A chemical solution that creates electricity from heat could become an economical solution to many of today's energy problems. Low grade heat represents a vast, but often unexploited, reserve of energy which could help power our modern economy if it could be changed cheaply into electricity. Sources of low grade heat include passively collected solar heat, waste heat from industrial processes (e.g. cooling water from fossil fuel or nuclear electric generation) geothermal energy or any other heat source which is not very hot relative to ambient temperature. The key to success in utilizing low grade heat is to focus on the economics of the process. The laws of thermodynamics dictate that conversion efficiencies will tend to be low. However, if the capital cost of the conversion technology is correspondingly low, a valuable energy resource will result. Heat can be converted to electricity by using mechanical devices, such as engines or turbines, to drive an electric generator. These systems can be quite efficient, but they are also complex, hence expensive, and they are not well suited to operation over small temperature differences (e.g. 100°C or less). The only presently used technology for direct, non-mechanical, conversion of heat to electricity is the thermocouple. Thermocouples made from junctions of dissimilar metals produce small voltages (10's of *V/°C) and are useful mainly for instrumentation (i.e. measuring temperature). They are too inefficient to be used as commercial power sources. Thermocouples made from semiconductors, most notably n-doped and p-doped bismuth telluride, can produce higher voltages (up to 0.35 mV/°C) and have found use as thermoelectric generators in some specialized applications. However, bismuth telluride devices are expensive and work best at temperature differences exceeding 250°C, limiting their present use to applications where long-term high reliability is paramount, such as on spacecraft. In principle, heat can be converted directly to electricity using an electrochemical cell (thermocell). Unlike a conventional voltaic cell, a thermocell would not deplete its internal chemicals as it operates (nor would it be resupplied with reactants, as in a fuel cell), but rather it would maintain a constant internal chemical composition, drawing energy solely from a temperature difference between its electrodes. Few chemical systems suitable for use in thermocells have been identified and none have yet approached commercial feasibility. The technology is tempting, however, because of its potential for large-scale realization at very low capital cost. The investigation presented below describes the identification and testing of a safe, inexpensive thermocell formulation for future development: the Energy Solution. The Energy Solution was discovered by accident. It was the original intention of this investigation to study chemical means for collecting solar energy. Water-soluble ionic compounds were selected from listings in the Merck Index on the basis of their indicated photo-sensitivity. Several compounds were tested for evidence of photo-dissociation in aqueous solution, the objective being to recombine a photo-dissociated compound in a fuel cell, and thereby capture electricity from sunlight. No candidate compounds proved suitably photo-reactive in preliminary tests. However, in testing copper (I) chloride, made by electrolyzing hydrochloric acid with copper electrodes, a voltage was noticed when bright light preferentially illuminated one of two copper electrodes in a beaker. Further testing revealed that it was warmth from the light, not the light itself, which produced the voltage. It was decided to redirect the investigation to the study of thermally induced electrochemical effects. A number of chemical reagents were obtained for use in tests. They were selected on the basis of availability (to the present investigation), safety, cost and similarity to copper (I) chloride. The list included copper (II) chloride dihydrate, iron (III) chloride, copper (II) bromide, sodium chloride, ammonium chloride, ammonium bromide, lithium bromide, ammonium iodide, and potassium iodide. As well, copper metal and iron metal were used in some tests as reducing agents. Carbon electrodes were used in all tests because of the relative inertness of carbon to the solutions. A series of preliminary tests was carried out. These tests were qualitative in nature so that they could be done quickly and systematically, thus allowing a large number of solutions to be evaluated in comparison with one another. The preliminary test setup comprised a pair of insulated carbon electrodes, one contacting the solution at the bottom of a 50 ml test tube, and the other contacting the solution near the top (see illustration). The test tube was set into a heavy brass cylinder which served as a heat sink at the bottom, while a 75 W floodlight was used to heat the top. More than 20 formulations were tested, each test commencing with the floodlight being turned on and continuing for one hour. During this time the voltage developed by the electrodes, connected across a 25 * load resistor, was logged using a digital multimeter reading millivolts. All results were plotted on voltage vs. time graphs to facilitate easy visual comparison of the data. The load resistor and extended test duration were used to differentiate thermoelectrically produced voltage and current from more transient effects as might result from contamination of the electrodes or imperfect mixing of the test solutions. Some tests were conducted under a layer of mineral oil to reduce any effect of dissolved atmospheric oxygen on the observed results. The preliminary test sequence was guided by the results observed as the tests were performed. Initial tests showed that only solutions containing copper or iron cations produced any measurable thermoelectric current. Both copper (II) and iron (III) solutions produced current, but the voltage developed across 25 * was greatly increased when a small amount of ammonium iodide was added. This effect was initially mistakenly attributed to the presence of the iodide ion. It was subsequently determined that the effect of the ammonium iodide was to partially reduce copper (II) or iron (III) to copper (I) or iron (II) respectively. The copper solutions gave three times the thermoelectric voltage of the iron solutions, so testing was focused on the copper solutions. Chlorides and bromides behaved similarly, with copper chloride solutions performing slightly better. Copper or iron solutions reduced with the respective metal, performed equivalently to those reduced with ammonium iodide. To solubilize copper (I), it was necessary to include an abundant source of excess halide ions in the solution. Hydrochloric acid, sodium chloride, ammonium chloride, copper (II) chloride, lithium bromide or ammonium bromide sufficed for this purpose. Best results were obtained with the ammonium salts. A comparative test of solutions of copper (II) chloride, copper (I) chloride and a mixture of these, showed much higher thermoelectric voltage across 25 * when the mixed copper (I) chloride / copper (II) chloride solution was used. Based on this and the other results of the preliminary tests described above, a solution comprising 21.2% ammonium chloride (4.0 M), 3.5% copper (II) chloride (0.26 M), and 3.3% copper (I) chloride (0.33 M) was selected for use in all further testing. The thermocell solution as described above, was further investigated in a set of tests designed to determine its behaviour at different temperatures. An apparatus was constructed consisting of a 50 ml test tube and pair of carbon electrodes as described above, along with a wrap-around resistance heater for the top of the test tube, and an ice-water cooled circulation jacket at the bottom of the test tube. Using this apparatus, the temperature of the solution contacting the upper electrode and the lower electrode could be adjusted over a wide range (18°C to 103°C top; 25°C to 4°C bottom). Two thermistor probes (one top, one bottom) were positioned in the solution in close proximity to the electrodes. Two series of tests were carried out, one under open-circuit conditions, and one under load conditions. In the open-circuit tests, the high side and low side temperatures were varied upwards and downwards in steps, while these temperatures, along with the thermoelectric voltage produced across the electrodes, were constantly logged . Following these tests, an AC resistance measurement was performed and the thermocell was found to have an internal resistance of 13.9 *. A second set of temperature dependence tests was carried out with the cell under load. A 12.9 * resistor, selected to approximately match internal resistance and hence maximize load power, was connected across the electrodes and high side and low side temperatures were adjusted in stages over a three hour duration. As in the open-circuit tests, temperatures and voltages were logged continuously at approximately half-minute intervals. Analysis of the temperature dependence data showed that the copper (I) chloride / copper (II) chloride thermocell system behaved remarkably like a simple thermocouple over the range of temperatures investigated. Plots of electrode voltage vs. temperature difference showed linearity to within experimental uncertainty. The open-circuit voltage of the thermocell was 0.90 mV/°C, while the voltage across the 12.9 * load resistor was 0.44 mV/°C. The internal resistance of the cell, as measured explicitly in the AC resistance test, closely accounted for the loss in thermoelectric voltage under load. It was noted that the equivalent to the Seebeck coefficient, 0.90 mV/°C for the thermocell, was higher than in any presently used metal- or semiconductor-based thermocouple system. Additional tests were performed to determine the extent to which reaction rate effects might limit the thermocell output. Reaction rate effects could include limitations on the diffusion rate of charged ions in the solution or limitations on the rate at which electron exchange occurred at the electrodes. A four-electrode arrangement was built by twinning double-electrode assemblies, the individual assemblies being constructed as described above. One pair of electrodes could be tested separately, or else the two pairs of electrodes could be tested in parallel, effectively doubling the electrode surface area. While maintaining a constant 82°C temperature differential, a series of forty measurements was made, using different resistive loads incremented in approximate steps of 2.5 * from 0.75 * to 50 * (20 measurements each for standard area and double area). The results were plotted, yielding a straight line with slope given in ohms (the implicit internal resistance). The explicit internal resistance of the two different electrode combinations was measured using an AC resistance apparatus, and the results were compared with the implicit measurements. The comparisons showed that the voltage vs. load relation could be accounted for entirely by internal resistance when using the double-area arrangement, but that the voltage drop-off rate was slightly higher than expected when using just one of the electrode pairs (standard area). This implied that reaction rate effects began to limit the thermocell current noticeably when the current density at the electrodes exceeded 2 mA/cm². A short circuit recovery test was conducted to determine how quickly the thermocell could replenish a population of ions sufficient to produce nominal open circuit voltage following an extended condition of short circuit. It was found that the cell could regenerate 90% of its open circuit voltage within five seconds, following a fifteen minute interval of short circuit condition. This implied that the chemical process driving the thermoelectric current reaches equilibrium rapidly. The principle of operation of the copper chloride / ammonium chloride thermocell was deduced from observations made during the solution identification phase of the research, augmented by thermal and electrical analysis data, and by data found in the literature. The illustration shows the basic concept. The action of the thermocell results from the temperature dependence of the equilibrium: The potential of this reaction relative to the standard hydrogen electrode is +0.153 V at 25°C, with a temperature coefficient of +0.776 mV/°C (complexing of the Cu ions can change this value significantly). Thus, considering the hot electrode relative to the cold electrode, a positive potential is developed at the hot electrode and the equilibrium is shifted to the right. When current flows in the external circuit, the concentration of Cu+ rises near the hot electrode, while Cu2+ is favoured near the cold electrode. The concentration gradient drives Cu+ toward the cold electrode and Cu2+ toward the hot electrode, thus balancing the flow of charge in the external circuit. The efficiency of a thermocell is found by dividing the electric power provided to the load by the total heat drain through the cell. When the efficiency, E, is low (less than a few percent) it can be shown to be given approximately by E = *²*T / 4** where * is the Seebeck coefficient, *T is the temperature difference, * is the resistivity, and * is the thermal conductivity of the solution. For the copper chloride / ammonium chloride thermocell, the efficiency is just under 0.1% when operating at a 100°C temperature differential. This efficiency is probably below the cutoff where the system would be economical for solar heat collection due to land-use considerations. It would have more chance to be economical in industrial waste heat utilization. However, future work can raise the potential! The copper solutions tested had 3 times the thermoelectric voltage of the iron solutions tested. A further factor of 3 would boost the efficiency to close to 1% where commercial success could become a reality. In its simplest form, a thermocell would consist of two layers of flexible carbon or graphite sheet, lying on either side of a blotter containing a thermoelectrically active solution. Such units, costing only a few dollars per square metre to manufacture, could be used as passive solar collectors or could be wound into heat exchangers and used to recover waste heat from power plants or other industrial processes. Sunlight can deliver more than 1000 W/m² to a horizontal surface, so that collecting heat from sunlight at 1% efficiency could yield 10 W/m² of electrical energy. Electricity produced at $1 /W (capital cost; zero fuel cost) is economical in today's market. Similarly, a 1000 MW thermal power plant, running at 25% efficiency, would produce 3000 MW of hot water. Converting this to electricity at 1% efficiency would add 30 MW to the capacity of the plant. Many energy solutions remain to be tested. They could be based on copper ions complexed in different ways or on other polyvalent ions such as tin, cobalt, or titanium. The methods and benchmarks set out in this work can help lead toward a new generation of thermoelectric heat recovery devices.

Back to the main page.