TheEEStory.com

CapacitorMan / CapConcern said BT is not usually "poled" (polarized) in capacitor usage.

Got me to thinking.

Heretofor, has the use of unpoled BT only permitted a tiny fraction of the crystals to have been statistically oriented in such a way to participate in electric fields / capacitors?

Maybe as poor an effective participation as 0.25%?
1/400'th or so?

A statistical jumble of 3 Dimensional orientations.
Versus all those lovely dipole moments all perfectly lined up.

Naw, can't be.
Can it?

nekote wrote:

Heretofor, has the use of unpoled BT only permitted a tiny fraction of the crystals to have been statistically oriented in such a way to participate in electric fields / capacitors?

Maybe as poor an effective participation as 0.25%?
1/400'th or so?

Last edited Sat, 13 Sep 2008, 4:58pm by Y_Po

I could try it in the lab, the problem is none of our high K dielectrics will take that high a field without breaking down. We routinely check our competitors as well, and have yet to see any breakdown on high K ceramics over 100v/micron. 47 is the average. To get to 500 would make headlines.

CapacitorMan, I don't know what is worth trying, in a lab or otherwise. Just offering a shot in the dark conjecture.

If I had the means and ability, I'd try what I had, at hand, within whatever limitations exist.

If only to see if, maybe, there is some difference - maybe only as much as Y_Po suggests - towards 15%.

The 500V aspect seems to be wrapped up in that insulating alumina shell that encases / protects the CMBT.

I'm traveling this week, but let me see what one of the techs can do...There is a phenomena with high K BT called ageing, which is about 5% per decade. If one subjects a part to over 150°C, it will jump up in cap, and then gradually lose cap at that rate.

So, if one has a 100uF cap right after the temp exposure, five hours later it will be 95uF, and 50 hours later 92.5 and 5000 hours later 81.45, etc. So, one can be fooled by just taking a capacitor thats been laying around for a year, and heating it...bang, a 20% increase. As I recall, hitting with high voltage will bring back some of that as well.

From http://www.theeestory.com/files/MLC_Capacitor_Primer.pdf

...The rate at which aging may occur can be influenced by “voltage conditioning” of capacitors. It is found that units stressed by a dc voltage at elevated temperature (below the Curie Point) will experience a loss of capacitance, but with a consequently lower aging rate. It is theorized that the voltage stress at the elevated temperature accelerates the domain relaxation process. This voltage conditioning effect is, of course, eliminated if the unit ever experiences temperatures exceeding the Curie Point.
Capacitor manufacturers compensate for capacitance loss of ferroelectric dielectrics by adjustment of the testing limits, such that units do not age out of tolerance over a long time period. For example, for a dielectric with a 1.5%/decade aging rate, the testing limits are raised 3%, i.e. two decades of time. Units tested 100 hours after last exposure to the Curie Temperature therefore will remain within tolerance for another two decades or 10,000 hours...

Y_Po -

I would have thought max polarisation ~ cos(theta) where theta = angle of dipole to plate perpendicular. In which case random alignment would be something like:

int^pi/2_0 [2pi.sin(theta).cos(theta)]/(2pi) = pi/8 of theoretical max with all dipoles aligned.

But practically things may be different, and also perhaps I have got this wrong.

Even in this optimistic theoretical scenario, the advantage is at most 2.5X. No way 400X.

Best wishes, Tom

ee-tom

nekote here - it was my topic query, not Y_Po's :)

BTW - again, thanks for your scholarly efforts.
still a tad bit over my head - but still trying!

It kinda' makes me smile.
More like rolling a 6 sided die, for which a (±x, ±y, ±z) vector would come up.
Then toss in the sin or cos, depending on which angle perspective (measured from parallel or perpendicular) is selected.

Last edited Fri, 21 Nov 2008, 10:06pm by nekote

I don't understand what you are talking about.

My understanding is that when you "pole" the barium titanate you set the domains in the fully rotated position. Therefore domain rotation is not available for energy storage. Now just compare the remnant polarisation with the value of polarisation at saturation. Even with "unpoled" barium titanate there is very little energy storage available.

Isn't that correct?

Consider that with a 10 micron dielectric polarisation saturation occurs at .26 C/m^2 with a voltage of 14 and the Eestor requirements are 60C/m^2 with a voltage of 3500.

Isn't it obvious that energy storage is not the ferroelectric effect?

PeterP

If you are saying it is obvious that eestor's claimed energy density requires some completely new physics - I would agree with you. No-one has come up with a machanism, and many people have come up with contradictions.

But the possible polarisation of a BaTiO3 lattice is greater in some directions than others, so aligning crystals can help increase overall plarisation as I have said. The point is that random orientation 9worst case) gives you quite a lot of polarisation so my figure is a best case estimate for the increase - and not significant whe compared with 400X.

The underlying question was how great a multiplier in the (net effective) number of dipoles could 100% "poled" BT provide, over statistically random orientations. And if that was a core aspect of the "secret sauce".

Whatever the exact number is, it is apparent to me, now, that it is a comparatively small multiplier.
1.5X, 2.5X, 2.8X, 5X, 6X
Nothing over 10X.

Which explains, for me, why the existing capacitor industry doesn't bother with a poling process.

I assume, to much time and expense, where simply using more (reletively inexpensive) BT accomplished the same thing - more dipoles available.

For the case of the EESU, however, that small multiplier is worth a lot, in the reduction of the quantity / mass / volume of (relatively very expensive) ultra-pure alumina coated CMBT required. If, hypothetically, if the multiplier were, say, 2X, *not* poling would mean needing twice as much of the CMBT, taking up twice the volume.

A 500 or 600 pound EESU that has a volume more like 40 gallons would be a lot less attractive, than ~300 pounds @ ~20 gallons volume.

So, my conclusion is that the EESU poling step is more about optimization / minimizing materials - thus minimizing cost, weight and volume. It would also further assist in increasing production yield by not including stuff with some potential defects occuring in (CMBT) material that is just "along for the ride", without out contributing ~100% to the energy storage task.

Rather than being directly about the scientific "secret sauce" that makes 3500V (or 5000V) @ ~10 µm possible.

ee-tom,

I don’t seem to have got my message across. Perhaps there is a flaw in my thinking. Let me state what I think then you could perhaps point out where I’ve gone wrong.

Barium Titanate consists of domains. Each domain has a lattice structure and all the cells of the lattice in a domain are in alignment. Each cell has an intrinsic polarisation of .26 C/m^2 (= Ps). If the domains were randomly orientated the average polarisation would be close to zero. If the domains were all aligned then the average polarisation would be Ps. The ratio of the two is a large number but this is not important as the maximum value must be limited to Ps.

Now let’s say we have gone through the poling process and the capacitor is back at room temperature. We remove the poling voltage and connect a load. The capacitor discharges into the load and the voltage will end up at zero volts. The polarisation will drop from Psat to Pr (remnant polarisation). The energy dissipated in the load is proportional to the difference (Psat-Pr).

But poling means that this difference (Psat-Pr) is even smaller than it would be in an unpoled capacitor.

Poling decreases the energy storage capacity of a Barium Titanate capacitor.

Therefore the poling must be done for another reason. I suggest one reason that the poling is done is to align all the domains. But this could be done using any voltage above 15 volts (Esat = 1.4 volts/micron, 10 microns). However 4000 volts are used. There is likely to be a reason. And that reason is likely to be important.

The heterogenous eestor dielectric consists of 73% Barium Titanate, 22% PET and 5% alumina (if spherical particles are assumed). Eestor claim that breakdown occurs above 5000 volts. Is there a post which explains this? Does anybody have a link that might help?

Regards,

PeterP

Last edited Sun, 23 Nov 2008, 12:02am by PeterP

nek: Yes, that did occur to me - The rationale for poling in the WIPO document being that the majority of domains are oriented in one direction, then used with the opposite polarity during operation.

But there's a problem with that. The total energy storage can't come from common intrinsic properties like dipoles. If different energy storage mechanisms are used, say dipole rotation and mobile charge carriers, it's very unlikely that the derivation of the "permittivity" curve from the charging/discharging measurments will be as extraordinarily flat as reported in the WIPO document.

PeterP wrote:

ee-tom,

I don’t seem to have got my message across. Perhaps there is a flaw in my thinking. Let me state what I think then you could perhaps point out where I’ve gone wrong.

Barium Titanate consists of domains. Each domain has a lattice structure and all the cells of the lattice in a domain are in alignment. Each cell has an intrinsic polarisation of .26 C/m^2 (= Ps). If the domains were randomly orientated the average polarisation would be close to zero. If the domains were all aligned then the average polarisation would be Ps. The ratio of the two is a large number but this is not important as the maximum value must be limited to Ps.

Now let’s say we have gone through the poling process and the capacitor is back at room temperature. We remove the poling voltage and connect a load. The capacitor discharges into the load and the voltage will end up at zero volts. The polarisation will drop from Psat to Pr (remnant polarisation). The energy dissipated in the load is proportional to the difference (Psat-Pr).

But poling means that this difference (Psat-Pr) is even smaller than it would be in an unpoled capacitor.

Poling decreases the energy storage capacity of a Barium Titanate capacitor.

Therefore the poling must be done for another reason. I suggest one reason that the poling is done is to align all the domains. But this could be done using any voltage above 15 volts (Esat = 1.4 volts/micron, 10 microns). However 4000 volts are used. There is likely to be a reason. And that reason is likely to be important.

The heterogenous eestor dielectric consists of 73% Barium Titanate, 22% PET and 5% alumina (if spherical particles are assumed). Eestor claim a breakdown occurs above 5000 volts. Is there a post which explains this? Does anybody have a link that might help?

Regards,

PeterP

I can't say I completely understand domain rotation. Some references I've seen describe it (IIRC) as the actual bulk substance of the domain rotating - ie domain walls "sliding" past each other. I don't see how that is possible. I mean we are talking about a solid here, regardless of whether it's polycrystaline or not. So yeah, I don't completely get it. I also don't get the distinction between a grain boundary and a domain boundary, since they are apparently (or can be) different, appearing at different scales.

To my mind, unless there is an inclusion of some other material, and that material delineates and defines a "grain" boundary, then fine. But if not, then what's the distinction?

So to my limited understanding at least, "domains" can rotate under the proper conditions (heat and E-field?), to the point where the spontaneous dipole moments (in a polar material such as BaTiO3) in each domain line up.

Upon releasing of the poling conditions (cooling & removal of E-field), the entire bulk of the dielectric will now manifest a larger dipole moment. This represents some energy taken from the E-field during poling.

Upon use, in reverse polarity, you now have available a maximized number of molecular dipoles that can be rotated 180 degrees, presumably providing more energy storage.

Again, I'm not sure about this. I posted about all these problems of concept, and apparent inconsistencies of definition a while back, maintaining that too much discussion was basically "talking past each other" due to differences in individual conceptualization, but got no reply.

Anyway, is this how you picture it?

Oh, one other thing - the WIPO document specifies these volume relationships:

CMBT: 94 %
PET: 4 %
aluminum electrodes: 2 %

... although as I said before I can't see this being possible unless the particles are nearly cubic (unlikely, as Christine has detailed), or there is a more gaussian distribution of (mostly) round shaped particle sizes, which is inconsistent with EEStor's PR purity data (shrug).

I agree about the poling - it's obviously important or they wouldn't have included it. I can't think of any reason that is consistent with other data though. On the other hand, I haven't been able to figure out the exact mechanism of energy storage, so one might inform the other, so to speak.

Last edited Sat, 22 Nov 2008, 10:11pm by Daniel R Plante

ee-tom wrote:

PeterP

If you are saying it is obvious that eestor's claimed energy density requires some completely new physics - I would agree with you. No-one has come up with a machanism, and many people have come up with contradictions.

But the possible polarisation of a BaTiO3 lattice is greater in some directions than others, so aligning crystals can help increase overall plarisation as I have said. The point is that random orientation 9worst case) gives you quite a lot of polarisation so my figure is a best case estimate for the increase - and not significant whe compared with 400X.

ee-tom:

Yes, I agree. Compared to 400x, anything you do to maximize dipole rotation as energy storage during use is tiny, unless I'm missing something critical.

Dan,

Thanks for your reply. Like you I have more question than answers but here is my attempt to respond to your points

The domain boundary is where different lattice orientations meet. Tom would probably be a good person to explain details of how domain rotation occurs. However domains do rotate and this can be seen using a microscope. Friction during rotation would cause losses. High K dielectrics are lossy but not that bad. The eestor dielectric has particles with alumina coating which defines the boundary, are these the grains you mean? Each particle is one domain.

Domains will rotate due to the field, heat is used to take the material above the Curie point so that when cooled below the Curie point they will “set” in the fully oriented position.

The dipole moment will be larger but my argument is that this represents stored but unrecoverable energy.

I don’t believe that the EESU is used in reverse polarity. Wouldn’t this mean that each charge/discharge cycle requires a change of polarity?

The volume relationship is that for spherical particles 78% is the total volume of tightly packed spheres, the rest is PET. The coating on the spheres is 6% of the volume of the spheres. So my figures are the result of my calculation and can be queried. I have used spheres because that’s what Christine said. ( I think this changes during sintering but I don’t know how)

Regards,

PeterP

Last edited Sun, 23 Nov 2008, 4:55pm by PeterP

I agree about using reverse polarity - really only possible to use it once, if at all.

Agree EESU seems to be trying to make each particle of CMBT coated with 100Å of alumina as close to a pure / single crystal / domain, all aligned perpendicularly to the plates.

That final poling step - at 180°C at 100 bar (1500 psi) and 4000V differential - might tend to make the spheres a bit elongated, a tad toward eggs or rods, ends oriented toward the plates.

As the energy it takes to reach full polarization is, apparently (according to the various SMEs), no where close to the amount of energy the EESUs are supposed to store, any additional energy has to be stored in another manner. I think that is called extrinsic. And fairly clear that the alumina would be involved. Which is how I get to my pet theory of the alumina cage, essentially as a mechnical spring, greatly increasing the amount of force (E(lectric) Field) it takes to approach polarization saturation, prior to voltage breakdown and destruction.

PeterP wrote:

Dan,

Thanks for your reply. Like you I have more question than answers but here is my attempt to respond to your points

The domain boundary is where different lattice orientations meet. Tom would probably be a good person to explain details of how domain rotation occurs.

You mean ee-tom? Yeah that would be great if he could clarify this.

PeterP wrote:

However domains do rotate and this can be seen using a microscope.

See, this really bothers me. I agree that a domain boundary is simply where different lattice orientations meet. My problem is this: a domain is still a solid region inside a solid. How can it possibly rotate? It's like putting a quarter into a coin slot, but only half way, then trying to rotate the coin.

I thought "well, maybe it's just the molecules/unit-cells themselves that rotate in place to re-orient, but that would take way too much heat since you have to overcome the much lower energy state of the (covalent?) bonds that give you your lattice. You're basically melting the crystal into an amorphous substance so the molecules can rotate freely, then coolling to allow the crystal to reform in it's new orientation under the E-field. No such temperatures are used though, and it would probably carbonize the PET if if was that high.

I'm just having a hard time picturing domain rotation.

PeterP wrote:

Friction during rotation would cause losses. High K dielectrics are lossy but not that bad.

You mean friction during domain rotation, right? Not dipole rotation? If that's the case, then I gotta disagree - my understanding is that the domain only rotates under poling conditions, not during everyday use conditions - that's only dipole rotation.

I look at it like this (warning - possibly bad analogy):

Take a metal spring that's say, 2 inches long. It takes a certain amount of energy to compress it full - say 2 joules.

Now use whatever amount of energy you need to physically stretch the spring with brute force, to the point that when it springs back, it's now twice as long - 4 inches - in its relaxed state. But, you've changed the gross morphology of the spring without changing it at the grain level - the level that gives it the elastomeric qualities. Indeed, if you now compress it maximally, you'd probbly find that it only takes the same amount of energy (2 joules) or even slightly less, to compress it, which is of course the same as saying how much energy it can store.

Now, if you take the same spring and heat it properly to the right temperature then stretch it out to 4 inches (relaxed length), then cool/anneal it properly, you'd now find that if you compress it maximally, it would take more than 2 joules (maybe 3.5 or something, probably not double) to maximally compress it. Again, this is also how much energy it can store.

The process of re-annealing the spring to re-orient & resize the grains for the longer length is not "lost" exactly, it has instead gone into changing the material so that it is able to store more total energy on compression and release it during decompression.

I picture poling the same way. The poling reorients the domains, but after poling, during regular use, the energy has simply gone towards allowing more energy to be stored and released by maximizing the number of dipoles that can be rotated a full 180 degrees under a opposite polarity field.

Anyway, that's how I understand it, but I also am not satisfied about other aspects, such as the molecular scale explanation of "domain rotation"

PeterP wrote:

The eestor dielectric has particles with alumina coating which defines the boundary, are these the grains you mean? Each particle is one domain.

I believe Christine explained this a while back...

A bulk crystal will normally be polycrystaline and composed of grains (I assumed these were delineated by boundaries consisting variously of lattice mismatch/imperfections, chemical impuries, etc), with a various number of domains inside the grain. But with EEStor's purity level the grains will typically contain only one domain. So, when you cryogenically fracture the crystal it will typically fracture along the weakest surfaces, which will be the grain boundaries. So you end up with particles consisting of only one grain (which is important for things like aging effects, breakdown, etc), but they also now consist of only one domain as well. IIRC. So the alumina coating will be coincident with the grain boundary, not define it (a niggling distinction, sorry). It works out to the same thing.

PeterP wrote:

Domains will rotate due to the field, heat is used to take the material above the Curie point so that when cooled below the Curie point they will “set” in the fully oriented position.

Yes, I'm just accepting this as fact and carrying on from there in spite of the fact that I don't understand "domain rotation" at the bulk/molecular level :) I know that's a dumb thing to do, but I've been dumber.

PeterP wrote:

The dipole moment will be larger but my arguement is that this represents stored but unrecoverable energy.

See above. I agree that it represents "stored" energy, but only in terms of how it was used to change the material at the grain/domain level, like re-annealing a spring. I also agree that it is "unrecoverable" in a sense, but the end result is that the material now has a higher energy storage density characteristic. I think this is the process used to make old standard electrolytic capacitors, but I can't recall exactly. CapMan would be the expert on that.

PeterP wrote:

I don’t believe that the EESU is used in reverse polarity. Wouldn’t this mean that each charge/discharge cycle requires a change of polarity?

No, because in everyday use, after manufacture (poling, etc) the applied field doesn't change the domain orientation, just the dipole orientation. Again, IIRC.

PeterP wrote:

The volume relationship is that for spherical particles 78% is the total volume of tightly packed spheres, the rest is PET. The coating on the spheres is 6% of the volume of the spheres. So my figures are the result of my calculation and can be queried. I have used spheres because that’s what Christine said. ( I think this changes during sintering but I don’t know how)

I know, that's one of the problems. The WIPO gives volume percent material relationships that are inconsistent with their narrow particle size figures in their 3rd party PR, and with Christine's and your analysis (shrug). I can't comment about changes during sintering I'm afraid.

Dan,

Thanks for your reply. You've obviously put a lot of work into it. I didn't mean ee-tom I meant Tom the materials guy. I think domains do rotate in a barium titanate capacitor in normal usage. We must clear this up before we can go anywhere. I include a link which I think may be helpful.

http://www.msm.cam.ac.uk/doitpoms/tlplib/dielec...

Please have a look and let me know what you think.

Regards,

PeterP

PS I reccomend the whole TLP particularly the section on AC behaviour.

Peter:

Ah, that Tom.

Well, not a whole lot of work, mostly just refreshing my memory. It's been a hell of a long time since college.

You're right, it definitely needs to be cleared up before continuing, although I still don't believe molecular dipoles are EEStor's gig. Thanks for the link!

Guys,

I think rotation is a bad analogy. Think of an square index card that you compress 2% inward from opposite edges. It warps slightly toward one direction. Then blowing (E field) on the bowed side can flip it toward the other. It is bent either one way or the other. No rotation. Stable states are only +-180 degrees. (The magnitude of applied electrostatic force would be the cosine between angle of dipole and applied field.) These are called the 180 degree unit cells. There are other cells at right angle (90 degree to the 180). Just flip. Now replace the image of the card with the titanium ions located at center of card/unit cell.

Below the Curie temperature, the cubes shorten/elongate by 2% into slightly tetragonal shape. The electrostatic and mechanical stress arrange to minimize the energy. You can use this model (thermodynamic-Gibbs-Free-Helmholtz energy, statistical)of energy minimization to construct math models of the behavior of domains/ phase changes. There are many way to construct hysteresis curves from these ideas.

At the domain boundaries, it might be useful to conceptualize the transition as a continuous rotation from one orientation to the other. But, this is probably just a mathematical device to crank out some numbers.

Tom

Last edited Sun, 23 Nov 2008, 7:15pm by Tom

Tom,

Thanks for that it makes sense to me and it agrees with the animation in the following link:

http://www.msm.cam.ac.uk/doitpoms/tlplib/ferroe...

The conclusion that the energy stored during poling cannot be recovered is still valid in my opinion. I think hysteresis loops are generally done for unpoled material. The charge/discharge curves for poled material would be limited to the rhs positive quadrant. Do you have any idea how it compares with unpoled material?

Regards,

PeterP

PeterP,

I believe you are correct in your assessment. Poling is usually done to approximate a "single domain crystal" type material for mechanical sensor or piezoelectric type applications. Removal of the field leaves a hysteretic remanent polarization and stress. This still leaves a 400X gap between known science and EEStor's claims. You can get either high saturation fields or high permittivity, but not both at the same time.

Last edited Mon, 24 Nov 2008, 9:00am by Tom

Worth 1.4X?

Isn't the statistical average of random uni-directional 3 dimensional vectors, in relation to any 2 dimensional plane (the plane of parallel capacitor plates) going to be (45°,45°,45°) ?

Thus the statistical average vector component perpendicular to the plates sin(45°) (same as cos(45°)) = √2 / 2 = .707

Random BT dipoles have a 70.7% effective orientation perpendicular to the capacitor plates?

So, if polled, there'd be 1.414X dipoles 100% perpendicular?
Alternatively, increase the quantity of BT by 1.4X (along with the plate area), to have the same number and strength of dipoles available?

nekote wrote:

Worth 1.4X?

Isn't the statistical average of random uni-directional 3 dimensional vectors, in relation to any 2 dimensional plane (the plane of parallel capacitor plates) going to be (45°,45°,45°) ?

Thus the statistical average vector component perpendicular to the plates sin(45°) (same as cos(45°)) = √2 / 2 = .707

Random BT dipoles have a 70.7% effective orientation perpendicular to the capacitor plates?

So, if polled, there'd be 1.414X dipoles 100% perpendicular?
Alternatively, increase the quantity of BT by 1.4X (along with the plate area), to have the same number and strength of dipoles available?

Nekote,

That seems a reasonable question,I wonder why nobody has answered it? Here is my humble attempt:

I think this would be correct if only 180 degree switching happened and the field wass strong enough to cause that switching. I think that 90 degree switching is possible so the maximum angle possible is 45 degrees. The increase due to poling is therefore about 15%.
I also think some physical reversible rotation occurs.

The important facts are that the polarisation is limited to .26 C/m^2 and that the remnant polarisation for poled barium titanate is about 80% of the saturation value.

Regards,

PeterP

No, the average is different because at an angle of theta from the plane there is a solid angle proportional to cos(theta) hence the probability of being theta from plane is proportional to cos(theta). The dipole moment is proportional to sine(theta) so the whole thing is integral of sine(theta)cos(theta). There is a constant factor, calculations above. (Correct I think).

Bottom line, you can't average since it is non-uniform.

Tom

ee-tom,

Thank you for your response to my post. Unfortunately I am unable to see where I have gone wrong. So I am setting down in more detail the steps I went through, in the hope that you can identify my mistake for me. Here is my reasoning:

Consider a unit cell (a tetrahedron) which has an arbitrary orientation with respect to the (yet to be applied) field. The titanium ion is in one of its six possible positions.
Draw a line through the centre parallel to the field (line A).

Apply the field with sufficient intensity to switch the Titanium ion (provided that it needs to be switched).

I think this ion will switch to the position which is closest to line A in the positive direction.

Draw a line through the centre of the tetrahedron and this position (line B). The dipole moment has a known value (due to the intrinsic properties of the material) and its direction is towards the positive ion. That is along line B. Line A and line B intersect at the origin and define a plane. There is an angle between them in this plane. Considerations of symmetry lead me to believe that the maximum value of this angle is 45 degrees. So the maximum and minimum values of the dipole component in the direction of the field are 1 and .707 and the average is .85. Integrating to find the average gives 2*sqrt(2)/pi = . 9 Yes I was a bit rough. But that’s not what you are talking about. Is it?

Regards,

PeterP