-------
72
[Blank Page]
-------
73
, MR. TELLIARD: Carrying on with the same theme, this year, we have
been working diligently on getting out a regulation on pesticide manufacturers, and part
of the work was involved with attempting solid phase extraction, basically looking at
Method 608.
Merlin Bicking from Twin Cities is going to talk about the work that they
have done looking at the application of pesticides analysis.
I have always kind of thought it would be neat to be a chemist and have a
first name like Merlin.
-------
74
MR. BICKING: All right. Like Craig, I would like to acknowledge Greg
Junk's contributions to solid phase extraction. I was a graduate student at Iowa State in
the mid to late seventies, but when much of his work was going on, and I certainly
remember Greg's contributions.
This presentation will discuss a modification to EPA Method 608, which is
organochlorine pesticides and PCBs.
[SLIDE 2] We had several primary objectives. The first one was to
evaluate EMPORE disks as a replacement for liquid-liquid extraction. The emphasis
was on wastewater samples in this particular method. We were interested in the
recovery of spiked samples. We were not analyzing native contamination levels in this
particular study. We were simply looking at a recovery of spiked samples. Finally,
because of the concerns that have been mentioned earlier, we are working with the 90
mm disks for samples which probably are going to contain particulates.
[SLIDE 3] There have been concerns about whether or not EMPORE
disks could be used for wastewater samples. Certainly, disk plugging from high
paniculate samples is one of the primary concerns. If we can't get the sample through
the disk, we are not going to do an analysis. The other issue with high paniculate
samples is adsorption of the analytes on the particulates. Most of these analytes have
very low water solubility, and adsorption on a paniculate is quite likely. The next issue
is the level of interferences compared to the standard methods.
Then, if we can solve all of those three problems, the big question is: "does
it work?" If it does work, is the precision comparable to what we get with, say, a
methylene chloride extraction (or, at least, not any worse)?
[SLIDE 4] We split up the Method 608 analytes into four groups, a
pesticide "A" and "B"mix, simply for convenience. We chose two PCBs, the lower level
of chlorination, Aroclor 1016, and then the upper end, Aroclor 1260. Toxaphene was
evaluated separately. We chose four peaks in each of those patterns. We used
decachloro-biphenyl as a surrogate. In this case, while we are trying to follow Method
608 as closely as possible, we also considered the CLP protocols, which employed a
surrogate, and wanted to adopt some of those procedures as well, or at least evaluate
-------
75
them at the same time.
[SLIDE 5] This study was divided, like Gaul, into three parts. The initial
evaluation simply was a qualitative evaluation of the matrices we were working with: do
they go through the filters, through the disks, and what are the levels of interferences?
The full method validation consisted of five replicates at approximately 100
times the MDL and one control or unspiked sample. We used three different
procedures, as you will see in the following tables; the Method 608 procedure which
involved methylene chloride extraction in a separatory funnel, and the two EMPORE
procedures. The "EMPORE-Baker" and the "EMPORE-Varian" refer to the
manufacturers of the C18 particles. We evaluated both sets of disks, and in this paper,
we are going to discuss those two disks separately.
Four matrices were evaluated: POTW, pulp and paper effluent, a pesticide
manufacturing effluent (from a facility that does not manufacture organochlorine
pesticides) and, finally, a petroleum facility effluent. We have completed this phase for
the first three matrices. We are just beginning the petroleum matrix and do not have
any data for that at this time. The other thing I should note is our pulp and paper
sample was as bad as the others you have heard about today, although ours looked like a
vanilla milkshake rather than a chocolate milkshake. Finally, the last phase is an MDL
study with the replicates. We have not completed this yet and won't be talking about
that today.
[SLIDE 6] For those of you who may not be intimately familiar with the
EMPORE disks and method 608,1 thought I would provide a little more detail on the
actual performance of this method. All of our spikes were into a 1 liter glass container.
This procedure was used for all three methods. In our case, we allowed the sample,
after spiking and shaking, to equilibrate for one to two hours. This was a suggestion
from EPA. It certainly represents a worst case scenario for studying the adsorption on
particulates problem.
The liquid-liquid extraction approach is conventional separatory funnel
extraction, K-D techniques for concentration, solvent exchange to hexane, and ECD
quantification.
-------
76
[SLIDE 7] The EMPORE procedure involves a disk preparation step.
You assemble the apparatus. (We used the disk and glass fiber filters from Whatman.)
You condition the disk, essentially cleaning it with the elution solvent, which is
methylene chloride in this case, allow it to soak, and then pull that solvent through.
Then the disk is conditioned with methanol, by soaking.
At this point, it is very important that you not let the disk dry out. Leave
some methanol on the surface. We believe this affects the wetting process, and allows
the water to go through the disk.
The sample extraction or filtration, depending on your terminology,
required about 20 minutes to filter a 1 liter sample. The range was probably 10 to 30
minutes for these matrices. The pulp and paper sample was particularly difficult in the
early work, and we discovered, much to our surprise, that if we adjusted the pH to about
2, it would filter much more readily. At neutral pH, I think we set the new record for
the 90 mm disk. It would have gone past Craig's three and a half day limit, but at a pH
2, it filtered quite nicely, and all of our data for pulp and paper will be at that lower pH.
[SLIDE 8] In the final step, we eluted with three 15 mL portions of
methylene chloride, allowing the portions to soak 3 minutes each time, and then pulling
them through the disk. The rest of the process is just like the conventional 608
procedure: solvent exchange and GC/ECD.
[SLIDE 9] Now, it is time to talk about the data. We have talked a lot
about what we think the EMPORE disk can do. Now, we can have a chance to see what
they actually can do.
[SLIDES 10-12] This is the kind of slide that will make your eyes glaze
over when you see it. I really don't intend you to read it. It is just up there to show that
we have done the work. These are real data points, and I will have another graph which
is a little easier to comprehend.
We have four or five data points for each analyte. We have looked at the
individual replicates, performed statistical tests to reject the outliers, and then looked at
the final sets. We have recovery, or accuracy, and precision data, then, for all three
experiments (liquid-liquid extraction, Empore-Baker, and Empore-Varian).
-------
77
[SLIDE 13] We feel it is much easier to evaluate the two techniques using
a graphical approach. The X-axis on this plot is the percent recovery for each analyte
for the liquid-liquid extraction method (Method 608). That is plotted against the
recovery for that analyte with the two EMPORE disks. The "square" is for the Baker
disksand the "pluses" are for the Varian disks.
If the recoveries are equal for each of those analytes using the two
techniques, all points should fall on that 45-degree line. You can see that there is a
cluster of points around that 45-degree line, weighted a little bit below it. That means
that the EMPORE recoveries are a little bit lower than the corresponding liquid-liquid
recoveries, and there are obviously some scattered points. However, it looks like the
points are clustering near the 45-degree line and are within about 10 percent of the
liquid-liquid recoveries. We are very encouraged by this result. For a large number of
data points, we see a good match in recoveries.
[SLIDE 14] This is a plot for the pulp and paper matrix. You can see
there are many points in the lower right part of the plot, where the EMPORE disk
recoveries are lower than the liquid-liquid recoveries. Most of the problems occur with
the PCBs and toxaphene. We are not sure why this happens, but 3M has indicated that
the PCBs are more difficult than the rest of the conventional organochlorines.
[SLIDE 15] The next graph gives the same type of presentation for the
pesticide effluent. Again, we see a pretty good cluster around the 45-degree line, but a
little bit lower for the EMPORE experiments.
One of the other important things to note here is that you do see a general
trend along the 45-degree line. In other words, if you have lower recoveries for the
Method 608 procedure, you also get lower recoveries for the EMPORE procedures.
This is telling us that whatever the reasons are for those recovery losses, they are the
same for either procedure. In other words, there have been concerns that we are going
to lose analytes on particulates with the EMPORE approach that we would normally
extract with the liquid-liquid approach. The fact that we are seeing a general trend
rather than a complete scatter tells me that that is not the case. Whatever the sources of
the recovery losses are, they are independent of the EMPORE step itself. This
-------
78
conclusion is important in evaluating the performance of the EMPORE disks.
[SLIDE 16] There were so many data points that we wanted to take a
more formal approach to evaluating it, and we really thought statistics offered a
convenient way to do that.
[SLIDE 17] This slide summarizes two statistical tests for the data. The
first test, a t-test, is shown in the left two columns. This test compares the means, or the
average recoveries, for liquid-liquid and the two different EMPOREs. This is really
answering the question: "are the average recoveries different for the two data sets or
being compared?" A "NO" means there is no difference, and that is good in this case. A
"YES"means there is a statistical difference between the two means. For the POTW
matrix, you see there are more NOs than YESs, and that is good. In other words, there
are no statistical differences for most of the data sets.
The right two columns summarize the results from an F-test on the
variance. This test is answering the question: "is there a difference in the variance or,
ultimately, the precision, for these two methods?" You can see that in most of those
cases, there are NOs in the table and that means that the variances are not statistically
different. In other words, we can expect comparable precision for the two techniques.
[SLIDE 18] This table gives a summary in more absolute terms. Where
there is a statistical difference, what is the difference in those recoveries? For the first
two columns, the numbers are the average recovery differences. If you see a minus sign,
it means that the EMPORE recovery is lower, and statistically different from liquid-
liquid recovery. Most of the numbers are in the 10 to 20 percent range lower. So, even
when there is a statistical difference, the difference is only 10 to 20 percent for the
EMPOREs. There are a few cases where the EMPORE recoveries are actually higher.
The third and fourth columns indicate where there was a significant
difference in the variances, which of the variances were smaller. In some cases, the
EMPORE procedures had a statistically significant smaller variance (or precision).
I think this result is encouraging news. It tells us that, from a precision
standpoint, the EMPORE disks are going to provide results which are comparable to the
standard procedure.
-------
79
[SLIDE 19] Similar results were observed for the pulp and paper matrix,
except we see more YESs here. For some reason, we had some low recoveries in the
EMPORE procedures. We are not sure why. This is the worst sample we have worked
with, although most of the variances were still comparable between the procedures. We
seem to have problems with the pesticide "A" and "B" mixes which are the top sets in the
table. We are not sure why at this point.
[SLIDE 20] The next slide is in a similar format. Where there are
significant differences, the magnitudes are listed in the table. These differences tend to
be a little larger, 20 to 30 percent, and for the top group, the liquid-liquid extraction
procedure is giving better precision. For the other sets, the variance test indicates mixed
results, with either technique showing a smaller varience, depending on the analyte.
[SLIDE 21] This table shows results for the pesticide effluent. There was
a split between whether or not there was a significant difference in the average
recoveries, although there were very few differences in the variances.
[SLIDE 22] We see a few larger differences in recovery, but many of them
are pretty close. Again, recognize taht the blank spots are cases where there are no
statistically significant differences between the data sets, and all of those recoveries are
usually within 10 percent of each other. Again, on the right two columns where there
are blanks, that means that the variances are not statistically different.
[SLIDE 23] A final comparison on solvent usage is provided in this table.
We reduced total volume from over 300 to under 140 mL. That reduces our purchase
cost by more than a factor of 2. It certainly helps with disposal costs also. So,
EMPORE is certainly doing one of the things that it is claimed to do, and that is
significantly reduce our solvent use in the laboratory.
[SLIDE 24] In conclusion, we feel we have demonstrated that we are
going to get comparable results between the two techniques. The recoveries are usually
within 15 percent, and even where there are statistically significant differences, many of
the recovery differences are less than 20 percent. Precision is comparable in most cases
with the exception of the pulp and paper matrix.
In general, the extracts from EMPORE disks have fewer interferences. We
-------
80
did not evaluate any cleanup steps for these samples. That is one important aspect to
remember. We found no substantial differences in interferences but somewhat lower
levels of interferences using the EMPORE disks. And I should note that all of the
recovery values we have are corrected for levels in the unspiked sample, so they are all
blank corrected. Certainly, solvent use is reduced, as we have demonstrated.
We have had no problems with plugging from the 90 mm disks.
Extractions are rapid, with the exception of that first set of pulp and paper samples at
pH 7 to 8.
We are also looking at other procedures which will further reduce solvent
use. For example, supercritical fluid extraction (SFE) can be used to remove analytes
from the disks, and that will eliminate much of the solvents, and take the total solvent
usage down to less than 50 mL total.
[SLIDE 25] I would like to leave you with a parting comment (and
warning) about statistics. Certainly, we have generated a lot of statistical information
here. I don't want to go overemphasize in our use of statistics, but I think they do
provide a valuable tool for us in evaluating these two procedures.
[SLIDE 26] I would like to acknowledge Craig and George at 3M, who
have been working with us closely on the experimental design and the operational
considerations. Bill and Harry have been also heavily involved in the planning stages
and the early execution and have had substantial input into the experimental design.
I would be happy to answer any questions.
-------
81
QUESTION AND ANSWER SESSION
MR. STANKO: George Stanko, Shell Development Company.
In your studies in the pulp and paper industry, your recoveries were lower.
Was there any effort to look in what came through the filter to see of the actual analytes
were in there, and is there a possibility that there were other polar solvents like DMSO
or acetone that may have accounted for this?
MR. BICKING: No, we have not looked at that issue. That is certainly a
good suggestion. Our primary objective was to see how total recoveries compared
between the techniques. I think now we have to go back and look at where recoveries
were different and find out why.
Another issue is that the type of particulates in the pulp and paper sample
are fibrous. I have concerns about whether or not there is some swelling or shrinking
that goes on when that organic solvent contacts the particulates in the elution step. It
could be that analytes are on the particulates or on the fibers, and when you add
methylene chloride, something happens that traps them. That is another possibility.
MR. PERTUIT: Bob Pertuit, PPG Industries.
When you are running garbage samples through these disks, is the water
coming out of the disk still turbid, showing that you are passing some ultrafme
particulates through the disks?
MR. BICKING: In almost all cases, the filtrate is clear. In most cases, we
remove any color with the disks also. So, we don't see any turbidity.
MR. PERTUIT: I have had some experience working with these
ultrainsoluble compounds like PCBs, and they are going to find anyplace besides the
water to reside. They are going to get out of solution.
If they attach to particulates, they will pass through a moderately porous
system on the particulates, and you won't be able to find them. I would suggest that you
look on the glass in the bottle and in the filtrate for your PCBs to see if they either
stayed in the bottle on the walls or have gone through with the particulates.
MR. BICKING: That is a good comment. I didn't mention that we rinsed
the sample container with the methylene chloride that we used for elution, and we also
-------
82
rinsed down the sidewalls of the glass reservoir. All of the glass contact surfaces have
seen methylene chloride at least once, and in the case of the reservoir, usually two or
three times. So, I hope that isn't an issue. Now, if an analyte adsorbs to a particle
which is smaller than about 0.1 microns, then you are right. Then the analyte is going to
go right through.
In looking at the extraction from particulates issue, if you consider the
amount of time that the solvent is in contact with the paniculate, it is actually longer
with the EMPORE procedure than with conventional separatory funnel extractions.
With the particulates, it is soaking, in our case, three times for 3 minutes each. So, it
has a total of 9 minutes of contact time, (direct contact) between organic solvent and
particulates. That is substantially higher than you are going to get from a separatory
funnel shake which is, at best, maybe 2 minutes for each step.
MR. PERTUIT: Thank you, sir.
MR. WESTON: Charlie Weston from ETC.
What significance level did you use in your statistical comparisons?
MR. BICKING: 95 percent. Alpha equals .05.
MR. WESTON: Thank you.
MR. LAW: Peter Law with Tighe & Bond Laboratory in Westfield, Mass.
Could you clarify the initial pHs for your sample? Was the pulp and paper
the only one at 2, everything else at neutral?
MR. BICKING: We checked the pH of all the samples, and they were
adjusted as specified in the method (7 to 9), to make sure they were in the range
specified by the method.
MR. LAW: But the pulp and paper was at 2?
MR. BICKING: That sample was initially at a higher pH, but we adjusted
it to a pH of 2 for all of the studies we reported here.
-------
83
MR. TELLIARD: Thanks, Merlin.
All right, it is break time. For all of those who are still napping, would you
please go out, get your coffee, your strawberry, and get back in here in ten minutes so
that we can continue on. Thank you very much.
(WHEREUPON, a brief recess was taken.)
-------
84
a
o
H
fc
O
^^j
F"1^
H
C/5
W
C/3
HH
^
^^—•^
O
1 1
>
'o
o
Q
c
• »—H
UJ
i-r-4
^
C
fa
^
fe§^J
^wa
H S
O
O
g
S3
(D
43
VH
(D
-4— >
C/3
O
OS
bO S §
c f 2
PQ
OH
i>
CO
cd
PH
O
O
«
d)
Q
(D
V^
PH
-------
85
S
j
CO
CO
^
J^
5
«
BQ
O
O
• f>^
ts
» j
X
o
o
(D ^j
oS
3 ^p
13 g
W CO
G
• I-H
OS
C
o
o
CO
Ix
E
OS
CO
VH
^2
s^^
IH
•*->
»— H
1C
J^
^
t3 ^3
^2 .$>
-_^ ^^^V
LQ ^^
^>
w a
-------
86
r CO S
feW £
°^ -i
ca fe
J|H ^H C
CO ^ •-
^ ^P
a M • s
Stf 1
^ [d (D
^^ ^^^ VH
&?5
»— >i ^^H fli
hJ ^ ^
On E
*A CxJ 5
QQ F*H cd
3 H
^ V5 «
co ^3 ^
7^ t^ P
S t> 2
n^ ^^ • ^^
P§ !
^ fe -a
Oco S
U^^ ^ft
^^ s
^^*» v-^K "
co ^2 2
ss *
W ^ co
^•^ yu (C (D
^2 "w> ts
^ o ^ S
r—M ^^ »— < £•*
W £ 0,0
* \ ^^^ * i >
H^ ^^ s/< cj
B O'i |£
S W Q«
s
•
.§
O
O
D
S
^
O
i— H
^H
C
• FH
• F-H
-4— >
1— <
s
rv
3
05
o
• f-H
O
CO
-^— >
^
C
<*H
O
C
o
• i— H
&
o
CO
CO
O
C
(D
J-H
CD
-4—>
C
•FH
«^H
O
13
(D
V-i
O
CO
ctf
O
•V
-------
87
tt
i
w
u
f ~\
W
H
O
O
S
WQ
Q Q U
PQ
CQ
W
Q
hH
X
w
Q
^ ">
w
H
i-J
a
O
o
r—>
w
VOVOVOVOOOOO
OOOO(N(N(N(NO
oooD
uuuuuuoT
OOO
a
-------
88
c
W)
(D
j^
'S
erferences
o
3
S^
1i ^
cd ^
(-H cd
S: c-2
a o «
Si*
H tJ >,
S S T3
P >
^ ^'^
5 s.s
fc-g!
w ex
Q
S
X
w
O
« ^
^
o
VALUATI"
w
Q
O
[V]
1
"H
o
UH
o.
&
^->
ed
CO
s
cd
O
• PH
'SL
D
VH
MM
|
CO
(unspiked)
*o
VH
•*->
c
0
o
c
.^•^
bO 3
VB^ ^^^^
cedures:
o
oo
VO «_ C
<3 .2
2 03 Cd
-S PQ >
^H »
-4— >
• f-H
• ^H
O
03
PU
e
3
JD
'o
M
+->
D
OH
O OH
CO
-------
89
O
w
u
o
VC
i
CO
OO
o
-o
o
J «>
D
(U
1 +-*
h-J OS
C/3
Q
5
a
CO
D
Q
-G
O
C
(D
JC
+->
(D
E
dVL/
">
£
O
P
CJ ~
3 *
* 1
W -a
Q u
g £
h«* f
» 5
X
W
-------
90
IM
(D
CO
CO
3
"5b
C
ctf
(D
c
C 3
(D O
>• H
o
S
OH
^
-------
91
g
fl
0
c
a
CJ
c
U
"c
I-H W
^H rtj
W *3
^k^0 MM
Uu JS
O cd
CO
CO
CO
c
UH
'S
a a
o-S
W)
^ 3
0 2
(D
P CO
^
*-" ^
o O
c
(U
j>
'o
00
c
o
03
O
CX
c
S
o
O
OO
O
PQ
Q
e
x
E
^-^ ^^
en -§
^ 2
c> 5
ta> ™
•S^
S £
^ O
Q t<
u S
W 60
U-S
O S
oo
I
CO
-------
92
c
• •—«
03
Ui
03
5S .
^j
-
w
e
-------
[SLIDE 10]
93
SUMMARY OF SPIKE RECOVERIES FOR
POTW MATRIX
Recovery . oercent
Analyte Spike Level,
(s.d.)
G-BHC
Heptachlor
Aldrin
Heptachlor Epoxide
Endosulfan I
Dieldrin
Endosulfan II
Endrin Aldehyde
DDT
DCBP (Surrogate)
A-BHC
B-BHC
D-BHC
A-Chlordane
DDE
Endrin
DDD
Endosulfan Sulfate
Endrin Ketone
DCBP (Surrogate)
Toxaphene-1
Toxaphene-2
Toxaphene-3
Toxaphene-4
DCBP (Surrogate)
Aroclor 1016-1
Aroclor 1016-2
Aroclor 1016-3
Aroclor 1016-4
Aroclor 1260-1
Aroclor 1260-2
Aroclor 1260-3
Aroclor 1260-4
DCBP (Surrogate)
1
1
1
1
1
2
2
2
2
3
1
1
1
2
2
2
2
2
2
3
24
24
24
24
3
4
4
4
4
4
4
4
4
3
n
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
4
5
5
4
5
LLE
Ave. (s.d.)
86.3 (8.9)
81.6 (8.8)
63.6 (6.6)
86.7 (6.1)
96.0 (6.9)
93.1 (6.4)
79.8 (5.2)
105.0 (5.8)
82.9 (9.3)
97.2 (4.0)
83.6 (4.3)
104.6 (9.0)
113.8 (7.3)
69.9 (6.7)
83.1 (6.0)
130.9 (7.2)
103.3(15.4)
79.3 (6.5)
78.8 (6.2)
106.9 (4.4)
129.2 (8.6)
126.2(16.2)
110.7(11.1)
102.3(10.1)
80.2 (8.7)
78.4 (3.0)
93.8 (8.3)
129.0 (3.9)
101.1 (6.0)
81.5 (1.5)
97.5 (6.1)
88.5(11.3)
95.4 (3.2)
84.0 (5.0)
Empore-Baker
n Ave. (s.d.)
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
84.3 (2.6)
72.6 (8.9)
55.8 (5.9)
78.9 (4.7)
87.8 (4.9)
82.5 (5.0)
74.3 (3.0)
81.1 (5.8)
80.6 (4.3)
86.1 (4.2)
82.9 (3.6)
103.1 (6.8)
116.3 (7.6)
85.9 (5.7)
81.8 (5.3)
131.4 (6.1)
112.5(12.5)
85.0 (4.0)
82.6 (5.0)
105.8 (5.0)
113.1 (5.9)
116.5(14.1)
103.4 (8.1)
95.2 (7.8)
83.3 (1.5)
52.6 (4.2)
79.6 (3.6)
76.8 (4.8)
90.5 (7.9)
63.8(12.8)
65.0 (7.0)
74.5(14.9)
84.8(22.7)
65.5 (3.5)
Empore-Varian
n Ave.
5
5
5
5
5
5
5
4
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
4
5
5
5
5
5
5
5
5
87.0 (5.2)
61.4(11.0)
56.0(10.0)
76.8(10.1)
77.6 (9.5)
78.7(10.2)
74.7(11.4)
85.5(14.3)
80.4(10.0)
80.1 (7.8)
87.9 (2.9)
109.9 (5.0)
123.0 (3.6)
94.5 (3.2)
88.5 (4.3)
140.6 (3.8)
111.6 (8.0)
89.0 (3.3)
86.0 (3.4)
109.2 (9.2)
102.3(18.6)
114.5(12.3)
100.8(11.7)
94.8 (8.0)
83.7 (2.9)
15.5 (6.2)
80.9 (3.8)
97.0(15.6)
82.9(12.3)
58.2 (5.1)
76.2 (2.5)
81.6 (6.0)
79.3 (7.6)
75.8 (2.7)
-------
94
[SLIDE 11]
Recoverv. nercent
Analyte Spike Level,
G-BHC
Heptachlor
Aldrin
Heptachlor Epoxide
Endosulfan I
Dieldrin
Endosulfan II
Endrin Aldehyde
DDT
DCBP (Surrogate)
A-BHC
B-BHC
D-BHC
A-Chlordane
DDE
Endrin
ODD
Endosulfan Sulfate
Endrin Ketone
DCBP (Surrogate)
Toxaphene-1
Toxaphene-2
Toxaphene-3
Toxaphene-4
DCBP (Surrogate)
Aroclor 1016-1
Aroclor 1016-2
Aroclor 1016-3
Aroclor 1016-4
Aroclor 1260-1
Aroclor 1260-2
Aroclor 1260-3
Aroclor 1260-4
DCBP (Surrogate)
1
1
1
1
1
2
2
2
2
3
1
1
1
2
2
2
2
2
2
3
24
24
24
24
3
4
4
4
4
4
4
4
4
3
n
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
LLE
Ave. (s.d.)
99.2 (1.3)
86.9 (1.5)
77.8 (1.4)
111.7 (2.5)
92.6 (1.1)
89.0 (2.0)
90.2 (2.1)
116.7 (2.5)
67.7 (1.1)
78.1 (1.3)
81.9 (7.5)
95.0 (7.7)
101.6 (8.6)
75.5(19.2)
89.5(10.6)
98.3 (9.4)
111.8 (9.6)
86.9 (9.4)
79.7 (9.8)
48.1(13.0)
63.7 (4.3)
82.3 (4.4)
62.8(12.7)
98.6 (3.3)
63.8(19.4)
52.9 (5.5)
55.1 (8.3)
96.8(11.9)
98.5 (9.5)
80.0 (8.6)
85.1 (8.8)
60.9 (6.6)
91.4(10.5)
99.7(10.7)
Empore- Baker
n Ave. (s.d.)
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
4
5
5
4
4
4
5
4
4
4
4
4
65.8(12.9)
60.9 (7.0)
53.0 (6.6)
75.1(14.0)
63.8(10.6)
61.9 (9.1)
62.1(12.1)
86.3(15.5)
54.2(11.9)
55.5 (5.2)
78.4 (3.5)
100.3 (6.8)
96.2 (6.0)
39.2 (3.8)
44.6 (5.1)
78.9 (4.3)
63.7 (6.7)
75.6 (3.3)
71.5 (2.9)
12.4 (4.6)
50.5(15.3)
36.1 (5.7)
21.7 (5.8)
92.5(10.0)
44.9(17.4)
48.2 (7.5)
59.3(11.0)
83.7 (9.0)
100.1(10.5)
41.2 (5.3)
44.8 (4.9)
30.4 (1.8)
52.6(14.9)
81.9(11.1)
Empore- Varian
n Ave.
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
3
3
3
3
3
4
5
5
5
4
4
4
5
4
65.7(10.8)
51.2 (7.4)
42.9 (6.5)
64.6(11.9)
57.9(10.5)
47.7 (8.2)
48.7 (8.0)
66.5 (9.9)
43.6 (6.7)
46.6 (6.9)
86.7 (4.7)
110.4(12.7)
104.4 (5.7)
55.8 (6.6)
64.3 (8.7)
92.9(10.1)
87.1 (7.5)
84.3 (5.7)
77.7 (4.8)
26.2 (4.9)
69.7(21.6)
50.4 (4.8)
45.9 (2.3)
93.3(26.2)
72.1 (3.7)
29.1 (5.2)
32.3(11.2)
44.1(17.7)
65.4(16.8)
14.9 (1.0)
12.6 (5.2)
12.9 (2.6)
33.9(17.0)
27.0 (6.6)
-------
[SLIDE 12]
95
SUMMARY OF SPIKE RECOVERIES FOR
PESTICIDE EFFLUENT MATRIX
Recoverv. oercent
Analyte Spike Level
(s.d.)
G-BHC
Heptachlor
Aldrin
Heptachlor Epoxide
Endosulfan I
Dieldrin
Endosulfan II
Endrin Aldehyde
DDT
DCBP (Surrogate)
A-BHC
B-BHC
D-BHC
A-Chlordane
DDE
Endrin
ODD
Endosulfan Sulfate
Endrin Ketone
DCBP (Surrogate)
Toxaphene-1
Toxaphene-2
Toxaphene-3
Toxaphene-4
DCBP (Surrogate)
Aroclor 1016-1
Aroclor 1016-2
Aroclor 1016-3
Aroclor 1016-4
Aroclor 1260-1
Aroclor 1260-2
Aroclor 1260-3
Aroclor 1260-4
DCBP (Surrogate)
1
1
1
1
1
2
2
2
2
3
1
1
1
2
2
2
2
2
2
3
24
24
24
24
3
4
4
4
4
4
4
4
4
3
LLE
n
5
5
5
5
5
5
5
5
5
5
4
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
Ave. (s.d.)
99.1
67.0
65.8
88.0
87.9
95.4
99.0
123.2
95.8
89.3
104.7
119.5
117.7
89.2
115.6
156.3
121.4
89.6
88.7
83.6
97.2
95.3
109.7
105.0
106.1
(2.7)
(5.2)
(4.8)
(1.8)
(2.0)
(2.0)
(3.3)
(5.3)
(3.9)
(3.8)
(2.5)
(4.4)
(3.6)
(0.9)
(5.7)
(5.4)
(8.2)
(3.7)
(1.0)
(2.2)
(4.7)
(4.1)
(6.5)
(7.6)
(3.6)
3886.2(186.8)
84.6
191.1
76.7
143.7
100.7
115.1
115.7
85.0
(5.5)
(24.4)
(8.7)
(7.4)
(9.2)
(15.6)
(40.3)
(1.5)
n
5
5
5
5
5
5
5
4
4
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
4
5
5
5
4
5
4
4
5
Empore-Baker
Ave. (s.d.)
92.3
69.1
61.7
85.6
83.1
84.8
91.3
90.0
89.4
81.2
94.0
105.9
104.5
82.2
93.2
135.4
102.3
87.8
88.5
71.2
72.3
77.3
51.2
98.3
81.5
(5
(5
(5
(3
(3
(2
(5
(6
(3
(4
(3
(3
(4
(8
(8
(5
(7
(1
(5
(4
(6
(9
•6)
.7)
•1)
.7)
.2)
.3)
.5)
.8)
•3)
.1)
.0)
-7)
•1)
.2)
.2)
.0)
.0)
•1)
•2)
•8)
•4)
•1)
(10.4)
(11.0)
(4.0)
3338.6(434.8)
99.6
147.5
82.4
101.3
86.1
103.7
81.5
62.1
(22.2)
(45.8)
(23.2)
(13.4)
(8.5)
(11.0)
(15.5)
(7.
9)
n
5
5
5
5
5
5
5
5
5
5
4
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
Empore-Varian
Ave.
91.7
67.4
58.1
78.0
74.5
82.5
86.1
91.1
89.9
92.4
97.7
111.9
108.8
96.4
98.5
140.4
112.2
92.1
88.9
71.7
60.6
51.2
94.9
76.6
77.2
(2.3)
(1.6)
(1.7)
(2.3)
(2.2)
(2.9)
(2.2)
(3.8)
(4.4)
(3.1)
(1.3)
(5.1)
(4.4)
(1.5)
(5.8)
(2.7)
(12.2)
(4.1)
(2.8)
(3.1)
(7.8)
(20.3)
(31.8)
(13.3)
(10.8)
3162.3(268.8)
88.3
128.4
62.7
98.5
87.1
115.6
160.6
66.7
(6.4)
(9.1)
(6.8)
(8.7)
(7.6)
(17.7)
(38.9)
(3.6)
-------
[SLIDE 13]
96
METHOD 608 RECOVERY STUDY
Matrix: POTW Effluent
Empore-Baker
Empore-Varian
so
50
70
90
110
% Recovery (LLE)
-------
[SLIDE 14]
97
I
r
METHOD 608 RECOVERY STUDY
Matrix: Pulp/Paper Effluent
Empore-Baker
Empore-Varian
% Recovery (LLE)
-------
[SLIDE 15]
98
METHOD 608 RECOVERY STUDY
Matrix: Pesticide Effluent
Empore-Baker 70
Empore-Varian
50
50
70
90
110
130
% Recovery (LLE)
150
-------
[SLIDE 161 99
"There are three kinds of lies:
lies, damned lies, and statistics."
Benjamin Disraeli
-------
[SLIDE 17]
100
SUMMARY OF STATISTICAL TESTS FOR
POTW MATRIX
Analyte
G-BHC
Heptachlor
Aldrin
Heptachlor Epoxide
Endosulfan I
Dieldrin
Endosulfan II
Endrin Aldehyde
DDT
DCBP (Surrogate)
A-BHC
B-BHC
D-BHC
A-Chlordane
DDE
Endrin
ODD
Endosulfan Sulfate
Endrin Ketone
DCBP (Surrogate)
Toxaphene- 1
Toxaphene-2
Toxaphene-3
Toxaphene-4
DCBP (Surrogate)
Aroclor 1016-1
Aroclor 1016-2
Aroclor 1016-3
Aroclor 1016-4
Aroclor 1260-1
Aroclor 1260-2
Aroclor 1260-3
Aroclor 1260-4
DCBP (Surrogate)
t-Test. a
LLE/Empore-BAK
NO
NO
NO
NO
NO
YES
NO
YES
NO
YES
NO
NO
NO
YES
NO
NO
NO
NO
NO
NO
YES
NO
NO
NO
NO
YES
YES
YES
YES
YES
YES
NO
NO
YES
= 0.05
LLE/Empore-VAR
NO
YES
NO
NO
YES
YES
NO
YES
NO
YES
NO
NO
YES
YES
NO
YES
NO
YES
NO
NO
YES
NO
NO
NO
NO
YES
YES
YES
YES
YES
YES
NO
YES
YES
F-Test.
LLE/Empore-BAK
YES
NO
NO
NO
NO
NO
NO
NO
NO
NO
NO
NO
NO
NO
NO
NO
NO
NO
NO
NO
NO
NO
NO
NO
YES
NO
NO
NO
NO
YES
NO
NO
YES
NO
a = 0.05
LLE/Empore-VAR
NO
NO
NO
NO
NO
NO
NO
NO
NO
NO
NO
NO
NO
NO
NO
NO
NO
NO
NO
NO
NO
NO
NO
NO
YES
NO
NO
YES
NO
YES
NO
NO
NO
NO
-------
[SLIDE 18]
101
SUMMARY OF STATISTICAL TESTS
MATRIX: POTW
Analyte
G-BHC
Heptachlor
Aldrin
Heptachlor Epoxide
Endosulfan I
Dieldrin
Endosulfan II
Endrin Aldehyde
DDT
DCBP (Surrogate)
A-BHC
B-BHC
D-BHC
A-Chlordane
DDE
Endrin
DDD
Endosulfan Sulfate
Endrin Ketone
DCBP (Surrogate)
Toxaphene-1
Toxaphene-2
Toxaphene-3
Toxaphene-4
DCBP (Surrogate)
Aroclor 1016-1
Aroclor 1016-2
Aroclor 1016-3
Aroclor 1016-4
Aroclor 1260-1
Aroclor 1260-2
Aroclor 1260-3
Aroclor 1260-4
DCBP (Surrogate)
Average Recovery Difference
Smaller Variance
LLE/Empore-BAK LLE/Empore-VAR LLE/Empore-BAK LLE/Empore-VAR
-11
-24
-11
16
-16
-26
-14
-52
-11
-18
-33
-18
-20
-18
-14
-20
-17
9
25
10
10
-27
-63
-13
-32
-18
-23
-21
-16
-8
EMPORE
EMPORE
LLE
LLE
EMPORE
LLE
LLE
-------
102
[SLIDE 19]
SUMMARY OF STATISTICAL TESTS FOR
PULP/PAPER MATRIX
Analyte
G-BHC
Heptachlor
Aldrin
Heptachlor Epoxide
Endosulfan I
Dieldrin
Endosulfan II
Endrin Aldehyde
DDT
DCBP (Surrogate)
A-BHC
B-BHC
D-BHC
A-Chlordane
DDE
Endrin
DDD
Endosulfan Sulfate
Endrin Ketone
DCBP (Surrogate)
Toxaphene-1
Toxaphene-2
Toxaphene-3
Toxaphene-4
DCBP (Surrogate)
Aroclor 1016-1
Aroclor 1016-2
Aroclor 1016-3
Aroclor 1016-4
Aroclor 1260-1
Aroclor 1260-2
Aroclor 1260-3
Aroclor 1260-4
DCBP (Surrogate)
t-Test. a
LLE/Empore-BAK
YES
YES
YES
YES
YES
YES
YES
YES
YES
YES
NO
NO
NO
YES
YES
YES
YES
YES
NO
YES
NO
YES
YES
NO
NO
NO
NO
NO
NO
YES
YES
YES
YES
YES
= 0.05
LLE/Empore-VAR
YES
YES
YES
YES
YES
YES
YES
YES
YES
YES
NO
YES
NO
NO
YES
NO
YES
NO
NO
YES
NO
YES
NO
NO
NO
YES
YES
YES
YES
YES
YES
YES
YES
YES
F-Test.
LLE/Empore-BAK
YES
YES
YES
YES
YES
YES
YES
YES
YES
YES
NO
NO
NO
YES
NO
NO
NO
YES
YES
YES
YES
NO
NO
YES
NO
NO
NO
NO
NO
NO
NO
YES
NO
NO
a - 0.05
LLE/Empore-VAR
YES
YES
YES
YES
YES
YES
YES
YES
YES
YES
NO
NO
NO
YES
NO
NO
NO
NO
NO
YES
YES
NO
YES
YES
YES
NO
NO
NO
NO
YES
NO
NO
NO
NO
-------
[SLIDE 20]
103
SUMMARY OF STATISTICAL TESTS
MATRIX: PULP/PAPER
Average Recovery Difference
Analyte
G-BHC
Heptachlor
Aldrin
Heptachlor Epoxide
Endosulfan I
Dieldrin
Endosulfan II
Endrin Aldehyde
DDT
DCBP (Surrogate)
A-BHC
B-BHC
D-BHC
A-Chlordane
DDE
Endrin
ODD
Endosulfan Sulfate
Endrin Ketone
DCBP (Surrogate)
Toxaphene-1
Toxaphene-2
Toxaphene-3
LLE/Empore-BAK
-33
-26
-25
-37
-29
-27
-28
-30
-14
-23
-36
-45
-19
-48
-11
-36
-46
-41
LLE/Empore-VAR
-33
-36
-35
-47
-35
-41
-42
-50
-24
-32
15
-25
-25
-22
-32
Smaller
LLE/Empore-BAK
LLE
LLE
LLE
LLE
LLE
LLE
LLE
LLE
LLE
LLE
EMPORE
EMPORE
EMPORE
EMPORE
LLE
Variance
LLE/Empore-VAR
LLE
LLE
LLE
LLE
LLE
LLE
LLE
LLE
LLE
LLE
EMPORE
EMPORE
LLE
EMPORE
Toxaphene-4
DCBP (Surrogate)
Aroclor 1016-1
Aroclor 1016-2
Aroclor 1016-3
Aroclor 1016-4
Aroclor 1260-1
Aroclor 1260-2
Aroclor 1260-3
Aroclor 1260-4
DCBP (Surrogate)
-39
-40
-30
-39
-18
-24
-23
-53
-33
-65
-73
-48
-57
-73
LLE
LLE
EMPORE
EMPORE
EMPORE
-------
104
[SLIDE 21]
SUMMARY OF STATISTICAL TESTS FOR
PESITICIDE EFFLUENT MATRIX
Analyte
G-BHC
Heptachlor
Aldrin
Heptachlor Epoxide
Endosulfan I
Dieldrin
Endosulfan II
Endrin Aldehyde
DDT
DCBP (Surrogate)
A-BHC
B-BHC
D-BHC
A-Chlordane
DDE
Endrin
DDD
Endosulfan Sulfate
Endrin Ketone
DCBP (Surrogate)
Toxaphene-1
Toxaphene-2
Toxaphene-3
Toxaphene-4
DCBP (Surrogate)
Aroclor 1016-1
Aroclor 1016-2
Aroclor 1016-3
Aroclor 1016-4
Aroclor 1260-1
Aroclor 1260-2
Aroclor 1260-3
Aroclor 1260-4
DCBP (Surrogate)
t-Test. a
LLE/Empore-BAK
YES
NO
NO
NO
YES
YES
YES
YES
YES
YES
YES
YES
YES
NO
YES
YES
YES
NO
NO
YES
YES
YES
YES
NO
YES
YES
NO
NO
NO
YES
YES
NO
NO
YES
- 0.05
LLE/Empore-VAR
YES
NO
YES
YES
YES
YES
YES
YES
NO
NO
YES
YES
YES
YES
YES
YES
NO
NO
NO
YES
YES
YES
NO
YES
YES
YES
NO
YES
YES
YES
YES
NO
NO
YES
F-TesL
LLE/Empore-BAK
NO
NO
NO
NO
NO
NO
NO
NO
NO
NO
NO
NO
NO
YES
NO
NO
NO
YES
YES
NO
NO
NO
NO
NO
NO
NO
YES
NO
YES
NO
NO
NO
NO
YES
a = 0.05
LLE/Empore-VAR
NO
YES
YES
NO
NO
NO
NO
NO
NO
NO
NO
NO
NO
NO
NO
NO
NO
NO
YES
NO
NO
YES
YES
NO
YES
NO
NO
YES
NO
NO
NO
NO
NO
NO
-------
[SLIDE 22]
105
SUMMARY OF STATISTICAL TESTS
MATRIX: PESTICIDE EFFLUENT
Average Recovery Difference
Analyte
G-BHC
Heptachlor
Aldrin
Heptachlor Epoxide
Endosulfan I
Dieldrin
Endosulfan II
Endrin Aldehyde
DDT
DCBP (Surrogate)
A-BHC
B-BHC
D-BHC
A-Chlordane
DDE
Endrin
DDD
Endosulfan Sulfate
Endrin Ketone
DCBP (Surrogate)
Toxaphene-1
Toxaphene-2
Toxaphene-3
Toxaphene-4
DCBP (Surrogate)
Aroclor 1016-1
Aroclor 1016-2
Aroclor 1016-3
Aroclor 1016-4
Aroclor 1260-1
Aroclor 1260-2
Aroclor 1260-3
Aroclor 1260-4
DCBP (Surrogate)
LLE/Empore-BAK
-7
-5
-11
-8
-33
-6
-8
-11
-14
-13
-22
-21
-19
-12
-25
-18
-58
-25
-548
-42
-15
-23
LLE/Empore-VAR
-7
-8
-10
-13
-13
-13
-32
-7
-8
-9
7
-17
-16
-12
-37
-44
-28
-29
-724
-63
-14
-45
-14
-18
Smaller Variance
LLE/Empore-BAK LLE/Empore-VAR
EMPORE
EMPORE
LLE
EMPORE
LLE LLE
LLE
LLE
LLE
LLE
EMPORE
LLE
LLE
-------
106
a
c/i
s
TD
C
(D
0)
VH
O
&
S
W
(D
"o
00
o
(N
8
(N
o
-------
107
co
CO
3
Q
CO
c
2
^
O
to
cr
G U3
c
os
">
X -*->
(D 0)
fc
2
ex
E
(U
T3
• i-^
.5"
"3
.?
y*^
l^^\
i-H
.S
.ti
J^.
•^
1^^
X*
Id
3
CO
3
-
2
cd
cd
Cu
E
o
0
(D cd
cd ^
CO -,
0) X
J-H d>
•4_>
C
« ^H
UH
cd
-C
CO
CO
'S
cd
t<
(D
^
O
3
(D
CO
;—]
— CO
S 52
Ecd
b/1
ON ^
(D •£
-c •ti
H £
o
IS
w
HH
00
co
O
CO
r\
CO
o
O
UH
(D
T3
(D
CO
3
s
o
(D
O
a
-------
108
C
o
•J
-------
109
s
en
•^^
D.
s
o
U
D C
* 1
cd o
8
U
fr
c/5
.2 U
-------
110
[Blank Page]
-------
Ill
MR. TELLIARD: Could we get started again, please? We would like to
continue. I forgot this morning to announce that...after 15 years, my mind is going or has
gone...the folks to my right are taking down the proceedings that end up on a publication
that generally comes out two to three hours right after this publication or concurrently,
you know, in about 11 months.
So, everything we are saying is being recorded for posterity. If for some
reason you would like to make a comment, call somebody a name or something along
that line, and don't necessarily want it in the record or the proceedings, simply state that
when you get to the microphone, and we will be glad to turn off the women. Turning off
women is one of my specialties. So, keeping that in mind, we will continue on.
Our next speaker is from Boise Cascade. They are the people who make
those milkshakes that you heard about earlier this morning.
Sarah Barkowski is with Boise Cascade and has been doing some work on
the analysis of dioxins and furans using solid phase extraction, and I hope she is using
Method 1613, and she is going to talk a little bit about their success with using solid
phase extraction.
Thank you.
-------
112
MS. BARKOWSKI: I would like to start off the record. Our milkshakes
are strawberry.
First slide, please. The analytes that I want to talk about are 2,3,7,8-
tetrachlorodibenzodioxin and 2,3,7,8-tetrachlorodobenzofuran or just dioxin and furan for
short.
As you guess from the structure of these compounds, they are very
hydrophobic. They don't present for use the problems that we have seen earlier this
morning with the phenols. 2,3,7,8-TCDDis an extremely toxic compound for certain
animals, and, unfortunately for the pulp and paper industry, very small quantities of
these compounds are made under certain bleaching conditions during the bleaching of
pulp.
Naturally, then, the EPA being right on the ball, they have written some
methods for analyzing for our effluents for these compounds at the part per quadrillion
level. Other agencies have written similar methods, and they all follow the same sort of
generic background.
Next slide, please. It is an isotope dilution method. A liter of sample is
spiked with C13 labeled 2,3,7,8-TCDDand TCDF which I will refer to as the internal
standard. When everybody writes a method, they all call these things different. They
assign different names to all these standards, and I have always called these the internal
standards.
The sample after spiking is then filtered, and the solids are Soxhlet
extracted. Some methods use toluene and a Dean Stark trap on the Soxhlet. Other
methods use a mixture of toluene and ethanol, and the air dry the sample. But the point
is that the sample is filtered, and the solids are extracted separately.
The filtrate is extracted by all the methods using a separatory funnel,
liquid-liquid shakes with methylene chloride. The sample bottles have to be rinsed, as
we talked about this morning, with other analytes. Methylene chloride is used to rinse
the sample bottle, and that just gets thrown in with the shakes.
These two extracts then have to be concentrated and combined, and then
that gross extract is spiked with Cl37 labeled cleanup standard.
-------
113
The extract then goes through a series of column chromatography steps
which serve to purify the extract, and finally, it is concentrated to 10 microliters, and just
prior to analysis by GD/mass spec, the extract is spiked with the C13 labeled recovery
standard which has the chlorines in the l,2,3,4positions.
You get five pieces of information from this process. Three of those are
the internal standard and the cleanup standard recoveries.
The internal standard recoveries will reflect losses that occurred after
spiking with internal standard and before spiking with recovery standard, whereas the
cleanup recoveries will reflect only losses that occurred after the extraction procedure.
So, if you run a sample and you get lousy recoveries across the board with
all these standards, you know that the losses occurred in the cleanup steps and not
during extraction. Alternatively, if the internal standard recoveries are poor but the
cleanup standard recoveries are good, then you know that the losses occurred during
extraction.
The two other pieces of information that you get from this are, of course,
the analyte concentrations, the 2,3,7,8dioxin and furan native to the sample.
As with any isotope dilution method, the measured analyte concentrations
will be correct only if the internal standard is recovered at the same efficiency that the
native is recovered. If both of these compounds are recovered at 10 percent efficiency,
the measured concentration will be correct. If they are both extracted and recovered at
100 percent efficiency, the measured concentration ' will be correct. This is true with any
isotope dilution method.
However, if for some reason the native is recovered at 100 percent and the
internal standard is recovered at 50 percent, the measured concentration will be twice
the actual concentration.
Now, you might ask how could you do this when the only difference
between these compounds is that one is C13 labeled and one is just the native
compounds. Well, back when you spiked the compounds or spiked the samples, there is
no guarantee that the internal standard will get incorporated into the matrix in a way
that mimics the native compounds.
-------
114
So, for example, if the native is...whydon't you go ahead and go to the next
slide. If the native is 90 percent attached to the solids and only 10 percent over there in
the filtrate portion and the internal standard gets evenly distributed between those two
phases, then if your Soxhlet extraction and filtrate extraction run at different efficiencies,
then the internal standard recoveries in this example would be 75 percent which is
generally accepted to be a good recovery, but the measured concentration will have
inherent to it a 27 percent positive bias.
So, basically, there are some assumptions that the method makes that must
be valid before the measured concentrations will be good.
Either the internal standard and native have to partition between these two
phases in exactly the same way, or you have to extract these two phases at exactly the
same efficiency, or else you will have an inherent bias in the procedure.
The reason that I wanted to bring attention to this feature of the method is
because what I am talking about is making a change to the conventional liquid-liquid
extraction part of the procedure, and it is very important when doing this, you know, you
i
make a change like this, to validate this.
i
Matrix spikes or native spiked blanks are not a rigorous validation
experiment. You have to run samples side by side, samples which contain measurable
quantities of these compounds. Side by side. Do them by liquid-liquid shakes and do
them by the alternative extraction procedure and compare the results.
We know that we can extract matrix spikes at the same recovery that we
extract the internal standard, because they are in the same...you know, they are
introduced in the sample in the same way.
The reason Boise Cascade decided to mess around with this procedure was
because about two or three years ago, we had about 200 or 300 effluent samples that
needed to be analyzed, and nobody in our lab had ever done a liquid-liquid shake on an
effluent before.
When we started doing that, we found out right away that we formed
emulsions, and not being experienced with handling these emulsions, we found ourselves
extracting a couple samples in the morning and spending the rest of the day trying to
-------
115
break up these emulsions. The backlog was not improving.
So, we abandoned our efforts altogether in analyzing the samples and just
focused our attention in developing an alternative extraction method.
Why don't you go to the next slide, please. After failing miserably with
some attempts at continuous liquid-liquid extractions, we started experimenting with
some solid phase extraction methods. We started out with the pre-packaged columns
and cartridges that were available. This is a couple years ago. They were plugging
within less than 100 ml.
Then we went on to the 47 mm disks. This was at a time when the 90 mm
weren't available. And we did get significantly more sample through. We could get 300
or 400 ml, depending on the sample, before it would plug, but they did plug before we
got a liter through.
We did one of those long, drawn-out experiments, though, where we just
let it drip and drip and drip and drip so that we could get a sample through the lab, and
we did get good recoveries, and the analyte concentrations matched very well with the
liquid-liquid shakes.
So, what that told us was that there is a good chance that the chemistry
was right for doing solid phase extraction of effluents for dioxins and furans. We were
just restricted by the physical limitations of the gadgets that were out at that time.
So, we had some conversations with the folks at 3M, and bearing in mind
the hundreds of samples in our backlog, we decided to go ahead and put together a little
homemade device.
Next slide please. This is a...it is kind of a poor man's disk. This is using
the conventional glassware for the filtration assembly. We put a GFF glass fiber filter in
the normal spot, and then we poured both phase RP silica on top and just leveled it out.
It was 10 ml of silica.
And then we wedged one of those chubby GFDs on top of that and kind of
made a sandwich out of it.
Next slide, please. These sandwiches didn't plug, so we went ahead and
worked out the details of running the samples through, and we have done some
-------
116
validation experiments.
The pre-extract steps that we used for the sandwich method was to adjust
the sample pH to 1 to 2. Sample pH adjustment is one of those optional steps in some
of the conventional methods, and we found that it was a necessary step in doing solid
phase extraction.
The conventional methods, the spikes are usually transferred in acetone,
and this is done in hopes to get that internal standard integrated into the matrix.
Acetone is a little bit stronger solvent than ethanol, so we chose to go ahead with
ethanol since we could get the spike dissolved in the ethanol, and we could get the
ethanol dissolved into the sample. We figured it would accomplish the same thing.
We sonicate all our samples right after spiking for an hour, just a little
added step that makes us feel better about getting that internal standard incorporated
into the matrix.
The samples are then pre-filtered through a GFC and through a OFF, and
the bottle rinse step which is normally done with methylene chloride, we were doing that
now with methanol, and then that methanol gets added to the filtrate, and it makes, if
you have got a 1 liter sample, it makes your filtrate 5 percent in methanol which will
help maintain the activity of the silica.
The filtrate extraction steps, then, are to assemble the apparatus from the
previous slide and then to pass methanol through the silica to activate it, and then pass
the filtrate through before the silica... or before the methanol has all gone through, and
then we were eluting with methylene chloride.
Next slide, please. We didn't optimize the parameters for this method
because of the pressure to get the samples run through the lab, so we just went ahead
with those and validated it.
To validate the method, like I said earlier, we didn't look at matrix spike
recoveries and those sorts of things, because they really don't reflect...they don't address
the issues that I talked about earlier where it is so important that the native and the
internal standard be proportioned between those two phases in the same way and that
the recoveries are the same through those two phases.
-------
117
So, we just went ahead and compared measured concentrations from runs
using the sandwich method to measured concentrations from runs using the liquid-liquid
extraction method. Also, we did look at the internal standard recoveries.
And to test for break-through, we took the filtrate after it had been passed
through the sandwich and did liquid-liquid shakes on that. So, I will show you some of
the results from that as well.
Next slide. You can see here that we did 12 runs with liquid-liquid shakes
and 12 runs using the sandwich method. All those sandwich runs were done at Boise
Cascade. The liquid-liquid shakes were done mostly at two different outside labs, labs A
and B, and then also some were done at Boise Cascade.
There are four samples in this experiment. These are all...well,they are
not milkshakes, but they are real samples. They have been collected from four different
Boise Cascade mills. Three of them are secondary effluents or effluents to the river, and
one of them is an untreated effluent.
You can see that the lowest level sample we ran, sample number 4, is
around 15 parts per quadrillion TCDD and that the highest level sample is about 130
parts per quadrillion TCDD. So, that is kind of the low end and the high end.
The calibration limit for the method is 10 ppq. Then, samples 1 and 2
kind of fall there in between.
Regardless of the statistical method that I came up with to look at these
two sets of numbers, the two methods yielded results which were not significantly
different at the 99 percent confidence limits. I ran these by lumping all the samples
together, but what that tends to do is just increase the sigmas for the two sample sets
because of the big differences in the samples.
So then I did it sample by sample, comparing the means of all the liquid-
liquid runs for sample 1 to the mean of all the sandwich runs for sample 1, and for both
analytes and for all four samples, the difference between the means was not statistically
significant at the 99 percent confidence limit.
Next slide, please. The next slide looks at the internal standard recoveries
for the two different runs. On the right-hand side is the average of the sandwich
-------
118
methods and 95 percent confidence limits for that, and then the bar on the left-hand side
is the average of the internal standard recoveries for the liquid-liquid shake method.
So, you can see that we got a little better recoveries. I think the point is
that the recoveries were, in general, good.
There is another factor in this experiment because it crosses labs, you
know.
If you go to the next slide, it shows a comparison of the furan internal
standard recoveries for the liquid-liquid on left versus the sandwich on the right.
Next slide, please. We did get through our backlog using this method, but
as you all know, 3M has come out with the 90 mm disk. This is a chromatogram from a
standard, and I want to explain this so I can show you what we got in our test for break-
through.
The top two peaks show the native analyte. The bottom two peaks are the
C13 labeled peaks. The ones on the right-hand side which elute the same time as the
native, that is the C13 labeled internal standard, and the one on the left-hand side is the
1, 2, 3, 4, the recovery standard that is put in. This is just what a standard injection
looks like, the chromatogram.
Would you go to the next slide, please? When we did the liquid-liquid
shakes on the filtrate after it had passed through the sandwich, this is a typical result of
what we got. The peaks that you see there are due to the recovery standard.
The location of where the internal standard would be...could you raise that
a little bit, a little higher? I put an arrow on there to show. That is where the internal
standard would be if we had found some.
And then the two chromatograms on the top would show where the
native...the native would be at the same place where the internal standard is.
So, basically, our tests for break-through on the solid phase extraction
method pretty much all looked like this. There was no significant break-through.
With the 90 mm disks available now, though, there are a lot of limitations
to this sandwich method, especially in comparison to the disk method. We are using 10
ml of the RP silica for our sandwich, and we are using it very inefficiently. We don't
-------
119
have it packed into a little teflon matrix. That is why we have to use so much of it. You
can get by with a lot less silica on a disk than you can with a sandwich.
It takes a lot of time to put the sandwiches together. The RP silica all has
to be pre-extracted with methylene chloride. The two filters have to be cleaned. The
top filter has to be trimmed. You know, it takes time to put all that together.
And the solvent used for the solid phase extraction method is not at all
minimized. As I said earlier, we really didn't optimize for minimal solvent use. We
optimized for the fastest method we could validate and run our samples.
The sandwich method worked very well in comparison to the liquid-liquid
shake method, but in comparison to the disk method, there is a whole lot of
opportunities that are in front of us there.
Next slide, please. The only differences between the pre-extract steps with
the disk method and the sandwich method is that you only pre-filter through a GFC, and
the methanol that is added, we reduced that from 50 ml to 5 ml.
In our tests for break-through on the disk method, we were seeing some
early experiments, and we cut back on this methanol thinking that it might reduce that
break-through, but I don't think it really had an effect.
The filtrate extraction steps, though, are a little bit different. We put a
OFF on top of the disk so the filtration steps, whereas in the other method, you have
two filtration steps, with the disk method, you also have two filtration steps, and the
extraction of the filtrate gets accomplished in that second filtration step. So, it really
streamlines the procedure.
Again, the 50 ml of methanol is passed through to activate the disk. And
then we put a water step in between the methanol step and the filtrate, and that kind of
helps make the method a little bit more user friendly.
If you add the methanol and then the sample, you have got to get that
sample in before the methanol all goes through but not too much before, or else you will
have a lot of methanol in there, and you will just be washing your analytes through. So,
you can get a little bit sloppy about when you add these things if you put a water step in
between.
-------
120
Finally...could you raise that a little bit? The real streamlining of the
procedure is in this last part here. The analytes are not eluted from the disk with
solvent bypassing it through the filtration assembly. Instead, we just take the disk and
the OFF and the GFCs from the pre-filtration and throw the whole mess into the
Soxhlet.
So, what we have...next slide, please...is the original method gets
streamlined quite a bit, because you don't have two separate flasks that have to be
rotovapped and combined. You know, the methylene chloride from doing shakes, you
have to dry it with sodium sulfate, rotovap or concentrate it by whatever means and
combine it, and all that is removed from the procedure.
I know everybody in our lab is delighted to run the disk method as
compared to any of the other methods.
Next slide, please. Again, to validate the disk method, we did run, of
course, you know, method blanks and spikes of the method blanks and spikes of samples,
and all those results, you know, we got clean method blanks, and our spiked recoveries
are all right around 100 percent, but just because you have accomplished that doesn't
mean you have a method that will give a right answer. You have to compare the results
of measured concentrations from liquid-liquid shakes compared to the disks, and again
we did the tests for break-through.
Next slide, please. If you will start out by looking at just the first eight
lines here, you can see that there were four runs with liquid-liquid shakes, four runs with
the disk method. Sample number 3 is evenly distributed between each of those four runs
and the same with sample number 1.
Again, statistical analysis of the concentrations for dioxin and furan show
that there is no statistically significant difference at the 99 percent confidence limits, and
the recoveries, again, for the disk method are very good in comparison to the liquid-
liquid shake method.
The dioxin issue, though, it isn't over with a method that measures down to
10 ppq. The proposed compliance levels at some of the mills in our final effluents are
well below the calibration limit in addition to being below, in some cases, the detection
-------
121
limits.
So, some of the regions are looking into going into the mill back at a
location where that dioxin is more concentrated and pulling a sample from there and
measuring the dioxin.
So, the method, if they are going to do that, the method will need to be
valid for samples other than just our final effluents.
Now, a final effluent really looks like a cup of tea, but a D-stage filtrate
looks like espresso, and it is about a pH 10. So, we ran some of the espresso and other
filtrates collected during the different stages of the bleaching process through the disks.
And we didn't...when we were running our final effluents, we didn't
measure how long it took to run them through, the disks. They ran through so fast it
didn't matter.
But when the D-stage filtrates and the E-stage filtrates were run through,
they were taking about 20 minutes.
What it kind of looks like when it goes through the disk, the first part that
is coming through is colorless, but after a couple hundred ml have come through, colored
compounds start breaking through, and by the time you get a liter through, you really
can't tell if you have removed a significant amount of that color.
Now, with the D-stage filtrates, they look just as bad coming out as they do
going in, but the analytes here, the dioxins and furans, are so hydrophobic as well as
being particulate-bound that if you reduce the pH of the sample prior to running it
through the disk, you will precipitate out a lot of these compounds, and then they will go
through the Soxhlet extraction rather than the filtrate extraction.
For the first four runs of the purple compounds, those are all collected
from one mill, and the recoveries were good for those samples, and the measured
concentrations compared well to runs that we did using solid phase extraction.
The last three lines on this overhead ' show samples which were non-detect
for dioxin. All the samples were. And there were some measured concentrations for
furan.
These concentrations compared very well with those from liquid-liquid
-------
122
shakes on the same compounds. I didn't include them in the statistical analysis, because
it becomes a little difficult on how to work with non-detects.
You can see, though, that the furan recoveries for two of these samples
were very low. They are down there in the 40 percent.
This was really kind of a puzzle, because you would normally expect the
furan internal standard to be recovered at the same efficiency as the dioxin internal
standard is. You can cut your cleanup steps wrong and lose some of your furan
compounds relative to the dioxins, but if that were happening through our cleanup steps,
you would think that it would happen to all the samples and not just two.
So, I really can't explain why, for two of our runs with the disk method, we
got low furan recoveries for the E-stage filtrates.
Next slide. In our tests for break-through with the disk method, in our first
experiments, we were seeing about 10 percent internal standard, and as I said, we
changed some things in the pre-extract steps to help reduce that, but it wasn't until we
ran a sample through at a reduced pH that we really got it down to where it is now.
You can see just a little bit of internal standard in there, and there are obviously no
peaks up in the native channels.
I guess, in summary, I think that the disk method is definitely an attractive
alternative to liquid-liquid shakes. It reduces solvent consumption, and it makes the
method much more efficient.
But there are a lot of things that will need to be done before you can
validate something like this. In particular, those validation studies will have to include
more work on bleach plant filtrates, the espresso type samples, and it will have to be
work that compares the measured concentrations from the disk method to those from the
liquid-liquid shake method.
Thank you.
-------
123
QUESTION AND ANSWER SESSION
MR. TELLIARD: Any questions?
MR. THOMAS: My name is Roger Thomas. I work for Viar and
Company.
I notice that concerning your solid phase extraction, you noted that there is
a pre-extraction step whereby you take 1 liter of sample plus your spiked compound
which is your internal standard, and then you sonicate it for 1 hour.
Have you done any studies on improved recoveries due to sonication based
on time of sonication, you know, like 15 minutes, 30 minutes, 1 hour, versus, you know,
samples not going through the sonication process?
MS. BARKOWSKI: The answer is no, we haven't done any studies to
compare the effects of sonicating and different lengths of time that you sonicate, but I
would imagine that all we are doing in that sonication step is trying to get the internal
standard integrated into a matrix which mimics the native.
If the recovery of the filtrate and the solids are both good, then the effects
of sonication will not be seen in any studies, but if the filtrate is extracted at a lower
efficiency than the solids, and if the sonication pushes the internal standard toward the
solids, then you would see an effect.
MR. THOMAS: Okay, thank you, but my biggest concern was an
additional hour of time being added, you know.
MS. BARKOWSKI: Well, during the time that the samples are being
sonicated, we are setting up our glassware.
MR. THOMAS: Okay, thank you.
MR. HALVORSON: My name is Jeff Halvorson from Burdick & Jackson.
I had a question for you about the glassware that you are using. You
mentioned that you take some special precautions when you are... for instance, the
reverse phase silica and the materials you are using, you clean them and prepare them
specially. I know looking at part per quadrillion levels, you are going to need to have
clean glassware and everything.
-------
124
Are there any problems that you noticed with the filtration apparatus itself,
any problems cleaning that?
MS. BARKOWSKI: We just used B&J solvents to clean it.
MR. HALVORSON: That is nice to hear.
MS. BARKOWSKI: Is that what you wanted?
MR. HALVORSON: I was more concerned with the frit itself. There is a
lot of surface area in there.
MS. BARKOWSKI: Yes, yes.
MR. HALVORSON: And these are.
MS. BARKOWSKI: That is a good point. The extractions that we did on
the filtrate after it was passed through, anything that we found in there did not prove
that we were getting break-through. Because the glassware goes through two filtration
steps, the first time through, you run it through, and that frit gets exposed to the sample.
The second time through when you have a disk on there, of course, the
first thing you run through the disk is methanol, and that sample that is stuck to the
fritted glassware will then get washed through with methanol down there into that flask
that we then take and test for break-through.
So, it could be that something like that is going on except for that wouldn't
explain why we got different results for break-through from the sandwich method versus
the disk method.
I think to solve the break-through with the disk method, it was important
to lower the pH of the samples.
MR. HALVORSON: Thank you.
MR. TELLIARD: Thank you, sir.
-------
125
2378-TCDD
"DIOXIN"
2378-TCDF
"FURAN"
ppq = pg/L = 10~12
-------
126
SOLIDS
SOXHLET (DS)
EXTRACT
(TOLUENE)
CONCENTRATE
EFFLUENT
SAMPLE
(ONE LITER)
|3c12-2378-TCDD
FILTRATE
C12-2378-TCDF
(INTERNAL STANDARD)
LIQ/LIQ
EXTRACT
(DCM-BOTTLE RINSES)
CONCENTRATE
COMBINE
37CIA-2378-TCDD
— (CLEAN-UP STANDARD)
COLUMN CHROMOTOGRAPHY
v
CONCENTRATE
13
Cir1234-TCDD
(RECOVERY STANDARD)
ANALYZE
GC/MS
-------
127
NATIVE ANALYTES PARTITION:
90% SOLIDS
10% FILTRATE
INTERNAL STANDARDS PARTITION:
50% SOLIDS
50% FILTRATE
SOXHLET EXTRACTION:
FILTRATE EXTRACTION:
100%
50%
(ASSUME 100% RETENTION THROUGH CLEAN UP.)
INTERNAL STANDARD RECOVERIES: 75%
MEASURED ANALYTE CONCENTRATIONS: 1.27 x ACTUAL
-------
128
SPE METHODS
o PRE-PACKAGED COLUMNS AND CARTRIDGES
PLUGGED RAPIDLY
o 3M EMPORE DISK (47 MM)
PLUGGED
IF PATIENT, GOOD RESULTS
o HOMEMADE SANDWICH
-------
129
BCC
Solid Phase Extraction Apparatus
2.7u Filter
RP Silica
0.7u Filter
Vacuum
-------
130
SANDWICH METHOD
PRE-EXTRACT STEPS
o ADJUST SAMPLE pH TO 1 -2
o SPIKE IN ETHANOL
o SONICATE 1 HOUR
o FILTER THROUGH GF/C AND GF/F
o RINSE BOTTLE WITH 50 ml METHANOL
FILTRATE EXTRACTION STEPS
o PREPARE SANDWICH:
10 ML RP SILICA
GF/F (BOTTOM); GF/D (TOP)
o PASS 30-50 ml METHANOL (TO ACTIVATE)
o PASS FILTRATE BEFORE SILICA GOES DRY
o ELUTE WITH 300 ml DCM
-------
131
VALIDATION OF SANDWICH METHOD
o MEASURED ANALYTE CONCENTRATIONS:
SANDWICH VS. L/L
o INTERNAL STANDARD RECOVERIES:
SANDWICH VS. L/L
o L.L SHAKES OF FILTRATES AFTER PASSING
THROUGH SANDWICH
-------
132
CO T- c\J CO CD CO CO
CD
C\J
•sfr oo to
00 CO O>
Q
CO
Q
O
H-
oooooooo
O
CMCM
CM CM CM CM
Oh-T-OOOO^QOO
COCOCO-i— CvJOJOJOOOO
Q
D
O
LU
_J
QL
^
<
O)
OO OOOOOOOOOOOO
OO OOOOOOOOOOOO
CDCQ<
-------
133
T 1 T
Q
D
O
H
I
CD
r-
o
cu
I
u
n
(U
z
u
H
O
-------
134
i—r
i—r
"i—r
i—i—r
(U
i i i
j L
j L
J L
-------
135
-------
136
CM
-------
137
LIMITATIONS OF SANDWICH METHOD
(AS COMPARED TO EMPORE DISK)
o INEFFICIENT USE OF EXPENSIVE RP SILICA
o SOLVENT USE NOT MINIMIZED
-------
138
DISK METHOD
PRE EXTRACT STEPS:
o AD JUST pH TO 1-2
o SPIKE IN ETHANOL
o ADD 5 ml METHANOL
o SONICATE 1 HOUR
o FILTER THROUGH GF/C
o RINSE BOTTLE WITH 50 ml TOLUENE (SOXHLET)
FILTRATE EXTRACTION STEPS:
o GF/F ON TOP OF 90 MM EMPORE DISK (C18)
o PASS 50 ml METHANOL (TO ACTIVATE)
o PASS 50 ml WATER (BEFORE DISK GOES DRY)
o PASS FILTRATE (BEFORE DISK GOES DRY)
o SOXHLET-DS EXTRACT WITH TOLUENE:
ALL GF/Cs
GF/F
EMPORE-DISK
-------
139
EFFLUENT
SAMPLE
(ONE LITER)
FILTER
SOLIDS
\/
SOXHLET (DS)
EXTRACT
(TOLUENE)
CONCENTRATE
13
13C12-2378-TCDD
C12-2378-TCDF
(INTERNAL STANDARD)
SPE EXTRACT
(DISK & GF/F)
37
Cl ^-2378-TCDD
-(CLEAN-UP STANDARD)
COLUMN CHROMOTOGRAPHY
CONCENTRATE
13
C12-1234-TCDD
(RECOVERY STANDARD)
ANALYZE
GC/MS
-------
140
VALIDATION OF DISK METHOD
o MEASURED ANALYTE CONCENTRATIONS:
DISK VS. L/L
o INTERNAL STANDARD RECOVERIES:
DISK VS. L/L
0 L/L SHAKES OF FILTRATES AFTER PASSING THROUGH DISK
-------
141
"3-
-------
142
COUDTCMrHCOOrOtnoOO
u< m m m m
w u w w
m in
W U Cd
o a") ao
-------
143
MR. TELLIARD: Our next speaker comes from PACE Laboratories.
Gabe is going to be talking about, again, solid phase extraction as it relates to the
analysis for explosives. We have seen some data on this. We got a real big bang out of
it.
MR. LE BRUN: First slide, please.
Good morning or good late morning. My name is Gabe Le Brun, and I
am from PACE, Incorporated. Basically, what I am here to talk about today is the solid
phase extraction disk method for the extraction of explosives from water.
Myself and Jim Madison developed this for the cartridge type technique
about a year ago. We have been doing explosives by solid phase for about a year.
You may notice there is not on this list of credits, Craig Markell. If it
wasn't to his dedication and perseverance and, mainly, his patience for my 6:00 a.m.
phone calls and weekend phone calls, we wouldn't have been able to get the material
that we were able to get for this particular presentation.
The first thing I want to do is I want to take a look at the analytes, the
analytes of interest, and they were these particular compounds right here. HMX and
RDX are, you may notice, the non-aromatic type compounds. TETRYL, the tri-
nitrobenzene, the di-nitrobenzenes, the nitrotoluenes, the di-nitrotoluenes, and then the
last two in the lower right-hand corner, PETN and nitroglycerin. You will notice that
those are the two smaller molecules. Therefore, they needed to be analyzed at a
different wavelength.
The two types of techniques that we have been utilizing, as I mentioned
earlier, was the cartridge technique which is essentially a 6 ml type self-packed cartridge
when we weren't able to get styrene divinylbenzene' in actual cartridges.
And then the disk which I am not even sure if it is on the market as of
right now, but at that particular time...that is why I was working so close with Craig, he
said that he had disks for it. So, we were able to get some of these disks from Craig.
We were using the 47 mm disks.
The biggest difference between these two methods is time. The amount of
-------
144
time is unbelievable, as you will end up seeing from the following slides.
For the disk type technique, we analyzed groundwater, reagent water, and
surface water. The groundwater went through in approximately 4 minutes. The reagent
water went through in about 10 minutes.
You may be wondering why the difference between groundwater and
reagent water. We think it is probably because in the groundwater, it has gone through
a much more filtered type system. They are much cleaner than the actual reagent water.
Kind of a scary thought, but it was much faster.
The surface water averaged about 70 minutes. We have heard about the
milkshakes, and, actually, if this was given about a week ago, I would say that most of
our samples have a little bit of paniculate matter, but Thursday and Friday of last week,
we had mosquito larvae swimming up and out, a lot of silicates, things like that.
I wish I could have had a slide, but believe me, I couldn't get it developed
fast enough for this presentation.
The cartridge rate that we are currently using, and this is primarily to get
the HMX recoveries that we need, is 1 to 2 ml a minute. Now, when you are doing 500
ml of water, that translates into 4 to 8 hours.
Those particular samples that I just mentioned with the high silicates and
the biological, they were going the next day when I came in. So, there is problems with
the cartridge technique, and I think that they can be eliminated with the disk technique
with the glass fiber filter, as we have seen earlier.
This is the setup that we used for the cartridge technique. It is basically a
homemade type apparatus, as you can probably see. I have got a manifold down there.
We use the sep funnels. Craig, you want to get rid of the sep funnels? We think it
makes a great container.
We can set this up for auto-flow. We are just stopping the top of the sep
funnel, and it is just gravity feed. It goes into those 50 ml reservoirs. It flows down and,
essentially, when the level goes down in the reservoirs, just enough is delivered that you
can end up displacing with the air in the sep funnels.
When you are dealing with 8 hours, you only want it to be set up. So, that
-------
145
is the particular technique that we used.
Here is just a little closer look. I am sorry about the overexposure, but you
can see that the 50 ml reservoir is attached to the 6 ml cartridges, and then all of the
water is collected right down in your suction flasks.
This is the disk manifold that we used. We obtained this from Craig. It is
very effective. You notice that there are six set-ups.
Those are for the 47 mm disks. The question was automatically asked,
what about the 90 mm disks when they end up becoming available? He goes well, then
you just alternate every other one. So, your six position holder turns into a three
position holder if you want to go to the 90 mm disks.
The disk technique, 75 to 125 ml a minute. I mentioned the item of time.
When you are dealing with 75 to 125 ml a minute, that translates into 4 to 6 minutes.
This is for groundwater and for reagent water. I mentioned earlier that if
you have a lot of paniculate matter...we did not use the glass fiber filters for any of our
stuff that we did. So, I would say that our 70 minutes would probably be drastically cut
down, but on average for the groundwater and the reagent water, we are talking some
very fast flow rates at 54 to 6 minutes for 500 ml of water.
Basically, what I want to do is I want to give you the opportunity to find
out how the method was developed and what we actually did. I will be going into the
disk preparation, the sample introduction. I will be touching briefly on the disk drying
and the elution and concentration.
First, let's take a look at the disk preparation. The disk preparation was
done in the following manner:
Essentially, what we were doing is we were adding acetone. We added 5
ml of acetone and brought through approximately 1 ml of it and allowed it to sit there
for 3 minutes. The purpose of this is to shock the material so that it will release any of
the contaminants.
I cannot emphasize enough how important this disk preparation is, and you
will end up seeing that, believe me.
Next, what we did after bringing the acetone to air, we did the same thing
-------
146
with the acetonitrile. 5 ml again, we brought 1 ml through, let it sit there for about 3
minutes, and then we brought that through to air. This was important, because that is
what we are going to be eluting with.
Finally, we started making that transition so we could end up adding our
aqueous. We added 5 ml of methanol, brought through about 1 ml, let it sit there for 3
minutes, and then we brought that down, at this point not allowing the disk to go dry.
We had about 1 ml there before we added 10 to 15 ml of water.
I mentioned that you can probably leave that there for about a minute
which is kind of option, but, actually, if you are bringing it through slow enough, you can
add an additional 15 ml and then you will be ready for your sample addition.
Here we go with sample addition. This is a real quick slide, and this is
how it looks.
People have mentioned that it is real nice. You can just tip the bottle
over. You may be able to see...I am not sure if you are able to see it. I can't see it from
my angle. Maybe you can't see it, either, but there is actually water in those bottles. It
is auto-feeding. You are essentially just going to set up the manifold very much like this.
You turn over the bottle. You don't have to worry about the disk going dry.
In fact, when you get into the disk drying, you would like to leave it sit
there for about 15 minutes. It can sit there an hour, it can sit there two hours. The
dryer the disks are, the better off you are going to be, especially for these particular
analytes.
As I mentioned, we go into disk drying, and I don't have a slide for this,
but there are two means of doing it. Essentially, what we did, we first, in our first
experiments, we were allowing it to dry for about 15 minutes.
What we found with the cartridge technique is that we essentially had to
spin-dry it in a centrifuge. We were putting the 6 ml cartridges into a centrifuge and
spinning the water off.
The disks, we actually took them off the manifold, and we dried them in a
desiccator.
The problem is if you don't soak the disks to get rid of all of the
-------
147
contamination levels, when you put it back on, inevitably, you are going to end up
putting it in such a position that you are going to end up just getting a little bit of
contamination from the disks. These disks, we found, were extremely dirty, and the disk
preparation was crucial.
The elution in concentration, I have got a slide here, and I don't know how
it is going to show up. It didn't show up very well in my room, but I am going to give it
a shot. And it doesn't. What a deal.
Okay, essentially what we were doing is we were adding like a 30 ml VGA
vial to the bottom of that manifold. We essentially took off the entire apparatus. We
put that 30 ml VGA vial down below, and that is what we collected it into.
We were essentially taking 3 ml of acetonitrile three times. In some of our
experiments, we took two 5 ml aloquats, but three 3 ml aloquats, we believed from the
statistics that we read, 80 percent the first time, 80 percent of the remaining 20 percent,
and so on and so on. Worked very effective.
At this point, we are putting it on an evap type system. We are taking that
9 ml, and we are bringing it down to approximately 1 ml. At that point, when we get it
to 1 ml, we will bring it up to 1 ml with water in a 2 ml pipette.
And you have to have that type of a mixture, because if you have pure
organic, your chromatography just does not look the way it should. You get a lot of
tailing, and the compounds start running together.
These were the particular levels that we went after. We had a high spike,
we had a mid-spike, and we did have a low spike. The low spike is at the detection
limit.
I am going to show you a little later on some of the chromatography that
we have got at the detection limit. You are going to see that it is much dirtier, but we
were able to get some pretty good recoveries even at the low limit.
If anybody is familiar with Method 8330, we are approximately five to ten
times lower than the 8330 detection limits which uses essentially /a dilute and shoot type
method.
If you are able to meet those particular detection limits, this is going to be
-------
148
a godsend. You are going to get 250 times lower.
We did not try Method 8330. We are approximately, like I said, five to ten
times lower. So, it has worked very well for us.
This is what we tried with one disk. This is a 47 mm disk using reagent
water. We spiked it at the high. Those are the way our recoveries looked. As you can
see, they are very encouraging.
Lowest one that we were able to have was 85 percent, I believe, for the 2-
and the 4-nitrotoluene.
I mentioned that we did have a contaminant that ended up coming off with
the 3-nitrotoluene, and that ended up accounting for some of our high recoveries.
Especially in this 2 disk that we ended up using, notice the recoveries are a little bit
better, but take a look at the 3-nitrotoluene, 3-nitrotoluene at 200 percent. That is
primarily because of a contaminant.
This is the groundwater that we had, and I mentioned that we did do...this
was the first run that we had through it, and, actually, Craig used this as some of the
preliminary results.
We did not use this one on the manifold. This was before we got the
manifold. We just wanted to have an idea of whether this had a potential of working.
We essentially did one point at the high, and the recoveries were such that we were like
hey, bring them on, bring on the disks, let's see what we can do here.
Now, if you are like me, like I was, I am like okay, what exactly were the
recoveries like? What did the chromatography look like? Was it clean?
As you can see by the following, this is a groundwater high. Jim Madison
did an excellent job on developing the HPLC type technique. This is at 254 nanometers.
Notice that the peaks are very, very well resolved. The highest peak in the
chromatogram, not the first one which is essentially a solvent type peak, but the one in
the middle, that is TETRYL.
Right after TETRYL, you will see a little blip down below. That is
essentially your nitroglycerin. And then way on the end is your PETN.
The nitroglycerin and the PETN have to be absorbed at a different
-------
149
wavelength, and this is the exact same sample at 210. Notice again how the separation.
It really worked well for us.
Now, things weren't all glorious. Things were not all glorious, and if you
got your chromatogram and you looked like that, you know you had some problem in
your disk preparation.
This is essentially a slide of our disk preparation if we didn't do the disk
preparation. Chromatography here had a little bit to be desired.
We essentially used a cyano column for confirmation. None of these peaks
come out on the cyano column, but you have to have a confirmation type column.
And this is a slide right here that I am hoping will emphasize the
importance of the disk cleaning technique. These disks, when you first get them, are
extremely dirty, and if you can perfect the disk preparation technique, your recoveries
and the rest of it is just a dream.
This particular slide indicates what the chromatography looked like at the
detection limit. These were spiked at, if you remember the earlier slide where I showed
you the three levels, this is a surface water which has got some contaminants to begin
with right at the detection limit.
Notice the nitrotoluenes on the end. The contamination in the
nitrotoluenes was primarily due to a disk problem in my inability to actually prepare
them correctly.
We have been working at it, and we have gotten much better, and with a
little bit of time and perseverance, you can bring that contaminant peak down to nothing,
and you can essentially have real good recoveries.
That is at 254; this one is at 210. Notice that even though you have the
contamination, the nitroglycerin, the PETN still very well resolved from all contaminant
type peaks.
This is a look at a high of the surface water, and notice again here, even
despite the biological type nature of the samples that you can have in there,
chromatography is good not only at the 254 but also at the 210 nanometer wavelength.
In comparison, we have been doing the cartridge technique for a year now.
-------
150
We got into the certification of this particular method for a client that we had. They cut
off certification. We would love to go to the disks, and as soon as they open up that
certification process, you can bet we are going to be doing that. We have a real good
feel for it.
The percent d's, if you notice, very, very tight. It is almost equivalent. The
only difference...and I mentioned this earlier and I will mention it again in summary...is
the time issue. The amount of time to run these things through disks is almost nothing.
Thank you.
-------
151
QUESTION AND ANSWER SESSION
MR. TELLIARD: Any questions?
MR. LE BRUN: Not to cut anybody off on questions, I am just going to
ask...I am going to give one little bit of information. I am sure that people are going to
ask where did you get the standards.
Standards are tough to come by, and, actually, if you are not doing
explosives for a particular part of the government, they are going to be very difficult to
get. So, if that is what your question is, I can't help you.
MR. TELLIARD: Anyone?
(No response.)
MR. TELLIARD: Thanks, Gabe.
-------
UJ
152
Ss
U. Z
wO
tt
u.
_ UJ
Q>
0.
OX
< UJ
UJZ
uj O
o
c/)
<
o o
(0 Q.
O CO
-------
• 153
CO
u
QC
W
U_
o
0)
U
-------
154
o
z
o
Z - O -
o ,
o
I
O
n
z
o
z
o
O
—Z
\.
o
/
Z
I-
LLJ
D.
o
i
o
i
*•
o
CD
Z
O
CO
Z
O
o
I
""
O
o
0
-------
155
CO
Q
O
X
LLJ
g
o
OC
UJ
UJ
CO
Q.
Q
o
CO
CO
O
LU
(3
o
OC
OC
<
o
-------
156
UJ
-------
157
(0
LU
(0
LU
UJ
DC
I
DC
U
CO
o
CO
<
Q
LU
1
O
DC
O
oc
o
li-
CC
UJ
I
CO
I-
|
Q
UJ
I
UJ
2
DC
DC
O
OC
UJ
I
UJ
Si
DC
D
CO
CO
LU
D
o
CO
<
O
UJ
o
3!
-------
158
LU
D
O
o
UJ
LJJ
DC
DC
LJJ
E
CM
(0
DC
D
O
X
GO
O
-------
159
UJ
cc
UJ
O
Q
O
X
UJ
^
ffl
z
g
cc
Q.
cc
DL
CO
O
g
o
D
Q
O
CC
l~
LU
_J
Qu
^
CO
O
Z
CC
o
CO
Q
Z
UJ
o
o
o
o
z
g
3
UJ
-------
160
O
O
I
UJ
^
V
CO
Q
Z
^^1
oc
a.
UJ
oc
Q.
^
CO
Q
-^^^
• ^
O
Q
O
OC
z
""""
UJ
^k
a.
g
CO
O
^«
oc
O
CO
5
UJ
o
o
o
a
z
o
H
UJ
-------
161
_
o
h"
^^
QC
0.
u
QC
a.
CO
5
QC
5
i
uj
D
i
CO
1
UJ
•H
D
J
^
QC
^
1
UJ
i
CO
1
Z
O
^
j
E
l
u
H
Z
CO
1
X
O
G)
Z
O
a.
O
u
D
Z
S™"
1
O
CM
X
E
in
2
O
•
t
O
UJ
H
Z
1
O
CM
X
E
in
-------
162
UJ
LU
Z)
O
CO
u
O
LJJ
CO
Q
in
CM
in
(O
-------
163
cc
UJ
I
Q
O
CC
O
UJ
CO
o
Q.
X
UJ
ons
O)
I 8
UJ
I
o
_j
in
x
CM
I
O
9
O
CC
CO
Q
CO
D
CO
E
CM
a
O)
a
o
O)
01
cc
CO
o
I
>
CC
W "
8
O)
fc
SIS
CO
•a
I
I
CO
iE
c* E
o
Q)
N
O CD
k_ ,Q
•is
i^l
X x ^^
5 Q cxT co
ITS n •>.w •> »
CC -^ T- H CM CM
O
Jg
I S
II
Ps
o
o
-2 2
£ 0
25
£ Q
.£ "5
O £
+ ?
cvTcvi
1
o o
c c
o o
5 5
2 S
II
0)
o
Z Q.
I
Q.
c
*
-------
164
88588888585
Csl CM
O> O)
CO UJ
z £j
Q-1
S£
Q.
Q
CVl
C\j
N
CO
HI <
OC rn
LU
1
CO
CO
o
X
O)
CMi-
OT-
CMCD
CM
CM C\J
0)
§
DC
3
I
in
CO
O
CO
! I
'<= £
i I
CM CM
V
I
I
CM
1
o
I
Q
CO
c
I
I
-------
165
CO
5
UJ
OC
o
Q.
U
8
LL
CC
*
Q
U
CC
O)
8 5
S
o
a
CC
in
CO
si
CO N
CM 5
CD
CO
co co o> ^F co
if i- CNJ CM CM
o
CO
111
CL
CO
o>
CO
CO CO
if! ^
CM CM
GO
N. T- T-
S3 5
CO
S SI Ki
X
I
RDX
O
^^
135 Trinitrobel
0
o
13 Dinitrobenz
TETRYL
0
246 TrinitrotolL
o
c
26 Dinitrotolue
o
24 Dinitrotoluel
2 Nitrotoluene
4 Nitrotoluene
3 Nitrotoluene
I
I
-------
166
Q
\Jw
^^H
Q
CM
i
W
DC
O
Q.
LU
Q
(/)
cr
I
v^
LI
DC
GO N CO
85 co N
Q
LJJ
DC
£
0
o
LLJ
DC
H
<
O) -r-. T-;
CM
CO
S CO
O5
i^
CO CM T-^ ^f ^
CO CO f^ ^3 ^^ QJ
^r ^r CA| c^i ^^ T™
H
2
C
u.
^v
CO
0) CO CO
O) O) LO
^- ^- CM
/i\
>*/
C
/l\
ffi
IN
2
.t±
x „ £
i i s
§
GO
CO CO N i- T-
^ 0) CO
CM O) CM
Q)
C
O
13 Dinitrobenz
o
o
N
TETRYL
246 Ttrinitrobe
T—
in
o
c
o
26 Dinitrobenz
S3
o
C
O
24 Dinitrobenz
CO
CO
s
N
S
2 Nitrotoluene
CO
CM
in
q
CM
CM
4 Nitrotoluene
00
Oi
CO
s
CO
^0
CM
3 Nitrotoluene
^rt
00
CM
CM
1
Nitroglycerin
8
CM
-------
167
o
i
\
r?
o
S-
CJ3
' 0 _ 3J
XWH
-------
TQ'T7 - -'
168
aO 0 i ~
• ,' T - .
7
0 ic' 0 i — -_
? -
RTi • /* '
-» JN —
C
3
O
s_
CT5
CS ' G -"'^
-------
as 0>
169
Cc, ~
iwr
IITU
Pi-
:0T-«tL
nv Q -
-------
170
r
V
V
^.
1N£
>
MM
I
I
** swiser
XXH
-------
171
\
wsi-A.
m
i
i
-i
i
J
j
_c—-' j M9E1
-
W r ->_
!
---"'I i
r--w i TUTO*.!fi
^-1 d.-!^ i _' V •_•'
( !
i —<
88'8-n
V.
i
.-,-^
"7
cu > !
o wf i
fO ——- ,' —'
M- -r— / XffH)
s_ "••
00 ^ ^
X 1
.11. ' A —f VMIl!
-'
-------
172
-------
173
"~— — v
'AT _— — ~~ \
4
IMb
Bb!TT-y
-I
j
I
-I
O)
98:P
80'S-i
-------
174
->\
Uj.ll
p r • n T _—-
en
CJ
03
3:
oo
'T
E/.'fi-
PT'TT -
- - •¥ "/
06 b'-:
CT ' ft - »tf -
TTTr—r:
V
/
98'TT-f?
-------
175
O
UjCO
§1
58
I
gs
"3
2
Q cc
CO
S
LU
QC
O
8§?
cs
3
gl
X x ^^
2 Q crT co
I DC T-" T-"
o
O o o
^ C C
o o o
"5 5 5 ° ° °
.fe 8 3 i S o £
.£ 2 2 -2 .2 .2 o
£ •« I 3 S 8 3
t— .£ .£ o o o o
CD Q Q ±= .-fe .-e O
«•" ci 4 Z z z .fe M
CM" CM CM" CM 4 A z S!
-------
176
[Blank Page]
-------
177
MR. TELLIARD: As we have heard this morning, solid phase extraction
offers a lot of flexibility in the laboratory. One of the questions that we are faced with is
an ongoing issue of taking samples...and some of you have heard rumors...taking samples
and sending them to the lab all at once. This is a rash lie.
We send three or four a week all year long just to make them happy.
However, in the summertime, we like to take a little more samples, because the water
i
isn't frozen, and we ship them off to the lab, and there is sometimes a protect about
getting 122 samples and having a 45-day turnaround time.
This shows a lack of real initiative and imagination on the laboratory's
part. Some people would consider this a problem. Other people with more initiative
would consider it a challenge. I generally get the guys who don't consider it a challenge.
One of the issues that we were looking at was the viability of using solid
phase extraction for such things as field application and, for example, shipping a solid
phase to the laboratory rather than 52 1-liter bottles in an ice chest rattling around
Federal Express.
The other issue was, of course, when you get these high numbers of
samples into the laboratory, could we extract them and then hold the samples on the
filters and analyze them, as they usually do, at their own damn well rate, you know.
So, one of the issues that had come up was stability and storage of these
samples as it relates to field application, and our next speaker is going to discuss that,
Scott Senseman from the University of Arkansas.
-------
178
STORAGE STABILITY OF SELECTED PESTICIDES
ON MEMBRANEOUS SOLID PHASE EXTRACTION DISKS
S.A. Senseman, J.D. Mattice, T.L Lavy, B.M. Myers, and B.W. Skulman
INTRODUCTION
The storage stability of various pesticides in water for prolonged storage
periods is a concern of many environmental laboratories. With increasing
numbers of samples being collected, it becomes important not only to be
concerned with storage stability but storage space as well. With the emergence
of solid phase extraction, alternative methods to resolve these problems may be
available by concentrating the pesticide on a solid phase extraction disk directly
after collection of a water sample.
Earlier applications of solid phase extraction in the cartridge form were
shown to be successful, although limitations such as slow flow rate and bed
channeling arose. To alleviate some of the limitations, more recent technology
has been introduced that involves similar solid phase material but contained in a
membrane filter or disk form. Figure 1 is a conceptual picture of a C-18 47 mm
diameter Empore™ disk that was used as the storage media in our experiment.
The disk consists of a teflon fibril network that suspends silica particles. These
silica particles have carbon chains emanating from them which creates a non-
polar environment. This non-polar environment is very conducive to "dissolving"
non-polar pesticides. Also, by holding pesticides in the solid phase, it is
1
-------
179
possible that some protection from hydrolysis and microbial decomposition over
long storage periods might be achieved.
Consequently, by concentrating the pesticides on the disk, the problem
of storage space could be greatly reduced in the laboratory while also giving
rise to substantial reductions in mailing costs of analytical samples being sent
from one laboratory to another. Past studies with SPE cartridges encourage
this approach and have shown that some hydrocarbons were more stable when
residing in an organic matrix (Green and Le Pape, 1987). However, the limita-
tions of the cartridges influenced further studies with membranous filters.
The objectives of this experiment, therefore, were to compare the relative
storage stability of some selected pesticides on solid phase extraction disks to
the storage stability of pesticides in water and also to determine the chemical
stability of these pesticides on solid phase disks under various temperature
storage regimes.
MATERIALS AND METHODS
General. The fortification solutions were prepared by dissolving all of the
pesticides of interest in, methanol with the exception of captan which was
dissolved in benzene for stability reasons. Prior to initiation of the experiment, it
was determined that captan was unstable in methanol and is also subject to
hydrolytic attack when stored in water. Consequently, it was decided that
-------
180
benzene would provide a more stable chemical environment for captan. A 2500
/^g/rnl solution of captan in benzene was made and then added to water
samples in a small quantity of 2 pL such that the concentration of benzene
would be insignificant. The final fortified water sample contained 250 mis of
deionized water, 1 ml of methanol with dissolved pesticides, and captan
contained in 2 /iL of benzene. All of these pesticides were fortified at 20 /*g/L
in water.
Storage treatments. The four storage treatments were replicated four times
and included storage of pesticides in amber glass jars and refrigerated at 4 C
representing the common method of storage for water samples. This treatment
was compared to three disk storage treatments in which the water sample was
fortified, filtered through a solid phase extraction disk, followed by storage of the
disk at a predetermined temperature. The three storage regimes for the disk-
stored treatments included frozen at -20 C, refrigerated at 4 C, and a combina-
tion of refrigerated for 24 hours then frozen at -20 C for the remainder of the
storage period.
The rationale for the frozen regime was that the temperature was consid-
ered to be optimum for chemical stability. The disks stored at 4 C would show
not only a comparison of temperatures on stability, but also a direct comparison
between disk storage and storage in water. Including the combination of
refrigerating and freezing treatments of the disk was representative of what a
-------
181
water quality researcher might attempt when extracting pesticides from water in
the field. This system might begin by filtration of the water sample to load the
pesticides onto the solid phase allowing for temporary storage of the disk in an
envelope at a cool temperature (approximately 4 C) until it could be more
permanently stored at the laboratory in a freezer (-20 C).
All pesticide loaded disks were stored in plastic bags until the storage
periods had expired. Upon expiration, the disk was taken out of the plastic
bag, replaced back onto the filter apparatus on which the pesticides were first
loaded, then, solvated with ethyl acetate which extracted the pesticide from the
solid phase. The ethyl acetate solution was then used to further quantify using
gas chromatography (GC) and liquid chromatography (HPLC). The pesticides
stored in water were extracted at the termination of the storage period in the
manner described above except the disk was not removed and stored after
filtration of the water sample but rather immediately solvated with ethyl acetate
after concentration onto the solid phase.
Storage periods. Five storage periods were included in the study including a
time-zero where pesticides were extracted immediately after pesticide loading
thus represented the extraction efficiency. The remaining storage periods
included 3, 30, 90, and 180-day incubations.
Pesticides. The twelve pesticides included in the experiment are shown in
Table I with retention times and method of quantification. The pesticides listed
-------
182
cover a wide range of chemical families and were chosen because of their
previous detections in surface or ground water. Also, it was known apriori that
trifluralin and captan were relatively unstable under certain environments and
that a direct comparison of disk storage versus water storage would be infor-
mative.
Statistical evaluation. The mean percent recoveries of the four treatments by
five storage period factorial design from GC and HPLC analysis were statistically
separated by the Least Significant Difference (LSD) at a 0.05 probability level.
-------
183
Table '• Retention times (min) of compounds quantified in storage study.
GC-ECD
f^r\\t inr»rt
Compound
alachlor
atrazine
benomyl
captan
fluometuron
methyl parathion
metolachlor
norflurazon
pendimethalin
profenofos
simazine
trifluralin
SPB-5 SPB-608
5.1
ND
ND
7.9
ND
4.8
5.9
16.9
6.5
11.0
ND
2.5
— retention time,
6.1
ND
ND
14.7
ND
7.1
8.4
24.9
10.7
15.9
ND
2.2
HPLC-UV
VARIABLE
C-18 Column
min
ND
8.4
2.5
ND
7.1
18.5
ND
ND
ND
ND
4.5
ND
-------
184
RESULTS AND DISCUSSION
Benomyl, fluometuron, and atrazine showed no interaction between
storage treatment and storage period (Figures 2, 3, and 4). In general, the
highest percent recovery for these pesticides occurred when the disk was
stored at -20 C; although, these recoveries did not always differ statistically from
the other disk storage treatments. In all cases each pesticide stored at -20 C
on the C-18 material gave better recovery than those stored in bottled water
(Figures 2, 3, and 4). Moreover, the lowest recovery was for the pesticides
stored in bottled water, but it was not always statistically different from the next
lowest value within a selected disk storage treatment. The combination treat-
ment gave the second highest percent recovery for each pesticide with the
exception of atrazine (Figure 4). Atrazine showed no differences among disk
storage treatments indicating that temperature treatment of the disk had no
effect on the recovery of this compound. Statistical differences for pesticide
recovery from disks were shown only for benomyl and fluometuron of the three
pesticides. Thus the consistent differences in percent recovery do not occur
between specific disk storage treatments among the pesticides tested. Never-
theless, these data show that these pesticides are at least as stable and often
more stable when stored on solid phase material as compared to storing the
pesticides in water over the same duration.
-------
185
Trifluralin, alachlor, methyl parathion, metolachlor, pendimethalin, norflur-
azon, captan, and profenofos showed a significant interaction between storage
treatment and time interval. The examples shown in Figures 5 and 6 are
alachlor and trifluralin data after 180 days of incubation and are representative
of data of the other compounds. After six months, alachlor and trifluralin were
fairly stable and recovery significantly higher when stored on the solid phase
disk. In most cases, the disk storage of pesticides was equivalent or superior
to pesticide storage in water.
Two distinct cases of statistical differences between disk storage and
water storage occurred in the trifluralin and captan data (Figures 6 and 7). The
various storage treatments of trifluralin did not demonstrate any differences until
the 180 day storage period. After 180 days of storage, there was 32% recovery
with trifluralin when stored in bottled water while recovery from solid phase
storage ranged from 54 to 64%, representing about a 50% loss of pesticide
when stored in water compared with disk storage. This loss would double the
sensitivity and detection limits of samples stored over this duration. Similar
trends of pesticide loss were shown for alachlor, metolachlor, methyl parathion,
pendimethalin, norflurazon, and profenofos, however, over the 180 days of
storage in water their loss was not as great as found for trifluralin.
Captan demonstrated the most dramatic results (Figures 7-9). The
recovery from water stored for 3 days at 4 C was 28% whereas the recovery
8
-------
186
from disks stored at 4 C was 114%. The differences were more pronounced
after 30 days when captan had all but completely dissipated in water while 32 to
54% was recovered from disk storage (Figure 8). This stabilizing ability of the
C-18 material has been observed by other researchers who stated that materi-
als bonded to a solid phase were more stable (Green and Le Rape). It is also
apparent from this data that disk storage does not totally solve the stability
problem of captan. Observation of the 30 and 90 day results for this compound
display significant loss of the parent captan over that time span even when it
was stored on the non-polar media of the solid phase (Figures 8 and 9). This
loss may be due to hydrolysis by water that could not be totally removed from
the disk by vacuum filtration. Therefore, captan would still be in an aqueous
environment and susceptible to hydrolytic attack even while it resides in the
non-polar matrix of the C-18 material. It may be possible to further stabilize
these compounds by removing the residual water through desiccation, lyophili-
zation, or blotting with anhydrous sodium sulfate prior to permanent storage.
Other limitations to disk storage may include microbial growth on the
disks over longer storage periods. At the 90 and 180 day storage period, disks
stored at 4 C exhibited some microbial growth. This may have been a factor in
the slightly lower recoveries from these treatments. Microbial growth was not
observed on the frozen disk-storage treatments, therefore, supporting the
advantage to freezing the disk after loading the pesticides.
-------
187
SUMMARY
The stability of these pesticides has been preserved and in most cases
enhanced by concentrating the pesticides on C-18 material. Trends appear to
favor storage of pesticides on the disk by freezing after extraction. These
results offer promising possibilities that could alter the way water samples
containing these pesticides are currently stored. The pesticides were both
stabilized and extracted by the non-polar media of the disk. Therefore, water
samples that formerly occupied the space necessary for 500 to 1000 ml bottles
may be reduced to a 0.5 mm thick X 47 mm diameter pliable filter. The reduc-
tion in storage space is clearly and quickly realized. Field extraction using SPE
disks appears to be the next logical step in developing and utilizing the potential
of this technology. Through field extraction, pesticides could be both concen-
trated and stabilized on a disk while in the field. The disk could then be stored
rather than a bulky glass jar. While increased time would be required for
sample collection a part of the normal laboratory extraction procedure would
have been performed in the field. Consequently, sample preparation time in the
laboratory would be decreased, thereby, allowing a quicker overall analysis
time. In addition, transport of pesticide samples through the mail from one
laboratory to another may be easier and cheaper if disks are used as the
storage container rather than bottles. Drawbacks of the this type of storage
scheme, such as hydrolysis found in the case of captan, must be studied more
10
-------
188
completely. Further studies need to explore more complete removal of water
from the disks as a means of stabilizing the bound pesticides.
Acknowledgments.
The author would like to express his appreciation to Dr. Craig Markell of
the 3M Corporation for his technical support in this endeavor. A special thanks
goes to Mr. Benjamin Myers for his dedication and input that was so crucial to
this project.
11
-------
189
LITERATURE CITED
Green, D.R.; Le Rape, D.L. Anal. Chem. 1987. 59, 699-703.
Hagen, D.F.; Markell, C.G.; Schmitt, G.A.; Blevins, D.D. Anal. Chim. Ada.
1990. 236, 157-164.
Markell, C.G.; Hagen, D.F.; Bunnelle, V.A. LC/GC. 1990. 9, 332-337.
12
-------
190
0 . 5 mm
C-18 chains attached
to silica particle
Silica partial
47 nun
Figure 1. Conceptual drawing of C-18
Empore™ disk.
-------
191
% Recovery
Diak -2OC
Disk 4C Disk 4C,-20C
Storage Treatment
Bottled 4C
mxn
(•'••'••••••••'••t Benomyl
Figure 2. Comparison of storage treatment on stability of
benomyl.
-------
192
% Recovery
Diak -2OC
Diak 4C Diak 4C,-20C
Storage Treatment
•Hi Fluometuzon
Bottled 4C
Figure 3. Comparison of storage treatments on stability of
fluometuron.
-------
193
% R«cov«ry
Disk -20C
DiBk 4C Diak 4C.-20C
Storage Treatment
Bottled 4C
Figure 4. Comparison of storage treatment on stability of
atrazine.
-------
194
% Recovery
Figure 5. Comparison of storage treatments on stability of
alachlor after 180 days of storage.
-------
195
100
80
6O
40
20
O
% R»cov«ry
73
_J i
\ :
: 1
*
Non -incubated Dick
k
\ i
14
i •; ; . ;
LSD
4
- 6
3
1 11
Storage Treatment
Illlllil Txifluralin
2
1 111
Niii
Figure 6. Comparison of storage treatments on stability of
trifluralin after 180 days.
-------
196
% R»cov«ry
Non-incub*t»d
Disk 4C Disk -SOC Disk
Storage Treatment
Captan
Boeti»d «c
Figure 7. Comparison of storage treatments on stability of
captan after 3 days of storage.
-------
197
% R«cov«ry
Cap tan
Figure 8. Comparison of storage treatments on stability of
captan after 30 days.
-------
198
% Recov«ry
Captan
Figure 9. Comparison of storage treatments on stability of
captan after 90 days.
-------
199
QUESTION AND ANSWER SESSION
MR. BICKEVG: Merlin Bicking from Twin City Testing.
Two questions. Particularly in trifluralin after 180 days, you had 50 to 60
percent left. My first question is, did you see a steady linear decrease over time, or was
it exponential? Did you see a rapid drop-off at the beginning?
And the second question is it appeared that the data labeled disk minus 20
which was, I think, frozen immediately always seemed to be lower than, for example, the
one that was stored at 4 degrees centigrade for a day and then frozen. Is there any
reason for that?
MR. SENSEMAN: There was an exponential decrease in trifluralin over
the storage period in both water and when stored on the disk but the data has not been
presented in terms of kinetics. The reason is because the main purpose of the
experiment was to study relative stability not necessarily kinetics. However, the data is
such that it could be used for a kinetic study.
MR. EPSTEIN: Paul Epstein, National Sanitation Foundation. Did you
look at all at room temperature stability on the disks?
MR. SENSEMAN: No, we didn't. That is a treatment that should be
studied. We felt that, in general, it would be just as easy, if we had the pesticides stored
on a disk, it could very easily be stored in a cold room and, generally, be more optimum
as far as stability of some of these pesticides is concerned.
MR. EPSTEIN: That is true, but if you have hundreds of samples in the
laboratory, even the disks take up a lot of refrigerator space.
MR. SENSEMAN: Yes. No, we did not involve room temperature, but
that could be future treatment in a similar study.
MR. EPSTEIN: Thank you.
MS. NOLAN: Lydia Nolan of Supelco, Incorporated.
My question is you alluded at the beginning of your talk to the fact that a
lot of these studies had been done in years past with cartridges also, and I am wondering
-------
200
if you are aware if the EPA is considering ever including this into their field
methodology. Are they really going to take this seriously?
MR. SENSEMAN: I'll have to direct that question to Mr. Telliard.
MR. TELLIARD: As we build a data base, the answer is yes, if we can
make it so that it cuts down on the shipping cost which, in our case, is quite extensive.
When you go out to do a facility, you have 20 ice chests that you are mailing around the
country. If you can get down to two boxes, that is going to be a significant cost savings.
For NPDS purposes for permit compliance, if the guy can do this at the
facility, put it in a little canister, and mail it to L.J. Slink & Associates to run the
analysis, that is going to be a cost savings.
So, as we build this data base and as we build the information, I think solid
phase extraction, certainly as far as we see in drinking water, is applicable, and in
wastewater, as we build a data base, will become more and more so.
MR. JUNK: Greg Junk, Iowa State University.
Relative to the previous question regarding stability at room temperature,
we had very limited results with organophosphates which are very labile kinds of
material. Phenotrion I remember specifically. At room temperature, we were able to
preserve them for long periods of time, but the speaker mentioned something that is
critical.
If that cartridge and/or disk is completely dry, there are good theoretical
reasons for most of the lability that exists for these pesticides being some sort of
sovolysis motivated. So, it is important for it to be perfectly dry.
But I do believe it requires further study, because if you don't have to
refrigerate it, particularly in shipment, this is a very definite advantage.
MR. TELLIARD: George:
MR. STANKO: George Stanko, Shell Development Company.
The experiments that you did were with compounds that probably would
not react with each other when they were next to each other. I would like to remind Bill
back to Conference 2304. When we tried to do this with tenax traps on the purge and
trap devices, we had very serious problems with the compounds reacting, disappearing,
-------
201
forming humpograms.
In the case of 8270 type methodology where you had bases and acids, when
you run them through this cartridge and you put them close to each other, I think you
are going to have some real storage stability problems.
For pesticides, this probably might work, but for the rest of the things, I am
not so sure it will.
MR. TELLIARD: We agree, George, and that is why we are trying to do
some of these things to take a look at it. Again, too, remembering that it may not be...it
is not a total panacea, but it may be, certainly, a field application, particularly if you
have a permit where you have one or two or three analytes and you are not looking for a
whole recipe here, that you could run it through and ship it off.
In our case, we were looking for everything creating by God and man. It
sometimes creates a bigger problem. We agree.
MR. TELLIARD: Thanks, Scott.
-------
202
[Blank Page]
-------
203
AFTERNOON SESSION
MR. TELLIARD: Continuing on with the same theme, we are going to
discuss solid phase extraction and applications of the system. Our next speaker is Marie
Brinkman from Battelle, Columbus.
MS. BRINKMAN: Before I start, I just want to make sure that everybody
can hear me through the microphone. Can you guys hear me in the back? All right.
I also want to say that I had the bad form to have vertical slides, so what
you are going to see on the screen is a little bit smaller version than the slides you saw
this morning, because Randy needed to change the lens in order for us to see the full
vertical slide. So, I apologize for that. If you all want to change your seats and come up
to the front, feel free.
I also wanted to point out that everybody has been saying solid phase
extraction, and I just wanted to mention that liquid-solid extraction is the same thing.
We just call it liquid-solid extraction.
If I could have, let's see, the first slide here. I hope the people that
expressed concerns about phenol recoveries are still here. I will be discussing a new
liquid-solid extraction anion exchange technique for the extraction of phenols and acids
from water that we developed in cooperation with USEPA-EMSL in Cincinnati.
This method includes the use of anion exchange resin and a novel
•derivatization method. So, I will also be giving some background information on both of
those.
Our program goals are listed here. Our first two goals, reduce volumes of
toxic solvents by applying liquid-solid extraction, tie in with the theme for this year's
conference, pollution prevention in the laboratory. Our next two goals were to develop a
single method for the analysis of diverse phenols and to achieve sensitive detection
appropriate for drinking water levels.
Some of the concerns that we were faced with about the established
methods included the fact that phenols are poorly extracted from water using liquid-
-------
204
liquid extraction techniques. For example, EPA Method 625 and 625.1 which, as you
know, are liquid-liquid extraction techniques, give recoveries for phenols as low as 25
percent.
Also, those procedures use about 400 to 500 ml of methylene chloride per
sample, and non- or singly-chlorinated phenols, even when methylated, are detected on
the order of 0.1 ng/ml, and we wanted to detect phenols at levels 10 and 100 times more
sensitive than that.
Hence, our approach, liquid-solid extraction with anion exchange and
pentafluorobenzyl bromide, or PFBBr, derivatization and GC/ECD.
Liquid-solid extraction is not new. We have been hearing about it all
morning. There are many commercially available cartridges, and as discussed this
morning, disks that contain different liquid-solid extraction media, including CIS and
silica.
However, analytes can desorb or partition off of these media after large
sample volumes are passed through the cartridge or over the disk. This behavior is
especially common for phenols, because they are so soluble in water. Therefore, they
are not strongly retained by these materials.
This phenomenon was discussed in a couple of the talks this morning.
Anion exchange solves this problem, because the phenols are retained by a
chemical bond to the resin. This reaction can then be reversed in order to elute the
analytes from the column.
One of the strong points for our method and, in fact, all liquid-solid
extraction techniques is that minimal quantities of toxic solvents are used. We were able
to reduce the amount of organic solvent used in our method to less than 20 ml per
sample.
And by using PFBBr derivatization, we were able to detect analytes at the
1 pg/ml level.
The first important aspect of our method is the anion exchange resin itself.
This slide is divided into three panels to show what is happening chemically between the
AG MP-1 resin and the analytes during retention and elution from the column.
-------
205
The resin has a polymeric backbone that is a styrene divinylbenzene co-
polymer which is represented in this slide as an orange wave. However, the AG MP-1
resin has two distinct features, the chemically bound quaternary amine groups and an
exchangeable anion which is the hydroxide bound to the amine.
We initially purchased the resin in the chloride form. We then exchanged
the resin from the chloride to the hydroxide form to produce the strongest possible
aqueous base, and that state is shown in the first panel in the slide.
As the water sample containing the phenols passes through the resin in the
column, that hydroxide anion abstracts a proton from the analyte, and this results in the
attachment of the analyte anion via an ionic bond to the quaternary amine functional
group.
You can see in the middle panel that the phenol has lost its hydrogen and
is now ionically bonded to the resin.
Neutral and basic molecules may be slightly retained by the polymeric
backbone, but they can then be eluted using neutral solvents without removing the
retained phenols.
Now, to elute the phenols from the column, a strong acid...and in this case
we used hydrochloric...will displace the analyte from the resin. The phenols are eluted
from the column and then collected as shown in the last panel there.
I would like to point out that this slide shows phenol as the only analyte
for simplicity's sake only. The resin retains many types of phenols and carboxylic acids,
and I will list the analytes we tested for this program a little later.
The second important aspect of our method, pentafluorobenzylbromide or
PFBBr derivatization, is not new. It has been reported in the literature many times.
PFBBr reacts with phenol in the presence of potassium carbonate and heat
to produce the derivatized analyte, thereby attaching five BCD-sensitive fluorine groups
to the analyte of interest.
And, again, only phenol is shown for simplicity's sake. The bottom of the
slide shows an ECD chromatogram of the diverse phenols derivatized with PFBBr for
this program.
-------
206
With those two main concepts introduced, I move on to our program
design. Looking back to the derivatization reaction shown in the previous slide, we
evaluated the effects of reaction temperature and duration, potassium carbonate
concentration, and trace amounts of water in the reaction solution on the formation of
PFBBr derivatives.
In addition, we evaluated the strong anion exchange resin, AG MP-1, for
collection of phenols and acids from water. Because AG MP-1 is an organic resin, we
needed to evaluate methods to remove residual organics from the resin prior to its use.
We were then able to measure recoveries of analytes spiked into 1-liter
and 100-ml volumes of water.
These are the compounds we evaluated in this study. We included both
weak acids, the alkylphenols, and strong acids, dichloroacetic acid and 2,4-D. We also
evaluated mono through pentachlorinated phenol.
This slide shows a schematic representation of a typical experiment.
First, we suspended the resin in the chromatography column. The resin
was then exchanged from the chlorine to the hydroxide form using a sodium hydroxide
solution. We then put the resin through the washing procedure to remove any residual organics.
We exchange and wash the resin after it is already in the chromatography
column to minimize passive sampling.
The syringe in the slide represents the spiking of 0.1 mg per analyte which
we spiked into varied water volumes for various experiments, 30 ml, 100 ml, and 1 liter.
The resin itself is pictured in the right-hand insert.
Once the aqueous sample is passed through the column and the analytes
are then chemically bound to the resin, we come to the analytical procedure depicted in
this slide.
I realize this is a complex slide with arrows going all over the place, so I
will start at the upper right-hand corner and go slowly.
Again, the sample has already been passed through the column, and the
analytes are now bound to the resin. We then eluted the analytes from the resin using a
2 percent hydrochloric solution of methanol/methylene chloride into a separatory funnel.
-------
207
The organic layer was then partitioned against an acidic aqueous solution.
The aqueous layer is discarded, and the organic layer containing the analytes is dried
using a sodium sulfate column. It is then concentrated to 1 ml by Kuderna-Danish, the
internal standards are added, and the extract is separated at this point.
900 ml branching to the left is solvent exchanged to methyl t-butyl ether,
methylated with diazomethane, and analyzed for acids using GC/ECD. 100 ml
branching to the right is solvent exchanged to acetone and diluted to 1 ml.
This is then derivatized using PFBBr and analyzed for phenols using
GC/ECD.
To isolate the issues, we focused on one pathway at a time per sample. I
will explain later why the extract is split and one derivatization technique is used for
acids and a different derivatization technique is used for phenols.
This graph shows analyte recoveries for 1 mg/analyte spiked into two
different water volumes. In general, recoveries are greater than 75 percent.
These results also show that the sample volume passed through the resin
does not affect analyte recovery. This is because of the ionic bond formed between the
resin and the analyte.
In fact, the recoveries are a little higher when the analytes were spiked into
1 liter of water, because silanized glassware was used for that experiment.
Because this was our first evaluation of the resin, we chose a spike level
amenable to methylation and GC/FID. Later, I will show results for lower phenol spike
levels using PFBBr derivatization and GC/ECD.
I should mention that each experiment was performed in triplicate and
included method blanks.
As I said before, numerous PFBBr derivatization methods have been
reported in the literature for selected phenols and acids. Our experience with these
methods is summarized by the top chromatogram which shows numerous artifacts in
addition to the three derivatized analytes highlighted in pink.
The bottom chromatogram demonstrates that it is possible to achieve
similar derivatization efficiency of those same three analytes without excessive artifacts.
-------
208
I should point out that the GC/ECD temperature programs are different for those two
chromatograms, and that is why the analytes don't exactly line up.
The original PFBBr derivatization procedure that we developed included
four steps. We start with 1 ml that contains all of the phenols and acids in acetone. At
the completion of step 2, the acids and pentachlorophenol are the only species
derivatized, and they are shown in the top chromatogram.
At the completion of step 4, phenols are derivatized, as shown in the
bottom chromatogram. However, the highly alkaline conditions of steps 3 and 4
hydrolyze the acid derivatives, so these species are no longer detected at this point.
Here we show that even at one of our lowest concentration levels, the 2.5
pg/ml per microliter standard shown in the bottom chromatogram, derivatized phenols
are still discernable from minor BCD detectable artifacts. The analytes are highlighted
in the pink rectangles, and this unhighlighted version allows you to see the baseline a
little bit more clearly.
These are examples of the calibration curves for one of the analytes,
pentachlorophenol, in the derivatized mix of 20 phenols and acids. Note the linearity
over the range of 1 to 100 pg/ml per microliter. All phenols and acids in the derivatized
mix exhibited similar linearity.
Now that I have shown you our standards, this is a chromatogram of the
phenols recovered from water using the anion exchange resin and subsequently
derivatized with PFBBr. As you can see, the resin does yield some other detectible
species. Remember, it is an organic resin. Yet, these species do not interfere
chromatographically with the analytes.
Again, the analytes are highlighted in pink here, and this unhighlighted
version lets you see the baseline.
This chromatogram shows the recovery of the acids from water using the
anion exchange resin. The blank below shows minimal interferences and artifacts,
however, 4-chlorobenzoic and dichlproacetic acid, had to be blank subtracted for the
analysis shown here.
These analytes were methylated prior to analysis; they were not derivatized
-------
209
using PFBBr.
Although we were able to derivatize acid standards using PFBBr, we were
not able to apply this technique successfully to resin extracts. We believe that this is due
to small quantities of HC1 that are carried through from the final partition step in the
separatory funnel.
This acidic environment did not appear to adversely affect methylation.
For this reason, we split the extract, and the two derivatization techniques are applied
for acids versus phenols.
In subsequent work, we have been able to identify slightly different PFBBr
derivatization conditions to compensate for this analytical procedure.
Shown here are the overall method recoveries for the phenols spiked into
water. These data represent recovery of 0.1 mg spike with PFBBr derivatization and
GC/ECD. The reason why we say simulated 0.1 mg/L is because we spiked 0.1 mg of
material into 30 ml or 100 ml volumes rather than a liter. This was done for
experimental efficiency.
Remember, the analytes are retained by an ionic bond to the resin, and
our experimental results, presented in a previous slide, showed that recoveries are
comparable whether analytes are spiked into water volumes of 100 ml or 1 liter.
These data show that overall recoveries are good, generally better than 70
percent. The exception is 1-naphthol, which is on the far right side there, which was
slightly less than 50 percent. We believe that that was due to its chemical instability.
I would also like to point out the large deviation for phenol, which is the
farthest bar on the left, was due to phenol contamination of the laboratory air from
roofing tar being applied to the roof near our hood exhaust. We would expect the
phenol standard deviation to be similar to the other analytes under normal conditions.
We would like to define and use a new term for the evaluation of liquid-
solid extraction media. We call this term LSE efficiency, and it describes the efficiency
of the collection onto and the subsequent release off of the resin.
This number isolates and defines the performance of the resin alone. This
involves isolating a portion of the analytical method which we call the post-resin method
-------
210
which is shown here in the green box.
In the post-resin method, the analytes are not spiked into the water sample
but directly into the separatory funnel and then carried through the rest of the analytical
procedure.
In the analytical method which is shown here in the purple box, the
analytes are spiked into the water sample. The sample is run through the resin, the
analytes are eluted from the resin into the separatory funnel, and on down through the
rest of the post-resin method.
To calculate an analyte's LSE efficiency, we divide its recovery value from
the entire analytical method by its recovery value from the post-resin method, and
multiply by 100 to get a percent value. An analyte's method recovery is the same as its
recovery from the entire analytical method.
If the method recovery for an analyte is low, the next number to look at is
the analyte's LSE efficiency. If the LSE efficiency is high, that means that the resin is
performing well and that losses are due to problems with the extractions or
derivatizations in the post-resin method.
However, if the LSE efficiency is low but the post-resin method recovery is
high, then that indicates that the resin is not retaining and eluting that analyte efficiently.
This graph shows the calculated LSE efficiencies for the phenols. They are
very high, averaging 92 percent. These data show that the AG MP-1 resin collects and
releases these analytes efficiently.
This slide shows the overall method recoveries and the LSE efficiencies for
the acids and two selected phenols. The post-resin method recoveries were nearly 100
percent for all of these acid analytes.
Because overall analytical method recoveries averaged about 80 percent,
we calculated LSE efficiencies that averaged 80 percent. It appears that the AG MP-1
resin's performance is slightly lower for the retention and elution of the acids.
Our data from this program for liquid-solid extraction with anion exchange
resin is listed on the left-hand side. Published data using liquid-liquid extraction
techniques are listed on the right side.
-------
211
For the acids, our recoveries are comparable, and our spike levels were
approximately the same. For the phenols, our recoveries are, again, comparable, and our
spike levels were almost 100 times lower than those used for the published data.
In conclusion, our method recoveries are comparable with those given
using liquid-liquid extraction techniques, and our method uses less than one-tenth the
amount of toxic extraction solvent used by liquid-liquid extraction methods. Also, PFBBr
derivatization gives more sensitive detection and will allow us to detect phenols at
drinking water levels.
Thank you. Any questions?
-------
212
QUESTION AND ANSWER SESSION
MR. TELLIARD: Bill?
MR. BUDDE: My name is Bill Budde. I work for EPA in Cincinnati.
I want to congratulate the speaker for using the terminology liquid-solid
extraction which, I believe, is a slightly more scientifically correct than the equally useful
solid phase extraction.
I would also just like to point out...I don't have any questions, but I wanted
to point out that this work which was sponsored by EMSL-Cincinnati is continuing, and
one of our goals is to eliminate entirely the use of solvents such as methylene chloride,
and one of the new approaches that we are looking at now is to use supercritical carbon
dioxide to elute the cartridges and disks and resins.
Thank you.
MS. BRINKMAN: Thank you.
MR. TELLIARD: Anyone else like to make an advertisement?
MR. BUDDE: We hope to report on that result with supercritical carbon
dioxide at the SFE meeting which is later this month in Cincinnati. If anyone would like
to know about that, I will be glad to tell you.
MR. MCCARTY: Harry McCarty, Viar and Company. I don't have
anything to advertise.
Two questions. What were you using for internal standards, and how many
relative to the number of analytes you had?
MS. BRINKMAN: The number of analytes totalled 18, and we had 2
internal standards, 4-chlorobenzoic acid for the acids and 3,4-dimethylphenol for the
phenols.
MR. MCCARTY: As I understood it, you added those after you went
through the resin and eluted it off. Would the addition of the internal standards prior to
passing it through the anion exchange resin eliminate the need to calculate the efficiency
factors? I mean, basically, you have some correction for the loss that was going on in
that efficiency process.
-------
213
MS. BRINKMAN: That is a very interesting idea, and we refer to that as a
surrogate, and that sounds like an excellent experiment to do. We didn't do any work
doing that, though.
MR. MCCARTY: Well, I am suggesting that you not use it as a surrogate.
A surrogate is what you would put in and say okay, we got this, we probably got as much
of everything else.
If you put it in and you use it for quantification, if it was a labeled
compound, you would call it isotope dilution, but what I am suggesting is it might
eliminate the need to even worry about the efficiency of the LSE itself and separate out
that step which might save, looking at other compounds, it might save a lot of time. All
you would need to do is find a compound that resembles the ones you are looking for.
MS. BRINKMAN: It is true calculating the LSE efficiency does involve
two separate experiments. Thank you.
-------
214
0)
O)
£
05
O
X
UJ
o
C/>
I
•o
Q)
C
O CD
CO _O
fi «
UJ
c
(Q
(0
••••
O
c
0)
o.
0)
*-»
(0
E
o
0)
o
o
X
o
(0
o
0)
c
ffl
3
o
CD
c
(0
(0
•- ^>
m "o>
o
0)
(0
(0
(0
m
0)
o
(0
"o
(5
c
o
0)
0
O)
•o ^
0 ^
I °-
UJ
E
E
-------
215
tt\
_
OS
o
(5
£
CO
D)
+•*
c
0
"5
CO
^^
o
JO
X
0
o
"o
CO
0
3
O
^^
0
o
3
0
^^
LU
(/)
— J
• 1MB
•^•^
CD
"x
0
2
"o
i
'5
_>
Q.
Q.
(0
phenols
0
(0
0
t>
^
^.
O
^^•M
O
r*
"S
£
"5)
c
"(0
Q.
o
0
0
•o
,O
0
Q.
O
Q.
Q.
(D
C
O
^L^J
O
0
0
TJ
0
'>
C
0
(0
0
.2
*fS
o
(0
^-^«
"01
3
^
O
1
o
Q
^i-
(0
0
(0
"<5
c
(0
0
ts
U)
_c
f"
•n
••^1^^^
Q.
-------
216
CO
c
CD
O
C
o
o
LU
j
•
JZ
•#•*
'i
i_
CD
l
E
§••
0
"0
CD
+•*
O
CO
X
CD
*T™
O
o
Q.
CD
L.
CO
CO
0
c
CD
1=
a.
CO
CD
E
3
O
>
C
CD
"o
CO
CD
D)
JO
0
Q.
E
CD
CO
£
^
CD
O
2
Q.
LU
J
-J
CD
JQ
2
O
0)
0)
•o
O
C
CO
o
c
CD
r*
•^B
Q.
C
CO
E
'o
CO
CD
>
0>
"co
5
O)
c
'H
o
Q
o
LU
6
(5
^
o
Q
LL
i
O
G
jC
1
c
mo
HH*
5
>*
0)
E
U)
'35
O
CO
o
Q.
Q.
Q
O
111
D) 6
i
x o
0) '^
(0
C N
c g
£ "o
LU CD
C/) LL
Q_
-------
217
0
x
Q
O
111
6
(0
C N
O '-5
C
(0
uj m
u-
a.
o
(0
O
Q.
Q.
c
o
JG
"5
BO
0
^
mLm
O
0
n
"35
0
>
0
C
O
c
0
Q.
0
"D
^
O
C
0)
"o
it
0
UJ
V)
^i
o
c
0
"o
£
0
C
o
"l"i
7
CO
Q,
C
(0
•*•*
0
"ffl
0)
c
0
^j
0
"x
o
Q
0)
B0
"c
(0
3
CT
n^, ^
(0
E
C"™
••••
0
(0
3
LJJ
0)
0
M*
')
C
0
0)
*©
>
IZ
!5
"(0
0)
o
Q.
0)
0
^
C
o
"£3
(0
~
(0
>
"J^
0
"D
m
m
LJ.
Q
o
m
6
o
O)
c
"55
^
"o
c
0
Q.
0
^
0
.>
•o
^
o
c
o
0
s
0
-------
218
(0
p
'o
o
O ,_
c •
.2
s>
UJ
CM
O
CD
o
o
'E T
cc a.
>
CO
X
O
+ zf
CM
X
O
0
CO
X
O
CM
O
X
^-
1
1
1
1
1
1
1
' \
is
^g
9
«
•g
o
«-
LU
O
•o <
CO
I 8 I
w JE "o
X
o
CO
CO
X
0
CM
X
o
1
^V
1
1
I
I
I
I
y \
\.
0)
1 «
is
o
«
•o
o
s
1 u
w
D
O
CD
O
E
75
0 5
CM
-------
219
C
o
to
OS
0>
O
ffl
CD
LL
Q.
Sw»
O
73
I
O
n
O
o
CM
N
C
.Q
O
m
CM
c
o
Q.
o
c
0)
Q.
0)
N
*3
(0
0)
Q
O
c
o
a.
O)
OJ
CM
IO
CM
CO
CM
C
I
CO
5 1=
C
JO
*3
C
0)
CO
o
o
CO
o
o
CM
O
o
(AUI) asuodsau QO3-OO
-------
220
C
mo
+3
CO
c
D)
iwmm
6t\
^JM
Q
E
(0
^^*
O)
o
+•»
(0
0)
TJ
i.
m
co
UL
^1
^
O
0)
it-
's
f
•••
BO
"JE
§
0)
i.
o
•Jjf
*w
E
(0
(5
Q.
4-*
(0
15
>
UJ
c
BO
"^
CO
LM
3
•o
^3
C
CO
0
&_
*-»
CO
0)
Q.
E
0)
4-*
C
o
CO
CD
or
1
c
BO
to
IM
HM
C
o
c
o
o
CO
O
O
CM
\^
l
solution
w
c
o
CO
CD
k>
C
• •^
1.
Q)
to
5
»*-
o
CO
^^J
^"*
c
Z3
O
E
CO
CD
0
JMBf
CO
£
1
(0
0)
0)
O)
c
(0
o
X
o
c
(0
u>
c
o
ts
0)
4-»
(0
"(5
UJ
c
CO
_a>
o
CD
o
12 «
o „
.E Q.
c
CO ft
o o
0 -o
c £
CO
c =
CO
CO
CO
_CD
CD
o
o
CD
CD
Q.
CO
0)
>
CO
0
CO
o
E o
o o
O T-
a.
-------
221
(0
•^B
O
Q.
(U
o o
f £
Q. Q.
(0 (0
i- CM
CO
U
U
0 O
ro =
c >
0 x
o o
(0 (Q
X O
O
o
N
C
0
.a
0 -L
S °
? =5
o 4
o
•
in
•^ »\ V\
TO CM CM CM
V)
o
c
0
I£H
Q.
o
o
T5«
o
0
c
0
CL
mj-m
O
CM
0
C
.c
Q.
O
0
c
0
Q.
i
0
"U
i
CO
CM"
o
c
0
a.
p
"•5
^T
•N
CM
O
C
0
Q.
a
"•5
IO
CM"
o
c
?•
Q.
O
"x.
HM
10
CM"
^_
o
c
0
Q.
O
J5
r*
MHl
O
4-*
C
0
Q_
(0
o
c
0
•MM
^.
<
O
0
a.
o
c
0
a.
•
CO
o
CM
0
c
0
Q.
CO
X
O
o
c
0
a.
CO
0
"•5
CM"
0)
O
C
/a
j^^i
Q.
O
Z
o
c
0
JC
CL
CM
0
CM
O
C
0
Q_
CM
O
»3-
-------
222
LSE of Phenols and Acids
from Water
Silanized glass
chromatography
column
0.5 g AG
Silanized
glass wool
1 mL of 0.1 ug/mL
analyte mix in MeOH
BioRad AG MP-1
100-200 mesh resin
70-150 urn, chloride form
CD\Reoov1H
-------
Analytical
Procedure
223
Elute AG MP-1:
6 mL 2% HCI in
MeOH:MeCI2 (20:80)
5 mL MeCI2 (x2)
10 mL H20
130 uL cone HCI
Na2SO4 (ca 2 g)
drying column
Discard
K-D concentration
Add IS: 3,4-diCH3-phenol for phenols
4-CI-benzoic acid for acids
Acids
Phenols
Solvent exchange to MTBE
Methylate
GC-ECD
Solvent exchange to acetone,
dilute to 1 mL
PFBBr derivatization
GC-ECD
-------
224
s
i
0)
-I S
1- O
P- T3
•E R
O) C
-J 3 _2
T- 1- "3
o
o
0)
O) o
<
c/)
E
C
O)
(B
2.2
w >.
II
a.
LU
CO
-------
225
Battelle PFBBr Method vs Literature
Literature Method*
~ 800
§ 600
o
Q.
8
400-
Q
O
III
8 200
I I I
I I
I I
VC/7rom469, 351-9, 1989
^ 800-
g 600-
o
Q.
0)
0)
400-
O
O
III
O 200
1 I I
Battelle Method
18
20 22 24
Retention Time (min)
I
26
I
28
-------
226
4-Step PFBBr Derivatization Method
Step 1
0 ug K2CO3
95'C for 1hr
Step 2
0.1 ug K2CO3
95'C for 20 min
250 uL
GC-ECD
J,
S
o
o.
CO
O
LU 150
o
O
50-
Acids
n 1 . . 1
1
,»
1 1 1 1 1 1 1-
12 15 18 21 24 27 30
Retention Time (mln)
Step 3
2 ug K2CO3
95'C for 20 min
Step 4
100 ug K2CO3
95'C for 20 min
CO
i
CO
CD
oc
o
LU
6
o
250-
150-
50-
Phenols
T T
12 15
'T-
IS
T
21
T
24
J
"T
27
"T
30
Retention Time (mln)
CO\pM«p1c
-------
227
PFBBr-Derivatized Phenols by GC-ECD
25 pg/uL
100-
2.5 pg/uL
13
15
17
19
Retention Time (min)
-------
228
PFBBr-Derivatized Phenols by GC-ECD
(0
o
Q.
(0
o
DC
6
O
100-
80-
60-
40-
20-
o
Q.
18
14
o
a 10
6
o
6 -
13
25 pg/uL
2.5 pg/uL
15
17
i
19
i
21
i
23
i
25
T
27
"T
29
Retention Time (min)
-------
229
GC-ECD Calibration Curve for
PFBBr-Derivatized PCP
40
I,
o
CO
c
o
Q.
CO
-------
230
O
UJ
6
(0 O
E
O
c
0)
UJ
CO
(AIU) asuodsay Q03-OO
-------
231
o
£ 6
o
£
^^^M /V\
O
c
LU
CO
8
CM
o>
CM
CM
— u>
CM
CM
_ O>
£
1
0)
c
-------
232
0)
3
1 T-
^^™ •
^L ^
2 ®
o *
< Q
^™ V^
£ LU
'i §
k. -Q
fli £••
"m ro
%\)
^>^ r™
^^^r -^
^^ .s
^^^
E f
0 £
^
• ^H
O
W^f^^^^
LJJ
0)
H
IJ/
CM" 4
O./ Q -^
^> - 2 /
Q. 0 ^7
C CM
0
mtfmt
JZ
Q.
CM
4
^^^mwmmmmmmmimmmmt
o /
"o /
c
0)
.a
•
O
•••
4 ••
•N
CM
I
O
'5
N
.2 O
8 2\.
? • "^
5 j
i i i i
^
c
CO
CD
•••••^^^^B
—
-
'
™
"i
i»
- a
_
— «
_ r*.
•• c
E.
_ m
" E
™* c
o
— CO A
T~ *••
0>
DC
- £
— O)
^
-
-
— in
J 1 1 1 1 ° 1
(AUJ) asuodsey 033-39
-------
233
(0
•••
o
c
0)
Q.
0)
o
o
0)
DC
yj
o
o
111
6
o
CO
m
m
0)
+*
(0
0)
c
12
o ;
-S5
3
E
Q.
CO
O
-------
234
LSE Efficiency =
(Analytical Method)
(Post-Resin Method)
x 100%
Analytical Method
Post-Resin Method
Discard
Acids
Phenols
-------
235
o
c
0>
CL
(5
O
C
LU
LU
03
Q
O
111
6
o
•o
c
(0
ffi
ffl
0)
c3
o>
c
0 s
0)
0)
^
D
E
CO
-------
236
o
c
0)
"o
S .g
LLJ 0 1
o
c
0)
O
O
ill
6
o
13
c
(0
O
«3
CO
£
05 1
CD
*«*
CO
O)
O ^= I
o O 5
o
0
a: ^ -
o
*-•
J5
3
E
GO
o
8
O
DC
*
O>
N
II
c
(0
CM
> "•"
o 41
.2 o>
O N
'£ II
LU
LU (0
O
o
10
o
in
10
CM
-------
237
LU
UJ
0
o
LU
LU
o
CO
O)
3
O)
(O
CM
O>
00
CO
O)
(0
TJ
"o
CO
in
CM
O)
D
CO
G>
o
O
C
O
3
0)
O
0)
CL
-------
238
o
w
O
c
o
o
UJ
CO
D)
_c
3
(0
•JO
"o
<
•o
c
(C
1
c
(D
.c
Q.
"o
O
O
0
DC
._
'T
Q.
O
<
0
O)
(0
^1^
o
X
UJ
c
o
'E
^
^
'I
>»
0
O
U
UJ
1
o
c
0
(C
_>
"5
o-
0
CB
(0
o
£
0
Q.
"o
c
O
•S*
0
T3
(0
75
U.
CD
CD
U.
Q.
_J
"o>
^
•
1
o
•
(0
0
>
0
™^
l_
0
1
D)
C
"
c
-------
239
c
0)
E
0)
D)
^m
O
0
•••
5
O
o
Q.
(0
(0
0)
D)
"73
_o
o
.*
o
(0
E
O
m
O
J
«.
o
z
o
CD
^jl
C
o
o
<
LU
CO
o
o
o
1
o
o
^•i^^
i
CO
M0^.
(O
o
£ >
o
4-1
«J
CD
-------
240
[Blank Page]
-------
241
MR. TELLIARD: Our next speaker is going to present data on some
groundwater analysis, using microextraction procedure, and Cathy Arthur from Waterloo
University.
MS. ARTHUR: Good afternoon, ladies and gentlemen. My presentation to
you this afternoon is on both the theory and mostly the practice of solid phase
microextraction.
The focus of the research in our group is not to reduce solvent in sample
preparation; we want to eliminate it entirely. That gives us some advantages as far as solvent
recovery and solvent expense. We also would like to see it very fast, and we would like to
see it automated.
For solids, we are using supercritical fluid extractions, and for liquids, we feel
that solid phase microextraction meets these objectives.
Solid phase microextraction is similar to solid phase extraction in that we are
taking the organic analytes out of the aqueous phase and onto a stationary phase, but instead
of the usual support, we have a stationary phase coated on a fused silica optical fiber. The
fiber has a diameter of about 200 microns. [SLIDE TWO]
The stationary phases that we are using right now are mostly polymethyl
siloxine which you can buy directly from the manufacturer. We have also used polyimide,
which doesn't work very well, uncoated fibers, carbowax, and liquid crystal polyacrybite.
At present, we are also working with Supelco, because they have licensed this
technology, to put GC stationary phases on the outside of the fiber.
There is one other major difference between microextraction and SPE, and that
is that we do not exhaustively extract the analytes. Instead, we form an equilibrium between
the aqueous phase and the stationary phase.
200 micron fibers are a little bit difficult to handle, so we have developed what
we call a solid phase microextraction device [SLIDE THREE], and what we have done is we
have inserted this fiber, 200 microns in outer diameter, inside 30 gauge stainless steel. 30
gauge stainless steel runs up through the needle and the plunger of a Hamilton 7000 series
syringe, and we put a blob of epoxy at the end of the steel so that we can move the fiber up
and down through the needle at any time we wish.
-------
242
When we do this technique, we draw the fiber into the needle, pierce the
septum of your sample vial, and drop the fiber into the sample. We leave it there until it
equilibrates. That takes about 5 minutes.
We then draw it back up into the needle, transfer it over to our GC, put it in
the injector, and thermally desorb it. You can use a split/splitless, a SPI, or an on-column
injector. Those are the ones we have tried. If you have got others, please try it, and we will
be interested to hear what happens.
One other nice aspect of SPME relative to, for example, purge and trap...and I
will mostly be talking about volatiles today...is that you can use any column you like. You
are not stuck to larger bore columns because of the high flow rates that come with purge and
trap devices.
I have mentioned several times that we form an equilibrium in a large volume,
and by that I mean above 25 ml. The amount that is absorbed on the fiber is related to K,
the distribution constant, the volume of the stationary phase Vs, and the aqueous concentration
Caq. It is linear over four orders of magnitude.
This is what we call an equilibration time profile [SLIDE FOUR] or an
absorption time profile. This was for the compounds benzene, toluene, ethyl benzene, and the
xylene isomers, and virtually everything that I talk about today will refer to the BTEX
compounds.
There are two things I would like you to note from this. First of all, there is a
lot more xylene absorbed than there is of the benzene. The topmost curve is double, because
meta and para co-elute on our column. More xylene is absorbed than benzene because of the
difference in the distribution constant of those compounds from each other.
The second point I would like to make is that the benzene has stabilized in
about 1 minute, and the xylenes took about 5 minutes. This is not, in fact, what mathematical
theory predicts, and we have developed the mathematics to describe the process. They
predict 15 seconds for a perfectly stirred solution, and 2 hours for an unstirred solution, the
difference being the time it takes to diffuse through water.
We have this difference from the ideal 15 seconds, because we have yet to
figure out how to perfectly stir a solution. We do not have mixing that would come from
-------
243
pouring through a cartridge or a disk, so we have to stir our samples.
Instead, what we feel happens is we have a stationary layer of water next to the
fiber, so our mixing is effective to a large degree, but the analytes come up, hit that stationary
layer, and have to diffuse through. It takes longer for the xylene than it does for the benzene,
because we have to flux more of the xylenes through that stationary layer than we do the
benzene.
Well, I hope I have teased your interest enough to find out how sensitive it is,
but before we could do that, we had to figure out how precise it was, because, initially, we
were having 20, 25 percent relative standard deviations, and that is not acceptable.
There are three factors that are particularly important in this. The first is the
desorption temperature and time. [SLIDE FIVE]
The xylenes in this mix are the highest boilers. They boil at 145 or so. We
can desorb at 150, so we don't have to worry as much about thermal degradation of either our
fiber or of our analytes. Because of the very simple geometry of that fiber, we don't have to
go to very high temperatures.
It takes 30 to 60 seconds at 150 degrees to desorb our materials back out of the
fiber. This is not a problem as long as you cryofocus your oven, because 30 seconds, 60
seconds is long relative to your peak width. So, we cryofocus, in this case, at minus 5. You
can go up to zero degrees centigrade, but I like to hedge my bets, so we use minus 5.
The last factor that we have found was very critical was the time between
exposing your sample in the vial and taking it over to your GC. If you go to your very worst
case which is an uncoated fiber, your benzene has gone 30 seconds after you have taken it
out of the sample, even though it is sheathed inside the needle.
If you have 56 microns of methyl silicone sitting on your needle, your benzene
is good for 2 minutes. Your xylenes are good for 5 minutes. Go to a thicker film, it is stable
for even longer.
Two minutes is more than enough time to go from your sample to your GC,
but you have to be aware of the fact that the problem may exist.
With the flame ionization detector, we can see limits of detection of 0.3 to 1
part per billion on a weight by volume basis [SLIDE SIX]. This is with a 56 micron coating.
-------
244
The linear range is almost four orders of magnitude, and we have a relative
standard deviation that averages around 5 percent, and this is over six grad students. So, I
am reasonably confident of that number.
If you go to a more sensitive detector, such as an ion trap mass spectrometer,
you can go down to 50 parts per trillion of benzene and still see it with a signal to noise ratio
of about 10 to 1 [SLIDE 7]. Here again, your signal to noise ratio is even higher for your
toluene and your xylene because of the difference in the K values.
When you compare for these five compounds to the method detection limits for
EPA 524.2, we are well within those method detection limits [SLIDE 8]. At the 50 part per
trillion level, our relative standard deviation is a little bit higher at about 7 percent, but it is
still, I think, more than good enough for most people.
The other thing I would like you to notice off this Table is I have compared
the log of the distribution constant with the log Kow. That is the octanol/water partition
coefficient, which is quite widely tabulated for a whole range of compounds, and they are
very similar.
So, if you wish to develop a method and you want to know is it going to be
sensitive enough, you can use log Kow as an initial estimate of what your distribution
constants will be if you are using the methyl silicone film. If you go to different films, of
course, your K values are going to be different, but this is not a bad initial guideline.
Well, the technique seems to be precise, and it has low limits of detection in
very nice clean reagent grade water. The question then comes what happens in the real world
when you have things like organic contaminants, salt, varying temperatures, and all that sort
of thing.
So, since we had good precision, we went on to look at matrix effects. These
are things that will change the distribution constant.
Above 1 percent methanol, we see a decrease in the amount absorbed [SLIDE
9]. Below 1 percent, there is no discernable effect on the amount absorbed by the fiber.
This can, in fact, be predicted by calculating out what the change in the
distribution constant would be using solvent selectivity parameters according to those
equations in the reference if you are interested. They would predict a 6 percent change in K,
-------
245
which we cannot see with the 5 percent RSD. So, you are good up to about 1 percent
methanol.
Similarly with salt [SLIDE 10]. If you go up to 1 percent salt, we are not
seeing any influence on the amount that is absorbed. Above 1 percent, you see a big increase
in the amount absorbed, because your well known salting out effect is kicking in.
In fact, for these compounds, if you go up to a saturated salt solution, you will
increase the amount absorbed by a factor of ten. You also, however, increase the analysis
time by a factor of two to five, depending on the compound.
We also looked at the effect of temperature [SLIDE 11]. The top graph is a 5
minute equilibration time, and we sampled our spikes at anywhere from 0 degrees up to 55
degrees, and you see an apparent influence on the amount absorbed.
What you are seeing is not a change in the distribution constant but a change
in the rate of diffusion through water. So, if you go to a 30 minute equilibration time, which
eliminates the effect of diffusion, you do not have any noticeable effect on temperature except
maybe a slight decrease at higher temperatures which1 suggests to us that we have an
exothermic reaction and it is shoving the equilibrium back in favor of the aqueous phase.
We have also looked at pH, and between pH 4 and 10, there is no influence on
the amount absorbed. Below pH 4, you are starting to dissolve the methyl silicone film.
That is all well and good. We also like automated methods. We are not in the
business of doing routine analysis, but we know most people are.
So, we adapted a Varian 8100 autosampler to work with the fiber method.
What we have done is taken out the Varian autosampler syringe and mounted our syringe
about 3 cm higher in the carriage so that the needle heights are identical. Otherwise, you are
going to smash out your inserts.
We also extended the plunger by about 1.5 cm so that it will fit nicely into
their device.
The other major change that we made is we had to add a stirrer, because our
samples must be stirred. And you see a little black box down at the bottom with a piece of
white tape on it. That is a microstirres that you can buy from Cole-Parmer. It is about 2 by
2 inches and .5 inches thick, and you can mount it where the wash cup normally is, and the
-------
246
carriage will, in fact, move and up and down in front of it.
The whole thing is controlled through Lab View software on a Mclntosh, and it
allows us to adjust the plunger movement so that the fiber is always contained in the needle
whenever it needs to pierce a septum, and it also controls the sampling time in the vial so
you can have whatever sampling time you choose.
Quite often, incidentally, even if we only need 5 minutes to equilibrate, if we
have a 20 minute analysis run on our GC, we just let the fiber sit in the sample for 20
minutes, because sampling time makes no difference once you have reached equilibrium.
One disadvantage to using an autosampler is that you only have a 2 ml sample
vial, and this fiber actually sucks up enough material that you will significantly exhaust the
vial. This, again, depends on the distribution constant of your compound.
So, for the xylenes, you are seeing a significant drop after two or three
injections, and I am using significant to be more than 5 percent, whereas with the benzene,
you can do 100 injections, and you are not going to exhaust the sample [SLIDE 12].
Oh, one other thing on that slide. We have calculated out what the actual
amount is. That is shown up in the middle of that slide. So, if you are using small volumes,
you can calculate out how much you are going to be extracting.
If you play around with that formula mathematically, you can work it so you
can calculate the number of injections you can do for a specific volume for a specific K
value, so you know how far you can go before you are going to exhaust your sample.
[SLIDE 13]
On the right-hand column, you can see that for compounds with a K of about
1000, which is typical of xylene, you can only do less than one injection before you are
pulling out 5 percent. If you go to benzene which has a K of about 126, then you can do 2
injections per autosampler vial.
The converse side of this coin is if you are going up to compounds with a K of
10,000 or more which would be typical of PCBs and PAHs and you want to extract that...if
you want to exhaust that vial, you can hold your oven at a cryotemperature, sample your vial,
desorb it in the hot injector, cryofocus and trap it in your column, and go back and do a
second injection, and you have got all your stuff out of your vial. So, you have done an
-------
247
exhaustive extraction.
So much for theory. What does it do in the real world? Well, I don't have
any "milkshakes" to analyze, but what we could do is we went out in the middle of January
to a nice culvert after a snowfall and got parking lot runoff.
As you can see, [SLIDE 14] we picked up the BTEX isomers and
trimethylbenzene and some naphthalene which is all well and good. We expected that. What
we were actually looking for was carry-over, because it is very easy to have a nice clean
sample and not see carry-over, but what happens if you get to a real sample?
And the lower RICs show you that we did not get carry-over on this particular
sample. So, for at least this sample and, presumably, for a lot of surface waters, carry-over is
not going to be a problem.
We went to a somewhat nastier sample [SLIDE 15]. This is a coal gasification
wastewater sample that someone sent us from New Jersey. I have no idea exactly where.
As you can see, we got everywhere from benzene all the way up to
acenaphthalene which just happens to be the length of our chromatographic run.
This technique, while I have focused so far on volatiles, works just as well for
the semi-volatiles, so that for all those analyses where you are doing liquid-liquid extractions,
you can replace it with the fiber technology and eliminate your solvents.
We didn't tackle semi-volatile, before this, because BTEX is a nice, fairly
simple range of compounds to study before we get into the 60-odd that the EPA is interested
in.
In summary [SLIDE 16], solid phase microextraction for BTEX compounds is
a whole lot less expensive than your solvents and your purge and trap device, and it is very
portable. One thing I didn't mention earlier is that because we are independent of sample
volume above about 25 ml, if you want to go field sampling, you do not have to measure the
volume of water. If it is a stream that is flowing fast, you don't even have to stir it, because
you depend on concentration, not total amount.
We have completely eliminated solvents from the sample preparation. We
have completely automated it so you do not have to handle it once it has been collected,
assuming it is collected in an autosampler vial or manual vial.
-------
248
We have a very short sampling time with excellent limit of detection with
either an FID or, if you want to do drinking water standards, you can go and use an ion trap
mass spec.
It is linear over four orders of magnitude. Incidentally, our mass spec is also
linear over four orders of magnitude.
We have shown it is independent of salt and organic solvents if those are both
below 1 percent. Also pH and temperature.
Lastly, I would like to acknowledge funding from all of these companies which
have contributed to our research program [SLIDE 17].
Any questions?
References.
1. Arthur, C.L.; Pawliszyn, J.; Anal. Chem. 1990, 62, 2145
2. Belardi, R.P., Pawliszyn, J., Water Pollution Res. J. Can. 1989, 24, 179.
3. Arthur, C.L.; Killam, L.M.; Motlagh, S.; Potter, D; Pawliszyn, J.; J. Env. Sci. Technol.
1992, 26, 979.
4. Louch, D.; Motlagh S.; Pawliszyn, J.; Anal. Chem. 1992, 64, 1187.
5. S.B. Hawthorne, D.J. Miller, J. Pawliszyn, C.L. Arthur "Solventless Determination of
Caffeine in Beverages Using Solid Phase Microextraction with Fused Silica Fibres" J.
Chromatogr. in press.
6. Arthur, C.L.; L.M.; Buchholz, K.D.; Pawliszyn, J. "Automation and Optimization of Solid
Phase Microextraction" Anal. Chem. in press.
-------
249
QUESTION AND ANSWER SESSION
MR. TELLIARD: Questions?
MR. SOLOMON: You mentioned you used different injectors in the analyses
that you did. Could you go into a little bit of detail on the differences you saw between the
split/splitless, the SPI, and on column?
MS. ARTHUR: Yes. Most of the stuff has been done on the SPI, because
our GCs get muddled around between both fiber and regular syringe injections, and we like
the SPI.
The difference in them is largely that the split/splitless is nicer because the
gauge on these needles is about 24 gauge, and you tend to core septa. So, with a
split/splitless, you are not plugging the restriction that occurs on a SPI injector, and that is
about the only difference we have seen.
MR. TELLIARD: Thank you, Cathy.
-------
250
Practical and Theoretical Aspects of Solid Phase Microextraction
for the Direct Analysis of Groundwater
Catherine L. Arthur. Lisa Killam, Karen D. Buchholz, David Potter,
Janusz Pawliszyn, The Guelph-Waterloo Centre for Graduate Work in
Chemistry and the Waterloo Centre for Groundwater Research,
University of Waterloo, Waterloo, Ontario, Canada N2L 3G1.
John R. Berg, Varian Associates, 2700 Mitchell Dr., Walnut Creek,
California, U.S.A. 94598
-------
251
OPTICAL FIBER
WATER WTTH
TRACE ORGANIC
MATERIAL
CHEMICALLY BONDED
ORGANIC PHASE
J
-------
252
cT
-------
253
0
F
(ON) Q38yOSQV SSVIN
-------
254
Factors Affecting Precision
desorption temperature and time
cryofocus temperature
time between absorption and desorption
-------
255
Optimized Precision with Spiked Reagent Water
Analyte
toluene
ethyl benzene
m,p xylene
o-xylene
and Flame
lonization Detector
LOD Linear Range
1
0.5
0.3
0.3
1-8000
1-8000
1-8000
1-8000
RSD
6
3
6
6
-------
256
50 PPT BTEX IN WATER
1.26-x
78
•can: 373
height: 32S
s'n: 8.3
height: 3,361
s/n: 158.3
91
*•
186
A
ETHYL BENZENE
i
XVLENE ISONEKS
scan: 714
height: 2.186
s/n: 128.9
A
5:88
6:48
• i •
8:28
1B:88 ll:48
RETENTION TIME (min)
-------
257
'S3 3
•3 ?
.
*<** "O
ft c
*•*
o
a
•2
£
o cr
o S
E a
o u^
a.
»v
o
vo
o
0
C
—
M
i
O
-a
c
E
'S
Q
^>
5
"S
i
-------
258
LOG PEAK AREA
CO
i
CD
I
3
CD o 3 m H
2 x -o 5 S.
i f
I
-------
259
LOG PEAK AREA
en
ui
en
b>
01
01
CO
en
b
o>
-------
260
5 MINUTE EQUILIBRATION TIME
1e+7
8e+6 -
6e+6 -
4e+6 -
2e+6 -
Oe+0
BENZENE
•— TOLUENE
I ETHYL BENZENE
MP XYLENE
OXYLENE
30 MINUTE EQUILIBRATION TIME
1e+7
1e+7
8e+6
6e+6
4e+6
2e+6
Oe-i-0
•5
5 15 25 35
Temperature (Celcius)
45
55
-------
261
MASS ABSORBED (NG)
8
T
IV3
8
Q
D
Q
- Q
§
O
CD
m
3)
P.
00
H 40 X
BHD X
O X + • H
(D
O
-------
262
Distribution Constant # injections # injections
to remove 95% to remove 5%
10 1016 22
100 103 2
1000 11 <1
10000 2
for a 1.7 mL autosampler vial.
-------
263
ORGANICS EXTRACTED FROM PARKING LOT RUNOFF WATER
188-x
tor
r
TOLUENE. ETHYL BENZENE* MID XYLENE ISONERS
8.72*
1BS , i TJUHETHYL BENZENE ISONERS
128-
NAPHTIMLENI
IBBvc
TOT-
91-
AftWWjS*^
•A»-HW—X*^
8.42^
185
8.19-x
128:
^jj\**^^
^^y^l^^
18:88 13:28 16:48 28:88
RETENTION TIME (min)
-------
264
H
en
<
2
O
H
3
u
<
O
OK:
fc,
a
y
z
-------
265
Solid Phase Microextraction of
BTEX Components from Water
inexpensive and portable
solvents completely eliminated
2-5 minute sampling time
LOD with FID 0.3 - 1 |ig/L
linear range 1-8000 |ig/L
independent of salt, organic solvent if < 1%
independent of pH 4 - 10
independent of temperature 0-40 °C
fully automated sample preparation
meets EPA 524.2 and MISA requirements with
ion trap mass spectrometer detector
-------
266
Acknowledgements
Natural Sciences and Engineering Council of Canada
Varian Canada
Varian Associates
Supelco Canada
Imperial Oil Canada
University Research Incentive Fund
Apple Canada
-------
267
MR. TELLIARD: That concludes our ramblings on solid phase extraction.
A number of questions came up at lunchtime on okay, or when can we use it and when
will EPA approve it.
I think we are all interested and excited about getting this procedure
accepted for wastewater. Drinking water, I have had the insight to move ahead and get
it into their methods. We would like to come back here next year and be able to use
solid phase extraction. I can't promise that, but we are going to try moving on that path.
George Stanko pointed out earlier that extraction in wastewater matrices,
could present problems, and we certainly have to build a data base before we give it a
blessing, but the data you have seen today looks very, very promising, and we think it
might be the way to go.
It relieves us of our solvent problem to a lesser extent. It also gives us
some more flexibility as far as field application is concerned, and I was very impressed
with what was presented in the storage studies.
So, we are hoping that, by next year, we will be able to see this in more of
our methods, and I am sure drinking water will continue on this path.
The Office of Solid waste have already mentioned that they are
incorporating solid phase extraction.
So, with that, I would like to introduce our next speaker. Jeanne Hankins
is presently Secretary of the methods consolidation group of the Environmental
Monitoring Methods Council. For those of you who don't know what all that is, the
Environmental Monitoring Methods Council is an agency-wide council which is dealing
with issues of methods consolidation, laboratory certification, quality assurance, and
quality control issues.
Jeanne, as Secretary, is going to be talking a little bit about what is going
on with the EMMC and, more specifically, what is going on with lab certification.
-------
268
UPDATE: COMMITTEE ON NATIONAL ACCREDITATION OF
ENVIRONMENTAL LABORATORIES
Jeanne Hankins
Executive Director
Committee on National Accreditation of Environmental Laboratories
The Environmental Protection Agency (EPA) has initiated an effort to investigate
the concept of a national program for accreditation of environmental laboratories. This
effort started with the Environmental Monitoring Management Council (EMMC) and its
Ad Hoc Panels. The Federal Advisory Committee on National Accreditation of
Environmental Laboratories (CNAEL), established on the recommendation of the
EMMC, has expanded the number and range of interests of the involved parties. The
history, accomplishments and plans for the future of these two groups are summarized in
the following discussion.
The seed for this effort was planted early in 1990 by EPA's Science Advisory
Board (SAB). The SAB observed that the development and direction of EPA programs
have been established in response to concerns of the public and Congress. As a result,
there is not necessarily coordination between programs and therefore, implementation of
Congressional statutes is not always efficiently accomplished. This lack of uniformity is
reflected in the monitoring requirements. Under Deputy Administrator Hank Habicht, a
committee of top managers was assembled into the Environmental Monitoring
Management Council to look at comprehensive multi-media solutions to environmental
monitoring. (Environmental monitoring has an all encompassing definition in this
instance and covers one-time sampling and analysis, as well as routine monitoring of a
relatively stable situation.) The EMMC recommended that coordinated Agency-wide
policies on environmental monitoring issues be established. The goal is to enable the
Agency to operate more efficiently and effectively. In this way it will be possible to
improve the monitoring data which is a linchpin in the decisions which are made in the
Agency everyday.
EMMC
The EMMC structure (see diagram) consists of a Policy Council, a Steering
Committee, and several Ad Hoc Panels. The Policy Council is composed primarily of
deputy administrators of the EPA program and regional offices, and is chaired by the
Assistant Administrator of Research and Development Eric Bretthauer and the Regional
Administrator of Region III Ted Erickson. Ramona Trovato serves as the Executive
Secretary. The Steering Committee members are mainly division directors from the
program and regional offices and represent the position on the technical and scientific
policies of their offices. The Ad Hoc Panels were established to address specific issues at
the implementation level. The Panels provide background information, and are a source
of technical expertise, and inter-agency input.
-------
269
The EMMC is charged with examining several issues. Several specific issues were
identified for investigation and were initiated in 1990. Five separate Ad Hoc Panels
were established to investigate 1) the development of integrated methods, i.e. methods
that can be used to generate data for any EPA program; 2) a comprehensive
computerized library of information on methods which are suitable for use in complying
with EPA regulations; 3) the institution of a process in regulation development that
assures consideration of monitoring requirements; 4) a system for assuring EPA
program offices continued support for quality assurance (QA) services and quality
control (QC) samples; and 5) the feasibility and advisability of a national program for
accreditation of environmental laboratories.
AD HOC PANEL ON ENVIRONMENTAL LABORATORY ACCREDITATION
The Ad Hoc Panel on Environmental Laboratory Accreditation completed Phase I
of its task, which was to evaluate the feasibility of a national multi-media program.
After approximately six months of deliberations the Panel determined that there are
numerous feasible options for a national laboratory accreditation program and that it was
advisable for EPA to examine the spectrum of possible options in depth and to evaluate
alternative solutions.
The Panel made several specific findings. First, the growing number of non-
reciprocal state and private sector accreditation programs indicates the need for national
accreditation. Each program is aimed at determining the capability of a laboratory to
perform analysis in accordance with EPA or state regulations. The manner in which this
is determined, the range of analyses covered, and the rigor applied to the evaluation of
the capability varies with the accrediting organization. Therefore, there is extreme
reluctance to grant reciprocity among the multiple accrediting organizations. Costs for
these programs range from the indirect costs of the time and effort of the laboratory
staff to comply with the requirements, tp over a hundred thousand dollars in
accreditation fees.
Second, there is significant evidence suggesting that laboratory accreditation
contributes to overall improvements in performance. These improvements are most
likely a result of information acquired during on-site inspections, specifically: 1) an
improved understanding of methods requirements (both understanding which method to
use and understanding how best to interpret and implement the published method), 2)
improvements to quality assurance programs and quality control practices, and 3)
improved data management practices.
Thirdly, the laboratory industry is promoting national accreditation as one
mechanism for addressing recent problems with laboratory performance and professional
ethics. Some professional organizations are making efforts to improve the industry's
reputation and strengthen the public's and government's confidence in its work. National
environmental laboratory accreditation is cited by these organizations as one tool for
achieving this goal. One group has approached both EPA and Congress to request that
a government sanctioned national program be created.
-------
270
The benefits of a national program were also examined by the EMMC Ad Hoc
Panel. Primary among the benefits was that a national program might help strengthen
the credibility and integrity of EPA's monitoring and compliance programs. Public
confidence is an important factor in the success of EPA's program, particularly those that
involve remediation of environmental contamination or control of point sources of
pollution. Although accreditation does not provide the public with guarantees
concerning laboratory performance, it would provide uniform baseline assurance that
practicing laboratories meet certain minimum capability standards.
A national accreditation program may help to reduce the potential for
misapplication of EPAs' monitoring methods. The on-site examination of laboratory
facilities, equipment, personnel, etc. provide the laboratory personnel with an opportunity
to seek clarifications regarding performance requirements. Therefore, accreditation can
provide an effective mechanism for ensuring that environmental laboratories are
appropriately applying EPA's required methods.
Users could be provided with assistance in selecting appropriate and qualified
laboratories under a national program. The majority of users currently rely, at best, on
available information from existing programs, their ability to interpret the information,
and on their own experience with laboratories. A national program would provide such
users with and up-to-date indication of laboratory capability and a source of information
concerning historical laboratory performance.
The economic and operating burdens imposed by the existing multitude of non-
reciprocal accreditation and certification programs could be reduced. A strong national
program with technical implementation at the state level should increase reciprocity
among state program, potentially reducing costs for laboratories and some accrediting
organizations.
The Panel recommended that steps be taken to further evaluate the usefulness of
a national program. Their recommendation to establish a Federal Advisory Committee
was based on the need to determine the needs and effects on the entire environmental
community. The EMMC concurred with this suggestion and the Committee on National
Accreditation of Environmental Laboratories (CNAEL) was established by Mr. Habicht
in July, 1992.
Next steps for the Panel will be to develop an independent cost estimate for
current programs and proposed options. Emphasis will be on that portion allocable to
EPA. An optimal program will also be developed which will take into consideration
EPA and state capabilities and responsibilities.
COMMITTEE ON NATIONAL ACCREDITATION OF ENVIRONMENTAL
LABORATORIES
The charter for CNAEL directs the committee to counsel the EMMC Policy
Council, Deputy Administrator Habicht, and Administrator Bill Reilly on the advisability
-------
271
of a national accreditation program for environmental laboratories. This program would
include any laboratory which provides services and analyses of environmental samples
which are collected in order to comply with federal and state environmental statutes and
regulations.
The objective of CNAEL is to advise EPA on whether a national environmental
accreditation is the most beneficial approach and, if not, what alternative solution would
best meet the needs. During this investigation, CNAEL is charged with characterizing
the needs of the affected community, including the laboratories, EPA, the states, the
regulated community, and the public. In addition, the committee is to identify and
evaluate alternate solutions which will satisfy those needs. The committee has also been
directed to provide a design outline for the selected program, to recommend modes to
implement the program, and to suggest appropriate roles for EPA participation.
The Committee on National Accreditation of Environmental Laboratories was
established following the requirements of the Federal Advisory Committee Act. The
composition of the members must fairly represent the affected community. In this case,
members were selected from state governments, trade associations for both
environmental laboratories and EPA's regulated community, public environmental
interest groups, academia, and other federal agencies, including EPA. All meetings,
proceedings, and correspondence are open to the public. The public is encouraged to
provide written comments for consideration by the committee members or to make an
oral presentation during one of the meetings. Meeting dates and locations are
announced in the Federal Register at least two weeks prior to the meeting. To further
disseminate information, a mailing list has been set up which provides basic information
on meetings as well as minutes of any meetings.
CNAEL has accomplished several of the tasks which were designated by EPA.
First among these was the description of the needs of the laboratories, regulatory
agencies, and laboratory users. The scope of any program was defined in terms of the
regulations covered, the media of concern, and the type of analytical procedures. A total
of fifteen alternative approaches were identified and ranked. Finally the committee
identified multiple options for implementing a program which will be evaluated in
relation to the selected alternatives.
The primary concern of all parties was to obtain data of the needed quality in a
cost effective manner. This takes into account the concept that the quality of the data
may vary, depending on the use. Purely qualitative information may be needed in a
confirmative analysis, while extremely precise and accurate data with low detection limits
may be needed when evaluating whether a particular source can be used for drinking
water. The goal of achieving the quality needed must take into consideration the most
efficient means for obtaining the data in order to avoid overburdening the reservoir of
capable laboratories in the face of expected increases of environmental analyses in the
coming years.
-------
272
Some of the specific needs identified, which stem from this overall goal were 1)
reciprocity among states, international organizations, private programs, and federal
programs; 2) standardization of sampling, analytical methods, and quality control; 3) an
objective means of evaluating laboratory performance in various sample matrices
(including provision of materials); 4) a comprehensive yet flexible program to cover all
environmental regulations and 5) elimination of technical redundancy and inconsistency
such as exists in the current system.
CNAEL recommends that the scope of the program would encompass testing to
serve all existing EPA monitoring, enforcement or other functions which are mandated
by statutes and pursuant regulations, including those which have authority and
responsibility invested in the states. Monitoring under the Resource Conservation and
Recovery Act (RCRA), the Comprehensive Emergency Response, Compensation and
Liability Act (CERCLA or Superfund), the National Pollutant Discharge Elimination
System (NPDES), the Safe Drinking Water Act (SDWA), the Federal Insecticide,
Fungicide and Rodenticide Act (FIFRA), the Toxic Substances Control Act (TSCA), etc.
would be included within the scope of a national program.
All types of environmental media would be within the scope, to include water,
soil, air, sludge, solid waste, liquid waste, and related samples, such as biological tissue,
body fluid, food and neat chemicals. Analytical procedures ranging from simple routine
analyses (such as turbidity determinations) to complex procedures (such as identification
and measurement of organic components using GC/MS) would be included within the
scope.
In addition, emerging technology and new environmental regulations would need
to be accommodated within the scope of any program. Flexibility is critical to allow for
adaptation to the rapidly changing rules and methodology.
One of the difficulties that surfaced in discussing a national accreditation program
was that there was disagreement on what constituted the elements of such a program.
For clarification purposes the committee defined the basic elements of a program for the
laboratory and the accreditation organization. Some of the critical items for the
laboratory would be performance evaluation samples, on-site audits, QA program
documentation, equipment maintenance, calibration, and record keeping. Accreditation
organization elements included, among others, record keeping requirements, a directory
of accredited laboratories, an appeals process, standard auditing methods, and a
technical advisory committee. The specifics for each category have not been completely
defined, pending the selection of the best alternative, but there was general agreement
that consideration of International Standards Organization (ISO) Guides should be given
high priority.
One important portion of the charge to CNAEL was to identify alternatives to a
national environmental laboratory accreditation program. The committee identified 15
alternatives, including accreditation, which were then evaluated. Some of the identified
alternatives were 1) analyst certification, 2) product certification, 3) resident inspectors,
-------
273
4) performance bonds, 5) reciprocity agreements among states, and 6) training for
analysts, lab managers, and/or auditors. In order to determine the benefits and
disadvantages, each alternative was evaluated in how well it met the identified needs of
the various affected parties. Using a matrix and a numerical scoring system the
committee graded each alternative and ranked them according to their individual merits.
After evaluation of the alternatives it was decided that the selection would require
a combination of some of the alternatives in order to obtain the best solution.
Therefore, the following three alternatives were selected as the top choices: 1) national
environmental laboratory accreditation, 2) a combination of performance testing and
certification of the QA program and the laboratory process (on-site audit), and 3) a
combination of performance testing and certification of the QA program and the product
(e.g. spot inspection of data records).
The means to implement any program on a national scale were discussed at
length. Various speakers informed the committee of what was currently being practiced
in similar areas, e.g. the clinical laboratories, the fastener industry, and weights and
measures, which is coordinated by NIST. Six implementation options were identified.
Within each of option there were various minor sub-categories. Those options were 1) a
centralized program operated by the federal government, 2) a program with federal
oversight but daily operation by states and/or private organizations, 3) a program
operated independently by states/private sector with guidelines provided by the federal
government, 4) a completely state run program with full reciprocity, 5) a private sector
program with input from the federal and state governments through an
advisory/oversight committee, and 6) a completely private sector program with no
government involvement at either state or federal level.
The next steps for the committee will be to examine the final three alternatives
under the six different implementation scenarios. The feasibility of each option and the
impact on current systems will be considered and weighed in the selection process
yielding a 6 X 3 matrix. The eighteen possibilities will be winnowed down to a more
reasonable number for recommended action to EPA.
Information on costs for operating an accreditation program and costs for
participating in such a program are currently being collected from the CNAEL members.
Extrapolating from these costs an estimate will be made for each of the alternatives
selected. This cost analysis is designed to compare the status quo with each alternative
and identify any savings or additional expenditures,
An overall evaluation of the pros and cons of each alternative will be compiled.
This will include not only cost to establish and operate the program, but also the ease of
implementation and the ability to apply an objective scientific process for evaluation. In
looking at the effects on the current systems in place special emphasis will be placed on
the effects on enforcement of regulations. Equitable disposition of any cases coming to
litigation would be a high priority for all parties.
-------
274
A final report with recommended action will be prepared. This report will outline
the selected alternatives, the recommended operating systems, and a complete
documentation of the benefits and costs for each plan. This counsel from the Committee
on National Accreditation of Environmental Laboratories will be considered by
Administrator Reilly, Deputy Administrator Habicht and the EMMC in developing their
plan for action. This report is expected to be presented in mid-summer, 1992.
Additional input will be provided from the Ad Hoc Panel on Environmental Laboratory
Accreditation, especially regarding estimated costs to EPA for implementation of any
program.
The public is encouraged to attend and participate in the advisory committee
process. The next meeting will be in the Washington, DC area on June 1-2, 1992. The
meeting location has not been finalized but will be available in the Federal Register and,
additionally, to all who have been enrolled on the mailing list.
Everyone is invited to provide written comments to the committee. These should
be submitted two weeks prior to the meeting in order that the committee members will
have sufficient time to consider the comments before discussion begins. A limited
amount of time is also available for any member of the public who would like to make
an oral presentation to the committee. Again, notification two weeks prior to the
meeting is needed to assure sufficient time on the agenda. Frequently during committee
meetings members of the public are requested to make comment when their expertise is
recognized by one of the members.
Anyone wishing to be included on the mailing list need only call Jeanne Hankins
at (202) 260-8454 or send a business card or brief written request to:
Jeanne Hankins
US EPA; (WH-550G)
401 M St. SW
Washington, DC 20460
Portions of this presentation were extracted from the "First Interim Report of the
Environmental Monitoring Management Council Ad Hoc Panel on the Feasibility of a
National Environmental Laboratory Accreditation Program", the "Summary of the
February 4 and 5, 1992 Meeting of the Committee on National Accreditation of
Environmental Laboratories", and unpublished paper "Environmental Monitoring in the
90's: Meeting the Information Needs of Environmental Programs".
-------
275
O
O
O
F°
Uj
O °C
g ^o
I?OOQ
2*C
5
Q
UJ
oc
o
O
O
OC
UJ
U.
O
Uj
uJ
i^s
.c^-g
(/) -rr *i
s^s
10 is'*
C *-J C
C
-------
276
*
0
(0
v
•H
CO
'c
• MM
E
T3
£
a
(D
LJ ;
: :
—
0
c
D
0
0
>^
*^^
0
0
A
Q.
/1V
-------
277
Uj
o
o
DC
<
to
c
T3 j- 0)
CD CD
•+-* o
O C!
^1^^ ^^^^
o
o
o
~O Q
o> co
o
o
c
0)
-------
278
O
•~^«iifc
Ul
c
o
"5
Hh*
*"U
O
o
o
2
o
jQ
c
CD
o
Hi
c
.o
CD
QC
O
t>
Q
05
0)
o
JT
O
6
O
co o
>*~»
c:
C
- C LU
O
^^^™
.2
0
***
o
c
j2
sP
\^^"f ^^^^_
"JAv-.
i*
-------
279
UJ
O
Q
O
g> c/j c:
QJ CD
.o «.
Q> -
Ch
c
***»
§
tft
iv
Jf* CQ
J9?
X? «>
.
\JF IF**
S S - *
O Q. S Jg ffi
S **^
^
c ^
O
ti
§
-------
280
uJ
i
O
,«t-«
5$
O
a
o
O
co
E
O 3
O O)
CO CD
fA VM.
CO in
CD
CD < Q.
•D Q- u
CO LU O
w C _ O
w CO CO C
o> en
E-g §*
O
O
.Q
CO
T3
(D
^ B =
W C J2
CD 3
c a.
2 S -
*t p
CO •,*"""' /•-",
• ^^^ ** *AMW^
(VJ JWP* *MWr
-9
-------
281
Uj
s:
Objectives
Advise EPA on
*B
b
03
c**
CD
o
^ ______ MUM .
HMM O
© GL
03
*nm
Q
(0
03
O
0)
E
Q.
g
HH>
03
E
b
03
O
jj^l
w
0)
IMM
Q_
UJ
O
o
Q.
O
Q_
Q
o
-------
282
1
1
c
CD
1
,<0
a
5
o
o
o
El
'S4S
"5*1
o
.
•
CD
CO
Q.
•a
0
a
0
H—
to
"CO
H— •
O
co
"O
0
0
c
*o
c
g
S
"CD
Q
, • i,
CO
0
O
co
o
0 O.
a. OL
o co
>
0 '^
%.....
O 0
§ 2
±i O
.E c
"CD °
Q to
o
-Hh;
C
0
73
E
CO
O)
o
Q.
C
o
to
"T?
q>
b
o
cc
c
cd
*o
^••^
0)
HM
C
CD
E
ID
H—
O
.O
Q.
CO
0
Q
' ' i '•
(^f
CO
c
g
Q.
o
c i
o
to i
C i
0
E !
0
.^ :
Q.
E i
B
c
o
'•§
o
"•*-••'.
C :.,-.
2 : :
f,
t'Ll.
l:j
jjl
,
y,i
ill
«,,«
tnr J
|'j
: , t.
!l
1
r
3 ^
**'.
•'2!
^1
iiylr
i
Kg;
S.
1
^
III
i
m
Hf
s
1
i
^»
B
sf
I
-------
283
CD
O
0
3=
0
>
o
o
as
•3
§•
0 03
"O f-
0 C
0
C
0
JZ
0)
"U
c
0)
C3)
Jtf (0
O
0
=
TO
f~ **—
^ O
V)
0)
-------
284
_
J2 3 DC
E D)O
(0
0>
05
•**»»
D
D)
•g
"o
^•» ®>
< T3
^ 3 6 Q
III
LU
QL
V»5
Cw
••MM Jhita
u. O O
o w 3
^ CO
CD
D_
Z
0)
*% «**r
-------
285
'
c
Q>
S3
e
E
f*\
Q
CX.
si
i
o
o
H—
05
3
b
*•"««.
O
CO
c
g
N
'c
05
D)
6
CO
•p
CO
"D
05
(55
__
05
C
g
to
j|
c
g
N
4MMM-
C
05
O)
O
D)
C
^
£
O
O
05
T5
05
W
«lM«f
1o
U.
JQ
JO
T3
HM
A-MWNk
"D
0
05
"o
0
to
O
Q.
^~
Q.
0
£
v_
.O
"0
T3
O
e
0
S
.—
ft\
\f M
05
T3
0
£
0
0
0
£
O
co
o
"co
2
c
0
E
c
o
V—
">
lo
c
o
to
c
E
^o
"I'l'-l'in- .. , , •' • 'i|.n.|i.r, „, , II, | )Mi|,|l|.|i|l|,|,|l|lii|,,l|,|l|,,l
E i
2
D)
D.
^
4—» 1
CO I
±f 1
=0
rtv j
E
o
0 ^
^ 1
•
';:. f
-------
286
-------
287
1
C/l
1
03
i
*tl
*^3-'
^
"o
c:
o
-4
0)
CO
H
Uj
^?^
O
I 1 "
CO
_>
c
0
1i
Q.
O
•*- •
,
b
1o
-—p
o
Xi
CO
-J
"co
•*-•
c
0
E
c
o
c
LJJ
"B
o
* MMfe •
JQ
cd
> w
— 1
. «
c
g
"«*-•
CO
g
^_
,
05
^
O
CD
c
CO
CD
O
C
CO
^ V
£
•*•••
V
-~
(D
DL
,:-'-T ••"'-.' !
X^iW)
( C^-»r^3f
V J^^P^^^Bn
NjJ^ ^^f\
t^v
c
o
'•Jd
g
K ^
'i
V
0
O
1J
13
TJ
O
ol
••':.':•••• i
i! , '
8!J'':
,,,iV
ft''
, J.J., .
t'i
£•
;,,.':
i'fii';
3 " !
: F «r !
tftei
! ;
'„ '
in- :
a*
F;'
-*£
?lr'''
f
' n'
'V
B1'
-^:
iiJV
frj
?Vi
C
fj
&
liipi
p!
mu
-------
288
uJ
o
co
C
a
c
co
^ O m <1>
> ,s. »u ——
O
"£ £ o o
Q.) /|^ i—
• MM \y HH-*
H— /^\ r*
***M VlMM! •**•
o o
— O
03 —
v— m
73
w S m
CM <-U
4MMM S^
*-X — - ^
.S cd o
o
U
0) 0
£
.a 2
-2
to
"S "O « a.
c a> >» o
"D CO w
^JL,
-------
289
CD
8 "s
Q, c
& 0
°> IS
t£! ti>
i 2
**C •*-»
0
.,« jt
"F""*
*B
c
o
*•*-*
CO
*"*J Z3
UJ g
5 ^
i o -. ,-• . ..
CO
c
.2
Q.
o
c
o
2
"c
e
j>
E
•*-•
c
0
0
?~
"O
%-.
"D
C
Zj
: '•••-.
O
D
C7
CO
1
*
C/)
CO
c
,0
*«*-*
CL
O
"a
0
o
0
0
&+_
o
CO
• MMM
CO
"co
c
co
,
o
o
0
.^
lo
C
0
"co
JZ
o
0
H—
o
CO
c
o
o
£-
CO
CO
o
•*•••
Q_
^
<
Q_
LU
2
tf 1
O
Q.
0
i
LL,
i
-------
290
-------
291
Status of EPA Studies on a
Replacement Solvent for Freon
for the Determination of Oil and Grease
William A. Telliard
Analytical Methods Staff
Engineering and Analysis Division
Office of Science and Technology
USEPA Office of Water
Introduction
The United States, as a party to the Montreal Protocol on Substances that Deplete
the Ozone Layer and as required by law under the Clean Air Act Amendments of 1990
(CAAA), is committed to controlling and eventually phasing out CFCs as well as other
listed chemicals, because the chlorine in CFCs has been shown to be a primary contributor
to the depletion of the stratospheric ozone layer. Under both the Montreal Protocol and
the CAAA, CFCs will be phased out by the year 2000.
To be consistent with its commitment on CFCs, EPA proposed on July 3, 1991 (56 FR
30519) to amend certain analytical methods under the Clean Water Act (CWA) and the
Resource Conservation and Recovery Act (RCRA) to allow the use of alternative solvents
in lieu of CFCs that are mandated in these methods. Of the CFCs to be regulated by EPA,
only CFC 113 (Freon 113) is used in laboratory testing.
The analytical methods that EPA proposes to amend include Method 413.1 in 40
CFR Part 136 and Methods 9070 and 9071A in 40 CFR Parts 260-270 (EPA Publication
SW-846 by reference). At issue will be the possible effects of amending analytical
procedures that are presently required for monitoring under the authority of the National
Pollution Discharge Elimination System (NPDES) and the Resource Conservation and
Recovery Act (RCRA).
Method 413.1 is used in the CWA programs to determine total oil and grease content
in surface and saline waters and in industrial and domestic wastes. This gravimetric
method involves the acidification of the sample, followed by serial extraction of the oil and
grease with Freon 113 into a separatory funnel, evaporation of the solvent from the extract,
and weighing of the residue.
Method 9070 is used in programs administered under RCRA and is essentially the
same as Method 413.1. RCRA Method 9071A is used to recover low levels of oil and
grease from sludges, soils, other solid matrices, biological lipids, and some industrial
-------
292
wastewaters. The method involves acidification or drying, extraction of oil and grease with
Freon 113, and weighing of the residue after evaporation of the solvent.
In all three methods described above, the result, termed "total recoverable oil and
grease," is a method-defined parameter. Therefore, any changes to the specific analytical
protocols will change the manner in which the analytical result is derived and potentially
change the numerical value of the results for a given sample.
The Agency's initial efforts to find a suitable replacement solvent for Freon 113 have
been conducted by the Office of Research and Development's Environmental Monitoring
Systems Laboratory in Cincinnati, Ohio (EMSL-Ci). EMSL-Ci first used laboratory-
prepared, synthetic samples containing materials that represent "oil and grease" compounds
covering extremely wide boiling ranges, such as No. 2 fuel oil, No. 6 fuel oil, Prudhoe Bay
crude, animal lard, and wheel bearing grease. Reagent water was fortified with these
materials dissolved in an organic solvent to simulate real-world samples. These samples
were then extracted using several different solvents in place of Freon 113, and the residue
was determined gravimetrically. Subsequent evaluations used a limited number of actual
industrial waste samples. It was on the basis of this work that EPA proposed the
amendment of the methods in question.
In response to comments on the July 3, 1991 Federal Register notice, EPA will publish
another notice that withdraws the proposed rule to amend the methods. That notice will:
o Allow laboratories to continue to use Freon 113 in the immediate future.
o Strongly recommend that laboratories reclaim and/or reuse the Freon 113 rather
than vent it up a hood.
o Give the details of the Draft Study Plan for Sampling and Analysis Activities to
Support the Freon Replacement Study.
Draft Study Plan
The study of possible replacement solvents for Freon 113 will be run jointly by the
Office of Water, the Office of Solid Waste, and the Office of Air and Radiation. The focus
of the study is to:
o Determine a solvent/extraction system that provides equivalent performance to
the current Freon 113 method across a rang of effluent and sludge sample types.
o Determine the solvent/extraction system that poses the least potential risk to
stratospheric ozone.
o Provide clear direction for further study of one or two of the solvent/extraction
systems across and even broader range of effluents and sludge sample types that
are regulated under NPDES or RCRA.
1 clhard 2 Freon Replacment Study
-------
293
Five solvents in addition to Freon 113 will be examined. They are:
o Methyl tertiary butyl ether (MTBE) and n-hexane in a 20/80 mixture
o n-Hexane
o Dupont-123, a hydrochlorofluorocarbon alternative to CFCs
o Methylene chloride (dichloromethane)
o Perchlorethylene
A total of four different extraction procedures will be employed, depending on the
sample matrix. Aqueous samples will be extracted by all six solvents using traditional
separatory funnel procedures. Samples representing final effluents will also be extracted
using solid-phase extraction cartridges. Solid samples will be extracted with all six solvents
using traditional Soxhlet procedures and by sonication as well. Each sample will be
extracted in triplicate by each solvent/extraction system.
The types of sample matrices to be included in this study will represent wastes from a
diversity of industrial categories and activities, and will involve a wide range of oil and
grease concentration levels. The intent is to select wastewaters and solid waste matrices
that are applicable for investigating the subject oil and grease methodologies.
Table 1 presents a list of suggested sample sources and types based on these
recommendations. A total of 15 solid wastes and 12 wastewater effluents have been
selected for study. Final site selection should also take into consideration the following:
o accessibility of waste streams,
o limited commingling of wastes,
o EPA's previous experience at selected facility,
o cooperation of facility personnel, and
o willingness to conduct voluntary self-sampling to minimize labor and travel costs.
Because the purpose of the study is to compare the six extraction solvents, and not to
develop interlaboratory performance statistics, the number of laboratories will be limited.
The Region III Central Regional Laboratory will perform the SPE work. A single
laboratory will be contracted for the aqueous sample separatory funnel samples, and a
single laboratory for the solid samples.
Data Evaluation
The statistical approach to the evaluation of the data from the study will be consistent
with the complete block design. Each of the solvent/extraction combinations will be
statistically compared to the Freon 113 data. As the objective of the study is to determine a
lclhard 3 Freon Replacment Study
-------
294
suitable replacement for Freon 113 in the context of a NPDES monitoring method, the null
hypothesis for the statistical comparisons will be that the solvent/extraction systems are not
equal. Thus, solvent/extraction systems that achieve higher or lower results than the Freon
113 will be identified and may be eliminated from consideration. Although achieving
higher results than the Freon 113 method might be seen as a method "improvement", such
enhanced recovery of an operationally-defined parameter such as oil and grease presents
significant problems for implementation of the "improved" method under NPDES, as
thousands of existing NPDES permits contain oil and grease limits based on the present
Freon 113 methodology.
Any solvent/extraction systems that have performance equivalent to that of the Freon
113 will be candidates for further evaluation. Given the complete block design of the
study, the data may be compared across all sample types and within sample types as well.
As a result, EPA can identify those sample types which pose the greatest analytical
challenge for any of the solvent systems, and thereby direct further research efforts to the
evaluation of those sample types.
Follow Up Actions
After the completion of the first phase of the study, EPA will publish a notice in the
Federal Register of the availability of the results. EPA will also sponsor a workshop to
discuss and take comments on the study results.
The second phase of the study, evaluating the ,one or two most promising alternatives,
will begin following the workshop.
Iclliard 4 Freon Replacement Study
-------
295
TABLE 1
SUGGESTED SAMPLE SOURCES FREON REPLACEMENT METHOD STUDY
Industrial Category Possible Location Wastcwaler Stream
1. POTW
2. Petroleum Refining
3. Organic Chemicals
4. Timber Products
5. Iron & Steel
6. Oil & Gas
7. Pulp & Paper
8. Meat Packing
9. Meat Packing
10. Leather Tanning
11. Misc. Foods
12. Coil Coating
13. Restaurant
14. Soil
15. Soil
Region 2
Oil Re fineiy, Reg 3
Polymer Plant. Reg 2
Wood Preserving
Plant. Reg 3
Coke Plant. Reg 3
Coastal Production
Facility. Reg 6
Paper Mill, Reg 3
Rendering Plant.
Reg 7
Slaughter House,
Reg 7
Tannery. Reg 1
Margarine Plant,
Reg 5
Can Manufacturing
Plant, Reg 3
Region 2
Prepared by OSW
Prepared by OSW
Final effluent
API separator
effluent
Final effluent
Oil/water separator
effluent
Oil/water separator
effluent
DAF effluent
Bleach plant
filtrate
Final effluent
Final effluent
Primary effluent
Primary effluent
Oil/water separator
effluent
none
none
none
Municipal sewage
sludge
API separator
sludge
Polyo.vyalkaline
sludge
Sludge from creosote
operation
Sludge from coking
operation
Used drilling mud
Dewatercd sludge
Rendering/cooking
waste (chicken fat)
Slaughter house
sludge
Tannery sludge
Vegetable oil waste
Oil/water separator
sludge
Filter cartridge from
vegetable oil use
Contaminated with
synthetic motor oil
Contaminated with
motor oil
Telliard
Freon Replacment Study
-------
296
QUESTION AND ANSWER SESSION
MR. TELLIARD: Hi, Merlin.
MR. BICKING: Hi, Bill. I guess the obvious question is, are you
assuming that SFE is not going to be a viable alternative in the near future and that is
why it has been left off the list?
SPEAKER: What about 418.1,TRPH?
MR. TELLIARD: I am sorry?
SPEAKER: The TRPH, total recoverable petroleum hydrocarbons, the 418
series which is the IR methods, any work on those?
MR. TELLIARD: I am not that familiar with that method. Can somebody
help me? Joe?
Oh, okay. No, we have not looked at that. That is an option. I talked to
somebody at lunchtime about IR. Right now, what we are trying to do is maintain the
standard method that we have in place and just look for a, quote, "solvent substitution".
MR. SCHRYNEMEECKERS: Rick Schrynemeeckers, Enseco.
I was curious as to when you think the workshop is going to be and how
will you be announcing that.
MR. TELLIARD: We can do it a couple of ways. We will probably
publish it, certainly, in the Federal Register, but we have the mailing list from this
meeting and if you are really interested, we can just send you a flyer.
MR. SCHRYNEMEECKERS: Approximately when do you think that will
be?
MR. TELLIARD: End of summer, November.
MR. SCHRYNEMEECKERS: Okay.
MR. TELLIARD: I hope earlier than that, but knowing all the other trash
we have on our plate, that will probably be it.
Hi.
MS. ROTHMAN: Hi. Nancy Rothman from Enseco.
-------
297
This might be a real naive question, but are we also looking at the
permitting process to determine why we need oil and grease numbers, or can we change
that process and eliminate freon that way?
i
MR. TELLIARD: Thank you for asking that. Thank you very much. I
think that would be a wonderful idea.
Getting rid of this measurement would probably be a giant step forward for
science, but there are a whole group of people who went to school to get sanitary
engineering degrees who just would not know what to do without this number.
So, we have to come up with something. Now, there is TOC and COD and
AOXs and MMLSSs, you know. I think that has certainly got to be a viable option. No
one wants to hear that, but I think it is. I think there are so many other ways that we
can measure whatever the hell we are measuring, because no one knows, that I think
that would be a wonderful idea.
MS. ROTHMAN: How do we get it going?
MR. TELLIARD: Well, we are going to have to get phase one done to
show...because there are people who believe that if you mix a little blue and a little
green, you will end up with a freon that will work just the same. So, that is part of this.
You have to preface that by saying fine, we looked, and there isn't, you know, and put
that on the plate.
That has been whispered in the halls but not spoken very loudly.
How are you doing?
MR. LEVY: Hi, Bill. Nathan Levy, A&E Testing.
As long as we are throwing out methods, can I put in my order to throw
out the BOD test? It is almost as worthless as the oil and grease.
I do have a question, though, and another comment. Let's not use n-
hexane, because that is going to increase the labor considerably, since it is going to float
on top of the water.
But my other question was on the liquid-liquid. Is your study going to
really include some liquid-liquid, or are they all going to be separatory funnel extractions
with those solvents?
-------
298
MR. TELLIARD: On the first shot through for the five that we are looking
at, we are going to use separatory funnel. Then, when we narrow it down, if we do
narrow it down, to one or two, we have archived a number of these samples, and then
we will run continuous liquid-liquid extraction on those to see what type of comparability
we get.
I don't envision that liquid-liquid would be a problem as an alternative.
MR. LEVY: Well, the time factor may be. Right? You know, it is going
to...liquid-liquid for the organics extraction is 18 hours. What do you expect a liquid-
liquid on a grease would be?
MR. TELLIARD: Don't know yet.
MR. LEVY: All right, thanks.
MR. THOMAS: Hello, Bill. How are you doing?
MR. TELLIARD: Hi.
MR. THOMAS: My name is Roger Thomas. I am from Viar and
Company.
I was wondering in the solvent selection process, have you taken into
consideration the boiling points of the solvents that you are selecting?
The reason why I said that is because if it is a high boiler, you have to boil
down the solvent at a higher temperature, thereby losing some of the low boiling oils and
greases and fats.
MR. TELLIARD: Right. You are absolutely right. No, we didn't.
MR. THOMAS: Okay.
MR. TELLIARD: Running concurrently with this study, the Canadians are
doing a similar study, and we met by accident in Colorado, and it turned out we all had
the same solvents which shows that this is a real warped community.
We are going to share data with them. So, you figure that we will end up
with about 30 different types of matrices. So, we will have a fairly substantial data base.
They are out sampling also at this present time using the same sets of
solvents. So, it looks like we will have a fairly stable data base to work from.
Thank you very much for your attention.
-------
299
MR. TELLIARD: Our next speaker is Joe Raia from Shell Development.
Joe and I did a similar dog and pony show in Colorado about a month and a half ago,
but Joe is going to again be talking about solvent substitution for freon.
MR. RAIA: I just want to say I am really glad to be able to have the
dialogue that we do with Bill Telliard, and that his sense of humor is a real help in all of
this work.
The work that I am going to present here, as Bill said in the introduction, I
have presented in part at a recent API meeting in Colorado. I will also present some
additional new data today.
We are looking at alternative methods for oil and grease which includes
the use of different solvents, and we have also looked at some solid phase extraction
techniques.
As you have heard before, what we have to do is replace freon in the oil
and grease tests. This has been dictated by the Montreal Protocol and the timeline by
the Montreal Protocol was by the year 2000.
The Clean Air Act Amendments, though, also require the same thing by
law, and recently, the President moved that deadline up some because of data from
NASA aircraft probes which reportedly showed a more serious problem with ozone
depletion.
The thing is the amount of solvents...the amount of freon used in
laboratory tests probably account for a very small portion of all the freon used out there.
I think the Federal Register notice that EPA issued estimated on the order of 1 percent.
I am not sure how you come up with a number like that, but if, while trying
to find this alternative method we do solvent recycling, that will essentially cut down on
any freon emissions from laboratory testing to something that we would not be able to
even measure.
Now, the number of EPA methods that are really involved in freon use, are
those that I have outlined up there on the slide under both the Clean Water Act and
RCRA. The gravimetric 413.1 is the main one that we have talked about here, in which
-------
300
you extract by separatory funnel and then weigh the oil and grease.
The IR method, 413.2, is really not a promulgated method, but it is in the
EPA Methods Manual and is used in some testing.
The TPH method someone just asked about is 418.1, and that is the
method where silica gel is used to treat the oil and grease extract, and then you finish
the test with an IR analysis. Freon works very well with that, because it has no
absorption in the IR region where the aliphatic hydrocarbons absorb.
The other methods, and I won't go through all of them, are shown up there
in the slide. Under RCRA solids, there is the Soxhlet extraction gravimetric finish, and
in the next slide, I will mention some concerns we have with changing the solvent away
from freon in that type of a procedure.
Now, I will say industry, and I know that this has been expressed in API
comments to the Federal Register notice that came out in July of '91 and then a later
one in October, the industry supports EPA and the need to comply with the Montreal
Protocol and the Clean Air Act Amendments to phase out and eliminate CFCs.
Alternatives need to be found, and the solvent change concerns are
equivalency, NPDES compliance, and RCRA methods. For RCRA, I am speaking again
about the Soxhlet method primarily. For the solvent mixture, hexane/MTBE, that was
proposed in the Federal Register notice, there was only a limited amount of data on
RCRA methods, and we wonder how such a mixture will really work in a Soxhlet type
apparatus.
The petroleum hydrocarbon methods use silica gel treatment to remove
polar organic compounds leaving the hydrocarbon fraction. When you change solvents,
what effect will that have on the silica gel treatment and how will it change that
number?
The safety and human health concerns have already been mentioned. This
includes flammability of the non-freon type solvents, such as hexane, and the human
health concerns of these solvents, methylene chloride toxicity, for example, and hexane
has some toxicity concerns in terms of nerve ending effects.
We need to have a cooperative and focused research effort with EPA and
-------
301
industry and vendors and anyone else interested in helping to solve this problem, and we
need to get it done at an accelerated pace to select the alternative methods.
The potential areas for research are alternative solvents testing of test
samples of different industrial categories, such as Bill has mentioned in EPA's work plan.
This includes produced water, wastewater, and solid waste.
Solid phase extraction, certainly, is a viable option, and it shows a lot of
promise, and we have seen a lot of data earlier today on that.
Petroleum hydrocarbon methods really come under this also,-the silica gel
method and also GC/FID methods for petroleum hydrocarbons. At the Colorado
conference, there were a lot of papers presented on the many different ways that
petroleum hydrocarbons were measured by GC/FID.
Supercritical fluid extraction is something that we should consider
especially for solid phase matrices.
And then, direct oil in water analyzer development. Wouldn't it be really
neat to not have to use an extraction solvent at all for an oil and grease type
measurement, if we are indeed going to have to make oil and grease measurements?
There is some work out there that is being done, but very preliminary at this point.
There are some commercial analyzers available, but all of these have not been evaluated
completely and really proved reliable in operations in a lot of different applications.
In the meantime, solvent recovery is what we need to be doing.
The work plan that we followed and that I will show data on here is
shown up there in the next slide. We compared solvents: freon, normal hexane, the
hexane/MTBE mixture, and dichloromethane.
We looked at liquid-liquid extraction, and by that I mean the separatory
funnel extraction and not continuous liquid-liquid extraction, and solid phase extraction
using an envirelute column type material that is available from Varian.
We looked at the solvent removal temperature effect on freon extracts and
dichloromethane extracts at 70 degrees C, which is the temperature used in the standard
method, and at 90 degrees C, which is the temperature that we used for removing
normal hexane and normal hexane/MTBE solvents.
-------
302
We did some characterization of the extracted oil and grease fractions.
The samples that we looked at in Shell were produced waters and some of our refinery
and chemical wastewaters. The objective here is to share this data with API and EPA in
providing input into the method selection.
The liquid-liquid extraction schematic is shown there in the next slide,
where you take 1 liter of the water, acidify it, extract with your solvent three times,
evaporate the solvent, and then weigh the residue.
In solid phase extraction, this was the column technique, take 1 liter water
sample, acidify, pass it through the solid phase, elute the captured oil and grease with
the solvents that we were testing, evaporate the solvent and weigh.
This shows the glassware that we used for the solid phase extraction. We
simply put the sample in the sep funnel and let that percolate through the cartridge, and
then rinse the bottle...well. Next, move the column cartridge to the other apparatus on
the right side of the slide and carry out your elution step, and then rinse the oil and
grease out of the bottle onto the cartridge in that elution step.
These results show comparisons of the different solvents with the samples
tested. The slide shows the solvents pretty much in the order of the polarity of the
solvent from hexane through dichloromethane. The lowest numbers there that you see
are for the refinery and chemical wastewaters. The right-hand slide shows levels for
produced water.
The trend, generally is the more polar the solvent, the more oil and grease
you extract. This is for the liquid-liquid extraction.
The next slide shows the same kind of results for solid phase extraction.
You will note that for the sample W, there were a couple of high oil and grease numbers
that we got for the non-polar solvents. We wondered about whether that was just a
sampling inconsistency or whether there was something else going on. We wanted to
take a look at that.
So, we looked at the oil and grease extract by GC/mass spec and found
some silane material contributing to the weight. Typically, the blank levels from the
solid phase cartridges were quite low, actually, on the order of 1 to 2 milligrams, but in
-------
303
that particular case, there was a lot more blank material that was contributing to the
weight, and it was identified by GC/mass spec to be silane. A large portion of the
weight was due to silane, a material evidently used in the manufacture of the column.
So, an important consideration that we are going to need to keep in mind
if we are going to use solid phase is to have high quality or appropriate quality columns
of consistency.
This one is a vertical slide. We got our statistics people to take a look at
the data. The top slide is for solid phase, and the bottom illustration is for liquid-liquid
extraction for the refinery and wastewater data. I am sorry, this is for produced water
data. The next slide will be for the refinery and wastewater data. Box plots are used
here to compare the results.
Now, for these box plots, basically, let's take the first one up there on the
dichloromethane. You can see this bottom whisker which represents 25 percent of the
data. The total box itself contains 50 percent of the data, and the top whisker would be
another 25 percent of the data. The line in the middle of the block is the median of the
values.
Those data are not for duplicate analyses but for all of the samples tested
which included different locations.
This slide shows results for the refinery and chemical plant wastewaters,
and in the case of freon on the top portion of the slide, the stars represent outliers.
Now, what does all this mean? We tried to make a summary of these
results in this next slide, - the top portion being the produced water data and the bottom
portion the refinery and chemical plant data.
The dashed line represents the freon value that you get with liquid-liquid
extraction, and the other solvents and techniques show how well or how different they
compare to the freon value and also gives you some notion about the variability between
the different matrices that we looked at in produced water locations and in refinery and
chemical plant locations.
Actually, the hexane liquid-liquid extraction, in the case of produced water,
comes out quite close to the freon liquid-liquid extraction. The more polar solvent,
-------
304
dichloromethane, is higher and shows more variability as does the hexane/MTBE
mixture.
The other two plots at the right of the slide are for the hexane/SPE and
hexane/MTBE/SPE. The hexane/SPE is not that different from the hexane liquid-liquid
extraction. So, hexane/SPE may be a viable alternative if we have to go to hexane.
Now, this next slide shows a comparison of the effect of the evaporation
temperature. If we start at the top and look at W70 and W90 for Freon, this compares
the effect of evaporation at 70 degrees and at 90 degrees, which shows a decrease in the
oil and grease value.
The other side of the slide shows a sample from the same location that was
tested with dichloromethane. These are all liquid-liquid extraction results.
In general, the trend of the slide shows, as you would expect, that if you
have to go to 90 C to remove your solvent, you are going to have a lower oil and grease
number, if you have volatile oil and grease present. These samples show varying
amounts of volatile oil and grease.
This next slide shows some characterization work that was done on a
produced water sample. The oil and grease was treated with silica gel. The non-polar
fraction is essentially total petroleum hydrocarbon done in the standard way.
The polar fraction was obtained by washing the polar material off of the
silica gel column and getting a weight on that. The heavy polar material represents the
difference number of oil and grease that was not able to be washed back off the silica
gel column.
Essentially, this shows that a more polar solvent is going to have a large
effect on an oil and grease value, for samples with different kinds of oil and grease
present.
This is data for a refinery wastewater done in a similar way which
compares freon, either by solid phase or liquid-liquid, with dichloromethane.
In solid phase extraction, this next slide shows a standard that was made up
containing mineral oil, hexanoic acid, and ortho-cresol, and this slide shows what each of
the solvents was able to do with the different type of organic compounds present. Again,
-------
305
as we have seen earlier today, the phenolic type compounds are not recovered well.
To summarize results, this study, with the limited samples that were
examined, shows all of these parameters will affect the oil and grease value: the
extraction solvent polarity; the extraction method; the oil and grease type, non-polar and
polar; and the solvent removal temperature.
The trend was that the oil and grease values increased with solvent
polarity, and that was observed by both LLE and solid phase extraction. The order was
dichloromethane was greater than hexane/MTBE, greater than or equal to freon, greater
than or equal to hexane.
The oil and grease values by solid phase extraction were greater than by
liquid phase extraction with the higher polar solvents like dichloromethane. The oil and
grease values are lower with higher solvent removal temperature.
SPE can yield column artifacts and consistent quality columns are
•necessary to use these materials for the oil and grease tests.
Another observation in this work was that the non-freon solvents tended to
have interferences from salt to a greater extent than did freon.
I would like to thank and acknowledge other people in Shell who have
contributed to this work. Some of these folks are here today. Tom Randolph is here
with Shell Offshore Environmental Affairs. George Stanko is here, and these other
people shown on the slide are from Shell Development Company who helped either with
statistics or other analytical work. Thank you.
-------
306
QUESTION AND ANSWER SESSION
MR. TELLIARD: Any questions?
MR. SCHRYNEMEECKERS: Joe, I guess this is a question for both Joe
and Bill. Is the purpose of the search for the missing solvent, so to speak, is part of the
mission to really find a solvent that, quote/unquote, is going to give the same numbers?
Because I know when you deal with certain permitting regulations and offshore oil and
grease numbers, a lot of things are driven off the number that you get being above a
certain level.
So, is the purpose of the substitute solvent one issue to try to find a solvent
that gives satisfactory numbers or similar numbers, and if you can't find one, what are
your thoughts on where we are going to go with that?
MR. RAIA: I think the purpose is to see if you can find one. Whether or
not that is going to happen I don't think any of us know yet, although we have seen some
data, limited data, here in a couple of industrial categories, that would show that hexane
might be a solvent that will come close to freon. Actually, hexane was used for oil and
grease up until about 1978.
How close we have to come to say that it is equivalent, I am not sure how
that is going to work out. Bill may have some input here on whether we may have to go
in and adjust permits.
MR. TELLIARD: I think one of the answers is that you have a situation
where normal hexane was our substitute for carbon tet way back in the dark ages, and
then, right away, we switched over almost immediately to freon.
Normal hexane will probably give you a lower number, thereby not driving
too many people into non-compliance. That is an assumption. But then again, we are
looking here at oil and grease in the sense of petroleum hydrocarbons. That is why we
are out looking at meat packers and margarine plants and rendering plants as to what
happens to mineral and animal fat when you get into this thing. We don't know.
Where do we go? We go have a meeting this fall and talk about it,
because I don't know. There is a hope that we are going to come close enough, quote,
-------
307
for government work. I am not sure that that is going to happen.
MR. SCHRYNEMEECKERS: Thank you.
MR. YOUNG: I am John Young with Westinghouse.
I have got a question about the use of more polar solvents. Would you not
: expect a significant contribution by pH of the sample influencing whether you would
extract the acidic compounds or the basic compounds?
MR. RAIA: Yes, you would.
MR. YOUNG: Yes. Are you going to try to control that?
MR. RAIA: In all this work, we adjusted the pH to less than 2.
MR. YOUNG: Okay.
MR. RAIA: As you do in the standard method.
MR. YOUNG: Thanks.
MR. TELLIARD: Thanks, Joe.
-------
CO
308
Z CO
< z
LU O
CO CD
< DC
UJ <
DC O
CD O
00
DC
O 3
CO O
Q ^
h- LLI
UJ CO
LU X
> ^
h- X
DC
UJ
h-
UJ Q.
CO
tjco
-3
UJ
O
LU CC
o>
Ul
LL -*-
2 •- co
OZ c c\j
°wffS
^Z5K-
UJ < j^"^
-I \~ O ^
CO
go
fo
10 CO
<
<
z
o
-------
309
CO
CO
LJJ
Q)
LLJ C
CO 03
< £
LLJ 0
CC O
"
co V
o
O
Q.
LJJ
DC
o
o
o
CVJ
o
o
o
CM
CO
0)
JD
CO
b
LL
O
CD
O)
O>
A
O
O
O
jo
0) D
i; < DO
-------
310
LU
Q_
•V.
|_
F^
LU
Q_
CO
oX
oo
CO
o
o
T
a™ ™
Q ^t CC ^t
"
CO
CO co
^0
1 °
<_^]
CC
CO
CO
,_r
"y* CO
-------
311
ETHOD CHANGE
OPERATIVE WORK
^ O
U) O
0) DC
> 0
< UJ
<£ oQ
OIL&GRE
CONCERNS
§|o
iig
Q. 1- <
^ oo
o
troleum Hydrocarbon M
0>
a.
^
X -J h-
ARCH WIT
"ANACCE
ALTERNA
\
UJ ^ UJ
CO
-------
312
CO
<
LU
DC
<
LU j:
o ^
^0
<5
X ^
o w
^ CO
Q LU
O CC
fs
si
LU DC
CO LU
< 0.
£0
cc o
OQ
oS _j
_i <
°2
LU
h-
0
CL
h- «>
cojy
LU a:
H 0
t o
O LU
°!<
F§
[S<
|_ £C
^^
SOLVEN
OM INDL
£E
P CO
• ^ iii
-------
313
o
2
CO
LU
t-
Q
O
I
K
LU
^
LU
_J CO
0- <
\x m
rr tt
go
06
LU
>
|
DC
LU
tj
<
O
CVJ
6
CO
CD
CO
o
0>
5
o
&
UJ
CD
D
+*
CO
CD
Q.
E
actions
o
CO
CO
CD
O
00
•o
CD
CO
Solvent
CD
CO
Q.
E
0
O
CO
I
I
c
o
CD
LL
CD
c
CO
X
CD
X
1
c
LU
ne/MTB
CO
X
CD
X
1
c
CD
>methan
^^
5
JC
O
b
c
o
Extract!
ompare
0
LU
0.
CO
T3
c
CO
LU
-J
-J
O
CD
LT
"c
CD
^•r
"5
CO
*-
CO
<11
F
O
O
o>
OD
O
0
^^^
co
i_
*-
X
LU
5
1_
haractel
o
Wastewaters
Chemica
X
L»
O
£
•»-
CD
QC
C
o
I*—
o
^CD
CD
CO
•o
O
^1
*-*
CD
2
L.
O
«^-
LU
•o
C
co
= "o
CO
CO
jj>
75
CO
*- > *-
*- > co
Produced
Tes
co
o
JOO
"55
CO
CD
CO
CO
-------
314
UJ
CE
15
O UJ
00
<
DC
gg
g<
UJ
Q UJ
=5 CE
go
\
-------
315
oS
9O
_J
O
CO
CO
LU
CD
LU
Q_
xli C/D
Z CO
p <
oz
DC ,.,
1 tn
><<£
III
LU CC
coo
•
1
O
LU
h-
co
1
o
LU
O
ID
0
LJLJ
QC
_,rf
LU
_J
Q.
2
4K>
oc<
LU CO
i
5g'
-g
^
CO
CO
LJ
<
z
<
9
a.
LU
^
o
CO
LU
GC
X
o
III
-------
316
CO
CO
LJJ
DC£
CO
o
co 2:
cc g
o
,- O CC
COo I-
•^UJ
Z Q
^5
o
CO^Q
< w 5
111 < O
DC 2 =i
O K >-
WUJ m
UJ
CO
135
o:
o
o0
m
™
CD
in cq CD CD
CD" co ^t" o
t- CM CO ^t
cq in CNJ
f-' CD CD
CD in i^
CO
cq in in oq
CD CD CD CD
CM CM CM CO
Q T- CD CO CM
Q CM CM" T^ T-*
DC O) CM _
Q co oi "^
CO
h-*
000000
O CO CD ^t CM
CO h* O> CO
co in CD o>
UJ
1
IIJ
OL
CO
T—
1
O
LU
tr
LL
H V
UJ
z
<
X
tL
s
0
cc
o
Zj
X
o
Q
19 M
UJ
S o
<° co
CO
CD*
uj a: in
X LL X Q
-------
317
1—
1
_ J
^^^N
M^/
rn
\jj
DC
C/l
wy
— — ^
I MM
>-
^*
^-
LU
nr
L^_
OS
I
1
O
rr\
\JJ
FT
k^_
Q_
^^
o
v«x
CO
LJJ
>
— 1
CO
LU
^
Z
m
cc
Lll
h-
— i
^?*
h-
O
n-
LL
•
X
LJJ
LU
CO
<
I
Q_
__
i
_j
0
CO
>-
CD
^
2
• it
LU
CO
wX
2S
^&J
or
o
00
i
•••M
O
: \////////A
: : : : 1 1] M CD
: : : tXW\ yj
; • : :KH n^
: : : •• U-
\S//////////s
1 1 1 M 1 It J 1 j CD
: ; V^miH
\'.'.'.'. \
'• '• ^/////A
: : : f ' 1 J 1 | f | DO
II 1 m £
M i M S 8
: i i i i •
: : i : i J
i : ! : i 0
: : : : : E ^*
MM! Q
N ! M r
: ; i i i jn
M M S
M M Q
: ' ' • ' MJ *C
i i i i i 1
CM
iri
«•/
^~
CO
CM'
CD
CO
iri
CO
CM
CD
*"
h*
CD
^~
o>
^;
"^*
T™"
^^
•
in
to
CM*
CM
CO
^
CM
CO
t
CO
Y1—
in
WJ
CM
O
o>
^
CD
^^t
CO*
CO
CO
CM
oi
\JJ
CO
^D
CO
^
f^
^»
CO
CD
o>
CD
\Jk^
N;
T-
T—
0.
h-
LLI
>
^J
CO
CO
^_
1
•z.
LU
cc
u.
^
LU
z
3
LU
X
I
LU
Z
J—
LU
O
DC
O
_j
T-
O
o
i
LU
K
p~
i
z
s
LU
X
1
o o o o o o o
CM O CO CD ^t CM
UJ
1
CO
* ^fc»
LU T LU
O
a: LU
LL x a
-------
318
Produced Water Data
Method: SPE
120
5^ 100
o>
E
-------
319
Refinery and Chemical Plant Data
Method: SPE
24 -
20
O)
16
0)
(0
S 12
o
03
4 -
0 -
DCM
Freoo-113 Hex/MTBE Hexane
Sdvwrt
Refinery and Chemical Plant Data
Method: LLE
24 -
20
O) 16
E
o
(0 12
ca
o
O 8
T3
C
03 4
I
DCM Freor>-113 Hex/MTBE
Solvent
Hexane
-------
320
Produced Water Data
120
^? 100
(D
co 80
CO
-------
321
LU
DC
> en
CD LU
H-
CO _j
LU < O
CC > o
CO O O)
CO -^ °°
> m O
1J DC
< o
LU CO
tr u.
o o
h-
O
LL
LU
LU
LU
CO
o
LU
or
LL
£L
2
LLJ
o
§
o
o>
m
o.
o
o>
m
a.
m
a.
o
o>
m
CL
m
LU
£L
o
o>
m
LU
o.
<
CO
-------
322
o
m
cc
o
< Q <
DC UJ >
< l~
I < Q
O
o
CO
o
CO
o
CVJ
DC
o
Q.
rr
UJ O O
X Q_ 2
O
DL
I
a:
o
a.
i
a:
o
a.
I
-------
323
N H EC
Q- O LJJ
^^J f^^ * ^^
i LL q
o^s:
< Q UJ
DC UJ f-
< h- CO
T «
UJ p >-
n^^|
^ CD ?
08 ^. DC
O
CM
J
CO
CO
eg
co
CS
J L
9 S
LU T-
9 S
LU CM
CO
2
O
LU
CC
LL
LU
CL
co
inoinoinomo
CO CO CVJ CM -r- T-
CVJ
CO
DC
O <
°- O
CL
2
LU O O
X CL 2
DC
3
O
CL
I
DC
S
o
a.
o
CL
I
-------
324
o
c/)
LU
LT
CO Ej <
< LiJ O
Z LU Z
< C/) <
LU < X
COIUJ
< CL X
H 0-T
o« CO
.J
O
<
LJJ
i- ••- w
O CD CO
CM -^ h-
O> CM O) GO
O) O) O) O)
O O O O O O O
CM O CO CD ^ CM
LU
O
LU
CD
O
CO
o
Q LU
— 7
O —
<2
-j _i
So
ss
in o
-------
325
CO
DC
<
CO
CO
CO
LJJ
DC
E MEASUREMENT DEPENDS ON
CO
^
LU
CC
O
oC
1
0
TION SOLVENT POLARITY
TION METHOD (LLE, SPE)
lEASE TYPE (NONPOLAR, POLAR)
T REMOVAL TEMPERATURE
O O 2
^ ^ CD uj
CC DC o0 >
X X — O
LU LJJ O CO
K
: INCREASE WITH SOLVENT POLARI
BY BOTH LLE & SPE
CO
LU
ID
— J
^
^
O
0
EXANE/MTBE >/ FREON >/ HEXANE
x
2
0
0
LU
0
CO
t*f
CL
3
O
CL
O
ofl
0
i
i~"
-^
ARE LOWER WITH HIGHER SOLVEf
EMPERATURE
CO 1-
LU -I
=><
_j ^
^ O
> 2
O UJ
oO DC
O
CC
<
_D COLUMN ARTIFACTS (SILANE)
T QUALITY COLUMNS ARE NECESS
Ml 2
— — iM
>~ h-
~z? CO
< 55
0 2
LU O
QlO
CO
X
1—
;ALT INTERFERENCE OBSERVED w
1 SOLVENTS IN BOTH LLE & SPE
u> ^
00
LUUJ
CO DC
LU2
DC Q
O "3*
2 Z
-------
o
326
EDGMENTS
_i
o
V
O
Offshore
0)
CO
co"
L-
"cd
s
75
c
0)
E
c
o
•—
c
LU
I*
a.
•
_j
o
Q
DC
2
K
•
O
O
c
(II Developmel
HI
^•r
CO
"5
.£2
75
<
1
^
CO
i
d
•
o
O
^
c
0)
E
Q.
0
Q)
0)
Q
"55
CO
CO
.J2
5
35
^
Z
LU
I
••••
o
o
H
<
d
O
Development
—
CD
CO
CO
.2
S
35
K-
LU
Lii
LLJ
C>
O
c
0)
E
CL
O
Q)
(D
O
"55
CO
75
.9
75
<
z"
LL
jj
O
Development
o>
CO
— ^
(d
mo
75
c
of
LU
CO
DC
CD
-------
327
ALTERNATIVE METHODS FOR OIL AND GREASE ANALYSIS
WHICH USE NO CHLOROFLUOROCARBONS
Authors: J. C. Raia
D. K. Lee
P. J. Staley
Shell Development Company
Houston, Texas
Presented at: 15TH Annual EPA Conference on Analysis
of Pollutants in the Environment
Norfolk, Virginia
May, 6-7, 1992
-------
328
ABSTRACT
Because the Environmental Protection Agency (EPA) must phase out the use
of chlorofluorocarbons (CFCs) in laboratory test methods, substitutes for
CFCs or alternative test methods need to be adopted. In July, 1991, the
EPA issued proposed analytical methods under the Clean Water Act (CWA)
and the Resource Conservation and Recovery Act (RCRA) to allow the use of
alternative solvents in place of chlorofluorocarbons (CFCs) in EPA test
methods. EPA proposed an 80:20 mixture of n-hexane and methyl tertiary
butyl ether (MTBE) solvent to replace Freon-113 in oil and grease tests.
In October, 1991, EPA issued a second notice which addressed the need for
additional research before a suitable replacement solvent can be selected
for oil and grease tests.
In anticipation of the EPA method change to eliminate Freon-113 in the
analysis for oil and grease, work was initiated by Shell Offshore Inc.
(SOI), and Shell Development Company Westhollow Research Center (WRC) to
evaluate alternate extraction solvents and also alternate procedures for
the oil and, grease test. This effort included the analysis of produced
water samples from the Gulf of Mexico and west coast, and also samples of
wastewater effluents from refinery and chemical plant locations.
Samples were analyzed for oil and grease by EPA Method 413.1 and with
modifications to allow comparison of different solvents which included
Freon-113, n-hexane, dichloromethane, and the new EPA proposed 80:20
n-hexane/MTBE solvent mixture. Data were also developed for a
potentially new technique for oil and grease, solid phase extraction
(SPE). The SPE method offers solvent waste reduction advantages.
Results of this continuing work are presented, which show that the oil
and grease measurement will give values which will depend on the
extraction solvent polarity, extraction method, and type of oil and
grease present. This poses a true challenge in finding an equivalent
solvent to Freon-113 for NPDES monitoring, since all existing permits
have limits based on the Freon-113 method. Further work is needed at an
accelerated pace with joint industry and EPA participation to evaluate
and select the most suitable test method for oil and grease which uses no
chlorof 1 uorocarbons.
-------
329
ALTERNATIVE METHODS FOR OIL AND GREASE ANALYSIS
WHICH USE NO CHLOROFLUOROCARBONS
by
J. C. Raia, D. K. Lee, P. J. Staley
INTRODUCTION AND SUMMARY
There continues to be increasing concern that the • emission of
chlorofluorocarbons (CFCs) into the atmosphere causes significant
depletion of the stratospheric ozone layer. In 1987, a multi-nation
treaty called the Montreal Protocol was signed to control and reduce the
production and use of ozonerdepleting substances. In June 1990, the
United States and some 100 other countries agreed to amend and strengthen
the Montreal Protocol and to phase out and eventually ban the production
and use of CFCs, halons, and carbon tetrachloride by the year 2000 [1,2].
The Clean Air Act Ammendments (CAAA) enacted in November 1990 (PL
101-549), by law require control and phaseout of CFCs and other listed
chemicals by the year 2000, similar to and in some cases more stringent
than the Montreal Protocol. Congress has recently enacted an excise tax
on all ozone-depleting chemicals listed in the Montreal Protocol and in
the 1990 CAAA. The phaseout is now being accelerated because of findings
by recent aircraft probes of the northern stratosphere by the National
Aeronautics & Space Administration.
Because the Environmental Protection Agency (EPA) must phase out the use
of CFCs in laboratory test methods, substitutes for CFCs or alternative
test methods need to be adopted. On July 3, 1991, the EPA issued
proposed analytical methods under the Clean Water Act (CWA) and the
Resource Conservation and Recovery Act (RCRA) to allow the use of
alternative solvents in place of chlorofluorocarbons (CFCs) in EPA test
methods (56 Fed. Reg. 30519). EPA proposed an 80:20 mixture of n-hexane
and methyl tertiary butyl ether (MTBE) solvent to replace Freon-113 in
oil and grease tests [3]. On October 8, 1991, the EPA issued a second
notice ( Fed. Reg. 50758 ) which acknowledged the public comments to the
earlier notice, and need for additional research before selecting a
suitable replacement solvent for Freon-113 in oil and grease tests.
In anticipation of the EPA method change for oil and grease, work was
initiated by Shell Offshore Inc. (SOI), and Shell Development Company
Westhollow Research Center (WRC) to evaluate alternate extraction
solvents and also alternate procedures for oil and grease measurement.
This effort included the analysis of produced water samples from the Gulf
of Mexico and west coast, and also samples of wastewater effluents from
refinery and chemical plant locations.
Results of this continuing work are presented in this paper, which show
that the oil and grease measurement will give values which will
-------
330
depend on the extraction solvent, extraction method, and type of oil and
grease present. This poses a true challenge in finding an equivalent
solvent to Freon-113 for NPDES monitoring, since all existing permits
have limits based on the Freon-113 method. Further work is needed at an
accelerated pace with joint industry and EPA participation to evaluate
and select the most suitable test method for oil and grease which uses no
chlorofluorocarbons.
EVALUATION OF ALTERNATE OIL AND GREASE METHODS
Samples were analyzed for oil and grease by liquid/liquid extraction
(LLE) by EPA Method 413.1 [4], and with modifications to allow comparison
of different solvents which included Freon-113, n-hexane, dichlormethane
(DCM), and the new EPA proposed 80:20 n-hexane/MTBE solvent mixture.
Data was also developed for a potentially new technique for oil and
grease, solid phase extraction (SPE). The SPE method requires no CFC and
offers solvent waste reduction advantages.
The boiling points of the different solvents tested required different
temperatures for solvent removal from the extracts: 70 C was used for
Freon-113 and dichloromethane, and 90 C for n-hexane and n-hexane/MTBE
mixture. The effect of solvent removal temperature on the oil and grease
value was tested.
The extracted oil and grease fractions were characterized and compared
according to their non polar and polar content. The hydrocarbon (non
polar) fractions were measured using silica gel treatment as in Standard
Method 5520F (Standard Methods for the Examination of Water and
Wastewater, 17th edition). The polar fractions were obtained as
material recoverable from the silica gel by washing with methylene
chloride; the heavy polar fraction was non recoverable material as
determined by difference.
TEST RESULTS FOR PRODUCED WATERS AND
REFINERY AND CHEMICAL PLANT WASTEWATERS
Results comparing the Freon-113 standard method with different solvents
by liquid/liquid extraction (LLE) and solid phase extraction (SPE) are
summarized for the samples shown in Figures 1 and 2. The data represent
the mean of duplicate analyses and reflect the combined variabilities for
the sampling and the analysis. Comparisons were difficult for some
samples because the oil and grease values were near or below the
detection limit of the method. A statistical analysis of all the data
was made using a complete block design. Data comparisons are shown in
box plots in Figures 3-5. The produced water data set were grouped
separately from the refinery and chemical wastewater data set.
The results show that generally oil and grease values by both LLE and SPE
increase with solvent polarity in the order dichloromethane > hexane/MTBE
>= Freon-113 >= hexane. A solvent system based on n-hexane may be a
suitable alternative to Freon-113. The single solvent gave better
-------
331
precision than the 'solvent mixture. Oil and grease values by SPE were
higher than LLE with higher polar solvents.
Sample W by SPE was atypically high for the less polar solvents relative
to dichlormethane. Further analysis of this sample's hexane and freon
extractable oil and grease fractions by gas chromatography-mass
spectrometry revealed a substantial contamination of silane material.
Blank runs of the SPE columns were generally in the 1-2 mg residue range.
A consistent acceptable quality of SPE material will be a necessary
requirement for application of this technique to the oil and grease
analysis.
The oil and grease results reflect the temperature used to remove the
solvent from the extract. Figure 6 illustrates the decrease in oil and
grease with increased temperature in the solvent removal step usinq 70 C
and 90 C.
The polar and non polar character of the oil and grease in a sample will
influence the oil and grease value. Produced water and refinery
wastewater samples were extracted and the oil and grease fraction then
treated with silica gel. The hydrocarbon (non polar) fraction and polar
fractions were measured as shown in Figures 7 and 8. The results show
the more polar solvent, dichloromethane, extracts more of the very heavy
polar oil and grease than does the less polar freon solvent. The amounts
of hydrocarbon and polar material are more comparable by each solvent by
both LLE and SPE.
FUTURE WORK
This study has shown that the oil and grease measurement will give values
which will depend on the extraction solvent, extraction method, and type
.of oil and grease present. Because the solvent change can impact
compliance to existing permit limits under the CWA, sufficient data
should be collected over time for permitted discharges using the
replacement solvent and method selected. Permit limits could be
adjusted if necessary to account for any bias caused by the solvent
change.
RCRA test methods which require the soxhlet extraction of solids such as
sludges and soils need to be tested with the replacement solvent [5].
Data are also needed to determine any effect of a solvent change on the
use of silica gel in oil and grease methods - Method 418.1 for Total
Petroleum Hydrocarbons (TPH), and Method 5520F (Standard Methods for the
Examination of Water and Wastewater, 17 th ed.).
Gas Chromatography techniques which use no CFC should also be
standardized for TPH analysis.
New solvents need to be found which are compatible with IR analysis
methods for oil and grease. Direct oil in water analyzers which require
no solvent extraction need to be evaluated and further developed if
necessary with improved reliability.
-------
332
The SPE technique, which does not require CFC and has reduced waste
solvent advantages, shows promise as a potential alternate technique for
oil and grease but needs further testing and optimization. Consistent
SPE cartridge performance is a necessary requirement.
The safety and human health concerns of alternate solvents need to be
addressed.
This work is needed at an accelerated pace with joint industry and EPA
participation to evaluate and select the most suitable test method for
oil and grease which uses no chlorofluorocarbons.
ACKNOWLEDGMENTS
The authors wish to thank the following people for their contributions in
this work: T. M. Randolph, Shell Offshore Inc.; A.T. Coleman, E. E.
Kettl, T. Fan, and G. R. Bear of Shell Development Company.
REFERENCES
1. 56 Fed. Reg. 30519, July 3, 1991: Guidelines Establishing Test
Procedures for the Analysis of Pollutants; Identification and Listing of
Hazardous Wastes; Test Methods
2. Chemical and Engineering News, July 9, 1990 pp. 6-7.
3. A Report on Additional Work Done by the Environmental Monitoring
Systems Laboratory-Cincinnati To Find the Most Suitable Solvent to
Replace Freon-113 for the Gravimetric Determination of Oil and Grease,
October 22, 1990.
4. Methods for Chemical Analysis of Water and Wastes, EPA-600/4-79-020
March 1979, Revised March 1983.
5. Test Methods for Evaluating Solid Wastes Chemical/Physical Methods,
SW-846, 1986.
6. Sax, N. I. Dangerous Properties of Industrial Materials, 7th ed., Van
Nostrand Reinhold, New York, 1989.
7. Proctor, N. H., et. al. Chemical Hazards of the Workplace, 2nd ed., J.
B. Lippincott, Philadelphia, 1988.
-------
333
LU
QC
D
O
LL
CO
t]z
3o
cow
LU
Z
o
CO
a: ^
^ *-
a. o
2<
Si o a
C00£
CO UJ
z o
£5
^g
LU o -i
coȣ
< UJ =^
Wb§
oc^2
Oc
-* U* m
00 i— CD
UJ
CO
<
LU
ac
o
CQ IO CO 0) Oft
w cd co ^ o
CL v- CM CO ^t
CO IO CM N.
^ CO CD ^t
CO IO K CO
CO IO IO CD ^
• • • • A
CO CD O CD U-
CM CM CM CO >-
O T CD CO CM H-
Q CM CM ^ ^ Z
LU
LU
Z
<
X
h-
LU
s i
i 3
o 5^
£ i
u. o
0> CM _ ^t
• • to •
co a> CD
O
CO
CD
GO
CD'
O O O O O O
O GO CO ^T 04
CO h» O> CO
co 10 CD a>
UJ
z
UJ
CD
UJ UJ
z z
X X
LU LU
X X
52 O
UJ
< o < ?
X UJ X £
UJ oc uj S
X U. X Q
-------
334
CO
COS2Z
UjffO
CD
10
Q. *- •*-
LU
1U
(0
CvJ
LLJ
CC
ID
g
UL
m£o
OC^w
O
01 o> iq co
O> CO
CM 00
CO -^ 0
oi f^ o "
CD IO O CD L
id oi si co Q-
CO CM ^t CD >-
O OJ CO CD O I—
Q ^ ^t W cb Z
UJ
o
CO
UJ
z
<
z
K
IU
i 3
O ^
Ui
m
UJ Ui
z z
< <
X X
UJ UI
X Z
o o o o o o o
CM O CO CD ^ CM
UJ
Z
UJ V uj 5
Z 2 2 O
< O < ^
X LU X S
uj cc in 2
X U. X O
-------
FIGURE 3
335
Produced Water Data
Method: SPE
120
O)
£
o
CO
CO
O
100
80
9 60
O
•o
c
(0
40
20 -
0 -
DCM Freon-113 Hex/MTBE
Solvent
Produced Water Data
Method: LLE
Hexane
120
100
CD
E
o
(0
(0
O
80
60
40
•o
c
CO
5 20
0 -
DCM Freon-113 Hex/MTBE
Solvent
Hexane
-------
FIGURE 4
336
Refinery and Chemical Plant Data
Method: SPE
24
20
16
D)
O
CO
eo 12
-------
FIGURE 5
337
Produced Water Data
120
100
O)
£
0 80
W
O
*- 60
0
•o
c
O
zo
0
24
20
\
O) 16
£
o
S 12
o
k.
8
—
(0
—
O 4
0
- J
~
-
—
-
—
-
i '.' i i I "-
—
Freon Hexane DCM Hex/MTBE Hexane Hex/MTBE
LLE LLE LLE SPE SPE LLE
Refinery and Chemical Plant Data
i , --
*
* —
j
~
___
h- r^-, j
i
-
" ^T ^ ~
• iji' i
Freon Hexane DCM Hex/MTBE Hexane Hex/MTBE
LLE LLE LLE SPE SPE LLE
-------
CD
LU
QC
ID
g
LL
LU
"ffi
uj
UJ
ao
CO
OCU.
CD O
O u.
u.
UJ
UJ
UJ
2
O
cc
O
T,
O
a
O
UJ
cr
u.
o
o
e o
00
•
0»
09
S
1
o»
8*
1
1
1
s
1
•K.-
1
1
§
1
•<»
6
S
1
Oft
K
f9
^
1
ia
cd
JO-
o
o
o
o
en
o
o
on .
V
?<
o <
to
O
CD
O
K
a.
2
UJ
H
v*
UJ
-J
Q.
O
K
O
9
£ £
o
K.
0
X
CL
O
o
CD
X
0.
o
K
m
a.
o
Oft
CD
Q.
O
K
CD
UJ
o.
o
0»
CD
UJ
0.
CO
-------
339
LLJ
QC
ID
g
LL.
DC ^ LU
W rr rv-
. DC QC
Q LJL QJ
-------
340
N H-
UJ
QCUJi-
< b eo
x < <
CO
uu
oc
ID
o
or =
a -^ u-
^
-------
341
MR. TELLIARD: One of the approaches that we are looking at is also a
reduction in solvent use by micro or semi-micro extraction. Paul Epstein is going to talk
about that now and see where it goes. Paul?
MR. EPSTEIN: When we started using the term microextraction, the first
thing that came to mind is an itty-bitty separatory funnel. We don't do that.
I would also like to say before I get into the body of the talk is that the
organization I represent used to be called the National Sanitation Foundation. It was
always referred to as NSF, however, we finally changed our name legally to NSF
International which causes some confusion with that other NSF that works out of
Washington, but we had the initials first.
I will be describing the analysis of several unused compounds compared to
what people have been talking about today. NSF runs a program called the Drinking
Water Additives Program.
In 1988, EPA turned over to a consortium led by NSF this program to
certify and test possible health effects of either chemicals or any indirect or direct
additives put into drinking water. This little cartoon kind of illustrates the process that
we go through.
The judge sitting up at the top represent the regulatory agencies. The tap
iat the end is the public water supply, and in between, you have people putting chemicals
into the water, people manufacturing pipes, and NSF examining everything that goes on.
We consider our client base to be tri-fold. It is the regulatory agencies, the
public, and the manufacturer of the items that we test and certify.
Just some quick definitions. An additive, a term which I will be often, is
any chemical added in any form to drinking water; and a contaminant is any physical,
chemical, biological, or radiological substance that is in water. That is the definition
from the Safe Drinking Water Act.
We deal in two types of additives. Direct additives are additives that are
added deliberately in the water treatment process. Indirect additives are contaminants
that may come out of the drinking water transmission process from the pipes, materials,
-------
342
coatings, paint, anything at all that comes into contact with drinking water.
We have two standards that deal with these additives. Standard 60 covers
the direct additives; Standard 61 covers the indirect additives. These standards, again,
examine only potential health effects. We don't test the performance of products.
Everything is driven by health effects, and it is all driven by a toxicology department.
I am going to give you some quick examples of additives. I don't want to
dwell on this, because there are many products. The direct additives include coagulation,
flocculation chemicals, scale, corrosion, softening, pH control, disinfection and oxidation
(ozone and chlorine), chemicals, and also possible additives in well drilling muds, and
antifreeze.
The indirect additives provide a much broader scope of products that we
test. We test these products to see if they are adding anything to the drinking water.
Again, we are not testing the products for performance. We don't care how well they do
what they say they are supposed to do. We are testing to see if there is anything at all
leaching out as the drinking water passes by these products.
There are many different categories of indirect additives. There is pipes
and related products, protective barrier materials, joining and sealing materials. The
protective barrier and the joining and sealing materials generate many unusual chemicals
that we have to identify, and later on in the talk, you will see the table of chemicals that
we have worked on in the past few months.
There are also process media that go into water purification. These are
generally media that are used at water treatment plants. And there is mechanical
devices. There are feeders, pumps, valves, meters.
There are two new sections to the standard that haven't been accepted yet.
Section 9 which is getting close to acceptance will be looking at faucets, valves, and
anything of that type, any mechanical valve material.
And there is a new section that we are looking at that doesn't affect that
many people. EPA asked us to go ahead and start working on it. This section covers
rainwater catchment materials for people who get their water from rain. There are
sections of the country where that is important. So, we will be testing paint that is put
-------
343
on roofs that the water runs down before it gets caught by cisterns.
Our process includes a complete formulation review by the toxicology
department. They get formulation information from the manufacturers and, in some
cases, go two and three levels deep. If a manufacturer submits a product, they may look
at it and say well, that product is manufactured, so we need that manufacturer's
.submission.
This particular process sometimes can take anywhere up to a year to get
complete formulation information. Once the toxicology department has this formulation,
they select analytes that they want the laboratory to test, and this is where the fun really
starts for the laboratory.
We then get the products, and we try and simulate real world use of the
product. We call this simulation an exposure. We treat the product with a very
aggressive water. It is usually either a pH 5, 8, or 10 buffered water that hss similar
characteristics to standard drinking waters, but it is as aggressive as possible toward the
analytes of interest.
When we are looking for metals, we use the high and low pH water; when
we are looking for organics, we use the pH 8 water.
The laboratory then reports back its results to the toxicology group who
then takes these results and normalizes them based on expected usage and determines
what the potential human exposure is at the tap from that chemical.
Potential analytes include metals, screens, and organics. The organics
include anything that you can imagine that goes into any plastic or metal product;
anything that comes into contact with drinking water.
We are testing a whole series of coatings now that are used to coat the
inside of water storage tanks. We test barrier materials. A hot issue right now is the
concrete that goes into concrete pipe, since there are some concerns that some of the
concrete is manufactured by burning hazardous waste in the concrete kilns.
The typical expected MDLs that the toxicology group asks us to provide
are 1 to 20 parts per billion. These are adjustable, because we have some control over
the amount of sample that we can expose. If we can't get a low enough detection limit,
-------
344
we try and raise the surface area to volume of the sample exposure.
We get 1 to 2 requests per week from toxicology to analyze for new
chemicals that we have never done before. They ire usually not amenable to the
standard methods.
There are two groups in our lab. There is the group that performs all the
regular methods and runs similar to a commercial laboratory.
Then we have the small group that does method development and tries to
do method development on a production line basis.
The resources of this particular group are limited to one full-time senior
chemist. She has limited resource available to her at all times. We try and develop
methods that are amenable to GC/FID and GC/EC.
The bright side of what we do is that we are not in the real world, and
greater than 95 percent of our samples are potable water or something that looks like
potable water. We don't test sludges, we don't test soils, we don't test oil and grease, we
don't test any of the things that people have been complaining about all day. So, there
are some advantages to doing our type of work.
The disadvantages are what we call the analytes from hell, the water
soluble glycol ethers and amines that we are trying to determine down at the low part
per billion level. We struggled with these for about a year and came up with the
methods that I am going to talk to you about which coincidentally use very small solvent
volumes and fit in with the topics that we are talking about today.
We have three different procedures. The glassware involved is usually a 40
ml volatiles vial. We don't have to clean sep funnels. We don't have to worry about
cross contamination, because we throw the vials away when we are done.
This is a neutral extraction method. It is really something we used to call a
shake and bake when I was in commercial labs where you just add a small amount of
solvent, shake it very hard for a couple of minutes. In some cases, for example, the
amines, we add sodium hydroxide to raise the pH. We add sodium chloride to salt it
out. We extract with 1 ml of methylene chloride. After 2 minutes, we take the
methylene chloride out with a Pasteur pipette and analyze it directly by GC/FID.
-------
345
It is a very simple method. You can test anywhere from 5 to 10 samples
an hour. It gives us the detection limits that we need, and in order to generate a high
confidence that the numbers are correct, we prepare what we call "standards prepared as
samples."
We will do our linearity set by spiking the standards into the same type of
water that we are extracting, whatever the buffer is, extracting the standards and then
doing external standard GC quantitation.
We also occasionally analyze 100 ml sample volumes in which case we get
a slightly greater concentration effect, but we found that, for the purposes of the
detection limits we need to reach, it doesn't make that much of a difference.
The primary aliphatic and aromatic amines were compounds that were
difficult to analyze. Again, the toxicology group wanted 1 to 20 parts per billion, and the
best we were able to do was about 1 to 5 parts per million by direct water injection
GC/FID.
One of the nice things about having somebody fresh out of school working
for you is that they come in with a totally different perspective on the world. Never
having done environmental chemistry, Kris Kurtz who is one of the co-authors, just
started to do real chemistry and looked for some possible derivatives of the aliphatic
amines that would solve the problem we were having, they just would not extract out of
the water.
We had tried some solid phase extraction, and what we found there was
that once it was on the column, you just couldn't elute it off. It was too polar and too
strong a hydrogen bonding material.
Kris went ahead and did some bookwork and came up with a benzaldehyde
derivative which is also a very simple technique. You add 20 ml of benzaldehyde and
allow it to react, and it forms an imine. I will show you the chemistry on the next slide.
You salt it out, extract it with methylene chloride, allow it to settle, and
sample the methylene chloride layer.
The only problem with this method is it is not very hardy. The derivatives
don't last a whole long time, so you really need to do your extraction and inject the
-------
346
sample as soon as possible.
The chemistry is fairly simple. If you take your primary amine, react it
with benzaldehyde, you get an imine which is a lot less water soluble and, therefore,
extractable.
Here are some chromatograms. The benzaldehyde derivatives are in the
upper right-hand corner. The one on the left is a n-butylbenzene sulfonamide which is a
neutral, underivatized example.
These are detection limit chromatograms. We consider our detection limit
the low standard. We don't do statistical detection limits, because we need to guarantee
to the toxicology group that if we need to see something at a level that they have said is
toxicologically significant, we could have seen it if it were present in the sample.
The other method that we use for primary aromatic amines is based on the
OSHA 57 method for methylenedianiline. It is a heptafluoro-butyric acid anhydride
derivative (HFAA). Again, it is a very simple method. It is also performed in a 40 ml
VGA vial.
We add 1 ml of half-normal sodium hydroxide, extract with 2 ml of
toluene, allow the toluene layer to separate, pull it out, add 25 microliters of the HFAA
reagent directly to the sample, add 1 ml of pH 7 buffer, shake, and then analyze the
extract by GC/ECD.
The big difference between this and the benzaldehyde method is that the
primary aliphatic amines are derivatized before they are extracted. These compounds
which are a little less soluble in the water can be extracted first and then derivatized.
This is a series of three injections of 5-chloro-ortho-toluidine using the
HFAA method. The first one is at 3.5 parts per billion. This middle one is at 1.8 parts
per billion, and the last one which is actually below our reported detection limit is .7
parts per billion. We report this one at a 1 part per billion detection limit.
I have got several slides now showing you some of the compounds that we
have analyzed in the last few months. They are not the normal compounds that you see
in environmental chemistry, but they are significant, because they are in products that
might be in the water transmission system.
-------
347
There are several glycol ethers which have a typical detection limit of 20
parts per billion. The one that is a fairly low boiler has a detection limit at 50. The
butylphenylglycidyl ether has a detection limit of 5 parts per billion.
Hydroxymethylpentanone is 50. That is a compound that can be analyzed
by other methods, but we chose to analyze it by this method.
The n-butylbenzenesulfonamide had a detection limit of 10 parts per
billion.
These are some of the amines that are done by straight extraction. On the
next slide, I will show you some of the benzaldehyde derivatives.
We have analyzed some pure hydrocarbons, the vinylcyclohexane, and
vinylcyclohexane dioxide. Again, you can see the third column over has the detection
limits, and they are anywhere from 1 to 50 parts per billion. Most of them are under 20.
These are the derivatized samples. We have done ethylenediamine, 1,6-
hexanediamine, and isopheronediamine...by the benzaldehyde method, and we get
detection limits of either 5 or 50.
The last five compounds were all done using the heptafluoro-butyric acid
anhydride derivative. All five of those have detection 1 part per billion.
Again, this derivatization method is based on OSHA 57 for 4,4/methylene
dianiline which is basically the same method. It is collected, I believe, either on a tube
or a filter, desorbed into water, and then derivatized and extracted.
That is the end of my talk. These methods are just one more tool in our
toolbox of different techniques. It has turned out to be very valuable, because, other than
the fact that it is quick, it does use a very small amount of solvents, it doesn't use
expensive glassware, one person can prepare the samples and analyze them fairly quickly,
and as you can see from the list, we developed methods for about 20 compounds in the
last 6 months using this technique.
Anybody have any questions?
MR.TELLIARD: Questions?
(No response.)
MR. TELLIARD: Thank you, Paul.
-------
348
J g
3 SP
^ cu
^ eu
d *
3 o
P5 fe
W *
Q
^ bJ
& e
2|
^ ^
« ^
o g
u X
u w
^S
* S
t-4 ^
S o
n *-*
w ^
o 5
05 ^
U H3
i^
O ^
HH 0Q
^3«
5 w
a H
S ^
^^
Forozesh
^
cd .0
•^ ^
N Q
B 5
5 |
CO ^
w ft
a
.3
w
CO
o m
X> x-^
•ss
o
O
B
-------
349
C/5
£
O
HH
H
E4
Q
00
c
c3
C
•8
cd
o
i-H
s
o .S
s-a
2 S
cd ^3
co
S §
• I-H ,;5
cd ^
c -o
§ 2
U &
roduce
g
O co
G
II
(D 3
& r°
CO "£3
•tn >
S-i
0
3
+-»
G
O
O
co n
i r
G O
•tn H
CO
CO
O
on
c £
O cd
U
CO
CO
o "3
& s
PH CO
o ^
CO
73
•i-H
VH
1 2
C W2
• i-H
>
J__i
^^
c
• i-H
S
£2
G
O
U
• r-H
4— >
• i-H
^§
^^^
^
4-J
O
o
•i-H
Q
•»-H
T3
<£
^^
4— >
o
• i-H
^O
G
I-H
-------
350
-------
351
•o
I.
C
CQ
C/5
O
C3
t» ""3
B fe
$ £
£ s
***H HH
fa ^
c» Z
rt w
O »rt
^ c
O
C«
^ 13
O>
DJD
C
v o .S2
o fe "°
a a
M^ r"l
^^^^^ |MBB^
c ••*
CQ ^
-------
352
C/5
H
U
I*
HH
O
fc
CO
(D
CO
e
• *-H
1—H
cx
o
U
CO CO
Jj VH
E 00
CO
(D
co
G
(D
O
00
<» 8?
G 8
S G
co rn +•*
60 G ^
S SP'S
'3 OH -§
• ?H ^5
CO <
0
U
^
"o
D
4->
§
G
O
U
CO
^—>
G
(D
I 2
ctf o>
(D i—H
TJ ^
&£
CD (D
4^ *->
S 8
VH VH
O O
G G
O O
U U
CO
VH
CD
G
CO
CD
S
00
co O
VH C
^
• 2 J3 3
Q H OH
CO
(D
X
co
VH
00
co
.Is
CD
ts
CO
(D
.> W)
co G
(D
€
CO
tf CO
3 S3
a "^
•c S
00
co
VH
CO
o|
VH OH
O *l
co C
o5
co cd
o M
CO
PQ
03 ^
U O
J
Cu
(D
CO
J3 t3
^4
CO
PH O
.^H VH
OH
s «
>^ i—H
T3 cd
^ "C
D D
p ^
VH
OH
c
ITS
W)
G S
i-H »»-H
1
00
-------
353
H
O
u
H
U
tivated Alumina Aeration
o
<
03
^
&
O
r/\
\jj
T3
<
Chelating Polymers
d
•
Packing Material
r/^
bonaceous Resin:
53
U
cd
^H
4-2
i— H
• i-H
tin
"g
b
Diatomaceous EJ
•*^J
*c
•i-H
4-J
cd
T3 T3
• i-H rt
>< g
O co
.2
1
O
•i-H
0>
c
>1
CO
cd
•i-H
D
o
2
cu
-------
354
O
U
H
U
CO
fe ^
O
•*-*
CO
H-H >
g CO
CH o
«>•§
*?3 -»-H
CO VH
» 1 ^ - —
0) co
S» f-H
Q
3
O
co
S C
o °
{J »f-H
d> H
Q ra
^ PH
C ^
0> C
s °
H -i—i
O
ta
•c
f-H
U
CO
CO
o
o
VH
-------
355
X
r-H
ctf
O
U
G
O
O
o
o
G
o
o5
O
U
r G
o ;
r—• r—\
o Q
ffi
G
O
G "C
(U oj
c 3
O x
O
G
O C
N
Q
o>
r-H
O
•rH
S S
"o
(D
O
Q
-------
356
a
fH
4-J
co
CO
+-»
o
^H
13
CO
-------
357
C/3
W
H
H
O
+->
d)
O)
O
0
c
(D
w
u
8
tt>
a
CD
%
o
00
-------
358
OH
O<
O
CN
i
2
03
Q
O
cx,
X
w
i-H
03
O
o
o
>i-H
x
o
e
o
(U
PH
05
>
3 (N
o3 in
-------
359
ed
CO
CO
-------
360
O
i
O
U
O
GO
J-H
% g
S O
Q
CX
-------
361
M a
o
S
^H
4-J
G
(U
O
a
"d o
§u
i-
O. f-H
A
as S
andards
mL
35
(U
a I
^"H
,0
G
O
^>
O
S s
X CO
(U
§ S
s
o
CO
CO
M
O
en
in
m
X
1~H r2
CO £3
CO
•d
13
G
i; o co
G pu
-------
362
O
U
S— •
C/5
U
O
P*
Cu
0)
s
S3
I
(U
i-H
OH
00
S
O
o
(U
X!
s
C/5
^ £
G ^
i^~\ o3
VH
03
PH ffi
D S
OH ^
CO
DH
o
o
S
CO
r ^1
O
CX
G
O
G
O
CO
CO
•i-H
-d
O
CO
G
03
U
03
03
§ 22
PH
CO
0)
£Hj
G
• i-H
a
(N
VH
<4H
r^\
•
CO
3±
4— »
3
G
• i-H
e
m
o
4-»
O
VH
O
CO
(N
(N Q
K^ hH
I 1
U
r
ffi
U
CO
O
03
(N
•S3 CJ
I &
X
2 ID
a, N
&
VH
03
U
ffi
U
r-H
U
ffi
U
(N
-------
363
H
Z
O
u
P
Q
U
O
03
CD
>
00
(D
0
T-H
^
O
i— i
VH
CN
05
0)
X
o
ex,
£
CN
CS
f 1
u
ffi
U
<+H
O
(U
(U
05
FH
0)
03
GC/FID
N
c
03
FH
(U
o
-------
364
t
ffi
o= u
o
ffi
-------
365
H
Z
O
p
c
u
o
at
CO
&H —^ »rH
X!
2
£
ex
(D
CO
o
X
w
cd
- H
(U
a
O
*-*
£
O
cd
• rH
•J
s
oo
a
o
cd
VH
X
rH
(D
for
igorous
CO
O
cd
X
(U
x
a
(D
bO
cd
<
ctf ^t^
O »
a ffi
(D
CO
H-H T3
o a
o
O
CD
^ ^
co CN O
a -o ^
S T>
uL
seconds
CO
T3
ffi
O
DH
ffi
cs
Q
U
O
(U
N
13
g
CD
O
hJ
S
(U
-H
DH
00
-------
366
CO
Q
O
ffl
H
O
HH
H
^
^ /^"
0 ^
if
£
^u
• TH 0
T-H
• TH
0
PQ
G
O
•TH
ts
VH
O
G
O
U
Compound
in
T-H
o
e
CN
T-H
0"
ON
T-H
CO
T-H
Js
1*^
glycol monome
fl
T^^
T-H
^*>
PH
O TH
(H 0>
f^( fH
• FH .+^
^ °
r-H
O
CN
T-H
CO
CN
r-H
in
CO
VH
-G
4— >
(O
1
1
PH
O
in
vo
VO
T-H
T-H
o
o
r-H
G
O
r^
G
i
CN
i
T-H
a
^r
X
o
•§,
f{
1
ON
CO
r-H
O
T-H
^
in
CO
enesulfonamide
a
d>
T-H
3
1-Q
. *
z
-------
367
Q
U
0
98 '8
£8'8
88'8
-------
368
Q
O
ffi
H
H
U
H
X!
O
I*
U
i
^
Q
c
o
• i-H
43
a
o
a
o
U
Compound
0
in
,-H
in
en
mine (Benzaldehyde)
ctf
• i-H
s
^-
^
T-H
(N
in
en
duidine (HFAA)
O
I
O
6
VH
O
f— H
O
in
^ A
Q
§
O
cd
=«=
T3
O
a
CO
O
ctf
*Q
C
O
• 1-s
•4— >
cd
N
-------
369
CO
Q
O
X
H
H
U
H
X
o>
^ "^
o ^
G
CD
O
a
0
U
T3
c
3
&
8
o
U
oo
T-H
T-H
o
1
in
en
crt
•i-H
-Methylene b
lohexylamine
^ >,
^ &
1
r-H
O
en
r-H
O
o
T-H
aphthylamine
G
i
PQ
Ox
eN
r-H
O
T-H
en
oo
r-H
T-H
in
en
e
^
i
r.
1
0
T-H
O
r— (
in
en
o>
T3
>»
^
(D
T3
13
G
O
^— >
a
u
1
o
in
in
en
>
G
(D
-dichlorodiphi
r>
^f
•s
^t-
1
in
o
en
^xj
o
o
T-H
zothiazole
G
4
en
0
T-H
0
eN
eN
eN
in
en
"rt
• S
X
.2
G
X
(D
0
T— H
o
o
T— H
G
4
-------
370
2.576
c
N-Butylbenzene
Sulfonamide
START
STOP
2.571
1.344
ZEt
Isophorone
Diamine
-------
371
MR. TELLIARD: Our last speaker for the day is from Columbia
Analytical Services, Inc. out in Oregon. That is true. They all look the same, trees,
water, grass, no delis, you know. And Mr. Anderson is going to discuss a system they
have installed for solvent recovery in the laboratory.
-------
372
[Blank Page]
-------
373
SOLVENT RECOVERY
presented by Jon Anderson
Columbia Analytical Services, Inc.
May 6, 1992
The 15th Annual EPA Conference
on
Analysis of Pollutants in the Environment
-------
374
The goal of the solvent recovery program at Columbia Analytical Services, Inc. (CAS)
is to recycle the solvents used in our extraction procedures and those used for rinsing
glassware. At CAS we are averaging 2,000 - 2,500 semivolatile extractions for our
semi-volatile laboratory, so the amount of solvent used and its cost was a factor in
our interest in recycling. The extraction procedures commonly employed at CAS are
3510, 3520, 3540, 3550 with Soxhlets, separatory funnels, continuous liquid-liquid
extractors and sonic horn. These extraction procedures and methods call for the use
of the solvents: methylene chloride, which will be abbreviated DCM; acetone; hexane;
and freon. It is these solvents that CAS is reclaiming and reusing.
We start out by segregating the solvents. The methylene chloride used in rinsing
glassware and that collected from the S-evaps must be segregated from the
methylene chloride waste. We also must segregate the 1:1 solvents used in our soil
extractions, methylene chloride/acetoneandhexane/acetone, which are collected from
the S-evaps during the concentration procedure. Another solvent that we reclaim and
reuse is the freon from Method 418.1, the oil and grease methods, and that used in
rinsing glassware. We collect, distill, and reuse the acetone used to dry glassware.
We do not reuse any of our solvents in sample extracts.
We make use of six Organomation S-evaps for solvent collecting during the
concentration procedure from the KD apparatus. We also have four stills, one 22-liter
still for DCM, one 22-liter still for the one-to-one solvents, one 12-liter still for freon,
and a small 3-liter still for acetone.
The reclaimed and then distilled 1 -to-1 solvents passed preliminary mass spectroscopy
tests, so we are in the process of redistilling almost fifty cases of reclaimed 1-to-1
solvent.
For the collection of the solvents to be recycled, we have clearly marked, color-coded,
and labeled 5 gallon and 2.5 gallon cans. We also use 2 liter nalgene beakers in our
hoods to collect the methylene chloride that is used to rinse glassware. The
color\coding used is: red for reclaimed, but not yet distilled solvents; yellow for
distilled, but not yet tested for purity; and green for solvents ready to be used in the
laboratory. Yellow and red are used on solvent containers that did not pass our
testing procedures.
We have found that the segregation and clear labeling is the key to making this
system work.
For DCM reclamation, we discard the first 500 ml that comes over the still. The
distillate is collected in clean, rinsed 4-liter bottles, which are labeled and stored for
testing. These bottles are very important. We have found that the primary cause of
contamination in the distillate is the collection containers, not the distilling process.
We then collect 100 ml from four of these bottles, place it in a clean KD and
concentrate it to 1 ml for GC/MS testing. Our acceptance criteria for the reclaimed
-------
375
solvent is that there is no detection of any of the target compounds tested for in
Method 8270, and that the total area for all non-target peaks is less than 10 percent
of the area of the internal standard. If no interferences are present, the DCM is then
labeled for use in the lab.
Hydrocarbon testing and the GPC cleanup procedure are a two tests we do not used
reclaimed DCM for:
We test the 1-to-1 solvents by Method 8270 using the same acceptance criteria as
we use for the DCM. We then dilute 10 microliters in one milliliter of iso-octane and
run it on the FID to determine the percentage of acetone versus DCM and hexane in
order to adjust it back to a 1-to-1 mixture. It can then be used for soil extractions.
To distill and test the Freon, we first filter off any silica gel, then collect and discard
the first 500 ml that comes from the still, collect the rest in 4-liter bottles and then
run an aliquot from each bottle on the IR. The pass criteria is less than 2 ppm of
hydrocarbons by Method 418.1 procedure.
In our glasswashing areas, we have installed a small still for the glasswashers to distill
the acetone they use to dry glassware. After distillation, they reuse the acetone.
For all of these reclaimed solvents we keep daily and monthly records of the reclaimed
solvent batches and the test results.
The system cost:
22-liter still
12-liter still
Organomation S-evaps, KDs,
condensers, various
glassware
5-gallon stainless steel
cans
$3,500
$3,000
$5,000/each
$ 200/each
During an average month we reclaim and distill approximately 30 cases of four, 4-liter
bottles of DCM, 15 cases of freon, 30 cases for 1:1 solvent at a total savings of
$8,300. In one year we have saved approximately $100,000. However, our
reclamation and distillation is on the increase, so we will be seeing even greater
savings.
-------
376
COLUMBIA ANALYTICAL SERVICES, INC.
SOLVENT RECOVERY PROCEDURE
-------
377
GOAL
To reclaim and reuse solvents used in
extractions and glassware rinsing for
Methods SW-846; EPA Methods 3510,
3520, 3540, 3550, 413.1, and 418.1.
Methylene Chloride; Acetone; Hexane; and
Freon.
-------
378
SOLVENT SEGREGATION
A. Methylene Chloride
1. Used in rinsing glassware.
2. Collected from S-Evap.
B. 1:1 Methylene Chloride: Acetone and
Hexane: Acetone
1. Collected from S-Evap.
C. Freon
1. Used in 418.1 hydrocarbon method.
2. Collected from S-Evap.
3. Used in glassware rinsing.
D. Rinse Acetone
1. Used for rinsing glassware.
Solvents from each are collected in 5 or
2Y2 gallon stainless steel cans. Clearly
marked.
-------
379
A. 2 L Nalgene beakers to collect rinse
solvent (i.e., DCM, Acetone, Freon).
B. Organomation S-Evap for collecting
solvent off K-D concentrators.
C. Stills:
1. DCM: Kontes 22L distillation
apparatus.
2. 1:1 Solvents: Kontes 22L
distillation apparatus.
3. Freon: Kontes 12L distillation
apparatus.
4. Acetone: Simple 3L heating
mantle/column use for glassware
rinsing.
-------
380
LABELLING
All canisters, bottles,
and boxes will be labeled accordingly
RECLAIMED DCM
(includes RINSE, to be
distilled, canisters)
DISTILLED DCM
Waiting to be Tested
OK FOR USE
Passed GC/MS Testing
RE-DISTILL
Did not Pass GC/MS Test
-------
381
DISTILLATION AND TESTING
DCM
1 . Collect and discard the first 500 ml_
that is collected after filling still.
2. Collect in clean, rinsed 4 L bottles, label
and store until after testing.
3. Testing - Collect 100 ml from 4-4 L
bottles and place in clean, rinsed K-D
and concentrate to 1 .0 mL for GC/MS
testing.
Redistilled DCM is not used for GPC.
Redistilled DCM is not used for volatile
hydrocarbon testing on water samples.
-------
382
DISTILLATION AND TESTING
1:1 Solvents
DCM/Acetone/Hexane use in Soil
Extractions
1. Collect and discard the first 500 ml
that is collected after filling still.
2. Collect in clean, rinsed 4 L bottles, label
and store until after testing.
3. Testing:
A. Collect 100 ml from 4-4 L bottles.
Place in clean, rinsed K-D and
concentrate to 1.0 ml for GC/MS
testing.
B. Dilute 10 jjl into 1.0 ml of
isooctane run on GC/FID to
determine percent acetone vs DCM
and hexane, in order to adjust to
1:1 mixture.
-------
383
DISTILLATION AND TESTING
FREON
1. Filter off any silica gel present from
418.1.
2. Collect and discard the first 500 ml
that is collected after filling still.
3. Collect in clean, rinsed 4 L bottles, label
and store until after testing.
4. Run an aliquot from each bottle, if
solution concentration of hydrocarbons
is <2.0 //g/mL by IR techniques, then
the bottle is posted and ready for use.
-------
384
DISTILLATION AND TESTING
ACETONE
Collect redistilled and use for rinsing
glassware after washing.
-------
385
IMPORTANT NOTES
1. Good, clear, labeling system.
2. Solvents to be redistilled, tested, and
used must be kept separate.
3. Daily and monthly records of batch
testing and use must be kept.
-------
386
SYSTEM COST
Kontes 22L still with 2 Columns $3,500
Kontes 12L still with 2 Columns $3,000
Organomation S-Evap with Glassware $5,000
5-Gallon Stainless Steel Cans $200
One Technician to run system(s)
30 to 40 hours per week.
-------
387
SYSTEM SAVINGS
DCM Average 30 Cases/Month $2,000
Freon Average 15 Cases/Month $4,200
Acetone Average 50 Gallons/Month $300
1:1 Solvent Average 30 Cases/Month $1,800
Total Average Savings/Month $8,300
Average Yearly Savings - Approximately $10D.nnn
-------
388
[Blank Page]
-------
389
PROCEEDINGS
Mav 7.1992
MR. TELLIARD: Good morning. We are going to start out this morning's
session with a number of papers discussing immunoassay techniques for analysis. This
particular procedure is quite common in the medical field, but it has just in the last few
years become more and more prominent in the environmental area.
Office of Water have done some work with immunoassay and we think it
shows promise. It could have application for field monitoring and for such things as non-
point source discharge monitoring programs.
Our first speaker is Kevin Carter, and Kevin is going to present data on
some of the basic field methods.
MR. CARTER: I am going to talk today about the performance of some
immunoassays for polychlorinated biphenyls and pentachlorophenol. I would like to first
thank Bill and the rest of |he conference organizing crew for recognizing the potential
value that immunoassays have in the environmental field and including a session on
immunoassays. Immunoassay methods can address a wide variety of contaminants
ranging from industrial contaminants, like I am going to talk about, to pesticides, which
are the subject of the next three or four talks.
The orientation of my talk will be around field methods. We feel that the
mitral application of immunoassays in the environmental field is probably best directed
toward field methods, because adequate laboratory methods already exist for the kinds of
compounds that we are talking about.
That is not to say that immunoassays can't be used in the laboratory,
because they certainly can be, but that the field is a better place to start.
To begin, I want to talk a little bit about the typical hazardous waste site
cleanup progression, going through four phases. Field analysis is valuable in a few of the
phases. It is really most valuable where many samples are collected and have to be
-------
390
analyzed.
Field methods can be used most valuably in the field to help you make
field decisions. Field analysis and laboratory analysis are complimentary activities. If
you use each where it provides the necessary level of information, the most value for the
dollar, you are optimizing your analytical efficiency on environmental projects.
One of the places where field analysis has a lot of value is in site
assessment or contaminant mapping. Whether you are working with soil to locate "hot
spots" or with water to1 follow a plume of contamination in groundwater, field testing can
help you cost-effectively increase your understanding of contaminant distribution.
Through the use of directed sampling more useful site assessment data can be obtained.
One of the greatest challenges in characterizing a contaminated site is taking samples
that are really representative of the site that you are trying to address and using field
analysis, you can do a better job of doing that.
I have heard it said that a greater number of measurements of lower
resolution, meaning field analysis methods, results in higher confidence with regard to
decisions that have to be made in the field. That has certainly proven to be true in the
projects that we have participated in.
The results of field analysis can be used in the area of site assessment and
contaminant mapping for several things. You can use them on-site for attempting to
define the actual site boundaries, not the property boundaries, but the contamination
boundaries. You can use them to define hot spots, clean areas, or both, depending on
what the goals of the project are. And you can certainly use them to optimize further
sample collection for laboratory confirmation. Obviously, it is not particularly valuable
to choose many samples from areas that contain either very low levels of contamination
or very high levels of contamination. Because you are trying to map the gradient across
the site. So, choosing the correct samples for laboratory analysis can be aided by field
analysis.
In addition, estimating the volume of contaminated soil can be done very
conveniently in the field using field analysis. You can take many samples, both laterally
and vertically, to get an idea of how much soil would have to be treated or how much
-------
391
water would have to be treated in a remediation project. This helps you establish a
remediation approach, as well as estimate a cost and time frame.
Another area where many samples are collected is in the pursuit of
remediation of a site. In the case of remediation, the same principles apply, but, in
many cases, when you are remediating a site, the thing that is of most value is knowing
when an area is clean. You want to know that what you have left at the site is actually
clean.
Now, there are some specific areas, where field analysis can play a role in
remediation: testing of influent soil to ensure that you are only remediating dirty soil;
Monitoring remediation efficiency; Verifying the completeness of
remediation; And guiding the collection of samples that will be needde for the closure
permitting process.
Throughout this process, QA is extremely important, and part of QA, as I
think I have already referred to indirectly is laboratory confirmation of a portion of the
samples. People who are using these techniques in the field are doing a certain amount
of laboratory confirmation to back up the results that they get from field analysis.
At this point I want to go further and identify the characteristics that would
be advantageous for a field analytical technique. It is highly desireable for the test to be
specific to the analyte that is doing the clean-up.
The test needs to be sensitive, because the kinds of limits that have to be
measured against these days are generally not getting higher, they are getting lower.
It needs to be easy to use, because in the field, conditions aren't optimum
for analytical techniques. It needs to be rapid, because, of course, the desired response
is quick decision making.
Finally, it is important that field analysis techniques, just as laboratory
analysis techniques, not be significantly affected by the matrix that is being analyzed.
There needs to be a lack of interferences so that when you think you are
measuring, for example, PCBs, you are not measuring something else.
Well, all of these criteria are met by immunoassay-based analytical
methods. I think that you will see as we go through the field performance data and see
-------
392
some of the data presented in subsequent talks in this session that these tests very useful
for field analytical purposes.
What is an immunoassay, for those of you who don't know? An
immunoassay is simply an analytical method that uses a biological molecule, an antibody,
which is simply a protein, to detect and quantify compounds in a test sample.
There is something special about the antibody in that it can be tailored to
specifically bind to the compound of interest that you are analyzing for. As well as
binding in a specific fashion, the antibody does so at very low analyte concentrations, and
that is what confers the specificity and sensitivity on immunoassays.
Now, I would like to review a generalized model of how an immunoassay
works. There are many formats for immunoassays. Some of them involve various other
components, but the most common type that is used and the kind that you will most
often find in environmental tests is the so-called competitive enzyme immunoassay. You
will see what competitive means in a second. The word ELJSA is an acronym for
enzyme linked immunosorbent assay. Two components are necessary for the test. You
see on the bottom of the tubes, tubes 1, 2, and 3; a little Y-shaped figure. That is a
representation of an antibody. An antibody has two "arms" that contain two individual
analyte binding sites. Those correspond to the two arms of the Y that are sticking out
into the tube. In the way that we run immunoassays the antibody is actually immobilized
to a plastic test tube.
You also need what is referred to as an enzyme conjugate. The enzyme
catalyzes a color-forming reaction that serves as the reporting or reading result in the
test. Each enzyme molecule has several analyte molecules chemically bound to the
outside surface of it.
Now, a practical immunoassay test for the environmental field is realized in
the following fashion. The tests I am going to be talking about are semi-quantitative
tests in which at the same time as you test samples that contain the analyte of interest,
you run a standard that contains the same analyte of interest. You compare the color
formed at the end of the test period with the standard to that developed with the sample
to determine whether the sample contains less than or greater than the amount of
-------
393
analyte that is present in the standard.
So, you get a presence/absence indication relative to a preset quantitative
level.
Referring abck to the slide, you test a standard which is represented in
tube 1 on the left where you add enzyme conjugate and standard. In sample tubes, you
add processed sample, as well as enzyme conjugate.
Once you have added these and mixed them, the analyte that is bound to
the outside of enzyme competes with the analyte that originated from the sample for
antibody binding sites.
It is easy to see that in samples that contain a greater amount of analyte
that originated in the sample, such as tube 3, that you get analyte from the sample bound
at a higher level to antibody immobilized on the tube.
In the case of tube 2, there was no analyte in the sample. Therefore, the
only thing that can bind to antibody is enzyme conjugate.
In the standard, which has an intermediate amount of analyte present, you
get both enzyme conjugate and analyte bound to the antibody.
So, at the point at which the equilibrium has been nearly established after
an incubation period and incubations typically range from 2 or 3 minutes to upwards of
30 minutes at ambient temperature...you stop the incubation by rinsing out the antibody-
coated tubes.
In doing so, you wash out all unbound materials. Only analyte and enzyme
conjugated analyte that are bound to antibody stay in the tube. Following the wash the
substrates for the color forming reaction are added. Color formation is catalyzed by the
enzyme part of the enzyme conjugate.
In our tests, we use an enzyme called horseradish peroxidase which, in the
presence of tetramethylbenzidine and hydrogen peroxide, catalyzes the formation of a
blue color, the starting substrates being colorless.
A short incubation proceeds, again, usually on the order of either seconds
or minutes up to 2 or 3 minutes. During this time, any enzyme conjugate that is
immobilized to the antibody catalyzes the formation of color, using the substrates.
-------
394
In the standard tube where there were a few molecules of enzyme
conjugate bound, a medium color intensity develops.
In a negative sample...and it doesn't show up very well in this slide where
only enzyme conjugate is bound, a deep color results.
In a positive sample that contains no enzyme conjugate bound because
there was so much analyte, you observe virtually no color development.
At the end of the incubation period, you stop the reaction through the
addition of acid which kills the enzyme activity, and compare samples and standards in
some kind of a colorimetric reader. A differential photometer is commonly used. There
i
are battery-powered instruments that can be used in the field very conveniently, set at a
fixed wavelength.
In cases where assays are in a multiple well plate format there are
automatic readers, which will scan several wells and give you a readout for each well, but
the principle is the same. It is reading the amount of color in the sample.
If you were to make a plot of the color intensity that was present at the
end of reaction versus the concentration, you would see over a certain range a log linear
response. This kind of a test is very capable of being used as a quantitative test.
But to make the tests field-usable, we employ a semi-quantitative approach,
to do this we set the standard in the middle of the linear range, as you see portrayed in
this slide, and then determine whether sample that has been processed through the assay
yields tubes with greater or lesser color. As you can see, this is an inverse relationship,
so less color implies more analyte; more color implies less analyte.
In a real world implementation, there are some things that I have left out
of the discussion so far. The first thing that I have left out is that, obviously, you need to
do some sample processing, because you can't dump soil into a coated tube and expect
an assay to work.
So, coupled to the immunochemistry that you have seen, there is also a
simple sample processing step that involves taking a weighed amount of soil, extracting it
with a solvent which is compatible with water, and then doing serial dilutions on the
sample extract to adjust the sensitivity to the appropiate semi-quantitative level.
-------
395
By making further dilutions on the sample extract one can actually use
semi-quantitative tests to range sample concentrations.
Immunoassay-based tests can be used very effectively as semi-quantitative
tools to zero in on analyte concentrations in samples in the field.
Now, I am going to switch gears and talk about some data that has been
generated in field trials for PCB and pentachlorophenol assays that use this technique.
The objective for field trials of these products was to document the
performance on a real world situation with real world samples using people, geologists,
engineers, and chemists that collect samples in the field.
There is also some data presented from spiked samples that were run in a
laboratory.
For the PCB soil test, two sets of data will be presented. First, a trial was
carried out at the Department of Energy's facility at Oak Ridge in their organic
chemistry analytical division.
And, second, the Gas Research Institute sponsored a trial of the PCB test
that was conducted by Roy F. Weston Analytics Division to assess its applicability to gas
pipeline sites where PCBs had gotten into the soil as a result of being used in
compressors.
In the case of the DOE trial, they had several objectives in looking at field
analytical methods. They were interested in documenting the degree of correlation that
could be obtained with the standard EPA GC Method 8080; looking at potential matrix
effects relative to various DOE sites; determining repeatability from test to test with the
same sample; and determining the degree of user variability that might have an impact
on the accuracy and precision of the results.
DOE took a soil from one of their facilities, the Rocky Flats facility, and
spiked it with several levels of Aroclors, 1254 and 1260, and had two different operators
run the immunoassay and compared the results with the spiked values.
What they found was 85 percent correlation between the results of the
semi-quantitative immunoassay and the spiked values.
In the cases where correlation wasn't observed, they observed that the
-------
396
immunoassay over-estimated the amount of PCBs in the sample. In fact, that happened
only 15 percent of the time. There were no false negative or underestimated results.
That has significance, because in the field when you are making decisions
based on a field analytical method, you don't want to say that something is clean that
isn't clean.
The user variability was shown to be very minor with this test in that
operators 1 and 2 both got similar false positive percentages when they ran the test.
There was also little variability with duplicate analyses. That variability
was all correlated with the false positive results.
They also analyzed contaminated samples that came from the Oak Ridge
area, samples of soil and sediment that had PCB contamination from historical
operations there. With these samples, almost 90 percent of the immunoassay results
agreed with the GC-ECD analyzed results.
Again, in this case, there were no false negative results. In this case, the
user variability was virtually nil. Duplicates also agreed well.
For the Gas Research Institute trial, 30 samples were collected from 6
different pipeline sites throughout the country. They represented several soil types, to
try to identify potential matrix effects.
All of the soils were analyzed for PCBs using EPA Method 8080, as well as
with the immunoassay-based test. GRI and Weston found that for nearly 90 percent of
the samples, the two methods agreed, and where they didn't agree, again, false positives
predominated over false negatives.
In fact, there was only one false negative result in this trial, and that was a
55 ppm sample that tested negative relative to a 50 ppm level. Given the standard
precision achievable with Method 8080, this is a questionable false negative.
Similar validation data have been generated for pentachlorophenol tests
that were applied to contaminated soil. The first trial was done in conjunction with
Mississippi State Forest Products Lab.
Mississippi Forest Products laboratory is associated with Mississippi State
University and focuses on the State's forestry industry and applications of timber to
-------
397
various uses. One of those applications is treated wood products using
pentachlorophenol to make a range of products, principally telephone poles.
As a result, they had an interest in the remediation of sites that are
contaminated with wood treating products and, as such, they have developed a very good
laboratory skilled in analyzing for pentachlorophenol by EPA Method 8270.
They conducted the immunoassay test in their laboratory. They also did
the GC/MS analysis to determine the pentachlorophenol concentration in a variety of
soil samples that they collected from various contaminated wood treating sites
throughout the U.S. Again, the goal was to get a number of samples that were
representative of a wide variety of soil types to examine potential matrix effects.
The other trial that was conducted with the pentachlorophenil testing
product was one that was conducted in conjunction with the EPA's Environmental
Response Team who, last summer, conducted the first phase of a removal action at a
number of pentachlorophenol contaminated wood treating plants, primarily in the
Southeast.
They needed a screening test that would allow them to look at these big
sites, document contamination, where it was and where it wasn't, and get an order of
magnitude idea of contamination levels.
To do this they used the EnSys pentachlorophenol soil test out in the field
to screen about 1000 samples for pentachlorophenol contamination. They compared
these results to GC analysis performed in a mobile lab. I will show you a summary of
the results of about 200 of those samples that were run at the Brunswick wood treating
plant in Brunswick, Georgia, by the EPA's contractor.
In the case of the Mississippi Forest Products lab trial, a high correlation
was observed between the GC/MS results run under Method 8270 on the soil samples
and the immunoassay results The immunoassay overestimated the amount of
pentachlorophenol in only one sample.
In the case of the Brunswick wood treating trial, 176 samples were tested
using both methods. The results for 84 percent of the samples agreed between the
immunoassay and their field lab. Where there was disagreement, there was about a 3:1
-------
398
ratio of false positives to false negatives in the field.
I think, in summary, it can be said that immunoassays, as field screening
methods, have very promising application and that with real world samples in the hands
of real world users, results that correlate very highly with conventional analytical
techniques operated back in the laboratory can be obtained.
These tools, present an opportunity for the environmental community to
use them selectively in field work and money on the analysis side where, instead of
spending $150, $200, $300 for a lot of laboratory analysis, do some of the testing in the
field and back that up with laboratory analysis.
But perhaps of greater value than that is the ability to delineate
contaminated areas in the field and make decisions in the field with no delay. Most of
the benefit that is going to accrue from a time and money standpoint will be had there.
Thank you very much.
-------
399
QUESTION AND ANSWER SESSION
MR. TELLIARD: Questions, please?
While they are getting to the microphone, I have one, Kevin. A lot of
times when you take a sample like we are talking about for water, the old bugaboo about
colorometric problems in that you have either a discolored sample or a lot of turbidity,
could you expand a little bit about how the color doesn't interfere with your... with this
type of analysis?
MR. CARTER: Yes. Actually, for water samples, this kind of a test
presents a unique tool in that because you are binding the analyte to the antibody and
washing the rest of the1 unbound sample out of the tube, you are also washing any color
out of the tube that you have placed in it.
I know in working in these wood treating sites, many of the water and soil
samples are essentially black because of the creosote contamination that is present in
concert with pentachlorophenol contamination. While that would interfere with a
normal colorometric method, it doesn't interfere with these colorometric methods
because of the wash step.
MR. TELLIARD: Thank you. Yes?
MR. YOCKLOVICH: Steve Yocklovich from Burlington Research.
I just wanted to get a clarification. Were all the samples you were talking
about that were tested contaminated, or was it a blind or double blind study where they
had some uncontaminated samples to compare against? Were they expecting them to be
contaminated and they found it?
MR. CARTER: Well, for the most part, these were single blind
experiments. Obviously, some of them were done in the field where nobody knew what
was in the samples, whether they were contaminated or not.
Laboratory investigations, such as the Mississippi Forest Products lab trial
and DOE trial were set up to have a mix of uncontaminated and contaminated samples,
and the contamination range was chosen so that the samples were contaminated, in the
concentration range close to the decision levels that were being tested.
-------
400
We restricted the choice of samples to things that were in range as
opposed to either outrageously contaminated or predominantly clean. So, it was a
mixture of the two intentionally.
In the case of the field trials where work was actually done in the field,
there was no way of knowing. But as is typical with field situations, a large percentage,
perhaps half or more, of the samples you collect are actually contaminated below the
action level, and that was the case with the Brunswick trial. I would say roughly 60
percent of those samples were below the action level that they were interested in and
perhaps half of those were non-detects by the GC method.
MR. YOCKLOVICH: Thank you.
MR. SCHRYNEMEECKERS: Rick Schrynemeeckers with Enseco
Corporation.
The question I have for you is twofold. One, when you say you had like an
88 or 85 percent agreement between the GC and the immunoassay tests, is that for an
agreement on the concentration or just an agreement on whether it is or isn't a hit?
And the second question is, is the response variable to the PCB Aroclor
that you are looking for, or do you get basically the same response no matter what
Aroclor you are looking for?
MR. CARTER: Well, in response to the first question, it was kind of a
mixture between the two alternatives that you gave me, because we ran the tests, in this
case, in a ranging fashion, meaning that each sample was tested, in the case of PCBs, at
5 ppm and at 50 ppm.
So, if the GC result was 33 ppm and the immunoassay test said it was
greater than 5 and less than 50, then that I counted as an agree.
MR. SCHRYNEMEECKERS: All right.
MR. CARTER: If the immunoassay test said it was greater than 50 but it
actually tested at 33 with the GC, then that was a false positive.
MR. SCHRYNEMEECKERS: Okay.
MR. CARTER: In answer to the second question, the test is relatively
insensitive to which Aroclor you are testing.
-------
401
In the case of this test, Aroclors 1248, 54, and 60 test with essentially
identical sensitivity. When you step down to 1016 and 1242, you get a decrease in
sensitivity such that if you were testing at a standard semi-quantitative level at 5 for
1260, you would be testing at a semi-quantitative level of 10 in the same test for 1242.
And as you get down to something like 1221 which really is very rarely
found as an environmental problem, the sensitivity drops fairly drastically because of the
low chlorine content and resulting poorer recognition by the antibody of that group of
congeners.
MR. SCHRYNEMEECKERS: For a specific analyte like
pentachlorophenol, what would you perceive the problem to be in just using a UV/VTS
spec and running a curve using your test as opposed just to show/no show at specific
concentrations?
MR. CARTER: There is really no issue in doing that. It certainly can be
done. There is a range about a tenfold or twenty-fold concentration range, over which
there is a linear relationship between absorbance and log concentration, and one could
certainly measure the color generated by the enzyme and correlate that to standards that
one runs in the test.
MR. SCHRYNEMEECKERS: Thank you. Appreciate it.
MR. PERTUIT: Your immunoassay is very...
MR. TELLIARD: Could you identify yourself, please?
MR. PERTUIT: Pardon me?
MR. TELLIARD: Could you tell us who you are and who you are with?
MR. PERTUIT: Oh, I am Bob Pertuit, PPG Industries, Lake Charles.
MR. TELLIARD: Thank you.
MR. PERTUIT: Immunoassay is generally specific to one particular
compound. Have you determined which one of the PCB isomers that the assay is
actually reacting to?
MR. CARTER: Well, this particular antibody was developed against one of
the pentachloro congeners that is present as a predominant component of Aroclor 1254,
but there is enough vagueness in its response with regard to recognition of, 4, 5, or 6
-------
402
chlorine containing congeners that, in fact, it has relatively broad specificity as a result.
MR. PERTUIT: But for something like decachlorobiphenyl or
monochlorobiphenyl, it would probably be totally insensitive?
MR. CARTER: Yes, I suspect that is true. We have not actually tested
that, but the results with Aroclor 1221 bear that out.
MR. PERTUIT: Okay, thank you.
MR.WITHAM: Mark Witham, Bio-Tek Instruments.
A two-part question. With the false positives, did you do anything to look
at the reason behind the number of false positives, number one? And number two, in
your field versus laboratory work, were the same methods used including the methods
and instruments for washing and doing the actual end reading?
MR. CARTER: I will answer the second question first. We used exactly
the same setup in the laboratory and in the field. All of the instruments, the washing
setup, everything was the same.
In answer to the first question, in most cases, we didn't go any further with
samples that were false positives. In some percentage of those cases, it is simply
analytical error or analytical imprecision, I should say, in the GC or GC/mass spec
results, because with real samples, you know that if you send off samples to two different
labs, you get two different results within some kind of an error window.
So, that was responsible for part of them. Part of them were probably due
to matrix interferences, things in the soil that resulted in an overestimation.
One of the properties of these tests that is useful is the way they respond
to interferences. Because of the inverse relationship of color and concentration, if you
have something in the sample that inhibits the formation of color in the assay in some
way, that results in less color and, therefore, a falsely positive reading.
So, failure modes with respect to matrix interferences give you false
positives.
MR. MYERS: Harry Myers from Keystone Environmental Resources.
A couple of questions ago, there was a question related to developing a
curve for using a spectrometer to get a more quantitative result. With the test kit that
-------
403
you have, the color development is time dependent. So, everything would have to be
done within the same time frame, all the operations of the tests to make use of a
calibration curve.
In the field, you can't always do those kinds of things. Therefore, the fact
that you develop a color from a standard for some finite number of samples every time
and you are comparing back and forth between the sample and the standard makes it
useful in the field.
MR. CARTER: Well, that was exactly the reason that we used in setting
this up as a field analytical method that was semi-quantitative. I mean, you certainly
could run a 3 or a 5 standard line and then use this quantitatively, but because all the
samples are timed and because, typically in the field, you are under adverse conditions, it
is just not very easy to get accurate and precise results under the circumstances.
MR. TELLIARD: Joe?
MR. VITALIS: Joe Vitalis, EPA.
More of a comment than a question. The history of the names of Aroclor
go back to the production process so that if you had an arochlor 1260, it really is 12.60
percent chlorine by weight in the manufacturing process.
This means that those are the...that would be a more stable
pentachlorophenol...! am sorry...abiphenyl than if you had a 1021.
So, also the congeners that you get depend on how much chlorination and
also the process and when the biphenyls were actually made. I think you made a good
selection in using the 1254.
MR. CARTER: Thank you.
MR. TELLIARD: Can we wrap this up? One more, and then we will get
on to our next speaker.
MR. BOTNICK: Eric Botnick from Electro-Analytical Laboratories in
Ohio.
A couple of questions with the color development. Is there any
temperature dependency as far as the development of the color goes at the time?
MR. CARTER: Yes.
-------
404
MR. BOTNICK: As in as the temperature goes up, the color development
increases, or have you established a...
MR. CARTER: And that is why you run a standard at the same time as
you run a sample, to basically cancel that effect out.
MR. BOTNICK: Okay. Once the color has been developed and you have
stopped that development with the acid, is that color stable, and if so, how long is it
stable for?
MR. CARTER: It is highly stable over an hour's period of time. We
haven't actually tested it longer than that, but I hear tell that it is stable for upwards of
12 hours.
MR. SMITH: At least two hours.
MR. CARTER: There is user comment.
MR. TELLIARD: Thank you.
-------
405
-------
406
-------
407
-------
408
-------
409
-------
410
-------
411
-------
412
-------
413
-------
414
-------
415
-------
416
-------
417
-------
418
-------
419
-------
420
-------
421
-------
422
-------
423
-------
424
-------
425
MR. TELLIARD: Our next speaker is an old timer here. Jim Smith has
been an ongoing presenter at this meeting. He is going to discuss the analysis of PCBs
and an economic analysis using immunoassay methods for PCB's.
MR. SMITH: This is all real world stuff. You are going to get the
numbers.
I am going to give you a little site history. It is going to be a very small
place. We haven't got much money. We are going to have a lot of fun.
We are going to attack it normally by gridding the small site. I am going
to give you the results. I am going to tell you what successes and failures we had and
what we are trying to do to work on other sites with this assay.
The site is about an acre and a half. It is owned by a family. It was used
for storing heating oil.
They wanted to sell it. The first thing they did was take down the above-
ground heating oil tanks, and they were in the small berm to the left-hand side of the
site. You can see the outline of the berm with the right-hand side torn out. Those tanks
are now gone.
To the bottom toward the fence next to those above-ground tanks were six
very small, 500 to 1000 gallon gasoline underground storage tanks. They were also
removed.
In the site assessment, soil samples were taken to conform with New
Jersey's regulations. In so doing, you take VOCs, the infamous total petroleum
hydrocarbons, and base neutrals, and, of course, for the base neutrals, you look for the
PAHs.
That is nice. Fairly clean site. Ready to roll. And some idiot had to look
at the TICs.
In that bottom corner where the gasoline tanks were, there was as TIC
named l,l-prime-biphenyl-2,2-prime-3,4-prime tetra...and there wasn't enough space for
the rest of the answer. No problem. We have those all the time.
Well, it doesn't take too much to look at the mass spectrum and say oh-oh,
-------
426
we have got problems here.
Well, if you tell a husband and a wife whose cash flow is not very great on
their heating oil business "you have PCBs out there and you own them," you get a lot of
tears, gnashing of teeth, calling of lawyers, and the next phase is, what are you going to
do for us now that you have given us this good information?
We decided to do some surface samples to find out where the PCBs were.
Use the immunoassays, because they don't cost as much as laboratories, they go a lot
faster, we can be very selective on site. We are going to work on that hot corner where
the gasoline tanks were.
There was an electrical substation next door. We thought we would work
along that border. The rest of them we will do at random, a little here, a little there,
and see what is dirty and what is clean.
Oh, yes, one must have quality control samples somewhere just to make
sure it works. We decided to take out a field GC and use Tom Spittler's Region I quick
and dirty PCB analysis just to keep us in the ball park, and then, of course, a laboratory
GC method to tell us whether anything we did was right.
If it worked, maybe we could clean up the site if the owners had money.
We gridded it. Nothing difficult there. Quality control. We used blanks
out of our back yard soil. We tested it to make sure there were no PCBs in our own
back yard. This is aa added benefit of running such things.
We then made a spike of our back yard soil sample. We thought we would
do some replicates and, of course, the field GC and the laboratory GC.
We are on site. We have three days. One of the days it rained. So,
therefore, we are working inside and outside of the back of a pickup truck with a cap.
The person doing the analysis, both the GC and all the immunoassays, is a
high school graduate who has worked in industry for 26 years in a GC lab. He is no
dummy, but he certainly is not a Ph.D.,either. Of course, that doesn't prove much, does
it?
We are using a test for positive PCB above 5 ppm in soil on an as is, not
dry weight, basis. All the blanks, and that is 2 a day, show positive numbers, and positive
-------
427
numbers mean that PCBs are not found by the immunoassay.
This made us very happy. Our blank spikes, 20 ppm, as determined by
GC, spiked into our back yard soil should give us all negative numbers which means that
it is above 10 ppm for PCB 1242, and we do get negative numbers in each case. As a
matter of fact, the technician was so happy with this result, he didn't run the last one,
because he was supposed to run 2 a day. But when you are happy and things are going
right, why not run an extra GC at the same time?
The replicates looked pretty good except for Kl, and he kept running it
and running it and running it until he gave up. It just gave us fits. We don't know
whether it meant it was positive or was negative.
Field GC, Tom Spittler's method. It is a beautiful method if you want not
to use very much solvent or very much sample. One gram of dirt. You put in 2 mL of
solvent, shake like the devil, pull a microliter and shoot.
You use a 1-point calibration curve and hope like heck all your peaks are
on your chart paper, if not, you have got to go through a dilution, but even with
dilutions with a syringe, you never have more than 10 mL of solvent.
The solvent that we used is 1 part water, 4 parts methanol, and 5 parts
hexane.
These are the hits, 9 of them. Looks pretty good. Happiness reigns.
The clean samples, 7 of them. Happiness still reigns, except we ran 20 of
them. 9 and 7 still doesn't make 20. So, we have 4 that we just don't understand very
well.
Apparently, Kl is going to give us fits forever. I would dig up that one and
haul it out just for fun. The rest of them seem to be false positives.
Let's go to the lab work. The lab tells me it is 1248 and 1260, mostly 1248.
Again, the hits, and note that the highest hit is very close to 1000 ppm which is about 10
times the total concentration of PCBs on the TIC page for the same location.
So, it is a hint. If you see PCBs, as TICs, add up the J values, multiply by
10, and you are going to be fairly close to what is actually in that sample.
The clean ones, 10 of them. Looks very nice. Please note as we have gone
-------
428
through the numbers, the pluses and the minuses from the immunoassay system that it
doesn't look as if it is quantitative. Some answers give you very negative values with very
high amounts; some give you very negative values with very low amounts. It does not
appear to be quantitative.
Oh, yes, we still have the infamous 4, except these 4 are 2 different ones
from the field GC work. So, if you like, we now have 6, and again, they apparently are
false positives.
Note that for all of these, whether it be field GC or laboratory GC work,
we are dealing with the values near the go/no go level of 5 ppm. There is no major
disaster there.
What did we do in three days? By the way, the technician told me if I said
they were 8-hour days, I had to pay him time and a half for the rest of it. So, therefore,
they were closer to 12-hour days.
One hundred seventy-two runs on the assay system, 20 on the field GC,
and he worked hard. It would be much simpler to have two people working the system
and setting up each. You would get a lot more done a lot faster without burnout.
Twenty-one samples were sent to a laboratory with full CLP-like data
packages.
Was it successful? The map again. The red X's are the hits for the
immunoassay. The red solid squares are hits for either the field or laboratory GC. The
green X's are the clean, and the fukk green boxes are laboratory or field clean samples
by GC.
The spot that seems to be most contaminated is where they dug up the
gasoline tanks. I have yet to hear anyone mention that PCB is a new additive for
gasoline.
When they dug up the dirt, guess where they placed it? On the concrete
right next to it, and, of course, that is the next place that is most contaminated.
And where did they push the concrete after they cleaned up some of the
soils they took out when they were removing the gasoline tanks? Well, down by S, about
two or three or four, and the S mark is where you will find the concrete pieces.
-------
429
So, it makes sense that the system worked very well in the field. We know
where the contamination is.
How did the contamination get there? The previous owner was a supplier
of fuel oil. He was also a hazardous waste hauler. New Jersey did indict him, so it is
said, for hauling fuel bil laced with PCBs to be burned in New York City apartment
buildings. Who knows? But the PCBs are certainly there where the gasoline tanks were.
It cost approximately $12,000 to do all the tests on site. That comes down
to about $30 on an immunoassay test and the technician's time. The field GC was free.
It was his time anyway.
The laboratory confirmation cost about as much as the 172 tests that we
made on site, and they came in about two weeks later.
This does not count the time and effort put in by the sampler whom we
kept very busy. The amount we saved is approximately an equal amount to what was
spent. If we had gone to a laboratory and received just sample results sheets, the cost
would have been about $25,000.
The success is that it worked. I think we did map the problem. It is a
success if you can do that and save money.
If there is a failure, it is the false positives and possibly a false negative
that occurs near the detection limit of the system.
Presently, we are trying to get approval to use the system on a Superfund
site to help guide test borings to determine where the PCBs are on about a 3-acre
Superfund site in New York State. We are trying to lower the detection limit, because
the United States EPA developed a ROD for a 1 ppm cleanup. We think we have
accomplished that.
We are not sure of the extraction efficiency which uses methanol on very
wet samples. We think it works well. We have just never tested it.
The other thing we would like to do is to compare, side by side in the field,
various companies' systems. They are not exactly the same, but we would like to
compare them to see which one is easier to run, which one works better at the detection
limit and gives the fewer false positives, and which one would be most cost effective to
-------
430
do a screening analysis for a site for PCBs.
I thank you for your attention. We had fun. I suggest you try it. It works.
QUESTION AND ANSWER SESSION
MR. TELLIARD: Questions? Are you going to let him off?
Okay, thanks so much. Oh, hold it. Got you.
MR. THOMAS: Yes, my name is Roger Thomas. I am from Viar and
Company.
I had a question concerning the false positives and false negatives. The
majority of times, false positives and false negatives are usually attributed to
interferences, and the way to remove interferences is through various cleanup
procedures.
Have you done any studies concerning various cleanup procedures both for
the field GC analysis and the colorimetric procedure that you used in your
immunoassays? Like, for example, you have a wash procedure that removes
interferences prior to performing the colorimetric procedure. Have you tried various
types of washes?
MR. SMITH: No.
MR. THOMAS: Okay, thank you.
MR. TELLIARD: Thanks, Jim.
MR. TELLIARD: Our next speaker is going to discuss some work that the
Office of Water sponsored looking at the implications of immunoassay.
Harry?
-------
431
COST EFFECTIVE PCB INVESIGATION
UTILIZING IMMUNO ASSAY
By:
Jim Smith and Gene Brozowski
Trillium, Inc.
and
John Rhodes
Rhodes Engineering
-------
432
PRESENTATION OUTLINE
1. Site History
2. Plan of Attack
3. Results
4. Successes and Failures
5. Tomorrow
-------
433
1 STOREY
MASONRY
BUILDING
D
1 STOREY
MASONRY
BUILDING
-------
434
PLAN OF ATTACK
1. Surface Samples
2. Immuno Assay
Hot comer
Electric substation
Random
Quality control (QC) samples
3. Field Gas Chromatograph (GC)
4. Laboratory GC
5. Option of Cleanup or Depth Profile
-------
435
2 4 6 8 10 12 14 16 18 20 22 24 26
V
1 f
V
T
,9
R I
Q
p
0
\ '
I/
L
K !""
// i ''
r i -
c • ;l
tt - :
< i
{ I- • •
q
•
• ; -
:-;.:.
(»:•••'<
• •
; • -
; •
p • •
• •
•
: • •
m
• • :
• : • ;
• ; • •
.A...
• ': '
• : •
: (
<
<
0 1 i ! j i ' j I : I <
• :«.•:«.. ».».••»:•••:•
O : .
• .••:.
4> ; : i
• : m • f •
• : • • i «
•
•
•
) :
•
•
•
•
•
* * •'*;•.
1 STOREY
MASONRY
BUILDING
1 T
. ; . . ;
•
j.
A . i : M
1 2-3 4\5\6 7
;**
^^-^^>-
""^^^^
• : •
• : i
i ; i <
•;'•'•.«
• i
•
: -
; •
•
m m \ • i • ':*.'•
r "
^
. .„ ...»
*
: . (
1
8 91011 1
* i •
• l
• •
• •
• ' •
1 i
: <
^; i<
'- <
<
<
(
£
1
1
4
}
»
>
i
• : •
* : *
• ; •
• • : •
• ' •
i ' :
:•;•=•:
1 STOREY
MASONRY
BUILDING
* •
i •
; •
: •
• • ;n
: •
(
)
(
<
<
C
u
A
H
0 <
D
0
C
P <
C
N
1
1 STOREY
MASONRY
BUILDING
16 18 20 \ 22 24
ELECTRIC
SUB-STAT ON
, . . .-> • « o — •& -er-
\ 2
':
<
(
«
" V
f- L
». T
'• i S
• R
M Q
• : P
•I o
: : /v
' i ^
' i L
* i /
( i H
i F
i Z?
i ^
>i
'5;
> :
}
i
1
-------
436
QUALITY CONTROL
1. Blanks
2. Blank Spikes
3. Replicates
4. Field GC
5. Offsite Laboratory GC
-------
437
BLANKS
All positive sample values indicate a "not detected1'
(ND) with a detection limit of 5 parts per million (ppm)
PCBs.
Blank No. Value
1 +0.18
2 +0.14
3 +0.32
4 +0.30
5 +0.27
6 +0.05
-------
438
BLANK SPIKES
20 mg/Kg (Air Dried) PCB 1242
All negative sample values indicate that the sample
contains 5 mg/Kg or more of PCBs.
Blank Spike No. Value
1 -0.33
2 -0.76
3 -0.11
4 -0.10
5 -0.23
-------
439
REPUCATES
(Co-located)
Values
Run 1
+0.04
-0.12
+0.15
+0.24
-0.07
-0.43
+0.77
Run 2
+0.22
-0.07
+0.22
+0.28
-0.07
-0.25
+0.27
Run 3 Run 4
+0.22 +0.09
Run £
+0.01
Sample Location
D-9
G-70
H-3
H-74
K-7
K-5
K-70
Positive values: <5 mg/Kg (wet weight)
Negative values: >5 mg/Kg (wet weight)
-------
440
FIELD GC
'HITS" (Negative Values)
Immuno GC Value
Sample Location Assay Value ma/Kg (wet wt.)
D-6 -0.47 >400
F-9 -0.09 16
E-1 -0.55 100
J-5 -0.39 18
J-4 -0.56 720
/•7 -0.23 24
D-7 -0.42 40
D-4 -0.51 400
1-5 -0.60 T50
-------
441
FIELD GC
"CLEAN" (Positive Values)
Immuno GC Value
Sample Location Assay Value ma/Kg (wet wf.)
D-70 +0.06 ND
B-6 +0.36 ND
C-8 0.00 2
B-70 +0.07 ND
B-8 +0.38 2
H-20 +0.37 ND
J-14 +0.28 ND
ND f's not detected with a detection limit of 1 mg/Kg (wet weight).
-------
442
FIELD GC
"OOPS"
Immuno GC Value
Sample Location Assay Value ma/Kg (wet wt.)
G-10 -0.12 and-0.07 ND
K-1 -0.01 and -0.01 6 and 8
and +0.22 and
+0.09 and +0.01
K-9 .Q.01 1
D-8 -0.06 4
3 False Positives
1 False Negative
ND is not detected with a detection limit of 1 mg/Kg (wet weight).
-------
443
LABORATORY RESULTS
(Sum of PCS 7248 and PCB 1260)
"HITS' (Negative Values)
Immuno GC Value
Sample Location Assay Value ma/Kg (dry wt.)
B-1 -0.51 730
D-4 -0.54 960
D-6 -0.47 700
E-1 -0.55 640
H-6 -0.45 210
J-5 -0.39 36
S-4 -0.22 5
-------
444
LABORATORY RESULTS
(Sum of PCS 1248 and PCS 1260)
"CLEAN" (Positive Values)
Immuno GC Value
Sample Location Assay Value ma/Ka (dry wt.)
A-15 +0.32 0.4
A-16 +0.33 0.8
A-17 +0.37 ND
A'18 +0.06 2.7
A-26 +0.34 0.5
B-B +0.36 t.3
B-10 +0.07 0.6
L-14 +0.15 0.6
O-24 +0.67 ND
V-12 +0.63 ND
ND is not detected with a detection limit of 0.1 mg/Kg (dry weight).
-------
445
LABORATORY RESULTS
(Sum of PCB 1248 and PCS 7260)
"OOPS"
Immuno GC Value
Sample Location Assay Value ma/Kg (dry wt.)
G-10 -0.12 and -0.07 ND
K-1 -0.01 and -0.01 1.5
and +0.22 and
+0.09 and +0.07
K-9 -O.Of 1.4
P-2 -0.23 1.7
3 False Positive Values
1 False Negative Value
ND is not detected with a detection limit of 0.1 mg/Kg (dry weight).
-------
446
2
8 I 10 \ 12 \ 14 I 16 ; 18 \ 20 \ 22 124 26
1 STOREY
MASONRY
BUILDING
1 STOREY
MASONRY
BUILDING
1 2 3 4 5\6 7'8 9W11 12 14 16 18 20 22 24 26
KEY:
• i Positive Result for tmmuno Assay
O NO for Immuno Assay
^3 Positive Result for Immuno Assay
and/or Field GC and/or Lab GC
ND for Immuno Assay
and/or Field GC and/or Lab GC
Oops
ELECTRIC
SUB-STATION
-------
447
SUMMARY
SAMPLES ANALYZED
Immuno Assay (3 Days)
# Samples: 151
# Blanks: 6
# Blank Spikes: 5
# Replicates: 10
Total: 172
Field GC
# Samples: 20
Laboratory GC
# Samples: 21
-------
448
SUMMARY
SAMPLE COSTS
Immuno Assay $5,160
Field GC Analyses 2,560
(including technician for field GC)
Laboratory GC Analyses* 4.050
Total Analytical Invoice: $11,770
*lncludes 2 duplicates, 2 matrix spike/matrix spike
duplicates, and data package.
-------
449
SUCCESSES AND FAILURES
Successes
1. Surface PCBs Mapped
2. Money Saved
Failures
1. False Positives and False Negatives
-------
450
TOMORROW
1. Compare immuno assay systems
2. Lower detection limit to 1 mg/Kg (ppm)
3. Determine extraction efficiency
versus sample water content
4. Field test at a Superfund site
-------
451
MR. MCCARTY: Good morning, folks. I would like to thank you for
hanging around for Thursday.
I would like to make three points real quick. One, I am not Bill Telliard.
Two, contrary to popular confusion, I am not with EPA, and three, I am not here to help
you.
[FIRST SLIDE] I am also not Cindy Simbanin, but I figured you could
figure that one out on your own. Cindy couldn't make it today to give you this
presentation. She asked me to give you her apology, and if Bruce is in the audience
somewhere, she particularly wanted to be remembered to Bruce Colby. Not sure why.
As Bill mentioned, we are going to talk a little bit about the work that was
done in conjunction with the pesticide effluent guideline work that Bill's office has been
doing in this past year or so. We looked at the traditional gas chromatographic methods
in concert with some immunoassay techniques that have become available.
[SECOND SLIDE] By way of introduction for those of you who haven't
been regulated yet or don't know what an effluent guideline is, under the authority of the
Clean Water Act, EPA develops guideline limitations for the 64 industrial categories, in
the U.S. that discharge wastewaters into surface waters of the United States.
The current slate of guidelines include the development of guidelines for
the pesticide manufacturing industry and the formulators and packagers, the people who
take what somebody else manufactures, mix it together, add the 87 percent inert
ingredients, and put it in a bottle that you can't get the cap off unless you are a five-
year-old.
[THIRD SLIDE] As a result of trying to develop these guidelines, EPA
collects a large amount of data. The number of samples will vary from facility to facility
and, certainly, is dependent upon the industry, but there are a large number of
compounds that they look for at any one time.
At a minimum, we are talking about looking for things on the priority
pollutant list which brings it up to about 126 compounds.
The sample types span the range of what is produced in an individual
-------
452
industry. The obvious importance of the final effluent that is being discharged is pretty
evident, but also, they look at the in-process waste streams, in part, to look at treatment
technology. If you have got a lot of it in your process and you end up with very little at
the end, then you have got a good treatment process, and there is some measure of that
treatment efficiency.
[FOURTH SLIDE] In terms of analytical methodologies, and traditional
methodologies in particular, you are talking about the need for relatively high sensitivity,
sub-ppb levels for the pesticides in particular. We are talking about a wide range of
concentrations, because you are dealing with not only final effluents that are supposed to
be pretty clean, but also in-process samples.
Given the matrix effects you would expect in in-process samples, you are
going to have to be able to deal with those kinds of challenges to the technique itself.
[FIFTH SLIDE] For the traditional GC approach, you are talking about
some sort of a solvent extraction. For the herbicides in particular and other related
compounds, you are often talking about a derivatization step which helps deal with some
of the matrix effects.
There are various cleanup techniques including florisil alumina, and silica
gel column cleanups, GPC in certain cases, etc.
And then the analysis is typically dual column GC with one of several
detectors, EC, the Hall or electrolytic conductivity, or an NPD (nitrogen/phophors) kind
of detector.
[SIXTH SLIDE] As any of you who have ever run those analyses know,
there is a whole suite of problems associated with those analyses. While an individual
analysis may not appear particularly costly, the cost to EPA for a large number of
samples adds up.
Waiting 35 to 40 days, in some cases, to get the data back is certainly an
issue in terms of going out and making sure that you have collected the right samples
from the right part of the industry.
The wide range of concentrations that you find complicates the problem
for both the agency and the laboratory. You often end up running a large number of
-------
453
dilutions to get 1 of 27 compounds in range for this dilution and 2 others at a higher
dilution, et cetera.
For the laboratory, certainly, you can easily contaminate your instruments
and other samples or the sample processing glassware by having really nasty, ugly, dirty
samples come in that didn't look that bad when you took them out of the bottle. And
this often leads to down time in the laboratory and, from the perspective of EPA, or
Viar and Company as one of their contractors, it leads to a certain amount of down time
as we wait to get the data back, and then determine that the lab was supposed to have
, run another dilution.
So, we are looking at ways to get arpund these problems in a cost effective
manner. We are looking at the promotional material that has been available for
immunoassay for the past couple years. I know I get it. I have seen it in Bill's office,
and most of the rest of you probably get swamped with 3rd class mail around the time of
the Pittsburgh conference, and occasionally you read some of it.
[SEVENTH SLIDE] We have talked with a number of the vendors for
immunoassay materials in the past year, and are looking at some of the materials that
were available, and the advantage that we could see was, obviously, as you have heard
already this morning, they are quick. They can be highly selective, depending on what
analytes you are looking at and what matrices.
They are relatively simple to use. We could even teach engineers to do
this, I suspect.
They are portable, or fieldable. You could carry some of the immonoassay
kits in a knapsack if you needed to go out in the field, but, certainly, given all the other
things that are being lugged around in coolers, this technology presents no problem in
terms of getting it to the site.
[EIGHTH SLIDE] The purpose of this specific study was to provide a
rapid, cost effective comparison of immunoassay techniques with traditional gas
chromatographic techniques on what everyone else here has been talking about, real
world samples. Nasty, ugly, awful, effluent and in-process samples.
If you know how to read the buzz words, rapid and cost effective, we are
-------
454
talking quick and cheap. Quick because we wanted to do it in concert with the existing
schedule for the pesticide manufacturing, packaging and formulating work that was going
on out of Bill's office; cheap because we didn't have a lot of money to do it.
[NINTH SLIDE] We designed a study, again, piggybacked on the existing
sampling plans for these industrial facilities. So, it had to be done on the fly in that
regard, and we were fortunate in that it didn't take a lot of work to set up something for
immunoassay. It doesn't present any difficult problems in terms of sampling.
We were already using or getting ready to use the traditional GC methods,
and we added to that what we are calling the tube and plate immunoassay materials.
We were dealing with commercially available materials from Millipore.
Their tube kit looks very much like the material that Kevin Carter was
talking about this morning. I want to thank him for putting up the nice slides, because I
certainly didn't have the facility to do the diagrams specifically of the process.
There was some discussion of the plate kit. It is a little well kit like you
would see in a medical testing laboratory. 96 wells to a plate, 8 by 12, something like
that, and there is a reader that will do a quantitative, or certainly much more
quantitative, measure of the color development and give you a better number.
The tube kits, again, we used as a semi-quantitative screening tool. It is
essentially a 1-point calibration.
We were using a calculator-sized, you can stick it in your pocket, color
comparator that Millipore has. It will give you an instrumental reading much as Jim
Smith was showing, (minus point something to plus point something) which can be
related to concentration.
[TENTH SLIDE] The tube kits were run in duplicate, which is also a
recommendation from the manufacturer.
For the tube kits, we started out saying, we know this is semi-quantitative,
let's not go overboard on running 17 replicates of all of these things. We decided to run
a duplicate if running of a duplicate, and use that to figure out what level of dilution of
the sample was most appropriate to move forward with for the plate kit, which is alleged
to be a lot more quantitative. It certainly was in our case.
-------
455
As a result, we ran the plate kits in triplicate once we figured out the best
level of dilution using the tube kit, rather than going through the whole process again to
determine what is the best level of dilution. We said the immunoassay procedure is the
same regardless of the form of the kit, whether it is the plate or the tube. Therefore,
once we had established, through the use of the tube kit, how many hundreds of times
you had to dilute some of these samples to get them in the range you want, we just went
ahead with that level of dilution and ran the plate kit in triplicate.
The GC methods were performed as traditionally is done; as a single
analysis. We took the ERL Athens, Georgia approach " Anybody who runs a sample
more than once gets exactly what they deserve. "
We looked started with either a neat analysis for GC and then diluting
down, or making an intermediate level of dilution as a first estimate, and then going
back and either reconcentrating or diluting, in most cases, much further, in order to get
all of your analytes within the range of the GC method.
All the GC results were confirmed by a second column analysis.
All of this work was done in a commercial laboratory under contract to
Viar and Company, under our EPA contract.
As was discussed earlier this morning, these materials are fieldable, but we
wanted to have a single set of operators essentially doing everything.
[ELEVENTH SLIDE] We were looking at four specific analytes by GC
that were amenable to the immunoassay techniques that were commercially available.
The metolochlor, atrazine, the two endosulfans, and 2,4-D were the analytes that we
picked from the GC method and came up with corresponding immunoassay kits.
As you can see for the first three, you are not talking about a compound-
specific immunoassay. There are a lot of triazines out there. We were concerned about
atrazine because of the facilities we were working in.
There is an even larger suite of possible cyclodiennes compounds that the
immunoassay is sensitive to. 2,4-D was very convenient because it is a compound-
specific immunoassay.
[TWELTH SLIDE] Again, the work was all done in a single commercial
-------
456
laboratory. The sampling crews went out in December of '91 and February of this year
to sample two facilities, based on knowledge of what was going on at the planned time of
sampling.
We were looking at 2,4-D, atrazine, and metolochlor in the first facility.
There were four of what the facility called treated effluents. These were typically
materials that had been through some level of bulk carbon column cleanup or some
other treatment technology to which a final finish was applied before you got to the final
effluent.
We looked at two in-process wastewaters from the first facility, and then
one raw water, which is essentially their well water or city water, whichever it is that is
coming into the facility.
[THIRTEENTH SLIDE] At the second facility we were concerned about
endosulfan in particular, because of what they were doing. With packagers and
formulators in the pesticide industry, what they are doing this week may be radically
different from what they did last week or what they do next week, or even what they do
two days from now.
The second facility had five of what they call untreated wastewaters, two of
these partially treated effluents, and then three different final effluents that we looked at
at that facility.
The tests were all done at a single contract laboratory, which had little
experience with immunoassay techniques. We got the materials to them a couple weeks
ahead of time, and they played around with them and got some experience using some
leftover sample volume that they had in the laboratory.
They reported back some potential problems to us and some specific
concerns from the standpoint of the work we were trying to do.
We worked with the manufacturer, Millipore, at that point, to try to resolve
some of these issues so that we didn't start work on real samples with a bunch of
laboratory people who were essentially rookies at the technique. I want to thank
Millipore for their responsiveness to our time frame in getting answers back to the
laboratory. We were acting, in part, as intermediates here on the issues of the
-------
457
technology itself.
[FOURTEENTH SLIDE] I am actually going to present some real data.
These are the data as received from the laboratory. One of the biggest problems is we
forgot to tell the laboratory to treat the immunoassay data like everything else and not
give us every decimal place they could generate on an 8 bit computer chip.
So, as a result, 937 probably isn't a real significant number there, but I did
not make any attempt to round off these data.
The 2,4-D results; the first column of results is the results from the
duplicate tube kits. That is a mean of the two results and a relative percent difference
in parentheses. The second column is for the plate kits. It is the mean of the three
replicates and a relative standard deviation. The third column the GC is a single value
reported to essentially two significant figures there.
Some of these samples are pretty highly contaminated. For the in-process
waste streams at the bottom, I did not put the commas into the GC results on that last
one, but we are talking about percent levels of 2,4-D in the process itself.
The answer by any one of these techniques for the in-process samples,
there is a hell of a lot of 2, 4-D there. You are not going to discharge that, so from a
compliance monitoring standpoint, it doesn't matter if it is really, 33365 or 40,000, or
whatever. It is a lot of material, certainly.
If you look at the raw water results, there was very little, if any, 2,4-D
found. The GC detection limit is probably on the order of 0.1 ppb in this analysis.
So, you have a detectable level of 2,4-D there. It is confirmed quite well in
the plate test. For the tube test, it simply was not there below a 1 ppb level. All these
values are in ppb, micrograms per liter.
For the first of the various treated effluents, there is not very good
agreement at all among the tube results, the plate results, and the GC results. With a 29
percent RSD, there is a fair amount of overlap there, certainly.
For the second treated effluent, the tube and the plate come into closer
agreement; the GC results look to be higher. I am not going to make apologies or
excuses or explanations about these data at this point, but I think there is a fair amount
-------
458
of agreement, certainly, between the two immunoassay techniques and the GC results.
In particular, the 2,4-D results look fairly similar.
[FIFTEENTH SLIDE] For atrazine, the biggest problem we found was we
could not find it by the GC method because of sensitivity. The limit on the atrazine
method, as run, was about 20 parts per billion, and we weren't seeing it there. By doing
a little bit of work on the raw water, we were able to get down to a lower concentration
and confirm a number around 1 ppb.
The tube and the plate kits look fairly close, certainly within a factor of 2
to 3 for the triazines. If we had run the tube in triplicate, as opposed to duplicate, we
could have even done a T test and maybe shown a little bit better agreement.
[SIXTEENTH SLIDE] The metolochlor results are shown here. Again
the basic problem is that it was not detectable by the GC method. We came up with
quite close agreement for most of these samples. For the raw water, we couldn't detect
it in the tube kit. It was below the sensitivity of that, but we got reasonably decent plate
kit results there with a 10 percent RSD at a low level.
[SEVENTEENTH SLIDE] Endosulfan was another one where there was
clearly a lot of it out there. Huge results in some of these untreated effluents for the
tube and the plate kits. But if you look over at the GC results, you will notice very
quickly that the GC does not confirm, in many cases, either the presence of endosulfan I
or H, or the concentration. I will talk a little bit more about this later.
As you can see from this slide, this is the worst of it. There are the
untreated effluents, the five of them, and some of those are pretty heavily loaded.
[EIGHTEENTH SLIDE] For the treated effluents, clearly you got lower
results with the GC than with either the tube or the plate kit.
The first of the final effluents had non-detects by any one of the
techniques. In the second final effluent, the plate kit came up with a 12 ppb level where
the GC method did not confirm it.
[NINETEENTH SLIDE] Basically, there is a lot of comparison between
the techniques for certain analytes and certain samples. The cleaner samples were
clearly better. Looking at final effluents, and at raw water, where you have not got
-------
459
significantly elevated concentrations of any of these analytes, there was pretty good
agreement. Based on the reports from the laboratory there seemed to be a lot fewer
problems with the samples.
This is certainly consistent with the results that other people have shown
for agricultural runoff, for drinking water, things like that.
You don't expect a lot of problems with samples that are on the order of
10 ppb or less. These techniques are all relatively sensitive there, and you get good
reproducibility.
The tube kits are semi-quantitative by design. Nobody advertises them to
be significantly different. As we heard earlier this morning, the quantitation is based on
a 1-point calibration.
One could go out, presumably, and develop a multi-point calibration, but
given you are looking at a color development reaction in the field, I don't know that it
would really make a lot of sense.
In a fixed laboratory, you would have a lot more possibility to do that, but
you also have the possibility to do the plate kit, which is certainly intended to be more
quantitative.
[TWENTIETH SLIDE] Both the tube and the plate kit results gave pretty
good precision, judged by RPDs or RSDs, for some of these samples and some of the
analytes. For other samples and analytes, there is poor agreement. There is no hiding
that.
The 2,4-D results probably are the best, and, again, that is the instance
where we are talking about a compound-specific analysis. As one might expect, when
you get to higher concentrations, the agreement is worse, in part because of the dilution.
We are talking about samples that were diluted by a factor of 500,000 to 1
to be analyzed by almost any one of these techniques. Given percent levels of
endosulfan or some of these other compounds, you are going to have to dilute the hell
out of these samples to put them on a GC without blowing it and the laboratory away.
The typical dilution factors for the immunoassay kits were on the order of
100 to 1000 just to get a sample run. Even some of the treated effluents had to be
-------
460
diluted fairly heavily to get the analytes within range.
If you have got that kind of a dilution factor, you are going to have a fair
amount of error associated simply with the dilution. If you are doing it by three different
techniques, each of those samples was a separate sample, so you have a compounded
error there.
[TWENTY FIRST SLIDE] The agreement was clearly worse for the
atrazine and the endosulfan results where you are not talking about a compound-specific
immunoassay. Any one of the triazine compounds could have produced some level of
positive response in that kit. We were looking for specific compounds.
The kit that does the endosulfan is designed for any one of the cyclodienes,
and given the materials that could be in these samples that we were not specifically
targeting by the GC method, there is a significant possibility of false positive results.
[TWENTY SECOND SLIDE] We did look at the possibility that the GC
results were false negatives due to severe matrix effects in the in-process samples. So
far, from the evidence we have, we don't think that is the case.
We looked at matrix spike data and blank spike, (OPR) data from the GC
method, and there is no indication of false negatives at all. If you look at the spike
sample data for the tube and the plate kit results, we are getting recoveries on the order
of 700 percent for the atrazine or the endosulfan in some cases.
Obviously, there are other compounds in those samples that are providing
a positive response and that we are not measuring by the GC technique itself.
We were a little bit surprised by the metolochlor results. The samples
were too low to be quantitated by the GC technique, but we got very consistent results
between the tube and the plate kit. Certainly within a factor of 3 or better, between a
technique that is designed as semi-quantitative and one that is designed as a quantitative
technique.
So, we were fairly pleased with the results for metolochlor and for the 2,4-
D. Again, the metolochlor is not a compound-specific analysis, but either because of the
facility we were sampling, or the particulars of the immunoassay test, we did not seem to
have a lot of variability in those data.
-------
461
[TWENTY THIRD SLIDE] We set out to look at a relatively rapid
comparison of the techniques in the constraints of what effluent guidelines development
needs for analytical work, what Bill Telliard's program is going to do in terms of field
sampling.
We think the immunoassay kits worked reasonably well on some of the
effluents that we looked at. We certainly think they worked better for the less
contaminated samples than for the highly contaminated samples.
The compound-specific test kits clearly have an advantage when compared
to GC results. There is a lot better agreement there.
We think that one of the biggest uses of immunoassay kits in the context of
Bill Telliard's work is to identify those high concentration, awful, nasty, samples that you
only discover three days after they go to the lab and somebody calls up and says, "do you
realize we are not going to be able to analyze anything for a week until we get the lab
clean?" Some of the labs we deal with routinely know the kind of samples we are talking
about, the ones that crap up the columns and the glassware and everything else.
Even with the ability to do a quick "shake and shoot" solvent extraction in
the laboratory, that takes a lot more time than the 30 minutes to an hour that an
immunoassay would take to tell you that this is the really hot sample.
[TWENTY FOURTH SLIDE] Certainly, in the context of using them in
the field, the tube kits would be relatively straightforward for a field sampling crew to
deal with in the course of processing samples.
You don't require a large sample volume, so you do not have to take a
whole another one liter bottle of sample in order to do the test. As you heard earlier
today, you put the sample in a test tube, you add ,a couple of reagents. These come in
little squeeze bottles. It is not unlike doing a home swimming pool chlorine test in some
respects.
If you can live with a "presence or absence" kind of number, you don't even
need to use the small color comparator that you can get for these materials. You could
readily determine which samples were likely to be highly contaminated and which were
likely to be relatively clean.
-------
462
The plate kits could easily be used in a fixed laboratory for a screening
technique, the advantage being you are not talking about using solvents (which is one of
the things we would like to get away from where possible) and having to go through the
aggravation of extracting industrial effluents.
These are samples that cause emulsions that break about three and a half
days later. (By that time, you could have got the sample through the EMPORE disk.)
The immunoassay techniques ought to be a lot quicker, ought to be a lot more easy to
set up in the laboratory if you are going to do it on a routine basis.
The advantage to the laboratory, obviously, is to avoid unnecessary time
spent running diluted samples and the costs associated with contaminating the glassware
and the instrumentation.
From the standpoint of the Office of Water, we think that there will be an
advantage in terms of not only cost, but in the timeliness of the results that come back to
EPA.
[TWENTY FIFTH SLIDE] This is strictly a personal opinion, but I don't
think you are ever going to replace traditional GC methods with an immunoassay in
terms of developing a data base for an effluent guideline. I think either there is going to
have to be a lot more development of the quantitative results for these techniques and
the compound specificity or Bill is going to still be paying for an awful lot of traditional
GC analysis.
There is some potential, from my point of view, for compliance monitoring
at an individual facility if you are talking about compound-specific immunoassays.
In terms of the cost issues, if you have to use seven different immunoassay
kits to do your compliance monitoring for a suite of analytes for which you are permitted
on under NPDES, the cost isn't going to be as advantageous as if you can get away with
a single kit that is compound-specific. If you are talking $300 for a GC analysis and
maybe a half to two-thirds of that for immunoassay, there is probably not going to be a
rush to go out and use those techniques for compliance monitoring.
But, certainly, I can see a number of instances where it could be very
helpful.
-------
463
[TWENTY SIXTH SLIDE] Finally, I would like to acknowledge Don
McCarthy and Barbara Young at Millipore, in particular, who were not only very helpful
in terms of getting us information and getting us the materials we needed when we
needed them, but who were very responsive to the questions that we were having to
transmit to them from the laboratory.
And the analytical work was all done at Analytical Technologies in Fort
Collins, Colorado. Steve Workman was heading up that work and spent a lot of time
working the bugs out from his laboratory's end of how to do the immunoassays. We
actually were able to get some feedback from the laboratory that Millipore found very
helpful in terms of what a production laboratory is going to expect out of any sort of
technology, and we owe a lot to Steve for the time he spent doing all of this.
Thank you.
-------
464
QUESTION AND ANSWER SESSION
MR. TELLIARD: Questions? You know, a little historic note here, on
April 10th, 1992, the proposed reg went out for pesticides formulators or manufacturers.
We are looking at formulators and packagers at the present time.
Over the next 24 to 36 months, what we would like to do is generate a data
base that is large enough that when we go final with these rules allows for the purpose of
compliance monitoring, again figuring that your permit will not have 92 pesticides in it
but perhaps 3 or 4, that this type of technique...again, understand the agency can live
with a false positive.
I mean, people out in the field can't, but we don't care about you. We can
live with a false positive because what it says to us is then you have to do the whole
rigmarole. You have to run the GC/FID, LC/MS, whatever you are going to do.
So, what we are looking at here is trying to start building into the
regulations an opportunity to cut down on the compliance monitoring costs, and, of
course, since it will be $1.38, that means you can do it every hour which will save a lot of
money. Right?
But we are looking at... we don't want to overload all these commercial labs
with samples, you know. They are constantly having to push them off the benches. This
way, on a routine basis, you can turn the sample around. If you get a positive hit, then
you are required.
And that is kind of the thinking here. The in-process streams that Harry
was referring to here, I mean they basically when you sample these things and you send
them to the lab. They kind of rattle in the jar, and we affectionately refer to them as
water samples.
So, that is kind of the thinking, and a lot of this, as you notice, is
preliminary. So, we are going to, over the next 24 months, be spending a lot more time
trying to work on this data base as it relates to, again, treated industrial effluents is what
we are looking primarily at.
Got a question?
-------
465
MR. HARRISON: A comment first and then a question. My name is Bob
Harrison. I am from ImmunoSystems which manufactures the kits that they were using.
The comment first is that the cross reactivity issues that you were
addressing fairly well are significant to comment on here. The difference between the
metolochlor and the atrazine that you were seeing may be due to the fact that the
alachlor kit is cross reactive with a series of compounds. It is a very small group,
actually, about three or four. And the triazine kit is cross reactive with as many as a
dozen.
The presence of small amounts of several of that dozen may actually
account for the significant variation between the GC and the immunoassay results.
Another factor that may come in here...and I would like you to comment
on this...isthe dilution factor. Your alachlor detection level, I think, was 20 ppb and you
couldn't reach that, or you couldn't reach the level that the samples were at with the GC.
Can you comment on how the dilution factor might account for, for
example, the difference between the metolochlor and the cyclodienes where you had a
profound dilution?
MR. MCCARTY: We have gone back and tried to look at the whole issue
of diluting the samples and the sensitivity of the GC method. I will confess it has been a
low priority at some level simply because we wanted to get some of this information out
in the context of this conference.
The GC analyses were all done under what was then called Method 1618,
which is the Office of Water combined organochlorine/phenoxy acid/herbicide/pesticide
method.
Each of the three individual pieces of that method has a series of surrogate
compounds that have method-specified recovery limits. If you can't meet the recovery
limits in the analysis, and some of these samples were clearly well outside the expected
surrogate recoveries, you are instructed to dilute the sample itself, not the extract, but
dilute the sample itself, re-spike it, and go back and do another analysis.
Because of some matrix effects, we noticed. a relatively high number of
samples that needed to be diluted, and I think that ties into our inability to get down to
-------
466
those levels. If you dilute the sample itself to achieve acceptable surrogate recoveries,
you have a corresponding loss of sensitivity for the analytes of interest.
We did not spend a lot of time in this study going back and saying "well,
what can we do." The traditional cleanup techniques were certainly applied to these
samples, but we didn't go back and go an extra mile on any of these in order to achieve
the lowest possible detection limits.
That is certainly an aspect in the work that Bill is talking about in the next
two years or so, going back and looking at that more carefully.
We think that there is a significant problem with the lack of specific
requirements for the immunoassay kits. That is not to say that you don't get good
reliable results, but you are working with a system that has a lot fewer constraints on
what someone was going to accept as data.
In some cases, if you get a number, that is probably, the end of the quality
control. We got an answer; we wrote it down.
We looked at what we needed to do to add additional quality assurance to
those immunoassay techniques. We added the requirement that the laboratory perform
what we call the initial precision and recovery tests, blank spikes, essentially, at a
specified level. Those results certainly looked reasonable, so we went on from there.
The dilution issue is essentially, you have 14 percent recovery of DEC or
some surrogate, and the method says it has got to be above 37 percent (or whatever
happens to be in the method.) So, when you get in that situation, you have got to go
back and run a diluted analysis.
Part of the problem is that most laboratories tend to run order of
magnitude serial dilutions, when, in some cases, you might have gotten away with a 1:2
dilution instead of a 1:10. Therefore, the loss of sensitivity can be much more profound
than you would expect simply based on the matrix.
MS. CRANE: I don't have a question. I have just a comment.
MR. MCCARTY: Would you identify yourself, please?
MS. CRANE: My name is Laura Crane. I am with J.T. Baker.
Just a general comment and a word of caution to those of you who are
-------
467
presenting these studies. This is a very new technology in this particular field, and I
think it is extremely valuable to get specific experimental studies reported.
I think it is very dangerous to draw sweeping generalizations about the
capabilities or limitations of the technology based upon the limited experimental data
and studies that are presented so far. I think the limitations will become apparent as the
technology matures, but there are very large differences in assay format, in assay
approach, in antibody specificity among the kits and assays that have been developed so
far and will be developed in the future.
So, just a word of caution about making very broad generalizations about
the technique. I think it is just too premature to do that.
MR. TELLIARD: Thank you. Anybody else?
MR. MCCARTY: We had also discussed looking at other formats and
vendors, but given the time frame, that wasn't a possibility. That is also down the line.
Thank you.
MR. TELLIARD: Thanks, Harry.
-------
468
[Blank Page]
-------
469
Comparison of Immunoassay and
Traditional Gas Chromatographic Methods
for the Determination of Selected
Organochlorine Pesticides and Herbicides in
Wastewater Samples
Cynthia A. Simbanin and Harry B. McCarty
Viar & Company
Introduction
Under the authority of the Clean Water Act, EPA
has a responsibility to develop Effluent
Guidelines for US industries discharging
wastewater in US surface water
EPA is currently developing or revising
guidelines for a variety of industries, including:
- Pesticide Manufacturers
- Pesticide Formulators and Packagers
-------
470
Data collection activities associated
with these guidelines involve:
• Collection of a large number of samples from
a number of facilities
• Analysis for a large number of compounds
• Sample types that range from final effluents
to in-process
• Waste streams
Analytical challenges include:
• Need for sensitivity (sub-ppb for pesticides)
• Wide range of expected concentrations
• Great likelihood of matrix effects, especially
for the in-process samples
-------
471
Traditional analytical approach
involves:
• Solvent extraction
• Derivatization where feasible
• Cleanup techniques (column
chromatography, etc.)
• GC analysis with selective detectors
(EC, ELCD, NPD)
Traditional problems:
• Costly
• Time-consuming
• Complicated by wide range of
concentrations found
- Dilutions
- Instrument and sample contamination
- Down-time
-------
472
Potential advantages of immunoassay
techniques:
• Quick
• Can be highly selective
• Simple to use
• Portable
Purpose of Study
To provide a rapid, cost-effective comparison of
the capabilities of immunoassay techniques to
supplement traditional techniques for the analysis
of effluent matrices
-------
473
Study Design
Piggy-backed on existing plans for sampling and
analysis of pesticide manufacturing and pesticide
formulating and packaging industries
Involved use of traditional GC methods plus the
'lube" and "plate" immunoassay materials
commercially available from Millipore
- Tube kits were used as a semi-quantitative
screening tool
- Tube kits were run in duplicate
Study Design (cont'd)
Tube kit results were used to determine most
appropriate dilution of sample to use for plate kit
analyses
Plate kits were run in triplicate once the dilution level
was established
GC methods were performed as a single analysis of
each sample, with subsequent dilution of samples
and/or extracts as needed to achieve traditional results
Confirmation of GC results utilizing a second GC
column
-------
474
Compounds Corresponding
for GC Analyses Immunoassav Kits
Metolachlor Alachlors
Atrazine Triazines
Endosulfan I and II Cyclodienes
2,4-D 2,4-D
Sampling
Two industrial facilities were sampled in December
1991 and February 1992. Based on knowledge of
products being manufactured or packaged at that
time, the samples were analyzed for:
2,4-D, Atrazine, and Metolachlor at the first facility in:
4 treated effluents
2 in-process waste waters
1 raw water
-------
475
Sampling (cont'd)
and Endosulfan at the second facility in:
5 untreated waste waters
2 treated effluents
3 final effluents
2,4-D results (in
Sample Type
Treated effluent
Treated effluent
Treated effluent
Treated effluent
Raw water
In-process
In-process
ug/L)
Tube (RPD)
937 (25%)
2471 (18%)
2162(74%)
2038 (42%)
<1.0
33365 (65%)
679808(19%)
Plate (RSD)
23585 (29%)
3297 (23%)
4177(38%)
7992(10%)
0.22(14%)
382105(50%)
864046 (60%)
GC
18000
7400
4200
8800
0.19
380000
1400000
-------
476
Atrazine Results (in ug/L)
Sample Type Tube(RPD) Plate (RSD) GC
Treated effluent 386 (81 %) 154 (7%) < 20
Treated effluent 36(25%) 131(16%) < 20
Treated effluent 71 (25%) 118 (15%) < 20
Treated effluent 72 (13%) 165 (12%) < 20
Raw water 4.2(26%) 10(16%) 1.2
Metolachlor results (in ug/L)
Sample Type Tube (RPD)
Treated effluent
Treated effluent
Treated effluent
Treated effluent
Raw water
0.76 (22%)
0.68 (3%)
0.83 (6%)
0.96 (4%)
<0.1
Plate (RSD)
2.0 (32%)
0.9(17%)
1.3(16%)
1.0(12%)
0.2(10%)
GC
<20
<20
<20
<20
<0.2
-------
477
Endosulfan results (in ug/L)
Sample Type Tube(RPD) Plate (RSD) GC
Untreated effluent 52000000 (4%) 18500000 (34%) 50000
Untreated effluent 872000 (8 %)
Untreated effluent 69000 (4%)
Untreated effluent 67600 (6%)
Untreated effluent 6060 (14%)
221000(49%) 1044
12200(14%) <10
11400(10%) <10
683(45%) 16
Endosulfan results (in ug/L) (cont'd)
Sample Type Tube (RPD) Plate (RSD)
GC
Treated effluent
Treated effluent
Final effluent
Final effluent
Final effluent
410(48%)
732(10%)
4660 (0%)
<10
<10
50 (5%)
130(30%)
920 (50%)
<5
12(15%)
5
9
1.4
<0.2
<0.1
-------
478
Discussion
All three techniques are comparable for the majority of
low level (<10 ppb) samples studied here (final
effluents and raw water). This is consistent with the
results shown by other investigators for drinking water
samples, agricultural run-off samples, etc.
The tube kits are semi-quantitative, but still provided
good agreement with the plate kits and the GC
analyses for some samples. Similarly, the tube kits
gave good precision for some analytes and some
samples.
Discussion (cont'd)
For other samples and analytes, there is poor
agreement between the tube and plate kit results.
Excellent agreement of the plate kit and GC results for
2,4-D
Differences between the three techniques are
generally worst at the highest concentration levels (>
1000 ppb), where dilution of the sample is necessary
for all three techniques.
Dilution factors ranged as high as 500,000 for some
samples using the immunoassay kits, but were
typically 100 to 1000.
-------
479
Discussion (cont'd)
Agreement was worst for the Atrazine and Endosulfan
results, where the immunoassay kits are not
compound-specific.
The tube and plate kits generally had higher results for
Atrazine and Endosulfan, relative to the GC results,
and these higher numbers may be the result of "false
positives".
However, given the lack of specificity of these test kits
for the single GC analyte, the higher results may be
due to the presence of other triazines or cyclodienes
that were not targeted by the GC method.
Discussion (cont'd)
Alternatively, the GC results may be false negatives
due to severe matrix effects for the in-process
samples.
The Metolachlor results represent the other end of the
spectrum
The levels in these samples were generally too low for
quantitation by the GC method.
The immunoassay kits gave results for these samples
that were consistent with a factor of 3 or better.
-------
480
Conclusions
Immunoassay techniques work reasonably well
on some of the industrial effluents studied here
Immunoassay techniques performed best for low
level samples
Better agreement with GC results for test kits that
are compound-specific than for kits that test for a
class of related compounds
Immune-assay kits are potentially very useful for
identifying those samples with high
concentrations of analytes or interferences
Conclusions (cont'd)
Tube kits are probably reasonable to be used by
sampling crews at an industrial facility for
screening of samples and refinement of sampling
plans
Plate kits would be readily used by a fixed
laboratory to screen samples prior to extraction
and analysis, thereby protecting the
instrumentation and the laboratory environment
from contamination by particularly dirty samples.
They would allow laboratory to segregate highly
contaminated samples from low level samples
-------
481
Conclusions (cont'd)
Given the need to analyze a variety of in-process
waste streams as well as final effluents, kits are
not likely to replace traditional GC methods for
regulatory guideline development work
Kits may have utility for compliance monitoring
purposes if they are compound-specific, and the
monitoring involves a relatively small number of
analytes
Acknowledgements
Don McCarthy, Millipore
Barbara Young, Millipore
Steve Workman, Analytical Technologies Inc.
-------
482
[Blank Page]
-------
483
ANALYSIS OF ALACHLOR-ETHANE SULFONIC ACID IN WELL WATER
Robert O. Harrison, Carol A. Macomber, and Bruce S. Ferguson
ImmunoSystems, Inc., Scarborough, Maine 04074
Paper Presented at
15th Annual EPA Conference on Analysis of Pollutants in the Environment
May 1992, Norfolk VA
-------
484
ANALYSIS OF ALACHLOR-ETHANE SULFONIC ACID IN WELL WATER
Robert O. Harrison, Carol A. Macomber, and Bruce S. Ferguson
ImmunoSystems, Inc., Scarborough, Maine 04074
Abstract
Screening of ground water samples by Enzyme ImmunoAssay (EIA) for a panel of common
herbicides led to the identification of a major metabolite of alachlor, 2-f(2,6-
diethylphenyl)(methoxymethyl)aminol-2-oxoethanesulfonic acid (alachlor-ESA), in several wells.
Some water samples giving consistent positive results by alachlor EIA were consistently negative
when analyzed for alachlor by GC and HPLC. These samples were further analyzed using a high-
performance liquid chromatography (HPLC) method developed specifically for the analysis of
alachlor-ESA. Quantitative correlation of alachlor-ESA concentrations determined by HPLC and
EIA was demonstrated by using an alachlor-ESA standard for calibration of the EIA.
Immunoreactivity of the HPLC purified material was confirmed by EIA analysis of HPLC fractions,
indicating the presence of an alachlor-like structure. Absolute structural confirmation of the EIA
positive HPLC peak was obtained by HPLC/MS/MS. The above work is briefly summarized in the
following paper; a more extensive account is in press in J. Agric. Food Chem. (Macomber et al. in
tvroccA "'
press).
Introduction
In recent years Enzyme ImmunoAssay (EIA) has become a valuable tool for screening of
environmental water samples. Commercial kits for a variety of pesticides have become available
(Flecker and Cook, 1990; Van Emon and Lopez-Avila, 1992) and numerous groups have developed
cost-effective screening programs using such kits as the first component of a comprehensive testing
system (Goolsby et al., 1991; LeMasters et al., 1989; Thurman et al., 1990; Thurman et al 1991)
As numerous reviews have emphasized (Harrison et al., 1988; Jung et al., 1989; Hammock et al.,
1990; Van Emon and Lopez-Avila, 1992), there are important differences between immunochemicai
methods and chrpmatographic methods which support their complementary use. One of the most
critical and possibly the most undervalued of these differences is the potential of EIA methods for
class-wise specificity, including metabolite recognition (Thurman et al., 1991). The design of QA
procedures, primarily in the form of confirmatory analysis by approved methods, must take into
account the class-wise specificity and metabolite recognition of the EIA's used. Specific
crossreacting substances, such as metabolites or related parent compounds, behave differently in a
panel of EIA's than non-specific interferences, such as pH or humic materials. It is important for the
EIA user to understand these differences and react appropriately. The study summarized here
illustrates this point and confirms the effectiveness of EIA for screening of ground water samples.
Materials and Methods
Methods for sampling, HPLC-UV, and HPLC/MS/MS have been described (Macomber et al in
press). Kits for EIA were obtained from Millipore (manufactured by ImmunoSystems). All EIA
testing was performed according to the kit inserts.
-------
485
Results and Discussion
Heidelberg College of Tiffin, Ohio, has for the past few years offered a well-water testing service
using EIA screening for a panel of common pesticides, followed by GC confirmation as necessary.
During 1991, it became apparent that the Alachlor EIA Kit was demonstrating an unusually high
positive rate in comparison to the positive rate of the Triazine EIA Kit, prior use of the same Alachlor
EIA Kit, and expected alachlor concentrations based on past use patterns and survey results. For
more than 4400 samples tested by the Alachlor EIA Kit during 1991,22 (0.5%) were over 2.0 ppb,
145 (3.3%) were over 0.2 ppb, and 236 (5.3%) were over 0.2 ppb. In a selected subset of over
3400 of these samples which were also tested with the Triazine EIA Kit, the frequency of high and
mid-range positive samples for the Alachlor EIA Kit was about sixfold higher than the rate for the
Triazine EIA Kit (Table 1).
Table 1. Frequency of Positive Samples (total n > 3400) for Two EIA Kits.
Kit middle range (ppb) high range (ppb)
Alachlor 3.2% (0.2-1.0) 0.6% (>2.0)
Triazine 0.6% (0.3-1.0) 0.1% (>3.0)
This pattern varies significantly from the expected pattern because of the lesser use and lower
environmental persistence of alachlor. Selected samples from this set were tested for alachlor using
an approved GC method and shown to be negative. Many of these samples were retested by both
GC and EIA, confirming prior results. Because of the geographic clustering of some of these
apparent false positive samples, a metabolite of alachlor was suspected to be responsible, rather than
a non-specific interference. Matrix effects did not appear to be a plausible explanation because of the
lower rates for the Triazine EIA Kit as shown in Table 1. Nonspecific matrix effects would be
expected to affect the two kits roughly equally and to be more geographically homogeneous.
Because the false positive results appeared to be kit specific, a sample of the alachlor metabolite 2-
[(2,6-diethylphenyl)(methoxymethyl)aminol-2-oxoethanesulfonic acid (alachlor-ESA) was obtained
and an HPLC method was developed for this compound (Macomber et al., in press). Seventeen
false positive samples were selected for more extensive follow-up. These were analyzed by HPLC
for alachlor and shown to be negative. The same 17 samples were then analyzed by three labs using
the newly developed HPLC method for alachlor-ESA. The samples were found to have a peak
matching this metabolite in retention time and UV spectrum. Quantitation for these samples was
performed by comparison to the alachlor-ESA standard. Aliquots of these 17 samples were analyzed
again with the Alachlor EIA Kit, but using the alachlor-ESA standard, rather than alachlor, to create a
standard curve for quantitation. Partial results of both EIA and HPLC for these 17 samples are
shown in Figure 1. The regression equation for the ISI EIA data against the U of Maine HPLC data
(both from Figure 1) was y = 0.8x + 8 ppb, with a correlation coefficient of 0.94.
For verification that the Alachlor EIA Kit was detecting the alachlor-ESA in water samples, the
metabolite peaks from 6 of these 17 samples were collected and analyzed by the kit.
Immunoreactivity of the captured peaks was confirmed and approximate quantitation correlated with
both HPLC results and Alachlor EIA Kit results from direct analysis of the original samples.
Absolute confirmation of the identity of the metabolite was obtained for six samples by
HPLC/MS/MS. The mass spectra of all of the immunoreactive peaks closely matched the spectrum
of the alachlor-ESA standard. All spectra demonstrated a molecular ion at 314 and major fragments
at 79, 120, and 160 m/z. These and other data supporting the identification of alachlor-ESA in the
samples tested are summarized in Table 2.
-------
486
Table 2. Summary of Data Supporting the Presence of Alachlor-ESA in Well Water
1. HPLC peaks of standard and samples match in retention time and UV spectrum
2. Correlation of HPLC'and HA using water samples with an alachJor-ESA standard
3. Correlation of HPLC and HA using captured HPLC peaks with an alachlor-ESA standard
4. HPLC/MS/MS results
a. correlation of concentrations with HPLC-UV
b. molecular ion at 314, fragments at 79, 120, and 160
c. match to alachlor-ESA standard
Conclusions
This study illustrates the importance of understanding immunoassay crossreactivity in general
principle and for the specific test being used. The Alachlor EIA Kit described in this study
performed well, as validated by the extensive follow-up work described. The metabolite alachlor-
ESA is easily detected by this kit, at levels near the detection limit of the kit for alachlor. These
results indicate that the presence of alachlor-ESA should be suspected in areas of high alachlor use
and that EIA results should be treated accordingly. Thus confirmation by GC or other approved
methodology is essential for any effective monitoring program for alachlor or any other pesticide.
All HA methods have the potential for crossreactivity to a wide range of metabolites and other
compounds related to their targets. Not all HA kits are equal in their ability to detect metabolites or
related compounds, not will all of these potential crossreactants necessarily be detected. Each kit
should be tested individually, preferably by the manufacturer in the kit validation process. Ideally
this testing should include metabolites identified during the registration process, many of which may
be unavailable except through the registrant. Thus the cooperation of pesticide manufacturers in
evaluation of immunoassay crossreactivity is critical to proper use of EIA kits. The ideal situation is
to anticipate potential problems due to metabolite crossreactivity and provide the maximum data with
the kit.
As for the case described above, it is critical to the successful use of ajl HA kits that crossreactivity
considerations be understood by the user and that proper confirmatory analyses be done to follow up
positive results. Regardless of the methods used, the best results come from thorough knowledge of
the capabilities and limitations of all methods, whether immunochemical or chromatographic. This
study shows that when positive EIA results are confirmed by approved methods, HA can be a
valuable tool for water quality monitoring. Even in situations where only metabolites with unknown
health effects or with no legal limits are found, HA still may be an important tool for assessing well
or aquifer vulnerability.
Acknowledgement
This paper is a summary of work done by Macomber et al., as noted in the abstract and below.
Thanks to Dr. Luc Mattasa of Mann Laboratories, Mississauga, Ontario, for HPLC/MS/MS analysis
of the water samples and to Karen Larkin of ImmunoSystems for much of the preliminary
immunoassay data.
-------
487
References
Flecker, J.R.; Cook, L.W. "Reliability of Commercial Enzyme Immunoassay in Detection of
Atrazine in Water". Chapter 7 in Immunoassays for Trace Chemical Analysis, Monitoring
Toxic Chemicals in Humans, Food, and the Environment; ACS Symposium Series Vol. 451';
Vanderlaan, M., Stanker, L.H., Watkins, B.E and Roberts, D.W., Eds., American
Chemical Society: Washington, DC., 1990.
Goolsby, D.A.; Thurman, E.M.; Clark, M.L.; Pomes, M.L. "Immunoassay as a Screening Tool
for Triazine Herbicides in Streams". Chapter 8 in Immunoassays for Trace Chemical
Analysis, Monitoring Toxic Chemicals in Humans, Food, and the Environment; ACS
Symposium Series Vol. 451\ Vanderlaan, M., Stanker, L.H., Watkins, B.E. and Roberts,
D.W., Eds., American Chemical Society: Washington, DC., 1990.
Hammock, B.D.; Gee, S.J.; Harrison, R.O.; Jung, J.; Goodrow, M.H.; Li, Q.X.; Lucas, A.D.;
Szekacs, A.; Sundaram, K.M.S. "Immunochemical Technology in Environmental
Analysis, Addressing Critical Problems" Chapter 11 in Immunochemical Methods for
Environmental Analysis; ACS Symposium Series Vol. 442; Van Emon, J.M. and Mumma,
R.O., Eds., American Chemical Society: Washington, DC, 1990.
Harrison, R.O.; Gee, S.J.; Hammock, B.D. "Immunochemical Methods of Pesticide Residue
Analysis" Chapter 24 in Biotechnology in Crop Protection: ACS Symposium Series Vol. 379;
Hedin, P.A., Menn, J.J., Hollingworth, R.M., eds., American Chemical Society:
Washington, DC, 1988.
Jung, F.; Gee, S.J.;, Harrison, R.O.; Goodrpw, M.H.; Karu, A.E.; Braun, A.L.; Li, Q.X.;
Hammock, B.D. "Use of Immunochemical Techniques for the Analysis of Pesticides".
Pesticide Science 1989, 26, 303-317.
LeMasters, G.; Doyle, DJ. "Grade A Dairy Farm Well Water Quality Survey". Wisconsin Dept.
Agric. and Wisconsin Agric. Stat. Serv., 1989.
Macomber, C.; Bushway, R.J.; Perkins, L.B.; Baker, D.; Fan, T.S.; Ferguson, B.S.
"Determination of the Ethane Sulfonate Metabolite of Alachlor in Water by High-Performance
Liquid Chromatography". J. Agric. Food Chem., in >press.
Thurman, E.M.; Goolsby, D.A.; Meyer, M.T.; Kolpin, D.W. "Herbicides in Surface Waters of the
Midwestern United States: The Effect of Spring Rush" Environ. Sci. Technol 1991, 25,
1794-1796.
Thurman, E.M.; Meyer, M.T.; Pomes, M.L.; Perry, C.A.; Schwab, A.P. "Enzyme-Linked
Immunosorbent Assay Compared with Gas Chromatography for the Determination of Triazine
Herbicides in Water" Anal. Chem. 1990, 62, 2043-2048.
Van Emon, J.M.; Lopez-Avila, V. "Immunochemical Methods for Environmental Analysis" Anal.
Chem. 1992, 64, 79A-88A.
-------
488
V.
'35
• 1
"5
CO
o>
DC
^
CO
111
"5
c
O
.J2
CO
Q
E
o
O
CM
"S
o
— J
CO
>
<
UJ
>
(0
"5
E
CO
CO
Ni
CO
COIO
COCMr-
qdd
-------
489
MR. TELLIARD: Our next speaker this morning is Ken Robillard from
Kodak and a little change of pace. He is going to present data on the analysis of silver,
Ken?
MR. ROBILLARD: Thanks, Bill. The subject of my talk is the analysis of
bioavailable metal ions in waste waters. This truly is a multi-disciplinary area of
research, utilizing knowledge of analytical, inorganic and aquatic chemistry and
environmental toxicology. I will focus particularly on the speciation of silver and the
measurement of free silver ion. Developing an analytical method that can selectively
analyze free silver ion at concentrations below a part per billion has been a formidable
challenge. We have been working on this project for approximately twelve years.
Several analytical approaches have been tried; but very few have shown the requisite
sensitivity and selectivity. Often it seemed that trying to find the proverbial needle in a
very large haystack would have been much simpler.
Our story isn't without chapters of success, and it is those successes that I
wish to share with your this morning. I will begin by providing the credits, which I do to
emphasize where credit is due. A tremendous amount of creativity and hard work went
into the efforts which I am going to describe. Some of this work was done at Eastman
Kodak Company in Rochester, New York, by Mr. James Chudd and
Drs. Deniz Schildkraut and Edwin Garcia. And, development work was done at the New
Mexico State University by Dr. Joseph Wang and his colleague, Dr. Ruiliang Li. What I
will do now is describe to you the Why, the What, and the How of this project.
Why did we get involved in this project? (FIGURE 1) Because, for
metals (M) like silver there is usually a relationship between their chemical speciation
and their environmental toxicity. The water quality criteria data that exist for silver were
derived from laboratory tests with silver nitrate, a soluble form of silver that provides
exposure to the free silver ion. However, silver as well as many other metals and some
organics exist in a variety of forms in water (Ma, Mb,...). Silver does not exist just as the
free metal ion, but in complexed and adsorbed forms. There have been a number of
efforts over the years to describe the speciation of metals using both analytical
-------
490
procedures as well as mathematical modeling.
In terms of toxicity, if each species of the metal (e.g. free ion, absorbed,
complexed, etc.) has its own unique toxicity (represented as Ta-Ca, Tb-Cb,...),then the
relationship between speciation and toxicity can be conceptualized as the summation of
individual toxicities. (FIGURE 1) This is a very simple description which assumes
additive toxicities for all the metal species. In fact, the relationships can become very
complicated both conceptually and mathematically if synergistic, antagonistic or non-
additive behavior occurs.
Mother nature may have anticipated the limits on our ability to deal with
these multiple and interactive variables. She made things somewhat easy for us by often
associating a higher level of toxicity to the free ionic form of metals as opposed to their
complexed or absorbed forms. (FIGURE 2) This isn't true in all cases. But, it certainly
is true in a large number of them. Available data indicate that it is particularly true for
silver.
Free silver ion is extremely reactive. There is ample evidence in the
literature to show that when placed into an aqueous solution that contains organics,
inorganic ions, sediment and particulates, ionic silver will rapidly associate with these
other chemicals and forms a variety of compounds. (FIGURE 3) The toxicities of these
silver compounds have been studied to a limited extent. The most extensively studied
silver species, other than free silver ion, are silver thiosulfate, silver chloride and silver
sulfide. Silver sulfide is the most commonly occurring form of silver in the environment.
Silver chloride is a prevalent species of silver in marine waters. And, silver thiosulfate is
the most prevalent form of silver in photoprocessing effluents. FIGURE 4 presents the
results of acute and embryo-larval aquatic effects tests in which fish were exposed to
these different species of silver. The differences in acute and chronic toxicities amongst
these species of silver varied from two to six orders of magnitude. In fact, no toxicity
was ever evident for silver sulfide. Referring back to the relationship shown in FIGURE
1, we should be able to ignore any of the toxicity terms beyond that of the free silver ion.
What we have tried to accomplish with this project is to develop one or
more analytical methods that respond selectively to the principal toxic form of the
-------
491
analyte, e.g. free silver ion. FIGURE 5 shows algebraically that the response should be
proportional to the concentration of the free metal ion. We have examined a number of
approaches, and subsequently focused on electrochemistry. Two electrochemical
techniques constitute the How of our story: potentiometry, using an ion selective
electrode; and, anodic stripping voltammetry (FIGURE 6)
The potentiometric methods is based on the Nernst relationship that the
potential of the electrochemical cell is proportional to the log of the activity of the
analyte. (FIGURE 7) The activity of the analyte is equal to the activity coefficient
times the concentration. Using the Nernst relationship and solutions containing known
amounts of free silver ion, it is possible to construct a calibration curve. (FIGURE 8)
The calibration procedure involves the use of both silver nitrate and silver nitrate with
various complexing agents. By adjusting the concentration of silver nitrate and of the
complexing agents, one can alter the activity of the free silver ion in the solution. The
activity of the free silver ion in each solution can be calculated using known equilibrium
constants and a knowledge of the amount of each added reagent. By analyzing each of
the standard solutions, we construct a calibration curve that is linear and quite
reproducible. Following calibration of the electrodes, the activity of free silver ion in the
test sample(s) can be measured. The electrodes are placed in the sample and the cell is
allowed to come to equilibrium. The measured potential is converted to a pAg value
using the calibration curve. Generally, we assume the activity coefficient in the sample is
unity, and therefore the numerical value of the activity is equal to the numerical value of
the concentration. This is a good approximation for a monovalent ion such as free silver
ion, especially if the sample does not have a high salt loading.
We have been using this potentiometric method for about 10 years at our
facilities in Rochester as well as at some selected wastewater treatment facilities in other
locations. FIGURES 9 and 10 contain data from a field study that was performed using
this analytical technique. We selected six wastewater treatment plants, dividing them
into three categories. Two plants received known quantities of photographic effluent;
two plants received effluent which contained some non-photographic industrial silver;
and two treatment plants received some silver from non-industrial non-photographic
-------
492
sources. In all cases there was some measurable amounts of total silver in the influents
and effluents of the treatment plants. When we used the potentiometric technique to
measure the free silver ion, we saw extremely small amounts. These results were
consistent with previous knowledge on the fate of silver during biological treatment.
Most of the silver is accumulated in the sludge. Any silver remaining in the aqueous
phase is believed to be present in complexed or colloidal forms. In this study and in
other work, the potentiometric method for measuring free silver ion has been useful and
informative.
I should mention that the ion selective electrode we used initially was a
silver wire. We are now using a commercial silver sulfide thin layer electrode, which
works very well, and is a bit more durable that the silver wire.
The potentiometric analytical method, like all analytical methods, has its
limitations. The diagram in FIGURE 11 is an attempt to graphically portray the working
boundaries of the method. On the right-hand side of the S-shaped curve (Region A),
you see a single flat line. This represents the situation where the concentration of free
silver ion is large enough to maintain a steady state (constant) potential across the
electrodes. This condition exists when the concentration of free silver ion is greater that
50 parts per billion. On the left-hand side (Region B), there is a series of flat lines,
which represents the presence of ligands and ions which complex strongly with ionic
silver. In the presence of these complexing agents, the activity of silver is buffered.
When this buffering capacity is sufficiently large, one sees a constant and reproducible
potential across the electrodes. This is exactly analogous to the conditions present in the
standard silver chloride, silver bromide, silver iodide and silver thiosulfate solutions that
are used to obtain the calibration curve. The series of lines in this region represent
varying activities of free silver ion. Each separate activity is determined by the
concentration and strength of the silver-complexing agents that are present in the
solution.
The difficulty comes when you approach the vertical incline (Region C),
where the concentration of free silver ion is less than 50 parts per billion and where
there is insufficient silver-complexing materials present. Under these conditions, the
-------
493
measurement does not work well. The analogy is trying to measure the pH of distilled
water. It is extremely difficult to do, and the results are very seldom reproducible.
We recognized the limitations of the potentiometric method and set about
to develop a supplemental or alternative method. The selectivity and sensitivity of
anodic stripping voltammetry prompted us to pursue this as an alternative analytical
approach.
FIGURE 12 shows the heart of the anodic stripping instrument, the cell in
which the measurement takes place. Anodic stripping is very similar to the technique of
microextraction, which we heard about yesterday. Only in this case, the driving force for
the collection of the silver is not a difference in fugacity but, rather, an applied potential
across the electrodes. A reducing potential is applied initially which forces the silver (or
whatever metal ion you are analyzing for) to plate out on the tip of the electrode. After
a period of extraction ranging from a few seconds to several minutes, the voltage is
changed to an oxidizing voltage, thereby forcing the metal ions to oxidize and return to
the solution. This oxidation creates a current that is proportional to the amount of metal
on the surface of the electrode. FIGURE 13 shows the process in graphical form. As
we change the voltage from a reducing to an oxidizing voltage there is some point where
the voltage will be appropriate to cause oxidation.
One of the first field trials of this method took place over a year ago, in
collaboration with the Department of Natural Resources, State of Michigan. They
provided us with samples from several wastewater treatment plants. These samples were
analyzed for free silver ion using the anodic stripping technique and for total silver using
a conventional atomic absorption method. The results are presented in FIGURE 14. In
most cases, we found the total silver was less and 0.5 parts per billion. I believe this was
pleasing to the operators of the treatment plants as well as to the Michigan Department
of Natural Resources. The free silver ion measurements by anodic stripping voltammetry
were performed in Dr. Joseph Wang's laboratory at New Mexico State University. They
reported an analytical detection limit of 50 parts per trillion.
After the initial analyses for free silver ion, the samples were spiked with
silver nitrate. The amount of silver nitrate added should have given a peak of about 30
-------
494
millimeters in height if the silver persisted as free silver ion. As expected, the samples
that were not preserved with nitric acid maintained a high buffering capacity for silver.
When spiked, the free silver ion was quickly complexed. Whereas, those samples
preserved with nitric acid, showed a much lower complexing capacity for ionic silver.
Many of the ions and atoms that bind strongly to silver, such as nitrogen, sulfur and
oxygen, become protonated in acidic solution and are no longer active (or adsorptive) for
the cationic free silver ion.
The most recent results I wish to share with you were obtained during the
past few months in our laboratory in Rochester. The anodic stripping voltammetry
method developed by Dr. Wang required trace amounts of mercuric ion. We found that
the reproducibility of the analysis could improve by eliminating mercury. FIGURE 15
shows examples of the traces that one would typically see when analyzing a series of
standards ranging from 0.3 to 3 nanograms per milliliter (parts per billion) of free silver
ion. An important observation we made, which confirmed similar observation reported
by others, was the plated silver was capable of being stripped from the electrode as a
series of peaks Sometimes we saw just one peak, sometimes two, and occasionally we
have even seen three peaks. Eliminating the mercury helped us to overcome some of
the early reproducibility problems. FIGURE 16 shows the results of analyzing two
effluent samples containing free silver ion. The curves on the left are for the effluent
samples. The curves on the right were obtained from calibration standards.
FIGURE 17 shows a typical calibration curve obtained using the anodic
stripping technique. The calibration is linear between 0.2 and 1 part per billion with a
good correlation coefficient.
FIGURE 18 provides information on the precision of the anodic stripping
method. These results were obtained by pooling the data from several of the calibration
curves. One sees a standard deviation of about 2 percent at the 1 part per billion level.
The standard deviation is only 12 to 17 percent at concentrations as low as 0.2 parts per
billion.
FIGURE 19 lists results for six separate effluent samples analyzed by
anodic stripping voltammetry and by the potentiometric method. In each sample, anodic
-------
495
stripping showed that the concentration of free ionic silver was less and 0.2 parts per
billion. The two analytical techniques gave consistent results for all six samples.
Because of the high silver-complexing strength in these effluent samples, the
potentiometric method was able to give more precise values for the concentration of free
silver ion. We spiked one of the samples with 10 part per billion of silver nitrate. After
spiking, the anodic stripping analysis showed the concentration of free silver ion
remained less than 0.2 parts per billion. As expected, the complexing strength of the
effluent was sufficient to essentially adsorb all of the added silver ion.
FIGURE 20 shows the results of spiking an effluent sample with varying
amounts of silver nitrate. The bottom curve (#1) is the sample before any silver is
added. We began adding silver nitrate, starting with a nominal concentration of 5 parts
per billion, and continuing up to a nominal concentration of almost 500 parts per billion.
For comparison, curve #6, shows the response for a standard solution containing 0.2 part
per billion concentration of free silver ion. Even after adding almost 500 parts per
billion of silver nitrate to this effluent sample, the concentration of free silver ion barely
exceeded 0.2 parts per billion. These are the type of results we would expect based upon
biological effects testing of effluent samples spiked with silver nitrate. Some effluent
samples can be spiked with considerable amounts of silver nitrate without any resulting
toxicity. This indicates the added silver was converted from free silver ion to other
relatively non-toxic forms of silver.
Presently, we continue to work on improving the anodic stripping method.
We would like to make it easier to use and more reproducible at sub-parts per billion
concentrations. And, we would like to encourage other laboratories to become familiar
with this analytical procedure.
Future work will include validating the anodic stripping method by
collecting a variety of effluent samples and analyzing them using the anodic stripping and
potentiometric methods, as well as analyzing them for total silver. Our overarching goal
of this work is to achieve a recognized and approved method for measuring free silver
ion in samples of effluent and receiving water. We believe the appropriate approach to
environmental protection for silver (and for many other metals) is to regulate the specific
-------
496
toxic form(s) of the metal, notably the free metal ion, and to monitor compliance on that
basis. THANK YOU.
REFERENCES
Chudd, J.M. "Measurement of pAg in Natural Water Samples," Environ. Toxicol. Chem.,
2:315-323 (1983).
Cooley, A.C., T.J. Dagon, P.W. Jenkins and K.A. Robillard. "Silver and the Environment," 7
Imaging Technology, 14(6);183-188 (1988).
Jenne, E.A., D.C. Girvin, J.W. Ball and J.M. Burchard. "Inorganic Speciation of Silver in
Natural Waters-Fresh to Marine," in Environmental Impact of Artificial Ice Nucleating Agents,
D.A. Klein, Ed., Dowden, Hutchinson and Ross, Inc., Stroudsburg, Pennsylvania, 1978,
Chapter 4, pp.41-61.
LeBlanc, G.A., J.D. Mastone, A.P. Paradice, B.F. Wilson, H.B. Lockhart, Jr. and K.A.
Robillard. "The Influence of Speciation on the Toxicity of Silver to the Fathead Minnow
(Pimephales promelas)", Environ. Toxicol. Chem.y 3:37-46 (1984).
Lytle, P.E. "Fate and Speciation of Silver in Publicly Owned Treatment Works," Environ.
Toxicol. Chem., 3:21-30 (1984).
Robillard, K.A. "Measurement of Silver in Effluents from Wastewater Treatment Plants,"
presented at the 6th International Symposium on Photofinishing Technology, Las Vegas, NV
February 21, 1990.
-------
497
Sikora, F.J. and FJ. Stevenson. "Silver Complexation by Humic Substances: Conditional
Stability Constants and Nature of Reactive Sites," Geoderma, 42:353-363 (1988).
U.S. Environmental Protection Agency (EPA), "Ambient Water Quality Criteria for Silver,"
Office of Water Regulations and Standards, Criteria and Standards Division, U.S.
Environmental Protection Agency, Report No. EPA-440/5-80-071, Washington, DC, October,
1980. PB81-117822, 221pp.
Wang, J., R. Li and H. Huiliang. "Improved Anodic Stripping Voltammetric Measurements of
Silver by Codeposition with Mercury," Electroanalysis, 1:417-421 (1989).
-------
498
[Blank Page]
-------
499
C3 _,
I ^
« I
& 8
3 >
M c
w
.2 .2
^ c o
cd G **^
O « X
O J3 O
PQ U H
0
o +
? U
+ H
01 II
^ K^_,
II
u
o
H
O
E
-------
500
« -a
s «
•o E
•8
X
£
ex.
I
o
•8
cd
H H
A A
A A
S
-a
*->
s
-------
501
cc
c*-<
O
|
O
Cf-i
13
o>
£
e
o
^
^N
>
fi
on
O
O
ffi
-------
502
Ill
o
UJ
go
Oof
-0
o
eg
QC
O
O
CO
O
O)
<
0)
0)
UJ
•a
a>
«*.*
CO
"to
CD
to
O
CO
— d
i
0)
o
-0
-------
503
ctf
T* Ctf
r+
2-3
00
•8
p.
S3
cd o
*C PL
^_^ ^^^
^^^*
'O °
^% ^"^
^ o
S H
o
>^
"^
.9
i
I
u
X
II
CO
o
OH
CO
W
-------
504
o
S3
-------
505
+
.<
W)
o
O
PH
u
-------
506
U
o
en
-------
507
4— >
C
•
a •
£ 8
• g
i^
t
Q.
|
|
CO
CO
o
H
o\
O
HH
fe
-------
508
4-»
c
CD
&
&
£
£
H
2 -
PH C
C §
•FH I-*
^-v£J
^W
§N
-3 a
n «*
(D
>
• FH
00
o
• i-H
c
o
KH
?
E
u
c
£
a
g^
&|
«
a
rf
*
P
•IT
J
,1
1
3
§
5
|
-
5?
^il
1|8
i^
a
q
|
5
|
M
2
|
g
|
d
r>
|
3
i
3
|
•i
I
*
•
3
|
S
1
O
u>
^1
5 8
II
o jj
2
|
^
|
d
4D
-------
509
I
B
FIGURE 11
-------
510
toe)
FIGURE 12
-------
511
u
o
e
o
z
^•^
2
H
O
I
HI
UJ
•o
2.
uf
X
£
uT
O)
<
^
UJ
111
cc
DC
o
g
-------
512
o E
||
o>
o>
^^^fc^
^^^^^^^
to
"1
O>
to
tf
(O h*
CO CM
in in
o o
o o
0 I
is
sr
00
<0
1
o _.
CO
in in
0 0
0 0
I:
is
00
m •
03 Q5
*" CM
1
oo o
CO CM
in in
O 0
d d
• i
9 '
l«
00
CM 0
t-
1
O 00
in in
o o
0 O
9 '
it
ST
00
co ri
i
**
in in
0 O
0 0
CO |
is
sr
00
•
CM
1
CM O
CO CM
in in
0 0
d d
• i
0 I
V
II
sr
00
m co
CM
1
co o
h» 00
in in
d d
CM 1
is
00
(O O
1
CM <°
og
*• o
• 1
9 '
l§
00
i
i
-------
513
G
O
S
U
CO
CO
ctf
o >
W)
(D .»
a ^
n ^
8
(D O
O
ex,
CO
O
• rH
eo
o tn o
cvi ^ T--
c> in
cJ ^
Iiu/Duex
-------
514
CO
Pk
03
CO
pq
o
73
G
\
\
y
*!
.
3223332:3 3"*
PUnQJ0M3Qfl
•
n
S Q.
s E
; CO
?
•? ? 5 •* *
233
i
-------
ro
c:
c.
co
-4->
OJ
en
CD
>
•—I
-r-l
CO
o
c_
CD
03
C.
03
CD
duiV ojoiiii /
-------
516
«*•*
O
3
•o
O
•» ^
s.a
s s
•O
CS
73
C C
cd O
\3 Q
0
O
'5
o
etf 00
o
c
o
U
c5 o
•*H CX,
. ON O en vo
' CS CS r-i
CN en ^t vo oo o
o o o o o
00
O
E
-------
517
Summary of Results Obtained
from Environmental Water
Samples
Sample
#
1A
2A
3A
4A
5A
6A
Silver
ng/ mL
by pAG
0.003
0.0146
0.0084
0.0075
0.0078
< 0.010
Silver
ng/ mL
by
ASV
<0.2
<0.2
<0.2
<0.2
<0.2
<0.2
i
amount
of spiked
Ag,
ng/mL
0
0
1 0
0
0
0
Total Ag
ng/mL
with
spike by
ASV
NA
NA
<0.2
NA
NA
NA
Comments
high
complexing
ability
Note: The historical TOTAL SILVER concentration in type A
samples between 01/03/91 and 12/18/91 was 76 ng / mL.
FIGURE 19
-------
518
Example of a Type A water sample
t-
0
o
I
-0.3 -0.1
0.1
0.3
0.5 0.7
Potential Volts vs. SCE
Type A sample, after addition of: (1) 0 ppb, (2) 5
ppb, (3) 10 ppb, (4) 200 ppb, (5) 465 ppb silver ion,
and (6) 0.2 ppb silver standard.
FIGURE 20
-------
519
MR. TELLIARD: I would like to thank the morning speakers. Continuing
with pollution prevention, let's look at a pollution prevention solvent recovery system, Jim
Stunkel with ABC Laboratories is going to tell us about a recovery system he has used.
MR. STUNKEL: Solvent purification by spinning band distillation with
specific emphasis on purification of methylene chloride is the subject of my discussion today.
I would first like to start out by reviewing some basic vapor-liquid equilibrium concepts as
well as different types of distillation.
Shown at right is a graphical representation of a mixture of two solvents which
I will designate as components A and B. As you know, the composition of the vapor above
the mixture is dependent upon the relative volatilities of the two components.
Shown at upper left is Raoults law which allows us to calculate the vapor
pressure of a component above a mixture. In words, this equation says that the vapor
pressure of component A above the mixture is equal to the mole fraction of component A in
the mixture times the saturation vapor pressure of component A.
If a mixture displays the type of behavior predicted by Raoults law exactly,
then it is considered to be an ideal mixture.
In order to calculate the actual composition of the vapor, we can use the
equation shown at bottom left. This equation states that the mole fraction of component A in
the vapor is equal to the mole fraction of component A in the mixture times the saturation
vapor pressure of component A all divided by the total pressure above the mixture caused by
all components.
The more volatile components, of course, have a greater tendency to be in a
greater abundance in the vapor phase due to their greater tendency to evaporate.
Now, the vapor in equilibrium with this initial mixture has at this point been
enriched in the more volatile component by one evaporation stage. Now, if these vapors are
condensed and the resulting liquid allowed to come to equilibrium with its
-------
520
own vapor, we would find that the vapor has again favored the more volatile component.
By repeating similar condensation and evaporation stages multiple times,
one can eventually obtain a small amount of condensed liquid that will be very rich in
the more volatile component.
Now, if we construct a diagram of vapor composition versus liquid
composition for each of these stages, we obtain a diagram similar to the isobar shown
here for a mixture of benzene and toluene. In this diagram, the top curve represents the
composition of the vapor; the bottom curve is the composition of the liquid.
Each stair step inside of the curve corresponds to one equilibrium stage.
Now, in some distillation columns, each one of these stages is an area in
the column where actual condensation and evaporation can be seen to occur, while in
other columns, each one of these stages is a more theoretical type more closely
resembling those in gas chromatography.
There are two basic types of distillation. The first type, called simple
distillation, involves a single equilibrium stage directly from the solvent flask to the
condenser. The liquid formed during condensation is then directed away from the
system in this type of operation.
The second type of distillation, called countercurrent distillation, on the
other hand, redirects a specific portion of the condensate formed to the column where it
flows back down to the solvent flask. The quantity of material returned to the column,
referred to as reflux, must then be re-evaporated in order to reach the top of the column
again.
Shown here is a rotary evaporator which is an example of simple
distillation. Again, this type of instrument essentially utilizes one evaporation stage
directly from the solvent flask to the condenser, and the condensate is then directed to
the receiving flask.
This type of system is not a very efficient type of a separating mechanism
and so is used primarily for evaporation of bulk solvent after sample extraction steps, for
example.
Here is a photograph of a Kuderna-Danish concentrator equipped with a
-------
521
Snyder column. This type of system represents a rudimentary form of countercurrent
distillation.
This particular example shows a 3 ball column, each ball joint representing
an actual equilibrium stage. Vapors rising up through the column are forced to come
into close contact with the condensate formed at each ball joint, and the more volatile
component is thereby enriched at each stage, and, more importantly, your analytes of
interest are left behind.
In addition, a small amount of condensate formed at each Ball joint had a
tendency to drip back one stage and results in countercurrent type distillation.
This is the system that I have been using to collect data that I am going to
present here today. This is a spinning band column, distillation system. The basic
components of the system from bottom to top are an electrically heated solvent flask, a
spinning band distillation column which is the silver portion of the system.
There is a column head and condenser, and then up at the top, there is a
motor which spins the spinning band which you will be able to see more clearly in the
next slide, and then over on the right, there is a small box which is a microprocessor
controller used to control the operating conditions of the system.
This is a closeup of the column head. I don't know whether you can see it
very well from where you are at or not, but there is a reflux valve on the top surface of
the silver portion of the column which is used to control the amount of reflux returned
to the column.
Now, unlike the Snyder column, this column does not depend on distinct
physical separating or equilibrium stages to achieve separation. Instead, this system uses
a helically-shaped teflon band through the length of the column which is spun at about
2200 rpm during system operation.
Several things occur due to the spinning action of the spinning band. First
of all, the band generates a lot of mixing throughout the column in the horizontal sense.
This gives fairly rapid vapor-liquid equilibrium of the components in the mixture.
Second, vapors attempting to rise up out of the solvent flask are pumped
downward by the spinning band which gives a longer residence time of the vapors in the
-------
522
column and helps ensure complete equilibration.
Thirdly, there is a tendency for the spinning band to scrape the inner
surface of the column, and this gives fresh surface area for liquid film formation.
Now let's move on to actual requirements for pesticide grade methylene
chloride. This data was taken from the Reagent Chemicals Handbook published by the
American Chemical Society and essentially lists all the parameters that they consider
important in order for methylene chloride to qualify as pesticide grade.
In this study that I am presenting today, we concentrate primarily on total
assay at the top right corner which must be greater than 99.5 percent. This essentially
shows the presence of other solvents in the mixture.
For our studies, we use flame ionization detection for this test.
The other thing that we concentrated on was total chlorinated hydrocarbon
content which is under the GLC interference category at bottom left. This was done by
electron capture detection, and it must be not greater than 10 nanograms per liter for
pesticide grade methylene chloride.
We did a series of experiments, actually. The first of these involved a test
mixture which was composed of methylene chloride and four chlorinated pesticides. The
top chromatogram shows an electron capture analysis of the test mixture diluted 100-fold
and exchanged hexane for analysis.
The first major peak you see is lindane, the second is 4,4-prime DDE, the
third large peak is 4,4-prime DDD. There are also multiple toxaphene components
which are seen as small peaks throughout the chromatogram.
Each one of these chlorinated pesticides was added at approximately 93
million nanograms per liter when quantitated against heptachlor epoxide.
The bottom chromatogram shows a flame ionization detector analysis of
the test mixture, also added some acetone at .6 percent by volume, which is the very
small peak on the trailing edge of the solvent front. The FID analysis was undiluted.
These are the operating parameters that we programmed into the
microprocessor controller to purify this test mixture. We programmed the distillate to be
collected in two fractions.
-------
523
The first fraction was collected when condensate temperature at the head
of the column was between 28 and 39.3 degrees C.
Unwanted low boiling components were to be eliminated during this
fraction by directing the distillate to a waste container. It turns out that these
parameters did not come into effect during this particular experiment, indicating the
absence of any low boiling components and also total suppression of the fortified
acetone.
The second fraction was collected between 39.3 and 39.7 degrees C. This
is the fraction during which we collected purified methylene chloride.
Other important parameters include the mantle rate and the reflux ratio
which I will talk about in more detail later on.
These two chromatograms compare an FID analysis of the purified test
mixture and Burdick & Jackson methylene chloride. Again, this test essentially is
intended to show the presence of any other solvents, and in this case, any FID detectable
compound.
Both of these solvents gave greater than 99.9 percent of total assay.
The top chromatogram is a chromatogram of the purified test mixture, and
the bottom chromatogram a sample of Burdick & Jackson methylene chloride. In both
instances, the samples were concentrated 100-fold using a Kuderna-Danish concentrator
and exchanged to hexane for analysis.
In the top chromatogram, we see some carryover problems from the
chlorinated pesticides. The second major peak in the top chromatogram is lindane
carried over. The third and fourth chromatogram peaks are 4,4-prime DDE and 4,4-
prime DDD respectively.
These three peaks amounted to about 196 nanograms per liter of carryover
contamination.
The bottom chromatogram of Burdick & Jackson methylene chloride had
about 59 nanograms per liter of ECD detectable contamination.
ACS specifications required less than 10 nanograms per liter.
Now, I realize that we are substantially over the ACS requirements, but
-------
524
please keep in mind that we started out with a very concentrated mixture, and we did
manage to bring contamination in the mixture down from 93 million nanograms per liter
each to a total of 196 nanograms per liter.
After this experiment, we designed another series of experiments. We
decided to dilute the test mixture 500-fold because of the carryover contamination
problems that we experienced in the last experiment. Also, we decided that the most
valuable information would be obtained if we optimized mantle heating rate and reflux
ratio.
The first factor, mantle heating rate, is simply the percentage of time that
the heating mantle is receiving power from the controller. This, in turn, controls the
boiling rate of the solvent in the flask and the amount of vapor being loaded onto the
column.
The second factor, reflux ratio, is related to the rate at which reflux and
distillate are being produced. As was stated earlier, reflux is the quantity of the material
returned to the column after condensation. Distillate, on the other hand, is the quantity
of material taken off as final product.
So, reflux ratio is equal to the reflux rate divided by the distillate rate.
Now, if we add reflux rate and distillate rate, we obtain another important
factor called column loading. This is simply the total amount of vapor passing through
the column in unit time.
This table shows four experiments that we did in an attempt to optimize
reflux ratio and mantle rate. By manipulating these two factors to produce opposite
effects, we were able to maintain a fairly constant distillate rate of about 520 ml per
hour for the first three experiment while, at the same time, reducing the amount of
vapor loaded onto the column.
The fourth experiment was done at a slightly higher distillate rate of 685
ml per hour.
And I will just run quickly through these chromatograms. The top one is
the chromatogram for the 35 percent mantle rate and 4:1 reflux ratio. We do have some
carryover problems here although less than what we had from the original undiluted test
-------
525
mixture.
The bottom chromatogram done at 28 percent mantle rate and 3:1 reflux
ratio also has some carryover but in reduced amounts. So, reducing the mantle rate
appears to have helped us.
This is really fairly unusual, since the higher the reflux ratio, generally, the
greater the purity of your final product.
These are the two chromatograms for the last two experiments. The top
one represents results from a 21 percent mantle rate and 2:1 reflux ratio, and here we
have very good results, really not very much carryover contamination at all.
Then, in experiment 4, we were back up to a 28 percent mantle rate, and,
correspondingly, the crossover contamination also increased.
This table summarizes the conditions and results from this series of
experiments. Over in the far right-hand column we have carryover which you can see
steadily decreases as the heating rate goes down.
I would like to call particular attention to experiments 2 and 4. Again,
generally, as the reflux ratio goes up, your purity also goes up. Well, we see here that
even though we lowered the reflux ratio from 3:1 in experiment 2 to 2:1 in experiment 4r
we obtained almost exactly the same results.
So, what this shows us is that at least for chlorinated pesticides, achieving
greatest purity is, by far, dominated by the heating rate of the solvent flask.
In experiment 3, we actually managed to achieve less than 10 nanograms
per liter of contamination.
The last experiment we did was to recover actual used solvent and re-
purify k to pesticide grade. This slide shows a chromatogram of GPC solvent which was-
used in the analysis of pesticides in corn oil. You see there is a fairly substantial
contamination in that.
This chromatogram is an unconcern rated electron capture analysis of that
solvent, It has simply been exchanged to hexane.
And these are the operating parameters. These parameters are identical to
the parameters which gave best results during optimization of reflux ratio and heating
-------
526
mantle rate.
These are the chromatograms of the final, purified GPC solvent. The top
chromatogram shows a 100:1 concentration with exchange to hexane. The bottom slide
shows hexane used for the solvent exchange concentrated 50:1. We concentrated the
hexane 50:1 because that is the quantity of hexane we used for the solvent exchange.
As you can see, almost every peak in the GPC solvent is also present in the
hexane used for solvent exchange. So, we found nothing in the purified solvent at all.
In conclusion, I would first like to say that we were very pleased with the
performance of the system. We had no problems in achieving pesticide grade methylene
chloride from actual recovered used material.
However, for extremely high levels, we may have to resort to double
distillation, meaning that we take the purified product and put it through the same
process again.
There is another possibility. We might try some pre-treatment such as
using activated charcoal or other adsorbents to remove some of the pesticides before
distillation.
We believe that tolerance for chlorinated hydrocarbons in the starting
mixture is around .18ppm fora single distillation operation. Optimum system efficiency
is, without a doubt, achieved at low mantle rates and low reflux ratios, at least for this
type of mixture.
The spinning band system produces high quality solvent from actual used
material, and we were able to produce purified product at at least 520 ml per hour.
And an important lesson that I learned is that no matter how well your
distillation system is working, if you are not extremely careful during the quality control
steps, it doesn't do you any good. It is very easy to accidentally introduce contamination
at the levels that we are working at.
Are there any questions?
-------
527
QUESTION AND ANSWER SESSION
MR. PIWONI: My name is Marv Piwoni. I am with Illinois Hazardous
Waste Research and Information Center.
I don't really have a question. I guess a commendation, first of all, for
your efforts, but I have a sort of a comment of a philosophical nature perhaps.
MR. STUNKEL: All right.
MR. PIWONI: The recovery process requires two components, that is,
gathering of the material that you are using in the lab, and we saw that demonstrated in
the presentation yesterday, and then the trickier part, I think, is the marketplace for
using that solvent.
You demonstrated here, I think, that you can with the spinning band still
which is on the order of $10,000 or $12,000,1 think, from BR Instruments, you can
generate solvent to reuse, but I have more trouble envisioning every lab in the country
doing this than I have them perhaps gathering the solvent in the first place.
I guess the comment maybe to Bill or to any solvent vendors in the
audience, if they are still around, is are there alternatives once you gather that solvent in
your little 4-liter jars to trying to process it in lab yourself?
The solvent manufacturers are equipped to do this, and if we could figure
out some mechanism to return that solvent to them, I think we would all benefit, and we
would benefit from the economies of scale, obviously. The still people don't like that,
but we would benefit, I think.
MR. STUNKEL: Thank you.
MR. HALVORSON: My name is Jeff Halvorson from Burdick & Jackson.
I just thought I would make a comment on the last person's observations.
B&J, I know, and my area...
MR. TELLIARD: Could you get closer to the microphone?
MR. HALVORSON: B&J is the solvent producer that a lot of people use,
and I am not really in that area, so I am not completely familiar with all the processes
involved, but I do know that they are very reluctant to take recycled solvent, because
-------
528
there is no history behind reclaimed solvents in most cases. We don't know what they
were used for, and we don't really know what clients are going to need to use it for once
it is recycled.
So...
MR. STUNKEL: Excuse me. Did you say they were reluctant to take used
solvent?
MR. HALVORSON: That is my understanding right now.
MR. STUNKEL: Okay.
MR. HALVORSON: There are companies that do recycle solvents, though,
not necessarily for laboratory use but for industrial use, so it is not necessarily something
that has to be disposed of.
MR. STUNKEL: Okay.
MR. HALVORSON: I had one question for you as well, though. I
wondered if you had done any calculations on the number of theoretical plates you have
for that distillation apparatus.
MR. STUNKEL: Yes, that is right. I did forget to say that. Under ideal
conditions, this system produces about 50 theoretical plates.
MR. HALVORSON: Thank you.
MR. STANKO: George Stanko from Shell.
Did you by any chance look at any of the other analytes in, say, Method
625 or 8270 with respect to recovering the methylene chloride?
MR. STUNKEL: No, we didn't.
MR. TELLIARD: Anyone else?
(No response.)
MR; TELLIARD: Km, thanks as lot.
-------
529
ctf
O
*H
PH
t3
-------
530
X
o
._* 6
ll
0-4 CO
a "3
* -5
|V| C
E?
^^
^ C/3
O • —
u- £
a^
£ u
r>
• c«
C
^^
, jjj
4—1
^O
"5
(N
OO
a\
o
cd
a
o
U
O
03
0)
*H
CO
CO
hJ O
a,
x
X
C
o
en
O
s
o
U
Ui
O
-------
531
<
e.a
2
w
-------
532
GO
00
a
o
Q
&,
00
X
(D
5-H
c
e
o
o
•o
O
-0
O
U
(D
s
c^
CO
C
•8
C
O
U
CO
C
o
4—>
CTJ
O
cx
aj
(D
a.
cd
O
Ou
cd
-------
533
a
o
s
o
PQ
W)
a
00
O
0)
W
o
w
o
W
X
o
_N
O
ffi
-s
•si
^ *-•
^£
CO)
g
'E.
t-i
o
00
-------
534
03
O
oo
a
o
O
O
(D
oU
a.
CO
CO
U
(N .
O O
co
VI
ON O
Al VI J2 g
ps^IsS
^3 ? •
^ CO ±3 O
H VI
CO
CO
CO
cd
VH
0
CX
cd
>
w
CO
G
(D
bX
^O
ffi
(D
(D
VH
I
£
o
•(—>
o
DH
O
O
-------
535
O
O
"O
-4—>
J3
s
H
<+-
o
CO
*00
Q
U
JI3
»i—<
C
X_x
<
S
^—>
oo
H
o
oo
• <—I
oo
^p
C
Q
h
it •«
till
JIB
ii
81
(£1
K I
I ' ' I ' I I ' i • '
-------
536
VH
O
(D
OH
C/)
• »-H
G
B
x
C/)
H
(N
a
o
• rH
-4-J
O
VH
T-H
a
o
'•£2
O
03
fc
VH
(D
Paramet
u
o
CO
ON
CO
U
o
O
oo
(N
VH
-4— >
ctf
VH
0
OH
s
1)
H
a
D
PH
O
o
o
£^
•
en
U
O
CO
ON
CO
(D
^— »
cd
rH
(D
DH
a
o
H
o
GO
o
o
Q\
__ _J
(N
KO
T-H
(N
£
d)
T— H
fl
cd
T-H
CN
T-H
• •
o
CO
o
• i-H
4->
cd
X
-------
537
X
s
4—>
oo
H
|
*c
(£
"o
oo
• i—H
00
G
Q
>—(
IX,
O'ii
It II
II 12
U'tl
II II
tt'K
CO'It
U'H
»«'[!
Ml
€>'»
so »
ni
K (
m
CC't
te t
•D
."2
"C
^2
IS
u
G
D
G
O
oo
^
O
cd
0
PQ
"o
00
"oo
C
Q
SCR
zo-zz
Zl
ICK
srsi
u»\
(AUJ)
-------
538
O
O
O
C
O
O
-------
539
a 2
S'^
ctf
PH X
N
PH
O
• w
Q
+
4
-------
540
I
D
—>
03
03
"
03
i
C3
5— i
X
0) 03
o O
a
o
l-H
-)->
03
N
a,
O
" T 1
Q
o
T—I
4—>
03
X
-H
+-H
-------
541
CO
55
Q
U
Cs.
O
• i—4
•*-»
ctf
cn
-------
542
00
'oo
*03
C
Q
U
D
CN
C
03
(N
cn
c
03
X
&
ri
C
03
oo
c
I
• •—^
Ui
X
W
i... t
-------
543
O
o o
^
„ r-.-,
Q
^H
C
X n
3 .2
^—>
cd
(D
s
• r-H
5-i
G.
w
(N
H-l
fl
vc
wo
(N
CO
CO
J
"5i
c
ON
-------
00
r— «
00
j>,
13
C
Q
cj
O
c
O
O
G
544
00
U
C^
O
-a
-------
545
.s
td
VH
0)
P
O
go
" 0)
a
o
• i— i
o
£
a
o
•i-H
O
aj
£>
ID
s
cd
tj
ctf
OH
U
o
o
00
CO
U
o
O
O
d
S
-------
546
i... i
C/D
-------
547
^o
*co
^^^B
^^^^
^^^^J
^^^^^
O
U
^H
3
3
o
T3
d)
VH
• I-H
2
cr
^
d
cd
d
• i-H
a
03
-4— >
d
o
o
.2?
:>•>
13
a
VH
"S
w
hydrocarbons in starting mixture
distillation.
•
rj
^H
o
• i-H
-4— >
JS
•i-H
-4— >
CO
. ^^
• ^^
T3
-7-t
orinate<
3
o
l-l
cB
^^™ <
% CO
co O
* *H^
w ^^^
a g
^^^\ ^**^
f^\ ^^Ij
OO
VH
CD
CO
+->
• i-H
13
g-
• i-H
(D
CO
(U
0
O
^
D.
a
3
CO
d
4D
t»o
d
'3
_d
"Q-,
C/D
»— H
.2
*£H
td
a
(D
CO
13
3
o
S3
a
o
^M
VH
O
d
'o
c«
-a
• f-H
0
CO
• i-H
U
o>
VH
o
o
PQ
PL,
-------
548
-------
549
MR. TELLIARD: Our next speaker is Greg O'Neil. He is from Tekmar,
the man who has purged you for years. I am sorry. Going to talk about some headspace
analysis work that they have done as soon as he masters the microphone.
-------
550
[Blank Page]
-------
551
Analysis of Volatile Organic Compounds in Soil
Using Static Headspace Extraction
Greg O'Neil
Tammy Cappel
Paul Kester
Denise Sherman
Tekmar Company
7143 E. Kemper Road
PO Box 459576
Cincinnati, OH 45242-9576
INTRODUCTION
There is a need to develop a better analytical method for the
analysis of volatile organics in soil. This study isolates the
problems associated with soils analysis. These difficulties can be
overcome by using static headspace techniques.
Currently, soils analysis is performed with purge and trap/gas
chromatography. There are fundamental differences between purge
and trap and static headspace. Purge and trap is a continuous gas
extraction where an exhaustive purge of the sample transfers the
analytes from the sample to the trap, which is then desorbed to the
GC column. Quantitation is based on recovery of internal
standards, as is typical within EPA methods as shown in equation 1:
(Cx) = (A,) (C,s) / (A,s) RF Eq. 1
Where Ax = Area of Compound
AIS = Area of Internal Standard
Cx = Concentration of Compound
CIS = Concentration of Internal Standard
Static headspace is a discontinuous gas extraction where the
analytes partition between the sample and the headspace. When the
rate of the analytes leaving the sample is equal to the rate of the
analyte returning to the sample from the headspace, an equilibrium
condition exists. At this time an aliquot of the headspace is
removed and analyzed by GC. It is very important within static
headspace methodologies to heat samples to constant temperatures
and for uniform periods of time. This ensures a fixed volume of
the headspace is injected, resulting in reliable quantitation.
The equation typically used for static headspace quantitation is
shown in equation 2:
Cs = CG [K + (VG / VM)] Eq. 2
Where Cs = the concentration of the analyte in the sample
CG = the concentration of the analyte in the gas phase
-------
552
K = the partition coefficient (CS/CQ
VQ = the volume of the gas phase
VM = the volume of the matrix
There are two important characteristics of this equation. The
first is the partition coefficient (K) , and second is the phase
ratio (VG/VM) . The partition coefficient is the ratio of the
concentration of the analyte in the gas phase (CJ to the
concentration of the analyte in the sample (Cs) phase. Phase ratio
is the ratio of headspace volume to the volume (or matrix) of the
phase in the vial. When K is very small the phase ratio becomes
extremely important. When K is large the phase ratio is relatively
unimportant.
In static headspace, the sample is placed in a vial and sealed with
a teflon-faced septum, and a crimp cap. The vial is placed in the
heated zone of the 7000 and allowed to equilibrate over time. The
vial is slightly overpressurized with pressurization gas, typically
helium, to the vial. The sample loop is then filled by opening a
vent valve downstream of the sample loop. The pressure in the
headspace drives the sample through the loop, filling the loop.
Sample loop pressure is controlled with a Variable Injection
Pressure Regulator (VIPR), set to a value approximately 2psi less
than the vial pressure. This increase in pressure exerts a certain
backpressure in the loop, compressing the sample during loop fill.
The contents of the loop are then swept into the GC injection port.
Figure 1 shows the process of equilibrating a two-phase sample. As
heating time increases, the concentration in the gas phase (XQ)
increases over time until equilibrium is reached, represented by
the flat portion of the curve. This is where the rate of the
analytes leaving the sample equals the rate of the analytes re-
entering the sample, giving maximum sensitivity and precision.
After some point in time a loss of response through degradation
will occur. This degradation is a result of a combination of
factors: thermal degradation, leaks from the septum, and adsorption
of analytes to the septum and walls of the vial.
Table 1 shows a series of partition coefficients for various
analytes. These are literature values showing very low partition
coefficients for alkanes, olefins, and aromatics, and much higher
partition coefficients for polar compounds, such as ketones,
aldehydes, and alcohols. The equation for the partition
coefficient is:
K = (Cs) / (CG) Eq. 3
Where K = the partition coefficient
Cs = the concentration of the sample matrix
CG = the concentration of the gas phase
For sensitivity reasons, a large analyte concentration in the gas
phase is desired. As that concentration increases, the partition
coefficient becomes very small. Therefore, the lower the partition
coefficient, the easier it is to get the analyte into the gas phase
-------
553
and analyze by static headspace.
For polar compounds, such as alcohols, it is a typical analytical
procedure to add salt to change the partition coefficient to a
small value, driving the analyte into the headspace. This
technique is called "salting out". In the homologous series of the
alcohols, as the aliphatic character of the alcohol increases the
partition coefficient decreases.
EXPERIMENTAL
This work was performed with a Tekmar 7000/7050 Headspace
Autosampler equipped with a GC/FID. Experimental parameters appear
as Tables 2 and 3.
The soil samples were weighed into a 22ml vial to which a matrix
modifying solution and internal standards were added.
The matrix modifying solution is an aqueous solution saturated with
sodium chloride adjusted to pH2 with phosphoric acid. It is
important that the salt selected be a non-buffering salt to ensure
the pH remains low. The salt chosen should not be a calcium salt,
as this can complex methanol. This complexation is detrimental if
methanol is an analyte of interest.
The main purpose of the matrix modifying solution is to control
sample to sample matrix variations which may induce quantitation
errors. By forcing the conditions of the sample to become
constant, results from sample to sample become reliable.
RESULTS and DISCUSSIONS
Figure 2a shows equilibration of aromatic compounds in water,
without mixing. A water sample was placed in a vial, heated to
85°C and allowed to equilibrate over time. An excess of one hour
was required for the equilibration to occur. By mixing the sample
with Optimix (Figure 2b), the time required to reach equilibrium
was vastly reduced to less than 10 minutes.
Optimix involves mixing the sample by tapping the vial with a
mixing rod from the outside. The sample tumbles in the vial so
analytes more easily reach the gas/liquid interface, which is
required for an extraction to occur.
Table 4 shows that area counts are increased and RSD values are
improved with mixing. These improved results occur due to
minimized thermal exposure time.
Standardization of the instrument was performed using the full
evaporation technique (FET) to produce a gas phase standard for
which response factors are generated. This was created by
injecting 1-20 microliters of a liquid phase methanolic standard
into a headspace vial, followed by heating the vial to 85°C. On a
daily basis a full evaporation check standard was run to ensure
-------
554
that the instrument was not drifting, and results had to be within
20% of initial values from the beginning of the study in order to
continue operation. The standard used was an AccuStandard 502.2A
standard. The internal standards/surrogates used for quantitation
were dibromofluoromethane, toluene-d5/ and bromofluorobenzene.
A soil sample was placed in a vial, along with a matrix modifying
solution that is equilibrated over time (Fig. 3a). This curve was
completely unlike that observed for an aqueous sample in Fig. 1.
Upon initial heating there was a rapid rise in the concentration of
the headspace, followed by a rapid loss, and then a very unstable
horizontal path along the plot, not really approaching equilibrium.
The total sample equilibration time was 1.5 hours. Mixing the
sample reduced the equilibration time to approximately one hour as
evidenced by a more horizontal path along the plot (Fig. 3b) .
However, there was still a significant difference in the appearance
of the equilibration curve observed in figure 1. The rapid rise
followed by the rapid fall was attributed to the three-phase system
contained in the vial.
Initially the analytes spiked into the soil partitioned into the
liquid phase of the matrix modifying solution, subsequently
analytes partitioned also into the headspace. There was a rapid
loss of the analyte from the spike on the surface of the soil
sample. The soil was broken up with mixing, increasing its surface
area. This increased surface area improved the efficiency of the
absorptive nature of the soil, shifting the partitioning to the
solid phase. This was represented by the rapid drop in response at
about 20 minutes. Eventually the analytes formed an equilibrium
with the three phases as shown in Figure 3b.
Quantitation was performed with the use of multiple internal
standards which tracked the pattern of their associated spiked
analytes very closely. Response factors for the internal standards
were derived using the full evaporation technique. Use of internal
standards in this fashion produced reliable, quantitative, and
reproducible results across the entire range of USEPA Method 502.2
analytes. Further, these internal standards were added prior to
headspace extraction and thus served as true surrogate compounds.
SOIL MATRIX VARIATIONS
The most significant challenge with analyzing soil samples is the
large number of variables involved. Seven have been identified.
Each soil sample can have a wide range of conditions within each of
the variables. Therefore soils can have a large number of
permutations and combinations of characteristics.
The first condition is ionic strength. Salt content of a soil
changes the partition coefficient, primarily of polar compounds.
Therefore as this partition coefficient changes, the concentration
of the analyte in the headspace changes. In order to control this
aspect of the analysis, a saturated salt solution is added to the
sample to control its ionic strength, making it constant for each
-------
555
of the samples to be analyzed.
The second variable is the pH. The pH is adjusted low to prevent
dehydrohalogenation reactions which occur spontaneously at 85°C
within a sample when pH is at 9 or greater. The matrix modifying
solution forces the pH to remain at 2 to prevent
dehydrohalogenation reactions. These reactions are the loss of
hydrochloric acid across a chlorinated alkane to form a halogenated
alkene.
The third variable is biological activity. Biological activity
within the sample can cause degradation of analytes during the
storage time. Samples are required to be maintained at a
temperature of 4°C. The lowering of the pH to 2 acts as a
preservative. Salt content of the matrix modifying solution also
acts to inhibit biological activity by killing bacteria.
The fourth variable is moisture. Samples vary in moisture content
as they come in from the field. The way to compensate for this is
to drive the concentration of water in the sample to remain
constant by saturating it with an aqueous matrix modifying
solution. The sample and headspace go to a consistent
concentration of water due to maintaining a constant temperature
within the sample and the headspace.
The fifth variable is to take the sample to a constant surface
area. Samples have a variety of surface areas, depending on their
composition. Some may be clumped, while others may be very fine
particulate. By mixing, the sample is broken apart, increasing the
surface area. This can be achieved by mixing within the instrument
and/or pre-sonicating the sample. Driving the samples toward
constant surface area reduces the average distance an analyte
travels to be extracted from the sample. This improves extraction
efficiency and reproducibility.
The sixth variable parameter is soil density. The soil density
will cause a vast difference in the phase ratio. A highly dense
material, such as a sandy soil, occupies a very small volume within
the vial, leaving an extremely large headspace. An equal mass of
a low density soil, such as top soil, occupies a large volume
within the vial, with a comparatively small volume of headspace.
If an analyte has a very low partition coefficient, this variation
in phase ratio between soil densities becomes extremely important.
Therefore, different results for non-polar organics in soil samples
can be obtained if phase ratios are not controlled. The addition
of the matrix modifying solution brings , the samples to a more
constant phase ratio. The effect of density of three types of
soils with and without a matrix modifying solution is shown as
Table 5. Without a matrix modifying solution there is an 8-fold
difference in the phase ratio. By adding an equal volume of the
matrix modifying solution to each sample, in this case 10ml, only
a 2-fold change in the phase ratio occurs between high and low
density soil samples.
-------
556
The last variable which affects soil samples results is the organic
content of the sample. Humic materials, such as top soil, tend to
be very absorptive for organic compounds, as opposed to sandy soils
which are inorganic and tend not to absorb. Comparable amounts of
sample containing equal concentrations of analytes would give
significantly different results. The way to control this is to
improve the extraction ability of the matrix modifying solution
with an organic modifier. This increases its solvating strength,
removing the volatile organics from the soil itself. Once the
anlaytes are in the liquid phase they are easily partitioned into
the gas phase.
Figure 4 shows the effects of the soil matrix type on partitioning.
The sandy soil shown at the bottom of the figure shows the best
response for all analytes. The middle chromatogram is the clay
soil, which shows somewhat similar recoveries although not quite as
good at the higher molecular weights, such as naphthalene and
hexachlorobutadiene. The top soil shows extremely low recoveries
across the entire range of analytes due to its absorptive nature.
The remaining factor to be established within the matrix modifying
solution is to select the appropriate organic modifier. Once
selected, the conditions for all sample matrices will be driven to
equivalency resulting in reliable data from any sample type.
CONCLUSIONS
There are seven variables which have been identified that affect
the partitioning of organic analytes out of soil samples. These
variables occur in many combinations which make it difficult to get
reliable data for soil samples. Use of a matrix modifying solution
combined with heated mixing/sonication can force the conditions of
the sample to become equivalent and override the inherent
differences of soil samples. This allows analysis of the analytes
within the samples to be accomplished successfully, reliably, and
with good precision and accuracy. The use of multiple internal
standards which act as true field surrogates combined with the
matrix modification produces quantitative and reproducible results.
The next step in this research project is to finalize the organic
modifier for the matrix modifying solution and to develop a
diffused standard for analytes and finalize the field test sampling
kit and procedures for its use. GC/MS BFB tuning will be automated
using the full evaporation technique. The end result will be a
reliable method for collecting and analyzing soil samples, where
the sample is handled once in the field and the vial is never
opened again. This will prevent loss of analytes through
volatilization or addition of impurities from the lab air.
Acknowledgements:
Mike Markolov, BP Research, Cleveland, OH
Tom Herman, Northern Lake Services Laboratory, Cramden, WI
-------
557
Tom Bellar, USEPA EMSL, Cincinnati, OH
Pedro Flores, TAI, Cincinnati, OH
-------
(/>
LLJ
C/)
>
o> co
o>
-------
559
TABLE 2. 7000/7050 SOILS PARAMETERS
Static Equilibration Study
Vial Pressurization:
V.I.P.R.:
Loop Size:
Sample Size:
Platen Temperature:
Platen Equilibration:
Sample Equilibration:
Vial Size:
Mixer:
Mix:
Mix Power:
Stabilize:
Pressurize Time:
Pressure Equilibration:
Loop Fill:
Loop Equilibration:
Inject:
Valve Temperature:
Line Temperature:
Injection per vial:
7 psi
5 psi
2ml
2g soil/10 ml salt solution
(spiked to 600ppb/soil)
85°C
0 min
10 min-160 min in 10 min
increments
22.5 ml
OFF
OFF
OFF
OFF
0.08 min
0.05 min
0.10 min
0.05
1.0 min
85°C
85°C
1
-------
560
TABLE 3.
GC CONDITIONS
Soils VOC Analysis By Headspace Extraction
Carrier Gas: Helium at 7 ml/min
GC: Varian 3300
Detector: Flame lonization Detector
Column: J&W DB-624 75m x .53
mm ID x 3.0 udf
Initial Temperature: 35°C; Hold 5 minutes
Rate: Ramp 3°C/minute
Final Temperature: 180°C; Hold 2 minutes
Injection Temperature: 200°C
Detector Temperature: 230°C
-------
561
TABLE 4.
REPRODUCIBILITY OF AROMATICS IN WATER
Without Mixing With Mixing
Mean! SD I RSD I Mean I SD I RSD
Benzene
Toluene
Ethylbenzene
o-Xylene
Bromobenzene
1,3-Dichlorobenzene
1,2,4-Trichlorobenzene
326
336
353
324
213
225
207
18
20
18
13
11
13
9
5.4
5.9
5.2
4.1
5.2
5.6
4.2
372
411
472
400
220
255
225
5
4
8
7
5
5
6
1.3
1.0
1.7
1.8
2.1
2.1
2.5
-------
562
TABLE 5.
EFFECTS OF MATRIX MODIFICATION/PRESERVATIVE
SOLUTION VOLUME ON PHASE RATIO
2g each in Soil only Soil + Solution
a 22.5ml Vial Phase Ratio PhaseRatio
Topsoil (0.45 gl/ml 4.06 0.56
Clay Soil (1.8g/ml) 19.20 1.03
Sandy Soil (3.0g/ml) 33.00 1.11
A Change = 8x A Change = 2x
-------
563
C
o
+5
CO
73
CO
O)
CD
V
LLJ
LLJ
DC
O
LU
X
-------
564
03
c
>
C rH C. X
Q) 0 -M 1
DO r- UJ 0
03
C
0)
0) N
C C
0) 0)
N n
c o
0) C_
.D 0
Q) O rH
C C. JZ
Q) 0 U
N rH -H
C JZ L.
Q) O H-
X3 -H |
o a -^
E 1 .
o m cu
c_ . .
CD •»-» -H
L
0)
JJ
O
3:
c
M O»
C
(0 ^H
0) X
E -H
•H 2
H
-M
C 3
O O
•H £
4J 4J
O '"-I
C_ 3
13
•H
cr
UJ
CO
ca
LU
DC
U)
0)
4->
i
4-o
i w
4-8
U-
o
o
cu
< C. 0) 10
-------
565
u
c
0)
0) N
C C
(U 0)
N JD
C O
0) L.
JQ O
tt) O rH
C L H
Q) o U
N rH -H
V C JZ L.
Q) 0) 0) C 0) U h-
c c n u JD -H i
0) Q) rH r-t O Q "*T
in
0)
E
•1-1
I- O)
C
C -H
O X
•H -H
4J Z
m
L JZ
D
CT
LU
JO
cu
LU
CC
O
--01
--in
--n
< L. OJ ID
-------
566
FIGURE 3a.
Total CGI vs Equilibration Time
No mix time/Variable equilibration times
9000 -r
10 20 30 40 50 60 70 80 90 100 110
EQUILIBRATION TIME (in ainut**)
120 130 140 150 160
-------
FIGURE 3b.
567
Tafget Compounds 1-3 vs Mixing Time
10 15
20
25 30 35 40 45
MIXING TIME (in »iaut«.)
50 55 60 65
Time
Temp.
Mixing
Time
Temp.
Mixing
Spike
Diffusion
Spike
Adsorption
(solidrliquid surface
area increasing)
Spike at
Equilibrium
(solid.-Jiquid surface
area stabilized)
-------
568
FIGURE 4. Effect of Soil Matrix Type Upon Partitioning
200.
•
160-
.
1 20-
80-
40-
0:
200.
•
160-
.
1 20-
80-
40-
o:
200.
1 60-
120-
80-
40-
0
d1190
1 ...
11111
d1 190
2_O_
I 1 71
d1 190
3_JU
56
502A/lnternal Std
in Topsoil
10ml Acidified
Solution
LJUAA
11111
)7
iiii
i
i i i i I
.1 L .il » ..
I I I 1 1 III
,
li i I I I i i i i i i i i i i i
502A/lnternal Std
in Clay Soil
10ml Acidified Salt
Solution
Ov_AjJLA
i f n~r
)8
JLil
TTTl
T
ll ,
"Mill
iL
i i i ii
_JVAA
— Mill
502A/lnternal Std
in Sandy
Soil
10ml Acidified Salt
Solution
^JLJuULJIl
.. li, ,
,,L
||
III i .. ii ,
1 II 1 1 1 1 1 II T'H
1
W, ' I H 1
0 6 12 18 24 30 36 42 48 54
-------
569
Analysis of Volatile
Organic
Compounds In
Soil Using Static
Headspace
Extraction
(Preliminary Study 5/92)
Greg O'Neil, Tammy Cappel
Tekmar Company
-------
570
GC CONDITIONS
Soils VOC Analysis By Headspace Extraction
Carrier Gas: Helium at 7 ml/min
GC: Varian 3300
Detector: Flame lonization Detector
Column: J&W DB-624 75m x .53
mm ID x 3.0 udf
Initial Temperature: 35°C; Hold 5 minutes
Rate: Ramp 3°C/minute
Final Temperature: 180°C; Hold 2 minutes
Injection Temperature: 200°C
Detector Temperature: 230°C
-------
571
7000/7050 SOILS PARAMETERS
Static Equilibration Study
Vial Pressurization:
V.I.P.R.:
Loop Size:
Sample Size:
Platen Temperature:
Platen Equilibration:
Sample Equilibration:
Vial Size:
Mixer:
Mix:
Mix Power:
Stabilize:
Pressurize Time:
Pressure Equilibration
Loop Fill:
Loop Equilibration:
Inject:
Valve Temperature:
Line Temperature:
Injection per vial:
7 psi
5 psi
2ml
2g soil/1 Oml salt solution
(spiked to 600ppb/soil)
85°C
0 min
10min-160min in 10 min
increments
22.5 ml
OFF
OFF
OFF
OFF
0.08 min
0.05 min
0.10 min
0.05
1.0 min
85°C
85°C
1
-------
572
ANALYSIS OF VOLATILE ORGANIC COMPOUNDS IN
SOIL USING STATIC HEADSPACE EXTRACTION
METHOD DEVELOPMENT SCOPE AND APPLICATION
A) Define the range of compounds which may be
accurately and reproducibly analyzed from a wide
range of soil matrices using static headspace
extraction.
B) Define the method detection limits using various
headspace extraction parameters as well as
separation and detection techniques.
C) Test the method using field sampling procedures
and devices.
-------
573
STATIC HEADSPACE ANALYSIS
QUANTITATION
(>o
1,
Sample Preparation
Mass Distribution (^
Constant [
Temi)erature 1
Vv \
Constant Time 1
,c°t
1 \
^ ^
2.
Sampling + Injection
Gas Law Effects }
T
— (V -V V-V
vvv VM'~VO
Sampling Needle ^^
Punctures Vial ^
Septa
Vv . ,.
M VM
VLNYci -fPy _V ,
^ II 1 ''in " m n
Co,
CM
p insiruiiiciu
Inject Mass ^
=V (@ P )*C
Start
Pressurize
and
Expand V into Sample
Loop Producing Cra and
Tekmar
5/5/92
-------
574
TOTAL AREA COUNIS
o
2000000 -
4000000 -
6000000 -
8000000 -
i
10000000 -
1
12000000 -
— "fcr —
14000000 •
1
16000000
Tj
(5'
C
CD
M
01
K»
O
N>
Ul
X °
M
H
H
Q Ul
^ O
E
P. *
P Ul
ji
Ul
o
tn
Ul
01
O
g
5?
o
r>
W
<
x
3'
(Q
H
3
-------
575
CONCENTRATION CALCULATIONS
SOLUTION SPIKE
CLO = Original concentration of standards in the
liquid phase (water or m.m.s.) after spiking.
_ NO spiked inio Hie liquid phase NO ,
o !—! = = ppb
VM ML KK
(200 NO) (6ul spiked into soil)
p _ ("') 1200 NO Jnn
PPb wgt
VM = Sample Matric Volume (liquids) or V = GMS SamP'e
M Density
-------
576
CONCENTRATION CALCULATIONS
SOLUTION SPIKE
Cso = Original concentration of standards in the
solid soil sample after spiking.
_ NG spiked into soil NO ,
80 " GMS soil = "GM" = ppb
(200 NG) (6ul spiked into soil)
(ul) _ 1200NG
2 GMS Soil ~ 2 GMS =
-------
577
Determining unknown sample concentrations using full evaporation
technique (FET) gas phase response factors (GRF)
1) Determine a known GRF using an FET standard
a) Inject no more than 10^il standard volume into a 22.5ml vial
and then seal vial
b) Run under FET standard conditions
c) Note the AC response of the compound to be quantified
d) Calculate the GRF: C raT
GRF = OE
AC
Example: 2\i\ of a 20ppm Parachlorobenzoic acid (in methanol)
measured before injection was 15ml. The injection
produced a detector response of 5.192 KAC
pr.T MASS IN VIAL 40 Ng
/~! I'll I _
V^* , ^ i. "~*
(ill
Vv VLF 22.5ml + 15ml
= 1.070ppb = .00107ppm
Parachlorobenzoic acid
GRF = C°E = .00107PPm = OOQ21PPM/
KAC 5.192 KAC' ' KAC
-------
578
Concentration Calculations
Full Evaporation Technique (FET)
Full Evaporation Technique (FET) provides gas phase
response factors, CGE/FET calibration curves as
well as target peak retention times for each
analyte in the standard.
C(3i/FE-r Concentration of gas phase analytes which are
injected into the analytical system
Vv Internal volume of the sealed vial
VLF Volume of gas expanded from the vial during
loop fill
ng analyte injected into vial ng
Gl/FET = " ~ =
Vvml + VLFml ml
22.5ml + 3.5ml 26ml
400ng
= 15.4ppb
-------
579
1) Non polluting extraction method
2) Rugged procedure:
a) Practical field sampling technique
b) Not subject to contamination
c) Wide range of sample concentrations
3) Techniques are available within this method
to account for variables in sample matrices
while producing both precise and accurate
data
4) Potential lower cost/sample through
increased capacity and shorter analysis
turnaround time
-------
580
Initial Selection of Analytes, Internal Standards
1) Full evaporation technique was used to produce
gas standards of known concentration (Cnc,cc:T)
* ot/r t1'
2) Analyte standard - 502.2A
Internal standard - DBFM, Tuluene-d8, BFB
3) Separation conditions were set to track
three target peaks and a non coeluting
internalstandard. The values derived for
R.T. and R.F. were used as a check
standard.
-------
581
FET Standard Conditions
7000/7050 Parameters
2jil of Standard Injected (200ppm each component)
Producing a CC|/FET of 15.4ppb
Vial Pressurization: 7 psi
V.I.P.R.: 5 psi
Loop Size: 1 ml
Sample Size: 400ng
Platen Temperature: 85°C
Platen Equilibration: 0 min
Sample Equilibration: 30 min
Vial Size: 22.5 ml
Mixer: OFF
Mix: OFF
Mix Power: OFF
Stabilize: 0.5 min
Pressurize Time: 0.08 min
Pressure Equilibration: 0.05 min
Loop Fill: 0.10 min
Loop Equilibration: 0.05
Inject: 1.0 min
Valve Temperature: 85°C
Line Temperature: 85°C
Injection per vial: 1
-------
582
7000/7050 SOILS PARAMETERS
Mixing Equilibration Study
Vial Pressurization:
V.I.P.R.:
Loop Size:
Sample Size:
Platen Temperature:
Platen Equilibration:
Sample Equilibration:
Vial Size:
Mixer:
*Mix:
Mix Power:
Stabilize:
Pressurize Time:
Pressure Equilibration
Loop Fill:
Loop Equilibration:
Inject:
Valve Temperature:
Line Temperature:
Injection per vial:
7 psi
5 psi
2 ml
2g soil/10 ml salt solution
(spiked to 600ppb/soil)
85°C
0 min
0 min
22.5 ml
ON
5 min-65 min in 5 min
increments
7
0.5 min
0.08 min
0.05 min
0.10 min
0.05
1.0 min
85°C
85°C
1
M.O.M.Parameter
-------
583
F.E.T. CHECK STANDARD
PASS = ± 20% OF INITIAL VALUES
BI RF
Target 1 10.4 Q.16
2 13.4 0.10
Target 3 22.8 0.79
Bromofluoromethane 162 10
IStd
RF = (Ax)
(A1S) (Cx)
Ax = Area of Compound
A(S = Area of Internal Standard
Cx = Concentration of Compound
CIS = Concentration of Internal Standard
-------
584
ANALYSIS OF VOLATILE ORGANIC COMPOUNDS IN
SOIL USING STATIC HEADSPACE EXTRACTION
PRELIMINARY METHOD OVERVIEW
II. Lab Analysis
Sample information will include site I.D., date, time and
temperature when sampled.
A) Trip Blank - to determine transport integrity
(Internal - contamination, degradation
standards check, R.T., R.F.
in MM/P (GC/MS: primary quantitation ions)
solutions)
B) Field Blank - to determine sampling integrity
(Internal - contamination, degradation
standards check, R.T., R.F.
in MM/P (GC/MS: primary quantitation ions)
solutions)
C) Dry Weight Sample - Determine sample dry weight
and use for sample
concentration calculations.
D) Analytical Sample - Run groups of target analytes
against appropriate internal
standards. Calculate
original analyte concentration
in Sample (ng —> ug Analvte \
^ gm Sample /
-------
585
ANALYSIS OF VOLATILE ORGANIC COMPOUNDS IN
SOIL USING STATIC HEADSPACE EXTRACTION
PRELIMINARY METHOD OVERVIEW
(.Field Sampling
A) Triplicate samples are taken with a soil plug gun
and placed into a screw top headspace vial. Two
samples are analytical duplicates and the third
sample is used for dry weight determination.
B) A matrix modification solution with preservative and
internal standards is delivered from a constant
volume/zero headspace dispenser into the two
analytical duplicates and one field blank. All vials
are sealed with screw top septa caps.
-------
586
ANALYSIS OF VOLATILE ORGANIC COMPOUNDS IN
SOIL USING STATIC HEADSPACE EXTRACTION
METHOD DEVELOPMENT SCOPE AND APPLICATION
A) Define the range of compounds which may be
accurately and reproducibly analyzed from a wide
range of soil matrices using static headspace
extraction.
B) Define the method detection limits using various
headspace extraction parameters as well as
separation and detection techniques.
C) Test the method using field sampling procedures
and devices.
-------
587
250 -r
200 -
150
8
100
Target CG| vs Equilibration Time
No mixing time/Variable equilibration time
TARGET 1
TARGET 2
TARGET 3
BROMOFLUOROMETHANE
EQUILIBRATION TIME (in minute.)
-------
588
Target CGI vs Mixing Time
No equilibration time/Variable mixing times
TARGET 1
TARGET 2
•* TARGET 3
BROHOFLUOROMBTHANB
MIZINO TIMB (in •inut«s)
-------
589
80'
60-
40-
20-
oq
80
502AStdFETat15.4ppb
J
d1250KP3a
60-
40-
Internal Std FET at 15.4ppb
Toluene-d8
Bromofluoromethane 4-Bromofluorobenzene
01250C2
60-
40-
20-
502A and Internal Std FET at 15.4ppb
Bromofluoromethane
3
0
LJ
12 18 24 30 36 42
liJLA
48 54
-------
590
[Blank Page]
-------
591
MR. TELLIARD: Our next presenter is going to talk about Method 524.2
which is presently undergoing a methods consolidation effort for the agency method for
VOAs, and Jean is going to talk about how successful they were at applying the VGA
method to the additional list of analytes.
MS. MUNCH: Today, I will be presenting the evaluation of 48 candidate
compounds using USEPA Method 524.2. As an introduction, I will give an overview of
the current method. [SLIDE 1ONE]
Currently, the method is used for the analysis of 59 volatile organic
compounds in water. It isolates those compounds by purge and trap, and then the
analytes are separated, identified, and measured using capillary column GC/MS.
[SLIDE 2]
The equipment we used for this particular project was a Saturn ion trap
system equipped with a purge and trap unit with a moisture control module. We used a
5 ml sample volume and the standard three-stage trap that is already outlined in the
method. [SLIDE 3]
We used capillary GC with a 75 meter DB624 megabore column. At the
end of the column, we had an open split interface set to split the column effluent at 20:1
before going into the ion trap detector.
The test compounds were selected because they are regulated or have the
potential to be regulated. Many of them are on the 1991 drinking water priority list, and
many of those are water soluble compounds that are particularly difficult to analyze in a
water matrix. [SLIDE 4]
The steps we went through in the evaluation process were first, to create a
calibration curve for each potential analyte and assess its linear range. We then looked
at the chromatography, because potentially there could be 100-plus analytes in this single
method. Additional steps were to evaluate purging efficiencies, calculate method
detection limits, and do a time storage study in real environmental samples to see if the
sample holding times would be valid for the new analytes, and take a look and see if
-------
592
there were any potential matrix effects. [SLIDE 5]
The first thing we did in calibrating was to attempt to run a six-point
calibration curve for all of the potential analytes. The concentration range was from .2
to 100 parts per billion.
The first part of the calibration step really acted as a screening procedure,
because the first thing we found out was that 18 of the original 48 analytes gave no
response at 100 parts per billion or less. As we look at this list, there are amines,
alcohols, and high boiling compounds. So, it wasn't really a surprise that these didn't
work very well. [SLIDE 6]
Therefore, at this very first step, they were eliminated from any further
study.
Also during the calibration phase, we observed that 2 analytes had
extremely short half-lives in water, propylene oxide and bis-2-chloroethyl sulfide. So,
they too were eliminated from any further study. [SLIDE 7]
Finally, you get a look at a partial list of the compounds that we kept for
further study. As you can see, there are ketones on this list that are on the drinking
water priority list. Also of special interest was methyl-tertiary-butylether. [SLIDE 8]
All of these compounds worked surprisingly well in the initial calibration.
They were all linear over at least two orders of magnitude.
1,4-dioxane and vinyl acetate had some non-reproducible data, but we will
talk more about them later.
Here is the second page of compounds that were kept for further study.
Of special interest on this list is acrylonitrile which is targeted for regulation in Phase VI
B of the Safe Drinking Water Act. [SLIDE 9]
Again, all of these compounds were linear over at least two orders of
magnitude and showed generally good response during the calibration.
One observation during the calibration phase of the project was that some
of the standards didn't hold up very well in methanol which is the prescribed solvent for
Method 524.2. [SLIDE 10]
These particular five analytes were found to need weekly standard
-------
593
preparation and, of course, to be stored refrigerated. It causes a little bit more work, but
it is not really a problem, because we already have a number of analytes in the method
that require weekly standard preparation.
The next step after the calibration was to take a look at the
chromatography. Even though we were using mass spec, we wanted to minimize the
number of co-eluting peaks. When you are talking about possibly close to 100 analytes
in a method, you really want to minimize multiple co-eluting peaks. [SLIDE 11]
We had already decided at the beginning of the study that we were going
to use the 75 meter column to try to give ourselves a little more room to minimize the
number of co-eluting peaks, and we used cryogenics, because it gave better overall
performance for the existing analytes and for the analytes that we were attempting to
add, although it may not strictly be necessary. [SLIDE 12]
Starting at minus 10 degrees C and using a multi-ramp temperature
program that lasted approximately 40 minutes, we ended up with seven pairs of co-
eluting peaks and only one case where three analytes eluted together. In each case, the
mass spectra of the co-eluting analytes were sufficiently different to allow independent
quantitation using unique quantitation masses.
Now that we had verified the chromatography that we were going to use,
we wanted to look at purging efficiencies. Although this information isn't strictly needed,
it has traditionally been used as a diagnostic tool and as a measure of how rugged the
method is for certain classes of analytes. [SLIDE 13]
In order to calculate the purging efficiency, we ran a series of standards at
a particular concentration and kept track of the raw peak areas. Then we took the trap
out of the purge and trap unit and spiked that same standard directly onto the trap,
reinstalled the trap, put reagent water in the purging device and ran the sample as usual.
Then, by dividing the peak area of the purged analyte by the peak area of
the standard spiked on the trap, we obtained a direct measure of purging efficiency.
These are the results of the purging efficiency experiments. The range was
1 to 100 percent for the compounds we were looking at. The mean and the median were
both around 50 percent, indicating that maybe half of the proposed analytes were really
-------
594
good candidates, and the other half could have potential problems with precision.
[SLIDE 14]
At this point, it was time to do our traditional calculation of method
detection limits. For this, we used a series of seven replicates analyzed at the estimated
detection limit. Because we already had all the data from the original calibration steps,
it was fairly easy to choose a concentration at which to run the MDL study. [SLIDE 15]
The range of the method detection limits was .02 to 1.9 micrograms per
liter with a mean of .36 and a median of .18 micrograms per liter. The mean is brought
up somewhat by the very few compounds that had MDLs greater than .5 micrograms per
liter. [SLIDE 16]
I believe there were only four compounds that had MDLs above .5. When
we look at the purging efficiencies of those compounds, their somewhat higher MDLs
are directly related to their low purging efficiencies. But for some of the compounds we
are talking about, chloroacetonitrile and THF, for example 2 parts per billion isn't bad.
One of the final experiments was to do a time storage study for the 28
analytes remaining in the study in some real environmental matrices. We needed to find
out whether or not we could meet the sample holding times allowed in the method and
if there would be any problem with the preservation techniques. [SLIDE 17]
Currently, the method allows a 14-day holding time when the samples are
acidified, dechlorinated, and stored at 4 degrees C. The matrices we selected for the
time storage study were a raw surface water and a chlorinated drinking water preserved
and dechlorinated according to the method directions. [SLIDE 18]
Only two analytes showed losses during the 14-day holding time allowed in
the method. These were benzyl chloride and acrolein.
This particular slide is from the tap water data, but we saw a similar loss in
the raw water data. So, these two compounds were eliminated as candidates for the
method. [SLIDE 19]
During the time storage study, we observed two analytes with poor day-to-
day precision: vinyl acetate, because, as it turned out, we were actually outside its linear
range and for 1,4-dioxane, because it had 1 percent purging efficiency and a response
-------
595
factor of .001, both of which can cause precision problems. [SLIDE 20]
The last thing we wanted to look at in this evaluation process was to see if
there were any apparent matrix effects, and because we had all of the data from the time
storage study in raw water and chlorinated tap water, it was kind of easy to do a review
of that data. [SLIDE 21]
What I did was to bring together the precision and accuracy data from the
preparation day of the matrices from the time storage study. The grand mean of the
precision and accuracy were calculated for each matrix from four replicates. We also
calculated the grand mean of the precision and accuracy for the calibration standards
which were prepared in reagent water.
The mean precision data didn't make it on this slide, but for each
individual matrix, the grand mean for all 24 remaining analytes was less than 6 percent.
[SLIDE 22]
The accuracy data which you see on the slide, the grand mean for 24
analytes was 97 percent for reagent water, 107 percent for raw water, and 105 percent
for tap water.
These data seem to be very comparable, and there is no apparent matrix
effect for these analytes. '
In summary, we were pleased to find that 24 of the 48 compounds tested
appeared to be suitable for inclusion into Method 524.2. The purging efficiencies of the
suitable candidates ranged from 5 to 100 percent. The MDLs ranged from .02 to 1.6
parts per billion. There was no apparent matrix effect, and we were able to select
chromatographic conditions that minimized the co-elution of analytes. [SLIDE 23]
The next step will be to include these compounds into interlaboratory
validation studies so that we can evaluate the ruggedness of the method for each of the
analytes and, to compare this single laboratory data with data that we can get from
typical multiple laboratories.
Thank you. Questions?
-------
596
QUESTION AND ANSWER SESSION
MR. STANKO: George Stanko, Shell Development Company.
Method 524.2 has undergone ASTM round robin testing. Were these
analytes included in that round robin testing, or will this have to be done in the future?
MS. MUNCH: Most of the preparation for the round robin study was
done in advance of these experiments. Some of these compounds of particular interest,
for example, the ketones that are on the drinking water priority list and methyl-tertiary-
butyl ether were included in the recent round robin study, and we are currently
evaluating the data.
Most of that data for the analytes that were included and are common to
these studies turned out very well.
MR. STANKO: The other thing is I would like to make a comment that
Method 524.2 was being round robin tested as part of the method consolidation effort of
the EPA to be used across all matrices, including Clean Water Act effluents and also
RCRA groundwater.
The matrices that you studied here are very simple and would have little
impact on the purging efficiency. I think if you find when you have more complex
matrices like an effluent or some groundwater samples, you are going to find that the
purging efficiency changes, and not only does it change, it is quite variable depending on
what else is present in the sample.
MS. MUNCH: That may very well be true. As I said, some of these
compound have already been in the recent study, and that did not appear to be the case
for those particular analytes, but when these go to round robin, all matrices will be
checked to see if there is a matrix effect.
This is simply single laboratory data to give us a starting point.
MR. TELLIARD: George, the issue was addressed Tuesday in a methods
consolidation meeting, and we feel that probably this data is certainly supportable for
ambient water and...perhaps ambient water and drinking water. We don't feel that the
data supports any effluents or in-process streams.
-------
597
Any further inclusion of those into the method would require additional
testing.
MR. WESTON: Charlie Weston from ETC.
I have two questions. Number one, could you tell me what the spiking
level was for your MDL study, and, number two, did the Saturn ion trap give the option
of using a jet separator instead of a straight split?
MS. MUNCH: Okay. The MDL studies were done at different levels
tailored to the estimated detection limit that we had calculated from the calibration
studies. Most of them were done at .2 or 2 parts per billion.
The Saturn ion trap, at the time we purchased it and, I believe, also at this
point in time, does not come with a jet separator option. We have not hand any
problem with the open split option, and it has worked very well for us.
MR. WESTON: Okay, thank you.
MR. THOMAS: My name is Roger Thomas. I am from Viar and
Company.
I notice with 524.2 that you had the ketones as part of your 48 compounds,
the 2-hexanone, the 2-butanone, the 4-methyl-2-pentanone, and the acetone. Historically,
those compounds have poor purging efficiency when you switch from the regular 5 ml to
the 25 ml purging.
Is there anything in particular that you have done to improve their purging
efficiency? And, number two, did the ion trap significantly improve your response
factors for those compounds and your low response factor compounds?
MS. MUNCH: You are right. We observed purging efficiencies for the
ketones in the neighborhood of 25 to 30 percent, with a 5mL sample. Traditional
wisdom would have it that they would not have sufficient precision for the method.
It is my personal opinion that the quality of purge and trap devices over
the years has gotten better, so that low purging efficiency compounds may now
demonstrate acceptable precision and accuracy.
This is single laboratory data, but we did not take any special pains to
obtain this data. We just ran the system as we would normally run it.
-------
598
The ion trap is a very sensitive system, and its use contributed to the low
MDL's achieved for poorly purged analytes.
MR. THOMAS: Okay, thank you.
MR. TELLIARD: Thanks, Jean.
-------
599
Q3
Q
LJJ
CM
O
LU Q
—m —5P
' ^Q
og
DC O. ±
9^
U-O
U-° ^
om<
co
o
c
05
0
>*
1_
o
>>
0
CD
o
CO
•*-»
O
»-
a.
"TO
0 O
E ±:
g§
CO
_: c
c
LLJ
-------
600
C\J
^st"
CM
LO
Q
O
X
LJJ
^
^~
—
LL.
LU
•
CO
HMBHH^
ID
••••^
H—
O
CO
M f\
CO
"co
c
CO
0
H— »
£ 0
"cO
0 ^
CO c
13 .=
_>, CO
"r~ ^
0 "^
v_
^ O)
O 10
Q.
CO
•£
03
* >^
0
D)
Q.
^
TJ
0
^_j
CO
O
CO
CO
0
H-J
"cO
c
<£
T5 ^
^^^ ^ ^
co O
O
0 c
E E
•*~> ZJ
0 "O
!2 °
j> j|
2 'Q-
co ^
Vw f ^
Q. ^
0 r-p,
CO ^
0 "co
%, —^
CO
Vw
w ^
B *-
.>» CO
co cp
c 0
< E
-------
601
LJJ
CO
CO
Q.
O
CL Z
LJJ 03
CO
c
03
03
T3
C
__ 03
2 0
3 a
co E
.^ b 03
O
§~E
Q.
03
0
O D 03
1g 00^
O ^ CO
JD
03
Q.
03
O
€O
0
D)
O
03
C
E
o
o
.
03
o
O
^
~
>
O
O
o
o
CO
/—
CNJ
CD
CO
Q
E
E
CO
10
CD
V
O
C\l
0
O
«£
5
•—
"5.
CO
i_
o
•«— »
o
0
-t->
0
Q
Q.
03
E
in
0
a.
O
-------
602
CO
Q
2
mmmtm
O
Q.
O
o
h-
CO
LU
h-
LL
O
O
O
LU
LU
CO
c
o
mmm^m
' evaluat
••••
o
H—
elected
CO
0
0
^>
CO
T3
C
O
CL
E
o
O
ro
c
0
O
d
0
03
O
£>
JS
D)
0
0
03
0
0
CO
Z5
03
o
N«X
0
.Q
T3
0
J9
D
D)
0
0
O
•*-•
•4-J
'^
O
ol
0
"03
D)
C
C
Q
CO
^•^Px
CD
0
-i— >
C
o
0
03
c
03
y
0
-1— »
O
O
"co
0
>1X
5
"c
"o3
'o
0
Q.
CO
H—
O
0
03
O
•4—"
CO
CO
E
03
O)
O
Q.
nmenta
o
0
0
03
-i— »
03
CO
T3
C
o
Q.
E
o
o
0
.Q
Z5
O
CO
0
-#— »
03
0
03
c
03
matrix.
0
-*— »
03
03
0
N
>
03
03
O
•1— '
•+— »
"5
o
• ••••
^^^^
-------
603
CO
CO
LU
o
o
DC
Q_
O
LJJ
LLJ
I
CO
Q.
LU
0
D)
C
CO
DC
CO
0
Li
c
_g
.g
"co
O
>
Q.
CO
D)
O
"CO
E
o
£
O
CO
0
'o
C
0
'o
«4—
LU
D)
C
O)
D
Q_
CO
E
Li
c
o
o
0
0
Q
O
0
^
in Environme
>»
Z5
•4— '
CO
0
D)
CO
0 CO
en 0
^ 0
0 '£
E-*-«
CO
i- 2
CO
o
0
»•___
*t
UJ
X
"CO
^
CO
-------
604
m
o_
a.
o
o
LU
CO
O
CL
CO
LU
DC
o
0
malononitril
0
-i— •
03
"a
.c
"2
o
us
._
E
__y
J5
^^
>
-^
0
E
N-nitrosodi
o
c
03
.C
-!-•
0
O
o
o
1
CM
0
C
E
_03
1 >,
Q.
O
Q.
N-nitrosodi
o
03
Q.
o
^^
Q.
1
CM
I
O
~
O
T3
I
CO
t-T
2-picoline
0
c
Q
r"*
03
0
oroproi
~
o
T3
1
CO
T- "
O
O
o
propargyl a
0
c
E
03
O
c
03
"0
73
0
C
-4— »
B-propiolac
0
c
03
^^H
Z5
JD
X
O
Q.
0
73
1
CO
CM
T-"
0
'E
_cg
Q.
o
Q.
1
c
0
rotolue
•*-•
'E
• m^mm
m*m**m
73
I
CM"
pyridine
0
c
rotolue
•*-•
'E
i
CD
CM"
o-toluidine
0
c
o
v_
o
Q.
O
CO
-------
605
DC
LLJ
m
co
>•
LJJ
DC
X
LLJ
CO
03
CO
0
0
LJJ =
0
0
CO
X
o
0
c
0
mammi
CL
O
Q.
>
0
O
£
O
I
CM
i
CO
-------
606
O
^E
o5
o
0
C
0 O
Q
ID
h-
CO
DC
LLJ
I
1-
DC
Z)
LL
DC
O
LL
CO
Q
^
CO
0
"0
C0~
0
to
0^
V)
alcoho
co"
c
o
Q.
E
o
VX
O
:anone
x
0
i
C\J
Q)
CO
o
S2
0
E
-t— •
_03
O
OJ
* w
"0
E
0
E
s^
£5
c
0
Q.
I
CM
I
mLmmJ
0
E
i
0
C
03
Q.
O
>*x
^^^
+-*
JD
I
I
0
E
03
D -2
I"
>* O3
ffimm ^ Vr
tf -^
i_ *><
0 ~
•*-» P*
_0
]S
j^
Q.
O
0
(D "• 0 0 03
Cv^ /«• ^^ r"™
i~ \— 7-: -p.
2 _° 5 ^ 0
C £- r— ^> c
C -TT 03 - O — C
O ^ -H-
S 2 JD I . n) j-
00 i T- ^ .02 ^
O3 O3 (M T-^ T-" "D 0
I
0.0
"O r- T-
-------
607
DC °
UJ P
2 £-S § Q.^
Ill^eS
CO •£--£- n .— •*-•
x •£ •£ t: c c
0 ^ ^ ±: i 0 ^
c E E c CM Q. Q.
OS
• •
LL w
>
°%
0=0 CO
CO CO & O O
O
-------
608
LU
LU
^mmm^^^m
o
LU
a*
O
Q. (T)
0
0)
•§•••••
O
o
03
o
o g «
O
03 O
o
O
-------
609
CO
CO
LU
O
O
VALUATION PR(
LJJ
LJJ
_
CO
QL
LU
0
D)
C
03
CC
0
C
c
g
5
.Q
"03
o
Q.
03
O)
"03
E
o
£
O
ncies
t0
^j^^^
^i^^^
LJJ
O)
C
D)
^
Q_
eo
E
Li
c
_g
"o
0
•i— »
0
Q
TJ
O
JZ
0
^
*
Study in Environme
O)
03
i_
o
•+- •
CO
0
E
[__
*
CO
0
_g
•i— •
03
U)
"o
0
• •_
H—
LJJ
X
"03
*
CO
-------
610
Q_
DC
O
O
CC
O
c
•o
co 0
C
E
CO
o co~
°
CO
O c
ffl ^<°
Q O CO
E £ S
E._ CO
CO
iq
d
X
^_y
o
E c o 0
10 o o o
N. o co o
co
co
co
0
CO
Q. CO
CO-rr
CO .2
ZJ
0 CO
O 0
0
= CO
CO
C
CD
ZJ
or
0
O o «
CO
0
o ±:
0
Q.
co
- B
—T >»
P §
CO
o
CO
0
CO c C
CO CD CO
P 5 CT 5
«B^ a • ^^^*"1"
0 D
£ W
~ 0
0 *-
CO ^
CO ^
o
CO
^1 0
O +-
CO >>
0 CO
_ c
J=. CO
de
CO
CO
CO
c E
0 r-
S!
O IS
r- CC
O cC
= zs
CO D"
-------
611
CO
CO
LU
O
o
x«^
DC
Q.
Z
O
ID
— •
5
LU
LU
I-
__
CO
Q.
LU
h-
CO
0
D)
03
OC
i__
03
0
Zi
2
J5
"03
O
_>
Q.
03
D)
"03
E
o
_£=
O
CO
0
'o
c
_0
H^
LU
D)
D)
^
Q_
CO
•»— >
I
O
1 *
^^^^^
O
0
Q
O
JZ
0
^
*
ly in Environme
u
CO
0
O)
0 CO
CO g
0 'i-
£•*-•
03
j_ ^
*
^0
"o
0
K 1
*l
LU
X
'm
"03
*
-------
612
CO
LLJ
o
LU
O
U_
LU
O
O
DC
o
o
I
CD
O
03
DC
03
03
CD
-------
613
CO
CO
LLJ
O
O
vLUATION PR(
5
LU
LU
h-
CO
CL
LU
0
D)
C
03
DC
03
0
C
c
o
2
.Q
"03
O
>>
a
2
D)
"03
E
o
£
O
*
CO
0
'o
C
0
'o
LU
D>
'o)
^
CL
*
CO
+-•
E
o
o
0
Q
T3
O
JZ
0
^
*
idy in Environme
_>
-t—>
CO
O)
03
0 CO
CO g
0 "£
E-t-"
._ «o
•
CO
"o
LU
x
to
^
•
CO
-------
614
CO
o
o
LLJ
LU
Q
Q
O
h-
LJJ
O)
•
1
CM
o
•
o
0
D)
C
03
DC
^-S
O)
13
CO
CO
•
o
03
0
•••••
^
O)
13
00
•
o
c
03
T3
0
•P""""
^
-------
615
CO
CO
LU
O
O
•^•^
DC
Q_
Z
O
ID
~J
^
LJJ
LU
1-
Z
CO
Q.
LJJ
0
D)
C
CO
DC
i_
CO
0
C
o
£
.9
"co
o
>>
Q.
CO
D)
O
-1— »
CO
E
o
£
O
CO
0
'o
C
0
'o
^1^^
^•^^
LU
D)
C
D)
Q.
*
CO
•+-*
E
Li
g
"o
0
Q
O
JE.
0
^
*
ly in Environme
QJ
•— »
13
"1— '
CO
D)
CO
0 CO
CO g
0 '5=
E-t->
CO
*
V)
o
LU
X
•*-»
03
•
CO
-------
616
Q
CO
LU
O
DC
O
CO
LU
_
m
"O .2 0)
T3
"~ '5 co
CO CO
CO 0
o E
CO 0
CO O
0
o
co
£
-------
617
LJJ
Q CO
:0
DQ O
CO
CO
O
in
C\J
O
CM
IO O
-OUOQ
05
Q
03
Q
03
Q
CO
>*
03
Q
03
O
i_
O
o
O
N
C
-------
618
Q
LU
Q
Q
z
••^^••B
I!
LU
CO
Q
2
••"•X
ID
O
CL
•^
O
o
^
o
-4— »
£—
0
jg
M—
«^-
•••^
13
CO
C
* "^™
1
0
^1 ^
^*r» ^^^^
I^^^B C 1 1
0 c
O CO
CO »-
^> CO
C 0
O
c
_0
M—
0
O) T-
.^ O
o>q
H
&~- v_
^ O
_ M a
' 0
^. rrt
0 -i
CO 0
X CO
o c
Q °
LJ Q.
_* CO
0
-------
619
CO
CO
LLJ
O
O
DC
0_
z
o
LLJ
LLJ
CO
CL
LLJ
0
D)
C
CO
DC
^_
CO
0
C
o
"co
.9
"co
O
_>
Q.
CO
D)
O
"cO
E
o
_c
O
CO
0
'o
C
0
'o
H—
M—
LU
O)
C
O)
D
Q.
^0
E
o
o
0
•4— »
0
Q
T3
O
JZ
0
^
ly in Environme
u
CO
0
D)
CO
1—
O
•+-»
CO
0
E
i-
co
0
O
'C
to
^>
CO
•f- •
o
^0
LU
X
•4— »
CO
^
CO
-------
620
o
DC
O
o
o
co
LU
LU <
LL LJJ
O LU
DC
O
CO
DC —
Q.
03
03
DC
O
O)
03
0
DC
LO
o
O)
CO
CD
-4—»
03
03
O
4—
c
03
CD
E
O
-------
621
oc
CO
•^^
.Q
O
0
CO
»t
•D
_0
"co
0
CO
-n
compounc
CO
*o
CM
1
^
_C
c
o
"co
Z3
O
c
1_
^o
_0
3
"3
CO
Die candidates
m*—+
"5
CO
o
fficiencies
0
"o>
Q.
•
o
o
1
IO
E
o
0
O)
2
D)
CD
i
CM
O
O
E
o
**—
T3
0
D)
CO
CO
_J
Q
^>
*0
X
"S
E
r~
no apparer
CO
0
0
JZ
H-
i
"E
E
TJ
0
"o
0
0
CO
CO
g
^*^m
• mm^
^««k.
graphic cc
o
"5
o
6
CO
"^
co"
'o
o
o
o
0
N
"E
c
o
"5
~CO
"5
o
_CO
_0
"c
CO
Q.
0
•i— >
CO
"x
0
C
0
1-
-------
[Blank Page]
-------
623
MR. TELLIARD: Our next speaker is, I guess, a founding father of this
conference. He was here when we opened the meeting, and he has been here ever since.
George Stanko is from Shell Development. He is going to talk a little bit
about a project they have developed on TQM which we all know means that there is a
threshold quality of misery. But George is going to talk about how he applies it on the
contract lab program.
MR. STANKO: I probably won't need the microphone, as usual. I want to
dispel a rumor that has been going around. Bill and I are really not twins.
I would like to share with you the results of an environmental analytical
contract laboratory study where the quality improvement process was used to correct
problems and to improve performance. This is something unusual for me. I usually
come here and complain how bad things are. This is the first time I have come here and
said how we have improved something.
A blind performance study of 24 environmental contract laboratories was
conducted late in 1990, and this study was conducted on a real matrix sample, a bunch of
samples, and they were submitted to these commercial labs, a select group, totally blind
to them. Dummy engineering firms were set up, and the samples had a real background
matrix. We put a little gasoline in it just to make things a little interesting.
The major goal of this PE study was to assess the performance of a select
group of laboratories for the analysis of groundwater samples for volatile organics by
GC/MS, metals by ICP, and also a select group of general parameters, the BODs, the oil
and grease, and pHs.
Results of that study were reported at the Norfolk conference last year.
Results showed that there were really no problems with the ICP metals
analyzed by these 24 laboratories. The precision and the accuracy for the ICP metals
was outstanding for all participants, and the overall mean recovery for all 11 metals was
97 percent.
Among the individual metals, the mean of all laboratories ranged from 89
-------
624
percent to 107 percent, and with such good precision, no corrective action was really
needed for any ICP metals analyses.
Most of the labs that were included in the study performed well for the
volatile organics by GC/MS, but there were some problems noted at two particular
laboratories. Both laboratories did not find approximately half of the analytes that were
spiked in the samples. These would be considered false negatives, and we consider false
negatives important.
The problem was such that we contacted the labs by phone, provided them
the necessary information they needed to go back to their raw data, and to search out
what we identified as the root cause for their poor performance.
One of the labs quickly responded after they had reviewed their raw data
and identified the problem that the sample had been diluted on the basis of their
sacrificial lamb. They normally screen all samples with a GC/FID technique, and
because we put a little gasoline in these samples, they saw the humpogram, and they
calculated a dilution factor based on the humpogram and totally ignored what was totally
in the samples.
They were the only laboratory out of 24 that found it necessary to dilute
samples.
That identifies the root cause, and the corrective action, the samples
should not be diluted unless it is absolutely necessary, and if it is necessary to dilute
samples, the quantification levels should be adjusted according to the dilution factor.
We also had some concerns about what dilution might do to samples.
There is very little information in the literature to indicate what that might do to
analytical variability.
What we did internally in Shell is to devise a study to measure the impact
of diluting samples for GC/MS analysis using the purge and trap technique. These are
the particular analytes that were selected for the study, and here again, you have seen
these before, and MTBE is one of them.
That is one we heard from a previous speaker, and there is an ether
included in there. And these were some of the analytes that were originally spiked
-------
625
blindly into the sample.
Special care was taken with the samples, and no other samples were run on
the instrument during the time of this study. The spike levels are indicated at the
bottom of the slide.
To reduce all the variability we could in this study, we used one particular
analyst from start to finish.
These are the tabulated data for the study that we did on the dilution.
Across the top, the compound concentration, and we show on the slide the mean
recovery as well as the standard deviation.
This is the result of eight particular runs. Why not seven? Why not ten?
For some reason, we chose eight runs.
The last sample which is the diluted 40 part per billion sample had a
nominal 2 parts per billion of each of the analytes. For the statisticians who are still up
in the audience, these are the corresponding coefficients of variations in large enough
numbers so everyone can read.
We also had our statistician look at each individual analyte and compound,
and we did some additional figures that appear in the paper. After this talk is
concluded, I have approximately 100 copies of the paper available, so you can pick them
up when that occurs.
The results for the 5, 10, 20, and 40 part per billion samples and the spiked
samples showed purge and trap method has the capability to measure concentration of
the eight selected analytes with good precision and accuracy in the distilled water matrix.
The coefficient of variation for all concentrations and analytes were less than 10 percent,
with toluene being the only exception.
Based on the observed CVs for this study, one would conclude that the
method is quite capable of measuring the concentrations of these analytes down to the 5
part per billion level, and the analytical variability is relatively constant over a range and
under the ideal conditions of this study.
I often told Bill Telliard it would be a cold day in Norfolk before he would
ever hear me say that. I think today qualifies.
-------
626
MR. TELLIARD: Is a cold day, right.
MR. STANKO: We observed in the last two columns of the previous slide
the estimated mean standard deviation shown for the diluted samples from table 1. The
actual diluted sample analytes contained only the 2 parts per billion levels of the
analytes. The corresponding coefficient of variation shown in table 2 are also on the
same basis.
Comparison of the means for the 40 part per billion spiked sample and the
diluted sample for each of the analytes revealed that in every instance, the mean from
the diluted sample were biased higher. Comparison of the estimated standard deviations
also showed a similar bias with toluene being the only exception.
Since the spiked and the diluted samples should be compared for each
compound, the statistician told me this is a paired comparison. A two-tailed sign test
can be performed to assess the probability of observing eight out of eight higher means
and seven out of eight higher standard deviations purely by chance.
The test results indicated that the chance probability or the p value for the
means is less than .01 for the standard deviations and less than .05...or.01 for the means
and .05 for the standard deviations. Thus, we concluded that diluting of samples prior to
analysis resulted in less precise and accurate results.
The reason for this less precise and accurate results is quite simple. In the
case of the 40 part per billion spiked sample, all the concentration estimates were
calculated from response levels that were near the center of the calibration curve.
The uncertainty and the inaccuracy associated with establishing the location
of the calibration curve are at a minimum near the center and increases considerably at
the two extremes. The response level for the diluted sample which was a nominal 2
parts per billion was close to the one extreme or 1 part per billion standard.
There is another important consideration associated with the dilution of
samples. In the above study, all of the concentration levels were designed to be within
the calibration range.
For real samples or unknown samples, one never knows what the
concentrations of these materials are, and in the case of lab number l,they diluted the
-------
627
sample and diluted the concentrations of the analytes below the detection limit.
In the second laboratory that had problems with organic volatiles by
GC/MS, it was more complicated. The first response we got from the laboratory said
that they had reviewed all their raw data, and they had concluded that their reason for
missing more than half of the analytes that were spiked into these samples was that they
missed the holding time by two days.
I had a little difficult time accepting that as the root cause for their poor
performance. They assured me that that was their root cause, and now they are back in
analytical control, would I please send them another sample.
With great delight, I sent them another sample and let them fall flat on
their face.
Following their second poor performance, because the lab was deemed to
be essential to certain operations, we opted to send two analytical chemists to the
location to work with the lab for two days to establish what the root cause for their poor
performance was and to get it corrected.
In this particular case, there wasn't a root cause. There were 13 root
causes, and I will not go through the list. I will only zero in on one.
They were using a muffle furnace to clean their syringes, and when you
check the calibration of the syringes, that can throw you off by 1 ml. In the 5 ml sample,
that could introduce 20 percent error before they do anything else wrong.
The two Shell analysts worked with the laboratory to correct each
individual root cause here to a point when they finally left this laboratory, the laboratory
could at least get the correct answers on the standards that were prepared by our people.
The laboratory also agreed to analyze whole volume samples that are
commercially available from ERA and to provide Shell with the data so that we could be
assured that they were in analytical control at that time and for the next year.
I can report that the laboratory has performed well, and there have only
been a few minor problems that they have had using these ERA standards.
The results for the blind study for the general parameters was rather
disappointing, for those of you who recall, particularly the oil and grease, pH, and things
-------
628
like that. In general, the performance was so poor at more than half of the labs that we
didn't know exactly how to address this on a one-on-one situation.
Instead, what we thought we would do is propose a voluntary program
where Shell would pay for whole volume samples from ERA that would be sent to the
labs three times every other month, and the laboratory would analyze the samples at
their expense and share the data with Shell and ERA so the data could be statistically
analyzed.
All 24 labs agreed to participate in the voluntary program, and their
response really indicated that laboratories are concerned with data quality and do try to
improve performance. They are also willing to be active participants in programs that
should lead to improved performance, and it also demonstrates the use of the quality
improvement process where the laboratory and the laboratory customer work together to
achieve some mutually agreed upon goal. In this case, it was improved performance.
The performance from the initial blind study which I reported on here last
year was used as the benchmark for the voluntary program. The results from the initial
blind study for the three whole volume samples were statistically analyzed, and the data
were plotted in a number of ways.
The number of outliers for all these data is shown in the next slide. The
data were further summarized with outliers removed into the next table. This table
compares the results from the blind study or, as we call it, before, to the after which was
the voluntary program. It also included on this slide, it shows the average recovery from
all,laboratories and the standard deviations for the means.
The question, was there a significant reduction in the variability, was a yes
for all parameters, and the corresponding p values are shown in the final column and
were included in this table along with a definition of what p stands for on the bottom. I
don't want to explain it. For the benefit of the statisticians, that is how we derived the p
values.
There are a number of ways to plot this data, and here again, I think a
picture is worth a thousand words. The statistician initially did it for me, and she says
which format do you like. I said I like all three formats, and will show you them.
-------
629
This particular format is called the box plot with whiskers. Essentially,
what is in the box is 50 percent of the observations are within a box. The line that goes
through the box identifies the median value, and the whiskers identify the 25 percent
above and below the boxes. This is for BOD.
If you compare the two blue boxes on the left which don't show that well
to the three red boxes, you can see that there is an improvement in the mean recovery
for BOD, and the boxes get smaller which is an indication of improved precision.
This is a similar plot for COD, and it is quite dramatic. This is the
difference for COD, and here again, you look at the size of the boxes and the size of the
whiskers, and that is an indication of improved precision as well as accuracy.
In this case, we have poorer accuracy but closer to 100 percent, or lower
accuracy, I would say, but the precision was a very good improvement.
This is my old friend, oil and grease, which I was really concerned about
the last time. If you look at the size of the boxes and the size of the whiskers before, the
left two, compared with the three on the right, you can see that we are in analytical
control, and there was a tremendous improvement in both precision and accuracy.
This is for pH, and I would like to have you remember those four dots on
the bottom in the last column. That was the last performance evaluation sample. I will
get back to that, but just remember where they are.
This is another format that the statistician came up with, and this actually
plots each data point for each lab for every observation. We calculated the 95 percent
confidence interval for the before or the blue color and for the red or the after, three
samples.
You can see in this case the 95 percent confidence interval shrinks
considerably, and, more importantly, the mean recovery goes from about 65 percent all
the way up to about 95 percent. This is for BOD.
This is the corresponding format for the COD recovery, and here again, it
is the same typical story. It is taking that same data that I showed in that table and
showing you in different formats, and each format tells you a little bit something
different.
-------
630
This is TOC, and this is my old friend, oil and grease. You can see right
now where I think the oil and grease recoveries and precision at these laboratories is
quite acceptable, and now Bill Telliard wants to screw it up by changing the solvent on
me.
This is the similar plot for the pH values. If you look...I told you to look at
those dots on that previous format. This is what we term in East Texas as backsliding.
And for the real pure statistician in the group, I just had to show you
histograms with curves superimposed over them. The top of the chart, the red, is the
before situation or the blind study that I reported on last year, and the bottom or the
blue is the result from the performance evaluation follow-up where we demonstrate the
improved performance.
In this case, the top doesn't even look like a distribution, and now the
bottom is starting to look like a normal curve. That is for BOD.
This is for COD. This is for TOC. Look how tight that thing gets, and
here again, just like I say, it is the same data presented in different formats. Each
format gives you a little different view of what you really accomplished.
And here is oil and grease. It is starting to look like a normal distribution.
MR. TELLIARD: Yeah, we are going to fix that, George.
MR. STANKO: This is the particular format that I like to show our
backsliding. You see those four guys on the left? It is the same point, but here again, it
points out something just a little differently.
The examples presented in this paper demonstrate the us of the quality
improvement process to determine the root cause for analytical problems and the nature
of the corrective action that may be required. In some instances, the root causes may be
difficult to establish and require thorough on-site investigations and assistance, but root
causes can be identified, and proper corrective action can be taken.
The follow-up investigation to assess the effects of sample dilution revealed
that there are two important considerations associated with the dilution of samples. For
real samples, the possibility of false negative observations is an important consideration.
The study also revealed that the dilution of samples down to the detection
-------
631
limit or away from the middle of the portion of the calibration curve may result in less
precise and accurate results, and these are beyond the control of the analyst. There is
nothing he can do about it.
Based on the two problems that can result from the dilution of samples for
organic volatiles by GC/MS purge and trap, it is recommended that samples should not
be diluted unless it is absolutely necessary, and when dilution does occur, one should
bump up the detection limit or quantification limit to reflect the degree that the sample
was diluted.
For the voluntary and cooperative program for the general parameters,
statistical analysis followed by innovative ways to depict the statistical information
resulted in a clear understanding of what had been achieved through the quality
improvement process. The bottom line is that there was considerable improvement for
all general parameters with few exceptions.
The willingness on the part of laboratories to improve their performance,
to participate in voluntary and cooperative programs with well-defined goals should be
noted. The work presented also illustrates the importance of good communications
between contract laboratories and their customers for mutual benefit.
I would like to give recognition to Lesa Rice-Jackson who is a co-author of
this paper. She was also the individual involved in going to the laboratory, and she also
was responsible for conducting the dilution study.
Joyce Wellman is my favorite statistician at the moment. She is a co-
author, and she did all the analysis and plotting of the data in the different formats.
Ron Claybon is an analyst. He ran the GC/MS dilution study. He was
also the analyst that went to the laboratory that had the particular bunch of problems.
And Lillian Thompson is my secretary who prepared the slides for the talk.
Thank you. Any questions?
-------
632
QUESTION AND ANSWER SESSION
MR. MCCARTY: Harry McCarty from Viar and Company.
George, I want to thank you, too, for coming and saying something nice
about something Bill Telliard did.
I am concerned, though, because in October of last year at the IATO
meeting in Denver, I saw a presentation by Mark Carter, and about two-thirds of the way
through, Mark had not identified the client for whom he was doing it, but it became very
apparent it was the same round robin study.
At that point, he let on some of the details of how the study was done.
The problem I see in what you have presented here is that the base
problem in this whole study was that you never or not you but the dummy company that
Mark set up with dummy letterhead and checks, et cetera, never specified what they
wanted the labs to do beyond I need volatiles, I need metals.
As Mark explained the story, he bought what the lab offered. Of the 24
labs, a significant portion, probably on the order of half, had a customer service type
person who said well, why do you want volatiles? Oh, it is for NPDES monitoring or it
is for groundwater monitoring? And talked whoever was on the phone through the
selections, the choices, the menu, if you will, of what they could get.
Other laboratories said sure, send us the sample. Apparently, the labs
were asked to provide shipping containers. Some of them came in cardboard boxes
wrapped in tissue paper and were subsequently broken. There was a wide range of
service that was offered.
The other problem is that, apparently, there was no specification
whatsoever of deliverables as to what the client expected to get.
Not that your quality improvement process didn't help, but I think the root
cause of the problems is you didn't ask for anything very particular in this study. When
people know you want 524.2 or 624 or 1624 or any one of a number of other methods,
they either do that, or they say you do it, and then you deal with the fact well, they really
ran 8260 not 624 or something like that, and you go back and you can deal with those
-------
633
kinds of problems.
But when you don't tell them what you want, it is kind of tough to beat on
the labs for giving you what they felt like sending out that week if you didn't think to
ask.
I am not exonerating the labs for not being good client relations type
people, but I think a lot of the problems that you talked about last year and a lot of the
root causes you are talking about here may be related to the fact that you didn't ask for
something in particular.
MR. STANKO: The response to that is, first, Shell has never identified the
public domain who had done the work, and we choose not to.
MR. MCCARTY: Oh, no, I am not asking that. I am just saying Mark
didn't identify you in his presentation.
MR. STANKO: Okay. Shell has never identified who actually did this
study for us.
The situation was that the study was set up to mimic a filling station that
had leaking storage tanks. The engineering firm who contacted the laboratories told
them or more than implied that this was a cleanup operation around a filling station for
a client. That is why we spiked in gasoline in the samples. Any laboratory has probably
run a lot of gasoline samples, because this happens quite often.
The other thing in defense of what we did, we think what we did was
defined well enough, because we had no problems with the ICP metals.
We had very few problems with the GC/MS for volatiles. There was only
two laboratories, and if you look at the root causes, that had nothing to do with who we
talked to at the lab. These were laboratory problems associated with running the
methodology itself and nothing with defining what was required for the program.
When it comes to the general parameters, the general parameters have
been run for the last 20 or so years, and no one thought we ever had problems with
things like BOD and COD and pH and our friend oil and grease. Anybody in the
petroleum refining industry lives and dies with oil and grease. We didn't know the study
was going to turn out that poorly.
-------
634
Here again, I don't think there was anything we could have said to the labs
in the way of being more specific of what our data quality objectives were to improve
their performance. I think what we got is what you would get on every-day samples, and
this is how it is done every day when engineering firms do work for Shell.
I wanted to assess the data that we have to look at and interpret at times,
and which is very difficult, under realistic conditions. We think that the study that I
reported on last year was conducted in a realistic fashion to really represent what we
have to go through on a daily basis.
MR. MCCARTY: I won't argue that some aspects of it weren't realistic.
My concern is that if your engineering firms that are actually doing work for you aren't
specifying or knowing what methods they expect to get back, whether it be a gasoline,
you know, distribution system that has got to use an underground storage tank type
method from RCRA or a compliance monitoring method for Clean Water Act, if the
engineering firms don't know that, then that is clearly a problem, because if you go in
and tell somebody what you want, you have at least got a better shot, from my point of
view, of getting what you are intending.
MR. STANKO: Most engineering firms don't know to ask for method 8240
or 8260 when they are looking for volatiles. They tell the lab they are interested in
volatile organics for a cleanup operation at a Shell filling station someplace.
That is the best you are going to get from the engineering firms. Not all of
them. Some of them are more knowledgeable than others, but that is the real situation
that you face.
I am looking at it from the lab point of view. If somebody says this is a
groundwater sample associated with a cleanup site, that clearly identifies this as RCRA,
and there should be no ambiguity about whether you are using Method 524.2,624, 1624,
or 8240, or 8260. This is clearly RCRA.
MR.MCCARTY: I would argue...
MR. STANKO: You are not putting the responsibility on the laboratory.
MR. TELLIARD: Excuse me.
MR. MCCARTY: Yes, I am not going to belabor the point.
-------
635
MR. TELLIARD: Time is precious here. We are not going to belabor the
point, but let's point out the fact that, Harry, you just reviewed a whole bunch of
documents from New Jersey all of which were sent in for NPDES, and they all ran
SW846 methods which is against the law. Okay?
MR. MCCARTY: Yes.
MR. TELLIARD: And we told them they were water samples, guys. Now,
this is not the cutting edge of science. This is called economics, outside of the scope of
this conference. Thank you very much.
Moving on.
MR. STANKO: Marlene?
MR. TELLIARD: Hi.
MS. MOORE: I don't know if I can follow this. I am Marlene Moore with
Advanced Systems.
MR. TELLIARD: Hi.
MS. MOORE: George, I know that some of the earlier studies that one of
the things we thought of is that since these samples originally were blinds, I assume that
the WPs and the 02, 03, 04 were knowns to the labs? They knew that these were
essentially performance evaluations of some type?
MR. STANKO: They were known, but they were whole volume samples.
We did not choose to go with concentrates.
So, the only difference between the whole volume samples and the blind
samples is the blind samples had a little clay and dirt and a little gasoline thrown in it.
These samples were still water with the same type of material spiked in. That is the only
difference.
The laboratories did know they were performance evaluation samples, and
there is always some possibility that they put their best people on it.
MS. MOORE: That is what I am concerned about, of course. One of the
things is when a laboratory knows you are paying attention to it and that a particular
client is paying attention to it, it will do very good work for a short period of time.
I am wondering if we can look forward to another blind survey next year at
-------
636
this conference with some more statistics.
MR. STANKO: We feel that the blind performance evaluation studies
conducted the way we have done them is the most effective way to assess what the
laboratories that you are using for routine analysis are doing. That is the only way you
can assess their performance.
MS. MOORE: Should we look for some more next year?
MR. STANKO: I hope so.
I would like to make one comment and one pitch. We have heard a lot
about solid phase extractors and immunoassay, the leading edge. I really wish EPA
would promulgate a paper method for pH. Narrow range pH paper is still the best way
to measure pH, and I have no idea why we can't get it on the books.
MR. TELLIARD: Thank you. Thanks, George. Thanks so much.
-------
637
THE QUALITY IMPROVEMENT PROCESS
AND
ENVIRONMENTAL ANALYTICAL CONTRACT LABORATORIES
Authors: G. H. Stanko
L. M. Rice-Jackson
J. M. Wellman
Shell Development Company
Houston, Texas
Presented at: 15TH Annual EPA Conference on Analysis
of Pollutants in the Environment
Norfolk, Virginia
May 6,7, 1992
-------
638
ABSTRACT
A performance evaluation (PE) study of environmental analytical contract
laboratories used by Shell was conducted late in 1990 and the results from
the study were reported at the "14th Annual EPA Conference on Analysis of
Pollutants in the Environment" on May 9, 1991. The major goal of the PE
study was to assess the performance of a select group of laboratories for
the analysis of groundwater samples for volatile organics by GC/MS, metals
by ICP, and a limited number of general parameters. Results from the PE
study revealed no problems with ICP metals analyses by any of the 24
laboratories. Problems were noted at two laboratories for the analysis of
volatile organics by GC/MS and performances for general parameters such as
oil and grease, pH, BOD, etc. were disappointing. Results from the PE
study were shared with the laboratories to assess their performances and
to initiate corrective action. Using the quality improvement process, the
root causes for the problems at the two laboratories with volatiles
analysis by GC/MS were established and corrective action was taken. The
quality improvement process was also followed to develop a voluntary and
cooperative program with all 24 laboratories to improve performance for
the five general parameters. The results from the cooperative program
showed dramatic improvement in precision for most parameters and
improvements in accuracy for some parameters. Details for the resolution
of analytical problems using the quality improvement process are presented
in the paper.
-------
639
THE QUALITY IMPROVEMENT PROCESS
AND
ENVIRONMENTAL ANALYTICAL CONTRACT LABORATORIES
INTRODUCTION
A blind performance evaluation study of 24 environmental analytical
contract laboratories was conducted in late 1990. The study was conducted
with real matrix samples and the samples were submitted to the commercial
laboratories without identifying them as performance evaluation samples.
The major goal of the PE study was to assess the performance of a select
group of laboratories for the analysis of groundwater samples for volatile
organics by GC/MS, metals by ICP, and a limited number of general
parameters. Results from the study were reported at the "14TH Annual EPA
Conference on Analysis of Pollutants in the Environment" on May 9,
Results from the PE study revealed that there were no problems with ICP
metals analyses by any of the 24 laboratories. The precision and accuracy
for the ICP metals was outstanding for all participating laboratories. The
overall mean recovery for all eleven metals was 97%. Among the individual
metals, the means for all laboratories ranged from 89% to 107%. With such
good performance, it was not necessary to take action to improve
performance at these laboratories for metals analysis by ICP.
Most of the laboratories included in the study performed well for the
volatile organics by GC/MS, but some problems were noted at two
laboratories. Both laboratories did not find a number of the target
analytes in the blind samples. These would be considered false negative
observations. The problem was serious enough to require immediate
corrective action. The laboratories were contacted by phone and told of
their poor performance. They were also provided with the necessary sample
identification and the true values for the blind samples.
Volatile Organics - First Laboratory
One of the laboratories quickly established the root cause for their poor
performance. Their review of the raw data revealed that the samples were
screened using a GC/FID method and a decision was made to dilute the
samples prior to GC/MS analysis. The laboratory routinely screens samples
and dilutes them to reduce instrument contamination and associated
analytical problems. The blind samples used for the study were prepared
in such a way to have a natural background that was picked up by the
GC/FID screening procedure; however, these samples should not have been
diluted. This was the only laboratory that had diluted the samples prior
to analysis. In this instance, the laboratory had diluted the samples to
a point where the analyte concentration was near detection limits and in
some instances below detection limits. The root cause for the false
negative observations was that the laboratory diluted sample to a level
where approximately half of the analytes were below their detection limit
-------
640
Dilution Study
The problem with dilution of samples for volatile organics by GC/MS was of
concern to Shell. A study was designed and conducted to determine if
there were other problems that could result from dilution of samples. It
has been widely recognized that there is some degree of analytical
variability associated with all measurements. As the analyte
concentration approaches the detection limit, the analytical variability,
expressed as the coefficient of variation (CV) (Note: CV = 100 x standard
deviation/mean), usually increases at a rapid rate. While dilution of
samples is allowed by EPA approved protocol[2], little information was
available to assess how dilution might impact laboratory performance or
analytical variability.
Eight common volatile analytes were selected for this study. The analytes
were methyl-tert-butyl ether (MTBE), diisopropyl ether (DIPE), chloroform,
benzene, toluene, ethylbenzene, para-, meta- and ortho-xylene. Spiked
solutions of the analytes in distilled water were prepared at 5, 10, 20
and 40 ppb concentration levels. The 40 ppb spiked solution was diluted
by a 1/20 dilution factor to represent the "diluted" sample. Each
concentration level was analyzed eight consecutive times, and the mean,
standard deviation, and coefficient of variation were calculated for each
set. Systematic errors were minimized as much as possible. Only one
trained technician performed all the analyses.
Special care was taken with the samples and no other samples were run on
the instrument during this study. The calibration curves used for the
study were prepared using calibration standards at the 1, 5, 10, 20, 50,
100, 150, and 200 ppb levels. The concentrations of all samples were
within the calibration curve including the "diluted" sample.
The data from the study are summarized in Tables 1 and 2. The means and
standard deviations of the analytes are given in Tables 1. The
coefficients of variation are listed in Table 2. As can be seen from
Table 2, the CV's are all < 10% except that of toluene. Unusual problems
with toluene have previously been noted during several PE studies and no
adequate explanation has been found for its unusual behavior [1,3,4,5].
To better illustrate the trends observed, comparative plots of the data
are shown in Figures 1 and 2. In Figures la and Ib, the results for MTBE
are shown. As expected, the estimates of the standard deviation increase
with increasing concentration. However, for the corresponding CV plot, no
particular trend was observed. This means that precision, when expressed
as a percent of the mean, is fairly constant over the concentrations used
for the study.
Figures 2a and 2b show similar trends for benzene, except for the 5 ppb
standard. The 5 ppb points did not seem to fit as well. In this case, the
estimate for CV was less than those calculated for the three higher
concentrations. This is not what is normally observed or expected [6]
The observed estimate of CV for the 5 ppb point may have resulted from the
greater uncertainty in both the estimate of mean and standard deviation
which led to a somewhat lower CV.
-------
641
The results for the 5, 10, 20, and 40 ppb spiked samples show that the
purge and trap GC/MS method has the capability to measure the
concentrations of the eight selected analytes with good precision and
accuracy in the distilled water matrix. The coefficients of variation for
all concentrations and analytes were less than ten percent with toluene
being the only exception. Based on the observed CV's from this study, one
would conclude that the method is quite capable of measuring the
concentrations of these analytes down to the 5 ppb level and the
analytical variability is relatively constant over this range and under
the ideal conditions of the study.
The estimated means and standard deviations shown for the "diluted" sample
in Table 1 were on the basis of observations that were corrected for the
dilution. The actual "diluted" sample contained nominal 2 ppb levels of
the analytes. The corresponding CV's shown in Table 2 are also on the
same basis.
Comparison of the means (see Table 1.) for the 40 ppb spiked sample and
the "diluted" sample for each of the analytes revealed that in every
instance the means for the "diluted" sample were biased higher.
Comparisons of the estimated standard deviations also showed a similar
bias with toluene being the only exception. Since the spiked and diluted
samples should be compared for each compound, this is a paired comparison.
A 2-tailed sign test can be performed to assess the probabilities of
observing 8 out of 8 higher means and 7 out of 8 higher standard
deviations purely by chance (see Table 1.). The test results indicate
that the chance probability (p-value) for the the means is less than 0.01
and for the standard deviations is less than 0.05. Thus, we concluded
that diluting of samples prior to analysis resulted in less precise and
less accurate results.
The reason for the less precise and less accurate results is quite simple.
In the case for the 40 ppb spiked samples, all of the concentration
estimates were calculated from response levels that were near the center
of the calibration curve. The uncertainty and inaccuracy associated with
establishing the location of the calibration curve are at a minimum near
the center and increase considerably at the two extremes. The response
levels for the "diluted" sample were all very close to the lower extreme
(1 ppb) of the calibration curve.
There is another important consideration associated with dilution of
samples. In the above study, all concentration levels were designed to be
within the range of the calibration curve. For real samples or for the
blind samples used for the recent PE studyfl], the concentration of
analytes are rarely the same. What actually was observed in the PE study
was that a laboratory had diluted the sample to the point where some of
the target analytes were present in the diluted samples at concentrations
below the detection limits. In these cases the laboratory had false
negative observations because of the dilution. For real samples the
possibility of false negative observations is an important consideration.
It was concluded from the results from this limited study that the
analytical variability was relatively constant over the concentration
range studied and with the conditions of the study. The study also
-------
642
revealed that the dilution of samples down to the detection limit or away
from the middle portion of the calibration curve may result in less
precise and less accurate results for reasons that are beyond the control
of the analyst. A second problem associated with the dilution of samples
is the possibility of diluting the sample to a point where target analytes
are at concentrations below the laboratory detection limit. Based on
these two problems that can result from dilution of samples it is
recommended that samples for volatile organics by purge and trap GC/MS
analyses should be diluted only when it is absolutely necessary because of
a matrix or interference problem. When dilution of samples is necessary
the quantification limits need to be adjusted by the dilution factor to
reduce the possibility of false negative observations and to reflect the
true capability of the analysis.
Volatile Organics - Second Laboratory
Establishing the root cause for problems with the volatile organics
analysis at the second laboratory was much more complicated The
laboratory reported they had reviewed their raw data and had established
the reason for the false negative observations was that the holding time
had been exceeded by two days. Their corrective action was to make sure
to analyze samples within holding times. This explanation was not
acceptable to Shell. In discussions that followed Shell agreed to
provide some additional spiked standards so the laboratory could
demonstrate they had established the root cause and was now back in
control. The performance of the laboratory for the standards was just as
poor as with the blind samples.
Shell agreed to have two of their analysts visit the laboratory in an
It i u ?stabllsh the ro°t cause of the poor performance and to assist
the laboratory get back into analytical control. The two analysts were
on-site for two days and did eventually establish a long list of root
causes. Improper vials were used to store standards. Only one set of
standards was available and the set did not have all analytes Poor
technique was employed to load the volatile samples. The samples were not
at room temperature when loaded. Improper syringes were used and these
were not calibrated Errors from 0.5 to 1.0 ml were noted Too many
files were in the data system which led to confusion. Calibration and
quantification files contained different response factors and the wrong
procedure file was being used to update response factors. Variation in
response factors as high as 60% were observed. The calibration range was
±i!?T (22 ,t0, 8°°TuPpb)' Poor Jud9*nent was used when a callbKS
should be updated. There were problems with missing peaks which had to
manual y be identified and this Was not always being done Reviewinq and
whTh^iH^65111!15 WaS inadequate' There was a general lack 7f QVQC
which would have eliminated many of the observed problems.
The two Shell analysts worked with the laboratory analyst to correct the
reUsul?sUS Pfrn°rblT t0 a r11? lm the ^boratory couldyachieve acclptabVe
results for known standards. The laboratory agreed to Include
whole-volume PE samples from Environmental Resource Associates (ERA) nto
their QA program and to share the results with Shell. Results from these
PE samples have demonstrated .analytical control over the past yea™ with *
-------
643
few minor problems being noted. This example of the quality improvement
process demonstrated the extent of involvement that may be required to
achieve data quality objectives particularly if a given laboratory has
been recognized as being essential for Shell's operations.
General Parameters
The results for the PE study [1] for the general parameter were
disappointing. In general the performances at many of the laboratories
for parameters such as pH, oil and grease, BOD, TOC, and COD were
sufficiently poor to require corrective action to eliminate the obvious
problems and to improve the overall performances of all laboratories for
these parameters. Because of the number of laboratories involved and the
total number of general parameters where poor performance was observed, it
was decided to ask laboratories to participate in a voluntary and
cooperative program where Shell would pay to have whole-volume PE samples
shipped directly from Environmental Resource Associates (ERA) to the
laboratories for analysis, and the laboratories would cover the cost of
the analyses. Both Shell and the laboratories would share the resulting
data which would be statistically analyzed by ERA. The program would run
for six months and the PE samples would arrive every other month. The
overall objective of the program was to allow laboratories the opportunity
to establish root causes for poor performance and allow them to take the
appropriate corrective action necessary to improve performance.
Hopefully, the third PE samples would establish good performance at all
laboratories.
All 24 contract laboratories agreed to participate in the voluntary
program. Their response indicated that laboratories are concerned with
data quality and do try to improve their performance. They are also
willing to be active participants in programs that should lead to improved
performance. It also demonstrates the use of the quality improvement
process where the laboratory and laboratory customer work together to
achieve some mutually agreed upon goal. In this case, the goal was
improved performance.
The performance from the initial blind study was used as the benchmark for
the voluntary program. The results from the initial blind study and the
three whole-volume samples were statistically analyzed and the data were
plotted in a number of different ways. The number of outliers for all of
the data are summarized in Table 3. The data were further summarized
(outliers removed) in Table 4. Table 4 compares the results from the
blind study (Before) with the results from the voluntary program (After).
Table 4 shows the average recoveries for all laboratories and the standard
deviations for the means. The question of whether there was a significant
reduction in variability was yes for all parameters and the corresponding
11P" values were included in Table 4.
There are a number of ways to plot the data and each way provides some
perspective for laboratory performance. The data were plotted as
"boxplots" and Figures 3 to 7 illustrate the data in that format. A
boxplot divides the data into quarters. Fifty percent of the values fall
in the interval between the lower and upper edges of the box, with 25% of
-------
644
the data above (and 25% below) the horizontal line within the box. One
fourth of the values are below the lower edge of the box. This format
summarizes the performance for all laboratories for each of the five
samples. The boxplots showed improved performance for the "Before" and
"After" comparison.
Another way to show the results is by plotting the percent recovery for
each observation for the five samples. For example, percent recoveries
for BOD are displayed in Figure 8. Note that this is not a control chart,
but it is an effective display of the improved performance of the
laboratories after corrective action was taken. It provides a more
detailed visual comparison of the variability of the laboratory results
before and after corrective action as well as portraying any shift in mean
recovery (accuracy) that may have occurred. Figures 8 to 12 show the data
for each of the general parameters. Also shown on the plots are the means
and 95% confidence intervals that were calculated on a "Before11 and
"After" basis. Again, there is clear evidence of improved performance.
The data were also plotted as histograms with normal distribution curves
superimposed onto the histograms. Figures 13 to 17 show the data in that
format. Again, the comparison was made on a "Before" and "After" basis.
CONCLUSIONS
The examples presented in this paper demonstrate the use of the quality
improvement process to determine the root causes for analytical problems
and the nature of the corrective action that may be required. In some
instances, the root cause(s) may be difficult to establish and require
thorough on-site investigations and assistance, but root causes can be
identified and proper corrective action can be taken.
The follow-up investigation to assess the effects of sample dilution
revealed there are two important consideration associated with dilution of
samples. For real samples the possibility of false negative observations
is an important consideration. The study also revealed that the dilution
of samples down to the detection limit or away from the middle portion of
the calibration curve may result in less precise and less accurate results
for reasons that are beyond the control of the analyst. Based on the two
problems that can result from dilution of samples, it is recommended that
samples for volatile organics by purge and trap GC/MS analyses should be
diluted only when it is absolutely necessary because of a matrix or
interference problem. When dilution of samples is necessary, the
quantification limits need to be adjusted by the dilution factor to reduce
the possibility of false negative observations and to reflect the true
capability of the analysis.
For the voluntary and cooperative program for general parameters,
statistical analysis followed by some innovative ways to depict the data
resulted in a clear understanding of what had been achieved through the
quality improvement process. The bottom line is that there was
considerable improvement for all general parameters with few exceptions.
The willingness on the part of laboratories to improve their performance
-------
645
and to participate in voluntary and cooperative programs with well defined
goals should be noted. The work presented also illustrates the importance
of good communications between contract laboratories and their customers
for mutual benefit.
REFERENCES
1. G. H. Stanko, "Performance Evaluation Study of Environmental
Analytical Contract Laboratories", Proceedings of EPA 14TH Annual
Conference on analysis of Pollutants in the environment, Norfolk,
Virginia, May 8,9, 1991.
2. "Test Methods for Evaluating Solid Waste:, EPA Methods Manual SW-846,
Third Edition, November 1986.
3. G. H. Stanko, and R. W. Hewitt, "Performance Evaluation of Contract
Laboratories for Purgeable Organics", Proceedings of the EPA 12TH
Annual Conference on Analysis of Pollutants in the Environment,
Norfolk, Virginia, May 10,11, 1989.
4. G. H. Stanko, "Round Robin Study of EPA Methods 624 and 1624 For
Volatile Organic Pollutant", Proceedings of the EPA 6TH Annual
Seminar for Analytical Methods for Priority Pollutants, Norfolk,
Virginia, March 16,17, 1983.
5. G. H. Stanko, "Analysis of Petrochemical Wastewaters for Volatile
Organic Pollutants", Proceedings of the EPA Seminar for Analytical
Methods for Priority Pollutants, Hershey, Pennsylvania, April 9,
1981.
6. "Calculation of Precision, Bias, and Method Detection limit for
Chemical and Physical Measurements", Issued by Quality Assurance
Management and Special Studies Staff Office of Monitoring Systems and
Quality Assurance Office of Research and Development United States
Environmental Protection Agency, Washington, D.C. (March 1984).
-------
646
[Blank Page]
-------
647
li
mm*
O
(/)
Z
o
1-
H
^ i/)
UJ Z
O 3
Q
tt 00
O Z
Z O
< a
H U.
oo
(/)
0 0
z z
< 3
o
u) a.
z z
< 0
UJ CJ
J*^
CO
Q.
Q.
"•^
z
o
M
H
^
a
H
z
UJ
{_)
z
o
o
^
•5
^^
o
•^
o
o
^
o
o
fs
^
0
o
in
U9
x
M
X
M
f^t
V)
X
(A
X
^
in
*""*
»-H
in
ON
r>
ON
0
^
so
so
'
SO
SO
O
r-
i~^
r-^
1-H
t~-
<*>
f>
O
r^
ON
•^
*^
O
ON
"•J
m
m
•
o
SO
ON
1-H
in
^H
00
ON
so
^H
fv|
1-H
W
f*l
in
^
1-H
^'
f)
<*>
00
f)
IS
1-H
1*)
•^
0
fM
f*^
00
o
f>
in
o
1-H
-------
648
I
0
I-
u.
u.
Ill
o
o
CM
LU
-J
CO
ec
H
Z
UJ
O
Z
o
o
X— ~
^_
o
o
0
o
p
c
o
o
p
Compound
O oo ^ w>, — «
ON OO rs| ON O
rj" — CTN r~- rsi
-^ NO oq csj TJ-
CO <"s| fs| ^ £5
— < r- ON oo —•
r-^ co co »/•» oo
co cO cO if) ^3"
— • «0 (— fsj ON
OO
-------
649
TABLE 3
Outliers by Study and Substance
BLIND1 BLIND2 ERA-WP02 ERA-WP03 ERA-WP04
BOD
COD
TOC
O&G
pH
3
3
1
0
0
2
2
2
0
0
0
1
1
3
1
1
0
1
3
0
1
0
0
2
0
-------
650
TABLE 4
Summary Results (Outliers Removed)
#of
Test Study Samples Avg. % Rec.
Significant
Reduction in
SD Variability p-value*
BOD
BOD
COD
COD
TOC
TOC
O&G
Q&G
Test
pH
pH
^^^^^
Before
After
Before
After
Before
After
Before
After
Study
Before
After
42
66
42
69
47
68
43
70
#of
Samples
50
74
69.9%
96.8%
103.4%
95.9%
101.0%
99.4%
61.6%
89.8%
Avg. DifT.
0.15
0.02
32.2% yes
15.7%
36.8% yes
11.9%
25.1% yes
5.2%
32.6% yes
10.7%
Significant
Reduction in
SD Variability
030 yes
0.17
f T »*m *«^»
< 0.000001
< 0.000001
< 0.000001
< 0.000001
p-value*
0.000005
P-value is from testing the hypothesis that the variance Before is
equal to the variance After. A small p-value (typically less than
0.05) means that It is highly unlikely that the variances are the
same. The smaller the p-value, the stronger the evidence that the
variance of the lab results After is smaller the variance Before.
-------
651
Figure la
Standard Deviation vs. Cone, of MTBE
Standard Deviation (ppb)
a
a
Concentration (ppb)
Figure Ib
Coefficient of Variation vs. Cone, of MTBE
Coefficient of Variation (%)
30
Concentration (ppb)
-------
652
FIGURE 2
Figure 2 a
Standard Deviation vs. Cone, of Benzene
Standard Deviation (ppb)
Concentration (ppb)
Figure 2b
Coefficient of Variation vs. Cone, of Benzene
Coefficient of Variation (%)
Concentration (ppb)
-------
653
0>
0)
u
0)
a:
O
m
175
150
125
100
75
50
25
0
FIGURE 3
BOD Recovery Percentages
Comparisons Study
Blind 1 Blind2 WP02 WP03 WP04
(n=21) (n=21) (n=24) (n=22) (n=20)
-------
654
FIGURE 4
250
190
! 160
§
I 130
10 -
COD Recovery Percentages
Comparisons Study
^ ^
Blind 1 Blinda WP02 WP03 WP04
(np»21) (n=2!) (n=23) (n
-------
655
0)
0>
s
u
0)
a:
o
o
150
125
100
75
50
FIGURE 5
TOC Recovery Percentages
Comparisons Study
25
0 -
Blind 1 Blind2 WP02 WP03 WP04
(n=24) (n=23) (n=23) (n=22) (n=23)
-------
656
FIGURE 6
Oil and Grease Recovery Percentages
Comparisons Study
150
125
I ioo
-------
657
FIGURE 7
pH Differences Between Observed and Made—To Values
Comparisons Study
1.2
0.8
0.4
0)
u
c
a>
L_
0>
X
a.
-0.4
-0.8
-1.2
Blind 1 Blind2 WP02 WP03 WP04
(n=25) (n=25) (n=25) (n=25) (n=24)
J
-------
658
TJ
T^
.n. • i
JJI | |9
' •«? i! Q' i ,'u'j G
i!! i! %, \ •»! ;
* &\! •! J I • A s;
IA'I iu v r^
6 * ^ s;
6 1 ^6
O (v>
+2S
±
'
0
27
54 81
Observation
Blind 1: Obs. 1-27
Blind2: Obs. 28-54
WP02: Obs. 55-81
WP03: Obs. 82-108
WP04: Obs. 109-135
rr Sample Avq,
!!<& ^
-2S
108
135
-------
659
FIGURE 9
240
200
^a 160
0
L-
0)
o
I 120
Q
O
O
80
40
0
COD Recovery Percentages
Comparisons Study
h
£
»f*
0
27
!
+ 2S
Sample Avg.
-2S
i
»
1 . . 1 . , 1 , . 1 , ,
1 1 1
54 81
Observation
108
135
Blind 1: Obs. 1-27
Blind2: Obs. 28-54
WP02: Obs. 55-81
WP03: Obs. 82-108
WP04: Obs. 109-135
-------
660
FIGURE 10
0)
v_
(D
O
u
180
150
120
90
O
2 60
0
0
TOC Recovery Percentages
Comparisons Study
+ 2S
Sample Avg.
-2S
27
54 81
Observation
108
135
Blind 1: Obs. 1-27
Blind2: Obs. 28-54
WP02: Obs. 55-81
WP03: Obs. 82-108
WP04: Obs. 109-135
-------
661
FIGURE 11
Oil and Grease Recovery Percentages
Comparisons btudy
-Q
0)
0
QJ
a:
j.
^
0
o
150
120
QO
y \j
60
30
0
m»
-
J
•
•
*
\
i j
,':
) ; :
' '
,
ii '
;i ;,' Oij
i * 'ii
' d i:!
; 'id
: i ^
^ 6 i
w
j C
i 9
• 1
1 I
1 1
!' •
'A
I
1
1
ii
!|
1
1
>6
: f f
O?. /I Q O
? 9 rt 99 T? .'',?%• Q
1 c»* 9 5 1 -. ! i ' ', 'to rf «' T
> o ?." o,"i ^15 %# '"ii >jJfi:»9il9^ ®J*#o
> W ji« wii ^|3> ^»ij fl' liml'f Si' *"P®1 } P" *?P* \i4*l \
?!i r n'ljl?! ^v K iili^^***
,'• •' : i • . O •I1'!1 ''* /k I1
• " ! ' ' CA i' ' ™ JL
;': i! . : j JP.L1 J 6
, ;! «. : / ^ j |
1 1 ' x •' I ' • •
i t > ^j J Vi) •
; 'j 6 1 '•
,' 16 ® :
i i !
; ' $ |
i , (J> .
: i
ID :
+2S
Sample Avg,
-2S
-30 -
0
27
54 81
Observation
108
135
Blind 1: Obs. 1-27
Blind2: Obs. 28-54
WP02: Obs. 55-81
WP03: Obs. 82-108
WP04: Obs. 109-135
-------
662
FIGURE 12
pH Differences Between Observed and Made—To Values
Comparisons Study
1
0~7R
. /D
0.5
o> 0.25
u
c
0)
k_
Q
31
°- -0.25
-0.5
-0.75
-1
t
i
t
~ t
' t j
: 1 ?,? ii i
:" ? it i I ii
,«, •'! 5 •' :! ,': f
n ,'; ^ l '! • i'
"" 1 1 " ' • i i ' : T 1 ' ®
* 'T^*CN W1 '! •
-------
663
FIGURE 13
BOD Recovery Percentages
Comparisons Study
0)
0.18
Blind 1 and Blind2
20 40
60 80 100 120 140 160 180
BOD (% Recovered)
WP02, WP03, WP04
u
C
Q>
0)
0)
0.18 -
0.15 -
0.12 -
0.09 -
0.06 -
0.03 -
0 -
60 80 100 120
BOD (% Recovered)
140 160 180
-------
0.4 -
u
c
o>
D
CT
0)
0)
0.2
•« 0.1
WP02, WP03, WP04
i » » » i * * * i » » » i » » » i » «
* i ... i
0 20 40 60 80 100 120 140 160 180 200 220 240
COD (% Recovered)
-------
665
FIGURE 15
TOC Recovery Percentages
Comparisons Study
Blind 1 and Blind2
20 40
60 80 100 120 140 160
TOC (% Recovered)
0.4 h-
£ 0.3
o>
cr
0)
a>
0.2
0.1
okr
After
"n"="68
WP02, WP03, WP04
J ! ! ! 1 i » i t » I i i i I i i t I i
0 20 40 60 80 100 120 140 160
TOC (% Recovered)
-------
666
FIGURE 16
Oil and Grease Recovery Percentages
Comparisons Study
Blind 1 and Blind2
0.3 -
c
0)
3
cr
0)
0)
0)
0.1 -
0.05 -
0 20 40 60 80 100 120 140 160
0 & G (% Recovered)
WP02, WP03, WP04
c
a>
D
O"
Q)
a>
.>
*-M
_g
0)
0.1 -
0.05 -
0 L
0 20 40 60 80 100 120 140 160
0 & G (% Recovered)
-------
667
FIGURE 17
pH Differences Between Observed and Made-To Values
Comparisons Study
c
a>
0)
.>
U->
_D
0)
0.3 t-
0.25
0.2
0.15
0.1
0.05
0 L
-1.2 -0.8
Blind 1 and Blind2
-0.4 0 0.4
pH Difference
0.8
1.2
WP02, WP03, WP04
-0.8
•0.4 0 0.4
pH difference
0.8
1.2
-------
668
[Blank Page]
-------
669
MR. TELLIARD: We knew at this point in the program you would have
been dulled into kind of a maze by numbers, facts, and science. So, now we are going to
go into witchcraft and alchemy. We are going to talk about methods detection limits,
levels of quantitation, and other artificial forms of reality that you can do.
Larry Keith is here, and he is from Radian Corporation, and what hat are
you wearing today, ACS or Radian, or what uniform have you got on? A contractor for
EPA? Which one?
MR. KEITH: Let's do Radian today.
MR. TELLIARD: Okay, he is a Radian today. Larry, you are on.
MR. STANKO: Excuse me. I would like to break in. There is a box that
says Xerox on here that has 100 copies. Just leave your money on the side.
MR. TELLIARD: Okay.
-------
[Blank Page]
-------
671
MR. KEITH: Bill and I first talked about the problems with MDLs about
two years ago over in Germany when we were at the Dioxin '90 meeting. We discussed
this science and the problems with MDLs, and we recognized that there are a lot of
problems with them. Since that time, I have been working with the American Chemical
Society to help resolve these problems.
The definition for MDL came about in 1983, and it came out of the Office
of Drinking Water. The entire EPA essentially adopted the term.
However, the problem is that what was designed for drinking water
perhaps doesn't always work well for other water matrices (for example sewage).
Sewage, usually, does not closely resemble drinking water. Soils and other matrices are
even farther away in terms of their resemblance to drinking water.
Thus, we are going to try to come up with a generic definition of method
detection level, (which includes a name change) that, hopefully, everybody can agree
upon. It will be a generic definition, one that doesn't tell you how to do it, so the
protocol won't change, just the definition of the terms that everybody can agree upon.
What are the problems? [FIGURE 1] First of all, there is some unclear
language in the definitions, and there are multiple definitions [FIGURE 2].
The ACS reliable detection level and the ASTM's limit of detection are the
same concept. And the ACS term limit of detection and EPA's MDL are very closely
related.
There is a general lack of sufficient guidance in using statistically based or
related terms by statistically inexperienced users.
Our objective is to derive some revised consensus definitions that will
provide: clearer definitions, fewer definitions, some accompanying recommendations,
some accompanying usage guidance, and be clearly interrelated terms so that we keep
things simple. [FIGURE 3]
We want these definitions to be widely encompassing, broad in their scope,
so that all of the environmental situations are covered. And we want them to have
broad support and acceptance within the entire scientific and technical community; those
-------
672
who are the regulators and those who are the regulated.
First, let me review the difference between the level of detection (or the
limit of detection) and the method detection limit. [FIGURE 4] Really, the only
difference is the point of reference.
With the limit of detection (or level of detection), a blank that is the
lowest concentration that can be determined to be statistically different from a blank at a
specified level of confidence is used and the method detection limit, MDL is the lowest
concentration that can be measured with 99 percent confidence if the analyte
concentration is greater than zero.
Thus, there are two differences between the MDL and the LOD: (1) a
confidence level is specified, and (2) zero is as the reference point with the MDL. The
LOD does not specify a 99 percent confidence level (it recommends one) and it uses a
blank as the reference point.
The recommended level in each case is 3 sigma, (3 standard deviations).
These are basically levels where there is a binary yes/no decision that is made
concerning whether or not an observed signal represents the presence or absence of an
analyte.
Thus, the LOD takes into account any background interferences; the MDL
doesn't.
There are three kinds of problems with the EPA definition of MDL. (1) It
doesn't say what it means, (2) it is too specific and limiting and can paint EPA into a
corner), and (3) it doesn't consider matrix effects when these are a problem. [FIGURE
5]
Let's discuss each of these problems. First, the use of "limit" is a
misnomer. [FIGURE 6] When limit, is used it usually implies that you can't or
shouldn't get below something, and that is not a true reflection of its use here.
Webster defines "limit"as "...impliessetting a point or a line beyond which
you cannot or are not permitted to go" whereas "level" is "...themagnitude of a quantity
considered in relation to an arbitrary reference value".
The values calculated by an MDL are arbitrary reference values. They are
-------
673
not magic numbers. They are calculated, and they are also indirectly selected. They are
selected based on the amount of confidence you want in elimnating false positives.
One person's confidence level of 99 percent might be a different person's
confidence level of 95 percent. So, they are not magic numbers written in stone. They
are selected and they are arbitrary.
Thus, "L"should stand for "level",because that is really what we mean, and
I think we understand that when we are talking among ourselves. [FIGURE 7]
However, although we may understand that we are really talking about "level",not
everybody does.
The recommended 3 sigma above zero was arbitrarily chosen, because it
provides greater than 99 percent confidence of eliminating false positives, However, for
some cases less confidence is needed.
Therefore, we are going to try to change the name to "method detection
level".
The second problem is that EPA's definition of MDL is too limiting. It
doesn't take into consideration situations where there are statistically significant
background concentrations of an analyte, so you can't subtract background interferences.
[FIGURE 8]
that really wasn't the intention when the 1983 definition was promolgated.
But, that is what it has evolved to, and that is the way that people use it. One does not
usually subtract background in making MDL calculations.
We believe we can derive a single definition that allows subtraction of
background signals when they are significant and that will use the arbitrary zero
concentration when they are not. And we propose not to have a 99 percent confidence
level in a generic definition.
The third problem was one of not considering matrix effects. [FIGURE 9]
The procedure for making MDL calculations is to take an analyte of interest and spike it
into reagent water at 2 to 5 times above the instrument detection level (or estimated
instrument detection level) and analyze it about seven or eight times. Then practical
quantitation limits (PQLs) are estimated by multiplying the MDL times arbitrary values
-------
674
that are dependent on how complex the matrix is.
EPA has come under criticism for generating PQLs in that rather arbitrary
way.
Thus, EPA's desire is to replace PQLs with some revised definitions that
have a stronger technical basis.
Thus, the objectives are: (1) to replace "limit" with "level";(2)
accommodate both zero and background-subtracted signals; (3) accommodate various
confidence levels, (rather than only 99 percent); (4) use a representative matrix when
making analytical measurements when it is appropriate (and don't when it is not
appropriate or when you can't); and, (5) provide some guidance on the use and
limitations to accompany the revised definition. [FIGURE 10]
The draft revised definition is: A method detection level is the lowest
concentration at which individual measurement results for a specific analyte are
statistically different from a blank, that might be zero, with a specified confidence level
fora given method and representative matrix. [FIGURE 11]
The MDL is based only on the risk of false positive detections (i.e., is the
analyte correctly identified as present or not). There are also two other definitions that
need to be considered: (1) a reliable detection level (RDL) which is based on the risk
of false negative detections and (2) a reliable quantitation level (RQL). [FIGURE 12]
These definitions, the correctness of an analyte identity. That is a different
question not addressed by MDLs.
When is an analyte present? [FIGURE 13] that is one of the most
important decisions in low-level analyses. The first question that has to be answered is if
an analyte is there or if it is not.
This is a binary decision, and there are two types: detected or not detected.
[FIGURE 14] There are also only two true solutions: an analyte is either present or it
is absent.
If an analyte is detected and it is present, then that is a correct decision.
Likewise, if an analyte it is not detected and it is not present, that is also a correct
decision.
-------
675
The other two situations are where there is a false positive, (an incorrect
decision that an analyte has been detected) or a false negative (an incorrect decision that
an analyte is absent).
Consider a sample where an analyte of interest is absent. [FIGURE 15]
Theoretically, if the instrument used for anaylsis can give negative values, an equal
probability of positive and negative results will be obtained assuming a normal Gaussian
bell-shaped distribution. [FIGURE 16]
The next example has a concentration of an analyte, equal to the
recommended MDL of 3 standard deviations. The same situation will occur where there
is a 50 percent probability of detect or non-detect above or below that concentration to
provide a similar Gaussion distribution.
That means that there is a 50 percent probability of a signal falling below
the MDL, in which case, it wouldn't be reported. Thus, there is a 50 percent probability
of an incorrect decision, (a 50 percent probability of a false negative). [FIGURE 17]
Therefore, although the false positive probability is less than 1 percent, the
false negative probability is 50 percent. [FIGURE 18] That is not a very reliable
detection level!
What would be a reliable detection level? It is twice the method detection
level. At that concentration, there would be a very low statistical probability of either a
false positive or a false negative detection. [FIGURE 19]
Therefore, if reliable detection levels (RDLs) are desired, you have to
establish them at a higher concentration than the MDL. It may not have to be twice the
MDL, but the recommendation is going to be that an RDL be twice the MDL (in the
absence of appropriate data) so that there will be a very low probability of a false
negative decision [FIGURE 20].
If an RDL is set at twice the MDL and the recommended 3 standard
deviations as an MDL is used, then the RDL would simply be 6 standard deviations.
That is pretty simple and everyone can understand it.
In Figure 21, the true analyte concentration is to 6 standard deviations, and
there is an approximately equal overlap in the small black areas which represent less
-------
676
than 1 percent of the alpha and the beta type errors, (false positive and false negative
errors).
The table in Figure 22 shows that there is nothing special about the values
selected. The probabilities of false positives and false negatives are all statistically
based. If 3 standard deviations is selected for an MDL, the risk of false positives is
approximately 0.1 percent, but the risk of false negatives is 50 percent, as previously
discussed.
If one wanted a 1 percent risk of false positives, 2.33 standard deviations
could be selected. Then a corresponding RDL would be 4.66 standard deviations. If a
95 percent confidence level was sufficient, one could select with 1.64 standard deviations.
Thus, you can select any appropriate confidence level -- it does not have to be based on
3 standard deviations for an MDL (although that is the recommended value).
Therefore, for a reliable detection level when using a given method and
representative matrix, a single analysis should consistently detect analytes present in
concentrations equal to or greater than the reliable detection level. [FIGURE 23]
When sufficient data are available, which will probably be infrequent, the
RDL is the experimentally determined concentration at which false negatives and false
positives are specified. Otherwise, (most of the time), the RDL is the concentration
which is twice that of the method detection level.
The RDL is the recommended lowest level for qualitative decisions based
on individual measurements, and it provides a much lower statistical probability of a
false negative determination than the MDL.
To this point, we have only discussed qualitatively identifying an analyte of
interest. The next consideration is how to reliably quantitate low concentration levels.
[FIGURE 24]
In general, the higher the concentration of an analyte in a representative
environmental matrix, the more reliably you are going to be able to calculate its
concentration or to measure its concentration. Unlike the definitions of the MDL and
the RDL, statisticians remind us that when a quantitation is based on an individual
measurement, there is no direct statistical basis for its derivation. That is a significant
-------
677
differce from the situation with the MDL (and the corresponding RDL).
In 1983, the American Chemical Society recommended 10 standard
deviations for establishing a reliable quantitation level (RQL) at or above an average
blank signal. [FIGURE 25]
The proposed revised definition is that the RQL be 2 times the RDL when
based on individual measurements. This accommodates various RDLs that can be used.
It also translates to a value of 4 times the MDL. It is an arbitrary factor, but at least it
provides a consistent relationship between the proposed definitions of MDL, RDL, and
RQL.
Thus, the draft definition for the RQL is: The RQL is the recommended
lowest level for quantitative decisions based on individual measurements for a given
method and representative matrix.
The RQL is the concentration which is 2 times the reliable detection level,
and it recognizes that the RDL estimates produced at different times by different
operators for different representative matrices will not often exceed the RQL. [FIGURE
26]
The EPA plans to publish these proposed definitions along with some
additional historical and background information and some questions in the Federal
Register in August, 1992.
Notice that throughout the definitions of MDL, RDL and RQL, we have
used the words "representative matrix." That was placed in the definitions because it is
recognized that complex matrices may cause some significant analytical effects. When
this occurs, a matrix related problem can be accommodated, if that is possible. Thus,
reagent water will not have to be used to try to represent, for example, a petroleum
refinery or a paper mill refinery or a sewage effluent. One can use a sewage effluent, or
a petroleum refinery effluent to represent the appropriate matrix and that should be
more accurate than using reagent water to represent those or other kinds of industrial
effluents and other comoplex matrices (including soils and sediments).
The proposed definitions also will allow flexibility within EPA to vary the
confidence levels to meet individual Agency needs. These proposed definitions are
-------
678
sufficiently generic so that they shouldn't hamper various EPA offices or other
government agencies from meeting their respective needs as long as everyone agrees on
the them.
-------
679
QUESTION AND ANSWER SESSION
MR. STANKO: George Stanko, Shell Development.
One of the problems has been this detection limit thing that has been
going on for years, and we probably will never solve it here.
I would like to point out that, first, you cannot establish a distribution
curve at zero. It is impossible. It has to be something measurable so that you can
calculate standard deviation and then create a distribution curve. So, in theory, what was
shown here is impossible to do in practice.
Another thing, there was an error in Figure 16. If you have one less
molecule than the MDL, you have a 50.99 percent probability of a false positive and a
49.99 percent probability of a false negative. If the precise concentration in the sample
is one molecule less than the listed MDL, you have a 50/50 probability of a false
positive or a false negative, and not 50 percent false positives and 1 percent false
negatives. That is a flaw.
Another problem I am having is that going from RDL to RQL does not
include interlaboratory variability. It is done at a single lab, and it is multiplied by a
factor of 2. It does not include interlaboratory variability.
Whenever you have a situation where you have exceeded your permit,
somebody has caught you, there are always at least a two and/or three or four labs are
doing samples, and it really throws you into the realm of this variability associated with
interlaboratory measurements.
If you will take RDL and multiply it by 5, then I will accept your RQL.
DR. KEITH: Those are good comments to consider, George, and let me
say that I understand that the curve in the Figure that I showed are only theoretical, but
I was trying to show a simple concept.
So, even though, most instruments don't give, negative values (which I also
pointed out), it was simply to try to pictorially show a concept.
MR. SNELLING: Ron Snelling from LSU.
I think this is a comment instead of a question. We had a project we were
-------
680
working on where we were doing information theory to evaluate methods. I think it
would have some real use in these, because information theory will help you relate your
probabilities of false positives, false negatives.
There is a relationship between false positives and false negatives
described by a curve, and you figure out your costs that are incurred with false positives
and false negatives and try to find an optimum balance. With that, you can help
determine how you set your detection limits and quantitation limits based on the effects
of a false positive or a false negative.
DR. KEITH: Okay, thank you.
MR. DRIEDGER: Art Driedger from Wayne Analytical.
I was just curious if you were thinking of changing the way in which you
would establish the spiking level?
DR. KEITH: No.
MR. DRIEDGER: Okay, and the other thing is earlier this afternoon, we
had a talk where the spiking level was either done between 0.2 and 2 parts per billion,
and I thought that would be difficult to compare on a given list and wondered if there is
a way of using the relative standard deviation to more or less normalized some of the
values out. In other words, you have, say, your MDL at 3 sigma, and then divide by the
average value or, say, the average value recovered to account for the difference, you
know, in absolute spiking value. Also even, for that matter, in recovery from the
extraction procedure or what have you.
DR. KEITH: Doesn't that begin to get a little more into the protocol of
how to do it which is what we are trying to stay away from? We don't want to say how
to do it.
MR. DRIEDGER: Yes.
MR. YOCKLOVICH: Steve Yocklovich from Burlington Research.
I agree that determining the MDLs on reagent water samples has nothing
to do with the matrices we work with, and you said this will allow us to look at the
effects in different matrices, but working for a commercial lab, I wonder, does that mean
we will be required to determine MDLs in every different matrix that we look at?
-------
681
MR. TELLIARD: Yes.
MR. YOCKLOVICH: I mean in single samples.
DR. KEITH: If it is practical and EPA wants more realistic numbers, then,
yes.
MR. YOCKLOVICH: Well, I am not concerned about the cost. It is not
my money.
MR. TELLIARD: Eventually, it will be.
One point of clarification. Larry is talking about the Office of Water, and
you should be aware that this is the Office of Drinking Water putting forth this proposal
in a defense to do away with the PQL which, of course, is the "political quantitation
level".
The dirty water people, us guys, haven't been involved, and we don't
necessarily agree with all this, but we think it is really neat, you know. Again, your
agency together.
DR. KEITH: Well, the other thing, though, is that there is a group within
EPA called EMMCI which is kind of an overall Agency coordinator for harmonization of
methods, etc. and it is also involved with these redefinitions to try to make sure that the
various parts of EPA, like the dirty water folks and the Superfund folks, can have
something that they can all buy into.
MR. TELLIARD: I think the point I am trying to make here is if you have
a comment, it ain't over with yet. Get them in.
DR. KEITH: Right.
MR. LEVY: Nathan Levy, A&E Testing in Baton Rouge.
From an independent laboratory perspective, I would like to say thank you.
We consider MDLs as making our day lousy. We are really tired of being pushed
against the wall for these MDLs that were generated, usually, in DI water, and it has
been a long time since we analyzed DI water for money.
As far as having to worry about establishing PQLs or RQLs or whatever
you want to call it in each matrix, I think that is almost quite easy to do, because in
order to establish whether or not you do have a positive hit, you have probably got to do
-------
682
some dilutions or concentrations, etc.
And I appreciate George Stanko uncovering a farce in some of the
laboratory community in which they do not adjust their MDLs by their dilution factors.
That is another battle that some laboratories have to face when compared to
laboratories who do practice the art of reporting MDLs in DI water against samples they
have diluted by 10 and 100.
But I am really glad to see the Agency trying to work with us and trying to
understand that we have a big problem with matrices and that MDLs and the old
methods aren't appropriate.
DR. KEITH: Thanks for your comments. I have heard many, many people
echo similar sentiments. I think there is a lot of discontent with the MDLs.
MR. STANKO: George Stanko. I would like to make one further
comment.
Lloyd Curry told us that we should not report any value less than LOQ
(limit of quantitation) in the ACS publication, and that the area between LOD and LOQ
is totally unreliable. We, the regulated community, have to report a number on a
permit, and most of us have been standing very tall and screaming that we will not
report a value less than the current PQL.
If you change the rules from MDL to RDL and RQL, there needs to be
some guidance, because, here again, we don't think it is realistic to report any value, a
numerical value, less than RQL, and recognize the fact that the difference between RDL
and RQL is an area where there is highly variable and uncertain data, and one should
not be forced to report values that are totally unreliable.
And no value should be reported less than RDL, and, in fact, you have no
physical evidence that has ever been detected.
MR. TELLIARD: There is a matter of faith, George.
Thanks, Larry, so much for coming in and doing this.
DR. KEITH: You're welcome.
-------
683
(O
E
Q>
o
oZ
o
E
o
CO
CD
O
O
OT
*= ••= o)
r£ .£ CD
CO
— CO
Eo co cz
° CZ .0
Q ^5 -4= "CO
CD
CD S5
CD
O
O C
O O
O
CO
CD _CD "g CO
E-S-o E
o e c o
CO CLCO
CD
•4—'
CO
CD
CD
Q. O •=
CD
CO
CO
E
CD
CD
CO
CD
CO
CD
E
+—t
CD
O
CO
-------
684
o>
•••^•^^
o
O)
O
O
ol
CD
Cj
*a>
O
o
CO
c
,o
• •
CO
.g
^"•«
CD
• ^^•••H
CO
CD
2
Q.O
CD **~
CD
CO
CD
CO
Z3
CD
C3)
CO
CO
••••••
CO
CD
i^mmum
O
CO
O co
CD
+- E
CD CO.Q
E ^ o
CO CD CD
00
LU
03
O
CO
— co
E
O
CZ
CO
CD
O
CO CO
o
CO
CO CO
-------
685
CO
o
"O
03
CO
^
LJJ
<£
nversation
o
o
T3
CD
0)
03
O>
CD
CO
03
_LZ
( ^
consensus definitions that:
CD
_CO
O
CD
.9?
o
03
O
•4— •
CO
.1Z5
CD
"o
CO
1
CO
c:
_o
_c
"CD
o
CO
CD
0
CD
O
CL
CO
o
gj=
"CD
•a
CD
^
CD
">
o
ol
ying recommendations
c:
th accompa
'^
CO
c:
o
"-»— *
"c
"CD
"0
CD
o
ol
CD
O
03
T3
O)
CD
0)
03
CO
O)
_c
cz
th accompa
$
CO
o
H— *
%
"0
CD
0
ol
ily encompassing in their scope
; are covered.
•
CO
E
O
-*— •
CD
_m
Jr
52
-^
>\
CD
O
CD
JO
0
CL
CD
no
CD
_Q
O
CO
CO
.0
rt±
"CD
T3
•o
CD
CD
DC
CO
"03
Z3
"55
CD
E
o
• —
CD
'_'
O3
i *
03
CO
oad support and acceptance
omm unity.
£-
CD
03
"0
13
0
CO
CO
o
."ti
fc
*o5
•o
"0
CD
_co
CD
tr
o
id technical
i_
03
_O
~
/i^
CD
"o
CO
CD
|
"1
CO
in
-------
686
O
o
CO
Q
S
•*•* "O
-o o>
CD 5±=
CZ 0
•— CD
E Q.
CD w
0 ^
"o "co
CD v*
-Q c
cz _£S
s-°
•s *
Jz E
•«- o
cz4=
o
"2
"cz
CD
O
O
O
CO
CD
0
1
"cz
CD
|
.>-
"co
.0
.12
5
"co
CD
O
» CZ
CD
JO
H_
O
o
"o
n that can be measured with
t the analyte concentration is
o
"co
t concent
CO
CD
O
— J
CO
CD
O
CZ
CD
;o
*cz
o
o
^£
0^
O
CD
k.
CD
N
CO
.CZ
u.
3
"53
CD
0)
_.
*o
CD
JSI
CD
O
-Q
CO
O
CO
CO
cz
.0)
"co
J^
cz
JO
CD
CD
o
-Q
CO
O
CO
are made concerning whether
inalyte.
CO
cz
.2
CD
T3
*o
CD
M>^
^
CO
CO
CD
0
CZ
CD
CO
O.
, CZ
E
o
CO
-Q
CD
CO
CD
O
CZ
CD
1
"H
cz
Z3
O
CO
^^ '
Q
2
o
CD
"CD
Q
CD
CD
CD
a^ i§
Q x -° Q. 8^
•^ ^ P P CO § ^"CO
8 8 2 O Q -E
•*•*• -^"- «^^ f J W^B
-J co
o o c: vt»
CD CD CO U>
CD CD £ °
" — o cz
CO CO
CD
o
o
CD
-------
687
CO
E
o>
o
QL
CO
(O
s
LU
03
CD
E
03
03
CO
CO
CD
O
03
O
H—
O
CD
CO
O
CO
o
CD
CD
X
03
E
CD
;o
CO
o
o
CO
CD
o
CO
-------
688
CO
c
CO
CD
^••B
^l
CO
>
>
CO
CO
O
CO
CD
O
O
o
.5°
J^
LLI
"CD
"co
^N»
O>
s
:1
03
^
a misnomer.
nnot get belo'
tn &
.S2 o
=_, CD
"E §
LJ co
*o "o.
CD E
CO —
CO
o
^•^F
^
Q
te Dictionary
.03
O^
O
"o
o
JCO
CD
v*x
to
JD
CD
O
"§
Q 0^
: "°
CD.^
1= ^
og.
-E o
0 C
Q. .^
.S2
O) O
CD ^
^z: O)
d. m
E£
!±± CD
~ o
—1 CO
. .
If
s_
^
CD
Q^
^ m
CD
"co
O
O
~ Z3
o5§
I" 8
CO C
H- CD
CD^
T3 JD
C ^
0)2
03 ,t±
E<*^
^l—
(D ^
h- co
••M!
CD
CD
"o
C
1
1
"co
O)
"co
CD
CD
CO
O
CO
Q>=,*
N <2
"W 0
CD "-f=
-H- 03
CO 0
C^H^
d
"S °
la o
CD CD
^^ a^^
Q-3^
il
-------
689
o
CD
CD
^•^
Q.
CO
C
"o
"CD
C_J
CD
.>
"co
o
o.
CD
JO
O)
"co
"CD
ID
^D
O
>
^^^+
CO
CO
CD
"co
03
t ^\
CD
N
CD
O
CO
w— «.
nmended 3c
k_
o
o
o
^£
CD
o
Q.
CD
CO
45
^^^^
CO
-t— •
CD
0
VMI—
CD
\\
o
O
o^
o^
A
CO
CD
^^fc^ ^^^^^^
O CD
H
CD 2
CO
Z3 CD
03 -Q
O ^-»
CD O
JD C
O
UO
O^
H—
o
>
CD
O
CD
O
^N
cT
f^^
CO
•*— •
IQ
CO
03
CO
CD
CO
CO
0
CD
E
o
CO
o
U.
1
CD
JD
"o.
CD
o
o
CO
CD
JD
03
E
CO
CO
CD
T3
Z3
O
O
.CD
O
oT
o
2
^
*Q
"c
"o
^•^^
o.
03
03
03
o
05
03
CO
2
CD
M
CD
O
_03
03
"o5
>
CD
O
• •"•
•§
CD
O
O
CD
O)
CD
03
CD
CD Q"
"o5O
CO _J
CO
c
CO
CD
E
"03
2
1
.c
*+—^
CD
Q
CD
OJ
"1
CO
"53
CD
-------
690
O)
o
o
o>
CO
O
D
O
(O
LU
co
CD
CD
CD
i-B
co £0
go
CO O
~ o
CO O
2
CD
T3
8
o •£
o co
>
0>
•8'S
CD
O
CO
Q.
O
E
E
o
o
ended
CD
CD
CO
CO
*- 2
O ^ CD
C CO
O Q.
Q.
CO
0.0
CD 8>
-§CO CO
CO >* CO
03
CD
o _co
C OJ
o c:
^ cp o
c "co
CD
N
O
JD
CO
CO CD -55 o '7=Z
O3 -
O)
CO
03
03
CO
CD
CO
o
o.
CD
-------
691
CD
o
0>
111
o
o
O
BO
cS
03
a!
c
o
E
E
o
O
o
52
"CD
T3
CD
E
2
"co
CD
03
CD
C
CO
CD
03
03
CO
CD co
IS
CD
-rr* o
a 5
9^"°
CO
S 8
E ?2
= E
o
co
ca
a>
CO
CZ
03
CZ
o
CD
"O
O
"55
E
"co
CD
o
CD
"O
o
-------
692
o
4-»
O
o
o
o
C
4-^
"co
"O
^—
ID
O
^•B
\_jfy
O
03
-Q
TD
03
O
0)
jsj u
+-**
Jd CZ
o'o
JQ Q.
Accommodate
as a reference
^^* CJ}
00 f~
"*• ^rt
CC .zz
s e
CO Q^
O -i
> 0
•— ^>
8s
^•B •^ta^B
c5^
C J ^^
.z±. o
*C Q.
8 CD
O ^
CD ~
^. r*T^
.S5-S
c5 "$S
> E
^••B
15 ^
15
^^^
O3
O)
_
*v^
03
C
CD
$
X
^^ TT*
• ~ r—
S 03
Use a represer
measurements
"O
CD
.CO
0)
>^
f""
O3
Q.
E
o
o
o
**y
O3
^ ^^
CO
f**
.0
H-^
-CO
• —
..1
-o
CZ
03
CD
CO
C"
o
CD
O
Provide guidan
definition.
-------
693
CD
O)
t O
co
o
03
13=
-
co Q. cz - x
2 co w < 2 •=
_
^ S. -e -»-• ^ CD
*^— cj ^^ v— ' rr< /••• ^i^ ^"^
^^fecwj) $ S Q.O
I—
03 03 03 0
-------
694
positive detections
<0
_o
__
o
CD
O
CD
f f\
CO
45
*o
CO
^^ 0
CD
O
_>
O
"O
CD
CO
CO
a__
Q
CD
1 —
C^
CD
CO
CD
Q.
'I*
co
c
CO
CO
CO
CD
CD
^^
be considered:
33
o
CO
"co
CO
c
"c
T3
"o
O
^
^•••4
2
CO
2
CD
1 —
CD
CO
45
_co
CD
CD
CO
CO
i
******
Q
DC
^ ^
^^••i^^
"CD
CD
o
"•£5
"CD
Q
CD
JD
.co
"CD
DC
CD
CO
CO
CD
"co
CO
^_^
CO
;—
CD
CD
CO
JO
"o
"S
"O
^v^
CD
"co
O)
CD
0
to
"CD
>
CO
X— J,
o
DC
^.•^"
CD
CD
§
s
•^
CO
^f
_CD
CO
"CD
DC
od confidence,
^
o
O) -
CO
CO
' m
•£
CO
CO
"G
^t^
CD
•*— •
CD
^^^^
CO
*o
o
*^h""*
40
CO
13
CT
CO
2
,-£
CD
E
0
13
CO
co
CD
E
CD
O)
"co
cu
^^
O
T3
CD
CO
CO
-Q
CD
CO
CD
CO
CO
-£3
CO
I
o
DC
O
Q
DC
Q
CD
"CD
o
Z
TD
CO
C
o
"o
0)
o>
•o
_>
o
1
1
.^
•^
CD
CD
•£
CO
CO
CO
*g
CO
LJJ
^2
i
o
LJJ
DC
DC
O
O
k
k
k
o
CO
o
c:
o
o
T3
CD
to
CO
w ^
CD
-------
695
o
CO
CD
CD
CO
(O
CD
en co
CD
CO
cc.9'^1
O ^^_KI ^ ^^ .^^••k
CD m —
Q-o E.
o
CO
o
^ CD
t± "CD
CO
co
CD
co g-o
— S: CD
~ co
o
O CD
CD —
CL C ^r
CO O CO
CO O
CD ^
O CD
-------
696
CO
CD
•£
"oa <°
>2 CD
6
CD
^ CD
o o
°
8
Q
o
o
o
cz
8
O £r
t: ^5
o'o
O IM
CD
-------
697
®
o>
O
(O 5=
O
O
^^^^J W^""^"
go
»"S -g
CO
/-^ • ^ vw
m^S?^
^ .co .> ^r _
t= CD '1^ CO CD
CD CD
V/^ <— ^—
o £ E
~ O
CO
CO
£0 ^03
£ CD -^ S= CO
^ N= ^"f_
J.CO^^ §
P CO
CO 03
X C CO
CD CO 13 .^
W^^^MP ^^•••^
CO CO
CO
CD :
co c CD
-------
698
O
cn
GO
CO
LO
CO
CM
CVJ
CO
up
CO
co O
-
c
cu
-5=12
co
O O
-------
699
CO
£
CD —
> CO
CD ^
- o
CD =
0.
CD
JO
CO
o
CO
CD *- O
r o
•** jo o
o±: o
o
-2 D
< Q.CO
CD CD
-Q.>
= CD
45 £=
— CD
CO (O
c
co
CD
o
= o
-Q CD
CO fc
-§°
vO
5: co
m
co
CO :
CO
03
>*—
o
^:
,co
CD
n 'v
CD
I
g.
CD
CO
.
-------
700
CO
.t±
S -'o
CC c-^=
II
^ .2 ®
Q -5 ^
oB
Oco
O O
-------
701
S
» i'S
CD ^ CO
CO O -*£
CO X .2
^ . CO
-o J2>^
T^ *— O^
?-B"v
"go co
—
CD Q-^c
Q CD $
T3^2 .
0.03 D
kco
O
10 c
CO .Q
^O
O^
^D O
CO T3
to g
.<5 o5
CD —
CD CD
JZ Q.
c:
O
CD
CD ^
S o
? OQ
03
CD
CD
> CD
S -£2
'co co
o**-
CLCO
e
o
c:
•52
c:
o
O
co
CD
c:
co
H^.
O
^
IS
co
^Q
O
03
O
•4— »
CO
CO
^
O
CD
CD CD
> JO
CO CO
co TI
£S
si
^n ^
vy r^
*— 2
CD Q-
CD
CO
03
u_
CD
CD
Q.T3
CO :&
.^S-B
^ COO
C
.o
to
CD
Z3
O
CD
CD
O
CD
•*—»
CD
Q
"O
o
CD
CD
O
CD
CO
-------
702
0>
o
o
Q
-------
703
c
.2
.Is
CD
Q
Q "H
DC co
TJ^
cqO
.o
.o
03
O
c=
o
O
-------
704
CO
c
JO
to
c
E
&
OQ
^^^ ^}
• ••• ^^^
•U ^JJ
CO CD
.Q Q)
GL o
.—
JMB
CO
o
Q.
f
CO CO CO CO ^ CO CD
cvi cvi ^
Q
oDC
M ii
• • • •
d d
0 0
OO
CD CD
pz pz
T- c\J
-------
705
o
0)
o
CD
DC
o>
CD
~ =U>
CB^
^^S-o
S^-i "3
CO
CO
O>
X
CD
CO
i>kg><
.^ CO
CD CD 22 2 CD
O —
"-P3 73
8-8
o c:
r*\ CD —
LJ p co
~° c ^
O E "O
•c o •>
m ° =
05 CD
CO
Q
sis-22" -
CD co x:
CD
£2 *c
GO CO
CD *-
£ 2
CD
— *-.-= Q
Q. o »— rr -^ ^
p CD CD"- O CD
^^•CO^S^
c-o CD£ ^o
co '
CD
GO
— CD
_- —' GO
COQ CO
DC-0
GO
.Q
o co
I "|
co o5
w "55
CD
-o cz
O 03
:£.£
CD CO
e§
r o
^ 02
°^
Or=
**""* CO
§ s
CT CD
CO
CO
CD CD ^^
^ ^ ^™«" ••• ^-
O
O
CD CO
-^ GO
CD-^
CO^
>- CO
O C
CO
CO
CO
S§
So
O H_
15 '8
^2
o 2
CO
Q
o
'co
'o
(D
n Q.
CD
xi
OJ CO
o co
CD to
CO
O Z3
O CO
g.>
S"^
CD O
•— —I
"COQ
^DC
CD
8 >"o>
O CD >
O —I
CD
(D ~
CO CO
CO -O
CD O
E Q.
-------
706
o>
O
cc
CO
CD
•*-<
CD
"O
O ~
(D CO
F CD
CO
cr;S
T3 Cg
CO C3
Q CO
<§£
CD
CO
CD
& 0>
2 o
o o
o jj
^ (D
r ^
! °°
! co
I O)
P«. 5
; co
CO
cr
CD
o>
JO)
JQ
CO
-9
CO
to
CD
T3
O
CO it=
"
C0~ CO
o
E
O ±= -3
~ -^ co
co x co
£- 'C CD
"^ P
SEo
§|
CD CU CD
- E3
c co
o o
•-S-2
—r to -«±
S c 2
CD (D "c
§g§
0)Q.C
= 28
CO
O
o
CD
i w
^ §
J3
o
«^
CO d
.S2
rr* CO
tn
CO CD
e
CD
CD
Unlike
measu
CO ^3
2
a
CO
CD
CD
CD
O>
"co
CO
CO
CO^g
of .JS
CO
CD O
•£ CD
l
-------
707
CD
3
O CO
C\J
to
CD
^-* CD
E
0)
CO CO
~ CO
—I CD
s™
co
§>.£
(D C
CO CO
g 9>
"ll
Ql § RS
CO
o
T3
CD
T3
c^r^
CD O
Ec\j
o n
o o
£co
D x
OTT
> CO
(D —
— "O
"co 52
c o
CD
O CO
CD CO
CD
CD
co
-------
708
a
a>
CO
O
o>
OC
CD
CD
CD
CO
CD
CD
CO CO
si
-2 I
CD
Q
CD
O
DC ~
0)
CO ZJ
CD CO
> CO
51
CD CO
~O Z3
c -o
CD-5;
F :5=
^™" ^LJ
E c
o ~
CD O
CD
DC
CD
X +"
d CO
1-1
CD <-.
CO ~
.i v-o
•~ 7-5 >—•'
co M= DC
0 co CD
Q-§^
rr 2 -o
CD ^
^ cL#
-*— r^> cj
••- ° X
CD
CD
CD
W^
CD fc O
N T3 -H-
O
g C0==
8 g ^
*" = CO
CO '4= CD
CD
CO
CD ~
o
-------
709
a>
O
CO
O
O
a>
O
CO
CD
E
E
o
o
o
CO
CD
03
03
T3
CD
O
E
CD
03
E
CD
CD
CO
CD
CD
CD
O
OS
W.9
CD o
X —
CD-
O
05
CZ CO
o cz
o o
CO
CD
O
CO
'co
CO
£ £f 03
LU-g
.£ CO
.±± CD
^ ?D
CD.^
£•£
03 := :=
03 ^0
-S> 00
CD
O CO
CZ CD
CD >
CD
E
CD
CO
8*
Si
r£ <»
5£, co
CD
co E
03
"co -S
O) ~
o .>
z: -o
CD •=
CO .c
15 ^
CO
o^ CM
CD
E
° —I __J -J
CD Q Q O
or s DC or
-------
710
[Blank Page]
-------
711
MR. TELLIARD: We have saved the best to last. Dale is going to talk
about the new non-conventional pesticides methods compendium that has recently been
published.
-------
712
[Blank Page]
-------
713
METHODS FOR NON-CONVENTIONAL PESTICIDES IN WASTEWATER
Thomas E. Fielding and William A. Telliard*, Engineering and Analysis
Division (WH-552), Office of Science and Technology, U.S.
Environmental Protection Agency, 401 M St SW, Washington DC 20460.
Jim King, Lynn Riddick, and Steve Mitchell, U.S. EPA Sample
Control Center, 300 N Lee St, Alexandria VA 22314
D. R. Rushneck, Interface, Inc.,PO Box 297, Fort Collins CO
80522-0297.
* Author to whom requests for information should be addressed
ABSTRACT
On April 10, 1992, the U.S. Environmental Protection Agency (EPA)
proposed a regulation to limit the discharge of 122 pesticide active ingredients (PAI's)
into navigable waters and into publicly owned treatment works (POTW's) in the U.S. (57
FR 12560). This was a re-proposal of the regulation originally promulgated in 1985 (50
FR 40672) and voluntarily remanded in 1986 (51 FR 44911).
As a part of the proposed regulation, EPA proposed a compendium of
analytical methods (Compendium) for the determination of 228 PAI's in wastewater.
This Compendium includes the 600 and 1600 Series wastewater methods not
promulgated to date and contains at least one method for each of the regulated PAI's.
In addition to the methods in the Compendium, EPA also proposed to allow use of the
500 Series Drinking Water methods and the 200 Series metals methods for monitoring
the regulated PAI's in discharges.
This paper gives an overview of the regulation and details of the
Compendium.
-------
714
INTRODUCTION
The Federal Water Pollution Control Act (FWPCA), also called the Clean
Water Act, was passed by Congress in 1972 (PL 92-500).In 1976, several environmentally
concerned plaintiffs brought suit against EPA for failure to enforce certain provisions of
this Act.In a Consent Decree (reference 1), the Agency agreed to establish regulations
for 23 major categories of industry (reference 2). The pesticides manufacturing industry
is the last of the original 23 major industries to be regulated.
The regulation proposed on 10 April follows other EPA regulations to
control the discharge of pollutants into surface waters (reference 3), in that the Agency
studied the industry intensely to determine what pollutants are generated, which
pollutants may be discharged, and the most effective treatment systems for reducing the
concentration or amount of these pollutants in the discharge. The proposed regulation
controls many of the "toxic pollutants" (the 126 "Priority Pollutants"), but most
specifically non-conventional pollutants in the form of pesticide active ingredients
(PAI's). Three conventional pollutants [biochemical oxygen demand (BOD), total
suspended solids (TSS), and pH] were controlled by the original 1985 regulation.
The 10 April regulation is unique compared to other EPA wastewater
regulations in that a given PAI may be manufactured at one or very few plants; the
manufacturing process may be unique; the mix of pollutants generated may be unique;
and an unusual treatment technology or group of technologies may be required to reduce
or eliminate the pollutant in the discharge. The characteristics of the PAI, coupled with
the uniqueness of the discharge, present an analytical challenge because routine methods
used to measure the conventional or toxic pollutants cannot usually be used to measure
PAI's in the discharge. As a result, EPA has adopted or developed methods for
measurement of the PAI's.
THE PROPOSED REGULATION
EPA has proposed to establish effluent limitations guidelines based on,
"best conventional pollutant control technology" (BCT), "best available technology"
(BAT), "new source performance standards" (NSPS) based on "best demonstrated
-------
715
technology"(BDT), and "pretreatment standards for new sources" (PSNS) and
"pretreatment standards for existing sources" (PSES) for new and existing indirect
dischargers. Indirect dischargers are those manufacturing plants that discharge to a
POTW rather than discharge directly to a surface water. Effluent limitations guidelines
are limits on the amount of pollutant allowed to be discharged and are either in the
form of a concentration (mass per unit volume of water discharged) or of the maximum
amount of the pollutant that can be discharged per amount of product manufactured
(mass per unit mass of product manufactured). The discharger typically meets the
limitations by using some form of treatment system for pollutant reduction. However,
manufacturing process changes, re-use of process water, and improved housekeeping,
among other actions, could reduce the cost of, or eliminate the need for, end-of-pipe
treatment technologies.
The regulations are proposed for codification at 40 CFR Part 455, and are
supported by a development document (reference 4) and an economic analysis
(references 5 - 6). Legal authority for the regulation is under sections 301, 304, 306, 307,
and 501 of the FWPCA, as amended by the Clean Water Act of 1977 (PL 95-217) and
the Water Quality Act of 1987 (PL 100-4).
The preamble to the regulation gives technical data on the following major
topics:
EPA's data gathering efforts
Subcategorization of the industry
Water use and wastewater characterization
Pollutant control technologiesEconomic considerations
Water quality analysesNon-water quality environmental impacts
Regylatory implementation
Tables at the end of the proposed regulation list the PAI's and give the
proposed effluent limitations for those regulated.
-------
716
APPRO VED METHODS
Methods proposed to be approved for use in monitoring for PATs are
given in Table 1. Included are methods developed for determination of organic
pollutants in wastewater and promulgated at 40 CFR Part 136 (49 FR 43234), methods
for the determination of organic pollutants in drinking water (references 7-8), methods
for determination of metals in water and other sample matrices (reference 9), and
methods in the Compendium. Table 1 includes at least one EPA method for each
analyte. For 24 PATs, a single method is available; for 35 PATs, two methods are
available; for 26 PAI's, three methods are available; for 6 PAI's, four methods are
available; and for 4 PAI's, five methods are available. EPA has allowed this flexibility in
methods so that an analyst familiar with a given EPA water method will not be forced to
use an alternative method and can overcome matrix interference problems.
EPA has allowed flexibility within methods previously proposed and
promulgated. This flexibility is described in detail in the preamble to the 40 CFR Part
136 proposal and promulgation for the determination of organic pollutants in wastewater
[49 FR 43234], and permits the analyst to "improve separations and lower the cost of
measurements" provided all performance criteria in the method are met. Data
documenting that all performance criteria have been met must be retained on file for
inspection or later submission to EPA or the State, if requested. One of the objectives
in allowing this flexibility is to encourage method improvement, particularly to overcome
matrix interferences. EPA believes that promulgating multiple methods and allowing
controlled flexibility within these methods will result in reliable and high quality data.
METHODS COMPENDIUM
The Compendium (reference 10) contains 41 methods covering 228 PAI's
(analytes). Table 2 gives a list of the analytes, the Chemical Abstracts Service Registry
Number for each analyte, and the methods in the Compendium for each analyte.
The purpose of producing the Compendium was to assemble all pesticide
methods that EPA had developed and had not promulgated into a single volume so that
analysts would have access to these methods. EPA recognizes that this set of methods
-------
717
does not cover all PAI's ever manufactured. In developing the regulation, EPA studied
more than 650 PAI's and found that although methods are available for determination of
nearly all of these PAI's in product formulations, plant matter, food products, and other
food related matrices, methods were not available for the determination of a majority of
these 650 PAI's in environmental samples.
The Compendium includes methods developed by EPA's Environmental
Monitoring Systems Laboratory at Cincinnati, Ohio (EMSL-Ci), methods developed by
EPA's Engineering and Analysis Division (BAD) within EPA's Office of Science and
Technology, and industry method IND-01 for organo-tin compounds. The EMSL-Ci
developed methods have three digit numbers beginning with the number six (e.g.,622)
and the EAD developed methods have four digit numbers beginning with 16 (e.g., 1656).
Many of the methods included in the Compendium were listed in Appendix
E of EPA's original promulgation of the Part 455 rules (50 FR 40708). The methods
were withdrawn in 1986 as a part of the remand of these rules (51 FR 44911). In the
intervening years between the original promulgation/remand and the present, EPA has
developed additional methods, has updated several methods to include more analytes,
and has nearly eliminated dependency on industry and contractor developed methods.
DEVELOPMENT OF METHODS
Since the previous methods set was published, the trend of pesticides and
herbicides produced and applied in the U.S. has shifted from chlorinated compounds
toward phosphorus containing compounds and other substances found to be less
persistent in the environment. This change has necessitated the development of
analytical methods to measure these compounds in wastewater discharges and in other
environmental samples. EPA has therefore developed additional methods as a part of its
data gathering efforts for today's proposed rule.
Where possible, EPA avoids development of a new method by testing
existing methods to determine if an active ingredient can be measured by these existing
methods. If these tests are successful, EPA revises the method to incorporate the new
analyte.In addition, EPA has attempted to consolidate multiple methods for the same
-------
718
analyte by selecting a given method or writing a revised or new method and including as
many analytes as possible in this method. For example, EPA has used wide-bore, fused
silica capillary columns in recently developed gas chromatography (GC) methods for
PATs to increase resolving power so that more analytes can be measured simultaneously
and so that these analytes can be measured at lower levels. Drinking water methods
507, 508, 515.1, and wastewater methods 1656, 1657, and 1658 represent GC methods
that encompass large numbers of analytes.
On the other hand, it is frequently not possible to include an analyte or
group of analytes in an existing method because the nature of the compound(s) does not
lend itself to the techniques in the method. In these instances, an entirely separate
method must be developed. In the methods proposed on 10 April, Method 1659 for
Dazomet, Method 1660 for the Pyrethrins and Pyrethroids, and Method 1661 for
Bromoxynil represent examples of methods that were developed specifically for an
analyte or group of analytes.The method for Dazomet employs a base hydrolysis to
convert Dazomet to methyl isothiocyanate (MITC) and gas chromatography with a fused
silica capillary column and nitrogen/phosphorus detector for selective detection of
MITC. The method for the Pyrethrins and Pyrethroids employs acetonitrile extraction of
a salt saturated wastewater sample and high performance liquid chromatography (HPLC)
for selective detection of these analytes.The method for Bromoxynil employs direct
aqueous injection HPLC.
EPA solicited comments on the Compendium, and will attempt to use the
comments to aid in further improvement of the methods.Informal, constructive comments
on the methods may be submitted at any time to the EPA Sample Control Center (see
reference 10).
REFERENCES
1. Natural Resources Defense Council, Inc. et al. v. Train, 8 ERC 2120
(D.D.C. 1976).
2. L.H. Keith and W.A. Telliard. Priority pollutants H. A perspective
view. "Environ. Sci. Technol." 13: 416-23 (1979).
-------
719
3. William A. Telliard, Marvin B. Rubin, and D.R. Rushneck, "J.
Chromatog. Sci."25:322-327 (1987).
4. "Development Document for Best Available Technology, Pretreatment
Technology, and New Source Performance Technology in the Pesticide Chemicals
Industry", Engineering and Analysis Division (WH-552), USEPA, 401 M St SW,
Washington DC 20460 (1992).
5. "Economic Impact Analysis of Effluent Limitations Guidelines and
Standards for the Pesticide Chemicals Industry", Engineering and Analysis Division
(WH-552), USEPA, 401 M St SW, Washington DC 20460 (1992).
6. "Cost-Effectiveness of Proposed Effluent Limitations Guidelines and
Standards of Performance for the Pesticide Manufacturing Industry", Engineering and
Analysis Division (WH-552), USEPA, 401 M St SW, Washington DC 20460 (1992).
7. "Methods for the Determination of Organic Compounds in Drinking
Water" EPA 600/4-89/039, Revised July 1991, National Technical Information Service,
5285 Port Royal Rd, Springfield VA 22162 (PB91-231480) (December 1988).
8. "Methods for the Determination of Organic Compounds in Drinking
Water - Supplement I" EPA 600/4-90/020, National Technical Information Service, 5285
Port Royal Rd, Springfield VA 22162 (PB91-146027) (July 1990).
9. "Methods for the Determination of Metals in Environmental Samples"
EPA 600/4-90/010, National Technical Information Service, 5285 Port Royal Rd,
Springfield VA 22162 (PB91-231498) (June 1991).
10. "Methods for the Determination of Nonconventional Pesticides in
Municipal and Industrial Wastewater", available from EPA Sample Control Center, 300
N Lee St, Alexandria VA 22314 (April 1992).
-------
720
[Blank Page]
-------
721
Tabtol
Methods Required for Nonconventional
Pesticide Pollutants (1)
CAS
Registry PC
Name Number (2) Code (3) Method(s)
Acephate 30560191 103301 1656
Acifluorfen 50594666 114401 1656/515.1
Alachlor 15972608 90501 1656/505/507/645
Aldicarb (Temik) 116063 98301 531.1
Ametryn 834128 80801 507/619
Atrazine 1912249 80803 1656/505/507/619
Azinphos methyl (Guthion) 86500 58001 1657/614/622
Benfluralin (Benefin) 1861401 84301 1656/627
Benomyl 17804352 99101 631
Biphenyl 92524 17002 1625/642
Bolstar (Sulprofos) 35400432 111501 622
Bromacil 314409 12301 1656/507/633
Bromacil, lithium salt 53404196 12302 1656/507/633
Bromoxynil 1689845 35301 1661/1625
Bromoxynil octanoate 1689992 35302 1656
Busan40 51026289 102901 630/630.1
Busan85 128030 34803 630/630.1
Butachlor 23184669 112301 1656/507/645
Captafol 2425061 81701 1656
Carbam-S 128041 34804 630/630.1
Carbaryl 63252 56801 531.1/632
Carbofuran 1563662 90601 531.1/632
Chloroneb 2675776 27301 508/608.1
Chlorothalonil 1897456 81901 1656/508/608.2
Chlorpyrifos 2921882 59101 1657/508/622
Cyanazine 21725462 100101 629
Dazomet 533744 35602 1659
2,4-D 94757 30001 1658/515.1/615
2,4-D Salts & Esters 94757 1658/515.1/615
2,4-DB Salts & Esters 94826 30801 1658/515.1/615
DCPA (Dacthal) 1861321 78701 1656/508/608.2
DEF 78488 74801 1657
Diazinon 333415 57801 1657/507/614/622
Dichlorprop Salts & Esters 120365 1658/515.1/615
Dichlorvos 62737 84001 1657/507/622
Dinoseb 88857 37505 1658/515.1/615
Dioxathion 78342 37801 614.1
Disulfoton 298044 32501 1657/507/614/622
Diuron 330541 35505 632
Endothall Salts & Esters 145733 548
Endrin* 72208 41601 1656/505/508/608/617
Ethalfluralin 55283686 113101 1656/627
Ethion 563122 58401 1657/614/614.1
Fenarimol (Rubigan) 60168889 206600 1656/507/633.1
Fensulfothion 115902 32701 1657/622
Fenthion 55389 53301 1657/622
Fenvalerate (Pydrin) 51630581 109301 1660
Glyphosate Salts & Esters 1071836 103601 547
Heptachlor* 76448 44801 1656/505/508/608/617
Isopropalin (Paarlan) 33820530 100201 1656/627
KN methyl 137417 39002 630/630.1
Linuron 330552 35506 632
MCPA Salts & Esters 94746 30501 1658/615
MCPP Salts & Esters 93652 31501 1658/615
Malathion 121755 57701 1657/614
Merphos (Tributes) 150505 74901 1657/622/507
Methamidophos 10265926 101201 1657
Methomyl 16752775 90301 531.1/632
Methoxychlor 72435 34001 1656/505/508/608.2/617
Metribuzin • 21087649 101101 1656/507/633
Mevmphos 7786347 15801 1657/507/622
Nabam . 142596 14503 630/630.1
Nabonate 138932 63301 630.1
Naled 300765 34401 1657/622
Norfluorazon 27314132 105801 1656/507/645
Organotin(asTin=7440315) 0-192 200.7/200.9/IND-01
Parathion ethyl 56382 57501 1657/614
Parathion methyl 298000 53501 1657/614/622
PCNB 82688 56502 1656/608.1/617
Pendimethalin (Prowl) 40487421 108501 1656
Permethrin 52645531 109701 1656/1660/608.2/508
Tabtol (continued)
CAS
Registry PC
Name Number (2) Code (3) Method(s)
Pentachlorophenol* 87865 63001 525/604/625/1625
Phorate 298022 57201 1657/622
Phosmet 732116 59201 1657/622.1
Prometon (Promitol) 1610180 80804 507/619
Prometryn 7287196 80805 507/619
Pronamide (Kerb) 23950585 101701 507/633.1
Propachlor 1918167 19101 508/608.1
Propanil 709988 28201 632.1
Propazine 139402 80808 507/619
Pyrethrinl 121211 69008 1660
Pyrethrinll 121299 69006 1660
Simazine 122349 80807 505/507/619
Stirofos (Tetrachlorvinphos) 22248799 83701 1657/507/622
TCMTB 21564170 35603 637
Tebuthiuron (Spike) 34014181 105501 507
Terbacil 5902512 12701 507/633
Terbufos (Counter) 13071799 105001 1657/507/614.1
Terbuthylazine (Gardoprim) 5915413 80814 619
Terbutryn 886500 80813 507/619
Toxaphene* 8001352 80501 1656/505/508/608/617
Triadimefon (Bayleton) 43121433 109901 507/633
Trifluralin 1582098 36101 508/617/627
Vapam (Metam) 137428 39003 630/630.1
Ziram 137304 34805 630/630.1
(1 ) Method numbers have been updated since publication of the proposed regulation
[57 FR 12560]
(2) A number assigned by the Chemical Abstracts Service Registry
(3) Pesticide Code, formerly the Shaughnessy code— a code issued by EPA's Office
of Pesticide Programs
* Priority Pollutant
Table 2
Pesticides with Methods in the Compendium
CAS
Registry Applicable
Pesticide Number Method(s)
Acephate 30560-19-1 1656,1657
Acifluorfen 50594-66-6 1656
Alachlor 15972-60-8 645
Aldrin 309-00-2 617,1656
Allethrin (Pynamin) 584-79-2 1660
Ametryn 834-12-8 619
Aminocarb 2032-59-9 632
Amobam 3566-10-7 630,630.1
AOP — 630
Aspon 3244-90-4 622.1
Atraton 1610-17-9 619
Atrazine 1912-24-9 619,1656
Azinphos ethyl 2642-71-9 1657
Azinphos methyl 86-50-0 614,622,1657
Barban 101-27-9 632
Basalin (Fluchloralin) 33245-39-5 646
Bendiocarb 22781-23-3 639
Benfluralin 1861-40-1 627,1656
Benomyl 17804-35-2 631
Bensulide 741-58-2 636
Bentazon (Basagran) 25057-89-0 643
alpha-BHC 319-84-6 617,1656
beta-BHC 319-85-7 617,1656
gamma-BHC 58-89-9 617,1656
delta-BHC 319-86-8 617,1656
Biphenyl 92-52-4 642
Bromacil 314-40-9 633,1656
Bromoxynil octanoate 1689-99-2 1656
Bromoxynil 1689-84-5 1661
Busan40 51026-28-9 630,630.1
Busan85 128-03-0 630,630.1
Butachlor 23184-66-9 645,1656
-------
722
T«M«2 (continued) Q^§
Registry Applicable
Pesticide Number Method(s)
Butylate 2008-41-5 634
Captafol 2425-06-1 1656
Captan 133-06-2 617,1656
Carbam-S 128-04-1 630,630.1
Carbaryl 63-25-2 632
Carbendazim 10605-21-7 631
Carbofuran 1563-66-2 632
Carbophenothion 786-19-6 617,1656
CDN 97-00-7 646
Chlordane 5103-74-2 617,1656
Chlorfevinphos 470-90-6 1657
Chlorobenzilate 510-15-6 608.1,1656
Chloroneb 2675-77-6 608.1,1656
Chtoropicrin 76-06-2 618
Chloropropylate 5836-10-2 608.1,1656
Chlorothalonil 1897-45-6 608.2,1656
Chlorpropham 101-21-3 632
Chlorpyrifos methyl 5598-13-0 622,1657
Chlorpyrifos 2921-88-2 622,1657
Coumaphos 56-72-4 622,1657
Crotoxyphos 7700-17-6 1657
Cyanazine 21725-46-2 629
Cycloate 1134-23-2 634
Cycloprate 54460-46-7 616
Cyfluthrin (Baythroid) 68359-37-5 1660
Dalapon 75-99-0 615,1658
Dazomet 533-74-4 1659
2,4-D 94-75-7 615,1658
2,4-DB 94-82-6 615,1658
DBCP 96-12-8 1656
DCPA 1861-32-1 608.2
4,4'-DDD 72-54-8 617,1656
4,4'-DDE 72-55-9 617,1656
4,4'-DDT 50-29-3 617,1656
Deet 134-62-3 633
DEF 78-48-8 1657
Demeton 8065-48-3 614,622,1657
Diallate 2303-16-4 1656
Diazinon 333-41-5 614,622,1657
Dibromochloropropane 96-12-8 608.1
Dicamba 1918-00-9 615,1658
Dichlofenthion 97-17-6 622.1,1657
Dichlone 117-80-6 1656
Dichloran 99-30-9 608.2,617
Dichlorophene 97-23-4 604.1
Dichlorprop 120-36-5 615,1658
Dichlorvos 62-73-7 622,1657
Dicofol 115-32-2 617,1656
Dicrotophos 141-66-2 1657
Dieldrin 60-57-1 617,1656
Dimethoate 60-51-5 1657
Dinocap 39300-45-3 646
Dinoseb 88-85-7 615,1658
Dioxathion 78-34-2 614.1,1657
Diphenamid 957-51-7 645
Diphenylamine 122-39-4 620
Disulfoton 298-04-4 614,622,1657
Diuron 330-54-1 632
Endosulfanl 959-98-8 617,1656
Endosulfanll 33213-65-9 617,1656
Endosulfan sulfate 1031-07-8 617,1656
Endrin aldehyde 7421-93-4 617,1656
Endrin 72-20-8 617,1656
Endrin ketone 53494-70-5 1656
EPN 2104-64-5 614.1,1657
EPTC 759-94-4 634
Ethalfluralin 55283-68-6 627,1656
Ethion 563-12-2 614,614.1,1657
Ethoprop 13194-48-4 622,1657
Ethylene dibromide 106-93-4 618
Table 2 (continued) CAS
Registry Applicable
Pesticide Number Method(s)
Etridiazole 2593-15-9 608.1,1656
EXD 502-55-6 630.1
Famphur 52-85-7 622.1,1657
Fenarimol (Rubigan) 601 68-88-9 633. 1 , 1 656
Fenitrothion 122-14-5 622.1
Fensulfothion 115-90-2 622,1657
Fenthion 55-38-9 622,1657
Fenuron 101-42-8 632
Fenuron-TCA 4482-55-7 632
Fenvalerate 51630-58-1 1660
Ferbam 14484-64-1 630,630.1
Fluometuron 2164-17-2 632
Fluridone 59756-60-4 645
Fonophos 944-22-9 622.1
Heptachlor epoxide 1024-57-3 617,1656
Heptachlor 76-44-8 617,1656
Hexachlorophene 70-30-4 604.1
Hexamethyl-
phosphoramide 680-31-9 1657
Hexazinone 51235-04-2 633
Isodrin 465-73-6 617,1656
Isopropalin • 33820-53-0 627,1656
Kepone 143-50-0 1656
Kinoprene 42588-37-4 616
KN Methyl 137-41-7 630,630.1
Leptophos 21609-90-5 1657
Lethane 112-56-1 645
Linuron 330-55-2 632
Malathion 121-75-5 614,1657
Mancozeb 8018-01-7 630
Maneb 12427-38-2 630
MBTS 120-78-5 637
MCPA 94-74-6 615,1658
MCPP 7085-19-0 615,1658
Mercaptobenzothiazole 149-30-4 640
Merphos 150-50-5 622,1657
Metham 137-42-8 630,630.1
Methamidophos 10265-92-6 1657
Methiocarb 2032-65-7 632
Methomyl 16752-77-5 632
Methoprene 40596-69-8 616
Methoxychlor 72-43-5 608.2,617,1656
Metribuzin 21087-64-9 633,1656
Mevinphos 7786-34-7 622,1657
Mexacarbate 315-18-4 632
MGK264-A 113-48-4 633.1
MGK264-B 113-48-4 633.1
MGK326 136-45-8 633.1
Mrex 2385-85-5 617,1656
Molinate 2212-67-1 634
Monocrotophos 6923-22-4 1657
Monuron 150-68-5 632
Monuron-TCA 140-41-0 632
Nabam 142-59-6 630,630.1
Nabonate. 138-93-2 630.1
Naled 300-76-5 622,1657
Napropamide 15299-99-7 632.1
Neburon 555-37-3 632
Niacide 8011-66-3 630
Nrtrofen (TOK) 1836-75-5 1656
Norftorazon 27314-13-2 645,1656
Organo-tin — iND-01
Oryzalin 19044-88-3 638
Oxamyl 23135-22-0 632
Parathion ethyl 56-38-2 614,1657
Parathion methyl 298-00-0 614,622,1657
PCB-1016 12674-11-2 617,1656
PCB-1221 11104-28-2 617,1656
PCB-1232 11141-16-5 617,1656
PCB-1242 53469-21-9 617,1656
TabU 2 (continued) CAS
Registry Applicable
Pesticide Number Method(s)
PCB-1248 12672-29-6 617,1656
PCB-1254 11097-69-1 617,1656
PCB-1260 11096-82-5 617J656
PCNB 82-68-8 608.1,617,1656
Pebulate 1114-71-2 634
Pendimethalin 40487-42-1 1656
Permethrin 52645-53-1 608.2,1656,1660
Perthane 72-56-0 617,1656
o-Phenylphenol 132-27-4 642
Phorate 298-02-2 622,1657
Phosmet 732-11-6 622.1,1657
Phosphamidon 13171-21-6 1657
Picloram 1918-02-1 644
Polyram 9006-42-2 630
Profluralin 26399-36-0 627
Prometon 1610-18-0 619
Prometryn 7287-19-6 619
Pronamide 23950-58-5 633.1
Propachlor 1918-16-7 608.1,1656
Propanil 709-98-8 632.1
Propazine 139-40-2 619,1656
Propham 122-42-9 632
Propoxur 114-26-1 632
Pyrethrinl 121-21-1 1660
Pyrethrinll 121-29-9 1660
Resmethrin 10453-86-8 616,1660
Ronnel 299-84-3 622,1657
Rotenone 83-79-4 635
Secbumeton 26259-45-0 619
Siduron 1982-49-6 632
Simazine 122-34-9 619,1656
Simetryn • 1014-70-6 619
Sodium dimethyldi-
thiocarbamate 128-04-1 630,630.1
Stirotos
(Tetrachlorvinphos) 961-11-5 622,1657
Strobane 8001-50-1 617,1656
Sulfotepp 3689-24-5 1657
Sulprofos (Bolstar) 35400-43-2 622
Sumithrin (Phenothrin) 26002-80-2 1660
Swep 1918-18-9 632
2,4,5-T 93-76-5 615,1658
TCMTB 21564-17-0 637
TEPP 107-49-3 1657
Terbacil 5902-51-2 633,1656
Terbufos 13071-79-9 614.1,1657
Terbuthylazirte 5915-41-3 619,1656
Terbutryn 886-50-0 619
Tetramethrin 7696-12-0 1660
Thiabendazole 148-79-8 641
Thionazin 297-97-2 622.1
Thiram 137-26-8 630,630.1
Tokuthion 34643-46-4 622,1657
Toxaphene 8001-35-2 617,1656
2,4,5-TP 93-72-1 615,1658
Friadimefon 43121-43-3 633,1656
Frichlorofon 52-68-6 1657
Trichloronate 327-98-0 622,1657
rricresylphosphate 78-30-8 1657
'ricyclazole 41814-78-2 633
Trifluralin 1582-09-8 617,627,1656
nmethylphosphate 512-56-1 1657
Trithion methyl 953-17-3 1657
Vacor 53558-25-1 632.1
Vemolate 1929-77-7 634
ZAC — 630
Zineb 12122-67-7 630,630.1
Ziram 137-30-4 630,630.1
-------
723
CLOSING REMARKS
MR. TELLIARD: I would like to thank you all for coming. We have been
doing this for 15 years, and we are going to try to get it right one time. We will be back
next year.
I would like to thank Harry McCarty for putting together the technical
program, and I would like to thank Jan Sears arrangement for the aquarium and the
steak house activity and I would like to ask you for a round of applause for their people.
Next year's agenda is open for discussion. If you have any ideas of sessions
you would like to have or think would be advantageous, informative, and so forth, please
give me a buzz and drop me a line, and we will see if we can accommodate you. If you
have suggestions on papers or if you would like to present at this assembly, please give
me a call. You only have 12 months and 28 days left to get it in, because we are always
about a year late.
So, thanks so much for coming, thank you for your attention, and thank
you, speakers.
-------
724
[Blank Page]
-------
725
15th ANNUAL EPA CONFERENCE ON ANALYSIS
OF POLLUTANTS IN THE ENVIRONMENT
LIST OF SPEAKERS
Jon Anderson, Jr.
Project Chemist
Columbia Analytical Services
P.O. Box 479
Kelso, WA 98626
206-577-7222
Catherine L. Arthur
University of Waterloo
Guelph-Waterloo Ctr./Grad. Work in Chem
200 University Avenue
Ontario, Canada, N2L3G1
519-885-1211 x6823
Sarah L. Barkowski
Research Chemist
Boise Cascade
Paper Research and Development
4435 N. Channel Avenue
Portland, OR 97217
503-286-7441
Merlin K. L. Bickirig
Technical Director
Twin City Testing Corporation
662 Cromwell Avenue
St. Paul, MN 55114
612-659-7519
Marielle Brinkman
Battelle Memorial Institute
505 King Avenue, Room 7238
Columbus, OH 43201-2693
614-424-5277
Kevin Carter
EnSys, Inc.
P.O. Box 14063
Research Triangle Park, NC 27709
919-941-5509
Paul S. Epstein
Director of Laboratories
NSF International
P.O. Box 130140
Ann Arbor, MI 48113-0140
313-769-8010
Jeanne Hankins
USEPA-OWSER/OSW
401 M Street, SW (OS-300)
Washington, DC 20460
202-260-8454
Robert O. Harrison
Manager of R & D
ImmunoSystems, Inc.
4 Washington Avenue
Scarborough, ME 04074
207-883-9900
Larry H. Keith
Radian Corporation
P.O. Box 201088
Austin, TX 78720
512-454-4797
Gabe LeBrun, Supervisor
Semivolatile Organic Analysis
PACE, Inc.
1710 Douglas Drive
Golden Valley, MN 55422
612-525-3352
Craig Markell
Research Specialist
3M
Bldg. 201-1C-30
St. Paul, MN 55144
612-733-2813
-------
726
Harry McCarty
Viar and Company, Inc.
Sample Control Center
300 N. Lee Street, Suite 200
Alexandria, VA 22314
703-684-0610
Jean W. Munch
Research Chemist
USEPA-EMSL
Environmental Monitoring Systems Lab
26 W. Martin Luther King Drive
Cincinnati, OH 45268
513-569-7465
Greg O'Neil
Marketing Manager
Tekmar Company
P.O. Box 429576
Cincinnati, OH 45242-9576
800-543-4461
Joe Raia
Senior Research Chemist
Shell Development Company
3333 Highway 6, South
Houston, TX 77082
713-493-7693
Kenneth A. Robillard
Technical Associate
Eastman Kodak Company
Chemical Quality Services Div.
Rochester, NY 14652-3615
716-588-5941
Dale Rushneck
Interface, Inc.
P.O. Box 297
Ft. Collins, CO
303-223-2013
80522
Rick Schrynemeeckers
Technical Director
Enseco
1420 E. North Belt #120
Houston, TX 77032
713-987-9767
Scott A. Senseman
Research Assistant
University of Arkansas
Altheimer Laboratory
276 Altheimer Drive
Fayetteville, AR 72703
501-575-3955
James Smith
President/Chemist
Trillium, Inc.
7A Grace's Drive
Coatesville, PA 19320
215-383-7233
George Stanko
Sr. Staff Research Chemist
Shell Development Company
P.O. Box 1380
Houston, TX 77251-1380
713-493-7702
Jim J. Stunkel
Application Chemist
ABC Laboratories, Inc.
P.O. Box 1097
Columbia, MO 65205
314-474-8579 x394
William A. Telliard
Chief, Analytical Methods Staff
USEPA-OW/EAD
401 M Street, SW, (WH-552)
Washington, DC 20460
202-260-5131
-------
727
LIST OF ATTENDEES
Steve Adams
Monsanto
700 Chesterfield Parkway North
St. Louis, MO 63198
314-537-6166
G.M. Alsop
Union Carbide
P.O. Box 8361
South Charleston, WV 25303
304-747-5467
Windy Amundsen
Lab Director
Sequoia Analytical
680 Chesapeake Drive
Redwood City, CA 94063
415-364-9600
Clifford G. Annis
Task Force for Envrn. Qual. Assur
Merck & Co., Inc.
3517 Radium Springs Rd.
Albany, GA 31708
912-434-5399
David Armstrong
General Manager
PCE, Inc.
4764 First Avenue N
Birmingham, AL 35222
205-591-4350
David E. Ashkenaz
MW Regional Manager
VSPP
388 Forest Knoll Drive
Palatine, IL 60067
708-705-9629
Pete Ausili
Naval Investigative Service
Naval Station
Norfolk, VA 23511
804-444-8615
Mindy Baldwin
Environmental Labs
9211 Burge Ave
Richmond, VA 23237
804-271-3440
Michael E. Barber
Analytical Technologies
9830 S. 51st St. Suite B-113
Phoenix, AZ 85044
602-496-4400
Curtis Beck
Quality Assurance Assistant
EMS Heritage Labs
7901 W. Morris St.
Indianapolis, IN 46231
317-243-8304
Kevin Beltis
Senior Consultant
Arthur D. Little, Inc.
20 Acorn Park
Cambridge, MA 02140
617-864-5770
Beverly Blanchard
QA/QC Director
James R. Reed & Associates,
813 Forrest Drive
Newport News, VA 23606
804-599-6750
Inc,
Dan Bolt
Environmental Products Manager
Cambridge Isotope Labs
20 Commerce Way
Woburn, MA 01801
617-938-0067
Thomas Barber
Group Leader
CIBA-GEIGY Corp.
410 Swing Rd.
Greensboro, NC 27409
919-632-7297
Robert Beimer
S-Cubed
3398 Carmel Mountain Road
San Diego, CA 92121
619-587-8448
Derek R. Berger
Chemist I
Lancaster Laboratories, Inc.
2425 New Holland Pike
Lancaster, PA 17601
717-565-2301
Richard G. Bpgar
Scientist
Weyerhaeuser Company
Weyerhaeuser Technology Center
WTC-2F25
Tacoma, WA 98477
206-924-6521
Paul Bookmyer
Supervisor
Stewart Labs
R. D. #1
Strattanville, PA 16258
814-379-3663
-------
72X
Eric L. Botnick
Lab Director
Electro-Analytical Labs
7118 Industrial Park Blvd.
Mentor, OH 44060
216-951-3514
M. James Boyer
Texas Dept. of Health
1100 W 49th
Austin, TX 78756
512-458-7587
Geoff Brieger
Oakland University
Dept. of Chemistry
Rochester, MI 48309
313-370-2325
Ray V. Buhl
Senior Chemist
WW Engineering & Science, Inc.
5555 Glenwood Hills Parkway, SE
Grand Rapids, MI 49508
616-942-9600
Jennings R. Byrd
Materials Engineer
Maryland Dept. of Transportation
2323 West Joppa & Falls Road
Brooklandville, MD 21022
301-321-3536
Ann C. Casey
SUNY Research Foundation
D-219 WCL & R
P.O. Box 509
Albany, NY 12201
518-473-7298
Dan Caudle
Conoco, Inc.
Division of DuPont
P.O. Box 2197
Houston, TX 77252
713-293-1246
Ray Christopher
Finnigan Corporation
355 River Oaks Parkway
San Jose, CA 95134
408-433-4800
Frederick Clayton
Instrument Chemist II
MWRD of Greater Chicago
550 S. Meacham Road
Schaumburg, IL 60193
708-529-7700 x280
David Compton
Products Manager
Bio-Rad
237 Putnam Avenue
Cambridge, MA 02139
617-499-4509
Paul A. Bouis
Assistant Director Analytical Res
J. T. Baker Inc.
600 N. Broad Street
Phillipsburg, NJ 08865
908-859-9443
Joel C, Bradley
President
Cambridge Isotope Labs
20 Commerce Way
Woburn, MA 01801
617-938-0067
Nancy A. Broyles
Advanced Chemist
Union Carbide
3200 Kanawha Turnpike
South Charleston, WV 25303
304-747-4707
Anne Burnett
Quality Control Officer
Environmental Testing Svcs., Inc.
P.O. Box 12715
Norfolk, VA 23502
804-461-3874
Angelo Carasea
Chemist
USEPA-OSWER/OPM
401 M Street, SW (OS-230)
Washington, DC 20460
202-260-7911
Pat Castelli
Application Chemist
Hewlett-Packard Company
Rt. 41 & Starr Road
Avondale, PA 19311
215-268-5562
James Chambers
Laboratory Manager
General Engineering Laboratories
P.O. Box 30712
Charleston, SC 29417
803-556-8171
Roger Claff
Environmental Scientist
American Petroleum Institute
1220 L Street NW
Washington, DC 20005
202-682-8399
Bruce N. Colby
President
Pacific Analytical
6349 Paseo Del Lago
Carlsbad, CA 92009
619-931-1788
Sandy Conley
Water Pollution Control Div.
DES, Arlington County
3401 South Glebe Road
Arlington, VA 22307
703-358-6821
-------
729
Brooke Connor
Supervisory Chemist
USGS
5293 B Ward Road
Arvada, CO 80002
303-467-8170
Paul Cramer
Midwest Research Institute
425 Volker Boulevard
Kansas City, MO 64110
816-753-7600
Laura J. Crane
Director, Laboratory Products
J. T. Baker, Inc.
222 Red School Lane
Phillipsburg, NJ 08865
908-859-2151
Susan Croy
Chemist
EnviroTech Mid-Atlantic
1861 Pratt Drive
Blacksburg, VA 24060
703-231-3983
Linda Darrington
General Engineering Lab
2040 Savage Rpad
Charleston, SC 29414
803-556-8171
Anne M. Davidheiser
RMC Environmental Services
88 Robinson St.
Pottstown, PA 19464
215-327-4850
Rhonda Day
Staff Chemist
Environmental Health Labs.
110 South Hill Street
South Bend, IN 46617
219-233-4777
Ivan DeLoatch
Environmental Scientist
USEPA-OW/GWDW
401 M Street S.W. (WH-550D)
Washington, DC 20460
202-260-3022
Ashok D. Deshpande
Chemist
USDOC, NOAA, NMFS, NEFC
Sandy Hook Laboratory
Highlands, NJ 07732
908-872-3043
Linda S. Donald
Organic Section Manager
Commonwealth Technology, Inc
2520 Regency Road
Lexington, KY 40503
606-276-3506
B. Rod Corrigan
Environmental Consultants Inc
391 Newman Avenue
Clarksville, IN 47129
812-282-8481
Bruce Crane
Enviro Market Manager
E,M Science
480 Democrat Road
Gibbstown, NJ 08027
800-222-0342
John P. Criscio
President
Absolute Standards, Inc.
498 Russell St.
New Haven, CT 06513
203-468-7407
Zorah Curry
Organic Lab Manager
Westinghouse
P.O. Box 398704
Cincinnati, OH 45239
513-738-9262
Joe Dautlick
Marketing Manager
Qhmicron
375 Pheasant Run
Newtown, PA 18940
215-860-5115
Tom L. Dawson
Group Leader
Union Carbide
3200 Kanawha Turnpike
South Charleston, WV 25303
304-747-5711
Dominick DeAngelis
Mobil Oil Corporation
P.O. Box 1027
Princeton, NJ 08543
609-737-4925
Jerry DeMenna
Laboratory Manager
Buck Scientific, Inc.
594 Dial Ave.
Piscataway, NJ 08854
908-752-8664
Therese desJardins
Analyst
Northeast Laboratory Services
P.O. Box 788
Waterville, ME 04901
207-873-7711
Michael R, D'Onofrio
Senior Environmental Chemist
Technical Services Labs, Inc.
1612 Lexington Avenue
Springfield, MO 65802
417-864-8924
-------
730
Michael Dostillio
Environmental Chemist
MedLab Environmental Testing Inc
312 Castle Ave.
Claymont, DE 19703
302-655-5227 x 58
Art Driedger
Wayne Analytical & Envrn. Services
992 Old Eagle School Road
Wayne, PA 19104
215-688-7485
Preston Dumas
Department Manager
Environmental Science & Eng., Inc
P.O. Box 1703
Gainesville, FL 32602
904-332-3318
Rolla M. Dyer
University of Southern Indiana
8600 University Blvd.
Evansville, IN 47712
812-464-1701
Andrew Ecklund
Free-Col Laboratories
P.O. Box 557
Cotton Road
Meadville, PA 16335
814-724-6242
Kenneth Edgell
Section Chief
The Bionetics Corporation
16 Triangle Park Drive
Cincinnati, OH 45246
513-771-0448
Dave Fada
Trace Organics Supervisor
Metro Environmental Laboratory
322 W. Ewing Street
Seattle, WA 98119
206-684-2303
Toni Favero
Supervisor of Instrumentation Lab,
North Shore Sanitary District
P.O. Box 750 Russell Ave.
Gurnee, IL 60031
708-623-6060
Ron FitzGibbon
Lab Technician
Metropolitan Sewer District
4522 Algonquin Parkway
Louisville, KY 40211
502-540-6735
Willard Douglas
Sverdrup Technology, inc.
Building 2423
Stennis Space Center, MS 35929
601-688-3158
Joshua Dubnick
Principal Lab Technician
Bergen County Utilities Authority
P.O. Box 122
Little Ferry, NJ 07643
201-807-5853
Gregory Durell
Research Scientist
Battelle Ocean Sciences
397 Washington St.
Duxbury, MA 02364
617-934-0571
Susan Dzurica
Reasearch Support Specialist
SUNY Research Foundation
Wadsworth Labs, D-219 WCL & R
P.O. Box 509
Albany, NY 12201
518-473-7298
Dave Edelman
Lab Manager
Columbia Analytical Services
1317 South 13th Avenue
P.O. Box 479
Kelso, WA 98626
206-577-7222
Valerie Evans
Product Manager
Triangle Laboratories of RTP, Inc
P.O. Box 13485
Durham, NC 27709
919-544-5729
John E. Farrell III
VP & Gen. Mgr. Eastern Region
Enseco, Inc.
2200 Cottontail Lane
Somerset, NJ 08873
908-469-5800
Kirby R. Feldmann
Sample Prep. Section Manager
Environmental Science & Engineering
8901 N. Industrial Road
Peoria, IL 61515
309-692-4422
Gary Folk
Technical Officer
IEA, Inc.
3000 Weston Parkway
Gary, NC 27513
919-677-0090
-------
731
Julie Fox
Bionetics Corporation
20 Research Drive
Hampton, VA 23666
804-865-0880
Nancy Friederich
Chemist
Midwest Research Institute
425 Volker Boulevard
Kansas City, MO 64110
816-753-7600
Warren Gardner
Sverdrup Technology
Building 2423
Stennis Space Center, MS 35929
601-688-1446
Lisa Gatton
Chief, Organic Chemistry Section
USEPA Region II
2890 Woodbridge Ave.
Edison, NJ 08837
908-906-6875
Christopher L. Getchell
Source Control Supervisor
City of Tacoma, USTS Laboratory
2201 Portland Avenue
Tacoma, WA 98421
206-591-5588
Dean Gokel
President
GeoChem, Inc.
2500 Gate Way Center Blvd.
Suite 300
Morrisville, NC 27560
919-460-8093
James I. Green
Chemist
National Environmental Testing
100 Grove Road
Thorofare, NJ 08086
609-848-3939
John P. Gute
Laboratory Supervisor
LA County Sanitation Districts
1965 Workman Mill Road
Whittier, CA 90601
310-699-0405 x 3031
Don Haertel
Research Proj. Supervisor
Jim Walter Research Corp.
10301 Ninth Street North
St. Petersburg, FL 33716
813-576-4171
Drew Francis
Quality Assurance Officer
Hampton Roads Sanitation District
1436 Air Rail Ave.
Virginia Beach, VA 23455
804-460-2261
Guy Galleilo
Senior Staff Chemist
Analytical Services Corporation
16406 US Route 224 East
Findlay, OH 45840
419-423-3526
Jerry Garvis
Supervisor
Stewart Labs
R. D. #1
Strattanville, PA 16258
814-379-3663
Denise Sn Geier
Laboratory Director
Analytical Services, Inc.
390 Trabert Ave. NW
Atlanta, GA 30309
404-892-8144
A.J. Gilbert
Technical Director
V.G. Masslab
Crewe Road
Manchester, UK M239BE
061-946-1060
Harold M. Goldston, Jr.
Analytical Chemist
Environmental Laboratories, Inc.
9211 Burge Ave.
Richmond, VA 23237
804-271-3440
John C. Green
Research Associate
TN Tech University Water Center
Box 5033
Cookeville, TN 38505
615-372-3843
David Haddaway
Senior Chemist
City of Portsmouth
105 Maury Place
Suffolk, VA 23434
804-539-7608
Donald F. Hagen
Corporate Scientist
3M Company
3M Center, 201-1W-29
St. Paul, MN 55144
612-733-6978
-------
732
Gary Hahn
Laboratory Manager
Ecology & Environment,
4285 Genesee Street
Buffalo, NY 14225
716-631-0360
Inc.
Donald Hammer
NEESA
Code 112-E2, Building 835
Port Hueneme, CA 93043
805-982-2633
Bill Hardesty
Chemist
Viking Instruments Corp.
12007 Sunrise Valley Dr.
Reston, VA 22170
703-758-9339
Riazul Hasan
Supervisor
Bergen County Utilities Authority
P.O. Box 122
Little Ferry, NJ 07643
201-807-5855
Elaine T. Hasty
Applications Chemist
CEM Corporation
P.O. Box 200
Matthews, NC 28106
704-821-7015
Sheila Heath
Laboratory Scientist
State of NH Dept. of Envrn. Svcs.
Health & Welfare Bldg. Hazen Dr
Concord, NH 03301
603-271-3426
Michael F. Helmstettee
Laboratory Manager
Applied Marine Research Laboratory
College of Science
1043 W. 45th Street
Norfolk, VA 23529
804-683-4787
Michael Herbert
Technologist
Baxter Health Care
Rt. 120 & Wilson Road
Round Lake, IL 60073
708-546-6311
Kathy J. Dien Hillig
Manager Ecology Analytical Services
BASF Corp.
1609 Biddle Ave,
Wyandotte, MI 48192
313-246-6334
Geoff Hinshelwood
Laboratory Manager
Environmental Testing Svcs Inc
P-0, Box 12715
Norfolk, VA 23502
804-461-3874
Jeffrey W., Halvorson
Chemist
Baxter, Burdick & Jackson
1953 South Havery St.
Muskegon, MI 49442
616-726-3171
Rich Hamon
AECL Research
Whiteshell Laboratories
Pinawa, Canada, MB ROE1LO
204-753-2311
Ken Hart
Free-Col Laboratories
5815 Airport Road, Suite A-2
Roanoke, VA 24012
703-265-2544
David Haske
Chemist
Roche Analytics
8040 Villa Park Drive
Richmond, VA 23228
804-264-7100
Rex Hawley
VSPP
24201 Frampton Ave
Harbor City, CA 90710
310-539-6490
Jerry Hedrick
Laboratory Technician
Environmental Testing Svcs., Inc.
P.O. Box 12715
Norfolk, VA 23502
804-461-3874
Mike Heniken
City of Columbus, Div. of S&D, Lab
900 Dublin Road
Columbus, OH 43215
614-645-7016
Jenifer Hess
Group Leader
Lancaster Labs, Inc.
2425 New Holland Pike
Lancaster, PA 17601
717-656-2301
Aston Hinds
Vice President, Envrn. Services
Baroid Drilling Fluids, Inc.
P.O. Box 1675
Houston, TX 77251
713-987-4468
Paula A. Hogg
Hampton Roads Sanitation District
101 City Farm Road
Newport News, VA 23602
804-874-1287
-------
733
Dawn Holdren
Chemist
NASA
P.O. Box 44
Wallops Island, VA 23337
804-824-1761
Tim Holt
Laboratory Manager
Trace Analytical Laboratories, Inc
2241 Black Creek Road
Muskegon, MI 49444
616-773-5998
B. James Hood
Professor
Middle Tennessee State University
Box 68, MTSU
Murfreesboro, TN 37132
615-898-2074
George D. Howe11
Chemist
Naval Supply Center, Fuel Dept.
Quality Assurance Division
(Code 702)
Norfolk, VA 23512
804-444-2761
Greg Hudson
Lab Director
Envirocompliance Laboratories, Inc.
1 Maple Leaf Court
Ashland, VA 23005
804-550-3971
R. Tracy Hunter
Chemist Supervisor
Commonwealth of Virginia
1 North 14th Street
Richmond, VA 23219
804-786-4898
Nang Huynh
Lab Manager
National Lab, Inc.
3210 Claremont Ave.
Evansville, IN 47712
Tony Jarkowski
Section Supervisor
Eastman Kodak
Kodak Park, Chemical Quality Svcs
Bldg. 34
Rochester, NY 14650
716-477-5681
Martha Johnson
Environmental Engineer
Horizon Technology
P.O. Box 540, 25 Brown Ave.
Hampstead, NH 03842
603-329-5611
Anthony A, Holt
Johnson County Environmental Dept
P.O. Box 39/4800 Nail
Mission, KS 66201
913-432-3868
Ben Honaker
Chemist
USEPA, OW/OST
401 M Street, SW. (WH-552)
Washington, DC 20460
202-260-2272
Sonny Hopper
Associate Chemist
Eastern Municipal Water District
P.O. Box 8300
San Jacinto, CA 82581
714-925-7676
Han-Ping Huang
Chief Chemist
James R. Reed & Associates, Inc.
813 Forrest Drive
Newport News, VA 23606
804-599-6750
Frank Hund
Chemist
USEPA-OW/OST
401 M Street, SW (WH-552)
Washington, DC 20460
202-260-7182
Lisa Hutter
Stragetic Diagnostics, Inc
128 Sandy Drive
Newark, DE 19713
302-456-6789
Denny J. Ivey
President
Environmental Labs & Services
P.O. Box 1408
Carrollton, GA 30117
404-832-2171
Richard A. Javick
Research Associate
FMC Corporation
P.O. Box 8
Princeton, NJ 08543
609-520-3639
Robert Johnson
Horizon Technology
P.O. Box 540, 25 Brown Ave
Hampstead, NH 03841
603-329-5611
-------
734
Phanibhushan B. Joshipura
Chemist
Naval Supply Center, Fuel Dept.
Quality Assurance Division
(Code 702)
Norfolk, VA 23512
804-444-2761
Steve Kahl
Analyst
Fire & Envrn. Consulting Labs, Inc
1451 E. Lansing Dr., #222
East Lansing, MI 48823
517-332-0167
Michael Kauffman
Envrn. Testing & Consulting, Inc.
2924 Walnut Grove Road
Memphis, TN 38111
901-327-2750
R. Michael Kennedy
Laboratory Supervisor
City of Rock Hill/Env. Mon. Lab.
P.O. Box 11706
Rock Hill, SC 29731
803-329-8704
Brenda King
Senior Chemist
American Medical Laboratories
11091 Main Street
Fairfax, VA 22030
703-691-9100
William Kirk
CEO
Reliance Laboratories
P.O. Box 625
Bridgeport, WV 26330
304-842-5285
Dewey R. Klahn
Lab Manager
Environmental Science Corp.
1910 Mays Chapel Rd.
Mt. Juliet, TN 37122
615-758-5858
Margaret Knight
USEPA Region X, Manchester Lab
P.O. Box 549
Manchester, WA 98353
206-871-0748
Alan Kramme
Product Development
ACE Glass, Inc.
1430 Northwest Blvd.
Vineland, NJ 08360
609-692-3333
Gregor A. Junk
Associate
Ames Laboratory (USDOE)
Iowa State University
Ames, IA 50010
515-294-9488
Victor F. Kalasinsky
Chief, Div. of Envrn. Toxicology
Armed Forces Inst. of Pathology
14th Street & Alaska Avenue
Washington, DC 20306
202-576-2434
Kevin W. Keeley
Laboratory Director
Great Lakes Analytical
1380 Busch Parkway
Buffalo Grove, IL 60089
708-808-7766
Mohan Khare
Envirosystems, Inc.
9200 Rumsey Road, Suite B102
Columbia, MD 21045
410-964-0330
James R. King
Viar and Company
300 N. Lee Street, Suite 200
Alexandria, VA 22314
703-684-5678
Denni Kirtley
Group Leader
General Engineering Laboratories
P.O. Box 30712
Charleston, SC 29417
803-556-8171
Andrew Kluger
New Castle County Government
100 New Churchmans Road
New Castle, DE 19720
302-322-5897
Kathy A. Knowles
Environmental Chemist
Delaware DNREC
89 Kings Highway
P.O. Box 1401
Dover, DE 19903
302-739-4771
William'G. Krochta
Manager
PPG Industries
440 College Park Drive
Monroeville, PA 15146
412-325-5183
-------
735
Vincent Kuyawa
Lab Manager
Martel
1025 Cromwell Bridge Road
Baltimore, MD 21204
301-825-7790
Joan LaRock
3M Consultant
801 Pennsylvania Avenue NW #1213
Washington, DC 20004
202-628-4322
David Leonard
Applications Chemist
Fisons Instruments
1850 Lake Park Dr., Suite 105
Smyrna, GA 30080
404-438-7044
Joseph Loeper
Technical Manager
Roy F. Weston
208 Welsh Pool Road
Lionville, PA 19341
215-524-7360
Kristin K. Madden
Organics Dept. Manager
Stearns & Wheler Laboratory, Inc
7280 Caswell Street,
Hancock Air Park
N. Syracuse, NY 13212
315-458-8033
Carol Malone
QA/QC Coordinator
Jennings Laboratories, Inc.
1118 Cypress Avenue
Virginia Beach, VA 23451
804-425-1498
Michael F. Martin
Commonwealth of Va., DGS/DCLS
1 North 14th Street
Richmond, VA 23219
804-371-2874
Barbara 0. McCleary
Environmental Chemist
Delaware DNREC
89 Kings Highway
P.O. Box 1401
Dover, DE 19903
302-739-4771
Patrick McMahon
Vice President
Advanced Systems, Inc
P.O. Box 8090
Newark, DE 19714
Frank Lamb
Technical Manager
Burdick & Jackson
P.O. Box 214
Millersville, MD 21108
410-647-3905
Peter A. Law
Laboratory Manager
Tighe & Bond Laboratory
53 Southampton Road
Westfield, MA 01085
413-572-3200
Nathan Levy
A & E Testing, Inc.
1717 Seabord Drive
Baton Rouge, LA 70810
504-769-1930
Norman Low
Environmental Product Manager
Hewlett-Packard
1601 California Ave.
Palo Alto, CA 94304
4415-857-7381
James E. Maguire
Roche Analytics
8040 Villa Park Drive
Richmond, VA 23228
804-264-7100
Chung-Rei Mao
Chemist
Corps of Engineers
Missouri River Division Lab
Omaha, NE 68102
402-444-4304
Helen T. McCarthy
Supervising Public Health Chemist
RI Department Health Labs
50 Orms Street
Providence, RI 02904
401-274-1011
Robert J. McDaniel
Instrument Specialist
Applied Marine Research Laboratory
College of Science
1043 W. 45th St.
Norfolk, VA 23529
804-683-4787
John Melvin
President
PEL, Inc.
9405 S.W. Nimbus Avenue
Beaverton, OR 97005
503-671-0885
-------
736
Rodney T. Miller
Corporate Quality Assurance Officer
PACE, Inc.
1710 Douglas Drive North
Minneapolis, MN 55422
612-525-3465
Raymond Mindrup
Manager, Marketing
Supelco, Inc.
Supelco Park
Bellefonte, PA 16823
814-359-3441
Gregory B. Mohrman
Supervisory Chemist, Prog. Mgr.
Rocky Mountain Arsenal
Laboratory Support Division
Attn: AMXRM-LS
Commerce City, CO 80022
303-289-0217
Ken Mora
Viar and Company
300 N. Lee Street, Suite 200
Alexandria, VA 22314
Harry V. Myers
Senior Project Manager
Keystone Envirn. Resources, Inc
3000 Tech Center Drive
Monroeville, PA 15146
412-825-9818
John R. Nein
Chemist
Chesapeake Paper Products Co.
19th & Main Streets
West Point, VA 23181
804-843-5750
Anne D. O'Donne11
Group Leader, Organics
WMI Environmental Monitoring
Laboratories, Inc.
21 Cleanwater Drive
Geneva, IL 60134
708-208-3100
Robert G. Orth
Monsanto Co. - U4E
800 North Lindbergh Blvd.
St. Louis, MO 63167
314-694-1463
Veriti Overby
Chemist
Naval Supply Center, Fuel Dept.
Quality Assurance Division
(Code 702)
Norfolk, VA 23512
804-444-2761
Ian Milnes
Lab Manager
Wright Lab Services, Inc
3 4 Dogwood Lane
Middletown, PA 17057
717-944-5541
Jeffrey K. Mitchell
Market Division Manager
3M
3M Center, 220-9E-10
St. Paul, MN 55144
612-736-9365
Marlene Moore
President
Advanced Systems, Inc.
P.O. Box 8090
Newark, DE 19714
302-834-9796
Violetta F. Murshak
Vice President
FECL
1451 E. Lansing Dr. #222
East Lansing, MI 48823
517-332-0167
Linda Neal
Senior Research Chemist
Ashland Petroleum Co.
P.O. Box 391
Ashland, KY 41129
606-327-6755
Lydia Nolan
Research Chemist
Supelco, Inc.
Supelco Park
Bellefonte, PA 16823
814-359-5708
Steven Ortel
Senior Chemist
Potomac Electric Power Company
3300 Benning Rd., N.E.
Washington, DC 20019
202-388-2551
Nancy Osterhoudt
Ogden Environmental & Energy
Services, Inc.
3211 Jermantown Road
Fairfax, VA 22030
703-246-0596
Robert G. Owens, Jr.
Chief Chemist
Analytical Services, Inc.
390 Trabert Ave. NW
Atlanta, GA 30309
404-892-8144
-------
737
Susan E. Park
PPB Environmental Labs
6821 SW Archer Road
Gainsville, FL 32608
904-377-3-2349
Inc
Jay Perkins
Duke Power
13339 Hagers Ferry Road
Huntersville, NC 28078
704-875-5348
Robert K. Pertuit
PPG Industries, Inc.
P.O. Box 1000
Lake Charles, LA 70602
318-491-4700
William Pfeiffer
President
Ginosko Laboratories, Inc.
17875 Cherokee Street
Harpster, OH 43323
614-496-4051
Rebecca Plemons
Lab Manager
Reliance Laboratories
P.O. Box 625
Bridgeport, WV 26330
304-842-5285
Lee Polite
Research Chemist
Amoco Corp.
P.O. Box 3011 MS F-7
Naperville, IL 60566
708-420-3110
Jean F. Pugin
Staff Scientist/Project Manager
S-Cubed (Maxwell Labs Div.)
1800 Diagonal Road, Suite 400
Alexandria, VA 22314
703-838-0220
Margaret Randel
Senior Consultant
Arthur D. Little, Inc.
Acorn Park
Cambridge, MA 02140
617-864-5770 X2697
Katharine Raynor
Director Quality Assurance Div.
Naval Supply Center, Fuel Dept.
Quality Assurance Division
(Code 702)
Norfolk, VA 23512
804-444-2761
Leah Reed
Program Manager
Viar and Company
300 N. Lee Street Suite 200
Alexandria, VA 22314
703-519-1240
Joseph Peters
Sr. Field Marketing Manager
Millipore Corporation
397 Williams Street
Marlborough, MA 01752
508-624-8560
Marvin D. Piwoni
Laboratory Manager
Hazardous Waste Research and
Information Center
One E Hazelwood Drive
Champaign, IL 61820
217-244-9803
Roy W. Plunket, Jr.
Analytical Chemist Supervisor
Commonwealth of Va., DGS/DCLS
1 North 14th Street
Richmond, VA 23219
804-225-4007
Joseph Price
Environmental Manager I
Alabama Dept. of Envrn. Management
4043 Faunsdale Drive
Montgomery, AL 36109
205-261-2736
Phoko Ramarumo
Suny Research Fondation
D-219 WCL & R
P.O. Box 509
Albany, NY 12201
518-473-7298
Tom Randolph
Senior Staff Environmental Eng.
Shell Offshore, Inc.
P.O. Box 61933
New Orleans, LA 70161
504-588-6468
Sudhakar Reddy
Director of Research
Shrader Laboratories
3814 Vinewood
Detroit, MI 48208
313-894-4440
Stephen E. Reeves
Union Camp Corporation
P.O. Box 178
Franklin, VA 23851
804-569-4830
-------
John Reynolds
Section Manager
DataChem Laboratories
960 West LeVoy Drive
Salt Lake City, UT 84123
801-266-7700
Anita Rigassio
Environmental Chemist
COM Federal Programs Corporation
98 North Washington St., Ste. 200
Boston, MA 02114
617-742-2659
Nancy Rothman
Enseco, Inc.
205 Alewife Brook Parkway
Cambridge, MA 02138
John Roy
Project Leader
Dow Chemical
1602 Building
Midland, MI 48667
517-638-6912
Anna M. Rule
Chief Laboratory Division
Hampton Roads Sanitation District
1436 Air Rail Ave.
Virginia Beach, VA 23455
804-460-2261
Ed Saltzberg
Viar and Company
300 N. Lee Street, Suite 200
Alexandria, VA 22314
703-684-5678
George A. Schmitt
Program Manager
3M Company
3M Center, Bldg. 220-9E-10
St. Paul, MN 55144
612-733-0307
Warren Schultz
MN Valley Testing Lab.
1126 N. Front Street
New Ulm, MN 56073
507-354-8517
Lisa Secrest
Scientist
ManTech Environmental
Kerr Lab, Kerr Lab Road
Ada, OK 74820
405-332-8800
Heather Shandor
Tri-State Labs
19 East Front Street
Youngstown, OH 44503
216-746-8800
Lynn Riddick
Viar and Company
300 N. Lee Street, Suite 200
Alexandria, VA 22314
703-684-5678
Elsie Riggs
CCR, Inc.
124 East Cork St.
Winchester, VA 22601
703-667-0600
Hope Rovira
Stragetic Diagnostics, Inc
128 Sandy Drive
Newark, DE 19713
302-456-6789
Mariser Ruiz
Florida Power & Lights
6001 Village Boulevard
West Palm Beach, FL 33407
Phil Ryan
ATI - Colorado
225 Commerce
Ft. Collins, CO 80524
303-490-1511
Aisling M. Scallan
Senior Product Manager
EnSys, Inc.
P.O. Box 14063
Research Triangle Park, NC 27709
919-941-5509
William C. Schnute
Environmental Marketing Manager
Finnigan
355 River Oaks Parkway
San Jose, CA 95134
408-433-4800
Jan Sears
Project Manager
Ogden Environmental & Energy
Services, Inc.
3211 Jermantown Road
Fairfax, VA 22030
703-246-0306
Andy Sendelbach
Sales Manager
Varian
1829 Grande Oaks Road
Durham, NC 27712
919-477-1015
Cathi Sharp
Organics Section Manager
ETS Analytical Services
1401 Municipal Road
Roanoke, VA 24012
703-265-0004
-------
739
Charles G. Shaw
Senior Analyst
SAIC/CVR
8500 Cinder Bed Road
Newington, VA 22122
703-550-0430
Michelle Shearer
Tri-State Laboratories
19 East Front Street
Youngstown, OH 44503
216-746-8800
Chris Shumate
695' E. Patriot Blvd. No.
Reno, NV 89511
702-851-8110 Home #
Joseph Slayton
Technical Director/Sr. Scientist
EPA Region III
Central Regional Laboratories
839 Bestgate Road
Annapolis, MD 21410
410-266-9180
Daniel P. Smith
Zande Environmental Service
1233 Dublin Road
Columbus, OH 43215
614-486-4383
Terry Smith
Organics Section Manager
US PC I
4322 S. 49th West Avenue
Tulsa, OK 74105
918-446-1162
Dave Solomon
Varian Associates
2104 Stonequarter Court
Richmond, VA 23233
804-740-8907
David N. Speis
V.P. & Director QA and Tech,
ETC Corp.
284 Raritan Center Parkway
Edison, NJ 08818
908-225-6759
Jim Stave
Stragetic Diagnostics, Inc.
128 Sandy Drive
Newark, DE 19713
302-456-6789
Eric Steindl
Chemical Stds/Accessories Mgr
Restek Corporation
110 Benner Circle
Bellefonte, PA 16823
814-353-1300
Timothy A., Shaw
Analytical Services, Inc.
390 Trabert Ave. NW
Atlanta, GA 30309
404-892-8144
Christopher Shugarts
Accu-Labs Research, Inc.
4663 Table Mountain Drive
Golden, CO 80403
303-277-9514
David Singer
Sales Representative
Tekmar Company
P.O. Box 429576
Cincinnati, OH 45242
800-543-4461
Colleen Smith
Chemist
EnviroTech Mid-Atlantic
1861 Pratt Drive
Blacksburg, VA 24060
703-231-3983
Roy-Keith Smith
Analytical Methods Manager
Analytical Services, Inc.
390 Trabert Ave. NW
Atlanta, GA 30309
404-892-8144
Ronald D. Snelling
Research Associate
Institute for Environmental Studies
Louisiana State University
Baton Rouge, LA 70803
504-388-4305
R. Kent Sorrell
Chemist
USEPA, Region V
26 W. M.L. King Drive
Cincinnati, OH 45268
513-569-7943
Lisa Spinelli
Technical Representative
Millipore Technical Service
397 Williams St.
Marlboro, MA 01752
800-722-5998 x8654
Maureen Steadman
Laboratory Technician
Environmental Testing Svcs.
P.O. Box 12715
Norfolk, VA 23502
804-461-3874
Dennis Stocker
Group Leader
Agri-Diagnostics Associates
2611 Branch Pike
Cinnaminson, NJ 08077
609-829-0110
Inc
-------
740
William P. Stork
Environmental Analysis, Inc.
3273 N. Hwy 67 (Lindbergh Blvd.)
Florissant, MO 63033
314-921-4488
Cindy Stuefer-Powell
Research Technologist
University of Nebraska
567 PS
Lincoln, NE 68583
402-472-1633
Nancy A. Tanner
Process Chemist
E.I. DuPont
Chambers Works/QCL
Deepwater, NJ 08023
609-540-2810
Roger Thomas
Environmental Task Manager
Viar and Company
300 N. Lee Street, Suite 200
Alexandria, VA 22314
703-684-5678
David Tompkins
President
ETS Analytical Services
1401 Municipal Road
Roanoke, VA 24012
703-265-0004
Felicitas Trinidad
Environmental Supervisor
Hoffmann-LaRoche, Inc.
340 Kingsland Street
Nutley, NJ 07110
201-235-3131
Connie Van Dyke
Section Chief, Envrn. Ser. Program
MO Dept. of Natural Resources
P.O. Box 176
Jefferson City, MO 65102
314-526-3328
Theodore Varouxis
Associated Design & Mfg. Co.
814 N. Henry Street
Alexandria, VA 22314
703-549-5999
Al Vicinie
Supervisor, Industrial Laboratory
Deyor Labs
Southwoods Medical-Health Complex
7655 Market St. Ste 2500
Youngstown, OH 44512
800-365-3396
Andrew J. Strebel
Technical Specialist I
Lancaster Laboratories
2425 New Holland Pike
Lancaster, PA 17601
717-656-2301 x526
E. Louise Stunkard
Physical Science Technician
USAEHA
Analytical Quality Assurance Div.
Building E2100
Aberdeen Proving Ground, MD 21010
410-671-3268
Paul Taylor
President
Taylor Technology, Inc.
350 Alexander Road
Princeton, NJ 085040
609-921-6715
James Todaro
Matrix Analytical, Inc.
106 South Street
Hopkinton, MA 01748
508-435-6824
Lisa Traut
Chemist
Environmental Resource Associates
5540 Marshall Street
Aruada, CO 80002
303-431-8454
Mark Tuttle
MicroSep Membranes
632 N.W. Vicksburg Ave.
Bend, OR 97701
503-389-4525
Frederic B. Vanderherchen
Senior Chemist
City of Richmond, Pub. Ut./WWT
1400 Brander St.
Richmond, VA 23224
804-780-5338
Joe Viar
Viar and Company
300 N. Lee Street, Suite 200
Alexandria, VA 22314
703-684-5678
Orterio Villa
Laboratory Director
USEPA, Region III
Central Regional Laboratory
839 Bestgate Road
Annapolis, MD 21401
301-266-9180
-------
741
Joseph S. Vitalis
Chemical Engineer
USEPA-OW/EAD
401 M Street, SW (WH-552)
Washington, DC 20460
202-260-7172
Randy Ward
Chief Chemist
Environmental Science Corp
1910 Mays Chapel Road
Mt. Juliet, TN 37122
615-758-5858
Susan Weisheit
Product Manager
General Analysis Corporation
140 Water Street
S. Norwalk, CT 06856
800-327-2460
Richard White
Dow Chemical Company
734 Building
Midland, MI 48667
517-636-4896
Idelis Z. Williams
Project Manager
Betz Analytical Services
9669 Grogans Mill Road
The Woodlands, TX 77380
713-367-6201
Hugh Wise
USEPA-OW/EAD
401 M Street, SW (WH-552)
Washington, DC 20460
202-260-7177
Mark Witham
National Sales Manager
Bio-Tek Instruments
Box 998, Highland Park
Winopski, VT 05405
802-655-4040 x 227
Michael W. Woods
Manager of Services
Technical Services Labs., Inc.
1612 Lexington Avenue
Springfield, MO 65802
417-864-8924
Susan C. Wyatt
Technical Manager
Enseco - Rocky Mt. Analytical Lab.
4955 Yarrow Street
Arvada, CO 80002
303-421-6611
Tonie M. Wallace
President
County Court Reporters, Inc.
124 East Cork Street
Winchester, VA 22601
703-667-6562
James G. Ware
Sr. Technician/GC Specialist
Babcock & Wi^cox Co.
Nuclear & Envrn. Services Group
P.O. Box 785, Mt. Athos Rd.
Lynchburg, VA 24508
804-522-5188
Charles Weston
Technical Manager
ETC Corp
284 Raritan Center Parkway
Edison, NJ 08818
908-225-6784
David I. Wigger
AL Dept. of Environmental Mgmt.
2204 Perimeter Road
Mobile, AL 36695
205-450-3400
Greg Winslow
Senior Chemist
Texaco
5901 S. Rice Street
Belaire, TX 77401
713-432-3593
Dennis Wisler
Pretreatment Program Coordinator
Water Pollution Control Div.
DES, Arlington County
3401 South Glebe Road
Arlington, VA 22307
703-358-6881
Ira C. Woods
Environmental Chemist
Boeing-Renton SHEA
P.O. Box 3707
Seattle, WA 98124
206-965-0091
Areta Wowk
NJDEPE, Pesticide Control Program
Scotch Road, CN 411-380
Trenton, NJ 08625
609-530-5161
Larry Yen
Senior Consulting Scientist
Millipore Corporation
80 Ashby Road
Bedford, MA 01730
617-275-9200
-------
742
Steve Yocklovich
VP, Dir. of Chromatographic Science
Burlington Research, Inc.
615 Huffman Mill Road
P.O. Box 2481
Burlington, NC 27215
919-584-5564
Gunars Zikmanis
Ohio EPA
3671 W. 230th Street
North Olmsted, OH 44070
216-734-4783
John Young
Westinghouse Savannah River Co.
Savannah River Technology Center
773A
Aiken, SC 29808
803-725-3565
Nancy Zikmanis
Environmental Scientist
Ohio EPA
3671 West 230th Street
North Olmsted, OH 44070
216-734-4983
Rudolph Zsolway
USEPA
-------