Return to the list of clickable items
163) On reproducibility of cold fusion experiments
Ludwik Kowalski (7/26/04)
Department of Mathematical Sciences
Montclair State University, Upper Montclair, NJ, 07043
1) According to a very recent article (2004), in the online encyclopedia (Word IQ):
In the best experimental set-up, excess heat was observed in 50% of the experiment reproductions. Various fusion ashes and transmutations were observed by some scientists.
Most people know that random guessing, when someone is trying to predict one of two equally probable outcomes, should result in 50% of correct predictions. In other words they know that, in such situations, the ability to predict 50% of outcomes does not demonstrate that the instrument of prediction is reliable. On the contrary, it demonstrates that the claimed instrument (for example a theory) is totally unreliable.
From this somebody might conclude that the reported 50% reproducibility is the best argument against the above mentioned excess heat. Such reasoning would be incorrect. When 50% of competent scientists see a disputed effect then that effect is likely to be real. I am thinking about a situation in which the number of attempts to confirm the discovery is large. In my opinion, arguments for or against cold fusion should not depend on the number of confirmations; they should be made on the basis of examinations of procedures leading to particular claims. Scientific experiments are not predictions; they are fact gathering attempts. On the other hand, claims should not be taken for granted unless the level of reproducibility, by competent researchers, is close to 100%.
2) Reproducibility is the most essential requirement in experimental science. Let me mention some questions linked with it. Some phenomena, for example, are very difficult to observe. The required instruments (satellites, accelerators, ultra-clean labs, etc.) are expensive and only a limited number investigations can be performed. How many confirmations are needed to accept a claim? That is one question; and here is another. Suppose that results coming from two teams of researchers differ by 30%. Does it mean that the claim made by one team is confirmed by another team? It depends on the nature of the claim. If the purpose of the experiment is to confirm the existence of a phenomenon, such as generation of excess heat, then the difference of 30%, in the amount of released energy, would not be significant. But a 30% discrepancy would be very significant if the purpose was to measure a physical quantity, such as length or current, very accurately.
3) Irreproducibility usually indicates that experimentalists are not aware that something is not the same when apparently identical experiments are conducted. In the case of cold fusion, according to many, the outcome depends on hard to control impurities and defects. This became apparent when electrolytic experiments with palladium cathodes were performed to confirm reality of excess heat. Electrodes supplied by one manufacturer consistently resulted in confirmations while electrodes supplied by another manufacturer consistently resulted in failures to confirm. Furthermore, excess heat was consistently observed with some batches of electrodes supplied by the same manufacture, and not with others. Hard to identify differences, perhaps at the very low level of contamination, are likely to be responsible for the absence of 100% reproducibility. Let me show how this was described by E. Storms. In the 2001 paper, downloadable from the library at <http://www.lenr-canr.org>, and entitled Cold Fusion: Objective Evaluation, Storms wrote:
Presence of reproducible patterns within the data sets are as important as achieving reproducibility of the phenomenon its self. Such patterns are based on AE [anomalous energy] being produced only under certain unique conditions by all studies. . . . The work at SRI, lead by Dr. McKubre, studied the phenomena, first with $6M provided by EPRI (Electric Power Research Institute). After this program was terminated, work was continued by IMRA (Japan) at nearly the same level. A very small effort is now being funded by the US government. Over the course of this work, several designs of flow-type calorimeters were used that share the following characteristics:
A. The cells are sealed and contain a recombiner. As a result, no gas leaves the cell. Therefore, uncertainty in the amount of recombination is not an issue. Successful action by the recombiner is monitored using different methods including change in gas pressure.
B. The cells contain a heater, which maintains a constant inner temperature. Power to this heater can be adjusted to compensate for any change in temperature caused by electrolysis or by anomalous processes. This heater is also used to determine whether the power measurement, based on the flow rate and temperature change of a cooling fluid, is accurate. Sensitivity better than ±0.01 W (±0.1%) is claimed.
C. The electrolytic cell, its surrounding heater, and the cooling-fluid channels are all contained within a silvered, evacuated Dewar in order to isolate them from the environment.
D. The whole assembly is immersed in a fluid bath, which maintains a constant environment of 30±0.003°C. This bath is also the source of cooling fluid. Consequently, most studies are done at a temperature above 30°C.
E. A constant flow pump is used to circulate cooling fluid. Flow rate is checked periodically by weighing the fluid passing through the calorimeter. Better than 98% of power produced within the cell is captured in this fluid.
F. All aspects of the measurement are under computer control, which provides continuous monitoring, and redundant RTD sensors are used for temperature measurement.
G. The deuterium content of the palladium cathode is determined by measuring its change in resistance.
Nineteen samples of palladium were found to make AP with consistent results. One consistent behavior is shown in Figure 1 as the relationship between AP and average composition of the cathode. Figure 2 shows another pattern as the effect of applied current on AP. These two behaviors are found to be produced by all samples of palladium used in this work and by all samples reported in the literature when the necessary measurements were made. Variations in reported values are easily explained by the different shapes and sizes of the cathodes used, and by the amount of active material present on their surfaces. In addition, the following behaviors are also seen by everyone who has made suitable measurements.
A. The average D/Pd ratio must exceed a critical value. This value differs somewhat between studies because only the average composition can be determined and the value depends on the method used and the shape of the cathode. Typically, the value lies between D/Pd=0.85 and 0.90. Infrequently, compositions above this range are found to be dead for unknown reasons.
B. The current must be maintained for a critical time. This time is variable and presumably depends on how rapidly the surface can acquire the active structure. The time is zero for thin layers of Pd while it can be as long as months for bulk palladium. Failure to wait the necessary time is one reason some people have not seen the effect.
C. The current density must be above a critical value. Applied current determines the surface composition, hence the nature of the active structure. A value above 150 mA/cm 2is usually found for bulk palladium. No critical value appears to be necessary for thin layers of palladium.
D. Inert palladium can sometimes be activated by addition of certain impurities to the electrolyte. These impurities are found to help the surface achieve a higher deuterium content.
E. The effect occurs in only a small fraction of samples, but more often in certain batches than in other ones. In fact, all physical properties of palladium are found to be batch specific, making this metal highly variable in its general behavior, even in conventional applications.
These patterns of behavior add evidence that the observations are a real behavior of nature and not caused by error.
4) My accpted Letter to the Editorere of Physics Today (to appear in the September 2004 issue ?) ends in the following way. . . . . I also agree with chemist Allen Bard that being able to reproduce experimental results is not good enough; it is only a preliminary step. But wasnt poor reproducibility the central point of criticism when cold fusion was first investigated 15 years ago? In my opinion, experimental claims should not be disqualified solely on reproducibility; validation should depend on credentials of researchers and, above all, on methodologies they used in particular experiments.
Return to the list of clickable items