Four electrodes with 1cm spacing.
Today I decided to revisit the water-conductivity experiments for the course, now that I have an easy way to do proper impedance spectroscopy (including phase information as well as magnitude), using the network analyzer function of the Analog Discovery 2 USB oscilloscope. I wanted to look at 4-electrode measurements, as well as the 2-electrode measurements we’ve done in the past.
First, I made myself a 4-electrode device, by cutting some ⅛” stainless-steel welding rod (316L rod for TIG welding) into 15cm pieces, drilling 4 ⅛” holes in a scrap of cutting-board plastic, and driving the rods through the holes with a hammer.
I then immersed the short end in tap water (using a mason jar, so that the long end stuck out the top) and used alligator clips to attach wires from the electrodes to a breadboard.
I connected the function generator through a series 1kΩ resistor to one of the end electrodes and ground to the other end electrode. Channel one of the oscilloscope measured the voltage across the 1kΩ resistor (hence the current in milliamps).
Channel two of the oscilloscope was connected to either the two end electrodes (making a 2-electrode measurement similar to what we’ve done for years in the class), or to the two middle electrodes, for a 4-electrode measurement. The idea of a 4-electrode measurement is that there is an electric field established in the bulk material by the outer electrodes, and the middle electrodes can measure that field without interference from surface effects that occur on the electrodes that are providing the current.
I used the network analyzer function to sweep from 2Hz to 10MHz. I exported the data so that I could plot it as impedance (rather than as just the dB ratio of the two measured voltages). For the 2-electrode measurement, we are measuring the impedance of the water and electrodes (voltage across the electrodes divided by the current through them), but for the middle electrodes, we’re looking at the voltage across the middle electrodes, divided by the current through the end electrodes.
The voltage across the middle electrodes is nearly a constant, up to about 1MHz, where wiring inductance starts to matter. The surface chemistry interferes with measurement of bulk properties at low frequencies for the 2-electrode measurement.
The plot of the phase shows even better why 4-electrode measurement is useful:
The capacitive nature of the two-electrode system is seen at low frequencies, but the 4-electrode system has a resistive, nearly 0° phase shift (up to the point where the inductance of the wiring to the reference impedance starts to matter).
I don’t think I’ll switch to 4-electrode measurements this year (if for no other reason than that I’d have to make a dozen new electrode sets), but I’ll keep it in mind for next year.
Gas station without pumps 
Hello,
Why do you think that the 4 electrode method is better ? From the impedance vs frequency plot I can see that the impedance value is lower. How does this make the approach better ? Have you also plotted Nyquist plots for the measurements ?
Comment by Agnivo Gosai — 2017 February 24 @ 15:20 |
The 4-electrode method is flatter with respect to frequency—the electrolyte (water and ions) is basically resistive, all the capacitive effects come from surface chemistry at the electrodes. If we are trying to measure bulk properties of the electrolyte, then the surface effects are not signal, but undesired distortion.
The actual value of the resistance is going to differ, because I do not have the same electrical field in both cases. For measuring ionic concentration, one always has to calibrate with a known concentration, to correct for geometric effects, in any case.
Comment by gasstationwithoutpumps — 2017 February 24 @ 21:34 |
Thank you for the helpful post for my experiment.
I have one question.
the 4 electrode method at the high frequency(> 100000 hz), Why are the impedacne value lower compared to value at low frequency?
Comment by jae — 2018 July 29 @ 19:22 |
The impedance for electrodes in salt water can be roughly modeled as the conductance of the bulk salt water and surface effects that act like a thin insulating layer on the surface of the electrodes. That thin insulating layer can be approximately modeled as a capacitor with a resistor in parallel for the leakage through the layer. As the frequency goes up, the impedance of the capacitance goes down.
An impedance spectrum that is flat out to a particular frequency then drops as constant/f can be well modeled as a resistor and capacitor in parallel.
Comment by gasstationwithoutpumps — 2018 July 29 @ 19:29 |
[…] the stainless-steel rod and the spring are leftovers from previous projects: Two-electrode vs. four-electrode impedance spectroscopy and Physics Lab 4: spring constants […]
Pingback by Shakespeare cookie cutter | Gas station without pumps — 2019 July 14 @ 14:50 |
I know this is a very old post but if it’s still being monitored I was wondering if I could ask some questions about your setup and wiring? I ask because I am engaging in an experiment where we are thinking of doing a four-electrode approach to measure impedance across different biological tissues. This is also of course for my own personal edification as I learn how to use my Analog Discovery 2
Comment by RJ — 2022 December 14 @ 12:45 |
You can ask. I may not be able to help, as it has been a while.
Comment by gasstationwithoutpumps — 2022 December 14 @ 12:50 |
Thank you for responding so quickly and sparing some time to entertain my questions! I should preface this by saying that I am definitely a novice at electrical engineering concepts but I am learning/trying to learn, so actually I’m very happy to have stumbled upon your youtube videos as well and have found them helpful!
So for the experiment our basic setup idea is to have a four-electrode system with two of the electrodes being parallel to each other on one end of the tissue transmitting the signal and two other parallel electrodes at the end of the tissue receiving the signal. Unfortunately comparable studies have been a bit sparse on the details of the wiring for this setup especially using the Analog Discovery 2 (which is what my lab has) so my question is if you perhaps have a diagram of your 4-electrode wiring into the Analog Discovery that I could use as a basis to prototype our setup (I’m still learning how to conceptualize written circuitry to physical circuitry)?
I imagine that ultimately it will be a slightly different design since there are no “middle electrodes” in this case. But as part of the learning process I’m trying to understand the implications of different electrode setups so I’m sure this will at least get me started! I sincerely appreciate your time!
Comment by RJ — 2022 December 15 @ 10:24 |
For 4-wire measurements, two of the electrodes are used to put a current through the thing being measured, and the other two are used to measure the voltage drop due to that current. If you have an essentially straight-line thing being measured (a pencil lead, a metal rod, …) then using two outer electrodes to set up a current that is essentially the same everywhere along the thing. If the voltage measuring tool between the middle electrodes has a much higher input impedance than the thing being measured, then you can treat the current through the device as being the same everywhere, so you just measure the current between the outer electrodes (using an Analog Discovery 2 channel across a reference resistor in series with outer electrodes and the function generator) and the voltage across the middle electrodes (as a separate channel).
The test I did with 4 electrodes in solution is not very good for an absolute impedance measurement, because the current flow is along curving lines between the end electrodes, and only a portion of the current flow goes past the middle electrodes. If I has the 4 electrodes in a narrow trough, the measurement would be pretty accurate, but in a wider bath, the geometric factor should be determined by using a known reference concentration.
In your case, with two electrodes close together at each end of a sample, the assumption is that the voltage between the undriven electrodes is very close to the voltage on the sample at the driven electrodes (but removing any surface electrochemistry at those electrodes). So you just measure the current through the driven electrodes and the voltage across the undriven electrodes. That should work fairly well as long as the two electrodes at each end are very much closer together than the distance between the two ends.
Comment by gasstationwithoutpumps — 2022 December 15 @ 18:38 |
Thank you so much for your very thorough response! This has been supremely helpful!
Comment by RJ — 2022 December 19 @ 07:07 |