Particulate matters, aldehydes, and polycyclic aromatic hydrocarbons produced from deep-frying emissions: comparisons of three cooking oils with distinct fatty acid profiles

Cooking with oils, especially deep-frying foods emits a significant amount of particulate and gaseous pollutants. Cooking emissions have been found as one of the most important sources of organic particulate matter, contributing to 10–34% of total ambient primary organic aerosol and have been associated with adverse health effects, such as elevated risks of lung cancer and mutagenicity even in non-smokers. A higher incidence of respiratory diseases in cooks has also been attributable to frequent exposure to the degradation products of cooking. Various gaseous- and particle-phase toxic contaminants in the cooking emissions have been documented, notably PAHs and aldehydes. Kitchens are also an important source of black carbon (BC) and ultrafine particles, and range hoods may not always perform satisfactorily. Previous studies report that deep-frying generates higher magnitudes of air pollutants than other cooking methods, and emissions vary with oil properties. However, comprehensive investigations across oils with different fatty acid compositions during deep-frying are limited. This study compares multiple harmful emissions (particulate and gaseous-phase contaminants) from deep-frying using three popular cooking oils with distinct fatty acid compositions (palm, olive, soybean). It also explores relationships between emitted pollutant concentrations and oil characteristics (peroxide value, acid value, total polar compounds) to shed light on mechanisms.

- Cooking is a major indoor source of organic particulate matter, contributing 10–34% of ambient primary organic aerosol; exposure has been linked to lung cancer risk and mutagenicity in non-smokers and higher respiratory disease incidence in cooks. - Toxic cooking emissions include PAHs and aldehydes; kitchens are significant sources of BC and ultrafine particles, and range hoods may be insufficient. - Deep-frying emits more pollutants than other methods; emissions vary with oil properties. Prior controlled studies showed PM2.5, PAHs, and aldehydes vary by oil type, but comprehensive comparisons across oils differing in fatty acid profiles during deep-frying were lacking. - Prior work indicates saturated fatty acids (SAFAs), monounsaturated (MUFAs), and polyunsaturated (PUFAs) can influence particulate and aldehyde formation; PUFA-rich oils tend to generate more aldehydes upon heating; SAFA content and specific fatty acids (e.g., palmitic) have been implicated in particle formation. Mechanistic studies on lipid heating suggest pathways toward PAH formation depend on fatty acid structure.

Design and setting: Experiments were performed in a simulated kitchen (3.25 m × 3.10 m × 2.75 m) at the National Health Research Institutes (Taiwan). An electric fryer (5.0 L, 2000 W; WFT-4L, WISE Inc.) was placed on a table 1.1 m above the floor, beneath a range hood (89 cm × 52 cm; DR-7790ASXL, Sakura) mounted 70 cm above the fryer (1.8 m above floor). A backboard connected the hood to the table behind the fryer; other sides were open. Range hood flow was set to 4 m³/min to minimize turbulence effects in sampling. Oils and food: Three common cooking oils with distinct dominant fatty acids were tested: soybean oil (53.0% linoleic acid, PUFA-rich), palm oil (39.7% palmitic acid, SAFA-rich), and olive oil (72.5% oleic acid, MUFA-rich). For each test, 3.5 L of oil was used to deep-fry 12 consecutive batches of French fries (175 g/batch; Ya Fang Inc.). Protocol: preheat oil to 180 °C for 10 min; fry for 8 min; then 2 min off before next batch. Windows were closed to reduce drafts; an electrostatic precipitator was installed after the hood duct to remove exhausted oil fumes; make-up air came from room/outdoor. Sampling layout and timing: Sampling inlets were 15 cm beneath the range hood and 40 cm above the fryer to limit hood-induced turbulence. The kitchen air was purged for 30 min with door/window open before and after frying. Each test included 10 min background (BK), 10 min preheating (PreH), and 120 min deep-frying (12 batches). Each oil was tested in three independent repeats. Sampling tubes were cleaned and purged before switching oils. Real-time particle and BC measurements: Particle number and mass concentrations were measured using a scanning mobility particle sizer (SMPS; TSI classifier 3080, CPC 3775; size 0.02–0.54 µm, 93 bins) and an aerodynamic particle sizer (APS 3321; 0.55–19.81 µm, 51 channels). Particle sizes were categorized into Aitken (20–100 nm), accumulation (100–1000 nm), and coarse (>1000 nm) modes. Particle mass concentration was calculated from number distributions assuming effective density 0.9 g/cm³. Black carbon was monitored with an aethalometer (AE33, 880 nm, 60 s resolution). PAH sampling and analysis: Particle-phase PAHs were collected at 10 L/min on 37 mm quartz filters (1 µm; Whatman) in Personal Environmental Monitors (PM2.5; SKC). Gaseous-phase PAHs were sampled using XAD-2 cartridges (SKC) at 1.0 L/min with a linear air pump (Hiblow HP150). Extracts used dichloromethane:hexane (2:1) and were analyzed by GC–MS/MS in MRM mode. Method detection limits (MDLs) ranged 0.63–2.57 ng/mL; recoveries 73.6–128% (Supplementary tables/methods). Values below LOD were assigned LOD/2. Aldehyde sampling and analysis: Particle-phase aldehydes were collected at 10 L/min on 2,4-DNPH-coated glass fiber filters (37 mm, 1 µm; Supelco) in PM2.5 PEMs. Gaseous-phase aldehydes were captured downstream on 2,4-DNPH-coated silica cartridges (Supelco) at 1.0 L/min (Hiblow HP150). Flow rates were calibrated with a MesaLabs Defender 520 rotameter. DNPH derivatives were extracted with 5 mL acetonitrile and analyzed by HPLC (Jasco PU-2089) with a gradient program (Supplementary). MDLs ranged 0.008–0.058 µg/mL; recoveries 48.0–99.1%. Values below LOD were set to LOD/2. Oil characterization: Fresh oil fatty acid composition was analyzed by AOAC 996.06. Acid value (AV) and peroxide value (POV) by CNS 3647 N6082 and CNS 3650 N6085, respectively. Total polar compounds (TPC) by AOAC 982.27 (column chromatography). Oil quality indices and fatty acid profiles are in Supplementary materials. Statistical analysis: Non-parametric Kruskal–Wallis tests assessed differences among oils for particulate matter, BC, PAHs, and aldehydes. Spearman rank correlations evaluated associations between fatty acid classes (PUFA, MUFA, SAFA), oil quality indices (AV, POV, TPC), and emissions (particle mass/number, selected compounds). Time-series and correlation plots were generated with SigmaPlot 12.0. Replication: n=3 per oil (gas-phase PAH palm samples n=2).

- Particle number emissions: Palm oil produced the highest total particle number concentration: (3896 ± 1797) × 10^3 #/cm^3, significantly higher than soybean ((469 ± 476) × 10^3 #/cm^3) and olive ((400 ± 156) × 10^3 #/cm^3). Elevation was most pronounced in the Aitken (20–100 nm) and accumulation (100–1000 nm) modes. Particle mass concentrations were similar across oils. - Correlations with oil properties: Total particle number concentration positively correlated with palmitic acid (major SAFA) and with total polar compounds (TPC): palmitic acid r ≈ 0.73–0.78 (p < 0.05), TPC r = 0.68 (p < 0.05). Average BC concentration also positively correlated with TPC (r = 0.68, p < 0.05). - Black carbon (BC): Mean BC concentrations during frying were 0.93 ± 0.80 µg/m^3 (soybean), 1.99 ± 1.01 µg/m^3 (palm), and 1.74 ± 1.05 µg/m^3 (olive). Although means were not statistically different, time-series showed peaks during frying and positive association with TPC. - PAHs: Total (gas + particle) PAHs were 22.43 ± 18.62 ng/m^3 (soybean), 16.44 ± 7.09 ng/m^3 (palm), and 7.32 ± 10.5 ng/m^3 (olive). Naphthalene dominated gaseous PAHs (soybean 97%, palm 62%, olive 87%). Palm oil emitted significantly higher particle-phase PAHs than soybean and olive. Cyclopenta(c,d)pyrene was the predominant particle-phase PAH (62% soybean; 56% palm; 37% olive). Benzo(a)pyrene (IARC Group 1) was detected in particle-bound form at 0.039 ± 0.067 ng/m^3 (soybean), 0.11 ± 0.097 ng/m^3 (palm), 0.14 ± 0.023 ng/m^3 (olive). - PAH–fatty acid associations: Positive correlations included palmitic acid with acenaphthene (r = 0.74, p < 0.05) and benzo(e)pyrene (r = 0.79, p < 0.05); chrysene with total SAFA (r = 0.86, p < 0.01); and benzo(k)fluoranthene with oleic acid (r = 0.72, p < 0.05). TPC correlated with acenaphthene (r = 0.83, p < 0.05) and benzo(e)pyrene (r = 0.79, p < 0.05), though total PAHs did not correlate significantly with TPC. - Aldehydes: Total aldehydes were significantly higher for soybean oil (3636 ± 607 µg/m^3 gaseous; total 3655 ± 598 µg/m^3) than olive (2453 ± 1304 µg/m^3) and palm (2197 ± 841 µg/m^3), indicating PUFA-rich oil emits more aldehydes. • Soybean (top contributors): hexanal 1033 ± 206 µg/m^3; 2,5-dimethylbenzaldehyde 828 ± 195 µg/m^3; acrolein 674 ± 110 µg/m^3; also elevated propionaldehyde (157 ± 29 µg/m^3), crotonaldehyde (92 ± 15 µg/m^3), trans-2-nonenal (34.4 ± 11.2 µg/m^3). • Palm: hexanal 634 ± 277 µg/m^3; nonanal 444 ± 155 µg/m^3; trans-2-heptenal 399 ± 162 µg/m^3. • Olive: nonanal 2453 ± 1304 µg/m^3 (total nonanal incl. particle), hexanal 549 ± 477 µg/m^3; trans-2-heptenal 114 ± 135 µg/m^3 (gas) with small particle-phase contributions. - Aldehyde–fatty acid associations: Total aldehydes positively correlated with α-linolenic acid (ALA) percentage (r = 0.78, p < 0.01). Specific aldehydes correlated with ALA: acrolein (r_s = 0.72, p < 0.05), propionaldehyde (r = 0.86, p < 0.05), crotonaldehyde (r = 0.85, p < 0.05), hexanal (r = 0.83, p < 0.05), trans-2-heptenal (r_s = 0.83, p < 0.05), trans-2-nonenal (r = 0.78, p < 0.05). Hexanal (r = 0.73, p < 0.05) and trans-2-heptenal (r = 0.67, p < 0.05) also correlated with linoleic acid (LA). Nonanal emission correlated with MUFA content (r_s = 0.67, p < 0.05) and was highest for olive oil. - Particle size distributions: Unimodal for number and bimodal for mass across oils. Peak number diameters: soybean ~109.4 nm, palm ~57.3 nm, olive ~94.7 nm. Smaller emitted molecules corresponded to higher particle number concentrations.

The study addressed whether and how fatty acid composition and oil quality indices influence toxic emissions during deep-frying. Findings show that SAFA-rich palm oil generates substantially higher ultrafine particle numbers and particle-bound PAHs, with particle numbers correlating with palmitic acid and oil degradation marker TPC. This links saturated fatty acid content and oil degradation to elevated particulate and BC emissions, implicating increased exposure to ultrafine particles and particle-bound carcinogens. Conversely, PUFA-rich soybean oil emitted the highest levels of gaseous aldehydes, and aldehyde concentrations correlated strongly with α-linolenic acid and linoleic acid contents, consistent with faster oxidation of more unsaturated fatty acids and subsequent aldehyde formation from lipid hydroperoxide decomposition. MUFA-rich olive oil generally produced lower aldehyde totals than soybean and lower ultrafine particle numbers and particle-bound PAHs than palm, suggesting a comparatively favorable emission profile under the tested conditions. These patterns provide mechanistic insights: higher unsaturation promotes aldehyde formation; higher saturation (and specific fatty acids like palmitic acid) associates with ultrafine particle and PAH formation. The positive associations of TPC with particle number and BC further indicate that oil degradation progression contributes to airborne pollutant levels. Together, results inform selection of cooking oils and management of frying conditions to reduce indoor exposure to harmful pollutants.

Deep-frying emissions depend strongly on oil fatty acid profiles and oil quality. Under standardized frying of French fries at 180 °C: - Palm oil (SAFA-rich) produced the highest ultrafine particle numbers and particle-bound PAHs; particle numbers correlated with palmitic acid and TPC, and TPC correlated with BC. - Soybean oil (PUFA-rich) emitted the highest concentrations of gaseous aldehydes, with total aldehydes correlating with α-linolenic acid and specific aldehydes with linoleic/linolenic acids. - Olive oil (MUFA-rich) generally showed lower toxic emissions among the three, appearing preferable from an emissions standpoint. These findings highlight the role of fatty acid composition in health-relevant cooking emissions and suggest that choosing MUFA-rich oils and controlling oil degradation could mitigate exposure. Future research could expand to more oil types and blends, different foods, temperatures, and ventilation configurations, increase replication to reduce variability, and assess personal exposure and health implications in real kitchens.

- Limited replication (n = 3 per oil; gas-phase PAH palm samples n = 2) may reduce statistical power; authors note higher variability for soybean oil and recommend more replicates. - Simulated kitchen setup with a fixed range hood flow (4 m³/min) and an electrostatic precipitator may not reflect all real-world kitchens; emissions and dispersion could differ with other ventilation conditions or fuel types. - Single food type (French fries), single temperature (180 °C), and a specific fryer may limit generalizability to other foods, cooking methods, or temperatures. - Some analyte concentrations were near detection limits; values below LOD were imputed as LOD/2, potentially introducing uncertainty. - Study focuses on short-term emissions during 12 consecutive batches; long-term oil reuse effects beyond the test window are not assessed.