Peltier Effect

The intuition — electrons carry heat across the boundary

Picture two different conductors welded together, and push a current through the seam. On one side of the junction it gets cold; on the other side it gets hot. No fan, no gas, no compressor — just electrons crossing a boundary. That is the Peltier effect, discovered by Jean Charles Athanase Peltier in 1834.

Here is the mechanism. Charge carriers in a material don't just carry charge — they also carry heat. How much heat per carrier depends on the material: it is set by the Seebeck coefficient S, which you can think of as "entropy per unit charge." When an electron moves from a material where it carries little heat into a material where it must carry more, it has to grab that extra heat from the lattice around the junction — so that spot gets cold. At the other junction, electrons step down to carrying less heat, so they dump the excess into the lattice — that spot gets hot.

The beautiful part: it is fully reversible. Reverse the current direction and the cold side becomes the hot side. There is no thermodynamic cycle, no phase change — heat is pumped continuously and quietly as long as current flows.

The governing physics

The rate of heat absorbed or released at a junction is proportional to the current — not its square:

Q_peltier = Π · I = (Π_B − Π_A) · I

where Π is the Peltier coefficient of the junction (the difference between the two materials' coefficients) and I is the current. Q is in watts when Π is in volts and I in amperes. The sign of Q flips with the sign of I — that is the reversibility.

The Peltier coefficient is not an independent property. It is locked to the Seebeck coefficient by the second Kelvin (Thomson) relation:

Π = S · T

where T is the absolute temperature in kelvin. This single equation ties together the Peltier effect, the Seebeck effect, and the Thomson effect into one thermoelectric framework. It is why a good Peltier cooler and a good Seebeck generator are made of the same materials — both want a large S.

In a real device you must subtract two parasitic terms. The total heat removed from the cold side is:

Q_cold = S·T_c·I  −  ½·I²·R  −  K·(T_h − T_c)

Term by term: the first is the Peltier pumping (helps cooling), the second is Joule heating I²R generated inside the device — half of it lands on the cold side and fights you — and the third is plain thermal conduction leaking heat back from hot to cold through the device's conductance K. The fight between the linear pumping term and the quadratic Joule term is the whole story of Peltier cooling.

The optimum current — why "more power" backfires

Because cooling scales as I but Joule heating scales as I², there is a sweet spot. Differentiate Q_cold and you find the current that maximizes the cooling — the same current also gives the maximum temperature difference at zero load:

I_optimal (max cooling, max ΔT) = S · T_c / R

Push past it and the I²R term wins: the cold side stops getting colder and starts warming back up. The figure of merit that captures how good a thermoelectric material is bundles all three properties into one dimensionless number:

ZT = (S² · σ · T) / κ

where σ is electrical conductivity and κ is thermal conductivity. You want a big Seebeck coefficient, a material that conducts electricity well (low R) but conducts heat poorly (low K leak). Those goals conflict — most good electrical conductors are also good thermal conductors — which is why high-ZT materials are rare and thermoelectrics have stayed inefficient for nearly two centuries. Commercial Bi₂Te₃ sits around ZT ≈ 1 near room temperature.

How a real module is built

A commercial thermoelectric cooler (TEC) is not one junction — it is dozens to hundreds of them in series. Small pillars of n-type and p-type bismuth telluride alternate between two thin ceramic plates. They are wired electrically in series (the same current threads every pillar) but thermally in parallel (all the cold ends face the same plate). Using both n-type and p-type lets every junction pump heat in the same direction, so the contributions add instead of cancelling.

Component	What it does	Why it matters
n-type Bi₂Te₃ pillars	Electrons carry heat with current	Large negative S, large pumping per amp
p-type Bi₂Te₃ pillars	Holes carry heat with current	Large positive S; pairs with n-type to add up
Copper tabs	Connect pillars in electrical series	Low resistance, low parasitic Joule loss
Ceramic plates (alumina)	Electrical insulation, thermal contact	Keep current in pillars but pass heat to faces
Hot-side heat sink	Dumps Q_cold + electrical power	If it can't shed heat, the cold side warms up

Comparison — Peltier vs vapor-compression cooling

Property	Peltier (TEC)	Vapor-compression fridge
Coefficient of performance (COP)	≈ 0.4–0.7	≈ 2–4
Moving parts	None (solid state)	Compressor, fan, valves
Refrigerant	None	Required (HFC / propane / CO₂)
Response time	Seconds (electronic)	Minutes
Reversible heat/cool	Yes — flip the current	Needs a reversing valve (heat pump)
Practical cooling capacity	Watts to ~hundreds of watts	Hundreds of W to many kW
Noise / vibration	Silent (only the heat-sink fan)	Audible compressor
Best fit	Small, precise, spot cooling	Bulk, energy-efficient cooling

The takeaway: Peltier wins on size, silence, precision, and reversibility; vapor-compression wins decisively on energy efficiency and raw capacity. You pick a Peltier module when you need to cool a 5-gram laser diode by exactly 0.01 °C, not when you need to chill a warehouse.

Worked example — pump 30 W, pay how much?

Take a single-stage module with S ≈ 0.05 V/K (the whole stack), internal resistance R = 2 Ω, cold side at T_c = 280 K (7 °C), hot side held at T_h = 320 K (47 °C) by a heat sink. Run it at I = 6 A.

• Peltier pumping: S·T_c·I = 0.05 × 280 × 6 = 84 W of raw pumping.
• Joule penalty on cold side: ½·I²·R = 0.5 × 36 × 2 = 36 W working against you.
• Conduction leak (say K = 0.45 W/K): K·ΔT = 0.45 × 40 = 18 W leaking back.
• Net heat removed from cold side: 84 − 36 − 18 = 30 W.
• Electrical power drawn: the Seebeck back-voltage plus I²R ≈ S·ΔT·I + I²R = (0.05 × 40 × 6) + (36 × 2) = 12 + 72 = 84 W.
• COP: 30 W cooled ÷ 84 W in ≈ 0.36.

So to remove 30 W of heat you spent 84 W, and the hot-side heat sink must now reject 30 + 84 = 114 W. That last number is the gotcha every beginner misses: the hot side has to dump everything you pumped plus everything you paid for.

Where the Peltier effect shows up

• CPU and laser-diode cooling. Drives a hot spot below ambient for overclocking or to stabilize a laser's wavelength to a fraction of a degree.
• Portable car fridges and coolers. No compressor means they run silently off a 12 V socket and double as warmers when you reverse the current.
• Scientific instruments. Cooled CCD and CMOS camera sensors (astronomy, microscopy) drop sensor temperature 40–60 °C below ambient to kill dark-current noise.
• PCR thermal cyclers. DNA amplification needs fast, repeated heating and cooling — Peltier's electronic, reversible switching is ideal.
• Dehumidifiers and water-from-air devices. Chill a surface below the dew point to condense moisture.
• Spacecraft and detectors. Multi-stage cascades cool infrared sensors to below −100 °C with no vibration that would blur an image.
• Dew-point and chilled-mirror hygrometers. Precise, fast temperature control of a tiny mirror.

Common misconceptions and edge cases

• "It creates cold." No — it moves heat. The cold side only gets cold because heat is shoved to the hot side. Fail to remove heat from the hot side and the whole module saturates and the cold side warms up.
• "More current = more cooling." Only up to the optimum. Cooling is linear in I, Joule heating is quadratic, so past I_optimal you cool less. Running a module flat-out is usually a mistake.
• "It's just Joule heating." No. Joule heating I²R is irreversible and always heats. The Peltier term is linear, reversible, and can heat or cool depending on current direction. They are different physics living in the same device.
• "Metals work fine." Plain metal-to-metal junctions have tiny Seebeck coefficients and pump negligible heat. You need semiconductors (Bi₂Te₃) to get a usable effect.
• "The cold side can reach absolute zero." Far from it. Conduction and Joule leak cap a single stage at ~65–70 °C below the hot side; you cascade stages to go colder, and even then ΔT shrinks fast as you pump real heat.
• "It's the same as the Seebeck effect." They are inverses, linked by Π = S·T. Seebeck turns ΔT into voltage; Peltier turns current into ΔT.