Particulate-Bound Polycyclic Aromatic Hydrocarbons (PAHs) and their Nitro- and Oxy-Derivative Compounds Collected Inside and Outside Occupied Homes in Southern Sweden

Abstract This study presents indoor and outdoor levels of airborne fine particles (PM2.5), particle bound polycyclic aromatic compounds (PACs) including parent-, alkylated-, nitro-, and oxy-PAHs. Week-long simultaneous measurements were conducted inside and outside 15 occupied homes in southern Sweden during wintertime. The homes were single-family houses or apartments located in urban, semi-urban, and rural areas. The PM2.5 and PACs levels were low compared to studies worldwide. There was great variation in concentrations between sites, which likely is due to proximity to road and traffic intensity. The lower concentrations of nitro and oxy-PAHs compared to parent PAHs in this study, compared to other studies, could possibly be due to lower atmospheric photochemical formation outdoors because the cold climate. This assumption could not be confirmed and need to be further tested. The results point to that particle PAC levels found inside arise primarily from outdoor. This correlation was not as clear for PM2.5. The results of a comparison between residences before and after energy renovation did not indicate an improvement in indoor air regarding PACs. To understand exposure and assess risks it is important to measure wide range of PACs both in gas and particle phase.


Introduction
Air pollution continues to be a significant threat to human health globally. The World Health Organization (WHO) has estimated that around 7 million deaths per year globally are attributable to the joint effects of ambient and household air pollution. 1 The impact of air pollution is also significant in Sweden. It has been estimated that exposure to particulate matter (PM) can cause 7600 premature deaths in Sweden. 2 PM in the ambient air has been the subject of numerous studies and PM levels have been routinely monitored for many years. [2][3][4][5][6] Negative health effects of fine particles (PM2.5) include risk of lung cancer, asthma, as well as increased respiratory and cardiovascular mortality and morbidity [6][7][8][9][10][11] and the International Agency for Research on Cancer (IARC) has classified outdoor pollution particles as carcinogenic to humans (Group 1). 12 Gas-particle partitioning has great significance for the environmental fate of organic compounds. 13 Temperature, humidity, and wind speed are important environmental parameters for how the gas-particle distribution turns out. 13 Number studies point to that not only morphological characteristics of the particle but also the chemical composition, which is then controlled by the gas-particle distribution, plays an important role in the adverse health effects of PM. 4 Among the wide range of organic compounds that are associated with PM, polycyclic aromatic hydrocarbons (PAHs), and especially the 16 PAHs defined by the United States Environmental Protection Agency (so called 16 USEPA PAHs) has been studied intensively in different parts of the world. [14][15][16][17][18][19][20][21][22][23][24][25] PAH is a complex mixture of multi-ringed persistent substances with different molecular weights and physicochemical properties. PAHs are formed during the combustion of organic materials, and thus these compounds have many sources both outdoors and indoors, for instance motor vehicle exhaust, domestic burning of wood, coal or biomass, cooking, candle lightening and smoking, and various industrial processes. 15,20,[26][27][28][29][30][31][32] There are several processes that can influence PAHs concentrations; thus, measurements can be challenging. Temperature at sampling location will influence phase change between particle-bound and gas-phase PAHs, there are also several possible depletion mechanisms, such as atmospheric degradation due to photolysis and oxidation processes, where temperature was reported to play a major role. 33,34 A group of polycyclic aromatic compounds (PACs) that has been less addressed in environmental studies are the nitrated (nitro-PAH) and oxygenated (oxy-PAH) derivatives of PAHs. This is notable since many nitro-and oxy-PAHs are more toxic than their parent PAHs due to their mutagenicity and carcinogenicity. 35,36 These PAH derivatives, like PAHs, can be formed during incomplete combustion processes of organic matter. Once in the atmosphere, they can also be formed through secondary reactions with OH and NO 3 radicals and through homogeneous/heterogenous reactions with NO 2. [37][38][39] Although, several monitoring campaigns in various regions, have demonstrated the presence of these derivates in the atmosphere, 30,[37][38][39][40][41][42][43][44] to the best of our knowledge studies on the presence of these PAH derivates in ambient air in Sweden is missing. Furthermore, little attention has been given to indoor environments of these compounds 30 and this can be especially important since people spend most of their time indoors. 45 Moreover, several studies have shown that concentration of some air pollutants (e.g. PAH and persistent organic pollutants (POPs)) can be higher indoors than outdoors. 28,46 Therefore, in order to evaluate the risk to human health posed by exposure to air toxics, and to identify high-risk populations, it is important to acquire knowledge of contaminant levels and understanding of indoor sources for an increased number of PAH substances.
This study is a part of two bigger projects, one focusing on physicochemical characteristics and toxicity of indoor and outdoor particles 47,48 and People-Environment-Indoor-Renovation-Energy (PEIRE) project, which encompasses different effects of energy renovations and occupants' behavior on indoor environment quality. [49][50][51] The hypothesis in the bigger projects was that the toxicity of particles indoors (where we spend 90% of the time) is different in comparison to the toxicity of particles outdoors, whereas typically only outdoor particles' characteristics are used in the epidemiological studies to assess health effect due to exposure to particles. Herein we present the results of a multitude particulate-bound PACs as PACs are important PM component that relates to toxicity of particles and therefore to health effect of exposure to particles. PACs were measured indoors in residences in southern Sweden and were compared with levels measured just outside the buildings. Possible sources of the PACs found in these indoor environments are discussed. Moreover, repeated measurements were made in seven of the residences after energy renovation of the buildings. Levels measured before and after this renovation are compared and discussed.

Sampling campaigns
Week-long simultaneous measurements were conducted inside and just outside 15 occupied homes in southern Sweden during wintertime 2016/2017. The homes were single family houses or apartments. The places of the residences were considered rural, urban, or semi-urban. None of the sites were close to a known outdoor point source where PAH levels can be expected to be elevated, such as an industry, gasoline station or directly next to a motorway. Information of the sampling sites is provided in Table 1.
Two identical sets of instruments for indoor and outdoor measurements were used for a multitude analytical endpoint. Details of the entire instrumental equipment set, and analytical endpoints were presented elsewhere. 47 Only the parts relating to sampling relevant for this article are described here. Sampling equipment was placed in specifically designed enclosures to minimize noise and disturbance for the residents, and for the outdoor measurement to protect against weather conditions such as precipitation. Enclosure indoors was typically placed in the living room unless occupants due to practical reasons requested placement in the hall or kitchen. Outdoor enclosure was placed in close vicinity of the indoor sampling site, i.e. either on a balcony or in the garden in case of homes 1-5. In homes 6-15 as these were within the same vicinity, due to practical reasons the outdoor enclosure was kept in one place, i.e. on a balcony of home 9. The distance between indoor and outdoor sampling location for homes 6-15 was within 300 m radius.
Indoors PM2.5 particles were collected on Teflon filters (37 mm, pore size 2 mm, Teflo; Pall Corporation, Port Washington, NY) inside a DustTrack DRX 8533 (2 L min À1 flow, TSI Inc, Shoreview, MN). Outdoors, DustTrack 8520 (TSI Inc., Shoreview, MN) was used, it does not allow simultaneous collection of PM2.5 for post analysis (as in case of DustTrack DRX 8533, used indoors), thus outdoor PM2.5 were collected on Teflon filters (the same brand as describe earlier) in a separate collection line after PM2.5 cyclone at a flow rate of 4 L min À1 . Average total air sampling volume per filter was 21.5 m 3 for indoor samples and 40.7 m 3 for outdoor samples.
Week-long measurements were repeated one year later (winter 2017/2018) in seven of the residences after energy renovation. The renovation was described earlier. 49,50 In short, the renovation was focused on reduction of heat losses and balancing the existing mechanical exhaust ventilation (airflows mechanically forced through the kitchen and bathroom fans). Balancing the existing ventilation system was achieved by 1) replacing inlets supplying outdoor air (instead of wide inlets (window like openings) in the living room and bedroom, small inlet vents were installed); 2) better distribution of exhaust fans (situated on the roof) throughout the building (the single exhaust fan on the roof of each building before the renovation was replaced with three, each supplying the area served by one staircase); 3) The kitchen hoods were replaced in all apartments during the renovation. Both before and after the renovation a basic constant airflow was always exhausted via the hood and the residents could manually increase the airflow to a forced flow. Before the renovation the manual control was equipped with timer (switching for 30 or 60 min only), while after renovation there was no time restriction for forced flow control. To reduce the heat losses the following changes were done: 1) extra insulation to the building wall was added to tighten the building envelope and reduce heat transfer; 2) new thermostatic valves were installed on the hydronic radiators and the central heating system was balanced between buildings and flats; 3) the curtain walls, windows, and balcony doors on the first and second floors on the south façade were rebuilt or replaced during the renovation (the windows on the third floor had been replaced a few years earlier). The air exchange rates (AERs) were measured before and after the renovation (details in SI and in Omelekhina et al. 49 ) in 5 apartments out of seven the AER increased slightly after renovation. In the studied apartments AER before the renovation ranged from 0.31 to 0.64 h À1 (median 0.50 h À1 ) and after renovation from 0.41 to 0.73 h À1 (median 0.55 h À1 ).

Chemical analysis
The analytical procedures and instrumentation used for PM2.5 and particle-bound PAC analysis are described elsewhere 29,52 and in Supplementary Information (SI) and are given here in brief. The filters, diameter 37 mm, pore size 2 mm (Teflo; Pall Corporation, Port Washington, NY) were conditioned for 24 h prior to weighing in a climate chamber controlled for temperature and humidity (23 C, 50%). All filters were weighed using an MX5 balance (LOD < 11 mg) (Mettler Toledo, Columbus, OH). After gravimetric analysis, the filters were analyzed for their content of particulate PAC compounds. The target PAC compounds were 17 parent PAHs (the 16 priority US EPA PAHs plus perylene), 14 alkylated PAHs, 17 nitrated-PAHs (nitro-PAHs), and 10 oxygenated-PAHs (oxy-PAHs). All chemical classes with target compounds are presented in SI Tables S1 and S2. The filters were extracted by means of a Sonica ultrasonic extractor (Soltec, Milan, Italy) in three mL of dichloromethane for 12 min. 29 Two deuterated internal standard (IS) mixtures, containing 16 US EPA PAHs, and four nitro-PAHs, respectively, were added to the samples before extraction. The extract solutions were cleaned through Pasteur pipettes filled with 2 cm of silica gel and sodium sulfate on top, and finally concentrated by purging with gentle stream of nitrogen to a final volume of ca 30 mL. PAC compounds were separated and detected by means of high-resolution gas chromatography/ low-resolution mass spectrometry (HRGC/LRMS). The mass spectrometer (MS) was an Agilent 5975 C coupled to a 7890 A gas chromatograph (GC) (Agilent Technologies, Inc., Santa Clara, CA). The MS was operated in electron impact (EI) ionization mode for parent PAH and alkylated PAH compounds and in electron capture negative chemical ionization (ECNCI) mode for nitro-and oxy-PAHs. The MS was operated in selected ion monitoring mode (SIM) in both cases. Transitions and retention times for all included native and IS compounds are presented in Table S3 (parent and alkylated PAHs) and Table S4 (nitro-and oxy-PAHs). Every PAC target compound was quantified using the relative response factor (RRF) 53 to each native compound included in the standard mixtures. More details of the procedure used for quantification of the samples are described in SI.

Quality assurance
Consistent recoveries (40-110%) were obtained for all IS compounds that were added to and used for correction of the samples. Field and laboratory blanks were analyzed in parallel with the samples. Minor residues only of some 2-4 ringed parent PAHs occurred, although the amount occurring was <10% of the amount found in the samples. All results were corrected for the blanks. The limits of detection (LOD) were calculated as three times the standard deviation of the values for the blanks or the background noise of these blanks. Three standard reference materials (SRM 1649a, 1649b and 2975) were used as quality control (QC). The measured levels lie, for most part, within 30% of the certified levels. The results for SRM 1649b are given in SI for 20 parent-and 6 nitro-PAHs.

Introductory remarks on the data
All compound groups (PM2.5, parent-, alkylated-, nitro-, and oxy-PAHs) were detected in most of the samples. This confirms for the PACs in addition to parent and alkylated PAHs but also nitro-and oxy-PAH derivatives may be present in both outdoor and indoor environments in southern Sweden. Although comparison with literature data should be made with caution due to variations in the season sampling, the number of samples collected, and/or the number of targeted PAC compounds among studies, in general, the PAC levels reported herein are low.
The results (mean, median and range) of indoor and residential outdoor measurements can be found in Table 2, but also for indoor and outdoor measurements, the results are also shown in Tables S1 and S2 and Figure 1. Data are presented as range and median in further discussions since the values were not normally distributed. It is important to point out that the temperatures in the sampling enclosures was relatively high, which can have led to some loss of volatile species during sampling, hence some results can be underestimatedsee details in limitations.

PM2.5
The results (mean, median, and range) of indoor and residential outdoor PM2.5 measurements are given in Table 2 and Figure 1. The ranges for indoor respective outdoor was 2.1-24 mg m À3 (median 5.5) and 3.7-22 mg m À3 (median 5.3). Thus, there was a factor of 10 between the highest and lowest measured levels indoors and factor of 6 for measurements outdoors. This indicates slightly lower variations in outdoor levels, higher variation indoors with contribution of indoor sources to PM2.5 levels indoors, as reported by Omelekhina et al. 48 based on real time PM2.5 measurements indoors and outdoors and information from occupants' logbooks of activities. Particle mass concentrations in our study were low. They were lower than levels found in central and southern Europe 54 and Lithuania 15 but similar to those found in other Nordic countries viz. Helsinki, Finland 54 Copenhagen, Denmark 55 and Gothenburg, Sweden. 4 It is important to note

PACs
Parent PAHs were, as expected, the PAC group that made up the largest part in both indoor and outdoor particles, range 0.015-3.8 ng m À3 (median 0.40) and 0.068-19 ng m À3 (median 0.90). It varied slightly between samples, but parent PAHs accounted typically between 70 and 90% of the total PAC content. There was a large difference, two orders of magnitude, in comparison between the highest measured concentrations, outdoors and indoors, respectively. High molecular weight PAHs with five to six rings, such as benzo(a)anthracene, chrysene, benzo(b)fluoranthene, benzo(k)fluoranthene, benzo(a)pyrene (BaP), indeno(1,2,3-c,d)pyrene, benzo(g,h,i)perylene, and the 3-ring PAH phenanthrene predominated. High concentrations of the 4-ring PAHs, fluoranthene and pyrene, were also observed. These results are expected, since 90% of PAH with 5-6 rings are adsorbed on particles, PAHs with 4-rings are often in similar amounts in both gas (not measured here) and particle phases, while PAH with 2-3 rings is mainly found (>90%) in the gas phase.
When the levels of PAH in the gaseous phase are elevated in a place, the levels of PAH with three rings can also be noticeable on particles. Phenanthrene (3-ring PAH) was the dominant PAH in three of the indoor sites. The concentrations of parent PAHs observed in this study are low compared to other studies in Europe. Comparison of outdoor BaP levels in our study (range 0.0040-0.95 ng m À3 ) are lower than BaP levels in PM2.5 fraction in Western European cities (0.1-1.1 ng m À3 ) and in Eastern and Central European cites from 3.0 to 3.2 ng m À3 . 57 A study from Lithuania reported as high BaP levels as 6.2 ng m À3 . 15 However, the levels in our study are quite similar compared to another Swedish study, in Gothenburg that reported BaP levels between 0.011 and 0.14 ng m À3 . 4 It is worth noting that the European limit value for BaP is 1 ng m À3 in PM10. Increasing levels of ambient PM pollution from rapid urbanization and economic development in China and other places from Asia have drawn worldwide attention, especially with its significance for possible health effects. [16][17][18][19][20][21] A study from Beijing, indoors and outdoors, the BaP levels were only slightly higher than those from our study, 1.67 ng m À3 and 1.04 ng m À3 , respectively, 18 while another study from China (rural area of Guanzhong Basin), reported indoor and outdoor levels as high as 23 and 19 ng m À3 . 19 Two outdoor studies from the mega city of Bangkok Thailand, the levels were quite the same as in our study (approx. >0.4 ng m À3 ) 16,20 although a study from Chiang-Mai, Thailand where the objective was to study effects of agricultural waste burning on PM2.5bound PAHs, ambient air levels rendered in as high BaP levels as 9-703 ng m À3 . 21 The nitro-PAH concentrations ranged indoor and outdoor between LOD and 0.028 ng m À3 (median 0.0027) and 0.0018-0.11 ng m À3 (median 0.0076), respectively, and corresponding results for oxy-PAH were between LOD and 0.41 ng m À3 (median 0.019) and 0.0072-2.9 ng m À3 (median 0.051). The nitro-PAHs concentrations were a magnitude of two lower as those of sum parent PAHs and the corresponding results for oxy-PAHs were one order of magnitude lower. In a study from south of France oxy-PAHs were of the same magnitude as those of parent PAHs while nitro-PAHs concentrations were one to two orders lower. 37 Thus, the levels in our study of these PAH derivatives are low compared to total parent PAHs.
Nitro-and oxy-PAHs as well as parent PAHs partition between gaseous and particulate fractions. Most of these derivatives are particle bound; only 1-nitronaphthalene, 2-nitronapthalene and, napthalene-1-aldehyde, 2-naphthaldehyde and p-fluorenone are reportedly almost exclusively in the gaseous phase. 37 Out of 17 target nitro-PAHs only seven nitro-PAHs were detected in at least one sample. The highest detectability was 3-nitrofluoranthene, 9-nitrophenanthrene, and 4-nitropyrene, which was detected in almost all samples both indoors and outdoors. 2-Nitrofluorene, 9-nitroanthracene, and 2-nitropyrene were detected in between 40 and 80% of the samples, while 1-nitropyrene was detected in only one indoor and five outdoor samples. All oxy-PAHs were detected in most samples with a detectability of between 80 and 100% for both the indoor and outdoor measurements. Only 1,4-anthraquinone had low detectability and was found in low concentrations in only six samples. The levels of PAH derivatives are very low compared to other studies worldwide. For nitro-PAH, only the two highest levels in our study are in the same range as other studies, e.g. in France, 37,42 UK 58 and Spain. 59 Significantly, higher levels have been reported in Athens, Greece (2.3-5.9 ng m À3 ) 60 and very high levels have been reported indoors and outdoors in four cities in Nepal (4.7-18 ng m À3 ). 30 In addition, oxy-PAHs were low in our study compared to other studies. The two highest levels found outdoor in our study are comparable with previous studies, e.g. Nepal (0.45-6.5 ng m À3 ) 30 France (6.3 ng m À3 ) and Greece (3.5-6.9 ng m À3 ). 61 The alkylated PAHs were present in most samples both indoors and outdoors. They accounted for approximately between a few to about 10% of total PACs. It is important to point out here that out of the 14 alkylated PAHs that were included in this study, only four were expected to be found as particle bound. 62 Alkylated PAHs with two to three rings, which are mainly found in the gaseous phase, were also found in some samples. Alkylated PAHs ranged indoor and outdoor from LOD-0.24 ng m À3 (median 0.012) and LOD-1.6 ng m À3 (median 0.030).

Comparison of parent PAH levels before and after renovation
The results of the seven houses before and after energy renovation, indoors and outdoors, are depicted in Figure 2. PAHs levels were found to be two times higher outside (average 2.5 ng m À3 ) than inside (average 0.7 ng m À3 ) before renovation. After renovation, PAH levels were similar between outside (2.2 ng m À3 ) and inside (1.8 ng m À3 ). The PAH levels were similar outside from the sampling before and after renovation (2.5 and 2.2. ng m À3 , respectively). This was also confirmed by similar concentrations of BaP before and after renovation. Outside levels of BaP ranged from < LOD to 0.65 ng m À3 (average 0.15 ng m À3 ) before renovation, and from <0.02 to 0.48 ng m À3 (average 0.14 ng m À3 ) after renovation.
In indoor environments, PAH levels increased slightly after renovation, with an average of 0.7 ng m À3 before renovation and 1.8 ng m À3 after renovation. This was also confirmed by the increase of BaP. Before renovation, BaP levels ranged from < LOD to 0.13 ng m À3 (average 0.03 ng m À3 ), and after renovation, BaP levels ranged from < LOD to 0.36 ng m À3 (average 0.11 ng m À3 ).

PM2.5 and PAC levels
Ambient PM and PAHs are derived from a variety of sources and have both natural and anthropogenic origin. Anthropogenic sources are attributable to human activities, such as combustion processes, e.g. burning of biomass and fossil fuels, traffic exhaust (petrol and diesel vehicles), and industry. 20,27 Major indoor sources include smoking, cooking 29 wood stoves 28 and candle burning. 32 Nitro-and oxy-PAHs can be released together with parent PAHs in incomplete combustion processes but can also be formed through secondary reactions with OH and NO 3 38,39 . Moreover, diesel vehicles have been reported as a source of nitro-PAHs 63 and biomass burning has been reported an important source of oxy-PAHs. 64 Alkylated PAHs can also be formed as well as parent PAHs by incomplete combustion. Additionally, various motor fuels may contain elevated levels of alkylated PAHs, and elevated levels of these PAHs may occur through direct emissions from the fuel. 31 It cannot be ruled out that petrol stations may be a source of these PAHs.
The two outdoor measurement sites (H7 and H8), which are clearly illustrated in Figure 1, had the highest content of both PM2.5 and all PAC chemical groups. Indeed, for all measurement sites, as can be seen in Figure 3, both PM2.5, and the PACs alkylated PAHs, nitro-and oxy-PAHs correlated well with parent PAHs (R 2 ¼ 0.89 À 0.98; p < 0.001) (alkylated-PAHs not shown). In case of home H8 occasional smoking indoors could have influenced the results, however in case of home H11, where occasional smoking was also reported influence of smoking is not visible. Additionally, the outdoor sampling during measurements in homes H6-H15 were performed on a balcony of home H9, thus occasional smoking inside of H8 could not influence simultaneous outdoor measurements (at different location) at this time. In other words, the influence visible for homes 7 and 8 is due to elevated PM2.5 and PAC outdoor concentrations. Number of studies show that atmospheric photochemical reactions may significantly influence nitro-and oxy-PAH level in the environment (e.g. 30,38 ) In addition, tropical countries with temperature and stronger solar radiation could possibly favor the atmospheric transformation of PAHs to nitro-and oxy-PAH derivatives. 65 This may mean that the low levels of nitro-and oxy-PAHs compared to parent PAHs, in our study, that the dark and cold climate in northern countries does not favor the secondary formation of these PAHs in the atmosphere. Thus, the large variety of PM2.5 and PAC levels between the outdoor locations ( Figure 1) but still a clear correlation between all groups (PM2.5 and PACs) can indicate similar combustion sources and the variation in levels can be due to intra-urban variation between the sites that can be related to the different proximities to roads and dissimilarities in traffic density as well as varying in time contribution from long-range transport of pollutants.
Environmental exposures to substances in indoor air which may have toxic potential is important to evaluate, especially that we spend majority of our time in indoor environments. Also, to assess the importance how the outdoor environment affects the indoor air quality. We assumed that if the outdoor air has significantly affected the indoor air, this can be shown in pattern comparisons between indoor and outdoor, and that the different PM2.5 and particulate PACs, respectively, correlated between indoor and outdoor levels. Moreover, ratios between indoor and outdoor air concentrations (I/O ratios) may be used as a tool to indicate whether there are indoor sources (I/O > 1) or outdoor sources (I/O < 1).
There are pattern similarities between the indoor and outdoor samples (Figure 1). In the samples where the levels are elevated outdoors, they are also indoors. This pattern appears to be clearer for PACs than for PM2.5. This is better clarified by correlation comparison for the respective PAC group and PM2.5 ( Figure 4). Indeed, the PAC groups are correlated (R 2 ¼ 0.80-0.95, p < 0.001) (alkylated PAHs are not shown) indicating the significance of the outdoor air for the indoor air for all PAC groups. Contrary, PM2.5 indoors versus outdoors correlated poorly (R 2 ¼ 0.16, p ¼ 0.569), which confirms results reported by Omelekhina et al. 48,66 (based on real time monitoring indoors and outdoors and logbooks information) that indoor PM2.5 do not originate from outdoors only and indoor sources contribute to observed PM2.5 levels as well, hence the lack of the correlation. That the levels of particulate PAC indoors would have the same source as outdoors is also indicated by the correlation between parent PAH and other PAC groups indoors (Figure 3; R 2 ¼ 0.86-0.97, p < 0.001 (alkylated PAHs not shown)). There is a very weak indoor correlation between parent PAHs and PM2.5 (R 2 ¼ 0.21, p ¼ 0.452). Other studies have also shown that indoor and outdoor PM2.5 often do not correlate well and that PAH levels indoors do not correlate with PM2.5 indoors. 4,67 This again emphasizes that PACs have the same origin indoors and outdoors but PM2.5 also has significant sources indoors.
The I/O ratios for PM2.5 and all PAC groupings are shown in Figure S1. The I/O ratios were considerably lower for all parent PAHs in all homes except one (H15). However, the ratio for this home was around one for PM2.5. Thus, there has been indoor source(s) of particulate parent PAH in this home during the measurement period. The ratios of PAH derivatives were lower than one in most homes. However, in four homes, nitro-PAH was higher indoors than outdoors. This indicates that there may also be sources of nitro-PAH indoors. It is important to point out that the levels indoors even in these homes were very low. The PM2.5 I/O ratios were similar to one in most places but higher in some cases. There are many possible indoor sources of PM2.5, e.g. smoking, cooking, wood stoves, candle burning, cleaning, and presence of pets. 68 In this study, indoor sources that contributed to indoor PM2.5 namely, candle burning and cooking, were determined on basis of real time PM2.5 monitoring indoors and outdoors and logbook information of occupants' activities. 48,66 Two examples of such indoor sources contributing to PAHs indoors are cooking 69 and recently confirmed flickering (stressed) burning of candles. 32

Comparison of parent PAH levels before and after renovation
Results demonstrate that PAH levels were slightly increased inside after renovation. This agrees with I/O ratio of PM2.5 reported for seven of the studied apartments before and after renovation 48 namely I/O ratios of PM2.5 increased slightly after renovation (median 0.9) in comparison to before renovation (0.77). It can be linked to two processes: 1) increased contribution from outdoors due to slightly increased AER (median AER before the renovation 0.50 h À1 and after renovation 0.55 h À1 ); 2) contribution from indoor sources to observed PM2.5 levels indoors (indoor sources such as cooking and candle burning, identified in these apartments can contribute to indoor PAH air levels, as described earlier). Due to our small sample number, our results are not conclusive. We cannot show that the efficiency of the energy retrofits could reduce the health risks for the residents with regards to exposure to PM2.5 and PAHs.

BaPeq concentrations
In order to estimate the carcinogenic risks of PAHs mixtures, the equivalents of benzo(a)pyrene (BaPeq), with toxic equivalent factors (TEFs) for BaP set to 1, were evaluated by multiplying concentrations of each PAH with their TEF value. 27,70 These authors report TEF for 16 US EPA PAH compounds. TEF factors have been reported for five nitro-PAHs. 71 Thus, these parent PAHs and nitro-PAHs are ranked according to cancer potency relative to BaP. Alkylated-and oxy-PAHs, to the best of our knowledge, do not have TEFs but will likely be developed in the future. It is important to point out that alkylated and oxy-PAHs may be more toxic than corresponding parent PAHs and this can underestimate the overall cancer risk.
The European Community has suggested the BaPeq on PM10 at an annual standard in ambient air of 1 ng m À3 . 72 All measurement sites were below or well below this cancer potency value indoor 0.00038-0.31 ng m À3 (median 0.011) and outdoor 0.006-0.47 ng m À3 (median 0.073). The annual standard BaPeq value (1 ng m À3 ) have been exceeded in urban sites all over the world, e.g. Athens, Greece (1.6 ng m À3 ) 73 and Zabrze, Poland (7.9 ng m À3 ) 74 and Mexico City, Mexico (2.2 ng m À3 ). 75 None of the nitro-PAHs with TEF factors were found in this study. Thus, these PAH derivatives did not contribute to the cancer potency value indoor and outdoor in this study.
Contribution of the cancerogenic potency of BaP alone of the PAH mixtures is commonly >50% (e.g. 29,75 ). BaP, in our study, dominated the carcinogenic potency (>40%) in nine of the indoor and outdoor sites. In addition, the intermediate PAH fluoranthene, with a significant portion also in the gaseous phase accounted for over 50% of the cancer potency in some places. The total PAH content in air is often dominated by PAH with 2-4 rings (>90% in gaseous phase). Although these PAHs have a small TEF factor 69 the contribution of these PAHs can be significant if their levels are high. 31 This underlines the importance of including both particulate and gas phase PAHs when evaluating risk to human health. WHO 1 has set as an air quality guideline that estimates a unit risk estimate of a BaP level of 0.1 ng m À3 , which will cause an extra lifetime cancer risk of 1/100,000. BaP levels found in our study are somewhat higher than this air quality guideline in two indoor and six outdoor measurement sites.

Limitation and uncertainties
There are several limitations in this study. First, the sample size is relatively small. Second, due to the sampling protocol used in this study, the air concentrations were determined without considering gaseous PAH fraction, implying the corresponding conclusion might be unilateral to some extent, especially for low molecular weight PAHs. This can underestimate the concentrations in indoor environment even more as concentrations of gas phase PAH can sometimes be higher indoors than outdoors. Also, the sampling volumes were found to be too small to obtain sufficient mass to quantify, in particular, the nitro-PAHs which were in a very low concentration.
Even though a moderate flow rate was used for the air samplings, the sampling time and volume were still quite high, and the levels of the more volatile PAHs (2-4 rings) may have been somewhat underestimated in the particulate due to losses of these compounds from the filter during sampling. Additionally, the reported results can also be underestimated for both indoor and outdoor measurements, due to temperature difference of ambient air at sampling locations in comparison to temperature in the enclosures where the particles were collected. The average temperature inside the enclosures was relatively high which could have led to some loss of the volatile species. In the enclosures the average temperature was 38.5 C indoors and 31.3 C outdoors, while the average ambient temperatures were, respectively, 25.4 and 5.3 C. The temperature difference between indoor and outdoor week-long collection of the particles inside the enclosures was 7.2 C. This could have influenced both PM2.5 and PAC concentrations measurements, i.e. some volatile material could have evaporated. Sampled air volume indoors per filter (22 m 3 ) was approximately half in comparison to outdoor sampled volume (40 m 3 ). This volume difference can to some extents have affected the comparison between inside and outside with respect to the limitations mentioned above.
The measurements were conducted in real life scenarios, i.e. in occupied homes thus capturing the living conditions in the studied homes. Occupants were asked to keep logbooks which together with real time PM2.5 measurements allowed to track indoor source contribution to PM2.5 (reported in Omelekhina et al. 48 ) however due to time-integrated samples of PACs, it was not possible to identify specific indoor sources contributing to PAC levels. Location of the measuring equipment in proximity to strong indoor sources, e.g. cooking in the kitchen can result in higher contribution of cooking to PM2.5 and PAC levels in comparison to other sampling locations (living room or hall), measurements in the kitchen were performed in two homes on request of occupants due to practical reasons and might have influenced the concentrations measured. As the outdoor concentrations were monitored in the gardens or balconies, the potential influence of neighbors smoking, barbequing or accidental trash fires could not be controlled or prevented. In some locations, the outdoor measurement was on a balcony on a different floor than the indoor locations. This may have affected the comparisons and contributed to the uncertainty between the indoor and outdoor comparisons for these locations.
The sampling campaign was completed in >3 months. This means an uncertainty in comparing levels between the outdoor locations. This was not the aim of the study either.

Concluding remarks
In summary, the study shows that different PAC chemical classes, viz. parent PAHs, alkylated PAHs, nitro-and oxy-PAHs can be found on particles (PM2.5) both indoors and just outside these homes. The levels were low compared to other studies worldwide. There was, however, great variation in levels between the different sampling sites. Within the framework of this study, it was not possible to investigate specific sources outdoors of PM2.5 and PACs. The low concentrations of nitro-and oxy-compared to parent PAHs, compared to other published studies, could possibly be due to low atmospheric photo formation of nitro-and oxy-PAHs because the cold climate. It is likely that intra-urban variation between the sites, different proximities to roads, dissimilarities in traffic density and varying influence of long-range transport of pollution can partly explain the notable difference in levels found outdoors. However, the results indicate that the content inside the homes is strongly affected by the PAC levels just outside. A comparison of the PAH levels indoors and outdoors after energy renovation does not indicate an improvement of the indoor air regarding the levels of particulate PAC. Although the PAC levels were low, the air levels for BaP were higher inside and outside the residences at some of the measurement sites compared to the air quality guideline set by the WHO.