Gravity modification experiment using a rotating superconducting disk and radio frequency fields
| Scientific Paper | |
|---|---|
| Title | Gravity modification experiment using a rotating superconducting disk and radio frequency fields |
| Read in full | Link to paper |
| Author(s) | George D Hathaway, Blair M Cleveland |
| Keywords | Frequency, gravity, fields |
| Published | 2003 |
Read the full paper here
Abstract
An experiment is described which attempts to replicate the results of Podkletnov et al. concerning an alleged detection of a gravity-like force above a spinning superconductor. The experiment is based on Podkletnov’s published descriptions plus personal communications but found no evidence of a gravity-like force to the limits of the apparatus sensitivity. A full description of the apparatus and operation is given.
Overview
This is a replication attempt, and it is a null result. Published in Physica C 385 (2003) 488–500 by G. Hathaway, B. Cleveland and Y. Bao of Hathaway Consulting Services in Toronto (received 26 November 2001, accepted 23 September 2002), it reports three full-scale runs of an apparatus built to reproduce Eugene Podkletnov's claim of a "gravity-like force" above a rotating, radio-frequency-illuminated YBCO superconductor. No weight modification of the test mass was observed, at any RF frequency, levitation or RF power level, sample temperature or rotational speed, down to the 0.001 % level of the balance's repeatability. The authors state plainly that their weight detection was about 50 times more sensitive than that available to Podkletnov, "one would have expected to see a definite weight change."
The value of the paper does not lie only in the negative answer. Roughly two thirds of it is a detailed, reproducible account of how the disks were made, how they were tested, and what went wrong — material that is normally omitted from published work and that is precisely what a later experimenter would need. The authors were also in direct contact with Podkletnov throughout (a string of personal e-mail, fax and telephone communications is cited as references [3], [5]–[9]), and they acknowledge his assistance and that of the Institute for Advanced Studies at Austin. This is a replication conducted cooperatively with the original claimant, not against him.
The claim being tested
Podkletnov and Nieminen reported in Physica C 203 (1992) 441 that a test mass suspended above a levitated, spinning bulk YBCO ceramic disk illuminated by RF fields lost about 0.05 % of its weight. A single-layer sintered disk 145 mm in diameter and 6 mm thick with a specific grain-size distribution was Meissner-levitated up to 7 mm above a "pancake" coil driven at 50–106 Hz and spun by an edge-mounted magnetic stator, the whole experiment running below 60 K in helium vapour. A 1997 update posted on the Los Alamos preprint server raised the claim to 1–2 % weight loss, using a much larger ring-shaped disk (270 mm outside diameter, 80 mm inside, 10 mm thick) of small-to-medium grain size in a bi-layer structure — one layer superconducting, the other remaining a normal conductor at operating temperature — levitated above three or six solenoids with two toroidal coils threaded through the centre hole and driven two-phase to spin the disk.
Hathaway and colleagues combined attributes of both procedures, settling the final configuration in consultation with Podkletnov himself: bi-layer YBCO disks of 160 mm outside diameter × 40 mm inside × 7–10 mm thick, three levitation solenoids at 105 Hz, and three toroidal RF coils in a three-phase arrangement threaded through the centre hole at 2–5 MHz, all at 5–20 K.
Method
Making the disks
The bi-layer requirement drove much of the work. Podkletnov had attributed the effect to non-superconducting phases and specifically to the diffusion layer (about 0.5 mm thick) between the two layers. Three fabrication routes failed and are reported as failures: induction heating of the surface layer produced uneven heating and thermal-shock fracture; a double-chamber method exposing one face to oxygen and the other to an inert gas failed because of the different diffusivities of the gases; and simple co-pressing of superconducting and non-superconducting YBCO powders gave badly warped disks through uneven shrinkage on sintering.
The successful route was cation substitution: replacing part of the copper in YBCO with Zn2+, Fe2+ or Pr progressively destroys superconductivity. YBa2Cu2.85Zn0.15Ox, YBa2Cu2.6Fe0.4Ox and YBa2Cu2.3Fe0.7Ox ("Z15", "F40", "F70") gave critical temperatures of about 40, 10 and 0 K respectively; PrBa2Cu3Ox ("Pr123") was finally chosen as the non-superconducting layer material. Full calcining, ball-milling, sieving, pressing (120 MPa) and sintering schedules are given for both a multiple-sized and a triple-sized grain distribution, the latter following Podkletnov's own schedule. Sintered density was 4.5–4.9 g/cm3, porosity 10–15 %, the transition layer under 0.5 mm, the superconducting transition ≈85 K, and critical currents only ≈103 A/cm2 — low, and attributed to the high porosity. A 160 mm YBCO/Pr123 triple-grain-size disk was sent to Podkletnov in March 2001, who "declared its characteristics were acceptable".
What could not be reproduced
Two elements of the original protocol defeated the apparatus, and the authors say so.
AC Meissner levitation failed. Despite testing pancake and solenoidal coils of many geometries, windings and wire types at 60 Hz to 100 kHz, DC currents of 12 V/100 A or 135 V/48 A, and arrays of NdFeB permanent magnets, no sintered disk could be freely levitated at the required frequency. The only successful liquid-nitrogen levitation used a 60 Hz three-solenoid design with soft iron pole pieces and reached only 0.5–1 mm, with excessive vibration. At liquid-helium temperatures the maximum frequency for successful levitation was ≈5 kHz, and only for a fine-powder disk — but Podkletnov specifically required coarse grain structure and a levitation frequency an order of magnitude higher. A direct beam-balance measurement of the Meissner force on a 160 mm disk at 1000 W and 100 kHz found a levitation force no greater than 5 g. Since Podkletnov had noted that an effect of order 10 % of the 1992 result should persist even without adequate levitation, the team proceeded with an external motor drive instead — a belt from a variable-speed DC motor on top of the cryostat to a shaft on low-temperature bearings and a gear assembly.
Rotation speed was limited. Single-layer multi-grain disks shattered at only ≈1300 r.p.m. in cold tests, so speeds were held below 1000 r.p.m.; in practice bearing and gear chatter at helium temperatures limited runs to a maximum of ≈550 r.p.m., averaging 400 r.p.m. Mechanical stress analysis put the tangential stress at shattering at only ≈0.5 MPa, concentrated at the inner circumference, with radial stress about a third of that.
The "reverse Josephson junction effect" tests — passing 0.05–10 MHz AC through samples and looking for a DC microvolt signal, which Podkletnov held to be the signature of a "good" sample — were also inconclusive. Two microvolt signals appeared, at ≈1.6 and ≈4 MHz, but the 4 MHz signal persisted when a graphite dummy replaced the superconductor, marking it as a set-up resonance artefact; the 1.6 MHz signal was absent for the dummy but too small in amplitude to call a detection.
The balance
Detection used a modified Ainsworth type DL chemical analytical balance rather than electronics, deliberately, "which eliminated any spurious electromagnetic fields from interfering". A ≈50 g phenolic disk-shaped test mass, its diameter matched to that of the superconducting disk (because the claimed force was said to scale with the test mass's cross-sectional area), hung 35 cm below the balance beam and 1.2 m above the superconducting disk, with four thermal baffles of 3 mm aluminium plate and three 9.5 mm structural aluminium plates in between. The test mass was housed in an air-tight acrylic cylinder inside a superinsulated enclosure, itself inside a balance enclosure covered in 3 cm foam, and read by CCD camera onto videotape.
Calibrated with milligram weights, the balance had a repeatable sensitivity of ≈0.5 mg, i.e. a weight change of 0.001 % of the test mass — against Podkletnov's reported 0.05 %. A separate preliminary test with a commercial gravitometer above the apparatus found no anomalous gravity-like behaviour to one part in 108 g.
Results
Three complete experiments were run: disk (A), a YBCO/Pr123 bi-layer triple-grain-size 160 mm disk; a dummy disk (C), an acrylic disc of similar size with a 1 mm copper disk on top to give the three-phase system a good RF load; and disk (B), from the same batch as (A). The dummy run was identical in every other respect and provided the baseline against which everything else was judged — a control the original experiments lacked.
A typical run cooled the cryostat with nitrogen, energised the three-phase RF at ≈90 K to "trap in" magnetic flux, removed the nitrogen at 77 K, transferred helium vapour to 10–20 K (5 K in some runs), then spun the disk while raising the 100 kHz levitation system to 1000 W and the three-phase system to 200 W per phase at 4–5 MHz. Maximum power and rotation were held for 60–90 s, then all systems were cut simultaneously to check for transients. Eight to ten runs per experiment were completed.
The summary table (Fig. 11), each entry the mean of ten one-per-second readings in milligrams with its standard deviation, reads:
| Data set | Mean of 10 observations (mg) | Std. dev. |
|---|---|---|
| Dummy 80 K, no power | 0.1 | 0.082 |
| Dummy 20 K, no power | −0.02 | 0.063 |
| Dummy 30 K, 400 r.p.m., full power | 0.08 | 0.079 |
| Disk A 80 K, no power | −0.07 | 0.067 |
| Disk A 20 K, no power | −0.05 | 0.071 |
| Disk A 30 K, 400 r.p.m., full power | −0.03 | 0.048 |
Every value lies within a tenth of a milligram of zero, and every error bar falls well inside the balance's repeatability, so the authors "can draw conclusions with considerable confidence". Nothing distinguishes the superconducting disk under full power and rotation from the acrylic dummy, or from either at rest.
One real physical effect was seen: a dragging force that slowed and sometimes stopped the disk whenever three-phase power exceeded ≈100 W per phase, stalling the motor drive entirely at 200 W per phase or more. The RF phase sequence was reversed in some runs (A, B, C to B, A, C) to change the field direction, but the disk was always driven clockwise viewed from above and the rotation sense was never reversed.
The conclusions are stated as four numbered points: no weight modification or gravity-like force to the 0.001 % level, with rotation never above 550 r.p.m.; the method of detecting internal Josephson junctions needs clarification; true AC Meissner levitation at 100 kHz was not achieved because the disks were too large and heavy and the solenoid field too weak; and higher speeds (1000–2000 r.p.m.), both rotation directions, and greater levitation power at lower frequency should be tried. The authors describe the tests as "thus far, proved inconclusive as to the actual existence of the Podkletnov 'effect' which may still be indeed present at greater energy levels and higher disk rotation speeds than used herein" — while noting that at fifty times Podkletnov's sensitivity a definite change should have shown.
Assessment
This is a careful and unusually honest experiment, and it deserves to be read as such. Its strengths are exactly the ones normally missing from anomalous-force reports: a mechanical balance chosen specifically so that no electronics could be fooled by the RF environment; a dummy-disk control run under identical conditions; a stated, calibrated sensitivity (0.5 mg on a 50 g mass, 0.001 %) that is fifty times better than the claim it tests; a second, independent check with a gravitometer at one part in 108 g; error bars quoted alongside means; videotaped raw readings; and consultation with the original claimant on the configuration. The disks were made to the claimant's own recipe and one was inspected and approved by him. Failed fabrication methods and failed levitation schemes are reported rather than quietly dropped. Under these conditions the result is clear: no gravity-like force was found at the 0.001 % level.
The paper's own reservations are the right ones, and the authors do not overstate their case. Two required conditions of the original protocol were not met — true AC Meissner levitation at the specified frequency, and high rotational speed. The disk was motor-driven through a shaft rather than freely floating, and it turned at 400 r.p.m. against a claimed requirement well above that. If the alleged effect depends on free rotation of a magnetically suspended disk, or on speeds of thousands of r.p.m., this apparatus could not have produced it. Against that, Podkletnov himself had said (reference [7]) that an effect of order 10 % of the 1992 magnitude should appear even with insufficient levitation — 0.005 %, still five times the balance's threshold — so on the claimant's own account something should have been visible. The mechanical weakness of the sintered disks, which shattered at 1300 r.p.m. and had critical currents of only ≈103 A/cm2 because of 10–15 % porosity, is a material limitation rather than a design error, and the authors' proposal to wind the periphery with carbon fibre is a sensible fix.
Two further points bear on interpretation. First, the strong dragging force at high three-phase power shows that the RF system was coupling to the disk substantially — the drive was not idling — which weakens any suggestion that the null result came from the fields never reaching the sample. Second, the inconclusive Josephson-junction tests leave open Podkletnov's own criterion for a "good" sample; the 4 MHz signal turning out to be a set-up resonance visible with a graphite dummy is a useful cautionary result in its own right, and precisely the kind of artefact that an uncontrolled experiment would have reported as a detection.
What the paper does not, and cannot, establish is a general non-existence proof. It rules out an effect of the claimed size under the conditions achieved, and it does so with a sensitivity margin large enough that the burden now sits with the original claim. It leaves untested the specific regime — free AC-levitated rotation at thousands of r.p.m. — that the claimant identifies as essential. That is a fair and precise statement of what a null result of this quality means, and it is very close to how the authors themselves put it.
More broadly, the claim under test would, if real, conflict with the Equivalence Principle as verified by torsion-balance experiments of the Eöt-Wash type, which constrain composition-dependent deviations from universal free fall at the 10−13 level — far below the 0.05–2 % weight losses claimed. A local, switchable "gravitational shielding" of the reported magnitude would be a very large effect in a domain that is otherwise among the most precisely measured in physics. That does not make the search illegitimate; it does set the standard of evidence, and this paper meets that standard on the negative side.