Principles of operation

Seebeck effect

In 1821 the German physicist Thomas Johann Seebeck discovered that when two strips of different electrically conducting materials were separated along their length but joined together by two “legs” at their ends, a magnetic field developed around the legs, provided that a temperature difference existed between the two junctions. He published his observations the following year, and the phenomenon came to be known as the Seebeck effect. However, Seebeck did not identify the cause of the magnetic field. This magnetic field results from equal but opposite electric currents in the two metal-strip legs. These currents are caused by an electric potential difference across the junctions induced by thermal differences between the materials. If one junction is open but the temperature differential is maintained, current no longer flows in the legs but a voltage can be measured across the open circuit. This generated voltage (V) is the Seebeck voltage and is related to the difference in temperature (ΔT) between the heated junction and the open junction by a proportionality factor (α) called the Seebeck coefficient, or V = αΔT. The value for α is dependent on the types of material at the junction.

Peltier effect

In 1834 the French physicist and watchmaker Jean-Charles-Athanase Peltier observed that if a current is passed through a single junction of the type described above, the amount of measured heat generated is not consistent with what would be predicted solely from ohmic heating caused by electrical resistance. This observation is called the Peltier effect. As in Seebeck’s case, Peltier failed to define the cause of the anomaly. He did not identify that heat was absorbed or evolved at the junction depending on the direction of the current. He also did not recognize the reversible nature of this thermoelectric phenomenon, nor did he associate his discovery with that of Seebeck.

Thomson effect

It was not until 1855 that William Thomson (later Lord Kelvin) drew the connection between the Seebeck and Peltier effects, which was the first significant contribution to the understanding of thermoelectric phenomena. He showed that the Peltier heat or power (Q_p) at a junction was proportional to the junction current (I) through the relationship Q_p = πI, where π is the Peltier coefficient. Through thermodynamic analysis, Thomson also showed the direct relation between the Seebeck and Peltier effects, namely that π = αT, where T is the temperature of the junction. Furthermore, on the basis of thermodynamic considerations, he predicted what came to be known as the Thomson effect, that heat power (Q_τ) is absorbed or evolved along the length of a material rod whose ends are at different temperatures. This heat was shown to be proportional to the flow of current and to the temperature gradient along the rod. The proportionality factor τ is known as the Thomson coefficient.

Analysis of a thermoelectric device

Practically, the thermoelectric property of a device is adequately described using only one thermoelectric parameter, the Seebeck coefficient α. As was shown by Thomson, the Peltier coefficient at a junction is equal to the Seebeck coefficient multiplied by the operating junction temperature. The Thomson effect is comparatively small, and so it is generally neglected.

While there is a Seebeck effect in junctions between different metals, the effect is small. A much larger Seebeck effect is achieved by use of p-n junctions between p-type and n-type semiconductor materials, typically silicon or germanium. The figure shows p-type and n-type semiconductor legs between a heat source and a heat sink with an electrical power load of resistance R_L connected across the low-temperature ends. A practical thermoelectric device can be made up of many p-type and n-type semiconductor legs connected electrically in series and thermally in parallel between a common heat source and a heat sink. Its behaviour can be discussed considering only one couple.

An understanding of the thermal and electric power flows in a thermoelectric device involves two factors in addition to the Seebeck effect. First, there is the heat conduction in the two semiconductor legs between the source and the sink. The thermal flow down these two legs is given by 2κ(a/L)ΔT, where κ is their average thermal conductivity in watts per metre-kelvin, a (or w²) is the area in square metres of the base of each leg, L is the length of each leg in metres, and ΔT is the temperature differential between source and sink in kelvins. The second factor is the ohmic heating that occurs in both of the legs because of electrical resistance. The heat power produced in each leg is given by ρI²(L/a), where ρ is the average electrical resistivity of the semiconductor materials in ohm-metres and I is the electric current in amperes. Approximately half of the resistance-produced heat in each of the two legs flows toward the source and half toward the sink.

In a thermoelectric power generator, a temperature differential between the upper and lower surfaces of two legs of the device can result in the generation of electric power. If no electrical load is connected to the generator, the applied heat source power results in a temperature differential (ΔT) with a value dictated only by the thermal conductivity of the p- and n-type semiconductor legs and their dimensions. The same amount of heat power will be extracted at the heat sink. However, because of the Seebeck effect, a voltage V_α = αΔT will be present at the output terminals. When an electrical load is attached to these terminals, current will flow through the load. The electrical power generated in the device is equal to the product of the Seebeck coefficient α, the current I, and the temperature differential ΔT. For a given temperature differential, the flow of this current causes an increase in the thermal power into the device equal to the electric power generated. Some of the electric power generated in the device is dissipated by ohmic heating in the resistances of the two legs. The remainder is the electrical power output to the load resistance R_L.

The leg geometry has a considerable effect on the operation. The thermal conduction power is dependent on the ratio of area to length, while ohmic heating is dependent on the inverse of that ratio. Thus, an increase in this ratio increases the thermal conduction power but reduces the power dissipated in the leg resistances. An optimum design normally results in relatively long and thin legs.

In choosing or developing semiconductor materials suitable for thermoelectric generators, a useful figure of merit is the square of the Seebeck coefficient (α) divided by the product of the electrical resistivity (ρ) and the thermal conductivity (κ).