Popa-Simil: Science Contrarian review (Step 2)

Reviewer role: judge by experimental evidence, build the strongest honest case, examine how the idea was handled, and apply the same rigour to the inventor. Sources tagged [read] / [web] / [memory] (confidence). Web checks: I relied on the Investigator's saved notes and downloaded files (research/ and research/files/), which I re-read in part; I did not run new web searches, so items that come only from those notes are tagged [read: research note] with the original URL.

Source base: one 4-page conference paper (L. Popa-Simil and I.L. Popa-Simil, "Nano-Hetero Structure for direct energy conversion", NSTI-Nanotech 2007, vol. 4, pp. 614-617) [read]. No source gap for that paper. Everything beyond it (patents, Russian experiments, later papers) comes from the Investigator's files.

Corrections to Step 1

I re-rendered p.2 at 200 dpi and read Eq.1, Fig.1 and Fig.2 myself. 1. Eq.1 transcription in deep/ is confirmed [read: page image p.2]: the log argument is printed "2mc beta^2 / (1 - beta^2)" with c to the first power, the Barkas term is printed with a minus sign, and the Bloch term is printed Z2^2 L2 although the text calls it "Z2L2". The text also says "Bohr's electron radius" for r0, which should be the classical electron radius. 2. Layer thicknesses and the "ion loses about 9%" finding are consistent with what I see in Fig.2 (plateaus between about 700 and 2600 eV/Angstrom over 6000 Angstrom). I did not re-measure pixels. The step-1 figure is credible but is a hand reading, accurate to roughly 15%. 3. Claims.md omits two things in the paper that bear on the verdicts. (a) The paper itself says on p.2: "Finally, the energy is mainly transferred to the electrons in the lattice and nuclei and after secondary interactions dominated by the electrons, the whole energy becomes heat" [read: p.2]. That is the paper's own description of the unmodified case, and it sits beside the ">80%" claim without a calculation linking them. (b) The paper's own insulator-voltage statement: the parallel connection gives voltages "in the mV domain, given by the breakdown voltage admitted by the insulators" [read: p.2]. This is the number that decides claim 1 (see below). 4. The Investigator's lead that the patent's 84% is "simply the efficiency of ionization energy stopped in the C layer" is correct and I read it myself: US20100061503A1 para. 0121 says it "is simply the efficiency of ionization energy stopped in the 'C' layer", must be obtained by "experimental buildup", and "remains upper conversion efficiency estimator ... only because it did not includes the rest of the aspects in the device" [read: research/files/US20100061503A1.pdf]. So the inventor himself says 84% is an upper bound on a different quantity. The 2007 abstract's "may be higher than 80%" is worded as possibility, not result [read: p.1]. Claims.md states claim 1 as "can convert ... at over 80%"; that is slightly stronger than the paper's "may be". 5. The sibling Nanotech 2007 paper (3021) says "efficiencies higher than 90%" from "quantum effects" [read: research/files/Nanotech2007v4-3021.pdf, grep only; I did not deep-read it]. This is a further statement of the same unmeasured kind.

No factual error in Step 1 changes a verdict.

Evidence base in one paragraph

The paper contains no measurement. Its only quantitative content is a SRIM-2003 (Monte Carlo ion-range code) run of a 100 MeV Cs ion through an eleven-layer stack. The ion leaves with about 90% of its energy [read: deep/, my spot check agrees]. There is no calculation of how many electrons cross the insulators, at what energy, or what voltage they sustain. Everything about conversion efficiency is asserted.

Claim 1: a CIci nano-stack can convert fission-product kinetic energy to electricity at over 80%

a. Mathematics. The only equation (Eq.1) has four errors (above) and is not what SRIM uses at this velocity (beam is 0.75 MeV/u, only partly stripped) [read: deep/]. No efficiency formula is given. So the 80% is not derived. The number matches a known ceiling: fission fragments carry about 168 of about 200 MeV per fission, about 84% [memory, high]. The 2026 MDPI review uses the same 0.85 as an "ideal upper bound" and says "Neither value represents demonstrated commercial battery performance" [read: research/2026-10-04-mdpi.md; web https://www.mdpi.com/2227-9717/14/17/2834]. Popa-Simil's 84% and that bound are the same kind of number: the share of energy available, not the share delivered.

My own energy budget, which the paper lacks. Take one 100 MeV fragment. At roughly 30 eV per ionisation, it creates about 3x10^6 electron-hole events. Even if every freed electron were moved across an insulator and collected, the work delivered equals charge times voltage. At the paper's own millivolt scale (10 mV) that is 3x10^6 x 0.01 eV = 3x10^4 eV, or 3x10^-4 of the 100 MeV, i.e. 0.03%. To reach 80% the collected electrons would have to be held against a potential comparable to their birth energies (tens to hundreds of eV), not millivolts. In other words, with the paper's mV output the efficiency is set by e x V per electron, and 80% is not available unless a large fraction of electrons does work against a potential of order 100 V per stack or more through series connection. The paper offers series connection "for domains where the moving particles field is constant at micrometric level" [read: p.2-3], but gives no numbers. This is a rough order-of-magnitude argument, not a refutation: the ionisation cost per pair, the electron energy spectrum (delta rays up to about 1.6 keV at this velocity [read: deep/]) and series stacking could change it by factors of 10-1000. It shows the 80% needs an explicit calculation the paper never does.

b. Experiments. Supporting measurements exist but are far lower: - Anufrienko, Kovalev, Kulikov, Chernov (2006) and Anufrienko, Chernov, Mikhailov et al. (2008), reported "a conversion efficiency of 25%" in plane layered structures with U-235. This is documented in one published secondary source (Klassen et al., InTech 2013, p.272-273) [read: research/2026-10-04-issp-intech-klassen.md]; the primary papers are in Russian and were not read. It is the most important evidence for the principle and sits a factor of about 3 below the claim. I do not know what "efficiency" is defined as there (fraction of fragment energy, or of electron energy, or at what load). - Yoshida, Tanabe, Chen, US20050077876A1: laminated SUS304/Al with insulators, measured efficiency about 10^-5 % per Klassen [read: research note]. Thick (mm-scale) layers. - Klassen/Yershov et al. report "direct, sufficiently efficient conversion" in triple-layer structures, and that thinning the insulator from 200 um to 1 um raised efficiency "by 3 orders" [read: p.255-256, 272]. This supports the thin-insulator trend, with no absolute number in the passage read. - Fission electric cells in vacuum-gap form (Safonov RAND 1957, Krieve JPL 1966, Ursu et al. Bucharest 1978-88) are related work. I know them only through the Investigator's list; I have not read their reported efficiencies and will not assert any [memory, low]. - Stepanov et al. 2018 measured 1.15 V on asymmetric supercapacitors under neutron irradiation, a different device [read: research note]. Nothing found shows a Popa-Simil-built CIci structure measured at all.

c. Tested? The inventor's concept has not been built and measured, as far as the record shows. Related structures were tested by others (Russia, Japan, Romania) with efficiencies from 10^-5 % to 25%. It is not a case of "tested and failed": the 10^-5 % result is from deliberately non-optimised thick layers, and the nano-scale versions appear to improve by orders of magnitude. Replication of a >80% figure: none, since no such figure was measured.

d. If true: no law is broken. Fragment kinetic energy is 84% of fission energy; direct collection of charge carried by that energy is allowed in principle (the second law constrains heat engines, not electrostatic collection). The change needed would be in engineering assumptions (that most reactor energy must pass through a thermal cycle), not in physics. What would have to hold: most ionisation energy ends up in electrons energetic enough to cross a nm insulator and do work against a usable voltage.

Verdict: split. - Principle (charge separation between dissimilar conductors by radiation, increasing as insulators thin): Plausible, untested as a designed device, with partial experimental support from others (up to 25% reported in one source). - Numerical claim of over 80%: Never properly tested. It is not derived in the paper and the inventor's own patent calls 84% an upper estimator. As written, the abstract's "may be higher than 80%" is Not testable as stated until a voltage, load and definition of efficiency are given.

Claim 2: mechanism, electrons from the high-yield conductor cross a nm insulator and charge the low-yield collector

a. The qualitative picture (high-Z emitter gives more secondary electrons than low-Z collector; net current flows) is the standard secondary-electron-emission asymmetry used in betavoltaic and "emission" nuclear batteries. Internal inconsistency in details: the paper says the knock-on electrons "discharge their energy in showers down to energies of several eV" [read: p.2], while the same page says the insulator must pass electrons of "few eV". Electrons at a few eV carry only a few eV of work, so the cascade is the limit on output voltage. The "10 keV/nm" and "about 1200 electrons per nm^3" figures are loosely right: Fig.2 gives 7-26 keV/nm and 1200 electrons/nm^3 fits iodine, but U and Au have about 4400-4700 [read: deep/]. The mV insulator limit is also odd: ordinary oxide breakdown is about 1 V per nm [memory, medium], so "breakdown voltage of few milivolts" (Cleantech 2007 paper, per research note) would mean either leaky tunnelling films or a loose use of "breakdown". This should be asked of the author.

b. Experiments: Klassen et al. (insulator thinning gives 3 orders), Anufrienko (25%), Stepanov (charge separation, mV to V under neutron dose) all fit the mechanism [read: research notes].

c. Tested by others with positive but modest results; not replicated by an independent group on the Popa-Simil design. The Russian group cites Popa-Simil (refs 17-19) and works along the same line, so they are not fully independent of him in concept, although their experiments are their own.

d. Nothing in mainstream physics would change.

Verdict: Plausible, untested for the specific nanolayer design; the general mechanism has experimental support from others (low to moderate efficiency).

Claim 3: a reactor harvesting fission energy this way would run cold

a. Energy conservation gives a trivial version: every joule exported as electricity is a joule not left as heat. But "cold" cannot be more than the fraction harvested. Fission energy not in fragment kinetic energy (about 16%: neutrons, prompt and decay gammas, decay heat) [memory, high] still heats the structure. The paper itself states that without conversion "the whole energy becomes heat" [read: p.2]. And any electron that does not cross the barrier, any back-flow and the electron-lattice interaction it mentions (p.3) still heat the lattice. b. Experiments: no measured reactor or heat balance. c. Not tested. d. No physics changes. Verdict: Not testable as stated (no quantity of "cold" is given). Weak form (heat falls in proportion to the electrical fraction extracted): Plausible, untested, and it depends entirely on claim 1.

Claim 4: sub-10 mV modules stacked and raised to grid voltage via DC/AC and AC/AC cascades, possibly with Josephson and Andreev effects

a. Power is conserved by a converter, so stepping up voltage lowers current, which is fine. But the practical gap is stated by the claim itself: converters need a start-up voltage. Silicon junctions need about 0.6 V and MEMS switches need volts [memory, medium], so an mV source has to be stacked first (hundreds of layers in series) before any converter works. Josephson junctions are voltage standards and detectors at microvolt to mV levels; they do not multiply power, and Andreev reflection is a charge-transfer process at a superconductor-normal interface. The paper says they "may be used to increase the conversion efficiency" [read: p.3] without any equation. I treat this as speculation presented as design. b-c. No relevant experiment found. d. Nothing would change. Verdict: Never properly tested; the Josephson and Andreev part is Not testable as stated (no mechanism or number).

Claim 5: superconducting actinide compounds (PuCoGa5, Tc about 18.5 K; PuCoRh5, about 8.7 K) as harvesting structure, with radiation damage self-repaired

a. The material data are real: PuCoGa5 is a Pu-based superconductor discovered in 2002 (Sarrao et al.) with Tc about 18.5 K [memory, high]; PuCoGa5's Tc has been seen to fall with self-irradiation damage in later studies [memory, medium; I did not check]. The paper cites the "Pu-Based Superconductor PuCoGa5 Probed" reference [read: p.4 references]. The stated Tc values of 18.5 K and 8.7 K are the low-temperature ones: the structure would have to sit below about 18 K while sitting in a reactor core with fission power density. The paper's own thermal plan ("run cold") would have to remove heat at cryogenic temperature, where the Carnot cost is large. Self-repair of damage "over the reactor lifetime" is asserted. Fission-fragment damage in fuel is intense (each fragment displaces thousands of atoms [memory, medium]); "self-repair" would need to be shown. b. Experiments: Pu compound superconductivity measured, yes; as a reactor harvesting structure, none. c. Not tested. d. No physics changes; engineering viability would be. Verdict: Plausible, untested for the materials fact; the reactor-lifetime self-repair and the cryogenic operation: Never properly tested, with real doubts raised by the energy budget at 18 K.

Claim 6: fuel power density from 0.2-0.5 kW/cm^3 to a theoretical 5000 kW/cm^3

a. I computed what 5000 kW/cm^3 (5 MW/cm^3) requires for pure U-235 metal. Atom density about 4.8x10^22 /cm^3, thermal fission cross-section about 585 barn, so macroscopic Sigma about 28 /cm. At 200 MeV per fission, 5x10^6 W/cm^3 needs 1.6x10^17 fissions/cm^3/s, i.e. a thermal neutron flux of about 5.6x10^15 n/cm^2/s [computed; cross-section and densities memory, medium]. For comparison, the highest-flux research reactors run at a few x10^15 [memory, medium]. So the figure is not forbidden by an energy budget; it needs a neutron flux at the limit of existing machines, in a core of pure fissile metal, and cooling of 5 MW per cm^3 by electrical extraction only. Conventional values of 0.2-0.5 kW/cm^3 correspond to flux of order 10^11 n/cm^2/s in that same pure-fuel case, i.e. the quoted 10^4 gain is plain scaling with flux, not specific to the CIci structure. The paper gives no derivation; the TMS 2008 paper reportedly says "up to 1,000 times" instead (Investigator lead, [web] laro.lanl.gov; not read). The two figures differ by 10x without explanation. b-c. No test. d. Nothing would change in physics. Verdict: Not testable as stated (it is called "theoretical" with no derivation); the calculation shows it is not forbidden but requires extreme neutron flux, and the 5000 vs 1000 inconsistency between papers is an unexplained discrepancy.

How was the work handled? (record, without verdict)

• Patents: US20100061503A1 (2009 filing, continuation of a 2006 application) and US20130121449A1 (2011) are both listed as abandoned [read: research note]. Whether by examiner rejection, fee non-payment or choice is unknown; the USPTO file wrappers were not reached. US8067758B2 (shielding, with Muntele) was granted and has expired for fees.
• Publications: 2007-2021 sequence of conference papers and a Physics Procedia (2011) chapter; no peer-reviewed journal report of a measurement of a built CIci device was found. The 2012 roadmap asks for "appropriate funding, equipment and support" [read: research note]. The record is consistent with an unfunded concept (no fabrication or beam time) and does not by itself show suppression or negative results. Competing explanations (no funding; results unpublished; failed tests not reported; no ability to fabricate nm stacks with actinides) cannot be told apart from the record. Decisive records: SBIR/DOE/Army award lists for LAVM LLC, beam-time logs at the Alabama irradiation centre, and the file wrappers.
• Mainstream reception: a 2026 review calls the technology "mainly conceptual ... patent-stage devices" [read: research note, web URL above]. That is the reviewers' characterisation, not a test result. No hostile or controlled test of the inventor's design is on record, so nothing was "tested and failed" and nothing was dismissed after a test.
• Interested parties: the inventor owns LAVM LLC and the patents, so his claims of efficiency are the claims of a party with a commercial stake. The Russian group (Chernov, Anufrienko) has its own related programme and cites him; their results are their own.

Strongest honest case for the claims

1. Charge separation by dissimilar-Z layers under radiation is a measured effect in several labs; the best reported number (25%) is three orders above the macroscopic precursor and thin insulators were seen to give a 1000-fold gain.
2. 84% of fission energy is released as fragment kinetic energy, so there is an energy pool large enough, and ionisation is the dominant channel in Fig.2.
3. The idea sits in a long lineage (fission electric cells since 1957) and recent reviews still treat it as a live concept.

Where the inventor falls short (same standard applied)

• Eq.1 has four errors and is not the formula behind the figures.
• Text and figures disagree on layer thicknesses (about 10 nm thinner in the figures for layers 2-7), and the stack total (700 nm) exceeds the plot (600 nm) [read: deep/].
• The simulated ion loses about 9% of its energy, yet the caption says "stopping", and no calculation links the SRIM run to electrons crossing insulators.
• ">80%" and ">90%" efficiencies in two 2007 papers are not derived, and the patent calls 84% only an upper estimator.
• Power-density figure differs between papers (5000 kW/cm^3 vs "up to 1000 times").
• No error bars, no definition of efficiency, no load voltage.

Verdict summary

1. 80% efficiency: principle Plausible, untested (partial support, 25% reported by others); the >80% figure Never properly tested and not derived.

2. Mechanism: Plausible, untested for this design; general effect supported by others' measurements.
3. Runs cold: Not testable as stated; weak form Plausible, untested.
4. mV modules stacked to grid voltage: Never properly tested; Josephson/Andreev part Not testable as stated.
5. Superconducting actinide harvesters: Plausible, untested (materials real); lifetime self-repair Never properly tested.
6. 5000 kW/cm^3: Not testable as stated; not forbidden by energy budget but needs about 5x10^15 n/cm^2/s flux.

What would settle it: a beam test of a stack (as in US20100061503A1 FIG.15) reporting open-circuit voltage, load curve, and current per incident fragment, giving electrical energy out over beam energy in; plus the primary Anufrienko 2008 paper to learn how its 25% was defined.