TheEEStory.com

Eenigma wrote:

PNeilson 10 wrote:

So we are getting a semi consistent picture.

Breakdown is the limit on CCTO type material.

Some methods of limiting breakdown have lead to lowering k

So reducing oxygen vacancies lowers leakage but reduces k

So wrap the material in alumina and keep the oxygen vacancies stable and k hi with low breakdown.

I wonder what happens then? Guess I have to ask DW at EEStor.

Excellent post PN. I would take it on step further and say we are getting a consistent picture across the board.

If Northrop adds an alumina shell to this material it's off to the races. I'm sure the designers of this dielectric would also like to have molecular tweezers and align each and every particle for optimum ED results. Perhaps Northup could borrow DW's alumina coating technique as well as the molecular "poling" tweezers.

The other consistent picture we are seeing here is Tom will trash any patent that isn't consistent with his limited understanding of dielectrics.

Well Een, my understanding may be limited, but I will debate it with anyone here and either defend it, or be educated and make it better!

I think you are putting it too strongly. The Northrup patent is about a process for depositing high quality dielectric, and mathching lattice spacing with substrate. It is (I guess) worth something. Obviously the more useful CCTO and related crstals are, the more it is worth. And as is proper it will claim everything under the sun - no-one ever got into trouble ading claims to a patent.

I also think you are over-reading its significance. Does it imply they have somehow got CCTO with high k and 1000X the breakdown of all other high-k CCTO? Seems unlikely to me, and certainly not necessary given the patent.

Since I have a clear understanding of the physical mechanisms and difficulties here, which you and most others here don't share, we may disagree about this. But you would be over-egging the pudding vastly to say that this patent constitutes strong evidence (let alone proof) that CCTO can have ultra-high ED.

Eenigma wrote:

ee-tom wrote:

I would say (nastily) that DW, having engineered a high breakdown material, is using it in this amplified k context without realising that this will effect breakdown. that will not be accepted by those here who see the patent measurements as real :)

Wow, sure EEStor doesn't understand breakdown related to K. EEstor doesn't understand saturation. EEStor doesn't know how to use voltmeter. I'm fully convinced because you say so Tom.

ee-tom wrote:

LOL - the high leakage in CCTOs comes from the free carriers which provide the high k. So you can get much lower leakage by avoiding oxygen vacancies (like DW) - but that also reduces k to intrinsic value of <100.

Tom it's interesting to see you accept EEStor's use of a shell is responsible for lower leakage measurements. Seems to me low leakage was also in the patent you claim is fraudulent. Of course you seem rather upset with Northrup Gruman's claims as well.

I think for EEstor Al2O3 is about chemical stabilisation (those oxygen vacancies). Certainly that is what DW says.

BaTiO3 undoped is a decent insulator, unlike CCTO, so I don't see need for reducing leakage. DW says 350V/u, even 500V/u OK.

However in the CCTO paper the insulating layer had other effects - it limited the space charge etc. Those could be relevant for EEStor too, but only if their BaTiO3 operates, like CCTO, as an IBLC.

Whichever way, in EEstor's case if the Al2O3 limits leakage that must be becasue it reduces field in the grains. But that would mean the Al2O3 breaks (it won't, according to DW, withstand more than 4X the operating voltage). At this field, it reduces grain field by approx 4*10/320 = 10%. Any more, it will break. Is that helpful?

But lower leakage at low fields - yes - it will do that!

Best wishes, Tom

When a hung around the accelerator lab the physicists were
still writing their code in Fortran...

The Forth micro-controllers I was using to update some of the
control systems were a real curiosity to them.

tom

I am still on the first page of Okazaki. However, he has a remarkable experimental claim on top of the second page in the table.

After charging for 3,000 seconds at 20 KV/cm field (2 V/um) he has a charge of 2685 uC/cm^2 (26.85 C/m^2)

This is a battery!

Please check my reading of the table. I am not sure where the Q is!

ee-tom wrote:

But you would be over-egging the pudding vastly to say that this patent constitutes strong evidence (let alone proof) that CCTO can have ultra-high ED.

To the contrary, I believe Northrup needs to wrap the CCTO in alumina to achieve low leakage high ED results that do not rapidly degrade. For ultra high ED they need maximum contribution from each crystal (poling).

It seems you agree about the low leakage aspect but are unwilling to acknowledge the high ED aspect.

On page 2

Okazaki says higher voltage charges faster lower voltage discharges faster

So you control the flow of current over time by varying the voltage. So you want current control on you charger/discharge.

Very dynamic with nothing like a capacitor response.
Its AC coupled alright.

He explicitly say it charges and discharges like a battery!

PNeilson 10 wrote:

tom

I am still on the first page of Okazaki. However, he has a remarkable experimental claim on top of the second page in the table.

After charging for 3,000 seconds at 20 KV/cm field (2 V/um) he has a charge of 2685 uC/cm^2 (26.85 C/m^2)

This is a battery!

Please check my reading of the table. I am not sure where the Q is!

Yes. He (says) he calculates Q on the basis of the current discharged when the high poled sample is short circuited. I don't trust his figures because he says nothing about what raw data he used to get them. And I think he has made a mistake. The only direct followup paper (Chinese) does not claim high polarization. If nek can read japanese he might be able to decipher the 1958 paper referenced that Okazaki claims also shows high polarization.

The other perculiarity is that the model he uses for the charge decay leads to infinite total charge, for all of his values! So I just wonder how accurate is his numerical integration to get total charge and what data he uses to do this. Clearly his model is suspect! the followup paper goes I(from somone else) goes into this and points out that charge decays more quickly than okazaki's model, hence is not infinite.

Do you see why I do not trust this result - quite apart from the fact that no-one has been able to replicate it. Others, if no okazaki, must realise that such a high polarisation would be astonishing and valuable.

Maybe the paper missed off u from C in translation. :)

PNeilson 10 wrote:

After charging for 3,000 seconds at 20 KV/cm field (2 V/um) he has a charge of 2685 uC/cm^2 (26.85 C/m^2)

This is a battery!

DAP is right

k is just a number and the number means next to nothing to Okazaki

Cross is right

DW's explanation is outrageous. It is wrong. Wrong enough that you could believe that he cast the ED as if it was a cap on purpose, to correctly express ED and send everyone off track on figuring it out. And yes you can get DW's exact numbers in the patent. You just have to take your data at the correct time. Just don't tell any one that time is important.

Tom is right

It is not ionic polarization.

DW is right

If you know how it works you know why it is safe. The charge - discharge curves are wacky with plenty of time to deal with any problems. I accept the challenge to drive a nail through the EESU. Nothing will happen.

PNeilson is right

The ED is real. With 5,000 J/cc experimental confirmation from Okazaki almost 20 years ago.

Alumina is the key. Yes it keeps in the goodness - oxygen and charge.

Po is wrong.

No capacitor can have ED above 20 J/CC. WRONG WRONG WRONG.

PNeilson 10 wrote:

And yes you can get DW's exact numbers in the patent. You just have to take your data at the correct time. Just don't tell any one that time is important.

OK - PN - I see that I missed a dp, this read 26C/m^2!!!

battery indeed.

Okazaki wrote:

However, the total charge to be stored in BT above
T, during the charging and discharging process is
possibly more than 1000 UC/cm2

Funny translation, or doubt?

Let us work out from his figures:
I1 = 720E-10
alpha = 0.526.

So his model (which you remember gives an upper bound, because the current falls off more rapidly for large t), integrated, gives:
Qtot = (720E-10/0.474)t^0.474

so for t=10000 we get:
(720E-10/0.474)*100 = 1.5E-5 = 10uC/cm^2

I chose 10,000 because it is 3X longer than the poling time, and of course we know that this model overestimates for large t. But even with t=1E6 we get only a factor of 10 extra, or 100uC/m^2.

26 times lower than the number given here of 2685uC/cm^2

A bit fishy don't you think? Did Okazaki measure over a 1 year period? (His model would give an extra 3X, still 10X too low)? Is his own model underestimating the real measured charge by this larger figure? What?

My bet - he got it wrong!

Please check my working because it is easy to make mistakes...

PNeilson 10 wrote:

DAP is right

k is just a number and the number means next to nothing to Okazaki

Cross is right

DW's explanation is outrageous. It is wrong. Wrong enough that you could believe that he cast the ED as if it was a cap on purpose, to correctly express ED and send everyone off track on figuring it out. And yes you can get DW's exact numbers in the patent. You just have to take your data at the correct time. Just don't tell any one that time is important.

Tom is right

It is not ionic polarization.

DW is right

If you know how it works you know why it is safe. The charge - discharge curves are wacky with plenty of time to deal with any problems. I accept the challenge to drive a nail through the EESU. Nothing will happen.

PNeilson is right

The ED is real. With 5,000 J/cc experimental confirmation from Okazaki almost 20 years ago.

Alumina is the key. Yes it keeps in the goodness - oxygen and charge.

Po is wrong.

No capacitor can have ED above 20 J/CC. WRONG WRONG WRONG.

I'll just keep this as a record for later... :)

PN. This one okazaki paper is obviously important to you. Please check my calculations for what the charge should be, given the one minute current (I1) and the value of alpha. His quoted charge looks thousands of times higher than it should be.

The more who check this the better. You need:

I1 = 700E-10 A/cm^2
alpha = 0.526

Q = 2700uC/cm^2 (total charge out)

Id = I1*t^-alpha (discharge current)

Thus Q is integral of Id from t=0 onwards.

And note that followup paper claims current decreases faster than this model at large t.

I've assumed t in seconds (he does not say) but perhaps it is measured in minutes. That would make more sense, and make things a bit less bad.

As a check, if the current never decreased from I1 it would take 4E5 seconds to give the stated total charge. That is 100 hours. But of course according to his model the current decreases by factor of 2 after approx 4 time periods (minutes?). So this is not possible, it could only be over a much longer period.

I don't believe he measured over such a period. I think he measured over a short period to get his model. Than made some arbitrary assumption about long-term decay to get the charge. And this massively overestimated charge - his model gives infinite total charge which is clearly unphysical.

Last edited Tue, 31 Aug 2010, 1:16pm by ee-tom

Sorry all,

PN - you have the paper. It is entirely unclear but I think maybe Id and td are the logs of the relevant qtys, which makes the relationship:

log(I) = log(I1)*log(t)^-alpha

What do you think? Maybe this will give better charge figures?

If you persist in treating this paper as evidence I will investigate further. Or you can?

Before I do this - what does the equation mean? Is it log, or not? is it t measured in minutes (as implied by I1 = current at 1 minute) or is it t measured in some other unit?

Tom

Modesty - Japanese cultural trait.

I am replying too soon because I have not got far enough but I can't follow your math. According to Okazaki Q is a function of 6 independent variables. C, V, Esubp, Tsubp, tsubp, Tsubd. Lot of t's there. I only count 4 values in your equation.

I don't see Esubp anywhere?
I will try to replicate.

PNeilson 10 wrote:

Tom

Modesty - Japanese cultural trait.

I am replying too soon because I have not got far enough but I can't follow your math. According to Okazaki Q is a function of 6 independent variables. C, V, Esubp, Tsubp, tsubp, Tsubd. Lot of t's there. I only count 4 values in your equation.

I will try to replicate.

The only variables in the results measured are t and Id, the model assumes Id has the given form, constants I1 and alpha are given in the table as best match figures in this model.

I am just checking consistency of Q as stated here against his own model. But it is not clear since his quantities are not clearly defined (no units for t and maybe the model is a log/log relationship - which would make sense).

tom - it is not clear to me either.

Translating Japanese technical papers to English is very hard.

Ask nekote for an explanation of why!

The table data is clear. The model is not. This may take some concentrated time and I am in and out too fast for that. Maybe we will have some better answers tonight.

tom

Charging time is no issue for Okazaki . Just increase the voltage to reduce charging time. Increasing the breakdown voltage allows faster charging. His experimental data is at 2 V/um. 350 V/um will charge far far faster.

The only issue is the charging limit.

PNeilson 10 wrote:

tom

Charging time is no issue for Okazaki . Just increase the voltage to reduce charging time. Increasing the breakdown voltage allows faster charging. His experimental data is at 2 V/um. 350 V/um will charge far far faster.

The only issue is the charging limit.

PN - my issue is that his stated charge value is way inconsistent with his stated I1 & alpha value.

This relationship:

log(Id) = log(t)^-alpha

Is not possible since at t->0 it blows up, and t-> infinity Id is still large.

So we must have Id = I1t^-alpha

and in fact t must be measured in minutes, to make I1 equal to Id when t=1.

So, reworking, the model predicts total charge after time t2:

Q = int(Id) = (I1/(1-alpha))t2^1-alpha

Plugging in the numbers for the weird 27C/m^2 = 2680uC/cm^2 figure:

I1= 0.072uA/cm^2, alpha = 0.526

Take a time of 30,000s = 500 minutes (10X longer than charge time)

we have Q = 0.072*19 = 1.4uC/cm^2

this is 2000X smaller than the stated charge!!!!

Suppose we assume his relationship holds for 100X longer, that increases the charge by a factor of 9, still 100X too small.

The followup paper pointed out that charge is not infinite because the okazaki model fails to predict Id sharply decreasing at large t.

So we have the polarization figures stated in the paper are a factor of 2000 too high [b]given his own measurements & model

Something a bit fishy here! I smell a BigMig type windup.

EDIT - MISTAKE - units are minutes not seconds so X60
Only 30X too high.

Last edited Tue, 31 Aug 2010, 6:37pm by ee-tom

Tom - you went one step to far.

Neither you nor I can really read this paper.

So yes something fishy is going on. Is it our ability to read Japlish, a bad translation, or the model does not match the table and title and the model is right?

Need more scholarship before passing judgement. I will compare the Japanese version with the English version to see if that turns up any discrepancies.

If we end up that the title matches the experimental table does not match the model then we may never really know what Mr Okazaki means or if he is just wrong.

As usual you or I can make conjectures. But right now this is pretty ambiguous.

PNeilson 10 wrote:

Tom - you went one step to far.

Neither you nor I can really read this paper.

So yes something fishy is going on. Is it our ability to read Japlish, a bad translation, or the model does not match the table and title and the model is right?

Need more scholarship before passing judgement. I will compare the Japanese version with the English version to see if that turns up any discrepancies.

If we end up that the title matches the experimental table does not match the model then we may never really know what Mr Okazaki means or if he is just wrong.

As usual you or I can make conjectures. But right now this is pretty ambiguous.

PN - I can read the maths, and the values of I1,alpha,Q.

There is on closer reading no room for ambiguity in the equation:
I1 must be value for t=1, which means t must be measured in minutes.

The equation only works if Id,t are normal (not log) qtys.

You can check my maths. The Q value is 1000 X too high. I don't know how it was calculated, but it is totally inconsistent with the I1 current measurement which you must admit is closer to the measurements.

So this paper is not your smoking gun.

Sorry, Tom

Its the I1 that has me still puzzled - alpha and q no problems

You say "I1 must be" - and "t must be" well I am not so sure yet.

The log seems just for plotting purposes.

But is this a political process - do the polls close at 8 PM? Or do I get some time?

I will ask you why the table has different I1 values for the 4 values given at various times with the same charging voltage?

The table in the Japanese version has different data than the English paper in the temperature column

Translation difficulties are possible.

PNeilson 10 wrote:

Its the I1 that has me still puzzled - alpha and q no problems

You say "I1 must be" - and "t must be" well I am not so sure yet.

The log seems just for plotting purposes.

But is this a political process - do the polls close at 8 PM? Or do I get some time?

I am not political, you have all the time you want. I am just saying how i see it.

I1 is defined to be current at t=1 minute

From the equation therefore, t must be measured in minutes since from equation when t=1 Id = I1.

As you say logs do not make sense so that is just bad wording.

Tom

PNeilson 10 wrote:

The table in the Japanese version has different data than the English paper in the temperature column

Translation difficulties are possible.

Yes, maybe he meant nC/cm^2 instead of uC/cm^2?

Come on tom we know snarky-ness is required in English Academic circles. But really, here in the American West that kind of stuff can get you shot, quick! I know a few bars in Denver where I would be afraid to take you.

I have only had time time to calculate a bit so far and figure out better what is going on. Its very interesting how dependent this device is on the polling temperature, voltage and time. It is also striking how little data there is across the polling space.

Qd vs polling time for the set given is linear.

No response needed.

PNeilson 10 wrote:

Come on tom we know snarky-ness is required in English Academic circles. But really, here in the American West that kind of stuff can get you shot, quick! I know a few bars in Denver where I would be afraid to take you.

I have only had time time to calculate a bit so far and figure out better what is going on. Its very interesting how dependent this device is on the polling temperature, voltage and time. It is also striking how little data there is across the polling space.

Qd vs polling time for the set given is linear.

No response needed.

PN - I don't know what this is about. But I have not looked at depenence of Qd on poling time. It is not relevant to my calculations, which are a calculation of Qd(total) from Id.

Tom

Thanks for the second paper. It kept me from going way out on a limb. I would have put in some caveats but I probably would have run down the limb. Not in my nature to be careful.

Now we are talking scientific possibility not economic feasibility or practicality. So it would be reasonable to use infinite charging time for a theoretical max ED.

But since other researchers (Peng) have shown Okazaki's charging function to decay faster than he proposed or had any comprehensive data for that would be unreasonable.

So I could try to figure out Peng's function precisely or we could just eyeball it and say Okazaki's function has a limit at 10,000 min. So I took the easy way out. Using toms function (which seems correct) Okazaki at 10,000 min or 7 days gets a charge of 7.2 C /m*2. Okazaki himself says differently and gives a single data point as an experimental data point that takes 100 days of charging. So lets not debate it. Lets just use 7.2 C / m^2. This is a very exceptional number even with 7 days charge time.

When I use 500 min in your function with my conversion to Q I get a pretty different number than you do. I get 1.7 C / m^2. I notice you take Q = 0.072*19 = 1.4uC/cm^2. The 0.072 is just the Id (uA / cm^2) for 1 min not 500. And Wolfram tells me a uA - min / cm^2 is 0.6 C/m^2.

Do we agree or not? Lets decide this before we go further. My arithmetic needs checking as always.

Ah tom, see I had to read the paper many times to get this. One reason I am so slow.

The data table shows the effect of Polling Time and Voltage. You get better electrets as a function of Polling Temperature, Time and Voltage. The correct temperature for the material, the longer the time and the higher the voltage the better the electrect. Of course because this is unwrapped IBLC dielectric Okazaki only polled at 2 V/um max.

DW is significantly different in that he polls at a higher temp, 2,000 times higher voltage and shorter time.

So polling is doing much more than just aligning the Crystallite!!!!

The material with the best Q and Id and alpha was polled at the highest voltage