Introduction

Guaranteeing an adequate indoor air quality represents one of the major modern-day challenges for technical and scientific communities involved in designing and managing indoor environments1,2,3. For decades the main design parameters of efficient buildings were energy efficiency and, in case, the protection from outdoor pollutants4,5,6 whereas the possible health threat of indoor sources was mostly underestimated7. In fact, building occupants are not properly aware of the indoor air quality, of their exposure to pollutants and, consequently, of how much they contribute to reducing the indoor air quality through their habits and activities (e.g. using combustion sources, such as incenses, candles, etc.)8,9.

Airborne particles were recognized as one of the main hazardous pollutants emitted by indoor sources due to their capability to cross the respiratory system, carrying toxic compounds, depositing in the deepest airways, and provoking negative health effects10,11,12. To date, the attention of the scientific community is shifting from super-micrometric particles, such as PM10 and PM2.5 (defined as particle mass concentrations with diameters less than 10 and 2.5 μm, respectively), to sub-micrometric particles13,14. Indeed, airborne particle dynamics and composition are strongly affected by their size, with smaller particles recently recognized as most critical for human health15,16,17,18. Among the different airborne particle-bound toxic compounds, heavy metals (HMs) and polycyclic aromatic hydrocarbons (PAHs) were recognized as highly detrimental for human health. As an example, long-term exposure to HMs can cause several adverse health effects including human developmental retardation, kidney damage, etc.19,20,21,22. Analogously, exposure to PAHs can increase the risk of respiratory and cardiovascular diseases, cognitive development delay, genetic mutations, etc.21,23,24,25. Moreover, some HMs (e.g. As, Cd, Cr(VI), Ni) and PAHs are classified as Group 1 human carcinogens by the World Health Organization26.

Since airborne particle chemical composition and dynamics are size-dependent27, investigating the size-resolved chemical composition of the entire airborne particle size range28,29, i.e. including also sub-micrometric particles, is a key research issue in view of a more detailed risk assessment related to the exposure to indoor-generated airborne particles30,31,32.

Combustion sources are widely used worldwide in indoor environments for several purposes (from cooking activities to religious and esthetical purposes, e.g. incenses, candles and mosquito coils)33,34,35. The scientific community recently characterized most of these sources from physical and dimensional points of view also providing emission rates/factors in terms of particle number and mass. For example, PM2.5 emission rates were found in the ranges 5–250 mg h−1, 0.02–25 mg h−1, and 36.2–88.6 mg h−1 for incenses, candles and mosquito coils, respectively36,37,38,39,40, while particle number emission rates were in the range of 4.54–10.8 × 1014 part h−1, 4.07–4.85 × 1013 part h−1, 2.92–4.57 × 1014 part h−1 for incenses, candles and mosquito coils, respectively40. Such large deviations of the data are likely due to the source composition, e.g. presence of scent additives, essential oils, fragrances, etc.41. Emission factor data for cooking activities were found extremely high as well (typically > 1013 part. h−1 and > 1 mg h−1) and present large deviations due to the several influence parameters, e.g. cooking method (frying, deep-frying, grilling, boiling, baking, etc.), cooking temperature, type of food (vegetable, fat), type of stove (gas, electric, etc.)42,43,44,45,46,47. All these combustion sources present particle number distribution modes in ultrafine range (< 100 nm), with large deviations due to above-mentioned influence parameters, and a high contribution of the sub-micrometric particles to the airborne particle mass emitted40,42,43.

Studies focused on the chemical characterization of the airborne particles emitted by indoor sources are fewer; some studies revealed the presence of dibenzo-p-dioxins/dibenzofurans, PAHs and HMs during incense, candles and mosquito coil burning and cooking activities39,41,45,48,49,50,51,52,53,54,55,56, nonetheless, the data provided are still scarce also considering the several parameters potentially affecting the emission. As an example, we recently carried out an innovative approach to evaluate the lung cancer risk related to airborne particles emitted by indoor sources31 which relies upon particle and related carcinogenic compound emission rates of the sources: one of the main limitations we faced was the incomplete information on the chemical analysis of the particles emitted by such sources. Moreover, we also pointed out that detailed size-segregated chemical compositions of the particles emitted by indoor sources, currently not available, would be extremely important: they would avoid the oversimplifying assumptions considering the chemical composition of the emitted particles invariant to the particle size.

To summarize, even if the scientific literature has characterized the airborne particle emission rates of indoor sources, chemical characterization of such particles was poorly investigated; in addition, information on size-segregated content of HMs and PAHs in indoor-generated airborne particles, that would be extremely useful for proper risk assessments, is completely missing.

In the present study, the size-resolved chemical characterization of airborne particles emitted by indoor combustion sources, i.e. incenses, candles, mosquito-coils and cooking activities (grilling bacon), was carried out and size-dependent HM and PAH contributions to the PM10 were provided thanks to an experimental analysis based on airborne particle collection through an electric low-pressure impactor and consequent gas chromatography-mass spectrometry (GC–MS) and atomic emission spectrometry (AES) analyses.

Materials and methods

Emission sources and sampling site description

Particle sampling experiments were carried out at the Laboratory of Industrial Measurements (LAMI) of the University of Cassino and Southern Lazio, Italy. Samplings were performed in a 1.80 m × 1.20 m × 2.20 m plexiglass chamber presenting a small opening allowing supplying air for combustion phenomena as well as for electrical cables and ducts. The air exchange rate of the chamber in this condition was measured adopting a CO2 decay method (not reported here for the sake of brevity57) and resulted equal to 0.50 ± 0.03 h−1.

Four different emission sources were tested during the experimental campaign: candles, incense sticks, mosquito coils, and grilling bacon on an induction cooker. Candles, incenses, and mosquito coils commercially available were chosen. In particular, candles adopted in the experimental campaign were paraffin wax candles without any kind of additives, incenses were natural benzoin resin sticks, and mosquito coils were made up of prallethrin (pyrethroid insecticide) with co-formulants and colorants. The investigation of cooking activities was limited to the grilling of bacon (which represents one of the most emitting cooking activities42) on a portable induction cooker.

Particle sampling: apparatus description and procedure

In order to collect size-resolved mass amounts for post-hoc chemical characterizations, an electrical low-pressure impactor ELPI + ™ (Dekati Ltd.) was employed (it was placed inside the chamber, whereas its pump was placed outside) and polycarbonate collection foils (type 23,000, dim. 25 mm, Sartorius Stedim Biotech GmbH, Goettingen, Germany) were mounted at each ELPI stage. The ELPI + ™ is a particle spectrometer able to measure airborne particle number distribution (and total concentration) in real-time and simultaneously collect particles with a sampling flow rate of 10 L min−1. Particles are firstly charged with a known charge level in a corona charger, then, they are size classified in a low-pressure cascade impactor according to their aerodynamic diameter. Every impactor stage is electrically insulated, and particles collected in each stage produce an electrical current proportional to particle number concentrations.

The ELPI + ™ presents 15 stages whose lower–upper boundary diameter ranges (expressed as 50% efficiency cut-off diameters, D50) are: 6.0–16.7 nm for stage 1, 16.7–31.3 nm for stage 2, 31.3–54.7 for stage 3, 54.7–94.9 for stage 4, 94.9–156 nm for stage 5, 156–258 nm for stage 6, 258–384 nm for stage 7, 384–606 nm for stage 8, 606–952 nm for stage 9, 952 nm-1.64 µm for stage 10, 1.64–2.48 µm for stage 11, 2.48–3.67 µm for stage 12, 3.67–5.39 µm for stage 13, 5.39–9.93 µm for stage 14, 9.93–10 µm for stage 15. The upper boundary of stage 15 depends on the PM inlet sampling head adopted (if any); we have adopted an EN 12,341 PM10 inlet (downscaled to 10 L min−1), thus, as mentioned above, the upper diameter D50 of the stage 15 is actually 10 µm. Stage 1 (6.0–16.7 nm) doesn’t allow particle collection, whereas stage 15 (9.93–10 µm) acts as a pre-cut stage thus, it is not measured electrically.

Since HM and PAH determination was performed separately, adopting different methods and instruments, two different samplings were performed for chemical post-analyses: Sampling 1 for HMs and Sampling 2 for PAHs. As concern the source combustion, candles were normally burnt, while incenses and mosquito coils were lit by a flame and fanned out so that the glowing ember continued to smolder and burn away the rest of the materials. Grilling of bacon was performed by putting the bacon slices on a pan heated by an induction cooker. In order to collect an adequate amount of particle mass on each ELPI + ™ stage for chemical post-analyses high mass concentrations for long periods were needed. To this end preliminary tests were carried out to check the concentrations reached and estimate the number of sources to burn simultaneously and the total sampling period. Then, the following procedure for each indoor source and each sampling was adopted: several sources (e.g. candles) were lit up simultaneously and the particle collection and measurement with the ELPI + ™ was started as well, then, when the sources burnt out (or, in the case of the cooking activity, when the bacon was properly grilled), they were replaced with new ones. This procedure was performed several times for each source: the quantity of source burnt, the total sampling period, and the average number and mass concentrations measured by the ELPI + ™ during the samplings are summarized in Table 1 for each source type for both the samplings.

Table 1 Quantity of sources burnt, total sampling period, average particle number and mass concentrations for each sampling and for each source type (Sampling 1 for HM analysis, Sampling 2 for PAH analysis).
Full size table

To avoid any contamination between the sources, the charger and the impactor of the ELPI + ™ were cleaned according to the manufacturer suggestions at the end of each sampling and flush pump was used to keep the ELPI + ™ clean between the measurements. Moreover, surfaces of the chamber were cleaned to avoid contamination between the sources due to resuspension of previously deposited particles. Finally, in order to minimize the contribution of particles coming from the environment outside the chamber, a fan filter unit equipped with a HEPA H14 filter and a F7 pre-filtration stage was used at the maximum flow rate of 850 m3 h−1 to clean the environment before test. Nonetheless, in order to avoid measurement artifacts, two further samplings were performed to determine the possible presence of HMs and PAHs in the chamber when the sources are not in operation. These samplings (referred as Sampling blank 1, for HMs, and blank 2, for PAHs) were performed after having applied the abovementioned cleaning procedures (i.e. cleaning of the chamber surface and filtration of the ambient air of the chamber through the fan filter unit), both the samplings were carried out for 10 h consecutively. Concentrations of HMs and PAHs resulting from the analysis of particle mass collected during Sampling blank 1 and blank 2, respectively, resulted lower than the LODs, then guaranteeing that the HM and PAH determination due to the use of combustion sources was not affected by the presence of pre-existing airborne particles in the chamber.

HM determination: apparatus description and analytical method

The analysis of HMs in particle mass samples (Sampling 1 and Sampling blank 1) involved a preliminary digestion of polycarbonate filters in a microwave digestion system (Mars 5, CEM Corporation, Mathews, North Carolina, USA). The protocol applied was as follows: filters were transferred into Autovent HP-1500 vessel and digested with 10 mL of HNO3 according to the instrumental parameters previously set (600 W, 100% Power, 20 min). The digested solution obtained was diluted with ultrapure water (1:10 v/v) before the analysis (detailed information about chemicals and reagents is reported in the Supplementary Information). Heavy metal extract concentrations were determined by Inductively Coupled Plasma—Atomic Emission Spectroscopy using Agilent Technologies 4210 MP-AES Atomic Absorption Spectrometer (AAS). The MP-AES method was validated by the construction of calibration curves in the linear range of 0.05–5 ng mL−1. A multielement mix of heavy metals was used as a standard solution. The emission wavelengths selected for each metal analyzed were: 228.802 nm for Cd, 425.433 nm for Cr(VI), 324.754 nm for Cu, 405.781 nm for Pb, 231.147 nm for Sb, 213.857 nm for Zn, 193.695 nm for As, 340.512 nm for Co, 403.076 nm for Mn, and 361.939 nm for Ni. The limit of detection (LOD) and limit of quantification (LOQ) of the method for heavy metal determination are reported in Table S1 (please see Supplementary Information).

PAH determination: apparatus description and optimization and validation of the analytical method

The amount of PAHs in particle mass samples (Sampling 2 and Sampling blank 2) collected on polycarbonate filters was extracted by ultrasonic extraction (USAE) procedure. The protocol was performed as follows: filters were placed in a beaker, covered with 2 mL of n-heptane solvent and extracted in an ultrasonic bath (Ultrasounds Starsonic 18–35, Liarre s.r.l, Casalfiumanese, Italy) for 6 min at room temperature for three times. The final extracts were combined in the same vial, dried under the stream flow of nitrogen, and recovered with 500 μL of n-heptane before the GC–MS analysis. A mix solution of 15 PAHs was used as a laboratory standard to evaluate the feasibility and reproducibility of the analytical method proposed (detailed information about chemicals and reagents are reported in the Supplementary Information). PAHs were quantified by means of GC-Ion-Trap MS (GC-IT/MS) analysis using a gas chromatograph Finnigan Trace GC Ultra (ThermFinningan, Bremen, Germany) equipped with a Polaris Q ion trap mass spectrometry detector (Thermo Fisher Scientific, Waltham, MA). The column Meta-XLB™ (30 m length, 0.25 mm internal diameter, 0.25 μm film thickness; Teknokroma™, Barcelona, Spain) was used for the separation, and He (99.9995% purity) was adopted as carrier gas at a flow rate of 1.0 mL min−1. The sample injection (Vinj = 1 μL) was performed in splitless mode with an opening split for 4 min through AI/AS 1310 autosampler. After the injection, the Programmed Temperature Vaporizer (PTV) injector heated from 110 to 320 °C at 800 °C min−1. In order to ensure the cleanup of the line, the injector was kept at 320 °C for 10 min after the sample vaporization. The starting oven temperature was 100 °C held for 30 s, then, it was ramped to 150 °C at 20 °C min−1 for 2 min and to 290 °C at 20 °C min−1 for 10 min. Before the chromatographic separation of PAHs, an internal standard solution of dioctyl phthalate (DOP) at 1 ng µL−1 was added to the samples.

Detection of PAHs was achieved in selected ion monitoring (SIM) with detector temperature and transfer line set at 250 °C and 270 °C, respectively. Full-scan MS data were acquired over the mass-to-charge (m/z) ratio ranging from 100 to 400 amu to obtain the fragmentation spectra of the analytes while Thermo Xcalibur Sequence Setup software was used for the data interpretation. The mass-to-charge (m/z) ratios and retention times of PAHs identified by GC–MS are reported in Table S2 (please see Supplementary Information). The method adopted was standardized through the construction of calibration curves in linear ranges reported in Table S2 as well. Recoveries of PAHs from samples spiked with standard solution were found in the range of 87–98% with the error ranging between 8–15%. The limits of detection (LOD) and limits of quantification (LOQ) of the method for PAH determination are summarized in Table S2.

Data post-processing

Calculation of the particle and chemical compound concentrations

Data post-processing of particle distributions and total concentrations was performed through the ELPIVI software. Particle mass distribution and total concentration were evaluated on the basis of the number distribution data and applying the following particle densities: 1.1 g cm−3 for incenses and mosquito coils50,58,59, 1.5 g cm−3 for candles39, and 1.0 g cm−3 for grilling of bacon43,60. Moreover, corrections for diffusion and space charger losses were applied61. Results of particle mass distributions and concentrations hereinafter reported, were obtained by averaging values of the sampling periods (Sampling 1 and 2) for each source.

We also point out that, since mass concentrations at stage 15 are not available (and the stage width is much smaller than others; please see section “Particle sampling: apparatus description and procedure”), in this paper, we are referring to PM10 as the mass concentration in the range 16.7 nm–9.93 µm. Similarly, we are referring to PM1 as the mass concentration in the range 16.7 nm–952 nm.

HMs and PAHs particle mass distributions were obtained as follows: (a) firstly, HM and PAH absolute masses (ng) were calculated by multiplying the concentrations in the solution (i.e. extract concentration; expressed with respect to the solution volume, ng mL−1 or ng µL−1) by the solution volume, 10 mL and 500 µL, for HMs and PAHs, respectively, (b) then, they were divided by the total sampling volume (depending on ELPI + ™ flow rate, 10 L min−1, and sampling periods summarized in Table 1).

Mass fraction of the HMs and PAHs on PM10 emitted by the different sources were evaluated as well as particle mass distributions, such distributions were normalized with respect to the corresponding total mass concentrations then obtaining relative particle mass distributions.

Estimation of the deposited concentration of chemical compounds

The knowledge of the size-resolved chemical composition of airborne particles emitted by the indoor sources under investigation also allows a proper estimate of the actual dose of chemical compounds received by exposed people and the related health risks. Indeed, thanks to the size-resolved chemical composition, size-dependent mass fractions of chemical compounds can be applied; consequently, the oversimplifying assumption considering the chemical composition of the emitted particles invariant to the particle size (i.e. constant mass fraction on the entire particle size range) can be superseded. To this end, as an example, we have evaluated the deposited concentration of chemical compounds (i.e. the dose per unit inhalation rate and unit time of exposure) that an exposed person can receive when exposed to the airborne particles emitted by the sources under investigation considering two different approaches: (a) actual amount of chemical compounds per each channel, i.e. mass fraction obtained from the measured size-resolved mass distributions of the compounds, (b) constant chemical composition for each particle size, i.e. constant measured mass fraction of the compounds (i.e. those reported in Tables 2 and 3). The ratio between the deposited concentrations (dMj-dep) evaluated according the two different methods (size-resolved, SR, and constant, CONST) of the j-th chemical compound received by an exposed person for a specific indoor source can be evaluated as:

$$R_{j} = \frac{{dM_{j - dep\_SR} }}{{dM_{j - dep\_CONST} }} = \frac{{\mathop \sum \nolimits_{i}^{{}} \left( {\frac{dM}{{dlogD_{p} }}} \right)_{i} \cdot dlogD_{p,i} \cdot MF_{j,i} \cdot DF_{i} }}{{MF_{j} \cdot \mathop \sum \nolimits_{i}^{{}} \left( {\frac{dM}{{dlogD_{p} }}} \right)_{i} \cdot dlogD_{p,i} \cdot DF_{i} }}$$
(1)

where (dM/dlogDp)i represents the particle mass concentration of the i-th channel size measured for the specific source under investigation (as reported in Fig. 1), dlogDp,i is the width of the i-th channel, MFj,i is the mass fraction of the j-th chemical compound measured for the i-th channel size (needed for SR case), DFi is the deposition fraction of particles in the lungs (sum of deposition in the alveolar and tracheobronchial regions) as a function of their size, MFj is the constant mass fraction the j-th chemical compound (needed for CONST case). The DFi values were obtained from the International Commission on Radiological Protection62, in particular, we considered average curves between adult males and females (normal nose breather, sitting activity).

Table 2 Mass fractions of HMs on PM10 (expressed as ng ng−1) in the freshly emitted particles by the indoor sources under investigation.
Full size table
Table 3 Mass fractions of PAHs on PM10 (expressed as ng ng−1) generated by the indoor sources under investigation.
Full size table
Fig. 1
figure 1

Particle mass distributions, normalized to the total concentration, measured during combustion of candles, incenses, mosquito coils, and grilling bacon through the ELPI + ™: solid lines represent average distributions, dashed lines represent standard deviations of the measured distributions.

Full size image

Results and discussions

Particle mass distributions

Particle mass distributions averaged among the two samplings (Sampling 1 for HM analysis and Sampling 2 for PAH analysis) for the sources under investigation are shown in Fig. 1, distributions are presented as relative distributions, i.e. normalized to the total mass concentration measured during the samplings. For all the sources a not negligible contribution of PM1 fraction was recognized, in particular, the average PM1 fractions with respect to PM10 were equal to 18%, 18%, 57%, and 35% for candles, incenses, mosquito coils, and grilling bacon, respectively.

Candles present a bimodal distribution with a main mode at stage 13 (3.67–5.39 µm) and a second minor mode in the sub-micrometric range (at stage 9, 606–952 nm), on the contrary incenses are characterized by an unimodal distribution with a main mode at stage 11 (1.64–2.48 µm). Mosquito coils produced an unimodal mass distribution as well with a mode in the sub-micrometric range (at stage 9, 606–952 nm). Finally, the mass distribution related to grilling bacon resulted bimodal with a main mode in the super-micrometric range (at stage 10, 952 nm–1.64 µm) and a secondary mode in the sub-micrometric range (at stage 8, 384–606 nm).

Actually, very few studies reporting mass distribution of freshly emitted particles from indoor combustion sources were carried out by the scientific community. Particle mass distributions while cooking were measured through an ELPI by See and Balasubramanian63 in a Chinese-cooked food stall fueled by LPG, they recognized a bimodal distribution similar to that here presented. In our previous papers, we measured particle mass distributions indoor by adopting an SMPS-APS system while grilling bacon on a gas stove43 and while using candles, incenses and mosquito coils40: distributions for grilling bacon resulted quite similar to those measured in this paper, whereas a smaller mode (about 300 nm) was measured for candles, incenses and mosquito coils. Such variation can be likely due to the huge variability in the composition of the sources.

HM and PAH concentrations

HMs

The results of the analysis of heavy metals in freshly emitted particles by candle, incense, and mosquito coil combustions, and grilling bacon are summarized in Table S3 (please see Supplementary Information) where heavy metal extract concentrations (ng mL−1) analyzed by ICP-AES are shown. The data highlight the presence of Group 1 carcinogenic compounds, i.e. As, Cd, Cr(VI), Ni, in particle samples from mosquito coil combustion and grilling bacon, whereas in samples obtained from candle and incense combustions, they resulted below the limit of detection (LOD). On the contrary, the presence of Cu, Pb, Zn, Sb was recognized in particle samples collected for all the sources under investigation. In addition, traces of Mn were detected only in particle samples collected during candle combustion, whereas Co only was found in samples collected while grilling bacon.

From these data, and adopting the data post-processing described in “Data post-processing”, the HM mass fractions on PM10 (expressed as ng ng−1), were calculated and reported in Table 2: the calculated mass fractions were quite high as they were mostly in the 10–6–10–5 range, i.e. values similar to those typically found in outdoor urban environments30,64,65. A proper comparison with the few data available in the literature cannot be carried out: indeed, the presence of HMs during the combustion of such indoor sources was previously identified39,45,48,56,66 but, even if HM concentrations were provided, the corresponding PM10 were not67. This is the main difficulty we also experienced in our previous paper31 where mass fractions of toxic compounds (with respect to PM10) were needed to properly estimate the lung cancer risk in indoor environments. Actually, we reported mass fractions of carcinogenic compounds (Group 1) on PM10 emitted by wood and pellet combustion phenomena68; nonetheless, no exhaustive studies were available for other indoor sources here investigated.

Despite their overall contribution to the total particle mass a further key information is evaluating the size-resolved mass distribution of HMs. Indeed, as mentioned above, different sizes imply different ability of particles to penetrate in the lungs and transporting there such compounds. In Fig. 2 mass distributions of As, Cd, Ni, Cr(VI) (Group 1 carcinogenic HMs) and Pb, Sb, Cu, Zn in freshly emitted particles from mosquito coil combustion (Sampling 1) are shown. The distributions clearly show that the amount of As is exclusively present in the ultrafine particle range with a huge contribution in stage 2 (16.7–31.3 nm). Similarly, Ni contribution is exclusively due to the sub-micrometric particle range (stages 6 and 7) with a main peak in the 156–384 nm range. Cr(VI) contribution as well was almost completely related to sub-micrometric particles with a clear peak in the range 156–384 nm, whereas the sub-micrometric particle range contribution to the total amount of Cd is 67% as the distribution resulted unimodal with a peak in stage 9 (606–952 nm). As concerns Pb, Sb, Cu and Zn the sub-micrometric contributions are 48%, 64%, 26%, and 11%, respectively; in particular, the Pb mass distribution is unimodal with a main peak at 606–952 nm, the Sb distribution is almost uniform, Cu distribution presents two modes, one in the sub-micrometric (94.9–156 nm) and the other in the super-micrometric range (1.64–2.48 µm), whereas Zn is unimodal with a mode in the super-micrometric range (1.64–2.48 µm).

Fig. 2
figure 2

Mass distributions (normalized to the total concentration) of As, Cd, Cr(VI), Ni (a) and Pb, Sb, Cu, Zn (b) in freshly emitted particles from mosquito coils combustion (Sampling 1).

Full size image

In Fig. 3 mass distributions of As, Cd, Cr(VI) and Pb, Sb, Cu, Zn, Co in freshly emitted particles from grilling bacon (Sampling 1) are shown. As and Cd present a contribution of a sub-micrometric range equal to 67% and 61%, respectively, with a significant fraction in the ultrafine range (around 30% of their total mass), whereas Cr(VI) is almost exclusively present in the sub-micrometric range (95%) as it shows a main peak of the mass distribution in the range 31.3–54.7 (stage 3). As concerns the other compounds, a contribution of the sub-micrometric range of about 70% was detected for Pb, Sb, and Cu, whereas it is equal to 83% for Zn and 100% for Co (completely included in stage 4).

Fig. 3
figure 3

Mass distributions (normalized to the total concentration) of As, Cd, Cr(VI) (a) and Pb, Sb, Cu, Zn, Co (b) in freshly emitted particles from grilling bacon (Sampling 1).

Full size image

Pb, Sb, Cu, Zn and Mn mass distributions in freshly emitted particles from candle and incense combustion experiments are reported in Fig. 4; such distributions resulted quite uniform with sub-micrometric range contributions to their total masses around 60%; with the exception of Mn for candles whose contribution is completely due to the sub-micrometric range (main mode at 384–606 nm).

Fig. 4
figure 4

Mass distributions (normalized to the total concentration) of Pb, Sb, Cu, Zn and Mn in freshly emitted particles from candle (a) and incense (b) combustions (Sampling 1).

Full size image

We point out that the huge contribution of the sub-micrometric particle range emitted by mosquito coil combustion and grilling bacon is a critical aspect in terms of people’s exposure and related lung cancer risk since As, Cd, Ni, Cr(VI) represent the particle-bound compounds with the highest inhalation cancer slope factors69.

As mentioned above, size-resolved chemical compositions of airborne particles sampled in indoor environments or emitted from indoor sources are not common in the scientific literature. In a recent paper Martins et al.70 provided size-resolved chemical compositions of PM10 sampled both at outdoor and indoor environments of schools and homes: in this study, the presence of the HMs in sub-micrometric particle range was also detected, with some differences between schools and homes likely due to the different sources affecting such environments (mainly penetration from outdoor for schools, possible presence of indoor sources for homes). Similarly, Pekey et al.71 found a higher presence of some HMs in homes with respect to outdoor values due to the possible presence of indoor sources (especially in winter when indoor-generated pollutants can slowly exfiltrate). Nonetheless, in these papers, the contribution of indoor sources to such indoor chemical compound concentrations is not provided.

PAHs

Results of the analysis of PAHs in freshly emitted particles due to the combustion of the investigated sources are reported in Table S4 (please see Supplementary Information) where PAH extract concentrations are summarized. Particle samples collected during the combustion experiments showed the presence of several PAHs emitted by the sources under investigation; in particular, benzo(a)pyrene which is classified as Group 1 carcinogenic compound was detected during incense and mosquito coil combustion as well as grilling bacon experiments. Group 2 carcinogenic compounds (e.g. chrysene, benzo(b)fluoranthene) and three- (fluorene, phenanthrene) and four-ring PAHs (fluoranthene, pyrene), which are classified as Group 3 carcinogenic compounds (i.e. not classifiable due to insufficient information)69, were also detected. On the contrary, for all the investigated indoor sources concentrations of benzo(a)anthracene, indeno(1,2,3-cd)pyrene, dibenzo(a,h)anthracene, benzo(k)fluoranthene, and benzo(g.h.i)perylene resulted under of the limit of detection (LOD). PAH mass fractions on PM10 (expressed as ng ng−1), calculated as described in the section “Data post-processing”, are reported in Table 3. Such values ranged between 10–7 and 10–5 which are, again, not negligible and similar to those typically measured at outdoor urban sites72. As already mentioned for HMs, a proper comparison with the few data available in the literature cannot be carried out. Indeed, even if PAHs were measured in cooking-generated particles66, during candle burning event41, incense burning event67, and mosquito coils burning event53, the mass factions and size-resolved analyses were not provided.

In Fig. 5 mass distributions (normalized to the total concentration) of detected PAHs in freshly emitted particles from candle, incense, mosquito coil combustions and grilling bacon are reported. Particles emitted by candles present a predominant contribution of sub-micrometric range, indeed such sub-micrometric contributions for phenanthrene, anthracene, fluoranthene, pyrene, chrysene, benzo(b)fluoranthene resulted equal to 80%, 88%, 75%, 75%, 76%, and 65%, respectively, with a not negligible contribution (except for benzo(b)fluoranthene) of the ultrafine particle range (25–38%). For incenses, the contribution of the sub-micrometric range to the total mass of the emitted PAHs resulted in the range 43–50%. In particular, 46% of benzo(a)pyrene mass, which is the most critical PAH for human health, was detected in the sub-micrometric range. Benzo(a)pyrene was also detected in particles generated by mosquito coil combustion and grilling bacon with sub-micrometric particle contributions of 67% and 100%, respectively. Particles released from mosquito coil combustion also showed the presence of phenanthrene, anthracene, fluoranthene, pyrene, chrysene, and benzo(b)fluoranthene (with the contribution of the sub-micrometric particles from 56 to 80%). Similarly, a huge contribution of the sub-micrometric range (54–100%) was also detected for other PAHs emitted by grilling bacon (fluorene, phenanthrene, anthracene, fluoranthene, pyrene, chrysene, benzo(b)fluoranthene).

Fig. 5
figure 5figure 5

Mass distributions (normalized to the total concentration) of PAHs in freshly emitted particles from candle (a), incense (b), mosquito coil combustions (c) and grilling bacon (d) (Sampling 2).

Full size image

Estimated deposited concentration of chemical compounds

The results provided in the present study represent a novel finding since mass fractions and size-resolved chemical compositions of airborne particles emitted by some widely-used indoor sources are shown for the very first time. This is a key information in view of a more detailed analysis of the indoor air quality and related risk assessment. Indeed, a direct implication of the data here shown is the different deposited concentrations of compounds in the lungs when adopting actual mass distribution of the compounds. To this end, in Table 4 the ratios between the deposited concentrations, evaluated according to the two different methods (size-resolved and constant), calculated according to the Eq. 1 and the hypothesis of the section “Estimation of the deposited concentration of chemical compounds”, are summarized for each compound and indoor source: the ratios resulted larger than 1 for 32 of the 54 (59%) compounds detected, in other words, assuming constant mass fraction of the chemical compounds over the entire particle size range, the dose received by exposed persons would be underestimated. For some compounds the ratios resulted much larger than 1 (e.g. 3.3 for As in the case of mosquito coil combustion), then likely resulting in a significant underestimate of the actual dose. The reason for the large deviation between the two approaches (size-segregated mass fraction vs. constant mass fraction) is due to the deposition fractions (alveolar and tracheobronchial) of the particles which present a peak in the ultrafine range. Thus, chemical compounds with a significant contribution in that size range will present a higher deviation with respect to the dose evaluated considering a constant mass fraction. The As emitted by mosquito coil combustion is the most critical example: its contribution is mostly due to the size range 16.7–31.3 nm (Fig. 2a) where the deposition fraction is around 60%; indeed, the deposited concentration of As (with respect to the total concentration of As), when applying the size-resolved approach, resulted equal to 59%; on the contrary, considering the constant mass fraction (2.93 × 10–6, Table 2), the deposited concentration of As was 18% of the total concentration of As (then, the ratio is 3.3).

Table 4 Ratios between the deposited concentrations of chemical compounds (for each source) evaluated according the two different approaches adopted to estimate the mass fractions of chemical compounds: size resolved mass fraction and constant mass fraction.
Full size table

Despite the key findings provided in the study and the related very important implications, the reader should be aware of the criticalities and limitations of the study in terms of both methodological assumptions and result exploitability.

As concerns the methodology, we point out that, mass fractions of HMs and PAHs were calculated considering PM10 concentration measured by the ELPI + ™: despite good agreement was found in previous studies with gravimetric measurements73,74, the mass concentration it provides is an indirect estimate as it starts from the measurement of the particle number distribution. Thus, a larger uncertainty is expected with respect to the measurement of the particle mass concentration through the reference method (i.e. gravimetric method75). Nonetheless, in the experimental setup we adopted it was actually not possible, for practical reasons, to perform a parallel gravimetric sampling in the chamber.

Concerning the data exploitability, we highlight that the results we provided cannot be considered immediately applicable to other similar sources, indeed, the composition of the sources (e.g. additives to obtain a specific fragrance, scent, etc.) can strongly affect the emission of particle-bound compounds39,40,41. This is even more critical for cooking activities due to the abovementioned several influence parameters43,45,63 affecting the emission: considering all the possible cooking types and foods would have been unfeasible. Moreover, the cooking experiments would have been critical also from a methodological perspective, indeed, differently from other sources (that were left to burn with no handling needed), food preparation needs frequent handling, and its duration is shorter than other combustion sources here tested.

Future developments of the study should then be focused on trying to extend the analysis here performed to different compositions/types of the indoor sources here investigated (e.g. including different compositions, fragrances, scents of candles, mosquito coils, incenses as well as cooking activities with different stoves, foods, preparations). Moreover, in view of prospective risk analyses due to the exposure to indoor combustion sources, risk models able to account for the size-dependent chemical composition should be carried out. Indeed, to date, a lung cancer risk model able to take into account for the contribution of sub-micrometric particles was recently developed32, but it is still based on the assumption of chemical composition invariant to the particle size (constant mass fraction); it should be properly modified to include the size-dependent mass-fraction.

Conclusions

In the present paper for the very first time the size-resolved chemical characterization of airborne particles freshly emitted by widely used indoor sources, i.e. incenses, candles, mosquito-coils and cooking activities (grilling bacon), was carried out. In particular, size-dependent particle-bound heavy metal and polycyclic aromatic hydrocarbon contributions to the PM10 were provided thanks to a particle collection through an electric low-pressure impactor and consequent gas chromatography-mass spectrometry (GC–MS) and atomic emission spectrometry (AES) analyses.

The results of the study revealed that all the sources emit fresh particles with a significant contribution of PM1 fraction (with respect to PM10).

As concerns the heavy metal chemical analysis, Group 1 carcinogenic compounds (Cd, Cr(VI), As, and Ni) were only found in particle samples from mosquito coil combustion and grilling bacon. The calculated mass fractions were not negligible as they were mostly in the 10–6–10–5 range and, moreover, the contribution of sub-micrometric particles to the total mass of emitted heavy metals was found extremely high: as an example, As, Ni, Cr(VI) emitted by the mosquito coils were almost exclusively due to the sub-micrometric particle range as well as As, Cd, and Cr(VI) emitted by grilling bacon.

In terms of PAHs, different compounds, including Group 1 (i.e. benzo(a)pyrene) and Group 2 carcinogenic compounds, were found in the freshly-emitted particles of the investigated sources with mass fraction in the 10–7–10–5 range. A high contribution of sub-micrometric range was detected for most of the investigated PAHs.

The main implication of the results here shown is the possible underestimation of deposited concentration of chemical compounds (and then of the related dose of exposed people) when a constant mass fraction over the entire size range is adopted. Indeed, we showed that by adopting a constant mass fraction, instead of a size-dependent mass fraction of the compound here obtained, the deposited concentrations of chemical compounds were mostly underestimated: such underestimation resulted by a factor of 3 if the amount of the compound is mostly located in the ultrafine range (as we found for the As due to mosquito coil combustion).

The information provided can be extremely useful for detailed health risk assessments since they allow avoiding oversimplified assumptions based on considering the chemical composition of the emitted particles invariant to the particle size.