Field notePaper AnalysisAtmospheric ChemistryVolume 3, Issue 1, Fall 2026

The Six-Decade Blind Spot Above the Ozone Layer

High-altitude aircraft measurements find an organic-rich nanoparticle population supplying up to 90% of stratospheric aerosol surface area, a population too small for the satellites and balloon-borne sensors that six decades of sulfate-based ozone and geoengineering models were built on.

Written jointly by the Science Journaling Club

Study: Field note
Review: Peer-edited by the club review board
Length: 14 sectionsabout 20 min
Files: LaTeX source22 kB
Interactive companion

Abstract

Since Christian Junge found a persistent haze of sub-micron particles in the lower stratosphere with balloon-borne counters in 1961 [2], the layer has been treated as a sulfate story: droplets of sulfuric acid and water, replenished by volcanoes, sitting well inside the size range that satellites and balloon-borne optical particle counters can actually see. In April 2026, a team flying NASA's WB-57 high-altitude aircraft during the SABRE mission in February 2023 reported something that sulfate-only bookkeeping had never counted: a persistent mode of organic-rich particles, present below 19 km wherever the air still carries a nitrous oxide signature above 270 parts per billion, with a number-weighted mean diameter of only 0.03 to 0.11 µm [1]. Weighted by the surface area that actually does the chemistry, the same population runs 0.08 to 0.2 µm; weighted by volume, or mass, it runs 0.11 to 0.3 µm — three different "sizes" for one population, depending entirely on what you ask the size to explain. What it explains, per the paper's own accompanying institutional release, is up to 90% of the surface area available in the lower stratosphere for heterogeneous chemical reactions and for condensing trace gases, dwarfing the classic sulfate mode in the one property that actually drives ozone chemistry and particle growth. It went undetected for as long as it did because most satellite retrievals only resolve particles above roughly 0.2 µm, and balloon-borne optical counters generally start around 0.15 to 0.3 µm — both comfortably above where this mode's number distribution actually sits. A global chemistry-climate model run against the same data could not reproduce the population's properties, which the authors read as a warning for anyone using such models to project ozone recovery or to evaluate stratospheric geoengineering proposals that were designed, implicitly, around the sulfate picture alone. This Field Note works through why stratospheric surface area matters at all, does two pieces of arithmetic of our own directly on the paper's three reported size numbers, and asks plainly what changes and what does not.

A Number Nobody Had Budgeted For

Ninety percent is not a subtle correction. It is not the kind of revision that shaves a percentage point off a model's output and gets absorbed quietly into the next generation of parameterizations. It is closer to discovering that the load-bearing wall in a building everyone has been renovating for sixty years was never actually load-bearing, and the wall that was doing the work was the one nobody thought to measure.

That is roughly the position stratospheric aerosol science found itself in this spring. Ming Lyu and thirty co-authors, based mostly at NOAA's Chemical Sciences Laboratory, NASA, NCAR, and the University of Vienna, published a paper in Science reporting in-situ measurements of a mode of organic-rich aerosol particles in the lower stratosphere that, by the authors' own accounting, supplies as much as 90% of the surface area available there for chemistry and for the condensation of trace gases [1]. Not 90% of the mass. Not 90% of the particle count, though the count is dominated by this mode too. Ninety percent of the surface area — the one quantity that governs how fast heterogeneous chemical reactions happen on a particle's skin, and how efficiently it soaks up whatever gas is trying to condense onto it.

We are a school club, not a lab, and we did not fly anything or measure anything ourselves. What we did is sit with the paper, verify what it and its home institution actually claim against the primary sources, and try to explain why a number this large could have gone unmeasured for as long as it did, and what recalculating around it might touch.

Why Surface Area, Specifically, Is the Number That Matters

It would be easy to assume aerosol mass is the thing that counts in the stratosphere, since mass is what you'd load onto a scale, and it is the number climate models often report first. It is not the number that drives the two effects stratospheric aerosol is actually known for.

The first effect is chemical. Ozone destruction by chlorine and bromine compounds proceeds mostly through gas-phase catalytic cycles, but the reactions that convert relatively inert reservoir species, like chlorine nitrate, into forms that can actually attack ozone happen on the surfaces of stratospheric particles, not in the open air. This is exactly the mechanism behind the Antarctic ozone hole's dependence on polar stratospheric clouds, and it is why Susan Solomon's landmark 1999 review of ozone depletion treats heterogeneous surface chemistry as inseparable from the gas-phase story [4]. A particle population's importance to this chemistry scales with its total surface area per unit volume of air, not with how much it weighs.

The same authors have shown how large that dependence can get outside of polar clouds, too. In 2023, Solomon and coauthors, including Daniel Murphy, one of the co-authors on the new nanoparticle paper, reported that organic-rich smoke aerosol injected into the stratosphere by the 2020 Australian wildfires drove extreme heterogeneous chemistry: the smoke's oxidized organic surfaces made hydrogen chloride unusually soluble, and the stratosphere over the Southern Hemisphere subsequently recorded its lowest HCl and highest chlorine nitrate levels on record, with measurable knock-on ozone loss [5]. That paper is a case study in exactly the mechanism this Field Note is about: an organic-rich aerosol surface, not the classic sulfate droplet, driving heterogeneous chemistry that a sulfate-only picture would not have anticipated. Wildfire smoke is episodic and self-evidently anomalous. The new paper's claim is that something with a similar chemical character — organic-rich, transported up from the troposphere — is not episodic at all in the lower stratosphere. It is background.

The second effect is optical, and it is the whole basis of stratospheric aerosol geoengineering. Aerosol reflects incoming sunlight back to space, and for a fixed mass of material, smaller particles scatter light more efficiently per unit mass than larger ones, up to a point set by the wavelength of visible light. Every proposal to deliberately whiten the stratosphere by injecting sulfur, going back to Paul Crutzen's 2006 essay that reopened the topic as a serious research question [7], depends on getting the resulting particle size distribution right, because the radiative effect per kilogram of injected material is extremely sensitive to exactly that distribution. Jeffrey Pierce and coauthors showed in 2010 that this sensitivity cuts both ways: how a proposed injection scheme's material actually forms and grows into aerosol depends on the particles it has to compete and condense with, not merely on how much is injected [8]. A background particle population you did not know was there is exactly the kind of thing that competition depends on.

The Picture That Had Stood Since 1961

The reason a discovery like this can happen in 2026, and not 1975, is that the standard picture has been remarkably stable, and stable pictures do not invite re-measurement.

Christian Junge used balloon-borne condensation-nuclei counters to find a persistent vertical layer of small particles in the lower stratosphere, reported in 1961 in a paper whose title is, plainly enough, "Vertical Profiles of Condensation Nuclei in the Stratosphere" [2]. That layer, still sometimes called the Junge layer, became the anchor for essentially everything that followed: decades of satellite occultation measurements and balloon-borne optical particle counters, synthesized most comprehensively in a 2016 review, spanning more than thirty co-authors, of stratospheric aerosol observations, processes, and climate impact [3]. The background layer between major volcanic eruptions has long been understood, in the broad strokes every atmospheric chemistry course teaches, as predominantly sulfuric-acid droplets. It is a genuinely well-tested picture for what the classic instruments actually measure. The problem the new paper identifies is not that those sulfate measurements were wrong. It is that the instruments built to confirm the sulfate picture were, by design, not built to see much below it.

Where the New Mode Was Actually Found

The measurements behind the new finding came from NASA's WB-57 high-altitude research aircraft, flown during the SABRE mission in February 2023, carrying instruments developed largely at NOAA's Chemical Sciences Laboratory to measure particle size distributions and chemical composition directly, in situ, rather than by remote optical inference [1]. The organic-rich mode shows up specifically in air below about 19 km where the nitrous oxide mixing ratio still exceeds roughly 270 parts per billion, a chemical signature atmospheric scientists use as a proxy for how recently that air parcel arrived from the troposphere rather than having spent years circulating and aging higher in the stratosphere [1]. That detail matters for what the particles are: mostly organic-rich material carried up from the troposphere, not something manufactured in place from stratospheric sulfur chemistry. And the paper reports that the particles are not static once they arrive — they grow in size and drop in number concentration as the air parcel they are riding in mixes with older, more processed stratospheric air, exactly the kind of aging signature you would expect from a population being diluted and coagulated over time rather than continuously replenished from below [1].

Figure 1. The vertical domain the paper reports for the organic-rich mode: below roughly 19 km, in air still carrying a nitrous oxide signature above 270 ppbv of relatively recent tropospheric origin. Particles grow and become less numerous as their air mixes upward into older, more processed stratospheric air — consistent with a population fed from below rather than manufactured in place [1]. Schematic, not a measured altitude profile.

One Population, Three Different Sizes

The paper reports the mode's size three separate ways, and the difference between the three is not noise, it is the entire point. A "number geometric mean diameter" of 0.03 to 0.11 µm describes what you would get if you counted every particle and asked for the typical size — and by that measure the mode is dominated by genuinely tiny particles. A "surface geometric mean diameter" of 0.08 to 0.2 µm describes the typical size weighted by how much surface area each particle contributes. A "volume geometric mean diameter" of 0.11 to 0.3 µm does the same thing weighted by volume, which for a population of roughly similar density is also a mass weighting [1].

Three numbers, one population, and each one nearly doubling or tripling the last. That spread is not a sign of sloppy measurement. It is the ordinary mathematics of a size distribution with a lot of very small particles and a much smaller number of somewhat larger ones: the small particles dominate the count, contribute much less individually to total surface area, and even less to total volume, so each successive weighting shifts the "typical" size upward toward the rarer, larger particles that happen to carry disproportionate area or mass.

Figure 2. The same organic-rich mode's reported diameter range under three different weightings [1]. Chemistry and light-scattering care about the surface- and volume-weighted numbers; a raw particle counter reports something closer to the number-weighted one — which is also the range that sits furthest below what most detection methods can resolve (Figure 3).

Club arithmetic, not the paper's. The paper does not, as far as we found, connect its three reported diameters explicitly. If we idealize the mode as one lognormal size distribution, aerosol theory gives simple relations between the number-weighted, surface-weighted, and volume-weighted mean diameters and a single geometric standard deviation σ_g: d_surface = d_number · exp(2 ln²σ_g) and d_volume = d_number · exp(3 ln²σ_g). Feeding in the paper's own low-end triplet, 0.03 / 0.08 µm, solves to σ_g ≈ 2.0, which predicts a volume diameter of about 0.13 µm — inside the paper's reported 0.11–0.3 µm range. The high-end triplet, 0.11 / 0.2 µm, solves to σ_g ≈ 1.7 and predicts a volume diameter of about 0.27 µm — also inside that range. We take this as a sanity check that the paper's three numbers describe one internally consistent, fairly broad size distribution, not three unrelated statistics; we do not take it as a claim about the distribution's true shape, which the paper's own analysis almost certainly resolves in far more detail than a single lognormal.

Why Something This Abundant Went Unmeasured for Sixty-Five Years

It is tempting to read "90% of the surface area" and assume the earlier measurements must have been careless. They were not. They were built, reasonably, around what the available technology could resolve, and what the available technology could resolve had a floor.

Satellite instruments infer stratospheric aerosol from how they extinguish or scatter light along a long optical path through the atmosphere, and that technique loses sensitivity steeply for particles much smaller than roughly 0.2 µm in diameter, because Mie scattering efficiency falls off sharply once a particle gets small relative to the wavelength of visible or near-infrared light being used [1]. Balloon-borne optical particle counters, the other pillar of the historical record, count particles by their individual scattering pulses as they pass through a laser beam, and in practice that technique generally starts resolving particles somewhere around 0.15 to 0.3 µm in diameter, with the exact floor depending on instrument design and on assumptions, often an assumed lognormal or gamma size distribution, used to invert the raw optical signal into a size [1, 9]. Recent methodological work on exactly this inversion step, comparing lognormal against gamma distribution assumptions for optical particle counter data, underlines how much of a stratospheric aerosol "measurement" is actually a measurement plus a size-distribution assumption baked in at the analysis stage [9].

Now put those two floors next to the organic mode's own number-weighted size: 0.03 to 0.11 µm. The entire number distribution of the newly reported mode sits below where either classic method reliably starts counting.

Figure 3. A size ruler, log-scaled in micrometers. The organic mode's number-weighted diameter range (amber) sits entirely below the resolving floors of balloon-borne optical particle counters (grey) and satellite occultation retrievals (blue) [1]. The lighter amber band underneath shows the full measured particle range down to 3 nm reported by the paper's home institution [1].

Nobody mismeasured the sulfate layer. The instruments simply had a floor, and this whole population was living underneath it.

Doing the Club's Own Arithmetic: Why Small Particles Win on Surface Area

The mechanism behind "dominates surface area despite being small" is not exotic; it is a direct consequence of how a sphere's surface area and volume scale with its diameter, and it is worth doing the arithmetic ourselves rather than just asserting it.

Surface area of a sphere scales as its diameter squared; volume scales as diameter cubed. So if you take a fixed total volume, meaning a fixed total mass, of aerosol material and divide it into particles of diameter d instead of a single particle of diameter D, the number of small particles you get is (D/d)³, and the total surface area of all of them together comes out to exactly (D/d) times the surface area of the single big particle.

Club arithmetic, not the paper's. Take the paper's own volume-weighted upper bound, D = 0.3 µm, as a stand-in for a single "big" particle, and its number-weighted lower bound, d = 0.03 µm, as a stand-in for the small particles that actually dominate the count. The ratio D/d = 10. Chopping the same total volume of material into 0.03 µm particles instead of one 0.3 µm particle takes 10³ = 1,000 particles — and multiplies the total surface area by exactly 10×. That tenfold jump, from ordinary geometry alone and using only the two diameters the paper itself reports, is the whole mechanical reason a population that need not dominate total mass can still dominate the property, surface area, that chemistry and light-scattering actually respond to.

total surface area = 10× the single particle

(grid shown schematically, not at literal count)

Figure 4. The club's own geometry, using only the two diameters the paper reports. Splitting a fixed volume of material from one 0.3 µm particle into 0.03 µm particles multiplies the total surface area tenfold — the mechanical reason a numerous small-particle population can dominate surface area chemistry while being a minority of the mass. Grid shown schematically, not at literal 1,000-particle count.

What the Model Comparison Actually Says

The paper does not stop at reporting the measurement. It also runs a global chemistry-climate model against the same conditions and reports that the model fails to reproduce the observed population's characteristics [1]. It is worth being precise about what that does and does not mean. It does not mean the model is broken in some general sense, or that its ozone chemistry is wrong wherever it has been validated against other observations. It means that a specific process, the persistent injection and evolution of organic-rich nanoparticles from the troposphere into the lower stratosphere, is not currently represented in a way that reproduces what the aircraft actually measured. That is a gap in a specific module, not a global indictment, and the paper's own framing treats it that way: as a reason model improvement is necessary before such models can be trusted for the two downstream jobs that most depend on getting stratospheric aerosol surface area right [1].

What This Could Change About Ozone Recovery Projections

The Antarctic ozone hole's slow recovery under the Montreal Protocol is itself tracked using exactly the kind of chemistry-climate models this paper is warning about. Solomon and coauthors reported the first statistically robust signs of that healing in 2016, using both observations and modeled expectations of how ozone should respond as stratospheric chlorine slowly declines [6]. Those recovery projections lean on the same heterogeneous chemistry machinery discussed above: how much surface area is available, and what that surface is made of, feeds directly into how fast reservoir chlorine and bromine species get reactivated.

We want to be careful here, because the new paper does not itself publish a revised ozone recovery timeline, and neither can we. What it does establish is that a large fraction of the relevant surface area in the lower stratosphere has a different chemical character, organic rather than sulfate, than the models computing those timelines currently assume, and that the same models cannot yet reproduce this population's basic properties [1]. Given that organic aerosol surfaces have already been shown, in the wildfire smoke case, to host heterogeneous chemistry that a sulfate-only model would not predict [5], the open and legitimate question is whether background organic surface area does something chemically similar, at a much steadier and smaller-magnitude scale, everywhere in the lower stratosphere, all the time, rather than only during smoke events. That is a hypothesis the paper motivates. It is not a result either paper claims to have already shown.

What This Could Change About Geoengineering Proposals

Every serious stratospheric aerosol injection proposal traces back to the same basic move Crutzen proposed in 2006: put reflective material into the stratosphere deliberately, at a scale that could offset a meaningful fraction of greenhouse warming, using the same physical mechanism, sulfate aerosol reflecting sunlight, that large volcanic eruptions already demonstrate happens naturally [7]. The physical realism of that idea has always depended on getting the resulting particle size distribution right, because both the radiative benefit per kilogram injected and the atmospheric lifetime of the injected material are sensitive to particle size, and Pierce and coauthors showed as early as 2010 that how injected material actually forms and grows into aerosol depends on what it condenses onto in the existing atmosphere, not simply on how much material goes up [8].

An unaccounted-for background population supplying up to 90% of the ambient surface area is exactly the kind of thing that competition depends on. Newly injected sulfate precursor gases do not condense in a vacuum; they condense preferentially onto whatever particle surface is already there to receive them, and the more of that surface is organic-rich nanoparticle rather than classic sulfate, the more the resulting particle sizes, growth rates, and ultimately radiative efficiency of an injection scheme could differ from what a model built on the sulfate-only picture would predict. The paper's own framing puts this plainly: models need improvement specifically for assessing proposed geoengineering efforts [1]. We read that as a caution rather than a verdict. It says the size distribution injected material would actually settle into is less certain than modeled projections have implied, not that any particular proposal has been shown to fail.

Figure 5. A schematic of why one measurement touches two very different literatures. Both the projected pace of ozone recovery and the projected efficiency of proposed stratospheric injection schemes run through models that need an accurate accounting of background aerosol surface area and its chemical character — the two things this paper's central number revises.

What We're Not Claiming

We want to close the technical discussion by being explicit about the boundary of what a single measurement paper, however striking, actually licenses. The paper reports a real, large, previously unresolved surface-area contribution and a model's failure to reproduce it. It does not, by itself, report a revised ozone recovery date, a revised geoengineering efficiency number, or a mechanism by which the organic surface specifically alters halogen chemistry at background stratospheric abundances the way it demonstrably did during the 2020 wildfire smoke episode. Those are the natural next questions the finding raises, not answers it already contains. We have tried, in the two sections above, to be careful about which side of that line each sentence sits on.

What Would Change Our Mind

Our own arithmetic in this piece rests on treating the paper's three reported diameters as points on one lognormal distribution, which is a simplification the authors themselves may not need or endorse; if a follow-up paper shows the mode is genuinely bimodal or otherwise non-lognormal in a way that breaks that consistency check, we would want to revisit the tenfold surface-area comparison in Figure 5 as well, since it uses the same two boundary diameters. We would also want to see the finding reproduced by an independent campaign, ideally with a different aircraft or balloon platform and a different in-situ sizing instrument, before treating "as much as 90%" as a stable number rather than a value specific to the February 2023 SABRE flights over the conditions they sampled.

Why We Picked This Paper

Most of what makes it interesting to us is not the ozone or geoengineering implications on their own, real as they are. It is the shape of the discovery: a genuinely enormous, genuinely background feature of the atmosphere, not a rare event or a faint signal buried in noise, sitting in plain sight below a detection threshold that six decades of otherwise careful instrumentation happened to share. That is a specific, checkable kind of blind spot, not a vague appeal to "more research is needed," and it is checkable precisely because the paper reports its own numbers, and the numbers hold together under the club's own arithmetic when we go looking.

References

Lyu, M., Ahern, A. T., Schill, G. P., Lawler, M. J., Murphy, D. M., et al. (2026). An unrecognized mode of small particles in the lower stratosphere. Science 392(6796). doi:10.1126/science.adw8939
Junge, C. E. (1961). Vertical profiles of condensation nuclei in the stratosphere. Journal of Meteorology 18(4), 501–509. doi:10.1175/1520-0469(1961)018<0501:VPOCNI>2.0.CO;2
Kremser, S., Thomason, L. W., von Hobe, M., et al. (2016). Stratospheric aerosol—Observations, processes, and impact on climate. Reviews of Geophysics 54(2), 278–335. doi:10.1002/2015RG000511
Solomon, S. (1999). Stratospheric ozone depletion: A review of concepts and history. Reviews of Geophysics 37(3), 275–316. doi:10.1029/1999RG900008
Solomon, S., Stone, K., Yu, P., Murphy, D. M., Kinnison, D., Ravishankara, A. R. & Wang, P. (2023). Chlorine activation and enhanced ozone depletion induced by wildfire aerosol. Nature 615(7951), 259–264. doi:10.1038/s41586-022-05683-0
Solomon, S., Ivy, D. J., Kinnison, D., Mills, M. J., Neely III, R. R. & Schmidt, A. (2016). Emergence of healing in the Antarctic ozone layer. Science 353(6296), 269–274. doi:10.1126/science.aae0061
Crutzen, P. J. (2006). Albedo enhancement by stratospheric sulfur injections: A contribution to resolve a policy dilemma? Climatic Change 77(3–4). doi:10.1007/s10584-006-9101-y
Pierce, J. R., Weisenstein, D. K., Heckendorn, P., Peter, T. & Keith, D. W. (2010). Efficient formation of stratospheric aerosol for climate engineering by emission of condensible vapor from aircraft. Geophysical Research Letters 37, L18805. doi:10.1029/2010GL043975
Nyaku, E., Loughman, R., Bhartia, P. K., Deshler, T., Chen, Z. & Colarco, P. R. (2020). A comparison of lognormal and gamma size distributions for characterizing the stratospheric aerosol phase function from optical particle counter measurements. Atmospheric Measurement Techniques 13(3), 1071–1087. doi:10.5194/amt-13-1071-2020
NOAA Chemical Sciences Laboratory (2026). New measurements reveal surprising abundance of stratospheric nanoparticles. Press release accompanying the publication of [1].