RankingsPrimerArbiterPhysicistMathematicianContrarianQESites logClaimsSourcesResearch notes

Science Contrarian review: Haisch-Bernard

Reviewer: Science Contrarian (Step 2). Method: judge by experimental evidence, build the strongest honest case, apply the same rigour to the inventors and to the critics. Tags: [read] from a document (page given), [web] with URL, [memory] with confidence, [calc] my own arithmetic (Python venv, numbers shown). Sources read myself: patent US 7,379,286 pp.3-4 (page image) and the Step 1 deep read; Forward 1984 deep read (equations re-derived by the deep reader, I rely on it); Dmitriyeva and Moddel 2012 (Physics Procedia 38) and Moddel and Dmitriyeva 2019 (Atoms 7, 51) full text from research/files/; the Investigator notes; the SSE 2010 abstract. No source gap: all three project documents are readable. Interested parties: Haisch, Moddel and Jovion Corporation (assignee) hold a commercial stake in the patent; their statements are reported as their characterisation.

Plain-language summary of the project (for later steps)

Atoms of a noble gas are pushed through gaps about 0.1 micron wide (a "Casimir cavity"). The inventors say each atom drops 1 to 10 eV in energy on entering, shedding heat, and regains the energy from the zero-point field (the vacuum's ever-present fluctuating electromagnetic field) on leaving. Repeat for continuous power. The theoretical basis is stochastic electrodynamics (SED), a classical theory in which the hydrogen atom is held up by a balance between radiation lost by the orbiting electron and energy absorbed from the zero-point field. One experiment by the co-inventor's own lab saw an infrared signal, about three orders of magnitude below the prediction, with trends opposite to the prediction.

Corrections to Step 1

  1. The "1700 Pa versus 17 atmospheres" inconsistency is resolved, and the resolution matters. claims.md item 4 and the deep read leave the two figures unresolved. I redid the pumping calculation independently [calc]: 200 nm pores, 1 cm long, 0.25 porosity over 1 cm^2 gives about 8x10^8 pores. In free-molecular flow (argon, mean thermal speed 394 m/s) the conductance of one long tube is (pi/12) d^3 vbar / L = 8x10^-17 m^3/s, so the block has about 6.6x10^-8 m^3/s. Delivering 1 cm^3/s at STP (0.10 Pa m^3/s) needs a pressure drop of about 1.5x10^6 Pa, about 15 atm. A Hagen-Poiseuille estimate (viscous flow) gives about 1.2x10^6 Pa. So the patent's "17 atmospheres" is right to within my accuracy and "1700 Pa" is the error (it should read about 1700 kPa). The pumping power is then about 0.3 W (isothermal compression work, Q ln(p2/p1)) to about 1.5 W (pressure drop times 1 cm^3/s), not the printed 1.7 mW. The patent's claim that pumping "is much lower than" the output stays true only against its 2.2-22 kW cube estimate; against its own 1-10 W bonded-wafer estimate the corrected pumping power is 3% to 150% of the output. The 2019 paper's "more than three orders of magnitude lower" [read: Atoms 2019 p.14] inherits the same unexamined figure. Also: at 17 atm the argon mean free path is about 6 nm, far below 200 nm, so neither flow regime assumed is accurate; this is a rough check, not a design number. [calc]
  2. The deep read of the 2012 experiment is missing from Step 1; the Investigator's summary of it needs two additions (the paper is in research/files/). (i) The authors say in the same section as the power balance: "At this point it appears that the power out was less than the power in and there is no need to invoke an additional power source, e.g., from ZP energy emission, inside the chamber." [read: 2012 paper p.13] The Investigator quoted the later sentence ("ZP energy extraction ... remains a possible explanation") but not this one. Both belong together. (ii) The paper contradicts itself on helium. Section 3 says the highest radiated power was seen with He [read: p.12]. The Joule-Thomson section says "the signal from He was so small we were not able to obtain reliable measurements from this gas" [read: p.15]. One of the two statements is wrong, or they refer to different conditions (the paper does not say). The Investigator's "He above Xe" relies on the first.
  3. "About 1000x below the ZPE prediction" should read "thousands of times". The paper says "on the order of milliwatts, which is thousands times larger than the measured power" [read: 2012 p.16]. Note the 1 eV-per-atom figure would, for 5 sccm (2.2x10^18 atoms/s), give 0.36 W if every atom shed it [calc]; the "milliwatts" prediction therefore already assumes only about 0.3% of atoms contribute. The paper does not state that factor. This is a hidden adjustment in the prediction.
  4. The patent's wavelength line is correct as printed. The OCR in the text layer reads "220.529137", which would be 290 A. The page image shows "2pi0.529137 = 455 A" [read: page image, patent printed p.3]; 2 pi a0 / alpha = 455.5 A and c/(6.6x10^15 s^-1) = 455 A [calc]. No error; do not carry the OCR form forward.
  5. Patent claims paragraph is inconsistent with the Xe wavelength used later. The patent scales the supporting wavelength as r^(3/2) [read: page image, printed p.4], which for Xe (r = 2.05 A) gives about 0.35 micron, with the cavity required to be about 0.1 micron. The 2019 paper puts the Xe orbital wavelength at 0.1 micron and He at 0.05 micron [read: Atoms p.14]. The two documents of the same group do not use the same number. Minor, but a reader will meet it.
  6. Date of the abstract. "Haisch_abstractSSE10" is probably the June 2010 SSE meeting (Investigator's inference). Its text says "Recent experiments have shown that Lamb shifts can be controlled and enhanced" and "potentially greater than nuclear" [read: abstract p.1]; it gives no data. Fine as a statement of intent; it contains no result.
  7. Step 1 otherwise holds. The patent deep read arithmetic (2150-21500 W, 21-210 W) reproduces: 2.7x10^19 x 500 x 1.602x10^-19 J = 2163 W per eV, and 1.3x10^20 x 1.602x10^-19 = 20.8 W per eV [calc]. The water-bath remark (alpha = 10 per cm over 1 mm gives only 63% absorption, "nearly all" needs a larger alpha) is correct [calc].

Claim 1. Atoms entering a ~0.1 micron Casimir cavity lose about 1-10 eV by a "Casimir-Lamb shift"

a. Mathematics consistent? The patent is a verbal argument with two numerical steps and no derivation. The 455 A figure follows from the SED Bohr orbit [calc, above]. The step to "1 to 10 eV" is an assumption: "we assume that the amount of energy released ... would be on the order of 1 to 10 eV" and later "it is reasonable to expect" [read: patent printed p.4]. The scaling argument for heavier atoms assumes a fully shielded outer electron, which the patent itself calls "a crude assumption". Internally there is a geometric tension: parallel plates at gap d suppress modes with wavelength longer than about 2d, so a 250 A gap suppresses wavelengths longer than 500 A. The orbit-supporting wavelength is 455 A, so the gap sits right at the edge, and the patent then relies on a "stairway" of partial suppression at d/n and on the tail of the wavefunction to justify gaps of 0.1-0.2 micron [read: printed p.3]. No quantity is computed for how much of the supporting spectrum is removed at 0.1 micron.

b. Consistent with established experiment? Three points. - Established cavity effects are real but tiny. Atoms between plates do show level shifts and forces: the Casimir-Polder force on ground-state sodium atoms in a micron-sized cavity was measured by Sukenik, Boshier, Cho, Sandoghdar and Hinds, Phys. Rev. Lett. 70, 560 (1993), described as a modification of the ground-state Lamb shift in the confined space [web: https://digitalcommons.odu.edu/physics_fac_pubs/7/]. The inventors are right that this is a mainstream, measured phenomenon and "Lamb shifts can be controlled" is true. Its size is the issue. A retarded Casimir-Polder estimate for a helium atom 50 nm from one wall is about 8x10^-10 eV, and for xenon about 1.5x10^-8 eV; at 25 nm, 1.2x10^-8 eV (He) and 2.4x10^-7 eV (Xe) [calc, polarizability volumes 0.205 and 4.04 A^3 from memory, confidence medium; only an order-of-magnitude check]. The claimed 1-10 eV is 7 to 10 orders of magnitude larger. - Energy available in the removed modes. The Casimir energy per area at 0.1 micron is pi^2 hbar c / (720 d^3) = 4.3x10^-7 J/m^2. Over an atomic footprint of (0.3-0.4 nm)^2 that is 2x10^-7 to 4x10^-7 eV [calc]. The SED answer is that the energy is not "stored" in the removed modes but flows in at the speed of light to feed the orbit (the Larmor power for the Bohr orbit is 4.7x10^-8 W, which would drain 13.6 eV in 47 ps [calc]), so a static volume-energy count is not decisive within SED. I report it because it shows how far the claim sits from the energy that mainstream accounting assigns to the cavity. - Metals are transparent at the wavelength the theory relies on. The patent needs suppression of zero-point modes near 455 A (photon energy about 27 eV). Real metal walls reflect well only below their plasma frequency (gold roughly 9 eV, aluminium roughly 15 eV) [memory, confidence high]; at 27 eV a gold or aluminium film is largely transparent, and polycarbonate pores certainly are. The ideal-conductor mode counting that the "exclusion of modes" picture uses fails for those short wavelengths. Forward 1984 states the same limit (the Casimir law holds down to a_min = lambda_min / 2 pi, with lambda_min the shortest wavelength the conductor responds to) [read: Forward deep read, Eq. 4]. The patent does not address it. This is an objection from established optics, and the inventors have no reply on record.

c. Tested? By whom? Replicated? Four attempts, none neutral, none positive in the sense claimed: 1. Puthoff, Little and Ibison, synchrotron test of Casimir-cavity ground-state shifts, "so far without success" [web: arXiv:1012.5264, via Investigator note; JBIS 55, 137 (2002)]. No data published that I found. 2. Dmitriyeva and Moddel 2012: pyroelectric detector, polycarbonate membranes with 0.1-0.4 micron pores, He/Ar/N2/Xe. An infrared signal "clearly observed"; measured power in about 2x10^-5 W, power detected about 10^-7 W, about 4x10^-5 W after a solid-angle correction the authors then discount because of wall reflection [read: 2012 p.13]. Trends opposite to the prediction (strongest for helium and for uncoated plastic). The authors attribute the plastic/gold difference to emissivity (0.95 versus 0.01-0.1) and the gas order partly to thermal conductivity, "qualitative, not quantitative" [read: p.13-14]. They exclude Joule-Thomson cooling by a phase test with a 3 mm hole in Mylar (opposite lock-in phase). They conclude: "None of these clearly fit the data, but it is possible that a combination of effects could" [read: p.16]. 3. Moddel and Dmitriyeva 2019 add helium data on 0.1, 0.2, 0.4 micron uncoated filters; the 0.1 micron filter gave the largest signal, which the authors read two ways (better mode match, or larger pressure drop and so more frictional heating) [web: Atoms 2019 p.11, via Investigator]. "Tantalizing but ... inconclusive" [read: abstract]. 4. Henriques 2014 (MSc, Coimbra): Al-coated membranes, UV-visible PMT. "Some evidence of radiation" for Xe; nothing above noise with pre-excited Xe; detector judged too insensitive [web: https://estudogeral.uc.pt/handle/10316/28082, via Investigator; thesis PDF not obtained, the fetch failed with too many redirects]. Moddel and Dmitriyeva summarise it more negatively ("not able to detect any emitted radiation") than the abstract does ("some evidence"); both stand. No independent group has repeated the 2012 gas-flow experiment. The "replicated" answer is therefore: not replicated, and the one closely similar independent attempt (Henriques) did not reproduce a signal.

What was never done (design gaps in the only experiment): a scan of pore size across the point where a "Casimir cavity" ceases to exist (for example 0.1 micron up to 5-10 micron at equal pressure drop and equal flow), which is the control that would separate a cavity effect from a generic flow effect; a non-membrane control with matched flow impedance; spectral or imaging identification of the source (the authors list this as future work [read: Atoms p.15]); a calorimetric balance with the walls included; a test of whether the pyroelectric detector responds to pressure or acoustic modulation at the 0.5 Hz flow chopping frequency. The 2012 paper does include a single-hole Mylar phase control and a Reynolds-number check, which counts for something. Pore sizes in the track-etched filters are not uniform [read: 2012 Fig. 2 per the paper's text], so which pores, if any, were "cavities" is not known.

d. If true, what would have to change? Atomic ground states in QED are the lowest-energy states of the coupled atom-plus-field system; a mainstream reading allows no energy release from entering a gap beyond the potential-energy change, which is the Casimir-Polder energy of order 10^-9 to 10^-7 eV. A real 1-10 eV drop would require (i) that the atomic ground state in free space is not the lowest state accessible to the atom-field system, (ii) that cavity QED calculations of the ground-state shift miss a term 10^8 times larger than the ones tested by Sukenik et al. and by later atom-surface work, and (iii) that SED (a classical theory of the vacuum field) is the correct description of atomic stability. It would not by itself break energy conservation (the ZPF would supply the energy), but it would require a vacuum that acts as a usable energy reservoir for atoms, which the inventors argue is not forbidden because the ZPF is not thermal (Cole and Puthoff 1993) [read: patent printed p.3; web: calphysics.org via Investigator].

Verdict (split). - Strong form (1-10 eV per atom per transit at 0.1 micron): Tested, not supported. The tests were small, by the inventors' own lab, and under-controlled, so I do not rate this a decisive refutation; but in each test the signal was far below the prediction (2012: thousands of times), with the wrong trends, and the other two attempts found nothing above noise. That is what "tested, not supported" means here, with the caveat that the tests could not have detected a much smaller effect. - Weak form (a shift of sub-eV size in a suitable cavity): Never properly tested. Nobody has run a controlled spectroscopic search of noble-gas atoms in cavities of the right size with a detector sensitive to the relevant wavelengths. It was not dismissed without testing: it was tested by its proponents, who stopped.

Claim 2. On leaving the cavity the atom re-absorbs the energy from the vacuum, so cycling yields usable power "extracted locally, replenished globally" without violating thermodynamics

a. Mathematics. No energy-balance equation is given. The thermodynamic assertion is by citation (Cole and Puthoff 1993) and by the ocean analogy. The co-inventor's own 2019 paper puts two objections on the table and does not resolve them: (i) the "potential hill": if the atom's energy is lower inside the cavity, the atom must climb out, and the energy would come from the pump, so the "emission" on entry would be pump energy returned; the authors argue against it but say "we cannot totally rule out this potential hill argument" [read: Atoms 2019 p.14]; (ii) detailed balance: if the gas enters slowly compared to the equilibration time the process is quasi-static, and "there would be no excess energy to shed" [read: Atoms p.15]. They conclude it "is not clear whether equilibrium considerations render it incapable of providing harvestable energy" and, in the closing line, "Either fundamental constraints beyond what have been discussed here and the nature of ZPE preclude extraction, or it is feasible and we just need to find a suitable technology." [read: Atoms p.16]. (The patent claims "thermodynamic laws are not violated" as settled [read: printed p.3]; the later paper by the same group is less sure.) A simple bookkeeping check: the flow time of an atom through a 1 cm pore at 10 cm/s is 0.1 s, against orbit periods of 10^-16 s; the process is as quasi-static as can be, which is the case where the 2019 paper says detailed balance applies. The inventors have not shown why the entry is non-adiabatic from the electron's point of view; the patent concedes "no matter how quickly the atom is injected ... the process will be a slow one as experienced by the orbiting electron" [read: printed p.3] and then assumes gradual heat release anyway.

b. Established experiment. No experiment has ever shown net power from a vacuum-field cycle. Haisch's own site: "to date no such device has passed a rigorous, objective test" [web: http://www.calphysics.org/zpe.html, via Investigator]. In 2009, as quoted second-hand: "not yet been able to prove anything in the laboratory. The sporadic signals they have seen can't be ruled out as experimental error." [web: https://www.nextbigfuture.com/2009/02/jovion-corporation-gets-patent-for-zero.html, via Investigator]. Established: the Casimir force and the Casimir-Polder force exist and do mechanical work when plates or atoms move (that is the stiction problem in MEMS, mentioned in the patent). The step that atoms in the cavity are in a lower and different state, and are re-excited on exit by ambient vacuum fluctuations, has no experimental support.

c. Tested? Only through the 2012/2019 data and the two others above. The power balance in 2012 shows power out below power in; the authors say there is "no need to invoke an additional power source" [read: p.13]. No test has put the full device in a closed calorimeter with a pump-power meter.

d. If true. The vacuum would have to act as a usable energy source for localized systems. Haisch gives the dam analogy: the energy is "extracted locally and replenished globally". The contrarian point is that this analogy is Forward's (claim 5), and Forward's version needs an energy input to re-separate the plates; Haisch himself says of Forward's scheme "there is no practical application ... since ideally it would take just as much energy ... to separate the plates for a second cycle" [web: calphysics.org]. Whether the gas cycle escapes that objection is exactly what the 2019 paper says it cannot yet tell.

Verdict: Never properly tested (as a net-power cycle). The record: a proponents' experiment whose power out was below power in; no closed-loop calorimetry; no neutral test. It was neither dismissed without testing by outsiders nor tested neutrally. A mainstream reviewer's alternative reading, that detailed balance already forbids it, is an argument from theory not yet shown by calculation: I do not assign "Refuted by calculation", because the claim's own authors treat the question as open and I cannot show it ruled out by an accepted-law calculation.

Claim 3. SED: the hydrogen ground state is a classical orbit sustained by Larmor emission balanced by zero-point-field absorption (Boyer 1975, Puthoff 1987, Cole and Zou 2003-4); the same balance is asserted, not proven, for all atoms

a. Mathematics. The balance is a real computation in the classical model: Larmor power for the Bohr orbit is P = e^2 a^2 / (6 pi eps0 c^3) = 4.7x10^-8 W with a = (alpha c)^2 / a0 [calc]. The Boyer/Puthoff absorption calculation treats the electron as a harmonic oscillator rather than as an orbit in a Coulomb potential, which Haisch himself calls "a weakness in the analysis", and he notes "the need to cut off the particle-field interactions to avoid autoionization" [web: calphysics.org, via Investigator]. The patent also says openly that the all-atoms extension "has not yet been proven theoretically" [read: printed p.4].

b. Established experiment. SED is a legitimate research programme that derives some real results (the Planck spectrum with zero-point terms, Casimir and van der Waals forces, the harmonic oscillator ground state) [memory, confidence high]. For hydrogen, Cole and Zou 2003 report that a classical point charge in a Coulomb field with zero-point radiation gives a probability density close to the Schrodinger ground state [web: https://www.sciencedirect.com/science/article/abs/pii/S0375960103012696]. The later, longer simulations by Nieuwenhuizen and Liska find that the atom always ionizes at long times, with and without relativistic corrections [web: https://arxiv.org/pdf/1506.06787 and https://arxiv.org/pdf/1502.06856]. The same group that proposed the patent notes that an SED treatment of environmental effects on spontaneous emission and the Lamb shift (Cetto and de la Peña 1988) does not predict the cavity shift "under particular simplifying assumptions" [read: Atoms 2019 p.9, ref. 46]. I have not read Cetto and de la Peña; what their assumptions are is unknown to me.

c. Tested. Numerically (Cole-Zou, Nieuwenhuizen-Liska, Nickisch, Henriques 2D/3D); the outcome is mixed, with the stronger evidence being the long-time ionization. Experimentally, atomic stability is not a discriminating test between SED and QED in free space; the cavity prediction is the discriminating test, and it is claim 1.

d. If true. Quantum theory's atomic structure would be a statistical description of classical motion in a real zero-point field; QED's account of the Lamb shift and spontaneous emission would need to be recast.

Verdict (split). - SED hydrogen ground state as a stable, quantum-matching classical system: Tested, not supported (long simulations ionise the atom; the short-time agreement is real but not sufficient). - Extension to all atoms and to "sub-Bohr" states in cavities: Never properly tested; the patent itself says it is unproven. - Interpretation of the cavity effect as testable at all: depends on claim 1.

Claim 4. Patent estimates: 21-210 W (channels), 2150-21500 W (1 cm^3 block), 1-10 W (1 cm^2 wafer device at 8 atm), pumping 1.7 mW

a. Mathematics. - Arithmetic from the stated inputs is right: 1.3x10^20 transitions/s x (1-10 eV) = 21-208 W; 1.3x10^22 x (1-10 eV) = 2.1-21 kW [calc]. - Everything is proportional to the assumed 1-10 eV per transition and 500 interactions per atom (cube) or the assumed transition rate (channels). If claim 1's per-atom energy is reduced by the factor of order 10^3-10^4 suggested by the 2012 data, the cube output falls from kilowatts to watts and below. [calc, scaling only] - The 1-10 W wafer figure at 8 atm has no worked calculation in the patent [read: deep read]. - The pumping figure is wrong by about 1000 (see Corrections, item 1). With the corrected 0.3-1.5 W the comparison to the cube (2.2-22 kW) is unchanged; the comparison to the wafer figure is not favourable. - The porosity (0.25) is never used; the cube estimate uses 10% open area and 0.1 micron, the pumping check uses 0.25 and 0.2 micron, so the two are not for the same device [read: deep read]. - Reference: the body cites Roy et al. as "(1993)" and the list gives 2003 (J. Appl. Phys. 93, 4870) [read: deep read]; a typo.

b. Established experiment. None of the power numbers is a measurement. The nearest measurement is the 2012 experiment: detected power 10^-7 W against pump power 2x10^-5 W at 5 sccm and 4 torr.

c. Tested. Not as claimed. No prototype with measured output exists in the record.

d. If true. A 1 cm^3 device making kilowatts of heat from room-temperature gas flow would be a power source of the type the second law forbids unless the vacuum is the source; see claim 2.

Verdict: Internally inconsistent in the pumping figure (1700 Pa versus 17 atm, which I resolve in favour of 17 atm; the printed 1.7 mW is then about 1000 times too small); not supported as output estimates, which are conditional on claim 1. Strictly on the scale, the estimates are projections that depend on an untested claim, so for the output numbers: Never properly tested.

Claim 5. Forward (1984): a stack of charged foil leaves, held just below the Casimir force and then allowed to collapse, yields electrical energy; the vacuum-fluctuation field is probably conservative (hydroelectric dam analogy), the device is a rechargeable battery

a. Mathematics. Consistent. The deep reader checked the Casimir and Lifshitz expressions against the page images; the coefficients 1440 and 480 are right for h (not hbar), equal to pi^2 hbar c/720 and pi^2 hbar c/240 [read: Forward deep read, SymPy check]. One sign slip in Eq. (2) (a missing minus after the derivative) does not alter the argument. The cohesion energy from collapse to final gap a_f is about |u(a_f)| = pi h c / (1440 a_f^3): 4.3x10^-4 J/m^2 at 10 nm, 0.43 J/m^2 at 1 nm if the ideal-conductor law held, which Forward's Eq. (4) says it does not below a_min [read: deep read table]. The deep read's statement that the aluminium plasma wavelength (about 80 nm) puts a_min near 13 nm is [memory, confidence medium] and the same objection as for claim 1b.

b. Established experiment. The Casimir force between metals has been measured precisely at 0.1-1 micron since the late 1990s (Lamoreaux 1997; Mohideen and Roy 1998) [memory, confidence high; not rechecked on the web]. These post-date Forward and address his gap at 300 nm and above, not his 1-20 nm region; I know of no clean measurement of the retarded-to-non-retarded crossover for metals at those spacings [memory, confidence low]. The force doing work on plates is established (it is the cause of stiction in MEMS).

c. Tested. Forward proposed no device test and gave no energy estimate [read: deep read]. I found no record of anyone building a foil-stack battery. Never built, so never tested.

d. If true. Nothing in mainstream physics changes: Forward's own position is that the field is conservative and the device stores energy that someone first supplied when the foils were separated. Haisch makes the same point about it.

Verdict (split). - That the Casimir attraction does work on charged, nearly-balanced plates and the work can be recovered electrically in a single collapse: Plausible, untested (as a device); the underlying force law is Established for metals above about 100 nm. - That it is a source of net energy over a cycle: not claimed by Forward; if claimed by others, Refuted by calculation in the sense that a conservative force gives zero net work around a closed cycle. - The numbers at 1-20 nm for metals: Never properly tested (Forward's own point, 1984; unknown whether since filled).

How the idea fared: rejection, neglect, and what was done

  • Was it rejected by mainstream bodies? I found no published mainstream rebuttal of the patent. The scepticism on record is the inventors' own: Haisch (2009, calphysics) and Moddel (2019). Haisch states most physicists consider ZPE use "extremely unlikely" [web: calphysics.org]. Whether the USPTO challenged the novelty or utility is not in the record I saw; the patent issued (granted 27 May 2008; priority 2004-09-27) [read: patent p.1; web: Google Patents via Investigator].
  • Was the work blocked? No evidence of funding or publication being blocked. The record shows modest, conditional funding: DARPA via SPAWAR (N66001-06-1-2026, from 2006), CU proof-of-concept money contingent on milestones, private money (about $200,000 by Feb 2009), $10 million sought, HUB Lab contract to Jovion (2010-12) [web: Investigator notes]. A $10 million sub-ask never visibly funded is a fact; why is unknown.
  • Were the negative experiments sound? The only negatives are the proponents' own and are weak instruments: the 2012 test could not distinguish a cavity effect from heating; the Coimbra test's detector was too insensitive by its author's account; the synchrotron test left no data. None counts as a definitive disproof, and equally none counts as a test the claim passed. The strongest honest statement is: the claim failed to produce a clear signal in every attempt known to me, and the attempts were not designed to be decisive.
  • Inventor rigour. The inventors deserve credit for candour: the co-inventor's papers list the conventional explanations, state that power out was below power in, and call their data "inconclusive". The patent, however, states large power figures and "researchers have shown that thermodynamic laws are not violated" as settled and carries an arithmetic error that flatters the output/input ratio by a factor of 1000. These are the weak points on the inventors' side.

Investigator findings kept in the record (whatever the verdict)

  1. The Colorado experiment (2012) and its detailed numbers, including the contrary trends, and the later helium data (2019). Weighed above.
  2. Moddel and Dmitriyeva (2019): the thermodynamic position that the gas-flow scheme "does not appear to violate" the second law, unlike the nonlinear-rectifier and mechanical-Casimir schemes, with the open issues of detailed balance and the potential hill. Weighed in claim 2.
  3. Haisch's caveats (calphysics.org; 2009 interview). Weighed in claims 2 and 3.
  4. Cetto and de la Peña (1988), an SED analysis that does not predict the shift under simplifying assumptions. A mainstream SED source against the patent's reading of SED. Not read by me; flagged for the arbiter.
  5. Henriques (2014, Coimbra): independent SED simulation and spectroscopic experiment. Thesis not yet obtained (fetch failed).
  6. Puthoff, Little and Ibison (2002): synchrotron search "so far without success".
  7. Moddel et al. 2021 (Phys. Rev. Research 3, L022007; Symmetry 13, 517): Casimir-cavity MIM devices with "anomalous offsets in the current and voltage"; patent family US 11,133,758 B2 etc. with projected "up to 250 W" for a 10 x 10 cm array (a projection). A different mechanism (hot-carrier injection, with the authors invoking a Delta E Delta t limit), peer-reviewed in an APS journal. It is related work that shares the premise and the funding history. I did not read these papers; I only have the Investigator's abstract quotation. Kotov et al. (2025), a mainstream cavity-modified ground-state theory that cites Moddel et al. 2021, is unexamined: whether it supports or contradicts is unknown.
  8. Related work: Cole and Puthoff 1993 (thermodynamic argument); third-party patents US20190207537A1, US20180059704A1, US20230208321A1 (citing Haisch-Moddel); BlackLight Power hydrino claims (a commenter's connection, not the inventors'). The hydrino link is the commenter's characterisation.
  9. Funding chain and company history above; Jovion website (c) 2023 with an empty progress link; Hathaway Research (HRI) project page with no results; patent legal status unresolved ("2011-12" event; possibly a lapse for maintenance fees).

Absence of evidence: the record goes quiet after 2012

After 2012 no gas-flow follow-up from the Colorado group appears (the 2019 paper reuses 2010-12 data plus the 0.1 micron helium run). The Investigator's four explanations stand: (a) the effect did not hold up and was dropped quietly; (b) funding ran out; (c) trade secrecy; (d) the group moved to MIM devices under a different mechanism. Evidence weighing from my reading: (a) is supported by Haisch's 2009 caution and the 2012/2019 wording, and by the 2019 closing sentence that fundamental constraints may preclude extraction; (b) by the unmet $10 million ask; (d) by the new patents from 2019 and the new papers. For (c) the 2019 paper (published, with stock-ownership disclosure) describes the gas-flow work in open terms and states that Moddel "plays no active role" in Jovion, which makes active secret development at Colorado unlikely; Jovion's own activity is unknown. A fifth explanation: (e) the ambiguity of the pyroelectric data (heating artifacts) made a cleaner follow-up impossible with the available money. Not mutually exclusive. Records that would decide: USPTO maintenance-fee and assignment history for US 7,379,286; DTIC/DARPA final report for the SPAWAR grant; HUB Lab/HRI reports; Jovion corporate filings; CU technology-transfer records.

What would settle the main disagreements

  • Claim 1/2: a pore-size scan (0.05 to 10 micron) at fixed flow and pressure drop with the same membranes, a matched-impedance non-cavity control, a spectrally resolved or imaging detector, and a calorimeter in which the pump power is metered. A cavity effect would turn off when the pores are far larger than the supporting wavelength; a heating artifact would not.
  • Claim 1 independently: high-sensitivity spectroscopy (UV to soft X-ray) of noble-gas atoms in 25-250 nm gaps in a material that actually reflects at the relevant wavelengths.
  • Claim 3: a published treatment of SED hydrogen in a gap (Cetto and de la Peña's assumptions versus the patent's) and a long-time stability test.
  • Claim 4: redo the pumping-power calculation with proper Knudsen/slip flow and publish the figure; correct "1700 Pa".

Web verification log (budget about 6)

  1. Nieuwenhuizen and Liska SED hydrogen ionization: [web] https://arxiv.org/pdf/1506.06787 and https://arxiv.org/pdf/1502.06856 (checked via search summary, papers not opened).
  2. Sukenik et al. 1993 Casimir-Polder measurement: [web] https://digitalcommons.odu.edu/physics_fac_pubs/7/ (search summary).
  3. Cole and Zou 2003: [web] https://www.sciencedirect.com/science/article/abs/pii/S0375960103012696 (abstract in search results).
  4. Henriques thesis PDF: fetch failed (too many redirects); not read.
  5. Items still [memory]: plasma frequencies of gold and aluminium (confidence high); Lamoreaux 1997 and Mohideen-Roy 1998 (high); He and Xe polarizabilities (medium); Cole and Puthoff 1993 content (from the Investigator's note of Haisch's page). Not verified and not claimed: any replication of the Colorado experiment (none found); any independent theory paper supporting the cavity shift (none found).