The Volcano We Could Build Would Reshape the Air We Breathe — Just Not the Way You'd Expect

A thin haze of imagined sulfur, injected high into the stratosphere, is one of the most seductive and terrifying ideas ever proposed for fixing the climate. The logic is brutally simple: volcanoes already do it. When Mount Pinatubo erupted in 1991, it lofted millions of tons of sulfur dioxide into the sky, and the planet cooled by about half a degree Celsius for a year. If we could mimic that eruption on purpose, year after year, we might buy the world time while we decarbonize. Stratospheric aerosol geoengineering, as scientists call it, could blunt the worst of global warming.

But here's the thing nobody had really asked: what would all that injected sulfur do to the air we breathe?

Not the climate. The air. The microscopic particles that lodge in our lungs and contribute to millions of premature deaths each year. Sulfate aerosols are themselves an ingredient in PM2.5 — particulate matter smaller than 2.5 micrometers, small enough to slip past the body's defenses and into the bloodstream. So the intuition runs: if you inject enormous quantities of sulfate-forming gases into the sky, surely some of it must come down and worsen the air pollution that already kills far too many people.

Lili Xia of Rutgers University, Alan Robock, and Simone Tilmes decided to actually test that intuition — using a purpose-built ensemble of climate simulations called GLENS (the Geoengineering Large Ensemble). Their finding, published on arXiv in August 2026, is genuinely surprising and runs against the obvious worry. According to the study, stratospheric sulfate geoengineering would reduce global average surface PM2.5 concentrations compared to a scenario where we just let warming run rampant. Total PM2.5 goes down. But the composition changes — more organic carbon, more sulfate, less dust and sea salt. And digging deeper reveals a subtler, more uncomfortable truth: strip out the dust and sea salt, and geoengineering actually produces more of the human-caused fine particles that are worst for health. The answer to "does geoengineering clean or dirty our air?" turns out to be a complicated, both-at-once answer that depends entirely on which particles you're talking about and where in the world you're standing.

This is not academic trivia. It is a first look at the air-quality consequences of a technology that is no longer purely hypothetical — it is being seriously modeled, debated, and even field-tested by small-scale experimental programs. Understanding what it would do to the air people breathe is a prerequisite for any honest conversation about whether "the volcanoes we could build" is an idea worth taking seriously at all.

The Science

The engine behind this work is GLENS — a set of climate model simulations run on the Community Earth System Model, version 1 (CESM1). The experiment design is elegant in its simplicity and audacious in its scale. The researchers simulate a world following RCP8.5, the grimmest of the standard climate scenarios — a world where emissions keep rising through the century and global average temperature climbs several degrees. Then they run a parallel world where, starting from 2020, they inject sulfur dioxide (SO₂) into the stratosphere at rates calibrated to exactly cancel out the warming from the rising greenhouse gases. Four injection locations spread sulfur around the globe. The goal: hold the global average temperature at its 2020 value.

The genius of GLENS is that it's an ensemble — a whole family of runs with slightly varied starting conditions, so the results aren't hostage to one particular weather fluke. It also couples the atmosphere to interactive oceans, land, and chemistry, allowing researchers to see real physical consequences, not just temperature numbers.

For this study, Xia and colleagues did something nobody had done with the GLENS output before: they pulled out the simulated surface PM2.5 concentrations and their chemical composition. They compared the geoengineered world against the business-as-usual RCP8.5 world — a comparison that matters, because the honest question was never "is geoengineering cleaner than a clean world?" but "is it better or worse than the world we're actually heading toward?"

PM2.5 is a messy cocktail, not a single substance. The models track its components separately: sulfate, organic carbon, black carbon (soot), dust, sea salt, and secondary organic aerosol (SOA) — particles formed in the atmosphere when gases react and condense into particulate form. Untangling which ingredients change, and by how much, is the real content of the study.

What They Found

The headline result: global average surface PM2.5 drops under geoengineering. According to the study, the reduction in total PM2.5 under GLENS geoengineering relative to RCP8.5 comes almost entirely from less dust and less sea salt — not from any cleanup of the human-made pollutants that dominate health statistics.

Why would geoengineering cut dust? Because of soil moisture and vegetation. In the geoengineered world, cooler temperatures and altered precipitation patterns leave more water in the soil over desert regions, and leaf area index — a measure of how much living plant surface covers the ground — increases there too. More moisture and more vegetation means less bare, wind-scoured soil, and therefore less dust lofted into the air. Climate change, by contrast, is a dust-buster's friend: hotter, drier conditions pump more mineral dust into the atmosphere.

Sea salt falls for a different reason: slower winds. Geoengineering alters circulation, generally calming surface winds, and calm seas spray fewer salt droplets into the air. Combined, dust and sea salt dominate the total mass of PM2.5 in much of the world, so their decline drags the global average down even while the ingredients that actually cause the most health harm — the combustion-derived carbon and sulfate particles we generate ourselves — are increasing.

Because dust and sea salt are such enormous components of total particulate mass, removing them can mask a worsening of the rest. Xia and her colleagues did the honest accounting: exclude dust and sea salt, and the global average PM2.5 under geoengineering is actually higher than under RCP8.5. The culprit is secondary organic aerosol, or SOA. Here the mechanism is subtle and fascinating. The gaseous precursors that eventually become SOA are fixed in the simulation — the model prescribes how much of these volatile organic gases exist, regardless of the geoengineering. What changes is the partitioning: in the cooler geoengineered atmosphere, thermodynamics favor pulling more of those gases out of the vapor phase and condensing them into aerosol particles. Cooler means more condensation, more SOA, more PM2.5. Geoengineering doesn't create new pollution chemistry from nothing — it just shifts an existing reservoir of gases further into the particle phase.

Then there's the question that would occur to any skeptical reader: what about all that injected sulfate itself? Surely millions of tons of SO₂ injected into the stratosphere must come raining down as sulfate pollution at the surface? Remarkably, the study finds it does not — at least not in the form of PM2.5. The majority of the sulfate aerosol that eventually makes its way down into the boundary layer — the lowest part of the atmosphere where we live — arrives in the coarse mode, in particles larger than 2.5 micrometers. Those larger particles are excluded from the PM2.5 definition, so the strato-injected sulfur largely dodges the health-relevant size cut. It's a striking and counterintuitive result: one of the biggest self-created pollutants in the geoengineering system barely registers in the metric that matters most for human health.

Why This Changes Things

This paper rewires how we should think about the air-quality trade-offs of geoengineering. The naive worry — "inject sulfates, breathe sulfates" — is not supported by the simulation. But the story is not simply reassuring either. Reading it charitably, geoengineering might actually improve global average air quality relative to unmitigated warming, mostly by suppressing the natural dust and sea-salt loading. Reading it less charitably, the global average hides enormous regional differences, and the components that do increase — sulfate and organic carbon — are precisely the fine, deep-inhaled particles most strongly associated with cardiovascular and respiratory mortality.

The distinction between mass and harm matters enormously here. Dust and sea salt are the heavyweights of PM2.5 mass, but they are comparatively benign per unit of mass. Organic carbon and sulfate are lighter in total mass but punch well above their weight in toxicity, especially sulfate, which is strongly linked to premature death in epidemiological studies from China, India, and elsewhere. So a decline in total mass can conceal an increase in the mass that actually kills. The finding that global PM2.5 mass drops while the chemical composition shifts toward more sulfate and organic carbon is exactly the kind of trade-off that a global-average number would smooth over and hide.

The ecosystem angle adds another layer. Even though the injected sulfate mostly doesn't show up as surface PM2.5, it does come down somewhere — as deposition. The study emphasizes that the spatial pattern of sulfate deposition differs between the geoengineered world and RCP8.5. Where acid and nutrient deposition lands determines its impact on forests, lakes, and soils. Sulfur deposition is not just an air pollution problem; it acidifies ecosystems. If geoengineering shifts where that sulfur rains down, it could concentrate acidity in sensitive regions — a slow-moving, hard-to-see ecological cost that the global PM2.5 metric completely misses.

There's also a deeper structural lesson here that extends well beyond this one technology. The paper is a reminder that climate interventions have multiple, interlocking consequences that cannot be captured by any single headline number. Global average temperature tells you almost nothing about local air quality. Total PM2.5 mass tells you almost nothing about its toxicity. A technology designed to cool the planet turns out to rearrange the entire chemistry of the lower atmosphere — vegetation, wind, condensation, deposition — in ways that touch agriculture, ecosystems, and public health simultaneously.

The regional heterogeneity deserves emphasis. A global average of PM2.5 is built from dramatically different local stories: dust-dominated arid regions, sea-salt-heavy coastal zones, and pollution-dominated urban and industrial areas will each respond differently. Some places will breathe cleaner under geoengineering; others, particularly those already burdened by sulfate and organic aerosol from combustion, could see the human-made fraction grow. Global-mean numbers — including the ones in this paper — can flatten those differences into a single misleading reassurance.

What's Next

The authors are careful to flag the limits of what a single Earth-system model can tell us. CESM1 is one model; the partitioning of SOA between gas and aerosol phase, and the exact behavior of sulfate in the stratosphere-to-troposphere transition, carry real uncertainty. The findings need confirmation from other models before we treat them as robust. And crucially, the study models a fixed injection schedule tuned to cancel RCP8.5 warming; real deployment decisions, if they ever happen, would involve adaptive choices, regional targeting, and off-ramps that these idealized scenarios don't capture.

The call for health-impact assessment is the paper's most important next step. Knowing that PM2.5 composition shifts toward more sulfate and organic carbon in some regions is a first step, but a full accounting requires translating those concentration changes into mortality and morbidity estimates using the epidemiological concentration-response functions — the same tools used to quantify deaths from coal plants and diesel exhaust. No such head-to-head health calculus exists yet for geoengineering, and this study is the scaffolding on which it should be built.

At a deeper level, the paper signals something about the state of the geoengineering conversation. Once a taboo topic, stratospheric aerosol injection is now being studied by major research programs, with serious attention to governance, ethics, and — now — public health. The value of a study like this is not that it settles the question of whether we should do it. It's that it replaces hand-waving with mechanism, and intuition with numbers. It tells us the trade-offs are real, complicated, and in need of far more research before anyone should be confident that tinkering with the sulfur cycle of the whole planet is a bargain worth striking.

The most honest summary may be this: geoengineering would probably not turn the sky into a toxic soup of injected sulfate — the coarse mode and the boundary-layer dynamics save us from that particular nightmare. But it would subtly and unevenly re-engineer the chemistry of the air people breathe, shifting the balance of particles in ways that have yet to be translated into the language of years of life lost or gained. The right response to this paper is not fear and not relief. It is a demand for the next study: the one that tells us, in human terms, what this re-arrangement of the air actually means.