Sciencemadness Discussion Board

Bucket cell - interesting observations

Swede - 19-11-2009 at 07:38

My bucket cell adapter is running with a full data stream to a DATAQ electronic chart recorder, measuring voltage, current, and temperature. pH will not be possible due to the harsh environment, but I may execute a data run in the future using a cheap, disposable probe.

The original plan was to hopefully correlate current and voltage so as to determine EOR (End of Run) conditions without having to measure chloride ion concentration. It is also the first use of an upgraded HCl delivery scheme using an Auber Instruments inexpensive timer.

The cell is a 2.5 gallon HDPE bucket with 8.7 liters of pure KCl at 17.9% chloride, MMO anode, Ti cathode, and a good CC supply. The liquor was saturated at 50 degrees C, quickly added to the cell, and the current cranked quickly to keep the liquor hot. Here is where it gets interesting...

With no cooling, the temperature at 40 amps CC climbed to as high as 76 degrees C. At this point, I pointed a small fan at the bucket, and this proved dramatic in lowering the temperature. Fan on, the temp drops to 55 degrees C. from nothing more than blowing air over the bucket.

Between 55 and 75 degrees C, I noticed a very distinct correlation between temperature and voltage to maintain 40 amps. As the temperature climbed, the voltage dropped, and as the fan cooled the bucket, the voltage climbed once more. Several data points confirmed that higher temps require less power overall. Over approximately 18 hours, while the cell chemistry is still fairly young, 40.00 amps:

V------------T
5.05----60.49
4.80----71.02
4.76----76.32
4.84----70.86
4.99----61.74
5.05----59.71
5.16----55.81
5.26----54.17

The fan was turned off, the temp allowed to climb again:

4.96----70.63

A solid correlation that should have been obvious from other runs. It pretty much kills my notion that the voltage and current data can be used to determine EOR, unless a cell's temperature remains constant, which it rarely does.

I'm certain that the literature on this process describes the reduced power requirement at higher temps, but I had not run into it before, or if I did, it went over my head.

Temperature is your friend for the bulk process, just watch out for the materials! 75 degrees C is pushing it on a small HDPE bucket. And again, it shows that heating is always a problem when attempting a higher-current run measured in days rather than weeks.

Swede - 19-11-2009 at 07:43

Oh yes, I forgot to mention - the bucket cell hardware is performing admirably. As usual, I made the vent too small, and the liquor level was too high. I noticed some seepage, drew off 300 ml of liquor to lower the level of electrolyte, and since then, there has been nothing more than a slight salt creep. I am using a bubbler for agitation rather than a mechanical stirrer, which I do have fitted to my large cell.

Each electrode shank has a small heat sink on it, and at 40 amps, they are not too hot to touch. I'll get some pics up soon.

I also wanted to re-arrange the data, from cooler to hotter:

V------------T
5.26----54.17
5.16----55.81
5.05----59.71
5.05----60.49
4.99----61.74
4.96----70.63
4.84----70.86
4.80----71.02
4.76----76.32







[Edited on 19-11-2009 by Swede]

[Edited on 19-11-2009 by Swede]

12AX7 - 19-11-2009 at 09:00

I would be interested to see the V-I curve (for instance, at constant V instead of constant I). That would have to be plotted for different temperatures, which would take a long time per datapoint and an external heater/chiller.

Offhand, your datapoints line up fairly linearly as T(V) = -44.7*V + 287.38. Evidently, if you put it in an autoclave it'll proceed all its own around 287C. 5.87V would seem to be the initial voltage (i.e., room temperature).

Tim

tentacles - 19-11-2009 at 14:59

Tim, that's all we need, the pressure cooker chlorate cell!

12AX7 - 19-11-2009 at 15:37

What's the steam pressure at that temperature, 50 atm? Not bad... Too bad the majority reaction would run in reverse; I can't see hydrogen seperating from the steam, anyway. :P Good way to dispose of chlorate though I suppose? Nahh... there are far more entertaining reactions. :D

Tim

[Edited on 11-19-2009 by 12AX7]

watson.fawkes - 19-11-2009 at 17:56

Quote: Originally posted by Swede  
Between 55 and 75 degrees C, I noticed a very distinct correlation between temperature and voltage to maintain 40 amps. As the temperature climbed, the voltage dropped, and as the fan cooled the bucket, the voltage climbed once more.
Ion mobility is higher at higher temperatures, which leads to a negative temperature coefficient. This is because current flow is the electrolyte is a diffusion-drift process.

As temperature increases, a second process, inelastic collisions, start becoming significant. This is the driving process behind positive coefficients of solid conductors, since their majority charge carriers (electrons) are much more mobile that ions are in liquid. I am doubtful that you'd see much attenuation in conductivity at atmospheric pressures and sub-boiling temperatures.

Swede - 20-11-2009 at 08:52

I have allowed the temperature to stabilize, and at about 2.5 days, I am starting to see the behavior I noticed earlier... at a given temperature, as the chloride concentration decreases, the voltage required to maintain CC goes up. Right now the bucket is at 55 degrees, and at 14.2% chloride, vs a starting concentration of 17.9%, 5.55 V is needed to maintain 40 amps, compared to 5.16V when the chemistry was young.

Initial hacks at efficiency show 87% based upon chloride reduction over the first 24 hours. This will probably drop a bit as the run continues. The pH is remarkably stable at 6.8, give or take a few hundredths. This is based upon the acid delivery rule of thumb:

0.057 ml of concentrated (32%) HCl per ampere, per hour.

I have the pump scheduled to deliver an appropriate mass of acid, which I diluted to 16%, every two hours. After a few runs now with a dosing pump, I have come to the conclusion that a superior method is simply an appropriate solenoid valve (probably PTFE, available on eBay cheaply), a decent timer, and a simple gravity system.

Anyway, this has been a data-rich run, and it will be interesting to create and compare some of the curves generated by the data-collection scheme.

Contrabasso - 20-11-2009 at 13:19

Can you get a driver to adjust the fan speed to keep the temp at 50c then you can fix the temp and do V/I plots at that temp,

Swede - 21-11-2009 at 07:20

Quote: Originally posted by Contrabasso  
Can you get a driver to adjust the fan speed to keep the temp at 50c then you can fix the temp and do V/I plots at that temp,


Fortunately, after the first few hours with wild temperature swings, it has settled down to 55 degrees, +/- 3 or so. It should generate decent data. The voltage at 40.0 amps CC continues to climb.

This morning the pH was a bit low, at 6.4. I adjusted the timer a bit. This fits into the notion that as a cell matures, the Cl- decreases, less HCl is needed. But the HCl rule of thumb does come very close, and is a useful tool. Interestingly, I would have expected the voltage to dip temporarily after an HCl injection, but it does not move at all, that I can see. Perhaps the HCl volume is simply too low to affect the voltage.

[Edited on 21-11-2009 by Swede]

dann2 - 21-11-2009 at 11:18

Hello,

There is some reading here about the Voltages accross Chlorate cells. Page 14 states:

Reversible Potential
Anode 1.39
Cathode 0.39

Over Voltage
Anode 0.05
Cathode 0.88


Ohmic drop in electrolyte 0.29
Ohmic drop in hardware 0.10

Total cell Voltage 3.10

Above cell is a Pt/Ir Anode BTW


What varies as the cell runs towards completion?
I guess all above except the Ohmic drop in hardware and the Cathode Voltages (both Overvoltage and the Reversible Potential for H2 evolution).

The Ohmic electrolyte resistance will change as the relative concentration of ions (Chloride/Chlorate) are changed. Since your cell Swede has much larger spacing between the Anode and Cathode this effect will be greater in your cell (I presume) compared to commercial cells that have 3mm spacings.
The Reversible potential of the Anode will change (thats the Voltage theoretically needed to get a reaction to happen) as the cell progresses, since the actual reactions taking place will change as Chloride gets more and more scarce.
You will get more Oxygen (and less Chlorine) being produced which has a somewhat higher Voltage requirement that Chlorine. The cell gets it harder and harder
to make the Chlorine as the NaCl is getting less and less plentyful.
Don't know what the Anodic Overvoltage for O evolution is or if it increases as more and more Oxygen gets produced (cell becoming depleated of NaCl).
Thats my two three cents worth on the subject.

The above ref. states (page 15) that the reason commercial cells are never operated below 100grams per liter NaCl is because too much Oxygen is produced and also too much Perchlorate! is produced. (Amounts of Perchlorate are very small and it's over a long
period that they build up).



Can anyone download that ref.?

I was wondering if a refractometer could be employed to keep track on the progress of a cell.
A not too expensive one like here


Dann2


[Edited on 22-11-2009 by dann2]

Swede - 22-11-2009 at 07:29

Another possibility is some sort of conductivity probe or meter. It would need either a very large spacing or great sensitivity, given the concentration of ionic species in the liquor, but a good system should be able to determine end-of-run conditions, once "calibrated" using traditional methods, i.e. measuring chloride ion via titration or similar.

With modern MMO, at least, I have never noticed visual, significant erosion at any chloride level so long as temperature and current density are normal. I did try the MMO --> Perchlorate attempt a while back, starting with KClO3 salts. I hammered that cell with a high current for several days, but the pechlorate produced was trace only. Still, no visual erosion of the anode. Given that, I have been taking chlorate cells down to <6% chloride, and stopping only because the efficiency plummets, not because the anode is being abused in any sense.

MMO in its modern form seems to be exceptionally tough stuff. If only lead dioxide anodes or material was as common... perchlorate remains a real challenge.

Swede - 22-11-2009 at 08:20

Update: Cell (8.7 liters, 17.9% chloride) was started on the 18th. This morning, 22 Nov, chloride was measured at 9.7%, pH too low at 6.4. I have twice adjusted the HCl delivery downward. All of the evidence points towards a diminished need for HCl as chloride drops and chlorate accumulates. Rather than a too-basic rule of thumb of mass HCl per amp-hour, it may be necessary to include either accumulated amp-hours or chloride concentration as part of the delivery equation. The HCl has been cut to 1/3 of what was needed earlier in the run, and the pH is still too low.

I don't think it will be hard to come up with a basic modification to the rule of thumb, but I will wait until this run is over before working at it. The V continues its upward trend as Cl- depletes. I will probably pull the plug tomorrow morning, and I expect the Cl- to be around 8% to 9%. Nothing like a good, hard current to make for a short run. With good materials and modest volume, a run can easily be less than one week, if there is a power supply capable of delivering the juice.

This cell also reinforces the fact that a surplus 5V supply capable of 100+ amps (super-common on eBay) will make for an excellent PS for a typical hobby cell. My anode area is about 200 cm^2, with spacing of 2.5 cm. A close spacing of 1.0 to 1.5 cm will reduce the necessary voltage, so a typical hobby mesh MMO anode of 3" X 5", spaced at 1.0 cm, Ti cathode, with 5V across the electrodes, should result in a 40 to 60 amp system.

dann2 - 22-11-2009 at 08:26

Hello,

You could perhaps make a conductivity measurer from two small pieces of MMO. (Or just one piece of MMO (Anode) and Ti (Cathode)).
As you said space them as far apart as possible in the cell and only pass a small current through them when measuring the conductivity. You would perhaps need to cut power going to the cell when measuring. Or you could have a very long path (few feet) for the measurement by putting the two probes into a narrow glass or plastic tube and taking a sample from the cell and putting it into the glass tube.
An AC might be better than a DC (I don't know). You will need two MMO probes if using AC I would presume. Temperature would have to be taken into consideration. Dissolved gasses might be a problem. You could perhaps boil a sample just before measurement to get rid of them.

I think that you will have in effect a Chloride to Perchlorate system (Sodium salt anyways) with the MMO reducing the Chloride level to (say) 20? grams per liter and then the Lead Dioxide taking over from there. No seperation of Chlorate.

How many grams per liter is 6% Chloride (approx.)?

Blessed be LaserRed and all his sales!!

Cheers,
Dann2


[Edited on 22-11-2009 by dann2]

watson.fawkes - 22-11-2009 at 16:04

Quote: Originally posted by Swede  
All of the evidence points towards a diminished need for HCl as chloride drops and chlorate accumulates. Rather than a too-basic rule of thumb of mass HCl per amp-hour, it may be necessary to include either accumulated amp-hours or chloride concentration as part of the delivery equation.
This seems like the chlorate may be acting as a pH buffer for chloride. It wouldn't be a single-species buffer, but rather four, including hypochlorite and chlorite species along with the other two. It might include others, as well. I'm not currently motivated to work out all the consequences of the simultaneous equilibrium equations, but it's not completely intractable, either. Regardless, the upshot is that if there's a buffering effect, then the practice of adding acid proportional to total charge is going to yield overly acidic conditions (as observed). Doing pH control, then, is going to be more complicated that a constant rate dose, at the very least. I see three ways of doing the control:

First: Open loop control based on a theoretical model. There's an implicit function defined by equilibrium equations between the total charge from the supply and the resulting pH. That function could be computed from first principles and rendered in an algorithm.

Second: Open loop control based on an observed run. Instead of doing math, do an observation campaign on the cell. Run it for, say, an hour, then turn it off and add acid to the correct pH. Record the amounts of acid used and use it as a dosing schedule for the future.

Third: Closed loop control based on a pH sensor. It's been discussed; pH probes for this application are hard.

Swede - 23-11-2009 at 06:56

Good observations. WF, of your three methods, I like #1 the best, creating a "model" that, while not perfect, is close enough so that a reasonable pH control is exercised. Variations in everything from cell geometry to circulation are going to cause differences between cells, but I think the end result will be close enough to satisfy a garage hobbyist. There IS I believe some sort of buffering action that occurs as the cell matures, which can be used to our advantage, in that imperfections in the model will be small enough to allow the pH range to still remain reasonable, and I'd call reasonable anything from 6.5 to 7.1. If the pH can be kept in that region, then efficiency will soar relative to an uncontrolled cell.

An experimenter with a pH meter can fine-tune the model with his own observations for subsequent runs. A guy with no meter will at least improve CE well above an uncontrolled cell.

pH papers simply cannot hack the bleaching, and the interpretation of the skinny color band that flashes briefly during a test is too imprecise. I think modifications to an acid delivery based upon pH papers is like a dog chasing his own tail. You are going to cause more problems than you fix. So the delivery model must be able to stand on its own, yet still create acceptable results, while a guy with a decent pH meter and probe can do a bit of fine tuning. Even then, with a good model, little to no adjustments should be necessary.

Swede - 23-11-2009 at 07:08

@dann2 - I have very little experience with conductivity measurements of this sort. Wouldn't plain Ti work as well as MMO? Isn't the area of the electrodes critical? Perhaps you could use Ti wire perhaps 3mm diameter, immersed deeply at two points in the cell, and with the majority of the wire covered with PTFE heat shrink, so that the readings will not be altered by variations in the level of electrolyte.

dann2 - 24-11-2009 at 17:19

Hello Swede,

Since it is a relative conductivity you would be measureing then the electrode area will not matter much. If you want an absolute value then the electrode area will matter.
The reason why the Voltage rises accross the cell may not have much to do with the conductivity of the solution. (I don't know myself). The number of ions of Chloride that are destroyed are replaced with the same number of Chlorate ions so number of ions stays the same. The mobility of Chlorate ions may be less that Chloride so giving lower conductivity (or if they are more mobile they will give greater conductivity).
See what sort of a 'signal' (rising Voltage) you get from your set up and perhaps there will be no need for conductance measurements. It may be the changing reactions as the cell progresses (more O, less Cl at Anode) that causes most of the rise in V anyways.

If using Ti as a conductance probe it will not do connected to the + of the conductivity meter. With an ac it would not do as either electrode.


Some info. here on conductivity measurements:

http://www.apcs.net.au/kn/kn20002/kn20002.html

Dann2

Swede - 30-11-2009 at 08:05

TY Dann2. Odd thought - It would be wonderful if a simple conductivity system might also detect pH changes... it would certainly be very crude, not sensitive, but if it could keep the pH between 6.0 and maybe 7.6, it could be turned into a pH controller without the use of a real pH probe, with all its drawbacks.

So conductivity is wide open as a tool for cells. It'll simply take a bit of effort to see if it can be made useful. I need to educate myself a bit.

watson.fawkes - 30-11-2009 at 13:38

Quote: Originally posted by Swede  
So conductivity is wide open as a tool for cells. It'll simply take a bit of effort to see if it can be made useful. I need to educate myself a bit.
Any sort of analytical use of this is going to be affected by electrode polarization. Make sure you understand this. I have to imagine that the Nernst equation is going to come into play as well, although I'm not sure exactly how, as I haven't thought the problem through in any way.

One open idea I've had in this realm (in my case, I was thinking about copper chloride etchant baths) is that the bath acts not only like a resistor, but a capacitor, a very non-linear one. A pulse on one electrode seems unlikely to appear as a pulse on the other one, since it's not a matter of simple time delay. And reversing the current direction should yield different results with the same pulse, since the majority charge carriers have different mass and mobility. So the shape of a sharp pulse as modified by the bath may yield useful information.

12AX7 - 30-11-2009 at 15:26

It's a soup of ions, not a crystal, so I don't think you'll see fancy semiconductor effects. For instance, by forming a PN junction at one end of a bar, you can introduce excess charge carriers in that bar. By drawing a current along the bar, they drift to one end. So you can monitor conductivity at one end while pulsing the diode at the other end, watching the blob of charge as it moves along.

In a solution, you can't get the same effect because you aren't modulating ions, they're always there (and in extraordinarily high concentrations relative to the charge carriers in a semiconductor).

If you apply a pulse to the solution, I'm willing to bet you'll see that pulse appear everywhere else, in direct proportion to the resistance to that point (as a resistor divider), with the rising edge limited by the RC time constant of the system (there may be a sharp edge due to capacitance in the solution (capacitor divider), but it may not be full height, swamped by having more capacitance to other points).

Tim

dann2 - 30-11-2009 at 18:46

Hello,
One item not mentioned is that if you are making conductivity measurements in the cell you will have to turn off the power supply going to Anode/Cathode or it will swamp your measurement.

You could use Pt (very small pieces) as the probes if you like, I guess.
If you use AC to measure the conductance then probe polarization will not be an issue???
Don't know much about the subject though.

Dann2

watson.fawkes - 30-11-2009 at 19:54

Quote: Originally posted by 12AX7  
It's a soup of ions, not a crystal, so I don't think you'll see fancy semiconductor effects.[...] If you apply a pulse to the solution, I'm willing to bet you'll see that pulse appear everywhere else, in direct proportion to the resistance to that point (as a resistor divider), with the rising edge limited by the RC time constant of the system
I don't there will be semiconductor effects either. Nor do I think there would be steady state difference. I do think there might be a measurable difference in the transition between conduction and non-conduction. This is because you have to accelerate the charge carriers, which are not electrons in this case, but rather ions four or five orders of magnitude heavier. In modeling terms, this would imply the resistance isn't constant. It's already very much non-constant when the electrode polarization region is not in steady state. The mass of charge carriers creates a second kind of transitory effect.

12AX7 - 1-12-2009 at 08:51

Ah, but you aren't accelerating ions, they're already moving. You're just applying a bias, and they drift casually as a result. The actual motion of ions will still be >> 100 m/s and with so many ions, every charge added at the cathode will be removed instantly at the anode.

Under different levels of bias, there could be variable resistance. In semiconductors, this occurs when drift velocity exceeds thermal velocity; the material becomes a constant current element as charge flow saturates. (I don't know if this can actually be observed; the electric field required tends to strip electrons off the lattice, causing avalanche first.) But I don't see that happening for any reasonable bias, either; such a thick soup would boil before you could get anything moving fast enough through it.

The typical model for a chemical cell is a voltage source (at the redox potential) in series with a resistor (the ohmic/drift resistance). There may be an exponential part, but that could be a result of thermal heating.

Tim

watson.fawkes - 1-12-2009 at 18:42

Quote: Originally posted by 12AX7  
Ah, but you aren't accelerating ions, they're already moving. You're just applying a bias, and they drift casually as a result. The actual motion of ions will still be >> 100 m/s and with so many ions, every charge added at the cathode will be removed instantly at the anode.
You apply a force from an electric field, you accelerate ions. What you are saying is that this isn't their only interaction. Fair enough, but that doesn't mean that there's no work being done, or that you can't (in principle) observe the acceleration.

Now as to the 100 m/s figure, at the very least this needs some qualification. We're talking about a liquid here, where Brownian motion reigns supreme and "mean free path" is more-or-less meaningless. I did some searching for "liquid molecular velocity distribution" and there's just not much out there. It's not analytically simply like the Boltzmann distribution for an ideal gas. I thought I could look at the distribution for water, and I can't find that. I did find a series of papers "A General Kinetic Theory of Liquids" by Born and Green, but they're mostly theoretical. There's clearly thermal motion in a liquid, but I'm not sure what the mode distribution is. I would guess that there's far more rotational energy than in a gas, given all the intermolecular forces.

I did have a second thought about this. A good fraction of the pulse energy is going to go into polarizing the electrolyte, so there's a competing process going on here. I haven't done an estimate about energy partition in the pulse between this effect and ion acceleration. At this point I'm not even sure about what the orders of magnitude are.

As to my previous comment about asymmetry between pulse polarity, that will only happen if the electrolyte can be considered at constant potential, which is tantamount to putting in a third electrode. So while that idea isn't valid, it does point out some other possibilities, which is to use more than two electrodes to take the measurements.

Swede - 2-12-2009 at 06:34

How about killing the power to the system and using the primary anode plus a third electrode to take pertinent measurements?

watson.fawkes - 2-12-2009 at 15:06

Quote: Originally posted by Swede  
How about killing the power to the system and using the primary anode plus a third electrode to take pertinent measurements?
Suppose you're taking two-electrode measurements. One electrode is already in use and the other is reserved for measurements. If you use a floating power supply for the measurements, you don't have to turn off the main supply. The only real requirement is to ensure that your measuring apparatus has high impedance so it doesn't become (much) of a parallel current path or affect the measurement too much.

But this issue is premature. The first thing to do is to take some trial measurements in a test rig and figure out what the right correlations are.