The treatment of Pons/Fleischmann is, by far, the worst scandal in modern science. It is probably worse than Galileo. Their claims were true, were reproduced immediately in several labs (sporadically, many labs failed too), and good research continues to this day, without funding and without theory.
Their claim is not completely new. It goes back to the 1920's, when Paneth and Peters reported Helium production in Pd heavy water electrolysis. Paneth and Peters' claims were dismissed for essentially the same reason that Pons and Fleischmann's were, the theorists' imaginations were too stunted to think up a mechanism that could bridge the gap between chemical energies and nuclear energies. In the 1950s, there was a Soviet scientist who also claimed that Pd heavy water electrolysis leads to nuclear anomalies. That claim was also dismissed by the Soviet scientific establishment, but his career was rehabilitated somewhat after Americans reproduced the effect in 1989.
Many people (i.e. graduate students) who worked with Pd/deuterium system noticed anomalies in the system for decades, and it was folklore in the chemistry community that deuterated Palladium acts up, while Hydrogenated Pd does not. Pons and Fleischmann decided to get serious about the anomalies, and did extremely careful calorimetry on the system for many years, until they were certain they had a reproducible effect for which chemistry could be safely excluded. Then they held their press conference, and chaos.
Part of the problem is that once they claimed fusion, people insisted that the fusion should emit neutrons, just as hot fusion does. This is impossible, because, considering the energy released, the number of neutrons would have cooked Pons and Fleischmann. Then nuclear physicists demanded that they measure nuclear effects, and they tried to do this, but their nuclear measurements were riddled with errors, and it is possible that they fudged a plot that they showed at a conference (although considering Fleischmann's impeccable scientific integrity, I find it more plausible that they made an honest mistake). It is important to note that their published paper contains only calorimetry data, and no nuclear data of which they were unsure.
Some people speculated that the effect can be explained by chemistry, or by insufficient stirring, or by storing electrical energy for later release. These claims are all idiotic. The effect is not small, the only reason it requires instruments to detect is because Pons and Fleischmann deliberately used a tiny Pd wire as a cathode. When they used a bigger Pd plate, the thing melted the table, and blew a hole in the concrete floor below. There just is no source of chemical energy, nor a battery, which can store energy chemically at more than about 1eV per atom. Other people noticed similar runaway explosions too.
Aside from the explosion, there is possible heat from recombination, which has often been emphasized by critics. The electrodes separate H2 from O2, and if the two mix together and the hydrogen burns, you will see excess heat. To control this, groups used infrared cameras to locate the heat source at the cathode, rather than the water where bubbles of gas can mix. They also separted the anode and cathode. But most definitively, in 1994, Pons and Fleischman demonstrated heat after death in 1994, where they cycle the temperature in the anomalously heating cell up, then shut down the current entirely. The cell continues to produce heat for hours with no current, no oxygen, no hydrogen, and many times more heat than you can store in the cathode by any chemical means.
The effect is very sensitive to the metallurgy of the Palladium, and Pons and Fleischmann couldn't reproduce the finicky effect on demand once they ran out of the good palladium. The experiment sometimes takes weeks, and many people just didn't have patience. Still, the effect was reproduced immediately in a handful of places. MIT ran an infamous reproduction that noticed excess heat production, and were going to press with a reproduction. Then they realized that this effect was going to be labelled bogus, and they cut off their graph to show no excess heat. One of the graduate students who was involved in this experiment, Eugene Mallove, was so outraged that he quit his position and became a cold fusion promoter.
Several groups published reproductions. These groups were attacked in the most unscientific of ways. Several groups, Bocris at texas, but also reputable researchers at Bhabha institute and Los Alamos, reported low levels of tritium production in the system. Since tritium is radioactive, it has a clear signature, it can't be mistaken for anything else. Since it is very expensive and is ordinarily produced in nuclear reactors, the only way such a signal could be seen is if it was deliberately faked by spiking the heavy water with tritium. Bocris was accused of doing just this--- spiking his cells with tritium, so confident were the deniers that he had committed fraud. Despite the intense pressure, he never retracted his claim. Another colleague at Texas who claimed tritium, Wolf, did retract the claim when he saw what was happening to Bocris, and never spoke in support of cold fusion again. Bocris was investigated for scientific misconduct and exonerated. No plausible way he could obtain tritium (other than cold fusion) was ever found. The tritium observations would require all the researchers who observed tritium to be engaged in deliberate fraud. It is impossible to misidentify tritium.
Two extermely well respected theorists, Julian Schwinger (emeritus UCLA) and Peter Hagelstein at MIT, were convinced that the effect was real. Schwinger was not allowed to publish in the field, and Hagelstein, who was tenured, had all his funding cut and was moved into a closet.
In the early 90's, without any official funding, McKubre quantitated Helium production to correlate with the excess heat. Low levels of ordinary fusion are seen to happen in the system. The SPAWAR group in the U.S. Navy reproduced the effect in co-deposition experiments, where they plate Pd onto a surface in the presence of heavy water. Their experiments mostly detect nuclear products, because the plated surface is so small, but the effects are 100% reproducible. More recently, the navy presented evidence of sporadic high-energy neutrons coming from the co-deposition system.
In Japan, Mizuno noticed new elements being produced in the Pd system with abnormal isotope ratios and atomic number near Pd (this was also detected by Wolf, as reported by Eugene Mallove, but Wolf would not publish after the tritium fiasco). Abnormal isotope ratios cannot be fraud, because such materials are so difficult to make. In Japan, Arata reproduced the effect by using gas-loading of deuterium into Pd, which has no heat source, so there is no calorimetry error to blame things on. This was a foolproof version of Pons' and Fleischmann's "heat after death" experiments, and it also rules out calorimetry/recombination error entirely. The effect has been reproduced many hundreds of times, in far away labs with no mutual interests, and everybody should be certain by now that it is real. I am ashamed of myself, in that I couldn't bring myself to trust the experimental data until I came up with a reasonable theoretical story to explain it.
Focardi/Rossi claims are more dubious. Their effect is in Nickel/Hydrogen, which has some reported energy anomalies too, but not with the same level of confidence. In terms of theoretical craziness, Nickel Hydrogen fusion is to Palladium Deuterium fusion as Palladium Deuterium fusion is to standard hot fusion. The cold fusion community is taking a wait-and-see attitude, but I think consensus is that the device is not likely to work. In his demonstrations, Rossi measured heat production using steam, not water, and by understating the water-content of the steam, you can inflate the energy output by the latent energy of vaporization, which is huge. For me, it is most suspicious that his claimed transmutation products were analyzed and have natural isotope ratios. It is possible that his machine works, and it is possible that it does not produce any excess energy at all, we will know soon enough. What is impossible is that there are no nuclear effects in Pd/deuterium.
Here is an update regarding the e-cat, which, as people expected, is a sophisticated scam: Is the E-cat for real?.
One major difficulty for acceptance of the effect is that theoretical work in this field is not sound. There are several theories, each of which are more or less preposterous. The central difficulties are overcoming the Coulomb barrier somehow, and making energy without nuclear reaction byproducts:
- Hydrinos/little hydrogen: this theory states that the electron in Hydrogen can find a closer orbit than the ground state, and spends some times close to the nucleus. This requires that quantum mechanics is wrong, or that there is some new electron/proton force which has been missed, and somehow does not alter the ground state energy, but is capable of sucking the electron into the proton every once in a while.
- Bose-Einstein condensed deuterons/alphas: this idea is that the cross section for fusion is enhanced by identical-particle effects, since deuterons and alphas are both bosons. In theory, you can enhance reactions by having a coherent source of bosons all go through the same reaction to a coherent superposition. This theory fails both because the temperature is too high for coherence between deuterons, and because when it is implemented in specific cold fusion papers, the deuterons are treated as non-interacting particles in a product state, so that the amplitude for being at the same point is big. But this is ignoring the whole difficulty, because the electrostatic repulsion leads the wavefunction to be entangled, with little probability of any two deuterons getting to the same point.
- Lattice enhancement mechanisms: This was the focus of Schwinger and Hagelstein, neither of whom claimed to have solved the problem. The problem with such theories is only that the effects have to be collective over thousands of atoms to explain taking eV energies into KeV energies, and it is thermodynamically difficult to imagine how you can take such entropic energy into such an entropically unfavorable place as a single particle.
- Weak Force Neutron production: The Widom Larson theory claims that it is possible for a proton and an electron to do inverse beta decay on the surface of a metal, where there are large local electric fields. This is preposterous, because of the MeV difference in proton and neutron mass. It requires millions of volts to accelerate an electron to enough energy to be able to do an inverse beta-decay, and such energies are not available on the surface of a metal. Further, this theory will predict transmutations of plus/minus one mass unit predominanatly, which is not observed, and does not explain how a deuteron can absorb an electron.
The following lists are bogus theories I speculated would work, other people come up with these too every once in a while:
- sporadic atmospheric muon capture: The idea there is that muons are captured by the metal, and lead to fusion. This doesn't work, just because there are not enough muons, the deuterons are separated from each other in the lattice, and if the muon is captured by a Pd nucleus, it's wasted.
- tunneling with weird many-body enhancement: the idea is that the tunneling amplitude is always estimated, not calculated, and this is an impossible-to-solve many electron/many nucleon system, so perhaps the tunneling amplitude is just off by many orders of magnitude. This doesn't work, because there is a way of giving a lower bound to tunneling amplitudes which excludes any appreciable fusion reaction by tunneling. To do this, you exploit the fact that tunneling is a ground state property, and the deuterons which are imagined to tunnel are bosons, and their imaginary time ground state has no nodes. The electrons have nodes, since they are fermions, and at high energies, but when the electron states are all fully occupied, they might as well be a vacuum, with structure only near the Fermi surface (this follows from the particle-hole symmetric approximate description of the Fermi liquid). There were rigorous upper bounds on the tunneling probability for deuterons in a metal that claimed to prove cold fusion is impossible.
The previous failures leads one to expect that the effect is out of equilibrium, and involves highly excited atoms.
My Personal Theory
To bridge the gap between the scale of chemistry at eVs and of the nuclei at MeVs, one should take note of the fact that there are K-shell electrons orbiting very close to the nucleus at KeV energies. The K-shell electron of Pd has a 20KeV ionization energy, and if you have a K-shell hole in one Pd atom, it stores an amount of energy non-entropically in an amount sufficient to lead to deuteron fusion. While this energy is large, it is not large enough to knock a Palladium atom out of its lattice position, so it cannot dump its energy by locally breaking the lattice. The reason is that the Pd nucleus requires more than the 20KeV ionization energy to be knocked out of position without it's core, and you can't conspiratorially transfer the hole energy to the entire core in one step, it's phase-space impossible.
Such K-shell holes usually decay by X-rays, but this is an electromagnetic process which is suppressed by powers of v/c when the electron is nonrelativistic, as it is even in the K-shell. This is a well known effect--- it's the same reason that atomic spectral lines are narrow. Emitting a photon takes many orbits because of the mismatch in scale between the photon's wavelength and the size of the orbit. This is ultimately because the orbit is nonrelativistic. Because the emission takes so long, the spectral lines are sharply defined and narrow, and the emission is dominated by the matrix elements of the dipole moment of the atomic state between stationary states.
Other observed ways for K-shells to lose their energy is to kick out an outer shell electron from a neighboring atom. This process is electrostatic, and nonrelativistic, so it is not suppressed by 1/c factors. It is only suppressed by the smallness of the charge on the electron and the distance between electrons on neighboring atoms. There is a significant fraction of decays in K-holes in Pd in this channel.
In a metal with protons or deuterons, a K-shell hole should be able to also kick its energy into a proton or deutrons by electrostatic forces. The matrix element is exactly the same as for kicking an electron, but the density of states is 30-50 times bigger (depending on whether it's a proton or a deuteron) due to the heavier mass. The proton, unlike a Pd nucleus, will leave its lattice site under such a transfer. So, considering that the cross section for a K-shell hole to kick an electron is not small, I feel safe to conclude that the proton-kicking process is the dominant decay mechanism for K-holes.
These deuterons have exactly the same energy as the K-shell hole, which means that their classical turning point when approaching a Pd nucleus is exactly the same distance from the nucleus electrostatically as the K-shell is wide, about 100 fermis. These holes can then excite another electron coherently, and travel many steps in the lattice before decaying by X-ray to the ground state. These hole-deuteron states make bands of several KeV width at energies around 20KeV, and these bands are full of classical turning points at 100fermis from a Pd nucleus.
Now suppose that two of these accelerated deuterons happen to come close to the same Pd nucleus. This can easily produce a fusion event at the turning point, the deuterons have around 20KeV after all, and the fusion rates at 20 KeV in beams is not that small, let alone in cases where the wavefunction is concentrated near a nucleus with a classical turning point (where the wavefunction is enhanced).
This fusion does not necessarily happen in the usual hot-fusion way, since it is very close to a Pd nucleus. Let us suppose that the fusion transfers the excess energy/momentum to a nearby charged particle electrostatically, the obvious candidate being one of the protons Pd nucleus. Then the alpha particle and whatever it transferred its energy to are moving with 24MeV of energy together, and they go through the metal, ionizing Pd atoms. Energetically, they can make up to 1000 K-shell holes, all within a millimeter, since the penetration depth is so tiny. The true number is more likely a hundred or a few hundred, since all levels are excited during the Bethe process of charged particle ionization. These holes are then banded with deuterons, so they accelerate new deuterons, and this can easily lead to a chain reaction. I believe this explains the cold-fusion.
There are two problems with this idea:
- The cross section for fusion at 20 KeV is not that huge, and it does not lead to a chain reaction by itself through the usual hot-fusion channels. The multiplication factor is around .001 from beam fusion on deuterated Pd, which has a 1 in 100,000 success rate, not 1 in 100, at 20KeV.
- The actual observed reaction produces an alpha particle without an emitted neutron or proton nearly all of the time. This is a 1-in-a-million event in hot fusion.
I think that both problems are related to the fact that the reaction is happening inside a dense metal. The first problem is not present if two deuterons are banded and both turning around near a nucleus, the result is like a directed collision of two 20KeV beams with a very good focusing device (the nucleus) to concentrate the scattering wavefunction.
The fusion of deuterons always happens through unstable intermediate states, and the cross section to alpha particle is only small because of the same non-relativistic issue. To get an alpha, you need to emit a gamma-ray photon, and emissions of photons are suppressed by 1/c factors. When there is a nucleus nearby, it can be kicked electrostatically, and this process is easier than kicking out a photon, because it is nonrelativistic (the same holds for an electron, but with much smaller cross section due to the smaller charge, and there is no reason to suspect concentration of wavefunction around electron density, as there is for a nucleus).
The time-scale for kicking a nucleus is the lifetime of the two-deuteron resonance, which is not very long, in terms of distance, it is about 100 fermis, this is about the same size as the inner shell. If the deuterons are kicking about at random, this coincidence is not significant, but if the deuteron-hole excitations are banded, it is plausible that nearly all the energetic deuteron-deuteron collisions take place very close to a nucleus, as explained above.
There are conservation laws broken when a nucleus is nearby. The nucleus breaks parity, so it might open up a fusion channel, by allowing deuteron pairs to decay to an alpha from a parity odd state. Such a transition would never be observed in a dilute beam fusion, because these fusions happen far away from anything else. This hypothesis is not excluded by alpha particle spectroscopy (there are a lot of relevant levels of different parities), but it is not predicted either.
But since something ha