August 1998
Introduction
At the 7th International Cold Fusion conference in Vancouver earlier this year, Ohmori and Mizuno reported on a new electrolysis experiment in which a W cathode becomes incandescent under certain conditions. They made preliminary calorimetric measurements on cells operating in this mode and concluded that the phenomenon produced significantly more heat energy than the electrical energy required to stimulate it.
Then, in the latest issue of Infinite Energy magazine, it was reported that several other groups have succeeded in replicating the incandescent W experiment and that they, too, have observed evidence of excess heat production.
The experiment is relatively simple and the desired phenomenon can be produced on demand, in sharp contrast to traditional cold fusion experiments. We have successfully replicated the experiment and measured its heat output accurately with our versatile water-flow calorimeter. Unlike Ohmori and Mizuno, we did not perform any analytical studies on our experiment. We concentrated on the calorimetry.
As far as we could tell, our experiment functioned exactly like that of Ohmori and Mizuno. We repeatedly observed the incandescent cathode effect but we observed no sign of excess heat.
Description of Apparatus
The cell is a cylindrical borosilicate glass vessel, 60 mm in diameter and 105 mm tall. It is fitted with a machined G-10 (fiberglass-epoxy) cap that is sealed to the glass with a large O-ring. Passages in the cap allow the W cathode rod and Pt anode wire to enter the cell through O-ring seals. A thermistor temperature probe in a glass jacket also penetrates the cap, again O-ring sealed. A 4th hole in the cap conveys the gases produced by electrolysis out into a small rubber tube.
The cathode is a 1/16″ diameter pure W welding rod sleeved with thick-walled Teflon tubing that extends to within 1 cm of the end of the rod. These rods are available in two grades: (1) pure and (2) 2% thoriated. We conducted tests with both types of rods. The anode consists of a 0.5 mm dia Pt lead wire that is crimped to a 1 cm2 piece of Pt mesh. The two electrodes are positioned about 2 cm from the bottom of the vessel and are located about 3 cm apart.
The cell was filled with 150 grams of 0.5M K2CO3 solution.
During operation, the cell is contained in a heat exchanger made from ¼” Cu tubing as shown in the photo below. This heat exchanger is contained in a sawed-off 600 mL beaker that holds a liquid coupling agent (water) that provides much better thermal coupling between the cell and the Cu tubing than air.
At the very top of this photo you can see four 100 mfd capacitors that are connected in parallel across the cell. These capacitors snub some of the very high frequency currents that occur during the arc discharges in the cell. They also act as bypass capacitors and provide very high currents to the cell, actually intensifying the discharges.
Note the space between the coils of Cu tubing near the bottom of the assembly. This space provides a view of the interior of the cell, specifically of the W cathode. It is necessary to view the W cathode during operation in the calorimeter in order to confirm that the proper state of incandescence has been achieved.
Cooling water is circulated through the Cu tubing by an FMI precision metering pump at precisely 4.82 ml/sec. The temperature of this water is measured both before and after passing through these coils by thermistor temperature probes located in Tee fittings that are buried in the thick Styrofoam insulation that surrounds the cell.
The heat output of the cell is obtained by multiplying the observed cooling water delta-T by the water flow rate and the specific heat of water.
This photo shows most of the entire system. The cell is inside the rectangular Styrofoam enclosure. The round hole outlined in black on the front of this enclosure is a viewport that looks into the cell. This viewport is thermally insulated with 3 dead-air spaces between 4 plastic panes.
The FMI pump is turquoise-colored and located between the cell enclosure and the computer monitor.
The cylindrical object just below the cell enclosure is a temperature-regulated water bath. Under control of the computer, this bath serves to keep the temperature of the inlet cooling water constant. Since this experiment produces up to 200 watts of heat output, active cooling is required for this bath. This is achieved by controlled additions of cool tap water into the bath. The red solenoid valve can be seen just to the left of the water bath.
Just visible in the lower left corner of this photo is the inverted-graduated-cylinder apparatus for measuring the evolving gas flow rate.
Just out of the picture to the right are the Variac-rectifier DC supply and Clarke-Hess 2330 Power Analyzer used to monitor the input power. The Clarke-Hess is a wide-bandwidth power analyzer capable of accurate determination of the delivered power regardless of voltage or current waveforms up to about 400 kHz. It is ideally suited for the “spikey” current waveforms encountered in this experiment.
We also placed a 250 mfd filter capacitor across the output of the full-wave-bridge rectifier. This and the 400 mfd capacitor (mentioned above) across the cell inside the calorimeter enclosure, served to reduce the current spikes in the external circuit to a manageable level. NOTE: These capacitors did not reduce the current spikes in the cell…if anything they were intensified by the capacitors. Without these filter capacitors, the current waveform at the power analyzer was hopelessly noisy. We observed frequencies ranging out to the limit of our 100 MHz scope. Such a signal poses serious measurement problems and would be very likely to cause erroneous readings.
This oscilloscope display shows the typical voltage and current waveforms at the power analyzer during incandescence. The voltage trace (Ch1) is near the top of the screen (ground is the center of the screen) and the current (Ch2) is at the bottom. The erratic nature of the current trace is a result of the spark discharges occurring in the cell around the cathode. On the right you can see RMS values for the voltage and current which indicate that the electrical power was about 200 watts.
Results
Here are the results of Run 1. The entire time span is only 4 hours and the vertical dotted lines mark each hour. The run begins with about 30 minutes of power-off. Then about 80 volts was applied to the cell as indicated by the Vcell trace. Cell current (Icell) was about 1.6 amps during this “warm-up” electrolysis period. Cell temperature (Tcell) rose rapidly and leveled off at about 70° C.
After 1 hour you can see that the observed heat output power (Pout) matches closely the measured electrical input power (Pin). Actually, Pout runs about 2 watts less than Pin during this period, consistent with the caloric value of the evolving H2 and O2. A measurement of this gas flow rate at this time yielded only 80% of the flow rate expected from the cell current.
At 1.5 hours, Vcell was increased to about 155 v and the cathode became incandescent. Note that Icell dropped significantly due to the formation of a gas sheath around the cathode. Pin rose somewhat and, unfortunately, became rather erratic. However, for the next hour, Pin stayed in the vicinity of 110 watts and Pout followed it closely, indicating no significant excess heat. During this phase of the run, the measured gas flow rate was about 93% of that expected from the cell current. Note that Tcell runs about 75-80° C during this phase. The double-boiler arrangement in the calorimeter works great!
At 2.7 hours we increased Vcell to 193 v to invoke a regime in which the cell rumbles very loudly and the emitted light becomes “blindingly white”. The Pintrace became quite erratic at that point but Pout appears to be following it faithfully and never shows any signs of large excess heat. Tcell reaches about 91°C at the end of this phase. Also during this phase, the measured gas flow rate increased rapidly reaching 2.3 times more than expected from the cell current! Possibly this extra gas was coming from the Teflon insulation around the W cathode which was burning off at a rapid rate at this power level.
The run ended abruptly when the top of the cell popped off (harmlessly)! This happened despite the fact that the cell was vented continuously through the gas flow apparatus!
In Runs 2 & 3 we experimented with different electrical filtering arrangements. In Run 2 we tried a 1 ohm resistor in series with the cell (inside the calorimeter) but it dissipated too much energy when the cell was in warm-up mode at ~4 amps. We removed the 1 ohm resistor for Run 3 but added a couple of smaller low-impedance capacitors in parallel with the large electrolytics in the calorimeter enclosure. This arrangement provided very good suppression of the radio-frequency noise generated in the cell.
We also developed a robust insulator for the W rod made from a solid 6mm diameter Teflon rod drilled out to accept the W rod down the center. This insulator survived the incandescent cell operation very well.
However, neither Run 2 nor Run 3 showed any signs of excess heat.
For Runs 4, 5 & 6 we used a thoriated W rod instead of the pure W rod used in the earlier runs. Thoriated W rods are commonplace in TIG welding and the thorium oxide content provides improved arc stability and greatly reduces melting of the electrode tip.
This change, recommended by Dennis Cravens who has had significant experience with similar cells, made a surprising difference in the appearance of the incandescence phenomenon. With the thoriated rod, the rod temperature appears to be much higher for a given input power. The result is a very bright light emerging from the cell that looks just like ordinary firelight. Compared to the pure W rods, there is relatively little evidence of the blue light usually associated with electrical discharges.
This photo was taken with the thoriated W rod in full incandescence while the calorimeter enclosure was open. You can see the brilliant light coming from within the cell. Also visible in this photo is the outlet water temperature sensor, located in the Swagelok Tee fitting on the right.
The overall level of cell activity with the thoriated rods is quite high and the lifetime of these rods is correspondingly short. Runs 4 & 5 served mainly to demonstrate that fact and to convince us that our usual power balance calorimetry would not work well on such a short-lived and unstable experiment.
We therefore modified the data acquisition software to integrate both input and output power during the run and to display and plot the resulting energies. Using such integration we can measure the energy balance of the experiment without ever achieving the thermal equilibrium necessary for an accurate power balance measurement.
Here are the results from Run 6. The time scale is only two hours and all the action takes place over a 20-minute period towards the end of the first hour.
Note the erratic appearance of the Pin trace, which is due to the rapid erosion of the W rod during the run.
The integrated input energy, Ein, totals 134,505 joules and the integrated output energy, Eout, totals 131,376 joules…2.5% less. We feel that this energy balance is about as good as could be expected under these conditions (sudden application and removal of the input power) and it certainly does not indicate any excess heat being generated in the experiment.
Regarding the escaping gases, we observed a striking discrepancy between the actual gas evolution rate and the theoretical rate predicted by the average cell current. When the thoriated W rod was incandescent, we measured gas flow rates up to 4 times higher than expected…even with a good cold trap in the gas line to condense and remove water vapor escaping from the cell!
Run 7 was a repeat of Run 6 and the results were substantially the same. On Run 7 we observed up to 4.8 times more gas evolving from the cell than expected from ordinary electrolysis. Clearly we had to find out what was causing this “excess gas”.
Starting with Run 8, we added an H2+O2 recombiner to the system. This recombiner consisted of a chamber full of Pd-coated alumina catalyst pellets plumbed into the exit gas line from the cell. The recombiner was located outside the calorimeter enclosure and its main purpose was to remove H2 and O2 from the exit gas stream so any additional gases present could be collected and analyzed.
For Run 8 we simply connected the recombiner chamber directly to the hose leading out of the cell. Shortly after the electrolysis power was applied to the cell, the recombiner began working and the heat of formation of H2O caused portions of the catalyst pellets to become red hot. This heat source ignited the 2H2+O2 stream coming out of the cell and the flame raced back into the cell and exploded the gas in the cell’s head space!
Fortunately the cell is well contained inside the calorimeter and there was no external damage from the explosion. However, most of the internal parts of the cell and the glassware were shattered.
For Run 9 we added a flame arrestor between the recombiner and the cell. This device consisted of a tight roll of fine Cu screen stuffed tightly into a glass tube. The roll was about 8 cm long and was expected to cool the advancing flame front below the ignition temperature thus extinguishing the flame.
It didn’t work! Run 9 exploded just like Run 8, again shattering the glasswork, breaking the W rod, and generally irritating the investigators.
For Run 10 we removed the flame arrestor and installed an all-metal water bubbler trap in its place. The exiting gas has to bubble through the water in this trap
before reaching the recombiner. Due to the construction of the trap, a flashback cannot force its way backwards through the trap unless it expels all the water from the trap…very unlikely.
Run 10 proceeded without exploding the cell but, during the warm-up phase, the gas between the recombiner and the bubbler trap exploded disconcertingly on a regular basis. Fortunately the apparatus withstood these shocks without any damage and we were able to proceed with the run.
Once the air had been purged from the cell during electrolysis warm-up, the recombiner succeeded in reacting all of the exiting gases from the cell. This was dramatically demonstrated by a complete stoppage of bubbling in the gas volume apparatus downstream of the recombiner and cold trap.
However, when we switched the cell into incandescent mode by raising the voltage to ~150 volts, gas began to come through the recombiner, through the cold trap, to the gas volume apparatus and the bubbling resumed. Presumably this was the “excess gas” observed but not isolated on previous runs. Interestingly, the presence of the excess gas made the periodic explosions stop.
Using a 500 cc graduated cylinder we collected all of the excess gas generated by Run 10’s incandescent period, which lasted only about 10 minutes. The total was 215 ml at 34° C (yes, it’s hot in our lab).
We extracted some of the collected gas into a syringe and injected it into a flame. It burned with the characteristic nearly-colorless flame of hydrogen. Assuming it is hydrogen, the collected quantity represents 8.5E-3 moles. If this H2 is coming from the reaction W + 3H2O –> WO3 + 3H2 then 1/3 that many moles of W had to be consumed in the process. That’s 2.8E-3 moles of W or .52 grams. In our haste to rebuild the apparatus after Run 9 we forgot to weigh the particular W rod used in Run 10. However, it now weighs 0.46 grams less than our only remaining unused W rod…pretty close to the predicted weight loss from the reaction hypothesized above.
This little photo shows the rod used in Run 10 alongside a new W rod. The right ends of both rods were aligned so you could see how much of the used rod has been eroded. The rod diameter is 0.062″ or 1.58 mm.
The calorimetric results of Run 10 were very similar to those of Run 6. This time there was a 4% shortfall in the integrated output energy compared to the integrated input energy. Could this energy shortfall be explained by the 215 cc of H2 gas we collected? The formation of WO3 from W and O2 is exothermic and yields 841,000 joules/mole. To get the necessary O2 for one mole of WO3, however, 3 moles of H2O have to be dissociated and that costs 858,000 joules. Thus the hypothesized reaction is only slightly endothermic and, for the quantity of WO3 formed in this run, would absorb only ~50 joules. The observed shortfall is ~8,000 joules. A more likely cause of the shortfall is simply calorimetry error. This is a highly erratic experiment that places severe demands upon the input power measuring instrument and the overall data acquisition system. In addition, the temperature regulation system that keeps Tin nominally constant is unable to handle the rapidly changing heat loads perfectly. The resulting excursions in Tin cause corresponding errors in the Pout signal. In other words, agreement to within 4% on the energy balance in this experiment is probably as good as we can ever expect.
Conclusion
In 10 attempts, we have observed no sign of the large excess heat reported by Ohmori & Mizuno. This is particularly surprising in view of the fact that we have apparently succeeded in duplicating the fundamental phenomena that they were investigating: the incandescent W cathode.
We cannot explain the discrepancy between our results and those of Ohmori & Mizuno. We remain open to suggestions for improvement of our experiments.