FAQ Peltier Elements

These are the three basic thermoelectric effects. The Seebeck effect describes the generation of an electrical voltage when two junctions of dissimilar materials are held at different temperatures – it is the basis of thermocouples and thermoelectric generators. The Peltier effect is its reverse: an applied current transports heat from one junction to the other – it is the basis of Peltier elements. The Thomson effect describes the additional reversible absorption or release of heat along a single current-carrying conductor in which a temperature gradient exists; it is of minor importance in practical sizing.

Almost all common cooling modules are made of doped bismuth telluride (Bi₂Te₃), because around room temperature this material offers the best compromise between electrical conductivity and low thermal conductivity. Other material systems are used for other temperature ranges:

Bi₂Te₃ is processed at around 150 °C or 230 °C (high-temperature variant). The maximum operating temperature is about 30 K below the respective processing temperature.

A thermocouple (couple) consists of one p-doped and one n-doped semiconductor leg, connected electrically in series and thermally in parallel. A Peltier element typically contains from 7 to more than 250 such couples. Since all couples are electrically in series, their voltage drops add up: each couple drops around 0.12 V. A module with 127 couples therefore has a maximum voltage of about 15.2 V. The same cooling power can thus be realised either with many couples at high voltage and low current, or with few couples at low voltage and high current.

The figure of merit summarises the three relevant material properties:

with S = Seebeck coefficient, σ = electrical conductivity, λ = thermal conductivity and T = absolute temperature. Good Bi₂Te₃ modules reach ZT values around 0.8 to 1.0. The figure of merit directly determines the achievable ΔT_max – which is why high-quality and basic modules of the same form factor differ markedly in performance.

The COP (coefficient of performance) is the ratio of pumped heat to electrical power input:

It determines the power-supply size, energy cost and, above all, the amount of heat the heat sink must dissipate – because the hot side must always release Q_h = Q_c + P_el. At small ΔT the COP can be well above 1; at large ΔT it drops to around 0.1 and below. The COP maximum is usually reached below 50 % of the maximum current.

All figures apply for a defined hot-side temperature – usually T_h = 300 K (27 °C), in some datasheets 50 °C. Comparing two modules is only valid at the same reference temperature.

Both values are limits in two opposite operating points. Q_max applies at ΔT = 0 K, i.e. when the cold and hot sides are equally warm – then no temperature difference is produced. ΔT_max applies at Q_c = 0 W, i.e. without any useful load. Every real operating point lies in between: the more heat the module has to pump, the smaller the achievable temperature difference. The characteristic Q_c = f(ΔT) runs approximately linearly between the two limits.

ΔT_max rises with increasing hot-side temperature, because the Seebeck coefficient and material quality are temperature-dependent in the relevant range. A standard module that reaches about 70 K at T_h = 27 °C achieves about 75 K at T_h = 50 °C and around 83 K at T_h = 80 °C. However, this does not mean the cold side becomes colder – the absolute cold-side temperature still rises. For low target temperatures, the coldest possible hot side remains decisive.

The datasheet values for ΔT_max are measured in vacuum. In air, three parasitic heat paths act back from the hot to the cold side: heat conduction through the gas between the legs, free convection at the cold side, and condensation of humidity. In practice the achievable ΔT in air is therefore typically 3 to 8 K below the datasheet value, relatively more for small modules. Anyone who needs the full ΔT works in an evacuated or dry protective-gas enclosure.

Suitable is any module that delivers the required cooling power at the required temperature difference at all – there are usually several. Optimal is the one that solves the task with the best COP, i.e. the lowest electrical power and therefore the smallest heat sink and power supply. The optimal module is almost always slightly larger than the just-suitable one and is operated well below I_max. We check this point as standard during consulting.

Common and sensible is 50 to 75 % of I_max/U_max. The reason: cooling power approaches its maximum exponentially – the last 30 % or so is bought with disproportionately high electrical power, which has to be dissipated as additional Joule heat via the heat sink. Whoever wants maximum efficiency goes below 50 % I_max; whoever needs maximum heat-pumping power in a tight space goes higher – at the cost of efficiency and service life.

For a reliable design we need nine inputs:

1. Required temperature of the object to be cooled (T_c)
2. Heat to be removed (active power dissipation, Q_c)
3. Required cool-down or response time
4. Ambient temperature and its variation
5. Parasitic heat loads through conduction, convection and radiation
6. Available space for the module and the heat sink
7. Available electrical power (voltage, current)
8. Required control accuracy
9. Type of heat dissipation and expected hot-side temperature (T_h)

Sizing is an iterative procedure, because the hot-side temperature depends on the electrical power – and the electrical power in turn depends on the temperature difference. Procedure in six steps:

1. Estimate the hot-side temperature: T_h = T_amb + (Q_c + P_el) · R_{th,heat sink}
2. Form the temperature difference: ΔT = T_h − T_c
3. Read the required current at this ΔT from the Q_c characteristic field
4. Read the corresponding voltage from the U-I field → actual P_el = U · I
5. Recalculate T_h with the actual P_el
6. Repeat steps 2–5 until T_h changes by less than 0.1 K

Worked example

Given: useful load Q_c = 20 W, target temperature T_c = 5 °C, ambient T_amb = 25 °C, fan heat sink with R_th = 0.35 K/W.

The procedure converges after four passes: the module must pump 20 W at ΔT ≈ 40 K and needs around 37.5 W electrically for this – so the heat sink must dissipate almost 58 W, not 20 W. The COP is 0.53.

The decisive quantity is the permissible thermal resistance:

From the example above: if T_h must not exceed 45 °C, then at 25 °C ambient and 57.5 W of waste heat, R_th ≤ 0.35 K/W results. For orientation:

Blowing onto the front of the heat-sink base is most effective, because the highest heat-flux density is present there. Suitable air and liquid heat sinks as well as complete Peltier assemblies can be found under Temperature management.

Besides the active useful load, three parasitic heat inputs act on the cold side. On small modules they often dominate the design:

1. Conduction through connecting wires and mounts

Example: two copper strands 0.5 mm² (AWG 20), 50 mm free, λ_Cu = 400 W/(m·K), ΔT = 40 K → 0.32 W. On a miniature module with Q_max = 1 W that is already 30 % of the power. Remedy: thinner strands, manganin or constantan wire, longer free length, thermal anchoring of the wires to an intermediate stage.

2. Convection at the cold side

Example: 10 cm² cold surface, ΔT = 40 K → 0.2 to 0.4 W.

3. Thermal radiation

Example: 10 cm², ε = 0.9, T_h = 25 °C, T_c = −15 °C → around 0.18 W. Remedy: bare, polished metal surfaces (low ε) and radiation shields.

The mass to be cooled requires transient power in addition to the steady-state load:

Example: 100 g of aluminium (c_p = 900 J/(kg·K)) is to be cooled by 20 K in 60 s → 30 W extra. If a phase change is passed through (e.g. water/ice), the latent heat is added. Short cool-down times are almost always the most expensive part of a specification – check whether the time requirement is really necessary.

Yes. For larger areas or higher loads, several modules are connected thermally in parallel (side by side between heat sink and cold plate) and electrically in series. This way all modules carry the same current and work at the same operating point. Requirements:

• Modules from the same production batch with tight height tolerance (±0.03 mm)
• Uniform mounting pressure across all modules – ideally via a common, stiff pressure plate
• Sufficient spacing between the modules so the cold plate is not distorted

An electrical parallel connection is to be avoided: the smallest resistance differences lead to unequal current sharing.

As soon as the required temperature difference lies above what is sensibly achievable single-stage – i.e. from about 65 to 70 K under load. In a cascade, a smaller upper stage sits on a larger lower stage, which must dissipate all of the upper stage's waste heat. Typical guide values:

Rule of thumb: each additional stage adds about 20 to 25 K, but costs roughly a factor of 3 to 5 in heat-pumping power. Our cascade types can be found in the catalogue.

On large modules with moderate heat-flux density the difference is small, because the ceramic's share of the total resistance stays small. On miniature modules with high power density, AlN can bring several kelvin of usable ΔT – there the surcharge is worthwhile.

The decisive factor is the aspect ratio of leg height to cross-sectional area. Short, thick legs give low electrical resistance and high currents at low voltage – but also higher heat back-flow. Long, thin legs give higher voltage at low current and a better ΔT, but are mechanically more sensitive and build taller. For the same base area, roughly: low height = more Q_c per area, greater height = better ΔT and better COP.

Standard modules are usually manufactured to ±0.3 mm. For applications in which several modules are clamped together between heat sink and cold plate, or in which a defined clamping force is required, lapped versions with ±0.03 mm are available. Please state the required tolerance with your enquiry – it is price-relevant.

The legs are soldered to the copper bridges. The solder used limits the permissible temperature of the entire module:

The permissible temperature is 20 to 30 K below the melting point in each case. Regardless of this: in continuous operation above 120 °C, copper diffuses into the semiconductor material and the module permanently loses performance. If a module is soldered into an assembly, the process profile (reflow, hand soldering) must match the internal solder – please coordinate in advance.

Miniature and micro modules are available from a few square millimetres of base area and heights below 1 mm, some with only 7 thermocouples. They are typically integrated into TO packages (TO-8, TO-56, butterfly) to stabilise laser diodes, IR detectors or sensors. Note: at these sizes the parasitic heat loads dominate (see above) and the mechanical load capacity is limited. See miniature types.

In thin-film modules the thermoelectric material is deposited as a layer instead of being sawn from blocks ("bulk"). Advantages are a very low height and extremely short response times in the millisecond range, which makes them interesting for fast-switched applications. Disadvantages: low absolute heat-pumping power, usually smaller ΔT and considerably higher cost. For the vast majority of cooling tasks, bulk technology is the more economical choice.

The central hole allows a beam path, a sensor, a shaft or a screw connection to be led through the middle of the module. Typical applications are optical assemblies, detector cooling and cold plates that must be bolted in the centre. Note: the hole reduces the active area and thus Q_c; in addition, the bolt must not introduce shear forces. See types with a hole.

Yes. NTC thermistors can be applied directly to the cold-side ceramic and wired in. This considerably shortens the control path, because the sensor sits right at the controlled surface instead of somewhere in the assembly. For precise measurements, four-wire (Kelvin) measurement is recommended, since with thin connecting wires the lead resistance would otherwise enter the result. Sensor integration is also possible on multi-stage modules. Such versions run as special types – please enquire.

Recommended is 3 to 8 kg/cm² (around 30 to 80 N/cm²). In this range the thermal contact resistance is already clearly reduced without mechanically overstressing the module. The design upper limit is 15 kg/cm² (about 200 psi); good modules only fail mechanically well above this, but beyond this value additional pressure brings nothing thermally and only increases the risk of breakage. What matters is not so much the absolute value as the uniform distribution: spring-loaded screws with a stiff pressure plate are preferable to rigid bolting.

In principle yes, but with restrictions:

• Clamping is the standard solution and the most robust variant for temperature cycling as well as shock and vibration.
• Bonding with a thermally conductive adhesive makes sense when bolting is not possible by design. With epoxy adhesives, observe the curing temperature – it must not endanger the internal solder.
• Soldering is only possible on modules metallised on both sides and is practically limited to small form factors up to about 12 × 12 mm.

With a rigid connection on both sides (bonding or soldering) combined with large temperature swings, considerable shear stresses arise – then clamping is preferable.

The cold and hot sides of a Peltier element work at very different temperatures. If the adjacent parts have a different coefficient of thermal expansion (CTE) than the ceramic, shear stresses arise – especially at the corners of the module, where the distance to the centre is greatest. Countermeasures:

• Choose materials with a similar CTE (Al₂O₃ ≈ 7 · 10⁻⁶ /K; copper ≈ 17, aluminium ≈ 23 · 10⁻⁶ /K)
• Rigid connection on only one side, clamp the other side or provide it with a compliant interface
• Leave the corner areas of the bonding or soldering surface free so the corners can move freely
• Use the smallest possible modules – the offset grows linearly with the edge length

Yes, slightly. In operation the hot side expands while the cold side shrinks – the overall height thus typically changes in the range of a few micrometres to a few hundredths of a millimetre, depending on height and ΔT. For pre-loaded assemblies with tight tolerances, optical alignments and spring-loaded clamping, this effect must be taken into account.

Compressive forces perpendicular to the surface are uncritical as long as they are introduced uniformly (see above). Shear forces parallel to the surface, on the other hand, are critical and must be kept away by design – they load the solder joints of the legs directly. The connecting leads must also not carry any tensile force: they are soldered to the ceramic and come loose under tension or repeated bending. Therefore provide strain relief for the leads in the immediate vicinity of the module.

With correct connection of the supply voltage, the side to which the wires are soldered always gets warm. This side belongs on the heat sink. By reversing the polarity, the cold and hot sides swap – this is permissible for heating operation, but the assembly must then be thermally designed for both cases.

Peltier elements are a resistive DC load with a positive resistance temperature coefficient of around 0.5 % per K. The decisive factor is the ripple of the supply:

Background: the AC component produces Joule heat but contributes nothing to heat transport – the achievable ΔT drops. In measurement setups there is the added issue that AC components on the module leads can disturb the measurement signals, even though the Peltier element itself is electrically very quiet.

Yes, PWM is common and cost-effective, but it costs performance: the pulsed current produces more Joule heat than a direct current of the same average power, so ΔT and COP fall measurably. Recommendations:

• Choose a sufficiently high switching frequency (usually > 1 kHz) so the module integrates thermally
• Provide an output filter (LC) – this brings the behaviour close to that of a linear supply
• For the highest demands on ΔT and stability: a linear, current-controlled output stage

Suitable controllers can be found under Peltier temperature controllers.

Every ON/OFF cycle is a complete thermal cycle for the module: the legs and solder joints expand and contract again. Thermal cycles are by far the dominant wear mechanism of Peltier elements – in steady-state operation the MTBF is over 200,000 hours, whereas in hard cycling operation it is a few tens of thousands of cycles. Analogue control or a short, high-frequency pulsing is therefore considerably gentler mechanically and extends the service life significantly.

As soon as the cold side falls below the dew point of the ambient air, moisture condenses. This leads to corrosion of the inner copper bridges, to leakage currents and, in the extreme, to a short circuit. For orientation: at 25 °C and 50 % relative humidity the dew point is around 14 °C, at 70 % r.h. already about 19 °C. Measures, in order of effectiveness:

1. Encapsulate the cold area and use dry protective gas or evacuate it
2. Seal the module peripherally (silicone or epoxy) and additionally seal the edge zone with closed-cell foam tape
3. Provide a designed condensate drain where condensation is unavoidable

A seal costs around 3 to 4 % of performance through heat feedback via the sealing material and should therefore be used only where it is needed.

In vacuum, Peltier elements actually work better, because convection and gas heat conduction as parasitic paths are eliminated. However, outgassing must be considered: for high and ultra-high vacuum, choose unsealed modules, since silicone and in part also epoxy seals outgas. Likewise, thermal greases must be replaced by dry interfaces (graphite foil, indium foil, direct metal connection). Please coordinate UHV applications with us – material selection is decisive here.

In constant operation without significant temperature swings, the MTBF is over 200,000 hours – equivalent to more than 20 years of continuous operation. Peltier elements have no moving parts, no refrigerant and no wear in the classical sense. The service life is limited almost exclusively by thermal cycles, overheating and moisture ingress.

This depends on four factors: number of cycles, magnitude of the temperature swing, absolute upper limit and rate of change. Typical test results for high-quality modules:

Important: modules of a higher temperature class (e.g. 200 °C) are considerably more cycle-resistant even at cycles well below their limit temperature. Smaller modules (edge length ≤ 30 mm) behave better under thermal cycling than large ones. For pronounced cyclic operation we recommend our cycler modules.

Considerably less than is often assumed. Under continuous loading at 150 °C, the following degradation was measured over 42 months: around 2.5 % loss of ΔT_max in the first 12 months, then a further 1.3 % over 30 months, with the material parameters stabilising. Ageing is therefore degressive, not linear. It only becomes critical above 120 °C continuous operation, through copper diffusion, and near the melting point of the internal solder.

A high ambient temperature raises the hot-side temperature and thus the mean operating temperature of all solder joints – the ageing processes accelerate. At the same time the module has to work harder for the same ΔT, which increases the Joule heat and the current. A well-sized heat sink is therefore not only a performance question but, above all, a service-life question.

From the first digit of the series:

Approximately, yes. On uwe standard modules, one thermocouple occupies roughly 12.8 mm² of base area. This gives:

Examples: 63 couples ≈ 806 mm² ≈ 40 × 20 mm; 127 couples ≈ 1,626 mm² ≈ 40 × 40 mm. Please take the actual edge length from the datasheet – the packing density differs in particular for miniature, cascade and centre-hole types.

The letter is an internal configuration code from uwe electronic. It has no bearing on the module specification – two modules with identical remaining fields are functionally equivalent.

Note: the sealing is not encoded in every type designation. In case of doubt, the datasheet or our stock and order list is authoritative.

Depending on the series, standard is 120 mm or 133 mm. Deviating lengths are appended as a three-digit millimetre value, for example -450 for 450 mm or -150 for 150 mm.

From 127 couples × approx. 0.12 V, a maximum voltage of around 15.2 V also follows.