Source apportionment analysis of atmospheric particulate matter is a systematic technical approach that quantitatively identifies which emission sources—such as vehicle exhaust, coal combustion, industrial processes, dust, and biomass burning—contribute to ambient PM concentrations. By combining rigorous field sampling, multi-component chemical speciation, and validated receptor modeling techniques, this analysis provides the quantitative evidence base needed for targeted air pollution control strategies. This article explains the complete technical workflow, from sample collection and testing parameters to modeling methods and practical applications.
Technical Overview
Source apportionment analysis is the quantitative determination of the origins of atmospheric particulate matter (PM), including PM2.5, PM10, and total suspended particles (TSP). Rather than simply reporting ambient mass concentrations, the analysis resolves the measured particulate mass into distinct source categories—such as secondary sulfate and nitrate, motor vehicle exhaust, coal combustion, industrial emissions, soil and road dust, construction dust, biomass burning, and cooking—and estimates the percentage contribution of each category to the observed concentration.
The technical workflow integrates three pillars: representative filter-based sampling at receptor sites, comprehensive chemical speciation of the collected particulate mass, and statistical receptor modeling that mathematically decomposes the ambient chemical composition into source contributions. The approach follows internationally recognized methodologies, including the US EPA Compendium Methods (IO-2.1, IO-3 series, IO-4.2), EPA Positive Matrix Factorization (PMF) and Chemical Mass Balance (CMB) guidance, thermal-optical carbon protocols (IMPROVE_A, NIOSH 5040), and national standards such as GB 3095 for ambient air quality, HJ 618 for gravimetric PM determination, and HJ 777 for metal element analysis, together with the national technical guideline for particulate matter source apportionment.
- Quantified mass and percentage contribution of each emission source category
- Seasonal and episode-based variation of source contributions
- Assessment of primary emissions versus secondary aerosol formation
- Distinction between local sources and regional transport influences
- Traceable, quality-assured data suitable for regulatory decision-making
Testing Samples
Ambient Particulate Matter Samples
Field campaigns collect integrated PM2.5, PM10, and TSP samples onto quartz fiber filters (for carbon and organic analysis) and Teflon membrane filters (for mass, elements, and ions). High-volume samplers operating at approximately 1.1 cubic meters per minute, or equivalent medium-volume samplers, are deployed over 22 to 24 hour integration periods. A scientifically robust study typically covers all four seasons, with 15 to 20 valid samples per season, and includes multiple representative sites such as urban background, traffic-influenced, industrial, residential, and suburban locations. Size-segregated sampling using cascade impactors may be added when particle-size-resolved source information is required.
Source Profile Samples
Receptor modeling requires characteristic chemical fingerprints of candidate sources, known as source profiles. These are generated from representative emission samples, including resuspended soil dust, road dust, and construction dust; coal combustion fly ash from boilers and power plants; vehicle exhaust from dynamometer testing across fuel types and vehicle classes; industrial process emissions from metallurgy, cement, and petrochemical operations; biomass burning smoke from crop residue and firewood; cooking emissions; and regionally relevant materials such as sea salt. Each profile is analyzed for the same chemical species as the ambient samples to ensure model compatibility.
Episode and Special-Purpose Samples
During heavy pollution episodes, short-interval samples or online continuous monitoring data with hourly time resolution for elements and water-soluble ions can be collected to capture rapidly evolving source dynamics. Additional special samples may include single-particle specimens prepared for electron microscopy to support morphological source identification, and isotope samples for distinguishing fossil versus contemporary carbon.
Testing Items
A full speciation suite is essential for reliable source resolution. The core testing program comprises the following component groups:
- Mass concentration: gravimetric determination of PM2.5, PM10, and TSP on conditioned filters
- Elemental composition: approximately 20 to 25 elements, including Na, Mg, Al, Si, K, Ca, Ti, V, Cr, Mn, Fe, Ni, Cu, Zn, As, Se, Cd, Sb, Ba, and Pb
- Water-soluble inorganic ions: sulfate, nitrate, ammonium, chloride, fluoride, sodium, potassium, magnesium, and calcium
- Carbonaceous components: organic carbon (OC), elemental carbon (EC), total carbon, and the individual thermal-optical carbon fractions
- Organic molecular tracers: levoglucosan, galactosan, and mannosan for biomass burning; hopanes and steranes for lubricating oil and vehicle exhaust; polycyclic aromatic hydrocarbons (PAHs) for combustion sources; cholesterol for cooking emissions; and sugar alcohols for biological sources
- Supplementary parameters: water-soluble organic carbon, mineral dust markers, and stable isotope ratios such as lead isotopes and radiocarbon where advanced source discrimination is required
Testing Methods
Sampling and Gravimetric Analysis
Filters are pre- and post-conditioned in a controlled environment of constant temperature and humidity, and weighed on a microbalance with microgram-level readability under anti-static conditions, in accordance with gravimetric procedures consistent with EPA Method IO-2.1 and national gravimetric standards. Field blanks, trip blanks, and duplicate samples accompany every campaign to quantify contamination and precision.
Elemental Analysis
Elements are determined by non-destructive X-ray fluorescence (XRF) spectrometry following EPA Method IO-3.3, or by inductively coupled plasma optical emission spectrometry (ICP-OES) and ICP mass spectrometry (ICP-MS) after microwave-assisted acid digestion in line with EPA Method IO-3.1 and HJ 777. Multi-element certified reference materials and continuing calibration verification ensure traceability of every analytical batch.
Water-Soluble Ion Analysis
Ions are extracted from filter portions by ultrasonic-assisted water extraction and quantified by suppressed ion chromatography, consistent with EPA Method IO-4.2, with separate anion and cation channels. Anion–cation balance checks are applied to each dataset to validate analytical integrity.
Carbonaceous Component Analysis
OC and EC are measured by thermal-optical analysis using the IMPROVE_A or NIOSH 5040 temperature protocols, which resolve the carbon profile into multiple organic and elemental carbon fractions. The split between OC and EC is optically corrected, and minimum OC/EC ratio screening is used to evaluate secondary organic carbon formation.
Organic Tracer Analysis
Molecular markers are determined by gas chromatography–mass spectrometry (GC-MS) in selected ion monitoring mode following solvent extraction and derivatization, providing source-specific fingerprints that strongly constrain receptor model factor identification.
Receptor Modeling and Data Analysis
Source contributions are quantified using EPA-recommended receptor models. Positive Matrix Factorization (EPA PMF) decomposes the speciation matrix into factors using uncertainty-weighted, non-negative least squares, with solution stability verified through bootstrap resampling, displacement analysis, and multiple random seeds. Chemical Mass Balance (EPA CMB) is applied when reliable local source profiles are available, fitting ambient concentrations as linear combinations of source fingerprints. UNMIX and principal component analysis with absolute principal component scores serve as complementary or verification tools. Species are categorized as strong, weak, or excluded based on signal-to-noise ratios and model sensitivity.
Quality Assurance and Model Validation
A comprehensive quality system includes laboratory and field blanks, duplicate and split-sample analysis, certified reference materials, ion balance verification, and chemical mass closure reconstruction, with recovered mass typically required to fall within 80 to 120 percent of the gravimetric mass. Model outputs are validated by examining scaled residuals, factor profile interpretability, comparison of independent models against each other, and consistency with local emission inventory knowledge and meteorological trajectory analysis.
Applications
Source apportionment results directly support science-based air quality management across a wide range of scenarios:
- Development of annual and seasonal PM2.5 control plans for cities and industrial regions, prioritizing the sources with the largest quantified contributions
- Diagnosis of heavy pollution episodes, distinguishing adverse meteorological accumulation from sudden emission increases
- Evaluation of the effectiveness of emission reduction programs through before-and-after contribution comparison
- Environmental impact assessment and planning support for compliance with ambient air quality standards such as GB 3095
- Investigation of dust and odor complaints near industrial parks using tracer-based source discrimination
- Academic and regulatory research on secondary aerosol formation, regional transport, and chemical aging of particulate pollution
FAQ
What is the difference between receptor models and dispersion models?
Receptor models work backward from measured ambient chemical composition at the sampling site to infer source contributions, requiring no emission inventory or meteorological data as primary input. Dispersion and chemical transport models work forward, simulating how emissions disperse and transform in the atmosphere. The two approaches are complementary, and agreement between them substantially increases confidence in source conclusions.
How many samples are required for a valid analysis?
For robust multivariate receptor modeling such as PMF, a dataset of at least 60 to 100 valid samples is generally recommended, with each sample analyzed for 20 or more chemical species. Seasonal campaigns typically target 15 to 20 valid samples per season at each site, and more samples improve factor stability and reduce uncertainty ranges.
How long does a complete source apportionment study take?
A standard four-season campaign generally spans 12 months of field sampling, followed by laboratory analysis and modeling, which typically require an additional one to three months. Accelerated studies using only one or two intensive observation periods can be completed faster, but their conclusions are limited to the sampled conditions.
Can the analysis separate local emissions from regional transport?
Yes. By combining a multi-site sampling network with back-trajectory analysis and conditional probability functions, secondary and regionally transported components can be distinguished from locally emitted primary sources, allowing authorities to differentiate locally controllable contributions from externally transported pollution.
What level of uncertainty should be expected?
Bootstrap-based uncertainty intervals for individual source contributions commonly fall in the range of 10 to 30 percent relative deviation, depending on data quality, number of samples, and source profile similarity. Cross-validation between PMF, CMB, and other models provides an additional measure of result reliability, and reputable studies always report these uncertainty ranges alongside the contribution estimates.