Mishra 2014_Volatile Organic Compounds-Characteristics, distribution and...

                                                          N. Mishra et al. / Atmospheric Environment xxx (2014) 1e7                                                              3


an HP-5MS column (60 m  0.25 mm, 0.25 mm). Further details of
the GC/MS system are provided in Supplementary Table S2.
    Identification was based on a probability-based matching (PBM)
library search (McLafferty and Turecek, 1993). Mass spectra and
retention data were compared with those of measured reference
substances. All identified compounds were quantified using inter-
nal and external standards against original reference substances.
The calibration was made using external standards (containing
around 200 compounds commonly found in air samples). Stan-
dards were measured every three weeks and on demand,
depending on the sample. For the substances that were not covered
by mixed standards, single standards for external standardisation
were prepared and updated the database. Quantification with
toluene and other compounds similar to analytes was only done if
no reference standard was available. Internal standards were also
continuously used. For the calibration, the linear regression model
was used (Massold et al., 2005). In all cases the limit of detection
was <1 mg m 3.

2.4. Data analysis

    The geometric mean and standard deviation of the ambient
concentration of individual VOCs as well as total VOCs for all
                                                                                         Fig. 2. Average VOCs concentration at different locations (Michigan:(Godwin and
samples were calculated using Microsoft Excel 2007. All of the
                                                                                         Batterman, 2007); Minnesota:(Adgate et al., 2004); Malmo:(Willers et al., 1996);
concentration values were blank corrected. The average total VOCs                        Turkey: (Sofuoglu et al., 2011); Korea: (Yang et al., 2009); Sweden:(Smedje et al., 1997).
were plotted in a bar diagram against each sampling site, in order to
compare indoor and outdoor concentrations (Fig 2). The I/O ratio
for each sampling site was used in further analysis, in order to                         3. Results and discussion
determine the VOCs sources. Traffic density was calculated using
five minute averaging intervals. Average traffic count per hour was                        3.1. Meteorological conditions
calculated in all sites for sampling period.
    Principal Component Analysis (PCA) with Varimax rotation was                             The average temperature and relative humidity for each school
applied to the VOC concentration data to identify the VOC sources.                       during the measurement period (i.e. from 7 am to 5 pm over 4 days
Absolute Principal Component Score (APCS) technique was used to                          of sampling) are included in Table S3. Sampling at 11 schools was
quantify the source contribution. The detailed description of PCA-                       completed during the summer (November to April), with an
APCS method was followed as given by (Guo et al., 2004). IBM                             average temperature of 24  C, while Sampling at 14 schools was
SPSS (version 21) was used for the PCA eAPCS analysis. Before                            done in winter (May to October), with an average temperature of
applying PCA to the VOC data, individual samples were inspected                          16e18  C. Since Brisbane has a subtropical climate, there was a
for their percentage of occurrence in all schools. VOCs with 20% or                      relatively small difference in summer and winter temperatures. The
less detection rates in the samples were not included in the analysis                    average humidity varied from 30 to 71%. The wind direction data
and absolute coefficients less than 0.50 have been omitted from the                       were used to prepare wind rose plots for the entire sampling period
Tables to facilitate source identification. The missing values and                        at each school to identify the prominent wind direction either from
values below the detection limit were replaced by the geometric                          school to road or road to school.
mean and half of method detection limit, respectively.
                                                                                         3.2. Composition of air samples

                                                                                            More than 70 VOCs including very few Very Volatile Organic
                                                                                         Compounds (VVOC) and Semi volatile organic compounds (SVOC)
                                                                                         were identified in the air samples collected at the 25 schools. Those
                                                                                         compounds were separated into different groups according to their
                                                                                         chemical structure namely aliphatic hydrocarbons, aromatic hy-
                                                                                         drocarbons, aldehyde and ketones, terpenes, cyclo-aliphatics, al-
                                                                                         cohols, halogenated hydrocarbons, organic acid and esters. The
                                                                                         detailed composition of compounds within each group is provided
                                                                                         in Supplementary Table S4. Overall, more VOCs were identified in
                                                                                         indoor than outdoor air samples.

                                                                                         3.3. Ambient concentration of total VOCs and I/O ratios

                                                                                            This section presents the indoor and outdoor concentrations of
                                                                                         total VOCs (sum of all the quantified VOCs ranging from C6eC16), as
                                                                                         well as I/O ratios and comparison with other studies. In general, the
                                                                                         total indoor VOCs concentrations were found to be higher than
Fig. 1. Total VOCs (geometric mean, mg/m3) for each school. Error bar represents the     outdoor concentration, with only five schools having higher out-
std. deviation.                                                                          door concentrations as can be seen in Fig. 1. The geometric mean


 Please cite this article in press as: Mishra, N., et al., Volatile Organic Compounds: Characteristics, distribution and sources in urban schools,
 Atmospheric Environment (2014), http://dx.doi.org/10.1016/j.atmosenv.2014.10.052