PM2.5 is not enough. Why toxicity matters more than mass.

Imagine if we assessed water quality the way we currently assess air quality. You open your tap, and instead of information about lead, bacteria, pesticides, arsenic, PFAS, or microplastics, you are given one single number: “Water Quality: 47.”

No explanation of what that number means. No indication of whether the contamination is harmless minerals or lethal chemicals. No data about pathogens, metals, or organic toxins. Just a single mass-based value, averaged over everything from dissolved salt to industrial sludge. You would never accept that.

No regulator, no water utility, and no health authority would ever claim that one number captures everything we need to know about water safety. Water pollution is inherently about what is in the water — not just how much. And yet, in air quality, this is exactly what we do.

For decades, we have built entire regulatory frameworks, health advisories, consultancy reports, and community apps around a single number: PM2.5. It looks precise, authoritative, and scientifically valid. But in reality, it is little more than an aggregate mass of everything smaller than 2.5 micrometres. Whether the particles are benign sea salt or toxic metal oxides, the PM2.5 value treats them as identical simply because they happen to be small enough.

From a scientific and public-health standpoint, this is as irresponsible as measuring water pollution without knowing whether you are dealing with calcium carbonate or arsenic.

Particle composition is the true driver of health outcomes

Toxicology does not care about particle size alone; it cares about chemistry:

• Heavy metals such as arsenic, chromium, nickel, lead, and cadmium are strongly associated with DNA damage, oxidative stress, neurotoxicity, and cancer. They can accumulate in the body and have no safe exposure level.
• Black carbon (soot) from diesel and biomass combustion is highly inflammatory, carries adsorbed organics, and penetrates deep into the alveoli and even translocates into the bloodstream.
• Sea salt particles, by contrast, are relatively benign; they dissolve in water and are exhaled or cleared quickly.

A PM2.5 measurement of 45 µg/m³ tells you nothing about which of these you are breathing. Yet they vary in toxicity by orders of magnitude.

The scientific community knows this, but the regulatory system fails

Ask any atmospheric chemist or toxicologist: “Is PM2.5 mass a good proxy for health risk?” They will answer: “It’s a starting point — but it hides the most important information.” And yet, global air quality standards and regulatory frameworks still pretend otherwise.

Why? Because composition is harder, and much more expensive to measure. Regulators measure mass concentration because it is simple, repeatable, and legally defensible. Consultants build systems to meet the regulations. Vendors design sensors to satisfy bureaucratic requirements. And once mass concentration becomes the regulated value, the entire industry optimises for it.

Honesty, openness, and innovation

We need to stop pretending PM2.5 alone is enough.

We need:

A next-generation air-quality monitoring system must evolve beyond simple particle counts to understand what we are actually breathing. This begins with composition-aware monitoring networks, where regular elemental and chemical analyses are the rule rather than the exception.

To complement this, real-time indicators of source type should become commonplace. By leveraging proxies such as black-carbon absorption, UV absorption for wood smoke, or organic-to-elemental carbon ratios, networks can distinguish between pollution from traffic, biomass burning, or secondary formation. Combined with source-apportionment models, these indicators can operate in real time or near-real time, offering actionable intelligence for both policymakers and the public.

The next step involves hybrid modeling using open data. Integrating fire-hotspot maps, satellite aerosol optical depth (AOD), weather patterns, local emission inventories, and periodic speciation results enables powerful machine-learning models that can infer source-specific pollution continuously.

However, none of this progress matters if the data remain locked away. Open publication of composition data —particularly from supersites and research-grade stations—is essential. When communities can see not only how much pollution is present but also what it consists of, the conversation around health, accountability, and mitigation changes substantially.

Finally, these advances demand a rethinking of air-quality standards. PM₂.₅ mass concentration should remain a cornerstone metric, but it must be complemented with toxicity-weighted indicators, source-specific thresholds, and communication that reflects composition-dependent health risks. By aligning standards with scientific understanding, we can move toward a more precise, health-relevant, and ultimately more just approach to air-quality management.

Why it matters to us at AirGradient

Our origins — a volunteer project during the burning season in Northern Thailand — taught us something that the rest of the system often forgets: people live with the consequences of air pollution every single day. They are not abstract numbers in a regulatory report.

That is why we will continue advocating for open data, honest communication, and more scientifically grounded monitoring. Our mission is not to satisfy outdated frameworks; it’s to help communities understand their air and protect their health.

Let’s work together to make air quality data more health focused.

By the way, if you are interested to dive more into the issue of PM2.5 measurements, you can read a series of articles about this topic written by our in-house scientist Siriel Saladin.