-------
API ANALYTICAL SYSTEMS TASK FORCE (W-22)
FIGURE 9
ANALYSES OF FIELD BLANK SAMPLES - VOLATILE ORGANICS (yg/i)
Location
1
2
2
0
0
0
1
3
2
1
2
3
1
1
2
1
2
3
3
1
2
3
1
2
3
3
0
0
1
2
3
1
2
3
CODE
1
2
3
0
Dace
6
6
6
1
3
6
6
6
6
6
6
6
1
3
1
2
3
1
3
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
LOCATION
Intake water
Uafttewatur feed
Effluent
Blank Jsuoclatcd
Laboratory
EPA
EPA
EPA
Company
Company
Radian
EPA
EPA
EPA
EPA
EPA
EPA
Company
Company
Company
Company
Company
Company
Company
EPA
EPA
EPA
EPA
EPA
EPA
Company
Radian
Radian
Radian
Radian
Radian
Radian
Radian
Radian
to blotreatment
with no particular
Parameters Detected
In Blank Samples
Hethylene chloride
Hethylene chloride
Hethylene chloride
Hethylene chloride
Hethylene chloride
1,2 Dlchloroethane
Hethylene chloride
1,1,1 Tlchloroechane
Benzene
Toluene
Kthylbrnzene
Hethylene chloride
Hechylene chloride
Benzene
Hethylene chloride
Hechylene chloride
Chloroform
Hechylene chloride
Hechylene chloride
Chloroform
Hechylene chloride
Hechylene chloride
Hechylene chloride
Hethylene chloride
Hethylene chloride
Hechylene chloride
Hechylene chloride
Kethylene chloride
Hechylene chloride
Hechylene chloride
Hechylene chloride
Hechylene chloride
Hechylene chloride
Benzene
Tolueae
Echylbenzene
Hechylene chloride
Chloroform
Trlchloroechylene
Toluene
Ethylbenzene
Hethylene chloride
Hechylene chloride
Methylene chloride
Hethylene chloride
Toluene
Hethylene chloride
Trlchloroethylene
Kethylene chloride
Trichloroethylene
Toluene
Hechylene chloride
Trlchlirocchytene
Toluene
COPE
1
2
location within the refinery 6
Reported
Blank
Result
35
70
40
D (<1)
D (<1)
D (<0.5
50
3
2
5
2
10
40
10
30
D (<5)
70
D (<5)
D (<5)
D (<5)
800
11
740
280
25
365
179
G (100)
C (100)
G (100)
50
10
50
33
41
9
38, 78
1, 2
1
1, 1
1. 1
63
7
6
6
2
12
1
22
1
1
19
1
I
RATE
Plrst day
Second day
Third day
Compos ice
Sample Rvsulcs
Associated With Blank
50
10
50
0.6, 1, 1
0.4, 0.9, 0.5
) ND (<1). 51, D (<.5)
10, 30, 300, D (<5)
2, MD (<.3), 10, 3. 2, 3
1, 50, 30, 1, 1, 30
6, 130, 50, 3, 3, 60
2, 70, 100, 1, 1, 20
G (100)
NO (<40)
6
G (100)
c (so)
70
G (50)
G (50)
0 (<5)
32
80
10
16
10
44
10
50
10
10
D (<10)
70
D (<10)
ND {
-------
API ANALYTICAL SYSTEMS TASK FORCE (W-22)
FIGURE 10
CONCLUSIONS
THE EPA METHODOLOGY IS INCAPABLE OF DETECTING
AND ACCURATELY QUANTIFYING INDIVIDUAL ISOMERIC
POLYNUCLEAR AROMATIC HYDROCARBONS.
RECOVERIES FOR TRACE ELEMENTAL AND ORGANIC
PRIORITY POLLUTANTS VARIED WIDELY.
THE VARIABILITY OF THE RESULTS OBTAINED FROM
INTERLABORABORY AND INTRALABORATORY ANALYSES
DEMONSTRATES THAT INDIVIDUAL DATA POINTS
SHOULD BE VIEWED IN TERMS OF RANGES RATHER
THAN EXACT NUMBERS.
A REVIEW OF THE VOLATILE ORGANIC BLANK
ANALYTICAL DATA STRONGLY SUGGESTS THAT
INADVERTENT CONTAMINATION OCCURRED.
METHYLENE CHLORIDE, TOLUENE AND BENZENE
ARE THE MOST OBVIOUS PARAMETERS FOR WHICH
CONTAMINATION IS SUSPECTED.
17J
-------
Methods Development - Status
J. Longbotttom
U. S. Environmental Protection Agency
Mr. Longhottorn: For those of you that are not familiar with
our program, the Environmental Monitoring and Support Laboratory
has the responsibility under Section 304(h) of the Clean Water Act
to recommend test procedures for measuring pollutants in industrial
discharges. Our work began as soon as a firm list of the 129 pri-
ority pollutants came out. We took the 114 organic compounds and
issued an RFP for contract research to develop methods that would
be required by the Agency for monitoring discharges after the dis-
charge limitations are set. We arbitrarily divided the 114 com-
pounds up into 12 categories to make more workable groups for the
contractors and set up two-phase programs for each category. In
Phase I of their research, they were to work on tasks such as the
optimization of GC conditions. They compare different extracting
techniques, conducting experiments using a couple of different
solvents at 3 different pH's. They look at compound stability and
ways of preserving samples at different pH conditions. They eval-
uate the effect of chlorine on the stability of the compounds and
whether the sampler will have to dechlorinate the sample to pre-
serve it. As they end the first phase, they experiment with poten-
tial cleanup procedures that might have application to wastewaters.
In the second phase of the research, they will go out and collect
wastewater samples and trouble shoot the method. Finally, they
will use the samples to validate the application of the method for
industries that would be be expected to be regulated. The final
package will include accuracy and precision data that was obtained
in a single laboratory. In a follow-up program, we will be con-
ducting an inter-laboratory study to fully define sensitivity,
accuracy, and precision, but I am not directly involved with that
program.
At the time of this meeting, our contractors are just about
reaching the end of the Phase I studies. They have completed all
of the chemical studies: the chromatography; the extraction and
preservation studies. They are now preparing to evaluate and mod-
ify the method as required to handle wastewaters. We are trying
to coordinate their wastewater collection with ongoing Effluent
Guidelines programs to ensure that we get relevant wastes that
will allow our contractors to really trouble-shoot the methodology.
Our basic game plan for the methods development was twofold.
Our first priority was to develop the best procedure for a cate-
gory of compounds (e.g. phthalates). Second, we encouraged the
contractor to use approaches that may have more universal appli-
cation - to pesticides, for example. That way, the analyst who
must measure both phthalates and pesticides will be able to use
the information provided to assemble a compromise method. The
compromise method may result in five percent less recovery for
his phathalates because of the choice of solvents, but simul-
taneously he will be able to run the two parameters together and
it will cost less for the analysis. We are trying to work both
18
-------
approaches into our contracts: the best method for each of our
12 categories and the best analytical system that would include
the cost-effective approach for any combination of the organics.
I would like to review the research in terms of our twelve
categories that we set up - the kind of options we're exploring
and the types of methods we are considering for promulgation as
a result of this research. Bear in mind that although I am dis-
cussing them in terms of twelve categories, we are also trying
to mesh all of the compounds together into a complete analytical
system at the same time.
First, in our laboratory, we identify 26 of the priority
pollutants as purgeables, with 23 of them being halogenated.
We waiver on the classification of a few additional compounds,
such as the dichlorobenzenes, which can be purged but not well
resolved with existing purge chromatography. So, some lists
you see may extend the applicaiton up to 1,2,4-trichlorobenzene
in contradiction to our own twelve category classification.
What is being researched for the halogenated purgeable, of
course, is the use of purge-and-trap with a Hall detector. We
are also conducting parallel studies of liquid extractions - the
single extraction approach using pentane or some other hydro-
carbon. There is a lot of interest in the latter approach, as
you know. Many reseachers are working with drinking water sam-
ples using the simple small-volume extraction and an electron
capture detector. In-house, we have always felt that pentane
extraction would probably not be practical for a complex waste-
water, and that most analysts would rather use the purge-and-
trap to take advantage of the Hall detector's selectivity and
compatibility with column temperature programming. Our study
will look at the extraction procedures, however, to determine
their applicability and limitations and propose them as accept-
able alternatives where appropriate.
In our own well-equipped laboratory, we would analyze the
halogenated purgeables with purge-and-trap into a Hall detector.
The three non-halogenated priority pollutants (benzene, toluene,
and ethylbenzene) would be analyzed by using a splitter at the
end of the column and running a split into a photoionization
detector. The photoionization detector is more sensitive than
the FID and considerably more selective. This permits us to
analyze all the purgeables together, in a single run. Our
contractor, however, isn't permitted to work in that direction.
The ground rules do not permit exotic approaches. The mass
spectrometer must be ruled out as a general detector. We're
going generally with common low-cost GC detectors -- electron
capture, FID, Hall. We also do not want the contractors to
use capillary column technology because it requires a high
degree of laboratory skill, and we do not feel we should expect
above-average skills from the users of our procedures. So, al-
though our in-house approach to an analysis may involve a splitter
and a photoionization detector, these are not considered viable
19
-------
options for a method that we we would preapare for promulgation.
The contractor is forced to use a flame ionization detector for
his non-halogenated purgeables.
For the semivolatiles, we have proposed that the current EPA
pesticide methodology as a model to the contractors for all of
the halogenated neutral priority pollutants. The PCB's and the
eighteen priority pollutant pecticides can be analyzed using the
classical approach of extraction with 15% methylene chloride (or
ether) in hexane, concentration of the solvent with Kuderna-Danish
techniques, and injection into a GC with either an electron cap-
ture or Hall detector.
The pesticide approach is generally applicable for the nine
smaller chlorinated hydrocarbons, ranging from the dichloroben-
zenes up to chloronaphthalene in volatility. The pesticide method
has been used throughout the Phase I studies of pH and chlorine
phenomena, preservation techniques and solvent stability.
We grouped several compounds together into a haloether cate-
gory that we expected could be analyzed with pesticide methodology.
Some of these compounds are very volatile, however, and require
careful attention. The first thing we did with this family of
compounds was to eliminate bis(chloromethyl) ether from all fur-
ther research because of its 38-second half-life in water. Liq-
uid-liquid extraction techniques are still a problem with some
of the other haloethers because volatility losses lead to poor
precision. Chloroethylvinylether in particular has been a pro-
blem. The precision of the experimental results obtained with
this compound has made it very difficult to interpret the results
of some of the preservation studies in Phase I. If the volati-
lity problems can be overcome, the rest to the haloether test
procedure will resemble the pesticide procedure, with a Hall or
electron capture detector used for final measurements.
Moving into the area of non-halogenated neutrals, I would
like to first discuss the phthalates. As an analytical problem,
the phthalates were assumed to be straightforward - we could
always find them when we weren't looking for them. No one had
ever reported any problems measuring them in the past but they
seem to behave erratically. Everyone assumed that methods de-
velopment for the phthalates would be a trivial problem, but
you saw how they behaved in the recovery studies presented ear-
lier. They seem to appear out of nowhere and recoveries of 150
percent are not unusual. Although it is difficult to get a
clear handle on the compounds because of chronic interference
problems, the phthalates appear to chromatograph somewhat less
than ideally on the GC and appear to fractionate poorly on ad-
sorption columns such as Florisil. It appears the pesticide
methodology works as well as any we have tried.
We have a small family of three compounds that we call the
nitroaromatics: nitrobenzene, dinitrobenzene, and dinitroto-
luene. Again, we can extend pesticide methodology to include
these compounds. They are fairly easy to extract and can be
20
-------
sensitively measured with an electron capture detector. From
the data to date, it appears that there may be some problems
with the stability of these compounds in water and special pre-
servation may be required. All the data is not in yet to show
what is going on, but they seem to be sensitive to light and
chlorine. When we set up the contracts, we could not find a
place for isophorone, so we added it to the small group of ni-
troaromatics. At the time the project was being put together,
I had about an hour to try to chromatograph isophorone. When
I got a peak with an electron capture detector, I put the com-
pound in with the nitroaromatics. We later discovered that the
peak was caused by an impurity in my standard instead of isopho-
rone. When our contractor used electron capture, he was unable
to find the compound and ended up starting all over again with
a flame ionization detector. Once the detection was settled,
isophorone gave us no further problems chromatographically,
although some questions have arisen concerning its stability.
There are three nitrosamines on the priority pollutant list.
We are looking at three gas chromatographic detectors for these
compounds. The contractor is using a Thermal Energy Analyzer
and we expect to find a limited application for this very ex-
pensive detector. Two years ago, I had some experience with a
pharmaceutical waste containing low levels of nitrosamines, and
was unable to overcome the background interferences of this com-
plex waste to measure the nitrosamines with a nitrogen mode
thermionic detector. For this project, we equippped the con-
tractor with a TEA so he could always be able to measure his
compounds regardless of the complexity of the waste. He is
using the TEA as a referee technique while developing clean-
up approaches that will permit the nitrosamines to be isolated
from a broad spectrum of nitrogen or phosphorus-containing inter-
ferences that would complicate the interpretation with the alkali
flame detector. The contractor has included a modified Hall
detector that looks to be highly selective for the nitrosmaines.
Diphenylnitrosamine does not behave like the other two compounds
and will be discussed separately.
The big problem with the polynuclear aromatic hydrocarbon
family is, of course, the resolution of the coeluters. The con-
tractor prefers capillary column gas chromatography, but we are
trying to get thorough evaluation done of liquid chromatography.
We have hopes of doing a combination separation using both re-
versed phase and silica-type adsorption techniques. We visualize
a classical packed column fractionation followed by analysis by
HPLC using fluorescence or ultraviolet detection. I do not
know how this category is going to come out yet but we are pro-
bably not going to achieve the complete separation of the 16
PAH's that are on the priority pollutant list. We will likely
require high technology, such as capillary column GC. If this
fails too, we would have to come up with compromises in our
regulations. Our job is to define the state-of-the-art tech-
nology. Others in the Agency must decide whether the technology
will support proposed standards.
21
-------
Of the three bases on the priority pollutant list, diphenyl-
drazine has become known because it decomposes during gas chrom-
atography. For it, benzidine, and diclorobenzidine, we are
examining several approaches because of the problems with direct
GC. As was discussed this morning, data from the base-deactivated
phases seem to indicate some promise for the direct GC approach.
Liquid chromatography with the electrochemical detector appears
very promising. It gives a degree of specificity and sensitivity
for benzidine and dichlorobenzidine. Also, the UV detector can
be used instead if the inexpensive electrochemical detector is
not available. The approach to this category is still somewhat
flexible and a lot of the final decisions are going to depend
upon which approaches work best on wastewaters in the face of
interferences. We have no feel for this aspect of the LC-EC
approach at all. We will likely require cleanup for some of
the samples before GC or LC can be used.
For the 11 phenols, we set up the contract to look at deri-
vatization gas chromatography. Again, however, the problem has
been changed due to advances in the art of gas chromatography
with Supelco's acid-deactivated column. Since we are finding
that derivatization of the nitrophenols is complicated, and
direct gas chromatography can be used to separate all of the
compounds on SP-1240 DA, it will be difficult to justify the
derivatization approach if the only advantage is increased
sensitivity.
The other miscellaneous-type family compounds includes di-
chlorodifluoromethane, which escapes through the trap under
conventional purge-and-trap conditions. At worst, we will
modify the technique to do a separate, brief purge to collect
this compound. Acrolein and acrylonitrile are in this miscel-
laneous category also and we can develop a modified version of
the purge-and-trap tailored for these two compounds. As we
discussed this morning, Tom Bellar and the rest of us at Cin-
cinnati cannot recommend the purge and trap because of its low
purging efficiencies combined with the likelihood that matrix
effects are likely to be significant. Based on the progress of
our contractor to date, the modified purge will likely be re-
quired if you are unlucky enough to have these compounds in your
wastewater.
We have two compounds where we have been forced to consider
liquid chromatographic techniques, and these are diphenylnitro-
samine and diphenylhydrazine. With the nitrosamine, where gas
chromatography decomposes it to the diphenylamine, we are eval-
uating a chromatographic separation, such as a Florisil or silica
gel column, that will hold up the amine and allow the nitrosamine
to elute. Then, when we do the GC analysis, we would know the
amine that chromatographs would have to be from the nitrosamine.
While we have several liquid chromatographic techniques that can
be used, we still like to avoid HPLC wherever possible. The
diphenlhydrazines does not decompose cleanly in the GC. We are
22
-------
incorporating it into the HPLC method with the benzidine family
and the separation is pretty clean. The electrochemical detector
will again be used, although several others could be employed.
We are treating TCDD as a separate project, of course. We
are looking at both electron capture and low resolution GC/MS.
We are trying to develop a screening procedure with an indirect
quantitation step, where the analyst does not need to handle the
2,3,7,8-isomer to screen his samples. The concept is similar
to what we are doing in the screening phase except that we are
going to push the sensitivity down as far as we can go. We must
zero in on this compound. I believe I read somewhere the pro-
posed drinking water limitation is about 10 picograms per liter.
When we try to respond to these toxicity-based limits, we are
going to have some very special problems with this compound.
I mentioned that we dropped bis-chloromethylether from our
research. We have also noted that the endosulfan sulfate, which
you can only get from Nanogens, is not stable in the solution
they distribute, or any other solution, and this compound will
likely have limited future research directed toward it.
That is about where we are right now. As we move into Phase
II of the contracts, we are setting up a matrix of candidate
techniques for evaluation with selected wastewaters, then single
laboratory validation of the final techniques that will be pro-
posed. The wastewaters will, if possible, be selected on the
basis of their likelihood to contain the compounds of interest.
We will try to identify cleanup techniques that will work on
these wastes, and revise the method where necessary to handle
the samples. The methods are scheduled to be completed by Jan-
uary, when the interlaboratory studies will begin. Are there
any questions?
Question; Jim, is your dioxin very toxic? Why is everyone
refraining from handling the dioxin?
Mr. Longbottom; Right, it is so toxic.
Question; It's so toxic, you're saying, well, we're not
going to measure it, no, we're not going to look at it. We're
not going to look for better methods or what's the deal?
Mr. Longbottom; Well, we don't want a method that would re-
quire all of the laboratories to maintain a dioxin standard.
We figured out, if we shipped our little bottles containing
dioxin to all of the labs, we would probably spread dioxin 100
times more than at present. We're looking at methods other than
straight comparison to a standard of TCDD. We're looking, for
example, at the use of a 1,2,3,4-TCDD calibration standard as an
option instead of everybody handling the 2,3,7,8-isomer. Every-
body will be supplied with the 1,2,3,4 and use the GC/MS and a
relative retention time to identify the compound. It's either
that or we set up a laboratory or two and have national centers
for TCDD analysis.
23
-------
Question; I think your method better take into account the
possibility of 22 different tetrochlorodioxins, and your ability
to separate those as the toxicities are different. The 2378, of
course, if the one that's a problem. So, you know, concentrate
on that, if you can. We've been involved with this dioxin for
a few years now. We're going around the merry-go-round with
the problem.
Question; The more I gather from your talk, I conclude that
you see the role of GC/MS decreasing as we move into monitoring,
is that a reasonable conclusion?
Mr. TeJ.liard; We're trying like hell. Remember, when we
started, we're talking about local permits that may, some day,
contain a requirement for the permit holder to monitor itself.
Something in one of these compounds, and we don't see that it's
feasible that every car wash has a GCMS. I know from your point,
you'd be glad to do it for them at a small charge, but right now,
we're still trying to stay away from the most expensive procedure.
Mr. Telliard; As a monitoring, i.e., permit requirement,
yes. Not necessarily as it relates to our study.
Question; What techniques are you looking into as methods
of separation?
Mr. Longbottom; Within a family, such as the phtlalates,
you mean, or separating the phtalates from the families?
Question; Well, just in general.
Mr. Longbottorn; Well, inexpensive ones, classical column
chromatography and, of course, available GC technology. We're
looking, for example, at a column that would be suitable for the
optimum separation of the dichlorobenzene isomers, things like
that.
Question; Is there any way of segregating these various
families and running them separately rather than going into the
129 in a single shot?
Mr. Telliard; Well, I think that what Bob had said earlier
was that a lot of these permits may only contain two of the 129,
we don't know, so, he may just have to look for 2000. Jim's
staff are trying to come up with a method rather than having to
look for all 114. If I want to look for 2000, what do I use? Or
how do I look for volatiles and one phenol, what do I use?
Question: Being quite honest, what about the impact that
cutting down the facts of this presentation will have on the time
away from doing the chromatography. Consider the impact of some
of these policies, manufacturing, and the primary significant de-
velopment which you could possibly request — that kind of policy.
24
-------
Mr. Telliard; No, we are considering the inflationary end
of the scale, which is, what is it going to cost all the permit-
holders. It's easy to cross out the clarifyer and the aerator,
but now, another big piece of change is going to be what's it's
going to cost that guy in another five years? We have a classic
example, that is, nitrosamine. Where we have the TEA, a $30,000
detector, we can't put that into a regulation, but we've seen some
examples where it certainly is nice to have them, and they'd be
cheaper, too, for this particular company to go out and buy them.
Question; This approach certainly is, I think fortunately
is, a good one. It's necessary, but as far as cost goes for
laboratories who are already equipped and doing the method by
GC Mass Spec, it still could well turn out to be less expensive
if they got the equipment for them to continue to do it that way.
So, what you're trying to do, as I see it, is to help those that
aren't in a position to do them by the existing methods. But, I
don't see that there is going to stop anybody from doing it the
way they wanted it to.
Mr. Telliard; It's an alternative, but, at the same time
we're trying to learn a little bit each time, and we look at these
methodologies, for example, as if we're going to look at them in
real samples.
Question; You asked the question, "What are we going to do,
have a look at all 114 instead of the two?" The answer really is,
yes, you do, because if you don't, then they get a matrix that in-
terferes with it; they continuously see positive tests for certain
concentrations. They're going to spin the wheels more than you
are as well, finally proving to them that what they're really see-
ing is not the compound they're supposed to be seeing and probably
if they went to 100 laboratories across the country for their
analyses, that would be resolved, because then they would look at
all 129 or 114.
Mr. Telliard; They're not going to be looking when they get
this wonderful paper from Uncle Sam. It's going to tell them these
are the compounds you have to look for. They'll go look and if,
say, if they do have any problems, then they have their lab to go
pester.
Question; Yes, but how many car washes can even run any kind
of a test?
Mr. Telliard; True, but it might be easier for a car wash con-
tract to have a couple of samples run by GC, not that your prices
are outlandish, but, as opposed to GCMS, and when you say he has to
run it six times a year, or four times a year, or three times a year,
we can then figure in what the economic impact is going to be. I
don't think anyone is worried about the guy who wants to go out
and do it that way. We're looking at the guy who maybe can't or
shouldn't be made to.
25
-------
Question; I just hope that the economics doesn't overrule the
complexity of the matrix interference with all the rest and then
really throw it all into a big quandre because you cannot get us
a merciful look at all the interfering materials of one or three
compounds. So, I can see a severe problem with the guy trying to
confide in that service. He says, use this method, it will cost
$12. No one is going to provide a guaranteed data of this.
Question; What would develop, a similar analytical method for
pesticides that are accepted along these same lines?
Mr. Longbottom; Bill?
Question; What kind of arrangements are being built into your
methodology to take care of new technology? Suppose someone comes
out with a $15,000 GC Mass Spec System, then what happens to all
your contracts now?
Mr. Telliard; When 304G, which is now 304H, came out, we set
a baseline. We input the minimal analytical methodology; if a guy
wanted to run it with an NMR, we weren't worried about him, we were
worried about some of my colleagues here, with their hach kits. I
think the same thing is true here. There is an equivalency pro-
cedure in the regulations for coming in and saying, hey, we want
to use our magic black box here, and we think we could break num-
bers. It's faster, it's cheaper and it's really neat. What is
used, and if it's cost effective for you, there is a procedure to
handle it.
Question; You mentioned about false positives, but what about
when you said you would set up the permit, saying that these are
the compounds that you have to look for and then at a later time,
you find another compound? Does the distance validate our original
permit? This was mentioned in the December...
Mr. Telliard; Oh, I'm sorry, yes, I forgot. I'm an old permit
person, I'm too out of touch too much. Isn't it everything you see?
You've got to monitor it somehow?
Question; I don't know. If I understand it, when you test
something, then you're putting a discharge in that order. If you
detect something in your previous discharge, you have to report it
and I assume that's something like...
Mr. Longbottom; Yes, we're trying to keep track of everything
and form at least the same GC column for each one, for each type of
matrix. The contract work immediately responds to drinking water
because in this clean water experimentation that they've been doing,
they've been optimizing sensitivity in a clean matrix. The sediment
sludges, we have other contract efforts that are shooting out all
over for those right now. This is something I would not get into.
Question; How about future plans to expand the list of 129,
we want a comment on that.
26
-------
Mr. Longbottom; No, we have nothing going...
Mr. Telliard; Their whole world is still 129.
Mr. Booth; I think we could safely assume that as new com-
pounds are added, many of them would fall in with these classes
that we have broken this down into. So, that some of the same gen-
eral methods that we are coming up with would imply using the com-
pounds. Also, I intend to form these that you all take and the
samples you come up with and any other waste makers, that there is
a precise and accurate way, and we have separate contracts out right
now on the street, in how you process the sample, and hopefully you
will be using the same kind of laboratory techniques in the trend of
the adequate measurement compound.
27
-------
GC/MS Analysis of Sewage
M. Carter
U. S. Environmental Protection Agency
Mr. Carter; The sewage plant we collected samples from was
in Athens, Georgia. The Athens sewage system receives about 40
percent of its load from industrial sources. Some examples of
the industries contributing to the system are poultry processors,
transformer manufacturers, and electric motor manufacturers.
On the day the sample was taken, the BOD of the influent to
treatment was 160 and the final effluent had a BOD of 46. These
values are about average for the plant. Since influent to and
the effluent from treatment samples were taken at the same time
and residence time in the plant was six to ten hours, the results
don't give good information about the effectiveness of the treat-
ment plant.
The plant is primary-secondary with the secondary treatment
being trickling filter. The two samples were split and one ali-
quot of each was spiked. Blanks were run by GC-PID. GC/MS was
unnecessary. The spiking solution contained six base/neutral and
three acidic compounds. The compounds were added in an amount suf-
ficient to give a concentration of 200 parts per billion in the
aqueous samples. We were not interested in detection limits; our
interest was in whether the protocol works, in general, on sewage.
The base/neutral compounds were 1,2-dichlorobenzene, acenapth-
thene, 2,4-dinitrotoluene, isophorone, diethylphthalate and benzo
(g,h,i) perylene. The acidic compounds were phenol, 2-nitrophenol,
and 2-chlorophenol. The chromatographic columns were the same ones
used in the packing evaluation Walt discussed this morning. Since
the spiked and unspiked sewage samples were run before the packing
evaluation, the columns were subjected to samples before they were
evaluated.
The GC/MS/computer system used was a Varian MAT 44 interfaced
to an SS-144 data system. An SS-144 is an SS-100 with extra pro-
grams to handle the MAT 44 GC/MS.
The first slide shows two GC/MS runs. (These slides were not
available for publication). The top one is the unspiked influent
to treatment and the bottom one is the spiked influent. The frac-
tion is the base/neutral. All three isomers of dichlorobenzene
were present in the unspiked influent. The spiked influent was
examined for each of the spiked compounts and all were easily
detected. The last large chromatographic peak in both runs is
cholesterol.
Question: This was done purely by the protocol?
28
-------
Mr. Carter; Yes. The next slide shows the spiked and un-
spiked effluent from the treatment plant- the base/neutral frac-
tion. The spiked compounds were detected with no problem. The
treatment has apparently done a good job, but dichlorobenzene is
still detectable. Assuming 100 percent extraction efficiency, per-
fect linearity, etc.; the concentration of dichlorobenzene in the
effluent was calculated to be about 2 parts per billion.
Question; Mike, are you aware that ortho-dichlorobenzene was
used as a deodorizer on occasion by sewage treatment plants? Do
you know whether it was added in the plant?
Mr. Carter; I do not know if it was added at the plant or not.
The next slide shows the GC/MS runs of the acid fraction. The
SP-1240 DA column was used. Once again, the top trace is unspiked
and the bottom is spiked. The difference in peak heights of the
ambient compoinds is probably due to a foul-up in the extraction-
concentration procedure. The small, tailing peaks at the beginning
of the chromatogram appear to be short-chain acids. There is appar-
ently a little phenol also. The three spiked compounds were de-
tected quite easily. The two sharp peaks in the last half of the
chromatogram appear to be C3 and C4 phenol. Any questions about
this slide?
Question; Is the SP-1240 column supposed to do a better job
on sewage;
Mr. Carter; The only answer I can give on that is that the
SP-1240 does a good job on the phenols in sewage. It does not
chromatograph the short chain acids well.
Question; Did you come up with any quesstimate of the mini-
mum detection limit for such things as pesticides and the various
other priority pollutants; What order of magitude?
Mr. Carter; No. That was not the purpose of this experiment.
We just wanted a quick look at how the protocol works on Athens,
Georgia sewage. It appears to work satisfactorily.
Question; Mike, you say you found about 2 parts per billion
in the effluent; how much was in the influent?
Mr. Carter; About 300 parts per billion, but that's a ball-
park figure.
29
-------
COMPUTER SEARCH AND QUANTITATION OF EPA PRIORITY POLLUTANTS BY GCMS
D. R. Rushneck
PJSB Laboratories
Introduction
This report updates my Savannah talk on a computer program for
identification and quantitation of the priority pollutants using
GCMS coupled to a computer. The objectives which led to the develop-
ment of this computer program are given in Figure 1. These objectives
have remained essentially unchanged since the Savannah meeting. We
have now developed an operational history with the program which
shows that the objectives have been met. In addition, we have had
the opportunity to work out the bugs associated with the program,
to discover its limitations, and to optimize its use.
Background
To briefly review the background of what we have called the
"PJB Program", this program was written specifically for the
Finnigan-Incos GCMS Computer System, but the concepts used can also
be applied to develop a similar if not identical program for other
GCMS data processing systems. The program was written for PJB
Laboratories by Mr. Joel Karnovsky. Finnigan Instruments now has
acquired the rights to the program from PJB Laborabories, and the
software and detailed operating instructions should be available to
all Incos Data System users by the time these minutes reach you.
Program Details
Operating details of the program were developed using the
approach shown in Figure 2. The program is based upon a reverse
search of a spectral library built by the user. Actually, two
libraries are used. A Master Library contains all of the information
about each compound sought in a given analysis type (e.g., acid,
volatile, base neutral). Such information is: The spectrum of the
compound, its retention time (both absolute and relative), the ion
used for quantitation of the compound, and its response factor
relative to an internal standard. This Master Library is built by
analyzing a standard of each of the priority pollutants, and reading
the spectrum of this compound into the library. An alternative
is to construct a library entry based on data in the literature.
Editing of this spectrum can be performed to make the spectrum more
specific, and to preclude possible interferences. This editing
will be discussed in greater detail in a later section. A Daily
Library contains the same information as the Master Library but is
updated each time a standard is run.
30
-------
After the library for a given type of analysis is built (e.g.,
acids)/ a mixture containing all of these compounds (in known con-
centrations) is analyzed, and a MASTER data reduction is performed.
This MASTER run updates all of the retention times (both absolute
and relative) the a MASTER library and calculates the necessary
response factors. Analysis of a standard at the beginning of each
shift that an analysis of a given type is to be performed (again,
acids, e.g.) results in updating of a daily library, yet preserves
the data in the MASTER library. This daily STANDARD run is used
with each data set that accompanies it. The MASTER information is
retained so that long-term changes in response factors and retention
times can be tracked in order to assess system performance. Once all
of this information is cataloged, analysis of unknowns into which
the appropriate internal standards have been spiked proceeds.
Internal Standard Test
The actual search and quantitation of the priority pollutants
is based on the tests required by the protocol. After the analysis
of an unknown has been performed and the data have been recorded,
a search for the internal standard in a plus or minus one minute
absolute retention time window is made (see Figure 3). If the
internal standard is not found in this window, the program stops
and waits for a command from the operator to tell it what to do.
If the internal standard is found, its area is measured at a
specific ion (e.g., m/e 188 for D-10 anthracene). This area is
then tested against the reference area in the Daily Library for the
internal standard. If the area is within the specified limit
usually a factor of two) the analysis proceeds; otherwise, it is
halted as in the search test described above. Once the internal
standard has been found and verified, a relative retention time
table is calculated so that the retention times of the priority
pollutants relative to the internal standard are known.
Search and Quantitation of the Priority Pollutants
The search for the unknown compounds then begins (Figure 4).
This search proceeds in a similar fashion to that for the internal
standard. First, a search window is set up based on the retention
time relative to the internal standard. This search window is plus
or minus 30 seconds. The search itself is actually performed by
comparing the spectrum stored in the Daily Library to each spectrum
within the search window. A degree of match is calculated between
these two spectra. The yardstick by which this degree of match
is measured in the Finnigan Incos System is termed the "FIT". If
none of the mass peaks in the spectrum being searched match those
of the spectrum stored in the library, the FIT will be 0. If all
of the mass peaks match in their proper respective ratios, the
FIT will be 1000. A detailed description of the meaning of "FIT"
is given in Section 6 of the Finnigan Incos System Manual of 21
March 78 (also written by Mr. Karnovsky). If the FIT is above a
predetermined minimum, the scan number at which the best FIT
occurs is saved.
31
-------
Quantitation now begins. Quantitation is performed at the ion
specified in the Daily Library and within a plus or minus 30 second
window. An additional restriction is placed on the quantitation
in that the peak maximum must occur within a preset window (nominally
plus or minus three scans from the scan with the best fit). This
restriction precludes doublets from being quantitated when they
occur at the same ion as that being used for quantitation of the
compound of interest. The quantitation further uses mathematical
routines for noise rejection and for quantitation of tailed peaks,
thus making the quantitation mathematically reproducible. After
the peak area is measured, the quantity of the compound present
is calculated. This calculation relates the area of the peak of
the compound now identified to that of the internal standard and
ratios this area to that of its counterpart in the daily standard.
A report is now produced (Figure 5) which gives the quantity
of the material present and compares the retention times of the
peaks found with the retention times stored in the library on both
an absolute and a relative basis.
Internal Data Verification Tests
During the search and quantitation described above, the com-
puter program also performs some internal checks and balances. If
more than one compound is found by the search routine, a flag is set
as is shown in Figure 6. Similarly, if more than one peak is
quantitated, a flag is set. These flags notify the operator that
doublets or multiplets have occurred and therefore that the data
should be reviewed manually in detail to determine how they affect
the identification or quantitation.
Operator Verfication of Data
Operator interaction is mandatory in the correct use of the
program. First, the operator must review the FIT for all compounds
detected. Any FIT of less than 975 indicates a possible false
positive. Second, the scan at which the peak maximum occurs
should be compared with the scan at which that maximum occurred
when the standard was analyzed, and any difference between these
two numbers greater than 10 seconds should be suspect. Third, the
operator should review the flags that have been set by the internal
checks to preclude erroneous identification and quantitation.
Operational History and Optimization
We have now been using the program for approximately six months
and have performed over 1400 analyses of the acid, base neutral, and
volatile fractions of the priority pollutants. The program has
proven to be of much greater value that we had anticipated, in that
data processing now keeps pace with data acquisition, and we can
operate an instrument three shifts per day without data processing
falling behind. We now understand the subtleties and limitations
of the program and have added specific checkpoints at which operator
interaction must occur to ensure data quality. The program has
been further optimized in several areas. First, we have edited
32
-------
the libraries of both the internal standards and the priority
pollutants and optimized sensitivity, dynamic range, and specifi-
city for these compounds. This optimization consisted of elimi-
nating all library peaks in a given mass spectrum which are less
than ten percent of the base peak. This improvement makes the
search algorithm more specific, in that all ions greater than ten
percent of the base peak are used in their respective ratios in
order to calculate the FIT, and it increases the sensitivity of
the system to the compounds of interest in that the small peaks
which would normally be included in the FIT calculation are
eliminated. In this way, the priority pollutants are detected when
the base peak is a factor of 10-20 above the system noise level.
Similarly, the sensitivity and specificity of the internal standard
has been increased by retaining all masses above one percent of the
base peak. Another improvement which has been made has been to
eliminate ions at which common interferences occur; i.e., we teach
the system to be smarter with each analysis we perform. Similarly,
the ions used for quantitation are chosen based on their lack of
interference with the non-priority pollutants normally found in
the samples we analyze.
A second way in which the system has been optimized is to set
the FIT to a value to preclude any false negatives (and thus prevent
missing any priority pollutants). This value was originally set to
850 but has now been reduced to 750. We feel this is a good compromise
between the number of false positives produced (less than ten percent
of the total peaks tested to date) and the false negatives produced
(none that we know of in data files which were searched both manually
and with the program).
The third way in which we have optimized the use of the program
has been to verify all positives using the three peak ratio specified
by the protocol. This is a simple test by the operator which takes
less than 30 seconds per positive, and although this step could be
automated, we wish to retain the operator interaction as a final
verification step.
33
-------
CO
CO
Q
Z
H
O
CL
"=r
-•>
O
o
LL
o
h-
LU
/ /"\
CO
Q
LU
X
LL
LL
O
z
o
H
<
h~
H
Z
<
D
O
Q
Z
<
I
O
cc
LU
00
w
>-
_J
_J
<
o
<
^**s.
6
i-
D
VES:
PROCESS DATA A
i
"
0
LU
—-}
CD
O
V'
••>!.,
DC
O
5
LU
t
LU
Q.
LU
DC
DC
O
i i
LL.
DC
LU
h-
D
D_
^
O
O
LU
CO
D
•
>
DC
l»L»
LU
O
Q
D
CC
Q
^
O
DC
LL
FREE OPERATOR
•
Ij
CD
<
_J
LU
DC
_J
<
O
K
INCREASE ANALY
•
CO
z
DC
LU
O
Z
AM*.
O
o
CO
LU
H
<
O
LU
Z
Q
Z
<
FALSE POSITIVES
•
I
O
OPERATOR CRUT
•
w
<
<
Q- CO c\j
33A
-------
CO
Q
Z
ID
O
CL
O
O
h-
LLJ
CO
Q
LLJ
X
! i
LL.
O
Z
O
I
I—
z
D
a
Q
z
O
CL
<
LU
CO
h-
Q
I
. [d CO ^ LU
t < co z g
'_JH2OS
< co O ~ j>
D , — Z I—
pi -J -7 n ij.
I? .
^* ^-*. i j
U<^-0^
HCO
LULU
CO
CO
LU
O
O
-DC
O
CC
CL
D_
U_ D_
h-
LU
LT
LU
T.
Q
v-^
CO
^z
00o
O
Z
LU
CO
D
CO
z
6
o 2
O NX
nr
Z
Q ^
* — 11 it
Z CD
< Z
CO N
J ^ ^"j
< LU <
z^z
cc^<
gfgS
^£?|
7^ ~" LU O ^
^OCQ - LU co CL LU O
< en CL Q COH LU
en ^ co cc LU LU CL
cga:<< ^KCO
-I CO CC Q h- h- h-
C£ ~j 1- Z CO CO CO
LUM|0< DZ)D
h- b^ LU h- ~3 ~3 ~>
CO Z Q_ CO Q Q Q
< O CO LLJ < < <
^ N
CO
o•
c i -i ii~- ,;••••'
--'"-
mr,< LJ-; ,-•
^ ^; ~D < -4-
CC r- CL co CM
<
O
t»«A
cn
<
Q
Z
-------
i™
CO
G
a:
AV—
/
v »- /
\LZ/
Q
CO
O
LU
CC
Q.
Z
o
CO
T —
_J
•} O i
JLL.
ill
~s> ^^
•^ l~ —
LLI
~) <£
CL CO
,J
33C
-------
J
O
.Li
.U
.-J
Ol-
QCC
m
H-l
'V
"Z
ID
ZCO
— LLI
5H
00
DCh-
LL.CO
GC
O
I-
O
<
u.
>UJ
t.I
«'
h-C
X I,
m cr
O 0
SAVE FOR
OUTPUT
1 COMPOUND
INOT DETECTED
33D
-------
QUANTITATrON REPORT
FILE: A00062
?ATA: PB0062.TI
R7/18/7B 15:32:80
SAMPLE: 5 UL PHENOL STD OF 062978
CONDS.:
FORMULft: INSTRtlMPIT: HIL2
SUBMITTED BY: ANALYST:
AMOUNT-AREA * REF.AMNT/CREF.AREA* RESP.FftCT)
HEIGHT: 0.000
ACCT. NO.:
NO
1
2
3
4
5
6
7
6
9
10
11
12
NO
1
2
3
4
5
6
7
8
9
10
11
12
NO
I
2
3
4
5
6
7
8
9
10
11
12
NAME
D-10 ANTHRACENE
24ft 2-CHLQROPHENOL
57 A 2-NITROPHENOL
65A PHENOL
34A 2,4-DlMETHYLPHENOL
31A 2.4-DICHLOnOPHENOL
21A 2,4,6-TRICHtHROPHENOL
22A 4-CHLORO-M-CRESOL
59ft 2,4-DINITROPHENOL
60ft 4,6-DINITRO-O-CRESOL
64A PENTflCHLOROF'HENOL
58A 4-NITROPHENOL
M/E SCAN TIME REF RRT
188
128
139
94
107
162
196
142
184
198
266
365 12
85 2
1.1?' 3
142 4
179 5
188 6
248 8
291 9
374 12
381 1?
4iS 13
: 10 1
:50 1
:34 1
:44 1
:58 1
:16 1
: 16 1
:42 I
:28 1
:42 1
:52 1
1.000
0.233
8.293
0.389
0.490
0.515
0.679
0.797
1.025
1.044
1. 140
METH
A BB
A BS
A B8
ft B3
A 03
A BB
ft BB
ft BB
A BB
A BB
ft BB
AREA
58101.
5925.
2393.
7252.
3738.
3253.
1813.
2125.
114849.
32096.
650.
AMOUNT '/
500.
100.
100.
115.
100.
100.
100.
100.
100(30.
2000.
100.
000
000
088
000
008
000
080
000
80G
000
BBQ
UG/L
UG/L
UG/L
UG/L
UG/L
UG/L
UGL
UG/L
tiG/L
UG/L
UG/L
3
0
0
a
0
0
•j
0
i"j
13
n
i'or
.76
.75
.75
.86
. 75
.75
.75
.75
. If;
.02
. 75
NOT FOUND
RET(L)
12: 10
2:50
3:34
4:44
5:58
6: 16
8: 16
9:42
12:28
12:42
13:52
17:38
RflTIO
1.0B
1.00
1.00
1.60
1.80
1 .60
1.00
1 .00
1.00
1.08
1.00
RRT(L)
1.000
0.233
0.293
0.389
0.490
0 . 5 1 !5
0.G79
0.797
1.025
1 . 044
1 . 1 40
1 . 443
RATIO
1.00
1.00
1.00
1.00
1.00
1.00
1.00
1 .08
AMNT
500.90
100.60
100.00
115.00
100.00
100.00
100.00
100.80
1.00 10000.00
1.00
1.00
2000.00
100.00
AMNTCL)
500. eo
100.00
100.00
115. OB
100.09
100.00
100.00
100. SB
10006 . R0
2000. PO
100.00
100.00
R.FftC
1.000
0.591
0.239
0.623
0.373
0.325
0. 18t
0 . ? 1 ?.
0.115
0. 160
0. 065
R.FDCCL1
1
8
a
0
0
9
n
~-i
3
0
n
0
. 0'JG
. :-)9 1
. ?"-;D
. G'°['
. ;./ A
. 3,?:i
. ! 0 i
. S. ': ?.
. \ 1 S
. l'>8
. O';"
.C3--
RH f I
1 .t!
1 . '!
1
It'1
1
1 . <;>
' . ' '
1 . ' ^
t .0
1 . '
1 .
a
0
n
fl
0
t'l
'•'}
u
T
! "i
1
FIGURE 5
33E
-------
PROCEDURE: PJB
FILE: AQ0062
DflTE: 7/20/78 22: .1
flNftLYSIS TYPE: flCID
SmiPLE CftTECUlPY: Utll'MOL
IDEX
STD
1
2
3
4
5
6
?
8
9
ie
11
12
LIBRARY
fU, 1
fll, 1
fll, 2
fll, 3
fll. 4
fll. 5
fll. 6
fll. 7
fll. 3
fll. 9
fll, 10
fll', 11
fll, 12
PRED PEflKS
365
365
B5
10?
142
179
188
248
291
374
381
416
529
1
1
;
l
1
1
1
1
1
i
1
1
0
BEST
365
365
05
107
142
179
188
248
291
374
381
416
0
FIT
979
979
98S
955
976
987
995
989
995
984
972
774
0
QUflM
1
1
1
1
1
1
1
1
1
1
1
1
0
12 PEAKS PROCESSED.I-
,T.H PEftKr:
}Kr: FOUND
SEARCH
FLAG
QUANTITATION
FLAG
FIGURE 6
33F
-------
Data Systems
J. Ryan
Gulf South Research Institute
Jim Ryan; My name is Jim Ryan, and I'm with the Gulf South
Research Institute in New Orleans. We are in a unique position of
having both Hewlett-Packard and Systems Industries 150 GC/MS
data systems.
We use a manual computer method for data analysis which is
long and rather time consuming. We have been working with the
Hewlett-Packard Company in trying to devise a way to automate
the search procedures.
The data system is operated, first, by the acquisition of
a GC/MS run. The acquisition program is called up, and scan
time, scanning speed, mass range, start and stop limits of a
spectrum, etc., are specified. One thing I'd like to point out
here is that the Hewlett-Packard system operates with an acquisi-
tion routine that is different than the Systems Industries 150
acquisition routine.
Question; Excuse me, what Hewlett-Packard model are you
talking about, the 5985?
Dr. Ryan? Well, both Hewlett-Packard 5985 and Hewlett-
Pa ckaFd~5~98T~h~ave the same acquisiton routine. We have had both
instruments in our laboratory. Both the 5984 and the 5985 oper-
ate with an acquisition routine which samples the mass spectrum
every tenth of an AMU. As a mass spectral peak appears, the data
system can assign a precise one-tenth AMU number to the m/e value.
It operates very much like a high resolution mass spectrometer
in the determination of m/e peak centroid.
Question; Have you actually acquired the raw data on magnetic
disc, and can you go back to the uncalibrated data, or if you
decide that your original calibration was not adequate, are you
able to reevaluate your data?
Dr. Ryan; No, it's the peak centroid and peak area that are
calculated on the fly, and the raw l/10th AMU data is discarded.
Now, for the processing, the program that's used is called
SPEED, which displays up to 4 ions of the data file; that is, 4
extracted ion current profiles. It can take a long time.
We have come up with a reverse search library program which
I have been working on with the people at Hewlett-Packard. It
works by having a standard file which contains key ions and rel-
ative response ratios. This library file will automatically
search the GC/MS data file. As I say, that is not operational
yet, but soon will be.
34
-------
One thing that does come to mind, which people who use
Hewlett-Packard equipment might want to be aware of, when the
system acquires data, HP assumes that the only thing to be dis-
played is good data. They have declared that saturated peaks
are by their very nature bad data, and those saturated spectra
are not saved. This can have an effect on relative retention time.
However, HP now has extended the overall dynamic range, and
eliminated this saturated peak effect. I think they've made a
wrong decision to discard that saturated peak, and they have told
me that later versions of their software will have this changed.
Question; How long does it take you to process a base/neutral
data file?
Dr. Ryan; I would say approximately 45 minutes.
Question; You do all your required work on the HP data system?
Dr. Ryan; Either an HP or Systems Industries data system.
Hewlett-Packard has a very nice catalog system for their disc
data. If data is acquired on a particular disc, and all the discs
are cataloged, the HP system will tell you where all your GC/MS
data files are stored. It is very easy to retrieve data this
way.
35
-------
Data Handling
R. Kleopfer
U. S. Environmental Protection Agency
Mr. Telliard; The fellow here now is Bob Kleopfer, from Re-
gion VII. He has information on some recovery work that he did,
I presume on organics?
Dr. Kleopfer; What else?
Mr. Telliard; Right.
Dr. Kleopfer; Actually, after hearing Bob Booth's presenta-
tion, to be quite honest with you, I doubt seriously whether
there's a mass spec lab in the country, that has presently imple-
mented a quality assurance program which is that rigorous. Now,
this is not to say that this is not needed, and that it's not
necessary because I'm sure that it is. I think that the problem
is that in the past, the emphasis in mass spectrometry has been
on the qualitative aspects of the tool, but the emphasis is
changing. The emphasis is going to be on the quantitative aspects,
We have to document how well we are doing, how well can be really
measure these parameters. Up until now we really don't know
how well we're doing, or how bad we're doing. We're going to
have to emphasize this area. I received this document, this
quality assurance document, about three weeks ago, and I imme-
diately went into the laboratory and began the implementation of
this program.
One of the first things that we did was to determine the
precision and accuracy on how well we could do on an ideal sample.
How well can we do on blank water? What we did is that we added
known quantities of standards to distilled water. We went through
the extraction process, and the analysis process. We now have
data for the base neutrals and the acids.
Now, as Bill pointed out earlier, in order to implement a
program like this, one of the things that we need are large
quantities of pure standards and unfortunately, with the Radian
standard, we don't have sufficient quantities in order to imple-
ment this. So, it does require that the standard be available.
On the base neutrals our data covers approximately 35 of those
compounds.
Now, following the protocol we have nine data points for each
one of these numbers that we have accumulated, and rather than
going through the list one by one and giving you the recovery and
the standards deviation, I'm just going to summarize, and state
that overall, for the base neutrals, or 35 of the base neutrals,
the overall recovery using the EPA protocol on the standard
addition to pure water, was 81 plus or minus 17 percent. The
accuracy is represented by the 81 percent, and one standard de-
viation would be plus or minus 17 percent.
36
-------
On looking at the data, we did find out one thing, that we
didn't know previously, and this is that our standard deviations
were larger for the phthalate esters, than they were on some of
the other compounds. At least, I would not have had expected
that, because, phthalate esters would be one of the easiest
groups of compounds to do, but as it turns out, to give you an
example here, on diethylphthalate, we were getting 45 plus or
minus 31 percent on the recovery. I don't think this is due to
contamination effects, because when you look at a phthalate ester
which isn't all that common, for example the butylbenzylphthalate
on that one we got 57 plus or minus 32 percent. All of the
phthalates generally had that large standard deviation attached
to it.
On the other hand, let me pick out one of the better ones.
Well, let's take 1,3 dichlorobenzene, 64 plus or minus 6 percent
on that one. Again overall for the base neutrals we had 81 percent,
+_ 17 percent.
Now, instead of using the mass spectrometer as the detector,
we took the same set of samples and used flame ionization as our
detector, to see whether this large deviation was perhaps coming
from the instability of the detection system, rather than the
deviations that you would get in reproducing your extractions.
By flame ionization, the results were an overall recovery of 83
plus or minus 15 percent.
So, in other words, we were doing as well using the mass
spectrometer as we could do by flame ionization, which, you would
consider to be a more stable detection system. On the phenolics
we did somewhat less well in the recovery, which is to be ex-
pected, I suppose. For the acids that we looked at, by using EPA
protocol, again, the overall recovery was 52 percent plus or
minus 14. In that study we used 5 of the 11 phenolics.
I'm going to make this tabulation of data available that
could be put in the proceedings. This is a first step in docu-
menting how well we can do. Of course, the additional steps are
to document how well you can do on a real sample, by adding
standards to real samples. I really don't think this document is
all that severe. The time involved in its implementation probably
would turn out to be greater than the 15 percent, but not that
much more. We need to be able to document how well we are doing.
As I said, we don't know at the present time how well we're doing.
Question; What's your level of spiking?
Dr. Kleopfer; On the base neutrals, it was 120 parts per
billion, and on the phenols it was 200 parts per billion.
Question; Did you involve an exact quantitation as was in
the document?
Dr. Kleopfer; I was following the original EPA protocol for
the priority pollutants.
37
-------
Question; Did you quantify, according to the way that the
verification specifies?
Dr. Kleopfer; Would you rephrase your question? I don't
really understand what you're asking.
Question; On page 12, did you quantify according to the
studies you've just presented?
Dr. Kleopfer; I quantified using the internal standard
method that is described in the protocol.
Question; Which protocol?
Dr. Kleopfer; The big green document.
Mr. Telliard; If you look at what's in the protocol and
what's in your contract, and look at what kind of data, page 12,
it's about the same thing. It says you've got to run the internal
standard...
Dr. Kleopfer; It says the peak area should be integrated
using ionabundances from three of its characteristic ions, that's
carried out with the EPA protocol.
Well, in this study, the quantitations involved simply
calculating the response factors according to the protocol, and
then running your unknowns.
Dr. Kleopfer; I implemented the quality assurance, using the
existing EPA protocol. In other words, I'm generating data,
telling us how accurate and how precise the data is, using this
existing protocol that I've been using for the past year.
Question; Over what period of time did you acquire your nine
samples, nine data points?
Dr. Kleopfer; This was over a two week period.
Question; Is your FID all the same data that was used....
Dr. Kleopfer; The flame ionization work was done over a
couple of days, I believe.
38
-------
Analysis of Pesticides
A. Dupuy
U. S. Environmental Protection Agency
Mr. Telliard; The folks up here now are from the Pesticide Monitor-
ing Lab in Mississippi.
Mr. Dupuy; I'm Aubry Dupuy, this is George Sand, and we're
pleased to be here today. As Bill mentioned, the primary function
of our laboratory has been the pesticide analysis of environmental
samples. However, recently, we've gotten involved with the
Effluent Guidelines Division of EPA and are participating in the
priority pollutant program. So, we're glad to join the group and
we hope we can cooperate with you and participate and perhaps
make some contribution, especially in the area of pesticide
analysis.
Mr. Dupuy; Okay, the pesticide residue analysis of a water
sample consists of basically four steps. First of all, there's
the extraction of the residue from the sample matrix by means of
an organic solvent. Secondly, if necessary, you may need a clean-up
step which removes interfering co-extractives. Thirdly, there's
the identification and quantification step and normally we're
looking for 10"^ to lO"-"-2 grams of pesticides.
The method of choice is usually the gas chromatograph in con-
junction with the electron capture detector. Fourth, there's the
confirmation of the presence and identity of the residue. There
are a lot of classical methods for confirming pesticide residues.
However, in the last few years the GC Mass Spec, has been the
prime method of choice.
Now, of fundamental importance in pesticide residue analysis
is being aware that certain substances give rise to artifacts
which interfere with the gas chromatographic electron capture de-
termination of organochlorine pesticides. Among these substances
are polyethylene or plastic wash bottles, tubing, bags, lids, and
so on. Stay sway from tygon and rubber tubing. Your glass wool,
filter paper and sodium sulfate can be contaminated. In many
cases, the contaminants have been shown to be our friends the
phthalates and in some cases the polychlorinated biphenyls, and
in the latter case we're actually looking for these substances in
the pesticide fraction of the priority pollutant analysis. So,
obviously, we can't afford to have any contamination from these
substances.
I have assembled on this slide some recommended precautions
to prevent laboratory contamination. First of all, always use
glass and when necessary use teflon. Teflon wash bottles and
teflon-lined lids are recommended. Always use pesticides quality
solvents or equivalent. After washing your glassware in deter-
gent, rinse with tap water followed by distilled water and then
acetone. Allow the glassware to dry and then bake it out in an
oven at 350 degrees. We found that this bake out procedure is
39
-------
extremely helpful in removing trace quantities of the phthalates
and PCB's. It's always a good practice to rinse your glassware
with some Nanograde hexane immediately before use. You should
soxhlet extract your glass wool and sodium sulfate with methylene
chloride and store it in an over at about 130 degrees. You can
obtain contaminant free water by passage through activated char-
coal followed by distillation. If you still have problems with
contaminants in your water, you can extract it with two or three
portions of Nanograde hexane followed by boiling the water to
remove traces of hexane. This should remove trace contaminants
from the water.
Of course, the ultimate test for seeing how well you're doing
with laboratory contamination is your method blank determination
and according to the Federal Register you should carry 1,000 ml
of water through the complete extraction procedure. Extract the
water and proceed with it as a routine sample. In the final
extract, you should see no peaks greater than two percent full
scale deflection at the retention sites of the pesticides of
interest. If you do, you've got contamination problems and you
should get those squared away before you proceed.
Since the prime method of determination is the electron capture
detector in combination with gas chromatography, I would like to
briefly describe the operating principle of the electron capture
detectors. Usually, a radioactive source emits low energy beta
rays. These rays ionize the carrier gas forming an electron rich
carrier gas. There is a steady flow of this electron rich species
from the cathode to the anode of the electron capture detector.
This produces a steady state background current. However, when
a pesticide elutes from the GC into the electron capture detector,
the partial positive sights created by electronegativity differences
on the halgenated pesticides (between the carbon and the chlorine
atoms) tends to combine with or capture the electron rich carrier
gas. This creates a decrease in the standing current which is
magnified in the electrometer and finally read out on a recorder
as a peak representing the pesticides. This detector is used for
the general screening and identification of organochlorine
pesticides. When this detector first came out, it was thought to
be a cure-all for pesticide identification. However, it was soon
found that the detector wasn't as specific for chlorinated hydro-
carbons as originally thought. It does respond to other compounds
that have electronegativity differences in the molecule. The
sensitivity of this detector is in the picogram region.
You may not be able to see this one too well. The G.C. chroma-
tographic column is the heart of the GC system, so I'd like to talk
a little bit about the recommended column packing materials for
pesticide analysis. There are a wide variety of packing materials
available on the market today. However, a lot of them are of
limited value for pesticide analysis. I have assembled here a
table of some of the old standbys recommended for pesticide
analysis and they're good because they show good efficiencies and
elution characteristics for the chlorinated hydrocarbon pesticides
and with the exception of the three percent DECS column, they show,
40
-------
or offer minimum bleed, and this is a prerequisite for working with
the electron capture detector. But, if you can't read them, the
first one up there is 1.5% OV-17/1.95% OV-210 the second phase is
4% SE30/6% OV-210, the fourth is 5% OV-210, and the last one is
three percent DECS. I've also listed on this slide the recommended
solid supports and in all cases but the DECS column, the recom-
mended solid support is chromosorb W, H.P. or Gas-Chrom Q-that
should be 100 to 120 mesh on those. It's not a new type of mesh we
have there, it was stapled on the slide. For the DECS phase use
80 to 100 mesh Gas-Chrom P. The approximate operating temperatures
for these phases are shown along with the suggested flow rates of
50 to 90 mis per minute.
In order for the chromatograph column to function properly
for pesticide analysis, it should be conditioned properly. I have
outlined on this slide the three main steps for properly conditioning
a column for pesticide residue analysis.
The first step is the heat curing process and this involves
heating the column 50 degrees higher than it's recommended operating
temperature, and I have listed here the recommended temperatures and
times of heating for the various columns. I've added to the list
the 3% OV-1. This column is also recommended in the Federal
Register as a good working column for pesticides and if you want
a non-polar column, this is the recommended one. Pesticide
chemists are trying to get away from the DC200 series, because
they tend to bleed a little more than the OV-1 series and, as I
mentioned earlier, the DECS column does also tend to bleed some-
what, so be careful on your curing time. It's recommended not to
go more than about 20 hours for this particular column. Cure the
columns at a flow rate of @ 60 to 70 CC's per minute. I might
add that these steps are done, of course, with the column dis-
connected from the detector.
The second step is the silylation step. Use silyl-8 which is
commercially available. You should place four consecutive in-
jections of about 25 ul each, spaced 30 minutes apart, onto the
column, and then after your fourth injection you should allow about
three hours for all the silylated material to pass from the column
before you proceed. The silylation procedure helps to remove some
of the active absorption sights on the column and hence improves
the response of pesticides and it also minimizes the potential for
breakdown for labile pesticides such as DDT and endrin.
The third step is the priming step whereby you can inject
several successive injections of a chlorinated pesticide mixture-all
components in the microgram range. The components of this mixture,
of course, should be the ones that you're looking for in your
analysis.
This slide shows how endrin can be decomposed into a couple of
decomposition products on a column that hasn't been silylated, and
the bottom chromatograph shows how after silylation, the response
of endrin is improved and there are little or no decomposition
products showing. The chromatogram on the right shows that DDT
41
-------
can break down into as much as 17 percent DDE and 8 percent ODD
on a column which isn't conditioned properly and, of course, if
you're unaware of this breakdown, you'd be reporting DDE and ODD
when you shouldn't and you'd be reporting less DDT than you should.
You can see on a properly conditioned column you should get minimal
break down, that's the chromatogram on the left.
Now, unfortunately, a lot of chromatographers when they pack
and condition a column proceed to use it immediately before they
take the time to see whether the column is any good or not. So,
considering the fact that a GC column, especially for pesticide
residue analysis may last up to a year, I think it's worthwhile
that you take a couple of hours out, to systematically evaluate
whether the column that you have is worthwhile proceeding with.
Now I'd like to talk a little bit about this column evaluation
procedure. First of all, you should adjust your parameters for
the liquid phase in use. Adjust your oven temperature and your
flow rates and then chromatograph a standard mixture. A good one
to use is the one listed here on the slide and the concentrations
are in picograms per microliter and, after you've chromatographed
this mixture, you should check for the specific pattern of compound
elution and peak separation for the phase in use.
On this slide I show the chromatograrns of pesticides on the
four phases I've mentioned earlier. The most efficient column or
the column of choice, if you're going to pick one to start with,
would be the 1.5 percent OV-17 and the 1.95 percent QF-1. And,
a good backup column for that one, for confirmatory purposes is the
five percent OV-210 column. It separates the BHC isomers well,
however, it's not as efficient for some of the later eluting com-
pounds. Note the three percent DECS column offers a very interesting
pattern and that's the reason why it is suggested as a confirmatory
column for pesticide residue analysis, and I don't know if you see
it very well, but the Alpha BHC comes after the aldrin peak which is
unusual and the delta BHC comes way on out past endrin and O,P'-DDT.
In this column P,P'DDT comes out before P,P'-DDD. Because of this
unusual sequence, this column probably can offer you some advantages
in shifting some of your compounds around for confirmation.
Back with the column evaluation procedure. After you've
chromatographed your standards, you should look to see that your
absolute retention time of the P,P'-DDT should be about 15 to 18
minutes.- If you don't find this to be the case, then your operating
parameters are not right, the column is not the correct length or
perhaps your column is not properly packed or you may have some
combination of the above.
So, try to adjust your system such that you get the DDT peak
out in this time span. This step is to calculate efficiencies based
on the P,P'-DDT peak and this is done using the equation: N (the
number of theoretical plates) is equal to 16 times the quantity }C
squared where X is your absolute retention time of the DDT peak and
Y is the peak width of the DDT peak. Using this formula, your
column should be such that N comes out to be about 500 plates per
foot or for a six foot column this should be about 3,000 plates.
42
-------
We've used columns that come out a little less than these, 2750,
2700, but if you come much less than that, it's probably not worth-
while proceeding with the column. It's definitely worthwhile
checking the theoretical plates on the column. The last step on
your evaluation procedure is to evaluate the column break down by
injecting separately the analytically pure endrin and P,P'-DDT
standards. You should have no break down peaks exceeding three
percent for DDT and six percent for endrin.
I'd like to talk a little about the quantitation of pesticide
residues. First thing, make sure you start with accurate analytical
pesticide reference standards, and if you're weighing your own
standards, make sure that you use certified high purity reference
materials and that you have an analytical balance that's capable of
weighing plus or minus a tenth of a milligram. Undoubtedly, the
greatest single source of quantitative error in GC analysis is
inaccurate standard solutions. You can have your system working
properly, but if your standard solution isn't correct, then it's
all for naught. So, take good care of your solutions. We like to
weigh out our stock solutions in benzene and keep them stored in
a freezer and, of course, they remain frozen in the benzene matrix.
We make our dilutions in isooctane, which is a fairly high boiling
solvent that doesn't tend to evaporate readily. When not in use,
we try to store our standards in the refrigerator in the dark to
minimize photochemical and thermal break down.
Now, secondly, before you can proceed with the quantitative
analysis of pesticide residues, you should check to see that your
electrometer attenuation is adjusted to obtain a minimum sensi-
tivity level equivalent to a 50 percent full scale deflection for
the injection of 100 picrograms of aldrin. Now, don't cheat. Your
noise level shouldn't exceed two percent. If you can't achieve
this sensitivity you've probably got a dirty detector or you've
got something contributing to the contamination of your detector,
such as column bleed dirty carrier gas, or perhaps even a small
leak in your system.
Another point with quantitating pesticide residues, is to make
sure that the quantitation is performed within the linear range
of your detector. Each detector has its own linear range and the
linear range does vary somewhat between the individual pesticides.
So, to make sure where you stand, do frequent linearity plots on
your detectors so you can understand exactly where your linearity
lies.
On some of the old MicroTek detectors, the linearity range
is somewhere in the dynamic range of 1 to 200, but there are now
linear detectors available which extend the linearity range from
1 to 10,000 to as much as 1 to 20,000. So, if you have a linear
detector, you don't have to worry as much about the linearity but
you still should keep a periodic check on it.
I would now like to say just a few things about sample cleanup.
Florisil chromatography and acetonitrile partitioning are frequently
used for cleanup and are described in the Federal Register.
43
-------
I've listed microcoulometric and electrolytic conductivity gas
chromatography as a form of cleanup.
I look at them as a clean-up and analysis technique all rolled
into one because these detectors are specific for halogenated
compounds and don't respond to other organic molecules.
If you're really serious about pesticide analysis, you should
have one of these detectors as a back-up to your electron capture
detector.
Finally mercury can be used to remove sulfur from hexane
extracts. Sulfur does cause an interference problem on the
electron capture detector.
Also silicic acid-celite can be used for the separation of
PCB's from organochlorine pesticides.
Thank you very much for your attention.
44
-------
SLIDK 1
PESTICIDE RESIDUE ANALYSIS OF WATER
1) Extraction of the residue from the sample matrix.
2) Removal of interfering co-extractives ("cleanup").
3) Identification and quantification of residues in cleaned-
up extract (usually at 10~9 to 10"^2 grams).
4) Confirmation of the presence and identity of the residues.
Procedure Described in: Federal Register (EPA)
Vol. 38, Number 125, Part II,
17319 - 17323.
44A
-------
SLIDE 2
BE AWARE OF LABORATORY CONTAMINATION
Interferences -
The following have been shown to give rise to artifacts
which interfere in the GC-EC determination of organo—
chlorine pesticides:
polyethylene/plastic - wash bottles, tubing, bags,
lids, etc.
tygon/rubber-tubing
glass wool, filter paper
Contaminants - Phthalates and/or PCB's
44B
-------
SLIDE 3
SOME RECOMMENDED PRECAUTIONS TO PREVENT LAB CONTAMINATION
1. Use only glass and where necessary teflon (wash bottles, lid
liners, etc.)
2. Use Pesticide Quality Solvents
3. After washing, bake all glassware in oven at 350°C.
4. So'xhlet extract glass wool and Na2SO^ and store in oven at
130°C.
5. Contaminant free water can be obtained by passage through
activated charcoal followed by distillation. (Further puri-
fication, if needed, by extraction)
Ultimate Test - Blank, reagent determination
Extract 1,000 ml distilled H20
Analysis - No peaks > 2% f.s.d. should appear at retention
sites of pesticides
44C
-------
B
UJ
I
f
g
£
3
CO
fc
h 1
cr:
c_>
CO
T
LU
1
g
>—»
CO
LU
CO
CO
col
44D
-------
to
o
g
CM
i
e"
in
a
w
(3
0. X
a S
O E
68
i r-(
w \
<0 O
O 00
ex
e ,
o
•8
o
in
i
o o
I O
M O
IS
! J
w
I w
i "^
I ^*"
i
<
*
r^
' >— i
' i
j>
O
i «*"
' *n
•
! ^
j
i
o
rH
(N
|
>
0
(ft*
vO
^1
o
t\
i
>
o
dt>
m
c\
•
r-t
S
J3
w
o
to
o
to
*£
U
5
•*•
£
.a
tS
o
OT
Q
Q
O
jC!
u
e
o
x
u
i
in
>3
O
W
O
Ot
£
O
.c
u
1
in
n
O
VJ
o
JC
in
o
g
o
o
•-1
\ o
o o
00
o<
o
•
X
0 U
Vi o
a,
§«:
o
§
>
• o
&4
• •
X C
0 -H
^4 g
a, \
S"S
D1
•r~i
r-|
^
3
-a
•H
>
•H
C
H
>l
44E
-------
SLIDE 6
COLUMN CONDITIONING
1) Heat Curing
Phase Oven Temp °C Minimum Time, hr.
4% SE-30/620V-210 245 72
1.5% OV-17/1.95%QF-1 245 48
10% OV-210 245 48
3% OV-1 250 48
3% DECS 235 20 ( Do Not)
Exceed
Flow rate = 60 to 70 cc/min.
2) Silylation
Use Silyl 8
Four consecutive injections - 25 M! each (spaced 30 min.
apart)
Allow 3 hours
3) Priming
Several successive injections of chlorinated pesticide
mixture in the microgram range.
44F
-------
SLIDE 7
A
EEf-ORE SHYIATION/
A
3'?
B
SUYIATION
.
If
J9 JJ
1?
36
44G
-------
SLIDE 8
FIGURE 4-H
BREAKDOWN OF DDT
SE-30/QF-]
A - 1 i.o conversion
i-
O
Q
' 9-
: Q.
c
' C
: <
1
L.__ _::.._
n a - /'icximum Conversion
u
vjp
c
<
^
IX,
— ff?
UJ CO
? g
a- Q
^ V
|S «i
f;-
|Q
o
Ta.
la.
1
^
44H
-------
SLIDE 9
COLUMN EVALUATION
1. Adjust operating parameters for liquid phase in use.
2. Chroraatograph the following standard mixture (concentration
in pg/ul):
ex.- BHC - 10 Kept. Epox. - 30 O.P1 - ODD - 80
£ - BHC - 40 P,P' - DDE - 40 P,P' - DDD - 80
Lindane - 10 Dieldrin - 50 0,P' - DDT - 90
Heptachlor - 10 Endrin - 80 P,P' - DD1 - 10O
Aldrin - 20
3. Check for specific pattern of compound elution acwi peak
separation for phase in use.
4. Absolute retention time of P,P' - DDT peak should be 15 to
18 minutes.
If not: operating parameters are incorrect.
: column is not correct length.
: column is not properly packed.
5. Calculate column efficiency based on P,P' - DDT peak.
x \2
Efficiency: N = 16( x V
V77
N should be 500 plates/ft. = 3,000 plates fear 6 ft. column.
6. Evaluate column breakdown by injecting separately — analytically pure
endrin and P,P' DDT standards.
Breakdown peaks should not exceed 3% for DDT and 6% for endrin
of the amounts injected.
441
-------
SLIDE 10
L
*i
3"X OIGS
JJ1 Jl
44J
-------
SLIDE 11
QUANTITATION OF PESTICIDE RESIDUES
1) Start with accurate analytical pesticide reference standards.
Use certified high purity reference materials.
Use analytical balance capable of weighing _ 0001 g.
Undoubtedly, the greatest single source of quantitative
error in GC analysis is inaccurate standard solutions!
2) Electron capture electrometer attenuation should be adjusted
to obtain a minimum sensitivity level equivalent to a 50% f.s.d.
peak from the injection of 100 pg of aldrin. Note - Noise
level should not exceed 2 percent full scale.
3) Quantitation must be performed within the linear range of the
detector.
Each detector has its own linear range.
The linear range varies somewhat between pesticides.
Do linearity plots for new or renovated detectors.
4) Quantitation of high concentrations of pesticides should be done
by serially diluting the final extract and continuing to operate
at high electrometer sensitivity.
This: minimizes chances of being outside linear range.
: helps keep cleaner system.
5) Optimum Peak Heights for Quantitation
The ideal range of peak height response is 20 - 60% f.s.d.,
with a minimum acceptable height of 10% f.s.d. (5 times .>•
signal noise).
For highest accuracy, peak heights of the sample and standard
should not vary more than 10% (at most, 25%).
6) Injection Volume
Injection of 1 - 3 /il samples is not recommended because
of the large increase in error probability.
Inject 5 - 8/al from a 10 ,/ul syringe.
7) Intersperse standard mixture injections throughout the workday
so that response fluctuations can be noted.
44K
-------
SLIDE 12
• — i - v "
7-E. Aroclor 3254 (solid line) 2 H,?I. rro,:
-------
SLIDE 13
SAMPLE CLEAK-UP
General Methods:
1.) Florisil chromatography
2.) Acetonitrile partitioning
3.) Microcouloir.etric or electrolytic conductivity gas
chromatography
Specific Methods:
1.) Mercury treatment - removal of sulfur
2.) Silicic acid - celite - separation of PCB's from organo-
chlorine pesticides
Arrrour, J, A.,and Burke, J. A., J. Assoc. Cffic. Anal, Chem.,
53, 761 (1970).
44M
-------
Quality Control - Validation
B. Booth
U. S. Environmental Protection Agency
Mr. Booth; By way of introduction, I'm a chemist who is now
pushing a pencil. But I do have a story to tell today, and I
think based upon what I heard yesterday, it's a story that needs
telling. If you haven't had the opportunity to read the document
that was passed out yesterday, please do so at your leisure. For
now, please take my presentation in the spirit that it's given, and
be prepared to ask constructive questions at the end of the pre-
sentation.
By way of background, I think you should know who EMSL is,
what it stands for, and what we're doing in Cincinnati. EMSL-CI
is an acronym for the Environmental Monitoring and Support Labora-
tory. We are essentially responsible for two primary programs:
methods development and standardization of methods that are used
to meet the monitoring needs of the agency, and provision of
quality assurance tools that are needed to show that data are valid.
You have heard Jim Longbottom talk to you about what we're
doing under contract on the methods for the priority pollutants,
I'm not going to dwell on that. We do have a big chunk of money
in this area, and it will continue probably for the next two fiscal
years at least. We are contracting now close to one and a half to
two million dollars on this methods research. Concurrently, in the
area of quality assurance, we are spending an equivalent amount in
providing quality control check samples and unknown performance
samples, which I'll talk about later.
I would like to encourage you to make use of our staff. When
we first went into this, we developed the protocol. As I think
most of you know, it was done on a very short notice. It was done
with the thought in mind that this was our best shot at that point
in time, that we would welcome your comments, and we would welcome
receiving problem samples. That offer definitely still holds. We
want to improve the protocol and we want to work with you on the
problem samples. That's the end of the P.R. part.
Okay, lecture time is here. I'm going to give you the elements
of a basic QA program — the intra QC effort that you need to be
doing on a day to day basis to turn out valid data. If you will
recall in many of the presentations made yesterday, the speaker
was asked, how good are the data? With all due respect to the
speakers, most of them fumbled. They were not able to say the data
are precise and accurate within certain defined limits. Now I'm
sure if they were to go back to their records and check it out,
they'd probably have it somewhere. I hope they have it anyway, but
at that particular moment in time, they were not able to say to the
judge, if you will, these data are valid because I have documented
the validity of my data. That's really what we're going to talk
about today.
45
-------
The first thing you must do, and this is something that we
just heard in the last talk, is to prepare a standard curve. It
should cover the entire working range, it should consist of at
least five points, including one standard near the upper limit of
the concentration range, and one near the lower limit. Once you
have established this standard curve, you should analyze a minimum
of two standards to establish the validity of this standard curve
for each group of samples that you run.
So, with each day, or once a week, whenever you are running
samples, check out your standard curve with a minimum of two stand-
ards. Again, you will use standards that graph at the midpoint,
and if these standards fall outside of the established limits of
your curve, then you should stop your system, find out what the
problems are, and construct a new standard curve.
Method blanks must be determined for each set of samples
analyzed and whenever a new source or a new container of reagent
or solvent is introduced into the analytical scheme. You can't
take this for granted. Particularly, new reagents and new sol-
vents should be checked for purity prior to their use in the sys-
tem. The blank must be carried through the entire procedure. If,
for instance, you are doing trace metals, you take that blank and
carry it through the digestion procedure, to see what's happening
during the digestion step.
On a field blank, as contrasted to the method blank, you want
to analyze a field blank with each set of samples from a given
source. Again, these blanks should be carried throughout the en-
tire procedure. If you do find interferences, you should disgard
the analytical results, unless you have data to show that there
was some reason for the blanks being the way they were, and you
can indeed, correct the data. Having done this, the next thing
you want to do in the laboratory is to do some work on precision,
recovery and accuracy.
Precision is nothing more than the ability to reproduce your
results. We are suggesting that, initially, you should establish
your precision in the laboratory by running 15 to 20 sets of
replicate results. This should be done over a reasonable period
of time to reflect day to day operation. I can't stress more here,
what we're trying to do is to get control over what you're doing
on a day to day basis in the laboratory.
You choose samples that are most representative of the various
sample matrices and the types of interference potentials that you're
going to have from the sample types that you routinely run. Ul-
timately, your program should include samples representing the
entire concentration range and the entire sample matrix samples
received. Having done this, at least two replicates of a raw mix
sample must be analyzed with each set of 20 samples or less, that
are analyzed at any given time. In other words, whether you are
running one sample a day, or 20 samples a day, you need to run a
duplicate to determine the precision of your method on that par-
ticular day.
46
-------
Record all of the results, evaluate the data by comparing the
difference (called range (R)) to the established critical difference
at the concentration level observed. If your observed range is
greater than the critical value, your system is out of control. You
should find the source of the variability, correct it, resolve the
differences before continuing. It's not enough to practice quality
control and simply document the fact that on a particular sample,
you were out of control. You must do something about it.
Periodically, your controls should be updated. We have recom-
mended about every 25 to 30 sets a day. You should constantly be
striving to revise, update, and improve your limits. One of the
gentlemen asked me this morning about this, and I was saying, if
you think about it, the only thing that is really changing on a
day to day basis is the ability of your analyst to repeat his
analysis. I've got a man shaking his head no. I'm assuming that
all other things are constant in terms of methods blank, reagents,
standard curve, you're using the same method. The data that you
have reflects sample matrix. This may be something you will want
to talk about when I get to the questions.
Accuracy involves the use of check standards, in which you
are analyzing them according to the approved method. In this case,
according to the protocol, carrying them through the entire pro-
cedure. You would use concentration levels that approximate those
found in routine samples being analyzed by your lab staff, or if
you so desire, choose one standard above and one standard below
the midpoint of the range in method.
Percent recovery is calculated in the standard manner, where
you simply take the observed value over the true value, times 100.
After you have about 15 observations, you can calculate the average
percent recovery. Also, calculate the standard deviation. Then, in
your succeeding check standards, you should be within plus or minus
two standard deviations of your average percent recovery. Now, keep
in mind this is on check standards, i.e., spiked distilled water.
We have purposely set high limits here rather than the tra-
ditional plus or minus three sigma, because we feel it will result
in more accurate data on real world samples. Again, having done
this, we recommend that at least one out of every 20 samples or
less, should be analyzed at any given point in time. Record all the
results, and as with precision, update your average percent recovery,
so that you're contantly striving for improving the accuracy of
your methods.
Recovery gets into the business of looking at sample matrix
interferences. Essentially, you're doing the same thing that we
just talked about, except instead of spiking a pure distilled water,
you're spiking an environmental sample.
Some do's and don'ts in spiking samples. Make sure that suf-
ficient spike is added to approximately double background concen-
tration level. If you find that the original concentration level
is higher than the midpoint of the standard curve, then your spike
47
-------
should be about one half of this background. On the other hand,
if the original concentration was not detectable, as it would be
in some of your samples, some of the clean samples, then the spike
should be approximately five to fifteen times the lower limit of
detection.
The volume of standard added in the aqueous solution should
not dilute the sample by more than ten percent. The volume of
standard added in the organic solvent solution should be kept small.
We have recommended 100 microliters per liter, or less, so that we
don't have to worry about the problem of solubility of the standard
in the water.
Here, we're saying that the results should fall, again, plus
or minus, three standard deviations. If not, the system is out of
control, you should find the source of your problem, and resolve it
before going on with your analysis.
These are the kinds of things that we would expect, hope, that
you would be doing on a day to day basis. In addition to that, as
part of interlaboratory quality control, we have two sample types
that we can offer you, offer to the contractor, to the project
officer. One is known as a quality control check sample. This
consists of a sealed vial, prepared for the various parameters, in
which we tell you what's in the sample. We have recommended that
you run these at least twice annually.
We have a second series of samples, called performance evalua-
tion sample. These, again, are in sealed vials (ampules), but we
don't tell you what's in them. These we rejjK$©A&t_ should be run at
least annually. Both types serve as a separate, independent check
on your entire system. I would strongly recommend their use.
Samples currently available of interest to this group, are
pesticides, PCB's, purgable organics, and the trace metals. Our
work plans call for expanding the organic priority pollutants in
the remainder of fiscal year "78, and by the end of fiscal year '79,
which is about a year and a half away, to have QC check samples for
all of the priority pollutants.
In terms of requests and data handling for these samples, the
QC checks samples and the performance samples can be forwarded to
you upon request of any responsible authority. In most cases, this
would be the project officer. In terms of the handling of the data,
the QC check samples, where we tell you what is in it, is your sam-
ple to use as you see fit. It serves as an independent check stand-
ard, if you will, which you can use in the laboratory to check out
your entire system. The performance samples, on the other hand,
since they are unknowns and since we want to protect the true values
of those samples are handled in a different manner. We have you
send the data back to us, we process it through the computer, and
we tell the project officer how well you have done.
I'm told by our staff that organic chemistry is unique. Ac-
cordingly, we have a section in the protocol entitled, Operations
48
-------
Specific to Organic Analysis by GCMS. I'm not going to insult
your intelligence by my going through this in great depth, except
to say that it includes sections on calibration checks, on the use
of internal standards, and how we feel one can quantify results from
the GCMS. That is detailed for you in the protocol document.
The other thing we were asked to do was to address the question
of what to do when you are just starting out and trying to gather
data on a method that you're not really sure is going to work on
all sample types. You want to keep the requirements down to an
absolute minimum of the samples required. That's the little package
that you have, entitled, Preliminary Evaluation of Analytical
Methods to be Used in Verification Phase. We have some ground rules
on this. They are indeed, minimum requirements. They are to be used
only when the agency can not provide you with validated methods.
We encourage far more strigent programs, but recognize that there
are some basic economics involved, so we're trying to keep the
number of samples to an absolute minimum.
Any program you use should be documented, as I've said through-
out my talk. Use this system only for the initial establishment of
the precision and accuracy of the method. The reason I say this
is because, clearly, it does not account for sample matrix effect.
Very quickly, for doing precision, we have recommended that
for each parameter of interest, spike a minimum of four replicates
of laboratory blank water, at a concentration equal to about ten
times the limit of detection. Analyze these four samples and cal-
culate standard deviation in the normal manner. For accuracy,
using the same set of data, from the same four samples, calculate
the mean percent recovery by simply taking the sum of the observa-
tions, dividing it by four times the true value, and multiplying it
by 100.
These data should be recorded and submitted to the responsible
authority, in this case, the project officer, prior to the initiation
of sample analysis. You should do this before you run your first
sample. Then, having done that, you should institute the routine
day to day QC program that I have talked about. That should be
done immediately to revise, update, and improve the data that
you're using.
That constitutes my formal statements. Before I start taking
questions and letting them fire away, how long do we want to let
this go on before coffee, or do you want to take a coffee break
first?
Mr. Telliard; Sure, (WHEREUPON, a coffee break was taken.)
Question; Some of the requirements I think are excellent,
but the need for a higher concentration of standards is really
pointed out in these peaks, the levels we should be running and
the number of references that we should be carrying out. I think,
I would like to see them personally made up on something like
hexane, for this purpose, as to opposed to the methylene chloride.
49
-------
Question; At what we have at hand in the way of standard
replicates of the samples, we couldn't carry out this as we should
do, with the large quantity of actual samples. We need more stand-
ards and more...
Mr. Booth; All right, I'll take that answer back to John Win-
ter. Other questions? My god, I thought I was going to get shot
down and be up here for two or three hours. Everybody agrees with
what I said? I can sit down? It's all over with?
Question; I disagree. Some things that I would like to do,
you haven't answered. Do you know what's going to accomplish
Mr. Booth; Why don't you ask that question? Some of the
QC requirements. Would you grant me that most people do set up
the standard curve, to begin with?
Question; Well, assuming that you can set up a standard curve,
at three a week, you can't do that. With this system, if
you're going to do this, it better be done every day.
Mr. Booth; Well, we had said that every day or every time
you run samples, that you use at least two standards to check that
curve.
Question; That would be different.
Mr. Booth; Well, if so, then you get yourself up a new
standard curve. That's exactly what I've said in my lecture.
Question; Right.
Mr. Booth; Yes, so we agree. That's great. Okay, moving
right along.
Question; That's why I asked you. You know what I'm trying
to say~.
Mr. Booth; Yes, I'm giving it to you here, because I was
sure that would come up. You need at least two standards to check
your curve. You need one method blank. You need one field blank.
You need one duplicate for your precision check. You need one
check standard for your accuracy check, and you need one spike
sample for your recovery check. That totals up to be 7 QC samples
required.
Question; Per day?
Mr. Booth; Per day, per set of samples...
Question; Do you know how
Mr. Booth; Let me finish. I'm adding to your argument here.
I really am. I'm on your side. Just let me finish.
50
-------
Question; The EPA ...
Mr. Booth: He asked me to come here to help you. Seven QC
samples required, whether you're running one sample, or whether
you're running 20 samples, on that particular day. So, it's quite
conceivable that if you have only one sample to run that day, that
you would be in the terrible situation of having to run seven more
QC samples. I want to make that very clear to the group. If you
have 20 samples, it would still be seven. If you had 40 samples,
it would be three more. So, keep that in mind. Now, what you have
to stop and ask yourself is, recognizing the fantastic work load
that this is, what can I cut out, and still be able to say, I know
for sure beyond any shadow of a doubt, that I have generated good
data today, at this point in time, on this sample, with this analyst,
and this instrument, under these conditions. If there are any of
these that you can prove to me that we can reduce, I'll be very
pleased to do so.
Question; Do you know that it takes an hour,
Mr. Booth; Yes, I know cost wise, for instance, what we're
talking about here, yes, but I also know the kind of data that I
saw yesterday up on this board that had no business being up there.
Question; Okay, but you spent seven hours meeting the require-
ments for the QC work, which gives you time to run one sample in a
day. I presume that you gain a day every hour, so that you're run-
ning a shift, two shifts a day, yet
Mr. Booth; If you are changing lab conditions, to the point
where you're bringing in new analysts, new variables, yes.
Question; There may be a way to do this and still, you know,
not make the cost to the agency $10,000? Have you considered add-
ing other internal standards into these things, so that you have
internal controls on the samples?
Mr. Booth; Yes, we have. When we gave this assignment to
the staff — two of the staff are here, so they know this is a true
story — the first draft that came through, you would be amazed at
the amount of samples required. As far as back to them, you say
no, that isn't what we're after. We want the absolute minimum
number of samples that you honestly feel that's required to generate
valid data. So, everybody that worked on this document had those
groundrules. They've come up, if you can, with the best lowest price
system that will guarantee us valid data coming out of the study.
Again, reviewing this, there is perhaps one thing that you
could do to save yourself some, and this is something that I just
thought of when I was totaling this up yesterday, and that is, it's
quite conceivable that the check standard that you use for your
accuracy check, could also serve as one of your two standards to
check your curve. In so doing, you would reduce that seven down
to six, which is $800.
51
-------
Question; You're right.
Mr. Booth; I have seen your help.
Mr. Telliard; Well, wait, I think a good deal of the program
we're talking about is a program that the laboratories should main-
tain. Particularly if the lab is running the same type of samples
all the time. Now, as we know, a number of you folks get samples
that solidify, and some that are nice and clean. So, you have a
constant changing of samples.
Mr. Booth; Okay, let me go on with what Bill is saying to you,
and it's here in my notes. If you think about this, any good labora-
tory is going to have a valid standard curve. I think we would all
agree to that. Any good laboratory is going to be able to run method
blanks and field blanks. So, I don't really think that it is fair
to say this is an extra burden upon you. I think this is what Bill
was saying. So, at most, we are requiring you to run a duplicate and
a spike, and that really isn't a hell of a lot more.
It gives you a tremendous amount of information, in terms of
telling you how precise and how accurate the data were at that point
in time, and would have resolved a number of questions we had con-
cerning data, as we saw up there yesterday.
I think your point is well taken. Let me say for the record,
I wasn't necessarily singling out your data. I saw an awful lot
of data on the board yesterday, and if there had been the kind of
quality control practice that we are recommending here today,
there would not have been questions about the data.
Question; I agree with you 100 percent. The big question
that remains is what to use in the data, which was collected under
the conditions of quality control, under which we have it.
Mr. Booth; I think I would be safe in saying that we will try
our best to make use of all data, keeping in mind that we may not
have been able to document it as well as we would like to have done
at that point in time, and all of us, I think, are working under
very tight time constraints. We are writing methods as we go along.
We are writing the protocols as we go along. So, certainly I did
not mean to single out the API. I would make that very clear to
you. I would say to all of the speakers yesterday, that the kind
of protocol that we passed out to you, is certainly the kind of
protocol that you should have going in, not only from a standpoint
of giving you the kind of guidance that you need, as to what the
agency felt was a valid program, but also in terms of pricing it out.
Now, we do have, interestly enough, situations where we can
cut back on the number of samples, and increase the kind of quality
control that we are suggesting here, or we can increase by add ons
the cost of contracts, and the cost of sample analysis, but it
seems to me that with all of the data that we have to collect yet,
52
-------
that we need to be working towards goals that will give us more
reliable data. So, that when we do sit down and look at it, we will
know how precisely and how accurately the data would be gathered.
Question; I think one of the problems is the goals in the
progress. One gets numbers and the other is like a mass amount
of GCMS data which can be searched later. Now, the kind of
label requiring data from the mass spec is simply not the best
way to acquire quantitative data. There's just no way about it.
You really can't get too much better than a 20 percent deviation
acquiring data that we will acquire. Multiple scans, you know,
particularly one scan per second, you don't get enough data from
that.
Mr. Booth; But we want to know what the deviation is. So,
that when we sit down to set limits, we have some feel for what
kind of an acceptance range we're looking at.
Question; Well, I can tell you in our lab what it is, but
it's very hard to say in the internal standard. I know how they
do it.
Mr. Booth; Okay, well there's a question here in the front.
I don't know whether it has been answered yet or not.
Question; We really could spend at least a day on document,
going over the...
Mr. Booth; Yes, I'm sure you could.
Question; Are you talking, first, of the standard curve? Was
this the compound that you've detected, or for every compound in
that group?
Mr. Booth; Just the compounds that are being detected.
Question; We're running away the area curves, before you run
the sample, are you supposed to know then, what's in the sample?
Mr. Booth; I think we now have enough data, let me talk. If
you don't agree with me, you can but let me answer.
Question; All right.
Mr. Booth; Okay. If you were asked are the 114 organic prior-
ity pollutants in this sample. Then, your standard curve better
have the 114 organics. So, that when you're running purgables,
you want to include in your standards those purgables that you're
looking for. Now, what Bill is saying to you is that we know from
past history that there are certain purgables that are showing up,
routine, more often than others. When we get to the other frac-
tions, we know that there are some compounds that aren't showing
up at all. So, it is indeed, a labor saving device to you, not to
do it to begin with, but recognize that you're going to have to
come back later if you get something and run another standard,
53
-------
then so be it, but it would seem to me, it would be a lot simpler
to make up a standard curve for each fraction that doesn't seem
like continuous compounds that you're screening for.
Question; Well, initially you would run a five point standing
curve for every compound that's on the list that you were looking
for, and then on a daily basis you would run a standard curve to
the compounds that you would expect to find?
Mr. Booth; If you do not run the initial standard curve of all
the compounds, you have no reference, so you have to do that, okay?
Question; Okay.
Mr. Booth; So, that you've got some base reference. Then, on
your day to day basis, you're going to run some check standards to
check out that standard curve. As Paul says, he can't get the same
curve from day to day, so he has to run three points, to come up
with a new curve, as we suggested for the GCMS.
Question; You're not talking about the five point, you're
talking about the three point?
Mr. Booth; There's a section here that is special for the GCMS,
and it talks about three points, there. The other section on five,
is the general lab practices, and that's what you'd be using for
trace metals, and phenols, things like that, yes. I think if you
really take some time out to read this, and go through it, there's
an awful lot in here. There really is.
Question; I'm a little confused. If you run a field blank
for that day, that fast test is the one you need to run method
blank or the blank?
Mr. Booth; The method blank?
Question; The method blank, yes.
Mr. Booth; Okay, what you're trying to do there, is to separate
the two operations, and if one checks out and the other one doesn't,
you at least have a pretty good idea of where the problem lies.
Question; Can't you run your field blank first, the fastest?
Mr. Booth; He's saying that if you take the field blank and
run it first, and it's clean, then there would be no problem run-
ning to lab length. Okay, that sounds reasonable.
Question; We would not have to analyze this, the methods.
We'd have to run the field blank.
Question; That's what I'm saying.
54
-------
Question; But with the purgables, you ran a field blank on
the purgables, that may not represent the water that you're using
in the lab.
Mr. Booth; The field data is the same water that is used for
the lab blank.
Question; Well, it may represent a different time sequence.
I mean, that, it might have been sent out of that lab, Bob, and
but it might have been sent out a month
Mr. Booth; Okay, yes, good point there, yes. I'm, again,
assuming that we're working on a fairly short turnaround time
here. If it's not the same water, obviously you're going to have
to go
Question; Just one other question.
Mr. Booth; Sure
Question; Is it possible to get your laboratory precision data
that you would use?
Mr. Booth; He's asking probably what he thinks is a loaded
question, and it is. Is it possible to get our laboratory's pre-
cision data? The answer is, yes, as part of this hand-out, you have
a table at the end that has a table for the priority pollutants that
we have done some work on, showing our percent recovery on some com-
pounds. There's a fairly extensive table at the end of the docu-
ment, there. That is not complete and it certainly does not tell the
entire picture in terms of precision. It is only a single analyst
in a single laboratory. We as a research laboratory do not routinely
process samples. In fact, we have a hard time getting samples. One
of the things we had hoped to do through this study, was to resample,
was to receive samples from contractors, and to do some sample
comparisons, but this, this just hasn't worked out. So, we don't
have real world data.
Okay, those requirements are those that are used by Bill Budde
and his staff for the GCMS system that we have. They are definitely
not pulled out of the air.
Mr. Telliard; If you don't like this program or you wish to
make some suggestions, or you think you have a better one, and you
don't need it, we would like to see it, and Robert would be very
happy to have one of his folks or himself, look at it and say,
C, A, B, whatever the grade may be. Okay, the other thing is, for
all of my IFB contractors, this is not a specific requirement in
your present contract.
Question; What said that, they can go on home and have a good
time.
55
-------
Mr. Telliard; We consider it (QA/QC) a necessary part of your
program. Now, if you're not doing this, what we do need to know,
and some of you have already supplied us with it, is what you are
doing, and what you will continue to do. Now, particularly after
what you've heard this morning from the folks from the pesticide
lab, what you turned in from QC on a lot of that stuff on pesti-
cides, I want to see more of because until we get this thing
straightened out you're going to have to get hammered on. So,
think about it. If you don't like this, what else do you want?
We're easy.
Mr. Booth; May I make a statement too? There was a little bit
of confusion and I'm sure it's on my part because I have gone over
it hurriedly, because I didn't want to waste time on it. On page 12
of this document there is a section entitled, Operations Specific
to Organic Analysis by GCMS. It goes into the business of the
calibration check and the use of the internal standard and how you
go through your quantification. Some of the questions I was asked
during the break lead me to believe that I had misled you in terms
of what I had said about the standard curve. This is unique. You
remember I said, the organic chemist is unique. Here is a section
that applies to GCMS. This is how you go about doing the calibra-
tion check, the use of internal standards, in your calculation, as
far as the organic priority pollutants are concerned.
Mr. Booth; That's a good question. I don't know.
Question; Do you elect help?
Mr. Booth; Yes, whenever it's going to be included, we
probably will want to include more than we have now. I think,
particularly, on the pesticides. Right now we just say, do it by
the Federal registered method. There will be something added on
that, for sure.
Question; I see your point about ones that have actually two
but, in your consideration of this effort,
have you taken into how much extra time it's going to take NRDC and
all of those people, so then you've got to get these things out by
such and such a day, what the contractors have done to the...
Speaker; Speaking of that, sir, I would remind you Congress-
man Brown has gone on record in the house as saying that no piece
of data should be reported without a precision value attached to
it.
Mr. Booth; The point I would like to make is how much is
this going to cost. I made this same kind of presentation as far
back as two or three years ago, to the Regional S&A Directors. I
was asked the same question there. My God, how can we afford to
spend 15 to 20 percent of our time with QC? At that time, they
were in a very ambitious sampling program, where we were looking
at permits. The number two man in the agency, John Quarrels, was
there, and he answered for me, and said that I want to know once
and for all, what the problem data are, and I would rather see you
56
-------
run 15 to 20 percent fewer samples, in which we have valid data,
rather than run 100 percent samples, in which I don't know the
quality of the data. So, from a cost standpoint, from where I
sit, although I do deal in the real world, I still have to come
back and say that what we have asked here we feel is the kind of
program that represents a minimal baseline effort in any reliable
laboratory that is documenting this data would be doing routinely.
If you're doing something less than this, you need to document
why you're not doing this, or why you have supplemented it with
something else to take its place because in your final analysis,
you need to be able to answer the question, "How precise and how
accurate are my data?" So, in giving you a cost estimate, I could
say that the additional cost would vary anywhere from 100 percent
on some situations, to as little as 10 to 15 percent on other
situations.
As an agency guideline, we say for a laboratory first starting
out, that they should figure about 20 percent of their time will
be spent on quality control. That's the cost of doing business,
but once they are operational, after the first year, that they
ought to be budgeting 10 percent of their budget for quality con-
trol as a minimal baseline effort.
Question; I'd like to go over just the simple mathematics of
this again, especially for acids, base/neutrals, and purgables by
GCMS. This is a very important point because we're talking 15 to
20 percent of your time, and yet 6 or 7...
Mr. Telliard; Now, wait a minute, that's not what we're say-
ing. Now, I would recommend strongly that you read the thing, sit
down and you talk to Robert about it. It is now a quarter to
eleven. We have not gotten to metals yet, and we have a lot to
cover.
Question; Well, if you'd just sort out that one problem for me,
I'll be goad to move on.
Mr. Booth; Well, during the break I think Bill was bringing
up the point that most laboratories that are running these priority
pollutants, as far as the mass spec is concerned, are processing
samples on a weekly basis, where they have a number of samples
going. So, to be able to say that today I have run sample one,
tomorrow I'm running sample two, it's rather difficult to sort
that out. In terms of time, what I'm saying is that in terms of
your laboratory, you should be running a method blank and a field
blank, and for your GCMS, you should be following the protocol that
we have outlined for just special requirements for the GCMS, which
as one of the contractors noted to me, I don't know how else one
could obtain results, if he weren't following this procedure.
Question; Well, I think that's fine, except I think you ought
to talk about the time the fraction of the amount of time, that
is required for QC in realistic terms, and not in 15 or 20 percent
of the time, because that is not realistic.
57
-------
Mr. Booth; It depends on the number of samples that you
process.
Question; Well, now you just said whenever you change shifts,
you have to recheck the QC.
Mr. Telliard; It says that in the protocol. It says that in
your contract.
Mr. Schaffer; You're not doing that?
Question; Again, I bring up the point, you don't talk in
terms of 15 percent of your time.
Mr. Booth; Well, again overall and if you're doing nothing
but the organic priority pollutants, obviously it's going to take
more than 15 to 20 percent of your time. I don't think anybody
argues that, but most laboratories for their overall operation,
certainly for the kind of laboratory that we work with, are running
a wide variety of sample measurements, sample matrices, and con-
taminants. It will even out if you're doing that, to about 15 to
20 percent of your time.
Now, if your laboratory is doing just the organic priority
pollutants, obviously, you're going to be spending a greater per-
centage of your time doing quality control, but I would still come
back to you and say, how else do you know the quality of the data
coming out of your lab, if you don't do these things? I think
that's the bottom line really. I'm not trying to add additional
costs for you. Many of the contracts being written now, build this
into the contract, so the contractor charges for each sample that
he analyzes.
Question; Well, you're familiar with the protocol, and I
think you could go back and I think you should go back, and realisti-
cally look at a difficult shift -- what you require, and how many
samples you think you can run, and know how you could estimate, and
be paid with the particular programming, and how long it will take to
run a sample. I think you ought to go back and look at this.
Mr. Booth; Okay, I think if your laboratory were doing just
the organics, I guess you could reasonably assume to do roughly
two to three samples per day, for all of the organic priority
pollutants?
Question; Per shift?
Mr. Booth; Yes, per shift.
Question; You said one, because if he said one hour, then
you said seven times, and it's one hour to run one sample. That's
eight hours. One sample...
Mr. Booth; That's what somebody said out there. I didn't
say that. What I'm suggesting is that on an average day, spread
58
-------
out over a time frame work, how many samples can you reasonably
expect to run for the organic priority pollutants? Two to three?
Three to four?
Question; Well, that's a very difficult question, how long it
takes to run them. It's just hard to sort that out, but a volatile
run takes about 45 minutes to 55 minutes to, just to run. That's
just to run itself.
Mr. Booth; We don't want to take too much time on this,
but at the same time, I think it is a very important point. So,
I don't want to brush it off. If you're doing just the organic
priority pollutants, let's assume for the moment that you can proc-
ess three samples a day, completely. Now, maybe it will be more
or less, but let's day three for argument sake. Then, you would,
by this protocol, need to run on that particular sample, three
samples, a methods blank, a field blank, a replicate, and a spike.
So, you would have four extra samples to run, and your cost would
have at least doubled.
Question; But I think if you sit down and look at an eight
hour day, you'll find that it is much worse than that, because you
don't just cut the samples in half. The time fraction will not
work out that way, because you can only get a small fraction of
the amount sample run, before you have to do the entire certifica-
tion procedure all over.
Question; Bob, I believe that we went through this analysis
when the inorganic parameters include the protocol. 54
operations called for 17 percent, and for the organics, I believe
with cooperation of the studies percent, that each laboratory has
to digest their sample load in that term. Those are the two
numbers that came out of the program.
Mr. Booth; It depends upon the number of samples and the
kinds of parameters that you're being asked to measure. Okay, I
think we've beat that.
As far as the priority pollutants are concerned, it's a matter
of public record, since 1974, that there are public data available
on trace metals. There are data available in our methods manual,
and in ASTM and in standard methods for those metals. Their data,
I believe I'm referring to the same, are available from ASTM, and
upon request from us for the pesticides, PGP's, and phenols, and
we have single operators, single precision accuracy data available
on the purgables. In the coming fiscal year, we're going out to
do methods validation studies on real world samples for the remain-
ing organic priority pollutants.
Question; There's been a lot of talk about checking within
the laboratory operation, but in the field program that has gone
on during the past year, a part of the trouble has not been in
directly transmitting, handling the field blanks, in the field. I
think one of the problems is that this needs to be spelled out, in
59
-------
much more detail as to what our EPA, Federal and regional contrac-
tors do. We have experienced differences in the way they collect
samples and how that field blank is taken for the blanks, so that
they truly relate to the problems of contamination back in the
laboratory analysis. I'd say the sampling — I agree the labora-
tory was tested to be found right — has to be done correctly also.
That is the point I wanted to make.
Mr. Booth; I couldn't agree with you more, sir. It's quite
well taken. The problem, quite frankly, is we can get a better
handle on the lab than we can on the field. The sampling protocol
that we have for the priority pollutants is the best we've got at
this point in time, but your point is well taken.
I might add that for those of you that have the guidance docu-
ment for sediments — and some of you may or may not have them —
that it is in the process of being reviewed and revised. One of
the major revisions will have to do with the sampling end, as to
how to properly collect a sediment sample that is indeed a repre-
sentative of the problem area, and how to properly transport it
back to the laboratory, and how to properly process the sample
required to the lab analysis. We recognize that this is a very
serious weakness in the program.
Question; You talked about performance evaluation samples.
I'd liketo know what your rules are with these samples. Talk
about the composition of the performance evaluation samples.
Mr. Booth; No, I'll tell you why. Right now, we've got a
limited number of different sets of performance evaluation samples.
That's all, until we can get more, but until then I'd just as soon
tell you, you passed or you failed.
Question: That's fine, but I've heard three different grades,
from three different people, and I would like to know what the
Mr. Booth; I called the grade I can give you.
Question; Okay.
Mr. Booth; But I'd rather not tell you what was in there,
either you missed five or you got them all, or you missed two in
quantification. I'd like to keep it that way. Simply because we
don't have any other check pattern available. John asked me not
to tell.
Mr. Booth; Let me take a crack at this for you if I may.
What we do back at the laboratory is to take the raw data, it gets
assigned a number, a special code number. This gets fed into the
computer. The computer reviews your observed value against the
true value that was in the sample. Then, it checks to see if this
was in, what we call the acceptance range. The acceptance range
is very reasonable, in that it is based upon three standard devia-
tions. So, 99 percent of the time, in fact 99 percent of the time
plus, the data should fall within those limits. If they don't,
60
-------
you're clearly out of control. These limits have been established
based upon the same kind of samples being given to our regional
and state laboratories that are using the same general techniques.
Question; Bob, could you combine the path program with yours
and save the taxpayer an awful lot of money? They have a very
extensive program, a tax program....
Mr. Booth; I'm not familiar with the tax program.
Question; .., the performance evaluation testing for the
dioxin program and all of that, they go through a very extensive
computer to study it. We get samples on a quarterly basis, or a
monthly basis very much like yours. The same kind with low levels
for the GC, and the metals analysis, the whole ball of wax.
Mr. Booth; Is this the program that Bill Kelley is involved
in?
Question; Yes.
Mr. Booth; Okay, yes, I know Bill. We have compared notes
along the way.
The same manipulations of the data, in terms of how we compare
the observed values to the true values, and working with pair vol-
umes, was worked out jointly by their staff and our staff. So, we
do use the same general approach for the data handling.
Question; Well, that would give it too.
Mr. Booth; Let me make some comments that I was asked to
make at the coffee break, and then I'll wrap it up. Well, I did
see more hands.
Some of the questions that came up last night in a bull ses-
sion that we had on this, concerning some other uses of this proto-
col, I think would be of interest to the group. That is, what
happens when you have a laboratory that has to get some data as
soon as they possibly can? They're just starting out and they
want to get data. What you need to do is to arrange to have the
initial 15 samples coming into the laboratory, representing a wide
variety of sample matrix types, so that you have as much of the
population sampled out there in setting up your performance cri-
teria.
If you don't do this, if you take data from just one single
source, you may come up with values that exceed what your normal
day to day operations should be, or conversely, if you take a clean
sample, that you may very well come up with limits that are two
types. So, keep that in mind.
Also, if you are working with samples in which the natural
background is below the detection limit, you obviously have prob-
lems. Now, seriously, this is a very good question. If you get
61
-------
involved in fish and sediment analysis, we have concluded that
your best approach at this point in time, is to do your spiking
before you start processing the sample in the laboratory. In
other words, make sure that your fish sample has been properly
ground, properly homogenized, and you have taken subaliquotes of
representative samples.
Then, take one of those subaliquotes and spike it with the
contaminant of interest and run it. The same way with your sedi-
ment sample. We're suggesting that your solvent of choice here,
would be acetone or a similar polar type solvent. I think those
were the pertinent things that came out of that meeting that I
would like to share with the group.
I'd like to thank you very much for your time, and for the
spirit in which your questions were given. Let me simply echo what
Bill said, in that at this stage we don't consider this cast in
concrete. We are trying our best to provide a service. I might
point out to you that this is something that is not going to go
away. The agency has a special task group that is working on
quality assurance right now, and there are some basic QA guidelines
that must be followed by all persons at regional/state level, that
are generating data for storage, whether it be ambient air, or
ambient water. This is forthcoming in the very near future. There
are some special study groups that are looking at the whole problem
of quality assurance all the way across the board. CEQ has a
special task group on that, of which I am a member. There are some
big things happening there that will be forthcoming in the near
future.
There is also a very serious consideration being given to the
agency setting forth a policy on what should be written into all
contracts regardless of monitoring time. So, with the ambient
point source, we'll know what you have. So, QA is here to stay.
I hope we can work together in solving the mutual problems.
62
-------
Metals Analysis
B. Fairless
U. S. Environmental Protection Agency
Mr. Telliard; Bill Fairless has got all his numbers in
hand, and he'll come up here and lay it on you. So, if Chuck
Taylor and the metals folks would come up, we could talk about
the hard ones.
Dr. Fairless; The current procedures being used by Region 5
to analyze metal samples are to receive the samples in one liter
high density polyethylene bottles, using labels that we supply.
The numbers are coded so they identify the industry with ten digits,
two digits refer to the lab, and two digits identify the sample in
any given project. Again, I would like to encourage everyone to
use the six number system. It really helps us to keep track of the
samples.
When the samples arrive, all are now unpreserved. We add nitric
acid and let them stand for 24 hours. We then take one aliquot
for mercury, and one aliquot for metals. We use a strong acid
digestion for the ICAP metals and take them up using diluted hydro-
chloric acid for introduction into the ICAP instrument. We normally
transfer 100 milliliters of sample into a beaker, acid digest that
volumn, and then bring the volumn back to 100 milliliters with
dilute hydrochloric acid. We measure the volumns by weight dif-
ferences in all cases. The sample is exposed to one beaker and
one plastic bottle.
We normally run 50 samples a day, and 16 QC samples. There
would be two laboratory reagent blanks, two duplicates, two spikes,
and two check samples for each set of 25 samples in the group of
50 samples.
Question; Do you verify your initial calibration?
Dr. Fairless; Yes. We calibrate once in the morning, and
then we calibrate throughout the day. The calibration criteria is
to recalibrate when one or more of the 20 metals is more than five
percent from the design value.
We have analyzed approximately 3,000 samples. Therefore, we
have 300 to 600 reagent blank values, duplicate pairs, etc., on
record now. We are reporting summary values for these audits to
Bill Telliard each quarter. The summaries would be for a three
month period and would include standard deviation, average, etc.
for the duplicate pairs, spike recoveries, etc. for each parameter.
To get on to the background, I presented a considerable amount
of analytical summary data at the Denver meeting, and I don't want
to repeat that. The only thing that we have added since the Denver
meeting, is to begin a program of changing the concentrations of
our check standards, so that we can estimate the precision as a
function of concentration.
63
-------
I have also summarized some data, I guess it was Wednesday,
the day we came to the meeting, for several parameters, where we
had received questions about scatter. These include chromium,
lead, and some other parameters, at concentrations of 10,000, 8,000,
1,800, and 180 parts per billion. The relative standard deviations
for all of those metals were under 10 percent, until we get to the
100 parts per billion concentration range, and then the relative
standard deviations increase. However, it is less than 30 percent
at these lower concentrations. In fact, it is less than 20 percent,
at 100 parts per billion for the scatter that was measured on dis-
tilled water standards for approximately, a three month time period.
The other point I would make is that we also look at the
scatter from our duplicate analysis. These are two aliquot
taken out of a sample that we receive in the laboratory and analyze.
They're not duplicate samples but duplicate analyses of a single
sample. The precision on duplicate analysis at a 95 percent com-
petence level is around 30 percent for most of the parameters.
We are involved, at the present time, with the American
Institute of Iron and Steel, and are analyzing some split samples
using both atomic absorption and ICAP data. The results will
probably be available in a couple of months. I guess that's all I
have to say, unless we have any questions. Yes, sir?
Question; Are you running any samples two or three times?
Dr. Fairless; It depends on the concentration. At low con-
centrations, we're not. If the concentration is high, where you
get interference, we dilute and run the sample again separately.
I might add, we established our quality control procedures prior
to getting involved in this program during which time we had
relatively clean samples to work with. We have elected not to
relax those control limits because we're looking at dirtier samples.
So, on a given day, we normally process 100 samples, and report
finished data from about 50 of those. We are repeating a high per-
centage of the samples because many are out of control for the
initial analysis.
Question; Did you compare atomic absorption results with
ICAP results?
Dr. Fairless; Yes. When we first purchased our instruments
in 1973, we spent approximately five or six months comparing the
two methods. We have an alternate test procedure approved which
shows that data. We did get excellent agreement between the two
methods. Since that time, we have participated in, five or ten
studies where we compared data. To my knowledge, we've never got-
ten differences that could be attributed to the analytical methods.
We also participated in several of the drinking water programs.
We have a lot of results on blind unknowns from the surface water
programs, the IJC programs, the State programs, etc., so we have a
considerable volume of analytical data, probably over 500 samples,
over the last three or four years documenting the reliability of
the ICAP method.
64
-------
Question; I need a draft copy of the EPA digestion procedure.
Have you tried different digestion stages going from dryness, versus
not going to dryness?
Dr. Fairless: We're looking at that currently, and I can't
really comment to any extent on the water data at this time. We
are looking at sediments. I think it does make a difference in the
sediment data. I haven't had an opportunity to look at the drinking
water or surface water.
Question; What's the basis for not preserving samples? Was
that checked out or not?
Dr. Fairless; We have not checked it for all of the elements.
We have looked at it from recovery, and we have experienced that
when you have a clean sample, the mercury does disappear. We sus-
pected on the walls of the container. We are able to recover
it by adding nitric acid and dichromate. We normally used dichromate
as a preservative for the mercury. We didn't experience that
problem in samples having high suspended solvents. If you recall
the Denver meeting, it was decided not to use nitric acid as a
preservative due to specific shipping problems. This was the best
procedure available under those conditions.
Question; What method do you use for arsenic?
Dr. _Fairlessj We use flame atomic absorption, and we also
have used a hydride reduction procedure. We're not analyzing
arsenic as a part of this program. Those are the methods we use
for other programs.
Question; Have you compared the flame AA method with the
hydride procedure for arsenic?
Dr. Fairless; No, not in our laboratory. However, I have
recently evaluated some other laboratories, and it looked like
they were getting better precision with hydride, and a better
detection level, than with flame AA.
Question; Is it possible to get that information?
Dr. Fairless; Yes, why don't you see me after the meeting and
I can refer you to the people actually doing the work.
Question; Did I understand you to say that you are analyzing
field blanks?
Dr. Fairless; No, I'm talking about the laboratory reagent
blank. In our regular regional program, we do get field blanks.
In this program, we get a sample with a number on it. We analyze
laboratory blanks, and if the contractor sends us field blanks, we
analyze them as samples.
65
-------
Question; As far as I know, the procedure for collecting a
field blank for the metals would be to give sample bottles to the
crew which are doing the sampling and require that they follow in
general the sample collection procedure and I don't know if this
is a problem, but it could be.
Dr. Fairless; It very definitely could be. Several years ago
when we were evaluating various types of sample bottles, and pro-
cedures, we identified numerous problems. The one I recall was
the glue holding the liner in the cap which had high concentrations
of potassium. We had a hard time finding a bottle cap that did not
leak and which was also contamination free. So, it is a serious
problem. Again, I recommend that everyone use specific bottles. We
do have quality assurance data on those bottles and caps that
were recommended. We haven't experienced any problems with them.
I am sure some of the other bottles being used do contaminate the
sample. At the Denver meeting I showed a slide of a sample we
received in an orange juice bottle with a metal cap on it. We're
not getting that type of sample any more. It looks like most people
are using good bottles and good caps.
Question; I have a question along that line. Do you do the
analyses for the recent EPA laboratories and if so, what kind of
sample containers are they using?
Dr. Fairless; The answer to the first question is yes, we're
doing the analyses for the other regions. For the most part, they
are using the sample bottles that we recommended.
Question; Why not require that everyone use the recommended
sample bottle?
Dr. Fairless; We still get some bottles from some contractors
that they had on hand. As you recall, I wanted everyone to use
the same type of bottles in Denver, and I asked everyone to use the
same type of bottles, but some of the contractors said that they
had a large supply of sample bottles on hand. So, I told them to
continue using the bottles they had on hand, and when they ordered
a new supply, to order the type bottles that we recommended. There-
fore, the change to the bottles we recommended is going to occur
over a period of time.
Question; There's another source of contamination in the
metals procedure. I know of one lab that uses rubber policemen
to transfer samples from one container to another.
Dr. Fairless; Yes, the comment was that any time you have
rubber that's coming in contact with your samples, you will probably
observe zinc contamination. I assume everyone is aware of the
problem with rubber and that it's a source of contamination. I
guess all laboratories that analyze for low concentrations experi-
ence this type of problem. This is why we went to the gravimetric
measure of volume to try and minimize the items that the sample
comes in contact with. We've eliminated that down to the point
described. Another source of scatter is getting a representative
66
-------
aliquot. We shake the sample as received and try to get a repre-
sentative aliquots. That's very hard to do, in some of these
samples as you know. The fact that we take 100 milliliters helps
us get representative aliquots.
As a final point, I might add that we appreciate your comments,
If anyone has questions about data quality, we'd be very happy to
participate in programs to identify and resolve any problems that
you may have. We would be very reluctant however to participate
in programs where we were not involved in the planning stages. If
anyone has such problems, you can contact myself or Curtis Ross.
We'll be very happy to work with you and resolve any problems that
might occur, or any problems that you think we have.
67
-------
Metals Analysis
J. Michalovic
Calspan Corporation
Mr. Michalovic; My name is John Michalovic. Calspan Corpora-
tion has been involved very early in the Effluent Guideline Program
in the analysis of metals and as they relate to the Ore Mining and
Dressing Industry. The metals work performed covered a wide range
of concentrations. When I was asked to speak today, I was asked
to talk about problems in the analysis of metals. In the particular
industry which we were investigating, it seems the major problem in
the analysis of metals is related directly to the amount of sus-
pended solids in the sample. The samples that we were looking at
have suspended solids that ranged from less than one milligram per
liter to 40 percent by weight. The samples with high solids cannot
be digested completely by the standard procedures.
To sample in the field, high solid-streams to obtain duplicate
analytical samples is a problem, especially from streams which con-
tain very fine solids to very heavy sands. The heavy sands settle
extremely rapidly. In the sampling, much care has to be taken to
represent all sizes of solids in that particular product stream.
Once the sample is stablized and comes to the laboratory, the
problem is in resampling a portion of the sample for analysis.
Duplicate analyses on aliquots of the same sample vary to a larger
degree than "clean" samples because of inaccuracies in sampling
the solids. Our analytical data has differed in the past with other
laboratories' data, and I think the major difference is how the
sample is well shaken and the solids suspended while obtaining an
aliquot. Now the protocol says a well shaken sample. Now, what a
well shaken sample for one person is, is not necessarily what a
well-shaken sample is to another person. Samples with high,
heavy suspended solids that are well shaken start to settle
immediately upon standing. If a sample is poured off the top, it
will be different than if we stir vigorously and while it is being
stirred withdraw the samples. Here has been our biggest problem.
Currently, we shake the best we possibly can to keep everything in
suspension while sampling.
Even when taking every precaution, the analysis of replicate
samples, if you compare percentage differences to relatively clean
samples, the relative percentage difference is about a factor of
two greater in samples with large concentrations of suspended solids,
I think this is something which is going to have to be spelled
out in the protocol, as to what is well shaken, especially with
this particular type of sample. That's where I'd like to leave my
comments today with the solids sampling part of the analysis. I
have brought actual data with me but, just to move things along,
I will not show it at this time. But to summarize the data--the
overall relative percentage difference average was 20.5 percent
68
-------
for six different elements when the suspended solids are high.
When the suspended solids are low, 40 milligrams per liter or less,
the average relative percentage difference comes out to be 8.9
percent for the same six elements.
Mr. Telliard; We expect you to put a sample on a magnetic
mixer. It should be a magnetic mixer that slurries everything so
that all your solids are at least in suspension and you run a pipe
from the side of the container half way up to the liquid level.
That's well shaken.
Mr. Michalovic; What I'm speaking about is not necessarily
an analytical problem as far as instrumentation or the actual
analytical technique, but in the techniques prior to analysis.
I think the only thing that I have to add would be this, on
checking the ICAP instrument against other methods. We have
an instrument that was built by a different company than the one
that Billy Fairless has, and we feel that they are probably
pretty equal. We have checked our instrument against AA in
numerous private labs and regional labs. We have checked with
neutron activation, with spark source, using isotope dilution,
and in general. We have found that it will not only reproduce
numbers from other methods, but equal to or better than produced
by other kinds of instrumentation.
Question; With the solution of the sample, I'm kind of
aware of another problem, at least there are several points of
view. When one follows the protocol as a guideline, let's say,
there is a residue in the bottom of the container holding the
sample. My question is, what does one do upon taking an aliquot
up to this digestion center for analysis?
Mr. Michalovic; The only thing I can comment on is what we
are doing in the laboratory relative to that. As I understand
your question, you're saying what do you do with samples where
you have a residue after the digestion? If the residue is standard
type and settles to the bottom, there is no problem. If it does
not, we filter the sample. We do not aspirate the solid material
into the instrument, for obvious reasons.
Mr. Michalovic; As a final comment, we're looking at methods
for the metals by automated procedures, and we're specifically
interested in the production of good analytical data and new tech-
niques. I anyone happens to have any information, I'd appreciate
it if you would contact me.
69
-------
Cyanide Analyses
Problems with Analytical Technique
G. D. McKee
U. S. Environmental Protection Agency
Mr. McKee; The purpose of this discussion is to identify some
of the basic problem areas and difficulties that affect cyanide
analyses. The procedure is a classical wet chemistry technique
that involves several steps including an acid distillation, absorp-
tion of a gas in a liquid scrubber, a choice of two colorimetric
measurement techniques for sample concentrations below 1 mg/1 and
a titrimetric measurement technique for sample concentrations above
1.0 mg/1. The ion selective electrode for cyanide is currently
under investigation and is also proving to be a satisfactory detec-
tion system. Also included are a number of optional steps for
pretreatment of interferences. As with any analyses, the number
of manupulations increases the potential for error in the final
result due to the variability introduced at each step.
The first step that error may be introduced is immediately after
the sample is obtained. The alkaline-chlorination decomposition of
cyanide is well known and, therefore, chlorine, if present, must be
removed prior to preserving the sample with sodium hydroxide. The
procedure states that the sample should be checked with potassium
iodide-starch test paper (Kl-starch paper) for the presence of
oxidizing reagents. There are a number of reports about the insen-
sitivity of the Kl-starch paper. It was reported by one individual
(EPA-Region I) that the Kl-starch paper being used did not detect
less than 10 mg/1 chlorine, other users have verbally reported de-
tection limits of 5., 2., 0.5 mg/1 chlorine. The Kl-starch paper
must be kept dry and out of the sunlight if the sensitivity is to
be maintained with any degree of confidence. Certainly, there will
be significant differences in reported concentrations of cyanide if
these high levels of chlorine are present in some samples when the
sample is preserved with sodium hydroxide.
There are two other optional pretreatment steps that will af-
fect the final results. Sulfides if present in the original sample
will distill over under acid conditions as hydrogen sulfide and be
trapped in the alkaline scrubber and adversely affect both colori-
metric and the titration procedures. The 1974 edition of EPA's
"Methods for Chemical Analysis of Water and Wastes" (Methods Manual)
recommends checking for the presence of sulfide with lead acetate
test paper and if present, removing the sulfide from the sample by
treating with cadmium carbonate and filtering. Both of these steps
have a potential for producing error. The first error is the pos-
sibly poor sensitivity of the lead acetate test paper to detect very
low levels of sulfide that interfere and the other is the treatment
with cadmium carbonate and filtration step. Formation of the pre-
cipitate and filtration may sorb some of the cyanide complexes and
when this process is physical, mechanical or a loose chemical at-
traction, the porosity of the filter paper, rate of filtration, time
required for the cadmium sulfide (CdS) precipitate to form, solids
70
-------
originally present in the sample and the amount of precipitate (CdS)
formed will affect the amount of the cyanide present in the filtered
sample.
Another pretreatment step, if required, that may introduce error
is the removal of fatty acids. If present, these fatty acids will
distill over and form soaps with the sodium hydroxide. To remove
them, the sample must be acidified to between Ph 6.0 to 7.0 and ex-
tracted with one of several optional organic solvents. Lowering the
pH will cause HCN gas to form and may be lost from the sample during
the extraction. Length of contact time and degree of shaking will
affect the amount of cyanide losses.
When the cyanide is dissolved in the sodium hydroxide scrubber
solution, there are three alternatives for measuring the cyanide.
Either pyridine barbituric acid or pyridine-pyrazolone may be used
to develop a cyanide complex that may be measured colormetrically.
The pyridine-pyrozolone is not a stable reagent and therefore cali-
bration must be checked frequently. The time of reaction for the
cyanide and the chloramine-T and the time of reaction between the
cyanogen chloride and the color reagent must be kept identical with
the standards and samples. The "Methods Manual" recommends adding
the pyridine-pyrazolone immediately after addition of the Chlora-
mine-T; better precision is obtained when this time for reaction be-
tween cyanide and the Chloramine-T is held constant at about two
minutes prior to adding the pyridine pyrazolone. If the sample con-
centration is above 1 mg/1, the cyanide in the scrubber solution
should be measured by titration with silver nitrate and benzalrho-
danine indicator, experience with the endpoint is necessary for
precise answers.
The next area of the analysis that will cause differences in
reported concentrations is the use of different catalysts in the
distillation step. The procedure as written in the 1974 edition
of the Methods Manual states that copper chloride is to be used as
the catalyst. The December 1, 1976 Federal Register, Vol. 41, No.
232 references the EPA method but there is a footnote that recom-
mends the use of magnesium chloride for samples suspected of having
thiocyanate present. Other referenced procedures in the above cited
Federal Register recommend the optional use of both catalysts. The
use of different catalysts in some sample types will result in wide-
ly differing reported concentrations and the catalyst used must be
known when comparing data.
Another area of potential wobble in the determination is the
rate of distillation. The rate of the vacuum distillation changes
and requires readjustment as heat is applied to the sample and to
maintain a constant accuracy. Although this seems obvious to most
analysts, this area requires constant attention and the rate of dis-
tillation and subsequent absorption rate has been a problem.
Another area that produces an inherent wobble is the procedure
used for preparing a standard curve. Although the EPA "Methods Man-
ual" recommends preparation of the standard curve using distilled
standards, if the standard curve is prepared using non-distilled
71
-------
standards, the curve will be approximately ten percent more precise
and also more accurate since any error during distillation is elimi-
nated. Then after obtaining a standard curve using non-distilled
standards, the distillation technique should be checked by distill-
ing standards. Using this procedure will produce a more precise and
accurate standard curve and will also ensure the entire procedure is
working well. This procedure for standard curve preparation is also
less time consuming.
In summary, the procedure for total cyanides when used with
care and attention and the options are held constant has a relative
standard deviation of about + 10 percent at concentrations less than
500 ug/1 and above 1 mg/1, and about 10-15 percent between concentra-
tion levels of 500 and 1000 ug/1. The recovery of cyanide is matrix
dependent. Simple cyanides when added to complex samples containing
metals and metal complexes can form new cyanide compounds. In gen-
eral, recoveries are found to be aboue 85 percent when simple potas-
sium cyanide is added to samples.
The other aspect of cyanide analyses is the "Cyanides, Amenable
to Chlorination" procedure. This procedure was developed to evaluate
the effectiveness of the alkaline-chlorination treatment process and
should be used for that purpose. There are a number of relatively
recent procedures used to measure and define simple cyanides, free
cyanides, and easily dissociable cyanides.
Are there any questions?
Question; All right. I have observed that there's been a
great significant difference, small numbers as compared to large
numbers, between EPA analysis and the refinery analysis. So, there
is this problem and it actually has been serious in some of the
field work that's been going on. I wanted to set the stage a lit-
tle bit by making that observation. It would be awfully nice if
someone from the ASTM group in industrial work analysis, could take
part in that. It's part of ASTM, 1906, and at a meeting a short time
ago, Jim Lichtenberg from your laboratory was in charge of that com-
mittee. Now, there as a sub-group working under Jim's committee,
and I don't know the numbers of that. I want to ask you about how
active you have been with ASTM?
Mr. McKee; Well, I've been with the Agency six months, and I
am becoming active in ASTM. Our Agency has been represented in the
past by Bob Booth and Larry Lobring who know cyanide analysis very
well. We participate in every round robin that we possibly can.
Do you want to elaborate on that Bob (Bob Booth)? Yes, we have
worked very closely with this group, we have worked very closely
throughout the entire time on that problem. There has been a spe-
cial cyanide study group composed by members that are active in the
ASTM and also Standard Methods. We have all been working together
as a team to come up with the best available procedures for EPA,
ASTM, and Standard Methods so that they will all state the best and
same procedure for cyanide.
72
-------
Mr. McKee; Some of the problems that I've pointed out are due
to the three references containing slightly different variations
that affect the results. Would you like to address the magnitude
of problems and the magnitude of the differences in your studies?
Question; Yes. These values are in the development document.
That in one sample, we found 0.02 milligarms per liters, if I re-
call, whereas, EPA reported 0.80 milligrams per liter. Now, in two
other samples that were taken, the effluent and final effluent, we
were quite close together, and what happened in this third sample
there's just no way of tracking this. It could have been something
as simple as the sample. Beyond that, I can say that we, within our
own refinery laboratories, problems with interfering materials, and
the discussion you gave was very helpful to me, by saying that there
are things to look for, and things to do to try to correct, but I
think it's necessary for any group who is addressing themselves to
a really useful method, to look at the real world samples. That's
where the problem is, the interferences must be identified.
Mr. McKee; We attempt to analyze as many as we possibly can.
The problem comes in, as you know, that the sample matrix varies so
much even within one industry that even if we do a lot of different
types, we can't cover all of the possibilities of the sample matrix.
Question; I might add that for the round robin studies that
have been done in the past couple of years, they haven't been work-
ing with real world samples in this special study group. We have
the samples that they can obtain, and the samples that we use con-
tain sulfides, things of that nature.
Mr. McKee; Thank you, right back here?
Question; In ore mining, we're having problems with cyanide.
It seems to mainly stem from complex cyanide and sulfide. The sul-
fide which is generated from the samples, while the reaction will
form H2S we're generating H2S at the same time, for cyanide. We
haven't found any way to go around this. I don't know if anybody
else or your laboratory has had any.
Mr. McKee; We haven't attempted to solve that specific prob-
lem, but I am aware of the problem. My first approach to this would
be to redistill out of the scrubber solution, if that doesn't work,
treat the scrubber solution with cadmium carbonate. Because the hy-
drogen sulfide is not present initially, it is released upon acidi-
fication and the distillation, and therefore it's in the scrubber
solution. You can then treat the scrubber solution with cadmium
carbonate to remove the sulfide, whether or not this will be an en-
tirely satisfactory procedure, I'm not sure. I think that would be
a satisfactory method.
Question; For some time now, we've been examining pulp and
paper industry up close, and we generally see cyanide in paper in-
dustry effluents only if the mill has been practicing waste paper
use. Also, we've had some difficulty with the cyanide available for
chlorination tests. We've been employing the Woods River, Jackson
73
-------
and Roberts Method, and I think this information may have been com-
municated to you, via Doctor Lancy. Would you be interested in see-
ing our information on the Wood, River, Jackson, and Roberts Method,
in our effluents, is one question.
Secondly, I'm kind of concerned as to whether or not there's
any point in our pursuing the free cyanide situation at least from
the paper industry's point of view, because of perhaps my limited
knowledge of what happens when free cyanide and total cyanide data
is presented to regular people. It seems that the argument is free
cyanide is what we should be concerned about, and total cyanide
doesn't seem to be of importance. Please excuse my ignorance, but
it seems that we may be going to trouble to collect data that's not
going to be of use anyway. Would you comment on that?
Mr. McKee; The answer to your first question, is yes, defi-
nitely, and I am aware of some of the data from the pulp and paper
industry. I don't know if it's that specific data that you are re-
ferring to or not. As far as the regulatory bodies go, I think
that there is going to be a need for both, simple, and total cya-
nide. The simple being the immediate toxicity problem and the
total because it does degrade, it does become simple over a period
of time, under the conditions that exist on a normal stream. So,
I think that we are going to be concerned about total, because it
does not remain tied up as a complex cyanide.
Question; You can see that regulations ultimately including
both forms, assuming there's a test that's valid in those.
Mr. McKee; I can't assure of that in any way, not being a
regulatory person or having a definite particular insight to that,
but my opinion is yes I think so.
Question; Would the regulatory person comment?
Mr. Telliard; ...No, no. Most of the regulations and permits
have additionally addressed total cyanide, or total cyanide amenable
to chlorination, but I haven't seen any permit with free cyanides.
Question; Gary, do your permits all have CNA?
Speaker; No, some of them have. Did I answer your question
satisfactorily?
Question; Yes.
Mr. McKee; Are there any questions in back of the room?
Question; There's another industry, the photographic industry,
that has a cyanide problem too. I think it would be nice to take
a look at the micro diffusion method and there has not been a lot
of work on that.
Mr. McKee; What group is this?
74
-------
Question; The National Association of Photographic Manufac-
turers, NAPM.
Question: Gary, we have worked closely with that group.
Question; They have done a mini round robin on that and found
it is very operator dependent, in that the operator must be well
trained. If you're doing it for the first few times, you're going
to have a terrible time, but once you're trained, it does appear to
be a very promising procedure.
Question; It is available to anybody?
Mr. McKee; Bill, let me clarify one point. Bill, just asked
me, he just said well, which industries are the ones where CNA
does not work? It's the ones that we're hearing from. It's the
refinery, the pulp and paper, and metal finishing industries. I
think that the regulatory people and the technical people are gett-
ing closer together to solve this problem now. You had a question
over here?
Mr. Telliard; Anything else? Everybody is happy with CN?
Another problem with CN is preserving them. You said, how far do
I have to go, pH twelve or greater?
Speaker; Greater than pH 12.
Question; Part of this problem that we've faced in ore mining
and some other people have faced, is getting the things analyzed in
12 hours and 33 seconds, or whatever it is. When you're out in the
middle of Idaho it's not always easy to double distill a sample, and
even putting it in the cockpits of airplanes takes longer than the
time specified. That's a real pain in the you know what, because of
that. I don't know if there's any answer to get around it, but it's
nice and simple up in Philadelphia. You can generally get the sam-
ples back. If you're out in New Mexico, it's really not that simple
to take your sample and get it back in time. So, whatever methodol-
ogy you pick, (I don't know which preservation) the technique is
going to be looked at.
Speaker; I would collect the sample, preserve the pH to
greater than 12, ice if possible, ship them to the laboratory and
keep track of the time between collection and analysis. Do not
throw the sample out because if it takes seven days to get it to
the lab, it takes seven days. We would do well to find out what
the problems along that line are.
Question; I have a question. Does that mean that with a
variable amount of time, we'll have a variable amount of cyanide
with different samples?
Speaker; Possibly, yes.
75
-------
Dept. of Transportation - Shipping Regulations
D. Neptune
U. S. Environmental Protection Agency
Mr. Telliard; This is Dean Neptune. Dean has recently
joined our analytical group. He had to go over and talk with
the Department of Transportation about their shipping regula-
tions. As you know, we have had some shipping problems. Dean
will describe what we're doing in order to insure compliance
with DOT shipping regulations.
Mr. Neptune; Thank you.
As Bill indicated, there have been some problems associated
with the Air shipment of EPA samples. None of these problems
involved the Effluent Guidelines Division priority pollutant,
unextracted samples. The Department of Transportation has a
very low threshold for pain on Government Agencies, or their
agents, not following their regulations. Shipping samples that
do not meet the criteria for samples to be shipped by air is
illegal. That was pointed out, and pointed out very seriously
that we are subject to the same rules and regulations, and the
enforcement of those rules and regulations, as anybody else,
nobody is exempt.
I met with the Department of Transportation to try and deter-
mine if we had a problem, and if we do, how do we overcome any
problems that they might have with our samples. We provided them
with information that was descriptive of the type of samples that
we were taking and the industries in which we were taking them and
the data we collected to date.
This allowed us to narrow down and focus upon particular
concerns. Their concerns revolved around several points. Was
it flammable, poisonous, pathogenic to man or thermally unstable?
They have very specific definitions of what this means, and these
are addressed in a memo that we are sending to all of our EPA
contractors, all the S&A people, and anybody else that may be
involved in the shipment of unextracted water and waste water
samples.
Upon looking at how they have defined these particular para-
meters, I wondered how do we answer whether these parameters have
been met or exceeded, or whether they really apply to us or not.
We entered into discussions on how we could translate these para-
meters into field determinations since exemption was out of the
question. I'll take one for an example, corrosive materials. Any
liquid or solid that causes destruction of human tissue causes a
severe corrosion rate of steel. DOT, through a number of studies,
looked at just about every major class of compounds, at various
concentrations, to determine what is the impact of this particular
material, on steel and aluminum. They established criteria that
if the corrosion rate exceeds 0.25 inches per year, it's a hazar-
dous material because it has violated that criteria.
76
-------
Well, that doesn't help us a lot because we haven't done those
kinds of studies on our wastewaters. So we attempted to translate
the situations that we might have to what might be a violation of
their criteria. With their staff chemists we took each one of
these five parameters and developed a set of working criteria for
Effluent Guidelines Division sample shipment by air. It doesn't
apply to any sample in the EPA, just the Effluent Guidelines
Division, and it's only our water and waste water samples that are
unextracted. For each one of these DOT has set working criteria.
The people that are taking these samples in the field, can either
make measurements or ask questions to determine before the sample
is shipped, if the criteria are violated. If the criteria are
violated the samples can not be shipped as non-hazardous.
This leads to another situation. If sample is not hazardous,
DOT doesn't care about it. They don't provide any labels to so
designate, nor do they want to become involved with these samples.
But if it is hazardous, they want to have a description of what
the hazard is, and how it can be mitigated.
We have to depend a great deal on all the people that are
involved in this sampling process to do what is indicated in this
memo. It is their responsibility that all the people that are
involved in the sampling process, that specifically respond to them,
are aware of this, that they understand it, and that they do imple-
ment it.
In addition there are concerns by some of the carriers, due to
problems such as Bill mentioned, that EPA samples may be hazardous.
In some areas, airlines are reluctant to take EPA samples just
because they're an EPA sample. So, DOT has suggested that we de-
velop a label, an EPA type label, that will indicate if the sample
is not hazardous, and what parameters it has been measured against
to determine this hazard. But as I said before, if it's not
hazardous, they don't care about it, and they don't want to become
involved. So, with our Office of General Counsel, we laid out all
these facts, and tried to come up with something on which we could
all agree. On Page three of this memo, you will find sample label.
It will go on top of each and every sample carrying container.
Sticking this label on the top of the sample carrying con-
tainer does mean that you have to do certain things, and that
you're taking on certain responsibilities by putting this label
on top. So, it is important that you address the questions revolv-
ing around these criteria.
If the sample should fail any one of the criteria that have
been established, it doesn't mean that we should forget that sample.
There are other means of transportation and with appropriate pack-
aging and warnings most could still move by air. The rules and
regulations for surface shipment by motorcar, or rail, are fre-
quently less stringent than by air.
77
-------
If you have any questions at all relating to this memo and
our agreement, call either Gail Goldberg or myself (202/426-7770).
If you have any questions at all relating to the interpretation
of the DOT regulations, call them, not us. Questions?
Question; Does than mean for our metal samples that we cannot
preserve them? I would strongly urge you to make it abundantly
clear that this memo and its criteria are for unpreserved samples.
Mr. Neptune; Okay, I don't know that we can change the memo,
due to the fact that we've had so many people involved in getting
approval of this memo, but we can certainly add a caveat to it when
it goes out relating to Nitric Acid preservation. But samples pre-
served with other than Nitric need only meet the pH criteria.
Gail Goldberg; Dean, the samples for metals analysis are not
to be preserved. I think we've made that very clear.
Mr. Neptune; That's right. The metal's analysis samples are
not preserved prior to shipment, but cyanide and phenols are.
Statement; But as I read this memo, the sampler may assume
that he may now go back to it, and you don't want them to do that.
Mr. Neptune; We can caveat that particular point.
Question; How are the samplers going to measure pH?
Mr. Neptune; We asked them how they wanted it measured, DOT
indicated pH paper would be adequate.
Question; Do you believe DOT understands these criteria?
Mr. Neptune; They wouldn't have signed off on this if they
didn't.
Question; So you're saying then, that for running phenols
you can continue to purposely add acid to the sample, because the
sample pH only goes down to 4?
Mr. Neptune; Yes.
Mr. Telliard; As long as you check and it's 4, not 2.
Mr. Neptune; Yes, but you must check. The EGD memo is not
agencywide. It's not an exception to the DOT regulations but a
means to live within the regulations.
Question; Well, I realize that, but you've got a lot of people
out there taking samples and signing their name on them. Has it
been carefully explained to them that you are going to purposely
be adding acid to the sample?
Mr. Neptune; We told them that we do stablize samples and
we expressly told them what samples and what acid.
78
-------
Mr. Neptune; Okay, the question was, what is the position on
shipping extracted samples? Do you mean like in hexane, for instance?
Questioner; Methylene chloride?
Mr. Neptune; Methylene chloride? These samples can go by
air with an appropriate label, indicating the organic solvent that
is in there, the amount of the organic solvent, and if necessary
an appropriate hazard label. Any hazardous shipment goes only at
the pilot's discretion. In other words, if he says, no, then it's
no. So, you'll just have to wait until you find one that would
carry it. But it must be labeled.
Statement; Dean, It's not just hazardous materials. He has
the right to eject passengers, baggage and whatever?
Mr. Neptune; That is correct.
Question; What happens when he asks these questions and the
guy at the plant says, I don't know, we don't have any data? What
do you do then?
Mr. Neptune; Sometimes, as I pointed out, the determination
can be made by inference, and DOT has what they call the rule of a
reasonable man. In other words, has your waste water ever caught
on fire?
Question; I'm not thinking about that. I'm thinking about if
the sample does not have an anticipated LD50...
Mr. Neptune; Has anybody ever been poisoned by working around
your waste water or having been exposed to it? Has anybody ever
shown any symptoms of toxic exposure?
Question; That isn't what that sentence says.
Mr. Neptune; I recognize that. It says anticipated.
Statement; I don't know that I can answer that.
Mr. Neptune; You couldn't say whether any of your your people
ever become ill from working with and being exposed to your waste
water.
Question; Will a revised interpretation or clarification be
issued for this in the Federal Register?
Mr. Neptune; This is strictly an agreement at this point.
Whether this will be published in the Federal Register or an excerpt
thereto, I can't answer at this time. We are discussing this with
OGC, but the point has been brought up.
Question; Okay, would you address the issue, say you get a
sample that's not air transportable. What are the DOT regulations
for that?
79
-------
Mr. Neptune; Yes, I think that I tried to answer that ques-
tion in the last sentence here. Questions referring to the shipment
of hazardous materials should be directed to DOT, and the carrier.
We're not going to try and do their job for them, because they prob-
ably wouldn't like that, and we probably wouldn't do a good job.
Statement; The pilot's union has a little black book and it is
a listing of substances that the pilot can determine that he doesn't
want. It depends on how faithful a union member is, and if the
label has the name that compares to his black book, which doesn't
have an awful lot of chemicals in it.
Question; Is there any way of shipping the hexane extract?
Mr. Neptune; I don't know if I understand the question. We
have shipped it by air...
Question; Air Express?
Mr. Neptune; Yes, and we have shipped them by Air Courier
International.
Question; Is it preferred to ship sample extracts by surface
transportation?
Mr. Neptune; I don't think so. I wouldn't ship them by
ground. The quicker the better, I would say, and the quickest
is probably air unless you're in proximity of where you're shipping
them. If it's a couple of hundred of miles it's probably just as
fast to go by surface transportation, as it is to go by air.
80
-------
LIST OF ATTENDEES
Jeffrey W. Adams
A. D. Little, Inc.
20 Acron Park
Cambridge, Massachusetts 02140
Dolloff F. Bishop
U. S. EPA
Municipal Environmental Research Lab
Cincinnati, Ohio 47268
Dr. S. C. Blum
Exxon Research & Eng. Co.
P. 0. Box 121
Linden, New Jersey 07036
Robert L. Booth
U. S. EPA
26 W. St. Clair
Cincinnati, Ohio 45268
Theodore J. Bowling
Duke Power Company
Steam Production Dept.
Environmental Lab.
Charlotte, North Carolina
Allen Bradley
U. S. EPA
401 M Street, S. W.
Washington, D. C. 20460
Roderick A. Carr
Versar, Inc.
6621 Electronic Drive
Springfield, Virginia 22151
Mike H. Carter
U. S. EPA
401 M Street, S. W.
Washington, D. C. 20460
S. Charles Caruso
Carnegie-Mellon University
4400 Fifth Avenue
Pittsburgh, Pennsylvania 15213
Robert R. Claeys
NCASI
Engineering Experiment Station
Oregon State University
Corvallis, Oregon 97331
Carl S. Coen
Environmental Protection Agency
llth & Chapline Street
Wheeling, West Virginia 26003
Bruce N. Colby
Systems, Science & Software
Box 1620
LaJolla, California 92038
Charles T. Drevna
National Coal Association
1130 17th Street, N. W.
Washington, D. C. 20036
B. F. Dudenbostel, Ph.D.
Environmental Protection Agency
Region II
Edison, New Jersey 08817
Aubry E. Dupuy, Jr.
EPA - Pesticide Monitoring Lab
NASA/NSTL, Bldg. 1105
Bay St. Louis, Mississippi 39529
Douglas Eisner
Foremost Foods Company
6363 Clark Avenue
Dublin, California 94566
Bill Fairless
U. S. EPA - CRL
25 Funston Road
Kansas City, Kansas 66115
Robert P. Fisher
NCASI
P. 0. Box 14483
Gainesville, Florida 32604
Paul H. Friedman
Versar, Inc.
6621 Electronic Drive
Springfield, Virginia 22151
Robert Greenspun
EPA, Monitoring & Data
Support Division
401 M Street, S. W., WH 553
Washington, D. C. 20460
-------
Nancy Jean Gross
Versar, Inc.
6621 Electronic Drive
Springfield, Virginia 22151
Thomas Gunder
Burns and Roe Industrial
Services Corporation
283 Route 17
South Paramus, New Jersey 07652
Fred Haeberer
EPA Research Lab
College Station Road
Athens/ Georgia 30605
Clarence L. Haile
Midwest Research Institute
425 Volker Boulevard
Kansas City, Missouri 64110
Ernst Hall
U. S. EPA
401 M Street, S. W.
Washington, D. C. 20460
Jack R. Hall
Hydroscience
9041 Executive Park Drive
Knoxville, Tennessee 37919
J. D. Hallett
Shell Oil Company
P. O. Box 2463
Houston, Texas 77001
Philip A. Kamiin
Olympic Research Division
ITT Rayonier, Inc.
409 East Harvard
Shelton, Washington 98584
Carol A. Hammer
RETA/Envirodyne Engineers
12161 Lackland Road
St. Louis, Missouri 63141
Frank Hammer
RETA/Envirodyne Engineers
12161 Lackland Road
St. Louis, Missouri 63141
Stephen C. Havlicek
Georgia Tech
117 Baker Building
Atlanta, Georgia 30332
James E. Henderson
Carborundum Company
3401 LeGrande Boulevard
Sacramento, California 95823
Peter A. Hernandez
Interlake, Inc.
150 West 137th Street
Riverdale, Illinois 60627
Frank Hochgesang
Mobil Research and
Development Corp.
Billingsport Road
Paulsboro, New Jersey 08066
F. N. Hodgson
Monsanto Research Center
Station B, Box 8
Dayton, Ohio 45407
Charles Hunt
U. S. EPA
Region II
Edison, New Jersey 08817
Robert T. Iten
E. I. Dupont De Nemours & Co.
Experimental Station E-336
Wilmington, Delaware 19898
Richard A. Javick
FMC Corporation
Box 8
Princeton, New Jersey 08540
Richard Kearns
Hamilton Standards
Airport Road
Windsor Locks, Connecticut 06096
Larry Keith
Radian Corporation
P. O. Box 9948
Austin, Texas 78766
Robert 0. Kleopfer
U. S. EPA
25 Funston Road
Kansas City, Kansas 66115
William G. Krochta
PPG Industries
Box 31
Barberton, Ohio 44203
-------
Oliver J. Logdson II
U. S. EPA
Box 25227, Building 53
Denver, Colorado 80225
Jim Longbottom
EMSL
26 West St. Clair
Cincinnati, Ohio 45268
Bernard S. MacCabe
Carborundum Company
P. 0. Box 1054
Niagara Falls, New York 14302
David Marrs
Standard Oil of Ohio
Ohio Research Center
4440 Warrenville Road
Cleveland, Ohio 44128
Gerald D. McKee
U. S. EPA
26 W. St. Clair Avenue
Cincinnati, Ohio 45268
John G. Michalovic
CAL Corporation
P. O. Box 235
Buffalo, New York 14221
M. L. Moberg
Analytical Research
Laboratories, Inc.
160 Taylor Street
Monrovia, California 91016
Neil Mosesman
Energy Resources Co., Inc.
185 Alewife Brook Parkway
Cambridge, Massachusetts 02138
Robert Z. Muggli
W. C. McCrone Associates
2820 S. Michigan Avenue
Chicago, Illinois 60616
Janine Neils
Midwest Research Institute
10701 Red Circle Drive
Mannetouka, Minnesota 55343
Dr. Dean Neptune
U. S. EPA
401 M Street, S. W.
Washington, D. C. 20460
James D. Norris
COBA-Geigy Corporation
Analytical & Environmental
Technical Department
P. 0. Box 113
Mclntosh, Alabama 36553
Dr. D. J. Northington
West Coast Technical Service
17605 Fabrica Way, Suite D
Cerritos, California 90701
Roger Novak
Gulf South Research Institute
Box 26518
New Orleans, Louisiana 70186
W. M. Ollison
American Petroleum Institute
2101 L Street, N. W.
Washington, D. C. 20037
Dale D. Olm
Eastman Kodak Company
Building 34, Kodak Park
Industrial Laboratory
Rochester, New York 14650
Jon D. Onstot
Midwest Research Institute
425 Vaker Boulevard
Kansas City, Missouri 64110
Louis J. Papa
E. I. Dupont De Nemours & Co.
Jackson Laboratory
Box 525
Wilmington, Delaware 19899
Kanu C. Patel
Health & Corns. Prods.
Dow Chemical
1200 Madison Avenue
Indianapolis, Indiana 46268
Robert B. Pojasek
Energy Resources Co.
185 Alewife Brook Parkway
Cambridge, Massachusetts 02138
William B. Prescott
American Cyanamid Company
Bound Brook, New Jersey 08805
-------
Dennis G. Revell
U. S. EPA
College Station Road
Athens, Georgia 30602
John E. Riley
U. S. EPA
401 M Street, S. W.
Washington, D. C. 20460
Curtis Ross
U. S. EPA
536 South Clark Street
Chicago, Illinois 60605
Leon E. Rubin
Polaroid Corporation
750 Main Street
Cambridge, Massachusetts 02139
D. R. Rushneck
PJB Laboratories
373 South Fair Oaks Avenue
Pasadena, California 91105
James Ryan
Gulf South Research Institute
P. 0. Box 26518
New Orleans, Louisiana 70186
George J. Sand
EPA-PML
Building 1105
Bay St. Louis, Mississippi 39520
W. M. Shackelford
EPA
College Station Road
Athens, Georgia 30605
James 0. Smith
Allied Chemical Corporation
P. 0. Box 1021R
Morristown, New Jersey 07960
Robert C. Smith, Jr.
Versar, Inc.
6621 Electronic Drive
Springfield, Virginia 22151
Bill Sonnett
Environmental Protection Agency
401 M Street, S. W.
Washington, D. C. 20460
George Stanko
Shell Development Co.
Box 1380
Houston, Texas 77001
Warren C. Steele
Foremost Research Center
6363 Clark Avenue
Dublin, California 94566
Murray Strier
Environmental Protection Agency
Washington, D. C. 20460
Dr. R. F. Stubbemen
Celanese Chemical Company
P. O. Box 9077
Corpus Christi, Texas 78408
John H. Taylor, Jr.
PJB Laboratories
Division of Jacobs Engineering
373 S. Fair Oaks Avenue
Pasadena, California 91105
Paul A. Taylor
CAL Labs
3401 Le Grande Boulevard
Sacramento, California 95823
Kathleen Thryn
A. D. Little, Inc.
20 Acorn Park
Cambridge, Massachusetts 02140
John R. Tilstra
Environmental Protection Agency
Box 25366
Denver Federal Center
Denver, Colorado 80225
Marlene N. Wass
Hydroscience
9041 Executive Park Drive
Knoxville, Tennessee 37919
Joe C. Watt
Catalytic, Inc.
1500 Market Street
Philadelphia, Pennsylvania 19102
John W. Way
E. I. Dupont De Nemours & Co.
Experimental Station, Bldg. 336
Wilmington, Delaware 19898
-------