Cost-effective chemical speciation monitoring of particulate matter air pollution

October 28, 2022
In Progress
Share this project


Ambient particulate matter is one of the greatest environmental health risks to society, but cost-effective methods for chemical characterization and source apportionment necessary to inform robust regulatory strategies are lacking. Our team has demonstrated the possibility to use infrared spectroscopy as an inexpensive optical probe to obtain chemically informative spectra of particulate matter, and this analytical technique is increasingly being used in air quality monitoring networks and research campaigns worldwide. To obtain reliable predictions at such a prolific scale, advanced data-driven methods and an open platform where the global community can access them are necessary to fully take advantage of the rich but complex infrared spectra acquired from these samples.

CHEMSPEC brings together an unprecedented data set from multiple monitoring networks and laboratory-generated particles to learn the relations between chemical composition and spectroscopic signatures that can be used in quantitative modeling of chemical species concentrations and source profiles associated with fossil fuel combustion, biomass burning, dust, and other major contributors to particulate matter. The results of this project will advance the characterization of particulate
matter not only in existing monitoring networks, but in critical areas of the world where the particulate matter composition is virtually unknown.



SDSC Team:
Francois Kamper
Ekaterina Krymova
Guillaume Obozinski

PI | Partners:

EPFL, Laboratory of Atmospheric Processes and Their Impact:

  • Dr. Satoshi Takahama
  • Prof. Dr. Athanasios Nenes

More info

UC Davis, Air Quality Research Center:

  • Dr. Ann Dillner

More info



We aim to develop infrared spectroscopy as a method to monitor ambient particulate matter in the air as an inexpensive alternative to current approaches.  In this way we aim to improve the monitoring of air quality in established networks, as well as in other critical areas of the world, to aid with the formulation of emissions control policy and improve healthcare in general.

Proposed Approach

We propose removing background interference from field spectra through a prior obtained by fitting a distribution to blank spectra, i.e. spectra free from any ambient particulate matter. Once the background interference has been removed, we aim to learn profiles of chemical species using, for example, non-negative matrix factorization approaches. These profiles can then be used to obtain predictions of the concentrations of ambient particulate matter in the field, generalizable to a variety of locations.


The anticipated impact of this project is the expansion in capabilities of monitoring networks worldwide – especially in areas where particle composition is virtually unknown – to cost - effectively provide chemical composition information required to identify the key sources and constituents contributing to PM2.5 at every location critical for shaping necessary emissions control policies, evaluating air quality action plans, supporting air quality modeling, and remote sensing efforts.

Note. Here PM2.5 is the mass concentration of ambient particulate matter less than 2.5 micrometers in diameter.

Figure 1. Illustration of filter sampling and scanning procedure by IR spectroscopy and example IR spectra. The sampled filter spectrum contains the contribution from the filter media (PTFE membrane) and collected particles. The wavenumber is a unit of frequency the reciprocal of the wavelength of electromagnetic radiation. Absorbance is a unitless quantity calculated from the reciprocal of the ratio of transmitted beam intensity through the sample (particles + filter) and transmitted intensity with no sample present, and is strictly additive when no interactions are present among constituents.



Additional resources


  1. Solomon, P. A. et al. U.S. National PM2.5 Chemical Speciation Monitoring Networks—CSN and
    IMPROVE: Description of networks. Journal of the Air & Waste Management Association 64,
    1410–1438. eprint:
    1080/10962247.2014.956904 (2014).
  2. Reggente, M., Dillner, A. M. & Takahama, S. Analysis of functional groups in atmospheric aerosols
    by infrared spectroscopy: systematic intercomparison of calibration methods for US measurement
    network samples. Atmospheric Measurement Techniques 12, 2287–2312. https : / / www . atmos - (2019).
  3. Bürki, C. et al. Analysis of functional groups in atmospheric aerosols by infrared spectroscopy:
    method development for probabilistic modeling of organic carbon and organic matter concentrations.
    Atmospheric Measurement Techniques 13, 1517–1538. AMT – Volume 13, 2020 1517/2020/ (2020).
  4. Boris, A. J. et al. Quantifying organic matter and functional groups in particulate matter filter
    samples from the southeastern United States – Part 2: Spatiotemporal trends. English. Atmospheric
    Measurement Techniques 14. Publisher: Copernicus GmbH, 4355–4374. issn: 1867-1381. https :
    // (2021) (June 2021).


Related Pages

More projects


In Progress
Interpretable and Robust Machine Learning for Mobility Analysis
No items found.


In Progress
Feature Learning for Bayesian Inference
No items found.


In Progress
Personalized epidural electrical stimulation of the lumbar spinal cord for clinically applicable therapy to restore mobility after paralyzing spinal cord injury
No items found.


In Progress
Directed Imitation During Vocal Learning
No items found.


Latest news

Smartair | An active learning algorithm for real-time acquisition and regression of flow field data
May 1, 2024

Smartair | An active learning algorithm for real-time acquisition and regression of flow field data

Smartair | An active learning algorithm for real-time acquisition and regression of flow field data

We’ve developed a smart solution for wind tunnel testing that learns as it works, providing accurate results faster. It provides an accurate mean flow field and turbulence field reconstruction while shortening the sampling time.
The Promise of AI in Pharmaceutical Manufacturing
April 22, 2024

The Promise of AI in Pharmaceutical Manufacturing

The Promise of AI in Pharmaceutical Manufacturing

Innovation in pharmaceutical manufacturing raises key questions: How will AI change our operations? What does this mean for the skills of our workforce? How will it reshape our collaborative efforts? And crucially, how can we fully leverage these changes?
Efficient and scalable graph generation through iterative local expansion
March 20, 2024

Efficient and scalable graph generation through iterative local expansion

Efficient and scalable graph generation through iterative local expansion

Have you ever considered the complexity of generating large-scale, intricate graphs akin to those that represent the vast relational structures of our world? Our research introduces a pioneering approach to graph generation that tackles the scalability and complexity of creating such expansive, real-world graphs.

Contact us

Let’s talk Data Science

Do you need our services or expertise?
Contact us for your next Data Science project!