A semi-quantitative decision matrix for Polycyclic aromatic hydrocarbons source apportionment to support sustainable urban transport and air quality management

Polycyclic aromatic hydrocarbons (PAHs) in urban particulate matter originate from multiple sources, including vehicular emissions, biomass combustion, and petrogenic inputs. Their source identification is essential for designing effective air-quality and sustainable transport strategies; however, conventional diagnostic-ratio approaches are often limited by overlapping source signatures and the uncertainty associated with interpreting individual molecular indicators. This highlights the need for standardized frameworks capable of integrating multiple molecular indicators into a unified source-oriented scheme. This study proposes a semi-quantitative weighted diagnostic ratio matrix that integrates multiple particle-bound PAH ratios and complements multivariate statistical analysis. PM 2.5 and PM 10 were monitored over one year at three sites representing different traffic intensities in an intermediate-sized Latin American city. PAH concentrations were determined and evaluated using Principal Component Analysis (PCA), molecular diagnostic ratios, and the proposed weighted matrix. The matrix integrates diagnostic evidence from multiple PAH ratios into a standardized semi-quantitative scoring scheme for three source categories: biomass combustion, liquid fossil fuel combustion, and petrogenic. PCA identified a dominant combustion-related pattern in both particulate fractions, characterized by high-molecular-weight PAHs and accounting for more than 76% of the total variance in PM 2.5 and 82% in PM 10. The second components were primarily associated with lower-molecular-weight PAHs. The weighted matrix identified biomass combustion as the predominant source across the monitoring sites, followed by liquid fossil fuel combustion, while petrogenic contributions were comparatively minor. Biomass combustion accounted for 54.3%–73.9% of the overall source attribution, indicating that urban PAH levels were influenced not only by traffic-related emissions but also by a broader combustion background potentially associated with residential heating and biomass-burning activities. Agreement between the matrix-based source profiles and PCA patterns provided a consistency check of the source interpretation, while differences among individual diagnostic ratios highlighted the limitations of relying on isolated indicators. The proposed framework provides a transparent, reproducible, and computationally accessible approach for integrating multiple diagnostic ratios into a unified source-oriented assessment. As a screening-level decision-support tool, it may support urban air-quality assessment, emission-source management, and transport-related environmental planning.

Polycyclic aromatic hydrocarbons (PAHs) are ubiquitous organic compounds consisting of two or more fused benzene rings, characterized by high thermal stability, low water solubility, and lipophilic nature. These properties allow them to persist in the environment and bioaccumulate across food webs (Jakovljević et al., 2025; Rajeev et al., 2023). Atmospheric PAHs are present both in the gas phase and bound to particulate matter (PM), whose partitioning is governed primarily by their molecular weight and structure. In this sense, PAHs with 2–3 rings predominate in the gas phase, 4-ring PAHs partition between both phases, and PAHs with five rings or more are predominantly particulate-bound (Giuliani et al., 2021).

Due to their toxicological relevance, the United States Environmental Protection Agency (US EPA) designated 16 PAHs as priority environmental pollutants (Ishtiaq et al., 2021; Safo-Adu et al., 2023). Human exposure, primarily via inhalation in atmospheric contexts, poses severe acute and chronic health risks. Acute outcomes include ocular and respiratory irritation, nausea, and heightened susceptibility to acute respiratory infections, particularly in children (Omokungbe et al., 2026; Safo-Adu et al., 2023). Prolonged exposure causes profound systemic damage, including asthma exacerbation, chronic obstructive pulmonary disease (COPD), decreased lung immunity, and cardiovascular diseases such as hypertension and infarction (Chen and Jia, 2025). Furthermore, high PAH levels are correlated with reproductive toxicity, neurodevelopmental deficits in children, and congenital malformations, with children, pregnant women, and the elderly exhibiting heightened vulnerability. Crucially, compounds like benzo [a]pyrene (BaP) are classified as Group 1 human carcinogens by the International Agency for Research on Cancer (IARC), linked directly to lung, skin, and bladder cancers (Ishtiaq et al., 2021; Jakovljević et al., 2025).

In urban environments, atmospheric PAHs originate from two fundamental pathways: petrogenic and pyrogenic processes. Petrogenic sources involve uncombusted crude oil and refined petroleum products released into the atmosphere via fuel evaporation, storage leaks, industrial solvent usage, and wear of asphalt pavements or lubricating oils (Rajeev et al., 2023). Pyrogenic sources, which dominate urban atmospheric burdens, arise from the incomplete combustion or thermal cracking (pyrolysis) of organic matter under oxygen-deficient conditions at high temperatures. Natural pyrogenic contributions include wildfires and volcanic activity (Elzein et al., 2020; Soleimani et al., 2024); however, anthropogenic emissions are one of the main factors influencing urban air quality. Key anthropogenic sectors encompass stationary industrial activities (thermal power plants, steel manufacturing, and petroleum refining), residential heating and cooking (primarily biomass, wood, and coal burning), and vehicular traffic.

Vehicular transport represents one of the most critical and dynamic contributors to urban PAH pollution, significantly worsening urban air quality and contributing to severe adverse health outcomes and increased mortality (Mak and Ng, 2021; Shetty et al., 2023). This is particularly evident across Latin America, where mobile sources continue to exert a major influence on urban air quality (Guerra et al., 2023; Naveyra, 2023). Unlike high-income regions experiencing rapid relative declines in traffic emissions due to accelerated fleet electrification and strict standards, medium-sized cities in Argentina remain characterized by aging passenger fleets, high shares of diesel-powered light and heavy-duty vehicles, and delayed turnover of catalytic technologies (Puliafito et al., 2015; Sosa et al., 2026). The signature of mobile source emissions is strongly dependent on engine specifications, fuel formulation, operating conditions, and fleet characteristics. Gasoline-powered vehicles predominantly emit low molecular weight (LMW, 2–3 rings) PAHs (such as phenanthrene and fluoranthene) due to higher operating temperatures and fuel chemistry. In contrast, heavy-duty and light-duty diesel engines emit substantially higher quantities of high molecular weight (HMW, ≥4 rings) PAHs (including chrysene, benzo [b]fluoranthene, and benzo [a]pyrene) bound to fine soot particles, driven by heterogeneous compression-ignition processes (Lara et al., 2022; Mellado et al., 2022; Tobiszewski and Namieśnik, 2012; Zhang et al., 2008). Furthermore, urban driving conditions, characterized by frequent acceleration-deceleration cycles, traffic congestion, idling, and high proportions of aging vehicle fleets lacking modern catalytic converters or particulate filters, significantly elevate specific PAH emission factors. Identifying and quantifying the specific contributions of these diverse vehicular and non-vehicular sources is essential for designing targeted urban mitigation strategies, promoting sustainable urban transport activities, and fostering environmental resilience (Papadakis et al., 2024).

To identify emission profiles, diagnostic concentration ratios between specific PAH are widely used (Bi et al., 2020; Liu et al., 2017; Rajeev et al., 2023). Assuming that relative source signatures remain stable during short-range transport, these indices help distinguish gasoline, diesel, biomass, coal, and petrogenic inputs, an approach similarly applied to volatile organic compounds (VOCs) for air quality management (Colman Lerner et al., 2014; Giuliani et al., 2021). However, diagnostic ratios are indicative rather than absolute; variations in fuel composition, engine technology, catalytic converter usage, air-to-fuel ratios, atmospheric reactivity, and photochemical degradation introduce overlapping boundaries (Fang et al., 2020; Giuliani et al., 2021). Despite their widespread use, diagnostic ratios may yield conflicting source assignments when interpreted individually. Consequently, ratio analysis provides qualitative screening that requires integration with quantitative receptor models, such as Positive Matrix Factorization (PMF), Chemical Mass Balance (CMB), Principal Component Analysis (PCA), or Factor Analysis/Multiple Linear Regression (FA/MLR), to evaluate source contributions accurately (Hopke, 2016; Mellado, 2020; Mellado et al., 2022). Nevertheless, standardized frameworks that integrate multiple ratios into a single source-attribution approach remain limited, particularly for medium-sized cities requiring practical air-quality management tools.

While sophisticated receptor models are fundamental for public health strategies, urban policymakers and transport stakeholders often lack the computational resources, updated local emission inventories, or real-time expertise needed to run complex multivariate models continuously. This creates a critical operational gap between environmental monitoring and rapid regulatory decision-making.

To address this challenge, this study presents the development of an accessible and structured Decision-Support Matrix (DSM) based on multiple diagnostic ratios of particle-bound PAHs and complemented by Principal Component Analysis (PCA). The framework integrates molecular diagnostic evidence into a standardized semi-quantitative scheme to facilitate the identification and comparison of predominant emission sources contributing to urban particulate matter (PM 2.5 and PM 10). Rather than providing a comprehensive apportionment of PAH sources or absolute source mass contributions, the proposed approach uses particle-bound PAHs as molecular tracers to identify the predominant PM emission patterns that may warrant further investigation and management. The practical applicability of the approach was assessed using PAH data collected in Tandil, Argentina, a medium-sized urban center experiencing increasing transport activity and potentially mixed combustion sources. By translating multiple diagnostic indicators into a transparent source-oriented framework, the DSM provides a screening-level decision-support tool that can complement multivariate analysis and assist in prioritizing emission sources for further investigation. Its application may therefore support local air-quality assessment, resource prioritization, and the design of targeted emission-management strategies within resilient and sustainable urban environments.

Study area and monitoring network design

The study area is Tandil, a rapidly expanding medium-sized city located in the province of Buenos Aires, Argentina. This city serves as a case study for intermediate urban centers with emerging economies that are experiencing rapid motorization and spatial growth. With a population of 145,575 inhabitants, the city has exhibited a demographic expansion above the provincial average (Instituto Nacional de Estadística y Censos, 2023). Simultaneously, Tandil has consolidated its position as a major regional tourist destination. Post-pandemic recovery indicators show sustained tourism activity with pronounced seasonal peaks (January and July) that exceed pre-2019 baselines (Municipio de Tandil, s f.).

Regarding residential heating, natural gas constitutes the main heating fuel in Tandil; however, biomass combustion persists during cold periods. In Argentine cities situated near hills, mountain and forest environments, such as Tandil, the widespread use of wood-burning stoves and open fireplaces for supplementary space heating constitutes a non-negligible source of seasonal PAH emissions. This practice is expected to be more prevalent in peripheral and predominantly residential areas than in the city center, where commercial activities and vehicular traffic are more concentrated. The combined influence of these emission sources, together with the continuous fleet expansion, daily commuter movements, and seasonal tourist influxes exert increasing pressure on the urban atmospheric environment, underscoring the urgent need for agile decision-support tools.

To capture this spatial and operational heterogeneity in the distribution of emission sources across the city, a strategic monitoring network was established comprising three representative sampling sites characterized by contrasting vehicular traffic densities and flow dynamics: high- (HT), medium- (MT), and low-traffic (LT) areas (Sosa et al., 2026), as shown in Figure 1. According to Sosa et al. (2026), high-intensity segments are associated with main avenues and commercial areas, intermediate-intensity segments comprise mixed and transition areas, whereas low-intensity segments are predominantly located in residential areas. Consistent with this classification, it can be observed that the high-traffic site is located in the city center, with frequent circulation of passenger vehicles, public buses, and delivery trucks; the medium-traffic site is dominated by passenger vehicles, with occasional public buses and truck circulation; and the low-traffic site is located in a predominantly residential area with lower passenger-vehicle circulation.

This multi-tiered sampling strategy was designed to evaluate the gradient of mobile-source contributions, providing an empirical basis for applying the diagnostic-ratio decision-support matrix for urban transport policy and environmental management.

Air sampling and gravimetric determination

Particulate matter (PM) sampling was conducted from June 2025 to May 2026 across the three designated monitoring locations. Samples were collected using discrete low-volume air samplers (TDA-Airplus) operating continuously for a 24-h sampling period at a regulated constant flow rate of 16.7 L/min. Sampling campaigns for PM 2.5 and PM 10 were conducted non-simultaneously across the monitoring sites; therefore, calculations of the PM 10 /PM 2.5 ratios were excluded from this study. The sampling and chemical analysis were designed to characterize PM 2.5 and PM 10 and to use particle-bound PAHs as molecular tracers of the predominant emission sources contributing to atmospheric particulate matter. Suspended PM was collected on pre-conditioned 47 mm Whatman glass fiber filters. Following collection, filters were handled and transported, adhering to established protocol (Giuliani et al., 2021). The atmospheric PM mass concentration was quantified gravimetrically using a high-precision microbalance. Prior to sampling, the filters were stored in a container with silica gel for 48 h before being weighed for the first time. Then, at the end of each sampling, the filters were stored in the container with silica gel for another 48 h and finally weighed. This procedure ensures that the filters are free of any trace of humidity.

Extraction and GC/MS analytical determination of PAHs

Target analytes comprised the 18 priority PAHs designated by the US EPA: naphthalene (Nap), acenaphthene (Ace), acenaphthylene (Acy), fluorene (Flu), phenanthrene (Phe), anthracene (Ant), fluoranthene (Fl), pyrene (Pyr), benzo [a]anthracene (BaA), chrysene (Chry), benzo [b]fluoranthene (BbF), benzo [a]pyrene (BaP), benzo [e]pyrene (BeP), indeno [1,2,3-c,d]pyrene (IP), dibenzo [a,h]anthracene (DBahA), benzo [g,h,i]perylene (BghiP), benzo [k]fluoranthene (BkF) andbenzo [j]fluoranthene (BjF), the sum of the last two reported as BkF + BjF. Analytical quantification was conducted using a certified reference standard solution containing the 18 PAHs supplied by AccuStandard, USA (EN 17132 PAHs in Textiles Reference Standard).

Filter-bound PAHs were extracted via solvent extraction following validated methodologies (Colman Lerner, 2013; Giuliani et al., 2021), and extracts were stored under cold conditions until analysis. Recovery factors (R) and standard deviation (σ) for each PAH are presented in Supplementary Table S1 as Supplementary Material. Extract analysis was performed on a gas chromatograph coupled with a mass spectrometer (GC-2030 Nexis-GCMS-QP2020NX, Shimadzu) fitted with an SH-5MS capillary column (30 m length × 0.25 mm internal diameter, 0.25 μm film thickness; Shimadzu). Chromatographic separation was operated in electron ionization mode using selective ion monitoring. Operational temperatures were set at 300 °C for the injector, 300 °C for the transfer line, and 220 °C for the ion source. The GC oven temperature program initiated at 70 °C for 2 min, ramped to 200 °C at 15 °C min -1 (held for 1 min), increased to 280 °C at 5 °C min -1 (held for 1 min), and reached a final temperature of 310 °C at 10 °C min -1 (held for 2 min). Sample extracts (1.0 μL) were injected in splitless mode using high-purity helium as the carrier gas at a constant flow rate of 1.5 mL min -1.

The analytical scope was restricted to PAHs associated with the particulate phase collected on the sampling filters. In the proposed framework, particle-bound PAHs were considered molecular tracers of PM emission sources. This approach was adopted to maintain a practical and accessible methodology focused on identifying predominant PM emission-source patterns and supporting source-oriented management decisions.

Quality assurance and quality control (QA/QC)

Quality assurance was maintained throughout the analytical procedure. Calibration curves for individual PAH congeners showed linearity, with correlation coefficients (R 2) exceeding 0.994. Limits of detection (LOD) and quantification (LOQ) were determined following Cantwell (2025), with ranges of 0.30–1.19 ng L -1 and 1.00–3.98 ng L -1, respectively.

Principal component analysis (PCA)

PCA was applied to identify the major patterns of variation in PAH composition and to explore potential emission sources in PM 2.5 and PM 10 samples. Separate PCA models were developed for each particulate fraction using standardized PAH concentrations (z-scores) to discriminate the influence of differences in concentration magnitude among compounds. Only PAHs detected in more than 20% of the samples within each particulate fraction were included in the analysis in order to minimize the influence of variables with excessive missing values and improve the robustness of the multivariate model.

PCA was performed using the correlation matrix, and principal components (PCs) with eigenvalues greater than one were retained (Kaiser, 1960). Component loadings, squared cosine values (cos 2), variable contributions, and sample scores were calculated to characterize the relationship between PAHs and the retained components. Component loadings were used to identify the compounds contributing most strongly to each principal component, whereas sample scores were used to evaluate similarities among individual samples and were subsequently correlated with molecular diagnostic ratios to assess the consistency between both source-apportionment approaches.

Development of a DSM for PAH source apportionment

To integrate the results of individual PAH diagnostic ratios, a semi-quantitative decision-support matrix was developed. Unlike conventional diagnostic-ratio approaches, where individual diagnostic indices are interpreted separately against source-specific threshold values (e.g., Liu et al., 2017; Arias et al., 2022), the proposed methodology combines evidence from multiple diagnostic indicators into a single source-apportionment scheme. This system utilizes a discrete scoring mechanism designed to account for overlapping diagnostic boundaries among source categories.

Ratio selection and threshold harmonization

Five diagnostic ratios widely established in atmospheric source-apportionment literature were selected: Fl/(Fl + Py), IP/(IP + BghiP), Ant/(Ant + Phe), BaA/(BaA + Chry), and BaP/BghiP. Together, these indicators distinguish among petrogenic, liquid fossil fuel combustion, and biomass or coal burning sources.

Classification thresholds were defined based on published literature values, adopting established boundary limits and setting standardized criteria where reporting discrepancies exist among authors. For example, reported consensus establishes that Ant/(Ant + Phe) values below 0.10 denote petrogenic sources, whereas values above 0.10 correspond to pyrogenic processes (Pies et al., 2008; Yunker et al., 2002). Similarly, Fl/(Fl + Py) categorizes petroleum inputs (<0.40), liquid fossil fuel combustion (0.40–0.50), and biomass or coal burning (>0.50) (Yunker et al., 2002).

For ratios exhibiting operational overlap across studies, specific boundary conditions were applied. For IP/(IP + BghiP), values below 0.20 indicate petrogenic inputs, while pyrogenic threshold assignments vary in the literature; several authors assign liquid fossil fuel combustion to 0.20–0.40 (Akyüz and Çabuk, 2010; Dvorská et al., 2011; Safo-Adu et al., 2023), whereas Yunker et al. (2002) considered the broader interval of 0.20–0.50 indicative of this source. Furthermore, values as high as 0.70 have been reported for diesel-related emissions (Rajeev et al., 2023; Ravindra et al., 2008). To account for this variability in the upper limit of the interval, the matrix defines 0.40–0.50 as a transitional interval, where equal weight is assigned to liquid fossil fuels and biomass/coal, while values below 0.40 and above 0.50 are assigned to liquid fossil fuels and biomass/coal, respectively. The robustness of this transition interval was evaluated through a sensitivity analysis in which the lower and upper boundaries of the transition interval were shifted by ±5% and ±10%, while all other classification thresholds and scoring rules were kept unchanged (Supplementary Table S2).

Following the same logic used for IP/(IP + BghiP), BaA/(BaA + Chry) was treated as an ambiguous indicator when the literature provided inconsistent source assignments. Values below 0.20 reflect petrogenic origins, whereas values above 0.20 have been variously attributed to traffic emissions (Akyüz and Çabuk, 2010; Amarillo et al., 2017; Tobiszewski and Namieśnik, 2012) or biomass/coal burning (Liu et al., 2017; Vu-Duc et al., 2021). Consequently, values exceeding 0.20 were classified as a mixed pyrogenic category to reflect the ambiguity in source attribution. Finally, BaP/BghiP was included as an auxiliary indicator, where values above 0.60 differentiate liquid fuel traffic emissions from biomass/coal combustion (<0.60) (Akyüz and Çabuk, 2010; Bi et al., 2020; Kim et al., 2021; Safo-Adu et al., 2023; Tobiszewski and Namieśnik, 2012).

Semi-quantitative scoring scheme

Based on the criteria mentioned above, a discrete scoring scale (0–2) was assigned to evaluate three source categories: Petrogenic, Liquid Fossil Fuel Combustion, and Biomass Combustion, reflecting the main source categories relevant to the local context of Tandil, Argentina (Table 1). The Biomass Combustion category was adopted to reflect the local relevance of wood and biomass burning, while the diagnostic-ratio thresholds used to characterize this category were derived from the broader biomass/coal combustion classification reported in the literature. A ratio assigns two points when corresponding to a single source category, one point when falling within an overlapping transitional boundary, and 0 points when providing no attribution to that source category.

A discrete scale was selected due to diagnostic ratios operating only as qualitative indicators. The 0–1-2 scale was intentionally selected to preserve the qualitative nature of diagnostic ratios while allowing transitional source boundaries to contribute proportionally without implying a false level of quantitative precision. The matrix structure assigns an equal maximum potential score of eight points to each of the three source categories under complete diagnostic agreement across the four ratios.

Missing data treatment and source contribution calculations

Prior to the DSM calculation, concentration values were evaluated against analytical detection limits. For individual PAH congeners with values below the LOD or LOQ accounting for less than 15% of total observations, missing values were substituted with LOD/2 and LOQ/2, respectively (US EPA, 2006).

When analytical data were unavailable to calculate specific ratios, maximum achievable scores were adjusted per sample. For a sample i with a complete set of ratios, the relative contribution (R C i, k, %) of source category k (k ∈ {1,2,3}) is calculated as: where S i, k is the raw score obtained for source k. For samples with missing ratios, raw scores were normalized against the maximum achievable score (S i, k max) to obtain a relative score (R S i, k):

The final relative contribution (F R C i, k, %) across source categories is calculated as:

Multi-sample integration and spatial-temporal profiling

Single-sample diagnostics characterize conditions during specific sampling events. To evaluate spatial patterns across monitoring sites, relative scores (R S i, k) were aggregated across n observations and normalized:

This multi-sample integration results in the aggregated contribution (AC) of individual observations to represent spatial source profiles across the monitoring network.

Data statistical analysis

Descriptive statistics were calculated for each PAH and for the ΣPAHs, LMW, and HMW groups. Statistical analyses were performed in Python (v. 3.12.13) using the pandas (v. 2.2.2), NumPy (v.2.0.2), seaborn (v.0.13.2), SciPy (v.1.16.3), scikit-learn (v.1.6.1), statsmodels (v.0.14.6), re (v.2.2.1) and gspread (v.6.2.1) libraries.

Differences in PAH concentrations between PM 2.5 and PM 10 were evaluated using the non-parametric Mann–Whitney U test. Differences among sampling sites (high-, medium-, and low-traffic areas) were assessed using the Kruskal–Wallis test, followed by Dunn’s multiple-comparison post hoc test with Benjamini–Hochberg false discovery rate (FDR) adjustment when significant overall differences were detected. Statistical significance was established at p-value < 0.05.

To evaluate the consistency between the multivariate PCA approach and the molecular diagnostic ratios, correlations between PC scores and diagnostic ratio values for each sample were calculated. The normality of each variable was assessed using the Shapiro–Wilk test. Since PCA scores and several diagnostic ratios did not satisfy the assumption of normality, Spearman’s rank correlation coefficient was used. To account for multiple hypothesis testing, p-values were adjusted using the FDR procedure, and an adjusted p-value < 0.05 was considered statistically significant.

Particulate matter concentration

Particulate matter concentrations recorded across the monitoring sites and particulate fractions are summarized in Table 2. For PM 10, the mean concentration was 24.77 ± 10.07 μg m -3 at the HT site, 24.12 ± 13.00 μg m -3 at the MT site, and 21.49 ± 13.71 μg m -3 at the LT site. For the fine fraction (PM 2.5), mean concentrations were 17.77 ± 7.13 μg m -3 at site HT, 14.14 ± 9.85 μg m -3 at site MT, and 12.18 ± 9.11 μg m -3 at site LT.

With reference to World Health Organization (WHO) daily guidelines, PM 10 concentrations exceeded established thresholds in 4.35% of samples at site HT, 5.00% at site MT, and 10.53% at site LT. For PM2.5, non-compliance rates were 22.22% at site MT, 39.13% at site LT, and 57.14% at site HT.

PAH concentration

The occurrence and concentrations of individual PAHs varied among sampling sites and particulate fractions (Figure 2). Nap, Ace, and Flu were below detection limits in all analyzed samples, whereas Acy was detected only once in the PM 2.5 fraction at the HT site. The HMW PAHs were detected in both PM fractions, representing 98% of the total PAH samples (ΣPAHs) in both PM 10 and PM 2.5. The mean HMW concentration was 8.77 ± 13.02 ng m -3 in PM 2.5 and 9.87 ± 13.99 ng m -3 in PM 10. The LMW PAHs yielded mean concentrations of 1.27 ± 0.96 in PM 2.5 and 1.50 ± 1.71 ng m -3 in PM 10.

Mean ΣPAHs values across sites were 5.46 ± 4.13 ng m -3 (PM 2.5) and 5.47 ± 4.76 ng m -3 (PM 10) at site LT, 12.91 ± 20.69 ng m -3 (PM 2.5) and 16.25 ± 22.28 ng m -3 (PM 10) at site MT, and 9.71 ± 10.24 ng m -3 (PM 2.5) and 8.60 ± 7.73 ng m -3 (PM 10) at site HT. A corresponding distribution was obtained for ΣHMW, with mean concentrations increasing from 4.98 ± 4.28 ng m -3 (PM 2.5) and 5.37 ± 4.83 ng m -3 (PM 10) at site LT to 12.48 ± 20.44 ng m -3 (PM 2.5) and 15.58 ± 21.76 ng m -3 (PM 10) at site MT. Concentration of ΣLMW showed no directional trend among sampling sites. Individual sampling events throughout the year reveal higher total PAH concentrations during the colder sampling periods compared to the warmer months, as shown in the time-series distribution in Supplementary Figure S1 (Supplementary Material).

The individual compounds IP, BbF, BkF + BjF, BaP, BaA, Chry, BghiP, and DBahA exhibited the highest detection frequencies across the three sampling sites. Mean concentrations of these compounds were higher at site MT than at site LT, particularly in PM 10. For instance, mean IP concentrations were 1.24 ± 0.50 ng m -3 at LT site and 1.98 ± 1.93 ng m -3 at MT site; BbF concentration was 1.23 ± 0.63 at LT site and 2.14 ± 2.31 ng m -3 at MT site; and BkF + BjF concentration was 1.02 ± 0.61 ng m -3 at LT site and 2.07 ± 2.80 ng m -3 at MT site. The lower molecular weight compounds from our dataset were detected less frequently; nevertheless, Fl and Py registered higher mean concentrations at MT site in both particulate fractions, relative to the other locations.

No significant differences were observed between PM 10 and PM 2.5 for any individual PAH or for ΣLMW, ΣHMW, and ΣPAHs (Mann–Whitney test, p > 0.05). Kruskal–Wallis tests across sampling sites indicated significant differences for Fl, ΣPAHs, and ΣHMW. Pairwise Dunn’s tests with FDR adjustments confirmed significant differences for ΣPAHs and ΣHMW, with site LT exhibiting lower concentrations than sites HT and MT, while no significant difference was observed between HT and MT. Following the FDR correction, pairwise comparisons for Fl were not statistically significant.

Source apportionment and relative source contributions

PCA identified two principal components (PC) with eigenvalues >1 for both particulate fractions, explaining 90.31% of the total variance in PM 2.5 and 92.79% in PM 10.

For PM 2.5, PC1 accounted for 76.64% of the total variance and was characterized by consistently high positive loadings (0.973–0.998) for HMW PAHs (BaA, Chry, BbF, BkF + BjF, BeP, BaP, IP, DBahA, and BghiP). PC2 accounted for 13.67% of the variance and was characterized by positive loadings for Phe (0.929) and Ant (0.847). For PM 10, PC1 explained 81.95% of the total variance, with positive loadings (0.945–0.994) for BaA, Chry, BbF, BkF + BjF, BeP, BaP, IP, DBahA, and BghiP, and moderate positive loadings (0.807) for Fl and Py. PC2 explained 10.84% of the total variance, dominated by Phe (0.916) with a secondary loading for Py (0.526).

Application of the semi-quantitative DSM provided relative contribution estimates for three source categories across sites and fractions (Figure 3). Biomass Combustion: Calculated contributions ranged from 54.3% to 73.9%. The highest relative score was recorded for PM 2.5 at site MT (73.9%), and the lowest for PM 2.5 at site LT (54.3%).

Liquid Fossil Fuel Combustion: Relative contributions ranged from 20.0% to 38.3%. The maximum contribution occurred in PM 2.5 at site LT (38.3%), while the minimum was observed in PM 2.5 at site MT (20.0%).

Petrogenic Sources: Estimated contributions ranged from 2.8% to 12.3%. The highest value was determined for PM 2.5 at site HT (12.3%), and the lowest for PM 10 at site LT (2.8%).

Consistency between PCA and diagnostic ratios

Relationships between PCA component scores and calculated diagnostic ratios were evaluated using Spearman’s rank correlation with Benjamini–Hochberg FDR adjustments (Figure 4). It shows that, in PM 2.5, significant correlations after FDR correction were observed between PC1 and Fl/(Fl + Py), IP/(IP + BghiP), Ant/(Ant + Phe), and BaP/BghiP, whereas BaA/(BaA + Chry) was not significantly associated with either principal component. In PM 10, only IP/(IP + BghiP) remained significantly correlated with both PC1 and PC2 after FDR correction.

Source identification through principal component analysis

The predominance of HMW PAHs in PC1 for both PM 2.5 and PM 10 indicates that pyrogenic emissions were the major source of particle-bound PAHs in the study area. PC1 explained 76.6% and 81.9% of the total variance in PM 2.5 and PM 10, respectively, characterized by uniformly high positive loadings (0.945–0.998) for BaA, Chry, BbF, BkF + BjF, BeP, BaP, IP, DBahA and BghiP. This profile represents a general combustion signature associated with vehicular exhaust, biomass burning, and other thermal processes (Kim et al., 2021; Liu et al., 2017; Pies et al., 2008; Rajeev et al., 2023). Because all HMW PAHs are loaded onto the same component, PC1 likely represents a general combustion factor rather than a single specific emission source. In contrast, PC2 explained only 13.7% and 10.8% of the total variance in PM 2.5 and PM 10, respectively. This component was dominated by LMW compounds, specifically Phe (0.929) and Ant (0.847) in PM 2.5, and Phe (0.916) in PM 10. PC2 may represent lower-temperature combustion processes and, to a lesser extent, petrogenic emissions. Mellado et al. (2022) detected Ant and Phe in PM 2.5 in Gran La Plata (Argentina), where petroleum-related industrial activities represent an important emission source. Likewise, Kim et al. (2021) associated Phe and PAHs of 3–4 rings to petroleum evaporation, linked to petrogenic contribution.

The lower PC2 variance explanation regarding PC1 may reflect atmospheric processing of the more volatile PAHs, as Phe and particularly Ant are more susceptible to partitioning and photochemical transformation than HMW PAHs. In addition, Tobiszewski and Namieśnik (2012) reported that the Ant/(Ant + Phe) ratio is particularly sensitive to photochemical aging because Ant degrades faster than Phe, suggesting that part of the variability represented by PC2 may also reflect atmospheric processing.

Spatial patterns of PAH concentrations correspond with the multivariate structure. ΣPAHs and HMW were lower at the LT site than at the MT and HT sites, whereas no significant differences were observed between the latter two locations. In contrast, neither LMW nor the concentrations of individual PAHs differed between PM 2.5 and PM 10. The predominance of HMW compounds, together with their spatial variability, is consistent with the PCA results, indicating that combustion-derived PAHs largely control the observed differences among sampling sites. The similarity between the MT and HT sites also suggests that local traffic intensity alone does not fully explain the measured PAH concentrations and that additional combustion sources contribute to the urban atmospheric burden.

Apportionment of mixed urban combustion sources

While the PCA identified pyrogenic emissions as the dominant factor, it did not resolve specific combustion categories. The application of the DSM addressed this limitation by integrating multiple ratio thresholds into a single attribution scheme. Unlike the conventional interpretation of individual molecular diagnostic ratios, which frequently yields conflicting source assignments, the proposed approach integrates the information provided by multiple ratios into a single weighted source attribution framework. This methodology allows each sample to receive proportional contributions from liquid fuel combustion, biomass combustion, and petrogenic sources rather than being classified according to a single diagnostic ratio. Consequently, the method reduces the uncertainty associated with interpreting individual ratios independently and provides a more comprehensive representation of mixed emission sources.

The DSM consistently assigned the highest relative diagnostic score to biomass combustion across all sampling sites, ranging from 54.3% to 73.9%, followed by liquid fuel combustion (20.0%–38.3%). Petrogenic inputs received relative scores below 12.3% in both particulate fractions. These percentages describe the relative distribution of diagnostic evidence within the DSM and do not represent quantitative source contributions. These findings suggest that the combustion signature identified by PC1 does not originate exclusively from vehicular traffic but instead reflects the coexistence of traffic-related emissions and biomass combustion. The predominance of biomass combustion, even at the HT site, suggests the influence of a widespread urban combustion background beyond direct vehicular emissions. Such source mixing is expected in intermediate-sized cities, such as Tandil, where residential heating, domestic biomass burning, particularly during colder periods, and road traffic may simultaneously contribute to the atmospheric PAH burden. In addition, regional transport of combustion-derived aerosols and the diffuse spatial distribution of biomass-burning emissions may reduce the contrast among sites with different traffic intensities. Pereira et al. (2019) studied PAHs, among other air pollutants, in Latin American cities and identified biomass combustion and vehicular activity as one of the main sources across all the studied cities. Besides, González et al. (2024) found that biomass burning and vehicular emissions were among the primary sources in Mexican cities. Regarding Argentinian cities, the results from Amarillo et al. (2017) and Arias et al. (2022) show that vehicular emissions, both gasoline and diesel, heating and combustion activities were identified as direct sources of PAHs.

Consistency analysis and methodological integration

The consistency analysis between PCA and diagnostic ratios supports this interpretation. After controlling for multiple comparisons using the Benjamini–Hochberg false discovery rate procedure, the IP/(IP + BghiP) ratio was the only diagnostic ratio consistently associated with the PCA structure in both particulate fractions (r = −0.81, p < 0.001 for PC1 in PM 2.5; r = −0.51, p < 0.001 for PC1 in PM 10). The remaining ratios exhibited fraction-dependent or non-significant relationships following FDR adjustment.

This difference indicates non-equivalent performance among individual ratios. Ratios involving LMW PAHs are sensitive to gas-particle partitioning, photochemical degradation, and volatilization during atmospheric transport. In contrast, IP/(IP + BghiP), composed of higher-molecular-weight congeners, exhibits lower reactivity, maintaining alignment with the multivariate variance captured by PCA.

Overall, the combined application of PCA and the weighted diagnostic ratio matrix provided a more robust source apportionment than either method alone. PCA objectively identifies the dominant covariance structure of atmospheric PAHs but does not directly discriminate among combustion sources. In contrast, the weighted diagnostic ratio matrix converts individual ratios into quantitative relative contributions while accounting for boundary overlaps. This integration provides an approach for source apportionment in intermediate-sized urban environments where vehicular emissions, residential heating, and minor petrogenic sources occur simultaneously.

Study limitations and future research directions

Some limitations should be considered when interpreting the proposed source-apportionment framework. First, diagnostic ratios may be affected by atmospheric aging and differential photochemical degradation, with ratios involving lower-molecular-weight PAHs being more susceptible to such processes and potentially modifying their source signatures during transport potentially experiencing greater modification during transport.

Because atmospheric oxidative conditions may vary among monitoring locations, these processes could modify the behavior of some diagnostic ratios and, consequently, their source signatures. Second, the 24-h integrated sampling period limits the ability to resolve short-term variations associated with traffic peaks and residential combustion. Third, the number of samples available for each particulate fraction limits the statistical power and generalizability of the source profiles; therefore, increasing the number of observations for both PM 2.5 and PM 10 would be important for improving the robustness of the proposed framework. Finally, the proposed DSM provides a semi-quantitative assessment of the relative diagnostic evidence for different source categories rather than absolute emission contributions; therefore, its scores should not be interpreted as direct mass fractions. In addition, the weighting scheme is based on literature-derived threshold values and may require refinement when applied to regions characterized by substantially different fuel compositions, combustion technologies, or emission profiles.

Future research should focus on benchmarking the DSM against established receptor models, particularly PMF, using datasets with sufficient sample size, chemical resolution, and uncertainty characterization to support quantitative source apportionment (Belis et al., 2019). Higher temporal-resolution measurements and independent molecular tracers, particularly markers associated with vehicular and biomass combustion, could further strengthen this evaluation. (Belis et al., 2019). Moreover, future studies could incorporate simultaneous gas- and particle-phase PAH measurements to assess the influence of gas-particle partitioning on diagnostic ratios and further refine the DSM. Applying the framework to additional urban environments and different particulate or environmental compartments would also help evaluate its transferability and robustness.

Looking ahead, the integration of DSM into smart mobility strategies and digital urban management platforms offers a promising pathway for modern cities (Mak and Lam, 2021). As urban areas adopt advanced digital architectures, incorporating AI-driven transport models, open-data networks, and real-time sensor technologies will further refine PAH source apportionment and support circular urban mobility solutions (Kaginalkar et al., 2021; Aslam et al., 2026). The proposed matrix serves as a scalable, foundational tool that can be embedded into broader digital twin and smart city governance systems to foster long-term air quality resilience.

Such developments could support the use of integrated PAH-based diagnostics as a practical tool for identifying emission sources and informing targeted urban air-quality and sustainable transport strategies.

This study evaluated an integrated methodology for PAH source identification in PM 2.5 and PM 10 in an intermediate-sized Latin American city by combining PCA with a semi-quantitative weighted diagnostic ratio matrix. While PCA effectively identified the dominant combustion-related structure of particle-bound PAHs, its ability to distinguish among overlapping combustion sources was limited. The proposed DSM addressed this limitation by integrating multiple diagnostic ratios into a single, reproducible source-attribution framework and assigning weighted evidence to petrogenic, liquid fossil fuel, and biomass combustion sources.

The DSM consistently identified biomass combustion as the highest relative diagnostic score across the monitoring sites, followed by liquid fossil fuel combustion, whereas petrogenic inputs were comparatively minor. This result indicates that urban PAH levels were influenced not only by traffic-related emissions but also by a broader combustion background, likely associated with residential heating and biomass-burning activities. The agreement between the molecular diagnostic evidence and the PCA structure, particularly the consistent association of IP/(IP + BghiP) with the PCA-derived patterns, provided an independent consistency check of the source interpretation. This integration reduces the reliance on individual diagnostic ratios, which may provide ambiguous or overlapping source assignments when interpreted separately.

The main methodological contribution of this study is therefore the development of a practical decision-support approach that translates multiple qualitative PAH diagnostic indicators into a standardized semi-quantitative source profile. Beyond its application to the Tandil dataset, the proposed DSM provides a transparent and reproducible framework for integrating multiple diagnostic ratios into a unified source-attribution scheme, thereby reducing the ambiguity commonly associated with independent ratio interpretation. By combining multivariate patterns with weighted molecular evidence, the proposed framework provides a systematic basis for interpreting mixed combustion sources more effectively than conventional ratio-by-ratio approaches. Its relatively simple structure also facilitates comparison among sites and particulate fractions, making it useful for urban air-quality assessment and source-oriented environmental management.

From a sustainable transport and urban resilience perspective, distinguishing the relative influence of vehicular fuel combustion from other combustion sources is particularly relevant for designing targeted air-quality interventions. Further validation using high-resolution measurements and independent source-specific molecular tracers is needed to assess the transferability of the matrix across cities and emission environments. Because the methodology relies exclusively on routinely measured PAH concentrations and simple statistical tools, it may represent a practical screening approach for source assessment in medium-sized cities where advanced receptor modeling is not always feasible. Nevertheless, the proposed methodology provides a methodological foundation for integrating PAH source diagnostics into broader urban air-quality and transport management strategies. Although developed and evaluated for atmospheric PAHs, the conceptual principles underlying the DSM could potentially be extended to other environmental applications where multiple diagnostic indicators must be integrated into a unified attribution framework.