United States
Environmental Protection
Agency
Environmental Monitoring and
Support Laboratory
Cincinnati, OH 45268
Research and Development
EPA-600/S4-83-052 Jan. 1984
Project Summary
Optimization of Liquid-Liquid
Extraction Methods for Analysis
of Organics in Water
William C. Glaze and C.-C. Lin
This report describes a study of a
liquid-liquid extraction (LLE) procedure
for the analysis of volatile organic
compounds (VOC's) in water. The
method consists of a single step
extraction with n-pentane in a closed
vial, followed by gas chromatographic
(GC) analysis of the compounds in the
extract. Several aspects of the procedure
have been investigated, including
extraction efficiencies, precision, ac-
curacy, and speed. Solvent: water ratio,
pH, and ionic strength of the samples
were the principal variables. In initial
studies, four trihalomethanes (THM's)
were used as analytes; but in later work,
several other Ci and C2 organohalides
were added. Subsequently, the method
was extended to aromatic compounds
such as benzene, toluene, and the
xylenes. As the list of compounds grew,
packed column GC was found to be
inadequate, so open tubular capillary
columns were adopted. Combined with
electron capture (EC) and flame ioniza-
tion detection (FID), capillary columns
allow one to use LLE for mixtures of
these compounds. Method detection
limits (MDL) were below 1 fjg/L for
most organohalides, except for chloro-
form (2.9 A
-------
bromoform benzene
1,1.1-trichloroethane
The method may also be extended to
other nonpolar organic compounds,
particularly those with boiling points
above chloroform, but is not recommended
for very volatile compounds such as vinyl
chloride and the dichloroethylenes,
although with special care and some
modification of the method these sub-
stances may be included. The method will
also measure the dihaloacetonitriles and
dichloroiodomethane, but the analyst
should be aware of the difficulties
associated with the preparation and
preservation of standards, and with the
stability of these substances in water. If
trihalomethane formation potential is to
be measured, a procedure similar to that
described by Glaze et al. should be used,
in which the chlorination reaction is
carried out in the vial used for the
extraction. At the conclusion of the time
required for reaction, quench may be
added and the THM's analyzed as
described below.
The extraction procedure is carried out
in 25-mL (6 dram) borosilicate vials with
screw caps, the tops of which have a 10
mm i.d. perforation. A silicons rubber
liner with a Teflon coating on one side is
used in conjunction with the cap so that a
syringe needle may pierce the liner in
order to withdraw or add materials to the
vial. The Teflon face of the liner is placed
in contact with the sample inside the
bottle.
The water sample to be analyzed is
usually added to the bottle in such a way
as to avoid excessive loss of volatiles, i.e.
without much bubbling. The vial is
completely filled so that a convex
miniscus is seen above the lip of the
bottle. For samples which have residual
chlorine present or suspected, a chlorine
quenching agent is added to the vial
before it is filled with sample. The
nominal amount of quench used in this
work was 0.25 mL of a solution contain-
ing 0.04 M Na2SO3. Care should be taken
in filling the bottle to ensure that the
quench is not washed out. No buffer is re-
quired if the sample is to be analyzed di-
rectly, i.e. without further chlorination.
On the day samples are to be analyzed,
the GC system is properly calibrated
using standards prepared by the procedure
dictated by EPA. A solvent blank should
be run to ensure adequate purity before
the extractions are performed.
Extractions are carried out by first
removing 5.0 mL of water from the vial,
then adding 1.0 mL of n-pentane. These
manipulations are carried out using
syringe withdrawal and injection tech-
niques, while the bottle remains closed.
The pentane used for the extractions
should be free of excessive interferences
and should contain an internal standard
at an appropriate concentration level. For
THM's, the standard used in this work is
1,2-dibromoethane. Other internal stan-
dards which may be used with the ECO
are bromotrichloromethane, 1,4-dibro-
mobutane, and 2-bromo-1 -chloroprop-
ane. n-Decane may be used with the
GC/FID method.
Following addition of the pentane, the
sample is shaken vigorously in order to
reach liquid-liquid equilibrium. Shaking
for one minute by hand, or thirty seconds
with a vortex shaker is sufficient. The vial
is then allowed to stand for a few minutes
until the two layers thoroughly separate.
While awaiting analysis, vials should be
stored with topsides down. Calibration
standards and water blanks should be
treated and extracted by exactly the same
procedure as samples.
GC analysis is carried out by carefully
removing a 1-2 //L aliquot from the
pentane layer in the vial, and injecting it
into the GC. In case the laboratory is
warm, the vial should be cooled in a
refrigerator prior to taking of the aliquot.
For THM analysis, a 10% squalane on
Chromosorb A/AW column is recom-
mended. (Column temperature, 80°C;
injection port temperature, 100°C; 63Ni
EC temperature 300°C; attenuation, 212;
flow rate of helium, 21 mL/min; retention
times: 1.43 chloroform; 2.36 bromodi-
chloromethane; 4.30 dibromochloro-
methane; 5.29 1,2-dibromoethane (in-
ternal standard); 8.29 bromoform).
An SP-1000 or 6% OV-11/4% SP-
2100 mixed-phase column is recom-
mended for confirmation purposes.
(Column temperature, 45°C, 12 min,
1°C/min to 70°C, 63Ni ECD; retention
times: 4.91 chloroform; 5.72 1,1,1-
trichloroethane; 6.37 carbon tetrachloride;
7.00 1,2-dichloroethane; 9.11 trichloro-
ethane; 10.96 bromodichloroethane;
21.01 tetrachloroethylene; 23.14 dibro-
mochloromethane; 39.38 bromoform).
Organohalides, except THM's or in
combination with THM's, may be analyzed
with the mixed phase column, or in some
cases with the squalane column. For
capillary work, an SE-54 fused silica
column is used with splitless injection of
sample. (Temperature program; 0°C, 3
min, 3°C/min to 120°C; column: 30M x
0.25 mm, SE-54 wall coating; splitless
injection with 0.5 min purge delay).
Quantification of the analytes is carried
out by the internal standard method with
EC and Fl detectors for organohalides and
aromatics respectively. Peak areas may
be measured by one of several methods,
but an electronic integrator is recom-
mended. The detector used should be
shown to give a response which is
linearly proportional to the concentration
of the analyte in the range of concentra-
tions measured.
Effect of Matrix Variables
Extraction efficiencies at the 20:1
water to solvent ratio are in excess of the
60% recovery minimum recommended as
good laboratory practice. Table 1 shows
values for nine organohalides using a
drinking water matrix. Detailed studies
have shown that extraction efficiencies
are affected by added sodium chloride
only at salt concentrations above 0.5M
(Table 2). Methanol and sodium sulfite
had no observable effect up to concentra-
tions of 1.0% and 126 mg/L respectively;
extraction efficiency was invariant from
pH4to 10.
Precision and Accuracy of the
Method
Table 3 shows data which illustrate
typical precision obtained by the method
for THM analyses, using isothermal
packed column GC/ECD. In other studies,
the method has proven to be a speedy and
reliable method for THM analysis.
There is considerably less experience
available for judging the precision of the
capillary GC method with temperature
programming. Table 4 shows results of
replicate analyses of solutions near the
detection limit for each of 11 organo-
halides. These data show the capillary
GC/ECD procedure including the extrac-
tion step has a precision of approximate^
20%. Average values of S.D., the stan-
dard deviation of replicate analyses, was
19% with a range of 5% (o-dichloro-
benzene at 9.68 ;ug/L to 44% tetrachloro
ethylene at 0.20 /ug/L). Variances, de
fined as the absolute value of the spikec
value minus the mean of measuret
values (on a percentage basis), average<
10% with a range of 0% (carbon tetrachlo
ride at 0.16 /ug/L) to 27% dibromochloro
methane at 1.08 //g/L). Given the fac
that these data are near the detectioi
limit, one presumes the precision ani
accuracy of the method would be superio
at higher concentration values.
Another, measure of the precision an
accuracy of the capillary GC/LLE metho
may be obtained from the results of an Eft
Quality Control Check (WP 1278). Table
shows the data obtained for this checl
which consisted of two blind checks f
-------
Table 1. LLE Extraction Efficiencies (20:1) Of Nine Volatile Halides From Dent on Tap Water"
Spike
Blank* Concentration Average Efficiency
Compound
-------
Module, using the EC detector. The P/T
procedure is inherently more sensitive
than the LLE method, and combined with
the 63Ni detector, P/T is a powerfully
sensitive method for organohalides.
Studies of the P/T: ECD combination
show that the linear range for the
organohalides is quite limited. Therefore,
in the analysis of the samples listed in
Table 6, dilution of the water samples
was required. By necessity, this is a trial
and error procedure which is time
consuming and which leads to larger
uncertainties than the usual P/T: Hall
method.
Table 6 shows reasonably good agree-
ment between the P/T and LLE methods
with some notable irregularities. The
trichloroethylene values for sample
0880-35-09 are badly out of agreement,
and there are other similar cases involving
bromoform and tetrachloroethylene.
Conclusions
The pentane LLE method may be used
with confidence for the analysis of THM's
and certain organohalides and aromatic
compounds. Precision is generally better
than '±20%, and check samples show an
acceptable accuracy in most cases. For
Table 3.
Precision Of LLE Method For Trihalomethanes In Actual Water Samples*
CHCI3 CHBrClz CHBr£t CHBr3
Texas and Louisiana Cities Waters'1
Cone. Range (ppb)
Median RSD (%)
If
Cone. Range (ppb)
Median RSD (%)
If
0.7 to 56 0.1 6 to 40
7.4 14
8 6
Denton, Texas Water
4 to 90 2 to 50
7.8 5.7
8 8
1. 7 to 22
12
4
2 to 25
10
8
0.65 to 2.0
21
4
1.5 to 2.4
10
4
compounds which do not have a higher
electron capture sensitivity, the MDL's
are high (>5 yug/L). The MDLfor chloroform
is limited by the blank values in pentane
and in water used to prepare calibration
standards. The method is not recom-
mended for very volatile compounds such
as vinyl chloride, but with minor modific-
ations it can be used for less volatile
compounds such as chlorinated ben-
zenes. When complex mixtures are
suspected, the method must be used with
capillary GC techniques which requires
more precision on the part of the analyst.
Comparison of the method with the
Hewlett-Packard 7675A P/T module
usually gave similar values for nine
organohalides in drinking water samples
from six U.S. sources.
"Analyzed on the 12.5% squalane column isothermally at 80° with internal standard. 1,2-dibromo-
ethane.
t'Drinking and other water samples from eight cities in Texas and Louisiana.
c Number of pairs of samples analyzed.
Table 4. Variance and Precision Of Replicate Capillary GC Analysis of Spiked Standards of Organohalides In Water
Low Concentration Range
Upper Concentration Range
Compound
Chloroform
1,1,1- Trichloroethane
1 ,2-Dichforoethane
Carbon Tetrachloride
Trichloroethylene
Bromodichloromethane
1, 1,2- Trichloroethane
Dibromochloromethane
Tetrachloroethylene
Bromoform
o-Dichlorobenzene
N
7
7
7
7
6
6
5
6
7
nm
Si
2.76
0.35
0.57
0.16
0.40
0.46
0.57
0.54
0.20
0.40
4.84
Sm
2.53
0.32
0.53
0.16
0.39
0.47
nm
0.59
0.18
0.53
nm
S.D.%
7
16
9
31
23
6
nm
8
44
25
nm
V%
8
9
7
0
3
2
nm
9
10
18
nm
N
8
9
9
10
10
9
9
9
9
9
4
Si
5.52
0.70
1.14
0.32
0.80
0.92
1.14
1.08
0.40
0.80
9.68
Sm
5.36
0.74
1.24
0.37
0.82
1.08
1.19
1.37
O.47
0.98
10.01
S.D.%
18
28
12
27
26
19
18
19
15
20
5
V%
3
5
9
16
3
17
4
27
18
22
3
N: number of replicate analyses; Si: spiked concentration (ug/L); Sm: measured value (mean ofN determinations with GC data system calibrated at Si);
S.D.: standard deviation, %; V=Sm - Si S.D. V.: standard deviation of variance, %; nm: not measureable at this concentration.
-
Table 5.
Analysis of EPA Quality Control Check Sample #WP 1278 by the Pentane LLEMethod
With FSOT Capillary GC Analysis
V9/L
Sample No. 1
Sample No. 2
1 . 2 -Dichloroethane
Chloroform
1,1,1- Trichloroethane
Trichloroethylene
Carbon Tetrachloride
Tetrachloroethylene
Dibromochloromethane
Chlorodibromomethane
Bromoform
EPA Value
1.5
11.0
1.1
2.6
2.3
1.1
1.7
2.4
2.8
Sm
b
11.0
0.9
2.5
2.2
1.0
1.7
2.2
2.9
%Dev.
10
12
1
1
5
0
5
1
% Var.
0
18
4
4
9
0
8
4
EPA Value
2O.O
45.6
14.0
13.0
9.4
5.6
8.6
12.0
10.4
Sm
13.6
51.2
12.3
12.9
9.9
6.0
10.0
12.2
13.6
%Dev.
4
7
6
3
5
2
4
3
8
% Var.
32
12
12
o.a
5
7
16
2
37
Sm: mean of duplicate runs (separate extractions and analysis); % Dev: percent deviation of duplicate runs; %
Var: EPA Value-Sm/EPA value* 100; b: below blank tevel[FSOT capillary column with SE-54 phase. 0°C, 3 min;
3°/min 120°C; split/ess injection, other conditions in text.]
-------
Table 6. LLE/Capillary GC and Purge/Trap Analysis of Drinking \Water Samples
Detection
Limit
Compound
Chloroform
1, 1 ,1 -Trichloroethane
1, 2-Dichloroethane
Carbon Tetrachloride
Bromodichloromethane
Dibromochloromethane
Trichloroethylene
Tetrachtoroethylene
Bromoform
Benzene
Toluene
O-Dichlorobenzene
LLE
2.9
0.6
8
0.3
0.6
0.8
0.6
0.2
0.6
P/T
04
01
0.05
01
O.OB
02
0.2
001
0005
Blank
Value
LLE
08
0.08
8
001
_
_
0.04
p/r
0.4
0.1
o.os
0.1
0.06
02
0.2
001
0005
0202-13-13
LLE P/r
/ 9±01 1 7 ±04
O.25 ± 0.03
_ _
432±003 17±0.1
039 ±0.02 0.38 ±004
0.89 ±OOS
0618-07-02
LLE
1.7 ±0.1
1.6 ± 0. 1
_
12 3 ±0.6
P/T
20 ± 0.06
3.4 ±003
0 16 ± 0.004
07 ±002
020± 0.01
0. 12 ± 0.02
5.7 ±0.3
71 ±0.8
012 ±004
033 ±002
0737-38-13
LLE
10. 5 ±0.6
1.1 ±01
30 ±0.3
2.76 ±0.03
2 7 ±0.3
3.8 ±0.3
_
P/T
073± 0.02
9.0 ±06
0.1 ±0,02
1.23 ± O.O3
4.7 ±0
3S± 0.05
2.9 ± 0.4
3 4 ±0.1
1.0 ±0.1
008± 0.001
0314-16-11
LLE
36±04
_
2.3 ±009
3.9 ± 0.2
14.0 ±09
30 1 ± 1.8
/ 8 ± 0.5
28.1 ±1.7
_
P/T
4.2
a
O45 ± O.O2
4.0 ±02
1 1.9 ±0.1
33 3 ± 0.2
074 ±0.01
09 ±0.2
16 6 ± 1.6
0880-35-09
LLE
4.7 ±4
2.8 ± 0.2
7.2 ±0.1
6.2 ± 0.4
29 7 ± 0.8
028± 0.02
31 ±0.1
P/T
4.7 ±0.1
04 ±0.03
22±02
3.0 ±.1
5.6 ±01
4.1 ± 1
1.10±005
1.0 ±.1
LLE: 20/L Pentane microextraction with GC/ED or GC/FID, 30 m x 025 mm i.d FSOT/SE-54; 0°, 3 mm. 3°/mm to 120°
P/T: 5mLsample. 10minHepurgeat4OmL/min;desorption.250°:0.2%Carbowax1SOOonaO/100Carbopac,organohahdes.columnat60°,3min.8°/minto130°.ECD;aro
3 min, 15°/min to 160°. FID.
Blank values subtracted. No values reported below detection limits listed in Tables 30 and 31 of the full report
William C. Glaze and C.-C. Lin are with the University of Texas at Dallas,
Richardson, TX 75080.
Thomas A. Bellar is the EPA Project Officer (see below).
The complete report, entitled "Optimization of Liquid-Liquid Extraction Methods
for Analysis of Organics in Water." (Order No. PB 84-112 937; Cost: $14.50.
subject to change} will be available only from:
National Technical Information Service
5285 Port Royal Road
Springfield. VA22161
Telephone: 703-487-4650
The EPA Project Officer can be contacted at:
Environmental Monitoring and Support Laboratory
U.S. Environmental Protection Agency
Cincinnati, OH 45268
-------
United States
Environmental Protection
Agency
Official Business
Penalty for Private Use $300
Center for Environmental Research
Information
Cincinnati OH 45268
mi
A^^
1111 { f:B lr-8* \ E!s02)S
fill \ ......
'.
ft U.S. GOVERNMENT PRINTING OFFICE: 1964-759-102/8:
------- |