United States
Environmental Protection
Agency
Environmental Monitoring
Systems Laboratory
Las Vegas, NV 89193-3478
Research and Development
EPA/600/S4-90/026
March 1991
4>EPA Project Summary
Method for the Supercritical
Fluid Extraction of Soils/
Sediments
V. Lopez-Avila, N. S. Dodhiwala, and W. F. Beckert
The Environmental Protection Agency
is interested in new and improved
analytical methods that are (aster, better,
and cheaper than the present methods,
and that, at the same time, are safe and
generate little or no waste. For a number
of years, supercritical fluid extraction has
been publicized as a new and promising
technique for the extraction of organic
compounds from solid matrices; how-
ever, applications of supercritical fluid
extraction techniques to the extraction of
compounds regulated by the Environ-
mental Protection Agency from matrices
of concern to the Agency have been
rather limited. In late 1988, we began a
study to develop a supercritical fluid
extraction technique for environmental
matrices such as soil, sediment, and fly
ash. The approach taken in this study was
carried out in various phases. In the first
phase, we reviewed the literature pub-
lished on the supercritical fluid extraction
technique. While the literature gathering
and review took place, we contacted
manufacturers of supercritical fluid
extraction instrumentation to identify the
most suitable equipment for this study.
In the second phase of the study, we
purchased a supercritical fluid extractor
system and conducted a series of exper-
iments to familiarize ourselves with the
instrumentation and to determine the
feasibility of extracting various classes of
organic compounds from solid matrices.
Sand was the primary matrix selected for
investigation, although soils and sedi-
ments have also been tested. Upon com-
pletion of the second phase, we prepared
a protocol that describes in detail how
a solid sample should be extracted under
supercritical conditions with carbon di-
oxide or carbon dioxide containing
modifiers, the instrumentation require-
ments, etc. The third phase of the study
was devoted primarily to method opti-
mization, application of the supercritical
fluid extraction technique to various
environmental matrices (mostly standard
reference materials certified for certain
organic compounds), and instrument
modification to increase sample through-
put (e.g., installation of two and four
extraction vessels for parallel extrac-
tions).
This Project Summary was developed
by EPA's Environmental Monitoring
Systems Laboratory, Las Vegas, NV, to
announce key findings of the research
project that is fully documented in a
separate report of the same title (see
Project Report ordering information at
back).
Introduction
An important step in the analysis
sequence for organic pollutants is ex-
traction of analytes from the samples, that
means, their separation from the matrices.
The sample extraction technique should]
to the extent possible, yield quantitative
recoveries of the target analytes from the
matrices, be selective so that extraction
of interferants is minimized, not generate
large volumes of waste solvents, require
little sample and extract handling to mini-
mize analyte losses and contamination,
and be fast and inexpensive. For the ex-
traction of organic pollutants from solid
samples (soil, sediment, tissue, fly ash,
etc.), two methods are currently included
nS) Printed on Recycled Paper
-------
in the SW-846 methods manual (1):
Soxhlet extraction (Method 3540) and
sonication extraction (Method 3550).
For a number of years, a new and
promising technique for the extraction of
organic compounds from solid matrices
has been publicized: supercritical fluid
extraction (SFE). The actual, or claimed,
advantages of SFE over conventional
extraction methods include:
• SFE is much faster than Soxhlet extrac-
tion; extraction times of as little as 1
min can result in quantitative recover-
ies.
• No toxic and expensive solvents are
required; this results in reduced mate-
rials and waste disposal costs and in
reduced environmental pollution. No
solvent removal or glassware cleaning
is required.
• SFE conditions can easily be optimized
by just varying pressure and temper-
ature and by using modified supercrit-
ical fluids (SFs).
• Overall, the use of SFE techniques in
place of conventional methods results
in substantial cost and labor savings.
The most commonly used SF is carbon
dioxide; it is so popular because of its low
critical temperature (31.3°C) and pressure
(72.9 atm), and because it is nontoxic,
nonflammable, and its use does not result
in a waste disposal problem.
Utilization of the unique extraction
properties of SFs for the preparation of
analytical samples has been reported in
a number of publications in the recent
technical literature. However, applications
involving extraction of compounds reg-
ulated by the Environmental Protection
Agency have been quite limited. A more
detailed discussion of such applications
reported in the literature can be found in
Appendix A of this report.
The study described in this report was
carried out in various phases. In the first
phase, we reviewed the literature pub-,
lished on the SFE technique. While the
literature gathering and review took place,
we contacted manufacturers of SFE
instrumentation to identify the most
suitable equipment for this study. In the
second phase, we purchased an off-line
supercritical fluid extractor system from
Suprex Corporation and conducted a
series of experiments to familiarize
ourselves with the instrumentation and to
determine the feasibility of extracting
various classes of organic compounds
from solid matrices. Sand was the primary
matrix selected for investigation, although
soils and sediments have also been
tested. Upon completion of the second
phase, we prepared a protocol that
describes in detail how a solid sample
should be extracted by SFE, the instru-
mentation requirements, etc. This proto-
col is included as Appendix B of this
report. The third phase was devoted
primarily to method optimization, applica-
tion of the SFE technique to various
environmental matrices, and instrument
modification to increase sample through-
put (e.g., installation of two and four
extraction vessels for parallel extractions).
The fourth phase which will not be
discussed in this report will address the
evaluation of various SFE systems that are
available commercially.
Experimental
Apparatus
Supercritical fluid extraction system—
Suprex Model SE-50. The system was set
up either with one, two, or four extraction
vessels for parallel extractions. Supercrit-
ical pressures were maintained inside the
extraction vessels by using uncoated
fused-silica tubing as restrictor. Figure 1
shows a schematic diagram of the four-
vessel-extraction setup. Eight restrictors
were mounted in the 12-port valve to
allow collection of two fractions per
sample. Fractions 1 A, 2A, 3A, and 4A were
collected when the 12-port valve was in
the load position. •Fractions 1B, 2B, 3B,
and 4B were collected when the 12-port
valve was in the inject position. Collection
of the extracted material was performed
by inserting the outlet restrictor into a 15-
mm X 60-mm glass vial containing an
organic solvent spiked with a known con-
centration of an internal standard
(terphenyl-d14).
Hexane was used as collection medium
for the experiments performed with
carbon dioxide, and methylene chloride
or methanol was used as collection
medium for the experiments performed
with carbon dioxide modified with 10
percent methanol.
Procedure
A known amount of sample (typically
1 to 10 g) was weighed out in an aluminum
cup; if the sample had to be spiked with
the target analytes, then spiking was
performed directly in the aluminum cup
by using 100 to 1000 pL volumes of
concentrated stock solutions of the target
analytes. After the solvent had evaporated
completely (approximately 15 min), the
spiked sample was transferred to the
extraction vessel. Surrogates were spiked
directly into those samples already placed
in the extraction vessels and that were
to be analyzed by GC/MS. Details of
sample extraction are given in the pro-
tocol included as Appendix B of this
report. Extractions were performed with
both carbon dioxide and carbon dioxide
containing 10 percent methanol at con-
stant temperature and pressure or under
conditions in which both the temperature
and pressure were varied. Furthermore,
several fractions were collected per
sample to verify the completeness of the
extraction. For example, a standard
reference soil contaminated with poly-
nuclear aromatic hydrocarbons (PAHs)
was extracted with carbon dioxide with
10 percent methanol as follows: to collect
Fraction 1, the extraction was started at
150 atm and 50°C; after pressurizing the
vessel for 10 min, the pressure was
increased to 200 atm and the temperature
to 60°C; after another 10 min, the pressure
was increased to 250 atm and the
temperature to 70°C and maintained for
10 min. Fraction 2 was collected at 250
atm and 70°C (30 min), Fraction 3 at 300
atm and 70°C (30 min), and Fractions 4
through 9 at 350 atm and 70°C (30 min
each).
Method Optimization
The experimental design for method
optimization focused on seven variables,
each chosen at two levels (high and low).,
Two separate sets of tests were per-
formed. The seven variables were pres-
sure (P), temperature (D), moisture (M), cell
volume (V), sample size (S), time (T), and
modifier volume (C) for Test 1, and
pressure (P), temperature (D), volume of
toluene (F), collection volume (G), mois-
ture (M), glass beads (B), and static
extraction time (E) for Test 2. The group
differences for Test 1 (VP through Vc) were
calculated using Equations 1 through 7;
those for Test 2 were calculated in a
similar manner.
VP = 1 /4(w + x + y + z) -
1/4(s + t+u+v) = p-P (1)
VD = 1 /4(u + v + y + z) -
1 /4(s +1 + w + x) = d - D (2)
VM = 1 /4(t + v + x + z) -
' 1 /4(s + u + w + y) = m - M (3)
Vv = 1 /4(u + v + w + x) -
1 /4(s +1 + y + z) = v - V (4)
-------
Vial Vial
1A 4A
1 to 8 Port
Manifold Tee
*— Vial4B
*•— Vial 38
Inject position
CO,
Vial Vial
2B 1B
Vial Vial
2A 3A
Vial Vial
1A 4A
1 to 8 Port
Manifold Tee
Vial
2A
Vial
3A
Load position
Figure 1. Schematic representation of four-vessel-extraction setup.
Vs = 1 /4(t + v + w + y) -
1 /4(s + u + x + z) = s - S
VT =1/4(t + u+x + y)-
1 /4(s + v + w + z) = t - T
1 /4(s + v + x + y) = c - C
(5)
(6)
(7)
The relative changes as 100 X change/
average recovery at low level were
calculated and the data were sorted by
variable and compound. The variables
were ranked in increasing order of
absolute relative change for each com-
pound and the sum of ranks was calcu-
lated.
A full factorial design experiment was
conducted for four variables (time, pres-
sure, moisture, and sample size). All
experiments were performed in duplicate
using the standard reference soil contam-
inated with PAHs (SRS103-100 soil). The
values for the four variables were as
follows: time (1 hr and 2 hrs), pressure
(250 atm and 350 atm), moisture content
(10 percent and 25 percent), and sample
size (1 g and 2.5 g). The duplicate ex-
periments were performed in parallel by
splitting off the incoming carbon dioxide
to two extraction vessels. All other
-------
variables were kept constant as follows:
temperature70°C, collection solvent —
hexane 5 ml, restrictor 50 jum ID, no static
extraction, extraction vessel 2 mL
From the recovery data that we ob-
tained for the 32 experiments, we estab-
lished the effect of the individual varia-
bles, the interaction between the two
variables, and the interaction between
three variables. Analysis of variance was
performed using the general linear model
procedure of the Statistical Analysis
System.
Results and Discussion
Summary of the
SFE Preliminary Work
' The extraction efficiencies we achieved
for sand samples spiked with Aroclor
1232 and 1260, organochlorine pesti-
cides, organophosphorus pesticides,
PAHs, nitroaromatic compounds, halo-
ethers, and some of the base/neutral/
acidic compounds targeted by the U.S.
Environmental Protection Agency were
quite good (an example is shown in Table
1 for the organochlorine pesticides).
However, the extraction efficiencies we
achieved with the standard reference
materials (Tables 2 and 3) varied consid-
erably. Some of our recoveries were much
lower than the certified values or the
values published by others. However, on
closer examination, it turns out that many
of the studies reported so far in the
literature were more of a qualitative
nature, were often conducted with home-
made equipment, focused' mostly on
PAHs, and used small sample sizes (as
small as a few milligrams). To develop an
SFE method that can successfully be
applied to samples of interest to the EPA,
we have to use commercially available
equipment, consider a wide variety of
sample matrices and groups of pollutants,
and use sample sizes large enough to
compensate for the inevitable inhomoge-
neities of most real environmental and
hazardous waste samples. We, therefore,
reassessed our approach to SFE evalu-
ation and first looked more closely at the
many variables that affect SFE efficiency,
as summarized in the following para-
graphs.
Small sample sizes (as small as 20 mg
or even less) are adequate when homo-
geneous material is extracted, e.g., for the
extraction of stabilizers and plasticizers
from polymers, extraction of waxes, of
many pharmaceutical materials, etc.
Table 1. Percent Recoveries and Percent RSDs of 41 Organochlorine Pesticides from Spiked
Sand Using Supercritical Carbon Dioxide Modified with 10 Percent Methancl"
j. Spike Percent
Compound Level . ' Average Percent
No. Compound (ng/g) Recovery RSD
1 alpha-BHC 62.5 93.6 16.8
2 gamma-BHC 62.5 121 10.8
3 beta-BHC 62.5 93.6 25.2
4 Heptachlor 62.5 91.7 23.0
5 delta-BHC 62.5 103 19.9
6 A Mr In ft? ^ Qfi 7 7^ 0
r\IQitfi \j£.,\J J7O. / I \J,S3
7 Heptachlor epoxide 62.5 88.6 12.2
8 Endosulfanl 62.5 88.2 10.3
9 4,4'-DDE 62.5 95.4 10.6
10 Dieldrin 62.5 103 19.6
11 Endrin 62.5 92.8 13.6
12 4,4'-DDD 125 120 10.1
13 Endosulfan II • 62.5 89.6 6.9
14 4,4'-DDT 62.5 99.9 4.9
15 Endrin aldehyde 125 120 10.8
16 Endosulfan sulfate 62.5 125 8.8
17 Methoxychlor 62.5 107 2.4
18 Endrin ketone 62.5 98.1 6.7
19 Hexachlorocyclopentadiene 62.5 26.1 13.7
20 Etridiazole 62.5 87.9 14.5
21 Chloroneb 1,250 92.5 20.8
22 Propachlor 1,250 95.8 22.1
23 Hexachlorobenzene 62.5 114 7.3
24 Trifluralin 125 94.5 19.7
25 Diallate 1,875 109 9.9
26 , Pentachloronitrobenzene 62.5 97.8 14.8
27 Chlorothalonil 125 93.4 9.9
28 Alachlor 625 103 20.5
29 DCPA 62.5 107 . 5.9
30 Isodrin 62.5 • 98.6 17.9
31 trans-Permethrin 625 102 17.9
32 gamma-Chlordane ' 62.5 95.8 18.5
. 33 alpha-Chlordane 62.5 99.8 11.4
34 trans-Nonachlor 62.5 108 11.4
35 Captan 125 95.0 6.5
36 Perthane 1,875 79.0 8.9
37 Chlorobenzilate 125 28.4 , 12.6
38 Chloropropylate 62.5 109 10.3
39 Kepone , ' 62.5 106 19.5
40 Mirex 125 96.4 3.7
41 DBCP 62.5 35.4 29.2
a The number of determinations was three. The sample size was 8 g. Each extraction was performed
with carbon dioxide modified with 10 percent methanol at 150 atm/50°C/10 min static, 200
atm/60°C/10 min dynamic, 250 atm/70°C/10 min dynamic.
However, most environmental and waste materials that are available were certified
samples are inhomogeneous even after for only a very limited number of com-
mixing. Therefore, the sample sizes pounds, e.g., PAHs. It is therefore difficult,
specified in SW-846 are 10 g (Soxhlet) even currently impossible, to determine
and 30 g or 2 g (sonication), depending absolute extraction efficiencies for most
on the concentrations of the analytes in analytes because in most cases, removal
the sample. For this reason, samples for of a spike from a sample matrix is much
SFE in the environmental field should be easier than removal of "incorporated" or
in the 1- to 10-g range, preferably at least "native" compounds. This, however, is a
5 g. problem that hampers the evaluation of
There is a lack of standard reference all extraction methods, not just SFE, and
materials that cover the matrices and one is usually confined to comparing
pollutants of environmental concern. The relative extraction efficiencies.
-------
Table 2. Comparison of Recoveries Obtained by SFE with Certified Values Reported for SRS
103-100*
Compound
Naphthalene
2-Methylnaphthalene
Acenaphthylene
Acenaphthene
Dibenzofuran
Fluorene
Phenanthrene
Anthracene
Fluoranthene
Pyrene
\ Benzo(a)anthracene
Chrysene
Benzo(M-k)fluoranthene
Benzoi(a)pyrene
Pentachlorophenol
Certified Value
(vg/g)
32.4
62.1
19.1
632
307
492
1,618
422
7,280
7,033
252
297
752
97.2
965
±
+
±
±
+
±
_±
±
±
±
±
± •
± '
±
±
8.2
11.5
4.4
105
49
78
348
49
220
289
38
26
' 22
17.1
374
Percent
Average
Recovery
by SFE"
63.8
82.6
64.6
98.2
92.9
80.4
724
78.4
92.3
78.2
67.6
68.4
53.3
32.2
747
Percent
RSD
11.1
2.2
9.4
4.2
3.5
4.3
9.8
14.7
6.5
7.0
4.2
10.2
14.3
19.3
11.1
a The number of determinations was three. The sample size was 2.5 g. Each extraction was
performed with supercritical carbon dioxide at 300 aim and 70° C for 60 mm; 10 percent moisture
was added to each sample prior to extraction.
Ms percent of certified values.
Table 3. Comparison of Recoveries Obtained by SFE with Certified Values Reported for NIST
SRM 1941*
Compound
Phenanthrene
Anthracene
Pyrene '
Fluoranthene
Benzo(a)anthracene
Benzo(b)anthracene
Benzo(k)anthracene
Benzo(a)pyrene
Perylene
Benzo(g,h,i)perylene
lndeno(1,2,3-c,d)pyrene
Certified Value
fog/g)
0.577
0.202
1.08
1.22
0.550
0.78
0.444
0.67
0.422
0.516
0.569
±
±
±
±
±
±
±
±
±
, ±
±
0.059
0.042
0.20
0.24
0.079
0.79
0.049
0.73
0.033
0.083
0.040
Concentration
Found
(tig/g)
0.25
NO*
0.70
0.70
0.20
0.10
ND
ND
-
ND
ND
Percent
Recovery
43.3
—
64.8
57.4
36.4
12.8
-
_
-
-
-
Surrogate recovery
2-Fluorophenol
Phenol-d5
Nitrobenzene-d5 .
2-Fluorobiphenyl
2,4,6-Tribromophenol
64.1
71.5
86.1
85.1
68.9
a The experiments were performed with supercritical carbon dioxide at 350 atm/60° C for 20
min. The sample size was 2 g.
bND-not detected; estimated detection limit 0.1 ng/g.
Temperature and pressure changes
affect the density and viscosity of an SF
and therefore its solubilizing ability.
However, little is understood about what
happens on the surfaces of the solid
matrices during the extraction process,
and what the desorption, solvation, and
transport mechanisms are, and little is
known about how to optimize pressure
and temperature for specific matrices and
analyte groups. An understanding of the
desorption and transport mechanisms of
a solute under supercritical conditions
would provide clues if the use of static
or dynamic extraction conditions, or a
combination of the two, would be advan-
tageous; whether rapid pressure fluctua-
tions would improve extraction rates and,
maybe, extraction efficiencies and selec-
tivities.
Modifiers are liquids that can be added
to the SF or directly to the sample. They
usually are more polar than the basic SF;
for example, benzene, toluene, methanol,
and acetone have been added to carbon
dioxide. Their use improves solvent
strength and affects selectivity. However,
we are mostly confined to the trial-and-
error approach. We do not know to what
extent, and in what way, the modifier
changes the properties of the supercrit-
ical system and in which way solubility
is enhanced; the selection of modifiers for
specific applications is still largely
guesswork.
Draft Protocol
Based on our results from the extraction
of relatively simple matrices, we devel-
oped a draft protocol in the SW-846
format, "Extraction Procedure Using
Supercritical Fluids." Our goal was to
write a generic protocol that is applicable
to as many different SFE systems as
possible. This protocol was written for
solid matrices like soils and sediments;
the target analytes include organochlo-
rine pesticides, organophosphorus pes-
ticides, nitroaromatics, naloethers, base/
neutral/acidic compounds, and PAHs.
The protocol addresses interferences,
apparatus and materials, sample prepa-
ration (including extraction), and quality
assurance; recovery data for the above
groups of compounds are included. The
protocol will be refined and updated as
additional data become available.
{
Method Optimization
The experimental design of our method
optimization study focused on two sets
of seven variables, each chosen at two
levels. A full factorial design would have
required 128 experiments. Instead, this
design required us to perform only eight
experiments in order to estimate the main
effects of the two sets of seven variables;
however, we could not test for statistical
significance of any of these effects be-
cause there were no degrees of freedom
for the error terms. From these data, one
can calculate the effect on the recovery
of each variable at its low and its high
value using Equations 1 through 7. To get
an overall picture of which variables affect
-------
the recovery of the highest number of
compounds, the group differences were
ranked and the ranks across the 15
compounds were summed up.
From these sums, we concluded that
In Test 1 the recovery was most affected
by time (sum of ranks 88) and least
affected by the volume of the modifier
(sum of ranks 31), Pressure and moisture
ranked second (sum of ranks 70) and third
(sum of ranks 69) in importance after time.
The sum of ranks for sample size,
temperature, and cell volume are 65, 50,
and 47, respectively.
When the experiments were performed
at the same pressure, temperature, and
moisture conditions but toluene was used
as modifier and the extraction vessel was
pressurized for 15 or 30 min during the
60-min extraction, then recovery was
most affected by moisture (sum of ranks
88) followed by pressure (sum of ranks
70) and volume of toluene (sum of'ranks
68). The fourth variable to influence was
the static extraction time (sum of ranks
57). Temperature, volume of collection
solvent, and the presence/absence of
'glass beads were the least important.
From the experiments in Test 1, we
concluded that the four variables time,
pressure, moisture, and sample size
should be further investigated. Pressure
and moisture were also identified as being
important variables in the experiments
. conducted in Test 2. Therefore, we de-
signed a full factorial experiment in which
the four variables were investigated at two
levels in 16 experiments. The sample used
was again SRS 103-100. From the recov-
ery data, we established the effect of the
individual variables, the interaction
between two variables, and the interaction
between three variables. The significance
levels from the analysis of variance
(ANOVA) for each of the 15 compounds
known to be present in the SRS 103-100
are included in the report. The four-way
interaction (time X pressure X moisture
X sample size) was not considered in
ANOVA.
CONCLUSION
The results of this study indicate that
SFE with carbon dioxide or carbon
dioxide containing modifiers is an attrac-
tive method for the extraction of organic
contaminants from environmental solid
matrices. Potential advantages of the
method include less solvent use and
disposal, reduced manpower require-
ments, and increased speed and selec-
tivity (in combination with modifiers).
However, more developmental work has
to be done before SFE becomes an easy-
to-use, off-the-shelf method.
References
1. Test Methods for Evaluating Solid
Waste (1986), 3rd Ed., SW-846, U.S.
Environmental Protection Agency, Wash-
ington, DC.
-------
-------
V. Lopez-Avila and N. S. Dodhiwala are with Mid-Pacific Environmental Laboratory,
Mountain View, CA 94043. The EPA author, Werner F. Beckert, (also the EPA
Project Officer) is with the Environmental Monitoring Systems Laboratory, Las
Vegas, NV 89193-3478.
The complete report, entitled "Method for the Supercritical Fluid Extraction of Soils/
Sediments," (Order No. PB91-127803/AS; Cost: $31.00, subject to change) will be
available only from:
National Technical Information Service
5285 Port Royal Road
Springfield, VA 22161
Telephone: 703-487-4650
The EPA Project Officer can be contacted at:
Environmental Monitoring Systems Laboratory
U.S. Environmental Protection Agency
Las Vegas, NV 89193-3478
United States
Environmental Protection
Agency
Center for Environmental Research
Information
Cincinnati, OH 45268
BULK RATE
POSTAGE & FEES PAID
EPA PERMIT NO. G-35
Official Business
Penalty for Private Use $300
EPA/600/S4/90/026
------- |