Using the complexity of secondary organic aerosols to understand their formation, ageing and transformation in three contrasting megacities

Exposure to poor air quality is the top environmental risk factor of premature mortality globally. By far the most damaging air pollutant to health is particulate matter, with the greatest effects associated with particles less than 2.5 microns in diameter (PM2.5). In megacities, with large numbers of inhabitants/emissions sources, PM2.5 can often exceed recommended guideline values. The World Health Organization recommend an annual mean concentration of less than 10 micrograms/m3, as current evidence suggests lower health risks below this value. However, over 90 % of the worlds population live in regions where this value is exceeded, with London, Beijing and Delhi having values in 2016 ~ 1.5, 8 and 15 times higher. Secondary organic aerosol (SOA) can make up a significant fraction of PM2.5 in urban areas, and this may increase as many counties act to reduce emissions of ammonia and NOx. Current analytical approaches fail to provide sufficient chemical speciation to routinely apportion the contributing sources of SOA, limiting the opportunities to develop more targeted PM abatement strategies. High complexity approaches have revolutionised biomedicine, however uptake within the environmental community has been slower. In this project, we will embrace the atmosphere's complexity to make a step change in our understanding of the sources and transformation of SOA in urban atmospheres. This will be achieved through the combination of two state of the art research areas; high resolution mass spectrometry (MS) and machine learning. We will develop new tools to allow high throughput screening and quantification of SOA tracers in atmospheric aerosol samples. We will develop a mass spectral database of SOA tracer species built using a novel aerosol flow reactor designed at the University of York and supplemented with samples from 6 world-leading simulation chambers. The key here is to identify unique source specific tracer molecules that allow a direct link between the gas phase organic molecule that is emitted to the atmosphere and it's specific oxidation products that can be measured in ambient particles. The MS uses electrospray ionization, one of the most common approaches used in analytical labs throughout the world. This method is ideally suited to many SOA tracer molecules, however the ionization efficiency is strongly dependent on the chemical structure. We will carry out a systematic evaluation of the ionization efficiencies of a wide range of molecules with different functionalities to build a regression model to predict instrument response as a function of a molecular "fingerprint". We will combine these tools to carry out the most comprehensive quantification of SOA tracers in ambient aerosol and use machine learning methods to determine the factors that impact SOA concentration and estimate the relative strength of biogenic and anthropogenic sources of SOA to PM2.5. Our project will provide the first demonstration of such methods; the lack of sufficient chemical speciation and low time resolution in previous studies has so far restricted our proposed analysis. The timing of this project allows us to exploit recent investment in the NERC Air Pollution and Human Health program, providing access to an archive of PM2.5 samples and a wealth of co-located air quality data collected by leading groups from the UK, China and India. To communicate our results we will produce city specific policy reports, highlighting the main conclusions for each city, for use by government and regulatory agencies. This will be aided by a two month knowledge transfer placement in the Air Quality policy group at the Department for Environment, Food and Rural Affairs in London. This project will provide evidence of the key factors that control the amount of SOA in cities, using London, Beijing and Dehli as test cases. However, the methodology could be applied in cities across the globe to develop abatement policies that would target SOA reduction.

Air pollution is one of the top environmental risk factors globally, with the World Health Organisation (WHO) estimating 3 million premature deaths per year as a result of associated strokes, heart disease, lung cancer and chronic respiratory diseases. Nine out of 10 people worldwide breathe polluted air, where air quality levels exceed WHO guidelines. The largest socioeconomic and health impacts affect developing regions such as SE Asia. However even in developed countries such as the UK, air pollution costs the economy around £16 billion per annum and reduces the average person's life expectancy by ~6 months. As an example of impacts and benefits, a policy which aimed to reduce the annual average concentration of PM2.5 by only 1 microgram/m3 would result in a saving of approximately 4 million life years for people born in 2008 in the UK.

Many countries have introduced policies to reduce emissions of primary particulate matter and the precursors for secondary inorganic aerosol, such as ammonia, nitrogen oxides and sulphur dioxide, however in many places ambient levels have not reduced as anticipated. A study of PM composition in 15 cities across the globe indicated that on average 25 % was made up of oxidised organic aerosol. Current analytical approaches fail to provide sufficient chemical speciation to routinely apportion the contributing sources of SOA, limiting the opportunities to develop more targeted PM abatement strategies.

This proposal will exploit a novel high-resolution analytical methodology and integrate it with modern machine learning approaches to make a step change in our understanding of SOA in the atmosphere. To test the approach we have chosen to study three megacities (London, Beijing and Delhi) as they suffer frequent poor air quality and have a different mix of sources, emission controls and meteorology. In order to communicate our results with regional and national governments, we will produce city specific policy reports identifying the major sources of SOA and the key factors that control their concentrations and transformation in the urban polluted atmosphere. The researcher will carry out a two-month knowledge exchange placement at the Department for Environment, Food and Rural Affairs within the Air Quality group to produce the initial evidence statement related to SOA in London. Being embedded at DEFRA, will provide valuable insight into the policy making process, particularly how research is used to inform decision-making. This placement will also achieve impact by improving engagement between government and academia and allow the researcher to build a network of contacts, while working in an interdisciplinary environment. The knowledge gained will be used to produce policy reports for Beijing and Delhi and we will exploit our on-going collaborations developed as part of the NERC funded Air Pollution and Human Health program with Chinese and Indian scientists and their links with policy makers to disseminate the report widely.

The tools developed in this project would be applicable to other areas of atmospheric and environmental science and beyond. The software tools developed will be released under the GPU general public license along with training and test data to ensure transferability to other areas. The regression model developed could transform calibration methods for electrospray ionization mass spectrometry, which represents a significant fraction of the global mass spectrometry market ($~7 billion in 2017). We have on-going collaborations with a range of mass spectrometry manufacturers and will exploit these links should this methodology prove successful. We will ensure academic impact through peer reviewed publications and conference attendance. We will deposit our database of SOA tracer concentrations in CEDA, with links to the APHH projects, to allow subsequent follow on epidemiological or modelling studies to use this novel resource.