Turquoise Energy Ltd. News #40
Victoria BC
Copyright 2010 Craig Carmichael - June 2nd 2011

http://www.TurquoiseEnergy.com = http://www.ElectricHubcap.com = http://www.ElectricWeel.com

Spotlight: A new high energy, economical battery chemistry
: 2.1 volt V-Ni salty cell.

Month In Brief (summary)
  * Most inventors are "unemployed"? Well, duh!
  * Raising the bar - Mechanical torque converter - battery design - Hubcap motor production

Electric Hubcap System

  * A3938 V2 Motor Controller
  * Motor Production Setup

Electric Weel Motor Project (Electric Wheel Motor... Rim Motor...) (No activity or report.)

Planetary Gears Project
  * Disappointing test in April leads to renewed thoughts about dropping planetary gears, making a torque converter work.

Torque Converter Project
  * oscillating planetary gear torque converter: rotary version of Constantinesco's original design - an idea from chat list "half bakery.com".
  * My version of the planetary gear version
  * Types of MTCs
  * New idea! A spring-loaded MTC - designs thereof...

Turquoise Battery Project
  * Salty electrolyte battery research: stalled since the 1880s!
  * But not high budget lithium research!
  * Cell with Nickel (+1 v), Vanadium (-1 v) Negatrode ? ... didn't work...
  * simply reversed the leeds: +Vanadium, -Nickel cell ...WORKS!, 2.1 volts, should have 200+ WH/Kg!
  * Vanadium Pentoxide+: ~ +1v, double electrons: higher energy than nickel hydroxide or manganese dioxide as a positrode - works in salt electrolyte only.
  * (Potassium permanganate "+" may be even better than vanadium pentoxide: more electrons, same voltage.)
  * Nickel-: ~ -1 volt in salt (!)
  * Carbon fiber to improve conductivity.
  * "Diesel Kleen" Solvent/Hexadecane": fabulous carbon/graphite conductivity booster! - electrode surface turns from "sandstone" to "silver".
  * Epoxy with graphite powder to improve conductivity between electrode and terminal post.

Newsletters Index/Highlights:
http://www.TurquoiseEnergy.com/news/index.html (oops - I moved the index in March and neglected to change the link until now!)

Construction Manuals and information:
Electric Hubcap Motor
- Turquoise Motor Controller
- 36 Volt Electric Fan-Heater
- Nanocrystalline reflective rear electrodes to enhance DSSC Solar Cells
- Simple Spot Welder for battery tabs, connections

- Electric Hubcap Motor Kits, Parts - Build your own ultra-efficient 5 KW motor!
- Sodium Sulfate 4x longevity additive & "worn out" battery renewal.
- NiMH Dry Cell Car Batteries
(please e-mail me to order batteries)
- NiMH Custom Batteries
(EVs, E-Bikes, Scooters, etc. - no extra charge)
- NiMH individual Dry Cells (D - 10 AH, $10 -- AA - 2.5 AH, $2.50)
- Motor Building Workshops

...all at:  http://www.TurquoiseEnergy.com/

Spotlight: A New High Energy, Economical Battery Chemistry

Vanadium-nickel test cell initial charging,
in transparent, openable case, with pressure gauge and adjustable pressure relief valve,
carbon terminals top and bottom.

   It appears I've finally created a good battery chemistry based on the designs and chemistries I've been gradually working out over the past 3-1/2 years. In fact, I believe it is the highest specific energy working, rechargeable aqueous cell made so far.
   The fundamental chemistry is vanadium oxide (+) and nickel (-) in salt solution electrolyte, yielding (to my surprise) a 2.1 volt cell. Depending on the conductivities and current capacities eventually achieved, this may translate to 1.75 to 2.0 volts "nominal" voltage under load.
   Vanadium pentoxide, a somewhat lighter substance than nickel oxyhydroxide and moving two electrons instead of one (per metal atom), provides high amp-hours at about +1 volt, and, rather unexpectedly, the charge and discharge products appear to be solids in neutral solution. (V2O5 <=> V2O3) On later inspection the electrode appeared to be unchanged from when first installed.
   The nickel as a negative in neutral solution, also unexpectedly, seems to have a higher reaction voltage than in either alkali or acid, about -1 volt. (This seems to be a great alternative to the manganese negatives I was doing, that seemed to be just a little too high a voltage (around -1.37) and consequently had a high rate of self discharge.)

   Higher amp-hours times higher voltage means much higher energy density, both by weight and by volume.  Both sides are well over 500 WH/Kg. Considering the weight of the water, graphite, and all the things that must surround the active electrode materials, I tentatively estimate that V-Ni cells could be made with a specific energy as high as 200 to 275 watt-hours per kilogram. 200 is double the best NiMH dry cells. Lithium ion reaches up to 140 or so, but is typically less for larger cells that I've looked at actual specs for. The latest lithium-sulfur cells are about 200-290 watt-hours per kilogram. They're still experimental (with a chemistry problem yet to be overcome), and I have little doubt they're costly.

   If for example one wanted 10 KWH of battery in a vehicle, NiMH AA cells would weigh somewhere over 100 Kg. Lithium ion would be similar, and lithium iron phosphate would be heavier, perhaps 150 Kg. At 225 WH/Kg, the V-Ni battery would weigh under 50 Kg. If lithium sulfate becomes practical, it might be a little lighter yet... but surely costly.

   The exact workings of the cell were unexpected. It was intended to try +nickel, -vanadium, the other way around, expecting the very voltage that was in fact achieved, about two volts. However, the cell didn't seem to be charging well in that direction - the vanadium seemed to be charging into an unexpected soluble state, as it appeared to be migrating. (I made this cell from clear plexiglass and was able to see unusual activity within the cell -- one of my brighter ideas recently.) So I tried it the other way around... and it worked! I didn't think the substances used would produce such a high voltage charged in that direction, that the cell would be about 1 volt or 1.5. But the higher energy reaction of two possible vanadium reactions evidently applied, and the nickel negative voltage proved unexpectedly high, yielding exactly what I had intended... and more: the vanadium "+" has substantially higher amp hours than the nickel would have had, so less of it is required to match the negative.

  some theoretical positrode values:
  V2O5: 576.7 AH/Kg @ ~+1v  =  ~577 WH/Kg
  Ni(OH)2: 289 AH/Kg @ ~+1v  =  ~289 WH/Kg
  MnO2: 304 AH/Kg @ +.55v   =  ~167 WH/Kg (standard dry cell)

  nickel negative:
  Ni: 736 AH/Kg @ -1v  =  736 WH/Kg

   Here then is what appears to be a much superior, economical, green battery chemistry.

   To make an actually practical battery, I now hope to improve the internal conductivity at least tenfold (preferably closer to a hundredfold) for high current EV cells. Here I struggle with the problem that probably led past designers to choose alkaline over salt electrolyte: that all common metals quickly corrode away in a salt based positrode. This is why standard dry cells have a carbon rod for their positive electrode. Only conductive graphite and carbon products can be used. The conductance is generally lower, and unlike metal, graphite and carbon can't be soldered or welded to make convenient, secure connections.

   The current capacities didn't even seem to match the standard dry cell, though they were creeping into the neighborhood.

   But I've found an exciting new technique of compacting electrodes - wetting the substances with "Diesel Kleen" oily hexadecane solvent instead of water - to create (probably lamellar?) structures of graphite (graphene?) and electrode powder, creating a conductive network which gives a metallic appearance to the electrode surface, and lower resistances.
   Finally I experimented by simply mixing graphite powder with epoxy, and (again somewhat to my surprise) got much better contact between the terminal posts and the the graphite sheets with this "glue".

   And vanadium-nickel isn't the only possible chemistry. The nickel negatrodes seem to work well and have good voltage, giving them much more energy by weight than as alkaline positrodes (as in Ni-MH, Ni-Cd, Ni-Fe, etc), and hence making the nickel much more economical as well as smaller. They're a winner!
   However, I intend to try the manganese-nickel mix again as a positrode. If the manganese charges to permanganate as I expect, it should work about as well as vanadium or perhaps a little better. And unlike vanadium, manganese is "dirt cheap", improving economy. If not, vanadium will remain the choice, at least for vehicle batteries.

May in Brief

   Someone said recently in the news - and I confess I can't remember just who or where - that most inventors seem to be unemployed. Well, duh! If someone is working at a job, on top of maintaining a home, perhaps raising a family, when on earth would they be able to put in the investment of time needed to invent and develop something new? On the other hand, if they don't have a job, how will they remain solvent and able to do the work at hand? With no renumeration while working and virtually no prospect of reward even on successful completion, most people with ideas aren't free to develop them, even if they have all the skills necessary.
   Ovshinsky had to wait until he retired to create the metal hydride that made nickel-metal hydride batteries an excellent chemistry (and made GM's EV-1 fabulous), using his pension as R & D funding. He was a university chemistry professor, yet evidently even with such credentials and such a vision at such an institution, he wasn't permitted to do experimental work on paid time. Without his work, NiMH batteries today would probably still be just a scientific curiosity, not a fine product on store shelves. He did it all himself.
   I'm able to devote my time to inventing because I've managed to eke out a solitary existence since 1995 without a job, mostly by renting rooms in my house, and for several years now my elderly mother has been sending me money. In the last three years, Canada Revenue's SR & ED tax credit program has added a vital few thousand dollars a year -- a partial reimbursement of each previous year's R & D expenses -- and having hit 55, I am now deferring my property taxes, a major relief. (The city can have my house when I go.)
   With a Department of Progress in the government, the nation could work towards its goals of sustainability, soliciting proposals from would-be inventors and selecting (preferably from those with good track records) those proposals with the most promise and seeming to offer the most "bang for the buck". Paying individual inventors to invent and develop targeted products would probably cost the taxpayer far less money than is today being shelled out for "random" untargeted R & D funding to "innovative businesses" -- and produce tangible results and technical progress visible to the public. The department would naturally modify the patent system and take other action to recoup its investments and to see that technical advances don't go to waste. The government would have a stake in progress instead of in stagnation.

   Better motors, a superior green battery chemistry with new techniques for making salt electrolyte battery cells more conductive (after over 120 years with no progress!), and perhaps soon a good mechanical torque converter design (below) to finally allow a smaller motor to give more drive and batteries to go farther than is commonly thought feasible... In all this, an overriding goal is to point out that with just minuscule R & D effort, with a fraction of the resources expended daily in countless cities all over the world on petroleum engine maintenance and repairs, the performance bar can easily be raised significantly in key areas of transport technology -- and then to prove the point by raising it myself with my own R & D.

   Somehow I'd got my hopes up about using planetary gears, but the 2.8 to 1 ratio planetary gear performed in accordance with my lowest expectations, and I went back to my 7 to 1 estimate as being a good minimum ratio to put a car on the street. This would mean giving up on planetary gears or having a dual 4 to 1 system good for city driving only - and even at that, with two motors instead of one, using almost as much electricity as other systems. Two motors at 2.8 to 1 and full motor currents just might pull it off, and perhaps even hit 80 Km/Hr, but I'm dubious.
   I preferred giving up on the gears, and by the 4th had worked out a new mechanical torque converter design plus a variant, which seemed very exciting. Later I realized that neither would work.
   I'd also been sent an idea for a new version of Constantinesco's original oscillating masses MTC design. It used a planetary gear! It seemed the most promising idea. Surely in 2011 Constantinesco's 1923-26 design and product can be duplicated. But it hasn't been done yet!
   Then near the end of the month, I thought of another very simple converter idea, using springs to store potential energy instead of inertia weights. That would make for less "semi-sprung" weight on the wheel. It would also be easiest to put together, which naturally has a great appeal to me. After a number of sketches with improving designs and ideas, I'm proceeding with it.

   Finally, after last month's successful chemistry battery cell and a perusal of vanadium's electrochemistry, I put together one on the 6th with nickel and vanadium electrodes and found to my surprise, first that it didn't seem to charge as planned, and second that it not only worked if charged backwards - not a big surprise - but it that it made a 2.1 volt cell. That was a surprise - I expected 1 or 1.5 volts, depending on which reaction the vanadium decided to use. It seemed the nickel reaction (-1v) was twice the voltage (-.5v) expected from the charts.
   In fact it was even better than my plan, with more watt-hours per kilogram. I had evidently discovered a superior positive electrode. (...for salty batteries: Vanadium wouldn't work in acidic or alkaline cells.)
   The cell still has "the usual" problems with low conductivity, but this month I also created a new technique for enhancing it by compacting electrodes with solvent instead of water, and after a promising experiment I'm about to try out epoxied graphite powder to connect the terminal posts to the electrodes. The next cell may well see substantial improvement. If the present .5 amps becomes 5 or better, it should be a practical battery.

   The nickel (fried monel-lanthanum-bean sauce) negatrodes seem excellent. Now I'll try the mixed nickel-manganese again for the plus side - it seems to me those must have been charging to potassium permanganate, and so could work if coupled with good minuses. This would be even lighter than the vanadium as they'd move three or more electrons per Mn atom, and unlike V, Mn is cheap... especially from used dry cells such as those currently flooding the stores. (probably in an attempt to displace rechargeables from the shelves.)

   I didn't have time to look at the big Weel motor in May. After waiting weeks to have rotors cut by abrasive waterjet and paying over 350 $ for cutting to date, I'm thinking again that it would be nice to finish that pulsejet steel plate cutter and see how it works.
   And occasionally I think of my ideas for ocean wave power and for better DSSC solar cells, without having any idea when I might ever be able to get back to them with all the rest that's on my plate. Hurrah to Germany for its excellent decision to phase out dangerous nuclear power generation with its commitment to radioactive waste storage for longer than recorded history! Hopefully now everyone else will follow suit.

Electric Hubcap Motor System

A3938 V2 Motor Controller

   I decided I really needed to have a good, high current supply of about 18 volts - high enough to work, but not high enough that the MOSFETs' 20 volts maximum drain to source voltage could be exceeded to burn them out. I made a 5 volt, 30 AH NiMH battery of 12 D cells to add to a 12 volt one to obtain this. Nice to have the lab power supply to charge this odd voltage battery. Unfortunately the lab supply doesn't put out quite enough current to run the motors properly itself.

   I swapped in the last A3938 chip, having burned out all the rest, and tried a couple of times to get a motor running with it. Mostly no current was being drawn. Some odd things were happening and it didn't run. But (for once) I didn't get the impression there was anything wrong with the chip. There it sat for a while pending some inspiration.

   On the 29th, I had the thought that with the slow output slew rates, perhaps the current limiting was sampling too soon. I had had to increase the "dead time" to the max to prevent high 'shoot-through' switching currents. "Theoretically" the dead time should have been overkill. Perhaps also, although the theoretical blanking time capacitor was under 50pF and I was using 82pF, it might still be shutting the cycle off just as it started. 1000pF was the original value, burning out the first A3938 controllers when turned up too high. I put in 330pF as a compromise value.
   While I was at it switched the gate resistors back from 18 to 27 ohms, nervous about burning out the A3938's gate drivers again on my last A3938 chip. They seem to be a weak point. On the other hand, the first controller was working (until turned up too far) with 15 ohmers, and longer gate wires than on the new one.

   That didn't seem to do the trick. Probably I'll find changing some component value, somewhere,  will get it going. In the meantime, perhaps next I'll try it driving single MOSFETs instead of doubled up. (Single transistors should be fine as long as I don't put a big load on the motor.) If that works, I'll know the slower timings owing to the large gate charges are the source of the trouble, instead of just guessing that's what it is. If not, I should be looking elsewhere.

Motor Production

   My enthusiasm for setting up for motor production wained owing to the fact that no one has ordered a fabulous Electric Hubcap motor kit yet, because I got the flu, other projects (yardwork... batteries!), and because it seemed like about a month before some magnet rotors were finally cut by Victoria Waterjet. Of course I wasn't expecting many sales until they're pushing cars on the road.
   But on the 23rd to 27th I finally made some new mold pieces for the slightly smaller 11.25" diameter size shells, and made the 3 body ring pieces for another motor. This diameter was obtained by eliminating the short PVC pipe section (1/2" thick walls) from the outer wall of the rotor compartment and instead wrapping around several thicknesses of strong PP strapping-epoxy strapping composite to form a thinner outer wall.
   It's not the complete set of molds to allow me to cast all the parts for a motor in one session, but it's half way there and I can cast all the pieces, though one at a time. Between clear acrylic battery cases and the motor body molds, I seemed to have to cross town to the plastic supply shop for something almost every day in this period. I thought I already had everything, but somehow whenever I went to do something I seemed to be missing what was needed.
   My enthusiasm for the three mold system also wained somewhat as I discovered that only one would fit in the oven at a time, and realized that I'd need another 100 dollars worth of C-clamps to press three at once. I'd rather spend that on epoxy to make more motor parts, even though more slowly!

Routering the mold base. This is the second pass, deepening the dish

Pieces for mold(s) with 3" & 4" center posts and plywood backings, and one finished motor ring piece.

Stuffing the mold with PP fabric and epoxy resin.

   The first two ring pieces didn't come out very well and needed a lot of touching up. I got better by the third one, but I'd like to figure out a rigid outside wall setup - I end up having to push protruding pieces in with a chisel or screwdriver just before the C-clamps close the gap. Nobody makes pipe the right size.

Metal Protection

   Finally near the end of the month Victoria Waterjet had cut the magnet rotors and other parts I asked for some weeks previously. On the last day I received back a magnet rotor that I had powder coated to protect it from rusting. The polyester coating looked great, and evidently there should be no problem with gluing the magnets onto it with epoxy & PP strapping as before. I decided I should have the bearing holder parts coated as well. That would cover all the main metal parts except the axle, the SDS (magnet rotor) coupling, and the bearings, which would be hard to work with if they were coated.
   But the powder coating place had done the rotor as a free sample. On inquiry, it turned out it might be around 30 $ per magnet rotor, and 15 $ per bearing holder, which there are 4 of. Doubtless there's a lot of prep to turn them from oily and perhaps rusty steel to spotlessly clean and shiny for coating in addition to the primer and finish coating, but 90 $ 'extra' - on motor kits I hope to sell for 500 $ - isn't going to work. Even the 30 $ virtually doubles the cost of the magnet rotor. I wonder if there's any way to do it, or something similar, myself? Spray paint is probably the simple answer for the bearing holders and maybe the back side of the magnet rotor, but I wouldn't want flakey spray paint under the magnets. The powder coating is good. I think I'll do a little more reading about it...

Planetary Gear Project

   The results of the first experiment seemed rather disappointing. The motor and the 2.8 to 1 planetary gear turned fine, but it didn't budge the car, and though the motor power could have been doubled, it just didn't seem that could be enough to put the car on the street. The options appeared to be the same as ever: get a variable rate torque converter working, or the car would need two complete motor systems even to travel at city speeds -- if even that proved practical. (Bear in mind that most electric conversions use at least 15 KW or 20 KW motors rather than 5 KW - only with the exceptional performance of a mechanical torque converter have I been expecting to drive a car with such a small motor.)

Planetary gear spin test with the wheel jacked up. This was the only time it spun the wheel.
It would no doubt have at least moved the car on level pavement at full power,
but it surely wouldn't have gone up much of a hill of a hill or accelerated well enough for the street.
Movie: http://www.youtube.com/watch?v=Zihj3wbPe3A

   If a practical mechanical torque converter was already invented, the choice would be simple. But it was build the second motor system or invent the torque converter. Here I've been for two years, thinking the later would be the superior option, and quite possibly the faster as well. Well, it hasn't been the faster!

   But someone sent me a link to an idea for a rotary version of Constantinesco's original torque converter -- using planetary gears! This development seemed so promising, and the planetary gears used in the normal way so little promising, that I decided to try the planetary gear torque converter idea and set the 'regular' planetary gears system aside.

Torque Converter Project

  With the disappointing performance of the planetary gear, I started thinking again about the variable ratio mechanical torque converter (MTC), and at the same time, someone sent me a link to an MTC design concept someone thought of and posted to "halfbakery.com" a couple of years ago - just a couple of months before I started trying to make one myself. It was essentially a rotary version of Constantinesco's original design using oscillating planetary gears. The 'engine' oscillated the planet gear carrier back and forth, about a 90º cyclic rotation. The sun gear shaft had a flywheel, replacing Constantinesco's original pendulum as the oscillating mass. With my gear, that would have a speed increase of 2.8 to 1, magnifying the effect of the flywheel's mass by about 8 over a one to one speed arrangement. Various rotors could be tried to see what mass works best. On the outside, the ring gear was connected so that it would turn a one-way ratchet as it twisted back and forth, moving the car one direction.
   It looked very promising, if a bit complex to build. I have the gears, but still it would take a lot of changes, detailed design effort and work to implement.
   And moving the car "one way" would mean forward only. Was I to turn on the gas engine just to back up?

   Another feature of this converter idea was that it could be a double version, with two masses, gear sets, flywheels et al at 90º phase to each other, which would balance the load on the 'engine'. I think Constantinesco also used two 90º masses in his production model car engine. But probably for an electric motor, there's no need to have balanced drive at all points of rotation.

   But I started sketching and tried again to come up with something simpler. After a couple of weeks of thinking I was really onto something and being eager to start building, I realized neither of the two design variations I'd come up with would work - and that one of them I had thought of before and already gone through the same process.
   Later I thought of a couple of variations of my other new design (not oscillating masses) that just might work. But my confidence level wasn't high.

   At the same time, I was starting to think of ways to put together the planetary gear type. The planetary gears were in fact just one of the several components. It couldn't be a simple in-line form. It would be something like a giant size mechanical clock, with all the shafts borne on two parallel plates of steel and all the gears and levers between, with a few posts holding everything rigid. Those plates would have to be at least a couple of inches apart to hold everything, which would do nothing for thinness of the fit outside the car wheel. But... it ought to work! I decided I would simply have to do it - even if it couldn't back up. Perhaps some solution for that might develop as I worked.

Checking out planetary gear inertial effects: put weight on one element,
clamp another in a vise, twist the third back and forth with vise-grips...
and see how much the narrow workbench wobbles.

   The biggest sticking point to getting it to work at all is perhaps the "one-way valve" or ratchet that takes the oscillating motion and turns it back into rotary motion at the wheel. Some more recent MTC implementations have come to grief over this seemingly minor point: commercial one-way bearings evidently aren't made to handle rapid oscillations - they get hot and soon burn out. I'd really like to know what Constantinesco used in his 1926 cars! I do however have a couple of ideas for durable ratchets. One uses the outside of the planetary ring gear, which is the output piece and which has grooves... and both use the arms from my aforementioned sketches. I did up a drawing of a design that looked workable, if a bit complex.


   The same person who found the halfbakery.com idea also pointed me to the NuVinci bicycle hub, an entirely different MTC design using tilting rolling balls, and the various diameters attained at different rolling angles. It looked very complicated to do and needed special 'lubricant', which is the reason I'd passed the design over previously. But now a small version was actually in production, for bicycles. Should I have another look at that before I started building the oscillating masses type? A phrase from a blurb on the site spells out the advantages of mechanical torque converters that I've been mentioning:

"... it dramatically improves acceleration, hill climbing and top-end speed while extending range and battery life..." (www.fallbrooktech.com/09_LEV_Kit.asp)

   Those features are why I expect to make a mere 5 KW motor drive a whole car. But is there any reasonable way to make these things, big enough for a car, without a whole factory and without having the special traction lubricant? (or can the lubricant be obtained?) Somehow, I can't see trying to scale up their friction-drive unit from the weight of a person to the weight of a car. I think any version I could build is virtually bound to slip and go nowhere.

   But it shows there must be many ways to skin a seal (relative of cat)... The NuVinci uses "smooth gears" that can be varied. "CVT"s use a belt around conical spindles. The Constantinesco type stores up energy in an oscillating mass. Then I had a fresh idea:

   Couldn't energy be stored up by compressing or stretching a spring instead of in an oscillating mass? Or, couldn't energy stored in a rotating mass (rather than oscillating) be rapidly transferred by suddenly compressing or stretching a spring?
   For such an idea to work (and with all-live parts), this would mean the motor would turn one turn (or some integral division thereof - 1/2, 1/3... 1/16...), and then at a specific point(s) of rotation, a spring would suddenly transfer some of the momentum of the motor to the output, moving it however much or little that energy had in it to do. Naturally, the amount of energy would go up with the square of the motor speed, and below a certain point, it wouldn't overcome static friction.
   Getting the energy transfer at specific points is the thing. The fewer and stronger the hits, the more the chances of each one overcoming that static friction. My last "clock escapement" design might have worked if there'd been just 2, 4 or 5 hits per rotation instead of 25. (Unfortunately, with all the progress in motors, I finally scrapped the old early 2009 motor that had those fittings, without having tried this.)
   Some mechanism would have to disengage the contact to allow the motor to continue turning (though slightly slowed down) past the point of contact, and something would have to "trip" to disconnect the spring from the output, to prevent it shoving backwards as it decompressed.
   The sprung section on the end of the arm, or at the rim of the rotor, could hit square on and rebound, then twist sideways as it reversed, bypassing the contact point on the second pass. Once again, I'm trying to visualize mechanical things that should be possible, but which haven't been made before. I went off to the drawing board to try some sketches.

MTC Sketches of May. Top-left is one that wouldn't work from early in the month;
the rest develop the new sudden spring compression or release idea, with spring-loaded wedges.

   The first spring activated sketches (top right) seemed plausible, but a day or two later I realized it would kick backwards after the forwards kick, as the other slope of the wedge passed the contact point. The second used triangles that would flip just once to the next face at the contact points. Using curved faces I tried to get a design that would turn a little farther than flat to ensure it would flip to the next face and not back to the first one. Then I noticed that they would work the same whichever way they flipped! So why bother with turning triangles? Next was the bottom center sketch, where it just flicked it and it returned to the same position.
   Then I started thinking that all those ideas compressed the spring rapidly at the point of action, but it should be more energetic to have the spring suddenly release. In the top center sketch, I simply changed the "strike posts" to "slots" in the rim of the output drum.
   The wedge would be held pushed sideways over most of its travel, and then when the point reached the slot, it would suddenly snap down. If the slot was the right length, the wedge would hit the far side of the slot at high speed, like the sudden snap of a strong electrical switch or circuit breaker. Then, as the motor rotor moved on, the wedge would be twisted up again, compressing the spring, which would remain compressed until the next slot. Note that the compressing of the spring pushes the drum the same way as the release and immediately after it, though with less sudden impact. The parts will of course be greased, but any slight frictional drag of the wedge points along the rim will also act in the same direction. None of the force acts in the wrong direction and almost none is wasted.
   In the final sketch at the lower left, I varied the wedge shapes and spring attachments and shapes, trying to achieve best effect. After sketching #4 in that drawing, I went from paper to 1/2" plastic on the 31st:

"Mock up" of sprung wedge MTC.

   The number of wedges and slots will of course match so all wedges act at once. The diameters of the rotor and the drum rim are going to have to be pretty close together for the wedges to engage properly. My current idea to proceed with is to have two thin (eg, .1") 11.5" aluminum rotors on the motor axle (via an SDS bushing), with the pivoting wedges between them. The drum can be a (thicker) 12" aluminum plate on the car wheel, with the slotted rim screwed to it. (The prototype plate may be steel as I have one the right size - or I might re-use the 12" frying pan for the umpteenth time.) The axle of the motor should extend to the output drum, where a ball bearing race will hold it centered on the wheel center, though free to pivot a little to 'spring' the motor weight a little vis a vis the wheel. Or (probably preferable) the output drum may instead spin securely on the motor axle with two bearings, with a flexible link to the car wheel (similar to last month's wheel link for the planetary gear).
   The spring tension needed to twist the wedges must be substantially less than the torque the motor can provide to compress them. (If less torque is required than is needed to twist the wedges, the motor and wheel rotations will "lock" together.) I will doubtless do a lot of playing around with the materials, sizes, shapes, strengths, and numbers of the wedges, springs, and slots. But the basic idea - as far as I can tell at the moment - seems sound.

Turquoise Battery Project

Salt electrolyte battery research: stalled since 1880s!

   I think I've said that salt electrolyte seems to be about the least explored area of battery research, much less researched than acid and alkaline. I found an 1891 book (Primary Batteries - Henry Carhart.pdf) on primary batteries.
   It was a revelation to see the "carbon-zinc" (manganese-zinc) dry cell, constructed in its essentials the same as today, which (in combination with the fact that evidently no other type of salt solution electrolyte based cells are or have ever been available) means that no important improvements have been made to salt electrolyte battery cells in over 120 years! Even then "sal ammoniac" was apparently almost the only salt that was used.

   Small wonder then that an 'outsider' to the electrochemical field like me might today be able to achieve some fine results, when it would seem no one else has done any significant R or D on "mundane" salt batteries since Victorian times except to improve the packaging.

But no shortage of lithium battery research

   A Technology Review article (mobile.technologyreview.com/energy/37632/) on a new "pumped sludge" lithium battery project spells out the main problem with lithiums:

"A big problem with the lithium-ion batteries used in electric vehicles and plug-in hybrids is that only about 25 percent of the battery's volume is taken up by materials that store energy. The rest is made up of inactive materials, such as packaging, conductive foils, and glues, which make the batteries bulky and account for a significant part of the cost."

   This is the reason I've always felt that other chemistries can outperform lithium: other elements theoretically have less energy per weight, but they can be much less "diluted" in a cell than appears to be the case with lithium types. I don't think the "pumped sludge" approach sounds very practical for electric transport, if for anything... though you never know! The article itself ends on the same sort of note:

"It's a very clever device," says Dahn. "I don't know if it will ever be more than an idea in a paper, but Chiang has surprised people before."

   For a doubtful "idea in a paper", the project certainly doesn't suffer from lack of support, having been incorporated, raised about $16,000,000 and employed 20 people, according to the article. (I wonder how many of the 20 are actually involved in the development, and of those, how well their time and talents are being utilized?)
   The money is somewhere around $15,990,000 more than I've been able to put into this battery project so far. (Like I said in "month in brief", paying individual inventors - even if they get a real salary or equivalent plus expenses - should cost much less than funding "innovative companies" and is more likely to produce practical technical results. A separate problem is that even if Chiang's battery proves practical, it is likely to end up as a patent in some oil company's drawer (like Chevron's 125 nickel-metal hydride battery patents), rendered unusable by anyone until the whole technology has been long forgotten.)

Meanwhile, back at the ranch...

Real, Solid Electrodes

   In disassembling recent previous cells, I've been finding that the electrodes (particularly the negatrodes) have become quite hard, seemingly much harder than when installed. They are in line with what is found inside, eg, commercial Ni-MH or Ni-Cd dry cells. The "sandstone" texture I've been trying to achieve in the compactor is finally achieved in the cell itself. Evidently I've been doing better on this score than I thought.

Vanadium negative?: -1 volts, total 2 volts --- ?

   It seemed the main reason for the high self discharging was that the manganese reaction voltage
[Mn + 2 OH-  <=>  Mn(OH)2, e= -1.37 V(?) in neutral solution] was just a little too high, even with the egg albumin to raise the hydrogen generation voltage, so the discharge reaction happens fairly slowly but spontaneously and continuously until the negatrode is discharged. Iron also seemed to discharge spontaneously. This was unexpected since it works well in alkaline solution - Jungner and Edison manufactured nickel-iron batteries, and Ni-Fe dry cells have recently been created in India. It led me to assume the problem wasn't in the negative electrode, when in fact it was - with both substances.

   What then could be used? Nickel worked, as shown in my cell at the end of April, but it looked like it would be only 1/2 a volt. What was between -.5 and -1.37 volts?
   The next element down from manganese seemed to be zinc at about -1 volt, but the soluble zincate ion causes seemingly insoluble problems and short life. Chromium looked similar.
   Vanadium appeared to have an ideal reaction of about -1 volts with no soluble ions shown for alkaline solution. [V + 2 OH-  <=>  VO + H2O + 2e-, e= -.975 V(?) in neutral solution] Drawback seemed to be that like chromium its compounds are poisonous. But only to people -- to the environment it's "green". (Well, actually yellow.)

States and voltages of  Vanadium - shown as usual for acid and alkali but not in neutral salt solution -
suggesting it might be a good negative going between V and VO at maybe -1 volts.
Wikipedia info suggested it might dissolve into ions, notwithstanding the insolubility of VO.
In fact, the direct reaction right from valence 0 to 5 isn't seen in other elements.

   I bought a 1/4 pound bag of vanadium pentoxide at Victoria Clay Arts - 12.95$ ++. This price seemed similar to nickel oxide - not dirt cheap, but not really rare, either. (I'm guessing a 10 pound bag will have a better discount than nickel but I haven't tried pricing it yet.) It must be borne in mind that if double the energy density by weight is achieved, it effectively cuts the cost in half as well as the battery size and weight.

   I picked up their pound of neodydmium oxide while I was there, which had been sitting there for years. (As pottery glazes: intense yellow (V) and delicate yellow (Nd).) Neo is probably better than lanthanum for raising oxygen overvoltage, though not as good as samarium. He had finally put a price on it, and for 17.95 $ I could hardly resist - after all I paid over 200 $ for 2 Kg of lanthanum I bought early in the project and then had to turn into hydroxide myself. (it made about 7 pounds of it.) Samarium would doubtless be more.

   Next on the agenda seemed to be to make a new battery in a new case - hopefully that wouldn't leak. January's ABS case, reused a couple of times, was split wide open at a seam. I decided the case I'd made with a longer rod of carbon inside - and two holes - was much too likely to leak and abandoned it. I used clear acrylic plastic on all sides in case there was anything to see inside the cell at any time. Also it's harder and stiffer than ABS - perhaps that would help prevent leaks. The clear case proved absolutely valuable for seeing things happening inside and I won't make any other again, at least not for prototyping. In addition this stiffer material held pressure without leaking. I also made this one a little taller so there could be a pool of water over the electrodes for a sealed (or not) wet cell.

Clear battery cell case with carbon rod terminal posts,
carbon fiber mat to back the electrodes.

   For the collector sheets I cut 1.5" x 3" pieces of carbon (graphite) fiber sheets and boiled them in hexadecane (this time outdoors, as it reeks and I have no fume hood). It was still slightly damp with hexadecane when I put it into the compactor. The hexadecane seems to act as a "flux" that helps to join the electrode to the graphite for a better connection and I'll probably continue doing it this way. The fiber sheets didn't come loose from the electrodes. All or most of the expanded graphite sheets have done so.

Note: The bottle labelled "Diesel Kleen" that I've been calling "hexadecane" contains not only hexadecane but unspecified "petroleum distillates" (from MSDS... probably Ethylbenzene, Naphthalene, Xylene and 1,2,4-Trimethylbenzene) and "slick diesel"(?), in unspecified proportions. Whatever it exactly is, it seems to work great! (Likeliest best ingredient: methylbenzene?)

Pos (Neg):

- 15g of the monel/La(OH)3 mix
- 10g of MnO2/graphite mix (from a dry cell)
- 5g more graphite powder
- .25g Sb4O6
- 1.5g Sunlight
- bit of HOH

(No MnO2 2nd cell)

   This 'trode was compacted onto the treated carbon fibre, painted with lime, dried in the oven for 40 minutes at 250ºF, torched for 10 seconds, and then placed in the bottom of the new cell case.
   I was in a bit of a hurry and only dried it in the oven for 2/3 of an hour. In the torching, a chunk popped out. They get the full hour drying in the oven from now on!

Neg (Pos):

- 12g V2O5
- 9g graphite
- .15g Sb4O6
- 1g Sunlight
- bit of HOH

The "intense yellow glaze" vanadium pentoxide made a yellowish electrode
instead of the usual dark gray to black.

    Both 'trodes measured in the mid 10's of ohms when tamped down in the mortar, and one about the same after compaction. I neglected to measure the vanadium one. It should become more conductive as the vanadium charged from oxide to metal.
   For the separator I used only a sheet of cellophane, doped with osmium in acetaldehyde. If it's too thin and shorts out, I'll have to dig out the vanadium electrode and replace with cellophane and art paper, but thin should make for great ion flow - gotta get that internal resistance down.

   As an aside, there were separator sheet materials/textures I was thinking of long ago (and probably won't use now), that I couldn't think what it was I was thinking of. One of them I finally realized was coffee filter paper. The other was polypropylene non-woven fabric, which I didn't know is also called "landscaping fabric". However, the fabric is too porous and the electrodes would short. Coffee filters remain possible, but are probably not as good as the Arches watercolour paper... though it may depend on brand. So does the watercolour paper.

   Using clear plexiglass for the case was quickly vindicated. A visual inspection of the cell after a few hours revealed that some vanadium oxide had gone through to the lower electrode in one corner - leaked or fallen over the edge of the cellophane "tray". The rest of the negatrode had lost its yellow colour and turned black over a few hours, presumably as it charged to lower oxides or to vanadium metal, but distinct yellowish patches were there around the edge spreading from that corner. If this battery has any self discharge, fast or slow, there was the first item to suspect. Perhaps that was part of why it seemed to be charging so slowly, and why the voltage rapidly dropped off if the charge was removed.
   How many previous cells might I have seen and solved problems in if I'd made them all clear?

   It didn't seem to want to charge. After a while I took it apart and put in another sheet of cellophane, but that didn't help - the yellow edges spread.

   Perhaps there was something odd going on, with all vanadium's odd soluble reactions spoken of on Wikipedia? The redox chart did show a reaction straight from valence 0 to 5, which seemed unique. Perhaps it became some dissolved ion form not shown, went through the separator, and recharged at the positrode to valence 5? That might explain the migration seen at the edges.

   But if it wouldn't work as a negatrode, could vanadium make a good positrode? With all the strange "+" oxidizing reactions of vanadium in "strong acid" and "strong base" forming various soluble ions, it didn't seem very likely, but as usual there's no redox charts of vanadium in neutral salt solution. Vanadium pentoxide is "the most stable and common compound of vanadium", which "upon heating reversibly loses oxygen". I also gleaned from one item that the tetravalent state is, or at least sometimes is, V2O4 rather than VO2, and the trivalent state is V2O3. It is perhaps noteworthy that the cation in the main positive states of charge should be "V2", suggesting a good stability between charge and discharge.
   Conceivably in salt solution it might discharge and charge between V2O3 [III] and V2O4 [IV], moving one electron per reaction, or (better still) V2O3 and V2O5 [V], a valence change of two electrons? This last reaction, illustrating the "reversibly losing oxygen" property of V2O5, looked on the chart like it might occur at about the desired +1 volts, which would make vanadium an excellent positrode material. There are also various transition oxide forms between V2O3 and V2O5, such as V3O5 and V3O7 that vanadium can take... as well as various soluble ion forms. But it seemed more likely the V2 form would remain intact as two oxygen ions came and went, and solubles might well only occur in acidic or alkaline solutions - not in neutral salt and with a "+" charge.

   So I tried what I tried with the previous battery (and once in a long distance phone call): I reversed the charges. I now knew that the nickel works as a negative. This seemed to charge! After 12 hours, a bit of yellow colour seemed to return, so evidently it was charging back to valence five -- perhaps to V2O5 in accordance with my fondest hopes. And the cell voltage had risen to around 1.7 volts and was still rising.
   This was another surprise. It had seemed to me the most likely reason the previous cell was 1.5 volts was that the manganese was charging to potassium permanganate (+1v) and that the nickel hydroxide to metal would be -.5 volts, midway between the base (-.72v) and acid (-.28v) levels.
   But for the vanadium at about +1 to cause the cell to rise to a higher voltage meant that to explain the previous cell's 1.5 volts, the "-" side must have accounted for more of the voltage. The "+" then was actually charging to only MnO2 (+.5), and the nickel-manganese-lanthanum side in salt had to be about -1. There is, eg, a redox reaction for NiS <=> Ni that's -.99 volts, but it's a far cry from anything shown on the chart for oxides/hydroxides.

Nothing in the chart hints that Ni(OH)2 <=> Ni is apparently -1 volt in salt solution,
instead of being be the average of the acid and alkali levels.

   In vanadium, seemingly I've stumbled across an excellent battery positrode substance. How good? Wow!:

(a) The only reaction suggested by the voltage would be V2O5 <=> V2O3. Theoretically this reaction moves two electrons per vanadium atom. It looks to me likely it will actually approach this value in practice. Manganese dioxide is said to move one, and nickel (with the best mixes - which significantly dilute the actual nickel content) moves about 1.5, so 2 (and without additives) is a major gain.

(b) V2O3 (@4.87 g/cc) is a little more dense than nickel hydroxide (@4.10), and it is only slightly less dense than MnO2 (@5.03), but it has the lightest atomic weight, so it packs in more atoms -- more amp-hours -- both by weight and by volume. Even if NiOOH, MnO2 and V2O5 each moved just one electron per metal atom, the amp-hours of the vanadium both by weight and volume would still be highest.

NiOOH <=> Ni(OH)2 - 102<=>103 g/m, 1 e- (1.5 e- with best mixes charging some to NiO2)
MnO2 <=> MnOOH - 87<=>88 g/m, 1 e-
V2O5 <=> V2O3 - 91<=>75 g/m, 2 e- (grams/mole & electrons for each "V")

(c) With the dual vanadium center, it may prove to be the most stable and have the longest cycle life - if it has any cycle life limit at all.

(d) It's "the max" +1 volts instead of some lower voltage. (Though, in neutral solution, nickel should also be +1. V2O5 and NiOOH might even make a good mix, if there seems to be any reason not to use just V2O5.) The higher the voltage, the higher the energy density, and +1 and -1 seem to be about the limits in salt.

   Thus vanadium looks excellent for a "regular" type battery, ie one with solid electrodes, in salty electrolyte. (It's also used in the "vanadium redox battery" in liquid form.)

   I don't know what vanadium's problem is as a negative - perhaps with the right treatment that could work too, but I didn't seem to be getting good results. Or perhaps like nickel, and maybe iron, the hydroxide to metal voltage was higher in neutral salt solution than expected, which could make it too high. (Could this be the general case? It would explain why salt water is so corrosive, and why "low voltage" nickel is -1 volt.)

  If the nickel is about -1 volts, it's higher than in either acid (-.28) or alkali (-.72). But that would seem to be the best explanation for getting about 2 volts. It can't be the lanthanum, and copper has no reactions that look like any sort of match. I added no manganese to the nickel mix in the second cell, so it wasn't the manganese.

   What a round-about way to arrive at my "ideal" 2 volt salt solution battery cell - one reaction unexpectedly without soluble products in charge or discharge, and one reaction whose voltage in salt is quite unexpected, neither really hinted at or suggested in the charts and literature! But the result is the thing - a better battery chemistry! I'm going to make a preliminary estimate that it's good for 200-275 WH per Kg. Only the newest, experimental, lithium-sulfur cells (220-290) rival that, and it should be far more economical to produce than those.
   Also, batteries have specific causes for eventually wearing out or failing, and I haven't identified any so far except for potential eventual drying out of the electrolyte through poor seals. Since it's a flooded cell type, it can be refilled by drilling a tiny hole in the top, adding water, eg with a syringe, and putting in a plug. That means I'm guessing they should go until the plastic case finally cracks or the separator disintegrates. (I should probably put in a combo filler/relief cap that releases gas above a certain pressure.)

   After charging the cell, I let it sit and found it discharged overnight to about 1.1 volts. I charged it again and did a 50 ohm load test. The internal resistance was higher than previous recent cells and it didn't fare well.
   I took it apart and added sheets of expanded graphite, sanded to roughen and cleaned with hexadecane, and  hexadecane brushed onto the carbon fibre cloth, which did come apart from the electrode briquettes on disassembly. The brushing seemed to loosen the fibres - perhaps that will improve conductivity. The graphite parts were still wet with it when I put them back together with the 'trodes, which may also help.

But I didn't put the vanadium positrode back in. Instead I made a new one, adding some nickel hydroxide and neodymium oxide in hopes of reducing the self discharge.


- 9g V2O5
- 9g graphite
- .10g Sb4O6
- 4g Ni(OH)2
- 2g Nd2O3
- 1.5g Sunlight
- bit of HOH

   This upped the current capacity to about what I had with other cells, but it didn't seem to do a thing for the self discharge. Had I in vanadium oxide picked another element and reaction that was just a bit too high in voltage? Considering how close it must be, with the potential not only of almost double the voltage of manganese dioxide, but also of moving two electrons per metal atom instead of one, that would be frustrating.
   Days later I realized I'd forgotten the calcium layer for overvoltage improvement, as well as the idea of adding cobalt oxide to increase conductivity. I should really get into a better habit when doing electrodes - post procedure notes on the wall and check them as I proceed!

   A load test before and after seemed to show that the voltage stayed higher longer for half an hour or so, but then dropped off to even lower levels thereafter.
   I took it apart and put in a calcium layer and torched it (in fact I did two, unthinkingly placing one on the nickel electrode instead of the vanadium one... well, that was the original "+" side, on the bottom of the cell where I put the "+"es... and I had the flu.), but by this time the briquettes were crumbling and there was little hope of better current capacity or performance. I also replaced the cellophane separator sheet (ripped) and added some more osmium powder to the acetaldehyde mix before coating it.
   The reassembled cell seemed to have higher self discharge and took considerably longer to charge than before. A positive note to that is that the self discharge, being higher, seemed to be something I'd done rather than to be an intrinsic property of the materials.

   The chemistry seems good, and (for once!) the cell holds pressure: the conductance simply has to be improved. That may well solve the self discharge and the low utilization of the electrode substances. It's got under 20mA/sq.cm. even through 1 ohm, which is nearly a short circuit with such low values. Ten times that figure would be a great improvement. I could add bits of carbon (graphite) fiber to the electrode as well as the powder, but I hate working with it - it's worse than fibreglass and I end up itchy from the fine fibers every time I use it. I can also add cobalt oxide, but that's supposed to form a conductive network "in solid solution" with nickel hydroxide, in alkaline solution. There's no guarantee it'll help simply mixed with vanadium. There may be many possibilities for improving salt electrolyte cell characteristics, but salt solution batteries and ideas for them are hardly in the literature.

   On the 13th I chopped up some carbon fiber mat into small individual strands, maybe .05" to .25" long, and immersed it in hexadecane. I realized I could probably use hexadecane to wet down the electrode mix as easily as water. It would evaporate, and it seemed it might improve contacts - and also keep the lightweight fibers from becoming airborne while working.

   I suspected that nickel hydroxide might either complement the vanadium oxide or raise its oxygen overvoltage. So Vanadium Electrode Mix #3 was:

10g (66%) V2O5 (active)
5g (33%) Ni(OH)2 (overvoltage/active)
.15g (1%) Sb4O6* (O2 H2 => H2O recombinant catalyst)
5g graphite powder (conducts)
"a 3g wad of" carbon fiber (conducts) [much of the mass of the "wad" was hexadecane.]
1g Sunlight (glue)
?? hexadecane (flux)

*I want to try Sb2S3 (Sb4S6?), stibnite - might work better - but I haven't found a cheap and convenient source for it yet. It's sold by pyrotechnics supplys, but I couldn't find a Canadian source, and the shipping from the USA was much more than the price of the merchandise. I could make Sb2(SO4)3 by boiling the Sb2O3 in sulfuric acid - perhaps that might reduce to Sb2S3 in the negative electrode? Yuk!

   This electrode was compacted on a sheet of expanded graphite. Taking off the graphite sheet revealed an amazing working of the Diesel Kleen solvent: what would normally have a fine "sandstone" texture instead had a smooth metallic sheen and very low electrical resistance, adjacent points reading as low as 4 ohms. This appears to be a fabulous technique for improving conductivity! I suspect it's forming random lamellar structures of graphite (graphene?) and electrode active substance(s) as it evaporates, a random conductive network through the electrode. (Perhaps I can dispense with the nasty chopped carbon fibers?)

Backside of vanadium electrode briquette, with a metallic sheen, and fuzzy with conductive carbon fibers (left),
with its expanded graphite backing sheet (right), illustrates the effect of the solvent/hexadecane.
A more usual electrode surface texture is seen in the thin area along the top, which protruded past the backing sheet.
The rest has a metallic sheen to it and quite low electrical resistance.
(Note: The backing sheet is somewhat lumpy and the worse for wear after 3 compactions. The first 2 compactions failed because the briquette was too damp with hexadecane and oozed out of the form. "Diesel Kleen" seems to evaporate considerably slower than water.)

Top face of the hairy electrode. A few surface spots had a silver shine to them.
With the addition of nickel hydroxide to the vanadium pentoxide, electrode color went from ecru to camo.

   However, with only 5 grams of graphite powder in the mix, the resistance from the top face to the bottom seemed to be in the mid tens (~50) of ohms. I should have used more powder. Next time I will -- and that'll be soon if the cell doesn't perform well.

So Vanadium Positrode Mix #4 (barring further adjustments) will be:

10g (66%) V2O5 (active)
5g (33%) Ni(OH)2 (overvoltage/active)
.15g (1%) Sb4O6* (O2 H2 => H2O recombinant catalyst)
8g graphite powder (conducts)
"a 3g wad of" carbon fiber (conducts) [much of the mass of the "wad" was hexadecane.]
1g Sunlight (glue)
?? Diesel Kleen (solvent/flux)

   Next the layer of calcium hydroxide was painted on, then it was dried in the oven for about 90 minutes at 225ºF, and then torched for 7 or 8 seconds or so with a swirljet (very hot) propane torch, moving the flame around the surface to hit it all.

For the negative, I used the Ni-Mn-La mixture at the top of this article except I used 7.5g graphite, .15g Sb4O6, and a ~2.4g wad of carbon fiber, and wetted it with hexadecane instead of water. Only about 3/4 of the mix seemed to fit into the compactor, so some was left over.
   This time I was afraid to add too much hexadecane, and it came out of the compactor rather dry. The metallic sheen covered only a portion of the back surface. Unsatisfied, I dabbed on some more hexadecane and re-compacted it. This time, the briquette had to be pried away from the graphite sheet, and the whole surface was silvery. (A couple of silvery briquette surface spots broke away and stayed adhered to the graphite sheet - they weren't 'missed'.)

Left: First try - too dry. Not enough hexadecane, and silvery surface coverage was incomplete.
Right: Added a few drops to the bad areas and re-compacted. This completed the coverage.

The front surface, also looking a bit silvery. Resistances were quite low,
around 10 ohms from surface points to the graphite backing sheet.

   After compaction, resistances from any surface point to the backing sheet seemed to be about 10 ohms. In these silvery surfaces I felt, for the first time since being forced to adopt carbon instead of metal backings and conductivity additives, that high current densities might be within reach. ...as long as the electrodes haven't become more or less impervious to the electrolyte (unlikely), and as long as those nasty carbon fibers don't poke their way through the separator sheets and short out the cell! I suspect the fibers are now superfluous, though.

After a day of charging the cell, it was up to almost 2.2 volts, and it would source around 3/4 of an amp into a one ohm load, only a small improvement over previous cells. That would be maybe 30mA/sq.cm of electrode interface. An order of magnitude more would be better. It wasn't as good as a new regular manganese-zinc D cell (1.275 A), but it was headed the right direction. The D cell would be about 50 sq.cm instead of 29 though, so 26 mA/sq.cm. (I found it was capable of 40 mA/sq.cm while delivering above one volt. What a far cry such numbers are from NiMH D cells, delivering 70 amps at almost a volt to start the car!)
   I was hoping for better. Doubtless it would have done better with a little more graphite in the positive to reduce those 50 ohm readings, and probably if I hadn't split the carbon sheets away from the electrodes to peek in between. In a future cell I can also try using metal (expanded copper mesh?) in the negatrode structure to provide lower resistance in that side.
   After a second day's charging, green material had formed around the edges on the negatrode and started creeping its way up the folded separator sheet to the positive. It could be nickel hydroxide or copper chloride - they're virtually identical blueish green. On the other hand, nickel hydroxide shouldn't be moving, and copper chloride, even if it formed, should have converted to hydroxide (or dissolved). Furthermore, both Ni and Cu should be reducing to metal with charge, not oxidizing. Whatever it is, I don't like the way it's trying to do an end run around the separator!
   By the end of the day, black lines were forming at the tops of the green. Best guess: it was Ni(OH)2 and it had now touched the positrode, and was charging up to black NiOOH on top. A bridge around the edge of the separator sheet -- it's a sure setup for rapid self discharge! I let the cell sit overnight and charged it during the day, and sure enough, on the second night, the voltage in the morning was lower, 1.3xx versus 1.6xx volts.

   I disassembled the cell. The positrode came out without damage, and I wrapped it almost all up in new paper except where it had to contact the terminal post. Not only was this better insulation with less chance of a bridge, it turned out to be easier to fold than forming a dish around the outside. It did use a 3x4" piece of paper for a 1.5x3" area, but paper is neither pricey nor heavy. I added a couple more sheets of rubbery stuff, hoping a stronger press would get everything connecting together better inside.
   The old paper was green not only around the edges; much of it was impregnated with green, in many areas black on the positive side surface. If this was copper chloride, it could possibly be an advantage, just as zinc chloride proved to be a better electrolyte in standard dry cells, and it then became a specific additive to improve performance. If it was nickel, either hydroxide or chloride, it was probably bad news. Nothing for it but to put it back together and see what happened. And maybe figure out some chemical test for copper or nickel on the old paper.

   How could I have made all those cells without clear sides to see in!?!

   I put a pressure gauge on this cell, but I unscrewed it and let the pressure out whenever it hit about 3 PSI. I think a flooded type cell with an air reservoir on top and a pressure release valve will be optimum, especially for DIY construction. Somebody else can try completely sealed cells (dry or wet) later. This will also be about the safest battery going: it can't pressurize and explode, and salty water isn't corrosive to human flesh if it gets out. Short of actually ingesting it - or perhaps leaving it short circuited until things start to melt - it's unlikely to injure anyone.
   On the 19th I finally bought an adapter fitting for the adjustable pressure relief valve I had bought along with the pressure meter, well over a year ago now, and I installed the valve. Now it could do its own pressure reliefs when and as needed, at maybe 5 or 10 PSI.

V-Ni cell in clear case with pressure meter and pressure release valve.
Greenish layer is nickel negatrode, white is vanadium pentoxide positrode wrapped in watercolor paper.
Brown is spongy rubbery stuff to push the two together.
5/16" carbon posts top and bottom (from standard "D" cells) connect the electrodes to the outside world.

   On the 20th the voltage had dropped overnight again to 1.64 volts. I thought of opening the cell and adding eggwhite/albumin to the negative. But it already had the thiamin - surely a good amine itself. When connected, the voltage rose to 2.1 in a couple of minutes, more like a commercial battery. It seemed the 'intermediary' additives had been charged out and all that was left active was the intended chemicals. Something seemed to be changing, and one must expect very gradual results when charging a cell that's several amp-hours at 25mA. I decided to give it another day of charging and see what happened on the fourth night. It could take the rest of the month to really see results, but I'd rather take it slowly. If it appeared to be working I might make my 'full size' 3" x 6" cell.
   The next morning the voltage had dropped even more, to 1.54 volts. All that green stuff around the edges... it came to my mind that I hadn't added the zircon ion shield layer, at least not to the second separator sheet. In the afternoon I opened it and painted some zircon on. (I'd better not forget the zircon if I try the manganese-nickel mix for permanganate again!) I also added yet another sheet of rubbery mat stuff to push the electrodes together more.
   It didn't help. Conductivity seemed to be gradually dropping. It would only source 1/2 an amp into 1 ohm, and the following morning the voltage was 1.46. I decided to charge it backwards and see what happened, this time not to attempt to use it that way, but only to change or move chemicals that might have charged up on the wrong sides or in a wrong way. I charged it much of the day that way, then reversed it again.
   There was improvement - after 8 hours sitting the cell was still 1.65 volts, and 1.5 volts after 18 hours. At least it was back to how it was. I let it rest the whole of the 23rd.

   On the 23rd also, I was given a PDF copy of Battery Reference Book 3rd Edition by T.R. Crompton - a 774 page tome. Imagine working on making batteries for 3-1/2 years, doing web searches and looking at library catalogs, and not discovering that such a comprehensive reference work exists! Could it have saved me countless hours of work? Or would I have been "indoctrinated" with "standard" assumptions and not tried some of the things that are now starting to yield great results? Starting with a "clean slate" (also known as complete ignorance) and perhaps discovering some thing or things overlooked by previous battery research was after all one of my ideas.
   However, as I perused some of the pages, I found practical design and construction seemed to be about the least covered aspect. In particular, I could find no mention of why only carbon and graphite were used for common dry cells - that every metal placed in the positive electrode would corrode away. Essentially it didn't seem to contain any revelations that would have really speeded up my work.

   On the 24th, it finally dawned on me that it might simply take a few charge, rest, and discharge cycles to get everything into proper balance. I'd done lots of charging but until now little discharging or resting on this cell. Having rest it over 24 hours (it was still about 1.4 volts), I shorted it out for a few hours until the current dropped to almost nothing, then let it rest some more, and charged it in the evening. On the 25th it still wouldn't source more than 450 mA into a one ohm load, and a load test didn't seem very good, notwithstanding that it wasn't very fully charged.

The Conductivity Problem: Epoxy-graphite experiment

   The chemistry (and probably more than one chemistry) seems great, the electrode conductivity is surely much improved by compacting with Diesel Kleen. The external metal clamps seem to contact the carbon posts quite well. Where then is the performance problem? No amount of charging and discharging seems to improve conductivity, nor restore it from its slow decline in each cell. There seemed to be just one likely culprit: the contact between the graphite backing sheet and the carbon terminal post. These two pressure-contact-only joins, where the graphite sheet would gradually saturate, surely deform and get looser under pressure, were now the one unvarying "worst feature" of all the cells I've been making, which are all two or three internal ohms to start and which all only put out 1/2 an amp or so, and that get very gradually worse.
   I visited graphitestore.com on the web and started wandering around the pages. After looking at a few things of marginal interest, I found some "conductive epoxy" glues with metal particles in them - silver, nickel, or aluminum. But if the electrolyte soaks through the graphite sheet and comes into contact with a metal at the positive electrode, it'll oxidize into a low conductivity oxide. If the metal particles contacted each other - as they must to conduct - they would all eventually oxidize.
   Okay then, how about epoxy with carbon or graphite particles? What about simply mixing graphite powder with epoxy? Mixing graphite into pitch gave poor results, but I hadn't tried epoxy.

   As a preliminary experiment, I mixed a little epoxy, 3.65 g, and then stirred in a gram of graphite. It turned black, but the consistency was little changed. I added a second gram, and it became a little thicker. With a third gram it became a thin paste, and with a fourth, a thick paste. With a fifth gram it all stuck to the stir stick and came out with it. It also lost its shiny appearance - it was getting dry with graphite.
   Between each graphite adding, I placed a drop on a small sheet of expanded graphite. With the three grams and four grams pastes, I also stuck two little stub carbon electrodes onto the sheet. I "caulked" some excess around their bases for more contact area. For the five grams, I wasn't sure the epoxy would "wet" the carbon and the graphite and get a good grip... and I was out of stub electrodes and the little sheet was full.
   From a consistency perspective, the 4 grams of graphite in 3.65 grams of epoxy - about 1 to 1 or a bit more of the powder - seemed about right for gluing carbon terminals onto graphite sheets. But how was it for conductivity?

   Before the epoxy set, I measured some resistances. Unfortunately, a lot depends on contact pressure, readings vary, and such low resistances are virtually beneath my meter's scale. But here are typical readings for what they're worth:
.3 Ω - touch cleaned meter leeds together ("0 Ω" calibration)
.4 Ω - any(?) two points on graphite sheet
.5 Ω - top of "4g" electrode to graphite sheet
.7 Ω - top of "3g" electrode to graphite sheet

   After the epoxy set, readings were taken from each droplet and from the two carbon posts, to the graphite sheet:
.2 Ω - cleaned meter leeds (or .3 Ω)
.4 Ω - two points on sheet (or two points on the same carbon post, either post)
∞ - 1g droplet
∞ - 2g
100,000 Ω - 3g
1000 Ω - 4g
150 Ω - 5g
.7 Ω - 3g post
.5 Ω - 4g post

Epoxy mixed with graphite in varying proportions, glued to an expanded graphite sheet.
A little over 1 part graphite powder per 1 part epoxy resin by weight seemed like good glue.
A little more graphite (~4:3) and it all stuck to the stir stick (right), but seemed to adhere well
(top left corner of sheet, behind rear "4 grams" terminal post) and had the lowest contact resistance.

I tried to pry the 5g droplet off to see how well it had adhered. It broke off, but it seemed to have taken a surface layer of the graphite sheet with it. The same thing happened with the 4g droplet and the 2g droplet when I pried on them. Evidently the glue at any concentration sticks well and is stronger than the flakey graphite sheet it adheres to. Owing to the fragility of the sheet one evidently must be careful not to break the post off it during battery assembly.

    Conclusions: It works! A terminal post made out of graphited epoxy would have high resistance, but as a conductive glue to hold the carbon post to the sheet it seems to work quite well. Certainly the more graphite powder in the epoxy the better the conductivity, and there's a threshold at about 1 to 1 below which conductivity is very poor. Somewhere between 4 and 5 grams of graphite (per 3.65 grams of epoxy) may be an optimum ratio, eg, 5 to 4. With too much graphite the mix probably won't adhere well and at some point will become unusable, or porous. On the other hand, the more the graphite, the lower the resistance.

   I rolled the leftover 5g mix into a thin sheet, thinking it may make a good electrolyte-impervious conductor for separating bipolar cells to make multi-cell, higher voltage batteries in one case. It became quite shiny again. The contact resistance was high, in the kilohms (in the thick areas) and up to megohms and no reading at all (in the thinner). But perhaps I might coat the entire outer face of each graphite sheet to strengthen it and to help prevent the post from getting broken off easily. ...or for interior bipolar electrodes, with no post.

   Unfortunately, gluing a post to the sheet can't be done before compacting the electrode, unless the sheet isn't to be compacted with it. Gluing it on afterwards is risky as the electrodes are so fragile. Would compacting the electrode without the sheet diminish the "Diesel Keen" electrode conductivity process? It certainly wouldn't do it any good! One possibility would be to put a hole for the terminal post in the compactor plates so sheets with a post could be inserted, but this would have a several problems.
   For the lower electrode, a good technique might be to put the post into its hole in the case (hole sized for a loose fit, with an inner countersink to fill with extra conducting epoxy), spread the glue inside the bottom of the case, and then put in the electrode, thus gluing it to the case and to the post, and sealing the post hole with epoxy.
   For the upper electrode, on balance I think compacting first and then gluing the post and painting on the epoxy should be tried first. One idea could be to put the sheet on top during compaction instead of on the bottom. Then it wouldn't be necessary to flip the electrode before gluing and painting. Potentially, one could paint the edges as well as the top, and this would help hold the electrode together for later handling.

   Of course, in the negatrode, copper or other metallic structures might work instead of carbon. If so, the whole conductivity/corrosion problem could be virtually eliminated for that electrode. If this were the bottom electrode, the leed might even be bent around somehow to come out the top, eliminating having one sticking out the bottom and most of the potential for direct liquid leakage.

   The experiment indicates that the measured electrode to post resistances will probably be lower initially. And they're much less likely to degrade than the simple press-contact connections. I decided gluing the posts to the graphite sheets - and painting the outer electrode surfaces - was definitely worth a go on the next cell, which would also have a nickel-permanganate positive. I had made another case and was about set up to start as the month drew to a close.

The Three Types of Rocks: - sentimental - metaphoric - ingenious
The Two Types of Trees: - carnivorous - delicious

Victoria BC