The ozone layer was supposed to be on a steady path to healing. A new study suggests that path has a leak.
Researchers from MIT and the University of Bristol report that industrial chemicals escaping from manufacturing plants are reaching the atmosphere at roughly seven times the rate international regulators have assumed. The gap, detailed in a study published in Nature Communications in early 2026, is large enough to push back full recovery of Earth’s protective ozone layer by years, depending on how the world responds.
The chemicals in question are not rogue pollutants. They are ozone-depleting substances still produced legally in large volumes as feedstocks, meaning they serve as raw ingredients in the manufacture of refrigerants, solvents, and other products. The Montreal Protocol, the 1987 treaty that phased out most uses of these substances, deliberately exempted feedstock production on the assumption that very little material would escape during processing. That assumption, the researchers argue, is badly wrong.
A sevenfold undercount
The study’s central finding is specific: feedstock-related emissions involve a typical loss rate of about 3.6% of production volume, compared with the 0.5% figure that has served as the standard estimate in international assessments. When production volumes run into hundreds of thousands of tons per year, that difference translates into a substantial and largely untracked flow of chlorine- and bromine-containing gases into the stratosphere.
Carbon tetrachloride, or CCl4, is the clearest example. Direct uses of the chemical were banned decades ago, yet it remains widely produced as a feedstock for manufacturing hydrofluorocarbons and other compounds. NOAA’s Global Monitoring Laboratory has tracked atmospheric concentrations of CCl4 since 1987, and those measurements tell a stubborn story: the chemical is not disappearing as fast as it should be if feedstock losses were truly negligible. The slow, halting decline in CCl4 levels has puzzled atmospheric scientists for years. This study offers a direct explanation.
The research team combined reported industrial production data with atmospheric observations to back-calculate how much material must be leaking. Their modeling extends through the year 2100, projecting a range of outcomes depending on whether feedstock emissions hold steady, grow alongside industrial expansion, or fall under tighter controls.
MIT atmospheric chemist Susan Solomon, one of the scientists whose work helped build the case for the original Montreal Protocol, is among the study’s authors. She described the findings as evidence that unexpected feedstock emissions could delay healing of the ozone hole, according to an MIT institutional summary. Stefan Reimann of Empa, the Swiss Federal Laboratories for Materials Science and Technology, contributed measurement and analytical expertise, helping the team trace where the leaks originate across industrial sectors.
Why the treaty has a blind spot
The Montreal Protocol is widely regarded as the most successful environmental treaty ever negotiated. It phased out production and consumption of chlorofluorocarbons, halons, and other ozone-depleting substances for nearly all applications, and atmospheric levels of many of those chemicals have been declining for decades.
But the treaty drew a sharp line between emissive uses and feedstock uses. When a chemical is sprayed from an aerosol can or leaks from an old refrigerator, it reaches the atmosphere directly. When it is consumed as a feedstock, most of it is chemically transformed into something else inside a reactor vessel. Treaty negotiators in the late 1980s judged the atmospheric risk from feedstock processes to be minimal and excluded them from the consumption calculations that drive national compliance.
That judgment made sense when feedstock volumes were modest relative to emissive uses. As the phase-out succeeded and emissive production collapsed, feedstock production held steady or grew, particularly in countries with expanding chemical industries. The result is that feedstock processes now account for a much larger share of total handling of ozone-depleting substances than they did when the exemption was written. Even a small percentage loss from a large and growing volume adds up.
The 2022 WMO/UNEP Scientific Assessment of Ozone Depletion, the most recent official benchmark, projected that the ozone layer would return to pre-1980 levels by roughly mid-century over most of the planet. That projection assumed feedstock emissions were minor. If the 3.6% loss rate holds globally, those timelines will need to be revised.
What remains uncertain
The study raises as many questions as it answers. Global production volumes for individual feedstock chemicals are not fully transparent. Countries report aggregate figures to treaty bodies, and independent verification of emissions at the plant level is limited. Without granular data from factories, the total tonnage of ozone-depleting substances reaching the stratosphere through feedstock pathways carries a range of uncertainty rather than a single precise number.
Geography is part of the problem. The European Environment Agency’s 2025 inventory on ozone-depleting substances confirms that feedstock volumes remain significant within the EU, but European data alone cannot capture the global picture. Major chemical-producing nations in East and South Asia, where much of the world’s CCl4-derived manufacturing is concentrated, provide less detailed or less publicly accessible reporting. Whether the 3.6% figure applies uniformly across different processes, plant ages, and regulatory environments is something the study flags but cannot fully resolve. Some facilities may run far tighter operations than the average; others may be responsible for outsized leaks.
The political path forward is equally unclear. Tightening controls on feedstock emissions would require consensus among nearly 200 Montreal Protocol signatories. No public statements from treaty negotiators about potential amendments have surfaced as of May 2026, and countries with large chemical sectors may resist stricter rules over concerns about costs and competitiveness.
The study itself does not assign a single number to the delay in ozone recovery. It models a range of scenarios: in high-emission paths, additional chlorine and bromine entering the stratosphere stretch out the timeline significantly; in lower-emission paths, where leak rates are reduced through better technology or regulation, the impact is smaller but still measurable against earlier projections. Readers should treat any specific “years of delay” figure in secondary coverage with caution unless it is tied to a named scenario from the peer-reviewed paper.
Where the science goes from here
The strongest evidence supporting the study comes from two independent lines. The first is the peer-reviewed paper itself, which provides the 3.6% versus 0.5% comparison and a modeling framework that has passed scientific scrutiny before publication. The second is NOAA’s decades-long atmospheric monitoring record for CCl4, which offers an observational check that does not depend on industry self-reporting. The persistence of CCl4 in the atmosphere, documented continuously since 1987, is consistent with the claim that feedstock leakage is substantially larger than assumed.
If future WMO/UNEP assessments adopt the higher emission rate and adjust their scenarios accordingly, official recovery dates for the ozone layer could shift, and policymakers may need to revisit whether the Montreal Protocol’s feedstock exemption still makes scientific sense. The treaty has been amended before, most notably in the 2016 Kigali Amendment targeting hydrofluorocarbons as greenhouse gases. Whether feedstock emissions generate similar political momentum remains to be seen.
For now, the picture is coherent but incomplete: a category of chemical use that was once treated as a rounding error in ozone protection appears to be leaking enough material to matter. The question is no longer whether the blind spot exists. It is how long the world takes to close it.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
