Bench Degree·REFRIGERATIONchapter

Chapter 23: Cooling Without a Refrigerant

Four times now the fluid has been the problem. So the obvious question to end on is whether you can get rid of it, and the answer is yes, in three different ways, one of which is a piece of wire.

Chapter 3 told the same story four times. Ether and ammonia and sulphur dioxide and methyl chloride killed people. Freon solved that and destroyed the ozone layer. HFCs solved that and turned out to have global warming potentials in the thousands. The current answers give back the flammability and toxicity that Freon had bought.

Every round of that story is a problem with the working fluid. So the honest question to close on is: can the fluid be dispensed with entirely?

It can, and the reader is already equipped to understand how, because all three methods are Chapter 5 and Chapter 6 applied to something other than a liquid.

Section 1: The Common Principle

Every one of the machines in this chapter works the same way as the one in Chapter 10. Something is squeezed, gets hot, dumps its heat, is released, gets cold, and absorbs heat. The cycle is identical. Only the working substance changes.

What all three exploit is a caloric effect: a material that changes temperature when some external field is applied or removed. Four exist and three are practical:

The first three are genuinely the Chapter 10 cycle with a solid instead of a fluid. The fourth is different, and Chapter 11 already built one.

IN PLAIN ENGLISH: A refrigerator squeezes a gas to make it hot and lets it expand to make it cold. A solid-state cooler squeezes, magnetises or electrifies a solid to do the same thing. Same four steps, no fluid, nothing to leak.

Section 2: Elastocaloric, and a Wire You Can Test Yourself

Take a length of nitinol wire, a nickel-titanium shape memory alloy, and stretch it hard. It gets warm. Let it snap back and it gets cold.

Not by friction, and not by much: a few degrees. But it is a real thermodynamic effect and it is a complete refrigeration cycle in a piece of wire.

Why it happens is Chapter 5’s material in a solid. Nitinol has two crystal structures, austenite and martensite, and it transforms between them. That transformation is a phase change, exactly like boiling, with a latent heat attached to it. Stretching the wire forces the transformation one way and releases that latent heat; releasing the stress lets it transform back and absorbs it.

So the reader who understood the ice plateau in Chapter 5 already understands this. It is latent heat, in a metal, driven by force instead of by temperature.

The numbers are respectable. Nitinol shows adiabatic temperature changes of 15 to 25 °C (27 to 45 °F) per cycle, which is comparable to what a vapour-compression stage achieves. Theoretical efficiencies are competitive with vapour compression, and prototypes have reached a meaningful fraction of that.

The problems are engineering rather than physics:

Fatigue. The transformation is a repeated large mechanical deformation. Early materials cracked within thousands of cycles; a refrigerator needs hundreds of millions. Current research alloys reach millions, which is progress and not yet enough.

Force. Straining a metal takes real force, so the mechanism is a press rather than a pump. That is heavy and it has its own losses.

And heat transfer. A solid must be brought into thermal contact with the space being cooled and then with the heat sink, alternately, which means either moving the solid or moving a fluid past it. The irony is that most elastocaloric prototypes end up circulating water, so they have not eliminated a fluid, only eliminated a refrigerant.

What it buys. Global warming potential of exactly zero, no ozone depletion, nothing toxic, nothing flammable, and nothing that can leak, because a solid does not.

ON THE BENCH: Latent heat in a wire

Parts: nitinol wire, 0.5 mm (0.020 in) diameter, about 300 mm (12 in) long, sold for hobby and educational use, about $10. A sensitive thermocouple, or your lip, which is a remarkably good thermometer. Cost: about $10 if you own a thermocouple. Time: 15 minutes. Hazards: the wire is springy and can whip when released. Eye protection. Do not exceed about 8 percent strain or it will not recover. Method: hold the wire taut, tape a fine thermocouple to its middle, and stretch it firmly and quickly. Record the temperature. Hold for a few seconds, then release quickly and record again. What you should see: a rise of several degrees on stretching and a fall of several degrees on release, each decaying back to ambient within a few seconds as the thin wire exchanges heat with the air. Quickly is essential: this is the adiabatic argument of Chapter 6, and a slow stretch leaks the heat away as you make it and shows almost nothing. Then do it against your lip, which detects a two-degree change easily. Stretch, touch to lip: warm. Release, touch to lip: cold. What you have just done: run a complete refrigeration cycle, with no fluid, no compressor and no moving parts beyond your own hands. It is the whole of this chapter in a $10 piece of wire.

Section 3: Magnetocaloric

Same cycle, different field.

Certain materials warm when a magnetic field is applied and cool when it is removed. Gadolinium is the classic, with a Curie temperature conveniently near room temperature, and the modern research materials are gadolinium-silicon-germanium alloys and lanthanum-iron-silicon compounds.

Why it happens. In an unmagnetised sample the atomic magnetic moments point in random directions. Applying a field aligns them, which reduces their magnetic disorder, and because total entropy must be conserved that disorder has to go somewhere. It goes into the lattice, as vibration, which is heat. Remove the field, the moments randomise again, and the entropy comes back out of the lattice, cooling the material.

The cycle is then a matter of moving the material in and out of a magnet’s field, or moving the magnet, with a fluid, usually water, carrying heat to and from each end.

Where it stands. Prototype wine coolers and small commercial units exist. It works, demonstrably. Two obstacles:

The temperature change per cycle is small, typically 2 to 5 °C (4 to 9 °F) per pass, so a useful machine needs a regenerator, a cascade that multiplies the effect over many stages. That adds complexity.

And the magnets are the cost. Achieving a useful field means substantial permanent magnets of neodymium-iron-boron, which are expensive and are themselves a supply chain and mining question.

Where it genuinely wins. Magnetocaloric refrigeration is already the standard method for reaching temperatures below 1 kelvin in laboratories, by adiabatic demagnetisation. At the cryogenic end it is not a candidate technology; it is the incumbent.

Section 4: Thermoelectric, Which You Have Already Built

Chapter 11 built one, ran it as a cooler, then disconnected it and ran it backwards as a generator. So this section is short and it is mostly an honest accounting.

It is the only one of the four already in mass production, in camping coolers, in laboratory instruments, in laser diode mounts and in camera sensor cooling.

And it is by far the least efficient. A COP of 0.3 to 0.6 against vapour compression’s 3.0, for the materials reason in Chapter 11 Section 6: you need high electrical conductivity and low thermal conductivity in the same material, and in most solids those travel together. The figure of merit is ZT, the best commercial materials sit near ZT = 1, and seventy years of work has not moved it much.

Where it wins outright:

Precision. A thermoelectric can hold a set temperature in either direction by reversing the current. No compressor can do that, which is why every laboratory instrument that needs a stable temperature uses one.

Silence and no vibration, which is why they are in hotel minibars, in equipment near sensitive instruments, and in wine cabinets, where Chapter 11 Section 6 works out why a machine this inefficient is nonetheless the right answer: the temperature lift a wine cabinet needs is small enough that a poor coefficient of performance costs very little in absolute watts.

Reliability with no moving parts. Chapter 11’s example: Voyager 1 launched in 1977 and is still transmitting on radioisotope thermoelectric generators, forty-eight years later.

And very small capacities, where a vapour-compression machine cannot be built small enough to be worth it.

Section 5: Where the Old Ways Are Still the Answer

Three non-refrigerant methods that are not emerging technology at all, and are already the correct answer in specific cases.

Evaporative cooling. Chapter 2 and Chapter 14. In dry air a swamp cooler achieves genuinely useful cooling for a fraction of the energy and with water as the only working fluid. Bounded absolutely by the wet-bulb temperature, which makes it superb in Arizona and useless in Florida, and which is exactly the wall the Persians hit.

And it is not a marginal technology at scale. Every water-cooled chiller in Chapter 20 Section 8 rejects its heat through a cooling tower, and a cooling tower is an evaporative cooler. Much of the world’s large-scale air conditioning already rejects its heat by evaporating water, limited by the wet bulb, exactly as the yakhchāl was.

Absorption. Chapter 3 Section 4, Carré’s 1859 machine. It uses a refrigerant, so it does not strictly belong in this chapter, but it eliminates the compressor and runs on heat. Where waste heat is free, an absorption chiller is the right answer and always has been.

And passive design. Chapter 16. The cheapest cooling is the load never admitted, and orientation, shading and thermal mass require no working fluid, no electricity and no maintenance at all.

Section 6: The Honest Comparison

Read that table honestly. Vapour compression is not being displaced because it is inefficient. It is extraordinarily efficient, it is a century and a half mature, and Chapter 13 showed a real machine reaching about 71 percent of its own thermodynamic ceiling, which is far better than an internal combustion engine manages.

The pressure on it is entirely about the fluid. If a refrigerant existed with zero GWP, zero ODP, no toxicity, no flammability and a convenient saturation curve, none of the alternatives in this chapter would attract a research dollar. The problem is that Chapter 8’s nine requirements contain genuine contradictions and no fluid satisfies them all.

Which is why the solid-state methods are interesting despite being less efficient today. A solid has no GWP because it does not leak, and that single property may matter more than a few points of COP.

Section 7: The Loop Closes

Worth noticing where this chapter ends up.

Elastocaloric cooling uses shape memory alloys. The same class of material, doing a different job, that has been chased for decades as an artificial muscle and never succeeded, because its bandwidth is limited by how fast it can be cooled.

And a refrigerator does not need bandwidth. One cycle per second is useless in a limb and entirely adequate in a cooling machine. The constraint that defeated the alloy as a muscle is irrelevant to it as a refrigerant, which is a satisfying piece of engineering symmetry and a reasonable place to end a book about moving heat.

The material was never wrong. It was pointed at the wrong job.

Section 8: What This Chapter Bought You

All three solid-state methods are the Chapter 10 cycle with a solid instead of a fluid. Squeeze, dump, release, absorb. The cycle is not new; the working substance is.

Elastocaloric is latent heat in a metal, driven by force rather than temperature, and it can be demonstrated with $10 of wire and your own lip.

Magnetocaloric works by moving entropy between magnetic disorder and lattice vibration, and it is already the incumbent technology below 1 kelvin.

Thermoelectric is the only one in mass production and the least efficient, and it wins on precision, silence and reliability rather than on efficiency.

Vapour compression is not under pressure because it is inefficient. It is under pressure entirely because of the fluid, and the solid-state candidates’ real advantage is that a solid cannot leak.

And evaporative cooling, bounded by the wet bulb since the Persians, is still rejecting the heat from a large share of the world’s air conditioning, through every cooling tower on every roof.

One chapter left, and it is the ledger.

Bench Degree

Get the degree without the diploma.

Learn the material, not how to pass the exam.