Chemical Analyses of Particle Filter Deposits* - Aerosols: Measurement, Dosimetry, and Health Effects - page 194

Signals from electron bombardment of the particle include secondary and backscattered elec-
trons along with characteristic x-rays. These are detected in synchronization with the position of
the beam to provide size, morphological, and chemical information. The secondary electron sig-
nal yields an image with a three-dimensional perspective, high depth-of-ield, and the appearance
of overhead illumination. The backscattered electron signal provides compositional information
because the signal intensity depends on the atomic number elements in the feature being examined.
The x-ray spectrum is the same as that obtained by XRF, which uses a photon to ionize the atom in
place of the SEM electron. Due to different shapes of particles and different mixtures, SEM x-ray
spectra provide semiquantitative elemental concentrations; but the ratios of peak areas are similar
for similar particles types, thereby providing a “ingerprint” for that type of particle. In other cases,
particle shapes may be similar, but they are distinguishable by their elemental proiles. This is
typically the case for spherical particles that form during high-temperature combustion and occur
naturally as pollens.
It is not practical to characterize every one of the million or more particles on a single air ilter.
The number of particles within a category follows a Poisson distribution, for which the counting
error is proportional to the square root of the number of counts. At least 100 particles must be
counted within a category to obtain a counting error (1 standard deviation) of less than ±10%.
If 10 categories are desired, at least 1000 particles must be characterized, and this is not practical
by manual methods.
As shown in Figure 7.1, polycarbonate-membrane ilters are among the best choices for CCSEM
because they have smooth and uniform surfaces with uniformly sized holes that can be recognized
and ignored by the CCSEM. The ringed Telon-membrane ilter has a nonuniform, ibrous appear-
ance under the SEM, which interferes with automated particle detection. Inert particles can be
removed from the Telon-membrane ilter and redeposited on a smooth surface for CCSEM by
sonicating the ilter in a solvent, such as acetone, and iltering the suspension through another
ilter such as etched polycarbonate. This results in loss of soluble species, but it is appropriate for
nonsoluble minerals. Particles also must be coated with a conducting substance—usually carbon,
gold, or platinum.
7.4.14 d ata M anageMent and v alidation
Modern analysis software allows data to be output into comma delimited, Excel, or Access iles.
Microsoft Access (Redmond, WA) is a useful and commonly available relational database that
allows data from ield sampling and various laboratory analyses to be uniied. Each record should
contain at least the analysis ID, date and time of analysis, analyst ID, value obtained by the analysis,
an analysis lag (e.g., irst measurement, replicate measurement, standard, blank, or audit stan-
dard), and a validation lag (indicating an unusual situation). Simple programs can be written in
the database language to convert instrument output (e.g., μg/mL, μg/cm 2 ) to μg/sample. Replicate
analyses performed during the time period of the original analyses are used to estimate precision
of the measurement for different concentration levels (Watson et al., 2001b). Field blanks or backup
ilters (see Figure 7.1) are used to estimate biases for passive deposition or adsorption, with averages
subtracted and standard deviations propagated (Chow et al., 2010b). The remaining concentration is
then divided by sampling low rate times the sample duration to obtain ambient concentration, with
uncertainties propagated to the inal concentration (Watson et al., 2001b).
Once the concentrations have been calculated, data can be further evaluated for physical
consistency (Chow et al., 1994): (1) the sum of measured species must be less than the mass
concentrations; (2) water-soluble K + , Cl − , and SO 4 = —measured by AAS, IC, or AC on quartz-iber
ilters—should be less than total K, Cl, and three times S concentrations measured by XRF on
Telon-membrane ilters, respectively; (3) cations should balance anions when all water-soluble
ions are measured. Additional internal consistency tests can be applied depending on the overlap