Identification of sources of atmospheric PM at the Pittsburgh Supersite, Part I: Single particle analysis and filter-based positive matrix factorization

Natalie J. Pekney, Cliff I. Davidson, Keith J. Bein, Anthony S. Wexler, Murray V. Johnston

Research output: Contribution to journalArticlepeer-review

41 Scopus citations


During the Pittsburgh Air Quality Study (PAQS), July 2001-September 2002, three co-located instruments analyzed the composition of ambient particulate matter (PM): (1) A single particle mass spectrometer, RSMS-3, was deployed to obtain high-temporal-resolution measurements of single particle size (>1.1 μm) and composition which were correlated with meteorological data to identify sources; (2) PM2.5 and PM10 were collected on cellulose filters using high-volume (hi-vol) samplers, followed by microwave-assisted digestion and analysis by inductively coupled plasma-mass spectrometry (ICP-MS). Positive matrix factorization (PMF) was used to identify possible source categories; and (3) a micro-orifice uniform-deposit impactor (MOUDI) obtained size-distributed samples of PM. Several days of MOUDI filters were selected for microwave-assisted digestion and analysis by ICP-MS. In this paper, sources identified using the single particle data were compared to the PMF results for the hi-vol/ICP-MS data. The strengths of each method were combined to hypothesize the most likely sources of various elements in ambient PM in Pittsburgh. In the final results, Mo and Cr are attributed to local specialty steel facilities; Fe, Mn, Zn, and K are attributed to a steel mill SE of the monitoring station; internally mixed Pb-containing particles are attributed to a major source to the NW; and Ga is attributed to coal combustion sources to the NW. There is a notable lack of oil combustion sources. The MOUDI data were used to resolve discrepancies between the single particle and hi-vol/ICP-MS data concerning the detection of Ti and Se. The hi-vol data showed appreciable Ti and Se masses, but RSMS-3 was unable to detect significant numbers of Ti-containing particles because of their large size, while we hypothesize that the volatility of Se caused it to be distributed more evenly over all emitted particles such that the amount of Se in any individual particle is below the limit of detection.

Original languageEnglish (US)
Pages (from-to)411-423
Number of pages13
JournalAtmospheric Environment
Issue numberSUPPL. 2
StatePublished - 2006
Externally publishedYes


  • Ambient aerosols
  • Particle size-composition
  • Positive matrix factorization
  • Real-time single particle mass spectrometry
  • Trace elements

ASJC Scopus subject areas

  • General Environmental Science
  • Atmospheric Science


Dive into the research topics of 'Identification of sources of atmospheric PM at the Pittsburgh Supersite, Part I: Single particle analysis and filter-based positive matrix factorization'. Together they form a unique fingerprint.

Cite this