Objections from Paul Dietz

[JoeStrout]

Paul Dietz posted some comments about polywell fusion in the sci.energy newsgroup (about halfway down this thread ). To quote:

Let me summarize why Polywell is unlikely to work, at least with p-11B.

First, you CANNOT have a near-neutral plasma in the central interaction zone. If you
do, then either the electrons there are hot -- in which case, bremsstrahlung
exceeds fusion power -- or the electrons are cold, in which case the power of electron
drag on the ions far exceeds fusion power. For intermediate electron temperatures,
both effects will occur.

(Rostoker and Monkhorst recognized this and got around it by having the
boron ions and electrons both at low energy; these highly charged ions dominate
the interaction with the electrons.)

So, you need to have a non-neutral plasma there. But this severely limits
current density, due to the space charge limit. To get around this limit,
you need to focus the ions down to an extremely high density central peak
(the total power will scale as 1/r, where r is the diameter of the central
space-charge limited reaction zone).

But even slight lateral velocity will mean the ions miss this peak.
In particular, Ligon's notion that collisions out at the turn-around
point 'anneal' the distribution (never mind that getting one entropy-creating
process to cancel out another entropy-creating process doesn't make
sense thermodynamically) cannot then work, since collisions there will
add angular momentum to the ion orbits, causing them to miss the central
peak and take them out of the game.

In practice, non-uniformity of the Polywell configuration would prevent
any fine focusing of ions into the center to begin with.

I don't know what Paul's qualifications are, or if he's relying on authority, what those authorities might be. But his arguments are at least coherent, and ones we should consider carefully.

Anybody have any thoughts on these objections?

[drmike]

I guess the most important statement is "show me the math". This is something Bussard is working on, and until he's got a full theory for us to beat our heads on. we won't really be able to deal with descriptions either way. It's entirely possible his theory has flaws, but it would be good to know what they are so we can try different ways to fix them.

Plasmas are complicated beasts, and in a magnetic field they get grumpy. I can see how the math will be messy. But to categoricly say "it won't work" is sticking your neck out pretty far. Seems like Bussard must have measured some thing useful for the past 11 years.

[tonybarry]

Like drmike, I think the maths will go a long way to settling the argument; and a working model will go even further , but in the interim, I find Dietz's comments unconvincing. Note that I am an interested amateur here, and so what I write is worth very little in the fusion scientific community.

The positive ions will preferentially fall to the centre of the polywell due to the considerable electrostatic attraction there. Any radial energy will get wiped off pretty quickly by interactions with other species (whether near misses or direct hits).

Lack of fine focussing of the magnet geometry is not an issue as far as I can see; the practical result is that the central peak becomes a central region (some thing like a mountain range). As long as the gross focus is OK, the polywell will operate. I take comfort in the straight IEC fusors (e.g. run by UniWisconsin) which operate purely on electrostatic grids. They have no issues with fine focus. And their central grids have assymetric ion injection (i.e from one side).

I'm unable to comment on the hot vs cold electrons issue. I think that Bussard makes the note that the fusing zone does not operate in local thermal equilibrium, thus bypassing the hot vs cold issue entirely. I understand that when first injected, the species are all at one energy level (kV) and hence don't have a spread out Maxwellian distribution of velocities. But I do not know whether this mellows out over the course of time, or whether the lifetime of the species is short enough that they die before they lose energy (i.e. are consumed by absorption into the fusor structures, or fused into helium at high velocity and scavenged out of the tank).

Once again, my expertise is not in this field, and anything I write should be taken with this caveat.

Regards,
Tony Barry

[MSimon]

Where the electrons are in high density but low energy (center) their energies will thermalize. However since the energy is low this tends to re-peak the distribution.

For ions this happens near the anode (shielded grids).

It is good enough that the ions fuse before their energies completely spread out.

[cuddihy]

Paul Dietz posted some comments about polywell fusion in the sci.energy newsgroup (about halfway down this thread ). To quote:

Let me summarize why Polywell is unlikely to work, at least with p-11B.

First, you CANNOT have a near-neutral plasma in the central interaction zone. If you
do, then either the electrons there are hot -- in which case, bremsstrahlung
exceeds fusion power -- or the electrons are cold, in which case the power of electron
drag on the ions far exceeds fusion power. For intermediate electron temperatures,
both effects will occur.

(Rostoker and Monkhorst recognized this and got around it by having the
boron ions and electrons both at low energy; these highly charged ions dominate
the interaction with the electrons.)

So, you need to have a non-neutral plasma there. But this severely limits
current density, due to the space charge limit. To get around this limit,
you need to focus the ions down to an extremely high density central peak
(the total power will scale as 1/r, where r is the diameter of the central
space-charge limited reaction zone).

But even slight lateral velocity will mean the ions miss this peak.
In particular, Ligon's notion that collisions out at the turn-around
point 'anneal' the distribution (never mind that getting one entropy-creating
process to cancel out another entropy-creating process doesn't make
sense thermodynamically) cannot then work, since collisions there will
add angular momentum to the ion orbits, causing them to miss the central
peak and take them out of the game.

In practice, non-uniformity of the Polywell configuration would prevent
any fine focusing of ions into the center to begin with.

I don't know what Paul's qualifications are, or if he's relying on authority, what those authorities might be. But his arguments are at least coherent, and ones we should consider carefully.

Anybody have any thoughts on these objections?

Here's what I have to say about this -- no where in this entire argument does he mention a dependance on power level of these effects. In other words, the result is not just that breakeven is not possible, but that IEC shouldn't work at all --in other words, this reasoning should "disprove" any type of IEC fusion, including the small level seen by Bussard, and even Hirsh-farnsworth and fusor effects.

Even if all he's saying is with regards to Brehmstrahhlung, it should be easily provable or disproveable in a fusor --simply measure the level of fusion power by neutron detector vs. the elvel of brehmstrahlung,

I would think that this entire line of reasoning is invalidated by existing fusors!

[bcglorf]

Plasmas are complicated beasts, and in a magnetic field they get grumpy. I can see how the math will be messy. But to categoricly say "it won't work" is sticking your neck out pretty far. Seems like Bussard must have measured some thing useful for the past 11 years.

For Bremsstrahlung losses, relative particle densities and energies really matter. Studies are only just now starting to try and measure what the potential well and ion distributions look like in fusors. You can bet that the polywell distributions will be a more complicated/dynamic scenario still.

From Krall and Wong 92, potential well formation will repel the electrons from the centre of the machine, forming an electron shell. Thus from potential well edge to the centre, electron density actually heads to 0. Every critical assessment of Bremm calcs I've seen assumes that both ion and electron density is at it's max in the centre. That's a significantly different plasma there alone.

That the ion density in the centre should draw electrons back towards the centre again just makes the whole plasma more complex still. I'm inclined to give more merit to experimental results than theory when things get that messy.

As for the current direct observation of fusors(2004-present), they've already started making interesting observations. When using a spherical grid fusor, ion velocity peaks at the sphere's edge, rather than the centre. Not too shocking perhaps, but current models seem to assume ion velocity peaks at the centre.

[drmike]

You ain't kidding it is messy. The formula for plasma Bremsstrahlung in my "Principles of Plasma Physics" (Krall & Trivelpiece, pg 599) is

P(w) = (4 Z^2 e^6 n_i n_e)/(3 pi m_e^2) [2 / (3 c^3 w)] (w^2 - w_p^2]^0.5 *
..............Integral( f_e0(v_e) { dirac( w - k v_e) / (k^2 |D_L(k, w)|^2} dk dv_e)

w is frequency, Z is ion charge, n_i is ion density, n_e is electron density m_e is eletron mass, w_p is plasma frequency, f_e0 is electron velocity disribution, k is wave number and D_L is longitudinal dielectric constant defined by

D_L(k, w) = 1 + sum(w_pa^2 / k^2 integral{ (k dot grad f_a0) / ( w - k dot v) dv}
..............................a

where a is the different species and f_a0 is the velocity distribution of each species a.

What are the particle energy distribution functions for p and 11B that's he's claiming? Without that you can't compute the dielectric nor power distribution as a function of wavelength.

My bet is that it is far easier to measure in your basement than to compute - but I'm an experimentalist

[TallDave]

Paul Dietz posted some comments about polywell fusion in the sci.energy newsgroup (about halfway down this thread ). To quote:

Anybody have any thoughts on these objections?

Yes, I had this discussion with Dietz at Rand Simberg's place. I don't know much about him, except that he seems very convinced that the ion orbits can't allow for fusion. I argued that seems more like a parameter dependent on the well depth than a true objection to the concept (as cuddihy points out here).

Also, this

"(never mind that getting one entropy-creating process to cancel out another entropy-creating process doesn't make sense thermodynamically)"

doesn't seem to be true. You can't reduce entropy in a closed system, but you CAN buffer it (think of pushing iron filings ordered around a magnet: the magnet tends to push them back into place as soon as you stop exerting force), and of course with power inputs (which there are in Polywell) you can reduce it.

The WB-6 and other fusor results would tend to indicate he must be wrong.

[Roger]

Brehmstrahhlung Losses were 1/32nd in the HEPS device.

Don't Brehmstrahhlung Losses occur only when electron denisity is high, and electron energy is high too. The potential well would have hi densities of electrons, but @ low energy... NO ? Wouldnt hi energy electrons tend to scatter, and well formation degrade to zero. But we know the Well forms....

[MSimon]

Brehmstrahhlung Losses were 1/32nd in the HEPS device.

Don't Brehmstrahhlung Losses occur only when electron denisity is high, and electron energy is high too. The potential well would have hi densities of electrons, but @ low energy... NO ? Wouldnt hi energy electrons tend to scatter, and well formation degrade to zero. But we know the Well forms....

Roger,

You need to change your sig. pBj (the j standing for joules) takes 50 to 65 KV of DRIVE . The actual resonance peak (.1 barn) is around 140 to 180 KeV center of mass.

As drmike has said. The easiest way to compute Bremms losses is to let the experiment compute it for you. Then you work the math.

In other words first time around you need experimental results. After that you work out the math and hand it to the engineers.

Think of the experiment as a quantum computer :-)

[TallDave]

Dietz seems to be a programmer with some interest in science threads.

http://www.google.com/search?hl=en&q=Paul+F.+Dietz

His argument style (at least re polywell) is, shall we see, a bit pugnacious. I wouldn't take his objections too seriously. I think he's just the type who likes to fight online.

Again, the amount of neutrons Bussard's design produced at 5 KeV and record amount of neutrons produced at 12.5 KeV tend to argue the concept is sound.

I have a little Hirsch/Farnsworth fusor that puts out 3000 fusions per second at 18 kV on DD. At 12.5 kV, the highest drive voltage WB-6 was run at, most fusors put out so little you have to beat the counting statistics to death to even detect the output, but WB-6 actually put out (for about a quarter of a millisecond at a time) a screaming load of neutrons. Yesterday I read a report that noted that one of the tests was actually run at 5 kV and produced a neutron count (26000 fusions per count) in the 1/4 millisecond or so that the deep potential well existed. Realizing that the statistical significance of one count is +/- 100%, it still floored me. NOBODY does DD fusion at detectable levels at 5 keV. The reason it could happen is that the machine naturally produces head-on collisions at fusion energies in the region around the central convergence point.

http://forum.nasaspaceflight.com/forums ... =5367&mid=

[MisterX]

The power lost from electrons to bremsstrahlung is essentially irrelevant*; the electrostatic attraction of the ions to the electrons isn't dependant on electron velocity. This applies to his statement on "hot electrons".

As for "power of electron drag" - I have no idea where this notion comes from, or what it refers to exactly. I envision escaping charged fusion products pulling electrons in their wake, only to have the electrons turned around by the magnetic field. However I see nothing to suggest this would result in an overly large loss of energy from the charged fusion product.

*except when electrons bind to ions, an event that should occur with freqency negatively correlated to both ion and electron velocity. Low energy ions may form neutrals in the core, and may serve as vehicles for electrons to reach the surface electrode. As long as the ion energies are high enough, energies of electrons should be irrelevant.

[JD]

Dietz seems to be a programmer with some interest in science threads.

http://www.google.com/search?hl=en&q=Paul+F.+Dietz

His argument style (at least re polywell) is, shall we see, a bit pugnacious. I wouldn't take his objections too seriously. I think he's just the type who likes to fight online.

From my previous observations of Dietz's posting habits I'd say you're pretty spot on in your assessment. In fact, if I was unkind, I would state Mr. Dietz is a legend in his own mind. I'm sure he's a decent fellow and all that but being well versed in one's own field does not necessarily carry over into other areas.

[TallDave]

More substantively, the objection on the basis of "only three neutrons" seems a bit off. It appears they actually detected 12 neutrons -- four test runs in each of which three neutrons were detected in the .25 millisecond the potential well was active.

If it were only one test with three neutrons, you might be able to argue that wasn't too significant. But four sub-milli tests, when background detection is on the scale of minutes? I feel pretty confident saying that's very significant, assuming there's no systemic error.

[MSimon]

I don't have my statistics book handy but a count of 3 - 4 times might have a standard deviation of at least 1 and maybe 2.