electronic supplementary material
Life Cycle Impact Assessment (LCIA)
Indoor intake fraction considering surface sorption of air organic compounds for life cycle assessment
Yvan Wenger • Dingsheng Li • Olivier Jolliet
Received: 16 October 2011 / Accepted: 23 March 2012
© Springer-Verlag 2012
Responsible editor: Alexis Lauent
Y. Wenger • D. S. Li (*) • O. Jolliet
Department of Environmental Health Sciences, School of Public Health, University of Michigan, Ann Arbor, Michigan 48109-2029, USA
Y. Wenger
Centre d'Etudes en Sciences Naturelles de l'Environnement, Faculty of Sciences, University of Geneva, 1290 Versoix, Switzerland
(*) Corresponding author:
Dingsheng Li
Tel: +1 (734) 730 0564
e-mail:
Supplement S1. Considered principles governing the fate of a substance in an indoor environment
S1.1 Air exchange
Air exchange is a major factor that influences indoor pollutant concentrations. First of all, it drives the transport of the gas between indoors and outdoors, directly affecting the concentrations. Secondly, it determines the amount of time a substance has to interact with other species in the system boundaries or to be degraded. As shown by Weschler and Shields (1997a, 2000), this competition between export from the system and chemical elimination can be meaningful at typical air exchanges rates for indoor air pollutants. The ventilation rates applied to the model were selected to be consistent with the recommendations of international standards and regulations, supposed to be representative of North American or European conditions. A personal ventilation rate of 50 m3h-1 pers-1 has been used. This corresponds to the rate recommended by the ASHRAE (American Society of Heating, Refrigerating and Air-Conditioning Engineers 2003) for offices (average of total ventilation rate for adapted and un-adapted person in Smoking-Permitted offices) and also in the range for “Medium indoor air quality” for non-smoking areas or “Acceptable indoor air quality” for smoking areas by the European Committee for Standardization (2004).
S1.2 Gas-phase degradation
Degradation among organic pollutants in the gas-phase is mostlya consequence of a consequence of bimolecular reactions. In the troposphere, they are generally dominated by interactions with ozone (O3), hydroxyl radicals (·OH) and nitrate radicals (·NO3). Other chemicals can also interact, such as alkoxy radicals (RO·), but they are not considered here since they are believed to be of minor importance as organic compound loss processes (Atkinson and Arey 2003); this has been extensively qualitatively studied (Atkinson 2000) as much as quantitatively (e.g. Atkinson and Arey, 2003). In the present modeling, focusing on long-term exposure, realistic, but not time-dependant, concentrations of ·OH, O3 and ·NO3 were selected assuming a time-averaged situation. An overview of guiding principles which determines their concentrations indoors is presented below.
Tropospheric ozone is almost entirely produced outside by a set of reactions involving principally nitrogen oxides, VOCs and sunlight. Reactions between ozone and most common indoor organics are often thermodynamically favorable but are not fast enough to compete with other removal pathways. The exception is organic compounds with unsaturated carbon-carbon bonds, representing about 10% of indoor pollutants. Ozone has a significant half-life, meaning that it is able to enter buildings and establish meaningful levels inside. Introduction of ozone from outdoor air is actually the first contributor of its concentration in indoor environments and indoor concentration tracks the outdoor concentration (Hales et al. 1974). Various studies reported an indoor/outdoor ratio of ozone in a range of 0.2 to 0.7, excluding extreme scenarios (see Weschler 2000 and references therein). In studies of indoor ozone, the use of the indoor/outdoor ratio is often preferred to absolute concentrations because of the close dependence of ozone levels on urban produced pollutants, as well as sunlight, creates an important temporal and spatial variability. Furthermore, the amount of ozone introduced indoor is dependent on the air exchange rates (Weschler and Shields 2000). Since Life Cycle Assesment (LCA) aims to evaluate long term exposure over a large population, a mean value of ozone concentration had to be chosen instead of using the temporally variable values of a particular situation. This is a common methodology in LCA, as this approach focuses on long-term average impacts rather than extreme situations. A long-term mean concentration of 8 ppb has been selected to be representative of an average concentration between days and nights of ozone (summer) and non-ozone season (winter) (Weschler, per comm.). This corresponds to an average outdoor value of 40 ppb assuming an air exchange rate of 1 h-1 (Weschler and Shields 2000). This value is within the same range as the average between ozone and non-ozone season according to an extensive measurement campaign carried out in Southern California over 156 homes during one year (6-days integrated measurements by Geyh et al. (2000) and additional data by Lee et al. 2002).
Outdoor, the hydroxyl radical is a degradation product of photochemical oxidation, which reacts quickly with most atmospheric constituents. Because of small indoor light levels, the rate of hydroxyl radical production in buildings has been thought to be small and thus has received little attention. However, Weschler and Shields (1996, 1997b), Fan et al. (2003) and Sarwar et al. (2002) showed that the OH radical could be produced at significant rates indoors by reaction between ozone and terpenes. The correlation between ozone and hydroxyl radical leads to a derived OH average concentration of 3×10-6 ppb (Weschler and Shields 1996, Sarwar et al. 2002). This is lower than a typical midday outdoor concentration (~2×10-4 ppb) but higher than measured nighttime levels (4×10-7 ppb, Tanner and Eisele 1995), indicating that the hydroxyl radical is able to influence indoor chemistry.
Finally, reaction of ozone with nitrogen dioxide leads to the production of the nitrate radical (·NO3). This radical is photolytically unstable, but given the lack of direct sunlight, indoor concentrations may be comparable to outdoors at night (~1×10-3 ppb), following Sarwar et al. (2002).
Reactions involving these three reactants have been extensively studied, and reaction constants are easily accessible in literature. Those are described using 2nd order rates constants and pseudo 1st order constants were obtained assuming fixed concentrations of the OH radical, ozone and the nitrate radical. The overall removal rate by reaction is given by:
(S1)
where second orders rates constants ki,jg are expressed in ppb-1h-1 and the cumulative pseudo 1st order constant kig,deg in h-1.
Degradation rate constants used for the calculations were principally extracted from Atkinson and Arey (2003) and Nazaroff and Weschler (2004).When not available, additional values were collected from EPI Suite v3.12 software (2004), which provides experimental and calculated data.
S1.3 Boundary layer transport and gas-surface mass partitioning
Several models have been developed to assess the distribution of masses indoors. Tichenor et al. (1991) proposed a sorption model based on Langmuir isotherms, assuming a linear relationship between phases to describe interactions between vapor phase organic compounds and indoor sinks. This model is still used by most recent Indoor Air Quality models. Further investigations in chamber tests lead to thermodynamic constants describing the concentration ratios of compounds between the gas-phase and a given surface material. Throughout the adsorption literature, it is emphasized that adsorption dynamics, in most practical situations, are controlled by diffusion transport since the adsorption kinetic step is practically instantaneous, although exceptions to this general rule exist (Axley 1991). Therefore the introduction of a boundary layer enables a link between results obtained within experiments in the test chambers and in real rooms, where equilibrium between the concentration in the bulk air and the surface is not achieved. In boundary layer theory, the adsorbent surface may be thought to be separated from the bulk air by a film of air, which exchanges both with the surface and the bulk-air. The thin layer just above the surface is characterized by a specific concentration (Cigb) and the rate of mass transfer between the air and the air layer at the surface is proportional to the difference of concentration between the bulk and the surface layer (Cig-Cigb) multiplied by the convection rate. Mass transfer coefficients hm [m3m-2h-1] may be estimated from heat transfer coefficients by convection hc [W∙m-2K-1] according to the heat and mass transfer analogy (Axley 1991) and using the specific heat of air (CPair = 0.34 Wh∙m-3K-1):
(S2)
The international standard ISO 6946:1996 (2004) and data derived from the ASHRAE Fundamentals Handbook 2001 gives comparable values of convection heat transfer which are reported in Table S1 and S2.
Table S1 Derivation of the Heat transfer coefficient by convection using original data (htot) from ASHRAE Fundamentals 2001 in “Thermal and Water Vapor Transmission Data”. h is given in W·m-2·K-1. For buildings materials, a typical emittance of 0.9 is used
Position of surface / Horizontal / Vertical / HorizontalDirection of heat flow / Upward / Horizontal / Downward
(Effective Emittance) / 0.9 / 0.9 / 0.9
htot (ASHRAE) W·m-2·K-1 / 9.3 / 8.3 / 6.1
hrad calc. Contribution W·m-2·K-1 / 5.2 / 5.2 / 5.2
hc= (htot-hrad) W·m-2·K-1 / 4.1 / 3.1 / 0.9
The “Thermal and Water Vapor Transmission Data” chapter of Fundamentals gives a table of surface conductances. These conductances include both radiation and convection. The radiation component must be subtracted to yield the convection coefficient only. As a first approximation, a rounded value for the intermediary situations of horizontal flow of heat transfer by convection hc = 3 W∙m-2K-1 has been selected, corresponding to an air convection flux of 8.8 m3m-2h-1 according to eq S2. As a comparison, Lee et al. (2005) used a mass transfer coefficient of 6.5 m3m-2h-1 for their experiments.
Table S2 Heat transfer coefficients by convection (hc) given by ISO and ASHRAE and respective mass transfer coefficients (hm) derived using eq S2
Heat flow / hc,n[Wm-2K-1] / hm,n
[m3m-2h-1]
ISO 6946:1996 / upwards / 5.0 / 14.7
horizontal / 2.5 / 7.4
downwards / 0.7 / 2.1
ASHRAE Fundamentals / upwards / 4.1 / 12.1
horizontal / 3.1 / 9.1
downwards / 0.9 / 2.6
The product of 2.7 m2∙m-3 area per volume is adapted from a model room used by Singer et al. (2002). The value of 73.5 m3∙pers-1 volume per person is from ASHRAE (2003), yielding the overall surface area of 198 m2∙pers-1. Assuming an overall surface area of 198 m2pers-1 the air flow coming from the bulk air and passing through the boundary layer reaches a value of 2010 m3h-1. In case of very high adsorption and degradation on surfaces, all organic compounds carried by the air and entering the boundary layer could be removed. In this situation, the removal is limited by the rate at which the compounds enter the boundary layer and the apparent flow, which is determined by Kieq,n and thus by specific chemical properties. This means that the total apparent flow could be potentially strongly increased from 50 m3h-1 (Ventilation only when no adsorption occurs) up to 2060 (2010+50) m3h-1 in the case of high adsorption and degradation, leading to a potential reduction in intake fraction of a factor 40. Therefore, it is important to account for adsorption and degradation on surfaces.
As discussed above, exchanges can be assumed to be fast close to the surface and the gas in the boundary layer can be assumed to be in equilibrium with the surface. Thus the following partition coefficient between air and room surface can be defined:
(S3)
Where Cis,n [kg∙m-2] is the compound concentration in the material and Cigb,n [kg∙m-3] the concentration of the air boundary layer in contact with the surface. Note that Cis,n considers the concentration in the real or projected surface, as opposed to the actual surface which includes roughness and porosity, so that each measured Kieq,n is dependent of surface characteristics.
Work by Won et al. (2000, 2001) showed that the logarithm of this coefficient is well correlated with the logarithm of the saturation vapor pressure. Thereafter, various studies showed that sorption appears to be a significant indoor process for a large number of substances, especially those with a low vapor pressure. This indicates that for a broad range of chemicals, concentrations measured in the bulk can be significantly lower than expected without taking into account this effect. Although Kieq,n are specifics for the different materials their behavior are similar and can be correlated as presented in Fig.S1.
Fig. S1 Partition coefficient between air and room surface as a function of vapor pressure for a carpet with pad and a virgin gypsum board. Values for cyclohexane, isopropanol, toluene, tetrachloroethene, ethylbenzene, 1,2-dichlorobenzene from Won et al. (2000, 2001)
S1.4 Surface degradation
Degradation on surfaces is a more recent domain of investigation, and detailed studies are still limited. Various models were built on the results obtained in inert chambers, allowing description of processes that do not involve degradation on surfaces, such as adsorption and the influence of air renewal. However, modeled results systematically underestimated measured effective degradation (Pommer 2003 and Fick et al. 2002, 2003), strongly suggesting that a removal pathway was missing in the previous models. As a first approximation, and because values available in the literature are scarce, the surfaces degradation constant kis,deg [h-1] have been approximated here by a fixed fraction of 10% of the degradation rates in the gas phase (kis,deg=10% kig,deg). This suggests that the reactions occurring on surfaces were also due to interactions between hydroxyl radicals, ozone, nitrate radicals, and organic compounds but with slower kinetics or smaller concentrations of co-reactants. Mechanistically, this is consistent with experiments reported for building materials like linoleum (Jensen et al. 1996), PVC flooring (Knudsen et al. 1999, Wolkoff 1998), and carpet (Jolliet and Hauschild, 2005).
Supplement S2. Solving the system for n types of surfaces:
(S4)