Citation
Comparison of extraction efficiencies for spiked versus native PCBS in sedimentary water and PAHS in wet clay using EPA standard methods and a closed extraction system

Material Information

Title:
Comparison of extraction efficiencies for spiked versus native PCBS in sedimentary water and PAHS in wet clay using EPA standard methods and a closed extraction system
Creator:
Tilbury, Marshall Dean
Place of Publication:
Denver, CO
Publisher:
University of Colorado Denver
Publication Date:
Language:
English
Physical Description:
ix, 95 leaves : illustrations ; 28 cm

Subjects

Subjects / Keywords:
Extraction (Chemistry) ( lcsh )
Extraction (Chemistry) ( fast )
Genre:
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )

Notes

Bibliography:
Includes bibliographical references (leaves 93-95).
Thesis:
Department of Chemistry
Statement of Responsibility:
by Marshall Dean Tilbury.

Record Information

Source Institution:
|University of Colorado Denver
Holding Location:
|Auraria Library
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
36401686 ( OCLC )
ocm36401686
Classification:
LD1190.L46 1996 T55 ( lcc )

Full Text
COMPARISON OF EXTRACTION EFFICIENCIES FOR SPIKED
VERSUS NATIVE PCBS IN SEDIMENTARY WATERS AND PAHS
IN WET CLAY USING EPA STANDARD METHODS AND
A CLOSED EXTRACTION SYSTEM
Marshall Dean Tilbury
B.S., University of Colorado
at Colorado Springs, 1983
A thesis submitted to the
Faculty of the Graduate School of the
University of Colorado at Denver
in partial fulfillment
of the requirements for the degree of
Master of Science
1996


1996 by Marshall Tilbury
and
Quanrterra Inc.
All rights reserved


This Thesis for the Master of Science
degree by
Marshall Dean Tilbury
has been approved
by
Paul C. Winkler
Donald C. Zapien
L 9/
Date
John Lanning


Tilbury, Marshall Dean (M.S. Chemistry)
Comparison of Extraction Efficiencies for Spiked Versus Native
PCBs in Sedimentary Waters and PAHs in Wet Clay Using EPA
Standard Methods and a Closed Extraction System
Thesis directed by Adjunct Professor Paul C. Winkler and
Assistant Professor Donald C. Zapien.
ABSTRACT
EPA approved groundwater and soil extraction methodologies
are time consuming to prepare, consume large solvent amounts, and
utilize many glassware components. Extraction methodologies are
evaluated using analyte spiking techniques that may not simulate
real samples.
Solid phase extraction (SPE) is an accepted drinking water
extraction method that provides accepted precision and accuracy
and decreased glassware, solvent usage, and preparation time as
compared to 1iquid/1iquid extraction (LLE) but isnt approved for
sediment containing groundwaters due to clogging of the
apparatus.


Initial preparation of a standard soil extraction
methodology involves manual dispersion which doesnt expose the
full matrix to the extraction solvent and produces reduced
analyte recoveries.
Novel preparation methods were developed to extract PCBs
from groundwaters via a semi-automated closed SPE/solids
extraction system and PAHs from wet clay matrices via mechanical
matrix dispersion followed by a closed extraction system.
To evaluate the novel methods against EPA approved methods,
simulated PCB groundwater and homogeneously PAH spiked wet clay
matrices were synthesized for analyte percent recovery
determinations. To evaluate the applicability of utilizing
traditionally spiked matrices to validate a method,
uncontaminated wet and dry clay as well as particulate containing
groundwaters (0 to 5 gram solids) were spiked using traditional
methods and extracted using the approved and novel techniques to
compare spiking recoveries to the simulated matrixs analyte
recoveries. The simulated PCB groundwaters solid and aqueous
fractions were extracted over a week time period to determine the
PCB/ matrix equilibration time as well as the fractions
partition ratios with relation to solids loading.
Simulated groundwater extraction results indicated fast
aqueous/solids equilibrium time of PCBs and determined a


formidable PCB aqueous/solid partition coefficient (Kp=376). The
SPE methods PCB recoveries were superior to the LLE methods PCB
recoveries due to LLEs incapability to extract the solids
fraction.
The PAH soil extraction results indicated that a
homogeneously spiked wet clay matrix was achieved, and this
simulated matrixs PAH recoveries were much lower than
traditional PAH spiking recovery results. The closed extraction
system PAH recoveries were superior to the standard methods PAH
recoveries due to the mechanical dispersion step.
This abstract accurately represents the content of the
candidates thesis. I recommend its publication.
Signed
Paul C. Winkler
Signed
Donald C. Zapien


CONTENTS
List of Figures.............................................vii
List of Tables.............................................. ix
Chapter
1. Introduction.............................................I
1.1 LLE/ SPE Background..................................1
1.2 Soil Extraction Background...........................7
2. Preliminary Method Development...........................12
2.1 PCB/SPE Method Development..........................13
2.1.1 Initial PCB/SPE Studies on Particulate/
Nonparti cul ate Waters............................13
2.1.2 Preparation of a Simulated PCB Groundwater Matrix..19
2.1.3 SPE with Inverted Filter and Encapsulated
Cartridge..........................................20
2.1.4 SPE with Inverted Filter and Standard Cartridge-----23
2.1.5 SPE Custom Cartridges...............................26
2.1.6 SPE Solvent Elution Studies.........................28
2.1.7 Closed Extraction System for PCB Groundwaters.......29
2.2 Soil Extraction Method Development..................32
2.2.1 Soil Dispersion Studies.............................32
2.2.2 Soil Matrix Sonication Studies......................36
I V


2.2.3 Closed Extraction System for PAH Soils..............38
2.2.4 Inline Extract Concentration Module..................40
2.2.5 Preparation of a Simulated Homogeneously Spiked
Wet Clay Matrix.....................................41
3. Evaluation of the Closed Extraction System...............44
4. Closed Extraction System Setup and Operation ............46
4.1 Matrix Preparation..................................46
4.1.1 PCB/Groundwater Matrix Preparation...................46
4.1.2 PAH Soil Matrix Preparation..........................47
4.2 EPA Standard Extractions.............................49
4.2.1 PCB/Water Preparation Via EPA Method 3510............49
4.2.2 PAH/Soil Preparation Via EPA Method 3550.............50
4.3 Closed Extraction System Setup and Operation............52
4.3.1 PCB/Water In Situ Preparation........................52
4.3.1.1 Closed Extraction System Assembly and Operation
for PCB Groundwaters................................52
4.3.1.2 SPE Cartridge Preparation...........................53
4.3.1.3 PCB/SPE Extraction and Elution......................53
4.3.1.4 PCB Extract Cleanup.................................55
4.3.2 Closed Extraction System Setup and Operation
for PAH/Soil .......................................55
4.3.2.1 Matrix Spiking and Dispersion.......................55
4.3.2.2 Closed Extraction System Assembly...................56
V


4.3.2.3 PAH/Soil Extraction and Concentration..............58
4.4 PCB Analysis.......................................58
4.5 PAH Analysis.......................................60
5. Results and Discussion..................................62
5.1 PCB Equilibration Time Between Solid/Aqueous
Matrix.............................................62
5.2 PCB Solid/Aqueous Extraction Efficiencies..........67
5.3 PCB Solid/Aqueous Partition Ratio..................71
5.4 PCB/Groundwater SPE Vs LLE Extraction Efficiencies.73
5.5 PAH/Soil Standard Extraction Results...............76
5.6 Closed Extraction System PAH/ Soil Results.........78
5.7 PAH/Soil Standard Vs Closed Extraction System
Efficiencies.......................................80
5.8 Soil Dispersion/Sonication Comparisons.............82
5.9 Homogeneously Spiked Wet Clay Discussion...........86
6. Conclusions.............................................87
Bibl iography..............................................93


FIGURES
Figure 2.1 SPE with Inverted Filter and Encapsulated
Cartridge......................................21
Figure 2.2 SPE with Inverted Filter and Standared
Cartridge.................................... 25
Figure 2.3 Clogging of a Standard SPE Cartridge............27
Figure 2.4 Closed Extraction System of PCB Groundwaters..31
Figure 2.5 Custom Mechanical Matrix Dispersion Apparatus.34
Figure 4.1 Closed Extraction System........................59
Figure 5.1 Percent PCB Recovery Vs Contact Time for SPE
Aqueous Fraction...............................63
Figure 5.2 Percent PCB Recovery Vs Contact Time for SPE
Solids Fraction................................64
Figure 5.3 Percent PCB Recovery Vs Contact Time for SPE
Method.........................................65
Figure 5.4 Percent PCB Recovery Vs Contact Time for LLE
Method.........................................66
Figure 5.5 Percent PCB Recovery for SPE VS Summed SPE
Aqueous and Solids Fractions...................68
Figure 5.6 Percent PCB Recovery for SPE Aqueous Fraction
Vs Corrected LLE Recoveries....................70

VM


Figure 5.7 Percent PCB Recovery Vs Amount of Clay for SPE
and Summed SPE Aqueous and Solids Fraction______74
Figure 5.8 Percent PCB Recovery Vs Amount of Clay for
SPE and.........................................75
V II I


TABLES
Table 2.1 SPE Disk Extraction Results.......................17
Table 5.1 Native Vs Wet Spike Vs Dry Spike PAH Recoveries
Via Manual Dispersion; Horn Sonication...........77
Table 5.2 Native Vs Wet Spike Vs Dry Spike PAH Recoveries
Via Mechanical Dispersion; Bath Sonication......79
Table 5.3 Standard Method Vs Novel Method for Native PAH
Soil.............................................81
Table 5.4 Horn Sonication Vs Bath Sonication................83
Table 5.5 Mortar and Pestle Dispersion Vs Optimum Manual
or Mechanical Dispersion Using Horn Sonication.85
Table 6.1 Native Mortar and Pestle Vs Native and Wet
Spike Mechanical Dispersion......................91


1. INTRODUCTION
1.1 LLE/ SPE Background
The determination of the total amount of trace organic
pollutants contained in an aqueous matrix requires that the
organic compounds present in the matrix be manipulated to be
compatible with the preparative analytical method (and vise
versa), the analytical instruments capabilities, and the
regulatory requirements. To achieve the required method detection
limits, the compounds that are extracted from the matrix are then
concentrated to achieve the instruments detection limit. The
extraction method must be accurate and precise and would
optimally be simple, expediant, and inexpensive to prepare as
well as free of instrumental interferences.
The current aqueous sample preparations promulgated by the
Environmental Protection Agency (EPA) to determine organic
analytes primarily utilizes liquid-liquid extractions (LLE). LLE
employs a water immiscible solvent to partition the organic
compound of interest from the aqueous phase into the solvent of
choice. One type of LLE for the determination of PCBs is the
shakeout method EPA Method 3510 (1) in which a liter of aqueous
l


matrix spiked with a recovery surrogate and shaken for two
minutes with 100ml of methylene chloride in a two liter
seperatory funnel. The organic solvent is removed and the matrix
is extracted twice more with subsequent 100ml aliquots of the
solvent. The three solvent aliquots are combined, concentrated,
and exchanged to the instrument compatible hexane. Another
popular LLE method is continuous liquid/ liquid extraction
(CLLE), EPA Method 3520 (2), in which 250ml of methylene chloride
is continually refluxed through the surrogate spiked matrix. The
solvent is then concentrated and exchanged to hexane as in the
shakeout method. The success of the partition of the analyte from
the aqueous phase into the organic phase will depend on the
partition coefficient (Kp). The LLE partition coefficient is
defined as Kp= C1/C2 where Cl is the concentration (ug) of
analyte per extraction solvent volume (ml) and C2 is the
concentration of analyte (ug) per extracted aqueous matrixs
volume (ml) when equilibrium is achieved. It can be seen that a
high Kp value provides a more complete extraction and that
repetitive extractions with clean solvent will exhaust the
analyte concentration in the aqueous fraction. One problem
identified with the shakeout method on sedimentary waters is
emulsion formation that results in loss of analyte and increased
preparation time. The problem of CLLE is long preparation times.
2


Both LLE methods utilize large amounts of solvent and glassware
and are very labor intensive. EPA Method 3510 (LLE) does not
address the issue of settleable solids extraction. Should the
entire matrix be extracted, or should the sample be decanted
before extraction to exclude solids extraction? The problems of
suspended and settleable solids will be addressed throughout this
thesis.
Solid Phase Extraction (SPE) has been recently approved for
the extraction of drinking waters (sediment free) for the
determination of polychlorinated biphenyls (PCBs) as well as
other organic analytes in EPA Method 525.1 (3).The phase employed
for hydrophobic organic analytes is a monolayer of C18 bonded to
an end capped 40um silica bead or a disk membrane. The bead C18
phase is encased with 20um frits in a standard polypropylene
syringe barrel to form the cartridge SPE design. The basic
extraction method is comprised of pre-eluting residual
contaminants (primarily phthalates carried over from
manufacturing) off the hydrophobic phase with the final elution
solvent. The phase is then solvated by conditioning the phase
with a few bed volumes of methanol as the wetting agent. The
matrix is then ready to extract by the deposition step which
involves passing a one liter water sample (with a 0.5% MeOH
modifier added to the matrix to keep the phase solvated) through
3


the phase for a period of 10 to 120 minutes (depending on the
method employed) followed by a 5 minute air drying of the
residual water from the phase. The organic compounds adsorbed to
the phase are then eluted with 3-5ml of the appropriate solvent
which is then adjusted to a predetermined final volume for
subsequent analysis. The adsorption/desorption of the analyte on
the C18 phase is determined by a 1 iquid/solid partition
coefficient (Kp) which is defined as Kp=S (mass of compound(ug)
/g of solid (C18))/(mass of compound(ug)/ ml of solution). A
successful SPE extraction/ concentration of an aqueous matrix
will exhibit a very large Kp for the deposition step (using a
minimal amount of C18 for the maximum volume of water) and a very
small Kp for the elution step (a minimal volume of solvent).
Method 525.1 provides acceptable recoveries (>80% of PCBs from
drinking waters) while minimizing the volume of solvent needed
per sample (30ml). Commercial SPE cartridge manifold systems for
the extraction of organic analytes from drinking waters are
inexpensive, utilize minimal glassware, and are capable of
processing fifty aqueous samples a day if performed by an
experienced analyst. Disadvantages of SPE are low adsorption
capacity (5% w/w analyte capacity in relation to the mass of the
phase which exhibits analyte losses for high organic content
matrices), cross contamination from the glass frit when utilizing
4


the SPE disks, and analyte losses attributed to channeling of the
water immiscible elution solvent through the phase during the
analyte desorption step (due to incomplete removal of water from
the phase during the air drying step). SPE extraction is not
approved for surface waters and groundwaters due to suspended and
settleable solids clogging the SPE phase. An approved SPE method
for the preparation of organic compounds in aqueous matrices
would be cost effective due to higher productivity, inexpensive
equipment costs, and decreased solvent usage and solvent waste.
The development and validation of an SPE preparatory method
should be applicable for all of the matrices expected for
analysis (including the sedimentary waters).
The group discussions at the Second Environmental SPE
Symposium on July 29-30, 1993 at Denver, Colorado addressed the
shortcommings of SPE technology on extracting actual sedimentary
waters. The issues that predominated the symposium were
particulate clogging of the SPE apparatus, the analytical
integrity of filtering the samples before SPE is performed, and
the effectiveness of the passive solvent extraction of the fines
that are collected on SPE discs. The extraction efficiency of the
target compounds from the matrixs solids during 1iquid/1iquid
extraction (LLE) is also unanswered. The poor solvent extraction
recoveries of an analyte sorped to a solid particle in an aqueous
5


matrix may be explained by examining the opaline silica particle.
When the particle is integrated in an aqueous medium, the surface
and pores are hydrated initiating the clay sheets expand
anisotropically allowing the hydrophobic PCB to adsorb in and on
the particle while high water surface tension is created over the
pores and crevasses. When the water immiscible solvent used for
LLE (such as dichloromethane) comes in contact with the hydrated
particle, the solvent cannot break the waters surface tension
barrier and therefore may not extract the adsorbed PCBs from the
particle.
Articles (4-12) discussing organic analyte adsorption to
humic and fulvic acids as well as settleable and suspended
particles in aqueous matrices report a liquid/solid partition
coefficient as high as log Kp ~5 and also note the adsorption of
analytes to glass sample containers. The percent carbon contained
in the solids govern the magnitude of the Kp with the high carbon
content humic and fulvic acids having greater reported Kps than
the low carbon content clay sediments. Typical water samples
exhibit varied amounts of suspended and settleable solids.
Qualitative visual estimates indicates most samples contain some
particulate matter that very rarely exceeds 5 grams of suspended
and settleable matter per liter of matrix prepared. Since the
suspended and settleable solids present in the aqueous matrices
6


may be considered a soil matrix, further discussion of the
interactions of organic analyte with the matrix will be discussed
in the next section entitled "Soil Extraction Background". The
analyte adsorption indicates that spiking a non-sedimentary water
for method validation doesnt account for analyte losses due to
adsorption and the validated method will not apply to real
samples.
It is therefore imperative that to develop and validate an
SPE method for the analysis of PCBs, real sedimentary matrices
with a known amount of naturally sorbed analyte must be extracted
and evaluated against the standard LLE method results via PCB
percent recovery data for method equivalency determinations.
Method development and validation data should also provide
information as to the liquid/solid partition ratios of the PCBs
between the aqueous and solid fractions of typical matrices and
compare the SPE and LLEs effectiveness in the extraction of each
of the fractions.
1.2 Soil Extraction Background
Contamination, quantitation, and remediation studies of
PAHs and other organic compounds in solids/soils are prevalent
in the literature (13-19). The precise science of environmental
chemistry relies strongly on extraction and validation
7


methodologies. As discussed by Dzombak (14), PAHs interact with
soil matrices via advection, dispersion, and adsorption. That is,
the hydrophobic PAH molecule enters a hydrated alumnosilica sheet
on the soil surface by way of anisotropic expansion and disperses
into the matrix via mass transport where it will eventually
adsorp via hydrophobic bonding (nonactive adsorption), and
therefore, the analyte may be thoroughly spread throughout a
given matrix. Aqueous leaching studies (14) of the contaminated
matrix indicates that the adsorption is irreversible utilizing an
aqueous solvent. The extraction efficiency of a preparation
method governed by the extent of desorption of the analyte
adsorbed in a soil matrix into the extraction solvent is
determined by a liquid/solid partition coefficient (Kp) which is
defined as Kp=S (mass of compound(ug)/ g of soil)/(mass of
compound(ug)/ ml of solution)(10). As with the preparation of
aqueous matrices, soil preparation methodologies must be accurate
and precise and optimally simple, expedient, and inexpensive to
prepare.
The current soil preparations promulgated by the EPA utilize
an organic solvent extraction of the matrix. A widely used
extraction procedure for PAHs is the sonication method, EPA
Method 3550 (20) in which 30 grams of matrix is manually mixed
with sodium sulfate, a drying agent used to produce a dry,
8


freeflowing powder. The manual dispersion is performed by
stirring the matrix/ sodium sulfate contained in a glass beaker
with a steel spatula. The resulting dispersed matrix clumps when
a wet clay matrix is prepared that rarely produces the EPA
recommended free flowing powder. A 100ml methylene chloride
aliquot is then added to the mixture with subsequent spiking and
sonication for two minutes with a 350 watt, 20 Kz, 50% duty
direct ultrasonic pulse. The organic solvent is collected and two
subsequent 100 ml MeCl2 sonications are performed. The three
solvent aliquots are combined and concentrated for analysis.
Another widely used soil extraction method is soxhlet extraction,
EPA Method 3540 (21) in which 250 ml MeCl2 is continually
refluxed for an 18 hour passive extraction through the 30 gram
matrix/ sodium sulfate manually dispersed mixture with subsequent
concetration of the extract. Both methods consume >250 ml of
solvent except that the sonication method is much quicker (one
hour per sample as compared to 19 hours via soxhlet) due to an
increased analyte desorption rate. Sonication of a solvated
matrix produces numerous gas phase bubbles produced at nucleation
sites located on the surface of the soil matrix and desorps the
analyte from the surface/ near surface of the soil by the
physical shockwave induced from the cavitation of the nucleation
bubble.
9


Although the standard extraction and validation operating
procedures promulgated by the EPA are ubiquitously employed,
their application to.real matrices may not reflect the true
analyte content. Spiking a matrix with an organic analyte for
validation studies or surrogate recovery determinations will not
simulate the natural interactions with a matrix but will only
produce surface adsorption of the analyte. The true analyte
concentration in difficult matrices to produce a reference soil
standard are presently determined by performing an exhaustive
extraction procedure on actual contaminated soil matrices. NIST
PAH soil concentrations are determined on real matrices by an
exhaustive 48 hour soxhlet extraction (13,16), but those
certified amounts may not accurately reflect the matrices true
analyte content. The soil extraction of PAH spiked soils measured
against NIST PAH soils persuaded Hawthorne (13) to state that
"Despite the common practice of using spike recoveries to develop
and validate extraction/preparation methods, little data exist
that describe the relative behavior of spiked and native analytes
during extraction from complex environmental solids".
The most reliable method validation experiment would
optimally utilize a homogeneously spiked difficult matrix of
known concentration to generate percent recovery data. A
homogeneously PAH spiked wet clay matrix is currently unavailable
10


due to the inability of using water for the reconstitution step
of a homogeneously spiked dry matrix (the homogeneously spiked
dry matrices are available). When a homogeneously spiked dry clay
is reconstituted with water, the water not only separates the
fine from course particles, but also leaches the spiked analyte
throughout the matrix producing a heterogeneous matrix.
11


2. Preliminary Method Development
The purpose of this investigation is to build and evaluate a
preparation system that may be used on the wide variety of
groundwater and soil matrices received at an environmental
laboratory. The novel extraction system will be evaluated by
utilizing simulated difficult matrices to ensure all matrices
will be applicable to the method. Due to the problems associated
with the present technologies extraction and evaluation
techniques (discussed in the Introduction Section 1), each
problem was examined individually and is discussed in the
following Preliminary Method Development experiments. Once the
problems were solved, components were be constructed to evaluate
a semi-automated closed extraction system as proposed in
Evaluation of the Closed Extraction System Section 3. (the
technical specifications are presented in the Closed Extraction
System Setup and Operation Section 4.) The following preliminary
method development experiments listed in chronological order
addresses fractionating a groundwater matrix for SPE extraction
of the aqueous phase and sonication extraction of the solids
phase. The soil experiments develops a method to prepare the
matrix via dispersion for a novel bath sonication extraction.
12


Simulated matrices will also be developed for evaluation of the
novel extraction method. Each preliminary method development
experiment will specify the problem to be solved, the
experiments techniques and results, and the application (if any)
to the complete extraction system.
2.1 PCB/SPE Method Development
2.1.1 Initial SPE Studies on Particulate/ Nonparticulate Waters
PCB SPE method development was initiated to determine the
effectiveness of SPE on one liter matrices of clean carbon
filtered water. Traditional SPE methodologies following the
manufacturers suggested validation procedures and using
cartridges and disks were employed with agreeable results.
Initial SPE disk extractions studies of PCBs on real
sedimentary water samples was initiated to determine SPEs
effectiveness on real samples.
PCB groundwaters (with suspended/settleable solids) were
extracted following a modified procedure by ODonnel,etal
entitled "Evaluation of Liquid/Solid Extraction for the Analysis
of Organochlorine Pesticides and PCBs in Typical Ground and
Surface Water Matrices" (internal publication). Efficiency of the
extraction will be measured as well as cross contamination
aspects and cost/time analysis.
13


Actual duplicate groundwater samples (with varying amounts
of solids) that were determined to be PCB free (samples 6-19 and
26-31), four duplicate groundwaters with natural PCB
contamination quantified by the shakeout method (samples 22-25),
and two carbon filtered water samples used for method blanks
(samples 20 and 21) were extracted via disk SPE with the results
displayed in table 2.1.
Extraction Summary: The one liter ground water samples were
spiked with 5% MeOH (a C18 wetting agent), 0.25ug 3 Bromo
Biphenyl (3BrBP) and 0.05ug decachlorobiphenyl (DCB) as recovery
surrogates, and varying concentrations of AR 1254 (0.5-100ug).
Two carbon filtered water matrices were spiked with 0.25ug 3
Bromo Biphenyl (3BrBP) and 0.05ug decachlorobiphenyl (DCB) as
recovery surrogate to be used as method blanks. All samples were
shaken for 15 seconds to disturb the settleable solids and set
aside for one to three days to enhance PCB adsorption to the
fines to simulate sampling and transport conditions. A standard
47mm CIS Empore disk extraction workstation was erected using a
vacuum pump and glass fritted 47mm filter holders for the disk
support. A 300ml reservoir was implemented for sample
introduction and a 47mm Whatman graded density glass microfibre
filter was used on top of the SPE disk to provide a filtration
medium for the fines.
14


The disks were preconditioned with 10ml hexane and vacuumed
to dryness followed by 10ml MeOH, 2X10ml CF H2O, and then the
groundwater. After the initial hexane elution, the disks were not
allowed to go dry (small amount of liquid was always just above
the top filter) and the sample was carefully decanted as not to
disturb the settleable solids until the end of the extraction
when the remaining solids were applied to the disk. After the
sample eluted through the disks, a two minute drying time was
provided to remove bulk water left on the fines/disks. A 40 ml
VOA vial was positioned below the disks for solvent collection
and a 1 ml MeOH portion was eluted through the fines/disks to
remove any residual water followed by 5ml hexane (that was
previously used to rinse the original sample bottle) followed by
an additional 5 ml portion of hexane. The hexane was
intermittently pulled through the disk to provide a
hexane/fines/disks minimum residence time of three minutes.
The VOA vials were then shaken for two minutes followed by
Pasteur pipetting of the excess water/MeOH. The volume was
adjusted to 10.0ml hexane followed by a 2 minute 20ml H2SO4
drying/cleanup shake. 5 ml extract was transferred to a 12 ml
vial for subsequent analysis. The glassware was prepared before
each use by a hot water/soap wash followed by an acetone rinse.
15


The glass frit was cleaned in a beaker of hexane immersed in an
ultrasonic bath for 5 minutes.
The modifications made on this prep as compared to
ODonnels: hexane was substituted for ethyl acetate and acid
drying was substituted for sodium sulfate drying of the extract.
Analytical results for the SPE disk extractions of PCB
groundwaters are displayed in Table 2.1.
In a private communication with 0Donne! after the
experiment was completed it was discovered that they spiked
undisturbed groundwaters and extracted immediately (no
appreciable PCB/solid contact time). They did attempt extracting
spiked groundwater samples with 1-3 day contact times and
achieved poor recoveries.
Reproducing this SPE procedure of PCB groundwaters indicated
extended elution times of the aqueous matrix due to clogging of
the SPE disk (up to 140 minutes). The glass fiber filter and
glass frit is suspected of being a source of strong DCB
adsorption due to the low recoveries in the method blanks. The
high 3BrBP recoveries in the method blanks may be system
carryover from previous groundwater extractions. The low PCB and
surrogate recoveries exhibited in the spiked groundwaters may be
attributed to analyte adsorption to the fines, glass fiber
filter, and/or the glass frit. The real PCB contaminated matrices
16


Table 2.1 SPE Disk Extraction Results
SAMPLE SAMPLE ug AR 1254 SPIKE TO ELUTION /RECOVERY
# VOL (ML) SPIKED EXT.TIME TIME (MIN) 3BrBP DCB PCB
6 1000 0.5 0 DAYS 60 46 53 81
7 1000 0.5 0 DAYS 90 25 24 12 *
8 1000 0.5 0 DAYS 45 49 46 53
9 1000 0.5 2 DAYS 25 - 63 79
10 1000 0.5 2 DAYS 35 - 74 93
11 1000 0.5 2 DAYS 60 - 64 -
12 1000 0.5 2 DAYS 55 - 52 66
13 1000 100 2 DAYS 90 DIL OUT 81
14 1000 100 3 DAYS 35 DIL OUT 79
15 1000 100 3 DAYS 40 DIL OUT 81
16 1000 100 3 DAYS 65 DIL OUT 96
17 1000 100 3 DAYS 90 DIL OUT 44
18 1000 100 3 DAYS 30 DIL OUT 85
19 1000 100 3 DAYS 120 DIL OUT 59
20 1000 0 2 DAYS 25 132 59 -
21 1000 0 2 DAYS 30 126 59 -
22 925 10.2ug/L 1242 2 DAYS 55 DIL OUT 235
23 920 8.4ug/L 1242 2 DAYS 70 DIL OUT 220
24 980 14.6ug/L 1242 2 DAYS 85 DIL OUT 125
17


Table 2. 1 SPE Disk Extraction Results (continued)
SAMPLE SAMPLE ug AR 1254 SPIKE TO ELUTION /RECOVERY
# VOL (ML) SPIKED EXT.TIME TIME (MIN) 3BrBP DCB PCB
25 910 7.9ug/L 1242 2 DAYS 50 DIL OUT 359
26 1000 80 2 DAYS 45 DIL OUT 107
27 1000 80 2 DAYS 30 DIL OUT 79
28 1000 50 2 DAYS 140 DIL OUT 37
29 1000 50 2 DAYS 45 DIL OUT 56
30 1000 20 2 DAYS 50 DIL OUT ND
31 1000 20 2 DAYS 30 DIL OUT ND
Preperation comments: sample #7 disks went partially dry midway
through sample elution^
Sample #20 was a blank extracted after sample #12.
Sample #21 was a blank extracted after sample #19.
Samples #9-12 displayed unknown contaminants that obscured the
3BrBP peak.
Samples #6-31 took a total 32 hours to prepare; 26.5 hours for
sample elution; 5.5 hours for disk elution,set-up,teardown.
Estimate an analyst can operate four extraction stations at a
time (glassware @ $500/each).
The total mass of the solids in the groundwater samples residing
on the filtered were not determined.
18


displayed 125 to 359 percent PCB recovery as compared with the
shakeout results indicating SW-846 methods may exhibit incomplete
extraction of PCBs from groundwater samples due to analyte
adsorption to the fines although the actual amount of PCBs
naturally incorporated in the matrices is not known.
Due to the prohibitive extended elution times of most of the
groundwaters and unexplained losses of surrogate and spike
compunds, SPE disk development studies were terminated. Future
SPE experiments must seperate the aqueous and solids fractions
without clogging the phase for the extraction of each fraction. A
simulated groundwater must also be prepared that provides
conditions to naturally incorporates the analyte into the matrix
to provide percent recovery determinations.
2.1.2 Preparation of a Simulated PCB Groundwater Matrix
An aqueous matrix that physically resembled the groundwater
matrices observed at our laboratory was prepared with a known
amount of PCBs naturally incorporated into the matrix to perform
percent recovery determinations. Simulated sedimentary PCB waters
were prepared by grinding opaline silica in a mortar and pestle
and then passing the powder through a 120 mesh sieve. Unspiked
simulated groundwater matrices were prepared by adding 0 to 5
grams of the fine powder to 960ml of carbon filtered (CF) water
19


in a 11 sample bottle, shaken for two minutes to simulate
sampling/ transport conditions, and allowed to settle overnight
to allow settling of non-suspended particles. The matrix visually
resembled a typical groundwater. This unspiked matrix was
employed to determine different SPE systems tendencies for
clogging without exposing the chemist to PCBs. For analyte
recovery studies, 5ug of AR 1254 were spiked into the simulated
matrix after the addition of the 960ml of CF water, shaken for
two minutes to assimulate sampling/ transport conditions, and
allowed to settle overnight to allow settling of non-suspended
particles. This simulated matrix was used in all of the following
experiments and further preparation discussion in section 4.1.1
2.1.3 SPE with Inverted Filter and Encapsulated Cartridge
To reduce particulate clogging of the phase due to gravity
deposition of the solids laden matrix on the phase, an inverted
filtration/SPE delivery system was designed (figure 2.1) to allow
the solids to remain settled on the bottom of the sample bottle
while the aqueous phase was filtered through a 6cc polypropylene
syringe barrel containing a lOum polypropylene frit at the
effluent end and a glass wool plug packed in the influent end of
the barrel (a rough filter). 20 PSI nitrogen pressure drove the
system through 1/4" (3mm) O.D. Teflon lines. The sample bottle
20


Teflon Line;
Figure 2.1 5PE with Inverted Filter and Encapsulated Cartridge
21


was sealed with a custom Teflon cap that accepted the influent
(N2 and solvent) and effluent (SPE cartridge) Teflon lines
through the top. The SPE phase was an Alltech Associates, Inc
(Deerfield, II) 900mg C18 Hema inline cartridge equipped with a
female leur inlet and a male leur outlet encapsulated 2cm of
layered polypropylene frits surrounding the 900mg of C18. The
system was designed to push the matrix through the rough filter
and through the preconditioned C18 cartridge with the effluent
discarded. Post elution was achieved by introducing 10 ml hexane
into the sample bottle with subsequent 20 PSI N2 elution of the
cartridge with the bottles solvent into a 12ml glass vial. The
final extract was adjusted to 10.0ml hexane.
The preliminary extractions exhibited varying elution times
of the nonparticulate water matrices of 0.5 to 6 hours. It was
determined that the cartridges possessed inconsistent packing
densities that dictated the varying flow rates. The inconsistent
flow problem was discussed by phone with Alltechs technical
services and a new type of custom, cartridge was ordered which
replaced using the automated machine to tightly pack the
cartridges to manually loose packing the empty cartridges with
the C18 phase using 80% less frit material. The dozen new
cartridges displayed consistent 3 hour elution times at 10 PSI N2
on nonparticulate one liter waters with acceptable PCB spike
22


recoveries (>80%). One hundred of the custom Alltech C18
cartridges were then ordered for more rigorous validation
studies. A test of twenty of the new lot of cartridges displayed
0.5 to 8 hour elution times of nonparticulate one liter waters,
after which the manufacturer indicated that a fast, consistent
flow inline C18 cartridge cannot be made to our specifications
and the closed cartridge system was not developed further. The
concept of the closed extraction system and the rough filter was
appealing and will be explored in the following experiment.
2.1.4 SPE with Inverted filter and Standard Cartridge
The SPE extraction system now consisted of an inverted glass
wool cartridge rough filter in the bottom of the sample bottle
with an adjustable pressure N2 inlet and Teflon tubing with press
leur fitting apparatus for influent/ effluent mass transport. The
next SPE system investigated to provide a consistent matrix flow
rate was a manufactured 1 gram C18 SPE phase in a 6cc cartridge
with 20um polypropylene frits sandwiching the phase. The support
beads diameter was 50um with a manufacturers reported small
particle size distribution. The cartridge was placed downline
from the rough filter/sample bottle in the traditional upright
position to prevent the phase from going dry. The system provided
acceptable and consistent fast flow rates at 10 PSI N2 for
23


nonparticulate waters- The 5 gram fines/liter loading aqueous
matrix clogged the system after 100ml of the sample was eluted. A
similar design as displayed in figure 2.2 consists of packing the
remaining volume of the C18 cartridge with glass wool and
inverting the cartridge at the bottom of the sample bottle also
clogged after 100ml of the 5g/l waters were eluted.
The inverted C18 cartridge in the sample bottle design was
then modified by replacing the glass wool with a 18 hour MeCl2
soxhlet pre-extracted plug of polyurethane foam (PUF). The
cartridge was designed to filter out particles >20um (frits pore
size) by using a polyurethane foam plug used as a rough filter
followed by the top 20um frit of the SPE cartridge. The
compressed polyurethane foam plug was chosen due not only for its
filtering properties, but also for its reported analyte
adsorption properties (8) The PUF design did not clog for the
5g/l fine loading but experienced a 5 hour elution time at 20 PSI
N2.
The 1 gram C18 cartridges from either the PUF and glass wool
rough filter experiments were then dissected to discover that the
small fines passed through the 20um frit and deposited in the top
1mm of the tightly packed, uniform particle size C18 support
bead. Ten cartridges were then disassembled and loosely repacked
with a PUF plug and resulted in subsequent clogging in the C18
24


Teflon Lines
Cartridge
Figure 2.2 SPE iraith inverted Cartridge and Standard Cartridge
25


phase for the 5g/l particulate loadings. Dissection of the
cartridges exhibited clogging at the influent frit/C18 interface
as displayed in Figure 2.3. The PUF plug appeared to outperform
the glasswool plug as a rough filter and was used in all of the
following experiments. A custom cartridge designed to prevent
clogging at the interface was developed and is discussed in the
next experiment.
2.1.5 SPE Custom Cartridges
To prevent patriculate clogging at the SPEs frit/ phase
interface, the preparation of a large particle size distribution
SPE phase was attempted to permit the fines < 20 urn that pass
through the frit to also pass through the SPE phase. Ten of the
manufactured 1 gram C18, 6cc cartridges were disassembled and the
packing placed in a small aluminum weigh boat. 0.5 grams of 60-
100 mesh florisil was weighed and mixed with the one gram C18 and
loosely repacked in the 6cc barrel sandwiched between two 20um
frits to produce a large particle size distribution. A PUF plug
was then inserted in the cartridges remaining influent void. The
mixed mode SPE design also clogged at high loading levels so a
0.75 gram florisil/ 1 gram C18 loose pack cartridge with the PUF
plug was employed with no clogging and acceptable flow rates for
the 5g fines/1 loading. The spent cartridges were dissected and
26


aMS^irii^4^:4<'W#!
g^p^lf:
E^SSSslS^^
ESS!52S^^
Clogging at Interface ^

mmm*

lpt|
S*S

:j#j#!£ijS

L^;iL.L|g
ii|§|i
HH
PUF Plug
20 um Frit
1 g C18
20 um Frit
Female Luertip
Figure 2.3 Clogging of a Stadard SPE Cartridge
27


it was discovered that the minute amount of particulates that
passed through the PUF plug and the 20um frit also passed through
the mixed mode phase to the elluent container. The mixed mode SPE
phase with the PUF plug rough filter was deemed to satisfactorily
seperate the aqueous and solids fractions for subsequent SPE and
fines extractions and was implimented in all further experiments
and its preparation futher detailed in section 4.3.1.2. Water in
the SPE elution solvent indicated that channeling was occuring in
the custom SPE cartridges and will be addressed in the next
experiment.
2.1.6 SPE Elution Solvent Studies
Channeling of an SPE is defined as a fluid (water, gas, or
solvent) that passes through a selected portion of the phase
while the remaining mass of the phase will not have a fluid flow.
Channeling occurs if a phase exhibits a nonuniform packing
density, or if a water immiscible solvent or gas is passed
through a hydrated phase, the solvent or gas will follow the
initial fluids path and not displace the residual water.
Channeling of the N2 through the phase during the post
depositions drying step produced excess water in the hexane
extracts. Partition studies for the fractionation of hydrocarbons
from PAHs in oil matrices involved binary solvent mixtures, a
28


water soluble and water insoluble solvent mix, and the addition
of a small amount of water to facilitate the fractionation of the
solvents. Subsequent PCB spiking studies into 20ml of 1:1
hexane/acetone with the subsequent addition of 2-5ml of water
indicated 100% recovery of the PCBs in the hexane fraction.
Therefore, the 20ml of the 1:1 hexane/acetone mixture was chosen
to wash the sample bottles wall and to elute the cartridge. The
water soluble acetone enabled the water to be driven off the
phase and eliminated channeling while the residual water
collected with the elution solvent fully separated the binary
solvent into a water/acetone fraction and a hexane fraction
containing the analyte as confirmed with binary solvent spiking/
seperation studies. The binary elution solvent will be utilized
in section 4.3.1.3
2.1.7 Close Extraction System for PCB Groundwaters
Extraction of the solids fraction will be addressed in this
experimental section. The SPE disk extraction results displayed
in Table 2.1 exhibited low extraction recoveries of the
sedimentary waters possibly due to incomplete solvent extraction
of the solids fraction. EPA Method 8080 (22) is used for the
analysis of PCB extracts and allows a minimum 31% QC recovery of
AR1254 matrix spike and spike duplicate samples. The lower
29


accepted spike recoveries for the "real aqueous matrices is
suspected to be attributed to the analyte adsorption to the fines
(and is referred to as the matrix effect). Treating the residual
fines remaining after the SPE deposition step as a soil matrix
would require a more rigorous solvent extraction procedure such
as horn sonication as described in EPA method 3550. A custom 900
watt sonication bath was purchased to actively desorb the analyte
from the solid fraction into the SPE elution solvent. The custom
SPE filter/ cartridges connected to the caps were inserted in the
sample bottles are placed in the sonication bath before the
extraction is started as shown in Figure 2.4. After the aqueous
phase passes through the SPE phase, the binary elution solvent as
described earlier in Section 2.1.5 is introduced via the
manifold/Teflon line into the sample bottle and sonicated for two
minutes. The extract is then passed through the SPE cartridge
using <20PSI nitrogen pressure into the 40ml VOA vial for the
water facilitated binary solvent fractionation.
The individual problems associated with groundwater
extractions via SPE are addressed and the individual components
of the closed SPE system (custom SPE cartridges for the
seperation of the aqueous and solids fraction and subsequent
extraction of the aqueous phase, a binary extraction solvent
utilized on both fractions, and sonication extraction of the
30


Teflon Lines
Figure 2.4 Closed Extraction System of PCB Groundwaters
31


solids fraction) will be assembled and evaluated by comparing PCB
recovery on a simulated PCB matrix against the EPA approved
methods as described in Closed Extraction System Setup and
Operation Section 4. Since the fractions may be seperated and
extracted separately, equilibration time for PCBs to be
incorporated into the matrixs fractions as as well as the
partition coefficient as related to solids loading will be
determined in section 4.3.1.3.
2.2 Soil Extraction Method Development
2.2.1 Soil Dispersion Studies
As discussed in the the Soil Extraction Background Section
1.2, the hydrophobic organic analyte can be thoroughly dispered
in a soil matrix and conventional manual dispersion of the matrix
for subsequent sonication extraction displayed incomplete
dispersion. A alternative sample dispersion method must be
developed to completely expose the matrix to the extraction
solvent.
A method validation for the determination of nitroglycerin
and PETN in soils was investigated to study the different matrix
dispersion methods. A literature search of explosives extractions
revealed SW-846 Method 8330 (23) for the determination of
nitroaromatics and nitramines (explosives) in soils. Method 8330
32


consists of drying approximately 30 grams soil overnight at room
temperature, mortar and pestle the dried matrix and passing the
fines through a 30 mesh sieve. 2 grams of the sieved soil is
sonicated for 18 hours in a 30 watt ultrasonic bath in 10ml
acetonitrile. The slurry is finally passed through a 0.2um Teflon
filter before analysis. The basic principles of EPA method 8330
involving a dried, finely divided matrix was advantageous to
validate an accurate method and was explored for application in
PETN/ nitroglycerin soil extractions.
Mortar and pestling soils containing potentially high levels
of nitroglycerin would fully disperse the matrix but deemed
unsafe. Due to safety factors and the labor intensive mortar and
pestle step, a mechanical dispersion step was necessary. The
milling and ball mill systems were too expensive, could only
accommodate one sample at a time, and required rigorous cleaning
between samples rendering the systems unacceptable for production
work. The ball mill concept was appealing, so an inexpensive,
high capacity, low maintenance system was needed. After
investigation, a Red Devil model 5400 paint shaker was chosen for
its potential capacity and three dimensional elliptical shaking
ability. A 24 sample container adapter was fashioned with guide
rods and can slots that provided quick and easy loadup and tear
down as displayed in Figure 2.5. The paint shaker dispersion
33


Clamp Dorn
Figure 2.5 Custom Mechanical Matrix Dispersion Apparatus
34


method will be further be explained during setup and operation in
section 4.3.2.1.
The sample container needed to be inexpensive, clean, and
disposable to limit cross contamination, cleanup time, and cost.
Different size paint cans were investigated (1/4 to one pint
volume) and the 1/2 pint can was chosen for study. The cans were
initially extracted with MeCl2 and subsequent analysis displayed
high hydrocarbon can contaminations. An optimum 2 hour baking
time at 280C of four hundred cans and tops produced a clean can
for GC-MS/ECD/FID analysis without melting the solder in the can.
Initial dispersion experiments involved placing randomly chosen
30g samples and 60g Na2S0^ into the cans with varied size
[0.25"(7mm)-0.75"(20mm)] steel ball bearings and dispersing for
one to five minute intervals. The resulting mixture would be a
muddy clump for the wetter matrices. An optimum method for all
matrices was devised by excluding the ball bearings for the first
minute of the dispersion, and adding two 1/2" (13mm) ball
bearings for the final one minute dispersion.
The matrix seemed warm to the touch after the mechanical
dispersion, so the ball bearings were immersed in liquid nitrogen
prior to dispersion. The cold dispersion appeared to freeze the
water contained in the matrices and facilitated dispersion due to
fracturing the soil. Studies indicated that spiked semivolatile
35


losses were the same for the warm and cold ballbearing dispersion
(40-60%), so the warm ball bearings were used in the following
experiments. Hundreds of soil samples were mechanically dispersed
using this method for subsequent standard horn sonication
extraction methods for the determination of nitroglycerin and
PETN with acceptable dispersion of multiple matrix
characteristics. The quality control criteria for the surrogates
(erythritol tetranitrate) recovery of 31-134% and matrix spike/
matrix spike duplicate recoveries of 48-110% for nitroglycerin
and 48-108% for PETN were almost always achieved for the
explosives preparations utilizing the mechanical dispersion
technique utilizing traditional spiking techniques before
dispersion. The mechanical dispersion steps acceptable results
on the explosive soils indicated that it could be utilized in the
closed extraction systems methodologies section 4.3.2.1.
2.2.2 Soil Matrix Sonication Studies
The organic analyte hydrophically bonded to a soil matrix
must be rigorously extracted to overcome the high sorption
energy. Conventional horn sonication of the matrix provides the
energy needed for organic extraction but is time and equipment
consuming and has the potential to introduce cross contamination
36


of extracts. A fast and effective sonication system needed to be
developed.
Initial soil extraction studies consisted of replacing the
conventional 350 watt probe horn sonicator with a 250cc capacity
350 watt cup horn sonicator for micro PCB/soil extractions. The
PCB soil method was validated for small sample amounts (2g soil
extractions for PCBs). The matrix dispersion technique developed
in the last section resulted in the ready to extract matrix
contained in a clean steel can. The matrix dispersion container
could be immersed into a bath sonicator to prevent sample
transfer and to reduce glassware usage as compared to horn
sonication. To achieve the desired reporting limits for most
organic analytes in soil matrices, a 30 gram sample must be
extracted, and therefore, a more powerful ultrasonic bath system
is required.
After vendor literature investigation, a custom
900watt/40Khz Ney ultrasonic bath was obtained. The pulse time
was set to 0.2 seconds and a timer was also installed for
accurate extraction times. The bath may accomadate over 25
samples at once and is operated with an inch of water to
facilitate power transfer from the 15 transducers to the samples
containers and deemed satisfactory to explore for the soil
37


sonication extraction step and is further discussed in section
4.3.2.3.
The methylene chloride extracts produced from the
mechanically dispersed matrices via the nitroglycerin/ PETN
method using horn sonication or initial bath sonication were
dried using a glass funnel containing a filter paper and sixty
grams of sodium sulfate. Studies of grab sample extracts
displayed severe clogging of the sodium sulfate funnels at the
filter paper which must be overcome. The labor intensive steps of
manually transferring solvent to each matrix before sonication
and extract transfer after sonication must be eliminated to
produce a faster extraction method.
2.2.3 Closed Extraction System for PAH Soils
Since the mechanically dispersed fines usually clogged the
standard Na2S04/ filter paper drying/ filtration step performed
before concentration, the lessons learned during SPE development
of using an inverted filter would be necessary to implement to
filter the post sonicated soil extract. Using the cans physical
dimensions, a custom machined can lid was ordered (Battery Power
Unit, Golden, CO) that produced an airtight seal on the can
utilizing minimal manual application (press-on). To facilitate a
semi automated solvent/ extract transfer, the custom lid was
38


tapped out to install three airtight stainless steel fittings
that accepted an easily replaceable (finger tight) 1/4" O.D
teflon tubing through a brass ferrule. One of the ports accepted
N2/ solvent influent, the second port was a air vent cap for
solvent load, and the last ports tubing accepted the inverted
filter. The inverted filter was made by cutting an empty standard
6cc SPE cartridge 1" below the effluents lOum frit and packing
it manually with a clean glass wool plug. The connections on the
manifold, solvent reservoir, custom cap lids 1/4" Teflon tubing,
inline inverted filter, and other periphery modules were
performed using press-on male and female leur fittings with press
on 1/8" O.D. Teflon tubing (the pressurized leur fittings/tubing
never disconnected during all experiments). Initial experiments
displayed consistant solvent flow rates through the glass wool
filter using coarse sand and finely dispersed matrices sonicated
under a low 5 PSI N2 pressure differential and deemed sufficent
to incorporate into the soil extraction method as described in
section 4.3.2.2.
The closed extraction system presently consists of
mechanically dispersing the sample in a clean steel can. The
steel cans lid is removed and the custom lid is pressed on that
consists of manifold solvent introduction, solvent bath
sonication of the matrix, and subsequent filtration of the
39


extract via N2 pressure through an inverted glass fiber rough
filter into a receiving beaker for subsequent manual transfer to
a Kuderna-Danish concentrator. The closed system must incorporate
a coupled semi automated concentration module to complete the
apparatus.
2.2.4 Inline Extract Concentration Module
To decrease the chemists sample preparation time, an inline
concentration module must be developed. The conventional glass K-
D apparatus consists of a 350ml solvent reservoir with two ground
glass joints to accept a three ball Snyder column on top and a 15
ml concentration tube on the bottom. The K-D apparatus is driven
by immersing the concentration tube in a hot water bath (90OC).
To produce an inline closed system connected with the extraction
can effluent tubing, a custom glass sleeve with two ground glass
joints was ordered to fit between the K-Ds solvent reservoir and
the concentration tube. The custom sleeve also incorporated a
glass tube that accepted a stainless steel fitting with a Teflon
ferrule that accommodated the 1/8" Teflon tubing from the
extraction cans effluent port. The Snyder column had a ground
glass joint on the top that accepted a custom ordered "U" tube
that accepted a 60cm length corrugated FEP tube. A circulating
cold water Allihn condensor with a custom manifold accepted up to
40


six of the effluent end FEP tubes. A 2 liter beaker was placed
under the condensor to accept the spent solvent for recycling.
Initial extractions using the complete extraction apparatus
developed thus far did not operate to expectations due to
extracts being concentrated to dryness before the extraction was
complete which produced losses of the analyte. It was observed
that the constant solvent flow system could not produce enough
solvent volume delivered to the K-D over the extraction period to
overcome the K-Ds concentration rate. A pulsed solvent system
was employed to extract the sonicating soil utilizing a three
MeCl2 aliquot pulses (100ml,50ml,50ml respectively). After the
solvent aliquots are collected in the K-D, the K-D is immersed in
the hot water bath for concentration. The completed closed
extraction system is deemed ready for study in section 4.3.2. As
discussed in Soil Extraction Background Section 1.2, a wet clay
homogeneously spiked with PAHs must be created to test the novel
extraction methods performance against EPA methodologies via
percent recovery determinations.
2.2.5 Preparation of a Simulated Homogeneously Spiked Wet Clay
Matrix
Dispersion experiments were conducted on five previously
analyzed semivolatile organic soils via traditional manual
41


dispersion/ horn sonication techniques. The samples were re-
extracted using the mechanical dispersion technique with
subsequent horn sonication and analyzed via GC/FID (a screening
instrument). Comparison of the screening chromatograms indicated
that the samples that were mechanically dispersed exhibited 3X-5X
higher semivolatile organic compound concentrations than the same
matrices dispersed via manual dispersion. The results suggest
organic analytes naturally incorporated in a soil extract at
varying efficiencies depending on the dispersion method employed
and that the total analyte content may not be known due to
insufficient extraction methods.
To evaluate the effectiveness of a soil extraction method,
PAH percent recovery studies from a "real" wet clay matrix must
be performed. The conventional PAH spiked dry clay certified
matrix must be reproduced and reconsistuted to produce the
consistency of a wet clay. Previous soil dispersion studies on
the explosive matrices displayed complete homogenization of
spiked analyte in dry solids. The analytical screening of seven
subsamples of nitroglycerin spiked and dispersed 50 gram matrix
displayed excellent recoveries and a very low standard deviation
between the subsamples. A 500 gram sample of a wet clay was
obtained locally and the percent moisture was determined
gravimetrically. A 105C oven dried the matrix overnight which
42


was then sieved < 710 um. Crushed ice was purchased and an amount
weighed to return the dried matrix to its original % moisture.
The sieved matrix and crushed ice (~ 4mm diameter) was placed in
a one gallon paint can and dispersed on the paint shaker for ten
minutes. The resulting matrix was a heterogeneous mixture of
clumps of mud/ice and regions of dry matrix. A smaller diameter
ice structure would be required to disperse more evenly with the
sieved clay. A cooler of Colorado "dry powder" snow was obtained
and kept frozen until use with dry ice. A test hydration of the
dried seived clay with snow was attempted in a large paint can
cooled with dry ice, and after dispersion, the homogeneous mix of
dry soil and snow mixture was brought to room temperature to
facilitate the phase change of ice to water. Compacting the
resulting matrix to its original density would also simulate a
"real matrix" which was performed in Closed Extraction System
Setup and Operation Section 4.1.2. Due to time constraints,
further initial study of this procedure was terminated and the
preparation of a homogeneously PAH spiked wet clay was performed
during method validation and is further discussed in section
4.1.2.
43


3. EVALUATION OF THE CLOSED EXTRACTION SYSTEM
The purpose of this chapter was to apply knowledge gained
during the SPE and soil extraction method develpment phase
explored in section 2 and build a closed extraction system and
test the equipment on prepared simulated PCB groundwater samples
and compare its extraction efficiencies to EPA Method 3510 and
EPA Method 3520 accepted methodologies (llLE shakeout and
continuous 1iquid/1iquid extraction respectively) via percent
recovery determinations. The prepared 5 PPB PCB groundwaters will
provide natural interaction (traditional spiking studies were
proved unrealistic for "real" groundwater method validation from
Section 2.2.ls results) of the analyte and matrix containing
varying amounts of ~ 10 urn opaline silica (0, 0.5, 2.5, and 5.0
grams/liter) to determine the sediment loading effects on PCB
extraction efficiencies. Duplicate spiked PCB/groundwater
matrices with varying amounts of solids loading will also be
prepared and extracted via in situ SPE from both the solid and
aqueous fractions in the simulated groundwater matrices to
determine the aqueous/solid equilibrium time and PCB partition
ratios. Unspiked groundwaters will also be prepared to determine
system/matrix contamination.
44


The soil extractions method development techniques and
equipment will also be utilized to extract 16 different PAHs from
a prepared simulated homogeneously spiked wet clay matrices and
compare its extraction efficiencies to the conventional
sonication extraction method through percent recovery
determinations. The percent recovery determinations standard
deviations will determine if a homogeneously spiked wet clay was
indeed achieved as well as the reproducibility of both extraction
methods. Unspiked dry and wet clay matrices will also be prepared
to determine system/matrix contaminations. The prepared unspiked
dry and wet matrices will be spiked with PAHs utilizing
traditional spiking techniques to determine if their recoveries
are comparable to the simulated PAH contaminated wet clays and
the traditional spiking techniques validity in determining method
performance.
45


4. CLOSED EXTRACTION SYSTEM SETUP AND OPERATION
4.1 Matrix Preparation
4.1.1 PCB/Groundwater Matrix Preparation
As determined in section 2.1.2, a simulated PCB groundwater
must be prepared for extraction method validation. Approximately
150 grams of opaline silica (Excel Mineral Co.,Goleta,CA) was
ground and sieved through a 120 mesh screen and stored in a
covered glass jar. A sample of the ground matrix was sent to
Jones Chromatography (Lakewood,CO) for a particle size
distribution determination. The opaline silica particle size
range was between 2um and 22um with a mean of lOum. One liter
clear glass bottles were utilized for the shakeout method while
one liter amber glass bottles (I-Chem, New Castle, DE) were used
for the solid phase extractions. 960ml of carbon filtered water
(CFH2O) were poured into the sample bottles and then varying
amounts (0,0.5,2.5,and 5.0 grams) of the sieved clay were added.
The matrix was then spiked with 1.0ml of 5ug/ml AR1254 in
acetone. The samples were then shaken ten times to simulate
sampling and transport conditions. The samples were then placed
in boxes, closed, and then stored in a 4C refrigerated room
until the time of extraction. Blank samples of 0 and 5 gram
46


solids/960ml without the AR1254 spike were also prepared for each
test (LLE,SPE,and SPE duplicate). Note: SPE samples needed to
settle for a minimum 16 hours for the 2.5 and 5 gram matrices or
the filter system clogged.
4.1.2 PAH Soil Matrix Preparation
As determined in section 2.2.5, a homogeneously spkied wet
clay must be prepared to evaluate an extraction methods
performance. The matrix chosen for this study was obtained from a
construction site in Arvada, Colorado. The matrixs percent
moisture was determined gravimetrically by baking five
representative 50 gram subsamples overnight at 110C. The
matrixs density was determined empirically by immersing four
representative 150 gram subsamples in a one liter graduated
cylinder containing 500ml water (the raw matrix was a tightly
packed, fine particle size clay with no visible air pockets).
Eight Titers of the bulk matrix was manually sectioned and baked
to dryness for 24 hours at 110C. The dried matrix was then
mortar/pestled and sieved through a 710 micron mesh screen. A
portion of the dried and sieved matrix was removed and labeled
"DRY SPIKE" for subsequent traditional spiking studies. Another
portion (492g) of the dried and sieved matrix was placed in a
clean one gallon paint can and spiked by spraying the matrix with
47


3 X lOOOul aliquots of a verified 200ug/ml PAH mixture (Ultra
Scientific, North Kingstown, RI) in MeCl2 with intermittent
manual shaking and let stand at room temperature for 90 minutes
to evaporate the solvent. The can was then sealed and placed on a
Red Devil Equipment (Minneapolis,MN) model 5400 paint shaker for
15 minutes.
Reconstitution of the matrix was achieved by cooling the
dried matrix with 500g dry ice wrapped in aluminum foil. Snow
(108g) was cooled in a 600ml beaker immersed in a 1000ml beaker
containing an acetone/dry ice islurry. The chilled snow was
sprinkled on the cooled matrix and sealed with the paint can lid.
Snow was utilized as the reconstituting agent as it behaves as a
dry dispersible powder that doesnt exhibit the aforementioned
reconstitution problems with a liquid. The paint can was then
dispersed on the paint shaker for 10 minutes. Twelve 1/2" (13mm)
clean steel ball bearings were added to the paint can and
dispersed for an additional 20 minutes. The reconstituted matrix
was transferred to a clean quart paint can with a 300cc volume
mark indicating the desired final density of 2g/cc. A clean 1"
(26mm) diameter punch and hammer was used to compress the matrix
to the final volume. The bottom of the one quart paint can was
opened with a can opener and the matrix was cut into quarters
with a clean knife for easy removal. The matrix was transferred
48


to a one liter wide mouth glass jars, labeled "1 PPM PAH NATIVE
MATRIX", and stored in a -lO^C freezer until sample preparation
was performed.
An additional 600 grams of matrix with 2g/cc density and an
18% moisture content was also prepared utilizing the above
procedure but without the PAH spiking step. This matrix was
labeled "WET SPIKE" for subsequent traditional spiking studies.
4.2 EPA Standard Extractions
4.2.1 PCB/Water Preparation Via EPA Method 3510
To evaluate the closed extraction systems performance
against standard methods for PCB groundwaters, EPA Method 3510s
(1) analyte recoveries on the simulated matrix will be evaluated
against the novel methods recoveries.
EPA Method 3510: All glassware is prerinsed with methylene
chloride (MeCl2)* A 2 liter separatory funnel with a Teflon
stopcock and stopper was used to pour the entire LLE matrix into.
The matrix was then spiked with 1.0ml of 0.05ug/ml
decachlorobiphenyl (DCB) in acetone. 60ml of MeCl2 was poured
into the original sample bottle, capped, shaken 10 times, and
this fraction poured into the separatory funnel. The separatory
funnel was shaken ten times and vented. The separatory funnel was
then placed on an automatic shaker (Glas-Col, Terre Haute, IN) at
49


speed 4 (to mimic manual agitation) and shaken for two minutes.
The separatory funnel was then allowed to settle out the emulsion
and then the stopcock was opened to allow the bottom fraction
(MeCl2) to collect in a 250ml glass beaker. This solvent
extraction step was performed two more times with the exclusion
of the sample bottle rinse. Of the 180ml MeCl2 used, MeCl2
recoveries from 140ml to 160ml were recorded.
The MeCl2 extracts were then concentrated by preparing a
300ml Kuderna-Danish (K-D) evaporator with 12ml receiver, 3 ball
Snyder column, and three Teflon boiling stones. The extracts were
poured into the K-D through a MeCl2 prerinsed 20 gram Na2S04
drying funnel and then placed on a boiling water bath until the
MeCL2 volume approached 8ml. 50ml hexane was then added to the K-
D for a solvent exchange to hexane. The sample was volume was
reduced to about 6ml and removed from the water bath to cool. The
sample was then adjusted to 10.0ml hexane and transferred to a
40ml VOA vial and shaken for two minutes with 20ml concentrated
sulfuric acid as a cleanup step. The hexane fraction was then
pipetted to a 12ml screw top vial and placed in a refrigerator at
4C until analysis.
4.2.2 PAH/Soil Preparation Via EPA Method 3550
50


To evaluate the closed extraction systems performance
against standard methods on PAH soils, EPA Method 3550s results
on the analyte recoveries on the simulated matrix will be
evaluated against the novel methods recoveries.
EPA Method 3550: PAH soil preparation (20) consists of
weighing 30 grams of the soil matrix into a clean 400 ml glass
beaker and spiking the appropriate samples (wet and dry spikes)
by spraying the matrices with 150ul of the 200ug/ml PAH mixture
and allowing the spike to be in contact with the matrix at room
temperature for 90 minutes to evaporate the solvent before
proceeding to the next step. No spike was added to the simulated
matrix. 60 grams Na2S04 (sodium sulfate) was then added to the
matrix and the contents were then dispersed manually with a clean
stainless steel spatula to achieve a dry, free flowing powder.
100ml MeCl2 was aliqouted into the beaker and horn sonicated
for three minutes at full power, 50% duty cycle, and one second
pulse modes. The liquor is then poured through a glass funnel
containing a glass wool plug and 30 grams of MeCl2 prerinsed
Na2S04, and into a clean 400ml glass beaker. The sonication step
was repeated twice and the extract collected for concentration.
Concentration of the extract involved pouring the extract
into a 500ml Kuderna-Danish (K-D) concentrator equipped with a
10ml receiver tube and a 3 ball Snyder column. The K-D was then
51


immersed in an 85C water bath and allowed to concentrate to
~5ml. The 5ml extract was transferred to a 10ml calibrated
concentrating tube and placed on a 30C nitrogen evaporator (N-
evap) and evaporated to 1.0ml MeCl2 under a gentle stream of
nitrogen with occasional MeCl2 rinsing of the tubes walls. The
extract was then Pasteur pipetted into a 1.8ml screw top vial
with subsequent marking of the meniscus and storage at -4C.
4.3 Closed Extraction System Setup and Operation
4.3.1 PCB/Water In Situ Preparation
4.3.1.1 Closed Extraction System Assembly and Deration for PCB
Groundwaters
In situ extraction: The novel extraction method utilizes a
tubing system to transfer the matrix from the original sample
bottle, through the SPE cartridge, and to collection bucket
driven by 8PSI nitrogen pressure. The Teflon lines are 1/8" O.D.
and capable of accepting polypropylene Leur fittings. A two hole
Teflon disk with a Viton 0-ring was inserted into the sample
bottles drilled cap that accommodated an airtight seal with the
inserted Teflon tubing and the sample bottle. Male and female
Leur fittings were used to connect tubing to the cartridge and
the manifold with male fittings for effluent flow and female
52


fittings for influent flow. The manifold was an 8 port Teflon
baffle.
4.3.1.2 SPE Cartridge Preparation
The cartridges developed in section 2.1.5 were one gram C18
in a 6cc polypropylene barrel with a 20um polypropylene frit at
both ends (International Sorbent Technology LTD, Mid Glamorgan,
U.K). The cartridges were unpacked into an aluminum weigh boat
and mixed with 0.75 grams of 60-100 mesh florisil (Mallinckrodt
Specialty Chemical Co, Paris, KY) baked at 400C for 18 hours.
The mixed packing was poured back into the cartridge and the top
frit replaced. The cartridge was then cut 2cm above the top frit
and a semicircle of polyurethane foam (1cm heightx6cm diameter)
was rolled into a cylinder and packed tightly on the top The
cartridge was preconditioned on a 12 port vacuum manifold
(American Burdick and Jackson, Muskegan, MI) at 15"Hg with 5ml
hexane to dry, 5ml MeOH to the top frit, and 5ml CFH2O to the top
frit. The cartridge is then inserted in the leur fitting of the
Teflon cap apparatus.
4.3.1.3 PCB/SPE Extraction and Elution
The PCB/SPE extraction procedure developed in sections 2.1.6
and 2.1.7 is performed by opening the sample bottle and the pH
53


was adjusted to 1 with 1:1 H2SO4/H2O. 5ml MeOH was pipetted into
the bottle with subsequent gentle stirring. The Teflon
cap/cartridge system was screwed on tightly with the influent
line plugged into the N2 manifold and the effluent line into a
waste bucket. The manifold was opened at 8PSI and the sample
elutes in less than two hours (including at least a two minute N2
dry of the cartridge after sample elution). To perform the
aqueous and solids fractionation studies, the assembly was
dismantled and the SPE cartridges containing the aqueous fraction
were placed in the vacuum manifold, else the elution was
performed in the sample bottle. 20ml of 1:1 hexane/acetone was
added to the sample bottle which contains the solids fraction and
was capped and shaken twenty times. The sample bottle was then
placed in a custom 900 watt ultrasonic bath (J.M. Ney Co,
Bloomfield, CT) with one inch of water in the bath and sonicated
for five minutes. The slurry was poured into a 40ml VOA vial and
enough CFH2O added to bring the volume to 35ml, capped, shaken
twenty times, and centrifuged for five minutes. The resulting
mixture displayed the 10ml of hexane as the top fraction with the
acetone/water/solids as the bottom fraction. The hexane fraction
was removed and adjusted to 10.0ml hexane and analyzed as the
solid fraction for the SPE duplicate samples.
54


4.3.1.4 PCB Extract Cleanup
To effectively extract the analyte from the SPE phase as
dwetermined in section 2.1.6, 40ml VOA vials were placed in the
vacuum manifold as the SPE elluent receiving tubes and the vacuum
was adjusted to 12 inches Hg vacuum. The cartridge was then
eluted with 0.5ml acetone followed by the 10ml hexane from the
solid fractions VOA vial. The aqueous fraction was extracted in
the same manner using 10ml of fresh hexane. The cartridge was
then post eluted with 5ml acetone. Two ml CFH2O was added to the
VOA vial to facilitate the separation of the hexane from the
acetone/water phase. The hexane was pipetted off into a 12ml
screwtop vial and adjusted to 10.0ml hexane. Subsequent acid wash
of the extract was performed as for the LLE method and the final
extracts stored at 4C.
4.3.2 Closed Extraction System Setup and Operation for PAH/Soi1
4.3.2.1 Matrix Spiking and Dispersion
The novel PAH soil preparation method begins by baking 1/2
liter steel paint cans with press on lids in a 280C oven for 2
hours. After the cans have cooled, 30 grams soil matrix was
weighed into the can and the appropriate samples were spiked (wet
and dry spikes) by spraying the matrices with 150ul of the
200ug/ml PAH mixture and letting the spike incorporate into the
55


matrix at room temperature for 90 minutes before proceeding to
the next step. 60 grams Na2S04 was then added to the can, and the
can was sealed with the baked lid by tapping with a rubber
hammer. The cans were then mechanically dispersed for one minute
on the paint shaker with a custom sample receptacle that provided
the center of the cans to be 14 cm from the center axis and able
to accept 24 samples. The cans were then opened and two 13mm
steel ball bearings were added to the can. The can was then
sealed and dispersed for an additional minute.
4.3.2.2 Closed Extraction System Assembly
To extract the matrix as determined in section 2.2.2-2.2.4,
the dispersed matrix was then readied for extraction by opening
the can and inserting the custom steel lid. The steel lid was
machined to insert snugly into the cans lip via a slight bevel.
The extraction lid also had three threaded holes that accepted
Swagelok stainless steel male connectors with PTFE ferrules. The
finger tightened fitting allowed a 6.3mm O.D. x 4mm I.D. PTFE
tube to be inserted into the extraction can and accepted a press
luer male fitting to fit snugly up to an 80 PSI pressure rating.
1/8" (3mm) PTFE tubing with male, female, and male plug press
luer fittings (Cole-Parmer, Niles,IL) was used to transfer
solvent, nitrogen, and extract through the system. An eight port
56


manifold controlled the sample loop while a 25 PSI nitrogen
regulator controlled the flow rates.
The extraction cans contents are sonicated by placing the
cans into a custom 900 watt ultrasonic bath (J.M. NEY,
Bloomfield,CT) with one inch of water in the bath.
The concentration system consisted of a conventional K-D
evaporation system as used in the above standard method with the
exception of a custom ground glass insert with male and female
joints inserted between the K-D and the receiver tube. This
insert has a hand tightened Swagelok fitting that accepts the
1/8" (3mm) O.D. PTFE tubing from the extraction can. The Snyder
column also has a 24/40 female ground glass joint to insert the
solvent recycling U-tube.
The solvent recycling system consisted of a 24/40 ground
glass joint with a tapered 13mm O.D. U-tube that accepted a 30cm
corrugated FEP tube. The effluent end of the FEP tubing was
inserted into a custom eight port glass manifold with 13mm
tapered glass tubing that accommodates the FEP tubing. The eight
port glass manifold is equipped with a 45/50 ground glass joint
that is inserted into an inverted Allihn condensor operated at
4C. A 41 empty solvent bottle is used for condensate collection.
Care must be taken to insure that the FEP tubing is not crimped
57


and always insuring a downward slope to prevent explosion of the
K-D.
4.3.2.3 PAH/Soil Extraction and Concentration
The extraction is performed by opening the manifold and
delivering 100 ml MeCl2 into the extraction cans at 12PSI. The
contents were sonicated for five minutes and the manifold was
adjusted to deliver the extract to the down line concentrator via
12 PSI N2. The extraction step was repeated two additional times,
but using 50ml MeCl2 aliquots instead of the 100ml as used in the
first sonication. The extracts were then concentrated in the K-Ds
to 5ml MeCl2- The recycled MeCl2 is measured and disposed of at
the end of the preparation. Four sample extractions were
performed simultaneously using this system. Final concentration
of the extract was achieved by N-evap as described in the
standard method. Figure 4.1 displays the closed extraction
system.
4.4 PCB Analysis
The PCB analysis was performed on a Hewlett-Packard model
5890 series II gas chromatograph with an electron capture
detector. The column was a 30 meter Restek RTX-5 with the
following temperature program: 1 minute at 55C, 40C minute
58


FIGURE 4.1
CLOSED EXTRACTION SYSTEM
' SOLVENT LOAD......
SAMPLE EXTRACTION-----
SAMPLE CONCENTRATION-----


temperature ramp to 200QC, a 4C minute temperature ramp to
230C, a 25C minute temperature ramp to 315C, with a final 3
minute hold at 315C. The injector temperature was set at 260C,
while the detector temperature setting was 320C. The AR 1254
working curve ranges from O.Olug/ml to l.Oug/ml while the DCB
curve ranges from 0.00125ug/ml to 0.025ug/ml.
4.5 PAH Analysis
The finished PAH extracts were analyzed on a Finnigan (San
Jose, CA) 4600 gas chromatograph/ mass spectrometer (GC/MS)
utilizing the selective ion monitoring (SIM) mode with a 26
microsecond dwell time. The injection temperature was 275C, a
250C transfer oven, and a 115C detector temperature. The 2ul
injection volume was eluted through the 30 meter, lum film
thickness Restek DB-5 column at a 45C initial temperature for
one minute followed by a 10C/minute temperature ramp for 28
minutes to a final 325C hold for 15 minutes. The initial four
point calibration curve concentrations were 100,500,1000,and
5000ng/ml of the PAH mixture and a lOOOng/ml continuing
calibration midpoint. The initial and continuing calibration
acceptance criteria was 25% RPD. The compounds were quantitated
from one of three internal standards and verified using a
confirmation ion.
60


The "wet spike" matrix was analyzed at 0.286% total organic
carbon at our lab. Professor N.Y. Chang at the UCD civil
engineering department classified the matrix as a silty sand.
61


5. RESULTS AND DISCUSSION
5.1 PCB Equilibration Time Between Solid/Aqueous Matrix
The PCB analyte equilibration time between the water and
solid matrix is determined by analyzing the seperate aqueous and
solids fractions produced in section 4.3.1.3. Comparing the
average of the four SPEs aqueous and solid phases5 PCB percent
recoveries results with relation to analyte contact time as shown
in figures 5.1 and 5.2 respectively. The graphs indicate that the
PCB recoveries over time do not display a trend for both the
aqueous and solids fraction over the 16 to 209 hour time span
indicating the PCB equilibrium between the aqueous and solid
fractions is less than 16 hours. Figures 5.3 and 5.4 exhibit the
average of the four SPE and LLE percent PCB recoveries
respectively versus contact time and supports the observation of
a fast equilibration time. The parallel fluctuations in the data
of figures 5.1-5.4 may be attributed to daily analyst technique
or different batches of sieved clay used on separate holding time
batches since all SPE and LLE extractions for the same contact
time were performed on the same day.
The fast equilibrium time allows the aforementioned PCB
recovery data for a specific particulate loading mass to be
62


Percent PCB Recovery
FlgureT.I: Percent PCB Recovery vs
Contact Time for SPE Aqueous Fraction
O Grams lOum Clay
0.5 Grams lOum Clay
2.5 Grams lOum Clay
5.0 Grams lOumClay


Percent PCB Recovery
Figure 5.2: Percent PCB Recovery VS
Contact Time for SPE Solids Fraction
0 O Grams lOum Clay
--- 0.5 Grams lOum Clay
2.5 Grams lOum Clay
*---- 5.0 Grams I Oum Clay


120
100
80
60
40
20
0
Figure^3: Percent PCB Recovery
Vs Contact Time for SPE Method
13 0 Grams lOum Clay
* 0.5 Grams IOum Clay
* 2.5 Grams 10um Clay
o-- 5.0 Grams 10um Clay
i 1 i i 1 i -----
50 100 150 200 250
Hours of Contact Time


Percent PCB Recovery
Figured: Percent PCB Recovery
Vs Contact Time for LLE Method
O Groms lOum Cloy
0.5 Grams lOum Clay
2.5 Grams lOum Clay
5.0 Grams lOum Cloy


averaged throughout the residence time intervals. This time
averaged data will determine the percent PCB recovery versus
amount of fines in the samples for the SPE and LLE methods and to
profile the extractions effectiveness for the aqueous and solid
fractions of simulated PCB groundwater matrices.
5.2 PCB Solid/Aqueous Extraction Efficiencies
The extraction efficiency of the closed extraction system is
determined by analyzing the seperate aqueous and solids fractions
produced in section 4.3.1.3. Figure 5.5 displays the four data
point time averaged PCB percent recovery versus the mass of fines
for SPE, the SPE aqueous fraction, the SPE solid fraction, and
the summed SPE aqueous and solids fractions respectively. The
data displays a fairly linear loss of analyte with the increasing
mass of fines for the SPE aqueous fraction, while the analyte
percent recovery increases fairly linearly with the increasing
mass of fines in the SPE solid fraction indicating that the
analyte is being adsorbed onto the fines and whose percent of
adsorption is directly related to the amount of fines. For
accuarate PCB determinations in a groundwater matrix, the fines
must be extracted.
Since the bottles interior surfaces were solvent extracted
for both LLE and the SPE solid fraction, it is observed from the
67


Percent PC5 Recovery
Figure £&~: Percent PCB Recovery
for SPE Vs Summed SPE Aqueous
and SPE Solids Fractions
*1 SPE Combined
--- Averaged SPE Aqueous
* Averaged SPE Solids
Averaged SPE combined


0 gram fines in the SPE bottle fraction data that about 12% of
the PCBs are adsorbed to the glass wall of the sample bottle. If
this 12% loss is subtracted from the LLE percent recoveries for 0
to 5 grams of fines per liter data, the LLE adjusted recoverys
results may be directly compared to the SPE aqueous fractions
recoveries as shown in figure 5.6. The 12% analyte adsorption to
the sample bottle stresses the need to solvent rinse the sample
bottles wall after removal of the matrix.
The similiar percent PCB recoveries for the LLE and SPE
aqueous phase versus fines loading as displayed in figure 5.6
provides strong evidence that LLE extracts the PCBs in the water
fraction almost exclusively while analyte sorbed on the fines are
minimally extracted. The observation of low LLE PCB recoveries
for sedimentary samples was supported by producing six 5ug
AR1254, 5 gram fines/1 matrices and allowing a 150 hour residence
time before extracting with the shakeout method and another EPA
LLE method. EPA Method 3520 for the continuous liquid/ liquid
extraction (CLLE) of PCB waters (2) extraction of triplicate
simulated groundwaters produced an average recovery of 39% and
standard deviation of 1% as compared to the average recovery of
41% and 1% standard deviation for the triplicate shakeout
samples. The results suggest that whichever LLE method used,
69


Percent PCB Recovery
Figured: Percent PCB Recovery
for SPE Aqueous Fraction Vs
Corrected LLE Recoveries
Averaged SPE Aqueous
Corrected LLE


similiar PCB recoveries are produced from similiar simulated
matrices.
5.3 PCB Solid/Aqueous Partition Ratio
Having determined that LLE generates minimal recovery of
PCBs from the particulates due to analyte adsorption, the success
of a sedimentary water LLE extraction will be governed by the
partition ratio. The four time averaged SPE aqueous and solid
fractions percent PCB recovery data from figure 5.5 as discussed
in section 5.2 provides the data to determine the partition
coefficients (Kps) of the sedimentary matrices which will
predict the absorptive effects of particulates in the sample.
Because we are interested in the sedimentary effects and not
sample bottle adsorption to determine the partition coefficients,
the aforementioned 12% PCB adsorption to the sample bottle was
subtracted from the SPE solids recovery data. The partition
coefficients (Kp) generated were as follows:
Kp (0.5g fines/960ml)= 624, log Kp(0.5) = 2.80
Kp (2.5g fines/960ml)= 279, log Kp(2.5) = 2.45
Kp (5.0g fines/960ml)= 225, log Kp(5.o) = 2.35
And the averaged partition coefficient (Kap) as:
Kap ( x g fines/960ml)= 376, log Kp(ave.)= 2.58
71


The partition ratio decreases according to higher loading of
fines due to the decreased surface area of the fast settling
fines (the 5g/l loading measured an 8mm fines depth at the bottom
of the sample bottle). The large partition coefficient for the
0.5 gram/liter matrix indicates that even a minute amount of
fines adversely effects the PCB recovery of a LLE extracted
water.
The PCB partition ratio between the water and solid
fractions also predicts the effects of filtering the water sample
before extraction. Filtration is needed to perform a successful
traditional SPE due to particulate clogging of the disc or
cartridge and to prevent excessive generation of emulsions during
a shakeout method. A filtered water sample cannot exceed the PCB
recovery of the SPE aqueous fraction because the fines are not
extracted, therefore rendering a diminished estimation of the PCB
concentration in a raw sedimentary water sample. The PCB log
Kp(ave.) of 2.6 for opaline silica fines/water is over two orders
of magnitude below the PCB log Kdom of ~5 for dissolved organic
matter/water as discussed in the LLE/ SPE backgroung Section 1.1
indicating a lower PCB affinity to silica particles compared to
humic and fulvic acids but was within expected range due to the
opaline silicas low carbon content. While dissolved organic
matter concentrations are in the ng/1 concentration range, the
72


suspended/settleable solids are routinely in g/1 range indicating
settleable solids should be addressed when developing a
groundwater extraction method. The SPE method described in this
paper was developed as a means of extracting PCBs from the
samples settleable solids along with the aqueous fraction to
provide the samples total PCB concentration.
5.4 PCB/Groundwater SPE Vs LLE Extraction Efficiencies
To determine the reproducibility of the SPE phase
fractionation method, the time averaged PCB recoveries for the
SPE aqueous and SPE solid fractions generated in section 5.3 were
summed and compared to the time averaged SPE combined method and
displayed in figure 5.7. The similar results indicate the SPE
methods are acceptable methodologies to measure partition
coefficients and total PCBs in sedimentary water samples. Figure
5.8 exhibits the time averaged PCB recoveries for both the LLE
and SPE methods. Observe the substantial analyte losses for LLE
on clay laden samples, while SPE recoveries are near 100%
indicating a superior PCB in groundwater extraction method. EPA
Method 8080 (22) for the analysis of PCBs in aqueous matrices
publish accepted lower and upper quality control limits (29-131%)
and accepted standard deviations (27.6%) from 50ug/l PCB spiking
studies utilizing four LLE clean water spike recovery results.
73


Figure £1: Percent PCB Recovery VS Amount
of Clay for SPE and Summed SPE Aqueous
and Solids Fractions
0- SPE Combined
---- Averaged SPE combined


Figure 6*8: Percent PCB Recovery Vs
Amount of Clay for SPE and LLE
o SPE Combined
---- LLE Combined


Although both LLE and SPE percent recoveries and standard
deviations are within EPAs acceptance criteria, the SPE method
recoveries display that the method provides superior accuracy and
precision for PCB recovery than LLE. Blank water matrices were
also extracted with the SPE and LLE methods and no contamination
was present in either methods analysis.
5.5 PAH/Soil Standard Extraction Results
The effectiveness of EPA Method 3550 on the extraction of
the simulated PAH soil as described in section 4.2.2. and section
4.1.2 respectively is displayed in Table 5.1. The performance of
the standard extraction method on the simulated homogeneously
spiked wet clay, wet spike, and dry spike matrices through
analyte percent recovery determinations with each component at
lppm indicating a 100% recovery. With exceptions, the wet and dry
spike recoveries were acceptable, while the simulated matrixs
PAH recoveries were unexpectedly low. Although we have determined
the importance of a binary extraction solvent utilized in the
PCB/SPE experiments discussed in this paper, MeCl2 was used as
the extraction solvent in the PAH/soil studies to simulate
Hawthornes (13) PAH spiking and extraction experiments on MeCl2
soxhlet extracted NIST soil. The NIST soil Hawthorne utilized was
a 1% moisture railroad bed soil matrix whose spiking methods and
76


Table?.! Native vs. Wet Spike vs. Dry Spike PAH Recoveries via Manual
Dispersion; Horn Sonication
Analyte Matrix: Native ^Recovery VSD Matrix: Wet Spike %RecovetyXj/SD Matrix: Dry Spike %Recovery Xj/SD
Naphthalene 8.7/1.3 12/0.6 92/6.5
Acenaphthylene 0-5/0.1 41/5.5 0.6/1.0
Acenaphthene 8.6/1.9 42/5.9 23/2.6
Fluarene 16/3.7 58/5.2 87/4.9
Phenanthrene 17/3.6 69/6.1 93/6.7
Anthracene 8.0/1.8 68/5-5 51/2.5
Fluoranthene 14/3.2 70/4.9 85/4.9
Pyrene 10/2.1 67/52 35/32
Benzo(a)Anthracene 13/3.4 86/4.0 94/6.7
Crysene 14/3.7 89/3 J 101/8.5
Benzo(b)Fluoranthene 12/2.9 82/42 95/62
Benzo(k)FIuoramhene 11/3.3 81/4.9 90/6.7
Benzo(a)Pytene 4.6/1.4 67/52 15/1.5
Dibenz(aii)Anthracene 7.9/22 75/5.7 84/4.4
Indeno( l 22-cd)Pyrene 7.8/1.8 76/62 60/1.7
Benzo(g,h.i)Perylene 7.9/2.1 73/62 46/2.0
77


labelled spike recoveries were similar to our dry spike
recoveries via manual dispersion and horn sonication. The high
spike recoveries and low naturally occuring PAH recoveries
observed in this experiment and reported by Hawthornes
supercritical fluid extraction results support Hawthornes
observation that traditional spiking procedures do not accurately
reflect the extraction profile of native analytes and should not
be utilized for method validation experiments.
Higher recoveries for acenaphthylene and benzo(a)pyrene were
noted in the wet spike as compared to the dry spike. This anamoly
may be attributed to decomposition of the compounds in the dry
spike environment according to personal communication with
Jeffery Lowrys experiments (Environmental Resource Associates,
Arvada, CO). The low recoveries of naphthalene in the wet spike
may be attributed to volatilization of naphthalene during the
spike solvent drying step (24). These observed anomalies were
also observed in the following extraction data.
5.6 Closed Extraction System PAH/Soil Results
The effectiveness of the closed extraction system on the
extraction of the simulated PAH soil as described in section
4.3.2. and section 4.1.2 respectively is displayed in Table 5.2.
The simulated, wet spike, and dry spike PAH recoveries for the
78


Table MLNative vs. Wet Spike vs. Dry Spike PAH Recoveries via
Mechanical Dispersion; Bath Sonication
Analyte Matrix: Native ^Recovery VS D Matrix: Wet Spike %Recovery 23/SD Matrix: Dry Spike %Recovery Xj/SD
Naphthalene 18/2.6 25/6.6 66/3.5
Acenaphthylene 1.1/0.4 36/5.0 0.4/0.4
Acenaphthene 15/2.9 36/4.5 72/2.4
Fluorene 26/4.8 39/3 2 68/2.9
Pbenanthrene 25/5.3 40/3.1 723.1
Anthracene 15/3.7 42/2.5 35/5.6
Fluoranthene 18/5.6 35/3-5 67/2.9
Pyrene 13/4J 32/3.0 13/3.6
Benzo(a)Anthracene 15/3.2 47/3.5 77/6.1
Crysene 14/3.4 47/2.6 81/4.0
Benzo(b)Fluoranthene 12/3.1 37/32 76/2.1
Benzo(k)Fluoranthene 9.7/22 44/2.6 64/5.7
Benzo(a)Pyrene 5.9/1.1 41/3-5 4.9/1.4
Dibenz(aJi)Anthracene 6.3/0.9 40/2.9 65/1.5
Indeno( l,22-cd)Pyrene 6.5/0.9 37/3.5 33/3.6
Benzo(g,h.l)Perylene 6.0/0.9 35/2.9 15/2.0
79


novel extraction method. The wet and dry spike recoveries
outperformed the simulated matrix PAH recoveries but exhibited
lower recoveries as compared with the standard extraction
procedure. This phenomena is due to the incorporation of the
spike into the matrix during the mechanical dispersion of the
matrix which will be explored hereinafter.
5.7 PAH/Soil Standard Vs Closed Extraction System Efficiencies
Comparing the standard extraction method and the closed
extraction system method PAH recoveries from the simulated matrix
as determined in section 5.6 and 5.7 respectively, Table 5.3
verifies that the novel extraction method is equivalent or
superior to the standard extraction method for the simulated PAH
wet clay. The minute PAH recoverys standard deviations for both
methods recovery data indicates the precision and
reproducibility of the novel method are acceptable. The
preparatory differences between the standard and novel methods
are the dispersion and sonication techniques. Due to the
migratory patterns of organic compounds through soil matrices and
the strong affinities of the analytes to the matrices (17,18,25-
27), maximum surface area of the matrix must be exposed to the
extraction solvent for optimum recoveries of the organic analyte
of interest. The mechanical dispersion technique provides a finer
80


Table {T.3 Standard Method vs. Novel Method for Native PAH Soil
Analyte Matrix: Native Dispersion: Manual Extranon: Horn %Recovery xy'SD Matrix: Native Dispersion: Mechanical Extraction: Bath y. %Recovery VSD FTest TTest
Naphthalene 8.7/l.S 18/2.6 2.1 7.91
Acenaphthylene 0.5/0.1 1.1/0.4 16 NA
Acenaphthene 8.6/1.9 15/2.9 22 4.99
Fluorene 16/3.7 26/4.8 1.7 4.40
Phenanthrene 17/3.6 25/5.8 2.6 320
Anthracene 8.0/l.S 15/3.7 42 4.76
Fluoranthene 14/3 2 18/5.6 3.1 1.70
Pyrene 10/2.1 13/4-3 4.4* NA
Benzo(a)Anthracene 13/3.4 15/3 2 1.1 1.13
Crysene 14/3.7 14/3.4 12 0.00
Benzo(b)FIuoranthene 12/2.9 12/3.1 1.1 0.00
Benzo(k)Fluoranthene 11/3.3 9.7/2 2 22 0.88
Benzo(a)Pyrene 4.6/I.4 5.9/1.1 1.6 1.92
Dibenz(aJi)Anthracene 7.9/2 2 62/0.9 6.5* NA
Indeno( 122 -cd)Pyrene 7.8/1.8 6.5/0.9 4.0 1.74
Benzo(g,h.i)Perylene 7.9/2.1 6.0/0.9 5.4* NA
*=95% probability level F Test = 428, Ties: 95% = 2.1S, Test 90% = 1.78
81


divided matrix than the manual dispersion. The exposed matrices
hydrophobic bonding sites not only facilitated the higher PAH
recoveries for the mechanical dispersion technique, but also
explains the lower spike recoveries due to incorporation of the
spike compounds to the active sites. The sonication technique
needs to be powerful enough to desorp the analytes from the
adsorption sites and it will be determined that the sonication
bath does provide the needed performance to extract the analytes
as well as the horn sonicator.
5.8 Soil Dispersion/Sonication Comparisons
The explaination of the different PAH recoveries from the
simulated soil matrix via EPA Method 3550 and the closed
extraction method as determined in section 5.7 is determined by
the preparatory differences between the standard and novel
methods are the dispersion and sonication techniques. To support
the hypothesis that the dispersion technique is the recovery
determining step, the horn and bath sonication procedures were
compared. The simulated matrices were extracted by mechanical
dispersion as discussed in section 4.3.2.1 and sonicated by way
of the ultrasonic bath (section 4.3.2.3) or the ultrasonic horn
(section 4.2.1) with their respective PAH percent recoveries
displayed in table 5.4 The similar recoveries indicates that the
82


Table Horn Sonication vs. Bath Sonication
Analyte Matrix: Native Dispersion: Mechanical Extraction: Horn %Recovery Xj/SD Matrix: Native Dispersion: Mechanical Extraction: Bath %Recovery Xj/SD FTest TTest
Naphthalene 13/1_5 18/2.6 3.0 3.5
Acenaphthylene 0.7/03 1.1/0.4 4.0 1.9
Acenaphthene 12/0.6 15/2.9 23* NA
Fluorene 22/2.1 26/4.8 53 13
Phenanthrene 23/23 25/5.8 6.7 0.7
Anthracene 10/13 15/3.7 8.1 1.4
Fluoranthene 16/3.1 18/5.6 33 0.7
Pyrene 12/2.1 13/4.3 1.9 0.5
Benzo(a)Anthracene 15/3.0 15/33 1.6 0
Crysene 15/2.6 14/3.4 1.7 0.5
Benzo(b)Fluoranthene 13/1.7 IZO. 1 3.3 0.6
Benzo(k)Fluoranthene 12/1.8 9.7/23 1.5 1.7
Benzo(a)Pyrene 5.1/0.7 5.9/1.I 2.5 8.9
Dibenz(a.h)Anthracene 7.7/13 63/0.9 2.1 1.8
lndeno( 1 33-cd)Pyrene 7.7/1.0 6.5/0.9 13 1.8
Benzo(g,h.i)Perylene 7.7/1.1 6.0/0.9 1.5 2.5
95% probability level for F Test = 8.94, T Test 95% = 2.31,T Test 90% = 1.85
83


sonication methods are equivalent and, therefore, the dispersion
technique must be the determining factor for the observed
differences in the standard methods lower PAH recoveries as
compared to the closed extraction systems PAH recoveries.
To further support the importance of the dispersion step for
the successful extraction of PAHs from wet clay, the simulated
matrix was exhaustively dispersed with 60 grams Na2S04 in a large
mortar and pestle to produce a very fine dry powder. Extraction
of this ground matrix was performed with horn sonication. Table
5.5 compares the mortar and pestle recoveries with mechanical/
manual dispersion with horn sonication PAH recoveries. While
mortar and pestle does perform well on the control matrix
(previously ground and sieved), utilizing mortar and pestle for
actual field samples is unrealistic for the following reasons: 1)
matrices with high moisture content tend to clump. 2) matrices
containing debris (pebbles, sticks, etc.) will not pulverize by a
manually operated mortar and pestle and, therefore, the soil
matrix on the bottom of the mortar will also not be dispersed. A
mechanical pulverizer/crusher will disperse the matrix, but may
expose activation sites in the matrix not naturally occurring in
the field sample and skew the analyte recoveries low due to
sorption. 3) labor intensive 4) potential of cross contamination.
The dispersion step was shown to cause the observed differences
84


Table SffMortar and Pestle Dispersion vs. Optimum Manual or Mechanical
Dispersion Using Horn Sonication
Analyte Matrix: Native Dispersion: Manual & Mechanical %Recovery x3 (or*)/SD Matrix: Native Dispersion: Mortar & Pestle %Recovery Xj/SD
Naphthalene 13/12 Mechanical 19/1.7
Acenaphthylene 0.7/02 Mechanical 1.1/02
Ace naphthene 12/0.6 Mechanical 16/12
Fluorene 22/2.1 Mechanical 31/32
Phenanthrene 23/22 Mechanical 34/4.0
Anthracene 10/12 Mechanical 16/2.6
Fluoranthene 16/3.1 Mechanical 27/4.0
Pyrene 12/2.1 Mechanical 20/3.0
Benzo(a)Anthracene 15/3.0 Mechanical 26/42
Ciysene 15/2.6 Mechanical 28/4.6
Benzo(b)Fluoranthene 13/1.7 Mechanical 24/3.5
Benzo(k)Fluoranthene 12/1.8 Mechanical 22/4.0
Benzo(a)Pyrene 5.1/0.7 Mechanical 8.9/1.9
Dibenz(aJi)Anthraceae 7.9/22 Manual 14/22
Indeno( 1 22-cd)Pyrene 7.8/1.8 Manual 14/2.5
Benzo{g,h,i)Peryiene 7.9/2.1 Manual 14/22
*=x7
85


in PAH recoveries. Even though the mortar/pestle dispersion was
superior, it is not recommended for real samples and so we
support the mechanical dispersion technique for routine
laboratory extractions.
5.9 Homogeneously Spiked Wet Clay Discussion
The narrow standard deviations exhibited for the simulated
native PAH matrixs extraction recoveries displayed in Tables
5.1-5.5 (generated in sections 5.5-5.8) indicates that a
homogeneously spiked wet clay was created. The synthetic matrix
not only possessed the field matrixs percent moisture and
density characteristics, but the matrices were also visually
indistinguishable.
86


6. CONCLUSIONS
The new SPE cartridge design and an extraction method was
developed to extract the total sedimentary water matrix. The
cartridge was designed to filter out particles >20um by using a
polyurethane foam plug used as a rough filter followed by the top
20um frit of the SPE cartridge. The compressed polyurethane foam
plug was chosen due not only for its filtrating capacity, but
also for its reported analyte adsorption properties (8) It was
discovered during method development that the SPE cartridges were
clogging due to the particulates passing through the first 20um
frit and depositing on the SPE phase. The SPE cartridge that was
developed also employed a mixed C18 and florisil packing that
exhibited a large size distribution range of the phase while
providing an online florisil cleanup. Therefore particles <20um
that do pass through this filter will pass through the SPE phase.
A closed extraction system was also devised to perform eight
extractions simultaneously in the original sample bottles. The
cartridges were located inside the bottle in an inverted position
utilizing gravity to prevent settleable solids clogging the
system. A binary extraction solvent of 20ml 1:1 hexane/acetone
during a powerful 900 watt bath sonication allowed the acetone to
87


remove the residual water from the solid fines, extract the
hydrated pores of the particles, as well as removing the residual
water and channeling pathways remaining on the SPE phase. After
the extraction is complete, the water that was removed from the
apparatus facilitated the complete phase separation of the
acetone/water fraction from the analyte bearing hexane fraction.
This work demonstrates the fast PCB equilibrium between the
aqueous and solid fractions and indicates that the time between
sampling and extraction is unimportant for sedimentary matrices
but thorough mixing of the spike into the matrix must be
performed for valid method development procedures.
The amount of fines contained in a LLE matrix governs the
percent recovery of the PCB found in the total matrix and
demonstrates the importance of extracting both the water and
solid matrices for a successful PCB extraction.
Partition coefficients demonstrate that when dealing with a
large partition ratio, the fines must be considered when choosing
an extraction method. A small partition ratio for an analyte and
matrix system indicates that fines are not important during an
extraction.
The validated solid phase extraction system provides
superior accuracy for PCB recoveries as compared to liquid/liquid
extraction methodologies. The novel SPE method also significantly
88