohms, 14.5mV
  • 99.5 ohms, 127mV rises at rate of 2mv/s. (127mV is initial reading)
  • 150.9 ohms (wirewound resistor), 182mV*
  • 325.8 ohms, 347mV rising

*The 150.9 resistor started out at 110mV and rose at a rate of approx. 0.5mV/s for over 7mins, at which point I took the 182mV reading and stopped. While I recorded the readings and looked up, it had already reached into 188mV's and still was rising.

  1. Using the Diode setting (to see how it would charge). (leads attached in the same fashion)

Result: 1.715 rising rapidly to 2.0 at which point my multimeter went to O/L. I quickly went to μA setting and shorted the leads. still 0 μA

  1. Back at Voltage Setting.

Result: now at around 1.71V and slowly dropping until it reaches 1.35V over twenty seconds

  1. Use Diode setting to 'recharge' to 2.0V. Return to Voltage Setting and Short

Result: Immediate discharge from 1.98V to 1.35V

  1. Capacitance Setting

Result: Multimeter read 0.000 [indicating seeking] then eventually O/L [Too large] Max 2mF for this multimeter.

And that made me think of the Pi model for a transistor: enter image description here

This I have modelled in LtSpice, though I don't know the various C or R values and guessed the Current Value and rbb' value based on actual output. And my resulting pnp transistor operates fairly close to the results I get.

Anyhow, I would love you to smash this theory to bits!

Or, provide a wonderful circuit to disprove it in.

Thanks, Robert

  • 1
    $\begingroup$ Magnesium in crystal silicon is a donor (even a double donor just for fun). The rest seems to be trying to model a complex non-ideal electrochemical system as a single entity, an approach that is unlikely to succeed. $\endgroup$
    – Jon Custer
    Apr 5, 2022 at 0:36
  • $\begingroup$ @JonCuster Can you make a full answer expanding on why it's unlikely to succeed? I have been using it as a battery cell for the past eight months but it doesn't act at all like one. For one it amplifies earth's telluric current - proven by rising and falling current peri-sunrise and sunset. Second, adding two cells (1.8 V + 1.8 V <> 3.6 instead = 2.0 V) Whereas an Ltspice transistor model like above acts more like the latter. I need a definite proof. $\endgroup$
    – RobMcN
    Apr 5, 2022 at 0:48
  • $\begingroup$ Would Electrical Engineering be a better home for this question? $\endgroup$
    – Qmechanic
    Apr 5, 2022 at 4:07
  • $\begingroup$ @Qmechanic EE asked if Physics might be better. Somewhere in between, I gather. $\endgroup$
    – RobMcN
    Apr 5, 2022 at 4:26
  • $\begingroup$ Crossposted from electronics.stackexchange.com/q/614573/52589 $\endgroup$
    – Qmechanic
    Apr 5, 2022 at 4:30

2 Answers 2


You have a dirty system, with tens of signals. If you want to draw any meaningful conclusions, measure each effect separately.

Telluric current? Use the same material for electrode to remove earth battery effects

Earth battery? Use different direction of the gap between electrodes to find the least affected by telluric current to exclude it

Immunity to thermal effects? Measure telluric currents and earth battery at different temperatures but at the same time of day

Immunity to bio processes? Measure telluric currents and earth battery effects at different time of the year, but the same temperature and the same time of day

Geography related anomalies? Measure everything in different locations.

This will help you to decompose all of the effects into separate ones. And then some of effects will become predictable. Once you will be able to predict them, you can create and test models. Currently there is a lot of options available, and your data is not sufficient to exclude a lot of models.

  • $\begingroup$ I understand that there are many ways to, as you say, 'decompose all of the effects'; its also a lot of leg work and I'm disabled. I was hoping instead of theorizing the properties then creating a circuit that would make use of those properties. I have actually made tables and tables of data already while attempting to to characterize it as a battery and as an earth battery. And I have made use of it in circuits as a cell, but its very unpredictable. Hence, exploring its properties as a 'device' of some kind. $\endgroup$
    – RobMcN
    Apr 5, 2022 at 3:55
  • $\begingroup$ @RobMcN either this, or test it like a black box. Dont assume anything, and just make graphs, and find suitable curves to extrapolate points. If you do ask about details about how it works, and you have no idea how it works, you have to decompose all of its effects into known ones. $\endgroup$ Apr 5, 2022 at 3:58
  • $\begingroup$ I'll post all my load test and peri-sunrise peri-sunset graphs tomorrow. If you have suggestions for tests, let me know and I'll do them and collect the data. I have a great oscilloscope and multimeter here. Lots of electronic parts/ics for building anything might be needed. $\endgroup$
    – RobMcN
    Apr 5, 2022 at 4:00
  • $\begingroup$ @RobMcN and in general, in limited resources scenario, consider the payoff. Earth battery just wastes the metal electrodes. Telluric currents are so tiny they are not worth the work of installing the electrodes. Even if there is a transistor-like effect, it is for sure to be much worse than existing transistors in terms of amplification. What possible payoffs can you expect to achieve? $\endgroup$ Apr 5, 2022 at 4:01
  • 1
    $\begingroup$ Thats about 1$ per 1kwh. Order of magnitide more than what it is sold for from a socket. Because metals just cost more than their energy content. Unless you expect to get more energy from metals than there is in these metals. $\endgroup$ Apr 5, 2022 at 4:07

I received an answer to this question from user U235 in a semiconductor forum which I believe is an end-all to the theory.

In reference to the above:

"I imagine you don't normally want a transistor to have its own voltage potential compared to ground, which is why the materials are called "semi"-conductors."

This is incorrect.

"We have Carbon, which is a Group IV element, so in silicon that would make it a p-type material"

This is incorrect. Carbon doped Silicon is an intrinsic semiconductor -- neither n nor p-type.

"... but I can't seem to find the electric potential of SiGe (Silicon Germanium used in most p-junctions) except that it is a semiconductor."

This doesn't make sense, to be honest. SiGe is a semiconductor alloy, and it's used in a number of applications. The electric potential of a SiGe region/material lump could be anything, depending on the circumstances. i.e. what it's connected too. You also need to be careful of the subtleties of electric potential vs the electro-chemical potential (fermi level) in semiconductor devices.

Proper Mg doped silicon would be a double donor (I think). So, it'd be n-type. It's not a common thing to do, however.

This is not an attack on your work: subjects like semiconductor and solid state physics aren't straight forward. They need a significant amount of study. Simply speculating, guessing, or theorizing how semiconductor devices work will mislead you.

IMO: you cannot make a transistor in the way you suggest.

Good luck with your work.

  • User U235 -SemiWiki.com
  • $\begingroup$ Never heard of an intrinsic semiconductor before. Thanks pointing out the difference between fermi level and electric potential. What this has done, is helped to dissuade me from the transistor theory and put more effort into the earth battery cell characterization. In that, I consider your answer to be complete and the subject closed. However, if you felt like expanding my knowledge in some of those points above or pointing me to some good reads, I would be grateful. I love knowledge. $\endgroup$
    – RobMcN
    Apr 5, 2022 at 18:58

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.