# Minimal temperature achievable by vanilla Peltier element?

I wonder, are there any fundamental issues leading to reduced performance of Peltier elements at cryogenic temperatures (-100C and lower)?

What is theoretical/practical minimum temperature achievable by a cascade of Peltier elements, provided that each next element have about 3 times less power than previous in cascade, so that they are not overwhelmed by selfheating? Let's say first element is water-cooled down to 20C when dissipating 150W.

Update: After extensive tests, I've found out that in any setup I cannot get below -19C using any number or combination of Chinese (r) Peltier elements (I've tried alot of different ones in different combinations).

• Update from myself: vanilla Peltiers suck big time below -25-30C. At -60C they are barely can make it to -65C with no load @5V (on 12V it just heats up). So no luck this time :-) Dec 17, 2011 at 19:02

I do not know any fundamental minimum of Peltier elements operation temperature.

However, there are serious technical issues: as long as Peltier effect relies on an interaction between electrons and phonons, there should be enough phonons to interact with. Decreasing temperature dramatically (though quantitatively, not qualitatively) changes effectiveness of the elements. As long as heat pumping should overcome heat produced by electron current which produces this pumping, going to low temperatures is a challenging task.

To sound more scientific, I've found this relatively new paper where Peltier effect is discussed for some rather standard system and where curves at room and liquid nitrogen temperature (which is low by human standards but pretty warm from cryogenic point of view) and its effectiveness may be found (see Fig. 3). As can be seen from the curves, Peltier element is able to give $\Delta T$ around few degrees at nitrogen temperature. So, it stil works at 77K but definitely far less effective than at room temperature. With numbers given on graph I can hardly imagine the cascade which will do these 77K out of room temperature.

In this paper authors claim that effect may be observed at 6K, but the numbers they give show that this effect may be hardly used in practice.

To conclude, it seems there is no definite theoretical limit, but practical limit is around -100C

• I see... Papers looks interesting, but looks like they both are for 'non-consumer' grade peltier elements... Jul 26, 2011 at 4:01
• If you are interested in real devices, it is better to read cite of a company which produces such devises. E.g., here you can buy device which cools down to -100C. Papers show that there is no theoretical limit, it is a matter of how much you can pay for such device and its operation. Jul 26, 2011 at 5:20
• Hehe, got 8 Peltier's, tried 2-stage system out of 2 elements - got -19C. Will try 3-stage system out of 6-8 elements tonigh :-) Aug 3, 2011 at 8:35
• I know this is an old answer, but different thermoeletric materials have different Zt ( see graphs page below ) values at varying temperatures. 'Vanilla' TEC elements are usually bismuth telluride, which is most efficient at ~room temperatures. The maxium achievable COP drops off rapidly at lower temps. For stacked elements and lower temperatures, specific cryogenic temp thermoelectric materials would be used. Some new developments in InAs/GaSb lattices claim achievable temps down to 4K. thermoelectrics.matsci.northwestern.edu/thermoelectrics/…
– kert
Nov 9, 2015 at 18:31
• To add, claimed low bar temp for stacked 'vanilla' tec's is about 150K, and you can buy ready-made 3-stage modules by Laird technologies on Mouser, with maximum delta-T of about 120C. Part number 9340006-301. Incredibly inefficient though
– kert
Nov 9, 2015 at 18:42

A got -28°c. I used only one TEC-12710(160W) with a modified water cooler, I changed the fan of radiator for a industrial blower.

I believe that one TEC-12730 with a more powerfull cooling device will reach extreme low temperatures.

• I got the same module (TEC1-12710), I was able to push it to -31°c with a big CPU cooling block (8 copper pipes) touching the 4x4CM hot side. Oct 9, 2017 at 21:37