Segment 1 — The Cold Open

In the spring of 1994, automobile plants from Michigan to Mainz began charging new car air conditioners with a refrigerant advertised as ozone-safe. The chlorine was gone. The chemistry that had been chewing a hole in the sky over Antarctica could not do that job anymore. This was designed to close the most famous wound human industry had ever opened in the atmosphere. Instead, the replacement gases — hydrofluorocarbons, harmless to ozone — turned out to be greenhouse gases hundreds to thousands of times more potent than carbon dioxide, installed in a technology the warming planet was about to demand in the billions.

Segment 2 — The Good Intention

The people who built the Montreal Protocol were not fumbling in the dark, and they were not chasing a fashionable cause. They were answering a specific, measured emergency. In June 1974, Mario Molina and F. Sherwood Rowland published a paper in Nature showing that chlorofluorocarbons — the miracle compounds Thomas Midgley Jr. had demonstrated in 1930 by inhaling a lungful and blowing out a candle — could drift intact into the stratosphere, shed chlorine, and catalytically destroy ozone. A single chlorine atom could scavenge tens of thousands of ozone molecules. For a decade the finding was contested. Then, in May 1985, Joe Farman, Brian Gardiner, and Jonathan Shanklin of the British Antarctic Survey reported in Nature that October ozone over Halley Bay had collapsed. Their aging Dobson spectrophotometer was measuring what NASA’s satellites had been discarding as instrument error. The hole was real. Skin-cancer projections, crop-damage estimates, and the fate of phytoplankton were no longer thought experiments. Diplomats already had a framework from the 1985 Vienna Convention. What they needed was a control treaty, and they needed it before the chemistry ran away from them. In that light, the brief was rational and narrow: stop putting chlorine and bromine into the stratosphere, as fast as sovereign states would accept, without grounding the world’s refrigeration, foam-blowing, and electronics-cleaning overnight. Climate change was on the intellectual map — James Hansen would testify before the U.S. Senate in June 1988 — but it was not the mandate on the table in Montreal, and the atmospheric lifetime and heat-trapping power of the substitutes had not been the axis on which a replacement would be judged.

Segment 3 — The Implementation

On 16 September 1987, twenty-four countries and the European Economic Community signed the Montreal Protocol on Substances that Deplete the Ozone Layer. The first text was cautious, more freeze than funeral: production of key CFCs would be cut 50 percent by 1998. That restraint was not timidity so much as coalition-building. Mostafa Tolba at UNEP and negotiators including Richard Benedick of the United States needed chemical companies, developing-country governments, and skeptical ministries of industry in the same room. The science then did the accelerating. The 1990 London Amendment moved developed countries toward a full CFC phase-out by 2000; Copenhagen in 1992 pulled that date forward to 1996 and put the transitional hydrochlorofluorocarbons on a longer leash. A Multilateral Fund, created in London, began paying the agreed incremental costs for Article 5 countries to convert factories. DuPont, which had resisted controls through much of the 1980s, announced after the 1988 Ozone Trends Panel that it would stop making CFCs, and the industry’s research engine turned toward substitutes that had no ozone-depletion potential. Hydrofluorocarbons — HFC-134a for car air conditioners and domestic refrigerators, later high-GWP blends such as R-410A for building cooling — looked, on the criteria that mattered in those rooms, almost ideal: nonflammable, low toxicity, thermodynamically close to the chemicals they replaced, and a chlorine atom of zero. Proponents described the switch as the proof that a global environmental treaty could work. Skeptics of the ozone science had mostly gone quiet by the mid-1990s. What almost no one treated as a design constraint was the heat those fluorine-rich molecules would trap if they leaked, or if they were manufactured at the scale of a planet buying its first air conditioner. The Kyoto Protocol would list HFCs among its six greenhouse gases in 1997, but Kyoto bound only developed-country totals, and it did not stop Montreal’s fund from financing the very conversions that locked HFCs in.

Segment 4 — The Unintended Consequences

The ozone layer began, slowly, to heal. That part of the story is not a prelude; it is a genuine success, and any honest account has to hold it in the same hand as what followed. Assessments by the World Meteorological Organization and UNEP now project that mid-latitude ozone will return to 1980 levels around 2040, the Arctic around 2045, and the Antarctic around 2066. Modeling of a “world avoided” — the phrase researchers use for the counterfactual in which CFCs were never controlled — suggests a twenty-first century of catastrophic ozone loss and dangerous ultraviolet radiation. Kofi Annan called Montreal perhaps the single most successful international agreement to date, and on its own terms he was not wrong. The trouble lived in the replacement rule. HFCs do not eat ozone. They do, however, absorb infrared radiation with startling efficiency. HFC-134a has a 100-year global warming potential in the neighborhood of 1,300 times that of carbon dioxide; the blend that became standard in residential air conditioners, R-410A, sits near 2,000; HFC-23, a byproduct of other fluorochemical manufacturing, is in the range of 12,000 to 14,600. Because the molecules were screened for ozone-depletion potential, toxicity, and flammability — not for radiative forcing — the industry had optimized a refrigerant for one atmospheric problem and delivered a concentrated dose of another.

The causal chain tightened because the world was not holding still. As CFC plants closed, HFC production scaled to fill every niche the old chemicals had occupied, and then some. The Multilateral Fund, doing exactly the job it was written to do, helped convert production lines in China, India, and elsewhere toward ozone-safe gases. Those conversions were victories at the factory gate. They were also path-dependence: once a nation’s appliance industry, service technicians, and safety codes are built around a nonflammable HFC blend, the cheaper, climate-friendlier alternatives — hydrocarbons such as propane, ammonia, carbon dioxide — look like a fire-code problem and a retraining bill. Meanwhile the demand curve bent upward. The International Energy Agency estimated in 2018 that the global stock of air conditioners in buildings stood at about 1.6 billion units and could reach 5.6 billion by 2050, with most of the growth in countries that were hot, urbanizing, and, thanks in part to Montreal’s own success, newly equipped with HFC supply chains. Every leaked kilogram during manufacture, installation, use, and disposal was a pulse of high-GWP gas. Every intact kilogram still implied a compressor drawing electricity, often from coal, in the places where cooling demand was rising fastest. That is the second-order effect hiding inside the first: the ozone fix did not merely swap one molecule for a worse climate molecule. It standardized a cooling architecture just as climate change itself was making cooling feel less like a comfort good and more like infrastructure.

By the late 2000s the numbers were no longer a footnote. Work led by Guus Velders and colleagues projected that, without new controls, HFC radiative forcing could become a substantial fraction of the climate problem by mid-century — in high-growth scenarios, on the order of several hundred billion tonnes of CO2-equivalent over coming decades, with papers in that period suggesting HFCs might contribute on the order of 0.25 to 0.40 watts per square meter of forcing by 2050, or emissions comparable to a significant percentage of global CO2. The irony had a human texture. A refrigerator factory that had been paid to stop using CFCs could be, in the same decade, a rising source of climate pollution while its government received credit for protecting the ozone layer. A family in Guangzhou or Hyderabad buying a first split AC was making a rational decision about heat and health. The molecule inside the copper was the blind spot of a treaty that had been asked to solve one crisis at a time. Institutional siloing made the miss durable. Montreal’s institutions measured ODP tons. Kyoto’s measured basket greenhouse gases in countries that were not, by then, the center of HFC growth. No single secretariat owned the interaction. The third-order effect was political as much as chemical: success itself became a reason to stop looking. When a treaty is called the greatest environmental agreement ever written, the substitutes it blessed inherit some of that halo — until the atmospheric record says otherwise.

Segment 5 — The Aftermath

The correction, when it came, used the original machine. Scientists, the EPA’s old ozone hands, advocates such as Durwood Zaelke, and a coalition of island states and African governments spent the early 2010s arguing that Montreal — not a new climate treaty — was the competent body to phase down the replacement. The legal hook was real: HFCs are not ozone-depleters, but they exist as a direct consequence of ozone policy, and Montreal already had universal ratification, a Multilateral Fund, and a habit of tightening itself. On 15 October 2016, in Kigali, Rwanda, the parties adopted an amendment to phase down HFC production and consumption by more than 80 percent over roughly thirty years, on staggered schedules — developed countries first, China and most developing countries next, a later group including India and the Gulf states after that. Estimates from UNEP and partner assessments suggest that full implementation could avoid about 0.3 to 0.5 degrees Celsius of warming by 2100. The amendment entered into force on 1 January 2019. The United States wrote a domestic HFC phasedown into the American Innovation and Manufacturing Act in December 2020 and the Senate ratified Kigali in September 2022. The ozone hole’s architects had, unusually, been allowed to debug their own unintended consequence.

The meta-irony is already visible at the edges. The next-generation substitutes, hydrofluoroolefins, have very low global warming potentials, which is the point — but some break down into trifluoroacetic acid, and researchers are still arguing over how much TFA accumulation in water and soil should worry us. Hydrocarbons and CO2 avoid that fluorine tail, and then collide with fire codes and training. Illegal trade in phased-out gases has cropped up more than once, including a well-documented burst of unexpected CFC-11 emissions later traced largely to eastern China and then, after enforcement, reduced. None of that undoes Kigali. It does mean the thermostat problem was not a single bad swap. It is a recurring test of whether a fix is evaluated on every axis the Earth actually cares about.

Segment 6 — The Lesson

Three principles sit in the chemistry. First, a solution scored on one environmental axis will migrate risk onto the axes you did not measure — ozone-depletion potential was the right emergency metric in 1987, and it was still an incomplete spec for a molecule that would be manufactured by the megatonne. Second, the substitute inherits the original system’s growth curve: once you bless a chemical that fits existing factories, safety codes, and business models, you have not merely replaced a gas, you have underwritten the expansion of the machine that uses it. Third, institutions can correct themselves, but only if someone is responsible for the interaction between treaties, funds, and markets rather than for a single pollutant. The useful question is not whether the diplomats in Montreal should have seen 2016 from 1987. They were preventing a different catastrophe with the evidence they had. The question is which of today’s celebrated replacements — in cooling, in batteries, in carbon removal, in “forever chemical” swaps — we are currently grading on a single axis, and who, exactly, is assigned to watch the other one.

💬 Reply to this email — Patrick reads every one.

Share: X · LinkedIn · WhatsApp

Forwarded this email? Subscribe here — it's free.