Huffman et al.
elemental composition of PM2.5, there has been relatively
little research on its molecular structure and microstruc-
ture. Many scientists believe that both the effects on hu-
man health and the source apportionment of PM2.5 are
closely related to parameters such as particle size distri-
butions and morphology and the valence and solubility
of critical elements. It is therefore essential to identify and
evaluate analytical methods that can provide such struc-
tural information.
Four residual fuel oils were combusted, with sulfur
contents ranging from 0.53 to 2.33 wt %. The ultimate
analyses of these oils, together with the concentrations of
the metals of interest for this paper, are given in Table 1.
Table 2 contains the loss on ignition (LOI) and metal con-
centrations for the PM2.5 and PM2.5+ fractions. The metal
analyses were performed using acid digestion and ICP/MS.10
Although burnout was fairly complete (> 99.7%), the inor-
ganic content of the oils was quite low (0.02–0.10 wt %
ash), and the LOI results indicate that the dominant ele-
ment of the ROFA is carbon (64–87 wt % for PM2.5 and 88–
97 wt % for PM2.5+). V is present at relatively large
concentrations (~0.5–5 wt %); Ni, Fe, and Zn are present at
moderate concentrations (~0.1–0.5 wt %); and several met-
als (Pb, As, Cr, Cu, Mn, Sb, and Cd) are present at concen-
trations that are rather low (~20–1000 ppm) but could still
be significant for health considerations. The metals are typi-
cally more concentrated in the PM2.5 samples than in the
PM2.5+ samples by factors from ~3 to 6.
X-ray absorption fine structure (XAFS) spectroscopy is
a synchrotron radiation-based technique that is uniquely
well-suited to characterization of the molecular structure
of individual elements in complex materials. In previous
research, we have used XAFS spectroscopy to determine
the molecular forms of environmentally important ele-
ments (e.g., S, Cl, As, Cr, Hg, Ni, etc.) in coal, oil, fly ash,
and sorbents.1-7 Our initial investigations of PM indicate
that XAFS will also be a powerful tool in this area.8,9
In the current work, XAFS spectroscopy and a num-
ber of additional analytical techniques were applied to a
suite of residual oil fly ash (ROFA) samples separated aero-
dynamically into PM2.5 and PM2.5+ fractions. Briefly, the
characterization data obtained included
XAFS SPECTROSCOPY RESULTS
The samples were investigated by XAFS spectroscopy at
the Stanford Synchrotron Radiation Laboratory (Menlo
Park, CA) and the National Synchrotron Light Source at
Brookhaven National Laboratory (Upton, Long Island,
NY). All measurements were carried out in the fluores-
cent mode using either a Lytle detector or a multi-element
Ge array detector, as described elsewhere.1-4 The X-ray ab-
sorption near edge structure (XANES) regions of the spec-
tra were analyzed by deconvolution, derivative, and
comparative analysis methods, as discussed in earlier pa-
pers.1-7 The results for the elements investigated to date
are summarized below.
(1) XAFS analysis of the molecular structure of S, V,
Ni, Fe, Cu, Zn, Pb, and As;
(2) computer-controlled scanning electron micros-
copy (CCSEM) analysis of particle size, composi-
tion, and morphology;
(3) 13C nuclear magnetic resonance (NMR) analysis
of the molecular structure of carbon, the domi-
nant element;
(4) X-ray diffraction (XRD) identification of crystal-
line phases; and
(5) inductively coupled plasma/mass spectrometry
(ICP/MS) determination of metal concentrations.
Sulfur
Typical S K-edge XANES spectra of ROFA PM2.5 and PM2.5+
samples are shown in Figure 1. The spectra are deconvoluted
by a least-squares computer analysis into a series of peaks
(50% Lorentzian–50% Gaussian) and two rounded arctan-
gent step functions, as discussed elsewhere.1,2 Most of the
peaks represent 1s→3p transitions of photoelectrons ex-
cited from the K-shell by X-ray absorption. Both the posi-
tion and relative intensity of these peaks vary significantly
with the electronic state of the S atom, increasing with
increasing valence. By using calibration data generated from
mixtures of standard compounds, the peak area percent-
ages can be translated into percentages of the total S con-
tained in different molecular forms.1,2
COMBUSTION PROCEDURE
The combustion experiments were carried out in a North
American three-pass fire tube package boiler, which is a
practical, commercially available heavy fuel oil combus-
tion unit. A detailed description of this boiler is given
elsewhere.10 Samples were separated aerodynamically by
a cyclone into PM2.5 and PM2.5+ fractions. The sampling
system consists of a large dilution sampler capable of
isokinetically sampling 0.28 m3/min (10 ft3/min) of flue
gas using a Source Assessment Sampling System (SASS)
cyclone. Details on the construction and operation of this
sampling system are available elsewhere.11 The SASS cy-
clone produces 50 and 95% collection efficiencies at ap-
proximately 1.8 and 2.5 µm diameters, respectively. The
resulting PM is collected on large (65-cm) Teflon (DuPont)-
coated glass fiber filters, transferred to sampling jars, and
made available for analysis.
The results of this analysis for the ROFA PM samples
are summarized in Table 3. The dominant molecular forms
of S observed are sulfate and thiophenic S. Sulfate was
greater in the PM2.5 samples than in the PM2.5+ samples,
reflecting the greater degree of carbon burnout for the
Volume 50 July 2000
Journal of the Air & Waste Management Association 1107