Characterizing PM_2.5 Emissions and Temporal Evolution of Organic Composition from Incense Burning in a California Residence

Abstract

The chemical composition of incense-generated organic aerosol in residential indoor air has received limited attention in Western literature. In this study, we conducted incense burning experiments in a single-family California residence during vacancy. We report the chemical composition of organic fine particulate matter (PM_2.5), associated emission factors (EFs), and gas-particle phase partitioning for indoor semivolatile organic compounds (SVOCs). Speciated organic PM_2.5 measurements were made using two-dimensional gas chromatography coupled with high-resolution time-of-flight mass spectrometry (GC×GC-HR-ToF-MS) and semivolatile thermal desorption aerosol gas chromatography (SV-TAG). Organic PM_2.5 EFs ranged from 7 to 31 mg g^–1 for burned incense and were largely comprised of polar and oxygenated species, with high abundance of biomass-burning tracers such as levoglucosan. Differences in PM_2.5 EFs and chemical profiles were observed in relation to the type of incense burned. Nine indoor SVOCs considered to originate from sources other than incense combustion were enhanced during incense events. Time-resolved concentrations of these SVOCs correlated well with PM_2.5 mass (R² > 0.75), suggesting that low-volatility SVOCs such as bis(2-ethylhexyl)phthalate and butyl benzyl phthalate partitioned to incense-generated PM_2.5. Both direct emissions and enhanced partitioning of low-volatility indoor SVOCs to incense-generated PM_2.5 can influence inhalation exposures during and after indoor incense use.

Short abstract

Incense particle emissions are chemically similar to organic aerosol from biomass burning. SVOCs in residential indoor air readily transfer to incense-generated PM_2.5. Both direct emissions and sorptive partitioning can influence occupant exposures.

1. Introduction

Exposure to fine particulate matter (PM_2.5) is a substantial risk factor for human health.¹ Indoor conditions² and air quality can strongly influence one’s personal exposure to PM_2.5 as studies have shown that indoor PM_2.5 concentrations can sometimes exceed outdoor levels.³ Given that people spend an average of about two-thirds of their time at home,⁴ combined with the health risks associated with PM_2.5 exposure, it is crucial to improve our understanding of PM_2.5 indoors. Incense burning is known to be a significant combustion source of residential indoor particulate matter^5,6 with emission factors similar to cigarette smoke.⁷ Incense-generated particles are predominantly in the PM_2.5 size range,^8,9 most concentrated in the accumulation mode,^6,10,11 and characterized by large surface area per unit mass, which allows them to sorb gas-phase organic and inorganic compounds, potentially affecting inhalation exposures.

Incense burning is a traditional practice in many cultures that serves as an integral part of worship and ceremonial functions.⁸ Beyond its religious significance, incense serves various other purposes, including therapeutic, aesthetic, insect-repelling, and producing pleasant fragrances.¹² Incense use is most prevalent in the Asia-Pacific, Middle East, and Sub-Saharan African regions, which is reflected in the majority of incense studies and associated adverse health effects emerging prominently from these regions.^{8−10,13−20} Incense use is also popular in the West with the US as the largest global importer.²¹ While estimates of US domestic incense use are limited, residential settings likely contribute a significant proportion of total incense consumption. Market research shows that incense products saw a surge in popularity and sales during the COVID-19 pandemic, which was driven by demand for enhancing in-home ambiance amid growing mindfulness practices (e.g., meditation and yoga).²² Despite this increasing trend, incense use remains largely overlooked as a source of indoor PM_2.5 pollution in Western literature.²³ This oversight is concerning given that a quarter of the US population comprises diverse racial-ethnic and cultural populations whose religions²⁴ incorporate incense burning into worship practices, potentially influencing household incense use. Furthermore, the religiously unaffiliated, making up another sizable portion of the population, may also frequently burn incense at home for aromatherapy or spiritual reasons, a practice especially common in Black²⁵ and Latino²⁶ communities. For instance, one study on polycyclic aromatic hydrocarbons (PAHs) exposure during pregnancy found that 28% of Black and Latino women reported burning incense at home.²⁷

Incense-generated PM_2.5 has been extensively studied indoors with characterization efforts focusing primarily on the physical characteristics of incense emissions, including emission factors, particle mass concentrations, particle size distributions, and particle-associated PAHs in incense smoke.^{6,7,11,13−15} Conversely, the speciated chemical composition of incense-generated PM_2.5 remains largely uncharacterized in indoor environments, despite organic aerosol (OA) constituting the majority of emissions.^8,28−30 Some studies have elucidated the chemical profile of particulate organic matter in incense smoke, particularly from incense sticks, by utilizing chemical characterization techniques such as aerosol mass spectrometry (AMS) and gas chromatography (GC). Yet, such investigations have been restricted to targeted chemical analysis of selected compounds or chemical groups,^{16,18,30−32} qualitative explorations of identified species,³³ and organic compositional analysis with low chemical resolution owing to limitations in the speciation capabilities of the instrumentation used.^6,34 Furthermore, these studies have only been conducted in laboratory or controlled environments with some results indicating substantial fractions of OA mass that are unidentified or unresolved.^30,32

Two-dimensional GC coupled with high-resolution mass spectrometry (GC×GC-HR-MS) is a chemical characterization technique that can provide more comprehensive separation and identification of diverse and complex particle-phase organics present in incense smoke.³⁵ GC×GC has been applied to speciate and quantify thousands of individual chemical species spanning a wide range of volatilities and polarities present in atmospheric matrices.^36,37 Its strengths also lie in separating isomers and several chemically similar compounds that would normally coelute in a one-dimensional GC column. A recent GC×GC-MS laboratory study characterizing total suspended particles from incense emissions could only chemically speciate a small fraction of total particulate organics, owing to quantification challenges that resulted from a lack of chemical derivatization in their analysis approach.³⁸ When applied to augment a characterization technique like GC×GC, chemical derivatization significantly enhances recovery of polar organic compounds,³⁹ which dominate incense particulate emissions,^33−35,38 and serves to adequately resolve separated compounds in a manner essential for reliable quantification. Employing online chemical derivatization coupled with GC×GC to characterize incense-generated PM_2.5, as is done in the current work, provides insight into the abundance and behavior of incense-derived particle-phase organics in indoor environments.

Assessing air quality impacts of indoor-generated PM_2.5 requires characterizing common sources. Whereas cooking, candle burning, and smoking have been relatively well studied as indoor combustion sources,⁴⁰⁻⁴² exploring the chemical composition of incense burning and its contributions to indoor air pollution has been more limited.⁴³ The current study focuses on evaluating incense PM_2.5 emissions and the associated chemistry indoors in a normally occupied residence in the San Francisco Bay Area during a period of vacancy. Incense-generated PM_2.5 was collected on filters for offline organic compositional analysis by GC×GC with supporting semivolatile organic compound (SVOC) measurements using semivolatile thermal-desorption aerosol gas chromatography (SV-TAG). As this work focuses on incense burning indoors, its significance lies in enhancing our understanding of combustion-generated PM_2.5 in a residential environment, thus offering insights for exposure assessment and source mitigation.

2. Methods

2.1. Incense Experiments and PM_2.5 Sample Collection

The analysis explored in this work was conducted in November 2021 during an unoccupied period in an otherwise normally occupied single-family residence in Oakland, California. During incense experiments, the studied home had interior doors open and exterior doors and windows closed. There was no mechanical ventilation, and the central air-handling system was off. Detailed descriptions of this H3 field campaign, incense burning protocols, and the rationale behind choosing the study residence and incense scents are provided in the Supporting Information. Briefly, five incense burning experiments were carried out in H3, each with two sticks of either lavender or Douglas fir incense. Incense burns occurred twice a day around noon and 5:30 pm with incense sticks ignited in the kitchen and allowed to burn through a ∼60 min duration before removing the incense source from the residence (see Figure S1 for an illustration of experimental timing).

Indoor incense filter samples were collected in the kitchen over 6-h periods on average, along with three outdoor filters simultaneously sampled in the backyard of the H3 residence during incense experiments. Incense smoke was sampled at 10 L min^–1 on 47 mm quartz filters using a custom-designed aerosol sampler³⁷ fitted with a cyclone to exclude particles larger than 2.5 μm. Before incense experiments, an indoor air background sample was collected in the kitchen over a 22-h period under closed-home conditions. The H3 occupants reported never burning incense in this residence; thus, the background sample is assumed to be representative of the H3 indoor environment, unperturbed by chemicals related to incense combustion. All H3 samples were collected on heat-treated quartz filters and were stored frozen (−20 °C) prior to chemical analysis.

2.2. PM_2.5 Filter Analysis: GC×GC

H3 PM_2.5 filters were analyzed by offline thermal desorption two-dimensional gas chromatography with online derivatization coupled with electron ionization high-resolution time-of-flight mass spectrometry (TD-GC×GC-EI-HR-ToF-MS). A detailed description of instrument methods and specifications as well as quantification, uncertainties, and compositional analysis of H3 samples can be found in the Supporting Information (Tables S1–S4). Briefly, filter punches (0.07–0.82 cm²) from H3 samples were initially heated in the instrument’s thermal desorption unit under continuous helium gas flow saturated with a derivatization agent, N-methyl-N-(trimethylsilyl)trifluoroacetamide. Desorbed organics were then separated by volatility and polarity, in sequence, by two serially connected GC columns separated by a thermal modulator followed by subsequent ionization under 70 eV EI and detection by HR-ToF-MS with a resolution of 4000. Analyzed PM_2.5 organics spanned between C₁₃ and C₃₆n-alkane volatility equivalents. GC×GC chromatograms were analyzed using the GC Image software, and quantification of detected compounds was performed using 110 external standards. Compounds were identified and classified into chemical families through matches with authentic external standards, searches against mass spectral libraries (NIST-20 and UCB-GLOBES^36,37) utilizing linear retention index (RI), and analysis of second dimension retention times (RT2) and specific molecular ions indicating chemical functionality. Henceforth, chemically ‘speciated’ PM_2.5 mass concentrations reported and discussed throughout the text are based on H3 filter samples analyzed by GC×GC unless otherwise stated.

2.3. Differentiating Indoor and Incense-Attributed Compounds in Incense Samples

The indoor air background sample allowed differentiation between existing conditions at H3 and distinct new compounds observed following incense burns. Compounds detected in the indoor background sample were assigned to a template in GC Image that could be searched against and compared to samples with incense PM_2.5, through location in GC×GC space (RI and RT2) and EI mass spectra information. In this way, ubiquitous indoor SVOCs such as phthalate ester plasticizers^44,45 along with other compounds identified in the indoor background were characterized as particulate organics not, directly, related to incense combustion, despite their detection in samples collected during incense burning events. Some compounds present in the indoor background air at H3 increased substantially during incense experiments and were likely emitted in incense smoke while other species such as indoor SVOCs may be significantly influenced by incense-generated particle mass as a result of partitioning that enhances their particle-phase abundance.^42,46−49 Compounds observed in the indoor background and the dynamics of their enhancements are discussed in Section 3.3.

2.4. Supporting Analysis for Incense Experiments

Hourly real-time measurements of combined gas and particle-phase organics (gas-plus-particle) smaller than 2.5 μm were collected using an online semivolatile thermal desorption aerosol gas chromatograph (SV-TAG). A detailed description of SV-TAG (e.g., design, operation, uncertainties) can be found elsewhere.^39,50 SV-TAG separately sampled air from the kitchen and the outdoors with measured SVOC volatilities corresponding roughly to the range C₁₄ to C₃₅n-alkanes. SV-TAG is a GC-based instrument utilized in prior indoor air chemistry campaigns.^47,51 During the H3 campaign, SV-TAG was housed in a temperature-controlled enclosure in the garage-basement. Real-time PM_2.5 mass was quantified in the kitchen during incense experiments by a Particles Plus (8306) particle counter. Time-resolved particle number concentrations were measured at 1 Hz in six diameter channels between 0.3 and 25.0 μm, and mass concentrations were calculated under a particle density assumption of 1 g cm^–3. A Sunset Laboratory Model 5L OCEC instrument coupled with the NIOSH870 thermal protocol was used at the Air Quality Research Center at UC Davis to analyze H3 PM_2.5 filters, including field blanks, for organic carbon (OC) and elemental carbon (EC) content.

3. Results and Discussion

3.1. Chemical Composition of Particulate Organics from Incense Burning

Incense-attributed compounds from both lavender and Douglas fir incense burns exhibited diversity in both structural and chemical properties as shown for a representative incense burn sample in Figure 1b. New compounds following incense burns in H3—i.e., distinct ‘incense-attributed’ compounds not including the indoor background (see Section 2.3)—were distributed over the range of measured species with more compounds inhabiting higher mass (lower volatility) and more oxidized (higher polarity) regions in GC×GC space compared to the indoor background (Figure 1a). Approximately 1600 distinct incense-attributed compounds were separated from incense-generated PM_2.5 OA samples, of which ∼300 compounds were classified into chemical families with 86 of those compounds being positively identified across all five incense burns. The chemical families include acids (carboxylic acids), alcohols, alkanes, aromatics (compounds with at least one aromatic ring), cyclics/oxygenates (non-aromatic/cyclic aliphatics), methoxyphenols, other oxygenates (with two or more -OH groups), other terpenoids (mono- and sesquiterpenoids), resins/diterpenoids, sterols/triterpenoids, and sugars (with sugar derivatives including anhydrosugars and sugar alcohols). Compounds with signal responses above the selected intensity cutoff that could not be identified or classified during analysis were categorized as not identifiable while the remainder of the compounds under this limit were grouped as unclassified.

Figure 1.

GC×GC chromatograms of (a) the indoor background sample and (b) an incense smoke sample collected from a Douglas fir incense burn event. Compounds are separated by volatility on the x-axis and by polarity on the y-axis. Each point (panel (a) total = ∼460; panel (b) total = 590) corresponds to a unique compound with a full mass spectrum (m/z ∼ 35–650). Point size approximately scales with quantified organic aerosol mass concentrations (μg m^–3), and colors identify the compound’s classification. Differing colors between the same classification (e.g., acids and alcohols) in both chromatograms highlight indoor background compounds in panel (a) distinct from incense-attributed compounds shown in panel (b) (see Section 2.3).

Incense-attributed compounds were closely related to smoke particles found in biomass-burning organic aerosol (BBOA).^37,52 These findings were consistent with previous studies,^32,34,38 as the composition of incense sticks (e.g., aromatic woods, flowers, and resins) supports the observation of wood pyrolysis products as found in BBOA. The range of observed BBOA compounds demonstrated the elevated presence of polar and oxygenated organic compounds in incense smoke. This finding was confirmed by estimating the incidence of derivatization in GC×GC, which accounted for more than 88% of all incense-attributed compounds across incense burn samples. The process for determining the fraction of separated compounds that were derivatized is discussed in the Supporting Information (see Figure S2). Due to the derivatization process used in this work, many polar organics were both recovered and well resolved, making it possible to reliably integrate and quantify these speciated compounds. However, speciated compounds differ chemically from their original forms with derivatization, which can create some challenges for identification and classification as derivatized mass spectra may not be available or published in MS libraries or literature.^37,53 Overall, 29–39% of incense-attributed compounds was classified within each analyzed incense burn sample.

3.2. Speciated Organic PM_2.5 Mass Contributions from Incense Emissions

Total GC×GC speciated PM_2.5 OA concentrations are shown in Figure 2a for incense samples, with mass fraction contributions from both indoor background and incense-attributed compounds. Incense samples contained mass concentrations 4.6–7.9 times higher than the indoor background. Incense-attributed compounds shown in Figure 1b accounted for 3.3–7.9 μg m^–3 (48–73%) of total speciated mass in incense samples (Figure 2a). The lower limit of the incense-attributed mass fraction is likely related to incomplete incense combustion for the DougFir-4 sample (see Supporting Information). The total outdoor mass concentration, 1.1 μg m^–3, is the average of the three outdoor filter samples collected at H3. The incense-attributed mass fraction for each incense sample is summarized by chemical family in Figure 2c. Classified compounds accounted for 74–82% of incense-attributed mass across burns while the not identifiable and unclassified compounds represented on average about a fifth (18–26%).

Figure 2.

GC×GC speciated PM_2.5 organic aerosol mass concentrations for (a) total quantified mass in H3 filter samples, with designated mass fractions in incense samples, (b) indoor background mass fractions classified by chemical family in each sample, and (c) incense-attributed mass fractions classified by chemical family in incense samples. Note the different y-axis scales in (b) and (c) in contrast to (a), and that the sum of (b) and (c) equals (a) for incense samples. Each stacked bar in (b) and (c) shows the sum of concentrations of classified compounds in each sample (see Tables S6 and S7). Colors that differ between common chemical families (e.g., acids and alcohols) in (b) and (c) highlight that compounds within those chemical families are distinct between indoor background and incense-attributed masses. ‘Ind-Bkg’ refers to the indoor air background sample, ‘Lavndr’ refers to a sample with lavender incense burned, and ‘DougFir’ refers to a sample with Douglas fir incense burned.

Sugars contributed the highest proportion (27–49%) to incense-attributed mass for all burns with levoglucosan, an unequivocal BBOA tracer, being the most abundant sugar compound emitted (41–62%). Levoglucosan was also the highest mass contributor across all incense-attributed compounds with concentrations ranging from 0.4 to 1.3 μg m^–3. This finding aligns with reported high levoglucosan emissions from incense burning³⁸ and as pyrolysis products of primarily cellulose-rich plant materials⁵⁴ such as those used in producing incense. Other important BBOA tracers such as resins/diterpenoids were also abundant (19–28%) in Douglas fir burn samples as the second highest contributing incense-attributed chemical family. These compounds are indicative of conifer species^37,55−57 such as the Douglas fir incense burned in H3. Emissions of dehydroabietic acid, which is an identified marker of conifer combustion,⁵⁶ were also in agreement with BBOA studies^56,57 as the most abundant resin/diterpenoid (27–33%) with speciated mass concentrations of 0.2–0.7 μg m^–3 for the four Douglas fir burn samples. Palustric acid was the second highest mass contributor of the resin/diterpenoid chemical family (17–26%) in Douglas fir samples with concentrations of 0.1–0.5 μg m^–3. This compound is distinct to pine species^37,55 and likely originated from the addition of ‘piñon pine’ highlighted in the list of ingredients for Douglas fir incense.

For the single burn sample of lavender incense (non-conifer), resins/diterpenoids only made up 0.2% of the incense-attributed mass with a concentration <0.01 μg m^–3. Furthermore, only the lavender burn exhibited other terpenoid compounds such as mono- and sesquiterpenoids. These compounds accounted for 3.3% of incense-attributed speciated mass and are distinctive to fragrant plants like lavender. Also, compared to Douglas fir samples, lavender incense emitted much less mass of methoxyphenols (3.1–4.7% vs 0.9%). Since methoxyphenols are pyrolysis products of lignin and indicative of burned wood in fine aerosol,⁵⁶ this difference is likely a result of more woody biomass components being present in Douglas fir incense compared to the higher floral content used in producing lavender incense. These variations in the organic chemical composition of incense-generated PM_2.5 are likely influenced by components of the incense materials used^10,38 and are relevant given that the toxicity of BBOA, and incense by proxy, depends partly on chemical composition.⁵⁸ The vast majority of these compounds have unknown health impacts;⁵² nonetheless, some BBOA-related compounds with hazard codes outlined in the Supporting Information of Liang et al.⁵² were positively identified in this work as summarized in Table S5. Additional BBOA-related chemical families contributing to incense-attributed mass fractions in Figure 2c are provided in Table S6.

In this study, PAHs were not detected in PM_2.5 organic aerosol collected on incense filter samples analyzed by GC×GC, including lower volatility PAHs with four rings or more (e.g., fluoranthene), which are generally found in the particle phase.⁵⁹ This observation was surprising given the prominent role of PAHs as products of incomplete combustion of carbonaceous fuels as is characteristic of incense burning. While some studies have explored particle-associated PAHs in incense emissions,^13,14,16 others have encountered challenges in measuring these compounds indoors, such as in Lung et al.¹⁵ who estimated household incense PAH emissions through laboratory combustion.⁶⁰ Observed mass loadings of PAHs from their laboratory experiments were less than 0.1% of total particle mass, with similar ratios reported in later studies in a controlled experimental house⁶ and during chamber combustion.¹⁶ Yang et al.¹⁶ reported that emission factors of gas-phase PAHs were about three times higher than particle-associated PAHs, and Lin et al.¹⁴ found that 90% of indoor airborne PAHs from incense burning was in the gas phase. These studies suggest that open combustion indoors, incense composition, particle loading, and PAH abundance in the gas phase may have impeded detection of particle-bound PAHs in H3 incense samples.

3.3. Influence of Incense Burning on Indoor Compounds and SVOC Concentrations

The speciated chemical composition of compounds observed in the indoor background sample is depicted in Figure 1a. Approximately 460 background compounds observed in H3 were largely concentrated within more volatile regions (95th percentile RI < C₂₄) in GC×GC space compared to compounds attributed to incense in Figure 1b. This finding indicates that indoor background compounds were predominantly higher volatility SVOC species such as plasticizers (e.g., from floor coverings and personal care products) and surfactants (e.g., in detergents and cleaners), which are abundant and persist in indoor air from stable or recurring indoor sources.⁴⁴

Analysis of the indoor background sample resulted in 58% of observed compounds classified into broad chemical families: esters, flame retardants, other non-cyclic aliphatics/oxygenates, pesticides, plasticizers (non-phthalates and phthalates), and surfactants (non-ionic, alcohol ethoxylates). Chemical families of indoor background compounds matching those illustrated in Figure 1b for incense are described in Section 3.1. Compounds observed in the indoor background sample had a total mass concentration of 1.4 μg m^–3 (Figure 2a) and accounted for 1.5–3.5 μg m^–3 (24–52%) of total speciated mass in incense samples. Figure 2b displays the speciated mass concentrations of chemical families, describing the range of observed indoor background compounds and their abundance with respect to incense samples (Table S7). Classified compounds accounted for 75–86% of the indoor background mass fraction across all samples shown in Figure 2b. For indoor background compounds detected in incense samples, we treated the indoor air background sample as the baseline for these non-incense-attributed compounds. This approach allows for broad observations of changes in the abundance and enhancements of indoor background compounds across incense samples.

Using GC×GC speciated mass, enhancement ratios (ERs) were calculated from the increases in the time-averaged concentrations of indoor background compounds detected in incense samples relative to (i.e., divided by) the indoor background sample. In this study, ERs for a given compound had to be consistently greater than two (2) across all incense samples to be considered ‘enhanced.’ Additionally, compounds detected in incense samples that were completely absent from the indoor background sample were manually searched at levels below normal analytical limits of detection (LODs) in the background sample to ensure that they were not misattributed due to changes in LOD from differing sampling volumes. Figure 3 illustrates the range of ERs calculated for indoor background compounds detected in the DougFir-1 incense sample. Table S8 reports all indoor background compounds that were consistently enhanced across incense samples and were positively identified. Background indoor compounds within acid, alkane, alcohol, aromatic, ester, and plasticizer chemical families were found to be consistently enhanced across incense samples, with variability in ER values. The enhancements of acids, alcohols, and alkanes most likely relate to direct OA emissions from incense burns. Observed background compounds such as C₂₀ acid (eicosanoic acid) and C₂₇ alkane (heptacosane) were previously shown to be constituents of primary BBOA, emitted abundantly in samples collected during wildfires.⁵² This evidence is consistent with high mean and median ERs for eicosanoic acid (14.5 and 16.0) and heptacosane (46.5 and 42.8) in incense samples. Both eicosanoic acid and heptacosane as well as other enhanced indoor background acids, alkanes, and alcohols have been previously reported in incense emissions.^35,38

Figure 3.

Enhancement ratios of indoor background compounds detected in DougFir-1 incense sample. Point size indicates the quantified organic aerosol mass concentration (μg m^–3) in GC×GC, and the color scale identifies the enhancement ratio compared to the indoor background sample. ‘Undetected’ refers to indoor background compounds not detected in DougFir-1.

The enhanced aromatic, ester, and plasticizer compounds we observed in incense samples were mostly SVOC species related to building materials and consumer items such as personal care products. For example, homosalate is an SVOC used in the production of sunscreen,⁶¹ which has been found in residential indoor air during occupancy independent of cooking or cleaning activities.⁵¹ Isopropyl myristate, commonly used as a cosmetic ingredient in personal care products, was detected in 100% of household dust samples collected in a study in northern California.⁶² Bis(2-ethylhexyl) phthalate (DEHP) and butyl benzyl phthalate (BBzP) are widely used as plasticizers that make up large mass fractions of several construction and furnishing materials.⁶³ They have also been measured extensively in residential indoor environments.⁶⁴ Diethyl phthalate (DEP) is an abundant plasticizer present in furnishings as well as in personal care products.⁶³ Bis(2-ethylhexyl) terephthalate (DEHT) is a high production volume non-phthalate plasticizer, often used as a replacement for DEHP, in consumer products and building materials.⁶⁵ 2,2,4-Trimethyl-1,3-pentanediol diisobutyrate (trade name TXIB; Eastman Chemical) is another non-phthalate plasticizer that is used largely in vinyl products such as flooring and wall coverings.⁶⁶ TXIB, DEP, and isopropyl myristate were also shown in one study to dominate SVOC peak abundances in residential indoor air under closed conditions prior to window opening.⁶⁷ To better understand the dynamic behavior of these assumed indoor originating SVOCs detected and enhanced in incense filter samples, we provide relative abundances of gas-plus-particle samples of these compounds during incense burns using SV-TAG.

3.4. Influence of Incense-Generated PM_2.5 on Gas- and Particle-Phase Partitioning of Indoor SVOCs

Nine indoor SVOCs were compared to particle mass at hourly resolution during incense burning events using the SV-TAG and Particles Plus instruments (Figure 4). The first four compounds in the top and bottom panels in Figure 4 represent SVOCs with volatilities above and below C₂₃ alkane-equivalent volatility, respectively. This distinction is highlighted to explain dynamic SVOC behavior that has been reported in prior residential indoor air studies. Using the same instrument and methods in an observational study, Lunderberg et al.⁴⁷ reported that certain indoor SVOCs exhibited shifting gas-particle phase partitioning in response to changes in PM_2.5 concentration. SVOCs with volatilities lower than the C₂₃ alkane were observed to partition onto airborne particles when particle mass concentration increased. This effect was less prominent for higher volatility SVOCs. We observed similar effects in our experimental incense study with the strongest correlations for SVOCs with lower volatilities (Figure 4).

Figure 4.

Associations of time-resolved indoor SVOC abundance with PM_2.5 mass concentrations during incense burning experiments. Reported SVOC measurements are total (gas-plus-particle) internal standard normalized signal (equivalent to mass concentrations). SVOCs are arranged in order of increasing retention index (see Table S8). ‘TXIB,’ ‘DEP,’ ‘BBzP,’ ‘TEG-EH,’ ‘DEHP,’ and ‘DEHT’ refer to 2,2,4-trimethyl-1,3-pentanediol diisobutyrate, diethyl phthalate, butyl benzyl phthalate, triethylene glycol di(2-ethylhexoate), bis(2-ethylhexyl) phthalate, and bis(2-ethylhexyl) terephthalate, respectively.

For SV-TAG samples of indoor SVOCs with volatilities above C₂₃, the background concentrations were mostly gaseous as evidenced by the high intercepts of the regression lines for gas-plus-particle measurements shown in Figure 4. With increased PM_2.5 mass during incense burning events, concentrations of these higher volatility SVOCs generally remained above background levels. Similar behavior was reported by Lunderberg et al.⁴⁷ and Kristensen et al.⁴⁶ as these higher volatility indoor SVOCs were not expected to strongly partition to airborne PM_2.5. In contrast, for indoor SVOCs below C₂₃ volatility, as displayed in the bottom panel of Figure 4, strong associations between total (gas-plus-particle) SVOC abundance and incense particle mass concentrations were observed, with regression intercepts close to zero across displayed compounds. These relationships strengthen moving from left to right for these lower volatility SVOCs, as partitioning to airborne particles appears to drive the enhancement of their airborne particle-phase concentrations.

Some laboratory incense chamber studies have shown the presence of certain plasticizers and related indoor SVOCs in particulate incense emissions.^31,38 DEP is the only phthalate plasticizer that has been reported to be used as a binder ingredient for incense production in India.^17,68 In the present work, although DEP was enhanced in incense filter samples, there was no real-time association with airborne particles during incense burning experiments (Figure 4) when compared to similar higher volatility indoor SVOCs. For other plasticizers and SVOCs discussed, the gas-particle phase partitioning behavior presented in this work, and supported by SVOC measurements in prior indoor studies, provides substantial evidence that these SVOCs largely originate from other residential indoor sources, rather than directly from incense emissions. Furthermore, if it were the case that the discussed indoor SVOCs were emitted primarily from incense emissions, one would expect to observe different time-resolved behavior of their abundance in indoor air than was exhibited in the SV-TAG data. Specifically, associations should be very strongly correlated with incense-generated PM_2.5 as the increases and eventual decay of airborne PM_2.5 would be equal to the presence and relative abundance of emitted indoor SVOCs.

Overall, the findings in this work are consistent with prior studies in residences^46,47 and chambers,⁶⁹ augmented by modeling studies,⁷⁰ which specifically show enhanced DEHP emissions from surfaces being influenced by the presence of airborne particles. That lower volatility SVOCs sorb to incense-generated particles is important for recognizing how SVOCs can indirectly contribute to indoor organic PM_2.5 mass and how strong indoor particle sources can alter the mode and intensity of SVOC inhalation exposures for occupants.⁴⁴ Inhaled PM_2.5 with increased toxicity from sorbed SVOCs would deposit in the respiratory tract differently than would a gas-phase SVOC.⁷¹ These observations are not only relevant for indoor SVOCs such as DEHP and BBzP, which have been shown to contribute to adverse health outcomes in their roles as endocrine-disrupting chemicals,⁷² but also for other unexplored low-volatility indoor SVOCs with unknown health effects.

3.5. Emission Factors of Incense-Generated PM_2.5 in Indoor Air

Consistent with prior incense studies,^28,29 total organic carbon (OC) mass comprised the majority of PM_2.5 in incense filter samples with concentrations of 17.4–37.2 μg m^–3. An assumed OA/OC ratio of 1.6, as derived from field and laboratory primary BBOA AMS measurements,^73,74 was applied to estimate total PM_2.5 OA in incense samples based on OC measurements (Table S9). The speciated mass concentrations of indoor background and incense-attributed compounds quantified in incense samples (Figure 2a) can explain, on average, 20% of the total PM_2.5 OA mass in each sample. This extent of successful speciation is credited to online derivatization that facilitated enhanced recovery of more polar and oxygenated incense particle-phase organics compared to non-derivatized analysis of incense emissions.³⁸

To evaluate and more easily compare incense burning contributions in H3 with prior studies, PM_2.5 emission factors (mass of PM_2.5 emitted per mass of incense burned) were estimated using the integral mass balance approach for episodic emissions.⁷⁵Equation 1 determines the mass, M, of PM_2.5 emitted from an incense burn event in a well-mixed indoor environment with volume, V, where C_i is the time-averaged OA concentration in incense filter samples, C_b is the background OA concentration in H3, Δt is the sampling duration, and L is the total removal rate of incense-generated PM_2.5 during and following emissions. L was estimated by fitting time-resolved incense PM_2.5 data from the Particles Plus to a first-order decay⁷⁶ (see Supporting Information). For this calculation, a few assumptions were made: 1) incense-generated PM_2.5 is well-mixed throughout the interior volume of H3 during the burning and decay period; 2) the background OA concentration during incense burns is the same as that measured in the time-averaged indoor background sample collected prior to incense experiments; and 3) the loss rate estimated during the post-burning decay period applies throughout the incense emission event. Applying these assumptions, the PM_2.5 emission factor, EF, is then estimated using equation 2, where m is the mass of two incense sticks burned per experiment.

1

2

Evidence from analysis of episodic point source emissions⁷⁷ shows that the use of the well-mixed assumption should not result in large estimation errors if the gamma (γ; source inactive and well mixed) period is longer than the alpha (α; source active and poorly mixed) plus beta (β; source inactive and poorly mixed) periods. Figure S3 illustrates time-resolved PM_2.5 concentrations for one of the experimental runs, showing that the duration of alpha plus beta periods is much less than the gamma period. PM_2.5 EFs from incense experiments ranged from 7 to 31 mg g^–1 (Table S10) and were within the range of results from other incense stick burning studies,^{9,12,16,29,59} which is summarized in Table S11. Estimated PM_2.5 EFs in this work indicated that Douglas fir incense burns (average: 27 mg g^–1) produced more PM_2.5 emissions than lavender (7 mg g^–1), which is likely attributable to higher carbonaceous matter from woody biomass materials in Douglas fir incense compared to the higher floral content in lavender incense. The estimated EFs could be affected by both incense composition and burning conditions, which can vary in ways that are neither easily predicted nor effectively controlled in H3 incense experiments. Nonetheless, results suggest that the severity of indoor particle pollution from incense burning may be influenced by incense scent or materials. In particular, incense with lower carbon and higher calcium content has been shown to result in lower particle emission factors.^16,78 Incense burning is often an episodic or short-term source of PM_2.5 indoors; however, long-term habitual use by occupants could have cumulative effects on indoor air pollution and human health.

3.6. Implications

Exposure to incense-generated PM_2.5 has been shown to be harmful to human health.²⁰ In this study, we characterized in detail the chemical profile of fine particulate matter resulting from five incense burning experiments in the H3 residence. Organic PM_2.5 fractions from lavender and Douglas fir incense burns resulted in 4.6–7.9 times higher concentrations of speciated OA compared to background conditions when averaged over 6-h time periods. Incense-attributed compounds showed clear similarities to BBOA with abundant masses of tracers such as levoglucosan and dehydroabietic acid. Incense type also influenced the chemical composition of OA as well as EFs with Douglas fir incense generating more PM_2.5 than lavender (27 vs 7 mg g^–1). Furthermore, ubiquitous indoor SVOCs, such as plasticizers, observed in H3 indoor background air were influenced by increasing PM_2.5 concentrations during incense burns due to volatility-dependent gas-particle phase partitioning. Low-volatility indoor SVOCs were shown to strongly correlate (R² > 0.75) with time-resolved concentrations of PM_2.5 mass during incense burning events, thus indirectly contributing to organic PM_2.5 incense mass. These observations also demonstrate how the chemical composition of incense-generated PM_2.5 can be altered in indoor air and point to potential increases in occupant uptake of low-volatility SVOCs such as DEHP and BBzP, through shifts in their airborne concentrations and gas-particle phase partitioning. We showed that incense burning, when it occurs, is a substantial combustion source of indoor PM_2.5 pollution with diverse chemical complexity and interactions with components of the indoor environment. As such, more research directed toward chemical speciation would help to better understand the impacts of incense burning emissions on indoor air quality in different indoor environments. Given that this study was conducted in a single residence with only two types of incense tested, readers are advised to exercise caution in extrapolating toward generalized conclusions from this study.

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.est.3c08904.

• GC×GC methods and materials, quantification, compositional analysis, and derivatization; detailed descriptions of field campaign and incense experiments; speciated mass concentrations of indoor background and incense-attributed chemical families in incense PM_2.5 samples; enhancement ratios of indoor background compounds; incense PM_2.5, OC, OA, loss rates, and EFs (PDF)

Author Present Address

^∥ School of Chemical and Biomolecular Engineering, Georgia Institute of Technology, Atlanta, GA 30332, United States

Author Present Address

^▼ Environment Unit, CSIRO, Aspendale 3195, Australia.

Author Present Address

^⬢ Office of Energy Efficiency & Renewable Energy, AAAS Science, Technology and Policy Fellow, U.S. Department of Energy, Washington, DC 20585, United States

Author Contributions

J.O. wrote the manuscript, conducted incense experiments, and analyzed H3 PM_2.5 filter samples; M.R.A., Y.L., and E.B.F. supported data analysis and manuscript development; D.M.L. collected and analyzed SV-TAG SVOCs data; C.E.I. supported experimental design and manuscript development; B.C.S., W.W.N., and A.H.G. oversaw incense experiments and sample and data analysis. All coauthors reviewed and contributed comments to the manuscript.

The authors declare no competing financial interest.