To this point in the series on thermoelectric devices (related posts here, here, and here.) the discussion has focused on their use as a cooling/heat pumping mechanism. The underlying phenomena also work in reverse; if one maintains a temperature difference across a thermoelectric module (TEM) by external means, it’s possible to extract a portion of the thermal energy required to do so in electrical form, without need of any moving parts. When used in such fashion it is common to see TEMs referred to as thermoelectric generators or TEGs, though these terms may also refer to an overall system or module for thermoelectric power generation rather than the thermoelectric device specifically by itself.
This page discusses several factors involved in thermoelectric power generation and its practical limitations, offering guidance on TEG system modeling for those interested in the technology and insights into the design and behavior of products based on it.
A tool that’s useful for identifying the boundary between practical possibility and wishful thinking regarding thermoelectric generation (and gaining insight into a lot of similar topics as well) is the concept of Carnot efficiency. Leaving the details of its derivation and proof to a class on thermodynamics, the basic idea is that the “efficiency” of any heat engine has a theoretical absolute maximum value described by equation 1, where TC and TH are the absolute hot- and cold-side temperatures between which the heat engine is operating.
Equation 1: Theoretical maximum (Carnot) efficiency of any heat engine
Given a hot-side temperature of 100°C (373K) and cold-side temperature of 27°C (300K) for example, an ideal heat engine would display an efficiency of:
For every 100 watts of thermal power supplied to the hot side of a heat engine in such a case, at best only about 20 could be extracted in electrical or mechanical form. Real heat engines are not ideal of course, and actual results will be less than the theoretical Carnot limit; when speaking of thermoelectric generation a factor of ten or so is a reasonable starting point for making ballpark guesstimations, at the time of writing. Stated differently, a thermoelectric generator operating between these temperatures might reasonably be expected to convert about 2% of the thermal energy supplied to it into electrical form.
Pausing and backing up slightly for a (long) moment, this “heat engine” concept merits some discussion. It’s a generalized term for any device or system that extracts energy in mechanical, electrical, or other convenient form from the natural flow of thermal energy from a region of high temperature to one of lower temperature. It’s easy (unavoidable, in fact) to go the other direction; whenever one tries to store, convert, transport or do most anything else with mechanical or electrical energy, friction-like phenomena end up transforming some part of it back to thermal form.
The concept of a heat engine is quite analogous to that of a water wheel; used since antiquity (and still today in updated form) a water wheel takes advantage of the fact that water flowing under the force of gravity can be used to do useful work; directing a stream of water against a bucket- or paddle-rimmed wheel causes the wheel to rotate, and that rotation can be used to do things like mill grain, crush metal ores, or in more modern use produce electricity by spinning a generator. Similarly, as thermal energy passes from a region of high temperature to one of low temperature, it’s possible to convert a portion of it into a more convenient form. This analogy is helpful in that it provides some points of intuitive insight about the basic nature of heat engines;
- Water poured onto the top of the wheel also has to exit at the bottom.
- Larger flows of water over a water wheel translate to higher potential work output.
- Larger differences in height between the top & bottom of the wheel translate to higher potential work output.
Figure 1. Diagrams of a water wheel (left) and a heat engine right).
A more precise tool for conceptualizing the operation of a heat engine is analogy with an electric circuit. Ideal, infinite thermal reservoirs can be modeled as voltage sources, and an ideal heat engine can be modeled as a resistor. (An electric motor might be a better analogy, but for sake of simplicity here, the resistor will be used.) A circuit representing an idealized heat engine system in this way is shown in figure 2. A look at this circuit/model suggests a few things:
“Current” (in the conventional sense) flows out of the higher-voltage source (VH) through resistor (R), and into the lower-voltage source (VC). Likewise, thermal energy flows from a region of high temperature to one of low temperature.
If the low-voltage source VC is removed, the circuit is broken and “current” flow through the resistor does not occur, thus no “power” can be converted by the resistor. Similarly, a heat engine -must- have access to a low-temperature region into which thermal energy can be discharged in order to operate.
The total “power” delivered by the higher-voltage source is either converted by the resistor, or absorbed by the lower-voltage source. Similarly, the thermal energy supplied to the hot side of a heat engine is either output in some other form, or discharged to the low-temperature region.
The expression for the ratio of the “power” dissipated by the resistor (PR) divided by the “power” (PH) delivered from the higher-voltage source in this model has the same form as the Carnot limit.
The only way that the entirety of the “power” delivered by the higher-voltage source can be converted in the resistor is if the lower-voltage source has a value of zero. Most Earth-bound heat engines are stuck with using Earth surface temperatures (300K, give or take) as their low-temperature region, which limits practically achievable conversion efficiencies.
Figure 2. Circuit model of an ideal heat engine
To make a long story short, it is simply not possible to convert all the thermal energy supplied to a heat engine into a usable output, for lack of access to a cold-side thermal reserve at absolute zero temperature. By convention, “efficiency” in terms of a heat engine refers to the energy it outputs as useful work divided by the thermal energy supplied to if from the hot-side heat source. Under this paradigm even a perfect, ideal heat engine is said to have near-zero “efficiency” as the temperature difference across which it operates approaches zero. That same theoretically perfect and ideal heat engine operating between a comfortable room temperature and, for example, the melting point of aluminum could not possibly achieve more than about 70% efficiency.
It is perhaps unusual to equate the concept of perfection with something less than 100% efficiency, but this is the state of affairs when discussing heat engines. Real systems are not ideal of course, and therefore cannot reach even these lesser efficiency figures. This is a thing worth bearing in mind when encountering statements that power plants, combustion engines, and similar heat engine-based systems “…waste xyz% of energy as heat.” It seems likely that many such statements are simply repeated from other sources, without consideration that even a theoretically perfect system would incur a substantial degree of such “waste.”
A further complication of thermoelectric generation is the matter of electrical matching; for a given set of thermo-mechanical operating conditions, maximum electrical output power can only be achieved at one specific electrical operating point. Stated differently, there is one specific value of electrical load resistance that yields maximum output power for a particular temperature difference across a given TEM. Guesstimations of power delivery potential from a thermoelectric apparatus made using the “calculate Carnot efficiency and divide by ten” approach assume such matching; without it, the fudge factor of ten might increase to 20, 50, 100, or beyond.
This matching issue is all but universal, though not of major concern in many circumstances and therefore commonly overlooked. Practical electric power sources all posses some amount of internal series resistance, modeled as RInt in figure 3. In cases where this resistance value is relatively small compared to that of a load that one would like to supply, it’s often ignored. As it becomes comparatively large, as in the case of most TEGs, it becomes a limiting factor in the amount of power that the source can deliver.
Figure 3. Circuit model and chart showing effects of matching a load’s resistance to the internal resistance of the source. Maximum power transfer to the load is achieved when the load resistance and source resistance are equal.
From one perspective, the matter can be understood in terms of a bicycle analogy; a person is limited in terms of both how hard and how fast one can pedal. As a result a single-speed bicycle tends to be difficult to ride up steep hills and slow over level ground, because a fixed gear ratio doesn’t allow the rider to operate within their comfortable range as the terrain changes. A multi-speed bicycle provides a matching function between the limitations of the rider and the nature of the terrain, allowing a much wider variety of the latter to be traveled comfortably.
Electrically, a switch mode DC-DC converter can function similarly to the gear system on a multi-speed bicycle, transforming power at one combination of voltage and current on its input side to an equivalent power at a different combination of voltage and current on its output. (Minus a bit for conversion losses—nothing’s free…) In order to do so, a suitable control system must be provided; some means of selecting an appropriate gear on the bicycle is necessary. The concept of maximum power point tracking (MPPT) is likely familiar from discussions elsewhere on solar energy harvesting; the problem is largely the same in a thermoelectric context with similar tools and techniques being useful in addressing the matter.
From another perspective, the electrical matching issue can be considered as a bulk flow problem. A load requiring 10 watts to operate for example, simply cannot be continuously powered by a TEG capable of delivering only 5. On the other hand, a load consuming only 1 watt from a TEG capable of delivering 5 under prevailing conditions leaves a great deal of capacity unused.
Most potential applications for thermoelectric generation are likely to present with a fixed or poorly-controllable heat source; adjusting the thermal input to a TEG to match electrical demand is not a likely option. As a result, in addition to some sort of maximum power point tracking mechanism, gaining maximum benefit from a TEG system also generally requires some sort of energy storage mechanism; if the supplied load cannot function acceptably with whatever amount of power is produced by the TEG under varying conditions, a battery or similar storage device must be provided in order to buffer differences in the rates of energy production and consumption. This would permit the 5-watt TEG mentioned above to intermittently supply a 10-watt load for example, by storing the excess available during the periods that only 1 watt is required.
Thermoelectric generation applications are categorically different from most refrigeration/temperature control applications in terms of typical system temperatures. In most refrigeration applications, the thermoelectric device operates across some temperature span straddling a comfortable room temperature. The cold side of a TEM in such applications is below ambient temperature in order to draw heat out of some refrigerated zone, the hot side is somewhat above ambient temperature in order to pump that heat that’s being drawn out of the cold region into the ambient environment.
In contrast, the ambient environment is generally the lowest temperature found in a thermoelectric generation system, with system temperatures only increasing from that point. Increasing by a lot potentially, considering that common combustion processes can generate temperatures on the order of 2000°C. System temperatures in thermoelectric generation applications simply tend to run higher than those in refrigeration/temp control applications.
Depending on the actual temperature of the heat source one desires to power a TEG from, this can pose a problem. Commonly-available TEMs of the sort used for cooling purposes may be assembled using solder materials with melting points as low as 140°C, others may be rated for use at up to around 200°C. TEMs rated for higher operating temperatures are also available, but are relatively uncommon as an article of commerce at the time of writing. The obvious problem here is that if there’s a possibility of a TEM being subjected to temperatures at which it melts & falls apart, system reliability is going to suffer…
Beyond that, suitable options for thermal interface, sealant, and similar materials become much more scarce as operating temperatures approach and exceed about 200°C. Mechanical complications due to thermal expansion also grow as system temperatures become more extreme, particularly when discontinuous (start/stop) operation is intended.
Interest in thermoelectric generation appears to be increasing in recent years, due to a variety of factors. Advances in low-power electronics have made it possible to do something useful with ever-smaller power sources, interest in resource conservation is driving investigation of energy recovery from otherwise untapped sources, and the increasing consumption and reliance of modern cultures on electronic-based technology has created and expanded market potentials.
One significant barrier to broader adoption however is limited availability of necessary materials. The most commonly-used thermoelectric materials at the time of writing are based on tellurium, an element said to be similar to platinum in terms of availability and obtained primarily as a byproduct of copper ore refinement, with global annual production estimated to be on the order of 100 milligrams per person at the time of writing. By one rough estimation outlined below, current bismuth-telluride (Bi2Te3) based thermoelectric devices offer potential to generate roughly a watt of electrical power per gram of tellurium used. Complete diversion of annual global tellurium production to thermoelectric energy generation (leaving none left for other purposes) could thus be estimated to enable roughly 100mW of per-capita generation capacity, or about 0.876kWh of potential electrical energy production per year. Per-capita electrical consumption in nations regarded as technologically undeveloped is said to be roughly 15 times this figure; in industrialized nations the factor is closer to 10,000.
Put differently, it seems clear that there’s not enough of the stuff available to support adoption of tellurium-based thermoelectric generation at any significant scale, even assuming the estimate of production potential given above to be low by a decimal place or two. Research into alternative materials with better supply potential and/or performance is ongoing, but commercial availability of tellurium-free thermoelectric modules appears limited at the time of writing. Considering also that electricity production by thermoelectric means typically comes at a cost of a thermal energy input some ten to fifty times greater (at best) than the electrical output obtained, it would seem apparent that the technology’s applicability is limited to niche/specialty applications, rather than holding promise as a means of bulk energy production.
Estimated global tellurium production (2019) = 470 metric tons (470,000kg)
Approximate global population (2019) = 8 billion persons
470,000kg/8,000,000,000 persons = 0.000059kg/person (approx 0.1 gram/person)
Atomic weight of bismuth (Bi): 209
Atomic weight of tellurium (Te): 128
% Te by weight in Bi2Te3 = 3128/(3128+2*209) = 0.48 = 48%
A thermoelectric module of a nominal 40x40x4mm form factor that was close at hand was found to weigh approximately 25 grams. Assuming half of this to be attributable to inactive materials suggests active material content to be roughly 12 to 13 grams. Looking up the atomic weights of bismuth and tellurium shows that Bi2Te3 is slightly less than half tellurium by weight, making the guesstimated tellurium content of such a device roughly 5 to 6 grams. The similarly-sized TG12-8-01LSG TEG module is rated to deliver 5 to 6 watts under good conditions, or approximately one watt per gram of estimated tellurium content.
Part of the hassle in design of a thermoelectric generator is the difficulty of finding TEMs designed and characterized for this manner of use; the information provided for use in context of cooling/temp control applications does not readily translate into information useful for generator designs, though such modules (temperatures permitting) can certainly be used. Due to the higher temperatures typically found in generation applications, devices intended for use as such tend to be rated for operation at comparatively high temperatures; this fact can be useful for recognizing such devices when they are not clearly distinguished from those targeted toward refrigeration applications in product listings.
Figure 4. Annotated datasheet excerpts showing correlation between graphical and tabulated data.
The datasheet for one such device is excerpted and annotated in figure 4 to illustrate the correlation between the graphical and tabulated information provided. With hot and cold sides of the device maintained at 170° and 50°C respectively, the device can produce up to 4.17 watts of electrical power, at a voltage of 3.65V. Taking the temperature difference across the device and dividing by the indicated thermal resistance, it can be seen that the thermal energy input required to do so is approximately 103 watts.
Dividing the (best case) electrical power output by the thermal input, we come up with an efficiency figure. The figure below does differ slightly from the one shown in the table, though not by a large margin; this is presumably a result of rounding/truncation errors in the data provided.
From Ohm’s law, one can calculate current flow at the optimum load point from the power and voltage values given:
Using this value for current flow at the optimum power point, the drop in output voltage from the open-circuit case can be used to calculate an effective internal electrical resistance of the TEM at the stated hot and cold-side temperatures:
With that value in hand, the closed (short) circuit current output by the device under the stated conditions can be estimated. Here again, the calculated value agrees with the one shown in the table to within a small margin.
Taking the ratio of the optimal load resistance given in the table to the effective electrical resistance calculated above, one comes up with a resistance ratio value that agrees nicely with that suggested graphically in another chart from the datasheet, excerpted and annotated in figure 5.
Figure 5. Annotated datasheet excerpt showing optimal load resistance ratio.
The graph in figure 4 describes how device output voltage and power vary with the amount of thermal energy transferred through the hot side of the device under optimal electrical conditions. The thermal resistance of the TEM itself is a fairly stable quantity, varying by only about 6% over a hot-side temp range of 120°C. It should be noted that the indicated thermal resistance figures apply at the optimal electrical load point, and can be expected to shift somewhat as the electrical load is varied. But by how much? The efficiency figures, representing electrical output divided by hot side thermal input, are quoted as being on the order of 2.5 to 5%. If one assumes that an equal amount of electrical power is dissipated in the device’s internal electrical resistance, it could be estimated that some 90 to 95% of the thermal energy passing through the device is “leaking” through its bulk thermal conductivity rather than being “captured” by its thermoelectric properties. Thinking in terms of the water wheel analogy, it’s as if some 90% or so of the water being used to drive the wheel is missing the wheel entirely. Variations in effective thermal resistance with changes in electrical load then will probably not vary from the listed optimum-load conditions by more than about 5%, give or take.
For initial design purposes then, it’s suggested that most TEG elements available at the time of writing can be modeled as simple thermal resistances, neglecting the effects of thermal-to-electrical energy conversion in the thermal model. Doing so should tend to underestimate the amount of electrical power obtainable from a given apparatus; energy drawn out of the TEG in electrical form does not pass through the cold-side thermal assembly, meaning that the temperature difference across the TEG and resulting power generation potential will tend to be higher in practice than estimated by calculations that assume zero conversion of thermal energy to electrical form. Particularly in cases involving convection heat transfer, the errors involved are likely to be smaller than those associated with estimation of thermal resistances associated with the convection process.
Adding hot- and cold-side resistances to represent the interface, conduction, and convection thermal resistances and modeling the temperatures of the heat source and sink as voltage sources, one then ends up with a relatively simple thermal model that can be used to estimate the amount of electrical power that can be produced, shown as figure 6. By solving for the “current” flow through the “resistor” representing a TEG element, and referring to the latter’s datasheet for information relating heat flow to electric output potential, an estimate of thermoelectric generating capacity of the apparatus can be made. The concepts of thermal interface resistances, conduction thermal resistances, and convection thermal resistances are discussed elsewhere, and will not be duplicated here.
Instead, a number of product concepts based on thermoelectric generation will be discussed, and tradeoffs involved in their general design highlighted.
Figure 6. A simplified thermal model useful for initial analysis of TEG systems.
Thermoelectric-powered circulation fans designed to move air within a room when placed on a wood stove, fireplace or similar heating appliance have been marketed for several decades, and are perhaps one of the oldest consumer product concepts making use of thermoelectric technology.
The typical design consists of two aluminum extrusions, a thermoelectric module, and a DC motor to which a set of fan blades is mounted. When the placed on a hot surface, heat travels up through the lower extrusion, through the TEM, and into the surrounding air through the upper extrusion. Electric power generated by heat passing through the TEM powers the motor, spinning the fan blades, causing increased air flow past the cold-side extrusion, improving its ability to dissipate heat into the surrounding air and therefore helping to increase the amount of power generated by the TEM. The hotter the surface, the greater the heat flow through the TEM, the greater the power generation potential, and the faster the fan spins.
Figure 7. Model representing a typical stove fan product.
A common characteristic of such products is the use of a lower extrusion with a long, thin cross section. These products are typically produced with a view toward minimizing costs, likely using standard TEM modules of the sort that might more commonly be used for cooling or temperature control applications. Because such modules are not compatible with the 300-500°C temperatures one might find on the surface of a rustic wood stove, there’s a need to limit heat transfer between the stove and hot side of the TEM in order to avoid damage to the latter from excess temperature. Use of a lower extrusion having a long, narrow cross section achieves that.
The upper extrusion in these products will typically have a larger surface area than the lower one, and will often be formed into a decorative shape. Its purpose is to transfer heat from the “cold” side of the TEM to the ambient air, preferably with as little thermal resistance as possible in order to maximize thermal generation potential. It will have to assume a temperature above that of the passing air however in order to transfer heat from the TEM into that air, and as a result the “cold” side of the TEM will be at some above-ambient temperature. The lower the thermal resistance of this upper extrusion, the lower the temperature of the cold side of the TEM will be and the more effective the assembly will be, all else being equal. Better designs tend to have short, thick connections between the TEM contact plate and fins with large surface areas. More decorative designs tend to have longer, narrower thermal paths between the TEM plate and regions of the extrusion with significant surface area, and as such would seem likely to be less effective from a thermal standpoint.
Most examples of these products also lack any provisions for electrical matching; the TEM and motor are simply connected together electrically and function at whatever operating point they happen to agree on under the prevailing conditions. This is likely to be a point other than the one at which the TEM would deliver maximum power under those same conditions. This, in combination with the deliberate restriction of hot-side heat transfer significantly limits the power generation potential of these products and their resulting performance as air circulation devices. Potential buyers should note that these products typically have unguarded metal blades yet aren’t considered a safety hazard, and calibrate their performance expectations accordingly.
A product concept that seems promising at first, yet has some serious drawbacks when the details start to be considered is the idea of thermoelectric cookware. Sandwiching a few thermoelectric modules between the bottom of a small kettle and a heat transfer plate might allow a person to charge small electronic devices while cooking, potentially serving two needs at once for folks who’d like to do both.
One problem with this idea is that it implies exposure of a thermoelectric device more or less directly to an open flame; after all, there would be very little point in using thermoelectrics to generate power using an electric stove as a heat source… Flames from common cooking fuels can be expected to have temperatures on the order of 1000 to 2000°C, which is rather in excess of the 200° or so to which the more commonly-available “high temperature” TEMs and sealant materials are limited. It’s not necessarily an impossible situation; if the fire’s kept small enough so that the thermal energy it transfers to the assembly can be spread out and passed through the TEMs without causing them to exceed a safe temperature, the idea can work. From a human factors standpoint however, it’s all but inevitable that a substantial number of users will turn the heat up just a bit too far out of impatience or simply by accident, ruining the system.
Avoiding that result limits the utility of the device as a cooking utensil, because the amount of thermal power that can be applied is reduced compared to what would be permissible if the thermoelectric gadgetry were omitted. It simply takes longer to cook stuff if one’s pot forbids use of strong heat sources. Further, the temperature limitations limit the options of what can be cooked, typically to water-based fare; one might boil potatoes or make coffee for example, but operations such as frying or searing would likely be out of the question.
Functionally, the pot contents of such a system act as the heat sink for the cold side of the TEM assembly. The evaporation of water is a fairly energy-intensive process, and quite effective at maintaining temperature. Once the boiling point is reached, the temperature of a volume of water doesn’t increase further to a significant degree; adding more heat simply increases the rate at which the liquid is turned into vapor. Unfortunately (from a thermoelectric standpoint) that process occurs at around 100°C, which isn’t exactly “cold”. If relying on boiling water as one’s heat sink in a thermoelectric system, generation potential will not be very good compared to what it could be if, for example, ice at 0°C were available. There are some places on earth where ice is available and electric utilities aren’t; the concept is perhaps best suited to use in such.
From a mechanical design standpoint, the concept is also problematic. The hot-side heat plate will experience significant temperature changes relative to the cold side plate (the body of the pot) and as such will expand and contract dimensionally relative to it. Holding the assembly together in a way that is sufficiently flexible to keep such expansion from damaging the TEM elements, and sufficiently rigid to ensure reliable thermal contact, without making the assembly unwieldy to handle yet protecting the brittle TEM modules from damage due to impact and reliably preventing moisture intrusion when submerged or exposed to moisture (it’s a cooking utensil, after all…) is not a trivial task.
Finally, such a device seems ill-fated from a user-factors standpoint. A sampling of product demonstrations available at the time of writing suggests that many/most users do not understand the underlying physics well enough to use such devices to their maximum potential. Limitations on temperature tolerance do not seem to be well understood, and an incorrect association between boiling of water with thermoelectric production seems widespread. The idea that power production is linked with maintenance of the pot contents at as low a temperature is possible does not seem well understood, nor the importance of leaving the vessel uncovered as a means of doing so.
Several products of this general type appear to be available or offered at some point prior to the time of writing. None appeared to provide any indication of temperature limitations, or any substantive indications of the conditions required to achieve a specified degree of electrical performance. Potential buyers of such appliances are advised to limit their expectations, and to be prepared for disappointment if the device’s limitations and working principles are not thoroughly understood prior to use.
Another product available at the time of writing is a solid-fuel camp stove incorporating a thermoelectric system capable of generating small amounts of electric power. This product uses a thermally conductive structure exposed to direct flame to carry heat to the hot side of a thermoelectric module, on the other side of which is a heat sink. A fan in the device draws air past the heat sink, helping to extract heat from the cold side of the TEM. This air is then forced into the stove’s combustion chamber, providing a forced draft that aids combustion of the solid fuel being burnt. Importantly, this device also incorporates a battery for energy storage.
Though perhaps a better-conceived product in some respects than thermoelectric cookware, such a device also has limitations of which (potential) users should be aware. One of the advantages of this product’s design is the lack of (nearly) direct exposure of the thermoelectric element to open flame, as well as a practical limitation on the quantity of such flame that can be applied. It’s simply not as easy or tempting to damage the device’s TEM by going overboard with the with the fire, compared with the thermoelectric cookware concept.
Disadvantages include reliance on forced air as a cooling mechanism, and an inherent coupling of the cold-side cooling effect with the hot-side heat input. In order to remove heat from the cold side effectively and thereby protect the TEM from damage due to excess temperature, the device’s fan must be operational. The product’s built-in battery is essential to making that possible during startup, as a slow increase in hot-side temperature can potentially cause the TEM assembly to warm to the point of being damaged, without producing significant electrical energy in the process. Unlike the water evaporation process relied on by the cookware concept, direct air cooling cannot be relied upon to provide a strong temperature limiting effect.
Because the product recycles the air used for cold-side cooling as combustion air for the hot-side heat input, the heat input to the TEM assembly and cold-side heat extraction cannot easily be adjusted independently; in solid-fuel systems such as this, increases in combustion air supplied generally increase the rate at which heat is generated by the combustion process. The thermal resistance of the hot-side heat transfer element is therefore a crucial design factor in a system such as this. If it’s too low, it will not be possible to remove heat from the cold side at a sufficient rate to prevent the TEM assembly from overheating. Too high, and the assembly’s electric power generating capacity will be limited. Because that generating capacity is at some point responsible for cooling the TEM assembly, a too-high thermal resistance of this heat transfer element can also potentially lead to TEM assembly damage due to excess temperature. Complicating matters, the combustion and mechanical properties of potential fuel stocks will vary, leading to uncertainty in the relationship between cooling/combustion airflow and heat input to the TEM assembly. Use of inappropriate fuels therefore also has potential to damage such a device.