Emission factors for PM₁₀ and polycyclic aromatic hydrocarbons (PAHs) from illegal burning of different types of municipal waste in households

It is a common practice in developing countries and in some regions of Europe that solid wastes generated in households (e.g. plastic beverage packaging and other plastic wastes, textile wastes, fibreboards, furniture, tyres, and coloured-paper waste) are burned in wood- or coal-fired stoves during the winter months. In Europe, the types and volume of municipal waste burned in households is virtually unknown because these activities are illegal and not recorded, with the exception of a few media reports or court cases. Even though particulate emissions from illegal waste burning pose a significant hazard to human health due to the combination of excessive emission factors (EFs) and uncontrolled chemical composition, there is scarce information on the specific EFs for PM₁₀ and polycyclic aromatic hydrocarbons (PAHs) in the scientific literature. In this work, controlled combustion tests were performed with 12 different types of municipal solid waste, and particulate emissions were measured and collected for chemical analysis. Absolute EFs for PM₁₀ and PAHs as well as the benzo(a)pyrene (BaP) toxicity equivalent of the latter are reported for the first time for the indoor combustion of 12 common types of municipal solid waste that are frequently burned in households worldwide. It was found that the PM₁₀ emission factors from the combustion of wood-based waste samples were about twice that of firewood, whereas EFs in the range of 11–82 mg g⁻¹ (a factor of 5–40 times higher than that of dry firewood under the same conditions) were obtained for different types of plastic waste. The latter were also found to emit exceptionally high quantities of PAHs, by a factor of 50–750 more than upon the combustion of dry firewood under the same conditions. Since the more toxic 4–6 ring PAHs were predominant in the particulate emission from plastic waste burning, BaP equivalent toxicity was up to 4100 times higher than that from wood combustion.

1 Introduction

Billions of people use solid fuels (wood, coal or agricultural waste) as the main source of household energy worldwide (Anenberg et al., 2013). It is well known that fossil fuel combustion and biomass burning are the two most important sources of fine particulate matter in the atmosphere (Simoneit et al., 2002). Karagulian et al. (2015) estimated that domestic fuel burning (wood, coal and gas) might contribute up to 32 % of PM_2.5 emissions and up to 45 % of PM₁₀ emissions in central and eastern Europe (globally 20 % and 15 %, respectively). In Europe solid fuels (primarily wood but in some countries also coal) are extensively used for home heating. Residential wood combustion was found to be the main emission source of fine particles all over Europe in winter (Tissari et al., 2008; Puxbaum et al., 2007; Gelencsér et al., 2007; Marmureanu et al., 2020). Emissions from wood burning were estimated to contribute to about 30 % of the PM_2.5 fraction in Portugal (Goncalves et al., 2012). In Lombardy somewhat lower contributions (5 %–25 % to PM₁₀) were estimated (Pastorello et al., 2011), even in the city of Milan (Piazzalunga et al., 2011). Based on recent studies fine particles from wood combustion may cause severe health effects such as lung cancer, chronic lung and heart diseases (Rafael et al., 2015; Bai et al., 2015); contribute to visibility reduction (Pipal and Satsangi, 2015); and even modify synoptic conditions (Tiwari et al., 2015).

On top of the significant emissions from the burning of solid fuels in households, there is another existing yet largely unknown source of atmospheric pollution, the illegal burning of municipal solid waste in households in several countries of the world. In Hungary, two independent polls have recently revealed that 2 %–10 % of the population burn their wastes in stoves on a regular basis (Kantar Hoffman Company, 2020; Századvég Foundation, 2018). The major types of waste burned were treated wood (furniture, oriented strand board (OSB), hardboard, painted wood, plywood), clothes, plastics, tyres, and used oil. In many villages in Romania the burning of household waste is an everyday practice due to the lack of an organised waste collection system. Burning any type of waste poses excessive risks to the health of people living in those areas since a plethora of toxic, carcinogenic or mutagenic compounds are emitted in immense quantities compared to the burning of authorised solid fuels such as dry fuel wood or high-quality coal (Lemieux et al., 2004; Estrellan and Lino 2010; Gullett et al., 2010; Wiedinmyer et al., 2014). Since burning any kind of municipal waste in households is strictly prohibited all over Europe, understandably such emissions are not included in many emission inventories (Wiedinmyer et al., 2014). Being an illegal activity, even its magnitude is mostly unknown apart from in a few public reports of NGOs, some media coverage and a few documented court cases. More interestingly, there has been no single systematic study on EFs for burning abundant types of solid waste in households in the scientific literature. There are only a small number of scientific papers on the burning of different types of (mostly plastic) waste with the specific focus of finding organic tracers for their tracking in atmospheric particulate matter (Simoneit et al., 2005; Tomsej et al., 2018; Gu et al., 2010; Kumar et al., 2015). This is surprising in the light of the fact that worldwide solid-waste burning is a known and important source of particulate pollution. There are only a handful of studies that report EFs of PM_2.5 for the open burning of municipal mixed solid waste, on the order of 10 g kg⁻¹ (Christian et al., 2010; Park et al., 2013; Jayarathne et al., 2018).

Polycyclic aromatic hydrocarbons (PAHs) are among the most hazardous combustion products due to their carcinogenic and mutagenic nature (Bjorseth, 1983; Kim et al., 2013). Among them five-ring PAHs such as benzo(a)pyrene, benzofluoranthenes and dibenzo(a,h)anthracene are known as highly carcinogenic compounds. The EFs of PAHs may vary for different waste types, but such data are scarcely available in the scientific literature though absolute EFs are essential to assessing the impact of residential waste burning on air quality. To the best of our knowledge there has been only one study reporting PM and PAH emission factors for the co-combustion of polyethylene (PE) and polyethylene terephthalate (PET) waste with beech logs (Tomsej et al., 2018). In this study EFs for total PAH and PM₁₀ were determined for 12 waste types and compared to those of wood burning.

2 Material and methods

2.1 Waste samples for combustion tests

The types of solid-waste specimens for the combustion tests were selected based on their abundance in households and available information on illegal waste-burning practices in Hungary. PET, polystyrene (PS), polypropylene (PP), PE, polyvinyl chloride (PVC) polyurethane (PU), acrylonitrile butadiene styrene (ABS), tyre material (TR), OSB, laminated melamine low-density fibreboard (LDF), rag (RAG), paper (PAP) and firewood (WOOD) were selected for waste combustion tests (Fig. 1).

Figure 1 Solid-waste specimens prepared for combustion tests.

The PET waste samples were prepared from beverage plastic bottles (volume 1.5 and 2 L) without caps and labels. The PS waste samples were hard cups or pots of dairy products (yoghurt and pudding), without aluminium foil caps, and pieces of expanded PS insulation board (80 mm $\times \sim \mathrm{100}$ mm $\times \sim \mathrm{100}$ mm). The hard and expanded PS waste sample specimens were burned separately. The PP waste sample specimens were a mixture of plastic cups and pots of dairy products (sour cream and pudding), without paper labels and aluminium foil, and quartered plastic trays of meat. The PE waste specimens were prepared from a mixture of high- and low-density polyethylene (HDPE and LDPE). The HDPE and LDPE fractions consisted of plastic caps of beverage bottles and pieces of various foil and plastic bags, respectively. The PVC waste samples consisted of soft packaging, small pieces of vinyl flooring and hard plastic water pipes. The PU waste samples consisted of pieces of packing sponge (average size 120 mm × 100 mm × 15 mm). The ABS samples were shredded pieces of stands of computer monitors (average size 40 mm × 40 mm × 20 mm). The OSB samples consisted of slices of OSB material of different brands (average size 120 mm × 100 mm × 20 mm). The LDF samples were pieces (average size 130 mm × 100 mm × 20 mm) of different fibreboards including coloured laminated coating and plastic borders. The TR sample specimens consisted of pieces of a new and old tyres of a van and a passenger car, respectively (average size 80 mm × 40 mm × 15 mm). The RAG samples consisted of a mixture of cotton, polyester and polyamide fabrics from different clothes. Ball-shaped specimens (average weight ∼ 70 g) of two types of PAP samples (colourful glossy-coated paper and uncoated paper from advertising flyers and newspaper) were burned separately. The WOOD samples consisted of pieces of logs (average weight ∼ 130 g) of Turkey oak (Quercus cerris) and black locust (Robinia pseudoacacia). All sample specimens were weighted with an analytical balance before the combustion tests. Authorisation for conducting controlled waste-burning tests was granted by the Department of Environmental Protection and Nature Conversation of Veszprém District.

2.2 Experimental conditions of the combustion tests

Combustion tests were carried out in a commercially available cast-iron stove (type – Servant S114, heating power – 5 kW). The stove was heated up with smouldering charcoal for ∼ 1 h prior to the start of combustion tests which produced a very low particulate emission baseline throughout the duration of the measurements. During each combustion test 1–10 aliquots of solid waste were burned depending on the emission characteristics of the given waste type. The mass of each sample specimen was measured with an analytical balance and was recorded. The stove used was a commercially available model which allowed for setting the air supply to enter the combustion chamber of the stove through an adjustable slit. There were two endpoints of this slit; thus either a larger (high air supply ratio) or a smaller (low air supply ratio) volume of fresh air was allowed into the combustion zone. Each type of waste (except PAP) was burned at high, combined and low air supply, resulting in different temperatures and conditions inside the oven, but this was largely independent from the type of waste combusted in the stove at any given air supply ratio. The temperature of flue gas was measured before and after each measurement by a K-type thermocouple thermometer (maximum temperature 1000 ^∘C; testo 925) in the stack 11 cm above the exhaust opening of the stove. Between the different sample runs the stove and the stack were heated up to above 700 ^∘C for a minimum of 10 min to minimise cross-contamination between combustion tests with different waste types. The temperature of the flue gas served as an indicator of the experimental conditions of each combustion test run. The temperature values of the flue gas during the experiments were stable as the mean temperature values were 299 ^∘C (SD 11 ^∘C) and 233 ^∘C (SD 10 ^∘C) at high and low air supply, respectively (see Fig. S1).

The mixing ratios of CO₂ and O₂ in the flue gas were measured with a CO₂−O₂ analyser (Servomex) at the end of the stack. At the end of the stack a small amount of the flue gas was introduced through brass tubing (5 mm i.d.) into a dilution unit of a volume of 80 L and was diluted with ambient air. The concentration of CO₂ in the diluted flue gas was also monitored with a CO₂ analyser (SensAir) at the inlet of the filter sampling head. The dilution ratio was determined from the ratio of the measured concentrations of CO₂ at the end of the stack and in the dilution unit (taking into account the ambient CO₂ concentration as well). The mean dilution ratio was about 81.2 (SD 12) (see Fig. S2), which is independent of the air supply but may depend on the progress of the burning process. The physical processes that took place during the applied sampling procedure were very similar to those occurring after the emission of smoke from the chimney into open air as in both cases the smoke cools and dilutes rapidly.

Figure 2 Example (PS-F9 sample) of the variations in the measured CO₂ and O₂ mixing ratios at the end of the stack and the CO₂ mixing ratio in the diluted flue gas during combustions tests.

Figure 2 shows the variation in the CO₂ and O₂ mixing ratios at the end of the stack and the CO₂ mixing ratio in the diluted flue gas for a PS sample. Each peak corresponds to the burning of a single aliquot of the solid waste. The collections of each PM₁₀ aerosol sample were started after reaching stabile baselines of CO₂ and O₂ concentration values and were finished after a return to stable baselines at the end of combustion of all aliquots of solid wastes.

PM₁₀ aerosol samples were collected on quartz filters of 150 mm in diameter (Advantec QR100 quartz fibre, binder free) with a high-volume aerosol sampler (flow rate 32 m³ h⁻¹; Kalman System Co., Hungary) at the dilution unit. Blank samples were also collected for each waste type. The blanks represent background measurements during which only charcoal was burned. The sampling times of the blanks were comparable with those of the samples. For each waste type one blank sample was collected with combined air supply settings. The quartz filters were conditioned at a temperature of 20±1 ^∘C and relative humidity (RH) of 45 %–50 % for 3 d and were weighed in an isolated weighing room before and after the aerosol samplings according to the European standard (MSZ EN 12341:2014). The parameters (RH, temperature) were measured and collected by a data acquisition system. The weighted filters were stored in glass petri dishes (preheated at 450 ^∘C) prior to sampling, whereas the exposed filters were stored in the freezer in glass petri dishes wrapped in aluminium foil until conditioning and measurements.

Table S1 lists the key parameters of the combustion tests, including the type and mass of sample specimens, the air supply settings, the number of test burns (by air supply settings), and the measured blank-corrected PM₁₀ mass on each filter. The mass of the waste sample specimens burned was optimised in preliminary tests to yield PM₁₀ concentrations of about the same magnitude in each combustion test. Since different waste types yielded vastly different particulate emissions upon burning (e.g. PS, PP, PE, PVC, PU, ABS and TR were superemitters compared to PET, OSB, LDF, RAG, PAP and WOOD), this step was necessary to avoid massive overloading of the filters and the measuring instrumentation. Thus the measurements were comparable and the measured mass of PM₁₀ on filters was kept in the range of 5.4 and 37.2 mg for all combustion tests. It should be noted that the mass of the specimens may affect the burning mechanism and thus the measured EFs. However, the observed consistency with previous EFs indicates that the approach and the results are reliable. The EFs were calculated using the weight of the waste specimens put into the stove. Here we note that the ash content of the plastics reported by Zevenhoven et al. (1997) (LDPE, HDPE, PP, PVC, PS) is below 3 %. It can be assumed that the ash content of other combustible plastics is similarly low.

2.3 Analysis of PAHs in the filter samples

The quantity of PAHs in the filter samples was determined by analysing a filter spot with a diameter of 1.4 cm. First PAHs were extracted with 4.5 mL of hexane in an ultrasonic bath for 15 min; then the extract was filtered through ashless quantitative (Grade 44, Whatman, UK) filter paper and cleaned on 0.5 g of Florisil adsorbent. The cleaned extract was gently evaporated to dryness and re-dissolved in 1.0 mL of acetonitrile. The concentration of 15 PAHs out of the 16 Environmental Protection Agency (EPA) priority pollutant PAHs was determined in the final aliquot by high-performance liquid chromatography (HPLC; Series 200, PerkinElmer, Shelton, CT, USA) on an Inertsil ODS-P 5 µm, 4.6 × 150 mm (GL Sciences Inc., Tokyo, Japan) column by using water–acetonitrile binary gradient elution and time-programmed fluorescence detection.

The ambient temperature affects the gas-to-particle partitioning of the compounds especially those of the semi-volatile ones. During the sampling the ambient temperature varied between 16.1 and 26.0 ^∘C, on average it was 20.3 ^∘C; which is higher than typical ambient temperatures during the heating season. This implies that some semi-volatile species might be lost to the gas phase upon sampling; thus measured EFs for PM₁₀ and total PAHs are likely lower-bound estimates for real-life conditions. It should be added that the standard procedure of the gravimetric PM₁₀ measurements includes the conditioning of the exposed filters for 48 h at 20 ^∘C (and RH = 50 %) which might also result in some loss of semi-volatile compounds. Concerning PAHs, at the sampling temperatures some of the more volatile two- and three-ring compounds might have been lost to the gas phase as compared to the case at ambient temperatures in winter, and they were likely to some extent underrepresented in the filter samples. These potential losses, however, do not influence the conclusions regarding PAH relative toxicity due to the very low toxicity of the two–three-ring compounds.

The extraction efficiency was tested by sequential extraction of the same filter spots in two consecutive steps. In the second extract the quantity of PAHs was found to be only 1 %–3 % of that present in the first extract. These results evidenced that under the experimental conditions applied (solvent-to-filter ratio, time of extraction) 97 %–99 % of PAHs were extracted in the first step. Therefore, the extraction was not continued and the sum of the quantity of PAHs in the two extracts was considered to be 100 %. Because of the high efficiency of the first extraction step, the samples were extracted only once and the results were corrected with the extraction efficiency (97 %–99 %). The recovery of PAHs from the extract was also studied by performing the clean-up procedure on diluted PAH standard solutions. The average recovery of 98 % (SD = 8.9 %) was obtained for the 12 PAHs investigated in this study. The analytical results were corrected for both extraction efficiency and recovery of the sample preparation.

3 Results and discussion

3.1 Emission factors for PM₁₀ from waste burning

The PM₁₀ EFs of the fuel wood and different solid wastes were found to be highly variable (from ∼ 2.1 to ∼ 81.5 mg g⁻¹), the lowest values being representative of the wood burning. In the literature, the PM_2.5 EF of wood burning in traditional woodstoves varies between ∼ 2.8 and ∼ 13.3 mg g⁻¹, whereas the PM₁₀ EF from a chimney-type (eco-labelled) woodstove is smaller (∼ 1.1–2.9 mg g⁻¹, Querol et al., 2016). Kistler et al. (2012) investigated the PM₁₀ EF of 12 central-European wood types from a chimney-type wood stove and found that it varied between 0.3 and 3.6 mg g⁻¹. The values for turkey oak and black locust were 1.0 and 1.1 mg g⁻¹, respectively. For the burning of oak logs in two different stoves, Schmidl et al. (2011) reported a PM₁₀ EF of 1.2–1.4 mg g⁻¹. In our measurements the PM₁₀ EF from the burning of mixed turkey oak and black locust varied between 1.3 and 3.2 (on average 2.1 mg g⁻¹, SD 0.7 mg g⁻¹), indicating that the obtained EFs agree quite well with those reported in recent studies.

Compared to the EFs of burning plastics, the PM₁₀ EFs of burning wood-based materials were found to be largely similar to each other, though the burning of OSB and LDF (both contain glue, the LDF surface coating and edge tape as well) release twice as much PM₁₀ (∼ 3.2 mg g⁻¹, SD 1.3 mg g⁻¹, for LDF and 5.2 mg g⁻¹, SD 1.4 mg g⁻¹, for OSB) into the atmosphere as dry firewood under similar burning conditions (Fig. 3).

Figure 3 PM₁₀ emission factors for burning of different waste types relative to the burning of dry fuel wood with a fixed experimental setup.

The EF of burning RAG (8.7 mg g⁻¹, SD 0.22 mg g⁻¹) is between the EF of PET and wood-based materials. This is not surprising as on average 46 % of the weight of the burned RAG was cotton and 43 % was PET-based material. The EF for PET (on average 11 mg g⁻¹, SD 1.6 mg g⁻¹) is roughly 5 times higher than that of WOOD. The only available EFs reported for PET co-combustion in a boiler are between 1.8 ± 0.3 and 5.8 ± 0.5 mg g⁻¹ depending on the combustion conditions (Tomsej et al., 2018). The differences between our and the reported EF values can be explained by the vastly different burning conditions (20 and 5 kW nominal power output for the boiler and the stove, respectively, and no fuel wood was co-fired in our experiments unlike in those reported by Tomsej et al., 2018). Our findings indicate that the burning of a given mass of PE and/or PU releases on average 1 order of magnitude more PM₁₀ into the atmosphere than wood burning. The combustion of PP, tyre material and PVC is even more polluting as 16 times more PM₁₀ is released into the atmosphere when burned under similar conditions. Somewhat surprisingly, our results also agree well with the few available EFs (for PM_2.5) for the open-air burning of mixed (presumably mostly plastic) waste reported in the literature (10.5 and 7.37 mg g⁻¹) (Christian et al., 2010, and Jayarathne et al., 2018, respectively). The tyres already contain soot and inorganic fillers which might contribute to the increased PM₁₀ emission of burning. The styrene-containing materials yielded the highest EF among the investigated components. The PM₁₀ EF for the burning of PS was on average 53.1 mg g⁻¹ (SD 15 mg g⁻¹) and that of the expanded polystyrene was even higher (81.5 mg g⁻¹, SD 27 mg g⁻¹). On average the highest PM₁₀ EF was obtained for the copolymer of styrene with butadiene and acetonitrile. The absolute EFs for all waste types are summarised in Table 1.

Table 1 The absolute emission factors of PM₁₀ (mg g⁻¹), total PAHs (mg PAHs kg⁻¹ of fuel) and total PAHs expressed in BaP toxicity equivalent (mg kg⁻¹ of fuel) from wood burning and residential waste burning.

3.2 Emission factors of PAHs from waste burning

The EFs of total PAHs (sum of 12 EPA priority pollutant PAHs from three-ring phenanthrene to six-ring indeno(1,2,3-cd)pyrene) during the combustion of different waste types were determined (Table 1). Both wood and wood-based waste types were characterised by EFs below 5 mg kg⁻¹. The lowest EF was obtained for WOOD followed by PAP, LDF and OSB. EFs at least an order of magnitude higher were measured for the other waste types investigated in this study. The average values covered a wide range from 21 mg kg⁻¹ (SD 19 mg kg⁻¹) for RAG to 257 mg kg⁻¹ (SD 103 mg kg⁻¹) for ABS. The average total PAH EFs were similar for RAG, PE, TR and PET, while they were higher for PVC, PU and PP but remained under 100 mg kg⁻¹ for all of these waste types. The highest EFs were measured during the combustion of PS and ABS exceeding 100 mg kg⁻¹. The relatively high SD values are the consequence of varying conditions (air supply) applied during the burning experiments as described in Sect. 2.2.

The EFs of the individual PAHs (Table S2 in the Supplement) are different for wood and waste burning. While the combustion of WOOD resulted in the emission of primarily phenanthrene and four-ring PAHs (fluoranthene, pyrene, benzo(a)anthracene and chrysene) and the contribution of five- and six-ring compounds was only a few per cent, the latter compounds accounted for 14 %–45 % of the total PAH emission in the different waste types. This shift in the chemical profile has a considerable effect on the toxicity of the particles emitted as discussed later. The availability of PAH emission data from the burning of any given type of waste is rather limited in the literature. Tomsej et al. (2018) studied the co-combustion of wood and polyethylene plastics (PE and PET). They burned 93 w∕w % beech log and 7 w∕w % PE in a 20 kW boiler and obtained total PAH EFs of 11 mg kg⁻¹ (SD 1.2 mg kg⁻¹) and 16 mg kg⁻¹ (SD 1.6 mg kg⁻¹), in the particulate phase for the 12 compounds investigated in our study under nominal- and reduced-output conditions, respectively. The corresponding values for the co-combustion of 93 w∕w % beech log and 7 w∕w % PET were 8.5 mg kg⁻¹ (SD 1.0 mg kg⁻¹) and 19 mg kg⁻¹ (SD 3.6 mg kg⁻¹). These results indicate the influence of operational conditions on the emission factors of PAHs. Furthermore, the total PAH EF values obtained for the mixtures of beech–PE and beech–PET by Tomsej et al. (2018) were between the total PAH EFs obtained for WOOD (0.34 mg kg⁻¹, SD 0.19 mg kg⁻¹) and PE (31 mg kg⁻¹, SD 23 mg kg⁻¹) or WOOD and PET (32 mg kg⁻¹, SD 14 mg kg⁻¹) in our study, respectively. Maasikmets et al. (2016) studied the EFs of four PAHs emitted from burning municipal solid waste mixed with wood in domestic heaters. They reported emission factors of 0.41, 0.18, 0.12 and 0.10 mg kg⁻¹ for benzo(a)pyrene, benzo(b)fluoranthene, benzo(k)fluoranthene and indeno(1,2,3-cd)pyrene, respectively. As a comparison, Lemieux (1997) found EFs of 1.4, 1.86, 0.67 and 1.27 mg kg⁻¹ for the same compounds from the barrel burning of household waste. These emission factors are comparable to those found in our experiments and summarised in Table S2.

Figure 4 Emission factors of total PAHs for different waste types relative to the burning of dry fuel wood with a fixed experimental setup.

In Fig. 4 the EFs for total PAHs are shown for different waste types relative to wood burning. It is clearly visible that even the combustion of wood-based waste types (PAP, LDF, OSB) generated considerably more (by a factor of 3–8) PAHs than the burning of wood. The PAH EFs from the burning of RAG, PE, TR and PET were higher by more than a factor of 50 relative to wood combustion, while those for PVC, PU, PP, PS and ABS were higher by well over a factor of 100. These extreme emission factors underline the severe hazard associated with the illegal burning of solid wastes in households.

From the PM₁₀ mass collected and the quantity of total PAHs determined on the filter, the PAH content of the PM₁₀ fraction generated from each combustion experiment was calculated. The lowest mean PAH content of the PM₁₀ aerosol was obtained for WOOD burning (0.16 µg mg⁻¹, SD 0.08 µg mg⁻¹) followed by the wood-based wastes OSB (0.52 µg mg⁻¹, SD 0.20 µg mg⁻¹), PAP (0.53 µg mg⁻¹, SD 0.40 µg mg⁻¹) and LDF (0.53 µg mg⁻¹, SD 0.12 µg mg⁻¹). For the other waste types the mean PAH content increased from 0.70 µg mg⁻¹ (SD 0.16 µg mg⁻¹) (TR) through 1.1 µg mg⁻¹ (SD 0.52 µg mg⁻¹) (PE), 1.4 µg mg⁻¹ (SD 0.44 µg mg⁻¹) (PVC) and 1.8 µg mg⁻¹ (SD 2.0 µg mg⁻¹) (RAG) to 2.5 µg mg⁻¹ (SD 1.8 µg mg⁻¹) (PS). The highest PAH content was observed when burning PET (3.0 µg mg⁻¹, SD 1.5 µg mg⁻¹), PU (3.0 µg mg⁻¹, SD 0.92 µg mg⁻¹), ABS (3.1 µg mg⁻¹, SD 0.86 µg mg⁻¹) and PP (3.1 µg mg⁻¹, SD 1.5 µg mg⁻¹). It is clearly visible that the relative share of PAHs in PM₁₀ emitted is typically an order of magnitude higher in plastic combustion as compared to wood burning. The relatively high PM₁₀ but low PAH emission factors of tyre combustion (as shown in Table 1) can be explained by the presence of inorganic components in the material of tyres (e.g. inorganic fillers).

It is important to note that in addition to the total PAH emission factors, the emission profile (i.e. the relative contribution of individual PAHs) should also be investigated when health effects are studied since the toxicity of PAHs varies from compound to compound. Among the 16 EPA priority pollutant PAHs five-ring compounds such as benzo(a)pyrene, dibenzo(a,h)anthracene and benzofluoranthenes; the four-ring benzo(a)anthracene; and the six-ring indeno(1,2,3-cd)pyrene were found to be the most carcinogenic compounds (Nisbet and LaGoy, 1992; Safe, 1998, and references therein). In order to compare the toxicity of different samples, toxic equivalency factors have been defined for PAHs. On this scale benzo(a)pyrene (BaP) has a value of 1 and the toxicity of the other 15 EPA priority pollutant PAHs varies from 0 to 1 (Nisbet and LaGoy, 1992; Safe, 1998, and references therein; EPA, 2007). There are numerous such toxicity scales, but for the sake of comparability the toxic equivalency factors of the EPA (2007) were applied in our study similarly to the work by Tomsej et al. (2018). On this scale benzo(a)pyrene is taken into account with a toxicity equivalency factor of 1; benzo(a)anthracene, benzofluoranthenes, dibenzo(a,h)anthracene and indeno(1,2,3-cd)pyrene with a factor of 0.1; and chrysene with a factor of 0.01, and the toxicity of the other PAHs is neglected. By applying the BaP toxicity equivalency factors the overall toxicity of the PAHs emitted during wood burning and the combustion of different wastes was compared (Table 1). The EFs of total PAHs as expressed in benzo(a)pyrene toxicity equivalent covered a very wide range of more than 3 orders of magnitude. The burning of 1 kg of wood (oak) produced PAHs with a total toxicity equivalent of 4.2 µg of BaP (SD 5.5 µg). The combustion of the same amount of wood-based waste led to the emission of PAHs equivalent to about 30 times more BaP. The combustion of PE, RAG, TR, PET and PU resulted in the emission of PAHs 280–800 times more toxic than the PAHs released from the burning of the same amount of wood. The toxicity equivalent EFs for PE and PET reported in Table 1 (1.2 mg kg⁻¹, SD 1.1, and 2.2 mg kg⁻¹, SD 1.4 mg kg⁻¹, respectively) were similar to those found by Tomsej et al. (2018) for mixtures of the above plastics with beech. They reported toxicity equivalent EFs of 2.3 ± 0.5 mg kg⁻¹ and 1.1 ± 0.2 mg kg⁻¹ for the mixture of PE and beech under nominal-output and reduced-output conditions, while the corresponding values for the mixture of PET and beech were 1.8 ± 0.6 and 1.8 ± 0.3 mg kg⁻¹, respectively. Toxic PAHs in the highest quantity were emitted from the combustion of PVC, PS, PP and ABS. The emission factors of total PAHs expressed in BaP toxicity equivalent were more than 3 orders of magnitude (!) higher for the combustion of these plastic wastes than the value obtained for wood burning. The very high emission of toxic PAHs from the combustion of plastic wastes as compared to wood burning may follow from the synergic combination of three factors:

1. On a per mass basis significantly more PM₁₀ is emitted from the combustion of plastic waste than from the burning of dry fuel wood (see Fig. 3, Table 1).

2. PM₁₀ released from plastic waste combustion contains more PAHs per unit mass than PM₁₀ generated from wood burning.

3. The distributions of PAHs emitted from plastic waste burning differ markedly from those produced in wood burning. The combustion of plastic wastes results in the formation of the significantly more carcinogenic four–six-ring compounds.

It should be added, however, that the burning conditions affect the profile and total quantity of PAHs considerably. The low quantity of five-ring and six-ring PAHs in the particles emitted from wood burning resulted in very low toxicity equivalent EFs and, consequently, led to high relative toxicity of the other wastes as compared to wood.

4 Conclusion

The PM₁₀ emission factors, total PAH emission factors and benzo(a)pyrene toxicity equivalent total PAH emission factors were determined for waste burning under controlled combustion conditions for the first time. The EF values were established for the combustion of wood, paper, LDF, OSB, rag, PET, PE, PU, PP, tyre material, PVC, PS and ABS, which are all abundant waste types frequently burned in households and open fires in developing countries and even in Europe. The determined PM₁₀ EF for wood burning was about 2.1 mg g⁻¹ (SD 0.7 mg g⁻¹) which shows good correlation with the results of recent studies. The PM₁₀ EFs from the burning of other wood-based materials (PAP, LDF and OSB) were only slightly higher, while from the combustions of RAG and PET were about 5 times higher than the EF of wood burning. The EFs for combustions of PE and PU were about 18 mg g⁻¹ (SD 7.2 mg g⁻¹) and 22 mg g⁻¹ (SD 8.4 mg g⁻¹), indicating a nearly 10-fold increase in PM₁₀ emission from burning of these waste types. Even higher PM₁₀ EFs were determined during the combustion of PP, TR and PVC, which were 33 mg g⁻¹ (SD 18 mg g⁻¹), 35 mg g⁻¹ (SD 6 mg g⁻¹) and 35 mg g⁻¹ (SD 10 mg g⁻¹), respectively. Finally, the highest PM₁₀ EF values were established for the combustion of PS and ABS, which may reach 53 mg g⁻¹ (SD 15 mg g⁻¹) and 82 mg g⁻¹ (SD 27 mg g⁻¹).

The total PAH emission factors varied in an even wider range. Burning of oak resulted in a mean total PAH emission factor of 0.34 mg kg⁻¹ (SD 0.19 mg kg⁻¹), while wood-based wastes (PAP, LDF and OSB) produced mean total PAH emission factors from 1.2 to 2.7 mg kg⁻¹ (SD 0.83–1.4 mg kg⁻¹). Much higher total PAH emission factors were obtained for RAG, TR and plastic wastes, ranging from 21 mg kg⁻¹ (SD 19 mg kg⁻¹) to 257 mg kg⁻¹ (SD 103 mg kg⁻¹). These high total PAH emission factors are the consequence of the higher PM₁₀ emission factors of plastic wastes as well as of the higher total PAH content of the unit mass of PM₁₀ released during combustion. To compare the hazard of residential waste combustion with wood burning, the total PAH EFs were converted into benzo(a)pyrene equivalent emission factors. It was found that the toxicity of PAHs emitted during the combustion of wood-based wastes (PAP, LDF and OSB) was about 30 times higher than that of PAHs released during the burning of oak. For RAG, TR and plastic wastes this ratio ranged between 280 and 4050 as a consequence of the different composition of PAHs emitted during the combustion of different waste types.

These results clearly show that residential waste burning (especially combustion of tyres and plastics) poses a serious hazard to human health for numerous reasons: waste combustion may produce considerably more PM₁₀ particles than wood burning; the particles contain more PAHs; and the PAHs formed during waste combustion are more carcinogenic than those released from wood burning. These findings underline the importance of concerted efforts of municipalities, authorities and NGOs to phase out the illegal burning of solid wastes in households. This would help significantly improve local air quality and reduce the number of limit exceedances of PM₁₀ and PM_2.5. But more importantly, the health risks associated with air pollution would be reduced disproportionately more than would normally follow from such reductions in PM₁₀ and PM_2.5 concentrations due to the extremely hazardous composition of particulate matter emitted during the combustion of wastes in households.