A Significant Role for Nitrate and Peroxide Groups on Indoor

Environment Department, University of York, York, YO10 5DD, U.K.. ‡ Atmospheric Chemistry ... Publication Date (Web): August 10, 2012. Copyright © ...
0 downloads 0 Views 542KB Size
Article pubs.acs.org/est

A Significant Role for Nitrate and Peroxide Groups on Indoor Secondary Organic Aerosol Nicola Carslaw,†,* Tiago Mota,† Michael E. Jenkin,‡ Mark H. Barley,3 and Gordon McFiggans3 †

Environment Department, University of York, York, YO10 5DD, U.K. Atmospheric Chemistry Services, Okehampton, Devon, EX20 1FB, U.K. 3 Centre for Atmospheric Sciences, School of Earth, Environmental and Atmospheric Sciences, University of Manchester, Manchester, U.K. ‡

S Supporting Information *

ABSTRACT: This paper reports indoor secondary organic aerosol, SOA, composition based on the results from an improved model for indoor air chemistry. The model uses a detailed chemical mechanism that is nearexplicit to describe the gas-phase degradation of relevant indoor VOC species. In addition, gas-to-particle partitioning is included for oxygenated products formed from the degradation of limonene, the most ubiquitous terpenoid species in the indoor environment. The detail inherent in the chemical mechanism permits the indoor SOA composition to be reported in greater detail than currently possible using experimental techniques. For typical indoor conditions in the suburban UK, SOA concentrations are ∼1 μg m−3 and dominated by nitrated material (∼85%), with smaller contributions from peroxide (12%), carbonyl (3%), and acidic (1%) material. During cleaning activities, SOA concentrations can reach 20 μg m−3 with the composition dominated by peroxide material (73%), with a smaller contribution from nitrated material (21%). The relative importance of these different moieties depends crucially (in order) on the outdoor concentration of O3, the deposition rates employed and the scaling factor value applied to the partitioning coefficient. There are currently few studies that report observation of aerosol composition indoors, and most of these have been carried out under conditions that are not directly relevant. This study highlights the need to investigate SOA composition in real indoor environments. Further, there is a need to measure deposition rates for key indoor air species on relevant indoor surfaces and to reduce the uncertainties that still exist in gas-to-particle phase parametrization for both indoor and outdoor air chemistry models.

1. INTRODUCTION Outdoor air pollutants receive much attention and many are subject to tight regulation. Although urban dwellers in developed countries spend ∼90% of their time indoors,1 indoor air quality has provoked much less concern. Indoor pollutants can be generated through cooking (particles, carbon monoxide, nitrogen oxides, NOx), cleaning (volatile organic compounds, VOCs, such as monoterpenes) and smoking (carbon monoxide, particles, NOx, VOCs) and can also be emitted from building, furnishing and consumer products such as carpets, adhesives, paints, houseplants, and toiletries.2,3 Further, outdoor air pollutants can ingress via ventilation systems or infiltrate the fabric of the building. Consequently, indoor air contains a complex mixture of chemicals, some of which potentially exist at higher concentrations than found outdoors. A number of indoor species have recognized or anticipated adverse health effects, as documented by animal and human exposure studies.4,5 Although hundreds of species have been measured in indoor air including VOCs, NOX, O3 (ozone), and particles, recent work suggests that the products of VOC © 2012 American Chemical Society

degradation following reaction with hydroxyl (OH) or nitrate (NO3) radicals or O3, may be responsible for the reported symptoms rather than the primary emissions.6−8 Interestingly, a number of studies have highlighted links between asthma and other adverse respiratory effects and the use of cleaning products (many of which contain terpenes), although the predominant effect mechanism remains largely unclear.9,10 The oxidation of VOCs forms a wide range of gaseous products, as well as those which are sufficiently involatile to promote aerosol formation and growth, often referred to as secondary organic aerosol (SOA). In the absence of indoor sources, SOA concentrations indoors follow those outdoors: concentrations and numbers of particles indoors can be significantly elevated, however, through activities such as cooking and cleaning.11 Long et al. (2001)12 suggest that particles generated indoors may be more bioactive than those Received: Revised: Accepted: Published: 9290

April 10, 2012 August 2, 2012 August 10, 2012 August 10, 2012 dx.doi.org/10.1021/es301350x | Environ. Sci. Technol. 2012, 46, 9290−9298

Environmental Science & Technology

Article

mechanism could contain up to ∼10 8 species.32 The mechanism is designed to provide a representation of the most important degradation routes, and thus provides a basis for the initial simulation of systems where a representation of chemical detail is required. It includes chemistry initiated by reaction with O3, OH, and NO3, and formation of all the major first-generation products identified in a number of reported studies of limonene degradation.31,33−36 However, the methodology also produces a large number of additional (currently uncharacterized) product species, that are predicted to be formed on the basis of general understanding of VOC degradation chemistry, including those formed in subsequent generations of oxidation.27,29 A further improvement to the model has been to add gas-toparticle conversions for limonene chemistry. Although the oxidation of other species will contribute to indoor particle formation, the ubiquity of limonene indoors, as well as its propensity for forming SOA, makes investigating the particles formed by its oxidation an important starting point. The MCM does not currently consider gas-to-particle conversions, so the gas-phase precursors (GPPs) likely to form particles in the limonene scheme were identified in two ways. Leungsakul et al. (2005a)35 described a semiexplicit gasphase mechanism for limonene, with 25 particle production pathways. Seventeen have direct equivalents in the MCM, including limonic, limononic, limonalic, keto-limonic, ketolimononic and keto-limonalic acids, 7-hydroxy-limonaldehyde, oxy-limonaldehyde, limonaldehyde, ketolimonaldehyde, and 7 nitrated species. The 8 remaining species are not formed, representing only minor pathways through the MCM construction protocol. In addition, a scoping study using vapor pressures, calculated as described below (Section 2.2), for 200 closed-shell products of limonene oxidation in MCM v3.2, established that the vast majority (>95%) of the simulated SOA mass under both photooxidation and ozonolysis conditions could be described using a subset of approaching 30 species. Within these, there was some overlap with the seventeen species already identified (limonic, limononic and ketolimonic acids), but most of the identified species were new and largely contained peroxide and alcohol group functionality. In all, this provided 41 species for which partitioning between the gas and aerosol phases was represented in the model. There have been few experimental limonene-SOA composition studies to date, and many of them have been carried out under very different conditions (in a chamber, much higher precursor concentrations, with OH scavengers, in the absence of NOX etc.) than typically experienced indoors.14,15,35,37−40 Also, detection of specific species depends on the experimental techniques adopted, their sensitivity and the availability of relevant calibration standards. Consequently, the studies often identify only a relatively low percentage of the total SOA mass. Finally, the SOA constituents interact with UV light, water, each other and gas-phase species and generally “age” throughout the experiments further complicating interpretation of the results.39 Glasius et al. (2000)41 focused on carboxylic acids and carbonyl compounds formed through the oxidation of a number of terpenes and found that for limonene, the most important aerosol-phase products were limonic, ketolimonic, limononic, ketolimononic, 7-hydroxylimononic and 7-hydroxyketolimononic acids. They also identified limonaldehyde and

generated outdoors and in addition, people are exposed to them through high frequency short-term high-concentration events (cooking for example). This pattern of exposure indoors is of concern, as evidence from short-term ambient events shows that there is an association with acute health outcomes.12 A particularly important group of aerosol-forming VOCs is terpenes, emitted naturally by coniferous plants and used extensively in indoor settings as additives/solvents to many household cleaning products, air fresheners and furnishings.3 Oxidation of terpenes can lead to a wide range of polyfunctional moieties, including alcohol, aldehyde, ketone and carboxylic acid groups: the oxidized products tend to have lower vapor pressures than their parent terpene and can condense to form SOA.13 Since the pioneering work by Weschler and Shields (1999),1 many groups have observed particle growth indoors through ozone-initiated reactions with terpenes:11,14−21 the focus has been on particle mass number, concentration and growth rates rather than composition. Limonene is both the most frequently detected and most abundant terpene species indoors.1 The mixing ratio of limonene is typically around 4 ppb (20 μg m−3) indoors,1,22 but much higher concentrations (e.g., ∼200 ppb) arise following activities such as cleaning.23 With its two double bonds, it has a higher propensity for SOA formation when compared to singly unsaturated terpenes such as α-pinene.15 Moreover, limonene-ozone reaction products have been shown to induce pulmonary irritation in rats (particularly elderly animals),24 though not in mice.25 There are still many question marks over indoor air SOA composition and few relevant studies in the literature. In this study, the composition of SOA produced by limonene oxidation reactions indoors is investigated using an improved version of the model described by Carslaw (2007).26 The nearexplicit Master Chemical Mechanism (MCMv3.2) employed, allows the chemical nature of the aerosol products to be examined in greater detail than experimental conditions currently allow.

2. MODEL 2.1. Chemical Mechanism. The basis of the model used in this work has been described in detail by Carslaw (2007).26 Briefly, a detailed chemical box model was constructed based on a comprehensive chemical mechanism (the Master Chemical Mechanism, MCM v3.2, http://mcm.leeds.ac.uk/ MCM/) to investigate indoor air chemistry in a typical suburban residence in the UK. The MCM has been described in detail elsewhere.27−30 Version 3.2 consists of a near-explicit representation of the gas-phase degradation of ∼143 VOC and is constructed following a defined protocol, using the latest kinetic and product data where available, or structure activity relationships in their absence.27 A new feature of MCM v3.2 is a scheme for limonene, enabling a significant improvement to be made to the indoor air model given the importance of limonene indoors. The complete degradation chemistry of limonene was based on the MCM protocol rules,27,29 kinetic parameters for limonene initiation reactions reported in a VOC review and evaluation31 and, where applicable, the prevailing kinetics recommendations of the IUPAC panel (http://www.iupackinetic.ch.cam.ac.uk/). The complete mechanism consists of 1576 reactions of 539 closed-shell and radical species. It is therefore moderately detailed, but necessarily contains a number of simplification measures,27,29 without which the 9291

dx.doi.org/10.1021/es301350x | Environ. Sci. Technol. 2012, 46, 9290−9298

Environmental Science & Technology

Article

The vapor pressures (in Torr) were calculated using a modified version of the work by Nannoolal et al. (2008),46 which takes into consideration group contributions and group interactions. The modifications were made to account for the complex-multifunctionality of the limonene oxidation species, in particular, those of hydroperoxide, peroxy acid and PAN groups.47 Calculating the vapor pressure first involves estimating the boiling point for each species. The method of Nannoolal et al. (2004)48 has been recommended for calculating boiling points following a quantitative review of available methods.49 The value used for the vapor pressure has a critical impact on the calculated values of Kp. This area remains one of considerable uncertainty when considering the partitioning process, as very few quantitative data are available for checking the predicted values. The partitioning process was represented dynamically as a balance between absorption and desorption.43,45 A species and temperature independent value of 6.2 × 10−3 m3 μg−1 s−1 was used for kon, which was calculated based on the estimated collision rate of gaseous molecules with a monodisperse aerosol with a diameter of ∼50 nm.43 Variations in kon by an order of magnitude have been found to have no effect upon the simulated aerosol mass.45 The value of koff was then found assuming equilibrium conditions exist, since at equilibrium:35

7-hydroxy-limonaldehyde in the organic aerosol. In another study,14 a detailed analytical analysis of the 21.2% of limonene photooxidation SOA products that could be identified, showed that maleic acid, ketonorlimonic acid, 4-isopropenyl-1-methyl1-hydroxy-2-oxocyclohexane, ketolimonic acid, and 5-hydroxyketolimononic acid were the most abundant species. In a highresolution electrospray mass-spectrometry study to characterize limonene/ozone SOA,39 the most prominent first generation products were endo-ozonide, limonic, limononic, limonalic and 7-hydroxy-limononic acids, limonaldehyde, and 7-hydroxylimonaldehyde, though rapid oligomerization through subsequent reaction of some of these products with Criegee intermediates also occurred. In conclusion, while the 41 species selected here are by no means an exhaustive representation of the SOA composition for limonene-oxidation, they include many of the major species observed through the limited experimental evidence to date, and provide a good starting point to investigate the aerosol composition in more detail. The stabilized Criegee intermediates (SCIs), formed through reactions of ozone in the new MCM limonene degradation scheme, react to varying degrees with H2O, CO, NO, NO2, and SO2. However, mass spectrometry analysis of the particle phase has shown that hydroperoxides, peroxyhemiacetals, and secondary ozonides may form from the reactions of SCIs in the presence of water vapor, carboxylic acids, alcohols, and aldehydes.35,38 Further, it has been suggested that dimerization reactions can lead to a large reduction in volatility of the parent species and hence play an important role in the formation of SOA, though there is debate over whether such material is formed reversibly or irreversibly.42 Inclusion of such reactions in models is currently difficult, as the kinetics and reaction pathways are uncertain.42 Consequently, these reactions are excluded at present. Jenkin (2004)43 represented the participation of involatile dimers of bi- and multifunctional acids in the gas to particle process, through a simplified “acid chaperone” mechanism. Therefore, in the improved model, each relevant acid (limononic, ketolimononic, limonic, ketolimonic, limonalic, and ketolimonalic acids) reacts with the sum of the other acidic species in the model to produce its aerosol phase equivalent. The rate coefficient for this process has been optimized using experimental data to ∼6 × 10−11 cm3 molecule−1 s−1 at 297 K.43 2.2. Gas-to-Particle Partitioning. Gas-to-particle partitioning was defined for the 41 species according to the absorptive partitioning theory of Pankow (1994),44 whereby the phase-partitioning of a given species is defined by the thermodynamic equilibrium of that species between the gasphase and condensed organic-phase.35,43 The associated partitioning coefficient, Kp (m3/μg), is given by 7.501RT KP = MWom109γompLo

Kp =

(3)

Note that the thermodynamic equilibrium approach severely underestimates the transfer of organic material to the aerosol.35,43,45,50−52 Such an observation is consistent with the observed or apparent vapor pressures of complex multifunctional compounds, for example, product aldehydes, being 1−2 orders of magnitude lower than those predicted35 or unaccounted for condensed phase reactions.43 In order to try and account for these issues and in line with the recommendation of previous work using a similar mechanism,43 a scaling factor of 120 has been applied to all Kp values in this work. In reality, such an adjustment is likely to be species dependent as reactivity in the condensed phase will vary with functional group. It is clear that at present, absorptive partitioning theory is insufficient to explain SOA formation in current frameworks. In the absence of more detailed experimental data, the impact of varying this scaling factor on the SOA composition is explored in more detail in Section 3.2.2. 2.3. Operating Conditions. The box-model used for this work assumes a single well-mixed environment. The concentration of each species is calculated according to eq 5: ⎛A ⎞ Q dC i = −Vd⎜ i ⎟C i + λrfC o − λrC i + i + dt Vi ⎝ Vi ⎠

(1) −1

kon koff

n

∑ R ij j=1

(E5)

−1

where R is the ideal gas constant (8.314 J K mol ), T is the temperature (K), MWom is the mean molecular weight of the absorbing particulate organic material (g mol−1), γom is the activity coefficient of the species in the condensed organicphase (assumed to be 1.043,45) and PL0 is the liquid vapor pressure of a given species (Torr). The initial value of MWom was 120 g mol−1.17 Subsequent values of MWom were calculated as the model run proceeded, by accounting for the molecular weights and proportions of each individual component of the overall particle mass.

where Ci (Co) is the indoor (outdoor) concentration of a species; Vd is the deposition velocity of a species; Ai is the surface area of a room; Vi is the volume of a room; λr is the air exchange rate (AER) between indoors and outdoors; f is the outdoor-to-indoor penetration factor (assumed to be equal to 122); Qi is the indoor emission rate and Rij is the reaction rate between species i and j. There are few deposition velocities available relevant for indoor conditions: the literature is also surprisingly sparse for outdoors with the exception of a few key species such as O3, 9292

dx.doi.org/10.1021/es301350x | Environ. Sci. Technol. 2012, 46, 9290−9298

Environmental Science & Technology

Article

Figure 1. Simulated mass concentration of particles from exchange and chemistry as a function of AER.

SO2 and NO2 that have been studied for a variety of surfaces.53 Therefore, a new approach was adopted here. Zhang et al. (2002)53 provide deposition velocities for 31 species (based on observation where possible) on 15 different outdoor surfaces for different times of year. The values for “mixed broadleaf and needleleaf trees”, “grass”, “shrubs and interrupted woodlands” and “urban” for summer conditions have been averaged to give a typical “suburban” value of the outdoor deposition velocity for 25 species/types of species that appear in the model. For O3 and NO2, there also exists a range of indoor values on a variety of surfaces (e.g., fabrics, wood, plasterboard, linoleum).54 These indoor values have been averaged for a range of indoor surfaces to give deposition velocities for O3 and NO2 of 0.0345 and 0.0261 cm s−1 respectively. The calculated outdoor suburban values for O3 and NO2 of 0.607 and 0.566 cm s−1, respectively, are factors of 17.6 and 21.7 times higher than the indoor values. Therefore, the outdoor suburban values were all divided by 20 to give indoor values to be used in this work (Supporting Information Table 1). The sensitivity of the model to these values is discussed in more detail later. Note that a deposition velocity of 0.07 cm s−1 was used for NO3, N2O5, HO2, and OH22, and fine particles were assumed to deposit at a rate of 0.004 cm s−1.17 It has been assumed that 30% of the outdoor TSP (total suspended particulate) is organic in nature and provides a surface for indoor SOA to absorb on once indoors.17 Organic compounds can make up between 20 and 90% of atmospheric fine particulate matter,42 so this percentage value is varied in later sensitivity tests. In order to obtain realistic values for the GPPs of the particulate species, an outdoor version of the model was runoff-line. The outdoor concentrations of the GPPs were then set to be a proportion of the outdoor limonene concentration, according to the steady-state value obtained in the outdoor model runs. Photolysis in the model considers both artificial lights indoors and attenuated outdoor light.26 The improved model contains ∼20,000 reactions describing VOC emissions, gasphase chemistry, deposition to surfaces, exchange with outdoors and gas-to-particle partitioning for limonene.

3. RESULTS AND DISCUSSION 3.1. SOA as a Function of Air Exchange Rate. Sarwar and Corsi (2007)17 produced an indoor model that used a simplified scheme for limonene oxidation, whereby two semivolatile organic compounds formed from the reactions of limonene with ozone, OH and the NO3 radical, were assumed to partition between the gas and particle phases. They tested their model assuming that no pollutants other than ozone and limonene and the resulting gaseous byproduct and fine particles were present indoors. They assumed a relative humidity (RH) of 50%, a temperature of 297K and a surface area-to-volume ratio (A/V) of 2.7 m−1 and then tested the effects of varying AER on particle concentration for an indoor ozone concentration of 12 ppb and an outdoor TSP of 15 μg m−3.17 They found that for an AER of 0.5 ach−1 (air changes per hour), their model predicted a total suspended particulate (TSP) concentration of 13.7 μg m−3 with 5.3 μg m−3 (39%) ozone-limonene SOA. For the model in this work using similar conditions (unlike the Sarwar and Corsi model,17 NOX and other species will enter the room through the reactions of the GPPs, which are derived partly from outdoors), the TSP concentration was 19.3 μg m−3, with 10.9 μg m−3 (56%) SOA and 8.4 μg m−3 from exchange with outdoors (figure 1). As the AER decreases, the SOA concentration increases down to a value of 0.1 ach−1. Lower AERs enable more time for indoor reactions to take place.17,19 Interestingly, below AERs of 0.1 ach−1, the concentration of SOA decreases, similar to a previous study modeling α-pinene ozonolysis, when SOA concentrations decreased for AERs below 0.3 ach−1.55 In the current study, at an AER of 0.05 h−1, the O3 concentration is