Brown-Paul: Mathematician's review (Step 2)

Reviewer role: PhD mathematician (derivations, notation, units, hidden assumptions). Source folder (read only): source folder: Brown-Paul

Source gap (repeated from Step 1, amended). Only two documents exist: the 1990 Tesla Symposium bundle and a 1999 Wired news item. Neither contains measured data. The bundle has a patent cover page only, not the full patent. I obtained the patent text from the web (Google Patents); that web text is secondary to the folder and is used below with a [web] tag. Everything about the gamma-transmutation claims (4, 5) rests on a news item plus [memory]/[web] physics; no Brown document describes that scheme.

Tags: [read] = from the folder; [read: page image p.N] = from a rendered page; [web] = URL given; [memory] = recalled, with confidence. Page numbers are PDF pages of telsasymposium1990paperbypaulbrown.pdf.

Corrections to Step 1

1. Patent coverage. Step 1 said the bundle holds "a scanned copy of US Patent 4,835,433" with an unreadable text layer. In fact PDF p.4 holds only the patent's front page (bibliographic data, abstract, one drawing "Figure 1") plus figures 2 to 6 of the paper, not the body or claims. I rendered it at 300 dpi and the abstract is legible [read: page image p.4]: "A nuclear battery in which the energy imparted to radioactive decay products during the spontaneous disintegrations of radioactive material is utilized to sustain and amplify the oscillations in a high-Q LC tank circuit... The circuit inductance comprises a coil wound on a core composed of radioactive nuclides connected in series with the primary winding of a power transformer. The core is fabricated from a mixture of three radioactive materials which decay primarily by alpha emission..." It states "10 Claims, 4 Drawing Sheets". So the patent device uses alpha emitters in the inductor core, not the Sr-90 beta source of the paper. Step 1's reading that the claim concerns "beta/alpha" generically and its description of the abstract gist as incomplete are corrected accordingly.
2. Figures with equations were missed. Fig. 6 states R < 2*sqrt(L/C) and dU/dt = -i^2 R; Fig. 7 states Beta Flux >= i^2 R and f = 1/(2*pi*sqrt(LC)); Fig. 8 labels the load "50 KHz"; Fig. 11 shows a Be-Cu coil, an Ag element and a Sr-90 source with a "HVDC in to start / Class C Amp. to ground" start-up [read: page image p.4 and p.5]. Fig. 4 lists the proposed mechanism as 1 original current, 2 increased charge carriers, 3 flux cutting, 4 collapsing magnetic field, 5 reduced resistance and increased conductivity, 6 current proportional to emf. Fig. 4 is the only place the paper hints at a magnetic mechanism; the text never explains "flux cutting".
3. Wired byline and detail. Step 1 said no byline. The saved text ends "- Charles Platt" [read]. The item also says Brown had a patent pending, formed a company "Nuclear Solutions", would "soon run tests at the University of Illinois or MIT", and was negotiating with the DOE for a pilot plant. It also quotes "half-lives of 24,000 years". No test results are reported.
4. Beta Cell history. The paper credits the beta cell to "Moseley in 1913" [read p.2]. Step 1 omitted it. Step 1's "Burke Cell" description is right, with Burke patents 3,409,820 and 3,530,316 shown in Fig. 3 [read: page image p.4].
5. Step 1's other numbers (0.25 W/cm3, 1 MCi/yr, 42 MCi, 8,500 Ci/ton, 80,000 ions in 0.030 in, 10 to 5,000 mW) check against the text [read p.2-3]. The CV is on PDF p.6, dated after Aug 1992/Jan 1993 (it lists those meetings), and lists a 1991 Boston ICEC paper "Current status and future research with resonant nuclear generators": Step 1 is right that the CV postdates 1990.

Cross-cutting calculations used below (all mine; constants tagged)

Constants: 1 Ci = 3.7e10 decays/s [memory, high]; 1 MeV = 1.602e-13 J [memory, high]. Kr-85 half-life 10.728 y, mean beta 251 keV, max 687 keV, 0.43 percent branch with 514 keV gamma [web https://en.wikipedia.org/wiki/Krypton-85]. Sr-90 half-life 28.91 y, beta endpoint 546 keV [web https://en.wikipedia.org/wiki/Strontium-90]; Y-90 daughter mean beta about 0.93 MeV, endpoint 2.28 MeV [memory, medium, not web-verified].

• Power per curie, Kr-85: 3.7e10 s^-1 x 0.251 MeV x 1.602e-13 J/MeV = 1.49 mW per Ci, an absolute ceiling (100 percent conversion, ignores the small gamma branch).
• Power per curie, Sr-90 + Y-90: 3.7e10 x (0.196 + 0.934) MeV x 1.602e-13 = 6.7 mW per Ci [uses medium-confidence Y-90 mean energy]. Specific activity of Sr-90 is about 140 Ci/g [memory, medium], so about 0.94 W/g ceiling.
• Kr-85 specific activity: N_A ln2 /(T_half x 85 g/mol) = 1.45e13 Bq/g = 392 Ci/g (computed).

Claim 1. A tuned LC tank with a built-in radioactive source sustains itself and delivers net power after one start pulse (resonant nuclear oscillator, US 4,835,433)

a. Internal consistency of the mathematics. - The circuit mathematics stated is correct but trivial. For a series RLC, f = 1/(2 pi sqrt(LC)) is the undamped resonance and R < 2 sqrt(L/C) is exactly the underdamping condition (from (R/2L)^2 < 1/(LC)); dU/dt = -i^2 R is Joule loss of the stored energy U [read: page image p.4, Figs 6, 7]. All fine. - The text says resonance "causes the inductive and capacitive reactances to cancel leaving only ohmic losses" [read p.2]. True of series resonance, but it means resonance minimises the impedance, it does not remove loss. Energy still decays as U(t) = U0 exp(-R t/L) (Q-limited decay), so something must inject power. - The key stated condition is Fig. 7: "Beta Flux >= i^2 R". Read as a power inequality this is the correct energy budget: source power P_s must be at least the loss. It is also an admission that net output cannot exceed P_s: P_out <= P_s - i^2 R. So Brown's own equation limits the device to the decay power. The paper never evaluates P_s for any source. - The text instead says the current "gets amplified by an amount proportional to the activity" every cycle [read p.2]. This is a different, inconsistent model. If the source acted as a gain proportional to current, it would be a negative resistance -r, giving injected power r i^2, growing with amplitude. A decay source delivers power fixed by the nuclear decay rate (independent of circuit current); a constant P_s and a current-proportional gain cannot both be right. The paper switches between them without noticing. - Hidden assumption: that random decay events inject energy coherently in phase with the oscillation. Decays are a Poisson process, uncorrelated with the tank phase. Randomly timed charge pulses average to zero net work on the oscillator at first order (they add noise, not drive). Phase-correlated injection requires a nonlinear element (junction, tube, negative-resistance device) that rectifies or gates. None is described in the paper. The patent text [web] says that "the collapsing magnetic field from stopped alpha particles induces current pulses", which is asserted, not derived. - Brown's energy statement "neither the electric field, the electrodes or the medium... contribute any energy... the energy is contributed by the ion generator" [read p.2] is correct as bookkeeping, and again caps output at the decay power.

Patent performance statement (key numerical test). The patent text states: "Using a one millicurie radium needle 39, 200 grams of uranium 33 and 100 grams of powdered thorium 31 ... at 86 kiloHz, a continuous output of 23 amperes at 400 volts into a resistance load has been achieved." [web https://patents.google.com/patent/US4835433A/en; fetched twice; no efficiency, input power or measurement method is given]. This is not in the folder (the folder holds only the cover page), so it is a [web] claim about the patent. - Output into a resistive load: P = V I = 23 A x 400 V = 9,200 W (taking rms values; peak would not help the claim). - Decay power available (generous upper bound, my calculation): radium needle 1 mCi = 3.7e7 decays/s; counting the four alphas of the chain to Po-214 (4.78 + 5.49 + 6.00 + 7.69 MeV) plus 3 MeV of betas gives 0.16 mW. 200 g natural uranium (about 12.35 kBq/g each of U-238 and U-234, plus 0.57 kBq/g U-235, adding Th-234/Pa-234 betas) gives 4.5 uW. 100 g Th-232 (4.06 kBq/g) with the whole chain at secular equilibrium (about 42 MeV per chain) gives 2.7 uW. Total at most about 0.17 mW (1.7e-4 W). Activity constants are [memory, medium]. - Ratio: 9,200 W / 1.7e-4 W = about 5e7. Energy is violated by roughly seven to eight orders of magnitude. Even if the uranium and thorium were hugely more active than I assume, the sources are tiny: for 9.2 kW from alpha decay at 5 MeV one would need 1.1e16 decays/s = 300 kCi. - Conclusion for claim 1 in its strong form: the patent's own quoted figures contradict energy conservation; no known measurement method is described.

b. Consistency with experiment. Betavoltaic and alphavoltaic cells are real and give single-digit percent efficiency and microwatt to sub-milliwatt power [web https://en.wikipedia.org/wiki/Betavoltaic_device (that page mentions no resonance or oscillator circuit)]. The Fig. 4 mechanism ("flux cutting, collapsing magnetic field") has no counterpart in any known measurement of radioactive decay: decay rates and energies are not changed by circuit conditions (to better than parts in 1e-3 to 1e-4 for chemical and pressure effects) [memory, medium]. Radioisotope electrical output has never been seen to exceed the decay power.

c. Tested? By Brown himself only. The paper says the prototype "generated electricity" but showed "a frequency stability problem and signs of material degradation" and that "Our current program will generate engineering data in the coming months" [read p.3]; no numbers are given. The patent number above is a one-line assertion. Secondary advocacy pages (rexresearch, tripod, forum posts, e.g. https://www.rexresearch.com/nucell/nucell.htm, https://antigravitypower.tripod.com/FreeEnergy/p.brown.html) repeat figures such as 7,500 W per gram Sr-90 and "fifty kilowatts from a radium needle" [web, search-result snippet only; not read in full, not a test]. 7,500 W/g exceeds the 0.94 W/g ceiling by a factor of about 8,000. I found no independent replication, no independent measurement, and no refusal-to-test record in the sources I read. Whether independent labs were ever offered the device is unknown.

d. What would have to change. Conservation of energy (first law) would have to fail, or radioactive decay rates would have to be controllable by circuit resonance. Given that the evidence is internally contradictory rather than anomalous data, nothing in mainstream physics needs revising on present evidence.

Verdict, claim 1 (split): - Strong form (net power exceeding decay power, thousands of watts from milligram-scale sources): Refuted by calculation (energy budget above; also the patent's own Fig. 7 inequality caps it). Also Internally inconsistent (constant-power injection versus current-proportional gain). - Weak form (radioisotope-powered oscillator delivering up to the decay power, tens of percent of a few mW/Ci at best, via a conventional betavoltaic or alphavoltaic element in an oscillator): Plausible, untested by Brown (no data), though near-equivalent devices are Established.

Claim 2. Radiation in a current-carrying conductor "amplifies" current in proportion to activity and lowers its resistance (Beta Voltaic Effect, Burke Cell)

a. Mathematics. - Ohm's law, I = V/R, "reduction in resistance... current directly proportional to voltage" [read p.2]: consistent with an ohmic wire, but the claim is about a change in R. - Series-circuit check (Kirchhoff's current law): in the Burke cell (Fig. 3, battery in series with a radioactive conductor and a load) the same current I flows through every element. Battery power is emf x I. The statement "all of the power dissipated in the load is not drawn from the battery" [read p.2] therefore cannot hold unless the radioactive conductor acts as its own emf. If the wire's resistance falls, the battery simply supplies more current. Any extra power taken from the radiation is bounded by the energy the decay deposits, <= 6.7 mW/Ci for Sr-90/Y-90. - Carrier-count check. The paper's own figure: one Sr-90 beta makes "80,000 ions in .030 inches" in copper [read p.2]. Check of the figure: 0.030 in = 0.76 mm; with the empirical Katz-Penfold-type range formula R = 0.412 E^(1.265 - 0.0954 ln E) g/cm^2 [memory, medium], Cu density 8.96 g/cm^3 gives 0.21 mm for the 0.546 MeV Sr-90 endpoint and 0.70 mm for 1.4 MeV, so the quoted 0.030 in matches the Y-90 daughter electrons (about 1.4 MeV), and 1.4 MeV / 80,000 = 17.5 eV per "ion". The number is plausible and shows deposited energy is the only supply. - Now the effect size. A 1 Ci Sr-90/Y-90 source makes 2 x 3.7e10 betas/s, each freeing 8e4 excitations: 5.9e15 per second. Copper has about 8.5e22 conduction electrons/cm3 [memory, high]; a 1 mm diameter, 10 cm wire holds 6.7e21. Excited electrons thermalise in about 1e-14 s [memory, medium], so the steady excess is 5.9e15 x 1e-14 = about 60 extra mobile carriers out of 6.7e21, a fractional change of order 1e-20. Even with an absurd 1 s lifetime the change is under 1e-6. The conductor's resistance cannot change measurably by this mechanism. The paper's word "avalanche electrons" misuses avalanche (multiplication by impact ionisation in a field); no multiplication occurs in a metal. - Hidden assumption: that ionised carriers are additional carriers that are not already counted in the conduction band. In a metal they are not.

b. Experiment. Radiation-induced conductivity is a real, measurable effect in insulators and semiconductors (few free carriers) and in gases (ionisation chambers) [memory, high]. In metals it is negligible by the count above. The Burke cell patents (3,409,820 and 3,530,316, listed as references on the patent cover [read: page image p.4]) are earlier art; the paper does not report any measurement of the claimed resistance drop ("we also measure" with no data).

c. Tested? No data in the folder. I found no independent replication.

d. What would change. Metal conduction theory (Drude/Fermi-liquid picture) would need a mechanism giving ~1e20 times more effect than the carrier count allows. Nothing mainstream requires it.

Verdict, claim 2 (split): The mild form (radiation-induced conductivity in insulators, semiconductors and gases): Established. The claim as stated for a metal wire (measurable amplification and resistance drop proportional to activity, with the load power not drawn from the battery): Refuted by calculation (carrier count, KCL, energy budget) and Internally inconsistent (series circuit).

Claim 3. Direct conversion can reach about 0.25 W/cm3 with Kr-85, with better efficiency and less shielding than conventional generators

a. Mathematics and units. "Preliminary data suggests that energy densities on the order of .25 watt per cubic centimeter is achievable" [read p.3]; no data, no volume definition (source only, or the whole battery?). - Required Kr-85: 0.25 W / 1.49 mW per Ci = 168 Ci/cm3 at 100 percent conversion. - Pure Kr-85 gas at 1 atm and 0 C (85/22,414 g/cm3 = 3.8e-3 g/cm3 ideal gas) holds 392 Ci/g x 3.8e-3 = 1.49 Ci/cm3. Reaching 168 Ci/cm3 therefore needs about 113 atm of isotopically pure Kr-85 (ideal-gas estimate), with 100 percent conversion into electricity. At the single-digit-percent efficiency of real betavoltaics [web https://en.wikipedia.org/wiki/Betavoltaic_device], the pressure would be 10 to 50 times higher still. So 0.25 W/cm3 is at or beyond the thermodynamic ceiling for a gas source, not an "on the order" engineering target. (Liquid or compressed forms are not discussed in the paper.) - Scale check on the fuel numbers in the paper: 1 MCi/yr x 1.49 mW/Ci = 1.5 kW of total thermal power; the whole 42 MCi inventory is 62 kW. The paper's 1989 conference title "Resonant nuclear battery may aid in mitigating the greenhouse effect" [read p.3 reference 3] cannot be supported at this scale: world electricity use is of order TW. (The 1 MCi/yr and 42 MCi figures are Brown's, unchecked by me.) - "Better efficiency": conversion efficiency cannot exceed 100 percent; real RTGs convert about 5 to 7 percent thermally [memory, medium]; no efficiency figure is given for the resonant device. "No shielding" claims: Kr-85 is mainly beta (0.43 percent gamma at 514 keV [web Kr-85 page above]), so shielding is modest for any design; that is a property of the isotope, not of resonance.

b. Experiment. Compared with real devices: tritium betavoltaics, about 100 microwatts at 20 g [web https://en.wikipedia.org/wiki/Betavoltaic_device], i.e. power densities of order 1e-2 to 1e-1 mW/cm3 to a few mW/cm3, thousands of times below 0.25 W/cm3.

c. Tested? The paper says "Preliminary data", no values, no method, and states a program "will generate engineering data in the coming months" [read p.3]. No independent result found.

d. What would change. No new physics needed if the number were a thermal ceiling; but as an electrical output with a resonant gain, again energy conservation.

Verdict, claim 3: Refuted by calculation for 0.25 W/cm3 electrical from Kr-85 at any plausible density and efficiency (needs about 113 atm pure Kr-85 at a perfect 100 percent). The qualitative statement that radioisotope batteries have high energy density versus chemical cells: Established. The "better efficiency" statement: Never properly tested (no efficiency given).

Claim 4. Gamma irradiation can transmute long-lived waste into short-lived or stable species within months, for under a quarter of DOE's $150 billion

a. Mathematics. The Wired item gives no equations or numbers beyond cost. Hidden assumptions that fail a back-of-envelope check: - Photonuclear reactions are real: the giant dipole resonance is excited by gamma rays of 7 to 40 MeV [web https://en.wikipedia.org/wiki/Giant_resonance]. Peak photoabsorption cross sections are of order hundreds of millibarns (1 b = 1e-24 cm2) [memory, medium; the web check did not give a number, so unverified], over a width of a few MeV. Atomic processes (pair production) for a 15 MeV photon in a heavy target are of order tens of barns per atom [memory, medium; unverified]. So only on the order of 1 percent of the gamma energy goes into photonuclear reactions. - Energy cost per transmutation (order of magnitude): about 15 MeV of gamma per photon / 0.01 probability = about 1.5 GeV per transmuted nucleus, before the beam-generation efficiency (electron linac plus bremsstrahlung, perhaps tens of percent [memory, low]). A Cs-137 nucleus releases 0.5 to 1 MeV in its whole remaining life. So the energy spent per atom is about 1,000 to 3,000 times what that atom will ever release. Most of the beam energy appears as heat in the target. - "The excess heat could generate electricity, more than enough to run the plant" [read, Wired p.3]: that heat is the input beam power converted to heat. A heat engine returns at most about 30 to 40 percent of it, so the plant is a net consumer unless the targets are fissile actinides and fission energy (about 200 MeV per fission) is gained. Photofission cross sections are far below the absorption total, so the same factor of about 100 in cost per event applies; a net gain is not shown by any calculation in the source [memory, medium]. - Transmutation outcomes: (gamma,n) on Cs-137 gives Cs-136 (13 days, then stable Ba-136), but on Sr-90 gives Sr-89 (50 days), so "safe within a few months" is a statement about particular nuclides at best; "nuclear waste" as a whole is mostly other species. The statement about alpha particles turning beryllium to carbon [read] is a different (alpha, n) reaction and says nothing about gamma transmutation. - Cost: no cost model given; $5 million for a plant versus $1.3 billion for neutron-beam projects is an assertion.

b. Experiment. Photonuclear transmutation of Tc-99, I-129, Cs-135 etc. is a studied research topic (laser-Compton gamma sources, electron-linac photoneutron) [memory, medium; the search result list included "Expanding Nuclear Physics Horizons with the Gamma Factory" https://arxiv.org/pdf/2106.06584 but I did not read it]. Small laboratory transmutation yields are known; large-scale economic waste treatment is not.

c. Tested? Brown said tests would "soon" be run at the University of Illinois or MIT [read, Wired p.3]. I found no published result in my searches and do not state whether tests happened. Three physicists quoted in Wired (Park, Schiffer, Doolen) say only that it is "not far-fetched", "could convert", and is expensive [read].

d. What would change. Nothing in nuclear physics; only an economic and energy-balance conclusion (that gamma transmutation is an energy sink) would have to be wrong.

Verdict, claim 4 (split): Principle (photonuclear transmutation of some long-lived isotopes into shorter-lived ones): Established (Schiffer concurs [read]). "Within months" for waste in general at under a quarter of $150 billion: Never properly tested (no neutral test on record that I found; not shown dismissed without testing either). The cost half also fails the energy-per-event estimate above, so it is not credible as stated but I class it on the calculation as at best Plausible, untested only for selected nuclides.

Claim 5. Gamma-driven transmutation heat can power the plant, allowing a $5 million plant

a. Mathematics. As worked above: the heat is the beam's own deposited power; electrical recovery is capped by Carnot-type efficiency (at most a few tens of percent); a 1,000-fold or greater energy deficit per transmuted nucleus follows from the cross-section ratio. Gain would need photofission-driven chain energy, which means the system becomes a (subcritical) reactor, a different device from the one described. No $ figure is derived.

b. Experiment. Electron linacs and laser-Compton sources of the needed energy and power are large machines costing tens of millions to hundreds of millions of dollars [memory, low]; no known gamma source of this type is listed at $5 million for waste-scale beam power.

c. Tested? No published test found; the Wired statement of planned tests is unconfirmed [read].

d. What would change. No physics, but an energy accounting (input beam power versus recovered power) that fails by about three orders of magnitude on my estimate would have to be wrong.

Verdict, claim 5: Refuted by calculation for self-powered operation on generic waste (energy per transmuted nucleus about 1e3 times its decay energy; heat recovery cannot exceed beam input) [calculation rests on medium-confidence cross sections that I could not verify on the web; if the reader doubts the factor 100 absorption ratio the conclusion weakens to "Never properly tested"]. The $5 million cost: Not testable as stated (no cost basis).

Summary table

Web checks performed (budget about 6)

1. Google Patents US4835433A (claims, performance statement; twice) [web https://patents.google.com/patent/US4835433A/en]; the claims, as retrieved: apparatus with a radioactive-core inductor in a series LCR with a transformer primary; core of at least two (claim 4, three: radium, uranium, thorium) radioactive materials decaying primarily by alpha emission.
2. Kr-85 decay data [web https://en.wikipedia.org/wiki/Krypton-85]: confirmed.
3. Sr-90 data [web https://en.wikipedia.org/wiki/Strontium-90]: half-life and 546 keV confirmed; Y-90 energies and specific activity remain [memory, medium].
4. Giant resonance [web https://en.wikipedia.org/wiki/Giant_resonance]: 7 to 40 MeV confirmed; cross-section magnitudes not confirmed (check failed).
5. Betavoltaic efficiency and power [web https://en.wikipedia.org/wiki/Betavoltaic_device]: confirmed; that page does not mention Moseley or resonant circuits, so I could not verify the 1913 Moseley date beyond the paper's own statement [read p.2].
6. Search for tests or replications of Brown's devices: no independent test or replication found; only advocacy pages (snippets only, unverified, not used as evidence).