Brown-Paul: Science Contrarian review

Reviewer role: Science Contrarian (PhD-standard; judges by experiment, builds the strongest honest case, applies the same rigour to the researcher).

Source gap (repeated from Step 1, updated). The folder holds two documents. The 1990 Tesla Symposium PDF (6 pages) has a clean-text paper, a patent front page only (not the full patent), figure sheets and a CV. The Wired item (Feb 1999) is a 600-word news piece. Neither folder document contains a measured power, efficiency or raw-data table. I therefore (i) rendered the figure sheets and patent front page myself, and (ii) fetched the patent text from Google Patents, which does contain one quantitative performance statement. That statement was read through a web-fetch summariser, not as a page image, so treat the exact figures as [web], medium-high confidence, and check them against the primary patent. No independent test of any Brown device was found.

Source tags: [read] = from the folder PDFs; [read: page image pN] = rendered page; [web] = URL given; [memory] = recalled, confidence stated.

---

Corrections to Step 1

1. Patent in the bundle is only its front page. Page 4 of the PDF holds the patent front sheet (bibliographic data, abstract, Fig. 1 drawing). Step 1 called it a "scan of US Patent" with "full claims not read"; the claims and description are not in the folder at all [read: page image p4]. The full text is on Google Patents: https://patents.google.com/patent/US4835433A/en [web].
2. Step 1 said no measured power appears anywhere. That is true of the folder but not of the patent. The patent's description, as returned by the fetch, states: "Using a one millicurie radium needle, 200 grams of uranium and 100 grams of powdered thorium ... 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]. This is the key number in the project and Step 1 and claims.md missed it. See Claim 1.
3. Patent details Step 1 got incomplete or missed [read: page image p4]: filed Feb 8, 1988 (appl. 153,070), a continuation of an abandoned application Ser. No. 855,607 of 23 Apr 1986; "10 claims, 4 drawing sheets"; abstract says the oscillations are "sustained and amplified" in a "high-Q LC tank", with the inductor wound on a core made of a mixture of three radioactive nuclides "which decay primarily by alpha emission". So the patented device is an alpha-emitter-core design, while the 1990 paper's experiments list Sr-90 and Kr-85 (beta) sources; Step 1 treated them as one device.
4. Figures were not "garbled"; they are legible when rendered [read: page image p4, p5]. Contents Step 1 missed: Fig. 6 gives the underdamping condition R < 2 sqrt(L/C) and dU/dt = -i^2 R; Fig. 7 gives "Beta flux >= i^2 R" (Brown's own power condition: decay power must at least cover the resistive loss); Fig. 8 shows the load at "50 KHz"; Fig. 9 start-up (10 kV supply, Class C amplifier, spectrum analyzer); Fig. 11 shows an Sr-90 source beside a Be-Cu coil with a silver element; Figs. 10 and 12 are photographs of a prototype and a "RADIOACTIVE MATERIALS" crate on a lab bench. Fig. 4 lists as mechanisms "flux cutting" and "collapsing magnetic field", "reduced resistance & increased conductivity" (see Claim 2).
5. Wired item has a byline: Charles Platt [read: Wired PDF p.4]. Step 1 said no byline. The item also says Brown had a patent pending, had 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. Step 1 omitted these. The article's beryllium-to-carbon remark is a real reaction (alpha on Be-9 gives C-12 plus a neutron) [memory] high, so "loosely says" was too harsh; what is loose is "with heat as a by-product".
6. 0.25 W/cm3: Step 1 ties this to Kr-85. The paper's sentence ("Preliminary data suggests that energy densities on the order of .25 watt per cubic centimeter is achievable") follows the Kr-85 paragraph but names no isotope, does not say whether the watts are thermal or electrical, and uses "energy density" for a quantity with units of power density [read: p3 image]. Claim 3 in claims.md should carry that ambiguity.
7. Step 1's how_read: "full text" for the bundle conflicts with text_available: partial; the patent and figures were not text-readable. Minor.
8. Brown's date of death, 7 Apr 2002 (car accident, Boise), and the company's statement that "the fundamental mechanism works on the laboratory scale" are in https://www.power-eng.com/nuclear/nuclear-solutions-inc-regrets-loss-of-dr-paul-m-brown/ [web]. Not in Step 1.

Step 1 otherwise reads accurately. The CV was read by me and matches (the OCR header reads "Dr. Paul M. Drown", an OCR slip). The CV credentials (EuroTechnical Research University, "Association of Distinguished American Scientists", Who's Who entries) are self-reported and I did not verify them.

---

Shared calculations used below

All constants: 1 Ci = 3.7e10 decays/s; 1 eV = 1.602e-19 J.

Power available from a source (upper bound, 100 percent conversion). - Kr-85: mean beta energy 251 keV, half-life 10.76 y, about 400 Ci/g [web: https://en.wikipedia.org/wiki/Krypton-85 via search]. Per curie: 3.7e10 x 251e3 x 1.602e-19 = 1.5 mW per Ci. Per gram: 400 x 1.5 mW = 0.60 W/g. - Sr-90 plus its daughter Y-90, roughly 1.13 MeV mean beta energy per decay pair [memory] medium: about 6.7 mW per Ci. - Radium-226 needle plus daughters, uranium and thorium (patent example): see Claim 1.

Decay power cannot be exceeded by output power. This is energy conservation, and Brown himself states it: "Energy is contributed by the ion generator" and Fig. 7 "Beta flux >= i^2 R" [read: p2, p5 image].

---

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

a. Mathematics internally consistent? The circuit math he shows is standard and correct: resonant frequency f = 1/(2 pi sqrt(LC)) (L inductance, C capacitance) and the underdamped condition R < 2 sqrt(L/C) with loss dU/dt = -i^2 R [read: page image p4-5]. Resonance cancels the reactances so that only R limits the current, which is true but it only stores energy; it creates none. The inconsistency is between the circuit statement (Fig. 7: beta flux >= i^2 R) and the output claim. If the decay power must cover i^2 R, then the useful output is bounded by the decay power. That bound is milliwatts for any source in the paper (below). The text never gives L, C, R, Q, frequency (except "50 kHz" on a figure), or the source activity, so no check of his own numbers is possible from the folder.

b. Consistent with experiment? Energy conservation and decades of radioisotope-generator work (SNAP-27 RTG appears on Brown's own Fig. 2) say no. The patent's one quantitative statement is: 1 mCi radium needle + 200 g uranium + 100 g thorium, "continuous output of 23 amperes at 400 volts" at 86 kHz [web: https://patents.google.com/patent/US4835433A/en]. Output power if read as 23 A x 400 V: 9,200 W. Decay power available (my calculation): - Ra-226 1 mCi: 3.7e7 decays/s. Chain of four alphas (about 4.8, 5.5, 6.0, 7.7 MeV) is about 24 MeV; with beta/gamma extras multiply by 1.5 as a generous cap: about 2e-4 W. - 200 g U (natural, 2.5e6 Bq of U-238 plus equal U-234) and 100 g Th-232 (4.1e5 Bq), counting a generous full alpha chain with a 1.5 factor: about 2e-5 W and 4e-6 W. - Total at most about 2.4e-4 W, even assuming radioactive equilibrium and total absorption in the conductor. - Ratio claimed/available: about 4e7. Even if "23 A, 400 V" is a peak or a reactive circulating value (volt-amperes in the tank rather than real power to a load), the patent says "into a resistance load" in the fetched text. A tank can circulate large reactive current at low real power, which is the only innocent reading; the statement as written is not that. Caveat: this figure was read through a web summariser. If the patent text differs, the conclusion would change only if real power were under about a milliwatt. Beta-voltaic effect proper (radiation-generated electron-hole pairs collected across a junction) is real and used in commercial cells: efficiencies of about 0.3 to 9 percent, power density micro- to nanowatts [web: https://iopscience.iop.org/article/10.1149/2162-8777/abe423 and search summary on Ni-63 cells].

c. Tested? By Brown, Nucell and "an independent nuclear engineering firm" (unnamed) [read: p3]. The 1990 paper admits the early prototype "generated electricity" but showed "a frequency stability problem" and "material degradation", and promises "engineering data in the coming months" [read: p3]. Brown's cited 1991 paper "Current status and future research with resonant nuclear generators" (26th Intersociety Energy Conversion Engineering Conference, per CV) [read: p6] is not in the folder. I found no independent replication, no published input-output energy balance, and no neutral lab test. Verdict category: Brown's own tests exist but are unreported; no controlled neutral test is on record. It was not dismissed without testing so much as never tested independently, but because the prediction is excluded by energy conservation, no test is needed to settle the strong form.

d. If true: Energy conservation (first law) would fail: a device outputting about 9 kW from a 0.2 mW source would be a free-energy machine. The mainstream would not need to change if the effect is only a small, real enhancement of conduction.

Strongest honest case. (1) Brown states the right power criterion himself (Fig. 7, beta flux >= i^2 R) and his paper mentions a few watts at best in the market section (10 to 5,000 mW), so the paper's own use case needs only milliwatts to watts, not kilowatts. (2) A tank with a radioactive coil can in principle sustain oscillation if the decay power exceeds the loss, and a very high-Q tank has very small loss; whether decay-induced noise or conduction could sustain a clean oscillation at microwatt level is a legitimate, testable engineering question. (3) Brown reports operating hardware and photographs (Fig. 10 to 12), and the patent examiner granted it (granting does not test operation; US practice does not require a working model). Against Brown: The patent number (23 A, 400 V) is a factor of ten million past the energy budget. Source-agnostic "all worked" (Cs, Sr, Ra, Kr, tritium, Pm) for a mechanism proportional to activity is not credible when tritium and promethium betas are 5 to 60 keV; no figure of merit is given for any. The mechanism text ("collapsing magnetic field" from a stopped particle, Fig. 4) is not physics: an alpha particle of 5 MeV carries a magnetic moment effect of order picojoules at best, and no field collapse transfers the particle's energy to the circuit beyond its ionization energy.

Verdict: Split. Strong form (net output beyond the decay power, kilowatts from millicuries): Refuted by calculation (about 4e7 shortfall by the patent's own example; first law). Weak form (a radioactive coil adding microwatts to a tank): Never properly tested (no neutral, controlled test on record), and physically unremarkable.

---

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

a. Math consistent? No equations are given for the effect, only a count: one Sr-90 beta in copper makes about 80,000 ions in 0.030 inch [read: p2]. Check: the Y-90 beta reaches 2.28 MeV; at about 25 to 30 eV per ionization that gives about 8e4 ion pairs, so the number is plausible [memory] medium. The internal tension: the paper says "the energy contributed ... does not come from the ions themselves but rather from the work done ... to generate the ions" and that neither field nor electrodes contribute energy [read: p2]. So power gain is limited to the decay power, which contradicts calling it "amplification" of load current. "Current amplification" in the sense of more carriers in a leakage path is one thing; a gain of the current in a wire is another. Also "current is directly proportional to the voltage" is just Ohm's law, which he lists as a symptom of the effect.

b. Consistent with experiment? Radiation-induced conductivity in insulators, gases and semiconductors is well established [memory] high, and is the physical basis of ion chambers and betavoltaics. In a metal it is negligible. My estimate: 1 Ci of Sr-90/Y-90 (about 2 betas per Sr-90 decay) fully inside a 1 cm3 copper wire makes 3.7e10 x 2 x 8e4 = 6e15 excitations per second; with a carrier relaxation time near 1e-13 s [memory] medium (electron-lattice thermalisation in copper is of order 1e-13 to 1e-12 s), the standing excess is about 6e2 to 6e3 electrons, against 8.5e22 conduction electrons in 1 cm3 of copper: fractional change about 1e-19 to 1e-20. So "reduced resistance, increased conductivity" in a copper wire cannot be measured. The Burke patents (US 3,409,820 of 1968 and 3,530,316 of 1970) are cited in the bundle [read: p4-5 image]; I did not read them and do not know what data they contain. Honest uncertainty: if Burke cell experiments used gas or semiconductor in the current path, a real small effect is expected; the claim as applied to a metal wire is the problem.

c. Tested? No test data in the folder or in anything I found. Not dismissed without testing; simply no data presented.

d. If true: Nothing in mainstream physics would need to change for a tiny conductivity effect; a real large-gain "current amplification" would again violate energy conservation.

Strongest honest case. A radioactive conductor in series with a battery does give a small extra power from the decay, which is the Burke idea and is consistent with the beta-voltaic principle in the sense of direct charge collection. Brown's own stated source of energy (the decay) is correct.

Verdict: Split. (i) Ionisation by radiation changes conduction in insulators and gases: Established. (ii) Measurable resistance lowering and "amplification" of current in a metal conductor: Refuted by calculation (carrier fraction about 1e-20; output limited by decay power of milliwatts). (iii) The explanation by "flux cutting / collapsing magnetic field" (Fig. 4): Internally inconsistent with his own statement that the energy comes from ionisation work.

---

Claim 3. Direct conversion of beta/alpha decay this way reaches about 0.25 W per cm3 with Kr-85, better efficiency, less shielding

a. Math consistent? "Energy density" of 0.25 "watt per cubic centimeter" is a unit slip: watts per volume is power density. No derivation is given: "preliminary data suggests" [read: p3 image]. Budget check (my calculation): Kr-85 gives at most 0.60 W per gram (above). For 0.25 W/cm3 of electrical output at 100 percent conversion you need 0.42 g of Kr-85 per cm3, i.e. 4.9 mol per litre. Ideal gas at 0 C needs about 110 atm of pure Kr-85. For 5 percent conversion (best betavoltaics are a few percent to roughly 9 percent [web: IOP review above]) you need 8 g/cm3, more than liquid krypton (about 2.4 g/cm3 [memory] medium), so impossible. At liquid density, 100 percent conversion gives 1.4 W/cm3 thermal, so 0.25 W/cm3 thermal is arithmetically fine but 0.25 W/cm3 electrical would need an efficiency of at least 18 percent. Also reactor krypton is mostly not Kr-85 (needs isotope enrichment) [memory] medium. Inventory check: 42 MCi of Kr-85 [read: p3] is 42e6 x 1.5 mW = about 62 kW of total beta power, in the entire existing US/world inventory he cites. That is a real number worth the reader's attention: it cannot serve a "greenhouse effect" mitigation (one of his reference titles, [read: p3]) at any scale.

b. Consistent with experiment? The thermal reading is consistent with RTG practice. The electrical reading at the stated efficiency is not consistent with any measured betavoltaic or alpha-voltaic result (microwatt scale per cm3 for Ni-63, up to about 10 microwatts/cm3 [web: search summary above]; Kr-85 is much hotter than Ni-63, so hundreds of times higher power per unit is expected, but not 0.25 W/cm3 electrical).

c. Tested? Brown says "preliminary data" without presenting them. Not independently tested.

d. If true: Nothing, if thermal. If electrical at over 18 percent from betas, it would be a breakthrough in direct conversion but not a change in physics.

Verdict: Not testable as stated (undefined: thermal or electrical, no conversion efficiency, no data); the electrical reading at his implied efficiency: Never properly tested, and by budget would require at least 18 percent conversion at liquid-density pure Kr-85, which is unsupported by any source I found.

---

Claim 4. Gamma irradiation can transmute long-lived waste to short-lived or stable species within months, for under a quarter of the DOE $150 billion estimate (Wired 1999)

a. Math consistent? No numbers in the article beyond cost. Physics check: (gamma, n) photonuclear reactions remove one neutron above a threshold of roughly 8 to 12 MeV [memory] medium. Examples I would check: Cs-137 to Cs-136 (half-life about 13 d, then beta decay to stable Ba-136) and Sr-90 to Sr-89 (about 50 d, then Y-89, stable) are "months" outcomes [memory] medium; Tc-99 to Tc-98 (about 4 million years) is not. So "safe within a few months" would hold for a few nuclides and fail for others, and Wired states it for "nuclear waste" generally. The giant dipole resonance cross-section, which sets the yield, is tens to hundreds of millibarns (1 barn = 1e-24 cm2) over a few MeV width [memory] medium; most photons never hit a nucleus. Doolen's comment that high-energy gamma rays are expensive is therefore correct in direction.

b. Consistent with experiment? Photonuclear transmutation is real and studied: Eremeev (1999) proposes bremsstrahlung gamma/photoneutron transmutation of Sr-90, Cs-137 and actinides and claims economic viability [web: https://www.osti.gov/etdeweb/biblio/20064919], and recent work studies laser-Compton and "Gamma Factory" beams [web: https://arxiv.org/pdf/2409.12473 listed in search; abstract not read by me]. Park (AIP) and Schiffer (Argonne) in Wired agree it is not far-fetched [read: Wired p.3]. This is a legitimate idea; the claim of cost is the issue. No cost model is shown.

c. Tested? Wired says Brown "showed" the effect and would "soon run tests at the University of Illinois or MIT" [read: Wired p.3]. I found no published results of those tests. Nuclear Solutions' 2002 notice says "the fundamental mechanism works on the laboratory scale" with no data [web: power-eng link above]. Not independently replicated as far as I can find. Not dismissed; apparently not funded or published.

d. If true: Nothing in physics would change; it would be an engineering and economics claim.

Strongest honest case. The mechanism is textbook nuclear physics and was proposed independently the same year by Eremeev, so Brown is right that the physics is not exotic. Three named physicists, including a professional debunker, said the idea is not far-fetched. Against: The article cites no cross-section, no beam, no yield, no waste inventory. "Safe" is not defined (daughter nuclides, activation products). Brown's claim that he found no earlier proposal is not supported: photonuclear transmutation papers predate 1999 [memory] medium (the Eremeev paper is dated 1999 per OSTI; earlier literature not verified).

Verdict: Split. Physics of (gamma, n) transmutation of some fission products: Plausible, untested at the scale claimed (active research). "All waste safe within months": Not supported for nuclides like Tc-99 (memory, medium); as a general claim Never properly tested. Cost "under a quarter of $150 billion": Not testable as stated (no cost model).

---

Claim 5. Excess heat from gamma transmutation generates more electricity than the plant uses; a plant costs about $5 million

a. Math consistent? (gamma, n) on a long-lived fission product is endothermic: it costs at least the neutron separation energy, about 8 to 12 MeV per atom [memory] medium. Compare the decay energy released if the atom simply decayed: Cs-137 gives about 1.2 MeV (beta plus gamma) [memory] medium. So each transmutation spends about 8 times more energy than the nuclide would ever release, before counting beam inefficiency (most gamma photons miss, and bremsstrahlung converts electrons into useful photons with low yield [memory] low). All beam energy ends up as heat, but converting that heat to electricity at about 35 percent efficiency can return at most about a third of the electricity put in, so net electricity cannot exceed what the plant consumes. The only way to get net heat is a multiplying process (fission of actinides, as in accelerator-driven subcritical reactors), which Nuclear Solutions later described as "HYPERCON ADS" [web: power-eng link]; that is a different device from "textbook radiochemistry". This is the energy-budget argument Schiffer was pointing to: the "enormous excess heat" he mentions is the deposited beam energy, not a surplus.

b. Consistent with experiment? Accelerator-driven systems with fission multiplication do make net energy; photon-only transmutation does not [memory] medium. The $5 million figure is set against a $1.3 billion "neutron-beam research project" [read: Wired p.3]; a gamma beam of 10+ MeV at useful intensity needs an electron accelerator of megawatt class, whose cost is far higher [memory] low. I could not verify cost data on the web within budget.

c. Tested? No data. Not independently tested. A pilot plant was being negotiated with the DOE (Wired). No result found.

d. If true: Nothing, if it uses fission multiplication; the first law, if it does not.

Strongest honest case. If the waste contains fissionable actinides, photo-fission can add heat, and a subcritical system could in principle yield net electricity. Brown's description does not say so.

Verdict: Split. Photon-only net-electricity claim: Refuted by calculation (endothermic reaction, threshold 8 to 12 MeV per atom versus about 1 MeV released by the decay it replaces; thermal-to-electric below 100 percent). With fission multiplication: Plausible, untested at the stated cost. $5 million plant: Not testable as stated (no cost basis).

---

Rejection and handling

• Was the work tested and rejected? No. I found no controlled negative experiment on either project. The 1990 resonant device and the 1999 transmutation plant were never independently tested as far as the sources show. Category: never properly tested, not "tested and failed", and not "dismissed without testing" either, since no source shows anyone formally dismissing it. The one dismissal-adjacent detail is a patent example whose claimed output is about 4e7 times the available decay power; a patent examiner did not catch this, which says patent grant is no evidence of operation.
• Publication path. Brown published in magazines and conference talks (Magnets In Your Future, Raum & Zeit, ANS meetings, an IECEC paper) [read: p3, p6], not in a journal with refereed data; I found no blocked publication or funding story.
• Researcher rigour. Positives: Brown states the correct power criterion (Fig. 7) and the right energy source (ionization work); the betavoltaic principle and photonuclear transmutation are real physics. Negatives: unit slip (W/cm3 as "energy density"); "amplification" language that conflicts with his own energy statement; no measurement table in any folder document; one patent example that fails the energy budget by seven orders of magnitude; claims "all sources worked" including tritium and promethium; CV padding (Who's Who, "Association of Distinguished American Scientists", unverified).
• What would settle it. (1) Claim 1: put a calibrated 1 mCi radium source and a high-Q tank on a bench, measure real power (RMS voltage and current with phase) into a resistive load for 24 hours; expected result microwatts at most. (2) Claim 4: a published (gamma, n) irradiation of a Cs-137 or Sr-90 sample with measured Cs-136 / Sr-89 yield and the electrical energy per transmuted atom.

Summary table