United States
Environmental Protection
Agency
Environmental Monitoring and
Support Laboratory
Cincinnati OH 45268
Research and Development
EPA/600/S4-86/038 March 1987
Project Summary
Standardization of EPA
Method 86 10, Part 2
S. V. Lucas, A. Riggin, T. F. Cole,
K. B. Degner, and W. M. Cooke
U.S. EPA Method 8610, "Total
Aromatics by Ultraviolet Absorption"
was evaluated in conjunction with U.S.
EPA Method 3560, "Reverse Phase
Cartridge Extraction" for the separation
and semi-quantitative determination of
visible or ultraviolet absorbing organic
compounds listed in Appendix VIII of
the Resource Conservation and Re-
covery Act (RCRA). In Part 1 of this
program, reported separately, the fol-
lowing work was conducted:
• A data base of visible and ultra-
violet (UV) spectral data for the
Appendix VIII compounds was
developed and used to estimate
detection limits for those com-
pounds which absorb UV or visible
light in the region 220 to 700 nm.
• The reverse phase cartridge extrac-
tion procedure of Method 3560
was evaluated and modified for
the separation of polar and non-
polar subsets of 21 Method 8610
analytes using methanol and
hexane eluents. However, the ex-
traction procedure was found to
be unsuitable for group separation
in its present form, and the results
indicated that non-overlapping
group separation was probably
chemically unattainable.
• The spectrophotometric determina-
tive technique of Method 8610
was evaluated and found to be
very sensitive for a majority of the
compounds in the range of 220 to
400 nm.
Based on these Part 1 results, a Part
2 study was conducted to further in-
vestigate the applicability of these
methods in a variety of ground-water
samples and to refine method detection
limit estimates.
Seven ground-water samples were
supplied for the Part 2 study by ASTM
Committee D-34 members. These sam-
ples were evaluated for background UV
absorbance, and duplicate sample ex-
tractions were used to simulate down-
gradient versus up-gradient testing. An
estimated positive response decision
level was found to be 0.02 absorbance
units. Five Method 8610 analytes were
evaluated for spike recoveries from both
reagent water and a composite ground-
water sample. One analvte was found
to be unstable in water and the elution
solvents used. The remaining four
analytes had good total recoveries from
reagent water ranging from 79 to 108
percent with standard deviations of all
but one analvte ranging from 1 to 5
percent. Spike recoveries for com-
posited ground water were not repro-
ducible due, apparently, to a significant
variability in recovery of native UV
absorbing material. The cause of the
variability could not be specifically at-
tributed to, but may have been associ-
ated with, the presence of very finely
divided (<20 micron paniculate material.
This Project Summary was developed
by EPA's Environmental Monitoring and
Support Laboratory, Cincinnati, OH, to
announce key findings ot the research
protect that Is fully documented In a
separate report of the same title (see
Project Report ordering Information at
back).
Introduction
The U.S. Environmental Protection
Agency (EPA) has proposed an amend-
ment (October 1, 1984, Federal Register)
to its hazardous waste regulations under
the Resource Conservation and Recovery
Act (RCRA) consisting of a hierarchical
-------
analysis procedure for ground-water test-
ing. The proposed hierarchical procedure
will allow hazardous waste facility oper-
ators to screen ground-water samples
quickly for specific compounds using in-
expensive methods, thus reducing their
overall testing burden for regulated pol-
lutants without jeopardizing environ-
mental protection.
One of the proposed methods included
in the hierarchical analysis protocol is
EPA Method 8610, "Total Aromatics by
Ultraviolet Absorption." When used in
conjunction with EPA Method 3560,
"Reverse Phase Cartridge Extraction,"
Method 8610 will allow operators of
hazardous waste facilities to monitor
ground water beneath their facilities for
many regulated compounds. This screen-
ing data allows the operator to also make
decisions for advanced testing.
To evaluate the usefulness of Methods
3560 and 8610, Battelle's Columbus
Division, under Contract Number 68-03-
1760, conducted a research program to:
(1) generate a data base of ultraviolet and
visible spectral data for Method 8610
analytes; (2) evaluate Method 3560 for
the collection and separation of polar and
nonpolar Method 8610 analytes; and (3)
evaluate Method 8610 for the analysis of
total aromatic compounds. The results of
this work have been reported separately
(c.f., Project Summary EPA/600/S4-
85/052, December 1985). The results
reported in this Project Summary are for
a Part 2 continuation of the research
program. Part 2 efforts involved: (1)
acquisition of authentic ground-water
samples from ASTM Committee D-34
members; (2) evaluation of matrix effects
on ultraviolet determinative techniques
using Method 8610; and (3) evaluation of
recoveries and detection limits of selected
Method 8610 analytes spiked into
reagent water and ground-water samples
containing significant UV background. For
the purpose of this Project Summary, the
reader's familiarity with the earlier work
is assumed.
Experimental Procedures
Reagent water and ground-water
samples tested in this work were ex-
tracted with a Baker-10 SPE vacuum
manifold system and 6-ml disposal
octadecyl (C18) solid-phase extraction
(SPE) cartridges. Each SPE cartridge con-
tamed about 500 mg of sorbent specified
to have a mean diameter of 600-^m and
a mean pore size of 60 A. Samples were
eluted with both methanol and hexane in
all experiments. The methanol and hexane
were obtained from Burdick & Jackson
and were used as received. Reagent
water was generated from an in-house
Milli-RO/Milli-0® system.
UV analysis of methanol and hexane
SPE eluates was conducted with a Gary
17DX recording UV/VIS spectrophoto-
meter. This system has a short-term noise
level of 0.001 absorbance units (a.u.) and
a non-correctable baseline non-linearity
of about ±0.002 a.u. between 220 and
400 nm. Analyses of 20 SPE-extracted
reagent water method blanks indicated
the system's overall spectral variability to
be approximately 0.005 a.u. between 230
and 400 nm.
Results and Discussion
Evaluation of Ground-
Water Matrix Effects
Seven ground-water samples were
contributed by ASTM Committee D-36
members. Three samples contained no
paniculate matter and had the appearance
of ordinary drinking water. Two samples
contained a moderate amount of par-
ticulate matter, and the remaining two
samples contained a significantly greater
amount of particulate matter. In all
samples, the particulate matter appeared
to settle completely on overnight standing.
The UV spectra of methanol and hexane
SPE eluates of the seven ground-water
samples were examined two ways: (1)
versus a reagent water blank to determine
the UV spectra of the SPE recovered
material, and (2) versus a second replicate
of the same ground water to simulate a
down-gradient versus up-gradient moni-
toring situation. Up-gradient ground
water is that moving towards a disposal
site; down-gradient ground water is that
moving away from a disposal site. For the
methanol SPE eluates, three of the
ground-water samples were found to
have very low UV background (0.03 a.u.),
three had slightly higher levels (0.04 to
0.06 a.u.), and one had a distinctly higher
level (0.25 a.u.). All of the ground-water
hexane SPE eluates were indistinguish-
able from those produced with reagent
water. For the methanol eluates, the
simulated down-gradient versus up-
gradient UV spectra gave the following
average absorbance differences for the
seven ground waters. 0.012 ±0.005 a.u.
at 230 nm, 0.009 ± 0.005 a.u. at 250
nm, and 0.008± 0.004 a.u. at 280 nm.
Thus, the absorbance difference required
to conclude that a given down-gradient
water sample contains higher levels of
UV absorbing material than a correspond-
ing up-gradient sample varies with the
wavelength used. For wavelengths
between 230 and 300 nm, about 0.02
a.u. difference is indicated by these
limited results. Similarly, between 300
and 400 nm, about 0.01 a.u. difference
would be required.
Spike Recoveries From
Reagent Water
Percent recoveries from reagent water
for five Method 8610 analytes spiked at
two levels are shown in Table 1. In these
experiments, each analyte was spiked
individually into 100 mL of water using
50 juL of a methanolic spiking standard.
After mixing, the 100-mL sample was
extracted using the SPE cartridge, and
UV analysis was conducted on the result-
ing methanol and hexane eluates. Cali-
bration standards were prepared by
spiking methanol and hexane eluates of
reagent water blanks with the same
spiking mixture as for samples (i.e., 100
percent recovery standards). The results
shown in Table 1 support three con-
clusions:
• Overall recovery efficiency of the
SPE extraction is adequate for es-
sentially all purposes.
• Separate recovery of apolar analytes
in the hexane eluate is clearly not
feasible when the first elution sol-
vent is as strongly eluting as
methanol.
• There is apparently more variability
due to uncontrolled aspects of the
sequential elution than there is in
the SPE cartridge extraction effici-
ency (i.e., precision of the methanol
or hexane eluates versus that of the
total by individual replicate).
Spike Recoveries From
Composite Ground Water
To provide a sufficiently large homo-
geneous sample for testing ground-water
spike recoveries, four of the individual
ground-water samples were combined.
The samples selected for compositing
were the four with the highest SPE re-
coveries of background UV-adsorbing
material so that a maximum challenge to
Methods 3560 and 8610 would be ob-
tained. Sample compositing produced a
substantial change in the morphology of
the particulate matter. Whereas, the
particulate matter in the individual
ground-water samples settled rapidly,
particulate matter in the composite
ground water would not fully settle even
on overnight standing. Also, in contrast
to the individual ground-water samples,
a small but variable amount of the par-
ticulate matter was recovered in the
-------
TABLE 1. Recoveries of Selected Analytes From Reagent Water
Percent Recovery
Compound
Amount -
Spiked.18'.
M9/L
Methanol SPE"" Eluate
Hexane SPE Eluate
Sum of SPE Eluates
Replicate
Mean
±sdc>
Replicate
Mean
±SD
Replicate
Mean
±SD
Polar Aromatic Compounds
3-3'-Dichlorobenzidine
2,4-Dimethylphenol
1,2-Diphenylhydrazine1'"
Nonpolar Aromatic Compounds
23
116
150
750
15
75
88
95
94
93
100
98
85
88
94
89
100
87
85
95
86±1.7
93±4
100 96±4
87 90±3
100 100±0
98 94±6
0
0
3
2
10
5
0
0
3
3
10
10
0
0
0
4
0
9
2±2
3±1
7±6
8±3
88
95
97
95
110
103
85 85 86±1.7
88 95 93±4
97
92
110
97
100 98±1.7
91 93±2.1
110 106±6
107 102±5
1 , 4 -Dichlorobenzene
Methoxychlor
600
3000
30
150
71
68
64
53
79
67
72
55
86
70
55
51
79±8
68±1.S
64±9
53±2
19
10
64
13
14
12
28
39
10
10
42
33
14±5
11±1
45±18
28±14
90
78
128
66
93
79
100
94
96 93±3
80 79+1
97 108±17
84 81+14
la> The low and high spike levels were based on the Kmgx absorbtivity for each analyte so that the lower level was three to five times an estimated
detection limit; the higher level was five times the lower level.
SO: standard deviation.
This analyte is unstable under aqueous acidic conditions; the actual species measured is probably benzidine.
methanol SPE eluates from the composite
ground-water sample. The UV spectra
obtained from the methanol SPE eluates
of the spiked and nonspiked composite
ground water all supported the same
conclusion: variable amounts of matrix
UV-absorbing material were recovered in
the methanol SPE eluates which super-
imposed an absorbance baseline to shift
the observed absorbances to either higher
or lower values than the correct ones. In
some cases, baselines for nonspiked
samples, shifted with a positive bias,
were actually above a low-spike UV
maximum shifted with a negative bias
giving a negative net absorbance and,
therefore, a negative recovery. An
example of this effect is shown for
dichlorobenzene in Figure 1, along with
the corresponding result for reagent
water.
Since the spectral variations of
methanol SPE eluates of nonspiked
composite ground-water samples all had
similar wavelength/intensity profiles, a
baseline correction was made based on a
visually approximated baseline bias for
each spectrum of spiked replicates to
give a more accurate representation of
the actual spike recoveries obtained. The
results using these pseudo-baselines are
shown in Table 2. While the methanol
eluate recoveries shown in Table 2 for
the composite ground water compare
favorably with those of Table 1 for reagent
water, the use of pseudo-baselines pre-
vents any detailed comparisons between
these methanol eluate data sets. The
hexane SPE eluates from the composite
ground water did not produce a similar
0.72
0.70
0.08
0.06
0.04
0.02
Reagent Water
a: nonspike
b: low spike
c: high spike
nonreproducible baseline bias. However,
the recovery values obtained do show
some significant differences between
reagent water and composite ground
water for methoxychlor. Averages of 44
and 28 percent were recovered in the
low and high spike samples, respectively,
for reagent water, compared to 5 and 6
Composite Ground Water
a: nonspike with
extreme positive bias
b: low spike with
b: slight negative bias
c: high spike with
c: zero bias
220
250 300 350 400 220 250 300 350 400
Wavelength, nm Wavelength, nm
Figure 1. UV spectra of methanol eluates from reagent water and composite ground water.
3
-------
TABLE 2. Recoveries of Selected Analytes From Composite Ground Water
Compound
Percent Recovery
Methanol SPE"1' Eluate Hexane SPE Eluate
Amount
Kpii, *,!<•> Replicate MMn Replicate Mt>an
' pg/L 1 2 3 ±Sdc> 1 2 3 ±SD
Sum of SPE Eluates
Replicate Jtfpfv?
/ 2 3 ±SD
Polar Aromatic Compounds
3-3'-Dichlorobenzidine
2,4-Dimethylphenol
1,2-Diphenylhydrazine1'11
Nonpolar Aromatic Compounds
1' ,4-Dichlorobenzene
Methoxychlor
15
75
150
750
15
75
600
3000
30
150
81
87
62"» 58
74?' 85
67±12
82±7
1lde> 10^el 115*"' 110±5
115"" 91lel 121 109±16
NO
74
1431"1
74
17l
74
'el
160
74±0
86fel 86*" 71"" 81±9
72"" ad* 70"" 74±5
36«" 36fe> 32"" 35±2
52"" 47!el 47"" 49±3
0
0
0
0
25
19
14
14
7
6
0
0
0
2
25
21
14
12
0
7
0 —
0 1±1
37
19
14
13
7
4
29±7
20±1
14±0
13±1
5±4
6±2
81
87
110
115
25
93
100
86
43
58
62
74
105
93
168
95
1OO
92
36
54
58
85
115
121
208
93
85
83
39
51
67±12
82±7
110±5
110±15
13O±90
94±1
95±9
87±4
39±4
54±4
ta> See footnote (a). Table 1; the levels for 3,3'-dichlorobenzidine were adjusted downward for these experiments based on the reagent water results.
SO: standard deviation.
>d> This analyte is unstable under aqueous acidic conditions; the actual species measured is probably benzidine.
fe' Value obtained after making an approximate correction for baseline bias (see text).
percent recovered in the low and high
spike samples for composite ground
water. In contrast, there was no hexane
eluate recovery difference between
reagent water and ground water for the
other nonpolar analyte, 1,4-dichloroben-
zene, which averaged between 11 and
14 percent for all four data sets.
While the actual chemical mechanism
causing the wide variability in recovery of
the native UV absorbing material from
composite ground-water samples cannot
be identified from the data presently
available, the cause of the variability can
definitely be attributed to the ground
water itself rather than laboratory proce-
dural variations. The reasons are:
• The same lot of reverse-phase car-
tridges, elution solvents and spiking
media were used for all experiments.
• All extracts were produced by the
same laboratory technician using the
same apparatus and identical labora-
tory procedures in the same labora-
tory location.
• Sets of extracts were generated
sequentially with no time delay
between the reagent water set and
the ground-water set.
• Reagent water control blanks pro-
duced along with the ground-water
samples were uniformly identical to
those obtained with the reagent
water set.
Finely divided particulate material in the
composite ground water is the mostly
likely cause of the variable UV baseline
bias. Although the extraction cartridges
are equipped with a 20-jum polyethylene
frit at the top and bottom, the suspended
particulate matter was not completely
removed as some of the particulate matter
color was visible on the column bed. For
the composite ground water, the methanol
elution removed some of this particulate
matter from the cartridge while none
was recovered in methanol eluates of the
non-composited ground waters. The re-
covery of particulate matter into the
methanol eluate occurred for every non-
spiked and spiked composite ground-
water sample and was estimated to vary
in amount over a range of a factor of
three with the total amount present
averaging about 0.5 mg. Of the amount
of suspended particulate matter in the
100-mL aqueous sample, less than one
percent is estimated to have been re-
covered in the methanol eluates.
While quantitative conclusions are not
possible from the composite ground-
water data, the results do suggest that
removing particulate matter before ex-
traction might provide recovery results
approaching those obtained for reagent
water. Further experimentation on the
composite ground water used was not
possible, since it was consumed in the
spiking experiments. However, it was
possible to further examine the composite
ground-water methanol eluates for
evidence of the possible mechanisms by
which the inorganic particulate matter
could have caused the matrix recovery
variability. Centrifugation and filtration
(0.5 micron) of the three nonspiked
methanol eluates with the highest matrix
background recoveries were tested to
determine if suspended particulate matter
could be the cause of the variability
observed. Composited methanol eluates
of reagent water method blanks were
used as the UV reference sample. Each
of these three high background composite
water methanol eluates was examined
against the composite reference three
times: (1) after settling for about 1 hour,
(2) after centrifugation at 2300 rpm for
10 minutes, and (3) after filtration (0.5
micron). Centrifugation and filtration,
versus the gravity-settled trial, had no
effect on the UV trace for any of the three
nonspiked composite ground-water
methanol eluates. Therefore, particulate
matter in the UV measurement beam
-------
was probably not the cause of the re-
covery variability. To determine if the
variability was caused by dissolved car-
bonate aruon, these same methanol
eluates were acidified with HCI and then
purged with nitrogen to remove CO2.
However, since all three samples became
yellow and more UV-opaque, this experi-
ment provided no information.
Evaluation of Method 8610
Sensitivity
Because of the matrix UV recovery
variability observed for the composite
ground water, only the reagent water
results are useful for predicting Method
Detection Limits in the 230 to 400 nm
range. If one assumes that the cause of
the ground-water matrix problem could
be identified and the problem rectified to
provide reproducibilities comparable to
those found for reagent water, an ab-
sorbance difference (down-gradient
versus up-gradient) of 0.02 a.u. can be
taken as the decision limit threshold. This
decision level for a positive response is
four times higher than the proposed
Method 8610 decision limit of 0.005 a.u.
Using the 0.02 a.u. threshold value,
theoretical detection limits for Method
8610 analytes can be calculated. Addi-
tional assumptions made for these cal-
culations are: (1) the use of a 100-mL
ground-water sample, (2) an SPE eluent
volume of 5.0 mL of methanol or hexane,
(3) 100 percent recovery in a single SPE
eluent, and (4) analysis without further
concentration. Of the Method 8610
analytes for which estimated detection
limits are thus calculated, only 28 percent
of the 129 analytes have 10 /ug/L or
lower detection limits and 68 percent are
estimated to have 50 M9/L °r lower
detection limits.
Conclusions
The results of this work support several
conclusions regarding the performance of
EPA Methods 3560 and 8610:
• Method 8610, in its present form,
was not suitable for analyzing the
tested composite ground-water
sample. The tested water contained
fine sediment particles which par-
tially moved through the extraction
cartridge and possibly interfered with
the UV analysis. However, it was
impossible to determine how the
particles caused the poor reproduci-
bility or even that it was, indeed,
caused by some phenomenon as-
sociated with the particles.
• For reagent water, the simulated
down-gradient versus up-gradient
results suggest that a positive
response of 0.02 a.u. in the UV range
230 to 400 nm is sufficient to indicate
down-gradient contamination. This
result is four times higher than the
proposed Method 8610 decision limit
of 0.005 a.u.
• Using extinction coefficients reported
in Parti of this program, 0.02 a.u. as
the positive response decision level,
a 20-fold Method 3560 concentration
factor, and recovery of 100 percent
of an analyte in either the methanol
or hexane eluent, it is estimated that
on ly 28 percent of the Method 8610
analytes have a detection limit equal
to or less than 10 ng/L and 68
percent have a detection limit equal
to or less than 50 M9/L.
• Method 3560 was successful in re-
covering 3,3'-dichlorobenzidine,
2,4-dimethylphenol, 1,4-dichloro-
benzene, and methoxychlor from
reagent water with recoveries
ranging from 79 to 108 percent. The
first two analytes were almost com-
pletely eluted with methanol and the
last two analytes were partially eluted
in both methanol and hexane. The
data obtained from ground-water
spiking appear to support the same
conclusion, but the non-reproducibi-
lity of the background UV material
does not allow quantitative demon-
stration of that conclusion.
Recommendations
It is unlikely that any amount of method
development will enable Methods 3560
and 8610 to adequately address most of
the analytes assigned to them by the SW-
846 Heirarchical Analytical Protocol (HAP)
scheme. These methods may, however,
be adequate for a significant enough
subset to justify further method develop-
ment. If so, then the following recom-
mendations are offered for further method
development:
• The effect on accuracy and precision
of including a step in Method 8610
for removing particulate, either by
filtration or centrifugation, before UV
determination should be investigated
to determine whether or not Method
3560/8610 can actually be applied
to heterogeneous samples.
• Since the results of the previous
study showed that polarity class
separation with methanol and hexane
elution solvents was probably not
feasible, effort should be directed to
identify a single elution solvent, such
as ethanol, isopropanol or a mixed
system, which will give high recovery
for all analytes in a single elution.
• Further work to improve the present
20-fold concentration factor and
thereby increase the method sen-
sitivity will be required to obtain a
usable method. For example, use of
1.0 mL UV cells, 1.5 mL of elution
solvent, and a 150-mL sample would
provide a 100-fold concentration
factor.
• Other commercially available reverse
phase sorbents should be evaluated
to determine whether a particular
type provides superior recovery and
precision of the method analytes.
• If the cause of reproducibility prob-
lems for ground water identified in
this work can be identified and elimi-
nated, an automated Method 8610
approach using an HPLC auto-
sampler, UV detectors, and auto-
mated chromatography data system
quantification should be developed
since improved signal-to-noise and,
therefore, lower detection limits
should be possible than those
demonstrated in this work using a
double beam spectrophotometer.
-------
------- |