Mexico City aerosol analysis during MILAGRO using high resolution aerosol mass spectrometry at the urban supersite (T0) – Part 1Fine particle composition and organic source apportionment

March two thousand six, Mexico City. Somewhere in the Instituto Mexicano del Petróleo, nine kilometers northeast of the city center, a suitcase-sized instrument is pulling in air and tearing it apart molecule by molecule. Twenty million people are waking up below a ring of mountains that traps whatever they breathe out. The city's own accounting of what it emits is off by a factor of four. That number is not a rounding error. It is what Aiken and colleagues found when they actually measured it. Mexico City is about as difficult an air quality problem as exists on Earth. The Mexico City Metropolitan Area packs over twenty million people into roughly one thousand five hundred square kilometers, sitting at two thousand two hundred forty meters above sea level in a mountain basin near nineteen degrees north. The elevation means lower atmospheric pressure and altered combustion chemistry. The basin geometry means that at night and in the early morning, the mountains wall off ventilation. The low latitude means intense solar radiation driving fierce photochemistry during the day. The result is a city that is simultaneously a local health emergency and a major source of chemically evolving pollution affecting the wider region. The MILAGRO campaign, or Megacity Initiative: Local and Global Research Observations, was designed to probe exactly this system during March two thousand six. The T zero urban supersite provided the densest instrumentation, anchored by the High-Resolution Time-of-Flight Aerosol Mass Spectrometer, or HR-ToF-AMS. Understanding what that instrument does is worth a moment, because the whole study rides on it. Air enters the instrument through a PM2.5 cyclone, gets dried, and then passes into a chamber where a thermal vaporizer running at about six hundred degrees Celsius flash-vaporizes everything that isn't refractory — meaning everything that isn't black carbon, soil dust, or metals. The resulting vapor gets ionized and sent through the mass spectrometer. Standard aerosol mass spectrometers stop there. The high-resolution version adds the ability to separate ions that sit at the same nominal mass but have different elemental compositions. Take mass-to-charge forty-three: during traffic-dominated periods, the signal there is dominated by C3H7, a reduced hydrocarbon fragment. During photochemically active periods, the same nominal mass is dominated by C2H3O, an oxygenated fragment. A lower-resolution instrument sees one blurred peak. The HR-ToF-AMS sees two distinct chemical stories. That resolving power is what makes source apportionment possible. So what did they find in Mexico City's air? The campaign-average PM1 concentration at T zero was thirty-three point seven micrograms per cubic meter. Refractory species — black carbon, metals, soil — accounted for twenty percent of that. Organic aerosol made up roughly half of the submicron mass, with the remainder being inorganic species like sulfate, nitrate, and ammonium. The daily rhythm was clear. Morning traffic produced a sharp primary emissions peak between six and eight AM, amplified by the shallow morning boundary layer trapping pollutants close to the surface. Then, as sunlight intensified, secondary formation took over. The photochemically produced organic fraction reached a midday peak of about twelve point seven micrograms per cubic meter, with that window running from roughly nine AM to three PM. These patterns closely matched what Salcedo and colleagues had found at the CENICA supersite during the earlier MCMA two thousand three campaign, confirming that Mexico City's aerosol has a consistent character across years. The central analytical tool for untangling sources was Positive Matrix Factorization, or PMF. Think of it as a mathematical unmixing algorithm: given thousands of mass spectra measured over three weeks, PMF finds the smallest set of repeating chemical fingerprints that can explain all of them, along with how much of each fingerprint is present at each moment. Aiken and colleagues found four reproducible components. The first is HOA, or hydrocarbon-like organic aerosol — the chemically reduced, combustion-related primary fraction. It correlates strongly with black carbon and carbon monoxide, shows a morning traffic peak, and averages about twenty-nine percent of organic aerosol mass. Its oxygen-to-carbon ratio is just zero point sixteen, meaning it is mostly unoxidized. The second, and largest, component is OOA, or oxygenated organic aerosol, interpreted as mostly secondary. OOA correlates with particulate nitrate and with Ox, the sum of ozone and nitrogen dioxide that serves as a photochemical clock. It averages forty-six percent of organic aerosol across the campaign and dominates daytime composition, accounting for seventy percent or more of organic aerosol between eleven AM and five PM. Even at night, a persistent background of four point six micrograms per cubic meter of OOA remained — secondary material that had built up and not dissipated. The third component is BBOA, or biomass burning organic aerosol. Its signature ion is at mass-to-charge sixty, a fragment from levoglucosan, the molecular marker produced when cellulose burns. The AMS-derived levoglucosan-equivalent mass correlates with filter-measured levoglucosan at an R-squared of zero point seventy-nine, and the PMF BBOA factor itself tracks the levoglucosan signal with an R-squared of zero point ninety-three. BBOA averages sixteen percent of organic aerosol, with an afternoon and evening secondary peak that follows satellite fire counts from the GOES instrument — these are regional forest fires, sometimes arriving as large discrete plumes, alongside likely contributions from refuse burning. The fourth component, LOA, is the one that surprises. It is a small, local, nitrogen-containing primary organic aerosol, likely from amine or industrial sources. It averages just nine percent of organic aerosol mass. But its nitrogen-to-carbon ratio is roughly four times that of any other component, which means LOA carries about one-third of all the organic nitrogen detected in the aerosol. It shows up in short spikes, often overnight, and correlates with nitrogen-containing ions and with nitrogen-rich particle counts from a co-located instrument. Small in mass, outsize in chemistry. Now here is where the study gets its teeth. Aiken and colleagues compared these PMF results to a completely independent method: Chemical Mass Balance using Organic Molecular Markers, or CMB-OMM, derived from twenty-four hour gas chromatography mass spectrometry filter samples analyzed by Stone and colleagues. CMB-OMM works differently — it uses specific molecular tracers and known source profiles to partition organic carbon among sources. The two methods were applied to overlapping data, and the verdict is clear: they largely agree. Both assign the major fractions of organic aerosol to primary urban combustion, biomass burning, and secondary formation at similar magnitudes. PMF-AMS finds OOA at forty-six percent; once you subtract the LOA component from the CMB "Other" category, the implied CMB secondary fraction is forty-nine percent. That is not a coincidence — it is two orthogonal techniques arriving at the same answer. Where they diverge is also instructive. PMF-AMS resolves LOA as a distinct component. CMB-OMM has no molecular marker for amines and folds it into "Other." The biomass burning comparison shows the CMB approach likely underestimates woodsmoke, possibly because levoglucosan degrades in the atmosphere before the twenty-four hour filter is collected. These divergences are informative, not damaging. They show each method's specific blind spots. Which brings us to the practical verdict. Aiken and colleagues compared their measurements against the two thousand six MCMA emissions inventory and found three large mismatches. The inventory underestimates urban primary PM2.5 by a factor of roughly four. When secondary species are included, the inventory total runs about sixteen times lower than observed afternoon concentrations at T zero. The inventory's estimate of forest-fire PM contribution is at least an order of magnitude below what the measurements showed. These are not small calibration offsets. An inventory that is four to sixteen times too low, and that misses an order of magnitude of fire emissions, cannot support reliable source control decisions or atmospheric modeling. The city was trying to manage an air quality problem using an accounting system that dramatically undercounted what it was accounting for. That is the deeper value of what the HR-ToF-AMS made possible during MILAGRO. The instrument did not just characterize aerosol chemistry — it exposed the gap between what a megacity reports it emits and what its residents are actually breathing. High-time-resolution, composition-resolved, source-apportioned measurement is not an academic refinement. It is the diagnostic that reveals when the official numbers are wrong, and by how much. In Mexico City in March two thousand six, the answer was: by a lot. This lecture was created by ennepō. Go to https://ennepo.ai to Discover, Create and Follow the latest research in your field. Read when you can. Listen when you want to.