Preparing Soil Samples for
Volatile Organic Compound Analysis
ALAN D. HEWITT
nant error (i.e., volatilization and biological losses)
INTRODUCTION
while assessing indeterminate error associated
Methods for the collection and analysis of soil
with various methods of preparing samples for
samples thought to be contaminated with Volatile
analysis (Hewitt et al. 1992, Hewitt 1994, Hewitt
Organic Compounds (VOCs) continues to be an
et al. 1995, Askari et al. 1996, Minnich et al. 1996).
actively researched and debated topic. Over the
Studies addressing only sample preparation have
last several years, an accumulating body of scien-
used in-vial procedures, similar to what is now
tific evidence has shown that volatilization (Urban
being recommended in draft Methods 5035 and
et al. 1989, Siegrist and Jenssen 1990, Illias and
5021. With this approach, soil samples collected
Jaeger 1993, Voice and Kolb 1993, Hewitt 1994,
in the field or prepared in the laboratory are trans-
Lewis et al. 1994, Hewitt et al. 1995, Hewitt and
ferred directly to hermetically sealed vessels that
Lukash 1996) and preservation (Hewitt 1995a,b;
contain an organic solvent or from which the anal-
Turriff et al. 1995; Hewitt 1997) issues were not
ysis could be done.
adequately addressed by Method 5030 (Purge-
Results from using an in-vial approach have
and-Trap) and Section 4.1 (Sampling) of the U.S.
clearly shown that different sample preparation
Environmental Protection Agency's Test Methods
methods do not produce the same quantitative
values. For example, samples prepared by aque-
ous dispersionextraction often yield lower VOC
To address these concerns, the U.S. EPA has drafted
concentration estimates than MeOH extraction
two new sample collection and analysis proce-
(Hewitt et al. 1992, Hewitt 1996, Askari et al. 1996,
dures, Methods 5035 and 5021, for the third update
Minnich et al. 1996). Explanations for these differ-
of this document, while, independently, several
ences have often focused on the organic carbon
States have either considered or adopted the use
content in the soil matrix and the octanolwater
of an in-field methanol (MeOH) sample preserva-
partition coefficient of the specific VOC being
tionextraction method.
measured.
Distinguishing which method of sample prep-
This study examines potential differences
aration for VOC analysis is best is not a simple
among various methods used to prepare discrete
task. This is because most of the evidence show-
soil (grab) samples for VOC analysis. The extrac-
ing that quantitative VOC soil determinations de-
tion solvents studied were MeOH, tetraethylene
pend on sample collection, handling, and analy-
sis fails to identify and separate the determinant
glycol dimethyl ether (tetraglyme), and poly
(systematic) and indeterminate (random) errors
(propylene)glycol (PPG). In addition, three equi-
associated with each of the steps involved. On the
librium headspace analysis methods were also
basis of this total measurement error approach,
studied--direct heating (H-HS), dispersion
several methods of sample preparation (extrac-
extraction in water acidified with sodium bisul-
tionequilibrium) for analysis are in use without
fate (Aq-NaHSO4-HS), and dispersionextraction
a clear understanding of how each variable quan-
in water saturated with sodium chloride and acidi-
titatively influences the results. Only a few stud-
fied with phosphoric acid (Aq-NaCl sat'd-HS). The
ies, to our knowledge, have minimized determi-
H-HS (thermal desorption) procedure was consis-
Previous Page
