The Chemistry of Refrigerants

Originally posted 2022-07-13

Obligatory disclaimer: all opinions are mine and not of my employer

Refrigerants are fluids that are used in every air conditioning unit in the world. They transfer heat from cold to hot against the natural temperature gradient, by exploiting temperature and pressure dependent phase transitions.

In this essay I’ll walk you through humanity’s ongoing quest for the perfect refrigerant. A more in-depth review can be found in this paper

One Weird Trick

When you mix ice at 0° C into a drink at 20° C, you get a drink at 0 degrees with some partially melted ice cubes. That’s pretty weird, if you think about it.

Phase transitions like melting and evaporation take up and release a lot of thermal energy - far more than normal heating and cooling of a material. Crucially, the temperature at which the liquid-gas transition happens is pressure-dependent - higher pressures lead to higher boiling points. This allows the refrigeration cycle to release heat by compressing and condensing the gas, then absorb heat by expanding and evaporating the gas.

The ratio of energy transferred to energy used is typically greater than 1, making it a very efficient way to heat and cool. I anticipate that heat pumps will become increasingly common as they make great environmental and financial sense. If you want to learn more, this guy is really enthusiastic and will talk for hours about heat pump efficiency and how they’ll work even in the dead of winter.

A Refrigerant Wishlist

The ideal refrigerant should have a boiling point that is somewhat below the operating cold temperature. That way, a cheap compressor can provide the moderate (~5 atm) pressures needed to straddle the cold-hot temperature gap. Depending on the application, \(T_{cold}\) and \(T_{hot}\) will differ. A refrigerated truck might have to keep its contents at -30° C in an outside temperature of 30° C, whereas a vehicle’s air conditioner only has to output air at 10° C. Perusing the spec sheet for the “Thermo King V520”, we can see that a truck cooler unit rated at -20° F uses refrigerant R-404a with a boiling point of -52° F / -47° C, whereas a truck cooler unit rated at 0° F uses refrigerant R-134a with a boiling point of -15° F / -26° C. Overall, we’re looking for refrigerants with a boiling point between -50° C and -10° C for typical applications.

Another desirable refrigerant property is safety, in the event of a leak. Ammonia, \(\ce{NH3}\) (b.p. -13° C) used to be a popular refrigerant, but it’s a notoriously pungent gas, and not particularly safe to inhale. Propane, \(\ce{C3H8}\) (b.p. -20° C) is another formerly popular refrigerant, with the downside of being highly flammable and potentially explosive in the right mixtures.

It turns out that the space of potential refrigerants is not that large. Most refrigerants have \(\leq 4\) non-hydrogen/fluorine atoms; any more and the boiling point gets too high. (Why are hydrogens and fluorines omitted from this count? Because H/F have extremely low polarizability, and contribute minimally to the Van der Waals forces that help molecules stick together in the liquid phase.)

Based on these criteria, DuPont systematically explored this space in the 1920s and decided that Freons were the way to go. R-12, \(\ce{CCl2F2}\) (b.p. -20° C) is nonflammable and nontoxic. As chlorine and fluorine are similar in electronegativity to oxygen, the molecule is not flammable in air.

The Ozone Hole

Unfortunately, it turns out that carbon-chlorine bonds can be broken by UV light to form chlorine radicals. This wasn’t news to us; free radical halogenation is in fact one of the very very first reactions that organic chemistry students learn. What was news was that a single chlorine radical could catalyze the destruction of 100,000 ozone molecules in the stratosphere. Fortunately, the world agreed to phase out the use of chlorinated hydrocarbons in the Montreal Protocol. Supposedly, Ronald Reagan led the charge here because he blamed his skin cancer on the ozone hole!

Due to their ozone destruction potential, chlorine, bromine, and iodine were no longer acceptable for use in refrigerants. With this additional constraint, the set of allowable atoms shrinks to H, C, N, O, S, and F. HFCs like R-134a, \(\ce{CF3CFH2}\) (b.p. -26° C), were the winners in this generation. Again, fluorine is a key ingredient here thanks to its noncombustible properties.

Greenhouse Warming Issues

As it turns out, Freon’s replacement R-134a has a greenhouse effect that is 1300x (!!) more potent than carbon dioxide per molecule. Fluorine confers nonflammability but also has a potent greenhouse warming effect.

The next generation of refrigerants, like R-1234yf, \(\ce{CH2=CFCF3}\) (b.p. -30° C), manage to dodge this catch-22 by optimizing for atmospheric breakdown time. By creating “weak spots” in the molecule, R-1234yf degrades in the atmosphere over 11 days (compared to R-134a’s 13 years), mitigating the cumulative greenhouse warming potential to near-zero.

The downside is that these “weak spots” also enable a small amount of flammability. Testing revealed that this flammability would not be an issue even in a severe auto collision, and R-1234yf is today rolled out in cars and light trucks in the US and EU.

Of course, this didn’t stop Daimler, parent of Mercedes-Benz, Volkswagen (what is it with these execs?) for using this as an excuse to delay the HFO transition.

Nowadays, the greenhouse potential of refrigerants is well known - the Kigali amendment to the Montreal Protocol mandates a reduction in fluorinated gas emissions, and R-1234yf will undoubtedly play a major role in this transition once the patent expires and prices come down.

Something to ponder is that it took six years to discover R-1234yf, another four years to run exhaustive safety testing, another five years to gather political support for mandates (and deal with Daimler’s bullshit), and another five years to scale up and finish rolling out to automobiles. Over the next decade, R-1234yf will likely roll out to adjacent markets. This 30-year process seems fairly representative of new technology rollout, and may even be on the fast side, due to Honeywell/DuPont’s existing expertise and understanding of the problem space, the relatively small search space, and the centralized regulatory mechanisms of the US/EU. Basic research and materials discovery will likely take at least this long to scale to planetary impact, if it ever does.

Eliminating high-GWP fluorinated gases

There’s still a lot of work to be done in refrigerants: rolling out HFOs, broadening the adoption of heat pumps. One area I want to highlight is the cleanup of existing high-GWP refrigerants.

Emissions of high-GWP refrigerants can happen at the manufacturing plant, as a slow leak during typical usage, during improper air conditioner disposal, during a car collision, and even intentionally, in compressed air dusters (!!). Due to the ridiculously high multipliers here, I suspect there is a lot of low-hanging fruit for climate change mitigation in this space. I’m encouraged by Recoolit’s efforts in the area, and hope they find a good way to counter the cobra effect.

Postscript: PFAS?

Fluorine is weird. The stability of the C-F bond leads to its chemical inertia - think Teflon. The C-F bond is so unreactive that fluorous solvents won’t even interact with other solvents! This leads to a little-known fact that in addition to oil and water, there is a third mutually immiscible so-called “fluorous phase”, which sounds like a made-up science fiction plot device.

When working with the fluorous phase, fluorous surfactants (molecules with a fluorous end and a hydrophobic/philic end) like PFOA are needed to homogenize fluorous-aqueous-organic mixtures. These PFAS are stable, bioaccumulate, and persist in the environment, and rightly generate a lot of pushback.

Fluorinated refrigerants, being mostly gaseous in nature, don’t actually require fluorous surfactants in their manufacture. R-1234yf degrades in the atmosphere to trifluoroacetic acid, or TFA. TFA is not considered to be a PFAS, as its extremely short chain renders it highly water-soluble and non-bioaccumulating. It accumulates in the ocean, but you might be surprised to learn that TFA levels in the ocean are mostly homogenous throughout the water column. Given that the water column takes thousands of years to equilibrate, this suggests that human activity can’t possibly be the source of this TFA. Instead, deep sea vents are suspected to be the source of 95% of this small, inert molecule. I am not worried about R-1234yf when it comes to environmental risks, and even if scale-up generates TFA levels that are 1-10x background levels, it is probably a worthwhile tradeoff for its humanitarian and global warming wins.