Physicist review (Step 2): Popa-Simil, "Nano-Hetero Structure for direct energy conversion" (NSTI-Nanotech 2007)

Reviewer role: PhD physicist. Sources read: facts.json, claims.md, deep/ read of p.2, the full extracted text of the paper (4 pages) [read], the Investigator's sites.md leads and research notes, and the US20100061503A1 patent text (para. 0121, 0156) [read]. Web checks: 1 (fission fragment energies). No source gap: the paper has a full text layer.

Corrections to Step 1

1. Deep read says typical heavy fission fragments carry "about 60-80 MeV" and that 100 MeV Cs "is at or above the top of the range". Refinement: light fragments (A about 95-100) carry about 100 MeV; heavy ones (A about 135-140) about 67 MeV [web: https://www.sciencedirect.com/topics/engineering/fission-fragment ; search summary, kinetic energy about 100 MeV for light fragments, falling for heavy; total about 169 MeV]. Cs-133 is a heavy-group nucleus, so 100 MeV is about 1.5 times the energy a Cs fragment would normally have. The simulation is a slightly mismatched but order-of-magnitude fair example. The deep read's "about 10x range" of the stack is unaffected.
2. Claims.md says the simulated ion loses "about 9%" of its energy. Consistent with the deep read (about 8.7 MeV of 100 MeV). I did not re-measure the figures; I rely on the deep read's pixel measurements [read: deep/, page image p.2]. The comparison table in the deep read (Eq.1 with Z1 = 55 gives about 10 times SRIM) I checked for logic, not recomputed.
3. Investigator lead on "84%" (US20100061503A1): I read para. 0121: "The calculated efficiency of 84% is simply the efficiency of ionization energy stopped in the 'C' layer. To be a real accurate evaluation it has to consider all the electrons and excitations inside the material ... have to be obtained by experimental buildup" [read]. So the patent itself says 84% is not an electrical efficiency. Note that 168/200 = 0.84 is also the fraction of fission energy carried by fragments, which is a ceiling on any fragment-harvesting scheme, not a conversion efficiency [memory, high; web as above]. Claims.md claim 1 ("over 80% efficiency") should be read as "the paper asserts a possible efficiency above 80%"; it is not a measured or derived electrical figure anywhere in the paper.
4. Claims.md has no claim for the missing electrical analysis. I treat claim 1 and claim 2 below as needing an energy-per-electron budget, which neither the paper nor the facts file supplies.

Mainstream view, stated fairly

Direct conversion of fission-fragment kinetic energy to electricity is a real, old idea. The fission electric cell (FEC) collects the charged fragments themselves across a vacuum gap and reaches high voltages, since a fragment leaves with tens of MeV of kinetic energy and charge of order +20e. FECs were studied by Safonov (RAND, 1957), Krieve (JPL, 1966) and Ursu (Romania, 1980s), with small measured currents and limited efficiencies [web: sites.md leads; I did not open the primary documents]. Secondary-electron-emission batteries (Young et al., US5861701) are also real and give volts per cell. Mainstream science does not call the idea impossible. It says: (i) the fraction of fragment energy that ends up as electrical work depends on the voltage at which charge is collected, (ii) in a solid, most stopping energy goes to heat/excitations within a few nm of the track, and (iii) no device of this kind has shown a high efficiency. The most recent mainstream review (Gabbar et al. 2026) calls the CIci concept "mainly conceptual ... patent-stage" and uses 0.85 only as an ideal ceiling [web: https://www.mdpi.com/2227-9717/14/17/2834, characterization by the review authors, who have no stake in Popa-Simil's patents as far as I know].

Claim 1: CIci structure converts fission-product kinetic energy to electricity at over 80%

a. Mathematics: There is no derivation of efficiency in the paper. The only equation (Eq.1, stopping power) has errors: r0 mislabelled "Bohr's" radius (classical electron radius is meant), a missing c in the logarithm (dimensionally inconsistent), the Barkas term sign reversed, the Bloch term using the target charge Z2 instead of the projectile charge Z1, and no atom density or effective charge. These are documented in the deep read and I accept them [read: page image p.2]. They do not touch the central claim, because Eq.1 is not used to get any efficiency, but they show the paper's formulas were not checked.

Energy-per-electron budget (my calculation, [memory, high] for the physics, with generous assumptions). Electrical work equals the charge moved times the voltage it is moved across. An 80 MeV fragment makes roughly one electron-hole pair per W = 10-30 eV of deposited energy (W of order 25-30 eV in most solids and gases; a generous 10 eV is used for metals). Even if every pair gave one net transferred electron, the work delivered at the cell voltage V is N e V, so efficiency <= e V / W. With the paper's own statement that "maximum voltages will be less than 10 mV" [read: p.3] and the parallel-connection voltage "in the mV domain, given by the breakdown voltage" [read: p.2]: - V = 0.01 V, W = 10 eV: efficiency <= 0.1%. - V = 0.01 V, W = 25 eV: efficiency <= 0.04%. To reach 80% at 10 mV, each transferred electron would have to carry about 8 V worth of the lost energy per 10 mV, impossible: the energy of the shower electrons (few eV) is thermalised in the collector, and only e V per electron is available as work. The 84% in the patent, as noted, is just the share of energy stopped in layer "C", not electrical.

Series stacking does not rescue the bound: each ion crosses all cells, and per-cell work is still e V_cell times the electrons transferred in that cell; the sum remains bounded by the same ratio. Voltage multiplication by DC/AC and AC/AC converters (paper p.3) trades current for voltage and cannot raise power; it only loses a little.

Verdict on claim 1: the strong form (>80% electrical efficiency at mV output) is Refuted by calculation, as stated in the paper's own voltage limit. A weak form (some small, measurable electrical conversion in a layered heavy-conductor / insulator / light-conductor structure) is Plausible, untested for CIci proper; see tests below.

b. Experimental consistency: Measured direct-conversion efficiencies in layered structures are small: about 10^-5 % in the Yoshida/Tanabe millimetre-scale stack, and 25% reported by Anufrienko et al. for planar U-235 layered structures [read: research/2026-10-04-issp-intech-klassen.md, p.271-273 of the Klassen chapter]. The Anufrienko figure is documented in one published source, second hand (the chapter by Klassen et al., who are working on the same principle, so an interested group); the primary Russian papers (2006, 2008) have not been read. I do not dismiss it. It is unexplained against my budget above: a 25% efficiency would require either a high collection voltage or a definition of efficiency different from electrical output over total deposited energy (for example, fraction of the secondary electron flux collected, or an efficiency relative to the electron energy rather than fragment energy). I cannot decide which without the primary papers. This is the single most important open lead in the project. c. Tested? The paper presents only a SRIM-2003 simulation, no measurement [read]. The patent describes a beam-test setup (FIG. 15) but no results were found by the Investigator. Independent work (Anufrienko; Yershov/Klassen at ISSP RAS, 2011-2013) tested related triple-layer structures; Klassen et al. report that thinning the insulator from 200 um to 1 um raised efficiency "by 3 orders" [read: p.272]. These are independent supports for the design trend, not for 80%. Not replicated for CIci with nm layers under fission fragments. d. If true: the Fermi-level / work-function picture of a cell (volts, bounded by the contact potentials) would have to be circumvented. Specifically a way to extract work at energy per electron comparable to the shower electron energy. This would not break any law but would require a new mechanism the paper does not provide.

Claim 2: mechanism, difference in electron-shower intensity between conductors; electrons cross a nm insulator into a low-yield collector

a. Math: qualitative only; no cross-sections, yields or charge-balance equations. The numbers quoted on p.2 are partly wrong: "about 1200 electrons per nm^3, fewer than 30 atoms" fits iodine and UO2 molecules, not U or Au (about 4400-4700 e/nm^3), and "very few electrons" receive the energy is not true in absolute terms: with T_max about 1.65 keV, at least about 6 primary electrons per nm, realistically hundreds [read: deep read section 5, computed]. b. Consistent with experiment: Differential secondary electron emission between high-Z and low-Z foils is a real, measured effect (the basis of the Young et al. battery and of beta-voltaic / SEE detectors) [memory, high]. The claim that ballistic electron range over a few nm exceeds a thin insulator is consistent with electron mean free path data (a few nm at tens of eV) [memory, medium; textbook ranges]. Dielectric breakdown strength of about 1e7 V/cm over a few nm gives a few tens of mV to a volt, so the paper's "few mV" sets the voltage, as above. c. Tested: partly, by others (Klassen; Anufrienko; Yoshida/Tanabe; the FEC lineage), not by Popa-Simil in this paper. d. If true: nothing in mainstream physics changes; this is ordinary secondary-electron emission and charge transport. What would change is engineering feasibility.

Verdict: Plausible, untested as a power source (the mechanism is established; the energy balance at useful efficiency is not). Independent related tests exist and are kept in the record (Klassen et al. 2013; Yoshida et al. US20050077876; Anufrienko et al.).

Claim 3: reactor would run cold because less energy is left as heat

a. Energy conservation holds: energy extracted as electricity is not heat in the fuel (the electricity is eventually dissipated in the load outside). The qualitative statement is trivially consistent. But the paper's "cold reactor" is only as large as the electrical fraction, which by claim 1's budget would be at most a small fraction. Fragments carry about 84% of fission energy; neutrons, gammas and neutrinos (about 16%) still heat the core and the structure [memory, high; web as above]. Also in a thin-layer stack the 10% energy loss per 600 nm means the fragments must pass about 10 stacks; the stack is not itself a fuel. b/c. No test of a reactor cooled this way. d. Nothing changes. Verdict: Plausible, untested in the weak sense (some reduction in heat, scaling with real efficiency); the implied large effect (a "cold reactor") follows only if claim 1's strong form held, so the strong form is Refuted by calculation by the voltage budget above.

Claim 4: mV modules stacked and raised to grid voltage by DC/AC and AC/AC converter cascades, with Josephson and Andreev effects

a. Math: the Josephson constant K_J = 483 597.879 GHz/V quoted from CODATA is correct [memory, high]. But the Josephson effect converts a DC voltage V into AC at frequency K_J V; at 10 mV this is about 4.8 THz, and the junction carries microamps. It is a frequency standard and a detector, not a power converter. "Andrew effect" is Andreev reflection (the author's spelling); it is a charge-transfer process at a normal-superconductor interface that doubles the charge carried per quasi-particle at sub-gap energy; it does not raise efficiency of a DC supply as described [memory, high]. No equations are given for the converter. b. A DC-DC converter chain at 10 mV input is a known difficulty: semiconductor switches have thresholds of 0.3-0.7 V, so the converters cannot start from 10 mV without a separate bias; thermoelectric harvesting circuits typically start at 20-50 mV with special startup. This is [memory, medium]; I did not run a web check. Power density scaling: at the claimed 5000 kW/cm^3 and 10 mV the current density would be 5e8 A per cm^3 of fuel (calculated), carried by nm-thick conductors; the resistive heating there would defeat the "cold" claim. This is my calculation from P/V. c. Not tested. The paper's Cleantech 2007 companion also gives "breakdown voltage of few milivolts" [web: laro.lanl.gov; read via Investigator note]. d. Nothing. Verdict: Plausible, untested for the mV-to-grid chain in principle (each part exists), with the Josephson / Andreev role Not testable as stated (no equations or device). Converter efficiency at 10 mV input is a serious unaddressed problem.

Claim 5: superconducting actinide compounds (PuCoGa5, PuCoRh5) as the harvesting structure, radiation damage balanced by self-repair

a. Math: Table 1 gives absorption cross-sections summed over constituents; the "absorption decrease" percentages are not derived in the text. I did not recompute them. b. PuCoGa5 Tc about 18.5 K and PuCoRh5 Tc about 8.7 K are real [memory, high; the paper cites Thompson 2006 and Ohishi 2006]. The Ohishi muon-spin study the paper cites actually shows self-irradiation damage from Pu alpha decay that suppresses Tc over time [memory, medium; the paper itself reports "loosing the superconductivity"]. The paper's claim of a "dynamic equilibrium" with self-repair in a reactor flux is a hypothesis; fission-fragment damage is orders of magnitude above alpha self-damage. It also needs cryogenic operation (below 18.5 K) inside a reactor core, and the paper's superconductor-based voxel size claim (dm^3) is unexplained. Rh has 144 b, which the paper itself says makes it unusable [read: p.4], so PuCoRh5 is rejected by the author's own table in a reactor. c. Not tested in a fission flux, as far as the record shows. d. Nothing to change in mainstream physics; it is an engineering feasibility problem. Verdict: Never properly tested (untested; speculative materials-science claim). Weak form (these are superconductors with measured Tc) is Established.

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

a. Math: no derivation; the number appears only in the Conclusions. The companion TMS 2008 abstract says "up to 1,000 times higher" than 200-500 W/cm^3 [read via Investigator note], which is 0.2-0.5 MW/cm^3 = 200-500 kW/cm^3, versus 0.2-0.5 kW/cm^3 in the 2007 paper. Also 5000 / 0.5 = 10,000 times, not 1,000. The two documents are therefore inconsistent by a factor of 10 to 1000 depending on how they are read. Internal inconsistency of the figures; the units (kw vs Kw) also vary. Feasibility (my calculation, [memory, high] for constants): 5000 kW/cm^3 = 5e6 W/cm^3 = 1.56e17 fissions/cm^3/s at 200 MeV per fission. For U-235 metal (4.8e22 atoms/cm^3, sigma_f about 585 b thermal) this needs a thermal flux of about 5.5e15 n/cm^2/s, a high-flux research-reactor level (HFIR is about 1e15), and would burn out all the U-235 in about 3.6 days. So the figure is not a sustained fuel power density in any ordinary sense. The voltage budget also fixes electrical output at a fraction of this. b. Typical power-reactor core densities are 0.1-0.3 kW/cm^3 (100 W/L is 0.1 kW/L, i.e. about 0.1 W/cm^3 for LWR core average, and about 0.3-0.5 kW/L for fast reactor fuel pins...) [memory, medium]. I did not verify the baseline; the paper's "0.2-0.5 kW/cm^3" is of the right order for fuel (not core) power density, in the range of fast reactor pin power density [memory, medium, unchecked]. c. Not tested. d. Nothing, but the limit is heat removal and burnup, not conversion. Verdict: Refuted by calculation in the sustained sense (fuel burn-out in days, and the document's own factor inconsistent with its companion); Internally inconsistent across the author's two documents. As a transient or theoretical peak, Not testable as stated.

Other points from the deep read that stay in the record

• Layer thicknesses in text and figures disagree (stated 700 nm, plots 600 nm; thin layers about 10 nm thinner in the figure) [read: page image p.2].
• In the SRIM stack the ion keeps about 91 MeV; a full fragment range is several um, about ten times the stack [deep read; memory, medium].
• Eq.1 as printed with Z1 = 55 overestimates the SRIM plateau by about 10 times because it omits effective charge; SRIM uses Ziegler's effective-charge scaling.
• Recoil implantation work in Bucharest (Muntele et al. 1999) is real experimental background for beam-material interaction, not for conversion [read: p.2].

Investigator findings: weighing

Kept in the record whatever the verdict: US20100061503A1 (with its own caveat that 84% is not a real efficiency and a beam-test design in FIG. 15 without results found); US20130121449A1 (extension to fusion and transmutation through "nuclear entanglement and quantum non-localities" as quoted by the Investigator: a physics claim with no mainstream support and no stated mechanism, Not testable as stated); US8067758B2 (shielding sibling); the 2008-2015 conference series; the Nanotech 2007 Vol. 4 sibling paper claiming "higher than 90%" by "quantum effects" (not read by me; the same budget argument applies unless it shows voltages of volts); the later LENR turn (2012, 2013, no CIci measurements found); independent supportive work (Klassen/Yershov; Anufrienko 25%; Yoshida/Tanabe; Stepanov 2018, which measured charge separation up to 1.15 V in a different device); related FEC lineage; the 2026 MDPI review. Absence-of-evidence: no measured CIci results from 2007 to 2026 despite patents and conference output. Competing explanations: no funding for fabrication and beam time (supported by the 2012 roadmap's request for funding); unpublished or negative results; the physics budget above (low voltage makes any measured efficiency tiny, giving little to publish). The USPTO file wrappers (abandoned applications) and LAVM LLC contract records (SBIR/DOE/Army, Alabama irradiation center) would decide between them. I did not follow these.

Summary table

Source tags used

[read] paper text and page images via deep read; patent para. 0121 and 0156. [web] https://www.sciencedirect.com/topics/engineering/fission-fragment (fragment energies), https://www.mdpi.com/2227-9717/14/17/2834, Investigator notes (laro.lanl.gov, Klassen chapter p.255-273). [memory] constants, Josephson and Andreev physics (high); electron mean free path, converter start-up voltages, reactor power-density baseline (medium; not web-checked, budget limited; the verdicts above do not depend on them).