Starting in mid-June, Richard Arundal and I have completed 45 experiments, charging up capacitors 4x4 cm in area with either kapton or Polyethylene dielectrics between the plates. We have been carefully eliminating pesky artifacts, through trial and error, along with the appropriate amount of engineering, er, 'jargon', and taking readings with increasing accuracy. Here I summarise the cleanest data we have so far. The plot below shows the change in thrust up the y axis (a positive value is always a thrust towards the capacitor's anode) for eight runs. As we increase the voltage and move along the x axis, each point represents the weight change we saw (in grams) when we stepped the Voltage up by 100V. QI predicts that when we reach somewhere around the breakdown voltage for the capacitor, which is very uncertain but may be around 1 kV for the thin kapton and 1.5 kV for the thicker polyethylene, there should be a thrust towards the anode, ie: a shift up on the graph.
F = 0.00014IA/d^2
where I is the leakage current between the plates, A is the plate area and d is the plate separation. If we assume that the leakage current = 10^-9 x voltage (which fits what we have seen with our measurements of voltage & leakage current) then
F = 0.00014 x 10^-9 x VA/d^2
dF/dV = 0.00014 x 10^-9 x A/d^2
To convert force in Newtons to grams we multiply by 100 and then by another 100 to get from g/V to g/100V. For the 4x4, 30 micron dielectric then we have 0.002 grams/100V (100V was our increment). This fits the x27 and x28 data (blue squares) as the peak at 1.6-1.9 kV reaches just above 0.002 g/100V. The peak thrust from these two capacitors is at a likely breakdown point.
The other runs show the 7 micron kapton dielectric runs (blue-red circles). Those that do show a positive peak: x38, x43 and maybe x45 show it at 1-1.3 kV which fits because these capacitors are thinner and should break down at lower voltage. From the equation above, the kapton should produce 0.046 grams/100V and that is close to what we saw. The faint box represents the area I expect for the QI thrust. It is uncertain in the x direction as the quoted breakdown voltages of dielectrics have a large range.
One problem remaining is that we do not yet have a perfect example of turning the capacitor over and getting a reverse thrust. We can maybe see that in x37, x45 and especially x48 which were reversed but the peaks are small and these runs were not very clean (x45 was drifting well before 0.1kV was reached). x48 gave us a sustained peak at the right time, but there was a brief glitch while powering up.
In summary, we have a few positive peaks which are encouraging. So is the fact that the kapton runs show both a lower break down voltage and a higher thrust than the polyethylene. Are these peaks due to QI? If so, the world has changed, but time will tell. We need a good clean reversal of force. Richard Arundal has been excellent - he is a skilled engineer who does not give up. Please go and see his youtube channel linked below. The last six weeks have been a roller coaster, but with a slow geological scale uplift.
McCulloch, M.E., 2021. Thrust from symmetric capacitors using quantised inertia: https://www.researchgate.net/publication/353481953_Thrust_from_Symmetric_Capacitors_using_Quantised_Inertia
Arundal, R., 2022. https://www.youtube.com/channel/UCO00nPk5WkpV0ggLGLnHHiQ
Mike - congratulations on getting results! Actually, the world changed a while back, but few people actually noticed. The main difference with QI is that you've been publicising both the theory and experiments and thus allowing many people to profit if they want to. Sadly, kudos does not translate immediately into income stream, but that may yet happen.
I'd nitpick the relationship between voltage and thrust here, though. The leakage current is not linear with voltage, with a somewhat exponential dependency. To me, too, it appears that the way to get a higher thrust per watt is to use a thinner dielectric and lower voltage, and that's probably best achieved using a chip-fab and deposition of thin films. The dielectric needs to be extremely uniform without pinholes and where the insulating layer through-current doesn't get high enough to damage that layer - problems that have been solved in the semiconductor industry in the production of Flash memory. Maybe a good design here would be a large number of fairly small area capacitors, each driven by its own programmable current source. Going from 10 micron dielectric to 10nm instead reduces the voltage needed by 1000 and also increases the thrust per watt by 1000.
I think the power supply for this use (testing a 4cm square of hand-built capacitor) might need to be specifically designed to have an effectively very low output capacitance and a very fast current-limit as well. This would reduce the spark-through problems. Working just on the edge of enough voltage to get enough leakage, but not enough to cause a catastrophic spark, is a fine line to tread, and putting some intelligence in the power supply could help.
Exciting to see the progress and measured results.
Simon I think your proposal and comments are spot on, however maybe we should be looking into structures that promote electron emission on a very small scale. Mike has mentioned that a 'rough' surface seems to be required to promote the phenomenon, and this type of surface obviously creates localised high electric field strengths.
Are you suggesting if the dielectric can be made thin enough and uniform enough, this type of structure will not be required?
The other comment I think is worth focusing on is a high performance current control mechanism built into the power supply. It would need to be a very fast acting (low inductance) source, I am not sure of what timescales required here but I would assume sub microsecond?
I intend to try these experiments with 6um kapton film and will report back on my findings.
Still waiting on my precision DMM.....
cstrudwicke - I'm trying anodised Titanium to produce the dielectric. Nice thing there is that the colour changes tell you the thickness produced. Anodised Aluminium doesn't have the same advantage. Rather than use acid for the anodising, which can give pinholes and where the depth depends on time as well as voltage, better to use a Borax solution where the depth of the oxide layer depends only on the voltage.
If the dielectric layer is much thinner, then the surface irregularities needed to get local field concentration will also be smaller, so buffing the surface before anodising using a Garryflex block (or fine abrasive) should provide very fine scratches that may be sufficient. Maybe deposit a Silver layer on top using the normal way to get mirrors, or use some Silver or Gold leaf to get the counter-electrode and compress it in place using an elastomer (rubber or Silicone rubber). Figuring that out at the moment. Leaf metal is easier, but electroless deposition of Silver should stick and give no entrapped air. A side-point here is that the work-function of Gold or Silver is higher than that of Titanium (sucks electrons better) so that adds a bit of extra acceleration over having the same metal both sides of the dielectric. Thus might work better with Gold leaf if we set the Titanium as the -ve electrode (cathode) and the Gold leaf as the +ve electrode (anode).
Once I've sorted things out here, I'll put up a recipe for producing the capacitor and measurements of the leakage voltage/current, as well as any thrust measurements. Note the calculation above of dropping dielectric thickness by 1000 and multiplying the thrust by 1000 is wrong - it's dependent on 1/d² so it's a million times more for the same area and leakage current, and of course if the voltage is 1/1000 too that's another 3 orders of magnitude on the thrust per power ratio. Thus though I won't be doing 10nm dielectric here (too easy to damage that without a clean room), even a sub-micron dielectric should show quite an improvement.
For current control, initial tests should work with a series resistance that's large enough and a higher applied voltage. Wasteful, but gets the job done effectively and cheaply, leaving the more-complex design job for later.
If you really want a plastic dielectric, using a kitchen blender-base as a spinner could give a very thin layer using drops of diluted plastic in solution.
Regarding a cheap thin film for testing, mylar space blankets are on the order of 12 um, however if you do not remove the outer metallic coating, you will get arcing all around the edges of the prototype, the metal layers are just to close. A quick acid wash, removes the layers and allows for usable prototypes to be made. From this low cost thin film.
may be relevant (capacitor construction)
Report on results so far:
I used a 3cm by 4cm piece of grade 2 titanium 0.5mm thick, with a bit of extra material making a connection wire. This was burnished using a brown Garriflex block with water until clean and even - not polished but smooth matte surface. This was anodised at 20V in a Borax solution of about 10g in 200cc of DI water, raising the voltage slowly so as to keep the anodising current below around 20mA. Might be better to use a resistance in series to limit current, though. According to the tables I've found this results in an oxide layer around 50nm thick, and the colour is a purplish blue. Testing this using the 'scope in curve-tracer mode showed that this layer of oxide was insulating to around 10V or so before breaking down. I cast some Silicone sealant on a flat (and lightly-greased) plate to get a layer around 7mm thick, and cut this into 3cm by 2.5cm sections, which I then coated with Gold leaf. A bit of Kapton around one end of the Titanium to insulate it, with fine wire wrapped around this, gave a way of connection to the Gold-leaf covered block of Silicone, then taped together using a 3x2.5cm Acrylic pressure-plate on top to ensure the Gold leaf was pressed to the Titanium surface. Using a 6M8 resistor in series to limit current, looked like the leakage current was pretty-well the resistor with the Titanium oxide providing no blocking. Measuring the resistance of the capacitor showed only around 1kohm, so likely Titanium oxide is not a good dielectric for this use. Looking for any weight change using a scale precise to 10mg showed no change up to 40 microamps in either direction or orientation.
Basically, this experiment didn't fly. For interest, calculated thrust at 1 microamp for a 3x2.5cm area of 50nm dielectric would be 42N, so I wouldn't have needed a weight scale if it had worked. With the total weight of around 18g, it would have flown.
Other things to try: that was a Titanium alloy, and I also have some anode-grade foil that was sold as 99.9% pure and would thus produce a higher-resistance oxide layer. Could be the surface was too smooth and I need to use a coarser grit to polish it. I also have some Tantalum foil, and since Tantalum capacitors have low leakage, that may be a better material anyway. Currently I don't know the relationship between anodising current and dielectric thickness for Tantalum or Niobium, though it's likely to be of the same order. My Niobium isn't sheets, but 6x6x1mm squares that can, with effort, be beaten to around 15mm square. Might also need to think about heat-treating the oxide layers to make them more even and compact the surface. For the anodising, I had left the voltage switched on until I no longer saw any new gas produced on anode or cathode, which took several hours, so that any pinholes in the oxide layer were minimised anyway.
Though standardly anodising is done using Sulphuric acid, that tends to make the oxide thickness dependent on time as well as voltage, because the acid etches the oxide layer slightly and thus produces pinholes. Thus better to use the Borax solution, though maybe something else might be better still. I'm not certain about the Gold leaf - does this actually block enough Unruh waves? It's only around 0.5 microns thick, so maybe not enough, and the 0.5mm Titanium may also not block enough. Thus maybe add in a backing Copper plate for the foil. If the the Gold leaf is the cathode (electron emitter) then its thickness may not be critical, but then the work-function here works against the acceleration from the applied electric field. Thus may change that Gold leaf for Tin, Silver, Aluminium, or Copper leaf, depending on what metal and oxide is being used for the opposite plate and dielectric.
Ran over the character limit....
There's also a possibility of using Nickel, with the Nickel Oxide layer produced by heating in Oxygen (or air) and then changing to Argon to anneal it when the thickness is correct. Needs an annealing oven I haven't got, though may be able to build one.
Could be this works best with plastic dielectrics, and whether the electrons are tunnelling or in normal free flight. However, it's a lot easier to get very thin layers using oxidation, so see where this goes.
For titanium, maybe experiment nanotubes anodization?
For ex: https://www.researchgate.net/figure/Top-view-SEM-images-of-TiO2-nanotube-arrays-anodized-at-60V-for-1hour-in-ethylene-glycol_fig3_327140704
keywords: SEM TiO2 nanotubes
Nanotubes as electron tunnels?
I just discovered this blog from the Tim Ventura video.
I've noticed that a device that produces decent thrust would need to have the plate separation at most 10 microns with a 5kV differential between then. This would require a dielectric material with a dielectric strength of at least 500kV/mm to prevent arcing. CVD diamond is the only material I know of that is above this limit. CVD diamond is hard to make and is especially hard to make in large areas (its mostly made with microwave CVD process).
Are there other dielectrics with similar strength that are more easy and cheaper to make? It would look like you have a materials develop process on your hands to make this thing viable. I know Woodward and Fearn have been looking at advanced dielectrics for their Mach effect thruster work. I remember reading about some advanced dielectric and electrostrictive polymer materials being developed some years ago. I think these materials are called colossal dielectric and electrostrictive materials.
I think the colossal dielectrics would be needed to make the capacitors for this concept.
Update so far - it was an error to try anodised Titanium to make the dielectric. Turns out that TiO2 is actually a semiconductor. Thus the electric field ends up far too low to be useful. However, Tantalum should be a good oxide and the breakdown voltage is around 588MV/m. This is a lot better than Polyethylene (around 20-30MV/m according to data I've found so far) or Kapton at around 200-300MV/m. In fact the electric field at breakdown data is all over the place so I'm not that certain of a lot of it. However, since Tantalum is used in electrolytic capacitors that have low-leakage, the data there has a good chance of being accurate enough. Probably the best data comes from chip manufacturers who rely on it to manufacture working devices.
Silicon has a native oxide layer of around 1nm, which will withstand around 1V/nm (1000MV/m), and this can be made thicker by heating above 800°C in Oxygen or air. May also be able to anodise here, and produce a reliable dielectric layer with a defined breakdown voltage.
Diamond has a breakdown voltage of around 2000MV/m.
The counter-electrode needs some thought. I'm figuring on using a conductive oxide powder. Here I have a choice of Cerium Oxide, Tin Oxide, Zinc Oxide, or Manganese Dioxide. This has two effects. Firstly, the powder form gives a local field enhancement at the micron scale, similar to scoring the electrodes. Secondly, it provides protection for excessive local current in the event of dielectric breakdown at a point - the capacitor area over which current can be collected to feed an arc is reduced.
First test using Tantalum anodised at 20V and Cerium Oxide did not produce any obvious thrust but maybe the bulk resistance of the 1m thick layer of CeO2 was too high. Will re-try using either Tin Oxide or Manganese Dioxide. Tests using Silicon will follow.
Other notes - obviously much safer using voltages in the 50V range or less. Question is whether there's something else happening once we get into the multiple kV range that doesn't happen in the 10V range.
Anodising needs also a bit of care. Standard instructions state to use Sulphuric acid at around 0.1M strength. With Aluminium, the depth of oxide produced this way depends on voltage and time, since pits are produced. If instead you use a dilute solution of Borax to anodise, the depth depends on voltage alone (obviously on time too initially, but reaches a limit based on voltage), and the oxide layer should be pit-free if the current is not too high.
What I'm using is around 5g of Borax in 200cc of DI water, and raising the voltage slowly to keep the current below 10mA. Probably easier to use a large resistor to current-limit next time, and just leave it for several hours to reach near-zero current drawn.
Though data I've found for Aluminium Oxide gives breakdown voltage as 13.4-40MV/m, chip fab data says it reaches 400-900MV/m. Same disparity on looking at SiO2 in the tables of data versus the chip fab data. Possibly we'd need to use very pure Aluminium to get the best results, as you find on the +ve electrode of an electrolytic capacitor. However, note that that is already anodised with deep pits so would need polishing first. May be able to reach a dielectric strength of 1000MV/m with practice.
For anodised materials, that electrode needs to be the +ve one. Thus the conductive oxide powder becomes the -ve electrode and emits from the points of the powder, and powder size is in the 1 micron range.
Possibly the equation doesn't apply at the 1nm range. Thus may need a series of increasing dielectric thicknesses to see at what point we get thrust. For Tantalum, the Ta2O5 thickness grows at 1.7nm per volt, and for Si it's around 1nm per volt.
Post a Comment